Trace Chemical Sensing of Explosives

TRACE CHEMICAL SENSING OF EXPLOSIVES Edited by RONALD L. WOODFIN WILEY-INTERSCIENCE A JOHN WILEY & SONS, INC., PUBLICAT...

Author: Ronald L. Woodfin

281 downloads 2316 Views 4MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

TRACE CHEMICAL SENSING OF EXPLOSIVES Edited by RONALD L. WOODFIN

WILEY-INTERSCIENCE A JOHN WILEY & SONS, INC., PUBLICATION

TRACE CHEMICAL SENSING OF EXPLOSIVES

TRACE CHEMICAL SENSING OF EXPLOSIVES Edited by RONALD L. WOODFIN

WILEY-INTERSCIENCE A JOHN WILEY & SONS, INC., PUBLICATION

c 2007 by John Wiley & Sons, Inc. All rights reserved. Copyright Published by John Wiley & Sons, Inc., Hoboken, New Jersey. Published simultaneously in Canada. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permission. Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and speciﬁcally disclaim any implied warranties of merchantability or ﬁtness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Neither the publisher nor author shall be liable for any loss of proﬁt or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic formats. For more information about Wiley products, visit our web site at www.wiley.com. Library of Congress Cataloging-in-Publication Data: Trace chemical sensing of explosives / edited by Ronald L. Woodﬁn. p. cm. Includes bibliographical references. ISBN-13: 978-0-471-73839-8 (cloth) ISBN-10: 0-471-73839-5 (cloth) 1. Chemical detectors. 2. Explosives—Detection. 3. Terrorism—Prevention. I. Woodﬁn, Ronald L. TP159.C46T73 2006 662 .20287—dc22 2006027027

Printed in the United States of America. 10 9 8 7 6 5 4 3 2 1

To all those courageous men and women, worldwide, who put themselves in harm’s way to search for hidden explosives in order that others may not suffer harm. May the concepts discussed herein assist them.

CONTENTS

FOREWORD

xv

PREFACE

xvii

LIST OF CONTRIBUTORS

xxiii

PART I 1

FUNDAMENTAL CONSIDERATIONS

CHEMICAL SENSING

1 3

Ronald L. Woodﬁn

1.1 1.2 1.3

1.4

1.5

What Is Chemical Sensing? / 3 Types of Sensing Systems / 3 Sensing Possibilities / 4 1.3.1 Bulk Sensors / 4 1.3.2 Trace Sensors / 5 Aromas / 6 1.4.1 Biosensors / 6 1.4.2 Electronic Sensors / 10 1.4.3 Other Indirect Methods (Switch of Molecules) / 12 1.4.4 Target Possibilities / 12 1.4.5 Sensitivity and the Problem of False Positives / 13 Conﬁguring an Electronic Trace Sensor / 15 vii

viii

CONTENTS

1.6

2

1.5.1 Required Elements / 16 1.5.2 Integration and Packaging / 18 Issue of Concentration / 18 1.6.1 Nomenclature / 18 1.6.2 Source to Sample / 23 1.6.3 Catch, Count, and Release Cycle / 23 1.6.4 Sensor Sensitivity Versus Sampling Time / 23 1.6.5 The Concentration Gap / 26 1.6.6 Sensitivity Comparison / 27 References / 32

WHAT TO DETECT?

35

Jimmie C. Oxley

References / 41 3

DANGEROUS INNOVATIONS

43

Kirk Yeager

3.1 3.2 3.3 3.4

3.5 3.6

4

Introduction / 43 Theory of Improvised Explosives / 43 History and the Anarchist Literature / 45 Fertilizer-Based IEs / 51 3.4.1 Ammonium Nitrate IEs / 51 3.4.2 Urea Nitrate / 54 Peroxide Explosives / 55 The Next Wave / 63 3.6.1 Improvised Detonators / 63 3.6.2 Peroxide Main Charges / 64 3.6.3 Fringe Mixtures / 65 3.6.4 On the Horizon / 66 References / 67

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES? Ronald L. Woodﬁn

4.1

4.2

Introduction / 69 4.1.1 Where Did the Molecules Come from and How Did They Get Here? / 69 4.1.2 Objects Other Than Buried Landmines / 70 4.1.3 Questions That Beg for Answers / 70 Source of the Molecules / 71

69

ix

CONTENTS

4.3

4.4

4.5

4.6

4.7

5

4.2.1 How the Molecules Diffuse or Leak from a Munition / 71 4.2.2 Example of Landmines / 73 4.2.3 Other Munitions / 76 Transport of the Molecules / 76 4.3.1 Buried Sources / 77 4.3.2 Concentration Estimates from Buried Sources / 94 4.3.3 Other Environments / 96 4.3.4 Odor Plumes / 97 EF&T Implications for Search and Sampling Strategies / 99 4.4.1 Sources Buried on Land / 99 4.4.2 Sources Producing Plumes / 99 Open Questions and Fruitful Areas for Future Research / 101 4.5.1 Objects Buried in the Sea Bottom / 102 4.5.2 Sampling Plant Material / 102 Role of Computer Modeling / 102 4.6.1 Soil Transport Models / 103 4.6.2 Plume Transport Models / 104 4.6.3 Plume Search Models / 104 Conclusions / 104 References / 105

STRUCTURE OF TURBULENT CHEMICAL PLUMES

109

Donald R. Webster

5.1 5.2 5.3 5.4 5.5

Turbulent Mixing / 109 Instantaneous Structure / 111 Time-Averaged Characteristics / 115 Information for Tracking Chemical Odor Plumes / 118 Variation of the Plume Structure / 125 Acknowledgments / 127 References / 127

PART II FIELD EXPERIENCE 6

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT Mark Fisher and Matthew Dock

6.1 6.2 6.3

Introduction / 133 Overview of Fate and Transport of Explosives Released from UUXO / 134 Sampling and Sensing Methodology / 135

131

133

x

CONTENTS

6.4

6.5

6.6

7

SeaDog Sensor Conﬁgurations / 137 6.4.1 Prototype Integrated with a Robotic Crawler Platform / 137 6.4.2 Diver-Deployed SeaDog and Initial Integration with the REMUS / 139 6.4.3 SeaDog Miniaturization: The SeaPup / 142 Results of Sensor Tests Conducted in the Marine Environment / 143 6.5.1 Tests of the Sensor Prototype on a Crawler Vehicle / 143 6.5.2 Tests of the Diver-Deployed SeaDog Sensor and Initial Integration to the REMUS / 145 6.5.3 Tests of the SeaPup Sensor Integrated on the REMUS / 146 Conclusions / 148 Acknowledgments / 149 References / 149

EXPLOSIVES DETECTION USING ULTRASENSITIVE ELECTRONIC VAPOR SENSORS: FIELD EXPERIENCE

151

Mark Fisher

7.1 7.2 7.3 7.4

7.5 7.6

8

Introduction / 151 Relevance of Field Testing To Sensor Development / 153 Overview of the Vapor Signatures of Explosives / 154 Landmine Detection / 158 7.4.1 Introduction to the Mine Problem / 158 7.4.2 Discussion of Landmine Chemical Vapor Signatures / 159 7.4.3 Landmine Detection Field Test Results / 164 Comparison of Fido with Canines Using High-Volume Sampling Methods (REST) / 170 Conclusions / 172 Acknowledgments / 172 References / 172

REFLECTIONS ON HUNTING MINES BY AROMA SENSING Vernon Joynt

8.1 8.2

Editor’s Note / 177 Interview / 177 References / 192

177

CONTENTS

PART III EXAMPLE SENSING TECHNOLOGIES 9

EXPLOSIVES DETECTION BASED ON AMPLIFYING FLUORESCENCE POLYMERS

xi

193

195

Colin Cumming

9.1

9.2 9.3 9.4 9.5 9.6 9.7 9.8 9.9

Introduction / 195 9.1.1 AFP Principle of Operation / 196 9.1.2 AFP Technology / 197 History of AFP and Fido® / 201 Fido Sensitivity / 201 Fido Performance / 202 Performance Limitations / 202 Physical Parameters / 203 Latest Implementations / 203 Maturity of the Fido Technology / 208 Funding / 208 References / 209

10 ION MOBILITY SPECTROMETRY

211

Ronald L. Woodﬁn

10.1 10.2 10.3 10.4

Introduction / 211 Brief Description of Principle of Operation / 211 Some Recent Developments / 214 Some IMS Manufacturers / 216 Acknowledgment / 217 References / 218

11 MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES Jack A. Syage and Karl A. Hanold

11.1 Introduction / 219 11.2 Detection Methods / 220 11.2.1 Explosives Trace Detection / 220 11.2.2 Sampling Methods / 222 11.2.3 Quantitative vs. Screening Analysis / 222 11.3 Mass Spectrometry / 223 11.3.1 Primer / 223 11.3.2 QitTOF Mass Spectrometry / 226 11.3.3 Mass Spectrometry Versus Ion Mobility Spectrometry / 228

219

xii

CONTENTS

11.4

11.5

11.6

11.7

11.3.4 Other MS Analyzers Used for Explosives Detection / 230 Results / 231 11.4.1 Mass Spectral Signatures / 231 11.4.2 MS/MS Analysis / 232 11.4.3 Limits of Detection / 233 Detection Accuracy—A Model / 234 11.5.1 False-Positive Analysis / 234 11.5.2 Receiver Operator Characteristics (ROC) / 236 11.5.3 MS vs. IMS Accuracy / 236 Applications / 240 11.6.1 Personnel Screening / 240 11.6.2 Other Applications / 242 Summary and Conclusion / 242 Acknowledgments / 242 References / 243

12 EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS

245

Thomas Thundat

12.1 Introduction / 245 12.2 Modes of Operation and Theory / 247 12.2.1 Resonance Frequency / 248 12.2.2 Thermal Motions of a Cantilever / 249 12.2.3 Thermal Effects—Deﬂagration / 250 12.3 Apparatus / 250 12.3.1 Cantilevers / 250 12.3.2 Excitation Techniques / 250 12.3.3 Readout Techniques / 250 12.3.4 Selectivity / 252 12.4 Results and Discussion / 254 12.5 Deﬂagration / 256 12.6 Conclusions / 259 Acknowledgments / 259 References / 259 13 LAB-ON-A-CHIP DETECTION OF EXPLOSIVES Greg E. Collins, Joseph Wang, and Christopher A. Tipple

13.1 Introduction / 261

261

CONTENTS

xiii

13.2 Lab-on-a-Chip Explosives Detection by Electrochemical Detection / 265 13.2.1 Amperometry for Explosives Detection / 266 13.2.2 Contactless Conductivity Detection / 268 13.2.3 Dual Amperometric/Conductivity Detection for Simultaneous Monitoring of Ionic and Organic Explosives / 270 13.3 Lab-on-a-Chip Explosives Detection Utilizing Optical Methods / 271 13.4 Lab-on-a-chip Sampling of Explosives / 277 13.5 Conclusions / 281 Acknowledgments / 281 References / 281 14 NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

285

Hedi Mattoussi, Aaron R. Clapp, and Igor L. Medintz

14.1 14.2 14.3 14.4 14.5 14.6 14.7

Introduction / 285 Quantum Dot–Protein Bioconjugates / 286 F¨orster Formalism and Quantum Dots as Energy Donors / 287 Quantum Dots as FRET Donors / 290 Quantum-Dot-Based FRET Nanosensors / 294 Surface-Attached QD–FRET Nanoassemblies / 296 Conclusions / 300 Acknowledgments / 300 References / 300

15 REMOTE SENSING OF EXPLOSIVE MATERIALS USING DIFFERENTIAL REFLECTION SPECTROSCOPY ¨ Rolf E. Hummel, Anna M. Fuller, Claus Schollhorn, and Paul H. Holloway

15.1 15.2 15.3 15.4

Introduction / 303 Differential Reﬂectometry / 304 Results / 305 Conclusions / 309 Acknowledgments / 310 References / 310

303

xiv

CONTENTS

PART IV SUPPLEMENTARY MATERIAL APPENDIX: ORGANIZATIONS INVOLVED IN SEARCHING FOR HIDDEN EXPLOSIVES

311

313

Charles O. Schmidt

International and Nongovernmental Organizations / 313 Commercial Demining / 315 Governments / 317 Military Systems / 318 Equipment / 319 University Research / 321 Information/Data Bases and Links / 322 Emeritus / 323 DEFINITIONS, SYMBOLS AND ABBREVIATIONS

325

Acronyms / 325 Symbols and Abbreviations / 330 EXPLOSIVES DEFINITIONS

331

BIBLIOGRAPHY

333

INDEX

351

FOREWORD

Dr. Woodﬁn and his associates who authored Trace Chemical Sensing of Explosives have performed a useful service for both the scientiﬁc and nonscientiﬁc communities. This volume takes a systems approach to the problem of chemical sensing of explosives. Considering Dr. Woodﬁn’s long background at Sandia National Laboratories in system engineering research, his approach to this volume is not surprising. The volume does not pretend to undertake a detailed, fundamental examination of the elements of sampling, concentration, detection, and understanding and interpretation of the results, which will probably disappoint many individual technical experts. But each technical expert in these individual scientiﬁc ﬁelds of the system should look at the entire system to understand how their individual specialty interacts with other technical specialties to make the total system more effective and all readers will beneﬁt. This volume can best be understood by considering the olfactory senses of animals. Everyone has witnessed the ability of a dog to use its nose with its superb olfactory sense to detect trace odors. A “holy grail” of detection science has been to duplicate a dog’s nose electronically. Dr. Woodﬁn’s volume tries to assess where the scientiﬁc community stands in its search for this holy grail. The authors of this volume are using this chemical sensing approach to detect explosives. Explosives, as used in this context, are deﬁned as materials that concentrate releasable energy; controllable or uncontrollable. A major detriment to successful detection of any explosive in trace concentrations has always been the problem of too many false-positive alarms. This book attempts to compare various chemical sensing approaches to alleviate this longterm bane of detection. The volume compares various approaches to sampling, concentrating, and ﬁnally sensing small amounts of material. The volume also xv

xvi

FOREWORD

attempts to understand how these various components of a sensing system can be optimized into a major detection unit. Dr. Ronald L. Woodﬁn’s unique background makes him highly qualiﬁed to edit a volume of this nature. His career at Sandia National Laboratories, universities, industry, naval laboratories, serving as consultant, in addition to being an ordained minister, makes him highly qualiﬁed to organize such a volume as this. I hope this volume will alleviate or reduce the dangers of hidden mines in future times. Dr. Fred E. Saalfeld Former Chief Scientist, Director, Technical Director, Deputy Chief of Naval Research, and Executive Director of the Ofﬁce of Naval Research (ONR)

PREFACE

WHY THIS BOOK?

For all of recorded history mankind has located and identiﬁed certain items by their aroma. Whether it is a dead mouse in the closet or a freshly baking loaf of bread in the oven, we often make the identiﬁcation correctly without seeing or touching the item. We have considered this sense so useful that when we ﬁnd our own sense of smell to have inadequate sensitivity for a certain task we often “borrow” the more acute sense of smell from some animal. For centuries we have used dogs for hunting and pigs for trufﬂe harvesting. Recent research has been directed toward ﬁnding ways to replace some of these noses, especially the borrowed ones, with electronic devices. In this book we will examine some of the technologies that have been developed and some that are currently being developed. The ﬁeld is progressing rapidly enough that it is likely that other technologies that could be included will be announced while this book is in press. Nevertheless, we will provide here a basis for comparison of technologies, so that one who plans to develop or use a device will be able to make more informed choices. While many of these other applications have great value, this book will focus speciﬁcally on ﬁnding and identifying explosive-bearing objects. As is often the case in the history of technology, work of this nature began in several places at about the same time. Those of us working at Sandia National Labs were asked by a Navy/Marine Corps committee to adapt technology developed for other purposes to their need for better ways to search for mines, both land and sea mines. As we considered technologies currently applied to that effort, it became clear most of them found too many targets. The rate of false positives was so high as to greatly reduce the actual effectiveness of the devices being used. For xvii

xviii

PREFACE

example, a metal detector will ﬁnd most mines, but it also ﬁnds every piece of metallic battleﬁeld debris, bullets, fragments of exploded munitions, and general metallic trash. We recognized the need for a sensor that would focus on the real issue. The objects that one needs to ﬁnd are those that contain explosives! This rather obvious observation led us to begin work on chemical sensors that look for the actual explosive molecules. Undoubtedly, other researchers came to similar conclusions because several programs were begun at about the same time to develop this basic idea. Chapter 4 provides a brief glimpse at the history of some of these programs. Explosives, like many technological developments in history, provide us with a way to concentrate energy to accomplish tasks more quickly. The beneﬁts of explosives in industry, particularly in mining, in construction, and in agriculture, are substantial. However, many people associate explosives most often with military, paramilitary, or terrorist operations. As with any technology that uses high levels of power, explosives can be turned to dangerous, even nefarious purposes. When this happens it is often desirable to locate the explosives before they are actuated. This book explores some of the technologies that have been developed, or currently under development, for the purpose of locating explosives. This exploration concentrates primarily on the search for explosives by portable means. Chapter 9 forms an exception to this rule, describing some technologies that are directed to ﬁxed position screening operations. The group of technologies included herein is by no means exhaustive. They are somewhat representative of the directions that research and development has taken recently. Some of the research has been done or sponsored by government agencies. Some has been commercially led. Some of the work will result, or has resulted, in systems committed to production. Our emphasis is not on products but on technologies. Our anticipation is that work will build on this and products will become available to enhance the search for hidden explosives.

PLAN OF BOOK

This book is composed of four basic divisions. Part I, Fundamental Considerations, includes Chapters 1 through 5 and discusses the problems and issues of ﬁnding and identifying small quantities of explosive materials within a given environment, with the hope of then being able to locate their hidden source. Part II, Field Experience, Chapters 6, 7 and 8, describes actual conditions encountered and procedures used. Part III, Example Sensing Technologies, Chapters 9 through 15, provides descriptions of some of the technologies under development at the time of this writing, and seeks to display some of the many ways that innovative researchers have applied the general principles described in Part I. Chapters 12 through 14 of this part are discussions of sensors at microscale and smaller. Part IV, Supplementary Material, Appendix 5 and following, includes contact information for organizations, deﬁnitions of acronyms and explosives, and a collective bibliography and index.

PREFACE

xix

In all the technology chapters, 9 through 15, the authors were free to organize and discuss their technologies as they considered best. These authors are actively working to develop the technology they are describing. Consequently, it is natural that each may show somewhat more enthusiasm for his own technology than a dispassionate observer reviewing several technologies might show. Competing technologies that are not represented are not omitted in order to favor those that are included. In some cases researchers with other technologies declined to contribute, usually from press of other obligations. The editor takes full responsibility for omissions that are due to his own ignorance and offers apology to those whose work he failed to include for that reason. It was done in ignorance, not in malice. The technologies presented provide a sampling of the possibilities, and an opportunity to examine the systems issues that the developer of any trace chemical sensing system must confront. The editor would welcome contact from researchers who are developing other technologies. The broadest exposure of these new ideas is beneﬁcial to all except the terrorists; perhaps a sequel to this book is in order. WHAT IS IN THIS BOOK?

Part I, Fundamental Considerations Chapter 1, “Chemical Sensing,” introduces the subject with a general discussion of chemical sensing, exploring some of the possible avenues available. This chapter examines types and classes of sensing possibilities. It considers the required elements of a generic system and discusses the relationship of chemical concentration and sensor sensitivity. It includes a ﬁnal section on the nomenclature of concentration and related issues. Chapter 2 “What To Detect?” Serving as a refresher in the chemistry of explosives, this chapter provides a basis for considering what molecules we may need to seek. Chapter 3, “Dangerous Innovations” This chapter continues the discussion of the chemistry of explosives by discussing some nontraditional explosives that are now being encountered in connection with various terrorist and paramilitary groups. It may become necessary to include these chemicals within our universe of compounds being sought, especially in improvised explosive devices (IEDs). Chapter 4, “Where Should We Look for Explosive Molecules?” In this chapter we examine the processes that affect the release of explosive molecules from a munition and their movement to some location where a sensing system may encounter them. The effects of environment and transport through soil, air, and water are explored. Chapter 5, “The Structure of Chemical Plumes,” continues the thrust of the preceding chapter by describing in detail the way small concentrations of chemicals are transported through water or air. Part II, Field Experience. Chapter 6, “Detection of Trace Explosive Signatures in the Marine Environment.” Since the main subject is underwater experience with the SeaDog sensor,

xx

PREFACE

it might sound like a collection of nautical yarns. In reality, it describes the results of applying a trace sensing technology (AFP) in SeaDog for plume tracking in an autonomous underwater vehicle (AUV), named REMUS, in a search for explosives hidden under water. Chapter 7, “Explosive Detection Using Ultrasensitive Electronic Vapor Sensors: Field Experience,” applies some of the concepts of the preceding chapters by describing system development work in ﬁeld conditions. It describes some surprising head-to-head comparisons between dogs and electronic sensors. Chapter 8, “Reﬂections on Hunting Mines by Aroma Sensing” records some of the unique practical experience that Dr. Vernon Joynt has accumulated over many years of laboratory and ﬁeld work. Chapter 9, “Explosives Detection Based on Amplifying Fluorescence Polymers,” describes a relatively new technology with very high sensitivity that has been applied successfully to several real-world search efforts. This is the sensing technology used in Chapters 6 and 7. Chapter 10, “Ion Mobility Spectrometry,” describes one of the most commonly used technologies in chemical sensing systems, with speciﬁc application to explosives, the one perhaps currently most used in portable systems. Chapter 11, “Mass Spectrometry for Security Screening of Explosives” This chapter is a little different from the other technology descriptions since it describes systems that are not portable. These are systems used to locate explosives in containers, or on personnel, that pass by a ﬁxed point. The chapter also compares the features of mass spectrometry (MS) and ion mobility spectrometry (IMS) that often cause systems developers to choose one or the other for a speciﬁc application. Chapter 12, “Explosive Vapor Detection Using Microcantilever Sensors,” is representative of a general class of sensors “Surface Effect Microsensors” that use the change in some mechanical or electrical property of an ultra-small structure to sense and identify a wide variety of molecules. Chapter 13, “Lab-on-a-Chip Detection of Explosives,” describes one particular technology that has been successfully developed in a very small size. It too thus becomes representative of the possibilities that are offered by sensor systems of this genre. Chapter 14, “Nanoscale Sensing Assemblies Using Quantum Dot–Protein Bioconjugates,” also describes a kind of sensor technology in that ultra-small size, but one using a very different approach. Chapter 15, “Differential Reﬂection Spectroscopy for Detection of Explosive Materials,” describes a quite different type of technology that became available just as this book was in press. It is different from all the others described herein because it seeks to remotely locate and identify the explosive molecules in situ, whereas all the other trace sensing technologies require that some molecules be taken into the apparatus, ingested, in order to be sensed. This approach presents exciting possibilities, but is just emerging, with no ﬁeld experience yet.

PREFACE

xxi

Part IV, Supplementary Material, begins with an Appendix, “Organizations Representing Those Who Search for Hidden Explosives,” cataloging some of those organizations, with contact information. Also included are a few pages for quick reference to acronyms and abbreviations, plus one to deﬁne all the explosives mentioned in the text. This section is called Deﬁnitions, Symbols, and Abbreviations. The Bibliography collects all the references cited in the text and adds a few. A Subject Index completes the work.

ACKNOWLEDGMENTS

A number of persons assisted or encouraged me in this effort. My wife, Martha, in addition to patiently enduring months of clutter and my reduced attention to normal household repairs, has proofread the manuscript several times in its various revisions. Her advice was always on track and I appreciate all her help. RADM John Pearson, USN (retired) and Prof. Al Bottoms of the Mine Warfare Association (MINWARA) were instrumental in getting this project started when they provided me an opportunity to chair sessions at the MINWARA conferences in 2002 and 2004. I initiated contact with several contributors as a result. RADM Pearson also got me the opportunity to be invited to lecture at the Naval Postgraduate School in 2003. I am grateful for both those opportunities, doubly so because that gave us the opportunities to visit our grandchildren. Gunner Schmidt, in addition to compiling the Appendix, read and commented on early drafts and helped me ﬁnd references and people. Tom Obrien, my manager at Los Alamos Technical Associates (LATA), where I do occasional consulting, was kind enough to allow me the use of some computer software that I do not own. Paul Cooper of Sandia National Laboratories provided some valuable guidance in ﬁnding a publisher, directing me to Amy Byers, of Wiley, who has been most helpful. Jim Phelan of Sandia National Laboratories was kind enough to review the chapter I wrote about his work, Chapter 5. Since he was unavailable to write it himself, his assistance in “making sure I got it right” is greatly appreciated. The congregation of Cedar Crest Baptist Church has been encouraging, even though most probably regarded this as a distraction. Their patience is appreciated. Finally, I want to thank each contributor, particularly those whom I have never met face to face, for the work they have presented here. It is most obvious that their contributions are the true value of this book.

LIST OF CONTRIBUTORS

Aaron R. Clapp, Ph.D., Naval Research Laboratory Optical Sciences Division, Code 5611 Washington, DC 20376 Greg E. Collins, Ph.D., Naval Research Laboratory Chemistry Division, Code 6112 4555 Overlook Ave., S.W. Washington D.C. 20375-5342 Colin Cumming, MS, Nomadics Inc., an ICx Technologies Company 1024 S. Innovation Way Stillwater, OK 74074 Matthew Dock, Nomadics Inc., an ICx Technologies Company 1024 S. Innovation Way Stillwater, OK 74074 Mark Fisher, Ph.D., Nomadics Inc., an ICx Technologies Company 1024 S. Innovation Way Stillwater, OK 74074 Anna Fuller, Materials Science and Engineering College of Engineering University of Florida P.O. Box 116400 Gainesville, FL 32608 xxiii

xxiv

LIST OF CONTRIBUTORS

Karl A. Hanold, Ph.D., Syagen Technology, Inc. 1411 Warner Ave. Tustin, CA 92780 Paul H. Holloway, Ph.D., Department of Materials Science and Engineering Gale Lemerand Drive P.O. Box 116400 University of Florida Gainesville FL 32611-6400 Rolf E. Hummel, Ph.D., Department of Materials Science and Engineering P.O. Box 116400 University of Florida Gainesville, FL 32611-6400 Vernon Joynt, Ph.D., Force Protection, Inc. 9801 Highway 78, #3 Ladson, SC 29456 Hedi Mattoussi, Ph.D., Naval Research Laboratory Optical Sciences Division, Code 5611 Washington, DC 20376 Igor L. Medintz, Ph.D., Naval Research Laboratory Center for Biomolecular Science and Eng., Code 6900 Washington, DC 20376 Jimmie C. Oxley, Ph.D., University of Rhode Island Department of Chemistry 51 Lower College Road Kingston, RI 02881 Jack A. Syage, Ph.D., Syagen Technology, Inc. 1411 Warner Ave. Tustin, CA 92780 Charles O. Schmidt, CWO-4, USN (Ret.) Los Alamos Technical Associates Inc. 2400 Louisiana Blvd, Suite 400, Bldg 1 Albuquerque, NM 87110 ¨ Claus Schollhorn, Department of Materials Science and Engineering P.O. Box 116400 University of Florida Gainesville, FL 32611-6400 Christopher Tipple, Ph.D., U.S. Department of Defense 200 MacDill Blvd. ATTN: CLAR Washington, DC 20340

LIST OF CONTRIBUTORS

Thomas Thundat, Ph.D., Oak Ridge National Laboratory Nanoscale Science and Devices Group Mail stop-6123, Oak Ridge, TN 37831-6123 Kirk Yeager, Ph.D., FBI/EU 2501 Investigation Parkway Quantico, VA 22135 Joseph Wang, Ph.D., Arizona State University Departments of Chemical & Materials Engineering and Chemistry Ira A. Fulton School of Engineering Director Center for Bioelectronics and Biosensors, Biodesign Institute P.O. Box 876006 Tempe, AZ 85287-6006, USA Donald R. Webster, Ph.D., Georgia Institute of Technology School of Civil and Environmental Engineering 790 Atlantic Drive Atlanta, Georgia 30332-0355 Ronald L. Woodﬁn, Ph.D., Sandia National Laboratories, retired PO Box 55 Sandia Park, New Mexico 87047

xxv

PART I

FUNDAMENTAL CONSIDERATIONS

CHAPTER 1

CHEMICAL SENSING RONALD L. WOODFIN Sandia National Laboratories (retired)

We do a lot of chemical sensing. Mostly we do it without special equipment. But, whether it is the pleasant notiﬁcation of tortillas toasting or bread baking or the jarring realization of a skunk or a cigar on the golf course, sensing, analyzing, and cataloging chemical signals are a part of our daily lives. We also use other means for chemical sensing. When we travel, we judge the local water hardness by the amount of soap we need to wash or shampoo. We use test kits to measure the chemistry of a swimming pool or a ﬂower bed. In this chapter we will examine some basic ideas of chemical sensing as they apply to ﬁnding explosives. 1.1

WHAT IS CHEMICAL SENSING?

When we speak here of chemical sensing we mean the direct sensing of chemicals, rather than sensing of something else by means of chemicals. By that choice we will not consider those techniques often called “wet chemistry.” We then have a fairly clear idea that we mean to search for a speciﬁc chemical or suite of chemicals. Sometimes the search is for single chemicals like chemical warfare agents such as mustard or sarin. At other times combinations of chemicals, such as found in marijuana, are the object of the search. A short digression to consider the broader range of techniques and targets is appropriate in order to limit more precisely the scope of this book to the speciﬁc type of search indicated by its title. We broadly classify the possible types of sensing system according to activity and according to focus. 1.2

TYPES OF SENSING SYSTEMS

First, we may class sensors according to the activity required of the sensing system. Hence a system may be passive, active, or semiactive. Passive sensors Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

3

4

CHEMICAL SENSING

do not produce any emanations directed toward the target; they are based on sensing something emanated or released by the target without stimulation by the sensing system. Conversely, active sensors need no emanation or release of anything from the target; they provide some sort of emission that interacts with the target. The active sensor then senses the alteration in its emission by the target. Semiactive sensors operate in yet another way. They provide a stimulation that produces some sensible emission or emanation from the target, which they then sense in the manner of a passive sensor. 1.3 1.3.1

SENSING POSSIBILITIES Bulk Sensors

There are two basic ways to look for explosive material. They differ in their point of focus. Some sensors seek the mass of explosive material within a device. These are particularly useful when the device is well sealed and its surface is well cleaned of stray explosive molecules, or when the explosive being used is nonaromatic, that is, it does not readily release molecules from its bulk. We will refer to these as bulk sensors. They include X-ray techniques, both transmission and backscatter; neutron activation in several techniques; γ -ray excitation, in either transmission or backscatter modes; and nuclear resonance techniques, either nuclear magnetic resonance (NMR) or nuclear quadrupole resonance (NQR). Bruschini [1] has described these thoroughly. They are also described by the staff of the Jet Propulsion Laboratory [2]. The following forms a very brief synopsis. 1.3.1.1 X-ray Techniques X-ray techniques are familiar because of their use in medical diagnosis. The basic concept is that material of different densities or chemical compositions absorb and scatter X-rays differently. When the X-rays pass through the materials and strike the ﬁlm or detector, they form a gray-scale image. After proper calibration a bulk charge of explosive may be inferred from this image. In many applications, such as seeking buried munitions, it is not possible to place the X-ray source and the detector on opposite sides of the objects being investigated. In this case techniques have been developed to form images from the X-rays that are scattered back toward the source, or backscattered. 1.3.1.2 Neutron or γ Sensors Neutron- or γ -based sensors are similar in concept to the X-ray sensors. They use different forms of excitation and different detectors, but the basic forms of transmission or backscatter follow the pattern described above. Both normally rely on extensive computation for signal processing called computed tomography, where the detector signals are combined to synthesize an image of the irradiated object. 1.3.1.3 Electromagnetic Techniques Sensors based on the related principles of NMR and NQR have been successfully developed. These are active techniques that excite the electromagnetic interactions between the atomic nuclei

SENSING POSSIBILITIES

5

by means of an external magnetic ﬁeld. In NMR the magnetic ﬁeld is a dc, or ﬁxed, ﬁeld, whereas in NQR it is an oscillating, or ac, ﬁeld. The detector is tuned to respond to resonant frequency of a molecule of interest, such as RDX [3]. Both portable and ﬁxed systems have been demonstrated. It may be possible that ground-penetrating radar (GPR) could be tuned to preferentially indicate masses of particular chemicals, but such work is not known to have been reported. 1.3.1.4 Bulk Sensor Targets All the bulk sensor technologies have a common thread. They seek to ﬁnd a mass of material with certain physical properties and to distinguish its shape when other materials obscure it. The object may be buried in the ground or contained within a vehicle, a structure, or container such as a crate or luggage. Potential targets are mines, unexploded ordnance (UXO), improvised explosive devices (IEDs), or drug caches. We are perhaps most familiar with the medical search for tumors or foreign objects within a body using computer-aided tomography, or CAT scan, and magnetic resonance imaging (MRI). 1.3.2

Trace Sensors

Bulk sensors certainly have a role in chemical sensing of explosives, but the subject of this book is the other basic type sensor, one that seeks molecules released from the bulk of the explosive material in an object. We will refer to these as trace chemical sensors. They are sometimes called vapor sensors, but that seems a less accurate description when they are applied to explosive molecules, which may not always be found in a vapor state. As we shall see in Chapter 5, that requires us to understand where and how to look for these molecules. It will become apparent upon a little reﬂection that the two types of sensors are complementary and are best used in different situations. Furthermore, even when trace sensors are used, in some situations sampling of particles of soil or vegetation or sampling from surfaces may prove to be more productive that vapor sampling. For underwater sources the term vapor sensing is also inappropriate. In fact, with the very recent addition of differential reﬂection spectroscopy (DRS) to the suite of applicable technologies, as described in Chapter 15, we now have the possibility of sensing trace quantities of explosives where they are most often found in the environment, adsorbed to solid surfaces. Technologies that can, like DRS, locate these traces in situ offer a very different way to approach the problem. There have been several recent attempts to do this in situ detection from some distance away. To date the DRS seems the most successful. It has demonstrated detection at a range of a few meters. The marked advantage of an in situ detection system is that it does not ingest the molecules; hence it does not disturb the area, does not require as close approach to an explosive device, and it should produce a faster response time than ingesting systems. The disadvantage of in situ systems is also that they do not ingest the molecules; hence they have no means of concentrating the sample

6

CHEMICAL SENSING

for sensing. Thus they must have greater direct sensitivities in order to produce comparable results. This book was in press before the ﬁrst publication of the DRS results. Therefore, the following chapters were written with the expectation that most sensor systems would require ingesting samples. This condition is expected to continue, since in situ sensors will most likely supplement, not replace, ingesting sensors. Nevertheless, the discussion applies to the in situ systems by recognizing the high proportion of molecules that are normally adsorbed to surfaces, such as dust, plants, manmade objects, and the like. These form the target for the in situ systems. Trying to apply in situ systems like DRS to vapors and plumes is unlikely to lead to success; but, when vapors and plumes strike solid objects, there is a marked tendency for the entrained molecules to adsorb to the surface, producing opportunity for the in situ sensor. 1.4

AROMAS

The ability to recognize, locate, or distinguish among speciﬁc materials by their characteristic aromas can be useful in a variety of applications. Our ﬁrst warning of a ﬁre is frequently the smell of smoke. Wine and tea tasters regularly use their sense of smell to assist in quality control of these food items. Similarly, when we buy fruits such as peaches, we often sniff them to judge quality. We become aware of poor water quality before tasting the water because it “smells bad.” It is well known by medical personnel that certain diseases and conditions can be diagnosed quickly from the patient’s aroma. Farmers sometimes judge when to harvest from the aroma of the produce. We often see television news reports of trained dogs searching for contraband illegal drugs. Trained soldiers can recognize some chemical agent attacks from the ﬁrst whiff of the agent, hopefully in time to protect themselves. Dogs and other animals are being effectively used in searching for hidden explosives in a variety of situations, including humanitarian demining and antiterrorist patrols. In each of these situations, sensing the aroma, which, for the purposes of this book, we will consider to consist of a speciﬁc molecule or suite of molecules that is uniquely produced by its source, provides the means of identiﬁcation and/or location of the source. In each of these examples the user would gain signiﬁcant advantage if a very sensitive and speciﬁcally “tuned” electronic sensor, which could accurately and reliably identify the characteristic aroma for that application, were available. 1.4.1

Biosensors

The historical use of the biologically based sensors mentioned above might lead us to the conclusion that ﬁnding hidden sources of explosives simply means training some kind of animal. Indeed, that conclusion has merit, and a great deal of success has been recorded in that way. This is discussed somewhat more fully in Chapter 8. There are, of course, disadvantages to this technique. The

AROMAS

7

possibilities of biosensors have certainly not been exhausted in the search for explosives or other trace chemicals. As an aside, designed to pique a hope of even broader application for technologies developed primarily to search for explosives, consider a recent ﬁnding by researchers at Amersham Hospital in the United Kingdom. [4] They report success in using dogs to detect cancers. For centuries people, particularly nurses, have been aware that certain conditions, for example, gangrene, could be detected and diagnosed by smell. This latest ﬁnding offers hope that research will make us able to apply electronic vapor sensing to medical diagnosis as well as our immediate objective of ﬁnding hidden explosives. There have been efforts directed toward using bacteria in the search for explosives1 [5]. Whether this approach offers a potential for developing practical sensing systems based on bacteria is an open question. 1.4.1.1 Mammals Perhaps because we have a long history of domesticating mammals for various tasks, the preponderance of explosive search to date has been by mammals, mostly dogs [6, pp. 165–174], though some work has been done using rats [6, pp. 175–193] and pigs [7]. There have been two principal approaches used, which may be characterized as direct and indirect search. These techniques are discussed in detail in McLean [6]. Direct Mammalian Search In the direct search technique the animal, usually a dog, is brought to the suspected location of the hidden explosive. The animal is trained to go to the strongest scent of the aromas presented during training. The animal is trained to adopt a particular behavior when nearest the source. With dogs, this is usually to sit near the source. This technique has the advantage of immediate results; hence it is the method most often adopted by military personnel or police forces when the threat is immediate. This direct approach has several distinct disadvantages as well. When animals such as dogs are used in a ﬁeld environment, they become subject to the hazards of that environment. Approaching an explosive device puts both animal and handler in the danger zone. Of course, in such a zone there are often related dangers. Also, because of the danger, the animal is restrained, usually by training, from actual contact with the source. Hence there is always an area of uncertainty surrounding the actual location of the source, particularly a buried one. Animals ﬁnd a multitude of interesting distractions in a ﬁeld environment. Since they operate in a world of aromas undetectable to us, it is hard to realize how many distractions there may be. Certainly food aromas and sexual stimulation aromas are present. Animals are subject to physical ailments, just as we are. A dog with the equivalent of a “head cold” is not as effective in search. Providing proper care for the animal in such locations often becomes difﬁcult and expensive. Training is also time consuming and expensive. In spite of all 1 http://www.ornl.gov/info/ornlreview/meas tech/threat.htm and http://eerc.ra.utk.edu/insites/ins54.htm#critters. Both sites visited 9/28/05.

8

CHEMICAL SENSING

these difﬁculties, the use of dogs in direct search for hidden explosives remains one of the most effective, arguably the most effective, method for ﬁnding hidden explosives. The limit of detection2 (LOD) for explosive searching dogs has been a subject for active debate for years. Chapter 8 includes some discussion on this matter. One report [8] attempts, with partial success, to quantify this LOD. Indirect Mammalian Search In the indirect method, called in general terms the REST concept, for remote explosive scent tracking [6, pp. 53–107, 9], samples of chemicals in air are trapped in special ﬁlter cartridges by passing large volumes of air from the search area through them. In principle, the concept could be extended to water also. The ﬁlter cartridges are then taken to a laboratory, or other protected area, where they are offered to trained mammals, sometimes dogs and sometimes rats, for snifﬁng. When the animal detects the aroma for which it has been trained, it gives the proper response. The sample is correlated with its collection location so that more detailed sampling or direct search can be initiated. This technique was ﬁrst developed in the late 1980s in South Africa, largely through the work of Dr. Vernon Joynt and colleagues at Mechem Consultants, a division of Denel (Pty) Ltd. They called it MEDDS (Mechem Explosives and Drug Detection System). The technique has been called EVD (Explosives Vapour Detection) by Norwegian People’s Aid (NPA) and Checkmate elsewhere. The term REST is now applied generically. REST was conceived for solving the most pressing question in Humanitarian Demining, or in any postconﬂict ordnance cleanup operation: “Where are there no explosives?” The most immediate need is always to identify those areas that are free of explosives, so that they can quickly be returned to normal activity, such as agriculture and transportation. Then the detailed search and removal efforts can be concentrated on the areas that actually contain explosives. REST has been used with success in several postconﬂict areas, especially in Africa. A typical REST application mounts an apparatus on a vehicle that can move safely through the suspect area. The apparatus has a vacuum pump that pulls air through the removable ﬁlter cartridge. A material for which explosive materials have an afﬁnity is chosen for the ﬁlter. That is a material on whose surface the explosive molecules tend to tightly adhere. Most often, that tendency is a function of temperature, so the molecules adhere more tenaciously at lower temperatures and release as the temperature is increased. This characteristic is one often exploited in electronic sensing systems as well. When a certain region, say a predetermined length of a particular road, has been sampled, the ﬁlter cartridge is removed and stored. A new ﬁlter cartridge is inserted and the next area sampled. Filter cartridges are then offered to the trained mammal in a controlled environment. This technique is essentially similar to the one used with the 2 We

can conveniently refer to the smallest quantity of any given compound that an animal or a technology can detect as its LOD.

AROMAS

9

cancer seeking dogs [4]. Large areas can be surveyed quickly, safely, and inexpensively. REST has also been effective in searching for contraband at border crossings. REST need not necessarily involve any animals at all. If electronic sensors of adequate sensitivity are developed, they can replace the animals. Certainly, the history of electronic instrument development shows that the earlier generations of any device are more suited for laboratory than ﬁeld use, and that laboratory units can normally be expected to show better performance than portable ones. Calibration of an electronic instrument, which corresponds to training of a mammal, should become more precise and dependable than that training. 1.4.1.2 Insects It has long been recognized that many insects may have a more acute sense of smell than any mammal. A consensus found in articles by many researchers seems to indicate that only a few molecules are required for odor detection by some insects. We have little experience in domesticating insects. Usually, we simply gather the ones we deem useful, mostly bees and silkworms (the larvae of an insect, the Bombycid moth, Bombyx mori ), to a convenient location and let them do their normal activity. We then take their produce when it is available. Even within this limited process we can ﬁnd ways to use them for chemical search. Researchers have found that bees can be trained to seek out sources of explosive molecules by the simple expedient of feeding them sugarwater laced with the target molecules [10]. Training bees is thus much simpler and quicker than training mammals. Bees have an obvious advantage when entering a mineﬁeld or similar danger area. They are too small to actuate any sort of mine, so they are completely safe. Thus, they have one of the most desirable characteristics for sensing systems. They do not put the searchers at risk. With an appropriate method for tracking the bees and geographically ﬁxing the points where a concentration of foraging bees indicates a source of explosive molecules, a mineﬁeld can be mapped for efﬁcient search and removal. Such techniques have been offered [10]. After discovering that bees tend to forage along a particular vector from the hive, the researchers found it possible to “aim” the bees in a general direction and track them with a LIDAR3 system developed at Sandia National Laboratories [11]. Recently, wasps, Microplitis croceipes, have been conditioned to detect explosives [12]. Training is similar to that of the bees, but instead of allowing the wasps to range freely, a handheld, air-ingesting sensor has been developed with the wasps as the sensing element, with a video system for monitoring behavioral indications of detection [13]. The device, called Wasp Hound, was developed originally for other purposes, primarily agricultural, but has been successfully demonstrated in detecting 2,4 DNT. No detailed information on ﬁled LOD for explosive compounds is yet available, but comparison with one commercial electronic sensor, Cyranose 320, [14] seems to indicate somewhat better LODs for some chemicals. Thus, Wasp Hound can be classiﬁed as a hybrid sensor, part biological and part electronic. 3

LIDAR: light detection and ranging.

10

1.4.2

CHEMICAL SENSING

Electronic Sensors

The term electronic is used here in the most generic sense. Since the objective is to ﬁnd and identify small quantities of explosive molecules by extracting them from their environment, the problem is analogous to many solved with electronic devices. Common examples include radio and television receivers and electronic instruments for measuring physical quantities such as weight, acceleration, or temperature. The principal difference here lies in the sensing modules that form part of the systems we will consider. Of course, it is conceivable that a system can be developed that uses no electronics, perhaps using color changes, or something else, but the systems discussed herein all use some electronic modules to enable the observer to know when explosive molecules have been located. Whatever system may be developed, the basic objective is the same. Different technologies may be employed, but the reason for the system is always to locate and identify explosive molecules in air, water or soil, or on a surface, in order to pinpoint the source of these molecules. Ideally, the source is located before it can be actuated in a harmful way. Consequently, every system developed for this purpose will have some characteristics in common with any other developed for the same purpose. Therefore, it is valuable to consider those common elements in a general way. We are faced with locating molecules that are sparse within the environment, hence the term trace. There are several basic steps that are necessary, independent of the design of the speciﬁc system or the technology being incorporated. Table 1.1 lists those steps and correlates them with more common electronic instrumentation nomenclature. Some of the actions will involve TABLE 1.1 Common Processes in Locating and Identifying Explosive Molecules within the Environment Step Sample

Action

Introduce a quantity of explosive-bearing medium into the system Separate Pass the sample through some process that preferentially selects the explosive molecules Concentrate Collect the separated molecules while additional sampling and separation continues Identify and quantify Apply some signal processing procedure or algorithm to infer the in situ concentration of explosive molecules in the samples Present Organize the results in a form that can be interpreted by an operator or that can be used to guide further activities in autonomous devices

Electrical Analog Receiver or transducer Tuner

Filter/ampliﬁer

Signal processor

Display

AROMAS

11

physical processes, such as pumping air or water past a collector. Some may use chemical processes, such as tagging the explosive molecules with other molecules that can be sensed more easily or sensitively. Some will use electronic processing; some may use radiation, such as ultraviolet light. The variety of possibilities has not yet been fully explored. Several types of systems have been, or are being, developed. The following descriptions are presented as a brief summary. Detailed descriptions will be found in the chapters devoted to each technology. 1.4.2.1 Spectrometry “Spectroscopy4 is basically an experimental subject and is concerned with the adsorption, emission or scattering of electromagnetic radiation by atoms or molecules” [15, p. 1]. A wide variety of applications of this concept have been applied in analyzing many substances. In the particular case of explosive molecules the most prominent are several forms of mass spectrometry and ion mobility spectrometry. Each has certain advantages and disadvantages. Each is discussed in detail in a later chapter. The former is most often used in ﬁxed applications; the latter, in both ﬁxed and portable applications. 1.4.2.2 Surface Effect Sensors Surface effect sensors have been offered in several forms. In each design they rely on using a substrate, often a microstructure, coated with a material having an afﬁnity for adsorbing speciﬁc molecules. When the molecules adhere to it, they alter some property of the substrate. Depending on the particular concept the property exploited may be physical or electrical. Some designs measure altered resistance; some, altered capacitance. Some designs measure alterations in physical effects, such as the frequency of vibration of a microcantilever or the frequency or amplitude of an induced surface acoustic wave (SAW) [16]. Sensors of this type tend to be very good at identifying a broad range of chemicals, particularly complex aromas that contain suites of compounds. They have been used to sort fruits for ripeness and can distinguish among fruits. These sensors use an array of sensing elements, each coated with a different material, to adsorb different molecules. They then form signature patterns for different aromas in a calibration process not completely unlike a dog’s training. Signal processing algorithms then match patterns of sampled aromas to known patterns for identiﬁcation. Acoustic wave devices have been used, primarily as electronic bandpass ﬁlters, for more than 60 years, but the ﬁrst reported use as a chemical sensor appeared in 1979 [16, p7]. A form of surface effect sensor that exploits altered surface resistance, or chemiresistors, forms the surface from a mixture of tailored polymers and a ﬁnely divided conductive material, such as carbon black, as a thin ﬁlm on a substrate. They use a number of polymers, 32 in one implementation, with different properties to form an array of chemiresistors. When a vapor is passed over the array, 4 Spectroscopy

uses photographic means to record the spectrum; spectrometry uses photoelectric recording [15, p. 67].

12

CHEMICAL SENSING

the ﬁlm swells. Each ﬁlm swells differently. The pattern of altered resistances forms a signature for that vapor. Sophisticated signal processing matches these signatures to known aromas, much as ﬁngerprints are matched. Chemiresistors have been successfully employed to detect undesirable odors in food packaging materials and to distinguish among ﬁve varieties of diesel fuel [17]. When the design focus becomes this kind of speciﬁcity among aromas, the sensitivity for simple compounds such as explosives may be reduced. A design focused directly on explosive molecules will likely prove more adaptable to the search for hidden explosives. Chapter 12 describes one of these sensors in detail. To date none of these sensors has produced an LOD adequate for ﬁeld search for explosives. 1.4.2.3 Amplifying Fluorescence Polymers (AFPs) The development of materials that ﬂuoresce intensely in the presence of ultraviolet light when no nitroaromatic explosive compounds are present, but are prevented from ﬂuorescing when those compounds are introduced, has led to the development of a new class of sensors. Upon encountering a nitroaromatic molecule, the tailored ﬂuorescing polymer binds with it. The bound pair no longer ﬂuoresces when struck by a photon. This behavior is exhibited by a class of polymer compounds called chemophores, each having an afﬁnity for a particular class of analytes. The characteristic that distinguishes AFPs from other chemophores is their forming of long chains with as many binding sites as molecules forming the chain. A single explosive molecule, binding at any one of these sites, is sufﬁcient to induce quenching of the ﬂuorescence along the entire chain; thus, the appellation of “amplifying.” The intensity of the quenching produced by the binding of one explosive molecule is thus multiplied many times. Very low concentrations of explosive molecules are thus detectable. Chapter 9 describes the way AFPs have been incorporated into a family of ﬁelded sensing systems. Chapter 6 examines the experience gained in using one of this family in an unmanned underwater vehicle (UUV) called SeaDog. Chapter 7 describes ﬁeld experience with AFPs in air, as implemented in the Fido® sensor. 1.4.3

Other Indirect Methods (Switch of Molecules)

Methods similar to AFPs in that they introduce other molecules have been suggested. The concept is that molecules that are easier to detect than explosive molecules, but only when the target molecules are present, can be introduced to increase the sensitivity of the detector system. With the introduction of AFPs that concept is clearly proven. Other adaptations are to be expected in the future. 1.4.4

Target Possibilities

When we consider possible targets in the context of system design we may group them into three broad classes, all of which may be considered as unexploded ordnance, UXO. Classes 1 and 2, and sometimes all three, are termed explosive remnants of war, ERW by the United Nations, UN.

AROMAS

TABLE 1.2

Possibility Space for System Indication of Detected Explosive

System Indication

Explosive No Explosive

13

Actual Condition Explosive Present True Positive False Negative

No Explosive Present False Positive True Negative

Class 1 includes any munition, whether standard or IED, that is deployed with the intent of causing damage to an opponent during a belligerency of any nature. These form two subclasses, those that are “fresh” and those that are abandoned. Clearly, those fresh ones constitute the most urgent targets. Those that are abandoned may be functional, as with mines, or malfunctioning units. These “duds” may still constitute an immediate danger, depending on the reason for their malfunction. Class 2 includes those munitions that have become lost or discarded during military operations or transport. Major quantities of ordnance are lost when ships are sunk or blow up. When munition storage areas are attacked, some munitions are ejected to substantial distances. These munitions may be fully assembled and functional, as when an aircraft ejects them. Others may be incomplete, likely unfuzed. Class 3 includes those ordnance items that are intentionally buried, dumped at sea, or otherwise disposed of either in an intact or in a partially assembled condition. It also includes military or civilian storage and military practice ranges. Each class has distinct characteristics that should be considered in optimizing a system design. The level of urgency varies greatly. The consequences of false negatives also vary greatly. The mention of false negatives begs a bit of explanation. During a search for hidden explosives the sensing system continually operates within a two-by-two space of possibilities, as diagrammed in Table 1.2. When the system provides the correct indication, whether positive or negative, it is functioning properly. However, there are two different error conditions that can be realized. These are indicated by the shaded blocks. A false negative means a missed target. Obviously, that can lead to immediate danger for the operator or others. A false positive induces a different kind of danger, deferred danger, due to loss of operator conﬁdence in the system. 1.4.5

Sensitivity and the Problem of False Positives

One of the primary characteristics of any sensor is its sensitivity. In principle, increased sensitivity should lead to increased performance, that is, a greater likelihood of ﬁnding the object being sought. However, increasing the sensitivity in a cluttered environment often leads instead to an increase in “false positive.” A false positive occurs when the sensor system correctly detects an item of the

14

CHEMICAL SENSING

class it is designed to detect, but an item that is not the item being sought. For example, when a metal detector is used to search for buried landmines, it also detects odd bits of metallic debris in addition to the mines. This can greatly impede the search for the mines. It produces deferred danger when the operator, having encountered several false positives, becomes skeptical of a true positive. One solution is to reduce the sensitivity of the sensor. Such a reduction in sensitivity may also reduce the probability of detecting the mines. Therefore, in any search the probability of detection, PD , must be balanced against the probability of false positive, PFA . This is discussed in a little more depth in Chapter 11. For some sensors seeking explosive molecules, an increase in sensitivity can lead to the detection of molecules other than explosives. These can be interpreted as false positives. Clearly, if there is unburned explosive material present in an area of interest, then the trace chemical sensor will register a false positive. Similarly, some bulk chemical sensors may provide false positives when they encounter a mass of nitrogen-rich material, such as fertilizer or feces. In either case the rate of false positives is still likely to be less that the false-positive rate of a metal detector. Adjusting the sensor sensitivity is usually only partially successful. There is another way to overcome the problem of false positives. A suite of sensors that are “orthogonal in principle” can be used together. 1.4.5.1 Orthogonality In geometry lines or planes are called orthogonal when they meet at a right angle. Values or conditions on one are therefore independent of those on the other. We consider two sensors to be orthogonal when they depend on different principles of chemistry or physics. In that sense, a chemical sensor is clearly orthogonal to a metal detector. When searching for hidden, explosive-bearing objects, false positives can be dangerous, as ﬁnding too many can cause an operator to lose concentration. Therefore, when one needs to identify an object detected by any sensor, that means that indications from two or more sensors, each of which exploits a different physical or chemical property of the target, need to be compared before identiﬁcation can be considered deﬁnitive. For example, a set of three sensors that are orthogonal in this sense is illustrated in Figure 1.1.

Metal Detector (Senses changes in an induced electric field caused by the presence of metal)

Visual Sensor (Senses the shape, size, texture, and appearance of an object)

Trace Chemical Sensor (Senses presence of specific molecules)

Figure 1.1

Orthogonal sensor suite.

CONFIGURING AN ELECTRONIC TRACE SENSOR

15

It is normally advisable to adhere to the principal of orthogonality in some way. In the above example the three sensors could be independently applied, not necessarily part of a single package or system, especially if the object in question is where it can be seen directly. 1.5

CONFIGURING AN ELECTRONIC TRACE SENSOR

Before anyone begins development of a sensor system, the ﬁrst task is to determine what is being sought. That is not necessarily the procedure when developing a sensing technology. Thus, it is common for a technology to ﬁnd its most effective application in systems that were not originally envisioned. Perhaps the most apparent need that led to the development of some of the trace chemical sensing technologies was the need for a method of warning troops, or populations, when they were being attacked with chemical weapons. While this still forms a major application of these technologies we will consider one that was not as apparent originally. The original need that led to development of trace chemical sensors for explosives was the need to restore land that had been abandoned to public or private use. This land was abandoned because of the presence of, or perception of the presence of, mines or unexploded ordnance, often called UXO. These potentially lethal items could be the result of some earlier armed conﬂict. In that case it is now common to refer to them as explosive remnants of war, or ERW. In some cases the war that left its remnants was concluded many years ago. Dangerous ERW are still found on World War I battleﬁelds, and occasionally on U.S. Civil War battleﬁelds, though the latter munitions are usually deteriorated beyond holding any dangerous explosive residue. UXO are also found on former military training facilities, such as Ft. Ord, California, or Kahoolawe, Hawaii. Both these sites have undergone restoration in recent years. Since September 11, 2001, the realization that explosive devices might be deposited almost anywhere by terrorists or other malefactors has added a new group of devices to the list of targets. These are improvised explosive devices, or IEDs. An IED may take any form, subject only to the imagination of its fabricator. Many contain traditional munitions but are constructed to permit delivery and/or actuation in nonstandard, or improvised, ways. Examples are “dud” artillery shells or aviation bombs that have been salvaged and equipped with a fuze for use as a mine or booby trap. Car bombs are also common examples of IEDs. When IEDs are constructed using salvaged munitions, they will naturally include conventional explosives; hence they may be appropriate targets for trace chemical sensors. However, IEDs are not always made in this way. Chapter 3 discusses a variety of nontraditional explosives that are being discovered in IEDs. In this book we will focus our attention on applications that require portable sensors, sensors that must be transported to the vicinity of the target. These applications include mine and UXO removal and the search for and identiﬁcation of IEDs. Chapter 11 forms a notable exception to this focus, where ﬁxed screening of airport portals is the principal application to date. Fixed sensing stations

16

CHEMICAL SENSING

are considered more fully by Oxley [18]. Our focus is due to the very different conditions in the two applications. We will devote a chapter, Chapter 4, to the peculiar subject of the environment fate and transport, or EF&T, of that elusive quarry, the free explosive molecule. These free molecules are the trace chemicals that we are trying to sense. We will explore how they become free, the environmental factors that affect their mobility, and where we may expect to ﬁnd them in sufﬁcient concentration to be detectable.

1.5.1

Required Elements

There are some common elements that any trace chemical sensing system must have. Each system will have a deﬁned purpose and be subject to speciﬁc size, weight, power, mobility, performance, and cost constraints. These, along with the imagination of the system designer, will steer the designs of different systems in individual ways, but the basic elements will remain, though their forms may be very different among a group of systems. The terminology used here is that commonly used in electronic instrumentation, but in some cases the element may have no electronic components. The elements are described in sequential order. In each case, except as noted, the function of each follows that of the preceding one. Conceptually, one samples, then concentrates, then senses, and so forth. No inference about the movement of material from one stage to the next is implied. That movement is a unique property of the system design. 1.5.1.1 Sampling Unit Since we are seeking the free molecules that form a trace chemical “signal,” we must have a unit that collects them from their environment. For some systems that environment is air, for others water or soil. For some, it may be a surface, such as a parcel, a vehicle, a leaf, or a wall. It will be necessary for the system designer to devise a way to collect the molecules from that environment of interest. While most of the system elements are more or less independent of the search environment, the sampling unit is entirely dependent on it. Sampling units may be as simple as a piece of chemically treated paper used to “swipe” a surface or one with multiple parts such as pumps, valves, scoops, and heating units. It may be physically integrated into the rest of the system or operated separately, as in the case of the swipe paper. The sampling unit will always be required to accept some of the environment along with the quarry molecules. Because these bits of the environment obscure the trace signal, it is necessary to sort through the sample in some way to collect as many as possible of the quarry molecules together while excluding as much as possible of the environment. This process is the function of the concentrating unit. In many practical systems the sampling unit and the concentrating unit may be so integrated as to be physically inseparable. Nevertheless, their functions remain conceptually distinct.

CONFIGURING AN ELECTRONIC TRACE SENSOR

17

1.5.1.2 Concentrating Unit As described in the preceding paragraph, the concentrating unit sorts through the sample taken by the sampling unit to extract the molecules of interest. Sections 1.6.2 to 1.6.4 discuss some of the considerations that determine the design of a sampling unit and a concentrating unit. Concentrator materials of choice are often polymers. Polydimethylsiloxane (PDMS), or a polydimethylsiloxane/divinylbenzene copolymer (PDMS/DVB) are favored choices for explosive molecules. PDMS are often used in the form of solid-phase microextraction (SPME) ﬁbers. PDMS/DVB is often used in the form of microspheres with diameters in the 50- to 75-μm range. Detailed considerations for use of SPME ﬁbers is given on a website maintained by the University of Western England [19]. It references a more complete treatise [20]. Other geometries, such as stacked spheres, have also been used successfully [21]. Some concentrating units use chemical solvents; some use mechanical methods, sometimes coating a surface with a material for which the quarry molecules have a distinct afﬁnity. One of the most convenient characteristics of explosive molecules is the way temperature affects their adhesion to surfaces. They adhere readily to cool surfaces but are easily released by a modest rise in temperature. Concentrating units often exploit this characteristic by alternately chilling and warming a collection surface. The surface is chilled while sampling and warmed for sensing. 1.5.1.3 Sensing Unit The heart of the sensor system is the sensing unit. It is here that the wide variety of sensing technologies discussed in Chapters 8 through 14 are employed in different systems. As systems are designed and integrated, the various units may be merged physically, but the sensing unit is where the quarry molecules are identiﬁed as present or not in a concentrated sample. The success of the entire system is determined by the performance of the sensing unit, but does not function without the other units. We often associate the overall performance of the system with the sensitivity of the sensing unit. Design of the other units can either enhance or degrade that overall performance by large factors. Nevertheless, the search for more sensitive sensing technologies is ongoing and needed. Along with sensitivity, the other dominant characteristic of a sensing technology is speciﬁcity. A technology that is very sensitive but lacks speciﬁcity leads to many false positives, or a high probability of false positive (sometimes called false alarms), PFA ; one that lacks sensitivity provides a lower probability of detection, PD . False positives cause the search to become diverted in an attempt to positively identify each one. They slow the search and can become dangerous if they obscure the real target or lull an operator into complacency. A low PD becomes dangerous when a potential hazard, such as a mine, is left undetected. A design that maximizes the PD while minimizing PFA is the goal for sensing technologies. 1.5.1.4 Ampliﬁer Some systems may use an ampliﬁer. This unit may use a process of chemical replacement or reaction to produce secondary molecules from the few quarry molecules collected. These molecules may be more numerous,

18

CHEMICAL SENSING

easier to detect, or exhibit a characteristic that permits a different principle of detection to be employed. This kind of ampliﬁer may precede the sensing unit. Another kind of ampliﬁer may be used, one that simply increases the intensity of the sensing unit output when a quarry molecule is detected. This type ampliﬁer will follow the sensing unit. 1.5.1.5 Signal Processor Electronic instruments normally employ signal processing of some form. Sometimes the calculations are quite extensive and complex. Often the object is to extract a signal from its background. In our case, that is also the function of the concentrating unit. However, it may be possible to further improve the system performance by some form of signal processing at this stage. This process is sometimes termed chemometrics. 1.5.1.6 Output and/or Display Unit Finally, the results of the process to this point must be formatted in a manner appropriate to the purpose of the system, so that some action can be taken. If the system has a human operator the display may be presented in any number of visual and/or audible forms. The operator, upon recognizing the message of the display, decides on his next course of action. If the system is a part of an unmanned or robotic vehicle, the output may be in the form of a digital value or a ﬂag, depending on whether the decision algorithms are contained within the vehicle or are part of the signal processing of the sensor system. 1.5.2

Integration and Packaging

There are always practical considerations when designing a system. Some of these relate to optimizing its performance, and some derive from its anticipated ﬁeld service environment. In many instruments increased sensitivity to the target often corresponds to increased sensitivity to environmental effects such as temperature, humidity, electromagnetic interference, shock, and vibration. It sometimes requires heroic measures to mitigate these effects. Systems can become large or clumsy. The systems we are principally considering, portable, ﬁeld-operable trace chemical sensing systems, have some common constraints. Portable systems must operate on a portable power source. Often this simply means batteries. When heating becomes a part of the operating sequence, batteries can be consumed quickly. A well-deﬁned energy budget is necessary early in the design process. Portable systems are often introduced into small spaces in search of the quarry. They also tend to be handled rather roughly by their operators. Current trends in microminiaturization offer relief to the system designer in both these areas. 1.6 1.6.1

ISSUE OF CONCENTRATION Nomenclature

In this book we will often consider very small quantities of materials. In an effort to ensure understanding, Table 1.3 presents the notation used for mass.

ISSUE OF CONCENTRATION

TABLE 1.3 Abbreviation g mg μg ng pg fg ag

19

Mass Nomenclature Name gram milligram microgram nanogram picogram femtogram attogram

Scientiﬁc Notation 100 g 10−3 g 10−6 g 10−9 g 10−12 g 10−15 g 10−18 g

Decimal Notation 1g 0.001 g 0.000001 g 0.000000001 g 0.000000000001 g 0.000000000000001 g 0.000000000000000001 g

1.6.1.1 Concentration Most often these mass terms will be used to describe concentration of the chemical of interest in some medium, usually air, water, or soil. When we talk of concentration, we have in mind some sort of normalization, so instead of describing an absolute quantity we describe the quantity of the chemical of interest contained in a ﬁxed quantity of the containing medium. When that medium is air or water, we most frequently use a speciﬁed volume; when it is soil, we use a speciﬁed mass. Thus we will use terms like ng/L (nanograms of chemical per liter of air or water) in air or water and ng/g (nanograms of chemical per gram of soil) in soil, since soil varies in density, depending on weather and compaction. It is quite common practice to express concentrations as a ratio, using the “parts per” method of describing the ratio. Thus, in soil the ratio is simple, 1 ng/g, or (10−9 g/g), is a ppb, or one part per billion5 . In water or air, the issue is a bit more complex since it is necessary to reconcile the units of mass per unit volume to units of mass per unit mass. We make this reconciliation by dividing by the density of the ﬂuid. For water this is, for purposes of this book, taken as 10−3 grams per liter. Table 1.4 illustrates this reconciliation. It may be worthwhile to illustrate the actual meaning of these concentrations: 1 ppb concentration in soil means that there is one gram of chemical contained within one billion grams (1000 tonnes) of soil. The implication is that the chemical is distributed more or less homogeneously within the medium. In that case, how much soil is one billion grams? Using a nominal value of 1.4 for the speciﬁc gravity of soil, we ﬁnd that this comprises about 650 m3 of soil, which would ﬁll about a 100 large dump trucks. Similarly, 1 ppb in water implies that a gram of chemical is dissolved in the quantity of water contained in an Olympic-size pool, 50 m long by 10 m wide by 2 m deep. One ppt means the gram of chemical is dissolved in 1000 of these pools. Since the parts per nomenclature for air would be expected to be a volume/volume ratio, in order to reconcile these we must account for the molecular 5

Note the use of common U.S. practice, deﬁning billion as 109 ; other terms follow the same pattern.

20

CHEMICAL SENSING

TABLE 1.4

Reconciling Concentration Nomenclature in Soil and Water

Parts per. . . Nomenclature

In Soil

Parts per . . .

1 part in:

Abbreviation

million billion trillion quadrillion

106 109 1012 1015

1 1 1 1

ppm = ppb = ppt = ppq =

Mass/Mass 1 1 1 1

μg/g ng/g pg/g fg/g

In Water Mass/Volume

Mass/Mass

1 1 1 1

1 1 1 1

mg/L μg/L ng/L pg/L

μg/g ng/g pg/g fg/g

TABLE 1.5 Molecular Weights and Vapor Molar Densities for Some Common Explosives and Associated Compounds Compound DNB DNT TNB TNT RDX PETN Tetryl

168 182 213 227 222 316 287

Molecular Weight

Mass of Single Molecule (g), m0

Molar Density (Vapor) (g/L)

[22, [23, [22, [24, [24, [24, [24,

2.8 × 10−22 3.0 × 10−22 3.6 × 10−22 3.8 × 10−22 3.7 × 10−22 5.3 × 10−22 4.8 × 10−22

6.87 7.44 8.71 9.28 9.08 12.92 11.74

pp. 3–43] p. 1180] p. 3–66] p. 125] p. 125] p. 125] p. 125]

weight of the explosive molecule. Thus we divide the mass per mole6 for the molecule of interest by the molar volume. The result may logically be termed the molar density. First, however, since, as we shall see in Chapter 4, the search for explosive molecules at 0◦ C is normally futile, we will adjust the molar volume to a temperature more often encountered, 25◦ C. The molar volume at 25◦ C and 76 cm Hg is 24.45 liters: mass ratio volume molecular weight = molar volume molecular weight (g) = 24.45 L

Molar density =

(1.1)

Table 1.5 lists the molecular weights and the calculated molar densities for some of the most common molecules of interest. 6 A mole contains 6 × 1023 molecules. It has mass in grams numerically equal to the molecular weight. The molar volume is the volume occupied by a mole of any gas, measured at standard temperature and pressure (0◦ C, 101.325 kPa = 76 cm Hg). It has a value of 22.414 L [22, pp. 1–19, 1–5, 1–14, 1–16].

21

ISSUE OF CONCENTRATION

TABLE 1.6

Correlation of Concentration Nomenclatures

Mass/Volume

TNT in Air

TNB in Air

DNT in Air

DNB in Air

1 1 1 1

108 108 108 108

115 115 115 115

135 135 135 135

146 146 146 146

mg/L μg/L ng/L pg/L

TABLE 1.7 Parts per . . . 1 1 1 1

ppm = ppb = ppt = ppq =

ppm ppb ppt ppq

ppm ppb ppt ppq

ppm ppb ppt ppq

ppm ppb ppt ppq

Reconciling Concentration Nomenclaturesa TNT in Air 9.3 9.3 9.3 9.3

μg/L ng/L pg/L fg/L

TNB in Air 8.7 8.7 8.7 8.7

μg/L ng/L pg/L fg/L

DNT in Air 7.4 7.4 7.4 7.4

μg/L ng/L pg/L fg/L

DNB in Air 6.9 6.9 6.9 6.9

μg/L ng/L pg/L fg/L

a Since parts per must remain dimensionless, this table actually contains “volume/volume ratios” even though the concentrations are stated in mass/volume units. The correction is made using Eq. (1.2).

With the molar densities we can calculate the volume/volume ratios in air. Equation (1.2) [25, p. 10] yields the data tabulated in Table 1.6: ppt =

1000cA molar density

(1.2)

where cA is the concentration in ng/L. It is often convenient to make the correlations “in the other direction.” This is shown in Table 1.7, where the data are essentially the reciprocals of the correlations shown in Table 1.6, but can be calculated directly from Table 1.5 values. 1.6.1.2 Minimum Possible Concentrations Obviously, the absolute minimum possible concentration of any chemical within any medium occurs when there is only one molecule of the subject chemical in the sample. Knowledge of this concentration will provide a recognizable lower limit to possible sensing. We can ﬁnd that concentration [25, p. 11] as follows: If one gram of soil contains a single molecule of the subject chemical of molecular mass m0 (Table 1.5), then the mass/mass concentration is numerically equal to m0 . For example, using TNT (trinitrotoluene) at molecular weight (MW) of 227,

m0 =

227 grams per mole = 3.78 × 10−22 g 6 × 1023 molecules per mole

(1.3)

22

CHEMICAL SENSING

yields the mass/mass concentration for one TNT molecule in one gram of soil as c0 = 3.8 × 10−22 g/g. We can express this in terms from Table 1.3 in several ways: c0 = 3.8 × 10−22 g/g = 3.8 × 10−10 × 10−12 g/g = 3.8 × 10−10 pg/g = 3.8 × 10−13 ng/g = 3.8 × 10−7 fg/g

(1.4)

By reference to Table 1.4 we can also use the parts per nomenclature, recognizing that 1 ng/g is 1 ppb, and so forth: c0 = 3.8 × 10−22 g/g = 3.8 × 10−16 ppm = 3.8 × 10−13 ppb = 3.8 × 10−10 ppt = 3.8 × 10−7 ppq

(1.5)

We ﬁnd similar results in water if we consider one milliliter of water to contain one molecule. Since the 1 mL of water has a mass of 1 g, the mass/mass concentration is again c0 = 3.8 × 10−22 g/g, numerically the same as the mass/volume concentration, 3.8 × 10−22 g/mL. However, expressing this concentration in the units previously used, g/L, yields cL = 3.8 × 10−22 g/mL = 3.8 × 10−19 g/L = 3.8 × 10−7 pg/L

(1.6)

This concentration describes a concentration of one thousand molecules per liter. Alternatively, considering the case of one molecule in a liter of water, the concentration is then cL = 3.8 × 10−22 g/L. As before, this can also be expressed in several ways: cL = 3.8 × 10−22 g/L = 3.8 × 10−9 × 10−13 g/L = 3.8 × 10−13 ng/L = 3.8 × 10−10 pg/L = 3.8 × 10−7 fg/L

(1.7)

Or, from Table 1.4, using the mass/mass concentration, for the 1000 g of water in a liter: cL = 3.8 × 10−22 g/L = 3.8 × 10−25 g/g = 3.8 × 10−13 × 10−12 g/g = 3.8 × 10−13 pg/g = 3.8 × 10−13 ppt

(1.8)

ISSUE OF CONCENTRATION

23

Therefore, on the basis of one molecule of the target compound present in a standard sample, a gram for soil, or a liter for water or air, we can calculate the minimum possible concentration for that compound in that size sample of soil water or air. Table 1.8 presents those results. This provides an objective lower bound for the LODs. 1.6.1.3 Flux Rates We use the term ﬂux to describe the movement of material. Flux rate is then the quantity of material moved por unit time. In the literature we ﬁnd rates quoted in different time units, sometimes seconds, sometimes minutes, or hours, or days. In particular, ﬂux rates are sometimes presented in μg/cm2 -day and sometimes in fg/cm2 -sec. We can simplify this a bit by normalizing these to cm2 . Thus, Table 1.9 provides a correlation of the rates. 1.6.2

Source to Sample

There are a wide variety of possible sources of explosive molecules. No matter what that source is, or where it is located, a similar process may be recognized. First, the molecules must become free from the bulk of explosive at the source. Then they must travel through the local environment to a location where they are accessible to the sampling unit. This journey is full of pitfalls that stop many of the released molecules, so the initial small ﬂux rates may be reduced even more. Chapter 4 discusses the process in considerable detail. 1.6.3

Catch, Count, and Release Cycle

A useful strategy adapted in many sensing systems is to use what we might term “minibatch” processing. By this we mean that the sampling unit takes in material for a short period of time, delivering this “batch” of material to the concentrating unit. The batch may be quite small in order to enable more or less continual motion of the system in search. The batch is delivered to the concentrator where the quarry molecules are collected. Then, through either thermal or chemical cycling, the concentrated molecules are stripped off the concentrator and sent to the sensing unit. While this process is proceeding, another batch may be being collected by the sampling unit. Properly timed, such a strategy can increase overall response rate substantially. 1.6.4

Sensor Sensitivity Versus Sampling Time

Concentration is a most important issue. Each of the sensing technologies discussed in this book, and any other conceivable sensing technology, has some limit of detection, below which it will not provide reliable results. But for most sensors this sensitivity threshold, the lowest concentration of a given molecule that can be reliability detected, is not a static value. The threshold is also a function of the sampling time, in the most general sense. Heuristically, we consider that a certain number of molecules need to be ingested by the sampling unit to

24

Volume/volume, from Eq. (1.2).

a

−13

3.8 × 10 3.8 × 10−10 3.8 × 10−13 3.8 × 10−13 3.8 × 10−13 4.1 × 10−11

ng/g ppt ng/L ppt ng/L ppt

g soil g soil L water L water L air L aira

1 1 1 1 1 1

TNT

Units −13

3.6 × 10 3.6 × 10−10 3.6 × 10−13 3.6 × 10−13 3.6 × 10−13 4.1 × 10−11

TNB 3.0 × 10 3.0 × 10−10 3.0 × 10−13 3.0 × 10−13 3.0 × 10−13 4.0 × 10−11

−13

2.8 × 10 2.9 × 10−10 2.8 × 10−13 2.8 × 10−13 2.8 × 10−13 4.1 × 10−11

−13

3.7 × 10 3.7 × 10−10 3.7 × 10−13 3.7 × 10−13 3.7 × 10−13 4.1 × 10−11

−13

Concentration per Quantity of Medium DNT DNB RDX

Minimum Possible Concentration, Represented by One Molecule of Explosive

Medium

TABLE 1.8

5.3 × 10 5.3 × 10−10 5.3 × 10−13 5.3 × 10−13 5.3 × 10−13 4.1 × 10−11

−13

PETN

4.8 × 10−13 4.8 × 10−10 4.8 × 10−13 4.8 × 10−13 4.8 × 10−13 4.1 × 10−11

Tetryl

25

1 103 106 109 1.16 × 10−5 1.16 × 10−2 1.16 × 101 1.16 × 104

fg/sec

TABLE 1.9

10 10−3 1 103 1.16 × 10−11 1.16 × 10−8 1.16 × 10−5 1.16 × 10−2

−6

−3

10 1 103 106 1.16 × 10−8 1.16 × 10−5 1.16 × 10−2 1.16 × 10−1

ng/sec

pg/sec 10 10−6 10−3 1 1.16 × 10−14 1.16 × 10−11 1.16 × 10−8 1.16 × 10−5

−9

μg/sec

Correlation of Flux Rate Nomenclature (per cm2 )

4

8.64 × 10 8.64 × 107 8.64 × 1010 8.64 × 1013 1 103 106 109

fg/day 1

8.64 × 10 8.64 × 104 8.64 × 107 8.64 × 1010 10−3 1 103 106

pg/day

−2

8.64 × 10 8.64 × 101 8.64 × 104 8.64 × 107 10−6 10−3 1 103

ng/day

8.64 × 10−5 8.64 × 10−2 8.64 × 101 8.64 × 104 10−9 10−6 10−3 1

μg/day

CHEMICAL SENSING

Sensitivity

26

Sampling Time

Figure 1.2 Conceptual relationship between sensitivity and sampling time for a ﬁxed probability of detection and ﬁxed sampling rate.

provide the sensing unit with enough product to form an acceptable probability of detection, PD . A balance thus needs to be struck between the quantity sampled and the sensitivity of the sensing unit. For a given sampling rate and a given PD , the relationship is visualized as in Figure 1.2. In any real sensing operation the sensing system is either being moved to search different locations or different objects are being brought to the system for examination. In either case the system will operate in a sequential batch mode. That is, the system will cycle through the series of actions described in Table 1.1. It is necessary to process through the entire sequence of actions in order to extract a determination of each location or object. Sampling time is only one part of the sensing cycle. Detectors or sensing units must be purged of the molecules collected in the previous cycle before a new cycle can continue. Some systems may be conﬁgured with multiple, alternating, sensing units in order to reduce or eliminate the “dead time” in the sensing cycle. Operational application and economics often determine trade-off choices between sensitivity and sampling time. Since sensing units have historically tended to increase in cost and fragility with sensitivity, increasing sampling time may offer a desirable option for some system designs. 1.6.5

The Concentration Gap

As we shall see in Chapter 4 the anticipated concentration of explosive molecules in many search situation, such as for buried landmines, may be very low, perhaps 1 pg/L (or 100 ppq, or 1 in 1013 molecules). Most sensing systems are not capable of detecting such low concentrations directly. Hence there usually exists a gap between the available sensitivity of existing systems and our perceived needed sensitivity.

ISSUE OF CONCENTRATION

Projected IMS & AFP

Buried Munitions Vapor AFP

IMS

QitTof MS

SAW

27

Chemiresistors

Explosive Vapor Concentration 10 ag/L

1 part in: 1018

10 fg/L

10 pg/L

10 ng/L

10 mg/L

1015

1012

109

106

ppq

ppt

ppb

ppm

Figure 1.3 Illustration of concentration gap.

There are three potential ways to reduce this gap: (1) The basic sensitivity of the sensing unit could be increased or (2) the degree of concentration accomplished in the concentrating unit could be increased. While it is easy to make a statement like that, actually increasing the performance of either unit will require a great deal of innovative research and development. Alternatively, since explosive molecules tend to stick to solid surfaces such as structures, vegetation, or soil particles, (3) a sampling process that extracts them from those locations, rather than directly from the air, could exploit that natural concentration. See the discussion of this in Chapter 8. Figure 1.3 illustrates the current situation. Figure 1.3 diagrams the problem faced by system designers. If the sensitivity of the sensing unit is insufﬁcient to detect the available concentration of explosive vapor there is a concentration gap. This requires the system to include a concentrator unit to bridge the gap. In the example shown, assuming a part per quadrillion target vapor source, those sensing systems that have lower sensitivities fall to the right of the source concentration and must take action to concentrate the sample in order to successfully detect the target. 1.6.6

Sensitivity Comparison

Many different technologies have been adapted to construct detection systems for ﬁnding explosives. The result is a variety of instruments being developed or marketed. The potential user can become bewildered by the variety. Several reviews have been published to categorize the technologies and the available instruments. The following information is extracted from four [26–29] of those reviews. This provides a representative, though by no means exhaustive, comparison of the limit of detection7 (LOD) of these technologies. Three of the four references describe the technologies and present a reported [26] or advertised [27, 28] 7

As earlier noted, we refer to the smallest quantity of any given compound that a technology can detect as its LOD. The concept of LOD for mass spectrometery is more rigorously deﬁned in Chapter 11, Section 11.4.3. The concept is used here in a parallel way for the other technologies, without specifying measurement procedures.

28

2004 2004 2004 2003

Ion trap ion mobility spectrometer

Nonlinear dependence of ion mobility

Thermo-redox

Electrochemicalb

2004 2004

2004

Gas chromatography/mass spectrometer

Ion mobility spectrometer

2004

Gas chromatography/differential ion mobility spectrometer

Gas chromatography/surface acoustic wave

2004 2004

Gas chromatography/chemiluminesence

2004

Chemiluminesence detector a.k.a. thermal energy analyzer

Chemical reagent-based (color change)

2005 2005

2005

Ion mobility spectrometer

Quadrupole ion trap—time-of-ﬂight mass spectrometer

2005

Amplifying ﬂuorescent polymer

Lab-on-a-chip/high-performance liquid chromatography

Published

Electrochemical

TR

NLDIMS

ITIMS

IMS

GC/SAW

GC/MS

GC/DMS

GC/CL

Color

CL or TEA

QitTof MS

LOC/HPLC

IMS

AFP

Abbreviation

Published or Advertised Limits of Detection for Sensing Technologies

Technology

TABLE 1.10

1000

1000

1000

1000

215

LOD (ppt)

1000

1

1

0.001

1

1

1000

0.9

0.05

0.001

LOD (pg)

[29, p. 37]

[28, p. 42]

[28, p. 42]

[28, p. 42]

[28, p. 42]

[28, p. 42]

[28, p. 42]

[28, p. 42]

[28, p. 42]

[28, p. 42]

[28, p. 42; 23, p. 34]

Syage, 11.3.4

Collins, 13.4

Woodﬁn 10.2

Cumming, 9.3

Sourcea

29

0.25

1000 10000

100

10 10

LOD (ppt)

20 200

0.5

10 0.01

50

500

1 10000 1000000

LOD (pg) p. p. p. p. p. p. p. p. p. p. p. p.

37] 37] 37] 26] 26] 26] 26] 26] 26] 40] 66] 54]

[26, p. 60] [26, p. 73] [26, p. 73]

[26, p. 53]

[29, [29, [29, [27, [27, [27, [27, [27, [27, [27, [26, [26,

Sourcea

When a name and section number is quoted, the quoted datum is found in that section of this book. Otherwise, the datum is to be found on the given page in the noted reference work. b Presented as a general class, rather than for any speciﬁc system. c Text mentions only AFP.

a

ASGDI MS/MS ECNIMS IMS BL

1996 1995 1993 1989

Fluorescent Piezoelectric Spectroscopic FIS GC/ECD GC/IMS GC/MS IMS TR GC/SAW IRS API/TOF MS

2003 2003 2003 1999 1999 1999 1999 1999 1999 1998 1997 1996

Fluorescentb,c Piezoelectricb Spectroscopicb Field ion spectrometer Gas chromatography/electron capture detector Gas chromatography/ion mobility spectrometer Gas chromatography/mass spectrometer Ion mobility spectrometer Thermo-redox Gas chromatography/surface acoustic wave Infrared spectroscopy Atmospheric pressure ionization time-of-ﬂight mass spectrometer Atmospheric sampling-glow discharge-ion trap-tandem mass spectrometer Electron-capture negative-ion mass spectrometry Ion mobility spectrometer Bioluminescence detector

Abbreviation

Published

Technology

TABLE 1.10 (continued )

30

CHEMICAL SENSING

ppt 0.00001 0.0001 AFP (2005)

0.01

0.001

0.1

1

10

100

1000

ppq

10000 100000 1000000 1E + 07

ppb

ppm

GC/MS (2004) API/TOF MS (1995) IMS (2005) ASGDI MS/MS (1996) QitT of MS (2005) Fluorescent (2003) GC/CL (2004) GC/DMS (2004) IMS (2004) ITIMS (2004) BL (1989) IRS (1997) ECNIMS (1995) IMS (1999) FIS (1999) GC/ECD (1999) IMS (1993) GC/IMS (1999) Color (2004) Electrochemical (2003) GC/MS (1999) LOC-HPLC (2005) CL (2004) GC/SAW (2004) NLDIMS (2004) Piezoelectric (2003) TR (1999) TR (2004) GC/SAW (1998) Spectroscopic (2003)

Figure 1.4

Normalized limits of detection reported or advertised.

LOD for each. MacDonald et al. [29], however, aggregate all trace detection technologies into four broad categories. The rated technologies are presented in Table 1.10. In many reports and advertisements the LOD is presented as a range of values. In each of those cases the most favorable value offered is shown in Table 1.10. However, some are presented in “parts per” nomenclature and some in mass nomenclature. In order to present all in a similar form Figure 1.4 presents them normalized. Normalizing required an assumption that may not be valid in every case. It was assumed that

ISSUE OF CONCENTRATION

31

Average Volume of Air Required in Sample to Detect Buried Land Mine (L) [Explosive Concentration ~ 0.9 pg/L] ppt 0.0001

0.001

0.01

0.1

1

10

100

1000

AFP (2005) GC/MS(2004) API/TOF MS (1995) IMS (2005) ASGDI MS/MS (1996) QitTof MS (2005) Fluorescent (2003) GC/CL (2004) GC/DMS (2004) IMS (2004) ITIMS (2004) BL (1989) IRS (1997) ECNIMS (1995) IMS (1999) FIS (1999) GC/ECD (1999) IMS (1993) GC/IMS (1999)

Figure 1.5 Expected volume of air sample required to detect landmine signature.

when a LOD was quoted as a mass, that that mass of explosive was contained in one liter of air. No attempt was made to normalize all to the same explosive molecule, although it is recognized that there may be as much as an order of magnitude between the LODs of two different molecules using the same detector. Therefore, Figure 1.4 should be used with caution as a relative indicator of LODs. Furthermore, the author apologizes for the omission of those technologies not included; there are constantly emerging technologies that have not been included. A further caution is offered. Since the LODs are based on published or advertised values, there is no way to know whether the one who published or advertised the value was being conservative or optimistic in the disclosure. Likely, there are some of each in the group. 1.6.6.1 Required Sampling Volume Based on this normalization, we can get some idea of the quantity of air that needs to be sampled in order to detect a given concentration of explosive. Of course, this can only be an average value estimate. For an example we can use that concentration predicted for the TNT vapor signature of a buried landmine as given in Chapter 4, Section 4.3.2, Table 4.5, cm ≈ 10−3 ng/L. To estimate the necessary volume of air containing TNT at concentration cm that must be sampled by a system with a given LOD to detect the TNT, we merely divide the LOD of that system by the concentration, cm , as in Eq. (1.9):

32

CHEMICAL SENSING

Vol (L) ≈

LOD LOD (ng) = ≈ 1000 LOD (L) cm (ng/L) 9.4 × 10−4

(1.9)

Sample volume estimates for the technologies reporting the better LODs listed in Table 1.10 are shown in Figure 1.5. Examination of Figures 1.4 and 1.5 indicates the signiﬁcant progress made in lowering the LODs in the last 5 years or so. Only two technologies reported before 2000 are in the most sensitive group. If progress continues at this rate, within another 5 years trace chemical sensing technologies should be able to detect buried or hidden explosives much more reliably and quickly.

REFERENCES 1. Bruschini, C. Commercial Systems for the Direct Detection of Explosives (for Ordnance Disposal Tasks), ExploStudy, Final Report, Ecole Polytechnique Federale de Lausanne, Lausanne, Switzerland, Feb 17, 2001. 2. Sensor Technology Assessment for Ordnance and Explosive Waste Detection and Location, JPL D-11367, rev B, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, March 1995. 3. Hibbs, A. D., G. A. Barrall, P. V. Czipott, A. J. Drew, D. Gregory, D. K. Lathrop, Y. K. Lee, E. E. Magnuson, R. Matthews, D. C. Skvoretz, S. A. Vierk¨otter, and D. O. Walsh. Detection of TNT and RDX Landmines by Stand-off Nuclear Quadrupole Resonance, preprint given author by Andy Hibbs, Quantum Magnetics Inc., San Diego, CA, 1999. 4. Willis, C. M. Olfactory detection of human bladder cancer by dogs: Proof of principle study. Br. Med. J. 329 (7468), 712–714 (2004). 5. Fliermans, C. B. and G. Lopez-de-Victoria. Microbial Mine Detection System (MMMDS), A. C. Dubey, J. F. Harvey, and J. T. Broach, Eds. in Proceedings of the SPIE 12th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, Detection and Remediation Technologies for Mines and Minelike Targets III, April 13–17, 1998. 6. McLean, I. G., Ed. Mine Detection Dogs: Training, Operations and Odour Detection, Geneva International Centre for Humanitarian Demining (GICHD), Geneva, 2003. 7. Townsend, J. Pigs, a demining tool of the future? J. Mine Action 7.3, 43–46 (2003). 8. Phelan, J. M. and J. L. Barnett. Chemical Sensing Thresholds for Mine Detection Dogs, SAND2002-0692C, Sandia Laboratories Report, Albuquerque, NM, 2002. ˚ 9. McLean, I. G., H. Bach, R. Fjellanger, and C. Akerblom. Bringing the mineﬁeld to the detector: Updating the REST concept, Proceedings of EUDEM2-SCOT–2003, Vol. 1, 2003, pp. 156–161. 10. Bromenshenk, J. J., C. B. Henderson, R. A. Seccomb, S. D. Rice, R. T. Etter, S. F. A. Bender, P. J. Rodacy, J. A. Shaw, N. L. Seldomridge, L. H. Spangler, and J. J. Wilson. Can Honeybees Assist in Area Reduction and Landmine Detection?, J. Mine Action 7.3, 2003. http://maic.jmu.edu/journal7.3/bromoersheuk/bromenshenk.htm Visited 8/9/06.

REFERENCES

33

11. Bender, S. F. A., P. J. Rodacy, R. L. Schmitt, P. J. Hargis, Jr., M. S. Johnson, J. R. Klarkowski, G. I. Magee, and G. L. Bender. Tracking Honey Bees Using LIDAR (Light Detection and Ranging) Technology, SAND2003-0184, Sandia National Laboratories Report, Albuquerque, NM, 2003. 12. Tomberlin, J. K., M. Tertuliano, G. Rains and W. J. Lewis, Conditioned Microplitis croceipes Cresson (Hymenopteria: Braconidae) Detect and Respond to 2,4 DNT: Development of a Biological Sensor, J. Forensic Sci., Sept. 2005, Vol. 50, No. 5, 5 pages. Paper ID JPS2005014, Available online at: www.astm.org 13. Rains, G. C., S. L. Utley and W. J. Lewis, Behavioral Monitoring of Trained Insects for Chemical Detection, Biotechnol. Prog. 2006, 22, 2–8. 14. Rains, G. C., Tomberlin, J. K., M. D’Alessandro and W. J. Lewis, Limits of Volatile Chemical Detection of a Parasitoid Wasp, Microplitis croceipes, and an Electronic Nose: A Comparative Study, Transactions of the American Society of Agricultural Engineers, Vol. 47(6): 2145–2152, 2004. 15. Hollas, J. M., Ed. Modern Spectroscopy, 4th ed., Wiley, Chichester, West Sussex, 2004. 16. Drafts, B. Acoustic Wave Technology Sensors, Sensors Magazine Online, October 2000; visited 8/9/04; http://www.sensorsmag.com/sensors/articles/article Detail.jsp? id=327349. 17. Li, J. The Cyranose Chemical Vapor Analayzer, Sensors Magazine Online, August 2000; visited 8/9/06; http://www.sensorsmag.com/articles/0800/56/main.shtml. 18. Oxley, J. Detection of Illicit Chemicals and Explosives, Elseveir, in Press. 19. Garner, K. and S. Smith. Volatile Organic Compounds, the good, the bad and the analysis, SPME, University of the West of England, Bristol; updated 9/23/04; visited 9/19/05; http://www.chemsoc.org/exemplarchem/entries/2004/westengland smith/ExempWeb/methdev.htm. 20. Janusz, P. Solid Phase Microextraction: Theory and Practice. Wiley-VCH, New York, April, 1997. 21. Chambers, W. B., P. J. Rodacy, E. E. Jones, B. J. Gomez, and R. L. Woodﬁn. Chemical Sensing System for Classiﬁcation of Minelike Objects by Explosive Detection, in A. C. Dubey, J. F. Harvey, J. T. Broach, Eds. Proceedings of the SPIE 12th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, Detection and Remediation Technologies for Mines and Minelike Targets III, April 13–17, 1998. 22. Handbook of Chemistry and Physics, 80th ed., Chemical Rubber Publishing Co., Boca Raton, FL, 2000. 23. Handbook of Chemistry and Physics, 37th ed., Chemical Rubber Publishing Co., Cleveland, 1955–1956. 24. Cooper, P. W. Explosives Engineering, Wiley-VCH, New York, 1996. 25. Phelan, J. M. and S. W. Webb. Chemical Sensing for Buried Landmines—Fundamental Processes Inﬂuencing Trace Chemical Detection. SAND2002-0909, Sandia National Laboratories, Albuquerque, NM, 2002. 26. Yinon, J. Forensic and Environmental Detection of Explosives. Wiley, New York 1999.

34

CHEMICAL SENSING

27. Rhykerd, C. L., D. W. Hannum, D. W. Murray, and J. E. Parmeter. Guide for the Selection of Commercial Explosives Detection Systems for Law Enforcement Applications. NIJ Guide 100-99, NCJ 178913, National Institute of Justice, Ofﬁce of Science and Technology, Washington, DC, 1999. 28. Theisen, L., D. W. Hannum, D. W. Murray, and J. E. Parmeter. Survey of Commercially Available Explosives Detection Technologies and Equipment 2004, Document No. 208861, National Law Enforcement and Correction Technology Center, a Program of the National Institute of Justice, U.S. Department of Justice, Washington, DC, 2005. 29. MacDonald, J., J. R. Lockwood, J. McFee, T. Altshuler, T. Broach, L. Carin, R. Harmon, C. Rappaport, W. Scott, and R. Weaver. Alternatives for Landmine Detection, RAND Science and Technology Policy Institute, Santa Monica, 2003.

CHAPTER 2

WHAT TO DETECT? JIMMIE C. OXLEY University of Rhode Island

The process of detecting explosives is generally broken down into bulk and trace detection. Bulk detection looks for a mass with certain properties considered indicative of an explosive—high nitrogen and/or oxygen content, high bulk density. Naturally, there will be explosive compounds that do not match these target characteristics [e.g., triacetone triperoxide (TATP)]; and there will be innocuous compounds that do (e.g., sausage). In detecting the presence of an explosive compound at trace levels, the general approach has been to look for a speciﬁc chemical from a library of target compounds rather than for a general property. This means that the probability of a false alarm is signiﬁcantly lower than for bulk detection, but that a positive detection is limited to the provided library and makes no immediate allowance for terrorist innovations. A positive detection may also be misleading. The example that comes to mind is the reported positive detection on the wreckage of TWA ﬂight 800. In that case, it was explained that explosives had been present in the aircraft many days earlier for purposes of a training exercise, but that none were present upon takeoff. In theory, any chemical analysis scheme should be applicable to trace detection of explosives, but the realities of explosive detection require a degree of rapidity and robustness that limit the type of usable instrumentation. Perhaps the place to begin in a book about trace chemical sensing of explosives is to deﬁne trace and explosive. If trace is deﬁned as a submilligram quantity, then it should be noted that there are no chemicals that are explosive at trace levels. To clarify this point we need to elaborate on what makes a chemical an explosive. An energetic material is deﬁned as one that releases energy upon decomposition. This material could be an explosive, a propellant, a pyrotechnic, or a fruit cocktail. For an energetic material to be an explosive chemical or composition its must be capable of undergoing decomposition with extremely Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

35

36

WHAT TO DETECT?

TABLE 2.1 Energy vs. Power

Burning petroleum Detonating dynamite

J/g

W/cm2

50 5

103 1010

rapid release of energy and gas. An energetic material that burns may overall release signiﬁcantly more energy than one that detonates (see Table 2.1). It is not the amount of energy (joules) released but the rate of the energy release that is essential for explosive behavior. Many materials and devices “explode” (e.g., a hot water heater, pressure cooker). Only explosives can “detonate,” that is, form a special type of shock wave that travels through the material at supersonic speed. The problem, in terms of our deﬁnition of an explosive, is that the same chemical can be formulated or conﬁgured so that it detonates or it doesn’t. Because the rapid energy release is necessary to “support” the detonation front, much like a piston, conﬁguration of the chemical can be such that this is not possible. Typically, the problem arises at edges (regions of density change). A high-density region will bounce the shock wave back much as water hitting the wall of a swimming pool. The bounced back waves (rarefaction waves) degrade the shock wave, so that at such edges the shock wave is slowed and an overall curvature to the wave develops. If the width of the explosive is narrow, the rarefaction waves may be sufﬁcient to kill the shock wave. This leads to the concept of critical diameter, where the explosive sample is too small to support a detonation wave. Thus, 200 g of explosive in a cylindrical conﬁguration is probably detonable; the same amount sprinkled across the table top is probably not. This is, of course, the principle explosive disruptors employ—breaking a large charge into small ones without initiating detonation. With the size requirement in mind, let us consider the types of materials that may be explosive. The U.S. Department of Transportation (DoT) provides some guidance, specifying characteristics that, if met, require a material to be screened for explosive performance (see Table 2.2). Materials that typically fall in the explosive category are those where oxygen is incorporated within the molecule in nitro (NO2 ) groups. The presence of oxygen

TABLE 2.2

DoT Requirements for Class 1 Screening1

If a compound contains certain chemical groups: C=C unsaturation; M–C or M–N; contiguous N or O atoms; N–O; N–X or O–X (where X = halogen, M = metal) AND if it has an oxygen balance (O B) greater than −200, where O B = [−1600(2C + (H/2)–O)]/molecular weight; OR if its exothermic energy >500 J/g and onset below 500◦ C, then it must undergo DoT series one testing.

37

WHAT TO DETECT?

ONO2

ONO2 ONO2

O2NO

O O2NO

O O2NO ONO2

Nitroglycerin

ONO2

X

ONO2

O2NO

Nitrocellulose

Pentyerthritol tetrnitrate (PETN)

Figure 2.1 Nitrate esters.

within the molecule allows for rapid self-oxidation: C–H + O → H2 O + CO or

CO2

(2.1)

The nitrogen primarily ends up as gaseous N2 . The amount of oxygen available for these self-oxidation reactions is calculated as the oxygen balance. As can be seen from the DoT requirements above, a material can be quite oxygen deﬁcient [trinitrotoluene (TNT) has an oxygen balance (OB) of −80] and still be detonable. Nevertheless, we are comfortable using 2-ethylhexyl nitrate ester as a cetane additive in diesel fuels, whereas we would not consider using glycerin trinitrate ester (nitroglycerin) for that application. CH3 CH2 | CH3 CH2 CH2 CH2 CHCH2 ONO2

2-Ethylhexyl Nitrate Ester

CH2 ONO2 | CH–ONO2 | CH2 ONO2

(2.2)

Nitroglycerin

For reasons of safety, as well as performance, U.S. military explosives require a high level of stability and a low level of sensitivity. These requirements result in years of formulation and testing. Thus, most of the chemicals used by the military have been known for the last 100 years. Most military explosives can be classed by the attachment of the nitro group: nitrate esters (O–NO2 ), nitrarenes (C–NO2 ), or nitramines (N–NO2 ) (Figs. 2.1 to 2.3).1 Figure 2.4 shows a comparison of powders that are similar in appearance: sugar, salt, AN, HMX, RDX, TNT, PETN, TATP, and HMTD.2 1 In

structural formula the angle represents carbon with appropriate number of hydrogen. These are common abbreviations for the following: TNT 2,4,6-trinitrotoluene; AN ammonium nitrate; PETN pentaerythritol tetranitrate; HMX octahydro-1,3,5,7,-tetranitro-1,3,4,5-tetrazocine; RDX hexahydro-1,3,5-trinitro-s-triazine; HMTD hexamethylene triperoxide diamine; TATP triacetone triperoxide. 2

38

WHAT TO DETECT?

OH

CH3

O2N

NO2

O2N

H3C

NO2

NO2 N

O2N

NO2

NO2

Picric acid

NO2

2,4,6-Trinitrotoluene (TNT) Figure 2.2

NO2

Tetryl

Nitroarenes.

NO2 O2N

N

N

NO2

N N

N

O2 N

NO2

NO2

N N NO2

Figure 2.3

Nitramines.

Figure 2.4 Comparison of powders that are similar in appearance. Left to right: sugar, salt, AN, HMX, RDX, TNT, PETN, TATP, and HMTD.

Picric acid and tetryl, both yellow powders, are no longer used by the military, though “do-it-yourself” books outline the synthesis of picric acid for the wouldbe criminal/terrorist and tetryl is still found in old munitions. Most of the military explosives are white-colored powders (TNT is cream colored). Since all, but TNT, decompose upon or instead of melting, they require some sort of compounding in order to be shapeable. They can be blended into TNT in a variety of ratios to make the formulations listed in Table 2.3. They can also be formulated in wax or plasticizer. The use of plasticizer is preferred because less dilution of the explosive occurs. (In the world of performance, TNT, with detonation velocity of 6900 m/s is considered a dilutant of HMX, detonation velocity of 9100 m/s.) In the detonation process, temperatures reach 3000 to 4000 K so that residue, if any, consists of unreacted explosive and, possibly, carbon if the explosive is

39

WHAT TO DETECT?

TABLE 2.3

TNT-Based Explosive Formulations

Ingredient Added to TNT

Name

% TNT

Aluminum (Al) Ammonium nitrate (AN) RDX RDX + 1% wax Comp B(+18%Al) HMX PETN Barium nitrate

Tritonal Amatol Comp B2 Comp B Torpex-2 Octol Pentolite Baratol

20–80 20–60 40 39 12 25–30 50 24

greatly underoxidized. Presumably any unreacted explosive recovered is material that has been thrown out from the perimeter of the charge and not experienced the detonation front. The amount of unreacted explosive residue depends on the efﬁciency of the initiation. Studies focusing on the decomposition products of explosives are not studying detonation products; they are looking for products of thermal decomposition. These may form during long-term or high-temperature storage. These products can be produced in laboratory studies by heating the explosive to 200 to 400◦ C. For detection purposes, thermal decomposition products may be important, if they are sufﬁciently volatile. One thermal decomposition pathway accessible to most explosives is homolytic cleavage of the nitro group: O–NO2 → O• + • NO2 C–NO2 → C• + • NO2

(2.3)

N–NO2 → N• + • NO2 The energy required for this process depends on the exact linkage. In general, the C–NO2 linkage found in nitroarenes is quite strong. Other decomposition paths, which require lower energies, dominate. For nitrate esters, the homolytic cleavage of the nitro group is a fairly low energy pathway. Furthermore, the evolved nitro group can accelerate the decomposition of the nitrate ester; thus, nitrate esters such as nitrocellulose, are stored with stabilizers, which function to bind evolved NO2 . Regardless of whether evolution of NO2 is a dominant or minor thermal decomposition pathway, nitro-containing explosives can be detected by the use of chemiluminescence. In fact, chemiluminescence is the detection scheme used in the thermal energy analyzer (TEA), a detector used in conjunction with gas chromatographic analysis by forensic explosive laboratories. It is highly speciﬁc and sensitive to nitro-containing explosives. The difﬁculty is immediately obvious—what if the explosive does not contain a nitro group? There are four types of explosives that do not contain nitro groups: inorganic oxidizers (in combination with any fuel), peroxide explosives, primary explosives, and newly developed high-nitrogen explosives. Chapter 3 discusses both the inorganic oxidizers—principally nitrate or chlorate-based and peroxide explosives—TATP and HMTD. We have previously reported on the thermal stabilities

40

WHAT TO DETECT?

TABLE 2.4 Stable Sensitive Volatile

Deﬁnitions Does not readily decompose at elevated temperatures or over time Easily initiated to decomposition or detonation by ﬂame, impact, friction, or spark Transitions from solid to gas at a relatively low temperature; it can be returned to its original state upon cooling

and decomposition pathways of these materials [1,2]. Detection of the inorganic oxidizers is challenging since they have extremely low volatility, though if the cation is ammonium or urea, there may be trace amounts of ammonia or urea present. TATP is readily detectable because of its high volatility—0.05 torr at room temperature [3], about 10,000 times the vapor pressure of TNT, which can be detected by most “snifﬁng” instrumentation, including canines. Unfortunately, there is a tendency to equate “volatility,” “stability,” and “sensitivity” (see Table 2.4). Pure TATP is thermally stable; it does not decompose any faster than PETN. However, TATP is highly volatile, readily subliming to the atmosphere at room temperature if its container is open. For the terrorist, TATP and HMTD offer easy sources of primary explosives. Consulting the “do-it-yourself” literature, it can be seen that there are two other commonly recommended primary explosives—lead azide Pb(N3 )2 and mercury fulminate Hg(ONC)2 , but these are difﬁcult to prepare cleanly. The synthesis of diazodinitrophenol (DDNP) (Fig. 2.5), common in commercial detonators, is reported in such publications, but apparently is rarely attempted by clandestine chemists. Typically, the brisance of a primary is less than TNT, but the efﬁcacy is the fact that a shock wave can result from a relatively mild insult. Synthesis of new chemical explosives continues. The trend has been toward compounds with high positive heats of formation, high nitrogen content, and/or large ring strain (Figs. 2.6 and 2.7). However, synthesis of these requires sophisticated chemical expertise and, generally, is not attractive to the clandestine chemist. For the next decade we can expect that the challenge will remain to detect nitrated (X–NO2 ) materials and composite materials. As detection limits

O

N N

O2N

NO2

Figure 2.5

Primary explosive DDNP.

REFERENCES

41

N

N

N N

N

N

N

Tetrazolidine

[1,2,4]Triazolidine

Figure 2.6 Nitrogen rings are basis of new explosive compounds. O 2N

N H2N

NO2

O

X N N

H2N

N

N

N

N

N

O

O

X = O, SOn (n=0,1,2), N=N, N(O)-N, N(NO2)CH2N(NO2)

N

N

O2N

O

O2N

NO2

N

N

N

N

N

N

HN

N

+ N

NO2

N

NH

N

NO2

O2N

NO2

NH2

N

H2N

N

NO2 N

NO2

N

N

N N

N

Figure 2.7 Examples of fused ring systems.

become lower and lower, the signiﬁcance of minute levels compared to background or other environmental exposition will require careful consideration. REFERENCES 1. Oxley, J. C., J. L. Smith, and H. Chen. Decomposition of multi-peroxidic compound: Triacetone triperoxides (TATP). Propellants, Explosives Pyrotechnics 27, 209–216 (2002). 2. Oxley, J. C., J. L. Smith, H. Chen, and E. Ciofﬁ. Decomposition of multi-peroxidic compounds: Part II: Hexamethylene triperoxide diamine (HMTD). Thermochem. Acta 388(1–2), 215–225 (2002). 3. Oxley, J. C., J. L. Smith, J. Moran, and K. Shinde. Determination of the vapor density of triacetone triperoxide (TATP) using a gas chromatography headspace technique. Propellants, Explosives, Protechnics 30(2), 127–130 (2005).

CHAPTER 3

DANGEROUS INNOVATIONS KIRK YEAGER

3.1

INTRODUCTION

The use of explosives by terrorists is not a new innovation. Examples of groups attempting to violently address political grievances range from the Gunpowder Plot to blow up the British Houses of Parliament in 1605 to the attack on the USS Cole in Aden Harbor in October 2000. Explosives have been throughout the last half-century, and remain to this day, the primary tool of terrorists. What is fairly new in this arena is the sophistication of the disparate terrorist groups throughout the world in terms of producing their own explosive formulations. These homemade explosives are often referred to as improvised explosives (IEs). Broader experimentation with IEs has been coupled with more documented cross communication between terrorist groups and the systematic training of a new generation of bomb makers. Put plainly, the bad guys are getting better at what they do. To address the threat they pose, it is imperative that those who ﬁght this ever-evolving enemy also advance in sophistication. This chapter is designed to give a perspective on the improvised explosives being utilized by terrorist groups throughout the globe. It will cover the scientiﬁc theory behind manufacture of improvised explosives, the history of their use, and an overview of the most common materials encountered by law enforcement worldwide. Some details will be excluded out of necessity so as to not make this a “how-to” chapter to be co-opted by the wrong individuals. 3.2

THEORY OF IMPROVISED EXPLOSIVES

Typically, just the mention of “improvised explosives” leaves the reader with a picture of a bushy-bearded mad bomber. Stereotypes surrounding IEs have Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

43

44

DANGEROUS INNOVATIONS

been put forth by a sensationalist press, unrealistic Hollywood exaggerations, and even cartoons for years. This has not given the ﬁeld a great deal of credibility. If any area of explosives research requires meticulous scientiﬁc understanding, it is IEs. Both commercial and military explosives are designed to be safely stored, handled, and used. Dealing with these materials in many cases requires no signiﬁcant insight into their properties or chemistry. IEs are usually much less stable and forgiving than their military and commercial counterparts. To work safely with these materials, a much greater insight into their properties is required. Often the terrorist/criminal’s lack of understanding of these characteristics creates a self-solving problem. Those forced to deal with these materials (especially in law enforcement) should never make the same error. The goal of this section is to provide greater insight into the science of IEs, and, hopefully, help those forced to deal with them. Before anything else can be said about IEs, some rudimentary chemistry is needed. From a “cookbook” perspective, all explosives (be they military, commercial, or improvised) require the same chemical building blocks, which consist of a fuel and an oxidizer. Some explosives have the fuel and oxidizer as part of the same molecule, such as trinitrotoluene (TNT), and some explosives are comprised of mixtures of separate fuels and oxidizers, such as ammonium nitrate–fuel oil (ANFO). The oxidizer employed by the vast majority of explosives tends to be the NO2 (nitro) group. It is so predominant as an explosive ingredient that the primary focus of detection methods traditionally has been to look for nitroderived properties. IEs tend to utilize a more diverse range of oxidizers. Table 3.1 gives a list of the numerous oxidizer possibilities. Many of the oxidizers should be familiar to those experienced in investigating bombings. Prime examples are NH4 NO3 (ammonium nitrate) and KClO3 (potassium chlorate), both of which are utilized to make bombs on a regular basis. Other materials may appear more exotic but have equal destructive potential. It should be noted that some oxidizer families such as chlorates, nitrites, and peroxides are very reactive. When used in IE formulations, these oxidizers can create very sensitive mixtures, which need to be approached with extreme caution.

TABLE 3.1

Potential Oxidizers for Improvised Explosives

Oxidizer Name Chlorates Perchlorates Hypochlorites Nitrates Chromates Dichromates Iodates Permanganates Peroxides

Formula −

ClO3 ClO4 − OCl− NO3 − CrO4 −2 Cr2 O7 −2 IO3 − MnO4 − O2 −2

Common Counter Ions +

+

+

+2

K , Li , Na , Ba K+ , NH4 + , Na+ , Ba+2 , Ca+2 Ca+2 K+ , NH4 + , Na+ , Ba+2 , Ca+2 , Sr+2 Ba+2 , Ca+2 , K+ , Pb+2 K+ K+ , NH4 + K+ Na+ , Ba+2 , H+

Examples KClO3 , Ba(ClO3 )2 NH4 ClO4 , KClO4 Ca(OCl)2 NH4 NO3 , Ba(NO3 )2 K2 CrO4 , PbCrO4 K2 Cr2 O7 NH4 IO3 KMnO4 Na2 O2 , H2 O2 , BaO2

HISTORY AND THE ANARCHIST LITERATURE

45

To make an IE the oxidizer must be combined with a suitable fuel. The majority of fuels available for use in IEs are hydrocarbons (materials containing mostly carbon and hydrogen). These materials are often referred to as organic compounds due to their prevalence in living organisms. Potential fuels can be broken down into three categories: hydrocarbons, energetic hydrocarbons, and “elementals.” The category of pure hydrocarbons has too many materials to list. Anything that is typically burned to produce heat or energy can be applied as a fuel for an IE mixture. Some high-proﬁle examples would include petroleum products such as diesel fuel and kerosene, plant- and animal-based oils, sugars, glycerin, wax/parafﬁn, sawdust/wood pulp, Vaseline, shellac, and rosin. Energetic hydrocarbons are materials with carbon, hydrogen, and nitro groups all in the same molecule. Because they contain their own oxidizer premixed at the molecular level in the form of the nitro group, energetic hydrocarbons tend to make some of the more powerful IEs when mixed with a suitable oxidizer. Prime examples of energetic hydrocarbons are nitromethane (NM) and nitrobenzene (NB). Due to the incorporation of the nitro group, many of these energetic hydrocarbons can be made into explosives in their own right, without addition of an oxidizer. The liquid NM is such a material. The elemental fuels applicable to IE production are not as obvious as the hydrocarbons described above and simply have to be learned and memorized. Those most prevalent are powdered metals (aluminum, magnesium, titanium), carbon disulﬁde, phosphorus, sulfur, and antimony sulﬁde. One important side note is that, for the most part, these elemental fuels produce IE formulations that are very sensitive and often unstable. Mixtures incorporating any of them should be treated with extra caution.

3.3

HISTORY AND THE ANARCHIST LITERATURE

One of the best ways of understanding the improvised explosives being exploited by current-day terrorist groups is to become a student of history. It is a fair generalization to state “If it is simple, it has already been done.” This holds true for the formulation of explosives. Since the early 1800s, and the development of better techniques for chemical puriﬁcation, research into explosives has mushroomed. Countless concoctions have been prepared and examined, especially over the last 150 years. The best candidates prepared over the years have gone on to commercial and military applications. All those formulations that were unstable, untrustworthy, or easily substituted with better alternatives fell into obscurity. Their ghosts haunt us to this day as IEs. To fully appreciate this transition, an exploration of some of the explosives developed in the last century is useful. Many of the formulations presented in historical terms will be discussed in further detail in the subsequent sections dealing with current terrorist IE threats. In 1871 Herman Sprengel patented a series of mining explosives that consisted of an oxidizing substance mixed with a fuel. The novelty of this approach was

46

DANGEROUS INNOVATIONS

that the components could be stored separately as nonexplosives, mixed on site to produce an explosive, and then shot with a cap (detonator). Taking into account the educational levels of miners back in the late 1800s, the isolation of explosives was a very prudent safety measure. Sprengel utilized potassium chlorate (KClO3 ), strong nitric acid (HNO3 ), and liquid nitrogen dioxide (N2 O4 ) as his oxidizing agents. He chose nitrobenzene (NB), nitronapthalene, carbon disulﬁde, petroleum, and picric acid (PA) as fuels. A study of the oxidizers reveals the most historical signiﬁcance. Over the years, numerous formulations were prepared based on HNO3 . Usually, fuming nitric acid (HNO3 > 90% in water) was utilized in their preparation. Some historically signiﬁcant formulations were: • •

Oxanite: PA (58 parts)/HNO3 (42 parts) Hellhofﬁte: NB (28 parts)/HNO3 (72 parts)

These explosives could be quite effective. Hellhofﬁte had ∼70% the power of TNT and was more shock sensitive (easier to initiate). Although the theory was sound, the practice left a good deal to be desired. The main drawback of these mixtures was the nitric acid, which was very corrosive, and exceedingly unpleasant to work with due to its choking fumes and tendency to produce bad chemical burns. It also ate through the copper blasting caps of the time, sometimes creating a premature detonation of the fulminate in them. Then, as now, nothing is as unnerving as an explosive picking its own time to detonate. Numerous attempts were made to get around the nitric acid problems. At the time, dynamite was well understood, with one variety produced consisting of nitroglycerin absorbed on Kieselguhr (diatomaceous earth). Attempts were made to absorb Hellhofﬁte on Kieselguhr. Fumes were reduced, but the resulting ﬁll still attacked wrapping paper, wood pulp, metal, and skin. Eventually, the idea of nitric-acid-based explosives was abandoned due to the existence of kinder and gentler alternatives. The most successful of Sprengel’s formulations, from the standpoint of “historical longevity,” would be those based on KClO3 . Chlorates have a rich history in the energetic materials ﬁeld. Perhaps a commercial formulation called Rack-aRock is the most notable example. It consisted of 21 parts NB and 79 parts KClO3 . In 1885 a combination of 42,331 lb of dynamite and 240,399 lb of Rack-a-Rock was used to blast out the Hell Gate Channel in New York Harbor in one single blast. An original advertisement for this formulation is seen in Figure 3.1 [1]. The majority of Sprengel’s fuels were energetic hydrocarbons. Chlorate-based composite explosives tend to be very sensitive to friction and shock. When mixed with energetic fuels, the chlorate formulations are even more sensitive. To address this problem, historically two paths were taken. The chlorate was either mixed with a nonenergetic hydrocarbon or the mixture was phlegmatized (lubricated) if an energetic fuel was desired.

HISTORY AND THE ANARCHIST LITERATURE

Figure 3.1

47

Rack a rock advertisement.

Friction is one means of initiating energetic materials. When mixing solids, or handling and transferring powders, friction can be difﬁcult to avoid entirely. In general, friction can be minimized with the application of a lubricant. The lubricant helps particles, such as crystals of potassium chlorate, glide over each other with minimized heat production. This reduces the chances of accidental initiation. Initially, castor oil was utilized to phlegmatize chlorate mixtures. The advantage of using an organic oil, above its ready availability and low cost, was that it could also serve as a fuel component and a water repellant. A group of explosives known as Cheddites was based on this principle. •

Example: KClO3 (80%)/nitronapthalene (12%)/castor oil (8%)

Some experimentalists chose to maximize precautions by eliminating energetic fuels and phlegmatizing their mixes. A classic explosive group that took this approach was the Min´elites. • •

Example 1: KClO3 (90%)/heavy petroleum oil (3%)/parafﬁn (7%) Example 2: KClO3 (90%)/Vaseline (3%)/parafﬁn (7%)

48

DANGEROUS INNOVATIONS

The initial example utilized petroleum oil to lubricate, while the later employed Vaseline. Both lubricants served a dual function as a fuel. In the end, the sensitivity of the chlorate mixtures could only be reduced to a certain extent. Eventually, safer and more reliable explosives pushed the chlorate variants out of vogue. It should be mentioned that whereas chlorates faded away from the explosives scene, they found a broad range of applications in the pyrotechnic industry. Numerous commercial ﬂash powders consist of KClO3 and aluminum to this day. They are still exceedingly sensitive and are perhaps one of the biggest sources of accidents in the industry. Over time, numerous formulations started using potassium perchlorate (KClO4 ) as a substitute. Perchlorates tend to be much less sensitive to insult than chlorates. Before returning to the present day, one more group of explosives deserves some historical exploration. This group consists of organic peroxides. Although a tremendous number of organic peroxides exist, only a small number are detonable. A subsection of this group has the properties of primary explosives. Unlike the common nitrated explosives, which derive their explosive potential from the NO2 group, peroxides utilize the–O–O–group. This is a much more reactive group than nitrate and as such makes peroxides much less stable than the more common explosives. Due to their emerging popularity within the IE community, the materials that will be discussed include hexamethylene triperoxidediamine (HMTD) and triacetone triperoxide (TATP). Their signiﬁcance will also be addressed later in the chapter. HMTD was ﬁrst prepared by Baeyer and Villiger in 1900 [2]. It is the only organic peroxide that ever drew serious consideration as a viable explosive. In the early 1920s it was examined by U.S. Army ordnance as a potential military explosive [3]. HMTD’s ease of initiation, coupled with its capability to crush approximately 2.5 times the amount of sand as mercury fulminate led to its consideration as a primary explosive for detonators. Both HMTD and mercury fulminate [Hg(ONC)2 ] were placed in caps and pressed to 1000 psi (pounds per square inch) to compare their ability to initiate base charges. These caps were used to initiate 0.4-g portions of various high explosives. The amount of primary was varied to determine the minimum amount necessary to facilitate initiation. The results are summarized in Davis [4]. Important highlights are given in Table 3.2. It was obvious that HMTD was much more efﬁcient as an initiator. Although there is some dispute about the sensitivity of HMTD, the general consensus is

TABLE 3.2

Comparisons of Initiating Powers of Primary Explosives Minimum Charge (g) of Primary with Reenforcing Cap to Initiate

High Explosive TNT PA Tetryl

Mercury Fulminate (g)

HMTD (g)

0.26 0.21 0.24

0.08 0.05 0.05

HISTORY AND THE ANARCHIST LITERATURE

49

that it is more sensitive than mercury fulminate. Literature drophammer results for a 2-kg weight are as follows: HMTD (3 cm) [3] and Hg(ONC)2 (5 cm) [5]. Meyer also shows HMTD as being more sensitive [5], with an impact initiation of 0.6 Nm versus 1 to 2 Nm for Hg(ONC)2 . The power of HMTD, rated by lead block expansion, varies from 60% [3] to 116% [5] of TNT. Despite the good performance characteristics of HMTD, it had several signiﬁcant faults. As mentioned earlier, the peroxide bond is very reactive. This made HMTD incompatible with most metals. It actively attacked aluminum, tin, zinc, brass, copper, iron, and lead. HMTD was also very unstable when stored, exhibiting tremendous weight loss over short periods of time. In the end, it was judged both too reactive and too thermally unstable for any practical usage. It fell into obscurity in the explosives community in the early 1950s. TATP is unique in that it is a material that never received any serious consideration for military or commercial applications. It was studied by numerous groups, but primarily for academic reasons. Minimal literature references exist on it. TATP was ﬁrst reported by Wolfenstein in 1895 [6]. Since that time, numerous recipes have been developed for its preparation. One of the most useful studies of its properties as an explosive was conducted by Rohrlich and Sauermilch [7]. They determined that TATP had a TNT equivalency of approximately 88% based on lead block expansion. They prepared a ﬁring train consisting of 0.05 g TATP (pressed at 250 kg/cm2 ) in contact with pentaerythrital tetranitrate (PETN) to produce reliable blasting caps. Other experiments determined that a 0.16-g portion of the peroxide (density = 1.35 g/cm3 ) could initiate TNT. TATP displayed less reactivity with metals than HMTD, but it was less thermally stable. Sublimation (the direct transition of a material from the solid to the gaseous state) was the main pathway for weight loss in TATP. Even at room temperature, notable weight loss could be witnessed within a day. The high sensitivity and thermal instability of TATP precluded its use in any practical applications. Although of limited academic interest, it never took hold in the explosives community. Having shown that both Sprengel explosives and organic peroxides have been seriously studied, and in numerous cases applied as commercial energetic materials, their current status needs to be addressed. As stressed in each case, these materials were too unstable, unfriendly, and unpredictable for continued use in a ﬁeld populated with much better alternatives. Terrorists, however, often do not have ready access to the commercial and military alternatives in use today. They are not deterred by the factors that forced the Sprengel explosives and the peroxides into mainstream obscurity. An examination of the current “anarchist” literature enforces this point. The term anarchist literature is used to refer to the dozens of books that have historically been available through mail order companies such as Desert Publications and Paladin Press dealing with the production of explosives and bombs. Many of these sources are still copied today as primer material to train current Al Qaeda bomb makers. The same information has been disseminated over the Internet in numerous bomb-making websites.

50

DANGEROUS INNOVATIONS

To illustrate how improvised explosives are often related to historically rejected formulations; a sampling of the anarchist literature will be utilized. The small cross section examined will consist of Guerrilla’s Arsenal (GA) [8], Improvised Munitions Black Book Vol. 1 (BB1) [9], Ragnar’s Homemade Detonators (RHD) [10], and FMX The Revised Black Book (FMX) [11]. It should be stressed that this small cross section examined represents only a fraction of the readily available information accessible through the Internet. These examples are utilized solely to show the overlap of historical explosive formulations and current terrorist recipes. More examples of current-day usage of these improvised materials will be included in the next sections of this chapter. Sprengel explosives are very popular in this cross section of anarchist literature. The mixture of NB/HNO3 (Hellhofﬁte) is actually given by name in BB1 and FMX. Chlorate-based formulations also abound. Examples are provided in GA, BB1, and FMX. GA gives a recipe for a PIRA (Provisional Irish Republican Army)-based mixture consisting of sodium chlorate and NB called CO-OP. It then goes on to relate this mixture back to the potassium-chlorate-based variant Rack-a-Rock discussed earlier. The author goes to the extent of being helpful enough to notify the reader that if potassium chlorate is problematic to ﬁnd, sodium chlorate can be acquired in the form of “Solidox” oxygen pellets for home welders. Luckily Solidox is no longer on the market. BB1 gives instructions on how to prepare a chlorate-based mixture that is quite prevalent in the anarchist literature. This formulation consists of KClO3 phlegmatized with Vaseline. Typically this particular formulation is given the nickname “Poor Man’s C-4.” It does not have the power of C-4 but does posses

Figure 3.2

Poor man’s C-4.

FERTILIZER-BASED IEs

51

a degree of moldability. A sample prepared by the author is shown in Figure 3.2. The relationship of this mixture with the second of the Min´elite formulations provided earlier is obvious. Numerous terrorist groups still apply chlorate-based explosives to this day. Sodium chlorate is a principle ingredient found in many weed killers in Europe. As seen, the applicability of chlorates to IEs has been well documented. This has not been overlooked by bomb makers throughout Europe. One of the more popular mixtures involves combining the chlorate with icing sugar. This fuel will be encountered again in the next section. The recipes for organic peroxides also crop up throughout the anarchist press. All four references taken for the cross-section analysis have recipes for HMTD. Keeping with the established research, all four references indicate that HMTD serves as a good initiating explosive. GA goes as far as drawing a homemade detonator with a ﬁring train composed of ﬂash powder, HMTD, and PA. A preparation for picric acid based on aspirin as the crucial starting material is provided as well for the curious reader. Not to be outdone, Ragnar Benson includes pictures of the three readily available components (hydrogen peroxide, hexamine, and citric acid) needed to fabricate HMTD. With a fair bit of accuracy, he refers to HMTD as “the primer mixture they don’t want you to know about.” As a ﬁnal note on the anarchist literature, it should be mentioned that there is a tremendous amount of information available to the would-be bomber. A good portion of it is, at best, dangerous to the bomb maker. Some of it is excellent. In either case, the anarchist literature should never be written off as information for “crackpots” by “crackpots.” The best example of this is FMX. A quick look through the contents by an individual experienced in the ﬁeld of explosives shows a great deal of insight provided by the author. Upon looking at the bibliography provided, the source of this insight becomes apparent. Some of the books utilized to prepare FMX mirror texts used to prepare this chapter [4,5]. Numerous other references given would be found on the shelves of any researcher engaged in the serious study of explosives. The terrorist does have sources of well thought-out and documented work, and this should never be forgotten.

3.4 3.4.1

FERTILIZER-BASED IEs Ammonium Nitrate IEs

Terrorists have become exceedingly adept at utilizing any precursor native to their environment to produce IEs. For simplicity, bombings can be broken down into large scale [such as vehicle-borne IE devices (IEDs)] or small scale (such as package bombs). The mixtures utilized by terrorist groups to manufacture small-scale charges are far too numerous to cover in any detail. A sampling was provided above in the review of historical precedence for the anarchist literature. However, when terrorists wish to fashion a charge in the hundreds or thousands of pounds, their available precursors become fairly limited.

52

DANGEROUS INNOVATIONS

Historically, and to this day, terrorists tend to utilize fertilizer precursors in their explosive production for large-scale charges. The ready availability of these materials in large quantities, coupled with the low cost per pound, makes them attractive to many bomb makers. It needs to be stressed that bomb makers will utilize what they have at their ready disposal. Most of the ingredients they adapt to their nefarious purposes will have legitimate uses. Fertilizers are common materials found throughout the entire world. With minimal processing they have shown themselves to be effective weapons in the terrorist arsenal. The most common fertilizer that can be used to produce explosive formulations is ammonium nitrate (AN). In special circumstances, pure AN itself can be made into an explosive. It contains both a fuel (hydrogen atoms) and an oxidizer (the nitrate group) in the same molecule. Although AN is a very common fertilizer, it is important to remember that it is classiﬁed as an oxidizer. Like any other oxidizer, mixing a fuel with it will produce an energetic material. As mentioned earlier, AN brings a small amount of fuel with it in the form of hydrogen. Nevertheless, even if the available oxygen reacted with all the hydrogen present, the oxygen still would not be totally consumed. Adding a fuel to AN gives the excess oxygen something to react with to produce additional energy. The types of fuels that can be added to AN to produce explosive mixtures are limited only by the imagination of the bomber. ANFO, the most common explosive produced from ammonium nitrate, is made by mixing it with common diesel fuel. This mixture employs 94% (by weight) AN and 6% No. 2 diesel. The AN used to produce ANFO comes in a very speciﬁc 1 form. It is produced in small “beads” (called prills) which are approximately 16 inch in diameter. There are two types of AN prills manufactured. The prill type intended for production of explosive mixtures is referred to as explosive-grade (or “commercial”) prills. These prills are designed to act like miniature sponges. As such, they are relatively porous (6 to 10%) and will readily absorb liquids such as diesel fuel. AN prills produced to be utilized as fertilizer are referred to as FGAN (fertilizer-grade AN). These prills are much denser than explosivegrade prills and usually have less than 3% porosity. Due to their low porosity they will not absorb liquids efﬁciently. If 6% fuel oil is added to FGAN prills, a good portion of it will pool at the bottom of the mixing container. The same qualities that make ANFO an attractive explosive for commercial applications also draw terrorists to it. It is easy to prepare and can be made from cheap readily available ingredients. Because of this, ANFO has been utilized in terrorist attacks worldwide for many years. It has been one of the most common IEs utilized in large-scale bomb attacks. When producing an improvised explosive from AN, there are four basic options dictated by the nature of the starting materials. The AN used can either be prilled or powdered. The fuels used can be either solid or liquid. In theory, the use of molten AN or gaseous fuels is possible, but the odds are really not in favor of it. Thus the four obvious combinations for improvised formulations are: (1) prilled AN plus a liquid fuel, (2) prilled AN plus a solid fuel, (3) powdered

FERTILIZER-BASED IEs

53

AN plus a liquid fuel, and (4) powdered AN plus a solid fuel. Each combination has unique characteristics, which will be covered separately. Prilled AN will be the ﬁrst option examined. Prills have certain physical characteristics that produce similar behaviors in any explosives based on them. Most importantly they are little beads separated by air gaps that make it difﬁcult for a shock wave to travel from prill to prill. Anyone who has committed to a step without realizing that they were walking off a curb can understand how energy can be disrupted by empty space. To get any explosive based on prills initiated, a good deal of energy is needed to overcome this physical air gap barrier. This is why a booster is utilized in ANFO. The air gaps dissipate the “punch” that the cap is trying to deliver. Any improvised explosives based on AN prills and an absorbed liquid fuel will run into the same need for a strong initiation source. Detonation velocities of these mixes will also tend to be lower due to the energy extracted by the air pockets throughout the explosive. Again, the choice of liquid fuel is limited only by imagination. Energetic fuels such as nitrobenzene (NB) can provide additional energy to the mixture. Common hydrocarbons such as kerosene can also be used. In general, bombers will most likely stick with diesel as the fuel, as ANFO is the formulation that gets the most “press.” Life is not simple. As stated earlier, prills come in two forms. When using liquid fuels, it is best to utilize the porous (explosive) grade AN. Its ability to absorb liquids ensures good mixing. It is possible to utilize FGAN prills to make ANFO-like mixtures. FGAN will not absorb liquids effectively; so mixes made from it will not have the power of analogous formulations produced with explosive-grade prills. In either case, it should be realized that bombs can be made utilizing fertilizer-grade AN and liquid fuels. The author has made ton quantities of ANFO with the densest AN prills available (E2 prills). Even though a good amount of diesel pools at the bottom of the charge container in FGANbased ANFOs, these charges still give a respectable yield. Fertilizer-grade AN is far from benign. Probably the most rare combination is the mixture of AN prills with a solid fuel. To get a good reaction between a fuel and an oxidizer an intimate mixture is needed. This is why monomolecular explosives (with fuel and oxidizer on the same molecule) tend to have higher detonation velocities than composite explosives (where fuel and oxidizer are two separate components). This is also why grain dust explosions occur; yet, a pile of wheat is not easy to light on ﬁre. 1 inch in diameter) will Mixing any solid fuel with prills of AN (which are 16 obviously not produce a very good mixture of reactive components. The next set of explosives that need to be examined are improvised mixtures based on powdered AN. Once the AN is pulverized (powdered), the air gaps that dictated many of the properties of the prilled mixtures vanish. Thus any mixture based on powdered AN will take on different characteristics than a prilled mix. This has been noted by groups ranging from PIRA to Al Qaeda.

54

DANGEROUS INNOVATIONS

There are far too many potential fuels that could be added to ground AN to make improvised explosives. It would be impossible to discuss them all. Instead, a few important general principles will be covered. First, ground AN can be mixed with liquid fuels. A common example prepared commercially is ground AN mixed with nitromethane. This produces the binary explosive known as KinePak. Diesel fuel can also be used, as it was with prills to produce ANFO. This brings up the last potential combination: powdered AN and a powdered fuel. Again, any fuel capable of reacting with the excess oxygen released by AN will produce an effective improvised explosive. For this reason, one of the common solid fuels historically encountered is icing sugar. PIRA has used icing sugar as its fuel of choice to mix with AN since 1991. The resulting IE mixture is referred to as ANIS (ammonium nitrate icing sugar). 3.4.2

Urea Nitrate

The second fertilizer that is often used as the basis for an explosive is urea. Pure urea is neither an explosive nor oxidizer. In fact, it is primarily a fuel. It can be mixed with AN to form an explosive mixture. In portions of the United States urea is routinely sold to de-ice sidewalks during the winter months. It comes in prill form, just like AN. When urea is mixed with nitric acid it chemically reacts, and one molecule of urea attaches to one molecule of nitric acid. This forms the explosive urea nitrate (UNi), with the chemical formula NH2 CONH2 –HNO3 . In this explosive, the urea serves as the fuel and the nitric acid the oxidizer. The literature lists UNi as being 90% as powerful as TNT [5]. Small-scale air-blast work displayed a far-ﬁeld equivalency closer to 65% [12]. Literature detonation velocities listed range from 3.4 to 4.7 km/s (∼11,200 to 15,400 ft/s) [13]. The author has seen several different velocities in this region depending on the charge size and conﬁnement. UNi possesses a low loading density (typically around 0.7 g/cm3 ). It is highly corrosive and readily attacks most metals. In terms of sensitivity, UNi is typically a tertiary explosives such as ANFO and requires a booster to be initiated. Urea nitrate secured its place on the U.S. landscape by being the explosive used in the World Trade Center (WTC) bombing in February of 1993. Although it was not commonly seen in the United States at the time, terrorist groups in other parts of the world had long since known about it. Both the Shining Path in Peru and numerous mideastern terrorists utilized UNi in their attacks well before the WTC incident. The use by the Shining Path was so predominant that sales of urea were outlawed in Peru in 1992 to curtail explosive production. The Palestinians have utilized UNi as a main charge explosive for many years. Along with TATP (to be discussed in the next section), it remains one of the main explosives in the Palestinian arsenal. A photograph taken by the author in the Gaza Strip shows a sample of Palestinian UNi (see Fig. 3.3). Urea in fertilizer form is seldom pure. It carries a good number of impurities that create a beige color in the ﬁnal UNi.

PEROXIDE EXPLOSIVES

55

Figure 3.3 Palestinian urea nitrate.

Finally, urea nitrate can be taken one step further chemically. Through the process of dehydration (chemical removal of water), urea nitrate can be transformed into nitrourea (NH2 CONH–NO2 ). Typically UNi is added to concentrated sulfuric acid to facilitate this reaction. In theory, the loss of water should make nitrourea more powerful than UNi. There is not a great deal of experimental data available about this material. It was theorized that the group that bombed the World Trade Center may have planned to conduct this reaction. No conclusive evidence exists to prove this speculation. To date, nitrourea has not been seen in any terrorist bombs. 3.5

PEROXIDE EXPLOSIVES

The newest group of explosives to take hold in the arsenals of terrorist groups across the entire world has been from the peroxide family. Each year new groups add these materials to their bomb-making repertoire. The author has personally encountered them in the past couple of years in Morocco, Turkey, Israel, Jordan, and Uzbekistan. Much of the emergence of these materials can be traced back to the cross pollination of ideas from terrorists coming out of Al Qaeda training camps with smaller already established groups. As seen in the history section, the two main peroxides TATP and HMTD have been around for over a century. The information on their manufacture is far from new. Neither of these materials was encountered with any degree of regularity until about the mid-1990s. Of the two, TATP has been the predominant one adopted for reasons that will become clear during the HMTD discussion. Two forms of acetone peroxides are known to exist: a dimer called diacetone diperoxide (DADP) and a trimer called triacetone triperoxide (TATP). Figure 3.4 displays the chemical structures of each.

56

DANGEROUS INNOVATIONS

Figure 3.4 Chemical structures of acetone peroxide molecules.

Although both dimer and trimer are known to exist, the trimer is the more easily produced. Typically, the dimer is only seen as a by-product of the trimer synthesis. Unfortunately, over the past decade law enforcement agencies have encountered TATP more frequently both in the United States and abroad. Numerous questions regarding its synthesis and properties have surfaced in Internet news groups as well, and there is no sign of abatement in interest about the compound. TATP is prepared by the proper combination of three chemicals: hydrogen peroxide, acetone, and sulfuric acid. The exact preparation will not be detailed for obvious reasons. All the precursors needed to prepare TATP are readily available in most countries. Acetone is a very common solvent and industrial chemical. Sulfuric acid of approximately 35% concentration is utilized as battery acid. Aqueous hydrogen peroxide (H2 O2 ) can be procured from a number of sources. Most drug stores sell dilute solutions of H2 O2 ranging from 3 to 6%. Hairdressers, who bleach hair, can buy concentrated solutions of peroxide (up to 60 volume) to make their own dilutions. Recently, one of the trends that have cropped up in health circles is to “oxygenate” the body. To do this, it is recommended that baths be taken in dilute peroxide. To facilitate this, multiple Internet companies that will sell and ship 35% peroxide (commonly referred to as “food grade”) have been formed. All percentages of peroxide can be used to produce TATP. Higher concentrations obviously produce better yields. As mentioned earlier, TATP is currently becoming more predominant. The Internet has been accused of being the root of most modern evils. Luckily, it can also serve as a valuable tool for those following trends in improvised explosives. For example, it is possible to track how many times any given explosive is mentioned in forums such as newsgroups. Newsgroups are computer bulletin boards dedicated to a variety of subjects of interest where members can log in and exchange information. The ﬁeld of explosives has two main newsgroups dedicated to it known as alt.engr.explosives and rec.pyrotechnics. The ﬁrst group was designed for explosives engineers (such as blasters) and the second for hobby pyrotechnicians.

PEROXIDE EXPLOSIVES

57

700 600 RDX TATP

Number of Citations

500 400 300 200 100 0 1996

1997 YEAR

1998

Figure 3.5 Analysis of number of citations occurring in explosives newsgroups.

Using an Internet search engine called DejaNews it was possible to search any newsgroup for keywords. The search would show how many times a given keyword was mentioned. Figure 3.5 shows an analysis of the two explosiverelated newsgroup citations for TATP and RDX from 1996 to 1998. As mentioned earlier, TATP has been known for over 100 years. It was not until the mid-1990s that it began to emerge as a recurring terrorist and criminal threat. Figure 3.5 focuses on the time frame when this increase in incidents involving TATP was occurring. Before any conclusions are drawn, the reason for analyzing RDX must be explained. Each year more and more people get “hooked-up” to the Internet. Looking at the number of references for any one material, like TATP, in isolation could be misleading. It would be hard to tell if an increasing number of citations was simply caused by more people having computer access to newsgroups. RDX would be a very common explosive to be mentioned in such a venue. Using RDX as a control, the increased Internet access during any given time frame can be normalized. Between 1996 and 1998 there was a 323% increase in the number of references to RDX in the explosive-related newsgroups. This increase can be attributed to an increase in the number of computer users. During the same period of time, the references to TATP showed a 1372% increase. The increase in the interest in TATP can therefore be attributed to a real problem and not just the fact that more people have access to newsgroups. TATP was deﬁnitely an emerging threat at that time. It remains a strong threat to this day.

58

DANGEROUS INNOVATIONS

TATP is very sensitive to impurities. The primary destabilizing impurity present in the TATP synthesis is the residual sulfuric acid catalyst. TATP purity can be related to the extent of which the acid is removed or washed out of the ﬁnal product. The TATP produced by the author is made in an exceedingly meticulous fashion and is as free of impurities as it can physically be. The common criminal or terrorist cannot be counted on to be as careful in his or her TATP preparations. To better understand the dangers of TATP, the author has over the years made and studied the sensitivity of TATP of various purities. A “crude” TATP would be one in which minimal attempt to remove the residual acid catalyst was made. Crude TATP might have a simple water wash applied to the product crystals. A “washed” TATP would be one in which some chemical step has been conducted with the intent of trying to neutralize the residual acid. Utilizing a bicarbonate solution to wash the product crystals would produce a washed product. “Pure” TATP would refer to the product of a process where multiple steps were utilized to clean the material. Pure TATP would have undergone both a neutralizing wash and a recrystallization of the washed material. Heat, impact, friction, and electrostatic discharge (ESD) can all initiate explosives. Impact sensitivity can be tested with a device called a drophammer. This device employs a 2.5-kg weight raised to varying heights and dropped on 35-mg samples of the material being tested. How high the weight has to be raised to initiate the sample 50% of the time is used to indicate sensitivity to impact. This value is referred to as H50 (in centimeters). To judge sensitivity to friction, a line of sample (for lack of better analogy, like a line of cocaine on a mirror) is ﬁrst placed on a movable metal plate. A locked metal wheel is lowered onto the front of the line of powder. The wheel can have an adjustable pressure exerted on it with a “piston.” The plate is then impacted with a falling pendulum, forcing it to travel underneath the locked wheel. This creates a sliding frictional force on the material. Sensitivity to friction is measured by how much pressure is placed on the wheel to create initiations 50% of the time. This value is called P50 (in pounds per square inch). Finally, ESD sensitivity is measured by discharging a spark of known energy through a sample. The spark energy needed to initiate the material 50% of the time is calculated and referred to as E50 (in Joules). Table 3.3 displays the results from these sensitivity tests on the various forms of TATP. Data for PETN provides a baseline for all comparisons. Overall, the most pure form of TATP is just slightly more sensitive than dry PETN. Although the

TABLE 3.3

Sensitivity of TATP at Different Purities

Material Tested Crude TATP Washed TATP Pure TATP PETN (Reference)

Melting (◦ C)

Impact H50 (cm)

Friction P50 (psi)

ESD E50 (J)

73–79 Decomposes 85–90 94–95 141 Decomposes

6.5 ± 3.2 5.3 ± 1.0 10.0 ± 1.4 12.5 ± 0.2

∼80 − 100 21 ± 12 37 ± 3 645 ± 70

— — 0.16 ± 0.05 0.325

PEROXIDE EXPLOSIVES

59

sensitivity of pure TATP is very high (a quality that classiﬁes it as a primary explosive), the freshly prepared TATP does not exhibit “hair-raising” sensitivity. Unfortunately, this one pristine state is the only one that is not unnerving, since the less pure materials were even more sensitive to insult. As mentioned before, it is doubtful that the “home-brew” versions of TATP will be as clean as the pure samples. A brief clariﬁcation needs to be made about ESD sensitivity. As ESD is the most difﬁcult form of insult to guard against, great attention needs to be paid to materials that exhibit high sensitivity to it. ESD initiation sensitivity was measured for only the purest form of TATP. As in all other forms of insult, the pure TATP was only slightly more sensitive than PETN. A standard deviation was done for the measurements. Using simple statistical analysis, it can be shown that TATP will initiate approximately 1 to 2% of the time with energies as low as 0.06 J. Note: A good sized human body under the right conditions can generate up to 0.08 J. This is more than ample energy to initiate TATP a small percentage of the time. Although odds are better than at a roulette table, it still does not seem a wise gamble to take. TATP needs to be treated as if it is static sensitive. As the TATP decreased in purity, an increase in initiation sensitivity was noticed. For a point of reference, an increase in initiation sensitivity means that a material is easier to initiate. Simply skipping the recrystallization stage reduced the drophammer of washed TATP to half that of the pure sample. This simple difference also increased friction sensitivity substantially (by ∼43%). Generally speaking, TATP goes from being very sensitive when pure to being dangerously sensitive in a less puriﬁed form. The TATP variation most likely encountered in the ﬁeld would be analogous to the crude TATP. The amount of water used to wash the explosive would change from maker to maker, but high levels of impurities should be anticipated in the ﬁnal product. It has been mentioned to the author (who, having never seen crack cannot verify this) that certain forms of TATP resemble crack cocaine. Spot tests for drugs such as crack contain strong mineral acid solutions. These solutions, when brought into contact with TATP, produce impressive decompositions. Extreme care should be taken to use small amounts of any unknown material when testing for the presence of drugs. If any of the starting materials for TATP are present, the use of these spot tests is not recommended. Pictures of TATP prepared by the author are provided in Figure 3.6. TATP can have a wide variety of appearances that are effected by the way the product is washed and allowed to crystallize. In addition, its appearance will be altered by any additives placed in the material. TATP has been mixed with ammonium nitrate, glue, TNT, mineral oil, and numerous other materials by terrorist groups in an attempt to increase either its power or its handling safety. Figure 3.7 shows photos taken by the author of TATP in two different countries. Detection of TATP in the ﬁeld has been problematic. Technologies have been developed that focus on the peroxide group. Of these tests, the most commonly available are ﬁeld detection kits that rely on color changes brought on by chemical reactions. Mistral’s PDK (Peroxide Detection Kit) offers chemical screening

60

DANGEROUS INNOVATIONS

Figure 3.6

(Left) Washed TATP and (right) pure TATP.

Figure 3.7 (Left) TATP (Amman, Jordan) and (right) TATP (Ankara, Turkey).

for peroxides [14]. As in all colorimetric testing, the chance for false positives always exists. False positives also plague more advanced technologies such as ion mobility spectrometry (IMS) trace detection. This remains an area open for active study. While little serious research has been conducted on the properties of TATP, this is not the case for HMTD. As previously mentioned, tests conducted by U.S. Army ordnance illustrated that HMTD was a tremendously powerful initiating explosive, exhibiting between three and four times the strength of mercury fulminate. Unfortunately, HMTD was too thermally unstable and too chemically

PEROXIDE EXPLOSIVES

61

reactive to be utilized in any practical fashion. Due to this, no military or commercial application of HMTD ever occurred. The hazards associated with HMTD have not, however, dissuaded criminals, terrorists, or juvenile experimenters from manufacturing and using it. It has been recommended in numerous anarchist publications as the improvised explosive of choice for the preparation of homemade initiators. Over the past several years it has been encountered with increasing frequency by law enforcement ofﬁcers worldwide. For this reason, as noted for TATP earlier, a basic knowledge of its properties is essential for those in the explosive community. Of all the improvised explosives the author has made, HMTD probably has the simplest preparation. It requires three basic ingredients: hexamethylene tetramine (hexamine), citric acid, and hydrogen peroxide. Hexamine is a common fuel used in the pyrotechnic industry and can be purchased from numerous chemical supply houses. It is also the principle ingredient used by many camping stoves and can be purchased in tablet form for this purpose. Citric acid is a common ﬂavorant additive and can be purchased from many drugstores. Aqueous hydrogen peroxide availability was discussed earlier. HMTD has many of the same molecular constituents of other high explosives (such as nitrogen, oxygen, and carbon). It should be noted that, like TATP, there are no nitro groups associated with HMTD. The structure of HMTD is provided in Figure 3.8. Like TATP, HMTD is characterized as a primary explosive. It is highly sensitive to external inputs of energy. A summary of the sensitivities to small-scale testing exhibited by HMTD is provided in Table 3.4. Values for PETN, RDX, and TATP have been included to provide reference points. PETN was obtained from Reynolds Corp. It was Risi “Hi-Surface” Lot F-2961. RDX was Type B Class G 0206-00 from Holston.

Figure 3.8

TABLE 3.4

Chemical structure of HMTD.

Sensitivity of HMTD Compared to TATP and Reference Standards

Impact H 50 (cm) Friction P 50 (psi) ESD E 50 (J)

HMTD

TATP

PETN

RDX

18.1 ± 3.2 0 !!! 0.217 ± 0.027

10.0 ± 1.4 37.0 ± 3.0 0.16 ± 0.05

12.5 ± 0.2 645 ± 70 0.325 ± 0.032

22.5 ± 1.9 >1100 0.265 ± 0.064

62

DANGEROUS INNOVATIONS

HMTD displays some interesting characteristics. It was surprisingly fairly insensitive to impact compared to PETN and TATP. The impact sensitivity measured lay between the values for PETN and RDX. Even more surprising was the ESD sensitivity. HMTD is a very ﬁne powder when isolated. It has an appearance almost like that of talc. Fine powders are well known for being prone to static charge buildup. When transferring from one container to another, the static buildup in the HMTD was obvious by the way the powder stuck to the walls of the vessels involved. HMTD was more static sensitive than either of the standards tested, but it was less static sensitive than TATP. This was unexpected, and at the moment cannot be explained. The values obtained seem to indicate that HMTD has only moderate static sensitivity. CAUTION: Even though the experiments showed that HMTD was not overly sensitive to static, when dealing with improvised primary explosives, it is always wise to take static precautions. Terrorists do not always make the cleanest products. The truly unnerving characteristic of HMTD was its sensitivity to friction. Earlier, the test apparatus used to determine friction sensitivity was brieﬂy described. Usually, some form of pressure needs to be exerted on the wheel in order to create enough frictional force to initiate an explosive. In the case of HMTD, a sample could be placed on the plate and the wheel placed over it with no pressure exerted at all. The minute the plate was slid under the wheel, the HMTD detonated. CAUTION: HMTD is by far one of the most friction-sensitive improvised explosives currently being produced by criminals and terrorists. Only contact explosives such as Armstrong’s mixture (chlorate and red phosphorous) rival it in sensitivity. As mentioned earlier, HMTD is a very ﬁne white powder. Fresh, it resembles talc or ﬂour. As it ages HMTD begins to decompose and clumps together. Examples of both fresh and aged HMTD can be seen in Figure 3.9.

Figure 3.9 (Left) Fresh HMTD and (right) aged HMTD recovered from Ahmed Ressam (see 3.6.1).

THE NEXT WAVE

63

Many sources provide instruction on producing homemade detonators from HMTD. They recommend placing it in holders such a cartridge casings. Recommending that someone press an explosive as friction sensitive as HMTD into anything is doing quite a disservice to the reader. In addition, HMTD is very reactive with most metals. It will not sit well in something made with copper or iron. As cartridge casings are made of brass, there is an immediate incompatibility of HMTD with this type of housing. When found, homemade initiators of this type should be treated as potentially self-initiating. Like TATP, detection of HMTD is problematic. HMTD, being an amine, does give off a unique odor. The smell emitted by HMTD has been likened to that of ﬁsh. As it gets older, the odor gets more predominant and unpleasant. 3.6

THE NEXT WAVE

Terrorist groups have over the years shown the ability to adapt and learn through experimentation. PIRA was well known for many years of experimenting with and perfecting improvised explosives based on ammonium nitrate. The current group of Al-Qaeda-trained terrorists is carrying on that tradition. In addition, a broader interaction between what were previously isolated terrorist groups has led to an unprecedented cross pollination of ideas. As a result, today’s law enforcement agencies are presented with an increasingly complicated explosive arsenal that presents totally new safety considerations to address. 3.6.1

Improvised Detonators

One of the newer additions to the terrorist’s lexicon is the ability to produce reliable improvised initiators. Only a decade ago the blasting cap was one of the most difﬁcult to procure items necessary to produce an IED. This hurdle has been almost totally eliminated. Not surprisingly this change was precipitated by the “discovery” of TATP and HMTD. Both the millennium bomber Ahmed Ressam and the shoe bomber Richard Reid provide examples of terrorists utilizing this new “technology.” Both TATP and HMTD have been applied by terrorist groups worldwide in the production of homemade detonators. TATP is, however, seen in more instances. Examples of improvised caps can be seen in Figure 3.10. The photograph gives an illustration of the standard paper detonators (in this case produced from yellow lined paper and Christmas tree bulbs) and also shows a variety produced from syringes. The detonators in the photograph were recovered in Amman, Jordan. Jordanian ofﬁcials discovered in their investigation that the bomb-making factory also served as a training facility for those involved in the terrorist cell. The detonators recovered were ﬁlled with either TATP or HMTD and were manufactured by members of the cell learning how to produce such improvised initiators. The paper detonators matched those produced by known Al-Qaeda-trained terrorists. Similar detonators were seen in the Gaza Strip, Uzbekistan, and Turkey by the author.

64

DANGEROUS INNOVATIONS

Figure 3.10

3.6.2

Improvised paper and syringe detonators.

Peroxide Main Charges

Although highly sensitive, TATP has been utilized as the main charge in numerous attacks over the last 10 years. The Palestinians have experimented with TATP and created numerous formulations in an attempt to increase its handling safety. Many explosive production labs met with a violent end during the course of this experimentation. This has motivated more Palestinian bomb makers to return again to urea nitrate as a main charge, although TATP is still routinely encountered in Israel. In an example of how quickly technology travels from one terrorist group to another, the attacks in Casablanca, Morocco, in May of 2003 utilized a TATP formulation ﬁrst seen a couple years earlier in Palestinian attacks in Israel. TATP has also become one of the main charges utilized by Communist terrorist groups in Turkey over the last couple of years. TATP charges ranging from 1 to 20 lb are not uncommon. The proliferation of TATP as a main charge explosive remains relatively new. It provides a sobering illustration of how far bomb builders are willing to go to produce an IED with limited available starting materials. The risks of utilizing a highly reactive primary explosive for a main charge have not deterred these groups. Unfortunately, law enforcement is increasingly encountering labs producing these materials and having to engineer ways to safely deal with them. Due to the current trends in TATP usage, the need for reliable ﬁeld-use trace and

THE NEXT WAVE

65

bulk detection methods geared toward peroxides has become more of an issue throughout the globe. 3.6.3

Fringe Mixtures

The trend seen with TATP has been recently mirrored with other highly dangerous mixtures. As commercial explosives become more difﬁcult to procure, by both legitimate and illicit means, terrorist groups determined to conduct bombing operations are forced to work with more dangerous and unpredictable formulations. A prime example of this can be seen in the major bombings carried out in Indonesia over the last two and a half years. On October 12, 2002, a series of bombs shattered the tranquility of the Indonesian island of Bali. Three devices in all were detonated over the period of a few minutes. The devices ranged in size from a couple kilograms to approximately a ton. Of particular note was the vehicle borne IED (VBIED), which was initiated outside of Sari’s nightclub. The explosive ﬁller in this bomb would mark the ﬁrst time terrorists applied a chlorate-based improvised explosive in a large vehicle bomb. Figure 3.11 shows the aftermath of the bombing. The ruins in the lower right-hand corner mark the location of Sari’s nightclub. As discussed in the history section, chlorates have been long utilized by terrorist groups for small bombing attacks. Sodium chlorate is a very common weed killer throughout many European countries. The mixture of sodium chlorate and nitrobenzene (CO-OP) by PIRA was discussed earlier. Most commonly, chlorate is mixed with sugar to make an explosive ﬁller for pipe bombs. As mentioned earlier, chlorates are also a major ingredient in various pyrotechnic formulations. Typically potassium chlorate is utilized in this capacity due to the more hygroscopic nature of the sodium analog. Mixed with powdered aluminum and/or sulfur, potassium chlorate produces a mixture commonly referred

Figure 3.11

See color plates. Aerial view of Bali bombing scene.

66

DANGEROUS INNOVATIONS

to as ﬂash powder. The low ionization energy of aluminum creates mixtures when combined with chlorate that are very static sensitive. Addition of the low melting sulfur creates mixtures with high friction sensitivity. Combining both sulfur and aluminum with chlorate creates horrendously sensitive formulations. Interviews of suspects associated with the Bali bombings indicated that the bomb makers produced approximately 1000 kg of a chlorate ﬂash powder incorporating both aluminum powder and sulfur. This device was delivered via vehicle to the targeted nightclub. Precursor chemicals were easily purchased in Indonesia due to the geographical proximity to China, which remains one of the largest producers of pyrotechnics in the world. Before the Bali attack it would have been almost unthinkable for a group to produce a VBIED ﬁlled with a ton of ﬂash powder. The hazards inherent with such a device would have dissuaded most bomb builders. However, current trends show that terrorists have become more willing to take risks to produce explosive charges with whatever chemicals are readily available regardless of the inherent dangers. Bomb making has historically never been an overly safe occupation. As more “exotic” formulations begin to wend their way into the terrorist’s arsenal, the job of dealing with explosives and recovered IEDs will become more complicated. It should be noted that the use of chlorate has not been limited to Bali. Two more attacks using chlorate-based improvised explosives occurred in Indonesia followed the Bali bombing. On August 5, 2003, the J.W. Marriott hotel was bombed in Jakarta. In an even bolder attack the Australian Embassy in Jakarta was bombed on September 9, 2004. This continued reliance on a highly sensitive main charge in Indonesia illustrates that terrorists will utilize whatever tools are at their disposal regardless of the risks. 3.6.4

On the Horizon

The production of large charges based on chlorates (or primary explosives such as TATP) has raised the threshold of danger exponentially. With this trend, law enforcement can no longer discount any mixture as being too dangerous for terrorists to attempt to utilize. Past detection techniques, both bulk and trace, have tended to focus on nitrate groups. At the time this approach was sound and logical, as most bombing attacks utilized traditional explosive charges such as dynamite or TNT. This scenario has been forever altered. Materials that in the past were never considered to pose a viable threat have been successfully utilized in numerous attacks. If explosive screening technology hopes to match the growing threats, it will have to expand into the realm of reactive chemicals, which were previously discounted as nonviable for illicit use. The list of precursors that can be utilized to produce improvised explosives is large. Luckily oxidizers efﬁcacious for this purpose tend to fall into smaller groups or families. Thus the family of peroxides would constitute a logical next step for screening technologies to focus on. Chlorate, perchlorates, and permanganates would also need to be further addressed.

REFERENCES

67

Terrorists have shown an ability to learn, improvise, and adapt. What only 10 years ago would have been considered too improbable a threat to consider addressing has in many cases become almost routine. Technology and training must take their cue from this trend and attempt to counter the next wave of threats in a proactive fashion. The alternative is to allow the enemy to exploit a weakness that thoughtful anticipation might have been able to eliminate. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.

Photograph courtesy of Robert Hopler, Powderman Consulting Inc. Baeyer, A. and V. Villiger. Ber. Dtsch. Chem. Ges. 33, 2479 (1900). Taylor and Rickenbach, Army Ordnance 5, 463 (1924). Davis, T. L. The Chemistry of Powder and Explosives. Wiley, New York, 1956. Meyer, R. Explosives, 3rd ed., VCH, New York, 1987. Wolfenstein, R. Ber. Dtsch. Chem. Ges. 28, 2265 (1895). Rohrlich, M. and W. Sauermilch. Z. Gesamte Schiess Sprengstoffwesen 38, 97 (1943). Harber, D. Guerrilla’s Arsenal: Advanced Techniques for Making Explosives and Time Delay Bombs. Paladin Press, Boulder, CO, 1994. Improvised Munitions Black Book, Vol. 1. Desert Publications, El Dorado, AR, 1978. Benson, R. Ragnar’s Homemade Detonators: How to Make ‘Em, How to Salvage ‘Em, How to Detonate ‘Em. Paladin Press, Boulder, CO, 1993. Wallace, W. FMX the Revised Black Book, Paladin Press, Boulder, CO, 1995. Yeager, K. Data-Base of Range Evaluated Improvised Explosives (D-BREIE) Phase I: Ammonium Nitrate Based Explosives. EMRTC Report to TSWG, July 1999. Kage, S., Ed. Encyclopedia of Explosives and Related Items Vol. 10. U.S. Army ARRADCOM, Dover, NJ, 1983, p. U103. MISTRAL Detection Israel, www.mistralgroup.com.

CHAPTER 4

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES? RONALD L. WOODFIN Sandia National Laboratories (retired)

J. M. Phelan did a major part of this research. Much of this chapter, through Section 4.1, is drawn primarily from his work [1]. This work is also reproduced in [1a]. Since he was not available to actually write this chapter, the author, who merely coordinated this work as part of a broader project, gratefully acknowledges his debt for these important contributions to him and his co-workers. S. W. Webb, J. L. Barnett, and W. B. Chambers of Sandia National Laboratories and T. F. Jenkins, D. C. Leggett, and their colleagues of the U.S. Army Engineer Research and Development Center—Cold Regions Research and Engineering Laboratory (CRRDC), also made major contributions in this area.

4.1

INTRODUCTION

4.1.1 Where Did the Molecules Come from and How Did They Get Here?

We most often employ a portable trace chemical sensor in an effort to ﬁnd an object containing an explosive charge. Normally, when we ﬁnd explosive molecules, the immediate question becomes: “Where did these originate?” In choosing this kind of device we make a tacit assumption that some molecules will be released from the explosive charge and ﬁnd their way into the surrounding environment, where they may be located and identiﬁed, using the sensor. It becomes crucial then, when we employ a trace chemical sensor, to recognize the mechanisms that affect these molecules after release. Therefore, we need to understand the environment through which these molecules migrate, the transport mechanisms that produce this migration, and the ﬁnal condition of the molecules as they come to the positions where a searcher may encounter them. As we began Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

69

70

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

to study this part of the problem at Sandia National Laboratories (SNL), we came to refer to this study as “environmental fate and transport,” or EF&T. Of course, the object that contains and releases the explosive can be almost anything. However, since we began this study from the perspective of searching for mines, initially, landmines buried in soil, much of this chapter will use that speciﬁc work to illustrate the approach. We have gained a level of understanding for this particular application. From there we can adapt the methods to understand the EF&T of molecules in other situations. We can think of numerous examples of other applications where we might need to locate the object that is releasing explosive molecules. This is a problem that is most easily worked in reverse. That is, we ﬁnd it most convenient to establish a source of molecules in a particular environment and follow the migration of the molecules as they are released under changes in that environment. With this understanding we have a reasonable expectation that we can predict the location of the source in a similar environment when we locate some of the released molecules. 4.1.2

Objects Other Than Buried Landmines

These obviously include UXO,1 IEDs,2 and other ERW,3 but also may include similar objects underwater, buried in the seabed. There are also abandoned mining and construction sites that may need to be searched for old explosives. No matter which type object we may hypothesize as the source of molecules found, we need to follow similar reasoning to locate that source. In particular, the study of “plumes” of molecules released in water from a submerged source has been the focus of substantial research. The migration of the molecules from within or from the surface of a submerged, buried munition is similar to the case of munitions buried in the ground. Following that release there is little similarity. Almost no work has been done to date to examine the processes that affect the molecules released from an object buried in the seabed. This may form a fruitful area for research. 4.1.3

Questions That Beg for Answers

We ﬁnd a series of questions that arise in each situation: • • •

How and why do the molecules we ﬁnd with the sensor leave their source? What forces cause the molecules to move away from the source? How quickly does this process occur, and how long does it last?

1 Unexploded

ordnance, usually referring to items such as bombs or artillery shells that failed to explode when employed, but also referring to items containing explosives that were abandoned or misplaced. 2 Improvised explosive devices, usually one-of-a-kind items or adapted UXO. 3 The United Nations has adopted the term “explosive remnants of war” to describe a variety of items that includes mines and unexploded ordnance.

SOURCE OF THE MOLECULES

• • • •

71

Is it a transient process or does it reach a “steady state”? Do the molecules tend to move in predictable directions? Do the molecules tend to concentrate in predictable locations? Are there some times or conditions more favorable for detecting molecules?

This chapter will explore these questions in a general way. Because of the nature of the processes involved, much of the basic research used to develop experiments and models for the migration of molecules was originally done for other purposes. In particular, much comes from agricultural research and much from research into environmental pollution. More details may be found in the references listed at the end of the chapter. 4.2 4.2.1

SOURCE OF THE MOLECULES How the Molecules Diffuse or Leak from a Munition

Henceforth we will consider any object containing an explosive charge as a munition. These may take any of the forms noted above. For every munition we assume that the explosive charge is contained within some sort of case. The forms of these cases vary widely. Some are as simple as a rubber or plastic bag inside a wooden box. Some are specially designed plastic or metal cases. Some are metal shells sealed with rubber o-rings or gaskets. All have two things in common: They have some explosive molecules adhering to their external surface and they have paths by which molecules can escape from the interior, either through leakage or diffusion. (It may be that some sea mines are so well sealed that no paths exist. This has not been well established.) 4.2.1.1 Surface Contamination During the process of ﬁlling a munition with explosive, some molecules will adhere to the external surface. Since these molecules are inherently “sticky,” they tend to remain there. In addition, when munitions are stored in magazines, or handled in work areas where explosives are regularly exposed to the air, the effect is for all surfaces to have explosive molecules adhering to their surfaces. (See Fig. 4.1.) (This is also true of the people who work in these areas.) Munitions may also receive surface contamination from being ﬁred from an artillery piece or a rocket launcher, or from becoming damaged in the act of being “placed.” The term placed is used to designate any method involved in delivering the munition to where it is intended to function. This includes direct burial or hiding, ﬁring of guns or rockets, dropping of bombs, and the like. A munition loses this surface contamination within days or weeks after being placed. This time depends on many factors, but it is clear that there may be an initially higher level of molecules entering the local environment from a recently placed, or “fresh” munition. This is a transient effect that normally gives way to a more steady-state supply of molecules. In munitions that have painted surfaces,

72

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

Figure 4.1 mine.

TABLE 4.1 Mine Type TM62-P TM62-P TM62-P TM62-P TMM-1 TMM-1 TMM-1 TMM-1

Sampling for surface contamination on a Soviet-type TM-62P2 anti-tank

Surface Contamination on Mines Chemical

Case

Concentration (ng/cm2 )

TNT TNT DNT DNT TNT DNT TNT DNT

Bakelte Polyethylene Bakelte Polyethylene Metal Metal Metal Metal Average

6 3 28 5 10 10 62 20 18

Source [3] [3] [3] [3] [4] [4] [4] [4]

this transient effect is to be expected. Paint samples scraped from unused U.S. mortar and artillery rounds contained from 1 to 45 μg/g of trinitrotoluene (TNT) and dinitrotoluene (DNT) [2]. Researchers at Sandia [3] and at the Cold Regions Research and Engineering Laboratory (CRRDC) [1] measured surface contamination on mine cases. Some of the results are highlighted in Table 4.1, based on data summarized by Phelan and Webb [1, p. 20].

SOURCE OF THE MOLECULES

73

4.2.1.2 Diffusion through Cases A principal mechanism for this steadystate release is direct diffusion of the explosive molecules through the munition case. Clearly, the rate of this diffusion is dependent on the case material, through a property called the diffusivity. Metal cases offer essentially zero diffusion, but cases made of some polymers or natural rubber have high enough diffusivities to enable substantial diffusion rates. The diffusivity of cases varies widely. Phelan and Webb [1] quote Mark and Kroschwitz [5] to illustrate that a natural rubber case may have a diffusion rate of explosive molecules about 9 orders of magnitude greater than a polyvinylchloride (PVC) case under the same conditions. However, there seem to be many more munitions with PVC than natural rubber cases. Most nonmetallic munition cases fall in between these extremes. There are some general statements possible about diffusion rates. Diffusivity or permeability of the case material increases with temperature. A 5◦ C rise in temperature can result in a 30 to 50% rise in permeability [1 p. 18]. This immediately implies that we would expect more molecules available for detection in warmer environments, if all other factors were equal. 4.2.1.3 Leakage around Seals When munitions, such as those illustrated in Figure 4.2, are fabricated in metal cases, there is essentially no diffusion through the case. However, since a munition needs a fuze to actuate it, and since it must somehow be ﬁlled with its explosive charge, there must be penetrations in the metal case. These may be closed with threaded plugs that often contain o-rings or gaskets. While the leakage is usually small around these plugs, direct samplings indicate a higher level of explosive material around them. This is observed for both fresh and old munitions. While the leak rate may be too small to produce a supply of molecules sufﬁcient for detection, once the munition is found, it is sometimes possible to separate explosive bearing munitions from nonexplosive bearing munitions by sampling around these penetrations [6]. However, considering the consequences of misidentifying a loaded munition as inert, this technique is considered as insufﬁciently robust for that purpose [7]. Two groups of UXO are pictured in Figure 4.2. The upper group has been recovered from a burial site; the lower group has mixed duds and inert practice rounds from an artillery training impact area. 4.2.2

Example of Landmines

When we began this line of research at Sandia National Laboratories, we were primarily concerned with landmines. One of our ﬁrst efforts was to characterize the surface contamination on some typical mines. For this work we obtained some Soviet mines from the U.S. Marine Corps. Leggett and colleagues at the U.S. Army CRRDC also measured surface contamination. In addition, they measured ﬂux of molecules from mines in both air and water. Mines form one of the larger groups of explosive bearing targets for trace chemical sensors. Therefore, these measurements provide some insight into the concentrations at the source

74

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

Figure 4.2

Typical UXO items.

where the migration of the explosive molecules through their environment begins. Hopefully, they eventually arrive at a location accessible to sampling, whether the actual source or target is a mine or something else. Surface contamination was found to vary widely from mine to mine, but effective average contamination from a number of mines of several national origins seems to be about 15 to 18 ng/cm2 . Contamination in the painted surface of mortar and artillery shells was similar [8]. Flux measurements by Leggett and colleagues are brieﬂy summarized in Tables 4.2 and 4.3 [9]. Two features are immediately apparent: (1) The ﬂux of TNT is smaller that that of dinitrobenzene (DNB), while the greatest ﬂux is always for DNT, and (2) the ﬂux from any is greater when the mine is submerged in water, compared to the same mine in air. They also determined that,

SOURCE OF THE MOLECULES

TABLE 4.2 Mine TM5 PPM2 TMM1 PMA1A PMA2 VS-50

75

Landmine Flux into Air at 20◦ C (ng/mine/day) Construction

TNT

DNT

DNB

Polystyrene Unknown Metal Polyvinylchloride Polystyrene Unknown

1,380 128 740 207 24

15,100 12,800 1,720 1,550 282

4,500 3,480 282 358 332

RDX

14

From [9] Leggett, et al., 2001.

TABLE 4.3 Mine Type PMA-2 PPM2 VS-50

Comparison of Mine Flux into Air and Water at 22◦ C (ng/mine/day) Medium

TNT

DNT

DNB

Air Water Air Water Air Water

21 1,270 2,040 4,640

240 720 2,110 6,690

240 1,300 460 1,000

RDX

8 1,460

From [9] Leggett, et al., 2001.

at least for these mines, the ﬂux rate follows an exponential temperature dependence, where the exponent, surprisingly, is not dependent on the case material or mine type [1, p. 23]: φ ∝ e0.11T where φ is the ﬂux and T the temperature difference (in ◦ C). Summary of Landmine Flux Results Since no one has devised a method of directly measuring the ﬂux of explosive molecules from a mine, whether in situ or in the laboratory, several laboratory measurements have been reported in which the mine was placed in a sealed container, surrounded by soil, water, or air. The concentrations of explosive molecules in the surrounding media were then measured at intervals of several days and the ﬂux inferred from the total concentration divided by the elapsed time. This likely provides the best estimate that can be expected. The various measurements have substantial variation, depending on the techniques and media used. Phelan and Webb describe several experiments [1, pp. 23, 24]. It appears that a reasonable expectation of ﬂux of explosive compounds from a buried landmine that move into the surrounding soil will be in the range of 1 to 200 μg/day. There are some complications, of course, since the surrounding soil produces a level of resistance, or “back pressure,” to the ﬂux of the molecules. While the mechanisms are complex, the net effect is that wet soil permits a lower diffusive ﬂux than dry

76

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

soil [1, pp. 24, 25], primarily because biological activity is greater in wet soil. These, and other environmental effects, begin to act on the molecules as they leave the source. Transient and Steady-State Conditions From the landmine studies we readily conclude that the “source term” for these molecules has an initial “spike,” or increased rate, in the days or weeks after the mine is placed. This rate then decreases to some more or less constant level and may remain at that level for years. The initial spike comes from surface contamination, while the long-term rate is primarily from diffusion through the case and seals or leakage through imperfections or damage. The rates are clearly subject to environmental factors, principally temperature and soil wetness. Nevertheless, it seems clear that, at least in the case of landmines, there is a continuing ﬂux of molecules that provide a potential for detection. 4.2.3

Other Munitions

Most of the research accomplished to date on source ﬂux rates has been directed toward landmines. This is generally because global publicity has focused attention, and hence funding, on them. Other ERWs are acknowledged, but similar measurements are more limited. We can recognize that many of the same processes will be operative. However, since landmines are intended to be placed, most often, by hand, they can be manufactured effectively from various polymer materials, rather than steel. Munitions that are designed to be dynamically placed, such as bombs, mortar rounds, or artillery shells, must naturally have stronger, usually steel, cases. These munitions, when they appear either as duds or as abandoned UXO, will have no signiﬁcant diffusion, except through seals or o-rings. This diffusion should be expected to be considerably less than that available from mines. Duds often have case damage that can provide leak paths for the molecules. Naturally, this is not quantiﬁable in any predictive way. The other munition type of major interest, IEDs, are usually assembled a short while before they are placed, hence any molecules available for detection must be expected to come primarily from surface contamination and leaks, rather than from diffusive processes. Fortunately, for the operator of the sensing system, IEDs, by their nature, are seldom assembled with the care or precision of a manufactured device. Hence, we may perhaps harbor a reasonable hope that the source produces a ﬂux of molecules adequate to permit detection. 4.3

TRANSPORT OF THE MOLECULES

Once a source of molecules exists, that source becomes the beginning of a pathway that leads, hopefully, to the sampling portion of a sensor system. That pathway may be tortuous; it may be broad. There are a number of effects that alter the rate at which molecules become available to a detector. That rate, expressed as a concentration level per unit time within the medium where the detection

TRANSPORT OF THE MOLECULES

77

is being attempted, largely determines the success or failure of a given system in locating the source. The medium of search may be air, water, soil particles, or perhaps even plant or animal4 material, depending upon system design and the suspected location of the source. In this section we will examine some of the main processes that affect the concentration of molecules available for detection. 4.3.1

Buried Sources

4.3.1.1 Competing Effects When we began studying these effects at Sandia in the early 1990s we used a diagram similar to Figure 4.3. This was an aid to isolating the various effects so that they could be studied more effectively. Because we were concentrating our study on landmines, we diagrammed the process for a buried mine. Of course, the basic processes would be the same for any buried munition; but the source ﬂux rates would vary with the different munition construction. Figure 4.3 illustrates some of the conditions and processes that determine the concentration of molecules available to the sensor intake. This diagram was developed by Phelan and uses terms from environmental and agricultural science. Understanding of each of these processes is needed to make accurate estimates of the expected concentration at any given time and location. Phelan and Webb discuss these processes in some detail [1, p. 54ff]. We will consider them brieﬂy in the following sections. Figure 4.4 is a similar diagram that would apply to a mine or other munition buried in the bed of the sea, a lake, or a stream. In Figure 4.4 the munition is shown partially buried, both to illustrate the different possibilities and to remind us that, while partially buried objects on dry land are usually found visually, submerged objects demand different search techniques. The exploitation of plumes will be discussed in Section 4.3.4 and in Chapter 5. 4.3.1.2 Partitioning of Molecules in Ground Partitioning is used to describe the sorting of the molecules according to their relative absorption or solubility. The three double-headed horizontal arrows in Figure 4.3 indicate the partitioning opportunities available to the molecules. These processes are reversible; the molecules may reside in one state for a while, then move to the other, later to return to the former state. The vertical, single-headed arrow associated with the mine represents the irreversible efﬂux of molecules from the source. Some molecules will be adsorbed on the surface of soil particles, some will be in solution in the water present in the spaces between the soil particles, and some will exist in the vapor phase, as free molecules, neither adsorbed nor in solution. Partitioning is a dynamic process, changing as soil conditions change. The principle dynamic parameters at a given location are moisture content and temperature. 4 Researchers

have intercepted bees returning to a hive and pureed them for sampling in an attempt to ﬁnd molecules collected by the insects during their foraging.

78

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

Figure 4.3

Processes affecting molecules released from a landmine.

Of course, in a different location, with a different soil type the partitioning will form differently. The single-headed horizontal arrow at the lower right indicates the irreversible processes of plant uptake and biological decomposition of the molecules. These processes will be discussed in later sections. Partitioning takes on a more limited role in Figure 4.4, where the vapor state is assumed negligible. There is still some partitioning between the liquid and solid conditions, but even this is neglected when the leak is directly into the water above the seabed, as from the left end of the seamine. Since this condition has not been studied in detail, when someone does study it we may ﬁnd that Figure 4.4 needs substantial revision. Such a study is certainly a fertile ﬁeld for research. The processes diagrammed are constantly working, always seeking an equilibrium condition, but as local environmental factors change, so does that equilibrium. Hence the molecules released from the source are subject to continual change of their situation. In the following paragraphs we brieﬂy discuss the processes at work and how they determine the availability of molecules for sampling. It will become apparent that, much like the search for the “mother lode”

TRANSPORT OF THE MOLECULES

79

Figure 4.4 See color plates. Processes affecting molecules released from a sea mine, partially embedded in the seaﬂoor.

of a gold mine, the easiest place to sample is not always the richest source of molecules. Sorbing, Solution, and Vapor In the case of buried sources, such as landmines, the dominant processes are the sorbing of molecules onto the surface of solid particles and the dissolving of molecules into solution in the moisture available in the soil. Normal soil consists of particles packed together with spaces, often called pores, between them. The relative size of these pores varies with the compaction of the soil. In most soils these pores between particles are ﬁlled with a mixture of air and moisture, called vapor and liquid in Figure 4.3. The pores form potential paths for migration of molecules. The processes controlling migration through these paths are discussed in the following sections of this chapter. Fortunately, for anyone searching for these molecules as an indicator of the presence of the source, these processes drive most of them upward, toward the surface. Explosive molecules tend to be rather “sticky,” that is, they readily sorb, or adhere, to available surfaces. They are also soluble in water. When water is present in a pore adjacent to a particle, some molecules entering that pore will sorb to the particle and some will enter solution in the water. Since the pore is not entirely ﬁlled with water, but also contains some air, some molecules will also be found in the vapor state, simply as free molecules, neither sorbed to a

80

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

Mass Fraction, Solid 95%

Mass Fraction, Vapor 0.00001%

Figure 4.5

Mass Fraction, Liquid 5%

Explosive molecule distribution in soil at 50% pore saturation, 23◦ C, Kd ∼ 4.

soil particle nor in solution in the soil moisture. Figure 4.5 illustrates5 a typical equilibrium condition in a soil with 50% pore saturation, that is, for a soil that has half its maximum possible moisture content. Clearly, the great majority of the molecules is sorbed to soil particles, and very few are in the vapor state. Here is a point of major importance for anyone designing a system for ﬁnding these molecules. When any chemical is distributed between vapor and liquid in the soil pores, the proportion residing in each phase is an important parameter, perhaps more important in agriculture than in any other area of study. Researchers in agriculture have adopted a measure of this proportion called the Henry law constant, with symbol KH . It is deﬁned as KH =

CG chemical concentration in gas phase(g/cm3 air) = CL chemical concentration in liquid phase(g/cm3 water)

(4.1)

That characteristic of a chemical called vapor pressure determines the quantity of that chemical that can be contained in a speciﬁc volume of air at a given temperature and pressure. Vapor pressures of the explosive chemicals of interest are strong functions of temperature. See Chapter 2. In fact, it has been shown [10] that, for the entire temperature range of interest, the vapor pressure of DNT is about 20 times that of TNT, but that the vapor pressures of both show similar temperature dependence. The vapor pressure of each increases by about a factor of 4 for each decade increase in temperature, from about 5◦ C to about 50◦ C. Also, the solubility of these chemicals is weakly dependent on temperature below 20◦ C, but above 20◦ C the temperature dependence increases substantially [11]. In both ranges DNT is about one and a half times as soluble as TNT. 5 The

unpublished diagram in Figure 4.5 was developed during the research at Sandia in the 1990s, based principally on work by Phelan.

TRANSPORT OF THE MOLECULES

81

Phelan and Webb combined these effects [1, p. 27] to determine the dependence of KH on temperature. Their chart is reproduced as Figure 4.6. Examination of this chart reveals several facts of interest. First, we see that the molecule concentration in the vapor phase is always several orders of magnitude smaller than that in the liquid phase. Next, we see that there tends always to be a higher proportion of DNT in the vapor phase relative to the proportions for TNT, by about a factor of 10. At the higher temperatures we can expect to ﬁnd nearly 100 times the proportion of molecules in the vapor phase as at the lowest temperatures. Of course, these are relative, rather than absolute values. Also, as the proportion of molecules in the vapor phase increases, there must be a corresponding decrease in the proportion in the liquid phase. Just as the Henry law constant is used to quantify the partitioning between air and water, it is useful to deﬁne a parameter to quantify the partitioning between soil and water. The linear adsorption coefﬁcient, Kd , has often been deﬁned for this purpose: Kd =

CS sorbed concentration (μg/g) = CL aqueous phase concentration (μg/mL)

(4.2)

Kd has units of milliliters per gram (mL/g). They then summarize the work of several research teams and individuals working to experimentally determine values for Kd [1, pp. 28–30]. These efforts were made in a number of different soils, with different moisture levels, different organic content, and different soil chemistry. Published results for these conditions naturally differ widely; however, it appears that values of 0.5 ≤ Kd ≤ 4.0

Henry's Law Constant

1.E−04

1.E−05

1.E−06 KH DNT KH TNT

1.E−07

1.E−08

0

5

10

15

20

25

30

35

40

Temperature (°C)

Figure 4.6

Henry’s law constant for DNT and TNT.

45

50

82

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

describe most soil conditions. The differing experimental techniques account for some of the variation. No dominant parameter causing the variations appears, but TNT seems to always have a higher value of Kd than DNT for the same conditions. In general, though [1, p. 26]: Soils with greater amounts of organic matter (agricultural or forest soils) or minerals (compared to desert sand) will sorb greater landmine signature chemicals,6 leaving less available for transfer to the air for vapor sensing.

As discussed earlier, the proportion of the molecules in the vapor state within the soil near a source is around 6 orders of magnitude less than the sorbed or dissolved portions. This condition was stated for 50% moisture saturation. For extremely dry soils, this ratio becomes even more pronounced. Phelan and Barnett deﬁned a soil–air partitioning coefﬁcient, Kd , in a manner parallel to the soil–water partitioning coefﬁcient, Kd [12]: Kd (w) =

CS concentration sorbed on soil particles (g/g) = CG concentration in vapor state in soil pores (g/mL)

(4.3)

Note here that Kd is indicated to be a function of w, the gravimetric moisture content in grams/grams. That dependence on soil moisture is most pronounced in very dry soils, up to about 11% saturation, above which Kd seems to asymptotically approach moisture content independence. In the most dry soils Kd may be 8 or more orders of magnitude greater than at 50% saturation. A plausible explanation for this is that, as the soil becomes wetter, water displaces some of the explosive molecules sorbed on the soil particles, causing them to be released into the vapor state. In the driest soils almost all the molecules sorb to the soil particles. Dry soils, therefore, tend to act as reservoirs, or storage, for explosive molecules. These molecules can be released into the vapor state when the soil becomes moistened. Clearly, though, in totally saturated soil, where there is no air in the pores, there can be no molecules in the vapor state. Therefore, both at near-zero saturation and at full saturation the proportion of the molecules in the vapor state becomes vanishingly small. The concentration is substantially larger at some intermediate values of saturation. Therefore, we should expect to ﬁnd an optimum condition of soil moisture, at which there is the highest concentration of molecules in the vapor state. Phelan and Webb show results from algorithms they developed [13–15] predicting those optimum conditions. They offer partitioning charts like Figures 4.7 and 4.8 [1, pp. 30–35]. The charts shown represent only one soil type, that is, only one value of Kd . Still, they give us a sense of the dynamics of partitioning as moisture content changes. The charts are based on the concept that “the whole is equal to the sum of the parts.” They illustrate the relative concentrations in each phase as functions of soil moisture. Their results indicate that a saturation of about 17.5% provides the highest concentration of molecules in the vapor state. 6

Principally TNT, DNT, and DNB.

TRANSPORT OF THE MOLECULES

83

1.0E-05 DNT 1.0E-06

Mass Fraction

1.0E-07 TNT 1.0E-08 1.0E-09 1.0E-10 1.0E-11 1.0E-12

0

10

20

30

Figure 4.7

40 50 60 70 Soil Saturation (%)

80

90

100

Soil vapor mass fraction.

1.00 0.90 TNT Soil Solid Phase

0.80 Mass Fraction

0.70 DNT Soil Solid Phase

0.60 0.50 DNT Soil Liquid Phase

0.40 0.30 0.20

TNT Soil Liquid Phase

0.10 0.00

0

10

20

30

40 50 60 70 Soil Saturation (%)

80

90

100

Figure 4.8 Soil solid and liquid phase mass fractions, Kd = 0.9.

Phelan and Webb further show that while increasing temperature in the range from 5 to 45◦ C substantially increases the actual concentration in the vapor state, the maximum value still tends to be at about 17.5% saturation. In fact, the vapor state concentration at 23◦ C is about 10 times that at 5◦ C; at 45◦ C the mass fraction in the vapor state is about 5 times more than at 23◦ C [1, pp. 30–35].

84

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

Thus, it would seem that for anyone planning to sample for molecules in the vapor state, a very warm day with 17.5% soil moisture would be ideal. Figure 4.9 illustrates the results of another of Phelan and Webb’s [1, p 31, Table 13] calculations mentioned above. When we compare Figure 4.9 with Figure 4.5, the effect of soil type, as represented by the differing Kd values, becomes very apparent. As mentioned earlier, the foregoing discussion has signiﬁcant implications for anyone designing a system for ﬁnding these molecules. We have seen that the dominant factors include soil type, including organic material, and soil moisture and temperature. But this is not a static condition. The molecules are migrating, forced by several processes. 4.3.1.3 Diffusion One of these processes is diffusion, the dominant transport mechanism for that portion of the molecules that are in the vapor state. As discussed above, this is a minor proportion of the total number of molecules in the soil. However, it is the mechanism that brings molecules directly to the open air above the source, where they are available to the most straightforward collection techniques, whether by mechanical or animal sensor systems. This process is indicated in the left-hand oval of Figure 4.3. Diffusion also occurs in the liquid state, as indicated in the center oval of Figure 4.3. In this phase it forms a minor, rather than the major, mechanism. In fact, it is difﬁcult to separate vapor diffusion and liquid diffusion, and for this application separation is not really required. Therefore, Phelan and Webb [13], combining earlier work of Hamaker, Millington and Quirk, and Jury, deﬁned an effective diffusivity that combines the two processes. They produced Figure 4.10 to illustrate.

Mass Fraction, Solid 86%

Mass Fraction, Vapor 0.00003%

Figure 4.9 Kd = 0.9.

Mass Fraction, Liquid 14%

Partitioning of explosive molecules in soil at 15% pore Saturation, 23◦ C,

TRANSPORT OF THE MOLECULES

85

1 Vapor Diffusivity Solute Diffusivity Effective Diffusivity

Diffusivity (cm 2/day)

0.1 0.01 0.001 0.0001 0.00001 0.000001 0.0000001

0

Figure 4.10

10

20

30

40 50 60 70 Soil Saturation (%)

80

90

100

TNT vapor, solute, and effective diffusivity.

However, as we discussed above, when the soil becomes very dry, below about 11% saturation, the molecules tend primarily to sorb onto the soil particles. This reduces the effect of vapor diffusion. Therefore the validity of Figure 4.10 is reduced in those conditions. Realizing this, Webb and his colleagues explored it in more detail [14]. This work showed “a dramatic decline in effective diffusivity below 15% saturation” [1, p. 43]. This conclusion is of paramount importance. It means that there is no effective mechanism for transporting molecules from a buried source to the surface when the soil is very dry. The transport process can begin anew when rainfall, or perhaps artiﬁcial introduction of water to the soil, raises the saturation to levels where diffusion can again occur. The reservoir of sorbed molecules simply awaits a mechanism. 4.3.1.4 Convection In addition to the double-headed arrow labeled “Diffusion,” the center oval in Figure 4.3 contains a downward arrow labeled “Precipitation” and an upward arrow labeled “Evapotranspiration.” Both these latter terms refer to movement of the water within the soil. They form the major drivers of convection within the soil. In fact Phelan and Webb list three key conclusions concerning transport of the subject molecules within the soil [1, p. 42]: • Movement of landmine signature chemicals is controlled by chemical and soil

properties, and driven mostly by the movement of water in soils. • Water transports more TNT, DNT and DNB by convection than occurs by either vapor or solute diffusion. • Conditions that cause upward evaporation of soil water in proximity to the landmine will be most beneﬁcial to chemical sensing.

In general, water applied to the soil surface tends to move deeper under the inﬂuence of gravity, taking the molecules into solution and carrying them lower.

86

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

This would seem, at ﬁrst glance, to make them even more inaccessible. That would be true if the inﬂux of water never stopped. However, when it does stop, the process of evaporation begins, driven mostly by solar heating. This causes the water to move to the surface, carrying with it the molecules in solution. Of course, plants take in some of the water, which then moves through their systems. Molecules dissolved in that portion of the water seem to remain within the plants rather than being released into the air. Effects of Moisture, Rain, Dew, and Evaporation These weather effects become major factors in the transport of the molecules from the source to the surface of the soil, and thence into the air. Phelan and Webb again list some key conclusions [1, p. 26]: • Soil moisture has a tremendous effect on soil-vapor sorption. Dry soils will

sorb about 10,000 times more landmine signature chemicals than damp soils. This depresses the vapor levels the same amount. This process is reversible, so daily morning dew is valuable for vapor sensing, and afternoon drying is detrimental for vapor sensing. • The soil acts as a temporary storage reservoir for the landmine signature chemicals, releasing them when dew or rain falls, and collecting more as soil water evaporates.

When we began this study at Sandia, we did a number of experiments to gain an understanding of the migration of explosive molecules under ﬁeld conditions. We also did some laboratory experiments to extend this understanding. One of the more enlightening experiments is illustrated in Figure 4.11. Phelan and his colleagues described in detail the techniques of the experiment and the complete results [15]. They also developed a predictive mathematical model along with the experiments. The experiment vividly illustrated results that have been regularly observed in ﬁeld trials and are reﬂected in the last quotation above. These results clearly show the major effect of convection in the transport process. A direct implication of this experiment becomes apparent: It is quite possible to determine favorable and unfavorable times for vapor ﬂux sampling, based on moisture events. Furthermore, it may sometimes be possible to artiﬁcially enhance sampling by appropriate local introduction of moisture.7 Otherwise, it may prove more effective, in some environments, to sample surface soil particles rather than vapor. The simulated weather changes that produced the variations in surface vapor ﬂux rate shown in Figure 4.11 began with a soil sample with about 50% pore moisture saturation in a chamber under an artiﬁcial atmosphere. A quantity of DNT was introduced beneath the surface, at a depth typical of a landmine, about 3.5 cm. The atmosphere in the chamber was controlled to 50% relative 7 An

issue of curiosity, engendering some little debate, is whether dogs do this with their exhaled breath before each inhaled breath collects samples for the aroma being sought.

TRANSPORT OF THE MOLECULES

87

2, 4-DNT Surface Flux Artificial Drying Begins

1000000

2, 4-DNT (pg/min)

100000 10000 Relative Humidity Reduced to 0%

1000 100 Relative Humidity 50%

10

Artificial Wetting Begins

1 0

10

20

30

40

50

60

70

80

90

100

Time (days)

Figure 4.11 Effect of wetting and drying on surface vapor ﬂux of 2,4-DNT.

humidity. In a bit less than a week ﬂux was measurable above the surface. That ﬂux increased rapidly until it began to approach an equilibrium state. From prior experiments it was understood that equilibrium was expected after about 35 days. Therefore, on day 35 the relative humidity was abruptly reduced to 0%. Two effects are visible. There is the immediate reduction in ﬂux as the surface dries and molecules are sorbed to the surface particles. This is followed almost immediately by a marked increase in ﬂux as evaporation draws moisture into the atmosphere above the surface, bringing dissolved molecules with it, tending toward a new equilibrium. On day 44 the soil was subjected to a drying process. As the moisture leaves the soil, the molecules have more opportunity to remain in both the vapor state in the pores, as air replaces water, and in the solid phase, sorbed to the surface of soil particles. Drying was continued until day 69, simulating the normal process in soil between rain or dew events. At that point water was introduced to simulate precipitation. The dramatic spike in ﬂux rate reﬂects the ability of the newly introduced water to transport molecules to the surface. This is consistent with the previous discussion where we saw that very dry soil holds on to its molecules. The drying–wetting cycle was repeated with similar results on day 79. The latter two cycles exhibit behavior similar to a ﬁeld condition where there is a substantial diurnal cycle of heating and cooling, with morning dew. Once molecules reach the surface, whether from a buried source as in the foregoing discussion or by some other transport mechanism from a nonburied source, they tend to sorb onto the surface particles. Some will be in the vapor state and some in solution, but as any water evaporates, they must revert to vapor state or sorb to surface particles.

88

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

Figure 4.12

4.3.1.5

Unfuzed Soviet-type TM-62P2 being placed for experiments at Sandia.

Residue on Surface

SNL Results from Experiments Among the experiments we conducted at Sandia during the early stages of our investigations was the establishing of a sample mineﬁeld, using Soviet mines supplied by the U.S. Marine Corps. We buried them according to published Polish military doctrine for the particular mine type. They were buried, unfused, in the normal prairie loam found in Albuquerque, New Mexico. Figure 4.12 shows a mine being placed. The snap-in plastic cover over the fuze well is not a seal. It was left in place to approximate the presence of a fuze for overburden thickness control. The fuze well is a continuous part of the case and forms a watertight seal over the enclosed explosive. This mine was placed on February 7, 1997. After several months, on June 26, 1997, the concentrations were measured as shown on Figure 4.13. Soil samples for the vertical series of measurements were taken directly above and below the mine, at 1-inch intervals. The horizontal samples were taken at the upper surface burial depth, 1, 2, and 3 inches from the mine case. As noted on the diagram, TNT concentrated near the surface. In this case we ﬁnd the DNT has not migrated as far from the source as the TNT. As this was early in the process, no real explanations were offered for the difference, and insufﬁcient detail on weather effects were noted. Nevertheless, this diagram illustrates a condition found by several deminers as they used dogs to search for mines.8 As they studied effects of success and failure and realized that very often the molecules they were searching for were in the surface layer, rather than being available in the air. This led to some interesting realizations of the relationship between physical locations of surface concentrations and the buried source. 8

Personal communication with Dr. Vernon Joynt in April 2004.

TRANSPORT OF THE MOLECULES

89

Figure 4.13 TNT and DNT concentrations measured near the buried TM-62P2 antitank mine.

Dried Puddles Concentrate Molecules In our common experience, we have all observed the formation of puddles after a rain. We realize, without much analysis, that the puddle is not formed from rain that fell in only that location. It contains water that fell nearby and ﬂowed to that area, which is at the locally lowest elevation. If the rain falls on an area that has buried sources of explosive molecules, then some of those that were sorbed to the surface particles above the source will be dissolved and carried into the puddle. When puddles dry they leave a concentration of molecules on the surface soil particles. Thus, an irregularly shaped area of relatively high concentration of molecules may appear some distance from any buried source. See the discussion on Figure 8.2 p 182. This concentration in puddles does not require a buried source. If a source on, or above, the surface receives rain, concentrating puddles can form. Since the presumed reason for searching for trace explosive molecules is to locate the source, some ingenuity may need to be exercised to complete that task. In fact, Phelan and Webb [1, pp. 70, 71] report on work by Hewitt et al. [16] where they buried mines on a gentle slope. The signatures from these mines were found to form in patterns where concentration decreased with distance (a few feet) from the mine as the surface water ﬂowed down the slope away from the mine. 4.3.1.6

Above the Surface of the Ground

Boundary Layers Molecules may reach the surface in each of the three states. Those dissolved in water will, at some point, either be released into the vapor state or become sorbed to surfaces as the water evaporates, unless the water

90

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

becomes part of a larger volume that does not evaporate, but ﬂows into a stream or pond. Transport of molecules that remain dissolved in water will be discussed in a later section. Molecules that originate from a buried source, upon reaching the surface, may be carried away by wind currents. This is not always the case since they emerge from their upward journey through the pores in the soil into a region of relatively stagnant air called a boundary layer. Ludwig Prandtl introduced the concept of boundary layers in 1904. Since they play a large role in determining the parameters of ﬂuid ﬂow, they have been extensively studied. A boundary layer is that region near a surface where the ﬂuid ﬂow is dominated by the presence of the surface. The ﬂuid cannot ﬂow through the surface; but there is always some attraction between the molecules of the ﬂuid and those of the surface, the surface tension effect. In addition, at low velocities, the viscous forces in the ﬂuid dominate the kinetic forces. Therefore, the ﬂuid immediately adjacent to the surface is restrained in its normal tendency to move with the rest of the ﬂuid. The result of this restraint is a velocity gradient. The velocity increases from effectively zero at the surface to the nominal ﬂuid velocity at some distance away. The ability of the moving ﬂuid to produce motion of objects on or very near the surface, from molecules to grains of sand, depends moving that object out of the boundary layer into the main ﬂow ﬁeld. This normally requires turbulence on the appropriate scale. The thickness of the boundary layer and the turbulence within it are determined by a number of factors, including ﬂow velocity and surface roughness. Beginning with Prandtl’s original work, there have been many empirical equations proposed for estimating the boundary layer thickness. All recognize that the thickness is inversely related to some fractional power of the Reynold’s number. δ∝

1 (NRe )1/n

where d is the boundary layer thickness, NRe is the Reynold’s number, and n varies from 2 to about 7 among the theories. Since the Reynold’s number is deﬁned as NRe =

ﬂow velocity × characteristic length × ﬂuid density V lρ = μ ﬂuid viscosity

(4.4)

the thickness is observed to depend inversely on the velocity. That means that boundary layers are thicker for slower ﬂows. This explains our common observation that higher wind velocities raise dust from locations that lower winds do not. The higher winds producing a thinner boundary layer are able to move particles that remained within the thicker boundary layer at lower wind velocity. Since searching for trace vapor signals in other than nearly calm conditions is almost certainly a futile exercise, the boundary layers of interest are relatively thick, perhaps a few centimeters.

TRANSPORT OF THE MOLECULES

91

However, the molecules percolating up into the boundary layer from beneath the soil surface tend to become trapped in the stagnant laminar sublayer of the boundary layer. This sublayer is usually much thinner than the overall turbulent boundary layer, since it is dominated by viscous and surface tension forces, rather than by velocity. Phelan and Webb call this the “chemical boundary layer” and state categorically that “there will generally be no chemical signature” above this chemical boundary layer [1, p. 52]. Other factors do intervene. Signiﬁcant solar heating of the soil surface, so that the soil becomes warmer than the air, causes vertical thermal convection currents to develop within the boundary layers. This introduces turbulence or instability that acts to move the chemical signature up into the free air. When the molecules are moved into the free ﬂow of the air, the effect is to reduce the concentration by dilution. Conversely, when the soil surface is cooler than the air, thermal convection is inhibited, with the result that the molecules are effectively trapped in the boundary layer. This effect is strengthened by the cooling of the air adjacent to the surface, which increases its viscosity. Higher viscosity lowers the Reynold’s number, thus decreasing boundary layer thickness. Thus, we can expect that the ﬂux of molecules that form the chemical signature at any given location will vary with time and weather conditions. This can even include intermittent transients. In addition, mechanically disturbing the surface can result in moving particles or molecules out of the stagnant part of the boundary layer where they can be moved by the wind. This effect can often be observed in desert regions. As a wind increases, it is often possible to see dust being lifted from footprints left since the last wind. If there is vegetation, such as short grass, the boundary layer may be largely submerged in it. Molecules do not readily ﬁnd their way through this boundary layer; hence, it may be necessary to force them into a collection device by suction. That appears an integral part of a mammal’s technique. In most cases the molecules that form the desired chemical signature will be found very close to the surface. Realizing this will enable more effective sampler designs to be ﬁelded. Some of the molecules that do make their way into the free air above the boundary layer are likely to sorb onto the surface of any object that is in their ﬂow path. Once this happens, that molecule is effectively lost for collection by vapor sampling techniques, reducing the available concentration in a sample. In many search areas plants form the most available surfaces for molecules to fall upon. Hence, it is possible that plant surfaces near a source might form a reservoir for molecules that could be proﬁtably exploited by innovative sampling techniques. Certainly, it is well recognized that when plants take in water through their root system that they may be also taking in the molecules released from a nearby source [17]. 4.3.1.7 Plant Uptake When the molecules are taken into a plant’s system, they may become concentrated in parts of the plant or they may begin to break down. These processes have scarcely been examined. When they are, it is possible

92

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

that new opportunities will be revealed. This area needs thorough investigation. This process is represented in Figure 4.3 by the lower right oval. That oval also includes microbiological processes that cause breakdown of the molecules. 4.3.1.8 Soil Type Effects Those microbiological processes represented in that lower right-hand oval of Figure 4.3 depend heavily on soil types. They are even more strongly dependent on moisture level and temperature. In addition to microbiological degradation, there is abiotic degradation from chemical reactions with soil components. These processes are most pronounced when the soil contains iron-bearing minerals [18]. Phelan and Webb report work from several studies [1, p. 36] and summarize the results: • Landmine signature chemicals change form, chemical properties, and eventu-

ally become eliminated from soil systems by microbiological and soil mineral degradation reactions. • Both biologic and abiotic reactions require water for degradation reactions to proceed.

Degradation times are normally stated in terms of half-life.9 Measured results vary substantially, depending on experimental technique, but in wet soil at warm (>20◦ C) temperatures both TNT and 2,4-DNT can show a half-life of one day or less. At temperatures just above freezing, this increases to a couple of weeks; while at temperatures below freezing, the half-life extends to years. Different soils show varying rates of degradation. Phelan and Webb’s summary included both laboratory and ﬁeld measurements from multiple sources. Results from laboratory and ﬁeld results correlate reasonably well. Continuing their summary [1, p. 36]: • Laboratory measurements of TNT, DNT, and DNB found degradation rates

to be dependent on soil type, soil moisture content and temperature. • Higher clay content and organic matter content soils have higher degrada-

tion rates. • Soil moisture contents greater than 1% cause very fast degradation rates (half

the amount degrades over the period of one day).10 • Soil moisture contents less than 1% preserve landmine signature chemicals (half the amount degrades over the period of 3 years) • Only subzero (◦ C) conditions limit degradation. The greater the temperature, the greater the degradation rate.

Table 4.4 [1, p. 37] lists the principal degradation products of the signature chemicals. See the additional discussion in Chapter 2. It may be useful to consider searching for these compounds in addition to the parent compounds when conditions for rapid degradation are present, or when the source being sought may have been in place for some extended time. 9 Half-life is the time, usually quoted in days for this application, during which the concentration is reduced to one-half its previous value. 10 At temperatures above 20◦ C (author’s clariﬁcation).

TRANSPORT OF THE MOLECULES

TABLE 4.4

93

Parent and Degradation By-products of TNT, DNT, and DNB

Parent Compound

Degradation By-product (abbreviation)

2,4,6-TNT

4-Amino-2,6-dinitrotoluene (4A-DNT) 2-Amino-4,6-dinitrotoluene (2A-DNT) 2-Amino-4-nitrotoluene (2A-NT) 4-Amino-2-nitrotoluene (4A-NT) 3-Nitroaniline (3-NA)

2,4-DNT 1,3-DNB

4.3.1.9 Climatic Effects Weather conditions cause several effects on the concentration level of the chemical signature at the soil surface and immediately above it. Earlier discussion focused on the cycles of rain or dew and the associated evaporation. This is the largest effect, but it is closely associated with the diurnal cycle of heating and cooling of the soil. Direct solar radiation and long-wave radiation from the atmosphere heat the surface during the day, while long-wave radiative transfer to space cools it at night. Soil surface temperature may also be affected by air temperature as warm or cold fronts move through the area. The amount of temperature change in each diurnal cycle is greatly inﬂuenced by cloud cover and relative humidity. Soil color greatly affects the magnitude of temperature changes due to radiation. Plant coverage may provide enough insulation to greatly reduce that effect. In addition to the diurnal cycle, the seasonal cycle produces much variation in the surface temperature, hence in the processes that control concentration. Concentrations tend to be much higher in the warm summer temperatures when evaporation is driving percolation of the molecules to the surface, than in the winter, when activity may become halted by freezing. Even if no freezing occurs, the processes are slower at colder temperatures, producing lower signature concentrations. These weather changes affect the signature concentration in several ways. Evaporation increases with increasing surface temperature and decreases with increasing humidity. Boundary layers are less stable when the surface is undergoing direct heating, causing vertical convection currents that carry molecules through the boundary layers. Microbiological processes producing molecular degradation are much more pronounced in warmer soils. These processes are greatly restricted when temperatures fall below freezing (0◦ C). Continued heating reduces soil moisture. When this results in the saturation dropping below 8 to 10%, the available signature drops dramatically, for reasons discussed previously. Drought conditions may cause this to persist for long periods. Conversely, continual rainfall will drive the soil saturation to a level that precludes molecules in the vapor state. In either condition, without a change in weather to permit evaporation no signature will be available to vapor state collectors. Prolonged or heavy rainfall may produce erosion, moving the surface soil to other locations. It will almost certainly produce runoff that carries dissolved

94

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

TABLE 4.5 Estimated Vapor Concentrations in the Air Boundary Layer from Surface Soils Residue Data and Ks/a Values TMA5 Chemical 2,4,6-TNT 9.4 × 10−4 2,4-DNT 2A-DNT 4A-DNT

PMA1A

ng/L

ppt

ng/L

ppt

9.4 × 10−4 1.5 × 10−1 9.0 × 10−3 1.1 × 10−2

0.1 20 1 1

8.5 × 10−4 3.0 × 10−1 2.3 × 10−2 2.0 × 10−2

0.1 40 3 2

molecules elsewhere, as discussed above. In either case, this condition can produce signature concentrations that are difﬁcult to relate to a source. These can, in effect, produce a form of false-positive search results. The concentration produces a true chemical signature, but not in a location readily associated with a source. One vivid photo, no longer available, exhibited this extreme condition, showing a mine partially uncovered by erosion in the bank of a wadi in Israel. The chemical signature might well have been discernable in the bottom of the wadi some meters distant from the actual source, misleading at best. 4.3.2

Concentration Estimates from Buried Sources

At the time of the studies described, and up to the present, a vapor sensing system with sensitivity adequate to routinely discover the vapor signature from buried landmines has not been ﬁelded in quantity. Therefore, it is necessary to estimate the concentrations that may be expected, so that system developers may form realistic design goals. Jenkins and his colleagues estimated the air concentrations, for one kind of soil and two types of mines, as shown in Table 4.5 [8]. The quantity Ks/a is equivalent to the Kd previously deﬁned. It was calculated as the ratio of soil residue to vapor concentration in their experimental samples. They were working at the limits of sensitivity of available instrumentation to produce this work. It is well established11 within the demining community that mammals have a vapor sensing capability several orders of magnitude more sensitive than current electronic systems, even laboratory systems, though recent developments (Chapter 7–9) are beginning to change that. Most mammalian ﬁeldwork has been done using dogs, but some work indicates that rats may be even more capable [19]. Other biological sensors may be even more capable. The well-known12 ability of some insects to detect one or two molecules of some chemicals provides a goal for sensor developers. 11 Several

conversations, 1996 through 2004, with Dr. Vernon Joynt, who has researched the use of dogs and rats as well as electronic systems in South Africa for locating explosives for many years. See Chapter 8. 12 Mentioned by Paul Patton in lecture notes on website: http://soma.npa.uiuc.edu/courses/bio303/ Ch13.html.

TRANSPORT OF THE MOLECULES

95

Phelan and Webb [1, pp. 73–78] report experiments that tried to determine the available concentrations by using established mine hunting methods employing a dog as the sensor system. They used several indirect methods, where the vapor in a suspected or known mineﬁeld is collected on ﬁlters and presented to dogs at a convenient site. The experiments compensated for the concentrating effect of the ﬁlters inherent in these methods. Among others, they used the MEDDS [20] system and the REST [21, 22] system. When they did those experiments in 2001, they found the best laboratory sampling and analysis capable of about 0.5 ppt (0.005 ng/L). They further note that some dogs could sometimes sense concentrations as low as 10−10 ppt [1, p. 75]. See Chapter 8. Unfortunately, even with best efforts no one is yet able to provide a deﬁnitive value that a sensing system developer can use as the available concentration near a buried mine. It will continue to be necessary to develop more sensitive sensors. However, it also becomes increasingly valuable to use them more astutely, based on the behavior of the molecules as discussed here. Whether using artiﬁcial or biological sensors to search for buried explosives, a few things become apparent as important. Among them are: • • • • • • • •

Damp soil releases more vapor than very dry soil or saturated soil. The concentration of molecules may be expected to be higher in the soil surface than in the air above it. Vapor signatures become stronger or weaker at some times, depending on the moisture and temperature cycles. The air and chemical boundary layers tend to keep chemical vapor close to the surface; farther from the surface dilution becomes the dominant factor. There may be advantages to artiﬁcially inducing local moisture cycles, perhaps even in the immediate vicinity of the sample collection. There will be no vapor signature when the surface temperature falls below freezing. Molecules migrate on the surface after rains, following drainage patterns. Molecules persist in dry soils for a long time but degrade more rapidly in wet soils.

With these considerations, and others gleaned from the research, system developers can use innovative sample collection techniques to extend the capability of their systems to detect lower concentrations and locate or identify buried explosive bearing objects. The issue of buried objects has attracted a great deal of attention, especially the worldwide proliferation of landmines; hence, there have been funds for research. This research will have application beyond landmines. Much UXO is buried, some because it was buried for disposal, some because it became buried in the course of the conﬂict. However, understanding the way the molecules are released and how they migrate after release will also assist in applications where the munitions are not buried but are hidden in various ways. There are also other environments worthy of consideration.

96

4.3.3

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

Other Environments

We may reasonably expect IEDs to be disguised. Penetrating the disguise may frustrate the efforts of their user. Some objects may be hidden by debris. Expected vapor signatures from these items required search in air. Many ordnance items are used in water; others are disposed of there. Search for traces of explosives from them may be expected to be somewhat different. 4.3.3.1 Underwater Objects We made a brief allusion to another environment in commenting on Figure 4.4. That ﬁgure shows one example of an environment that is very different from a simple buried munition. Several things become apparent immediately. Clearly, the aqueous state is likely to dominate the entire process. There will be some sorbing onto particles of the bottom, but even for objects totally buried in the bottom that will be reduced, since the bottom will always be at a saturated state. Therefore, it is reasonable to focus development attention on ﬁnding molecules in the aqueous state, rather than the vapor state. The Ofﬁce of Naval Research (ONR) has sponsored substantial work of this nature. Some of it is reported in detail in Chapter 6. A look at Figure 4.4 illustrates some considerations when the object is submerged. It is worth noting that, while seamines are intentionally placed underwater for military purposes, they are not the only munition that may be found submerged. Of course, UXO is commonly found underwater, either having been dumped there, having been cargo on ships that sunk, or having fallen there as duds during conﬂict. The sea is not the only recipient of these ERWs. During conﬂicts some ordnance will be intentionally or incidentally placed in rivers, canals, streams, ponds, lakes, or irrigated ﬁelds. For locating these ERWs, the most important characteristic of a body of water is its current. Currents may be nearly nonexistent in some stagnant ponds or irrigated ﬁelds. Currents in streams and rivers are usually consistent in direction but variable in velocity. Currents in tidal areas oscillate in direction and vary in velocity. Rapid currents have higher Reynolds numbers, hence more turbulence. Even rather modest velocities in water produce turbulence because the Reynolds number is higher with the higher density than in air. Turbulence is not the asset in water that it is in air. In air, we saw that some turbulence is required to bring the molecules out of the chemical boundary layer. That may also be needed in water to move molecules away from the bottom surface. Turbulence away from the surface tends to break up the plumes of molecules that are diagrammed in Figure 4.4. Those plumes are the key to successful detection of an underwater object that is releasing the molecules of interest. One result of the ONR experiment at San Clemente Island, off San Diego, California, was a better understanding of the formation, persistence, and dissipation of these plumes. When a well-formed plume is available, it often becomes possible to follow it to its source; see Chapters 5 and 6. But how does the plume form? What processes control its persistence and its integrity? Are there other processes that control the concentration of molecules that provide the signature within the plume? Can we reasonably expect to ﬁnd

TRANSPORT OF THE MOLECULES

97

detectable concentrations outside a plume? There has been some research directed toward answering these questions speciﬁcally for explosive signatures. However, there has been much research into similar plumes by workers in biological studies. They refer to them as “odor plumes.” The following section discusses plumes in more detail. There has been almost no research for objects buried in the seaﬂoor13 that parallels the work of Phelan and Jenkins and their colleagues on land. Therefore, we can only surmise that processes similar to those they describe are at work. Clearly, as diagrammed in Figure 4.4, there will be diffusion and convection in the liquid phase. For buried objects, there must still be partitioning. There will be microbial degradation, in fact, almost certainly at a greater rate than for objects buried on land. That microbial degradation, much more than in air, can actively continue as the molecules enter the free water above the seaﬂoor, whether they are in a plume or not. In addition, some molecules in the water will be ingested by larger aquatic organisms. In stagnant water no plume may form. This could cause a condition where the signature is ample for detection but so distributed that no source location can be inferred. In tidal areas the plumes will reverse directions with the tidal currents. Clearly, there are enough unanswered questions to offer ample research opportunities. 4.3.3.2 Surface Source or Hidden Source Sometimes source munitions are hidden, either intentionally or unintentionally, with neither soil nor water above them. In this case, we may expect that chemical signatures will be found in the vapor state. It is possible that sampling of surfaces could lead to detection, but normally a search will be in air. Many animals routinely face this same task. Whether looking for a meal or a mate, a common approach is to follow an odor plume to its source. 4.3.4

Odor Plumes

A plume is a collection of ﬂuid molecules within a larger body of that ﬂuid that, with their suspended particles and dissolved chemicals, are moving as a more or less distinct ribbon through the containing body of ﬂuid. While some might call it tubular, it is often styled as a ribbon because of its high aspect ratio. The term “plume” is normally applied to a portion of the ﬂuid that is being rendered distinct by the injection, over a period of time, of the characteristic dissolved chemical or suspended matter. We think of plumes of smoke from a chimney or a stream of muddy water joining a ﬂowing stream of clear water. A classic form of plume, familiar to everyone, is the smoke from a just extinguished candle. Close observation of this plume in still air reveals that it is tightly constrained and very smooth for some distance above the wick. A few inches above the wick the character of the plume changes. It begins 13 The

term seaﬂoor is used generically to refer to the bottom of any body of water, even a ﬁeld ﬂooded for irrigation.

98

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

to twist, widen, and become less distinct. It rolls about itself and forms gaps, ﬁnally disappearing altogether. This behavior is characteristic of chemical plumes or odor plumes as well. If we think in very simple terms, we may regard the tightly constrained portion to demonstrate laminar ﬂow, with the introduction of turbulence causing the observed breakup of the plume. While that may be generally correct, things are, of course, not that simple. There is extensive literature on plumes as related to stack emissions and other environmental pollution issues in air, in water, and in groundwater. There are several government agencies [23–26] that maintain websites to direct inquirers to data, analyses, and models describing plume formation in all levels of detail. A web search [27] will quickly locate more references. Unfortunately, all these are at the wrong scale for the sensing of trace chemicals. They are focused on large-scale events where quantities are measured in grams to tonnes and distances in kilometers. In contrast, when seeking plumes of trace chemicals, the quantities are in nanograms or less and distances are in centimeters to meters. Fortunately, there is another body of literature, one that is focused on scales compatible with our interest. Researchers who are exploring the mysteries of behavior in a variety of animals have found that following odor plumes forms an essential part of the activity patterns of many, from crabs to moths to foxes to elephants. Odor plumes provide a path to food and to potential mates. They also provide warning of danger from predators. There is an extensive literature reporting this research. Much of it is speciﬁc to habits and behavior of the animal being studied, but much of it has direct application to locating chemical signatures. Attention is directed to only a few parts of this research that have immediate application to our subject. Perusal of the reference lists in the publications cited here will lead one into that larger volume of literature. 4.3.4.1 Plumes in Water Several ﬁne researchers have approached the study of underwater plumes with different objectives. While all this work is undoubtedly instructive, a series of articles [28–32] produced by Webster and Weissberg and their colleagues may apply most directly. They have examined the structure [28] of plumes in controlled experiments and produced photographs of dye plumes to study their development. They also took the point of view of a hungry crab [32]. In its attempt to ﬁnd the food source indicated by the plume, the crab manipulates its sensors within the plume. The structure of the plume makes it necessary. Chapter 5 is devoted to a description of these plumes. 4.3.4.2 Plumes in Air It is clear that, if the supply of molecules is adequate, plumes can form in air as well. While there may not be enough concentration in the air above buried sources to form plumes, in the case of unburied explosives, such as in IEDs or UXO covered loosely with rubble, there may be exploitable plumes. When such plumes develop, there will often be some level of urgency associated with locating the source. Sometimes, when logs seem to be following

EF&T IMPLICATIONS FOR SEARCH AND SAMPLING STRATEGIES

99

a plume they may be sampling from residue on plants on surfaces that act as concentrators of molecules deposited by very weak plumes over an extended period.

4.4 EF&T IMPLICATIONS FOR SEARCH AND SAMPLING STRATEGIES 4.4.1

Sources Buried on Land

The primary purpose for this discussion of EF&T of the molecules is to provide the one who employs a trace chemical sensor with means to increase the probability of locating and identifying the source. A thorough understanding of the transport processes presents options to the system designer and the operator to make a system more successful. Since “there is no effective mechanism for transporting molecules from a buried source to the surface when the soil is very dry” (see Section 4.3.1.3) when faced with such conditions, the system operator may need to become innovative. In addition, unless there is some mechanism for horizontal transport of molecules in moist soil, the operator should expect the strongest surface concentration to appear directly above the buried source. Search strategies that exploit the natural tendency of the molecules to remain adsorbed on surface particles and to become trapped in the boundary layers near the surface are likely to be most productive. If odor plumes from buried sources do develop, they are likely to remain close to the surface until they dissipate from turbulence. When seeking buried sources, the material presented here should make it clear that some times and some weather conditions are more favorable than others. Alternately, when circumstances do not permit waiting for a more propitious time or weather condition, it may be that artiﬁcially increasing either moisture or temperature will prove useful. 4.4.2

Sources Producing Plumes

4.4.2.1 Weather Effects When following plumes to a source, weather can be a major factor. Obviously, this applies in water as well as in air. When turbulence is heavy, the plume will become more dispersed. This turbulence can result from a temporary, weather-induced condition, but may force delay in plume tracing. Judging the degree of turbulence in water is somewhat easier than in air, simply because we can see the water. Substantial research on environmental pollution plume persistence and dispersion in air has produced a classiﬁcation system called Pasquill’s stability classes for plume stability in air. Unfortunately, the system does not apply as directly as we might wish since it applies at the much larger scale of stack exhausts and the like. However, some insight from the system is available.

100

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

TABLE 4.6 Stability of Air in Various Meteorological Conditions1 Daytime Solar Heating

Nighttime Cloud Cover

Wind

Much

Some

Slight

Sparse

Heavy

Calm

Most Unstable Very Unstable Somewhat Unstable

Very Unstable Unstable

Unstable

Moderately Stable Somewhat Unstable Somewhat Unstable

Most Stable Moderately Stable Somewhat Unstable

Gentle Breezy to Strong 1

Somewhat Unstable

Somewhat Unstable Somewhat Unstable

Lighter shading denotes more stable air; heavier shading, more unstable.

This system classiﬁes the local turbulence conditions into classes A through F, where class A denotes the most anticipated turbulence, or the most unstable conditions, and class F denotes the most stable conditions. The class is determined for a particular location at a particular time, based on wind speed, cloud cover, and sun angle. At least two websites currently present tables for determining the local stability class. A government site14 presents a step-by-step procedure to precisely determine the class. A commercial site15 presents a more generalized table. A much-simpliﬁed classiﬁcation, based on these tables, is shown in Table 4.6. The scale of interest is very different for trace chemical or odor plumes from that for which the original classiﬁcation system was developed. Therefore, it is proposed that the simpliﬁed classiﬁcation scheme of Table 4.6 will be sufﬁciently applicable to provide an indication of the more and the less favorable conditions for following plumes. In this presentation the lighter shading denotes those conditions that are generally more favorable; darker shading indicates more instability and hence less well deﬁned plumes. There may be some surprises when considering those conditions that are the most favorable for stable odor plumes. Best conditions occur on cloudless nights with the lightest breezes. Worst conditions for plume stability are on bright, sunny days with the lightest breezes. It appears that surface heating is the source of the daytime turbulence. Conversely, clear nights, with radiant cooling to space produce the least turbulence from the surface. Whether the plume is being sought in air or water, weather conditions will definitely affect the formation of the plumes. Appropriate scheduling holds promise for more success. 4.4.2.2 Effects of Plume Characteristics Chapter 5 describes the behavior of plumes in more detail, but a brief discussion of their effect on search is appropriate here. Two characteristics of plumes appear signiﬁcantly important 14 http://www.eglin.af.mil/weather/Pasquills.html. 15 http://www.air-dispersion.com/formulas.html#stability

heavier shading, more unstable.

Lighter shading denotes more stable air;

OPEN QUESTIONS AND FRUITFUL AREAS FOR FUTURE RESEARCH

101

in the search for chemical signatures. Since, in experiments, a dye plume often seems to consist of long, twisting ﬁlaments of color moving through clear water, we might call the ﬁrst “ﬁlamentation.” We use this term to qualitatively describe the degree to which the plume resembles a ribbon suspended in the water. This ﬁlamentation often dominates the plume’s appearance near its source, but as the plume encounters turbulence, it becomes less like a ribbon and more like a cloud. Since the plume appears to be rather uniform within its bounds, we might call the other characteristic homogeneity. Turbulence also effects plume homogeneity. Turbulence interacts with the plume on different scales. It may be induced by the plume’s ﬂowing past a surface, by thermal convection cells, or by cross currents in the water. When the scale is large relative to the diameter of the plume, the effect is somewhat like twisting or rolling a ribbon. When the scale is comparable to the diameter of the plume, it directly affects the cross-sectional homogeneity of the plume. If we were able to place a sufﬁciently capable, sufﬁciently small sensor at a spatially ﬁxed point within a plume and measure the concentration as the plume moves past us, we would observe a variation with time. If the degree of ﬁlamentation were high enough we would see the signal vary abruptly from strong to essentially zero, then back again to strong, continuing in a random manner. On the other hand, if there were little ﬁlamentation, in a more homogeneous plume, the signal would vary more gradually, but still might disappear completely at times. Therefore, any ﬁnite sampling time produces a time average of the concentration. The longer sampling times needed to compensate for lower sensor sensitivity may actually be collecting an integrated average of instantaneous concentrations that vary widely. For a broader look at the plume, suppose we form an array of those perfect sensors in the preceding paragraph, and orient the array in a plane normal to the axis of the plume. This is the approach adopted by Ishida et al. [33] In a plume with pronounced ﬁlamentation they observed the concentration to vary randomly among the quadrants of an eight-element array. They also photographed that plume. A single, stop-action photo vividly showed the ﬁlamentation, but a multiple exposure (6000 frames at 10 frames per second) from the same spot produced an image that seemed completely homogeneous. The crab [32] maneuvers its sensors in an apparent effort to capture the strongest signals. Time averaging does not seem to be a technique used by crabs, but it may be an effective strategy for artiﬁcial sensing systems

4.5 OPEN QUESTIONS AND FRUITFUL AREAS FOR FUTURE RESEARCH

As the technology for sensing smaller concentrations of explosive molecules develops further, and new systems are ﬁelded and used, the need for better understanding of where to ﬁnd these elusive molecules will become ever more

102

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

evident. This will be particularly evident when intelligent machines form part of the system. These machines must be programmed or instructed to follow certain rules and procedures. Better understanding of the EF&T of the quarry molecules enables more effective and efﬁcient programming. Some obvious open questions must certainly be addressed. 4.5.1

Objects Buried in the Sea Bottom

As mentioned in Section 4.1.1, the extensive work of Phelan and Webb has no parallel for objects buried in the seaﬂoor.16 Whether searching rice paddies or oceans, if a munition has been dropped from the surface, or lain in the water for very long, it is likely to have become at least partially buried in the seaﬂoor. The understanding of the processes at work in these conditions is rudimentary at best. Many sensors commonly used to ﬁnd objects on the seaﬂoor prove ineffective when the objects become buried. With a more thorough understanding of the EF&T of the explosive molecules in these situations, trace chemical sensors may be able to provide more success in locating these objects. 4.5.2

Sampling Plant Material

Except for extreme desert areas, most locations where munitions are buried will have plants. When the plants are growing directly over the munition, some of the explosive molecules released will certainly be drawn into the plant with its water intake. Whether these molecules are metabolized or concentrated has been studied for a few plants [17]. In either case, there may be opportunities for exploitation. If the explosive molecules are metabolized, the plant may produce some characteristic chemical signature that could aid in locating the source. If the molecules are concentrated, then it may be plausible to use the plant as a preconcentrator and sample the plant tissue to search for explosive compounds.

4.6

ROLE OF COMPUTER MODELING

Several types of computer models have been developed for estimating the expected concentrations of the chemicals of interest as they move away from the source. Soil transport models attempt to estimate the expected concentration at the surface above buried sources. Plume transport models attempt to estimate the concentrations within a plume, along with its shape and position. A different form of model is designed to guide a search pattern for employing a sensing system to trace a plume. 16 Recall

that the term seaﬂoor in this work designates the soil of the bottom of any body of water, whether sea, lake, pond, river, stream, or irrigated ﬁeld.

ROLE OF COMPUTER MODELING

4.6.1

103

Soil Transport Models

Webb and his colleagues [14] have had considerable success in modeling the relative concentrations of molecules in their migration through the soil. They developed a code called T2TNT to include the effects of weather. This type of analysis can provide ﬁeld workers searching for signatures with indications of expected success as moisture changes in a particular area, or as the search moves to a different type of soil. For example, they analyzed the conditions illustrated in Figure 4.11. Figure 4.14 shows the degree of success of the predictions. Therefore, we may infer that, if the soil and moisture conditions are well known, the surface ﬂux can be estimated with considerable accuracy. Phelan and Webb [1, pp. 65, 66] present results of diurnal concentration predictions for periods up to a year. The results indicate that diurnal liquid phase concentrations tend to oscillate by about 8 orders of magnitude; gas-phase levels, by about 6. This oscillation is superimposed on a weather cycle and seasonal variation. These longer cycle effects move the diurnal concentrations up or down by about 2 orders of magnitude. They often reduce the diurnal variations to only 1 or 2 orders of magnitude, sometimes on the high side, sometimes on the low. Hence, it would appear that ﬁeld workers would ﬁnd it to their advantage to employ these modeling techniques, realizing that they must input good data to expect good results. In any given ﬁeld of buried munitions there will be times that produce much higher concentrations than at other times. Searchers may be able to increase efﬁciency, and perhaps safety, by using these predictive tools.

1000000

Surface Flux (pg /min)

100000 10000 1000 100 10 1

0

10

Figure 4.14

20

30

40 50 Time (days)

60

70

T2TNT predictions compared to data.

80

90

104

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

4.6.2

Plume Transport Models

Much research has been applied toward understanding the ways that various animals use odor plumes. It is not within the scope of this book to exhaustively survey the resulting literature. One aspect of this research has direct bearing on the use of plumes to locate sources of chemical molecules such as explosives. As they are studying their animal subjects, these researchers construct two kinds of mathematical models to assist them. One of these models the plume itself, attempting to provide a predictive capability for describing the characteristics of the plume in a measured environment. Knowing the motion characteristics of the medium—air or water—the models seek to predict the concentration and dimensions of the plume at distances downstream. Of course, the nature of the problem quickly leads to use of statistical descriptions. This provides a model quite adequate for constructing tracking algorithms. An example of this work is that of Farrell and co-workers [34]. They present a rather complex model to attempt to account for the effects of ﬂuid motion and turbulence in three different levels of scale, relative to the plume. They begin with classical equations of motion, but by breaking their particle velocity vector into three components related to the three scales of interest, they are able to introduce appropriate statistical descriptions for the components. The result is a model that retains both the diffusive and the ﬁlamentary nature of the plume. 4.6.3

Plume Search Models

The other type model attempts to track the plume using techniques inspired by observation of animal subjects. Some of the models actually emulate the search behaviors observed. A variation on this type model, perhaps more appropriately called an adaptation, is that of Ishida and co-workers [33]. In follow-on papers to [34], Farrell and co-workers [35, 36] ﬁrst use hidden Markov methods (HMM) to develop a model for mapping a plume, presupposing a searcher moving into the plume’s vicinity in search of it. Their model guides the searcher from ﬁrst discovery on a path that enhances the searcher’s probability of maintaining continuous contact with the plume, while directing the search toward its source. They compare this mapping with the plume described in the model of [35] with satisfactory results. In the third paper [36] they present algorithms for use in plume tracing by an autonomous vehicle. They show their level of success by presenting experimental results. Their procedure differs markedly from those of Chapter 6 and of Ishida and co-workers [33].

4.7

CONCLUSIONS

When the value a sensor system is based on its ability to ﬁnd very sparse concentrations of a few very speciﬁc molecules, every available advantage needs to be taken to ensure even modest success. We have all heard the story of the drunk

REFERENCES

105

searching for his keys under the street light, just because the light was better, even though he knew he lost them down the block. Without proper application of the principles discussed in this chapter, someone employing a chemical sensor could be following the drunk’s procedure. It is important to search where the molecules are most likely to be found. It may be helpful to schedule the search to take maximum advantage of favorable conditions, but this also can be guided by knowledge.

REFERENCES 1. Phelan, J. M. and S. W. Webb. Chemical Sensing for Buried Landmines—Fundamental Processes Inﬂuencing Trace Chemical Detection, SAND2002-0909, Sandia National Laboratories, Albuquerque, NM, 2002 1a. Phelza, J. M. and S. W. Webb. Chemical Sensing for Buried Landmines: Fundamental Processes Inﬂuencing Trace Chemical Detection, in Mine Detection Dogs, Training, Operations and Odour Detection, Geneva International Centre for Humanitarian Demining, Geneva, 2003, p. 209–285 2. Phelan, J. M, P. J. Rodacy, and J. L. Barnett. Explosive Chemical Signatures from Military Ordnance. Sandia National Laboratories Report SAND2001-0755, Albuquerque, NM, April 2001. 3. Chambers, W. B., P. J. Rodacy, E. E. Jones, B. J. Gomez, and R. L. Woodﬁn. Chemical sensing system for classiﬁcation of minelike objects by explosive detection, in A. C. Dubey, J. F. Harvey, and J. T. Broach, Eds. Proceedings of the SPIE 12th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, Detection and Remediation Technologies for Mines and Minelike Targets III, April 13–17, 1998. 4. Leggett, D. C., T. F. Jenkins, A. Hogan, T. A. Ranney, and P. H. Miyares, External Contamination on Landmines by Organic Nitro-Compounds. U.S. Army Engineer Research and Development Center—Cold Regions Research and Engineering Laboratory, ERDC-CRREL Technical Report TR-00-2, Hanover, NH, March 2000. 5. Mark, H. F. and Kroschwitz J. Encyclopedia of Polymer Science and Technology. Wiley, New York, 1985. 6. Rodacy, P. J., P. K. Walker, S. D. Reber, J. Phelan, and J. V Andre. Explosive Detection in the Marine Environment and on Land Using Ion Mobility Spectrometry, A Summary of Field Tests, SAND2000-0921, Sandia National Laboratories Report, Albuquerque, NM, 2000 7. Phelan, J., P. Rodacy, and J. Barnett. Explosive Chemical Signatures from Military Ordnance, SAND2001-0755. Sandia National Laboratories, Albuquerque, NM, 2001. 8. Jenkins, T. F., M. E. Walsh, P. H. Miyares, J. Kopzynski, T. Ranney, V. George, J. Pennington, and T. Berry. Analysis of Explosive-Related Chemical Signatures in Soil Samples Collected Near Buried Landmines. U.S. Army Engineer Research and Development Center—Cold Regions Research and Engineering Laboratory, ERDCCRREL, Report ERDC TR-00-5, Hanover, NM, March 2000. 9. Leggett, D. C., J. H. Cragin, T. F. Jenkins, and T. A. Ranney. Release of ExplosiveRelated Vapors from Landmines. U.S. Army Engineer Research and Development

106

10. 11. 12. 13.

14.

15.

16.

17. 18.

19.

20.

21.

WHERE SHOULD WE LOOK FOR EXPLOSIVE MOLECULES?

Center—Cold Regions Research and Engineering Laboratory, ERDC-CRREL Technical Report TR-00-2, Hanover, NH, February 2001. Pella, P. A. Measurement of the vapor pressures of pressures of TNT, 2,4-DNT, 2,6-DNT and EGDN. J. Chem. Thermodyna. 9, 301–305 (1977). Phelan, J. M. and J. L. Barnett. Solubility of 2,4-dinitrotoluene and 2,4,6trinitrotoluene in water. J. Chem Eng. Data Mar/Apr (2001). Phelan, J. M. and J. L. Barnett. Phase Partitioning of TNT and DNT in Soils. Sandia National Laboratories Report SAND2001-0310, Albuquerque, NM, February 2001 Phelan, J. M. and S. W. Webb. Environmental Fate and Transport of Chemical Signatures from Buried Landmines—Screening Model Formulation and Initial Simulations. Sandia National Laboratories Report, SAND97-1426, Albuquerque, NM, June 1997. Webb, S. W., K. Preuss, J. M. Phelan, and S. Finsterle, Development of a mechanistic model for the movement of chemical signatures from buried landmines/UXO, in A. C. Dubey, J. F. Harvey, J. T. Broach, and R. E. Dugan, Eds. Proceedings of the SPIE 14th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, Detection and Remediation Technologies for Mines and Minelike Targets IV, April 5–9, 1999. Phelan, J. M., M. Gozdor, S. W. Webb, and M. Cal. Laboratory data and model comparisons of the transport of chemical signatures from buried landmines/UX, in A. C. Dubey, J. F. Harvey, J. T. Broach, and R. E. Dugan, Eds. Proceedings of the SPIE 14th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, Detection and Remediation Technologies for Mines and Minelike Targets IV, April 5–9, 1999. Hewitt, A. D., T. F. Jenkins, and T. A. Raney. Field Gas Chromatography/Thermionic Detector System for On-site Determination of Explosives in Soils. U.S Army Corps of Engineers, Engineer Research and Development Center, Cold Regions Research and Engineering Laboratory Technical Report ERDC/CRREL TR01-9, Hanover, NH, 2001. Thompson, P. L., L. A. Raner, and J. L. Schnoor. Uptake and transformation of TNT by hybrid poplar Trees. Environ. Sci. Tech. 32 (7), 975–980 (1998). Hundal, L. S., J. Singh, E. L. Bier, P. J. Shea, S. D. Comfort, and W. L. Powers. Removal of TNT and RDX from water and soil using iron metal. Environ. Pollution 97 (1–2), 55–64 (1997). Verhagen, R., C. Cox, R. Machangu, B. Weetjens, and M. Billet. Preliminary results on the use of Cricetomys rats as indicators of buried explosives in ﬁeld conditions, I. G. McLean, Ed. in Mine Detection Dogs: Training, Operations and Odour Detection. Geneva International Centre for Humanitarian Demining (GICHD), Geneva, 2003, pp. 175–193. Joynt, V. The Mechem explosive and drug detection system (MEDDS), in I. G. McLean, Ed. Mine Detection Dogs: Training, Operations and Odour Detection. Geneva International Centre for Humanitarian Demining (GICHD), Geneva, 2003, pp. 165–174. Fjellaner, R. The REST Concept, in I. G. McLean, Ed. Mine Detection Dogs: Training, Operations and Odour Detection. Geneva International Centre for Humanitarian Demining (GICHD), Geneva, 2003, pp. 53–107.

REFERENCES

107

˚ 22. McLean, I. G., H. Bach, R. Fjellanger, and C. Akerblom. Bringing the mineﬁeld to the detector: Updating the REST concept. Proceedings of EUDEM2-SCOT—2003, vol. 1, 2003, pp. 156–161. 23. National Atmospheric Release Advisory Center. http://www.nrac.llnl.gov. 24. Subcommittee on Consequence Assessment and Protective Action (SCAPA) of the Emergency Management Advisory Committee of the U.S. Department of Energy. http://www.orau.gov/emi/scapa. 25. Atmospheric Dispersion Modeling Liaison Committee (ADMLC). http://www.ad mlc.org.uk. 26. Federal Coordinator for Meteorological Services and Supporting Research, Directory of Atmospheric Transport and Diffusion Consequence Assessment Models, FCMI3-1999, Washington, DC, March 1999. http://www.ofcm.gov. 27. Dispersion Modeling Feature Articles. http://www.environmental-expert.com/articleindex.htp. 28. Webster, D. R. and M. J. Weissburg. Chemosensory guidance cues in a turbulent chemical odor plume. Limnol. Oceanogr. 46(5), 1034–1047 (2001). 29. Zimmer-Faust, R. K., C. M. Finelli, N. D. Pentcheff, and D. S. Wethey. Odor plumes and animal navigation in turbulent water ﬂow: A ﬁeld study. Biol. Bull. 188, 111–116 (1995). 30. Horner, A. J., M. J. Weissburg, and C. D. Derby. Dual antennular chemosensory pathways can mediate orientation by Caribbean spiny lobsters in naturalistic ﬂow conditions. J. Exp. Biol. 207(21), 3785–3796 (2004). 31. Dasi, L. P. Statistical Characteristics of Turbulent Chemical Plumes. Master’s Thesis, Georgia Institute of Technology, Atlanta, Georgia, July 2000. 32. Weissberg, M. J., C. P. James, D. L. Smee, and D. R. Webster. Fluid mechanics produces conﬂicting constraints during olfactory navigation of blue crabs, Callinectes sapidus. J. Exp. Biol. 206, 171–180 (2003). 33. Ishida, H., T. Nakamoto, T. Moriizumi, T. Kikas, and J. Janata, Plume-tracking robots: A new application of chemical sensors. Biol. Bull. 222–226 (April 2001). 34. Farrell, J. A., J. Murlis, X. Long, W. Li, and R. T. Card´e. Filament-based atmospheric dispersion model to achieve short time scale structure of odor plumes. Environ. Fluid Mech. 2, 143–169 (2002). 35. Farrell, J. A., S. Pang, and W. Li. Plume mapping via hidden Markov methods. Ieee Trans. Syst., Man, and Cybernetics—Part B: Cybernetics 33(6), 850–863 (2003). 36. Farrell, J. A., S. Pang, and W. Li. Chemical lume tracing via an autonomous underwater vehicle. IEEE J. Oceanic Eng. 30(1) 1–15 (2005).

CHAPTER 5

STRUCTURE OF TURBULENT CHEMICAL PLUMES DONALD R. WEBSTER School of Civil & Environmental Engineering, Georgia Institute of Technology

Sensing systems that track chemical plumes have great practical importance because they, for instance, facilitate locating sources of explosive compounds or improve detection of other hazards, such as pollutants leaking into the environment. An important aspect of understanding the process of sensing and tracking chemical plumes is characterizing the environment into which the chemical compounds are released. For instance, chemicals leaking from a point source are transported by the local ﬂuid ﬂow. Unless the velocity is extremely slow, the ﬂow is likely to be turbulent. Thus, the transport of chemicals is a combination of advection by the bulk ﬂow and turbulent mixing. The primary objective of this chapter is to introduce the basic characteristics and structure of turbulent chemical plumes. The second objective is to discuss the sensory signals that provide information for locating the source of the plume. 5.1

TURBULENT MIXING

Turbulent mixing is the process by which a ﬂuctuating turbulent ﬂow dilutes and homogenizes quantities such as chemical concentration, heat, and momentum. This chapter focuses on the turbulent mixing of chemical concentration, but there are strong analogies with other scalar quantities. Turbulence is difﬁcult to deﬁne precisely, but turbulent ﬂows have several common features. For instance, all turbulent ﬂows are characterized by random, unpredictable ﬂuctuations of velocity that are spatially organized into swirling packets called eddies. These velocity ﬂuctuations advect chemicals in a spatially and temporally random fashion. Note that advection means transport by ﬂuid movement (i.e., the ﬂuid motion carries the quantity) and is often used synonymously with the term convection for transport of heat. Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

109

110

STRUCTURE OF TURBULENT CHEMICAL PLUMES

Turbulent mixing can be thought of as a two-step process, but in reality both steps are occurring simultaneously and continuously. The ﬁrst step consists of advection of a ﬁlament (or patch) of concentration. Because the ﬂow ﬁeld is spatially varying, neighboring regions of the ﬁlament are advected away from each other. Thus, the ﬁlament undergoes a process called turbulent stirring, which consists of stretching and folding by the random velocity ﬂuctuations. Turbulent stirring continuously creates very steep gradients of concentration as the ﬁlament shape becomes more irregular. The second step of turbulent mixing is molecular diffusion. Fick’s law for molecular diffusion states that the ﬂux of concentration is proportional to the strength of the concentration gradient. Diffusion acts across the large concentration gradients created by turbulent stirring to smooth the gradients, which effectively dilutes the ﬁlament and expands its volume. Thus, turbulent mixing can be conceptualized as continual random stirring that creates sharp concentration gradients, which rapidly diffuse. Turbulent mixing is signiﬁcantly more effective than molecular diffusion alone. Consider the mixing process in a coffee cup. The time for molecular diffusion to mix a small patch of milk throughout the cup is several days, at which time the coffee would be also cold and unpleasant. However, by stirring with a spoon, and hence creating a turbulent-like ﬂow, the mixing process is complete within seconds. Turbulent ﬁelds are often described in terms of spatial length scales that relate to the size of the eddies and ﬁlaments in the ﬂow. The range of scales extends from the largest ﬂow structures to the scales at which molecular diffusion acts to smooth gradients of momentum or chemical concentration. The integral scale of the velocity ﬁeld describes the largest turbulent eddies in the ﬂow. In an open-channel ﬂow a good estimate of the integral length scale is half of the water depth, simply because the presence of larger eddies is constrained by the physical boundaries. In an atmospheric boundary layer, the layer thickness similarly constrains the largest eddy size. Diffusion of momentum of the velocity ﬂuctuations (or dissipation of turbulent kinetic energy) occurs at the Kolmogorov scale, which is estimated as 3 1/4 ν η= (5.1) ε where ν is the ﬂuid kinematic viscosity and ε is the dissipation rate of turbulent kinetic energy [1]. It is important to note that this is an order of magnitude estimate rather than equality. Thus, the Kolmogorov scale provides a rough estimate of the size of the smallest eddies present in the turbulent ﬂow. The length scales for the turbulent concentration ﬁeld range from the plume width to the scale at which molecular diffusion acts to homogenize the distribution (or dissipate the variance of the scalar ﬂuctuations). The smallest length scale is referred to as the Batchelor scale and is estimated as 2 1/4 νD LB ∼ = η(Sc)−1/2 (5.2) ε

INSTANTANEOUS STRUCTURE

111

where D is the molecular diffusivity of the chemical in the ﬂuid [2]. Equation (5.2) also shows that the Batchelor and Kolmogorov scales are related by the square root of the Schmidt number (Sc = ν/D). The majority of laboratory research on turbulent chemical plumes has occurred in water ﬂows, while ﬁeld research has occurred in both air and water environments. Plumes in these ﬂuid environments are similar in many regards, but they differ in some important ways due to differences in the ﬂuid properties. For instance, the kinematic viscosity of water is ν = 1 × 10−6 m2 /s at 20◦ C, whereas the kinematic viscosity of air is ν = 1.5 × 10−5 m2 /s at 20◦ C. Thus, even in the scenario of identical dissipation rates of turbulent kinetic energy, the Kolmogorov scale would differ between air and water ﬂows. Similarly, the diffusivity of many chemical compounds in water is around 1 × 10−9 m2 /s, whereas the diffusivity of compounds in air is often around 1 × 10−5 m2 /s. This difference has signiﬁcant effects on the turbulent mixing process because of differences in the Batchelor length scale. In water ﬂows, the size of the ﬁne-scale chemical concentration structure is roughly 35 times smaller than the Kolmogorov scale, whereas in air the ﬁne-scale structure of the velocity and concentration ﬁelds is roughly the same size. 5.2

INSTANTANEOUS STRUCTURE

Figure 5.1 shows a photograph of a chemical plume in a turbulent water ﬂow. The dye in this example has been released from a small oriﬁce upstream of

Figure 5.1 Closeup, overhead view of a plume released isokinetically into a turbulent boundary layer in an open-channel ﬂow. Flow direction is from left to right.

112

STRUCTURE OF TURBULENT CHEMICAL PLUMES

the photograph location. The photograph reveals the basic processes described in the previous section. The ﬁlament forms an irregular pattern that wanders randomly across the image, which is evidence of the turbulent stirring process. The concentration within the ﬁlament is clearly more dilute in some regions compared to others, which is consistent with molecular diffusion occurring at the boundary of the ﬁlament. The distribution of concentration consists of complex shapes that are continually evolving. A moment after this image was collected, the plume had a completely different appearance. The ﬂow visualization image (Fig. 5.1) is an integrated view of the threedimensional plume structure projected onto the image plane. The image provides a good qualitative sense of the plume structure but lacks quantitative information. Webster et al. [3] employed the laser-induced ﬂuorescence (LIF) technique to quantify the instantaneous concentration ﬁeld. For the demonstration data shown herein, the ﬂow was water in an open channel with a fully developed turbulent boundary layer on the bed. Fluorescent dye was released from a point source elevated 25 mm above the bed in the inertial region of the turbulent boundary layer. The velocity of the ﬂuid released from the source nozzle matched the ﬂow in the channel and is called an isokinetic release. The test section was illuminated by a laser sheet of monochromatic light with a wavelength in the absorption range of the dye. The dye was excited by the laser light and emitted light at a longer wavelength. A digital camera collected the emitted light, the intensity of which was proportional to the dye concentration and laser light intensity. Thus, after a calibration process, the intensity of the light recorded by each pixel yielded a concentration measurement in the ﬁeld. For the data shown in this chapter, a onemegapixel digital camera effectively collected a million samples of concentration in each image. Figure 5.2 shows an example concentration ﬁeld for the plane illuminated by the laser sheet. In this case, the horizontal plane is aligned with the source nozzle

Figure 5.2 See color plates. Sample instantaneous concentration ﬁeld near the plume source. H is the channel depth and equals 20 cm. The contour values are normalized by the source concentration. (Figure adapted from data in Webster et al. [3].)

INSTANTANEOUS STRUCTURE

113

height. It is important to remember that the data correspond to a planar slice through the three-dimensional plume structure. The coordinate system employed deﬁnes x for the streamwise direction and y for the transverse direction with the source nozzle located at the origin. Note that the contour levels have exponential spacing to fully reveal the broad range of concentration levels present in the ﬁeld. The concentration values are normalized by the source concentration and vary between zero and one. Near the source, the plume width was narrow, like the ﬂow visualization image in Figure 5.1. The peak concentration values in this sample ﬁeld are in the range of 30 to 40% of the source concentration. However, the highest contour shown is 10% of the source concentration because the regions of higher concentration are extremely small. The ﬁeld appears to have randomly distributed islands of relatively high concentration. In reality the three-dimensional plume ﬁlament is continuously connected outside of the illuminated plane, and the measurement technique reveals only the intersections of the structure with the measurement plane. Keep in mind that this ﬁeld is a snapshot at an arbitrary instant and that the concentration distribution changes continually. Also, note that the peak concentration values decrease signiﬁcantly over the length of the ﬁeld due to the turbulent mixing process. Figure 5.3 shows a sample concentration ﬁeld at a location farther downstream (the x/H = 10 location corresponds to 2 m downstream of the source nozzle). The upper end of the concentration range decreased by more than an order of magnitude compared to the ﬁeld in Figure 5.2 (note that the contour levels have changed). Also, the plume structure covers a much wider transverse distance. Thus, over a fairly short distance from the source, the concentration decreases and the plume size increases signiﬁcantly. These attributes are consistent with the turbulent mixing process described in the previous section. A ﬂow visualization image of this region of the plume appears to be a homogeneous cloud of dye [4]. However, the concentration measurements reveal that the

Figure 5.3 See color plates. Sample instantaneous concentration ﬁeld. H is the channel depth and equals 20 cm. The contour values are normalized by the source concentration.

114

STRUCTURE OF TURBULENT CHEMICAL PLUMES

Figure 5.4 See color plates. Fine resolution perspective of a sample instantaneous concentration ﬁeld. H is the channel depth and equals 20 cm. The contour values are normalized by the source concentration.

instantaneous ﬁeld continues to have patchy structure with signiﬁcant gradients of concentration. Molecular diffusion acts across strong gradients in the ﬁeld to dilute the concentration of a ﬁlament. Figure 5.4 shows a high-resolution view of the plume to demonstrate the ﬁne-scale concentration structure. The resolution of this measurement is close to the Batchelor scale, and hence the ﬁeld resolves the ﬁnest structure in the concentration ﬁeld. Conversely, Figures 5.2 and 5.3 have resolutions slightly coarser than the Kolmorogov scale and, thus, only reveal the large-scale characteristics of the plume. In the high-resolution image, the ﬁeld appears as disconnected islands of high concentration. Again, the individual islands are continuously connected outside of the measurement plane. At the edge of these ﬁlaments, the concentration gradient is very steep. In this particular example, the concentration changes from a peak of roughly 10% of the source concentration to zero over a distance of half of a millimeter or less. Molecular diffusion transports very effectively across such steep gradients, which explains the rapid decrease in peak concentration shown in the large-scale ﬁelds.

TIME-AVERAGED CHARACTERISTICS

5.3

115

TIME-AVERAGED CHARACTERISTICS

While the concentration ﬁelds help to describe the spatial structure of the plume, the instantaneous snapshots do not reveal the temporal variation. Figure 5.5 shows the ﬁrst 50 s of a sample time record that was collected along the plume centerline. The time record was highly intermittent with low or zero concentration for long periods separated by extremely high spikes of concentration, which correspond to a concentration ﬁlament meandering over the sensor location. The time-averaged concentration for this record is much smaller than the concentration peaks due to the extended periods of zero concentration. Thus, the signal consists of a small time-averaged value with huge ﬂuctuations. Since negative concentration values are impossible, the statistical record of concentration is a highly skewed distribution [4]. The time-averaged concentration ﬁeld is calculated by ensemble averaging a long series of individual concentration ﬁelds, such as Figure 5.2. In this example, 6000 consecutive concentration ﬁelds were averaged to produce the time-averaged ﬁeld shown in Figure 5.6. As surmised from the discussion of the time record, the magnitude of the time-averaged concentration is much smaller than the instantaneous values (a factor of 10 or more between Fig. 5.2 and 5.6). While the instantaneous concentration ﬁelds consist of random islands, the time-averaged concentration is a much smoother distribution. The distribution appears nearly symmetric about the plume centerline (y = 0). Because the plume was unbounded in the transverse direction, the transverse proﬁles possess this symmetry and, in fact, follow a Gaussian proﬁle shape. Figure 5.7 shows the transverse timeaveraged concentration proﬁles at four distances from the source. The transverse coordinate is normalized by the square root of the second central moment of the proﬁle (i.e., the standard deviation, σ ), which is a measure of the half-width of the time-averaged plume. The concentration is normalized by the value on the centerline for each proﬁle. With this normalization, the proﬁles are self-similar, which means that they are coincident and have the same shape. The proﬁles also agree very well with the Gaussian proﬁle shape shown with the solid line: y2 (5.3) c = ccenterline exp − 2 2σ where an overbar indicates a time-averaged quantity.

c/Co

0.2

0.1

0

0

10

20

30

40

50

Time (s)

Figure 5.5 Time record of concentration at a point along the plume centerline. (Adapted from data in Webster et al. [5].)

116

STRUCTURE OF TURBULENT CHEMICAL PLUMES

Figure 5.6 Time-averaged concentration ﬁeld. The contour values are normalized by the source concentration.

centerline

Gaussian profile

Figure 5.7 Proﬁles of the time-averaged concentration at four downstream locations. The proﬁles are self-similar and the solid line corresponds to a Gaussian proﬁle shape.

As described above, the plume becomes wider and more dilute as it evolves in the streamwise direction, thus ccenterline and σ are changing with x. The decrease of the time-averaged concentration along the centerline of the plume follows a x −1 proﬁle for x/H > 2 (Fig. 5.8). This power law decrease agrees well with the time-averaged concentration ﬁeld predicted by modeling efforts that assume

TIME-AVERAGED CHARACTERISTICS

117

Figure 5.8 Time-averaged concentration along the plume centerline. Also shown is a power law curve.

constant eddy diffusivity [6]. Closer to the source, the data do not follow the x −1 line, rather the proﬁle shows a greater rate of decrease. The change in behavior at x/H ≈ 2 resulted from a transition of the size of the plume compared to the size of the turbulent eddies in the ﬂow [3]. While the time-averaged concentration is decreasing rapidly in the x direction, the magnitude of the ﬂuctuations is decreasing even more rapidly. Figure 5.9 shows the standard deviation of the concentration ﬂuctuations along the plume centerline. The proﬁle shown only tells part of the story about the magnitude of the concentration ﬂuctuations. At the source, the concentration ﬂuctuations are zero due to the homogeneous nature of the plume at the release. As the ﬁlament begins to distort and mix, the concentration starts to ﬂuctuate and the standard deviation increases rapidly. This increase occurs upstream of the measurement region shown. As the mixing process continues, the plume becomes more homogeneous and the peak concentration in the ﬁlaments decreases. The combined effect decreases the standard deviation. As shown, the proﬁle follows a power law decrease (x −1.75 ). At x/H = 2 the standard deviation is roughly 2.5 times the value of the time-averaged concentration. Thus, at this location the time record consists of large ﬂuctuations above a small time-averaged value as discussed above. The standard deviation decreases at a faster rate than the time-averaged concentration as indicated by the exponent of the power law. Thus, the plume

118

STRUCTURE OF TURBULENT CHEMICAL PLUMES

Figure 5.9 Standard deviation of the concentration ﬂuctuations along the plume centerline. Also shown is a power law curve.

becomes homogeneous faster than the time-averaged concentration dilutes. The sample ﬁeld shown in Figure 5.3 provides support for this conclusion because the ﬁeld is more homogeneous compared to the ﬁeld closer to the source shown in Figure 5.2.

5.4

INFORMATION FOR TRACKING CHEMICAL ODOR PLUMES

The task of locating the point source of a turbulent chemical plume is a difﬁcult challenge. Even in the scenario that perfect sensors are available to detect the appropriate chemical compounds, using the acquired sensory information to rapidly locate the source with a high percentage of success is far from a trivial endeavor. Nevertheless, animals have developed an incredible ability to track chemical plumes to facilitate feeding and mating processes. In recent years, research has been performed to better understand the turbulent chemical plume tracking ability and strategies of moths [7], blue crabs [8, 9], lobsters [10, 11], and other animals (e.g. [12, 13]). The robust tracking ability of these organisms provides inspiration for engineers attempting to design systems to perform similar tasks [14, 15].

INFORMATION FOR TRACKING CHEMICAL ODOR PLUMES

119

The tracking problem can be distilled into three distinct tasks: (1) making contact with the chemical plume, (2) using information in the chemical plume structure to move toward the source, and (3) determining when the source has been reached. Making contact with the plume (1) is essentially a search for a randomly located goal, and the effectiveness of various strategies is discussed in numerous publications (e.g., [16]). Realizing that the source has been reached (3) initially seems like a simple task, but it is actually quite difﬁcult in practice and has received little research attention to date. The discussion in this section is focused on the second phase of the tracking problem: extracting information from the plume structure in order to move successfully toward the source. Animal performance indicates that they are effectively extracting useful information from the structure of chemical plumes. For instance, after contacting a turbulent plume similar to the example in the previous sections, blue crabs ﬁnd the source via chemosensory tracking with a success rate of 80 to 90% from a distance of 1.5 m within a period of 10 to 30 s [17]. Thus, the basic question is: What information in the intermittent and chaotic plume structure is providing cues to guide tracking? The smooth and predictable gradient in the time-averaged concentration ﬁeld seems useful for tracking at ﬁrst glance (Fig. 5.6). A tracker would simply move up the concentration gradient by sequentially comparing the response of a single sensor to eventually locate the position of the highest concentration, which is coincident with the source. This strategy is called chemical klinotaxis, and small organisms, such as bacteria and nematodes, are believed to employ this strategy in laminar ﬂow environments [16]. Unfortunately, this strategy is ineffective for rapid tracking of turbulent chemical plumes for the simple reason that the information shown in Figure 5.6 is unavailable. The time period necessary to determine the time-averaged concentration with sufﬁcient accuracy to perceive the gradient direction is many minutes [4]. To follow the gradient in Figure 5.6, a tracker would have to collect chemosensory information for several minutes, compare to the time-averaged concentration previously collected to assess the gradient direction, and then move and start the process again. Figure 5.10 demonstrates that the convergence of the time-averaged concentration calculation takes roughly 10 min for the example plume. To generate this ﬁgure, the total time record (example segment shown in Fig. 5.5) was divided into numerous shorter periods, and the time-averaged value was calculated for each of the subperiods. For sample periods of 4.7 s, for instance, the time-averaged values vary by a huge range, roughly 200% larger to 100% smaller than the converged value. Thus, to accurately assess the concentration value compared to the concentration at the previous location, a much longer sampling period is needed. The data scatter is not solely a result of the number of samples that are collected, as one might expect from statistical considerations. Rather, it results directly from the intermittent nature of the concentration ﬁeld. In other words, a sensor must sample over a time period sufﬁcient for numerous ﬁlaments to pass by in order to determine a converged time-averaged value. Thus, the intermittent nature of

120

STRUCTURE OF TURBULENT CHEMICAL PLUMES

Figure 5.10 Time-averaged concentration as a function of sampling period at an arbitrary location in the ﬂow. The total time record of 600 s was divided into numerous shorter periods to demonstrate the slow convergence of the time-averaged concentration calculation. (Adapted from data in Webster and Weissburg [4].)

the concentration ﬁeld necessitates very long sampling periods in order to evaluate any time-averaged quantity, including the time-averaged concentration shown here. Obviously, a strategy that employs the gradient in a time-averaged quantity is extremely tedious. It is clear that blue crabs and other organisms are employing strategies that are much more efﬁcient than waiting for time-averaged quantities to statistically converge. Thus, they appear to be acquiring information from the ﬂuctuating plume structure. Several hypotheses have been suggested about the nature of the sensory information available in turbulent chemical plumes. For instance, Moore and Atema [18] reported that the rise slope of a concentration burst (see time record in Fig. 5.5) varies with distance from the source and hence provides ranging information during tracking. Webster and Weissburg [4] conﬁrmed that the timeaveraged value of the rise slope changes systematically as a function of distance from the source. However, like the concentration value, the rise slope at any location varies as a random variable and a long period of sampling is required to achieve a converged time-averaged value. This is, in fact, the underlying problem with any cue that requires a time-averaged value: The convergence time in an intermittent plume is very long and the required monitoring prevents rapid tracking progress. The instantaneous concentration or rise slope at a particular location in the plume is any value within a broad range and hence does not provide predictable information. If sequential sampling seems incongruent with the performance of rapid trackers, then what other information is available? Several researchers have suggested

INFORMATION FOR TRACKING CHEMICAL ODOR PLUMES

121

that animals are employing a bilateral comparison to adjust their transverse position in the plume [19–21]. Blue crabs, for instance, have chemosensors located on their leg appendages. Thus, by comparing the signal received by the left and right legs, the location of the plume edge may be deduced. Weissburg [22] reﬁned this idea to include the fact that the sensor spacing (which he called the spatial integration factor, SIF) needs to be relatively large compared to the plume size. The plume structure data introduced in the previous sections provide the opportunity to assess whether multiple sensors that are spatially separated can rapidly acquire useful information [5]. The correlation function for the instantaneous concentration acquired by two sensors at locations p and q, separated in the transverse direction, is deﬁned as cp cq =

1 T

T

c(y = p, t) × c(y = q, t) dt

(5.4)

0

where c is the instantaneous concentration, and T is the total period of the time record. Figure 5.11 shows the correlation function for four distances downstream of the source for the case of one sensor ﬁxed to the plume centerline and the other transversely displaced. The correlation function is normalized by the value at y = 0 (i.e., when the sensors are coincident along the centerline), thus the maximum value of the normalized correlation function is one. With increased

Figure 5.11 Correlation function in the transverse direction at four downstream locations.

122

STRUCTURE OF TURBULENT CHEMICAL PLUMES

distance from the source, the proﬁle becomes wider and the value of the correlation function increases. This trend results from turbulent mixing increasing the plume width and creating a more homogeneous ﬁeld (see Figs 5.2 and 5.3). The integral length scale provides a simple measure of the plume width and homogeneity and is calculated based on the area under the correlation function proﬁles: ∞ c0 cy L= dy (5.5) c0 c0 0 Figure 5.12 shows the integral length scale as a function of distance from the source. As expected, the integral length scale increases in the downstream direction. Further, and more importantly, the integral length scale provides a measure of the plume width to which the chemical sensor separation can be compared. Figure 5.13 shows the correlation function for sensors that are separated by a ﬁxed distance equal to L. The inner sensor moves throughout the ﬁeld and its position is denoted by y, as shown in the sketch. The sensor location is normalized by the standard deviation of the time-averaged proﬁles (see Fig. 5.7). With this scaling, the proﬁles at the four downstream locations are coincident, which suggests that the integral length scale is the correct scaling length for the sensor separation and that the sensor location is properly scaled by the width of the time-averaged plume.

Figure 5.12

Integral length scale as a function of distance from the source.

INFORMATION FOR TRACKING CHEMICAL ODOR PLUMES

123

Figure 5.13 Correlation function with sensor spacing equal to L in the transverse direction at four locations.

For the process of identifying the edge of the plume during tracking, the ideal situation is to have high contrast between the sensors, which corresponds to a low correlation function value. The correlation function shown in Figure 5.13 for sensor spacing equal to L is fairly high, approaching 0.3 at the plume centerline. By increasing the sensor spacing to four times the integral length scale, the value of the correlation function is greatly reduced (Fig. 5.14). Again, the proﬁles for the four locations are coincident, which reinforces the validity of the scaling. Thus, high contrast is perceived by the sensors, if the sensors are separated by a “large enough” distance. The measure of “large enough” for the turbulent plume is provided by comparing to the integral length scale. The deﬁnition of the correlation function includes the same time-averaging procedure that requires long sampling periods and hence is not directly useful for rapid tracking. The correlation function is useful, however, to deﬁne the plume width and appreciate that large spacing leads to contrast between the instantaneous signal received by the sensors. Essentially, if the sensor spacing is large, as measured relative to the integral length scale, then the correlation function is low, and instantaneous comparisons of the sensor signal values provides insight to the location of the plume centerline. In support of this conclusion, Grasso et al. [14] reported that sensor spacing affected path tortuousity of their autonomous lobsterlike robot tracker. Note that the integral length scale varies with distance from

124

STRUCTURE OF TURBULENT CHEMICAL PLUMES

Figure 5.14 Correlation function with sensor spacing equal to 4L in the transverse direction at four locations.

the source (Fig. 5.12), thus the ability to adjust the sensor spacing or compare among multiple sensors in an array is potentially advantageous. Spatial comparison appears to be useful to guide a tracker to the centerline of the plume and the information appears to be quickly available provided the sensor spacing is large. But, bilateral comparison alone is not enough to perform a successful, rapid search of a turbulent plume source. For instance, it is believed that blue crabs employ an odor-gated rheotaxis with bilateral comparison strategy. Rheotaxis means that they are sensing and using the ﬂow direction to head upstream, and odor-gated means that they move upstream only when receiving chemosensory information of interest. Odor-gated rheotaxis alone does not explain the high performance of blue crabs because, without the ability to effectively correct their position relative to the plume centerline, the tracker frequently misses the source as it continues to move upstream. However, the added ability to steer and adjust their position relative to the plume centerline via a bilateral comparison improves tracking performance to be in accord with observed animal behavior [23]. Keller et al. [17] showed that upstream movement of blue crabs largely resulted from stimulation at the antennule chemosensors, while the leg chemosensors are largely responsible for the steering ability. Further, the simulations of Weissburg and Dusenbery [23] showed that tracking performance is critically affected by the balance between employing ﬂow and chemical signals, which conﬁrms the importance of both cues.

VARIATION OF THE PLUME STRUCTURE

125

The odor-gated rheotaxis with bilateral comparison strategy appears to be effective for rapid plume tracking in straight plumes and is congruent with blue crab performance. Of course, other strategies may be as effective, or perhaps more effective, for turbulent plume tracking. Indeed, other organisms employ alternate strategies due to constraints such as body size and sensor spacing. For example, moths cast in the cross-wind direction because they are much smaller than a typical plume width and hence cannot instantaneously compare sensory input across the plume [24]. In addition, the odor-gated rheotaxis with bilateral comparison strategy has not been tested as the ﬂow conditions become more complex. 5.5

VARIATION OF THE PLUME STRUCTURE

The sample data presented in this chapter were collected for fairly simple ﬂow conditions. The ﬂow was a unidirectional open-channel ﬂow without large-scale ﬂow meander, and the release condition was isokinetic in the direction of the bulk ﬂow. Thus, chemical ﬁlaments were advected by the bulk ﬂow in the streamwise direction, while turbulent mixing acted to expand the plume size and dilute the chemical concentration. Changes in the ﬂow and release conditions lead to signiﬁcant variation in the plume characteristics and structure. Several laboratory studies have contributed to our understanding of turbulent chemical plumes and the effects of various ﬂow conﬁgurations. Fackrell and Robins [25] released an isokinetic neutrally buoyant plume in a wind tunnel at elevated and bed-level locations. Bara et al. [26], Yee et al. [27], Crimaldi and Koseff [28], and Crimaldi et al. [29] studied plumes released in water channels from bed-level and elevated positions. Airborne plumes in atmospheric boundary layers also have been studied in the ﬁeld by Murlis and Jones [30], Jones [31], Murlis [32], Hanna and Insley [33], Mylne [34, 35], and Yee et al. [36, 37]. In addition, aqueous plumes in coastal environments have been studied by Stacey et al. [38] and Fong and Stacey [39]. The combined information of these and other studies reveals that the plume structure is inﬂuenced by several factors including the bulk velocity, ﬂuid environment, release conditions, bed conditions, ﬂow meander, and surface waves. The release location inﬂuences the vertical distribution of the time-averaged concentration and ﬂuctuations. For a bed-level release, vertical proﬁles of the time-averaged concentration are self-similar and agreed well with gradient diffusion theory [26]. In contrast, the vertical proﬁles for an elevated release have a peak value above the bed and are not self-similar because the distance from the source to the bed introduces a ﬁnite length scale [3, 25, 37]. Additionally, it is clear that the size and relative velocity of the chemical release affects both the mean and ﬂuctuating concentration [4]. The orientation of the release also appears to inﬂuence the plume structure. The shape of the proﬁles of the standard deviation of the concentration ﬂuctuations is different in the study of Crimaldi et al. [29] compared with those of Fackrell and Robins [25] and Bara et al. [26]. Crimaldi et al. [29] attributed the difference to the release orientation, which was vertically upward from a ﬂush-mounted oriﬁce at the bed in their study.

126

STRUCTURE OF TURBULENT CHEMICAL PLUMES

The Reynolds number is the ratio of inertial to viscous forces and depends on the ﬂuid properties, bulk velocity, and boundary layer thickness. Turbulence characteristics vary with Reynolds number in boundary layers [40]. Thus, variation in the contributing factors for the Reynolds number ultimately inﬂuences the turbulent mixing and plume structure. Further, the ﬂuid environment, air or water, affects both the Reynolds number and the molecular diffusivity of the chemical compounds. Bed conditions also inﬂuence the plume by altering the turbulence intensity and distribution in the boundary layer. Rahman and Webster [41] systematically varied the bed roughness to evaluate the effect of boundary layer characteristics on a plume from an elevated isokinetic chemical release. The average concentration was lower along the centerline for increasing bed roughness due to increased mixing and dilution. Similarly, the concentration ﬂuctuation intensity was smaller and decreased faster for the rougher beds. Shifts in the ﬂow direction lead to large-scale meander of the plume. The mixing and dispersion of chemicals in the plume are then dictated by meander of the plume centerline and mixing about the plume centerline [39]. Figure 5.15 shows two plumes in the wake of a circular obstacle and a plume for the same ﬂow

Figure 5.15 Overhead view of plumes released isokinetically into a turbulent boundary layer with and without (top image) an obstacle upstream. Meander in the middle and bottom images was created by the wake of a small and large circular obstacle, respectively. Flow direction is from left to right.

REFERENCES

127

conditions without an upstream obstacle. The addition of an obstacle induced a K´arm´an vortex street that periodically advected the plume in different directions and resulted in the meandering plume shown. The ﬂow pattern also modiﬁed the turbulence characteristics and hence affected the mixing process. The overall effect is a change in the shape of the centerline of the plume and a change in the mixing and dilution of the chemical ﬁeld. Similarly, the imposition of surface waves on an open-channel ﬂow, as expected in coastal ﬂows, affects both the bulk velocity and the turbulence characteristics [42]. The full effect of periodic meander and wave perturbations on the plume structure is not currently known. Based on the discussed example parameters that inﬂuence the turbulent mixing process, it is clear that plume structure can be described in general terms, but the speciﬁc characteristics are likely to be case dependent. Nevertheless, certain characteristics, such as those employed by the odor-gated rheotaxis with bilateral comparison strategy, may be similar enough to allow animals and engineered systems to track chemical plumes for a wide range of ﬂow conditions. ACKNOWLEDGMENTS

The research conducted in support of this chapter was ﬁnancially supported by ONR/DARPA (N00014-98-1-0776) and NSF (IBN-0321444). Special thanks to S. Rahman and L.P. Dasi for data collection and analysis and to M.J. Weissburg and C.B. Woodson for helpful discussion. REFERENCES 1. Kolmogorov, A. N. The local structure of turbulence in incompressible viscous ﬂuid for very large Reynolds numbers. Dolk. Akad. Nauk SSSR 30, 301 (1941). 2. Batchelor, G. K. Small-scale variation of convected quantities like temperature in turbulent ﬂuid. Part 1. General discussion and the case of small conductivity. J. Fluid Mech. 5, 113–133 (1959). 3. Webster, D. R., S. Rahman, and L. P. Dasi. Laser-induced ﬂuorescence measurements of a turbulent plume. J. Engng. Mech. 129, 1130–1137 (2003). 4. Webster, D. R. and M. J. Weissburg. Chemosensory guidance cues in a turbulent chemical odor plume. Limnol. Oceanogr. 46, 1034–1047 (2001). 5. Webster, D. R., S. Rahman, and L. P. Dasi. On the usefulness of bilateral comparison to tracking turbulent chemical odor plumes. Limnol. Oceanogr. 46, 1048–1053 (2001). 6. Rahman, S. Effect of Bed Roughness on Scalar Mixing in Turbulent Boundary Layers. Ph.D. Thesis, Georgia Institute of Technology, 2002. 7. Mafra-Neto, A. and R. T. Card´e. Fine-scale structure of pheromone plumes modulates upwind orientation of ﬂying moths. Nature 369, 142–144 (1994). 8. Weissburg, M. J. and R. K. Zimmer-Faust. Life and death in moving ﬂuids: Hydrodynamic effects on chemosensory-mediated predation. Ecology 74, 1428–1443 (1993). 9. Weissburg, M. J. and R. K. Zimmer-Faust. Odor plumes and how blue crabs use them in ﬁnding prey. J. Exp. Biol. 197, 349–375 (1994).

128

STRUCTURE OF TURBULENT CHEMICAL PLUMES

10. Moore, P. A., N. Scholz, and J. Atema. Chemical orientation of lobsters, Homarus americanus, in turbulent odor plumes. J. Chem. Ecol. 17, 1293–1307 (1991). 11. Koehl, M. A. R., J. R. Koseff, J. P. Crimaldi, M. G. McCay, T. Cooper, M. B. Wiley, and P. A. Moore, Lobster snifﬁng: Antennule design and hydrodynamic ﬁltering of information in an odor plume. Science 294, 1948–1951 (2001). 12. Moore, P. A. and D. M. E. Lepper. Role of chemical signals in the orientation behavior of the sea star Asterias forbesi. Biol. Bull. 192, 410–417 (1997). 13. Ferner, M. C. and M. J. Weissburg. Slow-moving predatory gastropods track prey odors in fast and turbulent ﬂow. J. Exp. Biol. 208, 809–819 (2005). 14. Grasso, F. W., T. R. Consi, D. C. Mountain, and J. Atema. Biomimetic robot lobster performs chemo-orientation in turbulence using a pair of spatially separated sensors: Progress and challenges. Robot. Auton. Syst. 30, 115–131 (2000). 15. Hayes, A. T., A. Martinoli, and R. M. Goodman. Swarm robotic odor localization: Off-line optimization and validation with real robots, Robotica 21, 427–441 (2003). 16. Dusenbery, D. B. Sensory Ecology, W.H. Freeman, New York, 1992. 17. Keller, T. A., I. Powell, and M. J. Weissburg. Role of olfactory appendages in chemically mediated orientation of blue crabs. Mar. Ecol. Prog. Ser. 261, 217–231 (2003). 18. Moore, P. A. and J. Atema. Spatial information contained in three-dimensional ﬁne structure of an aquatic odor plume. Biol. Bull. 181, 408–418 (1991). 19. Reeder, P. B. and B. W. Ache. Chemotaxis in the Florida spiny lobster, Panulirus argus. Anim. Behav. 28, 831–839 (1980). 20. Zimmer-Faust, R. K., C. M. Finelli, N. D. Pentcheff, and D. Wethey. Odor plumes and animal navigation in turbulent water ﬂows: A ﬁeld study. Biol. Bull. 188, 111–116 (1995). 21. Atema, J. Eddy chemotaxis and odor landscapes: Exploration of nature with animal sensors. Biol. Bull. 191, 129–138 (1996). 22. Weissburg, M. J. The ﬂuid dynamical context of chemosensory mediated behavior. Biol. Bull. 198, 188–202 (2000). 23. Weissburg, M. J. and D. B. Dusenbery. Behavioral observations and computer simulations of blue crab movement to a chemical source in a controlled turbulent ﬂow. J. Exp. Biol. 205, 3387–3398 (2002). 24. Vickers, N. J. and T. C. Baker. Reiterative responses to single strands of odor promote sustained upwind ﬂight and odor source location by moths. Proc. Natl. Acad. Sci. USA 91, 5756–5760 (1994). 25. Fackrell, J. E. and A. G. Robins. Concentration ﬂuctuations and ﬂuxes in plumes from point sources in a turbulent boundary layer. J. Fluid Mech. 117, 1–26 (1982). 26. Bara, B. M., D. J. Wilson, and B. W. Zelt. Concentration ﬂuctuation proﬁles from a water channel simulation of a ground level release. Atmos. Environ. 26A, 1053–1062 (1992). 27. Yee, E., D. J. Wilson, and B. W. Zelt. Probability-distributions of concentration ﬂuctuations of a weakly diffusive passive plume in turbulent boundary-layer. Bound.-Lay. Meteorol. 64, 321–354 (1993). 28. Crimaldi, J. P. and J. R. Koseff. High resolution measurements of the spatial and temporal scalar structure of a turbulent plume. Exp. Fluids 31, 90–102 (2001).

REFERENCES

129

29. Crimaldi, J. P., M. B. Wiley, and J. R. Koseff. The relationship between mean and instantaneous structure in turbulent passive scalar plumes. J. Turbul. 3(14), 1–24 (2002). 30. Murlis, J. and C. D. Jones. Fine-scale structure of odour plumes in relation to insect orientation to distant pheromone and other attractant sources. Physiol. Entomol. 6, 71–86 (1981). 31. Jones, C. D. On the structure of instantaneous plumes in the atmosphere. J. Hazard. Mater. 7, 87–112 (1983). 32. Murlis, J. The structure of odour plumes, in T. L. Payne, M. C. Birch, and C. E. J. Kennedy, Eds. Mechanisms in Insect Olfaction. Clarendon Press, Oxford, 1986, pp. 27–38. 33. Hanna, S. R. and E. M. Insley. Time series analysis of concentration and wind ﬂuctuations. Bound.-Lay. Meteorol. 47, 131–147 (1989). 34. Mylne, K. R. Concentration ﬂuctuation measurements in a plume dispersing in a stable surface layer. Bound.-Lay. Meteorol. 60, 15–48 (1992). 35. Mylne, K. R. The vertical proﬁle of concentration ﬂuctuations in near-surface plumes. Bound.-Lay. Meteorol. 65, 111–136 (1993). 36. Yee, E., R. Chan, P. R. Kosteniuk, G. M. Chandler, C. A. Biltoft, and J. F. Bowers. Experimental measurements of concentration ﬂuctuations and scales in a dispersing plume in the atmospheric surface layer obtained using a very fast response concentration detector. J. Appl. Meteorol. 33, 996–1016 (1994). 37. Yee, E., R. Chan, P. R. Kosteniuk, G. M. Chandler, C. A. Biltoft, and J. F. Bowers. The vertical structure of concentration ﬂuctuation statistics in plumes dispersion in the atmospheric surface layer. Bound.-Lay. Meteorol. 76, 41–67 (1995). 38. Stacey, M. T., E. A. Cowen, T. M. Powell, E. Dobbins, S. G. Monismith, and J. R. Koseff. Plume dispersion in a stratiﬁed, near-coastal ﬂow: Measurements and modeling. Cont. Shelf Res. 20, 637–663 (2000). 39. Fong, D. A. and M. T. Stacey. Horizontal dispersion of a near-bed coastal plume. J. Fluid Mech. 489, 239–267 (2003). 40. DeGraaff, D. B. and J. K. Eaton. Reynolds-number scaling of the ﬂat-plate turbulent boundary layer. J. Fluid Mech. 422, 319–346 (2000). 41. Rahman, S. and D. R. Webster. The effect of bed roughness on scalar ﬂuctuations in turbulent boundary layers. Exp. Fluids 38, 372–384 (2005). 42. Mead, K. S., M. B. Wiley, M. A. R. Koehl, and J. R. Koseff. Fine-scale patterns of odor encounter by the antennules of mantis shrimp tracking turbulent plumes in wave-affected and unidirectional ﬂow. J. Exp. Biol. 206, 181–193 (2003).

PART II

FIELD EXPERIENCE

CHAPTER 6

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT MARK FISHER AND MATTHEW DOCK Nomadics, Inc., an ICx Technologies Company

6.1

INTRODUCTION

Underwater explosive devices typically possess a combination of detectable expressions, including sonar, electromagnetic, and chemical. While signiﬁcant progress has been made in the areas of advanced high-resolution sonar, synthetic aperture subbottom sonar, magnetic anomaly detection, and electrical anomaly detection, the area of chemical detection of explosives is still in an early stage of development. The characteristics of explosives submerged in water have been studied, the dynamics of the underwater chemical plumes have been analyzed, and now the sensors capable of detecting underwater explosives have been developed. Many of these advances in underwater chemical sensing have been a result of the Chemical Sensing in the Marine Environment (CSME) program sponsored by the Ofﬁce of Naval Research (ONR). A goal of this program was to develop new tools for detecting underwater unexploded ordnance (UUXO). During the program, a number of trinitrotoluene (TNT) sensors were developed and evaluated by investigators. The explosive contained in UUXO is a potential source for generation of a chemical signature that could be exploited as a means of determining the presence and location of these devices. Because of the low solubility of explosives such as TNT in water, the concentrations of explosive in the water near UUXO are likely to be very low (possibly orders of magnitude lower than the solubility limit of the explosive in water). Once the explosive chemical signature is released into the water, it rapidly mixes into the water column. If there is a prevailing underwater current (which is frequently the case), the plume of explosive is dispersed away from the source, primarily through convective (turbulent) transport processes. This dispersion results in additional dilution of the signature as it is Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

133

134

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT

transported away from the source. If there is a prevailing current, dispersion of the plume will be highly directional, which makes it necessary to position the sensor downcurrent from the source in order to facilitate detection. The challenges in detection are further compounded by the large variety of naturally occurring chemical, biological, and particulate matter encountered in the marine environment. These materials represent potential interferents that further complicate the detection process. Hence, sensors with exquisite sensitivity and selectivity are required for this sensing application. This chapter discusses the development and testing of the Nomadics SeaDog underwater explosives detection system, developed through funding from ONR during the CSME program. This system is an adaptation of Nomadics Fido vapor sensing technology originally developed for the detection of landmines (described in Chapters 7 and 9). The sensor utilizes a proprietary amplifying ﬂuorescent polymer (AFP) technology developed by collaborators at the Massachusetts Institute of Technology (MIT). The ﬂuorescence of AFP is greatly reduced when exposed to target explosive compounds. In addition, the response of the polymer is selective, responding to a very limited subset of chemical constituents found in the environment. The selectivity and sensitivity of the sensor allow the AFP to detect the chemical signature emanating from underwater UXO in real time and without the use of preconcentrators. To further increase the utility of the sensor for underwater sensing applications, the sensor has been deployed on a number of autonomous platforms, enabling real-time detection of underwater explosive plumes. An overview of the sensor platform is presented, along with results of tests conducted at sites in the Gulf of Mexico and the Paciﬁc and Atlantic oceans. 6.2 OVERVIEW OF FATE AND TRANSPORT OF EXPLOSIVES RELEASED FROM UUXO

During World War II, copious quantities of ordnance were lost into the harbor at Halifax, Nova Scotia. Decades later, these UUXO now present a signiﬁcant environmental contamination problem. Studies conducted on this ordnance by Sandia National Laboratories [1] suggest that there may be sufﬁcient concentrations of explosive chemical signature compounds emanating from UUXO to enable detection with chemical sensors. Some UUXO in Halifax Harbor have been shown to produce parts-per-billion levels of explosives in the water near the ordnance. In addition to the parent explosive compound (TNT), other explosiverelated compounds such as 2,4-dinitrotoluene (2,4-DNT) were detected, as were degradation products of TNT such as 4-amino-2,6-dinitrotoluene (4-ADNT), and 2-amino-4,6-dinitrotoluene (2-ADNT). In addition to the contamination detected in the water column, contamination was more frequently detected in sediment from the seaﬂoor near UUXO. It was also noted that the casings of some of the munitions had corroded, exposing the explosive ﬁll material directly to seawater. In some rounds that had corroded, the explosive ﬁll had completely dissolved.

SAMPLING AND SENSING METHODOLOGY

135

In another study, Darrach et al. [2] reported that samples collected near intact UUXO targets contained traces of explosives at up to parts per billion (ppb) concentration levels. The samples were analyzed in the laboratory, using solid-phase microextraction (SPME) to extract target analytes from the samples. The samples were then processed using a reversal electron attachment detection (READ) technique. If the levels of contamination found in these studies are representative of that emanating from most UUXO, the implication is that sensitive chemical sensors such as the SeaDog may be useful for detecting UUXO. Another important consideration in chemical detection of explosives in the marine environment is the nature of the chemical plume emanating from UUXO. Work completed during the CSME program has shown that chemical signatures can persist for long distances away from a source because of the relatively low rates of molecular diffusion of explosives in the aqueous environment. A number of advanced algorithms for predicting how chemical signatures are mixed and transported through the water were also developed during this program. Results from these predictive models were consistent with quantitative experimental data collected from plumes of ﬂuorescent dyes released into the ocean. Plumes of ﬂuorescent dye were shown to persist at detectable levels (using sensitive ﬂuorometers) for over 1000 m from the source. These studies also pointed out that there is great variability in the plume structure. Most simple models of plume dynamics assume a Gaussian distribution of explosive concentrations across and downcurrent in the plume. However, this study reveals that most underwater plumes will be very heterogeneous in structure, consisting of dispersed “ﬁlaments” of explosive at high concentrations relative to the remainder of the water (which contains little to no explosive) within the plume. These ﬁlaments may contain concentrations of explosive signiﬁcantly higher than the concentrations expected assuming a Gaussian distribution of the explosive. 6.3

SAMPLING AND SENSING METHODOLOGY

As described in Section 6.2, it has been shown that convective transport is the primary mechanism for dispersal of explosive signatures in water, resulting in a plume of small ﬁlaments of dissolved explosive emanating away from the source. Hence, the challenge is to search a large volume of water for small, widely dispersed plume ﬁlaments containing dissolved explosive material. This places extreme demands on sensor technology used to detect underwater chemical plumes. The technology must be highly sensitive and selective. In addition, it must be amenable to deployment in seawater. Finally, the sensor must also respond rapidly to the analyte and quickly reset after a response. The SeaDog sensor utilized in this work is capable of near real-time detection of low concentrations of explosives in water. The sensor utilizes novel sensing materials originally developed by collaborators at MIT. These materials are ﬂuorescent polymers that are highly emissive when deployed as solid-state thin ﬁlms. When the polymers interact with nitroaromatic explosives such as TNT, the ﬂuorescence is quenched [3–5]. The response of these materials to target analytes

136

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT

is ampliﬁed because a single molecule of explosive binding anywhere along a single chain of polymer quenches the emission of multiple polymer repeat units, each of which is a ﬂuorophore [6]. The quenching effect is not conﬁned to the polymer chain to which a molecule of explosive is bound. Rather, the quenching effect is delocalized over multiple polymer chains, resulting in quenching of a three-dimensional volume of polymer comprising multiple ﬂuorophores. The sensor responds to other nitroaromatic species often found in the environment near UUXO, such as 2,4-DNT. In addition, the sensor also responds to TNT breakdown products such as trinitrobenzene (a photochemical degradation product of TNT), and amino-dinitrotoluenes (microbial degradation products). The response of the sensor to these compounds is rapid and reversible, enabling the sensor to quickly respond to multiple exposures to target analytes. The underwater sensor platform is derived from the Fido explosives vapor sensor, originally developed under the Defense Advanced Research Projects Agency (DARPA) Dog’s Nose Program. The vapor sensor, whose operation is discussed in Chapters 7 and 9 and in other publications [7–9], was developed for the task of landmine detection. The underwater adaptation of the sensor is very similar to the vapor sensor. In the underwater implementation of the sensor, thin ﬁlms of polymers are deposited onto glass or sapphire substrates. The emission intensity of these ﬁlms is monitored as water (rather than air) ﬂows past the substrate. If the concentration of TNT in the water beings to rise, the polymer will exhibit a measurable reduction in ﬂuorescence intensity. The reduction in emission intensity is proportional to the concentration of target analyte in the water. Because the sensor is small, lightweight, and consumes little power, it proved to be ideal for deployment on autonomous platforms. The initial challenge during development of the sensor was to determine the pros and cons of aqueous versus vapor-based detection architectures. Because the partitioning of explosive into the polymer ﬁlms from the vapor phase is much greater than from the aqueous phase, the sensor response to a given concentration of TNT vapor is orders of magnitude larger than to the same concentration of TNT in the aqueous phase. Hence, to maximize sensor response, the original approach was to develop methods of extracting explosive from the water, and to then vaporize the extracted sample into a stream of carrier gas (air) that would transport the explosive vapor into the sensor for analysis. Extraction of explosive from the water proved to be slow (on the order of 10 s to minutes depending on the method) and inefﬁcient, with only a small portion of the explosive being extracted from the sample in a reasonable length of time. Hence, the sensitivity advantage obtained by analyzing in the vapor phase was largely (but not completely) offset by poor extraction efﬁciencies from the aqueous phase. This unfortunate result had further ramiﬁcations. One of the goals of the CSME program was to develop sensors that could be deployed on autonomous platforms for the purpose of tracing explosives plumes to their source. The autonomous vehicle utilized in the CSME program was not designed to hover at a ﬁxed position. The vehicle was designed to move forward continuously through the water, which is advantageous for interrogating a large underwater area in a short time.

SEADOG SENSOR CONFIGURATIONS

137

However, because the vehicle could not pause or even move slowly through the water, sensors deployed on the vehicle were forced to respond to the presence of explosive very rapidly in order to provide accurate, high-resolution concentration maps of the plume. Assuming that sample collection, analyte extraction, and analysis is a serial process, data points can at best be generated at a rate equal to the total time required to perform each of these three steps. Because the vehicle is moving during all steps of the sensor duty cycle, the length of time required to achieve all three steps must be short or the resolution of the concentration map will be compromised. Since the extraction procedure described in the previous paragraph was a relatively slow process, it would not have been possible to achieve high-resolution concentration mapping using this method. Hence, the method of vapor-phase detection after extraction from samples was abandoned in favor of direct detection from water, which can be accomplished in near real time. Direct sensing from seawater was also attractive because the hardware required to sense explosives directly in water was much less complex than the hardware required to extract explosive and analyze it in the vapor phase. 6.4 6.4.1

SEADOG SENSOR CONFIGURATIONS Prototype Integrated with a Robotic Crawler Platform

A crude schematic of the ﬁrst prototype sensor design is shown in Figure 6.1. The prototype consisted of two watertight boxes housing the sensor hardware. The ﬁrst box contained the sensing head and associated hardware to facilitate sample (seawater) passage through the head. Inside the sensing head is a cylindrical glass waveguide coated on one end with a thin ﬁlm of AFP. The AFP ﬁlm is illuminated with light from a laser diode source that causes the polymer to ﬂuoresce. The light emitted from the ﬁlm is internally reﬂected down the length of the cylindrical waveguide. The emitted light exits out the opposite (uncoated) end of the waveguide, passes through an 80/20 mirror, and then impinges onto an optical ﬁlter that blocks passage of stray light from the laser excitation source, but allows light emitted from the AFP to pass. The light passing through the ﬁlter impinges on a photomultiplier tube, enabling the intensity of the light emitted from the polymer to be quantiﬁed. Water is pumped through the sensor head and across the AFP ﬁlm. When explosives and related compounds such as 2,4-DNT or amino-DNTs are present in the water, the intensity of the emission from the AFP ﬁlm is dramatically reduced in proportion to the concentration of target analyte in the sample. The change in emission intensity is measured with a photomultiplier tube (PMT). The output from the PMT is processed via an onboard digital signal processor (DSP). The processed signal is then transferred to a handheld computer housed in one of the watertight boxes. The handheld computer controls sensor functions and allows logging of the data stream generated by the sensor. The data can be downloaded for postprocessing after a mission. In addition, the handheld computer generates a graphical display of sensor response that can be viewed in real time through the enclosure by a diver if so desired.

138

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT

Figure 6.1 Schematic of the ﬁrst SeaDog sensor design.

The second box contains a peristaltic pump and a servoactuator. Both the pump and servo are controlled electrically from the sensing head and are powered from the same power supply. The separation of the sensing head from the pump and servo provide electrical and mechanical isolation and address space constraints associated with mounting the system on the autonomous underwater vehicle. The peristaltic pump enables operation at a variable ﬂow rate and has bidirectional ﬂow capability. The servo actuates a movable sample inlet tube that can be raised or lowered by remote control to enable precise positioning of the inlet relative to the source or in the source plume. A photograph of the entire system, mounted on a metal plate that attaches to the robot, is shown in Figure 6.2. Initial testing of the sensor was successfully performed in our laboratory prior to integration with the vehicle. Integration of the sensor with a Foster-Miller Talon was then performed in Panama City, Florida, at Coastal Systems Station (CSS). Figure 6.3 illustrates the sensor assembly, consisting of the two boxes and the robot cover plate, attached to the robotic crawler. In this conﬁguration, data from the SeaDog is transferred to the robot control computer via a dedicated RS-232 link. The sensor and crawler were connected to a towed ﬂoat that contained system batteries and communications hardware to transmit sensor and crawler data to and from remote computers. A video camera was also integrated with the system to enable visual monitoring of the underwater tests. After initial testing of system integration in the robotics lab at CSS, the system was taken to a sandy beach area for mobility and integration trials. Other than

SEADOG SENSOR CONFIGURATIONS

139

Figure 6.2 SeaDog sensor and pump housings mounted for robotic integration.

Figure 6.3

SeaDog mounted on the Foster-Miller crawler during testing.

a minor communications problem that was rapidly diagnosed and corrected, the integration trials progressed smoothly. The system was then tested for basic operation underwater. After moving the crawler and sensor into the water, the pump system was activated and the operation of the servo-controlled inlet was conﬁrmed. These tests were successful, enabling sea trials of the system. The results of these trials are presented in Section 6.5.1. 6.4.2

Diver-Deployed SeaDog and Initial Integration with the REMUS

After successful testing of the sensor prototype described in Section 6.4.1, the prototype sensor was reconﬁgured for mounting onto an autonomous underwater

140

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT

Figure 6.4

SeaDog deployed on the REMUS during a sea trial.

vehicle (AUV). The platform selected for integration was the Remote Environmental Monitoring UnitS (REMUS) platform developed by Woods Hole Oceanographic Institute (WHOI) (Fig. 6.4). This adaptable platform is capable of carrying a variety of payloads including sonar and sensors of various types. The initial sensor prototype described in Section 6.4.1 was modiﬁed to ﬁt into a sealed enclosure that was designed to interface with the REMUS vehicle. The enclosure contained a pump and an electronics package very similar to the prototype sensor. Minor changes were made to the optical setup. The cylindrical waveguide was replaced with a circular sapphire plate coated with AFP on its front face. The plate was positioned inside a sample chamber through which seawater was pumped at a rate of 180 mL per minute. The sensor module also contained a ﬂuorometer designed to detect marker dyes that were introduced into the water during some missions. The dye plume gave a visual indication as to the location of the plume of explosives emanating from the source. The integrated ﬂuorometer was able to detect the marker dye, signaling when the AUV passed through the plume. If the plume also contained explosives, the SeaDog and ﬂuorometer would respond simultaneously, conﬁrming the sensor response to explosives. An optical transmittance sensor was included in the package to measure turbidity of the water. A high-resolution temperature sensor was also onboard to monitor the temperature of the water during testing. Finally, a generic REMUS-compatible communications protocol was developed through collaboration with WHOI and the Space and Naval Warfare Systems Center (SPAWAR). This generic protocol enabled communication between sensors in the payload with the onboard computer in the REMUS. All data collected during a mission was logged by the system and was downloaded to an external computer for analysis after completion of a mission. The SeaDog sensor payload as deployed on the REMUS is illustrated in Figure 6.5. The REMUS sensor suite was also conﬁgured for diver deployment (Fig. 6.6). The sensor module was easily detachable from the REMUS and could be mounted

SEADOG SENSOR CONFIGURATIONS

Figure 6.5

141

SeaDog sensor payload conﬁgured for deployment on the REMUS.

Figure 6.6

SeaDog sensor package conﬁgured for diver deployment.

onto a plate equipped with a handle for the diver to grasp. A transparent box was also attached to the plate to house the control system, a handheld computer, and a battery to power the system. The data collected during a diver-deployed mission could be stored to the handheld computer, or viewed in real time on the graphical interface of the handheld computer. Audio output was also added to the system for use during diver deployment. The audio output consisted of a

142

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT

series of “clicks” whose frequency was proportional to the response of the sensor. The audio output was audible underwater, enabling feedback of sensor response to the diver during turbid or low-light operating conditions when the graphical interface was not visible. Test results for the diver-deployed and initial REMUS-deployed sensor tests are presented in Section 6.5.2. 6.4.3

SeaDog Miniaturization: The SeaPup

Lessons learned during sea trials with the SeaDog led to development of the SeaPup sensor, which is a miniaturized version of the SeaDog sensor. In addition to being signiﬁcantly smaller than its predecessor (see Fig. 6.7), the SeaPup was more sensitive (by a factor of approximately 6) than the SeaDog. More importantly, the new design exhibited greatly enhanced response kinetics, making it possible to better resolve the structure of plumes. The accelerated rate of response was due largely to a redesign of the sensor sample cell. The new proprietary design achieved these gains in performance by optimizing the mass transport of target analyte from the sample volume to the AFP-coated substrate. The ﬂuorometer and optical absorbance sensors were eliminated from this design, further reducing the size of the sensor. The new sensor design made it possible for the ﬁrst time to map a plume of explosive emanating from a source in the marine environment. The results for sea trials using the SeaPup sensor design are presented in Section 6.5.3.

Figure 6.7

Comparison of the size of the SeaDog (left) to the SeaPup (right).

RESULTS OF SENSOR TESTS CONDUCTED IN THE MARINE ENVIRONMENT

143

6.5 RESULTS OF SENSOR TESTS CONDUCTED IN THE MARINE ENVIRONMENT 6.5.1

Tests of the Sensor Prototype on a Crawler Vehicle

The ﬁrst tests of the prototype sensor described in Section 6.4.1 were conducted during February of 2002 in conjunction with CSS. During these tests, the prototype sensor mounted on a Foster-Miller crawler robot (provided by CSS) was used to detect traces of explosive emanating from underwater targets. Figure 6.8 is a photo of the test site, illustrating the sensor on the robot, and the towed ﬂoat (being carried by the two test personnel located in the center of the image) housing batteries and communications hardware. This photo was taken just prior to deployment of the system for testing. All targets in these tests were secured to a post located in 4 ft of water. Targets were suspended from the post by a rope so that the targets were approximately 1 ft from the seaﬂoor. For initial tests, a target was constructed using a commercially available TNT simulant. The simulant consists of TNT coated onto particles of sand, but the quantity of TNT in the simulant is insufﬁcient to sustain a detonation. This target was placed inside a water-permeable fabric bag and was afﬁxed to the pole as previously described. For the initial test, the crawler approached the target from a direction upcurrent of the prevailing current so that any chemical signature emanating from the target would be transported away from the sensor. No sensor response was noted as the sensor approached the target from upcurrent. The robot was then repositioned by executing a number of turns, placing the robot and sensor approximately 3 ft downcurrent from the target. Shortly after arriving in this position, two sensor responses (seen in Fig. 6.9 near the 230 time position) were observed. These responses were consistent with that of TNT in the water. Note that the response to TNT was rapid and reversible.

Figure 6.8

Sensor/crawler system prior to deployment at the test site.

144

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT

Figure 6.9 Response to TNT 3 ft downcurrent of target.

Immediately after these responses were noted, the robot was positioned closer to the target, approaching to a distance of approximately one foot downcurrent from the source. The response at this position is indicated in Figure 6.10. A comparison of Figures 6.9 and 6.10 reveal that the frequency of encounters between TNT-containing plume ﬁlaments increases closer to the target. This is consistent with the assumption that the distance between plume ﬁlaments containing detectable concentrations of TNT increases farther from the target as the plume ﬁlaments become more widely dispersed. Also note that while the number of responses per unit time is greater near the target, the magnitude of the sensor response to each ﬁlament is similar at both distances investigated. This result suggests that while the plume ﬁlaments become progressively farther apart, the concentration of explosive within the plume ﬁlaments does not rapidly decrease with distance from the source.

Figure 6.10 Sensor response 1 ft downcurrent of target.

RESULTS OF SENSOR TESTS CONDUCTED IN THE MARINE ENVIRONMENT

145

To conﬁrm that the response was to TNT and not a chemical interferent in the water, the sensor was moved to numerous locations around the target that were upcurrent from the prevailing current near the target. No responses consistent with the presence of TNT were observed unless the sensor was downcurrent from the target. Water samples were also collected at the site and were later analyzed in our lab for the presence of TNT. The samples collected were extracted into toluene and analyzed by gas chromatography (GC) using an electron capture detector (ECD) and by a second chromatographic system using a Fido vapor sensor as the detector. The results from both analytical systems conﬁrmed the presence of TNT in the samples collected near the target. The concentration of TNT in samples collected from 1 to 3 ft from the target ranged from 1 to 20 ppb by mass (as measured by GC-ECD). On the ﬁnal day of testing a TNT demolition charge was substituted for the test target constructed from the TNT simulant. The block of explosive was prepared in a manner consistent with how the block would normally be deployed for use. The block was then deployed from the post as previously described. The sensor was positioned at the downcurrent side of the post at a distance of approximately 3 ft. The sensor immediately indicated a strong response to the explosive. Initial tests of the Nomadics SeaDog sensor integrated with a Foster-Miller crawler were extremely promising. The sensor was readily integrated with the crawler, enabling tests to be conducted seamlessly. The sensor was able to detect plumes of TNT emanating from test targets. This was accomplished without sample preconcentration and in real time. A more detailed account of this test is presented by Dock and co-workers [10, 11]. 6.5.2 Tests of the Diver-Deployed SeaDog Sensor and Initial Integration to the REMUS

The SeaDog sensor was tested in September 2002 at a test site in the Paciﬁc Ocean. During this test, the SeaDog sensor module was tested in the diverdeployed conﬁguration described in Section 6.4.2. The diver-deployed sensor module successfully demonstrated detection of TNT near an underwater target during these tests. After completion of the diver-deployed portion of the testing, the sensor module was then reconﬁgured for operation on the REMUS. Due to time constraints (the REMUS was being utilized for multiple tasks during this test), it was not possible to test the SeaDog integrated with the REMUS in the marine environment. However, the sensor was integrated to the REMUS and was tested on the benchtop. The system integration was successful except for a minor software problem that made it impossible to download data from the system computer after a mission. This software issue was resolved prior to a second series of tests conducted at the same site 2 months later (November 2002). During the second round of tests conducted in November of 2002, detection of a plume of TNT at a signiﬁcant standoff distance was demonstrated for the ﬁrst time. During these tests, a TNT source was constructed and deployed in the ocean at a depth of approximately 30 m. Using the diver-deployed SeaDog module (shown in Fig. 6.11), divers were able to detect a plume of TNT at distances of up to 30 m

146

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT

Figure 6.11

Diver-deployed SeaDog sensor during sea trials.

from the source. For the ﬁrst 6 min of data collection (the upper data trace in Fig. 6.12), the diver was located upcurrent from the TNT source. As can be seen from the upper data trace, no sensor responses were noted, as would be expected when the sensor is located upcurrent from the source. No responses suggestive of chemical interferents in the water were observed. However, numerous responses were observed when the sensor was positioned downcurrent of the source in the TNT plume. The lower data trace in Figure 6.12 was collected at distances of up to 30 m downcurrent from the source as the diver transected the plume numerous times with the sensor. Large responses to the TNT plume were obtained. During the second round of testing, it was possible to deploy the sensor on the REMUS. Figure 6.13 is a photograph of the system while operational in the marine environment during the test. While no positive indications of TNT were observed during the test, the sensor was successfully operated on the REMUS in the marine environment. Lessons learned during this deployment led to development of the SeaPup sensor, which was constructed and tested in early 2003. 6.5.3

Tests of the SeaPup Sensor Integrated on the REMUS

In June 2003, the SeaPup sensor was tested at a site off the Atlantic coast. As was discussed in Section 6.4.3, the SeaPup sensor showed almost an order of magnitude improvement in sensitivity over the SeaDog due to design enhancements incorporated into the system. In addition, the SeaPup responds much more rapidly to TNT than the SeaDog, which is an important advantage for mapping chemical plumes in the marine environment.

RESULTS OF SENSOR TESTS CONDUCTED IN THE MARINE ENVIRONMENT

147

Figure 6.12 Data collected with diver-deployed sensor during sea trials.

Figure 6.13 SeaDog deployed on the REMUS AUV during sea trials.

In these tests the SeaPup sensor was mounted on the REMUS AUV and commanded to interrogate an area greater than 22,000 m2 . The REMUS was programmed to execute a back-and-forth “mow-the-lawn” style of mapping mission perpendicular to the direction of the prevailing current near the target. As the REMUS followed the programmed mission, the time-domain TNT concentration data is integrated with the longitude–latitude positioning data to generate a concentration map of the area. Figure 6.14 is a concentration map generated near one of the test targets. The TNT plume emanating from the target is clearly visible in the map, with the concentration of TNT decreasing with distance from

148

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT

Figure 6.14 See color plates. Detection of TNT plume with the SeaPup sensor mounted on the REMUS AUV off the Atlantic Coast of the U.S. in June 2003.

the source. The plume was detected at over 100 m from the source. To our knowledge, the SeaPup is the only sensor to demonstrate real-time detection of an explosive plume from an AUV in the marine environment from any distance. While not demonstrated during this work, another interesting capability of the REMUS AUV system is that a sensitive, real-time explosive sensor could supply explosive concentration information to an adaptive mission planner that is capable of modifying the course of the AUV, so that the AUV tracks the explosive plume to its source. This capability, if further developed, could enable detection of UUXO by tracking explosive plumes to source. During these tests, the SeaPup sensor had a third-party veriﬁed, in-ﬁeld TNT sensitivity of 4 ppb. While it is generally believed that a sensitivity of 4 ppb is not adequate for detecting all UUXO, the results of the studies of UUXO in Halifax Harbor [1, 2] suggest that this level of sensitivity will enable detection of some UUXO items. Further improvements in sensitivity of detectors may soon make detection of UUXO possible with chemical sensors, providing an orthogonal detection capability for detection methods such as sonar. 6.6

CONCLUSIONS

To our knowledge, this is the ﬁrst demonstration of a sensor capable of realtime detection of a TNT plume in the marine environment at standoff distances (up to 100 m from the source) while deployed on an autonomous underwater vehicle. The sensor has shown virtually no sensitivity to chemical interferent during testing in the marine environment. While the sensitivity of the detector is excellent, its sensitivity is not adequate at its present state of development to

REFERENCES

149

detect all types of underwater unexploded ordnance. However, based on published accounts of the concentration of explosive signature compounds in the environment near some UUXO, the system should have adequate sensitivity to detect some UUXO, especially if the UXO has been underwater for decades. While not discussed here, new AFPs are being developed for enhanced detection of RDX and other explosives not detectable with the sensors at the time they were tested. With further development, chemical sensor systems such as the SeaDog could be a valuable tool for detection of UUXO, especially when deployed on autonomous underwater platforms. ACKNOWLEDGMENTS

We would like thank Dr. Keith Ward and Dr. Angela Ervin, formerly at the Ofﬁce of Naval Research (ONR) for sponsoring this work, and Don Robeson, Rudy Arrieta, and their colleagues at USN Coastal Systems Station (Panama City) for their efforts in the planning and conduct of the trials using their test assets. We also wish to thank the Woods Hole Oceanographic Institute (WHOI) and the Space and Naval Warfare Systems Center (SPAWAR) for technical support during integration of the SeaDog sensor with the REMUS, and for support during sea trials. REFERENCES 1. Rodacy, P., P. K. Walker, S. D. Reber, J. Phelan and J. V. Andre. Chemical Sensing of Explosive Targets in the Bedford Basin, Halifax, Nova Scotia, Sandia National Laboratories, SAND2001-3569, November 20, 2001. 2. Darrach, M. R., A. Chutjian, and G. A. Plett. Trace explosives signatures from World War II unexploded undersea ordnance. Jet Propulsion Laboratory, Environmental Science Technology 32(9), 1354 (1998). 3. Yang, J. S. and T. M. Swager. Porous shape persistent ﬂuorescent polymer ﬁlms: An approach to TNT sensory materials. J. Am. Chem. Soc. 120, 5321–5322 (1998). 4. Yang, J. S. and T. M. Swager. Fluorescent porous polymer ﬁlms as TNT chemosensors: Electronic and structural effects. J. Am. Chem. Soc. 120, 11864–11873 (1998). 5. Williams, V. and T. M. Swager. Iptycene-containing poly(aryleneethynylene)s. Macromolecules 33, 4069–4073 (2000). 6. Zhou, Q. and T. Swager. Methodology for enhancing the sensitivity of ﬂuorescent chemosensors: Energy migration in conjugated polymers. J. Am. Chem Soc. 117, 7017–7018 (1995). 7. Cumming, C., C. Aker, M. Fisher, M. Fox, M. la Grone, D. Reust, M. Rockley, T. Swager, E. Towers, and V. Williams. Using novel ﬂuorescent polymers as sensory materials for above-ground sensing of chemical signature compounds emanating from buried landmines. IEEE Trans. Geosci. Remote Sensing 39(6), 1119–1128 (2001). 8. Fisher, M. and C. Cumming. Detection of trace concentrations of vapor phase nitroaromatic explosives by ﬂuorescence quenching of novel polymer materials, in Proceedings of 7th International Symposium on the Analysis and Detection of Explosives, Defense Evaluation and Research Agency, Edinburgh, Scotland, UK, June, 2001.

150

DETECTION OF TRACE EXPLOSIVE SIGNATURES IN THE MARINE ENVIRONMENT

9. Fisher, M., M. la Grone, and J. Sikes. Implementation of serial amplifying ﬂuorescent polymer arrays for enhanced chemical vapor sensing of landmines, in Proceedings of UXO/Countermine Forum 2003, Orlando, Florida, September, 2003. 10. Dock, M., M. Fisher, and C. Cumming. Novel detection apparatus for locating underwater unexploded ordnance, in Proceedings of the 5th International Symposium on Technology and the Mine Problem, Mine Warfare Association, Monterrey, California, April 2002. 11. Dock, M., M. Fisher, and C. Cumming. Sensor for real-time detection of underwater unexploded ordnance, in Proceedings of UXO/Countermine Forum 2002, September 2002, Orlando, Florida.

CHAPTER 7

EXPLOSIVES DETECTION USING ULTRASENSITIVE ELECTRONIC VAPOR SENSORS: FIELD EXPERIENCE MARK FISHER Nomadics, Inc., an ICx Technologies Company

7.1

INTRODUCTION

This chapter summarizes almost a decade of ﬁeld experience using trace-level ® electronic vapor sensors (the Nomadics Fido ) to detect explosives and explosive devices. This sensor was originally developed as part of the Defense Advanced Research Projects Agency (DARPA) Dog’s Nose Program. The objective of this program was to develop advanced sensor technology for the purpose of detection of low-metal-content (plastic) antipersonnel landmines that are difﬁcult to detect with conventional landmine detection systems. Since the completion of this program, the sensor has undergone further development for landmine detection applications with funding from the U.S. Army Night Vision and Electronic Sensors Directorate (NVESD). Experience gained as a result of ﬁeld testing conducted as part of these programs will be presented. In addition to landmine detection, the sensor has been evaluated and found to have utility for other explosive sensing applications, but due to the sensitive nature of much of this work it cannot be presented here. Signiﬁcant effort has been devoted to development of extremely sensitive trace detection systems capable of detecting a variety of explosive threats. While sensitive detectors are required to detect these threats, our experience and that of others working in this area have shown that a sensitive detector is not alone sufﬁcient to ensure detection of targets containing explosives. The methods by which trace samples are collected and presented to the sensor are critically important, and often determine whether or not a target will be detected. Sample acquisition is often mistakenly regarded as trivial by comparison to sensor development, and is hence largely ignored during sensor design. To neglect sampling issues during development of explosives detection systems virtually guarantees that the system Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

151

152

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

will be of little value operationally, regardless of the sensitivity of the detector. Our experience has been that a majority of the difﬁculties encountered during ﬁeld trials have not been related to sensor hardware, but rather to sampling issues. Hence, issues related to sampling will be presented throughout this chapter where relevant. Having a basic understanding of the nature of the explosive chemical signature emanating from a given target is also critical to achieving high probabilities of detection of explosive devices. The parameter most often considered is the concentration (or mass ﬂux) of explosive emanating from the target. Generally, the sensitivity of a detector can be measured in a laboratory setting and will have been established prior to use in the ﬁeld. Similarly, the rate of release of explosive chemical signature constituents emanating from an explosive-containing target can be measured in the controlled environment of a laboratory. However, when this target is translated from a laboratory to a ﬁeld environment, the inﬂuence of environmental factors can signiﬁcantly alter the rate of release of explosive from the target. Moreover, the difﬁculty in measuring the explosive “source term” for targets in the ﬁeld is compounded by the fact that environmental factors affecting the rate of release of explosive can change rapidly. Further, there is no “standard” explosive device. For example, the vapor signature produced for two apparently identical landmines can vary signiﬁcantly, even though they were both manufactured to the same speciﬁcations. Subtle differences in how the devices are deployed can also have an impact on detection. An understanding of how the vapor signature evolves after it is released into the environment is also important. For example, the concentration of explosive vapor released into the boundary layer of air in contact with an explosive device may be quite large, approaching the equilibrium concentration value of the explosive contained within the device. However, the concentration of the explosive signature can diminish by orders of magnitude over a relatively short distance from the target. Hence, the signature becomes progressively more difﬁcult to detect as the distance from the target increases. If the detection application requires tracing a “vapor plume” to its source from a distance, this factor can be a signiﬁcant limitation. In addition, environmental inﬂuences can degrade the chemical constituents into other substances, and if the detector being utilized cannot detect these degradation products, detection probabilities will be compromised. The “fate and transport” of explosives in operational environments is a complex subject that has not been extensively studied for many detection applications and should be a focal point of ongoing explosives detection research. As will be discussed, explosives fate and transport issues and the environmental factors that inﬂuence them play a key role in designing sampling strategies, and also must be considered during development of training materials used to teach sensor operators how to effectively use a sensor system. The emphasis of this chapter will be on vapor-phase detection of explosives. While the data and results presented here were obtained using the Nomadics Fido sensor, many of the lessons learned are applicable to other trace sensor systems and also to detection using canines.

RELEVANCE OF FIELD TESTING TO SENSOR DEVELOPMENT

7.2

153

RELEVANCE OF FIELD TESTING TO SENSOR DEVELOPMENT

In 1997, DARPA launched the Dog’s Nose Program. The program brought together researchers from a variety of science and engineering disciplines to address the problem of landmine detection. In this program canines, regarded by many as the “gold standard” in mine detection, were used as a model detection system from which to base sensor design and performance. A critical component of this program was ﬁeld testing. DARPA maintained a test mineﬁeld for the purpose of testing sensor performance. All systems were tested at the facility at least twice a year. Some investigators were not involved directly in sensor development, but ﬁeldwork was still a component of most of their statements of work. For example, experts in canine olfaction studied the canine olfactory system and the performance of mine detection dogs. Some of this work was done in a laboratory, but in the latter stages of the program, canines were taken to the ﬁeld for testing alongside of sensors in order to evaluate and compare their performance under ﬁeld conditions. There was a widely held belief prior to this program that chemical vapor sensors could never achieve the level of sensitivity required to detect landmines in the same manner as canines. This assertion was shown to be false, as an electronic vapor sensor developed during this program (the Nomadics Fido) demonstrated for the ﬁrst time detection of landmines at performance levels comparable to experienced mine detection canines during blind ﬁeld trials. The motivation provided by mandatory biannual ﬁeld testing, coupled with lessons learned in the ﬁeld, greatly accelerated development of the sensor, driving the achievement of canine-comparable detection capability with less than 2 years of sensor development [1]. It should also be noted that the work done by other researchers was instrumental in reaching the objective of detection of landmines with electronic sensors. In particular, the results of researchers investigating the mechanisms of explosive release from mines into the soil, and how these explosive “chemical signatures” are modiﬁed as they are transported through the soil were very useful [2–4]. These factors strongly inﬂuence probabilities of detection because explosive traces must be released from the mine and then transported to the surface of the ground where they can be sampled and detected. This is true whether the sensor is electronic or a canine. An important outcome of this work was the identiﬁcation of key chemical constituents of the mine chemical signature and their concentrations, as well as how the signature is inﬂuenced by environmental factors [5–8]. While some of these experiments were performed in laboratories, much of this data was collected from test mineﬁelds that differed from real mineﬁelds in only one respect—the landmines were ﬁtted with inert detonators (or a storage plug that sealed the detonator well) to enable testing to be conducted without accidental detonation of mines. From this data, sensor developers gained important insights into which explosive signature constituents were most useful as target analytes and how to effectively sample these signatures. It is understood that new technology must be validated through laboratory and ﬁeld testing before it can be ﬁelded operationally. However, sensor developers

154

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

are often reluctant to subject new technology to ﬁeld testing until the technology is regarded as reasonably mature. It is true that the rigors of testing a sensor in the ﬁeld will likely accentuate known weaknesses in the design, but ﬁeld testing will also reveal issues that would likely never manifest in a laboratory environment. It is often much easier to correct weaknesses in a system in the early stages of development while the design is still evolving rather than to address them late in the design process. An additional beneﬁt of ﬁeld testing that is often overlooked is that testing is often conducted with potential end users of the equipment present, many of which have considerable operational experience. These individuals are uniquely qualiﬁed to provide expert feedback on the sensor design and potential concepts of operation. Sensor developers rarely have operational experience using the equipment they are developing. Hence, a common problem that arises is that the equipment, while functional, is not always practical to use in the ﬁeld. This problem could be largely avoided if developers and end users are both involved in the design cycle. Involvement of end users in the design process increases the probability that the equipment will be suitable for operational deployment. Field exercises provide an excellent opportunity for this to occur. In fact, the feedback from experienced operators and the lessons learned from the ﬁeld tests have driven many of the system enhancements that have been incorporated into the sensor. 7.3

OVERVIEW OF THE VAPOR SIGNATURES OF EXPLOSIVES

The Fido sensor [9–13], described in detail in Chapter 9, is an extremely sensitive detector for trinitrotoluene (TNT) and explosives formulations containing TNT such as Composition-B (a mixture of 60% RDX and 40% TNT). The polymeric sensing material used in the sensor was originally developed for detection of nitroaromatic explosives (TNT) [14–17], but through additional development collaborators at the Massachusetts Institute of Technology (MIT) have developed new sensing materials that possess excellent sensitivity to RDX and pentaerythritol tetranitrate (PETN). These explosives are utilized in many types of plastic explosives including Semtex, C-4, and PE-4. TNT and plastic explosives are widely utilized in military and civilian applications. These explosives are also commonly used by terrorists. When a source of explosive is not readily available, terrorists are more frequently turning to the use of improvised explosives, including triacetone triperoxide (TATP) or hexamethylene triperoxide diamine (HMTD). These explosives can be synthesized from readily available materials. Hence, they are being used frequently in terrorist attacks. The explosives mentioned here are only a few selected from a long list of materials that can be used as explosives. This presents an unusual detection challenge. The chemical and physical properties of explosives vary widely, so it is a challenge to design a sensor that can detect all explosives equally well. One such property is the equilibrium vapor pressure of explosives. From Figure 7.1, which is a plot of the equilibrium vapor pressures of selected explosives at 25◦ C,

OVERVIEW OF THE VAPOR SIGNATURES OF EXPLOSIVES

155

1.00E+00 EGDN

Equilibrium Vapor Pressure

1.00E−01

NG

TNT

RDX

PETN

1.00E−02 1.00E−03 1.00E−04 1.00E−05 1.00E−06 1.00E−07 1.00E−08 1.00E−09 1.00E−10

Figure 7.1 Equilibrium vapor pressures of common explosives (in Torr) at 25◦ C.

it can be seen that the vapor pressures of these common explosives span 9 orders of magnitude [18]. The equilibrium vapor concentration of an explosive at a given temperature represents an upper limit to the concentration of explosive vapor available for detection. Many factors serve to lower the vapor-phase concentration of explosive near an explosive device to less than the equilibrium value. First, the ﬂux rate of explosive emanating from the device may not be adequate to maintain equilibrium concentrations in the boundary layer of air at the surface of the device. Regardless of the concentration of explosive in the boundary layer, the vapor is rapidly dispersed into the ambient air near the device, further lowering the concentration of analyte. In practice, the average concentration of explosive vapor emanating from an explosive device is orders of magnitude lower than the equilibrium value, and the average concentration at a given point in space decreases rapidly as the distance from the device increases. Hence, detection of explosives in the vapor phase requires extremely sensitive detectors and/or the use of sample preconcentration methods. It is often assumed that transport of explosive vapor away from a source results in a Gaussian-like distribution of explosive downwind of the source. This is approximately correct if one is considering the average concentration of explosive at a given point in the plume as a function of time. However, the instantaneous concentration of explosive in the plume at a given position in space away from the source can range from zero to values approaching the source concentration. Convective dispersion (due to turbulent ﬂow of ambient air) can quickly transport vapors a signiﬁcant distance from their source. When this occurs, the vapor plume, which may be fairly homogeneous very near the source, is fragmented into smaller and smaller ﬁlaments by the turbulent ﬂow. These ﬁlaments eventually

156

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

1.005

Normalized Response

1 0.995 0.99 0.985 0.98 0.975 0.97 0.965 0.96 0.955 0

50

100

150

200

250

Elapsed Time (s)

Figure 7.2

Response of Fido to a vapor plume 2 m downwind of a TNT source.

disperse due to the action of molecular diffusion. However, molecular diffusion is a slow transport process compared to turbulent convection, so before the plume ﬁlaments disperse via molecular diffusion, they can be transported a signiﬁcant distance from the source while retaining a signiﬁcant fraction of the source concentration. Molecular diffusion eventually becomes the dominant mode of dispersion, but this does not occur until the plume ﬁlaments become quite small. The heterogeneous nature of a TNT vapor plume is presented in Figure 7.2. This ﬁgure illustrates the response of Fido to a vapor plume encountered during an outdoor test of the sensor at an arid test facility. The source for the plume was three 1-lb demolition blocks of TNT. The blocks were stacked on the ground 2 m upwind of the sensor. The TNT source and the detector remained stationary throughout the sampling event. Wind during sampling was light (less than 2 m/s) and variable. Fido responded strongly when plume ﬁlaments were drawn into the sensor (at between 120 and 180 s into the sample), but a vast majority of the time the sensor did not respond because it was sampling relatively “clean” air. A good visual example of this type of plume behavior can be observed by watching smoke from a cigarette. Near the cigarette, the smoke is very concentrated and reasonably homogeneous in composition, but the smoke plume quickly fragments into ﬁlaments that can persist for a signiﬁcant distance from the cigarette. The concentration of smoke in the ﬁlaments is large relative to the concentration in the air surrounding the ﬁlament. An important point should be made here. If the average concentration of TNT in the air over time were computed at the point in space where the data in Figure 7.2 was collected, the average concentration would be at or below the minimum detection limits of the detector. However, the sensor was able to detect explosive because it was not measuring the average concentration but the instantaneous (real-time) concentration of explosive at that point in space, which at times is signiﬁcantly greater than the average concentration.

OVERVIEW OF THE VAPOR SIGNATURES OF EXPLOSIVES

157

The heterogeneous nature of vapor plumes has several important implications for sampling and sensing. First, plume ﬁlaments containing concentrations of explosive slightly less than the concentration at the source can persist a signiﬁcant distance from the source. This enables the possibility of standoff detection, as demonstrated by the data shown in Figure 7.2. However, the greater the distance from the source the ﬁlaments are sampled, the smaller the concentration of explosive in the ﬁlaments, and the number of plume ﬁlaments per unit volume of air also decreases as the distance from the source increases. This means that the greater the distance between the point of sampling and the source, the larger the volume of air that must be processed in order to ensure that the plume ﬁlaments have been sampled. Except for laboratory reference materials that have been rigorously puriﬁed, explosives are not pure materials. For example, TNT contains synthesis byproducts that are the result of incomplete nitration of toluene [19]. Depending on the synthesis process and how much effort was made to purify the explosive, mono- and di-substituted nitrotoluenes may be present in TNT at levels ranging from a trace up to a few percent by mass. This has important implications for detection because both mono-and dinitrotoluenes (DNT) have higher vapor pressures than TNT. Hence, the vapor-phase concentrations of these contaminants may be higher than that of TNT near an explosive device utilizing TNT as the explosive. Plastic explosives are also mixtures containing one or more explosives mixed with a variety of other compounds, including materials that may be more volatile than the explosive in the mixture. If these higher volatility compounds are unique to explosives, they could be used as a means of identifying the explosive. However, if the compound in question is widely used in a variety of nonexplosive products, the sensor could be prone to false alarms and would therefore be of little utility. The composition of the vapor signature of a given type of explosive can be quite variable. For example, the relative abundance of explosive signature constituents for a given type of explosive can vary signiﬁcantly depending on the source of the explosive [4]. For example, TNT from different sources can contain a range of concentrations of 2,4-DNT, which could affect probabilities of detection if 2,4-DNT were the preferred analyte for detection. In addition, the composition of explosives can change with age. Degradation of the explosive into other compounds does occur, and escape of more volatile constituents from an explosive can occur with age, changing the chemical vapor signature of the explosive over time. Environmental factors can also change the chemical signature of an explosive. For example, in soils near landmines TNT can be quickly converted to amino-dinitrotoluenes by the action of soil microbes, altering the chemical signature of the mine [3]. Ideally, any compound found in an explosive that is unique to the explosive, and is always found in the vapor signature of the explosive regardless of its source or age could be used as a means of detecting the explosive. Unfortunately, these requirements severely limit the potential analyte set for consistent, reliable detection. One could argue that the only compound that can be guaranteed to be present in an explosive headspace are vapors of the

158

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

actual explosive, and as such should be the target analyte for a vapor detector. However, if certain relatively uncommon signature compounds were consistently present in the headspace of a given type of explosive, and these compounds could be reliably detected when present, target analytes other than the parent explosive could be used to detect certain explosive devices, especially when the parent explosive has a very low vapor pressure. 7.4 7.4.1

LANDMINE DETECTION Introduction to the Mine Problem

The worldwide landmine problem is multifaceted, encompassing a myriad of humanitarian, political, and technical challenges. As a result of decades of armed conﬂicts around the world, it is estimated that 60 to 70 million mines are deployed in approximately 70 countries. It has been estimated by the International Committee of the Red Cross that a person is maimed or killed by a mine every 20 min [20]. In addition to the physical and mental trauma inﬂicted by landmines, landmines restrict access to land and critical natural resources, which can have a disastrous effect on the economy of mined areas. Clearing mineﬁelds is also expensive, making it difﬁcult for the governments of most countries affected by mines to subsidize a signiﬁcant fraction of needed demining operations. Demining operations are largely subsidized through humanitarian organizations and governments located outside mine-affected areas. In general, the level of funding has not been adequate nor consistent enough to clear mineﬁelds in a timely manner. Hence, mineﬁelds often remain long after the conﬂict that resulted in the mines being emplaced has ended. The landmine problem is also not without technical challenges. In particular, modern plastic-cased antipersonnel mines with very low metal content are much more difﬁcult to detect than metal-cased mines, which can routinely be detected with a metal detector or ground-penetrating radar. Hence, new technologies with enhanced capability for detecting plastic-cased mines is needed. Acoustic, infrared, and spectroscopic methods including laser-induced breakdown spectroscopy (LIBS) are examples of technology that have demonstrated detection of plastic-cased mines with limited success. In addition to new technologies, signiﬁcant improvements have recently been made in decades-old mine detection technology (metal detection and ground-penetrating radar in particular). Nevertheless, widespread implementation of the enhanced technology in demining operations has been limited due to the high cost and complexity of ﬁelding the new equipment. In spite of recent advances in detector technology, metal detectors, mine probes, and canines remain as the most widely utilized mine detection tools. All of these methods have been utilized with some success, but all have limitations. For example, metal detectors must be operated at extremely high gain settings to obtain adequate sensitivity to detect low-metal-content mines. This is an issue because mineﬁelds are often littered with metallic objects (e.g., fragments from exploded munitions). Hence, false alarms are common, limiting the utility

LANDMINE DETECTION

159

of metal detectors in these environments. “Probing,” a technique that involves “feeling” for mines by methodically inserting rigid rods into the soil, is also utilized. When a solid object is encountered in the soil, it is then excavated to determine if the object is a mine. This is an extremely dangerous and time-consuming method of detecting landmines. Canines have also been used for landmine detection for decades [21]. Even though canines can be very effective for this purpose, their use is not without problems [22]. Logistical problems can be signiﬁcant. Dogs and handlers are expensive to train and maintain. Dogs (and handlers) do not perform well under all environmental conditions. In addition, the performance of the dog–handler team can vary from day to day, and performance at any given instant in time can be difﬁcult to verify. A single solution to the problem does not exist. Hence, new mine detection technologies are desperately needed. In an effort to address the problem of detection of low-metal-content mines, DARPA funded development of new sensor technologies that focused on detection of the single material unique to all landmines—the explosive. The Fido sensor was one of several explosives vapor detectors developed under this program. Unless speciﬁed otherwise, the data presented in this section was collected during ﬁeld tests of this sensor. 7.4.2

Discussion of Landmine Chemical Vapor Signatures

This section is a brief literature review of laboratory results obtained by a number of investigators studying landmine chemical signatures. These results are relevant to a discussion of ﬁeld experiences because they have proven to be extremely beneﬁcial in understanding the performance of the sensor and sampling system under ﬁeld conditions. In addition, many of the characteristics of landmine signatures documented in the laboratory investigations have been observed under ﬁeld conditions, adding a measure of validity to both the ﬁeld and lab results. 7.4.2.1 Chemical Composition of Landmine Signatures Almost 80% of the types of mines manufactured worldwide contain TNT, or mixtures of explosives containing TNT such as Composition-B [23]. Approximately 85% of the number of mines now deployed contain at least some TNT as a constituent of its explosive charge. Of these mines, approximately 86% contain at least 50 g of TNT [24]. Because of the relatively low volatility of TNT, it would take many decades for 50 g of TNT to completely sublime. Hence, the quantity of TNT found in most landmines is sufﬁcient to generate a chemical vapor signature that can be released into the soil for decades. Like canines, the Fido sensor is an extremely sensitive detector for TNT, making it ideally suited for detection of landmines. As discussed previously, military-grade TNT is not a pure material, containing additional nitroaromatic constituents at concentrations of up to several percent by mass [4]. Some of these explosive-related compounds (ERCs) have been shown to be signiﬁcant contributors to the chemical ﬁngerprint of a landmine [5, 7]. Of the ERCs found in TNT, those that are most prevalent in the headspace vapor of TNT include 2,4-DNT, 2,6-DNT, 1,3-dinitrobenzene (1,3-DNB), and 1,3,5-trinitrobenzene (1,3,5-TNB) [5].

160

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

Many of the trace contaminants found in TNT have higher equilibrium vapor concentrations than TNT, which has an equilibrium vapor concentration of 70 pg/mL of air at 298 K [18]. For example, analysis of TNT from a PMA1A1 . (Soviet Block antipersonnel landmine) found the headspace concentration of 2,4-DNT to be 20 times that of TNT even though the 2,4-DNT accounted for less than 1% of the explosive by mass [5]. 2,4-DNT is also more stable in the environment than TNT [3]. It has also been shown that 2,4-DNT is the compound most likely detected by canines trained to detect TNT [25, 26]. For these reasons, some investigators have singled out 2,4-DNT as the analyte of choice for vapor phase sensing of landmines. However, our laboratory analysis (gas chromatography using electron-capture detection) of acetonitrile extracts of soil samples collected from mineﬁelds around the world has shown that soils found to contain detectable levels of TNT do not always contain detectable levels of 2,4-DNT. As will be discussed in the next section, explosive and ERC-contaminated soils near mines provide the source material for production of the landmine chemical signature that can be detected by canines and the Fido sensor. Hence, we conclude that the best approach is not to focus on detection of a single analyte but on detection of both TNT and ERCs derived from it. Our ﬁeld results with Fido support this conclusion. The sensor can readily differentiate 2,4-DNT responses from responses to TNT. Depending on ﬁeld conditions, responses consistent with that of TNT are sometimes more prevalent, while at other times 2,4-DNT appears to be the more prevalent analyte. In addition to synthesis by-products, the environment in the vicinity of a landmine may contain additional nitroaromatic constituents that are derived from other nitroaromatic species as a consequence of chemical degradation. In the environment, nitroaromatic compounds are subject to various forms of degradation. For example, in addition to being a contaminant in TNT, 1,3,5-TNB trinitrobenzene is a photochemical degradation product of TNT. 2-Amino-4,6dinitrotoluene, 2-ADNT, and 4-amino-2,6-dinitrotolune, 4ADNT, are microbial degradation products of TNT [3]. The amino-DNTs are frequently found in the soil over landmines, particularly during warm weather when microbial activity is high. None of these products are common in the environment. Fido is capable of detecting all of these compounds. Hence, detection of any of these compounds could be used to signal the presence of a landmine. 7.4.2.2 Factors Affecting the Release of Explosive into the Soil There are multiple sources for release of explosive signatures into the soil from a mine. The ﬁrst source is surface contamination that is on the outside of the mine casing at the time of emplacement. Mine casings are likely to be heavily contaminated on their outer surfaces at the time of burial. This contamination could have been placed on the exterior of the mine at the time of manufacture, or could be due to contamination from other sources during storage or handling of other explosives or explosive devices. This source of contamination is likely to be 1

Details of the mines mentioned are included in [23].

LANDMINE DETECTION

161

quickly dispersed into the soil matrix surrounding the mine once it is buried. This type of surface contamination on a plastic-cased Yugoslavian PMA-1A AP mine was measured and reported in the literature [5]. The surface of the mine was contaminated with 1,3-DNB, 2,4-DNT, and TNT. The total mass of these analytes distributed over the entire surface of the mine was 6.5 μg. This mass of analyte is insufﬁcient to sustain long-term release of signature compounds into the soil. This level of contamination is, however, possibly sufﬁcient to facilitate detection of recently buried mines for a short time after their burial until this source is depleted. The remaining mechanism for explosive release into the environment is via vapor phase diffusion of analyte through mine structural materials and by leakage through cracks, seams, and holes in the mine. Because of the low vapor pressure of the explosives, and because the quantity of explosive in most mines is large, these are both long-term release mechanisms. The total vapor-phase ﬂux of signature compounds from several types of landmine cases was also measured as part of the mine surface contamination study described in the previous paragraph [5]. The ﬂux in this case was the total ﬂux due to diffusion through structural materials and leakage through cracks and holes. For the PMA-1A landmine encased in polyvinyl chloride (PVC), the vapor-phase ﬂux of 2,4-DNT into air was 3.4 μg per mine per day at 296 K, while the ﬂux for the less volatile TNT was only 0.3 μg per mine per day. Flux rates were also measured into water. The ﬂux rate of 2,4-DNT from a polystyrene-cased PMA2 AP mine into water was a factor of 30 larger than into air, and for TNT was a factor of 400 larger. It was estimated that the ﬂux rate into wet soil would be intermediate between the air and water values. This has important implications for sensing mines since soil moisture content will likely inﬂuence the rate of release of ERC from mines. The vapor-phase ﬂux through a mine case also varies depending on the type of material the mine is constructed from and the structural design of the mine [5, 7]. Depending on the type of plastic used, ﬂux rates from plastic-cased mines are typically higher than from metal-cased mines. For a metal-cased Yugoslavian TMM-1 Soviet Block antitank AT mine, the ﬂux rate for 2,4-DNT was 2.3 μg per mine per day at 296 K. This compares with 3.4 μg per mine per day for the plastic-cased PMA-1A whose total surface area is approximately 6.5 times smaller than that of the TMM-1. Release of ERC through intact metal-cased mines is presumably through seams, seals, and other nonmetallic structural materials. 7.4.2.3 Fate and Transport of Signature Compounds in Soil In order for vapor detectors to detect buried landmines, there must be vapors of explosive or ERCs present in the air over mines. The source for this vapor is from soils at the surface of the ground that have become contaminated with explosives. This implies that once explosive is released from a mine casing into the soil, there are mechanisms that facilitate transport of the explosive from the surface of the mine casing through the soil, and onto the soil at the surface of the ground. The fate and transport of ERC released into the soil from buried landmines has been the focus of a number of recent experimental and theoretical studies [2, 5–7].

162

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

These studies have resulted in a better understanding of the conditions that are necessary to produce vapor concentrations of ERC at the surface of the ground sufﬁcient for mine detection using our sensor. Figure 7.3 illustrates some of the processes that occur in the soil near a buried landmine. Signature compounds escape from the mine casing and are quickly adsorbed onto soil solids or dissolve into soil water. Because explosives and ERCs have a high afﬁnity for sorption onto surfaces [27], a large fraction of these compounds quickly sorb onto soil solid phases. The remainder of the explosive partitions into either soil liquid or gas phases. The fraction of the signature partitioning into these three soil phases is primarily a function of soil moisture content, followed by temperature and soil type [5, 7]. For a typical soil, approximately 95% of the total mass of ERC is adsorbed onto soil solids, followed by approximately 5% into soil water, with a trace (approximately 1 × 10−6 %) partitioning into the vapor phase [2]. Hence, the concentration of ERCs in the soil is orders of magnitude greater than the concentration of ERCs in the vapor phase. Some of the ERC sorbed by soil solid phases is irreversibly bound, preventing it from reaching the surface of the ground. Some of the material is degraded by the action of microbes in the soil [3]. The half-life of TNT in certain soils when the conditions for microbial degradation are favorable is short, on the order of a few days [28], while 2,4-DNT is much more stable in the environment. Some of these degradation products (namely amino-dinitrotoluenes, ADNTs) are detectable by Fido, but some interact more strongly with the soil than the parent materials, further reducing the concentration of signature available for detection. Explosive-related compounds are transported to the surface of the soil primarily through the movement of water in the soil. Molecular diffusion plays only a

Chemical Detection System Precipitation

nsp potr a

Microbial / Chemical / Physical Transformation or Loss

Eva

ion

fus

Dif

iratio

n

Photodegradation

n

sio

fu Dif

Liquid Phase

Vapor Phase

Soil Solid Phases

Land Mine

Figure 7.3 Processes affecting the fate and transport of landmine signature compounds in soil.

LANDMINE DETECTION

163

minor role in movement of ERCs through the soil column [2, 6]. During periods of precipitation, water tends to carry the signature downward through the soil, or transports the signature at the surface of the ground away from the mine in surface runoff. When precipitation ends, water evaporating from the surface of the ground tends to transport contaminated subsurface water upward toward the surface of the ground (evapotranspiration). As the water evaporates, it deposits signature compounds onto soil at the surface of the ground. This mechanism also describes transport of low-volatility pesticides through soil [29]. The chemical properties of pesticides and herbicides are similar to explosives, so it is not surprising that the same transport mechanisms are involved. Once the explosive signature molecules are deposited on surface soils, they can escape into the boundary layer of air near the ground over the mine. The concentration of ERCs released into the air depends on the solid-to-vapor partition coefﬁcient, which is again primarily a function of soil moisture content, followed by temperature and soil type. Explosives sorbed on surface soils are also subject to photochemical degradation when exposed to sunlight. Some of the photodegradation products (such as 1,3,5-TNB) are Fido detectable. Due to the complexity of the vapor signature of military-grade TNT, and because of the microbial and photochemical degradation products that result from degradation of signature compounds, it is not surprising that 20 different ERCs have been detected in surface soil samples collected over landmines [6, 7]. Most of these 20 compounds are infrequently detected in land-mine-contaminated soils. Four of these compounds (2,4-DNT, TNT, 4-amino-2,6-DNT, and 2-amino-4,6-DNT) are found much more frequently than the other compounds. Soil moisture content is an important parameter affecting the concentration of ERCs in the headspace over explosive-contaminated soils. Lab experiments have shown that for a given soil contaminated with ERC, the concentration of ERC in the headspace over the soil can span ﬁve orders of magnitude as a function of soil moisture content [7]. Hence, the water content of surface soils is a very important consideration when sampling vapor-phase ERC to locate landmines. The concentration of TNT in the boundary layer of air over a landmine has been estimated at 3 to 6 orders of magnitude below its equilibrium vapor concentration [2, 6]. This places the concentration of TNT in the air over a landmine in the parts-per-trillion (ppt) to parts-per-quadrillion (ppq) range. In terms of mass of analyte per milliliter of air, the concentrations range from femtograms (10−15 g) to low attograms (10−18 g) per milliliter of air. More volatile constituents such as 2,4-DNT would be present at slightly higher concentrations. Nevertheless, vapor detectors must be extremely sensitive in order to detect these low concentrations of signature compounds. The distribution of signature compounds in the environment can be highly heterogeneous, often with small areas of relatively high contamination dispersed among a larger area of little measurable contamination. A surprising ﬁnding is that the mine chemical signature is not necessarily strongest directly over the mine. Further, the mine signature often extends past the perimeter of the mine, up to several meters from the mine center. The direction of dispersal of signature

164

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

has also been observed to be nonsymmetrical and can be inﬂuenced by factors such as the topography of the soil near the mine [30]. Consequently, it is extremely difﬁcult to pinpoint the exact location of the mine using trace chemical detection methods alone. This is analogous to the current use of dogs for mine detection. Virtually all mine dogs are trained to alert when they have the landmine’s scent, not when they are directly on top of the mine [31]. Nevertheless, evidence currently available indicates that it may be possible to isolate a mine location to within a small, well-deﬁned area. 7.4.3

Landmine Detection Field Test Results

7.4.3.1 Results of Sensor Blind Testing Blind ﬁeld testing of Fido against buried landmines has been conducted on several occasions in a test mineﬁeld at Ft. Leonard Wood, Missouri. Blind test lanes (one containing TMA-5 Soviet Block antitank mines and the other containing PMA-1A antipersonnel mines) were established by marking potential target positions in the test ﬁeld. At each test position, two surveying ﬂags were placed approximately 50 cm apart. At some of the test positions, mines were buried, centered between the ﬂags. Other test positions contained no mines. The mines used for the tests were authentic but had the detonators removed and the fuze wells capped with shipping plugs. Fido was used to sample between the ﬂags at each test location. The sensor responses at each location were recorded and submitted to DARPA for scoring. The best performance recorded at this facility was in the antitank mine lane, with a probability of detection (PD ) of 0.90, with a probability of false alarm (PFA ) of 0.10. Tests of this type have been conducted on numerous occasions, with sensor performance ranging from poor to excellent. There is a high degree of correlation between sensor performance and environmental conditions, which have been shown to strongly inﬂuence the strength of landmine chemical vapor signatures. Warm to hot temperatures with damp soil conditions and light winds appear to be most favorable, while cold temperatures, with dry soil conditions and high winds appear to be least favorable. Conversations with dog handlers (at the site and elsewhere) conﬁrmed that canines tend to perform well (or poorly) under similar operational conditions. 7.4.3.2 Landmine Detection Performance Comparison to Canines On one occasion, the sensor was tested in conjunction with two experienced teams of canines. One dog was trained to detect explosives, and the other was an experienced mine detection dog. The team trained to detect explosives (i.e., bombs) withdrew from the test because of extreme difﬁculty in locating the mines. The team of canines with actual landmine detection experience in Bosnia and Mozambique performed much better. The ﬁeld conditions at the time of the test were very hot with dry soil, conditions under which the dog handlers stated that canine performance was typically poor. Comparisons of Fido performance were made to that of the dogs. Probabilities of detection and false alarm rates were in general slightly better for the sensor than that of the experienced canine landmine

LANDMINE DETECTION

165

detection team. While the testing was limited in scope, the performance of the sensor was very promising. A more detailed account of this testing has been published elsewhere [7]. To the knowledge of the author, this is the ﬁrst time a chemical vapor “sniffer” had demonstrated the ability to detect landmines under ﬁeld conditions with performance comparable to that of canines. 7.4.3.3 Inﬂuence of Environmental Factors on Detection In general, the performance of the sensor correlates well with what are now understood to be the most relevant environmental factors that affect landmine vapor signature concentrations. In general, the sensor performs better when conditions were warm to hot, with the soil in a drying phase after rain had fallen. This is consistent with the results of the fate and transport simulations conducted by Phelan and ebb [2] and the experimental work of Jenkins et al. [5–7] During one ﬁeld test, a subset of mines at the Ft. Leonard Wood mineﬁeld were sampled at 10:00, 14:00, and 16:00 on three consecutive days in order to evaluate the variability in mine signatures as a function of changing environmental conditions throughout the day. For these tests, a high-volume sampling method known as Remote Explosive Scent Tracing (REST) was utilized. The REST method is derived from the MECHEM Explosive and Drug Detection System (MEDDS), used successfully for almost two decades by MECHEM, a South African demining company. MEDDS has been used to detect roadside improvised explosive devices (IEDs), weapons caches, landmines, and various types of contraband. Using this methodology, an area suspected of being mined is sampled by drawing large volumes of air and entrained soil particulates from a suspect area through a specially designed ﬁlter designed to trap vapors of explosives and soil particulates. Portable, highvolume air pumps are used to draw air through the ﬁlters. After collecting a sample on the ﬁlter, which is inexpensive and disposable, the ﬁlter is presented to highly trained dogs for analysis. These dogs are trained to detect traces of explosive that may have been collected on the ﬁlter during sampling of a suspect area. Standard REST ﬁlters are not compatible with electronic sensors, so a Fido-compatible version of the ﬁlter, shown in Figure 7.4, was developed. The ﬁlter is also compatible for use with canines. Because canines were not available, samples were analyzed only with the Fido sensor. On day 1, samples were collected at 10:00 and analyzed. Soil conditions were relatively dry, resulting in nonideal detection conditions. Only 60% of the mines were detected. Prior to the scheduled 14:00 sampling on day 1, heavy rainfall fell, resulting in no samples being collected the remainder of day 1. At 10:00 on day 2, sample collection was resumed. The percent of mines detected had fallen to 20%, consistent with the signature being carried into the soil by the heavy rainfall. Days 2 and 3 were sunny with highs in the upper 80s, resulting in rapid drying of the soil. As can be seen from the data in Figure 7.5, the percentage of mines detected rose steadily from a low of 20% detected at 10:00 on day 2 to 90% detected at 18:00 on day 3. This ﬁnding is consistent with an increase in the concentration of ERCs in surface soils due to evapotranspiration, resulting

166

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

Flow (Air)

Packed Bed of Beads

Trapped Analyte

Flow (Air + Analyte)

Figure 7.4 Nomadics version of the REST high-volume sampling ﬁlter.

100 Day 1

Day 2

Day 3

20 10

18:00

15:00

30

10:00

40

18:00

50

14:00

60

10:00

Percent Detected

70

10:00

80

Heavy Rain. No Samples Collected

90

0

Figure 7.5 Effect of changing environmental conditions on mine detection.

in enhanced transport of the mine signature to the surface of the ground. The surface soil remained moist during days 2 and 3 of the test, resulting in favorable conditions for release of ERCs from surface soils into the vapor headspace over the mines. The results of this test are consistent with the explosive fate and transport results presented in Sections 7.4.2.3, 4.3.1, and 4.3.2 of Chapter 4. Our experience suggests that during periods of low precipitation when soil conditions are dry, sensor performance was typically better early in the morning, presumably due to the fact that surface soils contain a higher level of moisture early in the morning. As the temperature rises during the day, the soil presumably dries out, resulting in a drop in headspace ERC concentrations, thereby reducing

LANDMINE DETECTION

167

performance of the sensor. Canines seemed to perform better during the early morning as well. 7.4.3.4 Distribution of Mine Signatures in the Vicinity of Mines Another interesting result that was observed consistently during ﬁeld tests was that mine signatures are not necessarily the strongest directly over the top of a mine. This counterintuitive result was especially true for large antitank mines. The signature was consistently stronger around the perimeter of the antitank mines, forming a “halo” around the mine. It was also observed that when a mine was buried on a slope, the signature was consistently stronger on the downhill side of the slope. This was presumably due to the runoff of surface water carrying the signature in a downhill direction. Mine signatures were heterogeneously distributed in the soil near mines. Areas of contamination resulting in large Fido responses were often very localized, and not necessarily over the top of the mine. Moving the sensor a few centimeters in any direction could change the sensor response dramatically. All of these ﬁndings are consistent with the results of laboratory analysis of soil samples taken from these sites during testing [30]. During one ﬁeld exercise, when mine detection dogs indicated the presence of a mine, the location at which the dog indicated was marked. The sensor was then used to search for the mine. The sensor consistently detected the strongest mine signatures in roughly the same area as the dog. Interestingly, the location of strongest response was not always directly over the mine. The strongest response was, however, usually within a meter of the mine position. 7.4.3.5 Investigation of Area Reduction Methods As pointed out in the previous section, over time mine signatures seem to spread over a fairly large area extending outward from the mine. As previously mentioned, this makes it difﬁcult to use trace chemical sensors to pinpoint the exact location of a mine. A study funded by the U.S. Army Humanitarian Demining Program at NVESD (Ft. Belvoir, Virginia) enabled an investigation of the extent of spread of contamination from mines. The study involved Nomadics and MECHEM, and utilized the high-volume sampling (REST) method previously described in Section 7.4.3.3 [32–35]. REST samples were anlayzed by MECHEM’s canines and were also analyzed by Fido. Three high-volume samples were collected at 3-, 7-, and 11-m radii from the center of four different mine types (metal AT, metal AP, two plastic AP) at three different depths (10, 15, and 20 cm). These were real mines with detonators installed but rendered inert. In tests conducted during the spring of 2003, 82% of samples from the test area were found to contain explosive traces. Explosive signatures were detected out to 11 m from the mine center. Tests were not conducted at larger distances due to the layout of the test ﬁeld. Data supporting the conclusion that mine chemical signatures do not remain localized in the immediate vicinity of mines were collected at an arid test facility in the United States. In this test, multiple soil samples were collected along a line that ran perpendicular to the length of a mine lane. This line was selected so that

168

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

Sample Position (Feet From Mine Center)

Sensor Response

−12 −10 −8 −6 −4 −2 2 4 6 8 10 12

0 0 0 0 1 3 1 0 0 0 0 0

Figure 7.6

Sensor responses for samples collected perpendicular to mine lane.

it crossed over the top of a mine in the lane. Twelve samples were taken at 2, 4, 6, 8, 10, and 12 ft north and 2, 4, 6, 8, 10, and 12 ft south from the mine. Solvent extracts of the soil samples were analyzed by gas chromatography using the Fido sensor as a detector. Explosives were detected in only four of the samples. TNT and 2,4-DNT were found in samples at 2 ft south (refer to Fig. 7.6). TNT was recorded in samples taken at 4 ft south and 2 ft north. Hence, samples collected at points along the line on both sides of the lane (but outside of the lane) were blank, but explosive signatures were detected in the samples taken from inside the mine lane. The data indicate that the detection radius of Fido around this mine was ∼1.75 m ± 1.0 m (with 95% conﬁdence). These test results suggest that while it may be difﬁcult to pinpoint the exact location of mines using vapor detection, it may be feasible to detect areas that are mined. This detection paradigm differs from most in use today that rely on various mine-detecting technologies to locate the exact position of targets. Approaches that seek to locate exact target positions require that every square meter of an area be screened for individual mines. In contrast, an area reduction technique could be used to determine whether or not dispersed explosive signatures are present in an area, indicating that a mineﬁeld may be present. Once an area is deemed suspect, technologies that are good at pinpointing the exact location of the mines can be used to facilitate removal of the mines. If successful, this will enable demining efforts to focus on areas that actually contain mines, rather than consuming signiﬁcant resources in areas that are not mined. This need for rapid area reduction has been identiﬁed by humanitarian demining organizations as a top priority. 7.4.3.6 Landmine Soil Probe Test Results The concentration of TNT and other ERCs are often many orders of magnitude higher in the soil near a landmine than in the air immediately above the mine. In the soil, the concentration of landmine chemical signature compounds is often highest in the soil that is directly adjacent to the sides, top, and bottom of the mine casing [6, 30]. See Figure 4.13,

LANDMINE DETECTION

169

p88. However, this soil is not easily accessible for vapor sampling analyses because of the overlying covering of soil in which the mine is buried. Hence, a prototype soil-sampling probe has been designed to sample TNT and ERCs in the subsurface soil near a mine where the concentrations are likely to be the highest. The ﬁrst test of the probe was conducted at Fort Leonard Wood, Missouri, in May 2003. The results of this test were very encouraging, which led to further testing of the probe design at Fort AP VA Hill. The probe at this stage of development consists of a rigid, tapered stainless steel tube that can be inserted into the soil near a mine. The tip of the probe is perforated to allow vapor to diffuse from the soil into the interior of the probe. The probe is designed to accommodate a standard solid-phase microextraction (SPME) sampler that is deployed inside the probe during sampling [36]. The functional portion of the SPME sampler is a fused silica ﬁber coated with a material that preferentially binds semivolatile organic compounds such as TNT. The sorbent coating is exposed to sample vapors that diffuse or are pumped into the tip of the probe through perforations in the probe tip. As vapors of explosives diffuse through the perforations in the tip of the probe, the coating on the SPME ﬁber sorbs (through adsorptive and absorptive mechanisms) constituents of the vapor sample. Because of the high afﬁnity of the SPME coating for target analytes, the SPME ﬁber serves as a vapor preconcentrator. The mass of TNT sorbed is proportional to the amount of time the ﬁber is exposed to sample, enabling accumulation of sufﬁcient analyte for detection even when the concentration of vapor-phase analyte is very low. To collect a sample, the probe with a SPME ﬁber installed is inserted into the soil. Air is pumped through the probe, drawing subsurface soil vapors into the probe tip and across the SPME ﬁber. Pumping air across the ﬁber increases uptake of target analytes by the SPME ﬁber relative to what is collected by molecular diffusion alone. Once a sample is collected, the SPME ﬁber is removed from the probe for analysis. To analyze the sample, the SPME ﬁber is inserted into a modiﬁed inlet system attached to the Fido sensor. The modiﬁed inlet serves to heat the SPME ﬁber, causing rapid and quantitative desorption of trapped molecules of analyte. The vapor-phase analyte is then swept into the sensor for analysis by a ﬂow of carrier gas. Figure 7.7 is a comparison of the response of Fido to a high-volume vapor (REST) sample and a sample collected with the soil vapor probe. The samples were both collected near a TM-62P3 landmine. The REST sample generated a weak sensor response, but the sample collected with the soil probe resulted in a strong Fido response. This data illustrates the improvement in sensor response that can be achieved with the soil probe. More importantly, all mines sampled with the probe (a total of four) were detected. Obviously, there are operational issues with using such an approach for detecting mines, speciﬁcally the danger associated with inserting probes into the ground near mines. However, a soil probe could be mounted on a robotic platform equipped with ground penetrating radar (GPR). The GPR could be used to locate a suspicious target. Once located, the GPR could be used to guide insertion of the probe into the soil near the target for sampling. This method could allow

170

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

High-Volume Vapor (HVV) Sample (0.55% Quench)

1.02 Normalized Response

1 0.98 0.96 0.94 0.92 Soil Probe Sample (13.3% Quench)

0.9 0.88 0.86 0

5

10

15 Time (s)

20

25

30

Figure 7.7 Comparison of Fido responses to samples from the soil probe and MEDDS samples of the same landmine.

for enhanced detection capability without unnecessarily endangering demining personnel. 7.5 COMPARISON OF FIDO WITH CANINES USING HIGH-VOLUME SAMPLING METHODS (REST)

A comparison of the performance of the Fido sensor to MECHEM MEDDS canines was performed at the MECHEM MEDDS facility in Pretoria, South Africa, in February, 2003. These tests were conducted using the Nomadics version of the REST ﬁlter previously described in Section 7.4.3.3. At the time of testing, the MECHEM canines had been trained on the Nomadics ﬁlter for a period of approximately 4 months. Two dogs, with previous REST experience, were retrained to detect explosives on the Nomadics ﬁlters. Positive samples were prepared by placing TNT vapor strips (ﬁlter paper coated with a few milligrams of TNT) inside cardboard boxes. Once the vapor strips were placed into a box, the box was sealed using shipping tape. After allowing the vapor to accumulate inside the box for a measured amount of time (typically 30 to 120 min), the top of the box was punctured to allow insertion of a sampling probe into the box. A sample was drawn onto a Nomadics ﬁlter inserted in the inlet of the sampling probe. Air was drawn through the ﬁlter at a rate of 1 L per second. Samples were collected for varying time intervals ranging from 30 to 120 s. After sampling was completed, the ﬁlter was inserted into its plastic storage container until analysis. The concentration of TNT vapor in the box was not determined. However, by increasing or reducing the number of vapor strips in the box, by increasing or reducing the “soak” time prior to sampling, or by varying the sampling time, it was possible to vary the quantity of TNT trapped in the ﬁlters.

COMPARISON OF FIDO WITH CANINES USING HIGH-VOLUME SAMPLING METHODS (REST)

171

Samples of potential chemical interferences were also prepared by placing samples of selected materials inside clean cardboard boxes. The potential “interferent” items were materials commonly encountered in demining operations such as petrol. These materials were sampled in an identical fashion as the positive samples. Blank samples were also prepared by sampling empty cardboard boxes. All samples were marked by sampling personnel in a manner that made it impossible for analysts to determine the composition of the sample during analysis. Nomadics personnel and dog handlers were not given any information on sample identity until analysis of samples was completed and results were submitted for scoring (i.e., the tests were conducted in a “blind” fashion). After samples were prepared, they were split into 2 separate batches, with each batch containing positive, blank, and interferent samples. The number of positive, blank, and interferent samples per batch was not revealed to the dog handlers or sensor operators. All samples were ﬁrst analyzed by the canines. After the canine analysis was completed, Fido was then used to analyze the same batch of samples. Batch 1 of ﬁlters contained a total of 25 samples, 4 of which were positive. Both Fido and the canines detected 3 of the 4 positive samples. Interestingly, the sensor and the dogs missed the same positive sample. All samples from batch 1 were analyzed at room temperature. In the second batch of samples, 3 samples out of 24 were positive. Fido and the canines detected all 3 positive samples. Prior to presentation of samples in the second batch to Fido, the samples were heated slightly to enhance the vapor-phase concentration of target analytes in the samples. As would be expected, responses to the positive samples that were heated were stronger than the samples analyzed at room temperature. The performance of Fido and the canines against interferents was also identical. The interferent samples were selected from materials that would be routinely encountered during a demining operation. Of the 20 potential interferents included in the test, Fido and the canines responded to the same interferents, giving responses to 2 of the 20 interferents. Although the sensor responded to 2 test interferents, the sensor response to these 2 interferents was not consistent with that of TNT or ERCs. For the 2 materials that false indications were registered (none of which were pure materials), it is not known which of the chemical compounds in the samples was causing the sensor and the dogs to alarm, nor was it known if they were responding to the same compound. Nevertheless, the response to interferences is interesting and should be studied further. The results of this comparison were promising. The performance of the sensor during this series of tests was comparable to that of the canines. One outcome of these tests was the notion that the Fido sensor could possibly be used as a canine training tool. For example, when positive samples are prepared there is currently no easy way to determine if the samples are actually positive. The sample that was missed by the canines and by Fido was prepared in exactly the same manner as the three samples that were detected, yet this sample was not detected. If the sample in question were used as a positive sample during training, but were actually blank, confusion of the dog could occur, reducing the effectiveness of the training session. In addition, a properly designed electronic sensor should

172

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

exhibit reproducible and quantiﬁable levels of performance from day to day. The performance of canines can vary for a variety of reasons, and it can be difﬁcult to determine when a dog is not performing at its best. The sensor could possibly be used to help verify the performance of canines. This is not to say that the performance of Fido is presently adequate to replace dogs in certain roles, but it may have a role in enhancing and complimenting the performance of dogs. Dogs can be trained to detect a wide range of substances in a short period of time. In this respect, dogs could be used to cover critical gaps in sensing applications while sensory materials for detection of other analytes of interest are developed. For sensing tasks in which Fido and dogs could both be used, the sensor could be used in a conﬁrmatory sensing role to improve the performance of both methods. 7.6

CONCLUSIONS

The ﬁeld experiences presented here represent a cross section of lessons learned during ﬁeldwork performed over a period of approximately 8 years. While laboratory testing of explosives detection equipment is necessary, laboratory testing cannot substitute for testing under realistic ﬁeld conditions. In addition, the phenomenology associated with chemical signatures of explosive devices is complex and cannot always be studied in a laboratory setting without signiﬁcant risk of inadvertently altering the characteristics of the signature. ACKNOWLEDGMENTS

We wish to thank the Defense Advanced Research Projects Agency and the U.S. Army Night Vision and Electronic Sensors Directorate for funding the work presented here and for providing technical and programmatic guidance during the execution of this work. We also wish to thank Professor Tim Swager at MIT for his contribution in the form of development of sensing materials for the Fido sensor.

REFERENCES 1. Cumming, C., C. Aker, M. Fisher, M. Fox, M. la Grone, D. Reust, M. Rockley, T. Swager, E. Towers, and V. Williams. Using novel ﬂuorescent polymers as sensory materials for above-ground sensing of chemical signature compounds emanating from buried landmines. IEEE Trans. Geosci. Remote Sensing 39(6), 1119–1128 (2001). 2. Phelan, J. M. and S. W. Webb. Environmental Fate and Transport of Chemical Signatures from Buried Landmines—Screening Model Formulation and Initial Simulations. Sandia Report SAND97-1426, Sandia National Laboratories, Albuquerque, NM June 1997. 3. Walsh, M. E. and T. F. Jenkins. Identiﬁcation of TNT Transformation Products in Soil. SR92-16, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, Hanover, NH 1992.

REFERENCES

173

4. Leggett, D. C., T. F. Jenkins, and R. P Murmann. Composition of Vapors Evolved from Military TNT as Inﬂuenced by Temperature, Solid Composition, Age, and Source. SR 77-16/AD A040632, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, Hanover, NH, 1977. 5. George, V., T. F. Jenkins, D. C. Leggett, J. H. Cragin, J. Phelan, J. Oxley, and J. Pennington. “Progress on Determining the Vapor Signature of a Buried Landmine”, Proc. SPIE, Detection and Remediation Technologies for Mines and Minelike Targets IV, vol. 3710, part 2, p258, 1999. 6. Jenkins, T. F., M. E. Walsh, P. H. Miyares, J. A. Kopczynski, T. A. Ranney, V. George, J. C. Pennington, and T. E. Berry. Analysis of Explosives-Related Signature Chemicals in Soil Samples Collected Near Buried Landmines. ERDC Technical Report, Cold Regions Research and Engineering Laboratory, Hanover, NH, 2000. 7. George, V., T. F. Jenkins, J. M. Phelan, D. C. Leggett, J. Oxley, S. W. Webb, P. H. Miyares, J. H. Cragin, J. Smith, and T. E. Berry. Progress on determining the vapor signature of a buried landmine. Proc. SPIE, Detection and Remediation Technologies for Mines and Minelike Targets V, 2000, Volume 4038, part 1, p. 290 (2000), Orlando, FL. 8. Cumming, C., C. Aker, M. Fisher, M. Fox, M. la Grone, D. Reust, M. Rockley, T. Swager, E. Towers, and V. Williams. Using novel ﬂuorescent polymers as sensory materials for above-ground sensing of chemical signature compounds emanating from buried landmines, in Proceedings of the UXO/Countermine Forum, New Orleans, Louisiana, April, 2001. 9. la Grone, M., C. Cumming, C. Aker, M. Fisher, M. Fox, D. Reust, M. Rockley, T. Swager, E. Towers, and V. Williams. Developments in the use of an ampliﬁed ﬂuorescent polymer-based sensor for the detection of landmines, in Proceedings of Aerosense 2001, Volume 4394-111, International Society for Optical Engineering, Orlando, Florida, April, 2001. 10. Fisher, M. and C. Cumming. Detection of trace concentrations of vapor phase nitroaromatic explosives by ﬂuorescence quenching of novel polymer materials, in Proceedings of 7th International Symposium on the Analysis and Detection of Explosives, Defense Evaluation and Research Agency, Edinburgh, Scotland, UK, June, 2001. 11. Fisher, M. and C. Cumming. Trace detection of nitroaromatic explosives by ﬂuorescence quenching of novel polymer materials, in Proceedings of the U.S. Federal Aviation Administration’s Third International Aviation Security Technology Symposium, Atlantic City, NJ, November 27–30, 2001. 12. Fisher, M. and C. Cumming. Utilization of novel ﬂuorescent polymer materials for trace level vapor-phase detection of nitroaromatic explosives, in Proceedings of the U.S. Federal Aviation Administration’s Third International Aviation Security Technology Symposium, Atlantic City, NJ, November 27–30, 2001. 13. Fisher, M., M. la Grone, and J. Sikes. Implementation of serial amplifying ﬂuorescent polymer arrays for enhanced chemical vapor sensing of landmines, in Proceedings of UXO/Countermine Forum 2003, Orlando, Florida, September 2003. 14. Yang, J. S. and T. M. Swager. Porous shape persistent ﬂuorescent polymer ﬁlms: An approach to TNT sensory materials. J. Am. Chem. Soc. 120, 5321–5322, 1998. 15. Yang, J. S. and T. M. Swager. Fluorescent porous polymer ﬁlms as TNT chemosensors: Electronic and structural effects. J. Am. Chem. Soc. 120, 11864–11873, 1998.

174

ULTRASENSITIVE ELECTRONIC VAPOR SENSORS, FIELD EXPERIENCE

16. Williams, V. and T. M. Swager. Iptycene-containing poly(aryleneethynylene)s. Macromolecules 33, 4069–4073 (2000). 17. Zhou, Q. and T. Swager. Methodology for enhancing the sensitivity of ﬂuorescent chemosensors: Energy migration in conjugated polymers. J. Am. Chem Soc. 117, 7017–7018 (1995). 18. Yinon, J. and S. Zitrin. Modern Methods and Applications in Analysis of Explosives. Wiley, New York, 1993. 19. Kirk-Othmer Encyclopedia of Chemical Technology, Vol. 9, 3rd ed. Wiley, New York, 1980. 20. U.S. Department of State. Hidden Killers: The Global Demining Crisis. U.S. Department of State Publication 190575, Washington, DC, 1998. 21. Field Engineering and Mine Warfare, Pamphlet No. 6, Detection and Clearance of Mines, British War Ofﬁce, London, UK 1947. 22. Nolan, R. V. and D. L. Gravitte. Mine Detecting Canines. Report 2217, U.S. Army Mobility Research and Development Command, Ft Belvoir, VA 1977. 23. King, C., Ed. Jane’s Mines and Mine Clearance, 4th ed., Jane’s Information Group, Surrey, UK. 1999–2000. 24. Mine Facts. CD-ROM, v. 1.2, U.S. Department of Defense, Washington, DC, 1997. 25. Harper, R. J., J. R. Almirall, and K. G. Furton. Identiﬁcation of dominant odor chemicals emanating from explosives for use in developing optimal training aid combinations and mimics for canine detection. Talanta 67, 313–327 (2005). 26. Johnston, J. M., M. Williams, L. P. Waggoner, C. C. Edge, R. E. Dugan, and S. F. Hallowell. Canine detection odor signatures for mine-related explosives. Proc. SPIE, Detection and Remediation Technologies for Mines and Minelike Targets III, 3392(1), 490–501 (1998). 27. Kenna, B. T. and F. J. Conrad. Studies of the Adsorption/Desorption Behavior of Explosive-Like Molecules. Sandia Report SAND86-0141, Sandia National Laboratories, Albuquerque, NM 1986. 28. Grant, C. L., T. F. Jenkins, and S. M. Golden. Experimental Assessment of Analytical Holding Times for Nitroaromatic and Nitramine Explosives in Soil. SR 93–11, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, Hanover, NH June 1993. 29. Spencer, W. F., M. M. Cliath, and S. R. Yates. Soil-pesticide interactions and their impact on the volatilization process, in Environmental Impact of Soil Component Interactions—Natural and Anthropogenic Organics, Vol. 1, CRC Press, Boca Raton, FL, 1995, pp. 371–381. 30. Hewitt, A. D., T. F. Jenkins, and T. A. Ranney. ERDC/CRREL TR-01-09, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, Hanover, NH May 2001. 31. Horwood, C. The Use of Dogs for Operations Related to Humanitarian Mine Clearance. Handicap International Mines Coordination Unit, Paris, France, 1998. 32. Fisher, M., M. Prather, and J. Sikes. Serial amplifying ﬂuorescent polymer arrays for enhanced chemical vapor sensing of landmines. EUDEM-2/SCOT Conference, Brussels, Belgium, September, 2003. 33. Fisher, M. and J. Sikes. Mineﬁeld edge detection using a novel chemical vapor sensing technique, in R. S. Harmon, J.H. Holloway, Jr., J.T. Broach, Eds. Detection

REFERENCES

175

and Remediation Technologies for Mines and Minelike Targets VIII, Proceedings of SPIE Vol. 5089, pp. 1078–1087, April 21–25, 2003, Orlando, FL. 34. la Grone, M. C. Cumming, M. Fisher, and E. Towers. Investigation of an area reduction method for suspected mineﬁelds using an ultra-sensitive chemical vapor detector, in Proceedings of Aerosense 2002, Volume 4742, International Society for Optical Engineering, Orlando, Florida, April, 2002. 35. Fisher, M. M. la Grone, C. Cumming, and E. Towers. Utilization of chemical vapor detection of explosives as a means of rapid mineﬁeld area reduction, in Proceedings of the 5th International Symposium on Technology and the Mine Problem, Mine Warfare Association, Monterrey, California, April 2002. 36. Pawliszyn, J. Solid Phase Microextraction—Theory and Practice. Wiley, New York, 1997.

CHAPTER 8

REFLECTIONS ON HUNTING MINES BY AROMA SENSING VERNON JOYNT Mechem Division of Denel, South Africa (retired)

8.1

EDITOR’S NOTE

This chapter is organized in the form of an interview with Dr. Vernon Joynt. Dr. Joynt has many years’ experience clearing mineﬁelds in southern Africa. In recent years he has brought his experience to bear in other areas, notably in the Balkans. He was a, perhaps better stated the, pioneer in using trace chemical sensing to locate mines in the ﬁeld. His techniques were developed with animals, ﬁrst dogs, later adding rats. The design of this interview is the highlighting of some of the practical considerations that only actual ﬁeld experience can teach. The following pages contain a very informal set of questions and answers, conducted over several months, mostly by e-mail.

8.2

INTERVIEW

Editor: You mentioned a workshop where you recently gave a lecture on a subject that sounds very much like the subject we want to include in the book. What was the most pertinent part of that workshop? VJ: The workshop that addressed the problem would have been quite interesting to you, as it was a shoot-out between chemical and physical test procedures and animal noses. I was asked to represent the noses. Editor: I would have found that discussion extremely interesting. I know you have compared the ability of animals to detect trace chemicals with the sensitivity of various chemical techniques. You have discussed this with Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

177

178

REFLECTIONS ON HUNTING MINES BY AROMA SENSING

many chemists. What is your personal assessment of the current comparison, and what trend do you see? VJ: To answer the question as a gut reaction just this: More and more biological detectors like bees, elephants, and others that I know less about are shown to be viable detectors. In all cases the best practical results are achieved by the addition of a scientiﬁc enhancement like an electronic eye to observe the bee’s proboscis ﬂick in and out, or the satellite thermal imager tracing elephants on the move. And then there are, of course, ﬁlters to help dogs and rats. So combination systems are the most useful. There are many more stories, but a key is to use combinations not shoot-outs. Animal sensors add a component of speed, sensitivity, and cost reduction to a system. The scientiﬁc instrument in turn enhances accuracy and credibility. Editor: You have, for quite some time, worked in the laboratory and the ﬁeld developing the MEDDS (Mechem1 Explosive and Drug Detection System) that has led to the REST (remote explosive sensing techniques) concept. What is the current state of development of these? VJ: In Africa I attended the REST workshop at Morogoro in Tanzania. This workshop was attended by top chemists from Nomadics2 (Dr. Mark Fischer), Sandia National Laboratories3 (Jim Phelan), and FOI4 (Lena Sarholm). The world REST fraternity was also well represented: The Geneva International Centre for Humanitarian Demining (GICHD), who was running the workshop under H˚avard Bach, APOPO5 under Christophe Cox, Mechem with Kip Schulz, NOKSH,6 and NPA7 with Rune Fjellanger. There was also an excellent summary of the workshop presentations issued by Dr. Ian McLean of the GICHD. At a large workshop of about 150 dog people in Ljubljana, Slovenia, some years ago, the case for mine detection dogs (MDDs) was debated. No consensus was reached. There were just too many opinions, stresses, and uncertainties. The work group at the GICHD started basically with the same problems but through the persistence of H˚avard Bach and others the problems became less. The old Mechem was part of that GICHD MDD effort and I pushed our MEDDS with a passion. It was only accepted once the South African government 1

MECHEM, a division of DENEL (Pty) Ltd (South Africa), was established in the late 1960s, originally as a research unit of the Council for Scientiﬁc and Industrial Research (CSIR) in South Africa, culminating into a division of DENEL (South Africa). 2 See Chapters 6 and 7. 3 See Chapter 4. 4 FOI, Totalf¨orsvarets forskningsinstitut, Sweden. 5 The APOPO project: African Giant Pouched Rats, SUA APOPO (prime contractor). 6 Norsk Kompetansesenter for Spesialsøkshund (NOKSH), Norway. 7 Norwegian Peoples’ Aid.

INTERVIEW

179

offered the use of the system to all humanitarian demining (HD) operations with no strings attached. This was done in Tokyo, during a United Nations meeting there, by the SA8 Foreign Affairs Department. The technology in this context was ﬁrst transferred to NPA who named it explosive vapor detection (EVD), and then later H˚avard Bach renamed and broadened the concept to REST. MEDDS was more scientiﬁcally controllable because the training and use of the dogs were separated from the ﬁeld and sampling problems. With MDD, setting up controlled tests was extremely difﬁcult when one is trying to do this in real operational demining conditions. With MEDDS this was easier and the results more acceptable. MEDDS resulted in our partaking in the DARPA9 Electronic nose project. This project delivered two signiﬁcant products: • •

The Nomadics/MIT FIDO®10 electronic nose The Quantum Magnetics NQR11 explosive detector

A NVESD12 contract over the past two years with the new Mechem and Nomadics attempted to repeat more scientiﬁcally what the original development and demining contracts had taught us about the system. They had some success. MEDDS taught us many things about MDD and their problems, but most important of all it taught us the problems of what happens to the explosives once they are out of the landmine and left to the environmental inﬂuences. A nightmare! MEDDS was often used to provide a dog-nose benchmark to a vapor sensor development or evaluation. Many unresolved arguments resulted as to the sensitivity levels that a dog’s nose can achieve. According to Jim Phelan and us, dogs in Mechem and Global13 (San Antonio) could do ppq (parts per quadrillion). This is detecting 10−15 grams of TNT [trinitrotoluene] or femtograms. A picogram is 10−12 grams, so dogs are at least 1000 times below classic chemistry, which can normally do picograms. Nomadics have, however, now cast some doubt on the way this ﬁgure was measured. And so it goes on. So it is fair to say that the whole ﬁeld of REST is improving and moving forward. Editor: Clearly, establishing what you call sensitivity, or others may call limit of detection, or LOD, is a very challenging task, one that is never ending as technology improves. In your opinion, what is the best way to quantitatively establish this LOD for any system? 8

South Africa. (U.S.) Defense Advanced Projects Agency. 10 See Chapter 9. 11 Nuclear quadrupole resonance, a bulk sensing technology. Quantuum Magnetics, Inc., San Diego, California. 12 U.S. Army Night Vision and Electronic Sensors Directorate, Ft. Belvoir, Virginia. 13 Global Training Academy, Somerset, Texas. 9

180

REFLECTIONS ON HUNTING MINES BY AROMA SENSING

VJ: My determination went back some years when the UK sent a team of forensic scientists under Dr. Dick Lacey and Tim Sheldon to come to South Africa in about 1988. My group was still in the CSIR14 of South Africa, and we had just started developing MEDDS, then called SOS, which in Afrikaans is Springstof Opsporing Stelsel, for tracing the smuggling of explosives into the country. For obvious reasons, the British police were interested. We hired them a chemical laboratory on the CSIR site, and they proceeded to prepare vapor samples using a technique of adsorptions onto polyethylene beads and a split-dilute-equilibrate-split again, etc., till they got down to vapor concentration levels of 10−14 grams RDX (cyclotrimethylenetrinitrzmine). This was the technique they had used to evaluate other chemical systems, like the Condor system. Our dogs almost broke their necks indicating the samples we had sucked onto ﬁlters from their measured source! (Exaggeration! All four dogs indicated, repeatedly, no errors.) I asked they drop the concentration. Equilibrations after further splits take exponentially longer. But, they went away, and after some days came back with a 10−15 grams RDX sample, which the dogs also found without error. However, from judgment of previous tests we had concluded with those dogs they would go tenfold down before some would start missing. Thus 10−16 grams is possible or even better with very carefully trained animals. Like with athletes, we found practice improves performance. Their ﬁnal conclusion was positive, and further visits took place in which operational facilities at country border stations of the SA Police and SA Defense Force were visited. Today some systems in the UK and France based on this original cooperation is still ongoing. Editor: In this set of experiments you used RDX. It is a more common practice to use TNT in such experiments. Do the MDDs distinguish between the two? VJ: The MEDDS dogs could never be taught to discriminate between RDX and TNT. This suited us ﬁne for mine clearance. In fact right at the onset in 1987 we had contracted a scientiﬁcally run dog school to train dogs for us that would discriminate between the various common explosives. After two years of trying they could only show the dogs were able to separate the nitro esters like PETN (pentzerythritoltetranitrate) and nitrocellulose on the one hand and organic nitro compounds like RDX and TNT on the other hand, but not between compounds inside a group. This made me conclude that the smell was a combination of the compound and its decomposition products. The nitro esters are decomposed when exposed to acids while the nitro explosives are decomposed by exposure to alkaline conditions. Maybe the alkaline nature of mucus in the mouth and nose produce the 14

Council of Scientiﬁc and Industrial Research, South Africa.

INTERVIEW

181

difference they can recognize. We only pursued that research for two years and it was inconclusive. Editor: Has there been an occasion in your experience where some electronic sensor has achieved an LOD comparable to those of the animals with which you have worked? VJ: Yes, some of my work included comparisons between dogs and chemical methods. I participated in an important one sponsored by NVESD. What makes this NVESD test important is that under some of those test conditions the Fido® actually gave more sensitive results than the dogs. My explanation, which was accepted by Prof. Vladimir Knapp of CROMAC15 who was also one of the scientiﬁc advisors on the project, was logical (says I!) but the conclusions were never or, as yet, have not been made public.∗ Editor: What is your explanation? VJ: The training of the Mechem MEDDS dogs in the test were rigidly controlled to ﬁnd only TNT and RDX, and none of the degradation products like the amino-dinitrobenzene products, while the Fido could ﬁnd that type of product. The life of these degradation products in wet soil was orders of magnitude longer than that of the TNT being exuded out of the test mines. The result was that the Fido could indicate the presence of the test mines at longer distances than the MEDDS dogs. In all probability, if the dogs were also trained on the other chemicals, they would have been at least as good, but the test records stand. It must also be said that the tests were typical of real-world conditions with storms and other factors spreading the chemicals around in the test site. Therefore, the comparisons were difﬁcult to scientiﬁcally duplicate, and the conclusion we made was only supported by percentage calculations. None of these were 90% or above. More like 65% for Fido and 35% for the dogs on speciﬁc mines in the test. Editor: You have found that the animals sense the chemicals in some ways that we don’t always anticipate. You mentioned vultures once. How do they ﬁt in? VJ: There are also some likely stories that vultures rely more on smell to alert them about a dead animal than their legendary eye sight that has encouraged some of us to ﬁt REST-based systems on RP16 helicopters to search for arms caches and mine ﬁelds. Editor: That brings two questions to mind. First, what success have you had with the RP helicopters? 15

Croatian Mine Action Center, Sisak, Croatia. Remotely piloted. ∗ Editor’s note: That comparison is discussed in Section 7.4.3, page 162.

16

182

REFLECTIONS ON HUNTING MINES BY AROMA SENSING

VJ: As yet we have only done wind tunnel tests of sampling and they look promising, no more. Editor: The other question brought to mind by the vulture reference concerns odor plumes. An entire, Chapter 5, of the book is devoted to the structure of plumes in air and water. Most animals, not just vultures, have an instinctive sense of how to follow a plume to its source. As you have used animals in the ﬁeld, have you observed any behaviors that seem to indicate plume following? How far have you seen animals effectively follow an explosive plume? VJ: Plumes work better in water. But in air, particulate sensing with dust and wind is more likely than vapor and diffusion. From the word go, Mechem and myself urged for the testing of vegetation instead of soils when searching for explosive vapors that originate out of the soil. In our mine clearance contracts in Mozambique and Angola, where we cleared more than 5000 miles of mined roads in UN contracts, we never failed a QA17 done behind us by manual deminers of another demining company, who had the QA contract from UN. Mechem’s own internal QA, however, found the MEDDS system missing mines. Then we would go back and look for the errors. Inevitably the misses would be on clean soil, mostly dry or when the sun was baking down. In all cases a retest on vegetation in the vicinity of the missed mine would indicate the explosive as positive, while a retest above the soil above the mine would still be negative. We changed our sampling to always include sucking ﬁlters in the vegetation on the roadsides. This worked well. What upset me, is that most of the scientiﬁc testing going on with dogs and instruments were on clean soil. Vegetation was not being included, because it was said that the problem then became too complex, and one step at a time was the way to go, before the vegetation steps would be brought to bear. What was good about the NVESD tests in Croatia was that vegetation was part of the test. When a local farmer accidentally cut a part of the test site’s long grass, we suddenly found no explosive near the mines; but when Mark Fisher, of Nomadics, took the Fido to the heap of cut grass, he had immediate positive tests in the cut grass. It is likely that dust particles get stuck to leaves and the TNT is protected there from both soil bacteria and intense sunlight. When one looks at studies on how the dog’s nose moves the dust particles [1] to the smell sensors with his nose hairs and mucus, then one realizes that the water in the mucus dissolves the TNT off the dust and offers it to the olfactory sensors. Fjellanger also quoted literature of how a dog’s nose uses side muscles to pulse air at a frequency of 3 to 5 puffs per second at the slits on the side of his nostrils when snifﬁng. The dog then sucks the stirred up dust with a long 17

Quality Assurance.

INTERVIEW

183

Figure 8.1 See color plates. Diagram of a dog’s snifﬁng passages. Substitute dust with TNT adsorbed for sausages! (Courtesy of Rune Fjellanger, NPA. Used by permission.)

suck through his main nostril into his nose. The olfactory sensors sit in the roof of the nose and the mucus moving past is ﬂushed down the back into his throat. Figure 8.1 is from Rune’s lecture in Tanzania. Editor: That seems to correlate with our observation that the molecules tend to adsorb on soil particles and dust, rather than remain in the air as explosive vapor. I take it that you mean that some wind is needed to move these particles in order to make them available above ground level to the animal. So, a pure vapor plume in air is not as rich a source as wind carried dust and soil particles. Is that correct? VJ: The story about the importance of vegetation seems to bear that out. Editor: In Chapter 4 of this book we examine the issue of leakage of explosive from mines. We attribute a comment to you about the resulting pattern of dispersal of the explosive molecules on the ground surface. Of course, the details of the local topography have a dominant effect, but you indicated that in nearly ﬂat areas you routinely notice a peculiar pattern. Please address this issue. VJ: The explosives leak and leach out of buried mines and then spread out into and onto the surrounding soil. How does it spread? Where does it ﬁnd itself after a time? What happens to it in different conditions? A million questions but the mere fact of MDD successes made it important to study this. This has been a bugbear all along, and in the early 1990s we tried to establish how the contours of detection around a buried mine, or mines, looked. The shape we established, I described later, could be as irregular as a camouﬂage pattern. Compare this with the lines determined during one NVESD test in Croatia recently (Fig. 8.2): The grid is in inches and feet. There were some deep ruts made by wheel marks along the service road through the site. This seems to have produced the 40-cm depth.

184

REFLECTIONS ON HUNTING MINES BY AROMA SENSING

Figure 8.2 Pattern of concentrated explosive residue, Croatia, Area Reduction Test Site, May/September 2002. Shaded areas represent standing and ﬂowing water to 40 cm deep (Courtesy of Mark Fisher, Nomadics, Inc. Used by permission.)

Although this represented standing water, tests done showed the TNT to be strongest in the water. This is to be expected because TNT is about a million times more soluble in water than in air. What is important is to note that when the water dried up the spread of TNT remained in that shape. This would explain the so-called camouﬂage patterns we had determined ten years before. The scientiﬁc tests and results obtained in the early 1990s by the old Mechem were never published and only recorded in broad detail. What was shared was the results and developed techniques. Editor: When you say “the old Mechem,” do you mean while it was part of CSIR? VJ: The term old and new Mechem refers to the split when the larger old Mechem in Denel had its R&D group split off and moved into the CSIR some four years ago. I then retired and was hired by CSIR to work with the 70 strong R&D group while the smaller new Mechem component that remained at Denel did mostly only mine clearance contracts for commercial gain. The NVESD contract started before the split but went on mostly after the split and I then had only an advisor role. CSIR did have an interest in MEDDS but not in mine clearance, which remained at new Mechem. CSIR was involved in using dogs and ﬁlters to do cancer detection in humans. Editor: This discussion brings to mind the question of how the animals sniff in the molecules. You have addressed this question from the point of view of ﬁeld practice more than physiology, which may help us understand how better to use electronic systems. You indicated that recent data from Norway18 indicates that the dogs more often make correct identiﬁcation of explosives when the humidity is low, but that the temperature seems to have no effect. How does this compare with the greater explosive vapor ﬂux rates measured after rain, as discussed in Chapter 4 and Chapter 7? VJ: We used to believe MEDDS worked better in humid conditions and avoided hot dry conditions. This came from our MDD experiences. Dogs smell better when the humidity is above 30%; but REST is sampled with a pump and ﬁlter that then is presented to a dog in a humidiﬁed room. 18

Norsk Kompetansesenter for Spesialsøkshund (NOKSH).

INTERVIEW

185

Editor: You used the MEDDS system to explore this in terms of the new ScentPrint REST ﬁlter system. Who developed this system and why is it an improvement? VJ: The CSIR R&D group developed the ScentPrint™ ﬁlter for the cancer work, but we then explored the possibility of using it more widely. New Mechem never adapted to its use. Mechem retained the dogs, as the CSIR ruled that they would not use dogs on their site; that is a registered game sanctuary. The ScentPrint ﬁlter is a two-stage system. It consists of a tubewithin-a-tube, about 3 cm in diameter by about 20 cm long. It comes with lids for each end that are removed for use. Air is drawn through the tubes to collect the sample. The inner tube is perforated to release the trapped vapor sample. The tube is also designed to house any vapor capturing material suited for the appropriate circumstance. The inner tube is a holding reservoir with strong absorption while the cotton wool-like wads have the correct surface adsorption properties. This is to hold enough transferred reservoir material ready on the surface of the wool for the hot moist animal breath to desorb the smell and get it into his nose. Editor: Your reference to “the appropriate circumstance” recalls that the system is used for vapors other than explosives. What are some of these? VJ: The CSIR, in collaboration with a U.S. company, is detecting cancer in humans by using dogs, and the ScentPrint two-stage ﬁlter system the CSIR has developed for multiple uses of REST. The U.S. company has already some trained dogs doing the cancer work and the addition of a ﬁlter step has shown an improvement. The CSIR is also involved with tuberculosis diagnosis but lacks an animal training company as a partner. One needs also to consider the APOPO rats and mention they are also using the rats to detect tuberculosis. So biological detection, or perhaps, the chemical signature produced by biological processes, is in the cards for REST. APOPO is also running a rat detection program to do TB diagnosis. For this they found a second-stage absorber that works better than the Mechem ﬁlter, to be polyester wool. The CSIR found that similar polyester wool coated with a silicone polymer was best for cancer detection’s second stage in the ScentPrint ﬁlter. Editor: So, do there seem to be materials that work better for speciﬁc vapors? VJ: Scientiﬁcally, the ﬁlter problem is that a good absorber is a good reservoir for large quantities of the active chemical or bouquet vapors. This seldom is a good adsorbent for releasing the ingredient to the animal when it smells. Good adsorbents often hold too little sample for a series of animals to make passes, and one then ﬁnds the second stage “fading.” So the second stage can have more than one “wool ﬁlter” to use in sequence of smelling. (Now you see why Jim Phelan has coined the word “sorption” so one need not specify!)

186

REFLECTIONS ON HUNTING MINES BY AROMA SENSING

Figure 8.3

Fixture for comparing ﬁlter materials using parallel adsorption ﬁlters.

APOPO uses a polyester ﬁber wad for their rats to smell so we ran a parallel adsorption ﬁlter conﬁguration (Fig. 8.3) and let the rats smell both the MEDDS and the APOPO ﬁlter. Figure 8.3 is a picture of the parallel ﬁlter system. Chemical detectors can and are using ﬁlters in mine detection to do a similar function, to concentrate samples, and also to separate the rough and tumble of ﬁeld sampling from the analysis step. The various companies use different solutions, but inevitably they have to have a holding stage followed by a quick release into the detector. Like the cold ﬁnger and ﬂash heater of commercial chemical analyzers. Editor: You discussed the camouﬂage patterns of leaked explosives in soil and vegetation. How long after the puddles dry are MDDs able to ﬁnd the scent? It would seem that, since animals must be trained with particular suites of molecules, or aromas, that these should match the expected ﬁeld aromas. Do decomposition products become more important and change over time? VJ: Both Sandia and FOI analyzed the decomposition products of TNT in various soils and conditions. An example of a FOI result is shown in Figure 8.4. These are products extracted out of soil contaminated with TNT, RDX, and TETRYL (Trinitrophenyl-n-methylnitramine). For TNT a decomposition product that had a longer half-life in the soil was 4-amino-dinitrotoluene, so in doing area reduction one should include this product. Editor: This begs for a much more extensive explanation and interpretation. What does it teach us? VJ: The only comment I would like to make is that of the hundreds of soil possibilities put into a bucket with the hundreds of landmine types put into

INTERVIEW

187

TNT

Real Sample from Afghanistan Kharga Area June-03 029AFG30

8e + 4

2, 4-DNA

Intensity

1e + 5

4-ADNT RDX 2-ADNT

6e + 4

4e + 4 5

10

15

20

Time (min)

TNT

3e + 5

Real Sample from Kharga Area Afghanistan 029AFG29

2e + 5

2, 4-DNT

1e + 5 5e + 4 5

10

15

4-ADNT RDX 2-ADNT

2e + 5 2, 4-DNA

Intensity

3e + 5

20

Time (min) (a)

Figure 8.4(a) Analysis of ﬁeld samples from the vicinity of abandoned landmines in Afghanistan. The ﬁgure shows the chromatograms of two samples where TNT, 2,4-DNT, 4-amino-2,6-DNT and 2-amino-4,6-DNT have been detected and quantiﬁed using GC and thermoionic detection. In sample B, the TNT concentration caused the GC detector out of range. RDX was also detected in both samples. (2,4-DNA = 2,4-dinitroanisole used as internal standard.) The detected sample background was in both cases very low. As an extra validation of results, the presence of TNT in sample B was veriﬁed with LC/MS analysis. In cases where samples only contained trace amounts, several samples with similar detected TNT and related compounds proﬁle were pooled to obtain concentrations high enough for LC/MS analysis. (Courtesy of Ann Kjellstr¨om, FOI.)

188

REFLECTIONS ON HUNTING MINES BY AROMA SENSING

0.065 LC

0.060

O2N N O2N N

UV 2B5 nm

4-NT NO2 N NO2 N

O

O−

0.055 N

0.050 0.045

O

0.040

N−

O

O

N−

0.025

N−

N+ O

2, 4-dinitro6-NT Tetryl 3-NT

O−

N

0.020 0.015 0.010 0.005

−

O

N−

N

O−

O

+

N+

N O−

O−

O

NB O−

0.030 O

O−

O−

N O

N+

3, 4-DNT

N−

N

O

N+ O

NO2

O

N− N

N+ O− O

−

N

−

0.035 A.U

N

O2N

N

N

N−

1, 2-DNB O 2-NT

N+ O 2, 5-DNT 1, 4-DNB 2, 4-DNT

3, 4DNT

1, 3-DNB

O−

O

2, 3-DNT

RDX HMX

CL 20

O−

2-amino-4, 6-DNT

3, 5-DNT TNT

4-amino-2, 6-DNT 2, 6-dinitro4-NT

TNB

0.000 5.00

10.00

15.00

20.00 25.00 30.00 Time (min) (b)

35.00

40.00

46.00

Figure 8.4(b) HPLC-UV chromatogram of organic explosives (standards) using a porous graphitic carbon (PGC) column in reversed phase mode (UV detection 290 nm). Gradient elution using a water–acetonitrile–methanol–toluene system. A quartenary HPLC system was used to deliver the solvent mixture. The complete separation of all the analytes, including the DNT isomers, was carried out on a porous graphitic carbon column (150 mm, 4.6 mm I.D., 5 μm particles). UV detection was conducted by using a Waters UV Detector Model 996 photodiode array detector. The best signal-to-noise ratio was found at 290 nm. (Courtesy of Erik Holmgren, FOI.)

that bucket of soil, you would never get results that would help you with a ﬁeld system. The only signiﬁcant result for me was the recognition that decomposition products are abundant and some have much better survival and mobility away from the source of explosive. Results we observed in the Croatian tests support this observation. Editor: Can you illustrate just how the REST system is used in the ﬁeld? VJ: While working with the company CGTVA19 in Mozambique we used a REST to sample and then using the Morogoro rats to analyze the ﬁlters. We surveyed 70 kilometers of road verge using the pump on a hand-pushed cart (Fig. 8.5).

19

Carlos Gassmann Tecnologias De Vanguarda Aplicadas Lda, Portugal.

INTERVIEW

189

Figure 8.5 See color plates. Searching for mines and UXO in Mozambique using REST procedures.

In the picture Osario Serveniano, the ex-director of the Mozambique mine action center and then the manager of the CGTVA operation, can be seen doing the sampling. He is standing in a footpath made by locals so is quite safe. The area to be cleared extended some 3 meters each side of the road and some UXO (unexploded ordnance) and mines were located. The road had to be widened with a new tar strip and the road building machines followed the demining operation. No mines or UXO were missed. Editor: You have mentioned these rats several times. Can you describe them and how you use them? How is it different from using MDDs? VJ: Figure 8.6 is a picture of me petting one. I suggest you get a copy of one of APOPO’s publications. APOPO had won the World Bank competition in

Figure 8.6 APOPO African pouched rat (Cricetomys gambianus), a friendly creature.

190

REFLECTIONS ON HUNTING MINES BY AROMA SENSING

the face of more than a thousand entries, I believe, for this system to help in the mine clearance problem. The story should be out there. APOPO has a website [2]. Editor: Your success with dogs, both in the mineﬁeld and using the MEDDS system, is well documented. You have also used the rats and most electronic systems. What does your experience indicate, overall, of the relative sensitivities of the animals and the electronic systems? VJ: Animals are more sensitive and a lot quicker but their communication and training is ﬂawed. Editor: Do you have a feeling, based on your extensive ﬁeld experience, of the relative values of sensing a single chemical compound or a group of related chemicals that form a complex aroma? VJ: What is important in the project we discussed is that, to my mind, it indicated what chemicals had staying power in soil and spread out certain distances using plants, rain, wind, and diffusion to be picked up by an animal or chemical detector under real world conditions. In the test, the dogs were trained to ﬁnd only TNT and nothing else. Because Fido could ﬁnd more chemicals, it could afford to lose on straight TNT detection in this real-world test. Editor: Based on your experience with animals and with some of the best current chemical sensing technology, what would you project as the necessary practical sensitivity needed in a chemical sensor: VJ: Around 10−15 grams of TNT for a detection. Less sensitive than that will require a concentration step that would take time. For Mechem to do clearance contracts looking for “wild mines” we had to move at about 20 kilometers a day while sampling a 10-meter-wide strip. Manual demining worked orders slower. Rough estimates showed that in Cambodia, Angola, and Croatia, manual demining will take 50 years or more to remove the mines, using the most optimistic calculations. Mechem’s methods could clear roads at a practical rate. However, wild mines are spread over open countryside. To do area clearance, Mechem’s techniques would have to speed up at least ﬁvefold to be really useful. Manual demining, using hand-held detectors, at present rates has no hope in humanitarian demining. Military objectives are much more restricted. Editor: You mentioned earlier that animals are faster. What kind of cycle time do electronic instruments need to have to become competitive? This may be best answered in terms of area coverage rate, or something like that. VJ: Quite a question. Let’s look at MEDDS as we used it: First, we do a survey checking 400 meters by 10 meters per ﬁlter. A ﬁlter suction sampler can reliably ﬁnd explosive vapors (on dust) on the vegetation at 5 meters distances away from the source of the explosive, so one pass with a vehicle gives a 10-meter-wide check. The use of an absorption ﬁlter concentrates the explosive vapor. We suck at a liter per second continuously through the ﬁlter.

INTERVIEW

191

A dog sucks maybe 100 milliliters per sniff, so he must move slower if he needs to pick up enough dust from say 5 meters away from the source. Surveying a 10-meter-wide path at 2 kilometers per hour gave 5 pairs of simultaneous samples in ﬁlters; we could check 400 ﬁlters a day with a team of MEDDS dogs. All samples were taken double. The MRV20 sampling vehicle worked 10 hours a day, thus doing 20 kilometers, and 200 samples in the 10 hours. Doing the test directly in the vehicle while on the move would require an analysis every 3 minutes. Each ﬁlter would then represent 400 meters times 10 meters. The positive sections are then rechecked using a normal MDD. A dog takes a sniff and in a second and gives the answer. The problem with MDDs is that they are not allowed to follow their noses but the handlers have to work them in a rigid pattern to satisfy contractual requirements. At that rate it takes them about a minute to cover a 10-meter by 10-meter section. So if an instrument that can work a 10-meter by 10-meter section and declare it positive or mine free in a minute, it can do the same. A manual deminer does the ﬁnal positioning and opening of the mine. Not all companies using MDD on squares work the same as Mechem, to let the dog only identify the positive section and then let a manual deminer use a handheld detector and search the whole 10-meter by 10-meter positive section to ﬁnd and position the mine. Some other companies try and rely on the dog indicating the mine within a 1-meter distance and then let the man just open the mine The dog process is then slower, but the manual deminer’s job much shorter. In my opinion this method is less reliable because of the movement of the explosive smells away from their source and allowing conclusions that dogs had given a false signal when there is no mine within a meter. They then move to another block often leaving a mine behind in the positive block. Alternatively for an instrument to ﬁt into our process, a third different approach can be used: Redo the long positive section with ﬁlters but now use much smaller areas to narrow down the position of mines, for example, sections of 10 meters long by 10 meters wide. If the answer would come directly through an analyses of the previous section while one is now sampling the next section, then the process would be continuous. The problem with this third approach was that the changing of the ﬁlters between sections becomes a real time-limiting step. Doing 10 meters at 2 kilometers per hour would require a test answer every 18 seconds. Editor: You estimated 3 minutes per analysis on board the MRV. You also state an analysis cycle time of 18 seconds. How do these two times relate to each other? VJ: They are supporting different steps in the process. In summary, the process is as follows: 20

Mine resistant vehicle.

192

REFLECTIONS ON HUNTING MINES BY AROMA SENSING

1. Survey the road or strip at 2 kilometers per hour in 400-meter lengths, 10 meters wide (3 minutes per strip). 2. Go back to the positive strips and ignore the negative strips. 3. Narrow the positive in the 400-meter strip to 10-meter by 10-meter blocks that are testing positive (18 seconds per block). 4. Use manual deminers in the 10-meter by 10-meter positive blocks to locate the precise position. Editor: You mentioned the MRV. You have spent a great deal of effort developing these vehicles. Having seen your videos of the Casspir moving through a mineﬁeld exploding mines as it goes, but without damage, is very impressive. Did you use them extensively to do the route surveys you mentioned? VJ: The key to the Mechem mine clearance operations have been the use of these vehicles because they can carry detectors through mined areas safely without exposing the operator. Why I liked rats is because the REST system using them could be built into a MRV like a Buffalo or Casspir using a team of rats on board. The new Fido can now also do this. MRV are also sometimes called MPV or Mine Protected Vehicles but I prefer MRV. Incidentally in contrast to LAV light armored vehicles like Bradley and Stryker the MRV used in Africa and built in South Africa, Namibia and Rhodesia numbered around 19 000 and are now the only vehicles that suit conditions in Afghanistan and Iraq. Editor: Thank you for the beneﬁt of your experience. Hopefully, these insights will help make the development and employment of new technologies more efﬁcient and more effective. REFERENCES 1. Fjellaner, R. The REST concept, in I. G. McLean, Ed. Mine Detection Dogs: Training, Operations and Odour Detection. Geneva International Centre for Humanitarian Demining (GICHD), Geneva, 2003, pp. 53–107. 2. http://www.apopo.org/newsite/content/, visited 12/11/05.

PART III

EXAMPLE SENSING TECHNOLOGIES

CHAPTER 9

EXPLOSIVES DETECTION BASED ON AMPLIFYING FLUORESCENCE POLYMERS COLIN CUMMING Nomadics, Inc., an ICx Technologies Company.

9.1

INTRODUCTION

One of the most promising technologies for explosives detection is built upon ﬂuorescent polymers developed at the Massachusetts Institute of Technology (MIT) by Professor Timothy Swager. When excited by a light source, the polymers ﬂuoresce. However, should a molecule of explosive bind with the polymer, the ﬂuorescence is quenched. Further, because the polymer can be structured so that the polymeric chains are electronically linked, a binding event anywhere along this “molecular wire” of conjugated polymer quenches the ﬂuorescence all along the polymer network. Consequently, even a single binding event can result in a reduction of overall ﬂuorescence that is readily detectable with relatively simple and inexpensive optics. For this reason, the technology is referred to as amplifying ﬂuorescent polymer (AFP) [1]. The ﬂuorescence-quenching phenomenon is clearly shown in Figure 9.1. Both vials contain AFP in water and are being exposed to an excitation source. However, the vial on the right also contains particles of TNT; as a result, the ﬂuorescent emissions have been shut down. The use of MIT’s AFP for explosives detection has been licensed to Nomadics, Inc. The company has developed a family of explosives detection systems based on the AFP sensor platform and is expanding this collection of devices for additional applications. Because these systems emulate the sensing of explosives demonstrated by dogs, the Nomadics detection technology is called Fido®. Thin ﬁlms of AFP materials coated onto a suitable substrate form the sensory element of the Fido explosives detection systems. Through extensive testing, Fido has demonstrated the detection of explosives at levels of concentration that are three orders of magnitude less than those detectable by the most sensitive commercially Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

195

196

EXPLOSIVES DETECTION BASED ON AMPLIFYING FLUORESCENCE POLYMERS

Figure 9.1 Two vials containing the amplifying ﬂuorescent polymer (AFP) that has been developed to detect nitroaromatic explosives. They ﬂuoresce when excited by light of a given wavelength as shown by the vial on the right. The vial on the left contains trace amounts of TNT, which has quenched the ﬂuorescence.

available laboratory instruments [2]. The Fido sensor is the only trace chemical vapor sensor that has demonstrated the ability to detect explosives with performance levels similar to that of trained canines. High explosives (HE) characteristically have very low vapor pressures. Thus, in order to detect explosive by means of its chemical signature, trace sensors typically sample large volumes of the low-concentration analyte in order to collect sufﬁcient material to ensure detection. Fido sensors use a different approach. Rather than collecting large samples of potential analyte or preconcentrating samples in order to obtain enough explosive to generate a detection signal, Fido relies on the ampliﬁcation provided by the AFP so that an extremely small amount of analyte—a few femtograms of explosive—creates a detectable signal. While many conventional detection methods use other approaches to ampliﬁcation, such as electronic ampliﬁcation, these are generally ineffective for detecting vapors from explosives. Because such ampliﬁcation occurs subsequent to the transduction step, both the signal and any noise present are ampliﬁed, providing no advantage in terms of sensitivity or minimum detection level. Fido’s AFPbased technology is highly signiﬁcant in that it exhibits intrinsic ampliﬁcation so that the transduction process is ampliﬁed. 9.1.1

AFP Principle of Operation

Structurally, the Fido polymers consist of a conjugated backbone with rigid, threedimensional pentiptycene groups (Fig. 9.2). The rigidity and three-dimensional

INTRODUCTION

197

C14H29 O

n

O C14H29

Figure 9.2 Chemical structure of the AFP (left) and a representation of the physical structure that has been engineered to bind to vapors of nitroaromatic explosives (right).

structure of the polymer forms cavities in the sensor ﬁlms that accommodate small molecules and helps to enhance diffusion of small molecules into the ﬁlms. Because of the overlap among polymer backbones, there is a risk that the polymer will quench itself due to π stacking. However, the AFPs have been engineered so that the same structural features that help create cavities for analyte capture also minimize self-quenching [3]. The polymer depicted in Figure 9.2 is only one of several dozen AFPs that have been synthesized for explosives detection, each with slightly different responses to target analytes. Because the performance needs vary for different explosives detection applications, these different AFP formulations seek to optimize the material’s performance characteristics for speciﬁc needs, such as the attachment of the material to the substrate used, the adsorption of analyte, duration of polymer operational life, temperature stability, and mechanical robustness. 9.1.2

AFP Technology

The sensory mechanism of AFP is not unlike that employed by conventional ﬂuorescent chromophores. Typically, individual chromophores ﬂuoresce upon excitation by a light source. However, if the chromophore binds with a target analyte, then excitation is not possible and the chromophore does not emit photons. This is illustrated in the top panel of Figure 9.3. The signal ampliﬁcation exhibited by the Fido polymers is largely the result of the ease with which excited electrons can migrate along the AFP backbone. When thin ﬁlms of AFP absorb photons of light, the excited state electrons or excitons form and begin propagating along the polymer backbones due to their higher energy. This energy transfer is highly efﬁcient in AFP because the polymer’s electrons are extremely delocalized. Consequently, a single exciton propagates through many polymer repeat units during its excited state lifetime. These polymer units contain electron receptor sites, and eventually the exciton is likely to become trapped in a receptor so that it transitions back to a ground state, resulting in ﬂuorescence. [3]

198

EXPLOSIVES DETECTION BASED ON AMPLIFYING FLUORESCENCE POLYMERS

Traditional chemosensor (isolated chromophores): Binding of an analyte molecule quenches only the chromophore to which it binds. hn

Legend

hn

hn′

hn′

hn′ +

= Emissive polymer = Quenched polymer

−

= Analyte molecule

hn′

hn = Light at excitation wavelength

hn′

hn′ = Light at emission wavelength

hn′

hn′

Receptors wired in series (molecular wire arrangement): A single molecular binding event quenches the fluorescence of many chromophores, amplifying the chemosensory response by orders of magnitude. hn

hn′ + n

−

hn n

Figure 9.3 Comparison of conventional ﬂuorescent chromophore sensors versus the molecular wire response of the AFP.

When an electron-deﬁcient (i.e., electron-accepting) molecule binds to one of the polymer’s receptor sites, a low-energy trap is formed. If an exciton migrates to the site of the bound electron-deﬁcient molecule, the exciton will become trapped and unable to return to a ground state. Thus, no ﬂuorescence will result from the excitation event. One key to the ampliﬁcation provided by the AFP technology is that once exciton migration has been halted anywhere along the polymer chain, ﬂuorescence is prevented along the entire chain. This is the mechanism illustrated in the lower panel of Figure 9.3. In both examples, the same number of chromophores are present when a single binding event occurs, but the reduction in ﬂuorescence is substantially greater in the case of the AFP chain. [3] Another factor in the enhanced sensitivity of AFP is the extent of exciton propagation. Individual excitons are not conﬁned to the polymer chain upon which the original photon absorption event occurred. Excitons propagate three dimensionally and follow a “random walk” path, which means that they can pass through the same polymer repeat unit more than once. Because an exciton migrates through many receptor sites, the probability that it will encounter an occupied receptor site is greatly increased. Hence, the likelihood that a photon absorption event will result in ﬂuorescence is greatly reduced even when small amounts of analyte have bound to the polymer receptor sites. While in theory an AFP chain of N polymer repeat units would be N times more sensitive than the same number of monomer chromophores, in practice the Fido polymers exhibit an effective ampliﬁcation in response of between 100 and 1000 times as compared to monomeric quenching mechanisms. [4] The Fido polymers have also proven to be quite selective in responding to speciﬁc analytes. This characteristic has been engineered into the materials via

INTRODUCTION

199

three principal paths. The ﬁrst path is through the reduction of steric constraints. Because ﬂuorescence quenching depends on the transfer of electrons from the electronically excited polymer to the target analyte, measures must be taken to enhance the transfer of electrons. The rate of F¨orster energy transfer is greatly inﬂuenced by the distance between the electron donor and the acceptor, with the rate varying with the distance between the donor and acceptor to the inverse sixth power. Small molecules, such as the target explosives, ﬁt into the cavities within the AFP ﬁlms and, as electron acceptors, come into close proximity to the electrons migrating along the donor polymer backbone. Larger molecules are excluded, greatly reducing the possibility that these nontargeted analytes will quench the polymer. A second mechanism to enhance selectivity is the electrostatic complementarity between the polymer and target analytes. The polymers, which are electron rich, appear to bind reversibly to electron-deﬁcient explosives through an electrostatic-type interaction. Compounds that have a strong afﬁnity for the polymer (i.e., those that have a large binding constant, Kb ) are more likely to bind strongly with the polymer. To enhance binding of target compounds, receptor sites that are electrostatic mirror images of the target molecule have been speciﬁcally synthesized into the polymer backbone. A third avenue to enhanced selectivity is the fact that, for electron transfer to the bound molecule to occur spontaneously, the overall free energy change (G0 ) for this process must be negative. Thus, the bound molecule must have a standard reduction potential large enough to cause G0 to be negative. In equation form, the magnitude of ﬂuorescence quenching (FQ) can be approximated as FQ ∝ (C)[exp(−G0 )2 ](Kb ) where (C) is the concentration of the quencher in the sample. Hence, for signiﬁcant quenching to occur, an analyte must bind strongly to the ﬁlm and have an appropriately high reduction potential. Consequently, compounds that bind weakly to the AFP ﬁlm and those without favorable reduction potentials are not likely to result in signiﬁcant quenches, except in very high concentrations. 9.1.2.1 Fido Detector Operation The Fido technology has been in development since 1998, leading to several products with more on the way. Implementing AFP as a sensor platform and integrating it into an electronic and optical system capable of highly sensitive and selective detection of explosives under a number of scenarios has been an iterative process. While numerous optical geometries and sampling conﬁgurations have been developed, Figure 9.4 shows the general conﬁguration of components prominent in most designs. The conﬁguration illustrated is built around a glass capillary whose inner surface is coated with AFP material. The device is conﬁgured in such a way that the system pump pulls air from the environment through the capillary. Because molecules of many explosives tend to stick to surfaces, the inlet is heated to ensure that analyte does not become trapped in the inlet. The heat also helps to improve system response [5].

200

EXPLOSIVES DETECTION BASED ON AMPLIFYING FLUORESCENCE POLYMERS

Detector

Filter Fluorescence

Contaminated Air LED Activated Chromophore Coating Capillary Waveguide

Contaminated Air

Figure 9.4

Block diagram of the Fido explosives detection system.

As air moves through the capillary, the molecules of explosives diffuse to the AFP where they are adsorbed. External to the capillary is a light source. Various Fido models have used blue light-emitting diodes (LEDs), ultraviolet LEDs, and laser diodes. The light source is matched to the particular AFP used for a speciﬁc analyte and application. The excited AFP emits ﬂuorescence, which the optics system waveguides to a detector aligned with the axis of the capillary. Both photomultipliers (PMTs) and semiconductor photodetectors have been used in various Fido models. The optics subsystem also includes a ﬁlter ahead of the detector to reject the excitation light. Reductions in ﬂuorescence are processed by the Fido controller and software, and data indicating whether explosives are present in the sample are presented to the operator. The entire process from sample collection to notiﬁcation of a sample containing explosives takes a matter of a few seconds [6]. Additionally, the AFP is recoverable—that is, once the analyte is removed (by airﬂow), the sensor is ready to analyze another sample. Recovery time is generally less than one minute. In ﬁeld experience, AFP ﬁlms have been exposed to analyte, recovered, and exposed again through dozens of cycles before the intensity of response diminishes enough to require replacement of the AFP cartridge [7, 8].

FIDO SENSITIVITY

9.2

201

HISTORY OF AFP AND FIDO®

The initial use of AFP for explosives detection came as part of the Defense Advanced Research Program Agency (DARPA) Dog’s Nose Program. The goal of this project was to identify technologies that could locate buried landmines based on the minute amounts of explosives that escape to the surface, emulating the sensory methods theorized to be used by dogs. Nomadics proved the feasibility of explosives detection with AFP in the laboratory using a rudimentary sensor. As a result of this genesis, the AFP platform became known as Fido. The ﬁrst ﬁeld tests of early prototypes took place in 1999 at Fort Leonard Wood, Missouri. Since that time, the technology has been further reﬁned and tailored for use in various other explosives detection applications [8]. In 2001, the Fido system was modiﬁed to operate underwater and became known as the SeaDog. The U.S. Navy Ofﬁce of Naval Research (ONR), under its Chemical Sensing in the Marine Environment (CSME) Program, funded the integration of the SeaDog with an autonomous underwater vehicle (AUV). The integrated system was able to map a plume of trinitrotoluene (TNT) in open water in real time. This was the ﬁrst demonstration of the mapping of an explosive plume underwater in real time [9, 10]. The SeaDog technology proved to have applications for environmental monitoring as well. In 2002, the Strategic Environmental Research and Development Program (SERDP) funded Nomadics to develop a Fido conﬁguration that could be used to monitor explosives contamination of groundwater. SERDP is a joint effort by the Department of Defense (DoD), the Environmental Protection Agency, and the Department of Energy focused on remediating former military sites for public use. SERDP has also funded development of Fido devices that can assist in locating unexploded ordnance in former range sites. Current Fido development efforts and new explosives detection applications are directed toward cargo, vehicle, and personnel screening for covert explosives and improvised explosive devices (IEDs). In short, Fido is being considered for use in nearly all explosives detection application in which dogs have been used [11]. 9.3

FIDO SENSITIVITY

The Fido system has demonstrated the ability to detect low (<10) femtogram levels of explosives in the vapor phase. This corresponds to parts per quadrillion (ppq) concentration levels [7]. These measurements were made in a direct sample mode without preconcentration. Fido’s extreme sensitivity beneﬁts potential users in many ways beyond direct detection needs. For applications where sensitivity is not as critical, reasonable trade-offs can be made in system design. For example, the use of less sophisticated optics can reduce cost and allow the system to be more rugged, while lower complexity in sample collection and signal processing supports portability through lightweight, low-power models.

202

9.4

EXPLOSIVES DETECTION BASED ON AMPLIFYING FLUORESCENCE POLYMERS

FIDO PERFORMANCE

Generally, the performance characteristics of greatest interest for an explosives detection system are sensitivity, selectivity, and response time. As used here, sensitivity is the ability to detect the target analyte in extremely small concentrations, while selectivity is the ability to distinguish the target analyte from other materials that may be present. In combination, good sensitivity and selectivity mean a high probability of detection when the analyte is present and a low false alarm rate when the analyte is not present. There are two dimensions of response time. First, the explosive must be detected in real time, essentially immediately upon collection of the sample. Second, the sensor must be ready to collect and process a subsequent sample with minimal delay and, certainly, without complicated procedures for resetting the system or replacing reagents. Because there are no real standards for evaluating the performance of explosives detection system in terms of these characteristics, the performance of dogs has served as the gold standard for explosives detection. A canine’s olfactory sense is capable of detecting miniscule amounts of explosives vapor, especially at a distance where the plume has become highly dispersed. Dogs can be trained to be very selective in what they respond to, and their response time for detection and recovery is nearly instantaneous. While a dog’s performance may be affected by fatigue, illness, boredom, distractions in the environment, and other factors, comparison with canines is a useful way of evaluating the explosives detection performance of a technology. The need for comparing such technologies to the abilities of dogs is further underscored when the complexity of the real-world environment is considered. While there is value in measurements such as detection thresholds being made in a pristine laboratory, performance in the ﬁeld against realistic targets under real-world scenarios is needed to provide an adequate level of conﬁdence for any system intended for the crucial and dangerous task of explosives detection. Fido has been tested in the laboratory and in the ﬁeld, including head-to-head comparisons with dogs. Field test results are described in Chapter 7 of this volume.

9.5

PERFORMANCE LIMITATIONS

Fido’s performance limitations are largely deﬁned by application requirements. For example, a particular application may require several hours of operation on batteries, minimal weight, and high-volume sampling. These requirements are at odds with each other, so trade-offs have to be assessed in order to determine the best overall conﬁguration for a particular application. However, when a detection system provides exceptional sensitivity, selectivity, and response time as the Fido technology does, more latitude is available in implementing trade-offs while maintaining adequate performance in these three areas.

LATEST IMPLEMENTATIONS

9.6

203

PHYSICAL PARAMETERS

As has been mentioned several times, the exact physical parameters of the Fido depend on how the system is conﬁgured for a particular application. Currently, the sensor platform, including optics and power components, weighs less that a kilogram with a volume of no more than 3.3 L. These ﬁgures are approximately one-third of the comparable physical parameters in the early Fido prototypes, and opportunities for further reductions in weight and size remain. This will continue to be the case as advances in other technologies take place. For example, the options for light sources have greatly expanded while the size and power requirements for those components are signiﬁcantly more conducive to portability than what was available when Fido development began. 9.7

LATEST IMPLEMENTATIONS

The commercially available Fido and Fido XT explosives detection systems are both handheld devices as shown in Figure 9.5. The Fido incorporates the sampling head directly into the body of the device. The XT version includes a tethered extension for the inlet that allows the sampling head to be separated from the rest of the instrument. The sampling head is mounted on a pistol grip so that sampling can be performed more conveniently (Fig. 9.6). In addition to the handheld models, Fido has been adapted for integration with robotic platforms most popular with those in the explosives detection community. As examples, these include the Foster Miller Talon, the iRobot Packbot, and the USMC Dragon Runner. The Fido XT tethered sampling head can thus be mounted on a robotic arm to accommodate a variety of operational concepts as shown in Figure 9.7.

Figure 9.5 The Fido X (left) and Fido XT (right) explosive detection systems.

204

EXPLOSIVES DETECTION BASED ON AMPLIFYING FLUORESCENCE POLYMERS

Figure 9.6 Fido XT sampling head.

Figure 9.7

Fido on robot.

The SeaDog is also available in two conﬁgurations. A handheld version can be easily used by divers, while the AUV model is speciﬁcally conﬁgured for integration with underwater autonomous vehicles, such as the REMUS (Remote Environmental Monitoring Unit) shown in Figure 9.8. The AFP detection technology is a sensor platform that lends itself to integration into a number of conﬁgurations while offering signiﬁcant potential for performance enhancements by taking advantage of advances in other areas. New developments in polymer chemistry and synthesis, innovative surface coating methods, breakthroughs in optics and semiconductor devices, and advances

LATEST IMPLEMENTATIONS

205

Figure 9.8 SeaDog on REMUS.

Figure 9.9

Multichannel array concept.

in microelectrical and mechanical systems (MEMS) all offer the potential for improved performance. One such innovation currently in development is the use of multiple bands of AFP in the sensor platform. By employing two or more detection bands comprised of the same or different polymer materials within the capillary, the selectivity of the detector can be increased. Figure 9.9 shows a schematic representation of the Fido array sensor concept. In this example, the sensor has two sensor channels. Samples are drawn into the capillary and past the array bands. A separate laser diode source is used to excite each polymer band. The lasers are modulated at different frequencies, enabling the emission of each band to be measured with a single photodetector. The capillary directs the light emitted from each polymer band onto the photodetector. As with the basic Fido, a digital signal processor is used to control the output of the laser diodes and to process the signal from the photodetector [12].

206

EXPLOSIVES DETECTION BASED ON AMPLIFYING FLUORESCENCE POLYMERS

The enhanced selectivity of the array sensor is achieved via what is in essence a crude chromatographic separation. Because molecules of different compounds interact with the AFP with different strengths, some constituents of the sample will bind more strongly to the AFP ﬁlm than others. Thus, the ﬂow rate of analyte through the capillary varies based on the binding strength between the analyte and AFP ﬁlm. Molecules with weak interactions rapidly sweep past the bands of polymer in the ﬂow of carrier gas, while those that are more strongly retained take longer to clear the sensor. In this way, the AFP ﬁlm serves as a chromatographic stationary phase [12]. If a given sample constituent quenches the polymer, the ﬁrst channel will quench in proportion to the concentration of quencher in the sample. At some time interval after the onset of response in the ﬁrst channel, the second channel will begin to respond as the analyte pulse reaches the second polymer band. Eventually, the responses of both bands return to the baseline value as the quencher is swept from the sensor. The time evolution of responses in both channels is characteristic of a given quencher, providing data that can be used to more speciﬁcally identify the substance and, thus, enhance the selectivity of the sensor. Figure 9.10 shows the sensor response to a nontarget compound that binds weakly with the AFP. The response by both channels is almost instantaneous. By comparison, Figure 9.11 is the response to 2,4-DNT, [2,4-dinitrotoluene] a compound frequently contained in TNT. With this sample, the delay between the response maxima of the two channels exceeds 1 s, which is signiﬁcantly longer than response of the potential interferent [12]. Using discrimination algorithms, simple signal processing methods can plot response curves such as that shown at the bottom of Figure 9.11. This trace represents the difference in response between the two channels (i.e., the response of the

1.1 Channel 2

Normalized Response

1 0.9 0.8 0.7 Channel 1

0.6 0.5 0.4 0

1

2

Figure 9.10

3

4

5 Time (s)

6

7

8

Interferent in multichannel Fido.

9

10

207

1.05 Channel 1 Normalized Response

1 0.95 Channel 2 0.9 0.85

Ch1 - Ch 2

0.8 0.75 0.7

2, 4-DNT 0

1

2

3

Figure 9.11

4

5 6 Time (s)

7

8

9

0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 −0.05 −0.1 −0.15 10

Channel 1- Channel 2

LATEST IMPLEMENTATIONS

Array sensor response to 2,4-DNT.

0.1

2, 4-DNT

Channel 1- Channel 2

0.08 0.06

TNT

0.04

Interferent

0.02 0 −0.02 −0.04 −0.06 −0.08 −0.1

0

1

2

3

4

5

6

7

8

Time (s)

Figure 9.12 Array sensor differential response for TNT, 2,4-DNT, and an interferent.

second channel is subtracted from the response of the ﬁrst channel). Figure 9.12 is a comparison of the differential responses for three different quenchers. The differential responses of TNT and 2,4-DNT can easily be distinguished from the differential response of the potential interferent [12]. In addition, the differential band shapes for each compound are different, providing another indication of the identity of the quencher. In the event that more than one quencher is presented to the sensor in the same sample, the array response will be a convolution of the response of each quencher, leading to a more complex response proﬁle that will be more difﬁcult to interpret. More elaborate discrimination algorithms are being developed to process responses involving mixtures of quenchers [12].

208

EXPLOSIVES DETECTION BASED ON AMPLIFYING FLUORESCENCE POLYMERS

New generations of the Fido technology currently under development will signiﬁcantly shrink both the size and weight of the system. Targeted at deployment on small robots, these new systems will see further reductions in volumes and weights. Another enhancement in current development is preconcentration subsystems appropriate for Fido robot-deployed and handheld explosives detection systems. The primary Fido operational concept does not incorporate preconcentration, but clearly there are scenarios and applications where preconcentration will contribute to performance. To this end, integrated systems that will tightly couple AFP-based detection with MEMS-based preconcentration are under development.

9.8

MATURITY OF THE FIDO TECHNOLOGY

The Fido technology is currently under evaluation for use by U.S. military forces. The Fido X and Fido XT are available as commercial off-the-shelf (COTS) items. Consequently, the technology is adequately mature for commercial deployment. However, as a platform technology, the AFP sensor and Fido detection system support broad application to meet explosives detection needs. Further, Nomadics has incorporated the ampliﬁcation features of AFP into other sensor mechanisms aimed at the detection of analytes that are not explosives related, including other chemicals and compounds of interest in the biomedical and food safety ﬁelds. Thus, while the technology is mature enough for commercialization, its potential is far from fully exploited.

9.9

FUNDING

Professor Swager’s early work on conjugated polymer-based sensing was funded by the Ofﬁce of Naval Research (ONR). The Defense Advanced Research Projects Agency (DARPA) provided funding for the initial implementation of AFP and also for the ﬁrst Fido prototypes conﬁgured for landmine detection. Subsequent funding to continue Fido development was provided by the U.S. Army’s Night Vision and Electronics Systems Division (NVESD). Work on the SeaDog marine explosives detection system was funded under ONR’s Chemical Sensing in the Marine Environment (CSME) Program. Other organizations that have contributed funding to further the AFP and Fido technologies for explosives detection include the U.S. Air Force, the Army’s Institute of Soldier Nanotechnologies at MIT, the Oklahoma Center for the Advancement of Science and Technology, the Oklahoma City Memorial Institute for the Prevention of Terrorism, Great Britain’s Defence Science and Technology Laboratories, SERDP, and the FBI. Current funding sources also include the Transportation Security Administration and the Homeland Security Advanced Research Program Agency.

REFERENCES

209

REFERENCES 1. Yang, J. S. and T. M. Swager, Fluorescent porous polymer ﬁlms as TNT chemosensors: Electronic and structural effects. J. Am. Chem. Soc. 120, 11864–11873 (1998). 2. la Grone, M., C. Cumming, M. Fisher, D. Reust, and R. Taylor. Landmine detection by chemical signature: Detection of vapors of nitroaromatic compounds by ﬂuorescence quenching of novel polymer materials, in A. C. Dubey, J. F. Harvey, J. T. Broach, R. E. Dugan, Ed. Detection and Remediation Technologies for Mines and Minelike Targets 1V, Proceedings of SPIE, Vol. 3710, April 5–9, Orlando, FL, 1999, pp. 409–420. 3. Williams, V. E., J.S. Yang, Lugmair, C. G. Miao, Y. J. Swager. T. M. Design of novel iptycene-containing ﬂuorescent polymers for the detection of TNT Proc. SPIE, 3710 402–408 (Oct. 1999). 4. Fisher, M. and C. Cumming. Fluorescent polymer chemosensors for explosives detection (Poster Session). Fourteenth International Forum, Process Analytical Chemistry, Las Vegas, Nevada, January 2000. 5. Cumming, C. Aker, M. Fisher, M. Fox, M. la Grone, D. Reust, M. Rockley, T. Swager, E. Towers, and V. Williams. Using novel ﬂuorescent polymers as sensory materials for above-ground sensing of chemical signature compounds emanating from buried landmines, in Proceedings of the UXO/Countermine Forum, New Orleans, LA, April, 2001. 6. Fisher, M. and C. Cumming. Trace detection of nitroaromatic explosives by ﬂuorescence quenching of novel polymer materials, in Proceedings of the U.S. Federal Aviation Administration’s Third International Aviation Security Technology Symposium, Atlantic City, NJ, November 27–30, 2001. 7. Fisher, M. C. Cumming, and M. Prather. Detection of ultra-trace landmine chemical signatures using novel sampling and sensing strategies. Poster Presentation, Gordon Research Conference, Il Ciocco, Italy, June 2003. 8. Fisher, M. and J. Sikes. Detection of landmines and other explosives with an ultratrace chemical detector. NATO e-Nose Conference, Coventry UK, October 2003. 9. Dock, M. M. Fisher, and C. Cumming, Novel detection apparatus for locating underwater unexploded ordnance, in Proceedings of the 5th International Symposium on Technology and the Mine Problem, Mine Warfare Association, Monterrey, California, April 2002. 10. Dock, M., J. Sikes, M. Fisher, and M. Prather. Chemical detection of underwater explosives. 2004 Mine Countermeasures & Demining Conference/MINWARA, Canberra, Australia, February 2004. 11. Fisher, M. Applications of sensors utilizing amplifying ﬂuorescent polymers for ultratrace level detection of explosives. Eighth International Symposium for the Analysis and Detection of Explosives (ISADE), Ottawa, Canada, June 2004. 12. Fisher, M. M. Prather, and J. Sikes. Serial amplifying ﬂuorescent polymer arrays for enhanced chemical vapor sensing of landmines. EUDEM-2/SCOT Conference, Brussels, Belgium, September 2003.

CHAPTER 10

ION MOBILITY SPECTROMETRY RONALD L. WOODFIN

10.1

INTRODUCTION

Ion mobility spectrometry (IMS) is in worldwide, daily use in laboratories in many ﬁelds of chemistry, medicine, food science, and manufacturing and is perhaps the most commonly used technology for detection of explosives at the present time. Its use in forensic laboratories is well known. In fact, it is so well established that there is a journal specializing in IMS [1]. A recent book [2] provides a very detailed development of the subject, with many references. Section 11.3 of this book provides a comparison among several forms of mass spectrometry (MS). Section 11.3.3 is devoted to a comparison of IMS with MS. Therefore, it was considered that with so much detailed information available, this chapter could be rather brief and simply reference more detailed accounts. Some recent developments are mentioned in Section 10.3, but a government security ruling while this book was in press has limited the material that can be included. The potential for extremely good limits of detection seems to exist, as projected in Section 10.3. Several manufacturers offer IMS instruments for sale. A partial list is included in Section 10.4. The author apologizes for any omissions. Some work has been done in recent years to make ﬁeld-portable units available. Many technical reports are available. A sampling is listed in Section 10.6.1. 10.2

BRIEF DESCRIPTION OF PRINCIPLE OF OPERATION

We will not attempt a detailed, or technically precise, description of IMS. That is well covered by Eiceman and Karpas [2] and others. The following description is for those who only need a cursory understanding of the principles of operation. Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

211

212

ION MOBILITY SPECTROMETRY

An ion mobility spectrometer,1 like the devices used in most of the sample technologies described in this book, requires “ingesting” a sample of the medium being searched for explosive molecules. When the medium is water or air, the process is straightforward, but when the sample is to be taken from a solid surface some solvent may be involved. Quantities can be quite small, so papers or cloths, sometimes called “swipes” are often used. These swipes are normally used dry, but sometimes are solvent saturated, then allowed to dry before sampling. In either case the material is passed into a chamber where an ionizing element, often 63 Ni, a radioactive isotope that produces β particles (electrons), converts the molecules in the chamber to ions, the same technique used in many household smoke detectors. Newer designs sometimes use 241 Am, which decays in α particles and γ rays. To avoid the regulatory inconvenience of radioactive material, several electronic ionizing techniques have also been proposed. Some molecules produce positive ions, some produce negative ions. The common explosive compounds of interest produce negative ions, while many chemical warfare agents produce positive ions. The investigator is thus faced with a decision: Am I looking for positive or negative ions? The polarity on most instruments can be adjusted to look for either, though not both from the same sample. However, in an effort to circumvent this condition, some recent instruments have modiﬁed their designs to analyze both positive and negative ions in the same sample. Two basic concepts are used, either by providing two drift tubes with opposite polarities that are fed from the ionization chamber, or by cycling the polarity on a single drift tube to provide alternating bursts of positive and negative ions. An IMS is a time-of-ﬂight (ToF) instrument. What this means is that there is a region in the instrument through which the time required for passage of a particle, ion, or molecule, is measured to determine its identity, or separate it from other constituents. Many forms of ToF spectrometers operate only at partial vacuum conditions; IMS operates at normal atmospheric pressure. The characteristic feature of the IMS is that it uses an electric ﬁeld to accelerate ions toward a detection plate, or grid, the ToF being measured between the gating grid and the guard grid. Clearly, the polarity of the electric ﬁeld will cause ions of one polarity to be accelerated through it, while it will repel ions of the other polarity. A typical IMS chamber arrangement is shown schematically in Figure 10.1. This ﬁgure is reduced almost beyond its simplest form to illustrate the general principles. The ions of the selected polarity will be accelerated down the drift region, left to right in Figure 10.1, while ions of the opposite polarity are repelled and pushed left. An important feature is the reverse ﬂow air, a moderate velocity stream that resists the movement of the accelerating ions and sweeps the nonionized molecules and the ions of the opposite polarity into the exhaust stream. 1 The acronym IMS is commonly used for the instrument or the process, interchangeably and without distinction.

BRIEF DESCRIPTION OF PRINCIPLE OF OPERATION

Drift Region Ionization Region

Gating Grid

213

Collector

Electric Field

Sample Ions

Exhaust

Figure 10.1

Non-ionized Molecules & Opposite Polarity Ions

Reverse Flow Air

Guard Grid

Simpliﬁed schematic of an IMS Chamber.

The result is somewhat akin to a tiny wind tunnel. The only measured quantities are the ToF through the drift region and the intensity of the signal produced on the collector. Since the ions all have the same charge, having lost or gained one electron in ionization, the force on each will be the same. The difference in ToF will therefore be determined by the, hopefully unique, combination of ion mass and ion drag. Drag is used here in a generalized aerodynamic sense, as the force acting opposite to the direction of ﬂight. It is dependent on both the shape and size of the ion. The force of the electric ﬁeld must accelerate the mass of the ion and overcome the drag to deliver the ion to the collector. This combination allows the correlation of the measured ToF with a speciﬁc specie of ion. The measured ToF and the intensity of the signal on the collector are independent quantities. Hence they are often presented on a set of orthogonal axes like those shown in Figure 10.2. The intensity of the signal at the collector is assumed to be directly related to the number of ions collected and therefore represents a

Intensity or concentration

Drift Time (msec)

Sample Time (sec) or Scans

Figure 10.2 Axes for presentation of IMS data.

214

ION MOBILITY SPECTROMETRY

TNT Reactant Ions

RDX

(Thousands)

15

10 Sample Time (s)

5

0 Drift Time (ms)

Figure 10.3 Example presentation of IMS data. (Courtesy Philip J. Rodacy, Sandia National Laboratories.)

measure of the concentration of the specie with that characteristic ToF. As the sample is introduced, the IMS begins to collect the data in a series of cycles, or “scans,” accumulating the ions for a set cycle time, then beginning a new cycle. Hence, one axis may be labeled either “Scans” or “Sample Time,” depending on the preference of the investigator. Sample times are normally such that the units are seconds. The axis labeled “Intensity” or “Concentration” is of relative scale. The other axis, “Drift Time,” is usually in milliseconds. It shows the measured ToF in the drift region. Two examples of this presentation are given in Figures 10.3 and 10.4. Both are from work done at Sandia National Laboratories. The picture in Figure 10.3 is produced from data taken with a commercial IMS; that in Figure 10.4 is from a new miniature IMS developed by Sandia. Like any other ingesting sensor, IMS must be purged between batches. The previously discussed heating–cooling cycle is often used. Often a solid phase microextraction (SPME) ﬁber is used to collect molecules during the cooling portion of the cycle; they are then released into the IMS during the heating portion. In Figure 10.3, the response noted as reactant ions is from the medium containing the explosive molecules. Some of the molecules of the carrying medium are ionized. These ions could be formed from air, water, or a solvent, depending on the procedure being used.

10.3

SOME RECENT DEVELOPMENTS

IMS development in recent years has been focused on increasing sensitivity and miniaturization for portability. Since the basic operation is rather well deﬁned,

215

SOME RECENT DEVELOPMENTS

0.55

Intensity

0.45 0.35 0.25

24

0.15

0

18

0

0.05

12

12

0

14

s

Tim e (m

20

s)

an

Sc

18

16

Drift

60 0

Figure 10.4 Another Presentation of IMS data. (Courtesy Philip J. Rodacy, Sandia National Laboratories.)

much of the recent development has been concentrated on signal processing to increase sensitivity and separate various species. Among the leading groups working on these developments, especially directed toward explosives, is a partnership between Sandia National Laboratories, Albuquerque, and the University of Arizona, Tucson. Figure 10.5 is from a Sandia graphic that illustrates their latest development. In Figure 10.5, the acronym CTIA means capacitive transimpedance ampliﬁer, illustrating the focus on signal processing. A new form of detector, or collector, a μFaraday plate, is a major deviation from existing IMS hardware designs. In the upper left picture, the coin is a U.S. quarter dollar, approximately 24 mm diameter, placed for scale visualization. The picture in the upper right is of the packaged unit under development. These illustrate the efforts toward miniaturization and portability. The picture of the packaged unit reminds us of the ancillary systems necessary to form a useful portable sensor. In this unit they project a 4-hour battery life with a lithium ion battery. They have a pumping system that supplies air at 1.5 L per minute. It also contains an analog-to-digital converter capable of 200,000 samples per second at 14 bits and a personal digital assistant for control and data analysis. The development plan is shown in abbreviated form in Table 10.1 and continued in Table 10.2 with current and projected limits of detection [3]. These are rather remarkable projections! If realized they will push the state of the art forward for any sensing technology.

216

ION MOBILITY SPECTROMETRY

Ionizer Drift Tube

Detector

Figure 10.5 See color plates. Handheld, Portable μ-Faraday Differential CTIA detector under development at Sandia National Laboratories in April 2006. (Courtesy Philip J. Rodacy, Sandia National Laboratories.)

TABLE 10.1 Sandia Development Path for μ-Faraday Array Miniature handheld detection system Battery powered, vapor sampling Estimated operating time -4 hours Results in 1–3 seconds Version 1—lab prototype Version 2—planned improvements Optimize performance Data acquisition and data analysis software Embedded processor and LCD New inlet conﬁguration

10.4

SOME IMS MANUFACTURERS

Since IMSs are used in so many applications, they are manufactured worldwide. Table 10.3 lists a sample of manufacturers in North America and Europe. These are companies that had IMS products available in May 2006. Almost certainly some companies have been omitted. For this the author apologizes. It is hoped

SOME IMS MANUFACTURERS

TABLE 10.2

Sandia’s Current and Projected Limits of Detection

Nanograms (10−9 ) Picograms (10−12 ) Femtograms (10−15 ) Attograms (10−18 ) Zeptograms (10−21 ) Yoctograms (10−24 )

TABLE 10.3

217

Typical Portable System Practical Detection Limits Best Commercial System Detection Limits Current Lab Detection Limits for μ-Faraday Array Attainable Detection Limits for μ-Faraday Array Theoretically Possible Limits for μ-Faraday Array Limit restricted by Fundamental Laws of Nature (actually in the hundreds of yoctograms)

Short List of Some manufacturers of Ion Mobility Spectrometers

Manufacturer Bruker Dr¨ager GE Security GE Security GE Security NUSS America Scintrex Trace Corporation Scintrex Trace Corporation Sibel Ltd., represented by Bahia 21 Corp. Smiths Detection Smiths Detection Smiths Detection Smiths Detection Smiths Detection Smiths Detection Thermo Electron Corporation

Type

Model

Country

GC/IMS IMS IMS IMS IMS IMS

IMS 5000/6000 EntryScan Vapor Tracer Itemiser Explorer 2000 LVDBS

Germany Germany USA USA USA USA Canada

GC/IMS

E 5000

Canada

IMS

MO-2M

Russia, USA

IMS IMS IMS IMS IMS IMS GC/DMA

Sabre 2000 Sabre 4000 Document Scanner Ionscan 400B Ionscan 500DT Ionscan Sentinel II EGIS Defender

Canada Canada Canada Canada Canada Canada USA

that this sample will enable inquirers to make a start in ﬁnding a device that meets their needs.

ACKNOWLEDGMENT

Much of the information in this chapter comes from Philip J. Rodacy of Sandia National Laboratories, Albuquerque, New Mexico, who was invited to write this chapter but was prevented by a series of circumstances beyond our control. Nevertheless, the author takes full responsibility for errors and omissions.

218

ION MOBILITY SPECTROMETRY

REFERENCES

The intent of the forgoing brief article is to acquaint those who do not need the speciﬁc details of this heavily studied device with the basic concepts of IMS. The references that follow have much more detail and offer anyone who needs to ﬁnd all the details a roadmap for doing so. The most comprehensive development is in [2]. Many of the newer works will be found in [1]. Descriptions of IMS operation and details of techniques are found in [4–7]. Reference [4] includes 19 references; reference [5], 35 references; reference [6], 132 references; and reference [7], 57 references. The two SAND Reports, [8] and [9], describe some ﬁeld experience with an IMS. They also contain extensive lists of references. An extensive bibliography [10] of the older IMS literature, from 1939 to 1994, exists as a pdf ﬁle. It has about 674 references in 58 pages. There is no retrieval information in the ﬁle. Inquiries can be made to P.J. Rodacy, Sandia National Laboratories, Albuquerque, New Mexico. 1. The International Journal for Ion Mobility Spectrometry is published as reviewed journal. It is the ofﬁcial publication of the International Society for Ion Mobility Spectrometry (ISIMS), c/o ISAS—Institute for Analytical Sciences, Bunsen-KirchhoffStr. 11, D-44139 Dortmund, Germany. 2. Eiceman, G. and Z. Karpas. Ion Mobility Spectrometry, 2nd ed., CRC Press, Boca Raton FL, 2005. 3. Personal communication with Philip J. Rodacy, Sandia National Laboratories, April 25, 2006. 4. Koyuncu, H., E. Seven, and A. C¸alimli. Examination of some organic explosives by ion mobility spectrometry. Turk. J. Chem. 29, 255–264 (2005). 5. Roehl, J. E. Environmental and process applications for ion mobility spectrometry. Appl. Spectrosc. Rev. 26(1&2), 1–57 (1991). 6. Hill, H. H., Jr., W. F. Siems, and R. H. St. Louis. Ion mobility spectrometry, Anal. Chem. 21 (5), 321–355 (1990). 7. St. Louis, R. H. and H. H. Hill, Jr. Ion mobility spectrometry in analytical chemistry. Crit. Rev. Anal. Chem. 62 (23), 1201–1209, (1990). 8. Rodacy, P. J., P. K. Walker, S. D. Reber, J. Phelan, and J. V Andre. Explosive Detection in the Marine Environment and on Land Using Ion Mobility Spectrometry, A Summary of Field Tests, SAND2000-0921. Sandia National Laboratories Report, Albuquerque, NM, 2000. 9. Phelan, J., P. Rodacy, and J. Barnett. Explosive Chemical Signatures from Military Ordnance, SAND2001-0755. Sandia National Laboratories, Albuquerque, NM, April 2001. 10. Bibliography for Ion Mobility Spectrometry (IMS), unmarked .pdf ﬁle.

CHAPTER 11

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES JACK A. SYAGE AND KARL A. HANOLD

11.1

INTRODUCTION

Security and protection against rogue activities is one of the greatest mainstream public concerns. A broad range of countermeasures are needed, one of which is detection technology to warn against an imminent attack. In this chapter we review the issue of explosives trace detection (ETD) and describe the method of mass spectrometry (MS) as an alternative to existing technologies. The performance criteria on which security screening technologies are judged are accuracy (high detection and low false-positive rate), speed, and cost. As terrorists become more sophisticated, there is a need to detect an increasing variety of threats (new explosives, chemical weapons, etc.). Systems must also be sufﬁciently automated for nontechnical security personnel to operate. Acceptable defenses are now being deployed for baggage scanning, including differential X-ray and X-ray tomography. Swipe methods of sampling using ion mobility spectrometry (IMS) are also being used for high-touch areas of baggage and other items. Other ETD analyzers that have been deployed include gas chromatography (GC) with a variety of detectors such as chemiluminescence and electron-capture detectors. Ion mobility spectrometry is the most commonly deployed method for ETD devices. Its advantages are compact size and relatively low price. For applications requiring a handheld detector, IMS is an excellent choice. For applications that are more stationary, other alternatives to IMS are often used. Mass spectrometry is recognized for its superior performance with regard to sensitivity and speciﬁcity, which translate to lower false-negative and false-positive rates. Active programs are now in place to develop routine MS technology for security applications in airports and other venues. Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

219

220

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

In this chapter we describe the application of MS to detection of trace explosives on people and baggage. We begin by summarizing a variety of candidate ETD detectors; however, we refrain from an in-depth discussion deferring instead to the many excellent review articles and proceedings that are available [1–4]. However, because of their similarities, we do compare the methods of MS and IMS and identify the strengths and weaknesses of each method. 11.2

DETECTION METHODS

The Transportation Security Administration (TSA) and the Federal Aviation Administration (FAA) have been responsible for ensuring safety of air travel and have invested signiﬁcantly in developing technologies to combat the potential for terrorist attack by explosive devices. This has led to deployment of two types of detectors for screening baggage and people. Explosive detection systems (EDSs) detect bulk explosives hidden in checked baggage and operate using dual X-ray tomography. Explosives trace detectors detect vapor or particles of explosives that have contaminated people and the surface of baggage. ETDs are also used to resolve alarms from EDSs. Currently ETDs are used on a selective basis to screen for personal items and carry-on bags but not for directly screening individuals. The explosives that are of greatest concern for screening purposes are the high explosives (e.g., trinitrotoluene (TNT), pentaerythritol tetranitrate (PETN), cyclotri-methylene trinitramine (RDX)) and their composition forms (e.g., C-4, Semtex, Detasheet), dynamites [e.g., nitroglycerin (NG) and ethylene glycal dinitrate (EGDN)], improvised explosives [e.g., ammonium nitrate (AN), ammonium nitrate fuel oil (ANFO) urea nitrate, triacetone triperoxide (TATP), and black and smokeless powder)], and the International Civil Aviation Organization (ICAO) taggant compounds (e.g., 2,3 dimethyl 2,3 dinitrobutane DMNB, EGDN). However, there are a host of other explosives threats that are in the hands of terrorists [e.g., Octahydro-1,3,5,7-tetranitro-1,3,4,5-tetrazocine HMX, hexamethylene triperoxidediame (HMTD), tetryl, black powder derivatives, etc.]. In fact the U.S. Bureau of Alcohol, Tobacco, and Firearms lists over 200 explosives materials [5]. 11.2.1

Explosives Trace Detection

There is a good reason why there are many different analytical techniques for the detection of explosives. No single detector is suitable for all applications. The level of accuracy is somewhat a function of size, weight, and power requirements. Fixed location applications can use more sophisticated technology, whereas manportability imposes compromises in performance. The following summarizes explosives detection technology that is being pursued for threat detection applications (e.g., early warning systems and emergency ﬁrst responders): •

Mass spectrometry (MS): MS offers high levels of sensitivity and speciﬁcity compared to other technologies for chemical detection. Its traditional disadvantages have been high cost and complexity. Over the last few years, however, the economics have greatly improved and MS is now capable of routine and

DETECTION METHODS

•

•

•

•

•

221

automated operation. MS has been used extensively for forensics applications [6] but only recently has been developed for early warning security screening of explosives. Companies developing deployable MS-based screening include Hitachi, Syagen, Mass Spec Analytical, and Grifﬁn Analytical. Ion mobility spectrometry (IMS) [7]: IMS is similar in concept to MS except that the ions are dispersed by gas-phase viscosity and not by molecular weight. The main advantage of IMS is that it does not use a vacuum system, which greatly reduces the size, cost, and complexity relative to MS. However, the trade-off is that the measurement accuracy is considerably less than MS. This is especially true for complex samples or when screening for a large number of target compounds simultaneously. Companies deploying IMS in security applications include GE/Ion Track and Smiths Detection. New developments, such as ﬁeld asymmetric ion mobility spectrometry (FAIMS) and commercialized by Sionex, are improving on detection accuracy of IMS [8]. Gas chromatography (GC): This method provides high-resolution separation of compounds that are then detected by a variety of means. The highest accuracy is obtained using an MS detector. However, for ETD applications, lower-cost detectors that are speciﬁc to explosives are often used such as electron-capture detection (ECD), chemiluminescence detection (for the NO2 group), or IMS. For other nonspeciﬁc detectors, identiﬁcation is dependent on accurate and reproducible retention times, which can introduce uncertainty in making deﬁnitive identiﬁcations. Companies that have commercialized GC for explosives screening include Thermedics, Scintrex, and Electronic Sensor Technology. Optical spectroscopy: Infrared (IR) and Raman spectroscopy can be used to make positive identiﬁcations; however, it is not well suited to complex mixtures or detecting compounds at very low concentrations. Long-wavelength absorption spectroscopy such as millimeter wave are becoming attractive options as they provide the potential for very high speciﬁcity for volatilized explosives; however, the sensitivity is not very high due to the low absorption cross sections at these wavelengths. Chemical sensors: This approach is attractive for applications requiring a handheld device and generally relies on an array of detectors, each of which responds differently to different compounds. The principal is for each target compound to give a different pattern of responses on the detector array. The detectors can respond to conductivity, wave propagation, ﬂuorescence, or other indications. Chemical sensors are not very speciﬁc or sensitive for complex mixtures and the devices can be prone to high false-positive rates. New technologies: The quest for broad-based explosives detectors that are fast, sensitive, speciﬁc, small, and inexpensive continues. Promising technologies include molecularly imprinted polymer (MIP) sensors. It will take considerable time and work to develop MIPs for a wide range of compounds; however, results on selected test compounds showed mid-parts-per-billion

222

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

(ppb) detection limits with a response time of 10 to 15 min [9]. Another interesting approach of a similar nature involves semiconductor organic polymers that can be made to undergo lasing action at very low thresholds. When speciﬁc compounds, such as TNT, bind to the polymer surface, the lasing action is quenched, giving a sensitive means for detection [10]. Though this method can be very sensitive its speciﬁcity against false positives from other nonexplosives compounds is not yet documented. (Editor’s note: Recent work on this subject is reported in Chapter 9.) 11.2.2

Sampling Methods

Fast and accurate screening for trace levels of explosives and other targeted compounds requires a sophisticated combination of components that work together as a system. The three basic functions of an ETD screening system are: • • •

Sample collection and compound enrichment Selective detection of targeted compounds Accurate measurement of targeted compounds

Sampling is difﬁcult because the vapor pressures for most explosives are very low. For example, the room temperature equilibrium headspace concentration of RDX in air is about 10 pptv (parts per trillion by volume). Collection of vapor is further compounded for explosives that are bound in matrices and wrappers and/or are concealed in wrappings or baggage. The prospects for trace detection of explosives are considered to be better when sampling objects for explosives contamination in the form of particles and residue. Sample collection in airports and other venues is typically done with cloth swipes that the operator rubs across high-touch areas of baggage and other properties. The swipe is used to collect particulate residue. It is then inserted into the detector where it is heated to thermally desorb the residue into the analyzer. Another approach is to collect an air volume that is suspected to contain explosives particles. This method is more practical than using swipes for applications such as screening people for contamination indicative of concealed explosives or inside of containers where physical access is limited. Air sampling is often preceded by a method for dislodging particles from suspected objects, such as the use of air jets to rufﬂe and loosen particles. The surrounding volume of air is then collected and sifted to remove the vast majority of air, while retaining particles, residue, and condensable vapors. One way to concentrate the sample is to use a high-throughput, high-surface-area metal mesh. The mesh is then heated to thermally desorb the collected particles and vapor into the analyzer. This method of collection/concentration/desorption is used in one form or another for screening people for explosives (see Section 5.1). 11.2.3

Quantitative vs. Screening Analysis

Quantitative analysis involves making measurements of samples of unknown concentrations and determining the concentration to some speciﬁed level of accuracy

MASS SPECTROMETRY

223

and precision. This is done by ﬁrst generating a calibration curve that relates the instrument-measured intensity vs. concentration for a standard of the compound being measured. The unknown sample can then be measured and the intensity matched up against the calibration curve to determine the concentration. The accuracy of the instrument is based on how accurate and repeatable (precision) the concentration is measured. Because the unknown concentration can conceivably span a large concentration range, analyzers with large dynamic range are essential for quantitative analysis. Screening analysis does not require determining precise concentrations, but instead must give a reliable indication of whether a target compound is present above some threshold concentration. The accuracy of a screening system is based on the probability of detection PD of the target compound at the threshold concentration relative to the probability of false-positive PFP due to background compounds giving signals that appear indistinguishable from the target signal. Although screening does not require as high a dynamic range as is needed for quantitative analysis, it is still an advantage because it can minimize the effect that a strong background signal can have on weak target signals, and it can help resolve between strong background and strong target signal that might overlap. MS has a very large dynamic range (typically 104 to 106 range from weakest to strongest detectable signal strength) and records intensities that are essentially linear with compound concentration. This makes it a preferred tool for quantitative analysis but as noted above is also beneﬁcial for screening applications. 11.3

MASS SPECTROMETRY

11.3.1

Primer

We brieﬂy summarize the variety of mass analyzers that are currently used for a wide range of chemical analysis applications and then discuss their speciﬁc attributes and how decisions are made as to what analyzer is best suited for a particular application. 11.3.1.1 Quadrupole The quadrupole mass analyzer consists of four cylindrical linear rods that act as a mass ﬁlter for the transmission of ions. Radio frequency (RF) and direct current (dc) voltages are applied to the rods in such a way that the amplitude of the RF and dc controls the mass-to-charge ratio m/z of the ion that is stable in passing through the rods and the amplitude ratio of the RF to dc governs the resolution. The resolution is generally set at unit mass, which means that all ion masses but one are rejected and not detected at any given time during an analysis. The characteristics of a quadrupole mass analyzer include: •

Sensitivity is high in single ion mass mode but diminishes as the number of ions that are simultaneously monitored increases because the duty cycle for collecting any particular ion mass decreases.

224 • •

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

Resolution is moderate and not capable of accurate mass analysis sufﬁcient to resolve different elemental compositions of the same nominal mass. Cost is low relative to other mass analyzers.

The quadrupole mass analyzer is a popular economical choice when known compounds are being analyzed and the ﬁlter can be set to a limited number of ions. 11.3.1.2 Quadrupole Ion Trap The quadruple ion trap operates under the same principal as the linear quadrupole ﬁlter except that geometrically it is wrapped around so that ions do not travel from one end to the other but rather are trapped in an electrode structure consisting of a ring electrode and two endcap electrodes [11]. Similar to the linear quadruple, RF and dc voltages are applied to the device but only to the ring electrode. The dc voltage determines the range of masses that are stable in both a linear quadrupole and a quadrupole ion trap analyzer. The latter device, however, seeks to maximize the range of ions that are stabilized or trapped; this maximum range occurs for a zero dc voltage. Mass analysis of the trapped ions is then achieved by ramping the RF voltage amplitude, which acts to sequentially eject ions of increasing mass that are then measured by a detector. A valuable capability of ion traps is that voltage waveforms can be applied to the endcaps to isolate certain masses and to induce collisional dissociation to fragment the ions. This capability is important for identifying ions by deducing the structure from the ion fragmentation pattern. The characteristics of ion traps include: • • • • •

Sensitivity is medium to high for all ions within the mass stability region. Resolution is moderate and not capable of accurate mass analysis sufﬁcient to resolve different elemental compositions of the same nominal mass. Scan rate is relatively slow. Cost is low to medium relative to other analyzers. It is capable of multiple MS analysis (MS/MS and MSn ) by selective ion fragmentation analysis for structure identiﬁcation.

The commercial ion trap mass spectrometer (ITMS) can be an economical solution for detection of multiple targeted compounds; however, repetition rate is relatively slow especially for MS/MS mode. 11.3.1.3 Time of Flight The principle of time-of-ﬂight mass spectrometry (TOFMS) is very simple and makes use of the energy equation E = mv 2 /2 where m is mass and v is velocity. As illustrated in Figure 11.1, all ions are accelerated to constant energy E. The ions then enter a drift tube where they travel at velocity v. Because each ion mass m travels at a different velocity, the ions of different mass separate; the lighter ones run ahead of the heavier ones. This separation means that each ion mass hits the detector at a different time. Ions of the same

MASS SPECTROMETRY

+

+

E

+

+

Acceleration Region

225

Drift Region

E

Detector

Figure 11.1 Separation of different ion masses by TOFMS.

mass arrive together and give a sharp peak in the TOF mass spectrum. Fast detection methods can measure the arrival times and the simple formula above can be used to transform the times to mass. In practice TOFMS has additional complexities to improve resolution. For example, a spatial spread in the initially formed ions can degrade resolution; however, this spread can be focused using a two-stage acceleration as shown in Figure 11.1. An initial energy spread can also degrade resolution but can be focused using a reﬂectron at the end of the drift tube (not shown in Fig. 11.1). A TOFMS instrument can achieve very high resolution and ion transmission efﬁciency, making it very attractive for many applications. It can also detect all ion masses at once, which is an advantage for sensitive and high-speed detection of multiple compounds. However, there tends to be ion transmission losses in getting ions into the TOF acceleration region because of the need to restrict the initial energy and spatial distribution, despite the focusing methods described above. This makes TOFMS less sensitive than a quadrupole when monitoring only a very few compounds. However, the sensitivity of TOFMS is constant no matter how many ions masses are being probed. The characteristics of TOFMS include: • • • • •

Sensitivity is medium to high for all ion masses. Mass range can extend up to several thousand mass units. Resolution is very high and is capable of accurate mass analysis sufﬁcient to resolve different elemental compositions of the same nominal mass. Scan rate is very high. Cost is medium to high relative to other analyzers.

Time-of-ﬂight MS is a method of choice for a combination of high mass range, high resolution, and speed and for multiple compound analysis. 11.3.1.4 Fourier Transform MS Fourier transform mass spectrometry (FTMS), which is a modern manifestation of ion cyclotron resonance, relies on the collection of ions in a high-vacuum cell and containment with a magnetic ﬁeld. The ions orbit about the magnetic ﬁeld axis. The ion masses and

226

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

abundances are measured by applying an RF signal to excite them to a nonrandom spatial orientation, such that their motions induce an image current in a set of receiver plates. The image current consists of a superposition of oscillations corresponding to the orbit frequencies of the ions. The Fourier transform of the time-dependent signal gives the mass spectrum of the ions. The characteristics of FTMS are: • • • •

Extremely high resolution (e.g., >100,000) Nearly unlimited mass range Slow scan rate Cost very high

Fourier transform MS is an indispensable analyzer for large molecules such as proteins but is not well suited for fast and sensitive screening of target compounds for the purposes of threat analysis. 11.3.1.5 Hybrid MS It is often desirable to combine mass analyzers into hybrid instruments that can perform additional levels of mass analysis. There are two major reasons for doing this: (1) to obtain structure analysis by isolating speciﬁc ion masses and then causing them to break into identiﬁable fragment ions and (2) additional sensitivity can be gained by removing all ions except the target ions, thus creating a clean background on which a known fragment ion can be sensitively observed. The major tandem or hybrid MS instruments include: •

•

•

Triple quadrupole: In the standard conﬁguration, the ﬁrst quadrupole is used to select a speciﬁc ion mass, the second quadrupole is a collision cell that induces the selected ion mass to fragment, and the third quadrupole is used to analyze the fragments. The triple quadrupole is a very popular choice for very sensitive and quantitative analysis of complex mixtures (e.g., drug analysis). Quadrupole/TOF: A version of the triple quadrupole is to replace the third quadrupole with a TOFMS. This provides potentially faster analysis and higher mass resolution. Ion trap/TOF: This combination is advantageous because it enables the MSn capabilities of an ion trap with the high mass accuracy and speed of a TOF analyzer. The ion trap can either be a conventional hyperbolic Penning-type device or a linear trap device. Below we describe the deployment of the former type ion trap/TOF for explosives detection.

11.3.2

QitTof Mass Spectrometry

The quadrupole ion trap, time-of-ﬂight (QitTofMS) analyzer was ﬁrst developed by Lubman [12,13] and co-workers. QitTofMS components were ﬁrst sold commercially by R. M Jordan Company and a full commercial instrument ﬁrst

MASS SPECTROMETRY

227

Ion Optics

Discharge Ionization Source

Ion Trap

DI source • Selectivity toward high electron affinity (−NO2) • Minimum charge competition (1000:1 for known interferents)

Detector

Quadrupole Ion Trap (QIT) • Storage of ions from continuous source • MS/MS capable

Reflectron

TOFMS • Detection of all ions at once • Exact mass if desired (10 mDa)

Figure 11.2 Schematic of the QitTofMS detection system showing the individual components and their respective advantages. DI = discharge ionization.

introduced by Syagen [14,15] and then by Shimadzu [16]. A schematic of the QitTof conﬁguration is shown in Figure 11.2. The principal role of the ion trap is to accumulate ions that are continuously generated by the ionizer. These accumulated ions are then pulsed into the TOFMS analyzer at relatively high repetition rates (e.g., 10 to 100 Hz). Because the TOFMS analysis is performed independently of the ion accumulation into the trap other than the short RF shutoff and pulse out time (about 50 to 100 μs), the duty cycle for collecting ions in the total analysis cycle is >99%. This is to be compared to an ITMS analyzer where duty cycle is typically <50% because the accumulated ions must be scanned out of the ion trap, which typically takes on the order of 100 ms. The main beneﬁts of QitTof relative to other MS systems are: • • •

TOFMS analyzes all ions at once vs. quadrupole mass spectrometry, which can detect only one ion mass at a time. QitTof enables fast TOF detection compared to ITMS only. QitTof enables mass-selective analysis by MSn analysis compared to TOFMS only.

These features translate into high levels of performance in terms of speed, sensitivity, and speciﬁcity. The ionizer is a key component of any MS detection system. The QitTofMS detector that we employ for explosives detection uses a discharge electron attachment source to form negative ions for explosives. This source is a variation of

228

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

IMS

e− Concentration The concentrator strips out air and enriches the trace mixture by reducing volume a million-fold Ionization Negative ionization methods preferentially ionize explosives (O), but can't avoid all background molecules ( )

Ion mobility time

MS

MS/MS Daughter mass

Cl−

Parent mass

Ion mass Primary Analysis High resolution means less overlap of explosives signal (---)with background signal ( )

Secondary Analysis Additional specificity is obtained by selected ion fragmentation

Figure 11.3 Comparison of ion mobility spectrometry (IMS) and mass spectrometry (MS) methods of detection in an overall high-speed screening system.

the glow discharge ionization source developed by McLuckey and co-workers [17–19]. Explosives are very electronegative compounds that readily attach electrons. Most other common compounds that are likely to be sampled in a portal system are less electronegative; hence the ionization process provides a stage of selectivity in differentiating explosives compounds from common background. However, as portrayed in Figure 11.3, other compounds besides explosives can be ionized, and, therefore, it is important to have the capability to further differentiate explosives signals. For our detector, MSn provides this capability. The QitTofMS detection system has the following capabilities: • • • •

High resolution High sensitivity Molecular weight identiﬁcation Secondary conﬁrmation by ion fragment analysis

The detector was designed for automated and unattended operation and for use by nontechnical personnel. We present performance results for the QitTofMS explosives detection system in Section 11.4. 11.3.3

Mass Spectrometry Versus Ion Mobility Spectrometry

Figure 11.3 is a starting point for comparing MS and IMS detectors in an overall screening system. Following the enrichment process described in Section 11.2.2,

MASS SPECTROMETRY

229

the collected sample is introduced into the detection system. For MS and IMS the compounds are further selected for the explosives compounds using an ionization method that preferentially ionizes the compounds of interest. For example, explosives have properties that make them very reactive to negative ions and electrons. IMS generally uses a radioactive 63 Ni source to create a plasma of ions and charges to ionize neutral molecules. For negative ionization it is typical to add a reagent gas, such as methylene chloride, to enhance the ionization of explosives compounds. Because IMS generally operates at atmospheric pressure, these ionization sources also operate at this pressure and are susceptible to unwarranted ion–molecule reactions due to the many molecular collisions. The same ionization methods used for IMS may be used for MS, however, other options are available because MS operates at vacuum conditions and ionization may be chosen to occur anywhere from atmosphere to the vacuum pressure of the MS analyzer. This choice involves a trade-off. Ionization at high pressure affords higher molecular densities and greater yield of ions, however, the ion distribution is prone to thermodynamic effects that can suppress ions of interest. Ionization at low pressure minimizes unwanted ion–molecule reactions, but the molecular densities are lower yielding fewer ions. The glow discharge ionizer described above operates at an intermediate pressure that gives adequate performance with regard to ion yield and minimum ion suppression. As noted above MS analysis is based on attaching a charge to individual molecules to form ions and measuring their molecular weights. The molecular ions often fragment to give fragment ions of varying masses. Extensive fragmentation can be a hindrance to trace detection, especially if the background compounds likewise extensively fragment and interfere with explosives fragment signals. However, if the fragmentation can be limited and even controlled, then it can be used as a powerful conﬁrmatory tool. The detection of a single ion signal may give excellent detection probability, however, it is susceptible to false positives depending on the masses of the background compounds. Parent and fragment ions of explosives compounds span a range of about 300 atomic mass units [e.g., PETN MW = 316, NH3 MW = 17]. MS provides unit mass resolution, so the overall resolution R for detection is about 300. Ion mobility spectrometry is a method that also separates ions in time. Separation is based on the different speeds that ions drift through a viscous gas under the force of an electric ﬁeld. The drift time depends on both molecular weight and size; they do not give a simple measure of molecular weight. The principal advantages of MS over IMS are higher resolution and molecular identiﬁcation. As shown in Figure 11.3, higher resolution minimizes the problem of interference due to overlapping signals from different compounds. Practically speaking, IMS resolution, deﬁned as the ratio of the total span of drift times for target ion signal, and the ion signal width rarely exceeds 30. The measure of molecular weight by MS is a very strong indicator of what the compound is, unlike the measure of mobility times by IMS. For example, TNT has a molecular weight of 227. The probability of another compound having that molecular weight is very small. However, to positively conﬁrm that a detected signal at molecular

230

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

weight 227 is actually TNT, Syagen’s QitTofMS technology can selectively fragment the molecular ion to determine its structure and deﬁnitively identify the detected compound. This secondary analysis portrayed in Figure 11.3 provides an unprecedented level of speciﬁcity for positively detecting targeted compounds. Another means to improve MS resolving power is to achieve resolution sufﬁcient to resolve different elemental compositions of the same nominal unit mass, which effectively can raise the resolving power to several thousand. A recent report from the National Academy of Sciences assessed the relative performance of IMS and MS and concluded that the latter provided from 10 to 10,000 better resolution, which translates into improved accuracy in terms of probability of detection and false-positive rates [20]. The higher end of this resolution range represents high-resolution instruments and MS/MS instruments. A model for assessing detection accuracy in terms of the probability of detection PD and the probability of false-positive PFP is given in Section 11.5. We summarize some of those conclusions here: (i) resolving power is very important for minimizing PFP , (ii) for threat scenarios requiring screening of an increasing number of compounds, PFP will rise, and (iii) to offset an unacceptable false-positive rate, one must increase threshold levels, which degrades detection probability. 11.3.4

Other MS Analyzers Used for Explosives Detection

Detecting explosives in complex environments requires very sensitive and speciﬁc analyzers. Though MS offers excellent detection limits and resolutions (e.g., sensitivity and speciﬁcity), general-purpose laboratory instrumentation is not ideally suited for detecting trace levels of explosives in the presence of large abundances of potential interferent compounds. Furthermore MS has traditionally required a skilled operator not only to acquire data but to interpret it. MS developments have tended to focus on the performance issues and less so on the operational issues. We review MS developments centering on achieving a trace detector for early warning and ﬁrst response as opposed to an analytical instrument for postanalysis of a suspected event. A key strategy has been to focus on ionization methods that preferentially ionize explosives relative to typical background compounds. McLuckey and co-workers developed an atmospheric sampling glow discharge ionization (ASGDI) source that could be operated in negative ion mode to achieve high sensitivity and speciﬁcity for detection of explosives (as also mentioned above) [17–19,21]. They demonstrated coupling it to quadrupole and ion trap MS systems and achieved sensitivities of <10 pptv (and <10 pg) for volatized TNT and RDX (without preconcentration), as well as demonstrated sensitive detection for a wide variety of other explosives compounds. The ASGDI source operates at about 1 torr, which is sufﬁciently high pressure to obtain reasonable densities of explosives, yet low enough to minimize deleterious effects due to ion–molecule reactions. Atmospheric pressure ionization (API) sources, particularly chemical ionization using discharge methods, have been adapted to explosives detection

RESULTS

231

and have achieved exceptional sensitivity. Lee et al. reported a detectable signal for TNT of 10 fg using an API/TOFMS instrument [22]. Work by a SCIEX/British Aerospace team demonstrated detection limits on the order of 1 pg for TNT, RDX, and PETN for an API/triple quad MS system in a program called CONDOR [23,24]. However, API sources, which are also used in IMS, are prone to ion–molecule reactions that can deplete the signal for explosives ions depending on the abundance of certain background compounds. More sophisticated MS conﬁgurations have been demonstrated for selective explosives ionization and detection taking advantage of the high electron attachment cross section for explosives. Chutjian and co-workers developed the reversal electron attachment detection (READ) method, which relies on slowing and reversing an electron beam in a reﬂectron device [25,26]. The electron energy at the turning point is close to zero where the electron-capture ratio is greatest. Laramie and co-workers developed the electron-capture negative ion MS (ECNIMS) system [27]. This system varies the electron energy, which enables differentiation of different types of explosives by monitoring the electron energies at which the dissociated NO2 − ion appears. Because the READ and ECNIMS systems employ ionization in the high-vacuum region of the MS, sensitivity is compromised by the low molecular densities in the ionization region.

11.4

RESULTS

In this section we present results to illustrate the capabilities of the QitTofMS analyzer coupled to an electron attachment source that have been developed in our laboratories for commercial deployment of an explosives detector [29,30]. 11.4.1

Mass Spectral Signatures

The key to an effective trace explosives detection system is the simultaneous detectability of a broad range of compounds at trace levels in the presence of background compounds at much higher abundance. Achieving this capability is aided by a selective ionization process that maximizes the ion signal of compounds of interest while minimizing the ion signal from the more abundant background compounds. Furthermore it is essential that each compound have distinct and highly resolved molecular signatures to enable positive identiﬁcation and differentiation from potential interferents. Figure 11.4 shows electron attachment QitTof mass spectra of common explosives. These compounds are observed to have distinct mass spectral signatures. Because the spectral signals are very sharp, the probability of overlap with other compounds, such as background compounds, is very low relative to lower resolution detectors such as IMS. Nitroaromatic explosives (e.g., TNT, DNT, etc.) generally give molecular ion signal. Nitramines (e.g., RDX and HMX) give distinct peaks at m/z of 176, 129, and 102, but not at the molecular ion. Nitrate esters give the least distinct spectral signatures due to extensive dissociative electron attachment to give a

232

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

TNT

Ion Signal (rel. units)

TATP

PETN

PETN

RDX

50

100

150 Ion Mass (amu)

200

250

Figure 11.4 MS spectral signatures for selected explosives. Each compound gives a unique spectrum. Nitrate esters all have a common peak at m/z 62.

predominant NO3 − ion at m/z 62. PETN leads to signiﬁcant signal at m/z 242 under optimized conditions to give the potential to distinguish some nitrate esters. The spectra in Figure 11.4 were recorded from headspace vapor either at room temperature (TNT, PETN) or elevated temperature (about 50◦ C for RDX). For TNT this corresponds to a saturated headspace vapor pressure of less than 10 ppb. At these levels strong signal is observed with relatively weak signal from room air. Explosives compounds that have been detected by the MS detector with high sensitivity include TNT, ADNT, DNT, NT, TNB, DNB, DMNB, RDX, HMX, EGDN, NG, PETN, and TATP. (see Explosive Deﬁnitions, page 329). 11.4.2

MS/MS Analysis

For molecular screening purposes a single stage of MS is usually sufﬁcient to alert to the presence of explosives compounds while maintaining a sufﬁciently low false-positive rate. However, if a more deﬁnitive identiﬁcation is needed the QitTof mass analyzer allows additional stages of ion mass identiﬁcation through the process of ion mass-selected collision-induced dissociation (CID). In this method the presence of a target ion signal can be used to trigger an excitation waveform in the QIT that excites the target ion to high kinetic energy. When these energetic ions collide with the buffer gas in the trap (typically ambient air), they dissociate to fragment ions. The fragment ion signals are typically very speciﬁc

233

RESULTS

TNT in Silica, CID on the Fly −3

MS/MS

TNT-227 TNT-210

Ion Signal

−2 −1 MS 0

50

100 150 200 Ion Mass (m/z)

250

0.0

0.5

1.0 1.5 Time (s)

2.0

Figure 11.5 TNT detection using MS/MS. (Left) MS and MS/MS spectra. (Right) Molecular and fragment ion signal vs. time showing speed and stability of MS/MS analysis (time in seconds).

to given compounds and greatly increase the conﬁdence in the identiﬁcation of target compounds. Figure 11.5 shows an example of detection of TNT headspace vapor at room temperature in MS mode and in MS/MS mode. In the latter example, the detector is programmed to trigger an excitation waveform when an ion signal at m/z 227 (TNT) is observed. The observation of fragment ions at m/z 210 and 197 is positive identiﬁcation of TNT. 11.4.3

Limits of Detection

The electron attachment source coupled to the QitTofMS analyzer achieves excellent sensitivity. A limit of detection (LOD) was measured for a wide variety of explosives compounds by depositing known quantities of compound onto a heated tube connected to the inlet of the MS detector. The tube was then rapidly heated to desorb the compound, which ﬂowed into the ionizer using room air as a carrier gas. Figure 11.6 shows a linearity plot for TNT. The LOD is deﬁned as the quantity of compound that gives a signal intensity that is a factor of 3 greater than the standard deviation of the background signal (i.e., 3σ ). The LOD for TNT is about 1 pg. Similar measurements give LODs for RDX and PETN of about 5 and 20 pg, respectively. All other compounds tested fall in the LOD range of 2 to 20 pg, except DMNB and ANFO, which are less sensitive. The ion source was described above as operating well below atmospheric pressure, which greatly reduces the incidence of undesirable ion–molecule chemistry. In other words, the ion source is kinetically controlled, which means that the ionization process is preserved by the short time the molecules are in the source and

234

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

TNT m/q = 227 QitTOF MS only

−0.40

LOD = 0.9 pg

Int. Signal

−0.30

−0.20

−0.10

3s 0.00

0

20

40 Mass (pg)

60

80

Figure 11.6 Detector sensitivity. 3σ limits of detection are 1 pg for TNT, 5 pg for RDX, and 20 pg for PETN.

that the ions typically do not have time to exchange charge or reach thermodynamic equilibrium. Because of this property, the detector is much less susceptible to masking agents than are detectors that use atmospheric pressure ionizers. To illustrate this advantage, Figure 11.7 shows the detection of TNT by itself and in the presence of a masking agent at 5000× the mass of the TNT. This shows that a large abundance of masking agent has minimal effect on the detectability of TNT.

11.5 11.5.1

DETECTION ACCURACY—A MODEL False-Positive Analysis

Resolving power R is a measure of speciﬁcity and is a primary factor contributing to the probability of false positives. The salient issue is the probability that a background signal will overlap with a target explosives signal. We can qualitatively express this probability using a Poisson distribution function of the form P (a) = e−m

ma a!

(11.1)

where m in this case is the fraction of all detection channels that are target channels, and P (a) is the probability that the presence of background signals

DETECTION ACCURACY—A MODEL

235

−10

Int. (arb units)

−8

−6

−4

−2

0 210

220

230

240

250

260

Mass

Figure 11.7 Spectrum showing injection of 2 ng of TNT (upper) and injection of 2 ng and 10 μg of 1-nitropyrene.

will fall on target channels by a times. A value of a > 1 is only relevant if there is more than one background signal. The value of m is given by m = Nb

Ne R

(11.2)

where Ne is the number of ion masses being detected for the target explosives compounds, Nb is the number of all background ion masses whose intensity exceeds the threshold intensities for the target explosives signals, and R is the detector resolution. PFP is related to P (a) by summing over the probability of background signals overlapping with target signals one, two, three . . . (a = 1, 2, 3 . . .) times by the expression PFP = P (a) (11.3) a

The probability of a background signal overlapping with a target signal once [e.g., P (a = 1)] will normally be the dominant term in Eq. (11.3), but the higher

236

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

a terms can become signiﬁcant if the number of background signals Nb becomes large. The analysis of PFP using the above Poisson analysis is approximate because it does not assume different threshold levels for different explosives signals nor strategies based on detection of multiple ion signals for a single explosive compound. However, the Poisson distribution is very useful for making qualitative assessments of how PFP depends on resolving power R and on the number of explosives compounds that can be screened simultaneously. 11.5.2

Receiver Operator Characteristics (ROC)

Sensitivity and speciﬁcity are the ﬁgures of merit for screening detectors. Sensitivity is related to the limit of detection and is important for maximizing PD for some required detection concentration. Speciﬁcity is related to how well the instrument can discriminate between different compounds and is important for minimizing PFP . The PD and PFP response is strongly dependent on the target threshold concentration required. By setting the threshold level low (to very sensitive levels), the PD will increase, however, PFP will also increase because the probability of signal noise or background exceeding the threshold increases with decreasing threshold level. Raising the threshold level to reduce PFP , however, increases the risk of missing a target compound (decreased PD ). A convenient way to represent the dependence of PD and PFP on threshold level is by making a series of measurements and plotting PD vs. PFP at the required threshold concentration. This type of signal detection analysis was advanced during World War II for the monitoring of radar images where it was necessary for radar receiver operators to make judgments on whether a blip was due to enemy or friendly objects, or just noise. Consequently the analysis is known as “Receiver Operator Characteristics” or ROC. Figure 11.8 illustrates the relationship between linearity plots of intensity (and intensity variation) vs. concentration to ROC plots or what are more often called ROC curves. Basically, what is desired is for the distribution of signal level at a required concentration to be completely separated from the distribution of signal level for the background. ROC curves can be generated from a series of measurements at one concentration or a series of concentrations as shown in Figure 11.8. The ability to discriminate target signal from background constitutes speciﬁcity. Figure 11.9 illustrates by way of ROC curves the characteristics of PD vs. PFP for prototypical high- and low-speciﬁcity detectors. 11.5.3

MS vs. IMS Accuracy

It is possible to assess the relative susceptibility of MS and IMS to false positives using the qualitative Poisson analysis in Section 11.5.1. We begin by assuming typical resolutions for MS and IMS of R = 300 and 30, respectively. First we evaluate the detection accuracy of MS and IMS for detection of varying numbers of target explosives ion signals Ne . For simplicity we assume that there is only one background ion signal that exceeds the threshold intensity for the target ion

DETECTION ACCURACY—A MODEL

237

1.0

Concentration

0.5

PD = 0.9

0

0.5

1.0

1.0

PD

0

PD = 0.5

PD = 0.9

0.5 PFP

Signal

0

1.0

1.0

0

0.5

PD = 0.5

PFP

1.0

0

0.5

PD = 0.9

0

0.5

PD = 0.5

Concentration

Figure 11.8 Basis for taking signal data at different concentrations to generate ROC curves for showing the PD and PFP dependence.

signals. We will then be assessing the probability that this background ion signal overlaps with a target ion signal to give a false positive. Keep in mind that the number of background ion signals that can potentially trigger a false positive is generally greater than one and is a function of the level of the threshold intensity used for detecting target compounds, as will be considered below. Figure 11.10 summarizes the value of PFP [∼P (a), Eq. (11.3)] for a small and large number of targeted ion masses (Ne = 3 and 10, respectively). These results show that with regard to false positives an analyzer with low resolving power may be ﬁne for a limited number of targeted compounds (e.g., Ne = 3); however, PFP becomes excessive for an increasing number of target signals. Higher resolution is clearly very important for enabling a greater number of ion signals that can be simultaneously monitored while still maintaining acceptable PFP values.

238

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

ROC Curves

1.0

0.8

0.6 xx

High Specificity Low Specificity 0.4

0.2

0.0 0.0

0.2

0.4

0.6

0.8

1.0

PFP

Figure 11.9 Illustrative example of ROC curves for a screening system with high and low speciﬁcity.

We now consider how PFP depends on the number of background signals Nb . This number will depend on the complexity of the sample but also as noted above is directly related to the threshold intensity of the target ion signals chosen for analysis. Figure 11.11 shows the dependence of PFP on Nb for the cases of R = 30 and 300 and Ne = 3 and 10. Not surprisingly, high R is directly correlated with low PFP . By increasing the threshold levels, the value of Nb will decrease as will PFP . However, as illustrated in Figure 11.9, the detection probability PD will also decrease. This trade-off leads to an important judgment call when setting screening instruments. In situations where the frequency of true positive events is very low, it is tempting to increase the threshold intensities in order to minimize the inconvenience of false positives. However, if a threat scenario demands a high P D , the importance of high resolving power becomes very clear. The above analysis may be summarized as follows: • • •

The false-positive rate is a function of the number of background signals. The probability of a background signal triggering a false positive becomes less likely for analyzers with higher resolution. The greater the number of explosives that are targeted for detection, the greater the probability that background signal will trigger false positives.

DETECTION ACCURACY—A MODEL

239

0.30

0.25

R = 30 R = 300

PFP

0.20

0.15

0.10

0.05

0.00

Ne = 10

Ne = 3

Figure 11.10 Dependence of PFP on number of targeted explosives ion signals Ne and resolution R. 0.8 R = 30; Ne = 10 R = 30; Ne = 3

0.7

R = 300; Ne = 10 R = 300; Ne = 3

0.6

PFP

0.5 0.4 0.3 0.2 0.1 0.0 0

1

2 Nb

3

4

Figure 11.11 Dependence of PFP on number of background ion signals Nb and resolution R.

240 •

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

False positives may be reduced by raising the threshold intensity for alarming on a target explosive; however, this comes at the expense of a lower detection probability.

11.6 11.6.1

APPLICATIONS Personnel Screening

The lack of a capability to screen for explosives hidden on an individual is a major vulnerability in aviation and general security. Personal privacy issues and perceived health risks have deterred the use of bulk detectors, such as X ray, X-ray backscatter, and millimeter wave, for screening of individuals for concealed explosives. Consequently, the TSA is focused on trace detection as the best solution for passenger screening in airports. The TSA has determined that individuals carrying as little as 1 lb of concealed explosives get sufﬁciently contaminated to be detectable by portal devices that use trace detectors. The level of contamination on an individual’s exterior clothing that can be routinely detected by the best portal devices is about 1 μg or about 1 part in 109 of the explosive mass. The method of sampling individuals is essentially “snifﬁng” their exterior (e.g., clothing and skin) for contamination from handling explosives. Because the vapor pressures of explosives are very low, the sample that is collected consists of particles and residue. Though it is desirable that the sampling of people not make contact with the subject, a contact portal is a viable option and has been demonstrated. Contact is usually done by using paddles that act by vacuuming particles from clothing and ﬂowing the sample to a concentrator [30,31]. Noncontact collection is usually done by impinging the individual with air jets to shake loose the particles and then ﬂowing the large air volume in the portal through a concentrator to remove the collect the particles [28,29,32]. Noncontact is obviously the preferred method provided it achieves sufﬁcient collection efﬁciency from people. Different methods of sample collection of the dislodged particles have been developed in current portals. One method developed by Sandia National Laboratories involves pulling the air volume through a metal mesh concentrator. The sample is then thermally desorbed by passing a current through the metal mesh [32]. The vaporized sample is then analyzed by a detector such as IMS or MS. Another method relies on the natural convective boundary layer surrounding individuals. This air volume rises and is collected and detected at the top of the portal opening [33]. Portals have been developed with and without doors. Open portals without doors permit easy passage for passengers and a less claustrophobic environment. Closed portals with doors offer access control and more efﬁcient sample collection. Figure 11.12 shows a portal with doors and Figure 11.13 shows the internal components for collecting and analyzing sample. For this particular portal the airﬂow is done in a closed loop using a blower and concentrator without interaction with external air that can dilute the sample concentration.

APPLICATIONS

241

Figure 11.12 Picture of a MS-based portal for screening people. This device uses doors to control access. Air duct Control electronics and computer Rough pump MS electronics MS Blower st

1 stage concentrator

Figure 11.13 Internal components of the MS-based portal. First stage refers to the concentrator.

242

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

Unfortunately, there are not many publicly released test results for portals (test results are generally classiﬁed). Testing is typically done by two methods. TSA testing is conducted using patches of cloth upon which calibrated quantities of explosives are deposited. These patches are adhered to various locations on people to test for collection and detection efﬁciency. The quantities of explosives placed on these patches represent levels of contamination the TSA has determined are typically for bomb carriers. Another method primarily used by overseas users is to place bulk explosives on individuals and test for the detection of actual contamination. In order for the test to be realistic, the concealed explosives must be on an individual sufﬁciently long for contamination to occur, though generally the “soak time” is only several minutes, rather than the more extended periods expected of an actual bomber. 11.6.2

Other Applications

Explosive trace detectors using IMS are ubiquitous in airports with over 6000 units purchased for U.S. airports alone since 9/11. These detectors use cloth swipes to collect sample as described earlier. Recently, Hitachi introduced an MS version of the ETD using the swipe method. Little data is available at this time. Another novel application of MS is the screening of boarding passes for explosives (and narcotics) [34]. The MS detector in that case uses APCI and a triple quadrupole MS analyzer operating in MS/MS mode. Detection is made by monitoring parent and characteristic fragment ions (e.g., those that differ in mass by the NO2 or NO fragment). The boarding pass screener was shown to operate at a throughput rate of up to 1000 boarding passes per hour and was able to detect between 10 and 50 pg of explosives residue from the surface of the card depending on the compound monitored. 11.7

SUMMARY AND CONCLUSION

Mass spectrometry is often referred to as the gold standard for molecular detection and identiﬁcation. Traditionally a lab-based research tool, MS is now being developed for automated operation in rugged environments by technically unskilled operators. It is therefore inevitable that MS will ﬁnd a major role in security applications, particularly where detection accuracy outweighs the requirements for size and cost. In this chapter we compared IMS to MS, but in fact the determining factor regarding which detector to deploy depends on application. When inexpensive, handheld analyzers are required, IMS is the best choice. For stationary applications where compactness is less important then the superior performance of MS offers compelling beneﬁts. ACKNOWLEDGMENTS

We are grateful to Dr. Tom Chamberlain and Dr. Richard Lareau of the TSA/FAA for their support of the Syagen/Sandia collaboration. Finally this work was made

REFERENCES

243

possible by the exceptional contributions of the Sandia team consisting of Kevin Linker, Chuck Rhykerd, and Frank Bouchier. REFERENCES 1. Moore, D. S. Instrumentation for trace detection of high explosives. Rev. Sci. Instrum. 75, 2499–2512 (2004). 2. Steinfeld, J. I. and J. Wormhoudt. Explosives detection: A challenge for physical chemistry. Annu. Rev. Phys. Chem. 49: 203–32 (1998). 3. Kahn, S. M., Ed. Proc. 1st Proc. Int. Symp. Explosives Detection Technology, FAA, Atlantic City, 1992. 4. Makky, W. H., Ed. Proc. 2nd Explosive Detection Technology Symp. & Aviation Security Technology Conf. FAA, Atlantic City, 1996. 5. Yinon, J. Mass Spectrom. Rev. 10, 179 (1991). 6. Buckles, B. A. Fed. Reg. 67, 20864 (2002). 7. Eiceman, G. A. and Z. Karpas. Ion Mobility Spectrometry. CRC Press, Boca Raton, FL, 1994. 8. Spangler, G. E. and R. A. Miller. Int. J. Mass Spectrom. 214, 95–104 (2002). 9. Jenkins, A. L., S. Y. Bae, J. M. Lochner, and J. L. Jensen, in Proc. 2nd Joint Conf. on Point Detection for Chem. Bio. Def., Williamsburg, VA, 2004. 10. Rose A., Z. G. Zhu, C. F. Madigan, T. M. Swager, and V. Bulovic. Nature 434, 876 (2005). 11. March, R. E. and J. F. J. Todd Practical Aspects of Ion Trap Mass Spectrometry, Vol 1. CRC Press, New York, 1995. 12. Michael S. M., B. M. Chien, and D. M. Lubman. Anal. Chem. 65, 2614 (1993). 13. Michael, S. M., B. M. Chien, and D. M. Lubman. Rev. Sci. Instrum. 63, 4277 (1992). 14. Syage, J. A. and M. A. Evans. Spectroscopy 16, 14 (2001). 15. Syage, J. A., B. J. Nies, M. D. Evans, and K. A. Hanold. J. Am. Soc. Mass Spectrom. 12, 648 (2001). 16. Ding, L., E. Kawatoh, K. Tanaka, A. J. Smith, and S. Kumashiro. Proc. SPIE 3777, 144 (1999). 17. Asano, K. G., D. E. Goeringer, and S. A. McLuckey. Anal. Chem. 67, 2739 (1995). 18. Chambers, D. M., S. A. McLuckey, and G. L. Glish. Anal. Chem. 65, 778 (1993). 19. McLuckey, S. A., G. L. Glish, K. G. Asano, and B. C. Grant. Anal. Chem. 60, 2220 (1988). 20. National Research Council, Opportunities to Improve Airport Passenger Screening with Mass Spectrometry. National Academies Press, Washington DC, 2003. 21. McLuckey, S. A., G. J. Van Berkel, D. E. Goeringer, and G. L. Glish. Anal. Chem. 66, 689A, 737A (1994). 22. Lee, H., G. E. D. Lee, and M. L. Lee, Proc. Int. Symp. Explosives Detection Technol., S. M. Khan, Ed. Atlantic City, FAA, 1992, p. 619. 23. Stott, W. R., W. R. Davidson, and R. Sleeman. Proc. SPIE 2092, 53 (1994). 24. Bennett, G. R., Sleeman, W. R. Davidson, and W. R. Stott. Proc. SPIE 2276, 363 (1994).

244

MASS SPECTROMETRY FOR SECURITY SCREENING OF EXPLOSIVES

25. Boumsellek, S., S. H. Alajajian, and A. Chutjian. J. Am. Soc. Mass Spectrom. 3: 243 (1992). 26. Boumsellek, S. and A. Chutjian. Anal. Chem. 64, 2096 (1992). 27. Laramee, J. A., C. A. Kocher, and M. L. Deinzer. Anal. Chem. 64, 2316 (1992). 28. Syage, J. A., K. A. Hanold, and M. A. Hanning-Lee. Mass spectrometry based personnel screening system. Proc. INMM—42nd Annual Meeting, Palm Springs, CA, 2001. 29. Chamberlain, T., K. Hanold, M. Hanning-Lee, Y. Liu, J. Syage, K. Linker, C. Rhykerd, and R.Bouchier. Multi-threat mass spectrometer detector. Proc. of the ICAO Workshop, DERA, UK, 2000. 30. Wendel, G. J., E. E. A. Bromberg, M. K. Durfee, and W. Curby, in W. H. Makky, Ed. Proc. 2nd Explosive Detection Technology Symp. & Aviation Security Technology Conf. FAA, Atlantic City, 1996, p. 181. 31. McGann, W., A. Jenkins, and K. Ribeiro. in S. M. Kahn, Ed. Proc. 1st Proc. Int. Symp. Explosives Detection Technology, FAA, Atlantic City, 1992, p. 518. 32. Parmeter, J. E., K. L. Linker, C. L. Rhykerd, and D. W. Hannum. in W. H. Makky, Ed. Proc. 2nd Explosive Detection Technology Symp. & Aviation Security Technology Conf. FAA, Atlantic City, 1996, p. 187. 33. Settles, G. S., H. C. Ferree, M. D. Tronosky, Z. M. Moyer, and W. J. McGann. 3rd International Symposium on Explosive Detection and Aviation Security, FAA, Atlantic City, 2001. 34. Sleeman, R., S. L. Richards, W. R. Stott, W. R. Davidson, J. G. Luke, B. J. Keely, I. Fletcher, and A. Burton. Proc. of the 50th ASMS Conf. on Mass Spectrom. and Allied Topics, 2002, Orlando, FL, 2002.

CHAPTER 12

EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS THOMAS THUNDAT Oak Ridge National Laboratory

12.1

INTRODUCTION

Sensitive and selective detection of explosive vapors using inexpensive sensors is a formidable task due to their low vapor pressure and the large number of materials that can be used as explosives. Recent advances in fabrication of microcantilever beams capable of detecting extremely small stress and adsorbed mass offer exciting opportunities for developing miniature sensors for explosive vapors. Three unique approaches of detecting of explosive vapors are demonstrated. In the ﬁrst approach a cantilever beam coated with a selective layer undergoes bending due to adsorption of explosive molecules. The second approach utilizes the resonance frequency variation due to mass of the adsorbed explosive molecules. The resonance frequency variation is due to mass loading while adsorption-induced cantilever bending is due to a differential stress due to molecular adsorption. In the third approach deﬂagration of explosives adsorbed on the cantilever by rapid heating causes the cantilever to bend. Deﬂagration of adsorbed explosive molecules causes the cantilever to bend due to released heat while its resonance frequency decreases due to mass unloading. Selective, highly sensitive sensors that can detect trace amounts of explosive vapors in real time are needed to help combat terrorism [1–4]. Trace detection of explosives, however, is a formidable task. Selectivity is difﬁcult to achieve because many chemicals can be used as explosives, and they differ from each other in their chemical properties. The extremely small vapor pressures of the explosives make it challenging to achieve highly sensitive vapor-based detection. Also, because the terrorist threat is very broad, combating it requires widespread deployment of inexpensive, low-power-consuming sensors. Therefore, devices Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

245

246

EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS

that are capable of high sensitivity and low-cost mass production are of great interest. Presently available technology for vapor detection does not satisfy many of these requirements. Mass-produced cantilever sensors, however, have the potential to satisfy the conditions of selectivity, sensitivity, miniature size, low power consumption, and real-time operation [5, 6]. Microcantilevers are micromachined from silicon or other materials and can easily be fabricated in multiple-element arrays. They resemble miniature diving boards measuring 100 to 200 μm long by about 20 to 40 μm wide by 0.3 to 1 μm thick and having a mass of a few nanograms. Their primary advantage originates from their sensitivity, which is based on the ability to detect their motion with subnanometer precision. Microcantilever sensor technology is an upcoming sensing technique with broad applications in chemical, physical, and biological detection [7–9]. Microcantilevers have many different modes of operation. All the modes rely on cantilever motion caused by variations in surface stress, adsorbed mass, thermal properties (calorimetry), and damping characteristics. Here we will focus on two modes of operation: adsorbed mass (resonance frequency) and surface stress (bending). The resonance frequency of a cantilever varies as a function of mass loading due to molecular adsorption. This mode of operation is very similar to the operation of other gravimetric sensors, such as quartz crystal microbalance (QCM) and surface acoustic wave (SAW) transducers. The number of molecules adsorbed on the surface can be increased by coating the cantilever with selective layers with high partition coefﬁcients. The bending mode of operation is unique to thin structures; adsorption-induced forces cause the cantilever to bend if the adsorption is conﬁned to a single side. Differential adsorption on the surfaces results in differential stress. Microcantilever-based sensing satisﬁes many requirements for an ideal explosive sensor. Microcantilever sensors have extremely high sensitivity and are compatible with array arrangement for simultaneous detection of multiple analytes. They have the advantage of low power consumption and miniature size. However, microcantilever sensors lack chemical selectivity. In general, chemical selectivity is accomplished using selective layers based on receptor-based detection. Receptor-based sensing of small molecules is still in its infancy, and more work is needed in design and synthesis of highly selective receptors for small molecules. Microcantilever sensors can be operated in modes in which receptor-based coatings are not needed; for example, deﬂagration of adsorbed energetic molecules can induce a measurable response [10]. Since cantilevers can be made extremely sensitive to temperature using bimaterial effect, calorimetric methods can be carried out on cantilevers with adsorbed molecules [11, 12]. Exposing the temperature-sensitive cantilevers with adsorbed species to different infrared (IR) wavelengths in a sequential fashion creates mechanical signatures that mimic the IR absorption spectra of the analyte [13].

MODES OF OPERATION AND THEORY

12.2

247

MODES OF OPERATION AND THEORY

The microcantilever is an ideal displacement sensor. The ability to detect motion of a cantilever beam with nanometer precision makes the cantilever ideal for measuring bending and resonance frequency. Cantilever bending can be related to adsorption/desorption of molecules through adsorption forces, thermal effects, and electrical and magnetic forces. The resonance frequency of a cantilever is directly related to changes in its inertial mass. Bending due to mass effects (gravitational forces) is extremely small. Here we will discuss two modes of operation that are relevant to detection of explosives. As molecular reactions on a surface are ultimately driven by free energy reduction of the surface, the free energy reduction leads to a change in surface stress. Although they would produce no observable macroscopic change on the surface of a bulk solid, the adsorption-induced surface stresses are sufﬁcient to bend a cantilever if the adsorption is conﬁned to one surface. Adsorptioninduced forces, however, should not be confused with bending due to dimensional changes such as swelling of thicker polymer ﬁlms on cantilevers. The sensitivity of adsorption-induced stress sensors can be orders of magnitude higher than those of frequency-variation mass sensors (for resonance frequencies in the range of tens of kilohertz). Moreover, the static cantilever bending measurement is ideal for liquid-based applications, where frequency-based cantilever sensors undergo huge viscous damping reducing sensitivity [14]. Microcantilever deﬂection changes as a function of adsorbate coverage when adsorption is conﬁned to a single side of a cantilever (or when there is differential adsorption on opposite sides of the cantilever). Since we do not know the absolute value of the initial surface stress, we can only measure its variation. A relation can be derived between cantilever bending and changes in surface stress from Stoney’s formula and equations that describe cantilever bending [15]. Speciﬁcally, a relation can be derived between the radius of curvature of the cantilever beam and the differential surface stress: 1 6(1 − ν) = δs R Et 2

(12.1)

where R is the cantilever’s radius of curvature, ν and E are Poisson’s ratio and Young’s modulus for the substrate, respectively, t is the thickness of the cantilever, and δs = σ1 − σ2 is the differential surface stress. Surface stress, σ , and surface free energy, γ , can be related using the Shuttleworth equation [16]: ∂γ σ =γ + (12.2) ∂ε where σ is the surface stress. The surface strain ∂ε is deﬁned as ratio of change in surface area, ∂ε = d A/A. Since the bending of the cantilever is very small compared to the length of the cantilever, the strain contribution is only in the part-per-million (10−6 ) range. Therefore, one can easily neglect the contribution

248

EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS

from surface strain effects and equate the free energy change to surface stress variation [17]. The radius of curvature, R, of cantilever bending is related to the deﬂection, z, and the length of the cantilever, L. Using Eq. (12.1), a relationship between the cantilever displacement and the differential surface stress is obtained: z=

1 L2 3L2 (1 − ν) = δs 2 R Et 2

(12.3)

Therefore, the deﬂection of the cantilever is directly proportional to the adsorption-induced differential surface stress. Surface stress has units of N/m or J/m2 . Equation (12.3) shows a linear relation between cantilever bending and differential surface stress. The minimum detectable signal for cantilever bending depends on the geometry and the material properties of the cantilever. For a silicon nitride cantilever of 200 μm long and 0.5 μm thick, with E = 8.5 × 1010 N/m2 and ν = 0.27, a surface stress of 0.2 mJ/m2 will result in a deﬂection of 1 nm at the end. Because a cantilever’s deﬂection strongly depends on geometry, the surface stress change, which is directly related to molecular adsorption on the cantilever surface, is a more convenient quantity of the reactions for comparison of various measurements. 12.2.1

Resonance Frequency

For a rectangular diving-board-shaped cantilever, the spring constant for vertical deﬂection as derived for a load at the end is given by Eq. (12.4): K=

Ewt 3 4L3

(12.4)

where E is the modulus of elasticity for the cantilever material and w, t, and L are the width, thickness, and length of the beam, respectively. The fundamental resonance frequency can be written as 1 K E t f = (12.5) = 2π m∗ 2π(0.98)L2 ρ where ρ is the density of the cantilever material and m∗ is the effective mass of the cantilever. The effective mass is related to the mass of the beam, mb , through the relation m∗ = nmb , where n is a geometric parameter. The value of n for a rectangular cantilever is 0.24. Assuming that surface-adsorbed molecules have no inﬂuence on the spring constant, K, the mass of the adsorbed material can be determined from the initial and ﬁnal resonance frequency and the initial mass of the cantilever as f12 − f22 m = 2 m f1

(12.6)

MODES OF OPERATION AND THEORY

249

where f1 and f2 are the initial frequency and the ﬁnal frequency, respectively, and m and m are adsorbed mass and initial mass of the cantilever, respectively. Here we have assumed that the adsorption is uniform on the surface of the cantilever. From Eq. (12.2) a mass resolution for a cantilever with a mass 25 ng and a resonance frequency of 50 kHz could be estimated to be a picogram. Mass detection sensitivity, therefore, is not the strength of a microcantilever sensor. Increasing the resonance frequency and decreasing the mass could improve the mass resolution much further. For nanocantilevers, mass sensitivity could be orders of magnitude higher, and therefore, its main advantage. In certain cases when adsorption is uniform over the entire cantilever, the spring constant of the cantilever can change. From Eq. (12.5) it is clear that the resonance frequency can change with changes in mass as well as changes in spring constant in a competitive fashion: ∗

d f (m , K) =

∂f ∂m∗

∗

dm +

∂f ∂K

f dK = 2

dK d m∗ − ∗ K m

(12.7)

In most cases, the change in spring constant due to molecular adsorption is small and the term related to variation in spring constant can be ignored. This term plays a signiﬁcant role when thick layers of polymers (compared to the thickness of the cantilever) are used as selective layers. The change in spring constant can be due to a number of causes, such as changes in the elastic constant of the surface ﬁlm and dimensional changes of the cantilever or coatings.

12.2.2

Thermal Motions of a Cantilever

Since the cantilevers are very small structures, they execute thermal (Brownian) motion. The longer the cantilever, the more sensitive it is for measuring surface stresses. However, increasing the length also increases the thermal vibrational noise of the cantilever, which from statistical physics is [18, 19] δn =

2kB TB πKf0 Q

(12.8)

Here, kB is the Boltzmann constant (1.38 × 10−23 J/K), T is the absolute temperature (300 K at room temperature), B is the bandwidth of measurement [typically about 1000 Hz for direct current (dc) measurement], f0 is the resonant frequency of the cantilever, and Q is the quality factor of the resonance, which is related to damping. It is clear from Eq. (12.8) that lower spring constant, K, produces higher thermal noise. This thermal motion can be used as an excitation technique for resonance frequency mode of operation.

250

12.2.3

EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS

Thermal Effects—Deﬂagration

If adsorption or desorption of molecules involves variation in temperature, it can be detected by the bending of a bimaterial cantilever. These real-time thermal effects, however, are transient, and, at present, the tools do not exist to interpret them as useful signals. However, since the thermal mass of a cantilever is extremely small, it can be heated to hundreds of degrees centigrade in a millisecond. This effect can be used for calorimetric detection of molecules. The lower thermal mass of the cantilever can be further exploited for detection of energetic molecules by using deﬂagration. Rapid heating of a cantilever on which explosive molecules are adsorbed results in deﬂagration of the adsorbed molecules, resulting in additional heat-induced bending of the cantilever. The bending direction of the cantilever can be used to distinguish between exothermic and endothermic processes. Scanning force microscopy and optical microscopy have both shown that adsorbed explosive molecules form islands on cantilever surfaces. The sizes of these circular islands (drops) increase with increased adsorption. An optical beam deﬂection method can be used to monitor cantilever bending that results from deﬂagration of the explosive droplets.

12.3 12.3.1

APPARATUS Cantilevers

Cantilevers are usually microfabricated from silicon by using conventional photomasking and etching techniques. Typical dimensions of a cantilever are 100 μm in length, 20 μm in width, and 1 μm in thickness. Silicon and silicon nitride cantilevers and cantilever arrays that utilize optical beam deﬂection for signal transduction are commercially available. Piezoresistive cantilever arrays are also commercially available. Piezoresistive cantilevers are 120 μm in length, 1 μm in thickness, and 40 μm in width. 12.3.2

Excitation Techniques

The cantilever is excited into resonance by electrically exciting the piezoelectric cantilever mount. The frequency of the excitation wave is scanned in a given frequency range, and the frequency of maximum cantilever amplitude is taken as the resonance frequency. The frequency spectrum of the cantilever response shows the fundamental frequency as well as the harmonics of cantilever vibration. The cantilevers, however, also resonate in response to ambient conditions such as room temperature or acoustic noise without requiring any external power. 12.3.3

Readout Techniques

There exists a number of readout techniques based on optical beam deﬂection, variation in capacitance, piezoresistance, and piezoelectricity. Piezoelectricity is

APPARATUS

251

more suited for a detection method based on resonance frequency than the method based on cantilever bending. The capacitive method is not suitable for liquidbased applications. The most common readout technique for cantilever motion is the optical beam deﬂection technique, which can detect cantilever motion with sub-Angstrom resolution. In the optical beam deﬂection technique a light beam from a diode laser is focused at the end of the cantilever. The reﬂected beam is then allowed to fall on a position-sensitive detector. The cantilever motion can be recorded by laser deﬂection with the output of the photodetector sent to a frequency-counting device. The bending of the cantilever changes the radius of curvature of the cantilever, resulting in a large change in the direction of the reﬂected beam. A single detection technique can be used for measuring the resonance frequency, resonance amplitude, and bending of the cantilever simultaneously. The alternating current (ac) signal in the position-sensitive detector corresponds to the frequency of vibration of the cantilever, while the dc signal is proportional to cantilever bending. However, since the position-sensitive detector measures curvature rather than displacement, higher modes of cantilever resonance will have higher amplitudes. Recently, a method based on piezoresistance has been gaining attention. Doped silicon exhibits a strong piezoresistive effect [20]. The resistance of a doped region on a cantilever can change reliably when the cantilever is stressed with deﬂection [21]. Boisen et al. developed piezoresistive cantilever sensors with integrated differential readout [22]. Each cantilever had a thin, fully encapsulated resistor made of doped Si fabricated, on top of which the resistance would change due to any load on the cantilever. Each sensor was composed of a measurement cantilever and a built-in reference cantilever, which enabled differential signal readout. The two cantilevers were connected in a Wheatstone bridge, and the surface-stress change on the measurement cantilever was detected as the output voltage from the Wheatstone bridge. The electrical readout technique has several advantages over optical beam deﬂection methods. For example, optical beam deﬂection probes the bending of the free end of the cantilever. It is assumed that the bending is uniform along the length of the cantilever. The piezoresistive method, on the other hand, measures the integrated bending of the cantilever. Piezoresistive cantilevers can be encapsulated in silicon nitride for operation under solution, thus avoiding the longstanding problems associated with optical path lengths and variations in refractive index. In addition, because no external optical components are required, the electronic readout is more amenable to miniaturization, and is ideal for portable devices. An electronic readout is compatible with array arrangements because both cantilevers and readout circuits can be fabricated simultaneously on the same chip. However, currently available piezoresistive cantilever sensors are an order of magnitude less sensitive than those using optical readout techniques. This discrepancy in sensitivity, however, is vanishing because of recent progress in piezoresistive cantilever development.

252

12.3.4

EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS

Selectivity

The key to achieving chemical selectivity using microcantilevers is the ability to functionalize one surface of the microcantilever with receptive molecules so that explosive molecules will preferentially bind to the treated surface. Choosing receptor molecules that can provide highest afﬁnity, therefore, can control the selectivity of detection. Another important requirement for a sensor system is fast regeneration (recovery), so that the sensor can be used repetitively. One way to achieve chemical selectivity is to modify the cantilever surface with self-assembled monolayers (SAMs) with functional head groups that will bind to explosive molecules. Carboxyl-terminated SAMs on a gold surface are acidic (pKa 5–7) and bind strongly with basic groups such as nitro-substituted molecules of explosive vapors [23]. SAMs of 4-mercaptobenzoic acid (4-MBA; also known as thiosalicylic acid) on gold are stable and efﬁciently provide surface carboxyl (–COOH) groups for acid–base reactions [24]. It is expected that highly oriented monolayers of 4-MBA could provide highly efﬁcient substrates for explosive molecule adsorption. Under proper conditions high surface coverage of 4-MBA could be achieved on cantilever surfaces. The formation of a 4-MBA SAM on the gold surface of the cantilever was achieved by immersing the cantilever into a 6 × 10−3 M solution of 4-MBA (97%, from Aldrich Chemical Company) in absolute ethanol for 2 days. Prior to SAM immobilization, the cantilevers were thoroughly cleaned and coated on one side with 3 nm of titanium followed by 30 nm gold. Upon removal from the solution, the cantilever was rinsed with ethanol and then dried before use in the experiments. The monolayer coating was shown to be quite stable for several months under normal operating conditions. Figure 1 shows the schematic diagram of the experimental apparatus. Opticalbeam-deﬂection was used to record the response while SAM-immobilized cantilevers were exposed to explosive vapors. The ﬂow rate of gas around the cantilever was kept constant to minimize bending due to gas ﬂow. A vapor generator developed at Idaho National Laboratory (INL) was used to generate the pentaerythrital tetranitrate (PETN), RDX (cyclotrimethylenetrinitramine), and trinitrotoluene (TNT) vapor streams. Flowing ambient air through a reservoir containing explosives (PETN, TNT, or RDX) kept at a constant temperature generated the vapor stream. The reservoir consisted of 0.1 g of explosive dissolved in acetone and deposited on glass wool contained in a stainless steel block. The temperature of the reservoir was controlled by two thermoelectric elements that cooled or heated the reservoir, generating a saturation vapor pressure within it. All the experiments were conducted at a constant ﬂow rate of 100 standard cubic centimeters per minute (sccm) to eliminate the parasitic cantilever deﬂection that would have been induced by variations in ﬂow. For the experiments involving RDX and PETN, both vapor generators were operated at 50◦ C. At that temperature, the vapor concentrations for PETN and RDX were 1.4 and 0.3 ppb, respectively [25]. Deﬂagration experiments were conducted only for TNT vapors. For deﬂagration experiments we have used a piezoresistive microcantilever; the piezoresistive

APPARATUS

253

Pulse Generator Detector Interface Box

DAQ Board

Spectrum Analyzer Board

PSD

Diode Laser

Piezoresistive Cantilever Stanless Steel Tip

Personal Computer

Perforated Tube

Temperature-Controlled Stainless Steel Block

Vapor Generator Controller (Temperature & Flow Control)

TNT Reservoir (TNT-Coated Glass Wool)

Tip Flow Reservoir Flow

Figure 12.1 Schematic diagram of the experimental apparatus used in the present experiments.

track provided an efﬁcient way for heating the cantilever. The cantilever motion was detected by the optical beam deﬂection method. This allowed simultaneous measurement of bending and resonance frequency. The explosive vapor generator produced controlled amounts of TNT vapor. The tip of the vapor generator was placed about 5 mm below the cantilever for optimum TNT loading. Commercially available piezoresistive microcantilevers (resistance ≈2.2 k) [26] were used for the experiments. A Stanford Research model DS 345 function generator provided the 10-V heating pulse with ≈50-ns rise time. Exposing cantilevers to TNT vapors at ambient conditions led to adsorption of TNT onto the cantilever surfaces. The mass of adsorbed TNT was calculated by measuring the variation in cantilever resonance frequency. After the desired amount of TNT is adsorbed on the cantilever, a voltage pulse (10 V, 10 ms) was sent through the cantilever for rapid heating. It is estimated that a temporal gradient of a million degrees was achieved with rapid heating of the cantilever, which led to subsequent deﬂagration of the deposited TNT. The deﬂagration of TNT caused the cantilever to move more than it did when it was heated without any adsorbed TNT.

254

12.4

EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS

RESULTS AND DISCUSSION

Figure 12.2 shows the bending and resonance frequency response of a SAMcoated cantilever when it was exposed to a PETN vapor stream. It is clear from Figure 12.2 that the bending response of the cantilever to the PETN vapor exposure is extremely sensitive and fast. Since the noise level of the bending response in these experiments is ≈2 nm (3× standard deviation of the noise level), the detection resolution corresponding to Figure 12.2 is ≈14 parts-per-trillion. Maximum bending of the cantilever is achieved within 20 s. The amount of PETN delivered by the generator in 20 s was ≈700 pg. The adsorbed mass calculated from resonance frequency variation was approximately 100 pg. Since the bending due to weight of the adsorbed mass is around 4 picrometers, almost all of the bending is due to surface stress. The 200-nm deﬂection for 100 pg of adsorbed PETN corresponds to a limit of detection (LOD) of a few picograms. The resonance frequency method gives an LOD that is an order of magnitude higher (tens of picograms), mostly due to noise in the resonance frequency measurements. It is obvious from Figure 12.2 that the cantilever responds within 10 s to a stream of PETN vapor at a sub-part-per-billion vapor concentration. Once the PETN stream is switched off, the cantilever returns to equilibrium position within

50

32480

Bending Signal (nm)

32440 −50 32420 −100 32400 −150 32380 −200

−250

Resonance Frequency (Hz)

32460

0

32360

0

200

400 Time (s)

600

32340

Figure 12.2 Response of a 4-mercaptobenzoic acid (4-MBA)-coated silicon cantilever to PETN vapors of 1.4-ppb concentration in ambient air. The solid curve depicts the bending response, and the dots depict the resonance frequency of the cantilever.

RESULTS AND DISCUSSION

255

50 s indicating the fast relaxation of the cantilever at room temperature. When the PETN stream is turned on for 10 s, a 40-nm deﬂection signal is observed. When the vapor stream is turned off, the cantilever is relaxed back almost to the original position within 60 s. Another important observation from the data shown in Figure 12.3 is that the resonance frequency of the cantilever does not change signiﬁcantly as a result of the small amount of PETN deposited in 10 s. The bending of the cantilever is still quite easily detected. The mechanism for cantilever bending is assumed to be adsorption-induced stress. The adsorption decreases the surface free energy and surface free energy density is surface stress. Hydrogen bonding between the nitro groups of the explosives molecules and the hydroxyl group of 4-MBA may be responsible for the easily reversible adsorption of explosive vapors on the SAM-coated top surface of the cantilever. This hydrogen bonding creates a differential surface stress since hydrogen bonding is conﬁned to only one of the surfaces. One of the important characteristics of a sensor, in addition to selectivity and sensitivity, is its ability for regeneration. For useful applications as an explosive sensor, the cantilever should be able to regenerate itself within seconds for continued operation. Therefore, techniques, based on receptors should utilize weak bonding that can be broken at room temperature.

20 31840

Bending Signal (nm)

0 31800

31780 −20

31760

Resonance Frequency (Hz)

31820

31740 −40

31720

0

200

400 Time (s)

600

31700

Figure 12.3 Response of a 4-mercaptobenzoic acid (4-MBA)-coated silicon cantilever to the periodic turning on (10 s) and off (60 s) of PETN vapors of 1.4-ppb concentration in ambient air. The solid curve depicts the bending response, and the dots connected by dashed lines depict the resonance frequency of the cantilever.

256

EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS

Using just a 4-MBA receptor layer for elective detection of explosives will bring many false-positive signals because the fundamental mechanism is an acid–base reaction. However, there are many other SAMs with different head groups that could be used to provide orthogonal signals. Once a number of these layers are identiﬁed, an array of cantilevers could be modiﬁed with different SAMs to create unique responses. Response from an orthogonal array could be analyzed with pattern recognition techniques to identify the explosive molecules. Experiments are presently under way to investigate the use of orthogonal arrays for selective and sensitive detection of explosives. 12.5

DEFLAGRATION

When a cantilever with adsorbed TNT is heated rapidly with a temporal gradient of 106 degrees or higher, the adsorbed TNT undergoes deﬂagration [23]. A slower rate generally results in evaporation. For deﬂagration experiments the cantilevers were ﬁrst exposed to controlled amounts of TNT vapor from the vapor generator. The mass of adsorbed TNT on the cantilever was calculated from the resonance frequency variation using Eq. (12.6). The initial mass of the cantilever was calculated to be approximately 130 ng. When a voltage pulse of 10 V is applied across the piezoresistive cantilever (corresponding to a current of ≈5 mA), the temperature of the cantilever increases to ≈500◦ C. This corresponds to temporal-time gradient of 5 × 106◦ C/s. This rapid heating resulted in cantilever bending, probably due to the difference in coefﬁcients of thermal expansion of different regions of the piezoresistive cantilever (doped, undoped, oxide regions). Typical deﬂection signals from the cantilever, with and without TNT deposited on it, are shown in Figure 12.4. The dotted curve of Figure 12.4 represents the bending of the cantilever in the absence of TNT. From Figure 12.4 it is clear that the cantilever comes to thermal equilibrium during heating within about a millisecond, and the bending signal reaches an equilibrium position. When the voltage pulse is turned off, the cantilever comes back to ambient temperature within about a millisecond and the bending relaxes back to its original position. When a voltage pulse is applied to the cantilever loaded with TNT, the rapid rise in temperature of the cantilever leads to the deﬂagration of the TNT. This deﬂagration results in a rapid, exothermic reaction. The heat released during deﬂagration induces an additional bending of the cantilever, which appears as a “bump” during the fast heating regime. This additional bending is shown as a solid curve in Figure 12.4. Since the amount of heat released can be expected to increase with the mass of TNT deposited on the cantilever, we can expect the area of the bump to increase linearly with the mass of TNT deposited on the cantilever. The mass of TNT deposited on the cantilever can be deduced from the shift in resonance frequency. A plot of the area of the bump versus the TNT mass measurement is shown in Figure 12.5. The linear behavior shown in Figure 12.5 is consistent with the notion that adsorbed TNT is responsible for the observed bump.

DEFLAGRATION

257

0.10 (a) 80 pg 0.05 0.00 −0.05 −0.10

TNT loaded

−0.15

Without TNT

−0.20 −0.25

0

10

20

30

40

50

10

20

30

40

50

10

20

30

40

50

0.10 (b) 230 pg Cantilever Bending Signal (V)

0.05 0.00 −0.05 −0.10 −0.15 −0.20 −0.25 −0.30

0

0.1

(c) 400 pg

0.0

−0.1

−0.2

−0.3

0

Time (ms)

Figure 12.4 Bending response of a cantilever (as measured by the voltage output from a position-sensitive detector) to applied voltage pulse with and without TNT adsorbed on the surfaces. The bending of the uncoated cantilever follows the time proﬁle of the applied voltage pulse (except the lengthening of the rise and fall times) and is presumably due to the difference between the thermal expansion coefﬁcients of silicon and the doping material. The exothermic nature of the TNT deﬂagration event is clear due to the enhancement in bending of the cantilever.

EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS

2000

300

1800

250

Area Shift (arb. units)

1600

200

1400

150

1200 100

1000

50

800

Area Shift (arb. units)

258

0 0 100 200 300 400 500 600

600

ΔM (pg)

400 200 0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

ΔM (ng) — From Frequency Shift

Figure 12.5 Relationship between the area under the curve obtained from the peaks shown in Figure 12.2 and the mass of adsorbed TNT calculated from the observed changes in resonance frequency of the piezoresistive cantilever.

Temperature increase can cause various physical and chemical processes for the adsorbed TNT molecules, including melting, vaporization, and decomposition. All these reactions are endothermic except deﬂagration. Since the resonance frequency of the cantilever returns to its original value after deﬂagration, it can be assumed that all the adsorbed TNT desorbed from the surface. The sensitivity of the present technique can be estimated as follows. In the experiments that have been conducted so far, the minimum amount of TNT that we were able to detect on the cantilever is ≈50 pg. Since the sticking coefﬁcient for TNT on the cantilever surface at low TNT concentrations is ≈0.1, the cantilever needs to be exposed to 500 pg or so of TNT to be detected. Therefore, the current LOD for this technique is better than a nanogram. The selectivity of the technique was determined by exposing the cantilever to common interferents such as vapors of water, gasoline, acetone, and alcohols. It was found that none of these interferes with our detection method. All these interferents (at comparable concentrations) desorb from a surface very quickly as compared to more “sticky” molecules such as that of TNT. Therefore, such interferences will not lead to a deﬂagration signal. In addition, during rapid heating these interferents (from extremely high concentrations) showed an endothermic behavior, indicating removal of heat from the cantilever. These endothermic signals are opposite to that observed for TNT. TNT vapor detection using deﬂagration of adsorbed TNT has an obvious advantage in that it does not utilize any receptor-based detection. Deﬂagration occurs only for energetic molecules, and, therefore, many of the selectivity challenges encountered with receptor-based methods could be overcome. At present, however, the limit of detection using deﬂagration appears to be much lower than that of detection based on selective layers. It has already been demonstrated that

REFERENCES

259

the 70-pg limit of detection for TNT could be much improved by designing small-spring-constant cantilever with lower thermal mass.

12.6

CONCLUSIONS

Microcantilevers can be used for detection of explosives vapors utilizing different modes of operation. One method exploits a receptor–analyte interaction that results in bending and resonance frequency shift. Receptive SAM layers provide selectivity to cantilever sensing. Experiments conducted using a selective SAM coating of mercapto benzoic acid show sensitivity in the parts-per-trillion range. Another method of explosive detection that does not use receptor layers is based on deﬂagration of adsorbed explosive molecules by rapid heating of the cantilever. The amplitude of deﬂagration signal depends on the amount of explosive molecules adsorbed on the cantilever. For TNT, a limit of detection of 50 pg can be obtained. Higher detection limits could be achieved using cantilevers with smaller spring constants and lower thermal mass.

ACKNOWLEDGMENTS

The experiments reported here were conducted with the help of Drs. L. A. Pinnaduwage, V. Boiadjiev, Fang Tian, and G. Muralidharan and D. Hedden, J. E. Hawk, T. Ghel, D. Ye, and L. Senesac. This work was supported by the Department of Homeland Security, and Alcohol Tobacco, and Firearms (ATF), and Federal Aviation Administration, TSA, and ORNL. Oak Ridge National Laboratory is managed by UT-Battelle, LLC, for the U.S. Department of Energy under contract DE-AC05-00OR22725.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9.

R. J. Colton and J. N. Russell, Science 299, 1324 (2003). A. Fainberg, Science 255, 1531 (2001). C. M. Harris, Anal. Chem. 74, 127A (2002). L. A. Pinnaduwage, H. F. Ji, and T. Thundat, IEEE Sensors Journal, 5, 774 (2005). T. Thundat, P. I. Oden, and R. J. Warmack, Microscale Thermophysical Engineering, 1, 185 (1997). V. Lavrick, M. J. Sepaniak, P. G. Datskos, Rev. Sci. Instrum. 75, 2229 (2004). T. Thundat and A. Majumdar, Sensors and Sensing in Biology and Engineering, Barth, Humphrey, and Secomb (Eds.), Springer Wein NewYork (2004). C. Ziegler, Analytical and Bioanalytical Chemistry, 379, 946 (2004). L. A. Pinnnaduwage, A. Gehl, D. L. Hedden, G. Muralidharan, T. Thundat, R. Lareau, T. Sulcheck, L. Manning, B. Rogers, M. Jones, and J. D. Adams, Nature, 425, 6957 (2003).

260

EXPLOSIVE VAPOR DETECTION USING MICROCANTILEVER SENSORS

10. J. Fritz, M. K. Baller, H. P. Lang, T. Strunz, E. Meyer, H. J. Guntherodt, E. Delamarche, C. Gerber, and J. K. Gimzewski, Science, 288, 316 (2000) 11. J. K. Gimzewski, C. Gerber, E. Meyer, and R. R. Schlitter, Chemical Physics Letters, 217, 589 (1994). 12. J. R. Barnes, R. J. Stephenson, C. N. Woodburn, S. J. Oshea, M. E. Welland, T. Rayment, J. K. Gimzewski, and C. Gerber, Nature, 372, 79 (1994). 13. E. T. Arakawa, N. V. Lavrick, S. Rajic, and P. G. Datskos, Ultramicroscopy, 97, 459 (2003). 14. P. I. Oden, G. Y. Chen, R. A. Steele, R. J. Warmack, and T. Thundat, Appl. Phys. Lett. 68, 1465 (1996). 15. G. G. Stoney, Proc. R. Soc. London Ser. A 82, 172 (1909). 16. R. Shuttleworth, Pro. Phys. Soc. (London) 63A, 444 (1950). 17. H. J. Butt, J. Colloid Interface Sci. 180, 25 (1996). 18. D. Sarid, Scanning Force Microscopy (Oxford University Press, New York, 1991). 19. M. V. Salapaka, S. Bergh, J. Lai, A. Majumdar, J. Appl. Phys. 81, 2480 (1997). 20. O. N. Tufte and E. L. Stelzer, J. Appl. Phys. 34, 323 (1963). 21. M. Tortonese, R. C. Barrett, and C. F. Quate, Applied Physics Letters 62, 834 (1993). 22. A. Boisen, J. Thaysen, H. Jensenius, and O. Hansen, Ultramicroscopy 82, 11 (2000). 23. Z. Dai and H. X. Ju, Physical Chem. Chem. Phys. 3, 3769 (2001). 24. E. J. Houser, T. E. Mlsna, V. K. Nguyen, R. Chung, R. L. Mowery, and R. A. McGill, Talanta, 54, 469 (1995). 25. L. A. Pinnaduwage, V. Boiadjiev, J. E. Hawk, and T. Thundat, Applied Physics Letters, 83, 1471 (2003). 26. L. A. Pinnaduwage, A. Wig, D. L. Hedden, A. Gehl, D. Yi, T. Thundat, R. T. Lareau, J. Appl. Phys. 95, 5871 (2004).

CHAPTER 13

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES GREG E. COLLINS, JOSEPH WANG, AND CHRISTOPHER A. TIPPLE

13.1

INTRODUCTION

Microfabricated microﬂuidic analytical devices, integrating multiple sample handling processes with the actual measurement step on a microchip platform are of considerable recent interest [1]. Such devices are typically referred to as lab-on-a-chip devices or capillary electrophoresis (CE) microchips. Complete assays, involving sample pretreatment (e.g., preconcentration/extraction), chemical/biochemical derivatization reactions, electrophoretic separations, and detection, have thus been realized on single microchip platforms. These analytical microsystems rely on electrokinetic ﬂuid “pumping”, thereby, eliminating the need for error-prone micropumps or microvalves. Highly effective separations combined with short assay times have been achieved by combining long separation channels and high electric ﬁelds [2]. The dramatic downscaling and integration of chemical assays make these analytical microsystems extremely promising for faster and simpler ﬁeld monitoring of explosives. Particularly attractive for on-site security, decontamination, and remediation applications are the small dimensions/portability, high degree of integration, minimal solvent/reagent consumption and waste production (to the nanoliter level), efﬁciency, speed, and disposability of lab-on-a-chip devices. Lab-on-a-chip devices have predominantly been fabricated in glass or quartz substrates, although polymeric devices have now become commonplace as well. The fabrication of glass devices is relatively straightforward, as shown in Figure 13.1. A glass or quartz substrate is ﬁrst coated with a thin layer of chrome, followed by a thin layer of a photoresist [Fig. 13.1(a)]. If desired, these coated glass substrates can be purchased commercially as a convenient starting point. Using a high-power ultraviolet (UV) light source (∼100 W), the Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

261

262

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

Lamp

(a)

(b)

(c)

(d)

(e)

(f)

(g)

Figure 13.1 Depiction of the glass-based lab-on-a-chip fabrication method. Shown in the ﬁgure is (a) the photoresist and chrome-coated glass substrate, (b) the coated substrate exposed to UV light through a mask (black rectangle), (c) removal of the exposed photoresist, (d) removal of the exposed chrome layer, (e) removal of glass by wet chemical etching, (f) removal of the bulk photoresist, and (g) removal of the bulk chrome layer.

microﬂuidic design is transferred from a previously prepared mask [Fig. 13.1(b), black rectangle] onto the photoresist. These masks are routinely produced using high-resolution printers or made commercially by coating glass substrates with chrome. The UV exposed photoresist is then chemically removed from the surface, exposing the thin chrome layer beneath [Fig. 13.1(c)]. The exposed chrome is electrochemically removed using an etchant solution, leaving an exposed glass surface [Fig. 13.1(d)]. The microﬂuidic channel is etched into the substrate using a buffered hydroﬂuoric acid solution [Fig. 13.1(e)]. Once the glass has been etched to the desired depth, as veriﬁed by proﬁlometry, the remaining photoresist and chrome layers are removed as described above. The ﬁnal step in the fabrication process is the bonding of the etched substrate to another glass substrate to seal the channels. This has been accomplished by a variety of methods, with thermal bonding being the most popular [3–5]. Similar methods have been employed for the fabrication of polymeric labon-a-chip devices. One method involves ﬁrst creating positive features of the microﬂuidic network on an appropriate substrate [6, 7]. A liquid polymeric material is then deposited onto these features and cured to form a solid. The hardened polymer is subsequently removed from the feature-forming substrate, leaving a negative image of the microchannels in the polymer. Finally, thermal bonding the patterned polymer to another polymeric material or other suitable substrate seals the channels. A second method involves hot embossing of polymeric materials to form channels in the material [8–10]. In this approach, a pattern is prepared by depositing nickel onto a substrate. This fabricated metal pattern is called a molding or electroform tool. The molding tool is coated with a release agent to

INTRODUCTION

263

aid in removing the embossed polymer from the molding tool. The molding tool is heated to an appropriate temperature and pressed against the polymer using a mechanical press. After a predetermined amount of time, the polymer and molding tool are removed from the press and cooled, at which time the molding tool is removed from the polymer. As with the glass devices, the embossed polymer is then sealed with a suitable cover plate. Although using positive masters and embossing tools are the most common methods for polymeric lab-on-a-chip fabrication, there are other possibilities, such as laser ablation [11] and lamination [12]. Lab-on-a-chip separations are reliant upon the pumping of solutions by electroosmosis and the separation of charged ions in an electric ﬁeld by electrophoresis. Each ion has an electrophoretic mobility, which is proportional to its charge and inversely proportional to the frictional forces that act upon it [13]. The velocity at which an ion migrates in the electric ﬁeld is dictated by its size, charge, and the applied potential, as seen in Eq. (13.1), where v is the velocity of the ion, μe is the electrophoretic mobility, E is the applied potential, q is the charge of the ion, η is the viscosity of the solution, and r is the radius of the ion: v = μe E

where μe =

q 6πηr

(13.1)

Ions in solution are separated as they travel down the length of the microchannel because each ion has a different electrophoretic mobility. As seen in the equation, highly charged, small molecules will exhibit the largest velocities, while lower charged, larger molecules will elute at later times. Neutral molecules have a charge of zero and are, therefore, not inﬂuenced by the presence of the electric ﬁeld. However, neutral molecules still migrate down the microchannel length due to a phenomenon called electroosmotic ﬂow (EOF). EOF arises due to the bulk ﬂow of ions inside the column. For glass microchips, at pH values greater than 3, the walls of the microchannel become negatively charged due to the ionization of surface silanol groups. In order to maintain charge neutrality, the cationic counterions are electrostatically attracted to the walls of the channel. This creates an electrical double layer along the walls of the channel, with a potential difference near the surface known as the zeta potential, ζ . The application of an external electric ﬁeld across the microchip causes the mobile cations within the diffuse double layer to migrate toward the cathode. Because the cations are solvated by a sheath of water molecules, they drag the bulk solvent along with them as they migrate down the microchannel. This bulk ﬂow of solution is the mechanism by which neutral molecules are carried down the column. The magnitude of EOF depends upon several factors, as seen in Eq. (13.2), where μEOF is the mobility due to EOF, and ε is the dielectric constant of the solution: μEOF = (ε ζ /η)

(13.2)

264

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

As can be seen from Eq. (13.2), EOF mobility is directly proportional to the zeta potential, which is strongly pH dependent. At pH values below 3, EOF is very low because there are fewer negative charges at the surface of the wall. Higher pH values result in dramatic increases in EOF. The magnitude of EOF is also ionic strength dependent, wherein high ionic strengths compress the electrical double layer and, thereby, reduce EOF. For pumping purposes, the main advantage to using EOF as opposed to pressure-driven ﬂow is that EOF generates a ﬂat ﬂow proﬁle. This ﬂat ﬂow proﬁle reduces analyte band dispersion and is the reason capillary electrophoresis-based methods obtain such high intrinsic efﬁciencies. Equation (13.1) can be modiﬁed to include the contribution of EOF, which yields Eq. (13.3): v = (μe + μEOF )E

(13.3)

From Eq. (13.3), it is clear that neutral molecules will have a net velocity. In normal electrophoresis, cations will migrate faster than neutrals, and neutrals will migrate faster than anions. Anions are electrophoretically migrating in a direction opposite to EOF. Separations of neutral molecules, such as organic explosives, can only be achieved by using buffer additives, such as micelles, ionic cyclodextrins, and bile salts. The interaction of neutral analytes with these ionic buffer additives results in a modiﬁed mobility that enables separation. One of the major advantages of lab-on-a-chip devices over capillary electrophoresis instruments, in addition to the signiﬁcantly smaller size and faster separation times, is the ease by which sample injections can be accomplished. There are different methods for injecting an analyte plug into the separation channel of a microchip. The four most common types of injections are ﬂoating, direct load, pinched, and gated injections. Figure 13.2 shows the four different types of injections. A ﬂoating injection is shown in Figure 13.2(a). In this type of injection, analyte passes directly through the intersection of the four channels by applying a voltage at reservoir 2, grounding reservoir 3, and ﬂoating the remaining reservoirs [3, 14]. In addition to ﬁlling up the intersection with analyte, however, diffusion will partially ﬁll the separation (lower) and buffer (upper) channels, giving an injection plug that increases in size with loading time. Injection of all the analyte in the intersection is performed by ﬂoating reservoirs 2 and 3, switching the high voltage to reservoir 1, and grounding reservoir 4. Figure 13.2(b) shows a direct load injection [15, 16]. In this type of injection, reservoirs 1 and 3 are ﬂoating, while voltage is applied to reservoir 2, and reservoir 4 is grounded for a speciﬁed period of time. The voltage is then switched back to reservoir 1, with reservoir 4 remaining at ground in order to perform the injection and separation. The magnitude of the applied voltage at reservoir 2 and the length of time the voltage is applied determines the amount of analyte injected. A pinched injection is shown in Figure 13.2(c). As can be seen in the ﬁgure, the ﬂow of analyte is pinched in the intersection by grounding reservoir 3 while applying voltages to reservoirs 1, 2, and 4 in such a way that diffusion into the buffer and separation channels is prevented [3, 14, 17]. This type of injection is an improvement on the reproducibility of the ﬂoating injection.

LAB-ON-A-CHIP EXPLOSIVES DETECTION BY ELECTROCHEMICAL DETECTION

(a)

1

2

265

(b)

3

4

(c)

(d)

Figure 13.2 Illustration of the various types of injection methods for lab-on-a-chip applications: (a) ﬂoating, (b) direct, (c) pinched, and (d) gated. The labeled positions are as follows for panels (a), (b), and (c): reservoir 1 is for buffer, reservoir 2 for sample, reservoir 3 for sample waste, and reservoir 4 is for buffer waste. For panel (d), reservoir 1 is for sample and reservoir 2 is for buffer.

An example of a gated injection is shown in Figure 13.2(d). In a gated injection, voltages are applied to reservoirs 1, 2, and 3, while reservoir 4 is grounded [18–21]. By adjusting these voltages appropriately, a continuous ﬂow of sample from reservoir 1 into reservoir 3 is established while there is a simultaneous ﬂow of buffer from reservoir 2 into reservoir 4. This essentially creates a valve or gate in the intersection region. Injection is accomplished by lowering the voltage at reservoir 2 for a set period of time, thereby allowing analyte to ﬂow from reservoir 1 into reservoir 4. Although other injection schemes exist, they are mainly derivatives of the aforementioned methods, coupled with different channel geometries. 13.2 LAB-ON-A-CHIP EXPLOSIVES DETECTION BY ELECTROCHEMICAL DETECTION

An electrochemical detector uses the electrochemical properties of target analytes for their determination in a ﬂowing stream. Electrochemistry (EC) offers great promise for microchip systems, with features that include high sensitivity (approaching that of ﬂuorescence), inherent miniaturization and integration

266

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

of both the detector and control (potentiostatic) instrumentation, low power and cost requirements, high compatibility with advanced microfabrication and micromachining technologies, and independence of optical path length or sample turbidity [22]. Such properties make electrochemical detection extremely attractive for creating truly portable (and possibly disposable), stand-alone ﬁeld-deployable microsystems. Proper use of CE-EC microchips requires knowledge of the redox reactions of the target analytes and their dependence upon the composition of the running buffer and the working-electrode material. Different parameters such as current, potential, or conductivity can be monitored by various electrochemical detectors. 13.2.1

Amperometry for Explosives Detection

Controlled-potential detectors are ideally suited for monitoring analytes that are electroactive at modest potentials. Electrochemical detection is usually performed by controlling the potential of the working electrode at a ﬁxed value (versus a reference electrode) and monitoring the current as a function of time. The applied potential is used to drive an electron-transfer reaction. It can be viewed as “electron pressure,” which forces the chemical species to gain (consume) or lose (liberate) an electron (reduction or oxidation, respectively). Accordingly, the resulting current reﬂects the rate at which electrons move across the electrode–solution interface. The resulting electropherogram displays multiple current peaks (over the baseline current), reﬂecting the concentration proﬁles of the analytes as they pass through the detector. Different detector conﬁgurations, based on different capillary/workingelectrode arrangements and positions of the electrode relative to the ﬂow direction, have been proposed to meet these requirements. The common characteristic of most of these is the alignment of the detector at the exit of the separation channel. Such placement of the working electrode results in isolation of the electrochemical cell from the high separation voltages, owing to the dramatic voltage drop across the separation microchannel (to a negligible value at its outlet). The working electrode material also has a profound effect upon the detector performance, with carbon, gold, or platinum being most widely utilized. See Figure 13.3 for a picture of a typical amperometric microchip platform. The inherent redox activity of nitroaromatic explosives [23], namely the presence of easily reducible nitro groups, makes them ideal candidates for amperometric detection on microchip devices. These nitroaromatic groups undergo low potential reduction to form an amine via hydroxylamine. To facilitate the separation of these neutral nitroaromatic explosives on a lab-on-a-chip device, a surfactant, such as sodium dodecyl sulfate (SDS), is commonly added to the run buffer. The groups of Wang and Luong have developed several effective CE/amperometric microchip protocols, based on different electrode materials and detector designs, for detecting organic explosives down to the parts-per-billion (ppb) level [24–27]. Early work illustrated the utility of low-cost screen-printed (thick-ﬁlm) carbon electrodes as effective detectors for end-column detection

LAB-ON-A-CHIP EXPLOSIVES DETECTION BY ELECTROCHEMICAL DETECTION

Power Supply

267

mChip Pocket PC Hand-held Analyzer (‘PalmSens’)

Figure 13.3

Photograph of an amperometric microchip setup.

on glass CE microchips [24]. Typical electropherograms obtained using the CEmicrochip/thick-ﬁlm detection system for mixtures containing increasing levels of (a) dinitrobenzene (DNB) and (b) dinitrotoluene (DNT) in 200-ppb steps are displayed in Figure 13.4. Hilmi and Luong [25] employed a gold working electrode, formed by electroless deposition onto the chip capillary outlet, for highly sensitive amperometric detection of nitroaromatic explosives [with a detection limit of 24 ppb trinitrotoluene (TNT)]. Analysis of a mixture of four explosives (TNT, 2,4-DNT, 2,6-DNT, and 2,3-DNT) was accomplished within 2 min, using a borate/SDS run buffer (pH 8.7) and an applied potential of −0.8 V. The same group reported on the use of a CE microchip with a gold-wire, working electrode for measuring the explosive content in soil extracts and groundwater [26]. Good agreement was obtained with a standard Environmental Protection Agency (EPA) highperformance liquid chromatography (HPLC) procedure. Wang’s group [27] reported on the use of diamond working electrodes for imparting high sensitivity and stability onto amperometric detection of nitroaromatic compounds following their CE microchip separations. Diamond electrode detectors offer many attractive properties, including a wide potential window, low and stable background currents, negligible adsorption of organic compounds, and low sensitivity to oxygen. The enhanced stability was illustrated from a relative standard deviation (RSD) (deﬁned as the standard deviation divided by the mean) of 0.8% for 60 repetitive measurements of 5 ppm 2,4,6-trinitrotoluene. A highly linear response was obtained for 1,3-dinitrobenzene and 2,4-dinitrotoluene over the 200- 1400-ppb range, with detection limits of 70 and 110 ppb, respectively. The versatility of lab-on-a-chip devices has been exploited for developing a novel and effective protocol for rapid screening/warning followed by detailed

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

Current / nA

268

6

a

4

b

2 0

0

400

800

1200

Current

Concentration ppb

5 nA E D C B A a 0

30

60

80 Time (s)

b 120

150

180

Figure 13.4 Microchip electropherograms for mixtures containing increasing levels of (a) DNB and (b) DNT in 200 ppb steps, along with the resulting calibration plots (inset). Thick-ﬁlm amperometric carbon detector held at −0.7 V; borate buffer (15 millimolar (mM), pH 8.7) containing 25 mM SDS. (Reprinted in part with permission from [27]. Copyright 2003 American Chemical Society.)

identiﬁcation/ﬁngerprinting of explosives. The realization of such dual (screening/identiﬁcation) mode protocol using a single microchannel chip manifold involved a rapid switching from a run buffer that did not contain SDS to an SDS-containing buffer (Fig. 13.5). Convenient distinction between “total” and “individual” explosive compounds has been accomplished in connection with chip-based ﬂow-injection (fast-screening) and the “separation” (ﬁngerprint identiﬁcation) operation modes [28]. As desired for various security-surveillance applications, such operation allows repetitive fast screening assays of the “total” explosive content, and switching to the detailed identiﬁcation once such substances are detected. Figure 13.5 illustrates such total and individual measurements for a mixture of nitroaromatic organic explosives. Assay rates of ∼360 and 30/hr have thus been realized for the total screening and individual ﬁngerprint measurements, respectively. 13.2.2

Contactless Conductivity Detection

In addition to organic explosives, CE microchips with electrochemical detectors offer great promise for separating and detecting ionic explosives [29, 30].

2-Am-4,DNT 4-Am-4,6-DNT

Current

10 nA

TNB DNA TNT 3,4-DNT

LAB-ON-A-CHIP EXPLOSIVES DETECTION BY ELECTROCHEMICAL DETECTION

269

Individual

+2 kV

0

50

100

RB with SDS

150

RB without SDS Sample

Current

Total

100 nA

0

50 100 Time (s)

150 +5 kV

Figure 13.5 “Total” and “Individual” measurements of nitroaromatic organic explosives using CE microchip with amperometric detection, based on rapid switching between ﬂow injection and separation modes, respectively. (Reprinted in part with permission from [28]. Copyright 2002 American Chemical Society.)

The contactless conductivity microchip detection system, developed in our laboratory [31], has been particularly useful for this task. Its popularity has grown rapidly in recent years. Conductivity is a universal detection technique for CE microchips, as it relies on the same property of the analyte as the separation itself, namely the mobility of ions under the inﬂuence of an electrical ﬁeld. Such a detector can thus sense all ionic species having conductivity different from the background electrolyte. The capacitively coupled contactless conductivity detector for microfabricated CE chips relies upon the placement of two external metallic ﬁlm electrodes on the thin cover of a plastic [e.g., poly(methylmethacrylate); PMMA] microchip to act as a planar capacitor. The detector is operated with sine wave excitation voltages of frequency around 200 kHz and peak-to-peak amplitude of 5 to 15 V. Conductivity changes in the solution in the microchannel area below the two electrodes can thus be monitored. This approach has several distinct advantages compared to direct contact conductivity detection since the solution does not come in contact with the external electrodes. These include the absence of problems associated with the electrode–solution contact (e.g., bubble formation or surface passivation), effective isolation from high separation voltages, a greatly simpliﬁed construction and alignment of the detector (including placement of the detector or multiple ones at different locations), and the use of narrow microchannels. Such coupling of low-cost polymeric separation chips and easily constructed contactless detectors offers great promise for creating effective and disposable CE conductivity microchips. Tanyanyiwa et al. [32] demonstrated that improved

270

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

sensitivity (and 100 nanomolar (nM) detection limits) can be achieved by using higher alternating current (ac) excitation voltages of up to 250 V. The low electroosmotic ﬂow (EOF) of the PMMA chip material facilitated the rapid switching between analyses of explosive-related cations and anions using the same microchannel and run buffer (and without an EOF modiﬁer) [29]. This led to a rapid (<1 min) measurement of seven explosive-related cations and anions down to the low micromolar level. The presence of an 18-crown-6 ether modiﬁer in the run buffer allowed separation of the peaks of the co-migrating ammonium and potassium ions. Previously, contactless conductivity detectors had always been ﬁxed at a given location, close to the end of the separation channel. We developed a movable contactless conductivity detection system for CE microchips [30]. Such a versatile system relies on positioning the detector at different points along the separation channel by “sliding” the electrode holder. The movable microchip detection system offers distinct improvements compared to common ﬁxed-detector formats. These include convenient visualization of the separation progress and improved optimization of the separation process, shorter analysis time (higher sample throughput), convenient switching between “total” and “individual” (ﬁngerprint) assays modes in the same channel and faster detection of late-eluting compounds (in connection to repositioning the detector during the run). In addition to sequential injection of anionic and cationic explosives, Wang’s group designed a special chip-based dual-end opposite injection protocol for the rapid and simpliﬁed simultaneous measurement of explosive-related anions and cations [33]. For this purpose, mixtures of anions and cations were injected simultaneously from both sides of the chip onto the separation microchannel, so that the anions and cations migrated in opposite directions and were detected in the center of the separation channel by a movable contactless conductivity detector (Fig. 13.6). Such movement of the detector provided optimal resolution for all ionic explosives. Simultaneous measurements of explosive-related ions and nerve-agent degradation products were also demonstrated (with the six explosiverelated ions detected within 1 min and the nerve-agents breakdown products requiring an additional 2 min). 13.2.3 Dual Amperometric/Conductivity Detection for Simultaneous Monitoring of Ionic and Organic Explosives

A dual electrochemical microchip detection system, based on the coupling of conductivity and amperometric detection schemes, was developed for simultaneous measurements of both nitroaromatic and ionic explosives [34]. The microsystem relied on the combination of a contactless conductivity detector with an endcolumn thick-ﬁlm carbon amperometric detector. Such ability to monitor both redox-active nitroaromatic and ionic explosives is demonstrated in Figure 13.7, which shows typical dual-detection electropherograms for a sample mixture containing the nitroaromatic explosives trinitrobenzene (TNB) (4), TNT (5), 2,4DNB (6), and 2-Am-4,6-DNB (7), as well as the explosive-related ammonium

LAB-ON-A-CHIP EXPLOSIVES DETECTION UTILIZING OPTICAL METHODS

271

d

b

i a

e +

g

h

c

f (a)

+ j k l

m n

o

(b)

Figure 13.6 Schematic diagram of the dual-end injection CE microchip system with the movable conductivity detector for simultaneous measurements of explosive-related anions and cations. (a) injection mode and (b) separation mode. (a,e) Running buffer reservoirs, (b,d) unused reservoirs, (c, f) sample reservoirs, (g) injected cation plug, (h) injected anion plug, (i) movable contactless conductivity detector, (j–l) cations 1–3, (m–o) anions 1–3. (Reprinted in part with permission from [33]. Copyright 2003 Wiley Interscience.)

(1), methylammonium (2), and sodium (3) ions. While the conductivity detector (a) proﬁles only the ionic species (1–3), the amperometric one (b) responds to the redox-active nitroaromatic components (4–7). The total assay of this seven-component explosives-related mixture was performed in less than 2 min. Dual-response ratios can provide unique characterization of the individual components based on their distinct redox and conductivity properties and can be used for improving the precision. Such peak ratios provide real-time ﬁngerprinting capability of each explosive and facilitate peak assignment (through comparison to ratios from a standard mixture). The dual conductivity/amperometry microchip detection system displayed a well-deﬁned concentration dependence for simultaneous measurements of both low- and high-energy explosives. 13.3 LAB-ON-A-CHIP EXPLOSIVES DETECTION UTILIZING OPTICAL METHODS

Of all the detection methods applied to the lab-on-a-chip, the most popular by far has been laser-induced ﬂuorescence (LIF) detection. Direct LIF detection beneﬁts

272

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

1 2

3

SP

6 mV

Response

(a)

4

5

15 nA

6 7

(b)

0

50

100 Time (s)

Figure 13.7 Electropherograms showing the simultaneous measurement of low- and highenergy explosives as recorded with the (a) conductivity and (b) amperometric detectors. Analytes, ammonium (1), methylammonium (2), sodium (3), TNB (4), TNT (5), 2,4-DNB (6), and 2-Am-4,6-DNB (7), system peak (SP). Explosive concentration, 2 mM (1,2,3) and 15 ppm (4,5,6,7). Conditions: MES/His buffer (20 mM, pH 6.1) containing 15 mM lithium dodecyl sulfate as the run buffer; separation ﬁeld strength, +250 V/cm; injection ﬁeld strength, +250 V/cm for 2 s; detection at 200 kHz, (a) 5 Vp−p and at (b) −0.5 V. (Reprinted in part with permission from [34]. Copyright 2002 American Chemical Society.)

primarily from its superb sensitivity, due to the low noise background associated with ﬂuorescence detection, strongly ﬂuorescent dye tags available, and high-powered photodiode lasers that are conveniently packaged into small dimensions. Although nitroaromatic explosives are not directly ﬂuorescent, LIF can still be utilized in their sensitive detection through the application of a competitive immunoassay, as shown by Bromberg and Mathies [35, 36]. For a competitive immunoassay, an antibody to a particular target antigen is introduced to a mixture of the sample antigen and a labeled antigen, which is intentionally introduced to compete for binding with the antibody. Providing the labeled antigen has a similar binding constant with the antibody as the target antigen, a mixture of labeled antigen–antibody and sample antigen–antibody complexes will be formed. In addition, a certain concentration of displaced labeled antigen will exist in solution. The proportion of each of these three components is determined by the concentration of the target and labeled antigens, and their competition for binding by the

LAB-ON-A-CHIP EXPLOSIVES DETECTION UTILIZING OPTICAL METHODS

273

FI

TNB-FI

Relative Fluorescence

complex

antibody. By applying this mixture to a lab-on-a-chip device, it becomes possible to separate the bound antigen–antibody pairs from the displaced labeled antigen [see Fig. 13.8(a)]. Bear in mind, however, that only the labeled antigen–antibody complex and the labeled antigen can be monitored by LIF. In the case of Bromberg and Mathies [35, 36], a monoclonal anti-TNT antibody was utilized in conjunction with 1,3,5-trinitrobenzene labeled with ﬂuorescein (TNB-Fl) as the labeled antigen. The assay was performed off-chip, and the reacted solution analyzed using a microchip device. By separating and determining the ratio of the two components, they were able to determine binding constants between TNT and the TNT antibody. These experiments were performed by varying the concentration of the selected antigen and observing the resulting peaks, as shown in Figure 13.8(b). The values for the binding constants range from (4.3 ± 2.6) × 107 M−1 for TNT to as low as (5.1 ± 3.0) × 104 M−1 for 2,4-dinitrophenol, which shows the selectivity of the assay for TNT. As can be seen in Figure 13.8(b), as the concentration of TNT is increased, the intensity of the complex peak diminishes. There is also a corresponding increase in the intensity of the displaced TNB-Fl peak. Similar results were obtained with the other explosives studied. Sensitivity and linear dynamic range are two important analytical ﬁgures of merit to consider. Using this procedure, Bromberg and Mathies [35, 36] were able to detect TNT at a level of 1 ng/mL over a range from 1–300 ng/mL. This superb sensitivity and broad dynamic range make this method extremely useful for a variety of samples.

= Antigen = Labeled Antigen = Antibody

30.0 ng/mL 10.0 3.0 1.0 0.5 0.0

Microchannel (a)

0

20

40 Time (s) (b)

60

80

Figure 13.8 (a) Pictorial of competitive immunoassay performed on a capillary electrophoresis microchip; (b) Electropherograms of mixtures of 10 nM antibody, 5 nM TNBFl, 0.5 nM ﬂuorescein, and indicated concentration of TNT in the range 0–30 ng/mL. (Figure 13.8b reprinted in part with permission from [35]. Copyright 2003 American Chemical Society.)

274

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

Wallenborg and Bailey have alternatively utilized indirect ﬂuorescence detection on a lab-on-a-chip device for quantitating mixtures of nitroaromatic explosives [37]. Indirect ﬂuorescence detection relies on the ability of the analyte to displace or quench a background ﬂuorophore contained within the buffer, thereby decreasing the ﬂuorescent signal from that obtained in the absence of the analyte. In their studies, Wallenborg and Bailey [37] used Cy7, a commercially available dye as the background ﬂuorophore. All experiments were performed using an epiﬂuorescence conﬁguration in which a single microscope objective was used for both focusing the laser light onto the separation microchannel and for collecting the emitted ﬂuorescence for delivery to a photomultiplier tube. Using a borate separation buffer containing 50 mM SDS and 5 μM Cy7, they were able to separate 10 different explosives on a microchip in less than 60 s, as seen in Figure 13.9. It is hypothesized that the decrease in background ﬂuorescence is due to a quenching of the ﬂuorescence from the dye, rather than a displacement of the dye from the solute band. Using this system, relative standard deviations (RSD) in migration times and peak heights of ≤1% and 1.7 to 3.8% were obtained, respectively. The RSD for peak height was apparently inﬂuenced by degradation of the dye following repetitive injections. This methodology has been applied to

Fluorescence Intensity (arb.)

4.8

4.6

4.4

10

8

3

7 9 6

4.2 4 2

4.0

5 3.8

1 10

20

30

40

50

60

Time (s)

Figure 13.9 Microchip-based micellar electrokinetic chromatography (MEKC) electropherogram of a mixture of nitroaromatics and nitramines. Analytes: 20 ppm of each (1) TNB, (2) DNB, (3) NB, (4) TNT, (5) tetryl, (6) 2,4-DNT, (7) 2,6-DNT, (8) 2-, 3-, and 4-NT, (9) 2-Am-4,6-DNT, (10) 4-Am-2,6-DNT. Conditions: MEKC buffer, 50 mM borate, pH 8.5, 50 mM SDS, 5 M Cy7, separation voltage 4 kV, separation distance 65 mm. (Reprinted in part with permission from [37]. Copyright 2000 American Chemical Society.)

LAB-ON-A-CHIP EXPLOSIVES DETECTION UTILIZING OPTICAL METHODS

275

soil samples contaminated with explosives. Soil samples were subject to solvent extraction to remove the explosives, and the extracted samples were then injected onto the microchip in the same manner as for the previous samples. The explosives were successfully detected at a level of 1 ppm, which demonstrates the capability of this method for separating and detecting explosives in environmental samples, as well. Unlike capillary electrophoresis, wherein absorbance detection is probably the most commonly utilized technique, absorbance detection on lab-on-a-chip devices has seen only a handful of applications. This can be attributed to the extremely small microchannel depths evident on microchip devices, which are typically on the order of ∼10 μm. These extremely small channel depths result in absorbance pathlengths that seriously limit the sensitivity of absorbance-based techniques. The Collins group has shown, however, that by capitalizing on low conductivity nonaqueous buffer systems, microchannel depths can be increased to as much as 100 μm without seeing detrimental Joule heating effects that would otherwise compromise separation efﬁciencies in such a large cross-sectional microchannel [38]. The addition of certain nitroaromatic explosives to basic nonaqueous solutions results in the formation of a strongly colored reaction product, as shown in Figure 13.10. When comparing a set of 13 different explosives, only TNT, TNB, and tetryl formed these visibly colored products at a concentration of 10 mg/L CH2−

CH2

0.25 O2N

NO2

CH3O−

O2N

NO2

NO2

0.20

NO2

TNT NO2

NO2

Absorbance

TNB

CH3O− NO2

O2N

0.15

TNB H 2C

N

H2C

NO2

O2N

NO2

0.10

−

O2N MeO

NO2 H

N

NO2

O2N CH3O−

NO2 −

CH2 H

NO2

NO2

Tetryl

Tetryl

0.05

TNT

0.00 400

450

500

550 600 Wavelength (nm)

650

700

Figure 13.10 Visible absorption spectra for 1 mg/L TNT, TNB, and tetryl in acetonitrile/methanol [87.5/12.5 (v/v)] containing 2.5 mM NaOH and 1.0 mM SDS. Inlaid are the chemical reactions of TNT, TNB, and tetryl in basic acetonitrile/methanol. (Reprinted from [38] with permission from Elsevier.)

276

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

or below. Using a green (505 nm) light-emitting diode (LED) as a light source for absorbance detection enables a simple, compact, and inexpensive method for monitoring the presence of TNT, TNB, and tetryl on a microchip, even in the presence of 10 structurally similar explosives or degradation products. The buffer utilized in these studies was a mixture of acetonitrile and methanol. Acetonitrile is amenable to both capillary electrophoresis and solid-phase extraction for the purposes of preconcentration. However, because NaOH is insoluble in acetonitrile, a small percentage of methanol was additionally required. Interestingly, the use of this buffer system caused a reversal of EOF. This was attributed to an excess common ion effect of Na+ ions surrounding the colloidal NaOH particles and the interaction of these particles with the negatively charged channel walls. This reversal in EOF allowed the use of negative potentials for injecting and separating the anionic explosives on the microchip in an extremely rapid timescale. The addition of surfactant to the buffer system was necessary in order to baseline-resolve the three explosives. At the concentrations used and because of the nonaqueous environment, it is highly unlikely that micelles are formed. Instead, it seems that weak hydrophobic interactions are responsible for the separations observed. A mixture of 13 different explosives, each at a concentration of 2 mg/L, was prepared and analyzed using the microchip, although only three analyte peaks

Photomultiplier Current (μA)

0.7651

0.7650

0.7649

0.7648 TNT TNB 0.7647 Tetryl 0.7646

0

5

10

15 Time (s)

20

25

30

Figure 13.11 Electropherogram of a solid-phase extraction sample in acetonitrile/ methanol [87.5/12.5 (v/v)] containing 2.5 mM NaOH and 1.0 mM SDS, derived from a seawater sample spiked with 0.5 μg/L of TNT, TNB, and tetryl. Applied separation ﬁeld strength, −506 V/cm, using a 10-s ﬂoating load. (Reprinted from [38] with permission from Elsevier.)

LAB-ON-A-CHIP SAMPLING OF EXPLOSIVES

277

were actually observed. Based upon the observed electropherogram, the detection limits for TNT, TNB, and tetryl were 160, 60, and 200 μg/L, respectively. In order to analyze real-world samples, a sampling step was included for the collection of explosive compounds. A LiChrolut packed microcolumn was utilized for performing solid-phase extraction (SPE) of explosives from a seawater matrix. Seawater samples were spiked with 0.5 μg/L of each trinitroaromatic explosive. After preconcentrating the explosives on the column, the explosives were rinsed off of the microcolumn using acetonitrile and analyzed using the microchip. Figure 13.11 shows the resulting electropherogram. As can be seen in the ﬁgure, the three explosives were baseline resolved in approximately 20 s. The limits of detection for TNT, TNB, and tetryl following solid-phase extraction were 0.34, 0.25, and 0.19 μg/L, respectively. This is an enhancement in sensitivity of 470, 240, and 1050 times for TNT, TNB, and tetryl, respectively. This demonstrates the ability of this method to sensitively analyze explosives in real-world samples by coupling to SPE.

13.4

LAB-ON-A-CHIP SAMPLING OF EXPLOSIVES

Lab-on-a-chip devices have been used for a wide variety of analyses. However, integrating on chip sampling into these analyses is still in the preliminary stages. There are very few published reports that address sampling issues using lab-on-a-chip devices [39–42]. Papers that address the sampling of explosives on a lab-on-a-chip platform are even fewer. Fair et al. recently published a paper that addresses the sampling of vapor-phase explosives onto a lab-on-achip device [43]. Their device consists of two separated plates, between which is an array of electrodes used for moving discrete microdroplets along the surface of a plate. This is essentially an electrowetting technique in which electric ﬁeldinduced changes in interfacial tension causes the microdroplet to move along the surface of the device. Individual droplets can be scanned across a surface and/or mixed with other droplets to carry out chemical reactions. Sampling is performed by impacting one of the plates with an aerosol stream. A small droplet is then electrically moved along the surface of the plate in a controlled manner in order to concentrate the aerosol from the surface into the droplet. They state that an area of 2.5 cm in diameter can be scanned in about 4 s using their device. The droplet is subsequently measured directly or reacted with a reagent in order to be measured. Since the volume of the microdroplet is so small (∼1 μL), the concentration of analyte in the drop can be orders of magnitude higher than when macro methods of sampling are employed. They demonstrate the ability of their device to quantitatively measure TNT when placed directly on its surface. A droplet of TNT on the surface can be mixed with a droplet of KOH to form a colored product that can be detected using a green LED at 505 nm. Fair et al. determined that their limit of detection (LOD) was 2.6 μg/mL using this method. They also note that TNT can be detected in the presence of DNT using this procedure due to DNT’s lack of an absorbance band at the excitation wavelength.

278

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

Efﬁcient assays of real-world samples will require the incorporation of a continuous sampling capability (from the external environment) or rapid sampling of multiple discrete samples. Such ability to continuously introduce real samples into micrometer channels would make lab-on-a-chip devices compatible with real-life applications. Recent efforts in several laboratories have thus focused on introducing macroworld samples to microfabricated separation devices without manual intervention. We have described a simple and effective route for rapid and reproducible introduction of multiple explosive samples into CE microchip devices [44]. This design facilitates the electrokinetic introduction of samples directly into the separation microchannel without the use of injection crosses, complex microchannel layouts, or hardware modiﬁcation. The new macroscopic–microﬂuidic interface relies on the use of a sharp sample-inlet tip placed alternately in the explosive sample and buffer vials. This sample introduction results in the insertion of highly reproducible sample plugs and allows rapid replacement of different samples with no apparent carry over. Employing an 8-cm-long separation channel and a separation voltage of 4000 V offers high-throughput ﬂow injection assays of 100 samples/h with a relative standard deviation of 3.7% for TNT (n = 100). Such an operation allows rapid and reproducible multicomponent sample analysis. Figure 13.12 illustrates high-speed simultaneous measurements of a mixture containing 20 ppm TNB (a), TNT (b), and DNT (c) obtained with the new sharp inlet interface. This series of 12 repetitive injections resulted in highly stable well-deﬁned and separated peaks, with relative standard deviations (for the current signals) of 1.3, 3.1, and 2.2%, respectively. Such response remained highly stable, with the same chip performing over 200 repetitive runs during 10- to 12-h periods without treating the channel surface. Continuous monitoring from a single sample source could be realized by adding a second sharp inlet (for the run buffer) in connection to an injection cross. Preliminary results in this direction are very encouraging. Such ability to continuously introduce ﬂowing samples into micrometer channels makes lab-on-a-chip devices highly compatible with real-life monitoring applications. Solid-phase extraction (SPE) has additionally been investigated as a means of sampling on a lab-on-a-chip platform. Initial studies utilizing microscale SPE performed by the Collins group have shown that explosives can be extracted and measured using a LiChrolut EN packed microcolumn [45]. It was demonstrated that enhancement factors for TNT greater than 500 times could be routinely achieved. This approach resulted in a limit of detection for TNT of 215 ng/L in seawater when coupled with high-performance liquid chromatograph (HPLC) detection. Efforts toward integrating SPE onto a lab-on-a-chip device are currently being investigated by the Collins group. Two complementary approaches are being pursued. One approach is to use small-diameter, C18 functionalized silica beads that are packed into a microchannel to form an extraction bed [46]. A sample solution containing trace levels of explosives is electrokinetically directed across the microcolumn bed, causing the hydrophobic explosive molecules to adsorb onto the stationary phase with nearly 100% efﬁciency. Subsequently,

LAB-ON-A-CHIP SAMPLING OF EXPLOSIVES

279

10 nA

a

Current

b c

100 s Time

Figure 13.12 Performance of the chip with the sharp inlet interface. Typical electropherograms of repetitive injections of sample mixtures containing (a) 20 ppm TNB, (b) TNT, and (c) DNT. (Reproduced by permission of the Royal Society of Chemistry [44].)

the explosives are eluted in a concentrated band by using an organic solvent containing buffer, injected onto a separation microchannel, and ﬁnally separated and detected by the lab-on-a-chip device. An alternate approach being investigated is to utilize an organo-modiﬁed sol–gel material that is deposited throughout the microchip channel to act as both an extraction and electrochromatographic separation medium. The sol–gel microchip can be used in an identical fashion to that of the packed-bed approach, with the exception that the explosives are separated electrochromatographically, instead of by micellar electrokinetic chromatography [47]. The packing elements of both microdevices are shown in Figure 13.13. The extraction characteristics of the C18 functionalized silica packed microbed have been investigated using ﬂuorescent dyes as model explosive compounds. Rhodamine B was used to study the effect of sampling time on the observed ﬂuorescence signal. A 100 nM solution of the dye was electrokinetically directed through the lab on a chip extraction bed for various lengths of time. As expected, the ﬂuorescent signal of the extracted organic was found to increase linearly with extraction time, as shown in Figure 13.14. Enhancement factors were large enough to permit detection of subpicomolar concentrations of Rhodamine B using this packed bed. Application of this micropacked bed for on-chip extraction of TNT, followed by direct analysis by indirect ﬂuorescence or amperometric detection is currently being investigated by the authors. Analogous experiments are being performed using the sol–gel ﬁlled microchip.

280

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

Figure 13.13 Optical images of a C18 functionalized packed bed (left panel) and a sol–gel ﬁlled channel (right panel). (Left panel reprinted in part with permission from [46]. Copyright 2005 American Chemical Society.)

90 s 200 Fluorescence Intensity (a.u.)

100 nM Rhodamine B

150

60 s

100 30 s 50

* 0 0

20

40

60

80

100

120

Time (s)

Figure 13.14 Laser-induced ﬂuorescence detection of 100 nM Rhodamine B dye after extraction on the packed bed. The peak marked with the * was an unidentiﬁed system peak. (Reprinted in part with permission from [46]. Copyright 2005 American Chemical Society.)

Toward the goal of improved sampling technologies for the lab-on-a-chip, we have also designed a liquid microimpinger capable of gas-phase sampling of vapor analytes directly onto a lab-on-a-chip [48]. The device relies on the bubbling of a stream of analyte vapor through a microliter-sized bed of liquid. The analyte partitions into the liquid trapping reagent and can be subsequently injected and analyzed on the microchip. In this way, future lab-on-a-chip devices will be amenable to both vapor- and liquid-phase sampling of explosives.

REFERENCES

13.5

281

CONCLUSIONS

Because of the potential portability, sensitivity, high resolving capability, and highly integratable capability of lab-on-a-chip devices, their application to explosives detection has shown great promise. Explosives analyses have been demonstrated in a variety of different detection formats on the microchip platform, including amperometry, contactless conductivity, direct and indirect ﬂuorescence, and absorbance. Current research efforts are now shifting from the back-end separation and detection protocols to the front-end sampling and preconcentration issues in order to establish a completely integrated and fully functioning labon-a-chip device. Sampling issues of interest include extraction, pretreatment, sample cleanup, and ﬁltering that will be required to analyze explosive samples in the real world. Ultimately, these activities will lead to a self-contained completely functional multichannel “counterterrorism” ﬁeld-portable (handheld) microanalyzer for providing early and timely simultaneous detection of different classes of explosives and chemical warfare agents. These developments could have a major impact on the prevention of terrorist activity, the protection of ﬁrst responders and emergency personnel, on decision making, diagnosis of the nature of the attack, and on the gathering of forensic data. ACKNOWLEDGMENTS

The authors gratefully acknowledge funding support from the Ofﬁce of Naval Research (ONR) and the Memorial Institute for the Prevention of Terrorism (MIPT). Points of view in this document are those of the authors and do not necessarily represent the ofﬁcial position of the U.S. Department of Homeland Security or MIPT.

REFERENCES 1. Figeys, D. and D. Pinto. Lab-on-a-chip: A revolution in biological and medical sciences. Anal. Chem. 71, 330A–335A (2000). 2. Jacobson, S. C., R. Hergenroder, L. B. Joutny, and J. M. Ramsey. Open channel electrochromatography on a microchip. Anal. Chem., 66, 2369–2373 (1994). 3. Jacobson, S. C., R. Hergenr¨oder, L. B. Koutny, R. J. Warmack, and J. M. Ramsey, Effects of injection schemes and column geometry on the performance of microchip electrophoresis devices. Anal. Chem., 66, 1107–1113 (1994). 4. Huang, Z., J. C. Sanders, C. Dunsmor, H. Ahmadzadeh, and J. P. Landers. A method for UV-bonding in the fabrication of glass electrophoretic microchips. Electrophoresis 22, 3924–3929 (2001). 5. Jia, Z-J., Q. Fang, and Z-L. Fang. Bonding of glass microﬂuidic chips at room temperatures. Anal. Chem. 76, 5597–5602 (2004). 6. McDonald, J. C, D. C. Duffy, J. R. Anderson, D. T. Chiu, H. Wu, O. J. A. Schueller, and G. M. Whitesides. Fabrication of microﬂuidic systems in poly(dimethylsiloxane). Electrophoresis 21, 27–40 (2000).

282

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

7. Mourzina, Y., A. Steffen, D. Kalyagin, R. Carius, and A. Offenh¨ausser. Capillary zone electrophoresis of amino acids on a hybrid poly(dimethylsiloxane)-glass chip. Electrophoresis 26, 1849–1860 (2005). 8. Kricka, L. J., P. Fortina, N. J. Panaro, P. Wilding, G. Alonso-Amigo, and H. Becker. Fabrication of plastic microchips by hot embossing. Lab on a Chip 2, 1–4 (2002). 9. Qi, S., X. Liu, S. Ford, J. Barrows, G. Thomas, K. Kelly, A. McCandless, K. Lian, J. Goettert, and S. A. Soper. Microﬂuidic devices fabricated in poly(methyl methacrylate) using hot-embossing with integrated sampling capillary and ﬁber optics for ﬂuorescence detection. Lab on a Chip, 2, 88–95 (2002). 10. Wang, J., M. Pumera, M. P. Chatrathi, A. Escarpa, R. Konrad, A. Griebel, W. D¨orner, and H. L¨owe. Towards disposable lab-on-a-chip: Poly(methylmethacrylate) microchip electrophoresis device with electrochemical detection. Electrophoresis 23, 596–601 (2002). 11. Klank, H., J. P. Kutter, and O. Geschke. CO2 -laser micromachining and back-end processing for rapid production of PMMA-based microﬂuidic systems. Lab on a Chip 2, 242–246 (2002). 12. Metz, S., R. Holzer, and P. Renaud. Polyimide-based microﬂuidic devices. Lab on a Chip 1, 29–34 (2001). 13. Oda, R. P. and J. P. Landers, in J. P. Landers, Ed. “Handbook of Capillary Electrophoresis, 2nd ed. CRC Press, New York, 1997, Chapter 1. 14. Zhang, C-X. and A. Manz. Narrow sample channel injectors for capillary electrophoresis on microchips. Anal. Chem. 73, 2656–2662 (2001). 15. Harrison, D. J., A. Manz, Z. Fan, H. L¨udi, and H. M. Widmer. Capillary electrophoresis and sample injection systems integrated on a planar glass chip, Anal. Chem. 64, 1926–1932 (1992). 16. Seiler, K., D. J. Harrison, and A. Manz. Planar glass chips for capillary electrophoresis: Repetitive sample injection, quantitation, and separation efﬁciency, Anal. Chem. 65, 1481–1488 (1993). 17. Jacobson, S. C., R. Hergenr¨oder, L. B. Koutny, and J. M. Ramsey. High-speed separations on a microchip. Anal. Chem. 66, 1114–1118 (1994). 18. Jacobson, S. C., L. B. Koutny, R. Hergenr¨oder, A. W. Moore, Jr., and J. M. Ramsey, Microchip capillary electrophoresis with an integrated postcolumn reactor. Anal. Chem. 66, 3472–3476 (1994). 19. Ermakov, S. V., S. C. Jacobson, and J. M. Ramsey. Computer simulations of electrokinetic injection techniques in microﬂuidic devices. Anal. Chem. 72, 3512–3517 (2000). 20. Ramsey, J. D., S. C. Jacobson, C. T. Culbertson, and J. M. Ramsey. High-efﬁciency, two-dimensional separations of protein digests on microﬂuidic devices. Anal. Chem. 75, 3758–3764 (2003). 21. Collins, G. E, P. Wu, Q. Lu, J. D. Ramsey, and R. H. Bromund. Compact, high voltage power supply for the lab-on-a-chip. Lab on a Chip, 2004, 4, 408–411 (2004). 22. Wang, J. Electrochemical detection for microscale analytical systems: A review. Talanta 56, 223–231 (2002). 23. Bratin, K., P. T. Kissinger, R. Briner, and C. Bruntlett. Determination of nitro aromatic, nitramine, and nitrate ester explosive compounds in explosives mixtures and

REFERENCES

24. 25. 26.

27.

28.

29.

30. 31.

32.

33.

34. 35.

36.

37.

38.

283

gunshot residue by liquid-chromatography and reductive electrochemical detection. Anal. Chim. Acta. 130, 295–311 (1981). Wang, J., B. Tian, and E. Sahlin. Micromachined electrophoresis chips with thick-ﬁlm electrochemical detectors. Anal. Chem. 71, 5436–5440 (1999). Hilmi, A. and J. H. T. Luong. Electrochemical detectors prepared by electroless deposition for microfabricated electrophoresis chips. Anal. Chem. 72, 4677–4682 (2000). Hilmi, A. and J. H. T. Luong. Micromachined electrophoresis chips with electrochemical detectors for analysis of explosive compounds in soil and groundwater. Environm. Sci. Technol. 34, 3046–3050 (2000). Wang, J., G. Chen, M. Chatrathi, K. Shin, and A. Fujishima. Microchip capillary electrophoresis coupled with a boron-doped diamond electrode-based electrochemical detector. Anal. Chem. 75, 935–939 (2003). Wang, J., M. Pumera, M. P. Chatrathi, A. Escarpa, M. Musameh, G. Collins, A. Mulchandani, Y. Lin, and K. Olsen. Single-channel microchip for fast screening and detailed identiﬁcation of nitroaromatic explosives or organophosphate nerve agents. Anal. Chem. 74, 1187–1191 (2002). Wang, J., M. Pumera, and G. Collins. A chip-based capillary electrophoresiscontactless conductivity microsystem for fast measurements of low-explosive ionic components. Analyst 127, 719–723 (2002). Wang, J., G. Chen, and A. Muck, Jr. Movable contactless conductivity detector for microchip capillary electrophoresis. Anal. Chem. 75, 4475–4479 (2003). Pumera, M., J. Wang, F. Opekar, I. Jel´ınek, J. Feldman, H. L¨owe, and S. Hardt. Contactless conductivity detector for microchip capillary electrophoresis. Anal. Chem. 74, 1968–1971 (2002). Tanyanyiwa, J., E. Abad Villar, M. T. Fernandez Abedul, A. Costa Garcia, W. Hoffma, A. Guber, D. Herrmann, A. Gerlach, N. Gottschlich, and P. Hauser. Highvoltage contactless conductivity-detection for lab-on-chip devices using external electrodes on the holder. Analyst 128, 1019–1022 (2003). Wang, J., G. Chen, A. Muck, Jr., and G. E. Collins. Electrophoretic microchip with dual-opposite injection for simultaneous measurements of anions and cations. Electrophoresis 24, 3728–3734 (2003). Wang, J. and M. Pumera. Dual conductivity/amperometric detection system for microchip capillary electrophoresis. Anal. Chem. 74, 5919–5923 (2002). Bromberg, A. and R. A. Mathies. Homogeneous immunoassay for detection of TNT and its analogues on a microfabricated capillary electrophoresis chip. Anal. Chem. 75, 1188–1195 (2003). Bromberg, A. and R. A. Mathies. Multichannel homogeneous immunoassay for detection of 2,4,6-trinitrotoluene (TNT) using a microfabricated capillary array electrophoresis chip. Electrophoresis 25, 1895–1900 (2004). Wallenborg, S. R. and C. G. Bailey. Separation and detection of explosives on a microchip using micellar electrokinetic chromatography and indirect laser-induced ﬂuorescence. Anal. Chem. 72, 1872–1878 (2000). Lu, Q., G. E. Collins, M. Smith, and J. Wang. Sensitive capillary electrophoresis microchip determination of trinitroaromatic explosives in nonaqueous electrolyte following solid phase extraction. Anal. Chim. Acta 469, 253–260 (2002).

284

LAB-ON-A-CHIP DETECTION OF EXPLOSIVES

39. Kutter, J. P., S. C. Jacobson, and J. M. Ramsey. Solid phase extraction on microﬂuidic devices. J. Microcolumn Sep. 12(2), 93–97 (2000). 40. Oleschuk, R. D., L. L. Shultz-Lockyear, Y. Ning, and D. J. Harrison. Trapping of bead-based reagents within microﬂuidic systems: On-chip solid-phase extraction and electrochromatography. Anal. Chem. 72, 585–590 (2000). 41. Ceriotti, L., N. F. de Rooij, and E. Verpoorte. An integrated fritless column for on-chip capillary electrochromatography with conventional stationary phases. Anal. Chem. 74, 639–647 (2002). 42. Broyles, B. S., S. C. Jacobson, and J. M. Ramsey. Sample ﬁltration, concentration, and separation integrated on microﬂuidic devices. Anal. Chem. 75, 2761–2767 (2003). 43. Fair, R. B., A. Khlystov, V. Srinivasan, V. K. Pamula, and K. N. Weaver. Integrated chemical/biochemical sample collection, pre-concentration, and analysis on a digital microﬂuidic lab-on-a-chip platform. Proc. SPIE 5591, 113–124 (2004). 44. Chen, G. and J. Wang. Fast and simple sample introduction for capillary electrophoresis microsystems. Analyst 129, 507–511 (2004). 45. Smith, M., G. E. Collins, and J. Wang. Microscale solid-phase extraction system for explosives. J. Chromatogr., A 991, 159–167 (2003). 46. Ramsey, J. D. and G. E. Collins. Integrated microﬂuidic device for solid-phase extraction coupled to micellar electrokinetic chromatography separation. Anal. Chem. 77, 6664–6670 (2005). 47. Giordano, B. C., C. L. Copper, and G. E. Collins. Micellar electrokinetic chromatography and capillary electrochromatography of nitroaromatic explosives in seawater. Electrophoresis, 27, 778–786 (2006). 48. Tipple, C. A., M. Smith, and G. E. Collins. Development of a microfabricated impinger for on-chip gas phase sampling. Anal. Chim. Acta, 551, 9–14 (2005).

CHAPTER 14

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES HEDI MATTOUSSI, AARON R. CLAPP, AND IGOR L. MEDINTZ U.S. Naval Research Laboratory

14.1

INTRODUCTION

Luminescent semiconductor nanocrystals (quantum dots, QDs) have generated intense interest in the past several years. This interest is motivated by both a strong desire to understand their unique fundamental properties and their potential for use in diverse applications ranging from electronic devices to cellular imaging. Colloidal QDs constitute a subset of a larger family of semiconductor nanostructures with dimensions on the order of, or smaller than, the intrinsic Bohr exciton radius of the bulk parent material. In this size regime quantum conﬁnement effects of the charge carriers dominate their properties. This leads to an increase in the effective energy bandgap coupled with the emergence of discrete energies for the carrier (electron and hole) excited states, and imparts upon these materials unique optical, electronic, and spectroscopic properties [1, 2]. These properties include high ﬂuorescence quantum yields (QYs) and high molar extinction coefﬁcients (∼10- to 100-fold greater than commonly used organic dyes) [3, 4]. QDs also have broad absorption spectra that span the region just below the ﬁrst absorption peak (band edge) and well into the ultraviolet (UV), and emission characterized by narrow, symmetric photoluminescence (PL) spectra that span the UV to the near-infrared (IR) region of the optical spectrum [1–11]. The location of the PL emission maximum depends on the combination of materials used and the nanoparticle size [1, 2]. In addition, QDs have high photobleaching thresholds and exceptional resistance to photo- and chemical degradation [1–7]. For optically based sensing strategies using QDs, two properties are of particular interest: (1) the ability to tune the ﬂuorescence emission as a function of core size (e.g., case of binary core materials) resulting from conﬁnement effects, and (2) the ability to excite individual or mixed QD populations at a single Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

285

286

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

wavelength far removed (∼100 nm or more) from their respective emissions; this also permits/produces large experimental Stokes shifts [1–11]. The above properties make luminescent QDs very appealing for use in biological investigations, including immunoassay development and live cell imaging. Quantum dot ﬂuorophores most commonly used in biological applications to date are composed of CdSe cores overcoated with a thin shell layer of ZnS (CdSe–ZnS core–shell). These materials have narrow, symmetric PL spectra (full width at half maximum, FWHM, ∼25 to 40 nm) that span nearly the full visible spectrum [4, 9]. The ZnS layer passivates the core surface states, protects it from oxidation, and prevents leaching of the Cd or Se ions into the surrounding solution. This also produces a substantial improvement in the ﬂuorescence quantum yield with minimal effects on the spectral characteristics [1, 3, 6–9]. As prepared, QDs typically have a hydrophobic capping shell composed of alkyl phosphines and amines and are exclusively compatible with nonpolar organic solvents such as hexane and toluene. However, several surface modiﬁcation approaches have been reported to solubilize these QDs in aqueous buffer solutions for use in biological applications [10, 11]. Existing strategies to stabilize the QDs in aqueous media include replacing the native capping molecules with hydrophilic ligands (known as “cap exchange”), forming an outer shell of silica, and encapsulating the QD with a capping shell of polymer or phospholipids bearing hydrophilic functionalized groups [10, 11]. We have shown in the past few years that due to their ﬁnite size (comparable to an average protein), CdSe–ZnS core–shell nanocrystals capped with a thin layer of dihydrolipoic acid ligands provide excellent nanoscale scaffolds (“nanoscaffolds”) for attaching several proteins on their surfaces. QD–protein conjugates were used to design multiplexed immunoassays to detect soluble toxins. In this chapter we apply these self-assembled QD–protein conjugates to design and demonstrate sensors based on F¨orster resonance energy transfer (FRET) for signal transduction. Two examples targeting either the nutrient maltose or the explosive trinitrotoluene (TNT) will be described. The sensor consists of a bioreceptor (a maltose binding protein, MBP, or a single-chain Fv anti-TNT antibody fragment, scFv, appended with an oligohistidine sequence) immobilized on the surface of CdSe–ZnS QD via metal-afﬁnity coordination. A dye-labeled analog, prebound in the receptor recognition site, quenches the QD PL in a process caused by proximity-induced FRET upon conjugation. Addition of the target, either maltose or soluble TNT, to the solution displaces the dye-labeled analog, resulting in progressive recovery of the QD PL signal that is a function of the target concentration. Various aspects of these sensor assemblies will be discussed. 14.2

QUANTUM DOT–PROTEIN BIOCONJUGATES

Two early reports introduced the ﬁrst biological demonstrations of hydrophilic QDs [12, 13]. In the ﬁrst demonstration, Bruchez et al. used the ubiquitous biotin-avidin chemistry to label cellular F-actin ﬁlaments [12]. In the second demonstration, Chan and Nie reported attachment of transferrin to

¨ FORSTER FORMALISM AND QUANTUM DOTS AS ENERGY DONORS

287

QDs that allowed them to undergo receptor-mediated endocytotsis [13]. The demonstration reported by Chan and Nie [13] used EDC-based [1-ethyl-3-(3dimethylaminopropyl) carbodiimide] condensation chemistry to react carboxy groups on the QD surface to amines. Other modiﬁcation chemistries often involve multiple steps to create the QD–protein bioconjugate [5–12]. Our group took a different approach where proteins were engineered to self-assemble onto CdSe–ZnS QDs capped with a layer of dihydrolipoic acid (DHLA) ligands. The self-assembly was realized via electrostatic attraction between a negatively charged DHLA-capped QDs and chimeric proteins that express a positively charged leucine zipper domain. The negative surface charge arises from deprotonation of COOH groups on the DHLA ligands in basic buffer solutions [14,15]. Self-assembly of protein on DHLA-capped QDs was also realized via metalafﬁnity coordination with bioreceptor proteins engineered to express a C-terminus oligohistidine attachment domain [15]. Our self-assembly approach was ﬁrst prototyped by engineering maltose binding protein to express a leucine zipper domain (MBP-zb). The resulting QD-MBP bioconjugates were found to tightly bind amylose resin and release from the resin using excess maltose. Not only was this a demonstration of the self-assembly approach and a simple assay for the sugar maltose, it subsequently permitted us to develop it as a useful puriﬁcation tool for QD–protein bioconjugates that are assembled with MBP (in a mixed surface conjugation) using amylose resin loaded on a gel column [14]. Two QD–protein ﬂuoro-immunoreagents were prepared using this approach. In one example, a chimeric streptococcal protein G engineered to have the positively charged leucine zipper attachment (PG-zb) was ﬁrst assembled on the DHLA-capped QDs, and the QD-PG-zb assembly was used to attach to IgG antibodies through speciﬁc interactions with the β2 binding domain of PG [16]. Alternatively, avidin, a positively charged protein, was self-assembled on the QDs (mixed with MBP-zb) then used as a linker to bind biotinylated antibodies [17]. Using either of these strategies, antibodies were linked to QDs along with MBP-zb, and the resulting bioconjugate puriﬁed over amylose resin to remove unconjugated antibodies [16, 17]. These antibody-linked QD immunoreagents were utilized in multiple ﬂuoro-immunoassay demonstrations including a recent four-toxin simultaneous “multiplexed” demonstration where cholera toxin (CT), shiga-like toxin 1 (SLT-1), ricin, and staphylococcal enterotoxin B (SEB) were detected at levels as low as 30 ng/mL [16–18]. There have been other demonstrations of QD–protein bioconjugates as reagents in biological assays, and the number continue to increase as interest in functionalized QDs grows [10, 11, 19–21]. ¨ 14.3 FORSTER FORMALISM AND QUANTUM DOTS AS ENERGY DONORS

Fluorescence (or F¨orster) resonance energy transfer (FRET) is a process by which energy is transferred nonradiatively from an excited donor to a nearby ground state acceptor. This process arises due to dipole–dipole interactions and is

288

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

very sensitive to changes in molecular orientation and donor–acceptor separation distance. The effective energy transfer rate also depends on the degree of spectral overlap between the emission spectrum of the donor and the absorption spectrum of the acceptor. By carefully choosing appropriate donor–acceptor pairs, the FRET rate can be adjusted accordingly. Due to an intrinsic sensitivity to molecular rearrangements on the 1 to 10-nm range (a scale that correlates well with the size of many biological macromolecules), researchers commonly use FRET to monitor intracellular interactions and especially biological binding events [22]. A few excellent reviews on FRET applications in biology have appeared over the past several years [22–27]. FRET interactions are typically characterized by either steady-state or transient ﬂuorescence emission signals from the donor or acceptor species. Efﬁcient nonradiative energy transfer results in donor PL loss associated with acceptor gain in photoluminescence intensity (if the acceptor is an emitter). The rate of this energy transfer is related to the intrinsic lifetime of the isolated donor and depends strongly on the donor–acceptor separation distance: kD – A =

τD−1

R0 r

6 (14.1)

where τD is the isolated donor excited state lifetime and the F¨orster radius, R0 , is a characteristic distance at which the FRET rate equals the isolated donor ﬂuorescence decay rate. R0 depends on the refractive index of the medium, nD , the donor PL quantum yield, QD , Avogadro’s number, NA , and a parameter, κp , that depends on the relative orientation of the donor and acceptor dipoles; it is expressed as

R0 =

9000(ln 10)κp2 QD

1/6

NA 128π 5 n4D I

(14.2)

The spectral overlap integral, I , deﬁned as, I = J (λ) dλ = PLD−corr (λ) × λ4 × εA (λ) dλ is a measure of the donor–acceptor energy overlap over all wavelengths λ, where PLD−corr and εA represent the donor emission (normalized dimensionless spectrum) and acceptor extinction coefﬁcient spectrum, respectively. The energy transfer efﬁciency (E) is deﬁned as E=

kD – A kD – A +

τD−1

=

R06 R06

+ r6

(14.3)

For the case of one donor interacting with several nearby acceptors, the total FRET efﬁciency of the system involves the sum of n individual interactions

¨ FORSTER FORMALISM AND QUANTUM DOTS AS ENERGY DONORS

compared with the donor radiative decay rate: n kD−Ai Etotal = n i=1 −1 i=1 kD−Ai + τD

289

(14.4)

In the special case where multiple acceptors are equidistant from a single centralized donor, the efﬁciency may be simpliﬁed to Etotal =

nkD−A nkD−A + τD−1

=

nR06 nR06 + r 6

(14.5)

where n is the number of acceptors surrounding each donor at a uniform distance r. This arrangement is particularly useful for improving the total FRET efﬁciency where potential conjugate distances are relatively large compared with the intrinsic R0 value. It can also dramatically improve the signal response in poorly overlapping systems. Experimentally, FRET interactions are observed and quantiﬁed by measuring the donor or acceptor emission signals. The most common and practical deﬁnition of FRET efﬁciency is E =1−

FDA FD

(14.6)

where FDA is the integrated ﬂuorescence intensity of the donor in the presence of acceptor, and FD is the intensity of the donor alone. Equivalently, the efﬁciency can be derived from changes in the donor lifetime using the expression: E =1−

τDA τD

(14.7)

where τDA is the ﬂuorescence lifetime of the donor in the presence of acceptor, and τD is the lifetime of the donor alone. This method is best for systems that display single exponential decay kinetics, which is often not the case for individual QD populations but can nonetheless provide a reasonable estimate of the efﬁciency. Alternatively, one can measure the change in acceptor photoluminescence in the presence and absence of donor to determine the efﬁciency. However, this deﬁnition of efﬁciency is much less common in practice in part because it is more complicated than simply monitoring the donor photoluminescence. With a precise measurement of the FRET efﬁciency, it is possible to deduce donor–acceptor separation distances by rearranging Eq. (14.5) (for multiple equidistant acceptors): n(1 − E) 1/6 r = R0 (14.8) E In the context of quantum-dot-based FRET, estimates of the efﬁciency as a function of the number of acceptors per QD provide replicate measurements of

290

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

the donor–acceptor distance. For each ratio of acceptor to donor in a symmetric arrangement, the calculated distance should be identical for each measurement of the FRET efﬁciency.

14.4

QUANTUM DOTS AS FRET DONORS

We used a systematic approach to understand FRET employing QD donors conjugated to proteins containing site-speciﬁcally labeled Cy3-dye acceptors [28]. Figure 14.1(a) schematically depicts the QD bioconjugate structure we used for this study. Each QD is surrounded by maltose binding proteins engineered to express a C-terminal pentahistidine (5 His) sequence (MBP-His) for attachment to the QD surface through metal-afﬁnity coordination. The total number of MBPs bound per QD remained ﬁxed (at 15) while we varied the number of Cy3-labeled MBPs from 0 to 10. This conﬁguration allowed us to control the dye-acceptor to donor ratio and also to maintain the acceptors(s) at a ﬁxed distance from the QD center [28]. These studies utilized different QD populations (colors) whose emission maxima were varied. This allowed tuning of the degree of spectral overlap with Cy3 absorption [see Fig. 14.1(b)]. As the ratio of labeled proteins per QD increased, we observed a decrease in the QD PL and concomitant enhancement of Cy3 PL (see Figs. 14.2 and 14.3). Titration of MBP-Cy3 on the QD surface in conjunction with F¨orster theory [Eq. (14.8)] allowed us to determine the distance between the QD and the proximal dye. Monitoring of ﬂuorescent lifetime as a function of n (for a given QD-Cy3 pair) showed shortening of the donor exciton lifetime only when conjugate formation took place (using proteins appended with the HIS tail), with lifetimes shorter for higher dye-to-QD ratios. These observations complemented the steady-state ﬂuorescence data observed above and conﬁrmed that efﬁcient nonradiative exciton transfer between QD donors and dye-labeled protein acceptors takes place in these systems. Further analysis of the FRET data collected from QD PL loss with increasing ratio, n, showed that the average values for the separation distance extracted for each pair were generally consistent with those anticipated using the QD core radius, the dimension of the proteins, and assuming a close approach between QD and ˚ for 510-nm emitting QDs protein–dye. Measured distances ranged from ∼65A ˚ to ∼71A for the 555-nm emitting QDs [28]. This investigation showed that our particular QD–protein–dye system (along with the conjugate conﬁguration) has two inherent advantages over conventional molecular ﬂuorophores: (1) the ability to tune the degree of spectral overlap between QD donor and a given dye acceptor and (2) the conﬁguration where a single QD donor interacts with several surrounding acceptors allows for a proportional increase in the effective acceptor cross section, resulting in higher energy transfer efﬁciency. This provides the additional ability (beneﬁt) of potentially overcoming weaker spectral overlap integrals and generates measurable FRET signals even with rather large R0 values [28].

QUANTUM DOTS AS FRET DONORS

291

r

DHLA

CdSe ZnS

MBP

Cy3

(a) 1.2

PL/absorbance (a.u.)

1.0 510 QD PL 530 QD PL 555 QD PL Cy3 absorbance

0.8 0.6 0.4 0.2 0.0

400 425 450 475 500 525 550 575 600 625 650 675 Wavelength (nm) (b)

Figure 14.1 QD-MBP-dye nanoassembly. (a) Schematic representation of the QD-MBPdye nanoassembly used (not drawn to scale). The distance r represents the average distance between the QD center and the Cy3-labeled residue on MBP. (b) Normalized absorption spectra of Cy3 dye and photoemission spectra of three CdSe-ZnS core-shell QD solutions demonstrating the ability of tuning the spectral overlap of the QD with a given dye acceptor. Adapted from reference 28 and reprinted by permission of the American Chemical Society.

292

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

Photoluminescence (a.u.)

1.0e+6 0:1 MBP-Cy3/QD 1:1 MBP-Cy3/QD 2:1 MBP-Cy3/QD 3:1 MBP-Cy3/QD 4:1 MBP-Cy3/QD 5:1 MBP-Cy3/QD 7:1 MBP-Cy3/QD 10:1 MBP-Cy3/QD

8.0e+5

6.0e+5

4.0e+5

2.0e+5

0.0 450

500

550 600 Wavelength (nm)

650

700

(a) 1.2

Relative PL; Efficiency

1.0 0.8 0.6 0.4 0.2

510 QDs MBP-Cy3 E (QD PL quenching)

0.0 −0.2

0

2

4 6 Ratio MBP-Cy3/QD (b)

8

10

Figure 14.2 Evolution of the photoluminescence spectra from the QDs and Cy3 dyes in the QD-MBP-Cy3 assemblies versus increasing dye-to-QD ratio n (a), along with the corresponding fractional donor loss, acceptor enhancement, donor-based efﬁciency, and a ﬁt of Equation (8) versus n (b). Spectra shown were corrected for direct excitation and deconvoluted. Case of 510 nm emitting QDs is shown. Adapted from reference 28 and reprinted by permission of the American Chemical Society.

QUANTUM DOTS AS FRET DONORS

293

Photoluminescence (a.u.)

3.0e+6 2.5e+5

0:1 MBP-Cy3/QD 1:1 MBP-Cy3/QD 2:1 MBP-Cy3/QD 3:1 MBP-Cy3/QD 4:1 MBP-Cy3/QD 5:1 MBP-Cy3/QD 7:1 MBP-Cy3/QD 10:1 MBP-Cy3/QD

2.0e+5 1.5e+5 1.0e+5 5.0e+5 0.0 450

500

550 600 Wavelength (nm)

650

700

(a) 1.2

Relative PL; Efficiency

1.0 0.8 0.6 0.4 0.2 555 QDs MBP-Cy3 E (QD PL quenching)

0.0 −0.2

0

2

4 6 Ratio MBP-Cy3/QD (b)

8

10

Figure 14.3 (a) Evolution of the photoluminescence spectra from 555 nm emitting QDs and Cy3 dyes versus increasing dye-to-QD ratio n. (b) Fractional donor loss, acceptor enhancement, donor-based efﬁciency, and a ﬁt of Equation (8) versus n corresponding to data shown in (a). Spectra shown were corrected for direct excitation and deconvoluted. Adapted from reference 28 and reprinted by permission of the Amercian Chemical Society.

294

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

14.5

QUANTUM-DOT-BASED FRET NANOSENSORS

Building upon the previous ﬁndings and a better understanding of the basic concepts of energy transfer applied to our QD–protein–dye assemblies, we constructed a prototype FRET-based biosensor using QDs as energy donors and an active scaffold to assemble the sensing structure (see Fig. 14.4). This sensor is designed to detect the nutrient sugar maltose in solution and is constructed by self-assembling MBP onto the nanocrystal surface via metal-afﬁnity coordination [15, 29]. An analog of maltose, β-cyclodextran (β-CD) labeled with a QSY-9 quenching dye, is preloaded in the MBP binding pocket allowing close proximity to the QD center. This results in efﬁcient energy transfer from the QDs to the proximal acceptor dyes, and the sensor is in an initial “off” state. The high FRET efﬁciency is due to the conﬁguration using several QSY-9 arrayed around a single donor and the favorable spectral overlap. As maltose is added to the solution, MBP-bound β-CD-QSY-9 analog is displaced from the protein binding pocket, disrupting FRET. This results in recovery of the QD PL in a concentration-dependent manner [see Fig. 14.5(a)]. Transforming the data permitted derivation of an apparent dissociation constant of 7 μM [Fig. 14.5(b)]. This value agreed well with those reported in the literature for solution-phase wildtype MBP between 1 and 10 μM [29]. The sensing assembly also demonstrated detection speciﬁcity by only responding to sugars having the MBP recognized α-1,4-glucosidic linkage. This shows that the proteins retain their native binding properties when immobilized on the QDs. FRET was conﬁrmed by time-resolved ﬂuorescence experiments, where data showed a decrease in the QD lifetime when the dye-labeled analog was bound to MBP, and substantial recovery of the QD lifetime when maltose was added.

Q

Analyte

Analog-quencher

FRET

Q

hn

QD Exc

Bound analog-quencher. Reduced QD emission

Add target analyte

QD Exc

hn

Analyte in recognition site. Recovery of QD emission

Figure 14.4 Function and properties of a QD FRET-based nanosensor. Generalized QD bioconjugate nanosensor schematic. Each QD is surrounded by an average of ∼10–15 protein molecules. Formation of QD-protein-analogue assembly results in quenching of the QD emission. Adding preferred analyte to the solution displaces dye-labeled analogue from the sensor assembly, resulting in an increase in direct QD emission.

PL (AU)

QUANTUM-DOT-BASED FRET NANOSENSORS

3×106

10 mM Maltose

2×106

10 μM Maltose

295

100 μM Maltose

2×106 1 μM Maltose 1×106

0 Maltose

500×106 0

480

500

520

540 560 580 Wavelength (nm)

600

620

640

(a) 2.5×106

Kapp = 6.9 +/− 0.2 μM

Saturation

0.75

2.0×106

0.50 1.5×106

PL (AU)

1.00

0.25 1.0×106 0.00 0 10−3

10−2

10−1

100

101

102

103

104

105

Maltose Concentration (μM) (b)

Figure 14.5 Case of QD-MBP sensor targeting maltose. (a) PL spectra measured from a solution containing QD-MBP conjugates (at 10 MBPs per QD) preassembled with 1 μM beta-CDQSY9 following exposure to increasing concentrations of maltose. (b) Transformation of the titration data to derive estimation for the binding constant. The data were ﬁt to a four-parameter Hill function (solid lines) appropriate for describing the binding equilibrium. The 50% saturation level was used to derive the maltose apparent dissociation constant (Kapp ) value. Adapted from reference 29 and reprinted by permission from Nature Publishing Group.

296

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

The applicability of this sensing scheme has been extended more recently to target an analyte of interest to the U.S. Navy, namely the explosive molecule trinitrotoluene (TNT) [30]. Using the same design criteria, an antibody fragment selected for its recognition of TNT (TNB2-45) was engineered to express a Cterminal 12 histidine (12 His) sequence. The antibody fragment was allowed to prebind an analog of TNT (TNB) that has been conjugated to the dark quenching dye BHQ10; the complex is referred to as TNB–BHQ10. The bound TNB–BHQ10–TNB2-45 complex was then allowed to self-assemble on the QD surface (as depicted in Fig. 14.4). As in the maltose sensing arrangement, due to close proximity of the QD–donor to the BHQ10 and good spectral overlap for the donor–acceptor pair, the QD undergoes efﬁcient FRET quenching [Fig. 14.6(a)]. As TNT is added to the assay solution, it competes for binding to the antibody fragment. Displacement of the TNB–BHQ10 away from the QD center reduces the FRET efﬁciency, and the QD ﬂuorescence increases in a concentration-dependent manner [see Fig. 14.6(b)). The speciﬁcity of this TNB2-45 antibody fragment for explosive analogs of TNT was retained in this QD sensing conﬁguration. In tests against TNT, Tetryl, 2-A,4,6-DNT and 2,6DNT, only TNT elicited a signiﬁcant response [30]. By comparison, the other analogs elicited responses nearly fourfold less than TNT. Both of these studies demonstrate that QDs not only function as efﬁcient FRET donors for nanosensors, but also a convenient “nanoscaffold” or structural template that coordinates the formation of the nanosensor. Using the same QD donor/nanoscaffold system with MBP self-assembled on the surface, we showed that QD emission could be modulated through the use of a photochromic acceptor dye [31]. MBP was labeled with a sulfo-N hydroxysuccinimide activated photochromic BIPS molecule [1 ,3-dihydro-1 -(2carboxyethyl)-3,3-dimethyl-6–nitrospiro[2H-1-benzopyran-2,2 -(2H)-indoline]] at a dye-to-MBP-protein (D/P) ratio of either ∼1 or ∼5; ∼20 BIPS-labeled MBP were immobilized on 555-nm emitting QDs. The ability of MBP–BIPS to modulate QD photoluminescence was tested by switching the BIPS from the colorless spiropyran (SP, nonabsorbing) to the colored merocyanine (MC, energy absorbing) by irradiating the sample with white light (>500 nm) or UV light (∼365 nm), respectively [31]. We measured quenching efﬁciencies of the QD emission of ∼25 and 60% in the presence of the MBP–BIPS–MC product at D/P of 1 and 5, respectively, compared to negligible loss in the presence of the SP version of the BIPS dye [31]. This demonstration shows that it is possible to use not only proteins but other functional dyes to reversibly control QD properties; this could have important implications for biologically based nanoscale devices.

14.6

SURFACE-ATTACHED QD–FRET NANOASSEMBLIES

Attempts to transition present and future QD-based nanosensors to ﬂow cells and integrated devices require that facile schemes be developed to tether these assemblies to surfaces while retaining their functionality and orientations. Bottom–up

SURFACE-ATTACHED QD–FRET NANOASSEMBLIES

297

Fluorescence Signal (a.u.)

50000 TNB-BHQ10:QD = 4 TNB-BHQ10:QD = 8 TNB-BHQ10:QD = 16 TNB-BHQ10:QD = 32 TNB-BHQ10:QD = 64 TNB-BHQ10:QD = 128 No TNB-BHQ10

40000 30000 20000 10000 0

450

500

550 600 Wavelength (nm) (a)

650

700

Fluorescence Recovery (a.u.)

10000

8000

6000

4000

2000

0

0

2

4 6 8 TNT Concentration (μg/mL) (b)

10

12

Figure 14.6 Case of QD-based TNT FRET sensing. (a) Progressive quenching of the QD emission with increasing concentration of TNB-BHQ10 prebound in the TNB2-45 binding pocket. (b) TNT titration of QD-TNB2-45 assembly. Data is plotted as the difference signal (with respect to the case where the conjugate is preloaded with TNB-BHQ10) versus concentration in logarithmic scales. The assembly was contructed using 5470 nm emitting QDs. Each data point is an average of three measurements; error bars represent the standard deviation. The data were ﬁt to a four-parameter Hill function (solid lines) appropriate for describing the binding equilibrium. Plots are adapted from reference 30 and reprinted by perimission from the American Chemical Society.

298

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

self-assembly approaches represent an attractive technique for tethering and deriving nanosensor architecture and function. It obviates the need for chemical conjugation, which may interfere with the protein structure and function. We have carried out a preliminary study to determine if our hydrophilic QDs can be attached to surfaces via protein arrangement and still function in a FRET conﬁguration with dye-labeled proteins [32]. Biotin-avidin-driven interactions and electrostatic self-assembly were combined to create these surface-tethered QD–receptor assemblies. Figure 14.7 schematically depicts the self-assembly and conﬁguration of one of the assemblies tested. Slides functionalized with a monolayer of avidin were used as the solid support and exposed to biotinylated MBP. The latter functions as a linker protein to capture mixed surface QD–MBP–dye/avidin bioconjugates assembled with a ratio of ∼1 avidin and approximately 15 Cy3.5 dye-labeled MBP. In this conﬁguration, the MBP is labeled with the dye acceptor and is used to purify the resulting bioconjugate over amylose resin [15, 29, 30]. By using appropriate manifolds to cover the substrates and control the “layer-by-layer” assembly, discrete QD–protein structures were created and tested for evidence of FRET interactions with bound dyes [32]. This is demonstrated in the ﬂuorescence panels shown in Figure 14.8. The structures were excited with the 488-nm line of an Ar ion laser. Emission was collected using either a 550-nm-long pass ﬁlter or a 590-nm-long

Self-assemble QD with dye-labeled MBP and Avidin on QD surface

Purify assembly over amylose column

Allow QD assembly to attach to surface and wash away excess

MBP Chemically attach Avidin to slide

Attach biotinylated MBP to Avidin as linker

QD

QD

Av

Av MBP-Bt Av

Av

Av Glass Slide

Step 1

Step 2

Step 3

Step 4

Step 5

Figure 14.7 Schematics depicting the assembly of QD-protein conjugates that engages in FRET near a surface. Step 1, the glass slide waveguide is coated with Avidin. Step 2, attach biotinylated MBP to Avidin on the surface as a linker. Step 3, self-assemble MBP-dye and avidin onto the QD surfaces. Step 4, purify the QD conjugate solution from 3 over amylose resin. Step 5, allows the QD assembly to attach to the MBP-Bt via its surface Avidin and wash away excess reagents. Adapted from reference 32.

SURFACE-ATTACHED QD–FRET NANOASSEMBLIES

[2*]

100% MBP555QD-AV

[6 ]

(b)

[6 ]

[5 ]

[4 ]

[3 ]

10 μg mL−1 MBP-Bt

[2 ]

[1 ]

[6 ]

PB

ST

MBP-Bt

[5 ]

mL−1 [4 ]

[3 ]

ST

[2 ]

PB [1 ]

[5 ]

[1*] [2*]

(a) 10 μg

[4 ]

[3 ]

ST PB

[1*]

[2 ]

[1 ]

[6 ]

[5 ]

PB

[2 ]

[1 ]

[4 ]

10 μg mL−1 MBP-Bt

[3 ]

590-nm Emission Filter

10 μg mL−1 MBP-Bt

ST

550-nm Emission Filter

299

100% [1*] MBPCy3.5[2*] 555QD-AV [2*] [1*]

(d)

] [6

] [5

] [4

]

[2

]

10 μg mL−1 MBP-Bt [3

T BS ]P

[6

]

[1

] [5

] [4

] [3

]

10 μg mL−1 MBP-Bt

[2

[1

]P

BS

T

(c)

MBP-555QD[1*] AV MBPCy3.5[2*] [2*] 555QD-AV [1*]

(e)

(f)

Figure 14.8 FRET studies using surface-tethered QD-protein assemblies. Fluorescence images collected using a CCD for three MBP-Bt functionalized slides (a, c, and e) imaged with a 550 nm long pass emission ﬁlter, then imaged with a 590 nm long pass emission ﬁlter (b, d, and f). The slides were exposed to the following solutions: (a) MBP-555QD-AV; (b) MBPCy3.5-555QD-AV and (c) MBP-555QD-AV (top row) and MBPCy3.5-555QD-AV (bottom row); 555QD, MBPCy3.5 and AV designate 555 nm emitting QDs, cy3.5-labeled MBP and avidin, respectively. The discrete ﬂuorescence images shown in (d) and (f) result from nonradiative energy transfer from the 555 nm QD to proximal Cy3.5-labeled MBP. Figure adapted from reference 32 and reprinted by permission from the American Chemical Society.

pass ﬁlter to isolate the contribution from Cy3.5 in the structures; the 590-nmlong pass ﬁlter also removes the QD emission at 555 nm. This permitted us to test whether or not these tethered QDs could engage in FRET interactions with Cy3.5. The panels shown in Figure 14.8 clearly indicate that only the discrete squares containing QDs conjugated to dye-labeled MBP generated PL contribution beyond 590 nm; signal from control structures with no QDs was negligible, since 488 nm is within the valley of the dye absorption spectrum. We attribute this result to nonradiative energy transfer between the QD and proximal Cy3.5 acceptor. The same structures were also assembled using biotinylated antibody instead of the MBP–biotin as the linker protein between the substrate and the QD–protein–dye assemblies and tested for FRET [32]. These results conﬁrm our

300

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

previous solution-phase observations of FRET between QD donors conjugated to protein–dye acceptors, where both steady-sate and time-resolved experiments conﬁrmed that nonradiative exciton transfer occurs, and it can be quantiﬁed within the F¨orster formalism [28, 29]. 14.7

CONCLUSIONS

We have shown in this chapter the use of QD-bioreceptor conjugates to carry out demonstrations of solution-phase sensing strategies based on FRET. In particular, we discussed two FRET-based prototypes speciﬁc for the detection of a sugar nutrient and the explosive TNT. We have further tested the ability of surfacetethered QD-protein-conjugates to engage in FRET interactions with proximal dyes. These results indicate that it is possible to implement these described strategies to design sensing assemblies with potential applications in screening and diagnostics. The sensing assemblies described could ﬁnd use for in vitro biological assays to detect analytes as well as in vivo cellular indicators. Flow cell and integrated devices that incorporate QD-based biosensors for environmental or industrial process monitoring are also possible. Additionally, bioconjugate assemblies can be exposed to a solution environment for continuous monitoring of ﬂow streams or incorporated into submersible devices. Regardless of the particular application, FRET-based sensing strategies that incorporate proteins and QDs will speciﬁcally beneﬁt from some of the unique QD photophysical properties. ACKNOWLEDGMENTS

The authors acknowledge ONR and NRL for funding. H.M. acknowledges A. Ervin and L. Chrisey at the Ofﬁce of Naval Research, grant # N001404WX20270. H.M. also acknowledges A. Krishnan at DARPA for support. REFERENCES 1. Murray, C. B. and C. R. Kagan. Synthesis and characterization of monodisperse nanocrystals and close-packed nanocrystal assemblies. Ann. Rev. Mater. Sci. 30, 545–610 (2000). 2. Yoffe, A. D. Semiconductor quantum dots and related systems: Electronic, optical, luminescence and related properties of low dimensional systems. Adv. Phys. 50, 1–208 (2001). 3. Leatherdale, C. A., W. K. Woo, F. V. Mikulec, and M. G. Bawendi, On the absorption cross section of CdSe nanocrystal quantum dots. J. Phys. Chem. B 106, 7619–7622 (2002). 4. Dabbousi, B. O., J. Rodriguez-Viejo, F. V. Mikulec, J. R. Heine, H. Mattoussi, R. Ober, K. J. Jensen, and M. G. Bawendi. (CdSe)ZnS core-shell quantum dots: Synthesis and optical and structural characterization of a size series of highly luminescent materials. J. Phys. Chem. B. 101, 9463–9475 (1997).

REFERENCES

301

5. Niemeyer, C. M. Nanoparticles, proteins, and nucleic acids: Biotechnology meets materials science. Angew. Chem. Int. Ed. 40, 4128–4158 (2001). 6. Murphy, C. J. Optical sensing with quantum dots. Anal. Chem. 74, 520A–526A (2002). 7. Parak, W. J., D. Gerion, T. Pellegrino, D. Zanchet, C. Michael, S. C. Williams, R. Boudreau, M. A. L. Gros, C. A. Larabell, and A. P. Alivisatos. Biological applications of colloidal nanocrystals. Nanotech. 14, R15–R27 (2003). 8. Alivisatos, A. P. The use of nanocrystals in biological detection. Nat. Biotech. 22, 47–52 (2004). 9. Hines, M. A. and P. Guyot-Sionnest. Synthesis and characterization of strongly luminescing ZnS-capped CdSe nanocrystals. J. Phys. Chem. 100, 468–471 (1996). 10. Michalet, X. F., F. F. Pinaud, L. A. Bentolila, J. M. Tsay, S. Doose, J. J. Li, G. Sundaresan, A. M. Wu, S. S. Gambhir, and S. Weiss. Quantum dots for live cells, in vivo imaging, and diagnostics. Science 307, 538–544 (2005). 11. Medintz, I. L., H. T. Uyeda, E. R. Goldman, and H. Mattoussi. Quantum dot bioconjugates for imaging, labeling, and sensing. Nat. Mater. 4, 35–446 (2005). 12. Bruchez, M., M. Moronne, P. Gin, S. Weiss, and A. P. Alivisatos. Semiconductor nanocrystals as ﬂuorescent biological labels. Science 28, 2013–2016 (1998). 13. Chan, W. C. W. and S. Nie. Quantum dot bioconjugates for ultrasensitive nonisotopic detection. Science 281, 2016–2018 (1998). 14. Mattoussi, H., J. M. Mauro, E. R. Goldman, G. P. Anderson, V. C. Sundar, F. V. Mikulec, and M. G. Bawendi. Self-assembly of CdSe-ZnS quantum dot bioconjugates using an engineered recombinant protein. J. Am. Chem. Soc. 122, 12142–12150 (2000). 15. Goldman, E. R., I. L. Medintz, A. Hayhurst, G. P. Anderson, J. M. Mauro, B. L Iverson, G. Georgiou, and H. Mattoussi. Self-assembled luminescent CdSe-ZnS quantum dot bioconjugates prepared using engineered poly-histidine terminated proteins. Anal. Chim. Acta 534, 63–67 (2005). 16. Goldman, E. R., G. P. Anderson, P. T. Tran, H. Mattoussi, P. T. Charles, and J. M. Mauro. Conjugation of luminescent quantum dots with antibodies using an engineered adaptor protein to provide new reagents for ﬂuoroimmunoassays. Anal. Chem. 74, 841–847 (2002). 17. Goldman, E. R., E. D. Balighian, H. Mattoussi, M. K. Kuno, J. M. Mauro, P. T. Tran, and G. P. Anderson. Avidin: A natural bridge for quantum dot-antibody conjugates. J. Am. Chem. Soc. 124, 6378–6382 (2002). 18. Goldman, E. R., A. R. Clapp, G. P. Anderson, H. T. Uyeda, J. M. Mauro, I. L. Medintz, and H. Mattoussi. Multiplexed toxin analysis using four colors of quantum dot ﬂuororeagents. Anal. Chem. 76, 684–688 (2004). 19. A. J. Sutherland. Quantum dots as luminescent probes in biological systems. Curr. Op. Sol. St. Mat. Sci. 6, 365–370 (2002). 20. Cox, J. A quantum paintbox. Chem. Brit. 39, 21–25 (2003). 21. Niemeyer, C. M. Functional hybrid devices of proteins and inorganic nanoparticles. Angew. Chem. Int. Ed. 42, 5796–5800 (2003). 22. Lakowicz, J. R. Principles of Fluorescence Spectroscopy, Kluwer Academic: New York, 1999.

302

NANOSCALE SENSING ASSEMBLIES USING QUANTUM DOT–PROTEIN BIOCONJUGATES

23. Krishnan, R. V., R. Varma, and S. Mayor. Fluorescence methods to probe nanometer-scale organization of molecules in living cell membranes. J. Fluorescence 11, 211–226 (2001). 24. Klostermeier, D. and D. P. Millar. Time-resolved ﬂuorescence resonance energy transfer: A versatile tool for the analysis of nucleic acids. Biopolymers 61, 159–179 (2002). 25. Jares-Erijman, E. A. and T. M. Jovin. FRET imaging. Nature Biotech. 21, 1387–1395 (2003). 26. Miyawaki, A. Visualization of the spatial and temporal dynamics of intracellular signaling. Dev. Cell 4, 295–305 (2003). 27. Miyawaki, A., A. Sawano, and T. Kogure. Lighting up cells: Labeling proteins with ﬂuorophores. Nature Cell Biol. 5, S1–S7 (2003). 28. Clapp, A. R., I. L. Medintz, J. M. Mauro, B. R. Fisher, M. G. Bawendi, and H. Mattoussi, Fluorescence resonance energy transfer between quantum dot donors and dye-labeled protein acceptors. J. Am. Chem. Soc. 126, 301–310 (2004). 29. Medintz, I. L., A. R. Clapp, H. Mattoussi, E. R. Goldman, B. R. Fisher, and J. M. Mauro. Self-assembled nanoscale biosensors based on quantum dot FRET donors. Nat. Mater. 2, 630–638 (2003). 30. Goldman, E. R., I. L. Medintz, J. L. Whitley, A. Hayhurst, A. R. Clapp, H. T. Uyeda, J. R. Deschamps, M. E. Lassman, and H. Mattoussi. A hybrid quantum dot-antibody fragment ﬂuorescence resonance energy transfer-based TNT sensor. J. Am. Chem. Soc. 127, 6744–6751 (2005). 31. Medintz, I. L., S. A. Trammell, H. Mattoussi, and J. M. Mauro. Reversible modulation of quantum dot photoluminescence using a protein-bound photochromic ﬂuorescence resonance energy transfer acceptor. J. Am. Chem. Soc. 126, 30–31 (2004). 32. Sapsford, K. E., I. L. Medintz, J. P. Golden, J. R. Deschamps, H. T. Uyeda, and H. Mattoussi. Surface-immobilized self-assembled protein-based quantum dot nanoassemblies. Langmuir 20, 7720–7728 (2004).

CHAPTER 15

REMOTE SENSING OF EXPLOSIVE MATERIALS USING DIFFERENTIAL REFLECTION SPECTROSCOPY ¨ ROLF E. HUMMEL, ANNA M. FULLER, CLAUS SCHOLLHORN, AND PAUL H. HOLLOWAY University of Florida, Department of Materials Science and Engineering

15.1

INTRODUCTION

A number of methods for detection of explosives have been proposed in the past or are already in service. They are described in other chapters of this book and thus will not be listed here. The technique described in this chapter involves differential reﬂection spectroscopy (DRS), also often called differential reﬂectometry (DR), which is a surface analytical technique. It uses light [in general ultraviolet (UV), visible, and near infrared] as a probing medium and reveals details about the electron structure around the Fermi surface [1, 2]. Speciﬁcally, the instrument allows the measurement of the energies that electrons absorb from photons as they are raised into higher, allowed energy states. Since each material has a unique electron structure, the measurement of the characteristic energies for “electron transitions” serves as a means (i.e., a ﬁngerprint) for identifying these materials. (Essentially, the differential reﬂectometer modulates the electron structure of materials and thus provides the derivative of the spectral reﬂectivity with respect to the perturbation. A line shape analysis yields the imaginary part of the complex dielectric constant, often called the absorption [3], as a function of the excitation energy.) The aim of our endeavors to detect explosives is sevenfold: 1. First of all, the detection of explosives should be contact-less; that is, the detection should be possible from a reasonable distance (and not involve sample collection whereby the sample has to be subsequently inserted into an instrument). 2. The evaluation should be fast without compromising the sensitivity. Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

303

304

REMOTE SENSING OF EXPLOSIVE MATERIALS

3. The method should exclude false positives; for example, it should not be triggered by fertilizers on shoes, soil, or golf balls. 4. The technique should not harm people and property (such as photographic ﬁlms) as may be encountered with high-intensity X-ray equipment. 5. The method should differentiate between various explosives, that is, it should indicate which species of explosive are detected. 6. The technique should have stealth capabilities, if possible, that is, the probing beam should not be detected by human eyes or night vision goggles. 7. The technique should be sensitive, preferably registering in the μg/mm2 range or better.

15.2

DIFFERENTIAL REFLECTOMETRY

The differential reﬂectometer has been described elsewhere [1, 2]. It measures the normalized difference between the reﬂectivities of two adjacent parts of the same specimen. Brieﬂy, unpolarized, monochromatic light having a continuously varying wavelength emanates from a monochromator, illuminated, for example, by a high-pressure xenon source, see Figure 15.1. The light is alternately deﬂected to one or the other part of the sample by means of an oscillating mirror. The total area scanned is, for example, 2 × 4 mm2 . A stationary mirror, which is placed after the specimen, focuses the reﬂected light onto the face of a photomultiplier

Light Source Photomultiplier Tube

Monochromator

Output Signal Oscilloscope

Samples

Oscillating Mirror

Low Pass Filter

Lock-in amplifier

Samples Divider

Computer

Figure 15.1 Schematic of the differential reﬂection spectrometer.

RESULTS

305

tube (PMT) or other light-sensitive device. Its output voltage is electronically processed to yield the normalized difference in reﬂectivity, that is, R/R where R = R2 − R1 is the difference of the reﬂectivities of the two sample parts, and R=

R1 + R2 2

is the average reﬂectivity. Measuring R1 and R2 at the same time and forming the ratio R/R eliminates possible inﬂuences from ﬂuctuations of the line voltage. It also eradicates intensity variations of the spectral output of the light source, the spectral sensitivity of the detector, and the spectral reﬂectivities of the mirrors or the substrate. Thus, the sensitivity of the spectrometer is quite high in that the minimal measurable R/R is, as a rule, about 0.01%. An improvement to this technique will be to use a continuous blue/UV light source and replacing the PMT by a compact spectrometer coupled to a CCD (charge-coupled device) for readout, thus accelerating the data acquisition. In the latter case, the scanning and evaluation takes a few milliseconds.

15.3

RESULTS

The explosives whose spectrograms will be shown below were either commercial specimens whose purities are unknown to us or “chemically pure” species, having no taggants (acquired from Chem Service Inc). Figure 15.2 depicts a characteristic differential reﬂectogram (which displays R/R versus the excitation wavelength; see above) for solid, crystalline TNT (trinitrotoluene). It is observed that with increasing excitation energies R/R begins to gradually rise at about 435 nm and reaches a plateau near 400 nm. A second structure is detected in form of a peak near 250 nm. In many cases (but not in all) this structure manifests itself as a double peak near 245 and 255 nm. Finally, a weak peak near 303 nm, which overlaps the plateau, is discernible. (The 420nm and the 250-nm structures have essentially also been observed in transmission measurements for high-concentrated solutions of TNT in acetonitrile but are blue shifted due to the solvent; see inset of Figure 15.2. However, transmission measurements are less suited for remote detection and may add supplementary features due to the spectral characteristics of the solvent or underlying substrate.) The features in Figure 15.2 are believed to be due to electron transitions from the highest occupied molecular orbital states to lowest unoccupied molecular orbital states (HOMO–LUMO). Speciﬁcally, for single molecules and dilute solutions, the absorption in the blue part of the spectrum is proposed to be caused by the aromatic structure of TNT [4], probably involving π to π* transitions.1 In the solid state, where the molecules are stacked up on top of each other, interactions between the molecules occur causing the energy levels to split into higher 1

This transition between two orbitals (or energy levels) is fully described in Pavia et al. [4].

306

REMOTE SENSING OF EXPLOSIVE MATERIALS

Absorbance (a.u)

4

7

5

3 2 1

− ΔR/R (a.u.)

0 200 250 300 350 400 450 500 Wavelength (nm)

3

1

200

250

300

350

400

450

500

3.10

2.76

2.48

Wavelength (nm)

6.20

4.96

4.14

3.55 Energy (eV)

Figure 15.2 Differential reﬂectogram of TNT crystals on carbon tape. Inset: UV–visible transmission spectrum of TNT in acetonitrile (concentration 3.3 g/L).

and lower levels, thus producing a red shift. This yields, for TNT, an absorption feature near 420 nm as observed. Molecular orbital calculations for veriﬁcation of the absorption mechanism for TNT are currently in progress. Figure 15.3 depicts differential reﬂection spectra of a large number of mostly organic substances that could be expected to be found in or on the luggage of an average traveler. It is noted that none of these spectra have all of the same features as TNT. In particular, the 420-nm structure is generally missing. In addition, the peak or peaks near 250 nm, observed for TNT, is conspicuously absent. Instead, absorption peaks are seen that are situated at higher or lower wavelengths. As an example, the spectrum for aspartame has a threshold wavelength near 260 nm and an absorption peak near 230 nm. The spectra depicted in Figure 15.3 have all been obtained by placing small quantities of the pertinent substances on a black carbon pad. The question immediately arises whether or not the same results can be obtained by utilizing different substrates. Figure 15.4 displays differential reﬂection spectra of TNT on leather, fabric, latex glove, aluminum alloy, and cardboard. As can be seen, the spectra are essentially alike, demonstrating that the kind of substrate is immaterial for TNT detection by DR.

307

RESULTS

Polyethylene Epoxy Graphite TNT Mothball PVC Polycarbonate Baking Soda

− ΔR/R (a.u.)

Aspartame Starch Polystyrene Sugar Polyester Soap Polypropylene Nailpolish Paper Aspirin Fertilizer

200

250

300

350 Wavelength (nm)

400

450

500

Figure 15.3 Differential reﬂectograms of various substances in comparison to TNT. The individual curves have been staggered for clarity.

Another important piece of information to know is how differential reﬂectograms of various explosives are distinguished from that of TNT. Figure 15.5 depicts R/R for RDX (1,3,5-trinitro-1,3,5-triazacyclohexane) (C4), HMX (1,3,5,7-tetranitro-1,3,5,7,-tetrazacyclooctane), nitroglycerin, pentaerythritol tetranitrate (PETN), Tetryl, and ammonium nitrate/fuel oil (ANFO). It can be seen that each type of explosive yields a different curve shape, caused by different critical energies for optical absorption. Speciﬁcally, the rise in R/R and the UV peaks for these substances are distinctly different and shifted to the near UV compared to TNT. It is therefore concluded that differential reﬂectometry can distinguish between various explosives and can thus positively identify the species at hand. In order to quickly discriminate between TNT and other organic substances, we developed a curve discrimination program based on the LabVIEW software (National Instruments). An average spectral distribution was derived from the R/R of several different TNT samples. An envelope region consisting of an upper 5% and a lower 5% deviation was utilized as the comparative standard. The computer program normalizes the spectrum of an “unknown” sample and compares it to the average TNT spectrum. This normalization procedure was done in order to not exclude weaker reﬂectors. The program has been shown to reliably detect TNT and clear other substances. Figure 15.6 depicts a static display of the discrimination procedure.

308

REMOTE SENSING OF EXPLOSIVE MATERIALS

9 8 7 6 5 − ΔR/R (a.u.)

On Metal 4 On Latex Glove

3 2 1 0 −1

On Cardboard

−2

On Fabric

−3

On Leather 200

250

300

350

400

450

500

Wavelength (nm)

Figure 15.4 Differential reﬂectograms of TNT on various surfaces. Individual curves have been staggered for clarity. Note: The lower and upper threshold wavelengths of the 400 nm shoulder are essentially alike. 10 9 8

− ΔR/R (a.u.)

7

HMX

6

TNT Nitroglycerin

5

RDX (C4)

4 PETN 3 Tetryl 2 ANFO 1 0 200

250

300

350

400

450

500

Wavelength (nm)

Figure 15.5 Differential reﬂectograms of several different explosives (commercial and pure). Individual curves are staggered for clarity.

CONCLUSIONS

Measured Curve

309

Wavelength range 225–300 nm (WL1) Left: Enveloping curves (green) and measured curve (white) Right: Difference between enveloping and measured curve (zero: within the enveloping curves) Calculated value for the difference

Warning lights : Product < 10,000 −> red (TNT) Product >10,000 but < 20,000 −> yellow (TNT?) Product > 20,000 −> green (no TNT)

Shows the peaks (location, amplitude, and 2nd derivative (negative=>maximum)

Figure 15.6

Result : Product (WL1*WL2*WL2)

See color plates. Screen shot of the LabVIEW curve recognition program.

The sensitivity of the present DR setup is estimated to be better than 10 μg/mm2 . This sensitivity can be probably improved in the lab, but it is unknown at this time if ﬁeld applications would yield better results.

15.4

CONCLUSIONS

It has been shown above that essentially all seven requirements for explosive detection listed in the Introduction can be met by differential reﬂectometry. In particular, scanning a suspected area with a light beam from a reasonable (but not yet determined) distance is paramount for safe and unsuspicious explosive detection. The technique can be developed into a device that is lightweight, rugged, and small. With the use of a CCD camera the method is fast (i.e., in the millisecond range). The probing light (dark-blue/UV) cannot be detected with infrared (IR) (night vision) goggles or with the human eye. The only drawback is that all optical methods can scan only the surface of suspected areas, that is, the light beam cannot penetrate containers, such as luggage. To accomplish this, one would have to sample the air inside a container by means of a vacuum and a ﬁltration device followed by the insertion of this ﬁlter into a differential reﬂectometer.

310

REMOTE SENSING OF EXPLOSIVE MATERIALS

ACKNOWLEDGMENTS

The Army Research Laboratory has partially supported the above-mentioned research. Views and conclusions contained in this document are those of the authors and should not be interpreted as representing the ofﬁcial policies, either expressed or implied, of the Army Research Laboratory or the U.S. government.

REFERENCES 1. Hummel, R. E. Phys. Stat. Sol. (a) 76, 11 (1983). 2. Hummel, R. E. Differential reﬂectance spectroscopy in analysis of surfaces, in R. A. Meyers, Ed. Encyclopedia of Analytical Chemistry, Wiley, Chichester, 2000, p. 9047–9071. 3. Hummel, R. E. Electronic Properties of Materials, 3rd ed. Springer, New York, 2001. 4. Pavia, D., G. Lampman, and G. Kriz. Introduction to Spectroscopy, Harcourt College, Fort Worth, 2001, pg.353.

PART IV

SUPPLEMENTARY MATERIAL

APPENDIX

ORGANIZATIONS INVOLVED IN SEARCHING FOR HIDDEN EXPLOSIVES CHARLES O. SCHMIDT CWO-4, U. S. Navy (retired)

The following pages list only a small portion of the organizations currently involved in demining, explosive ordnance disposal (EOD) and unexploded ordnance (UXO) clearance but are provided to give the researcher somewhere to begin. There are hundreds more organizations whose involvement ranges from providing a forum or database to humanitarian groups and technical websites. Our interest here is limited to those organizations with proven demining expertise. All listed Internet addresses were active as of late November 2005. Th authors can assume no responsibility for the contents or accuracy of these Internet sites. Mention of commercial entities, or lack of mention, does not imply any recommendation by the authors. INTERNATIONAL AND NONGOVERNMENTAL ORGANIZATIONS

United Nations Mine Action Service Department of Peacekeeping Operations 2 U.N. Plaza New York, New York 10017 Ph: 212.963.1875 www.mineaction.org or [email protected]

International policy, photos, meeting calendars, publications, and much more.

Norwegian Peoples Aid Norsk Folkehjelp. Storgt 33A, 9 etg 0028, Oslo Norway Ph: 22.03.77.00 www.Norskfolkehjelp.no

Detection dogs, mechanical clearance. Projects in Asmara, Eritrea, and Bosnia.

Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

313

314

ORGANIZATIONS INVOLVED IN SEARCHING FOR HIDDEN EXPLOSIVES

MgM (Menchen gegen Minen) Oberlinstr 8 D-40625, Dusseldorf Germany Ph: +49.(0).211.167.8841 www.mgm.org

Detection dogs, manual and mechanical demining, robotics, custom-designed machinery, many photos, training, networking, demining projects all over Africa.

InterSOS Via Nizza 154 00198, Roma Italy Ph: +39.06.8537431 www.intersos.org

Mine clearance projects in Iraq, Afghanistan, Angola, Bosnia, and Kuwait.

HELP (Hilfe zur Selbsthilfe eV) Reuterstr 39 D-53115, Bonn Germany Ph: +49.(0).228.91529.0 www.help-germany.de

Emergency aid and humanitarian demining, earthquake/tsunami relief. Projects in Afghanistan, Iraq, Sri Lanka, India, and Chechnya.

Danish Demining Group Borgerade 10 Postbox 53, 1002 Copenhagen K Denmark Ph: +45.3373.5112 www.danishdemininggroup.dk

Mine awareness and detection in Chechnya, Somaliland, and Afghanistan.

Humanitarian Landmine Disposal Fund—US 15496 Round Valley Drive Sherman Oaks CA, 90403 USA www.hldf.org

Airborne mine detection, armored demining vehicles, and explosive foam sprayer.

Association for Aid and Relief 5F Mizuho Bldg 2-12-2 Kamiosaki, Shinagawa-ku Tokyo 141-0021 Japan Ph: +81.3.5423.4511 www.aarjapan.gr.jp

Refugee support, ICBL Projects in Afghanistan, Angola, Cambodia, Myanmar, and Yugoslavia.

Mine Clearance Planning Agency House 58, Street 4, Phase 2 Hayatabad, Peshwar Pakistan Ph: +9291. 810.803 www.psh.paknet.com.pk

Mineﬁeld survey, marking, mapping, mine information systems in Afghanistan and Yemen.

COMMERCIAL DEMINING

315

Halo Trust PO Box 7712 London SW 1V 32A UK Ph: +44.207.730.2820 www.halotrust.org.

ERW clearance, manual and mechanical demining in Afghanistan, Cambodia, Mozambique, Angola, Eritrea, NW Somalia, Abkhazia, Nagorno Karabakh, and Sri Lanka.

Mine Detection Rats P.O. Box 649, Maputo Mozambique www.apopo.org Training Base: Rua Jasuna Machel, 512, Chimoio Ph/Fax: +258.51.23.977 [email protected]

Logistics and training of African giant pouched rats for explosive detection and tuberculosis diagnosis.

COMMERCIAL DEMINING

RONCO Consulting Corporation 2301 M Street/Suite 400 Washington, DC 20037 Ph: 202.785.2791 www.demining.com

Mine clearance, security services environmental remediation, and postconﬂict clearance operations.

Mine Tech International (Operations) 22 New York Avenue, Highlands, Harare Zimbabwe Ph: 00.2263.477.6216 www.minetech.co.uk

Manual, mechanical, and offshore demining, detection dogs, ordnance disposal, and security services. Projects in Sudan, Laos, Libya, Iraq, and 20 other countries.

MECHEM P.O. Box 14864 Lyttelton, 0140 South Africa Ph: +27.12.620.3403 [email protected]

Multiple, large-scale demining projects in Africa, detection dogs, training, manual and mechanical clearance and R&D.

Specialist Gurkha Services UK 171 Cove Road Farnborough, Hampshire, GU 14 0HQ UK Ph: +44.1252.510400 www.sqsukltd.com

Mine/UXO clearance in Sudan, Cambodia, Laos, Iraq, Angola, and China.

316

ORGANIZATIONS INVOLVED IN SEARCHING FOR HIDDEN EXPLOSIVES

MAAVARIM Demining MAAVARIM Civil Engineering LTD Moshav, Kidron P.O. Box 349 Israel, 70795 Ph: 972.8.8690636 www.maavarim.co.il

Humanitarian mine clearance, EOD, detection dogs’, Projects in Israel, Angola, and South Korea.

Milsearch Pty Ltd P.O. Box 687 Mawson, ACT, 2607 Australia Ph: +61.2.6286.8266 www.milsearch.com.au

Humanitarian demining, UXO clearance, underwater EOD and R&D. Projects in Australia, Bougainville, Manus, Tasmania, New Guinea, Laos, and Viet Nam.

Els (European Landmine Solutions) Redhill Chambers 2d High Street, Redhill, Surrey RH1 1RJ UK Ph: +44 (0).1737.789697 www.landmine-solutions.com

Mine clearance, EOD, training, diving, range clearance, detector dogs, mechanical clearance. Projects in Angola, Bosnia, Russia, Somaliland, Viet Nam, and Iraq.

UXB International, Inc. Blacksburg VA, 24060 Ph: 540.443.3700 www.uxb.com

Landmine/UXO surveys, clearance, awareness training, building local mine action, range clearance, site mapping, and GIS. Projects in Bosnia, Africa, and Middle East.

Bactec International Limited 37 Riverside, Sir Thomas Longley Road Rochester, Kent ME2 4DP UK Ph: 01634.296757 www.bactec.co.uk

EOD/landmine clearance, postconﬂict ERW removal, local training programs and mine awareness education. Projects for UK, the UN, and 28 other governments.

Planit EOD The Old Granary, Radwinter Road Ashdon, Saffron Waldon, Essex, CB10 2ET UK Ph: 0870.766.3210 www.planiteod.com

UXO clearance, site remediation, ERW clearance, independent QA, military range clearance, land mine clearance and demilitarization.

GOVERNMENTS

317

GOVERNMENTS

U.S. Department of State 2201 C Street NW Washington, DC, 20520 Ph: 202.647.4000 www.state.gov

U.S. policy and public programs.

Department of Homeland Security S&T-HSARPA Washington, DC, 20528 Ph: 202.254.6132 (Keith Ward/Angela Ervin) www.dhs.gov

R&D, sensors, counter-mine warfare, improvised explosives, airborne surveillance, bio-aerosol detection, container security, and vehicle IED protection.

Naval Research Laboratory 4555 Overlook Avenue SW Washington, DC, 20375 Chemistry Division/6112 Ph: 202.404.3337 (Greg E. Collins) [email protected] Optical Sciences Division/5611 Ph: 202.767.9473 (Hedi Mattoussi) [email protected],mil www.nrl.navy.mil

Laboratory on a chip for explosive detection, quantum dot-protein bioconjugates, biosensors, UAV radar, and much more in basic research.

U.S. Army RDECOM CERDEC NVESD (AMSRD-CER-NV-CM-HD) 10221 Burbeck Road/Suite 430 Fort Belvoir VA, 22060 www.humanitarian-demining.org

Detection, vegetation clearance, neutralization, personal protective equipment, land mine awareness, publications, and equipment testing.

Federal Bureau of Investigation Code EU 2501 Investigation Parkway Quantico, VA, 22135 Ph: 703.632.7643 (Kirk Yeager) www.leo.gov or www.fbiacademy.edu

IEDs, explosives, “dangerous innovations,” terrorist action coordination, and explosive physics.

Oak Ridge National Laboratory Nanoscale Sciences/MS 6123 Oak Ridge TN, 37831 Ph: 865.574.6201 (Thomas Thundat) www.ornl.gov

Surface effect microsensors, homeland security projects, and much in basic research.

318

ORGANIZATIONS INVOLVED IN SEARCHING FOR HIDDEN EXPLOSIVES

Naval Surface Warfare Center Panama City 110 Vernon Avenue Panama City, FL, 32407 Ph: 850.234.4784 (Mine Warfare) www.ncsc.navy.mil

Mine warfare, R&D, mine countermeasures, littoral and expeditionary warfare, test and evaluation of countermeasures, airborne, surface, and coastal mine hunting.

Naval Explosive Ordnance Disposal Technology Division 2005 Stump Neck Road Indian Head, MD, 20640 Ph: 301.743.3300 www.naveodtechdiv.navsea.navy.mil

EOD technical data and support.

Naval School, Explosive Ordnance Disposal 304 N. McCarthy Ave, Suite 117 Eglin AFB FL, 32542, Ph: 850.882.9080 www.npdc.navy.mil/ceneoddive/eods/

Navy portion of EOD training including NKO: Navy Knowledge Online (available on a restricted basis).

Ministry of Defense, Australia Business & Commercialization Ofﬁce PO Box 4331 Melbourne, VIC, 3001 Australia Ph: +61.3.9626.7247 www.dsto.defense.gov.au

Fourier transform ion cyclotron resonance mass spectronomy for chemical/biological toxins, distributed feedback ﬁber laser for SONAR arrays and the CORMORANT (demining) lift bag.

MILITARY SYSTEMS

The Boeing Company Boeing Integrated Defense Systems PO Box 516, St. Louis, MO, 63166 Ph: 314.232.0232 www.boeing.com

“Prospector” class III UAV for Army Future Combat Systems and land mine detection.

EDO Corporation New York, New York Ph: 212.716.2000 www.edocorp.com

Airborne mine countermeasures, AN/ALQ-219 Shallow Water Inﬂuence Minesweep System, (SWIMS), AN/ALQ-220 Organic Airborne and Surface Inﬂuence Sweep (OASIS).

EQUIPMENT

319

Lockheed Martin Corporation Business Development 6801 Rockledge Drive Bethesda, MD, 20817 Ph: 301.897.6000 www.lockheedmartin.com

Airborne Mine Countermeasures in MH-60S helo, Pathmaker Advanced Integrated Mine Warfare System, SEAFOX Airborne Mine Neutralization System and Littoral Combat Ship (LCS).

Northrop Grumman Systems Corporation Electronic Sensors and Systems PO Box 17319/MS A-255 Baltimore, MD, 21203 Ph: 410.993.6848 321.726.7526 (Jim Stratford) www.es.northropgrumman.com

Automatic Mine Detection System, AN/AQS-14 Post Mission Analysis System and Airborne Laser Mine Detection System (ALMDS).

Raytheon Company 870 Winter Street Waltham, MA, 02451 Ph: 781.522.3000 978.858.5246 (Guy Shields) www.ratheon.com

AN/SQS-20A mine hunting SONAR for MH-60S helo and Littoral Combat Ship and Undersea Surveillance System.

Thales Naval Division—UK Ocean House-Templecombe, Somerset, BA8 0DH UK, Ph: +44.(0).1963.370551 www.thales-naval.nl

Mine countermeasures (MCM) systems, UUVs and TSM 2022 Mk III minehunting SONAR.

Antheon International Corporation 3211 Jermantown Road Fairfax, VA, 22030 Ph: 703.246.0200 www.antheon.com

Support for USN MCM ships.

EQUIPMENT

Institut Dr. Foerster GmbH In Laisen 70 72766 Reutingen, Germany Ph: +49.7121.140.270 www.foerstergroup.de

Eddy current metal detectors and magnetometers.

320

ORGANIZATIONS INVOLVED IN SEARCHING FOR HIDDEN EXPLOSIVES

Minelab Electronics–US 871 Grier Drive, Suite B1 Las Vegas NV, 89119 Ph: 702.891.8809 www.minelabusa.com

Mine detectors, training, U.S. DoD Handheld Standoff Mine Detection System (HSTAMIDS), Australian Rapid Route and Area Mine Neutralization System.

Guartel Technologies Ltd. 4 Talina Centre Bagleys Lane London SW6 2BW UK Ph: +44.(0).207.384.3001 www.guartel.com

Mine detectors, EOD probes, safety equipment, R & D and destruction equipment.

Trevor Davies Engineering (PTY) Ltd PO Box 2474 Cresta 2118 Gauteng, South Africa Ph: + 27.11.952.1743 www.tde.co.za

“Pookie” lightweight mine locating road machines.

Nomadics Inc. 1024 South Innovative Way Stillwater, OK, 74074 Ph: 405.372.9535 (Colin Cumming) www.nomadics.com

Explosive detectors, chemical agent detectors, “Sea Dog,” biological agent detectors, and ﬂuorescent polymers

Syagen Technology Inc. 1411 Warner Avenue Tustin, CA, 92780 Ph: 714.258.4400 www.syagen.com

Personnel screening portal, trace detectors, and mass spectrometry of explosives.

QuinetiQ (US) 2345 Crystal Drive/Suite 909 IV Crystal Park Arlington, VA, 22202 Ph: 703.414.5454 www.quinetq.com

Defense systems, sensors, robotics, and port security systems.

Mistral Group Multiple ofﬁces worldwide See website Ph: 1.800.MISTRAL www.mistralgroup.com

Detection equipment for explosives, contraband, and drugs. Perimeter security, vehicle/personnel armor, blast containment trashcans, EOD, Site survey, blast protection, antiterrorism, and security training.

UNIVERSITY RESEARCH

321

UNIVERSITY RESEARCH

Swiss Federal Institute of Technology Development Center for Security Studies Zentrum FSK-SEI CH 8092 Zurich, Switzerland Ph: +41.1.632.63 65 www.imsma.ethz.ch

Information management system for mine action.

Cranﬁeld Resilience Centre Cranﬁeld University Royal Military College of Science Shrivenham, Swindon Wiltshire SN6 8LA UK Ph: +44.0.1793.782551 www.rmcs.cranﬁeld.ac.uk

Development of mine action training programs for UN and assistance with nation’s strategic planning for mine action.

University of Rhode Island Department of Chemistry 51 Lower College Road Kingston, RI, 02881 Ph: 401.874.2103 (Dr. J. Oxley) www.chm.uri.edu

Chemistry of energetic materials and explosives detection.

Georgia Institute of Technology School of Civil/Environmental Engineering 790 Atlantic Drive Atlanta, GA, 30332 Ph: 404.894.6704 (Dr. Donald Webster) www.ce.gatech.edu

Environmental ﬂuid mechanics and turbulence of chemical plumes.

New Mexico Institute of Mining and Technology Energetic Materials Research and Testing Center 801 Leroy Place Socorro NM 87801 Ph: 505.835.5011 www.emrtc.nmt.edu

Energetic materials research, ﬁrst responder training, safety, ﬁre academy, antiterrorism assistance program, aircraft safety, explosive materials classiﬁcation, and custom ﬁeld testing of hazardous material.

322

ORGANIZATIONS INVOLVED IN SEARCHING FOR HIDDEN EXPLOSIVES

The University of Western Australia Department of Mechanical & Materials Engineering Nedlands 6907 Western Australia Ph: +61.8.6488.3057 (James Trevelyn) www.mech.uwa.edu.au/jpt

Specialized tools and equipment for deminers. Excavators, waterjet cutters, vegetation removal, PPE, detector accessories. Tools designed for speciﬁc areas.

INFORMATION/DATA BASES AND LINKS

Geneva International Centre for Humanitarian Demining 7 bis, avenue de la Paix PO Box 1300 CH-1211 Geneva 1 Switzerland Ph: +41.22.906.16.60 www.gichd.ch/

International forums, international mine action standards, manual and mechanical clearance, dogs, risk education, equipment, EOD/UXO safety, accident database, victim assistance, and much more.

James Madison University University Boulevard Harrisonburg VA, 22807 Ph: 540.568.2756 www.jmu.edu

Humanitarian demining staff, links to other sites, Demining Index, electronic journal, conferences, Journal of Mine Action and Global Mine Action Registry.

Jane’s Information Group Jane’s Mines and Mine Clearance 110 N. Royal St/Suite 200 Alexandria VA 22314 Ph: 703.683.3700 www.jmmc.janes.com

Comprehensive review of mine warfare, mine clearance, and general EOD methods.

Mine Warfare Association c/o DMC Companies 824 Munras Avenue, Suite C Monterey CA, 93940 Ph: 831.373.0508 www.minwara.org

Mine warfare forum, information, meetings, symposia and Mine Lines newsletter.

EMERITUS

323

EMERITUS

Dr. Vernon Joynt PO Box 912454 Silverton, 0127, South Africa [email protected]

Decades of demining experience in Africa (see MECHEM), personal recollections on mine hunting.

This is a story that came while the book was in press. Dr. Vernon Joynt told it during the Chemical Sensors Technical Session at the 7th International Symposium on Technology and the Mine Problem, held at the Naval Postgraduate School, Monterey, California, in May, 2006. We were fascinated. Here it is in Dr. Joynt’s words: Theo van Dyk and some of MECHEM’s mine clearers spent a weekend out at a resort near Pretoria where the animal warden made a statement that elephants can smell mines and avoid unmarked mineﬁelds in the wild. Not only that, he said they teach other members of the herd the smell and then all of them will avoid wandering into the danger. When Theo related this to us on the Monday I asked him to take the man on and go check if there was any truth in the story. Theo was at that stage involved in having MEDDS dogs trained so he prepared some MEDDS ﬁlters as blanks and one with TNT smell on. We went out to the resort and were given a demonstration by the warden of how his two year old elephants, some ten of them, could smell what was on the ﬁlters. The key was that one young elephant was from Zimbabwe and had experienced mineﬁelds that had hurt fellow elephants. The warden was not told which ﬁlter was positive and which not so we simply put all on MEDDS stands and he told the trained elephants to pick up the ﬁlters off the stands. All did so willingly and enjoyed the game. Lo and behold, the Zimbabwean youngster violently shied away from the stand with the TNT ﬁlter on and refused to take the ﬁlter as soon as he extended his trunk to do as commanded. One for the warden! Now for the next. He asked us to leave the ﬁlters with him and he will get the Zimbabwean elephant to show the others which was the “dangerous smell.” Next weekend the warden won his second bet! All the other elephants would not go near a new TNT ﬁlter amongst a new batch of blanks! From there we then concluded that if IR records from satellites which were able to follow the movements of herds of elephants would show spots with good food that they avoid, then the presence of mines or arms catches should be checked for.

DEFINITIONS, SYMBOLS AND ABBREVIATIONS

This book uses many different acronyms, symbols, and abbreviations. Unless they are in wide common use, they are normally deﬁned at ﬁrst usage. These pages are inserted to assist the reader who enters the text at another point to recall their meaning. Items that appear on only one page or that are peculiar to a single chapter may not be listed. Since several ﬁelds of science and engineering are represented in the different chapters, it is possible that some commonly used symbols, especially single letter symbols, for example L and R, will have different meanings in different chapters. In these cases they are deﬁned within each chapter.

ACRONYMS

βCD μECD AFP ALMDS ANIS AP APCI API API/TOFMS

β Cyclodextrin - (C4 H10 O5 )7 Microelectron capture device Amplifying ﬂuorescent polymer Airborne laser mine detection system Ammonium nitrate icing sugar Antipersonnel (as in mines) Atmospheric pressure chemical ionization Atmospheric pressure ionization Atmospheric pressure ionization time-of-ﬂight mass spectrometer

Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

325

326

DEFINITIONS, SYMBOLS AND ABBREVIATIONS

APOPO ASGDI ASGDI MS/MS AT AUV BBI BIPS molecule BL CCD C-CD CE CE-EC CID CL CO-OP COTS CRRDC CROMAC CSIR CSME CSS DARPA DHLA DoD DOE DoT DR DRS DSP EC ECD ECNIMS EPA EDS ESD EF&T EOD EOF

APOPO project: African Giant Pouched Rats, SUA Atmospheric Sampling–Glow Discharge–Ion Trap Atmospheric sampling–Glow Discharge–Ion Trap–Tandem Mass Spectrometer Antitank (as in mines) Autonomous underwater vehicle Improvised Munitions Black Book Vol. 1 1 ,3-Dihydro-1 -(2-carboxyethyl)-3,3-dimethyl-6-nitrospiro [2H-1-benzopyran-2,2 -(2H)-indoline] Bioluminescence detector Charge-coupled device Contactless–conductivity detector Capillary electrophoresis Capillary electrophoresis-Electrochemistry Collision-induced dissociation Chemiluminescence detector Sodium chlorate-nitrobenzene Commercial off-the-shelf (U.S. Army) Cold Regions Research and Development Center Croatian Mine Action Center Council of Scientiﬁc and Industrial Research, South Africa Chemical sensing in the marine environment (U.S. Navy) Coastal Systems Station, Panama City, Florida (U.S.) Defense Advanced Projects Agency Dihydrolipoic acid (U.S.) Department of Defense (U.S.) Department of Energy (U.S.) Department of Transportation Differential reﬂectrometry = differential reﬂection spectrometry Differential reﬂection spectrometry = differential reﬂectrometry digital signal processor Electrochemistry Electron-capture device Electron-capture negative-ion mass spectrometry (U.S.) Environmental Protection Agency Explosive detection system Electrostatic discharge Environmental fate and transport Explosive ordnance disposal Electroosmotic ﬂow

ACRONYMS

ERC ERW ESD ETD EVD FAA FAIMS FGAN FIDO FIS FMX FOI FRET FTMS FWHM GA GC GC/DMS GC/ECD GC/IMS GC/MS GC/MS GC/SAW GC-CL GICHD GPR HD HIS HSTAMIDS HMM HOMO HPLC-EC HPLC-UV ICAO ICBL IE IED IMS INL IR IRS ITIMS LCS LED

327

Explosive-related chemical Explosive remnants of war Electrostatic discharge Explosives trace detection Explosive vapor detection (U.S.) Federal Aviation Administration Field asymmetric ion mobility spectrometry Fertilizer-Grade AN (ammonium nitrate) Trade name for Nomadics, Inc.’s Trace Chemical Sensor (not an acronym) Field ion spectrometer The Revised Black Book Totalf¨orsvarets forskningsinstitut, Sweden F¨orster resonance energy transfer Fourier transform mass spectrometry Full width, half maximum Guerrilla’s arsenal Gas chromatography Gas chromatography/differential ion mobility spectrometer Gas chromatography/electron-capture detector Gas chromatography/ion mobility spectrometer Gas chromatography/mass spectrometer Gas Chromatography/Mass Spectrometer Gas chromatography/surface acoustic wave Gas chromatography—chemiluminescence Geneva International Center for Humanitarian Demining Ground penetrating radar Humanitarian demining Histidine Handheld, standoff, mine detection system Hidden Markov method Highest occupied molecular orbital state High-performance liquid chromatograph—electron capture High-performance liquid chromatograph—ultra violet International Civil Aviation Organization International Campaign to Ban Landmines Improvised explosive Improvised explosive device Ion mobility spectrometry Idaho National Laboratory Infrared Infrared spectroscopy Ion trap ion mobility spectrometry Littoral combat ship Light-emitting diode

328

DEFINITIONS, SYMBOLS AND ABBREVIATIONS

LIBS LIDAR LIF LOC LOC/HPLC LOD LUMO MBA MCM MBP MDD Mechem MEDDS MEMS MIP MIT MRV MS MSn MS/MS MS-CL NESTT NLDIMS NMR NOKSH NPA NQR NVESD OASIS O.B. ONR PA PDK PIRA PL PMA1,2; PPMA PMDS PMDS/DVB PMMA PMT PPE PVC

Laser-induced breakdown spectroscopy Light detection and ranging Laser-induced ﬂuorescence Lab-on-a-chip Lab-on-a-chip/high-performance liquid chromatography Limit of detection Lowest unoccupied molecular orbital state Mercaptobenzoic acid (4-MBA is also known as thiosalicylic acid) Mine countermeasures Maltose binding protein Mine detecting dog South African company doing chemical research and demining Mechem Explosive and Drug Detection System Microelectrical and mechanical systems Molecularly imprinted polymer Massachusetts Institute of Technology Mine removal vehicle Mass spectrometer Multistage (n times) mass spectrometer Tandem mass spectrometer Mass spectrometer—chemiluminescence Nonhazardous explosive for security training and testing Nonlinear dependence of ion mobility Nuclear magnetic resonance Norsk Kompetansesenter for Specialsøkshund, Norway Norwegian People’s Aid Nuclear quadrupole resonance (U.S. Army) Night Vision & Electronic Sensors Division Organic airborne and surface inﬂuence sweep Oxygen balance (U.S.) Ofﬁce of Naval Research Picric acid Peroxide detection bit Provisional Irish Republican Army Photoluminescence Soviet Block antipersonnel mines Polydimethylsiloxene (polymer used to concentrate explosive molecules) Polydimethylsiloxene/divynlbenzene (copolymer used as concentrator) Poly(methylmethacrylate) Photomultiplier tubes Personal protective equipment Polyvinylchloride

SYMBOLS AND ABBREVIATIONS

QA QCM QD QIT QitTofMS QY R&D READ REMUS REST RF RHD ROC RRAMNS RSD SAM SeaDog SeaPup SDS SERDP SPAWAR SPE SPME SWIMS TEA TM62 TMM-1, TMA-5 TOFMS TR TSA UAV UNi UUV UUXO UV UXB UXO VBIED WHOI WTC

329

Quality assurance Quartz crystal microbalance Quantum dot Quadrupole ion trap Quadrupole ion trap, time of ﬂight, mass spectrometer Quantum yield Research and development Reversal electron attachment detection Remote environmental monitoring units Remote explosive scent tracking Radio frequency Ragnar’s Homemade Detonators Receiver operator characteristics (a graphical portrayal of PD and PFp ) (Australian) rapid route and area mine neutralization system Relative standard deviation = standard deviation ÷ mean Self-assembled monolayer Nomadics Inc. Adaptation of the Fido sensor Miniaturized SeaDog Sodium dodecyl sulfate (U.S. DoD, DOE, EPA) Strategic Environmental Research and Development Program (U.S. Navy) Space and Naval Warfare Systems Command Solid-phase extraction Solid-phase microextraction Shallow water inﬂuence minesweep system Thermal energy analyzer Soviet Block antitank mine Soviet Block antitank mines Time-of-ﬂight mass spectrometry Thermo-redox (U.S.) Transportation Security Administration Unmanned air vehicle Urea nitrate Unmanned underwater vehicle Underwater unexploded ordnance Ultraviolet Unexploded bomb(s) Unexploded ordnance Vehicle-borne improvised explosive device Wood’s Hole Oceanographic Institute World Trade Center

330

DEFINITIONS, SYMBOLS AND ABBREVIATIONS

SYMBOLS AND ABBREVIATIONS

μg μm μM ν ag E E fg g J L M mg mM MW or mw M/z N ng nM Nm or nm pg ppb ppm ppq ppt psi PD PFA

microgram micrometer aka micron micromole Poisson’s ratio attogram applied potential, Chapter 13 Young’s modulus or modulus of elasticity, Chapter 12 femtogram gram joule liter mole milligram 1 millimole = 1000 mole molecular weight mass-to-charge ratio of an ion newton nanogram nanomole nanometer picogram parts per billion parts per million parts per quadrillion parts per trillion pounds per square inch probability of detection probability of false alarm

PFP

probability of false positive

R

resolving power of an instrument or radius of curvature or reﬂectivity volt watt

v w

10−6 g

10−18 g

10−15 g 100 g

10−3 g

10−9 g 10−12 g 1 part in 1 part in 1 part in 1 part in

109 106 1015 1012

PFA , PFP are equivalent expressions PFA , PFP are equivalent expressions

EXPLOSIVES DEFINITIONS RONALD L. WOODFIN

Amatol

80–40% AN, 20–60% TNT

ADNT

Amino dinitrotoluene

C7 H7 N3 O4

AN

Ammonium nitrate

NH4 NO3

ANFO

Ammonium nitrate—fuel oil

94% AN, 6% diesel

Baratol

76% barium nitrate, 24% TNT

C-4 Comp-B

91% RDX, 9% plasticizer Composition B

39% TNT, 60% RDX, 1% wax

Comp-B2

40% TNT, 60% RDX

CO-OP

Sodium chlorate-nitrobenzene

Cyclotol DADP

70–75% RDX, 25–30% TNT Diacetone diperoxide

Detasheet

C6 H12 O6 PETN, NC, and binders

DDNP

Diazodinitrophenol

C6 H2 N4 O5

DMNB

Dimethylnitrobutane

C6 H12 (NO2 )2

DNB

Dinitrobenzene

C6 H4 (NO2 )2

DNT

Dinitrotoluene

C6 H3 (NO2 )2 CH3

DNT

Dinitrotoluene

C6 H3 (NO2 )2 CH3

Dynamite

NG and/or NG/EGDN with binders

EGDN

Ethylene glycol dinitrate

C2 H4 (NO3 )2

HMTD

Hexamethylene triperoxidediamine

C6 H12 N2 O6

Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

331

332

EXPLOSIVES DEFINITIONS

HMX

Octahydro-1,3,5,7,tetranitro-1,3,4,5tetrazocine

C4 H8 N8 O8

Lead azide

Pb(N3 )2

Mercury fulminate

Hg(ONC)2

MNT

Mononitrotoluene, same as NT

C7 H7 NO2

NB

Nitrobenzene

C6 H5 NO2

NC

Nitrocellulose (guncotton)

C12 H14 (ONO2 )6 O4

NG

Nitroglycerin

C3 H5 N3 O9

NT

Nitrotoluene

NO2 C6 H4 CH3

Octol

70–75% HMX, 25–30% TNT

Pentolite

50% PETN, 50% TNT

PE-4∗

UK equiv of C-4

∼91% RDX, 9% Plasticizer

PETN

Pentaerythritol tetranitrate

C5 H8 N4 O12

Picric acid

2,4,6-Trinitrophenol

C6 H2 (NO2 )3 OH

RDX

∗∗

Cyclotrimethylenetrinitramine C3 H6 N6 O6

Semtexa

RDX, PETN varying proportion

TATP

Triacetone triperoxide

(CH3 )6 O6

Tetryl

Trinitrophenyl-nmethylnitramine

C7 H5 N5 O8

TNB

Trinitrobenzene

C6 H3 (NO2 )3

TNT

Trinitrotoluene

C6 H2 (NO2 )3 CH3 70% Comp-B, 18%Alb , 12% TNT

Torpex-2 Tritonol UNi

80–20% Al, 20–80% TNT Urea nitrate

c

DMNB

NH2 CONH2 –HNO3 2,3-Dimethyl-2,3-dinitrobutane

a

Semtex is offered in several forms. See catalog at http://www.explosia.cz/en/trhaviny/download/ trhaviny.pdf, visited 12/3/05. b Aluminum. c Taggant compound for explosives. ∗

PE-4 is essentially the same as C-4, but with different plasticizen. Proportions may differ slightly from C-4. ∗∗ RDX is also known as cyclonite, hexogen, or T4.

BIBLIOGRAPHY

Alivisatos, A. P. The use of nanocrystals in biological detection. Nat. Biotech. 22, 47–52 (2004). Arakawa, E. T., N. V. Lavrick, S. Rajic, and P. G. Datskos. Ultramicroscopy 97, 459 (2003). Asano, K. G., D. E., Goeringer, and S. A. McLuckey. Anal. Chem. 67, 2739 (1995). Atema, J. Eddy chemotaxis and odor landscapes: Exploration of nature with animal sensors. Biol. Bull. 191, 129–138 (1996). Atmospheric Dispersion Modeling Liaison Committee (ADMLC), http://www.admlc. org.uk. Baeyer, A. and V. Villiger. Berichte der Deutschen Chemischen Gesellschaft 33, 2479 (1900). Bara, B. M., D. J. Wilson, and B. W. Zelt. Concentration ﬂuctuation proﬁles from a water channel simulation of a ground level release. Atmos. Environ. 26A, 1053–1062 (1992). Barnes, J. R., R. J. Stephenson, C. N. Woodburn, S. J. Oshea, M. E. Welland, T. Rayment, J. K. Gimzewski, and C. Gerber. Nature 372, 79 (1994). Batchelor, G. K. Small-scale variation of convected quantities like temperature in turbulent ﬂuid. Part 1. General discussion and the case of small conductivity. J. Fluid Mech. 5, 113–133 (1959). Bender, S. F. A., P. J. Rodacy, R. L. Schmitt, P. J. Hargis, Jr., M. S. Johnson, J. R. Klarkowski, G. I. Magee, and G. L. Bender. Tracking Honey Bees Using LIDAR (Light Detection and Ranging) Technology, SAND2003-0184. Sandia National Laboratories Report, Albuquerque, NM, 2003. Bennett, G., R. Sleeman, W. R. Davidson, and W. R. Stott. Proc. SPIE 2276, 363 (1994). Benson, R. Ragnar’s Homemade Detonators: How to Make ‘Em, How to Salvage ‘Em, How to Detonate ‘Em, Paladin Press, Boulder, CO, 1993. Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

333

334

BIBLIOGRAPHY

Boisen, A., J. Thaysen, H. Jensenius, and O. Hansen. Ultramicroscopy 82, 11 (2000). Bottoms, A. M. and C. Scandrett, Eds. Applications of Technology to Demining, An Anthology of Scientiﬁc Papers (1995–2005), in three volumes: Part I, Vols. 1 and 2, Lamdmine Countermeasures, Part II, Naval Mine Countermeasures. Society for Counter-Ordnance Technology (SCOT), http://demine.org, July 2005. Boumsellek, S. and A. Chutjian. Anal. Chem. 64, 2096 (1992). Boumsellek, S., S. H., Alajajian, and A. Chutjian. J. Am. Soc. Mass Spectrom. 3, 243 (1992). Bratin, K., P. T. Kissinger, R. Briner, and C. Bruntlett. Determination of nitro aromatic, nitramine, and nitrate ester explosive compounds in explosives mixtures and gunshot residue by liquid-chromatography and reductive electrochemical detection. Anal. Chim. Acta 130, 295–311 (1981). Bromberg, A. and R. A. Mathies. Multichannel homogeneous immunoassay for detection of 2,4,6-trinitrotoluene (TNT) using a microfabricated capillary array electrophoresis chip. Electrophoresis, 25, 1895–1900 (2004). Bromberg, A. and R. A. Mathies. Homogeneous immunoassay for detection of TNT and its analogues on a microfabricated capillary electrophoresis chip. Anal. Chem. 75, 1188–1195 (2003). Bromenshenk, J. J., C. B. Henderson, R. A. Seccomb, S. D. Rice, R. T. Etter, S. F. A. Bender, P. J. Rodacy, J. A. Shaw, N. L. Seldomridge, L. H. Spangler, and J. J. Wilson. Can honeybees assist in area reduction and landmine detection? Journal of Mine Action, Issue 7.3, Harrisonburg, VA, December 2003. Broyles, B. S., S. C. Jacobson, and J. M. Ramsey. Sample ﬁltration, concentration, and separation integrated on microﬂuidic devices. Anal. Chem. 75, 2761–2767 (2003). Bruchez, M., M. Moronne, P. Gin, S. Weiss, and A. P. Alivisatos. Semiconductor nanocrystals as ﬂuorescent biological labels. Science 28, 2013–2016 (1998). Bruschini, C. Commercial Systems for the Direct Detection of Explosives (for Ordnance Disposal Tasks), ExploStudy, Final Report, Ecole Polytechnique Federale de Lausanne, Lausanne, Switzerland, Feb. 17, 2001. Buckles, B. A. Fed. Reg. 67, 20864 (2002). Butt, H. J. J. Colloid Interface Sci. 180, 25 (1996). Ceriotti, L., N. F. de Rooij, and E. Verpoorte. An integrated fritless column for on-chip capillary electrochromatography with conventional stationary phases. Anal. Chem. 74, 639–647 (2002). Chamberlain, T., K. Hanold, M. Hanning-Lee, Y. Liu, J. Syage, K. Linker, C. Rhykerd, and R. Bouchier. Multi-threat mass spectrometer detector. Proc. of the ICAO Workshop, DERA, UK, 2000. Chambers, D. M., S. A. McLuckey, and G. L. Glish. Anal. Chem. 65, 778 (1993). Chambers, W. B., P. J. Rodacy, E. E. Jones, B. J. Gomez, and R. L. Woodﬁn. Chemical sensing system for classiﬁcation of minelike objects by explosive detection, in A. C. Dubey, J. F. Harvey, J. T. Broach, eds. Proceedings of the SPIE 12th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, Detection and Remediation Technologies for Mines and Minelike Targets III, April 13–17, 1998. Chan, W. C. W. and S. Nie. Quantum dot bioconjugates for ultrasensitive nonisotopic detection. Science 281, 2016–2018 (1998). Chen, G. and J. Wang. Fast and simple sample introduction for capillary electrophoresis Microsystems. Analyst, 129, 507–511 2004.

BIBLIOGRAPHY

335

Clapp, A. R., I. L. Medintz, J. M. Mauro, B. R. Fisher, M. G. Bawendi, and H. Mattoussi. Fluorescence resonance energy transfer between quantum dot donors and dye-labeled protein acceptors. J. Am. Chem. Soc. 126, 301–310 (2004). Collins, G. E., P. Wu, Q. Lu, J. D. Ramsey, and R. H. Bromund. Compact, high voltage power supply for the lab-on-a-chip. Lab on a Chip 4, 408–411 (2004). Colton, R. J. and J. N. Russell. Science 299, 1324 (2003). Cooper, P. W. Explosives Engineering. Wiley-VCH, New York, 1996. Cox, J. A quantum paintbox. Chem. Brit. 39, 21–25 (2003). Crimaldi, J. P. and J. R. Koseff. High resolution measurements of the spatial and temporal scalar structure of a turbulent plume. Exp. Fluids, 31, 90–102 (2001). Crimaldi, J. P., M. B. Wiley, and J. R. Koseff. The relationship between mean and instantaneous structure in turbulent passive scalar plumes. J. Turbul. 3(014), 1–24 (2002). Cumming, C., C. Aker, M. Fisher, M. Fox, M. laGrone, D. Reust, M. Rockley, T. Swager, E. Towers, and V. Williams. Using novel ﬂuorescent polymers as sensory materials for above-ground sensing of chemical signature compounds emanating from buried landmines. IEEE Trans. Geoscie. Remote Sensing 39(6), 1119–1128 (2001a). Cumming, C., C. Aker, M. Fisher, M. Fox, M. la Grone, D. Reust, M. Rockley, T. Swager, E. Towers, and V. Williams. Using novel ﬂuorescent polymers as sensory materials for above-ground sensing of chemical signature compounds emanating from buried landmines, in Proceedings of the UXO/Countermine Forum, New Orleans, LA, April, 2001b. Dabbousi, B. O., J. Rodriguez-Viejo, F. V. Mikulec, J. R. Heine, H. Mattoussi, R. Ober, K. J. Jensen, and M. G. Bawendi. (CdSe)ZnS core-shell quantum dots: Synthesis and optical and structural characterization of a size series of highly luminescent materials. J. Phys. Chem. B 101, 9463–9475 (1997). Dasi, L. P. Statistical Characteristics of Turbulent Chemical Plumes. Master’s Thesis. Georgia Institute of Technology, Atlanta, GA, July 2000. Darrach, M. R., A. Chutjian, and G. A. Plett. Trace Explosives Signatures from World War II Unexploded Undersea Ordnance. Jet Propulsion Laboratory, Environmental Science Technology, 32(9), 1354 (1998). Data-Base of Range Evaluated Improvised Explosives (D-BREIE) Phase I: Ammonium Nitrate Based Explosives, Kirk Yeager, EMRTC Report to TSWG, July 1999. Davis, T. L. The Chemistry of Powder and Explosives. Wiley, New York, 1956. DeGraaff, D. B. and J. K. Eaton. Reynolds-number scaling of the ﬂat-plate turbulent boundary layer. J. Fluid Mech. 422, 319–346 (2000). Ding, L., E. Kawatoh, K. Tanaka, A. J., Smith, and S. Kumashiro. Proc. SPIE 3777, 144 (1999). Dispersion Modeling Feature Articles, http://www.environmental-expert.com/articleindex.htp. Dock, M., J. Sikes, M. Fisher, and M. Prather. Chemical detection of underwater explosives. 2004 Mine Countermeasures & Demining Conference/MINWARA, Canberra, Australia, February 2004. Dock, M., F., Mark, and C. Colin. Novel detection apparatus for locating underwater unexploded ordnance, in Proceedings of the 5th International Symposium on Technology and the Mine Problem, Mine Warfare Association, Monterey, CA, April 2002a. Dock, M., M. Fisher, and C. Cumming. Sensor for real-time detection of underwater unexploded ordnance, in Proceedings of UXO/Countermine Forum 2002, September 2002b, Orlando, FL.

336

BIBLIOGRAPHY

Drafts, B. Acoustic Wave Technology Sensors, Sensors Magazine Online, October 2000, visited 9/19/05, http://www.sensorsmag.com/articles/1000/68/index.htm. Dusenbery, D. B. Sensory Ecology, W.H. Freeman, New York, 1992. Eiceman, G. and Z. Karpas. Ion Mobility Spectrometry, 2nd ed. CRC Press, Boca Raton, FL, 2005. Eiceman, G. A. and Z. Karpas. Ion Mobility Spectrometry. CRC Press, Boca Raton, FL, 1994. Encyclopedia of Explosives and Related Items Vol. 10, Seymour Kage, Ed. U.S. Army ARRADCOM, Dover, NJ, 1983, p. U103. Ermakov, S. V., S. C. Jacobson, and J. M. Ramsey. Computer simulations of electrokinetic injection techniques in microﬂuidic devices. Anal. Chem. 72, 3512–3517 (2000). Fackrell, J. E. and A. G. Robins. Concentration ﬂuctuations and ﬂuxes in plumes from point sources in a turbulent boundary layer. J. Fluid Mech. 117, 1–26 (1982). Fainberg, A. Science 255, 1531 (2001). Fair, R. B., A. Khlystov, V. Srinivasan, V. K. Pamula, and K. N. Weaver, Integrated chemical/biochemical sample collection, pre-concentration, and analysis on a digital microﬂuidic lab-on-a-chip platform. Proc. SPIE 5591, 113–124 (2004). Farrell, J. A., S. Pang, and W. Li. Chemical lume tracing via an autonomous underwater vehicle. IEEE J. Oceanic Eng. 30(1), 1–15 (2005). Farrell, J. A., S. Pang, and W. Li. Plume mapping via hidden Markov methods. IEEE Trans. Systems, Man, Cybernetics—Part B: Cybernetics. 33(6), 850–863 (2003). Farrell, J. A., J. Murlis, X. Long, W. Li, and R. T. Card´e. Filament-based atmospheric dispersion model to achieve short time scale structure of odor plumes. Environ. Fluid Mechanics 2, 143–169 (2002). Federal Coordinator for Meteorological Services and Supporting Research. Directory of Atmospheric Transport and Diffusion Consequence Assessment Models, FCM-I3-1999, Washington, DC, March 1999, http://www.ofcm.gov. Ferner, M. C. and M. J. Weissburg. Slow-moving predatory gastropods track prey odors in fast and turbulent ﬂow. J. Exp. Biol. 208, 809–819 (2005). Field Engineering and Mine Warfare, Pamphlet No. 6, Detection and Clearance of Mines, British War Ofﬁce, 1947. Figeys, D. and D. Pinto, Lab-on-a-chip: A revolution in biological and medical sciences. Anal. Chem. 71, 330A–335A (2000). Fisher, M. Applications of sensors utilizing amplifying ﬂuorescent polymers for ultratrace level detection of explosives. Eighth International Symposium for the Analysis and Detection of Explosives (ISADE), Ottawa, Canada, June 2004. Fisher, M. and C. Cumming. Trace detection of nitroaromatic explosives by ﬂuorescence quenching of novel polymer materials, in Proceedings of the U.S. Federal Aviation Administration’s Third International Aviation Security Technology Symposium, Atlantic City, NJ, November 27–30, 2001a. Fisher, M. and C. Cumming. Detection of trace concentrations of vapor phase nitroaromatic explosives by ﬂuorescence quenching of novel polymer materials, in Proceedings of 7th International Symposium on the Analysis and Detection of Explosives, Defense Evaluation and Research Agency, Edinburgh, Scotland, UK, June, 2001b. Fisher, M. and C. Cumming. Utilization of novel ﬂuorescent polymer materials for trace level vapor-phase detection of nitroaromatic explosives, in Proceedings of the U.S. Federal Aviation Administration’s Third International Aviation Security Technology Symposium, Atlantic City, NJ, November 27–30, 2001c.

BIBLIOGRAPHY

337

Fisher, M., C. Cumming. Fluorescent polymer chemosensors for explosives detection (Poster Session). Fourteenth International Forum, Process Analytical Chemistry, Las Vegas, Nevada, January 2000 Fisher, M. and J. Sikes. Mineﬁeld edge detection using a novel chemical vapor sensing technique, in R. S. Harmon, J. H. Holloway, Jr., and J. T. Broach, Eds. Detection and Remediation Technologies for Mines and Minelike Targets VIII. Proceedings of SPIE, Vol. 5089, pp. 1078–1087, April 21–25, 2003a, Orlando, FL. Fisher, M. and J. Sikes. Detection of landmines and other explosives with an ultra-trace chemical detector, NATO e-Nose Conference, October 2003b, Coventry, UK. Fisher, M., C. Cumming, and M. Prather. Detection of ultra-trace landmine chemical signatures using novel sampling and sensing strategies. Poster Presentation, Gordon Research Conference, Il Ciocco, Italy, June 2003a. Fisher, M., M. la Grone, and J. Sikes. Implementation of serial amplifying ﬂuorescent polymer arrays for enhanced chemical vapor sensing of landmines, in Proceedings of UXO/Countermine Forum 2003, September 2003b, Orlando, FL. Fisher, M., M. Prather, and J. Sikes. Serial amplifying ﬂuorescent polymer arrays for enhanced chemical vapor sensing of landmines, EUDEM-2/SCOT Conference, Brussels, Belgium, September 2003c. Fisher, M., M. la Grone, C. Cumming, and E. Towers. Utilization of chemical vapor detection of explosives as a means of rapid mineﬁeld area reduction, in Proceedings of the 5th International Symposium on Technology and the Mine Problem, Mine Warfare Association, Monterrey, CA, April 2002. Fjellaner, R. The REST concept, in Mine Detection Dogs: Training, Operations and Odour Detection, I. G. McLean, Ed. Geneva International Centre for Humanitarian Demining (GICHD), Geneva, 2003, pp. 53–107. Fliermans, C. B. and G. Lopez-de-Victoria. Microbial mine detection system (MMMDS), in A. C. Dubey, J. F. Harvey, J. T. Broach, Eds. Proceedings of the SPIE 12th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, Detection and Remediation Technologies for Mines and Minelike Targets III, April 13–17, 1998. Fong, D. A. and M. T. Stacey. Horizontal dispersion of a near-bed coastal plume. J. Fluid Mech. 489, 239–267 (2003). Fritz, J., M. K. Baller, H. P. Lang, T. Strunz, E. Meyer, H. J. Guntherodt, E. Delamarche, C. Gerber, and J. K. Gimzewski. Science 288, 316 (2000). Garner, K. and S. Smith. Volatile Organic Compounds, the Good, the Bad and the Analysis, SPME. University of the West of England, Bristol. Updated 9/23/04; visited 9/19/05; http://www.chemsoc.org/exemplarchem/entries/2004/westengland smith/ExempWeb/ methdev.htm. George, V., T. F. Jenkins, J. M. Phelan, D. C. Leggett, J. Oxley, S. W. Webb, P. H. Miyares, J. H. Cragin, J. Smith, and T. E. Berry, Progress on determining the vapor signature of a buried landmine. Proc. SPIE, Detection and Remediation Technologies for Mines and Minelike Targets V, Volume 4038, part 1, p290 (2000). George, V., T. F. Jenkins, D. C. Leggett, J. H. Cragin, J. Phelan, J. Oxley, and J. Pennington. Progress on determining the vapor signature of a buried landmine. Proc. SPIE, Detection and Remediation Technologies for Mines and Minelike Targets IV, 3710(2), 258 (1999). Gimzewski, J. K., C. Gerber, E. Meyer, and R. R. Schlitter, Chem. Phys. Lett. 217, 589 (1994).

338

BIBLIOGRAPHY

Giordano, B. C., C. L. Copper, and G. E. Collins. Micellar electrokinetic chromatography and capillary electrochromatography of nitroaromatic explosives in seawater, Electrophoresis, 27, 778–786 (2006). Goldman, E. R., I. L. Medintz, A. Hayhurst, G. P. Anderson, J. M. Mauro, B. L. Iverson, G. Georgiou, and H. Mattoussi. Self-assembled luminescent CdSe-ZnS quantum dot bioconjugates prepared using engineered poly-histidine terminated proteins. Anal. Chim. Acta 534, 63–67 (2005a). Goldman, E. R., I. L. Medintz, J. L. Whitley, A. Hayhurst, A. R. Clapp, H. T. Uyeda, J. R. Deschamps, M. E. Lassman, and H. Mattoussi. A hybrid quantum dot-antibody fragment ﬂuorescence resonance energy transfer-based TNT sensor. J. Am. Chem. Soc. 127, 6744–6751 (2005b). Goldman, E. R., A. R. Clapp, G. P. Anderson, H. T. Uyeda, J. M. Mauro, I. L. Medintz, and H. Mattoussi. Multiplexed toxin analysis using four colors of quantum dot ﬂuororeagents. Anal. Chem. 76, 684–688 (2004). Goldman, E. R., E. D. Balighian, H. Mattoussi, M. K. Kuno, J. M. Mauro, P. T. Tran, and G. P. Anderson. Avidin: A natural bridge for quantum dot-antibody conjugates. J. Am. Chem. Soc. 124, 6378–6382 (2002a). Goldman, E. R., G. P. Anderson, P. T. Tran, H. Mattoussi, P. T. Charles, and J. M. Mauro. Conjugation of luminescent quantum dots with antibodies using an engineered adaptor protein to provide new reagents for ﬂuoroimmunoassays. Anal. Chem. 74, 841–847 (2002b). Grant, C. L., T. F. Jenkins, and S. M. Golden. Experimental Assessment of Analytical Holding Times for Nitroaromatic and Nitramine Explosives in Soil, SR 93-11, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, Hanover, NH, June 1993. Grasso, F. W., T. R. Consi, D. C. Mountain, and J. Atema. Biomimetic robot lobster performs chemo-orientation in turbulence using a pair of spatially separated sensors: Progress and challenges. Robot. Auton. Syst. 30, 115–131 (2000). Handbook of Chemistry and Physics, 37th ed. Chemical Rubber, Cleveland, 1955–1956 Handbook of Chemistry and Physics, 80th Ed., Chemical Rubber, Boca Raton, FL, 2000. Hanna, S. R. and E. M. Insley. Time series analysis of concentration and wind ﬂuctuations. Bound.-Lay. Meteorol. 47, 131–147 (1989). Harber, D. Guerrilla’s Arsenal: Advanced Techniques for Making Explosives and Time Delay Bombs. Paladin Press, Boulder, CO, 1994. Harper, R. J., J. R. Almirall, and K. G. Furton. Identiﬁcation of dominant odor chemicals emanating from explosives for use in developing optimal training aid combinations and mimics for canine detection. Talanta 67, 313–327 (2005). Harris, C. M., Anal. Chem. 74, 127A (2002). Harrison, D. J., A. Manz, Z. Fan, H. L¨udi, and H. M. Widmer. Capillary electrophoresis and sample injection systems integrated on a planar glass chip. Anal. Chem., 64, 1926–1932 (1992). Hayes, A. T., A. Martinoli, and R. M. Goodman. Swarm robotic odor localization: Off-line optimization and validation with real robots. Robotica 21, 427–441 (2003). Hewitt, A. D., T. F. Jenkins, and T. A. Raney. Field Gas Chromotography/Thermionic Detector System for On-site Determination of Explosives in Soils. U.S Army Corps of Engineers, Engineer Research and Development Center, Cold Regions Research and Engineering Laboratory Technical Report ERDC/CRREL TR-01-09 May 2001. Hibbs, A. D., G. A. Barrall, P. V. Czipott, A. J. Drew, D. Gregory, D. K. Lathrop, Y. K. Lee, E. E. Magnuson, R. Matthews, D. C. Skvoretz, S. A. Vierk¨otter, and D. O. Walsh.

BIBLIOGRAPHY

339

Detection of TNT and RDX Landmines by Stand-off Nuclear Quadrupole Resonance. Preprint given author by Andy Hibbs, Quantum Magnetics Inc., San Diego, 1999. Hill, H. H., Jr., W. F. Siems, and R. H. St. Louis. Ion mobility spectrometry. Anal. Chem. 21(5), 321–355 (1990). Hilmi, A. and J. H. T. Luong. Micromachined electrophoresis chips with electrochemical detectors for analysis of explosive compounds in soil and groundwater. Environm. Sci. Technol., 34, 3046–3050 (2000a). Hilmi, A., and J. H. T. Luong. Electrochemical detectors prepared by electroless deposition for microfabricated electrophoresis chips. Anal. Chem. 72, 4677–4682 (2000b). Hines, M. A. and P. Guyot-Sionnest. Synthesis and characterization of strongly luminescing ZnS-capped CdSe nanocrystals. J. Phys. Chem. 100, 468–471 (1996). Horner, A. J., M. J. Weissburg, and C. D. Derby. Dual antennular chemosensory pathways can mediate orientation by Caribbean spiny lobsters in naturalistic ﬂow conditions. J. Exp. Biol. 207(21), 3785–3796 (2004). Horwood, C. The use of dogs for operations related to humanitarian mine clearance. Handicap International Mines Coordination Unit, France, 1998. Houser, E. J., T. E. Mlsna, V. K. Nguyen, R. Chung, R. L. Mowery, and R. A. McGill. Talanta 54, 469 (1995). Huang, Z., J. C. Sanders, C. Dunsmor, H. Ahmadzadeh, and J. P. Landers. A method for UV-bonding in the fabrication of glass electrophoretic microchips. Electrophoresis 22, 3924–3929 (2001). Hummel, R. E. Electronic Properties of Materials, 3rd ed. Springer, New York, 2001. Hummel, R. E. Differential reﬂectance spectroscopy in analysis of surfaces, in R. A. Meyers, Ed. Encyclopedia of Analytical Chemistry. Wiley, Chichester, 2000. Hummel, R. E. Phys. Stat. Sol. (a), 76, 11 (1983). Hundal, L. S., J. Singh, E. L. Bier, P. J. Shea, S. D. Comfort, and W. L. Powers. Removal of TNT and RDX from water and soil using iron metal. Environ. Pollution 97(1–2), 55–64 (1997). Improvised Munitions Black Book, Volume 1. Desert Publications, El Dorado, AR, 1978. International Journal for Ion Mobility Spectrometry is published as a reviewed journal. It is the ofﬁcial publication of the International Society for Ion Mobility Spectrometry (ISIMS), c/o ISAS—Institute for Analytical Sciences, Bunsen-Kirchhoff-Str. 11, D-44139 Dortmund, Germany. Ishida, H., T. Nakamoto, T. Moriizumi, T. Kikas, and J. Janata. Plume-tracking robots: A new application of chemical sensors. Biol. Bull. 200, 222–226 (2001). Jacobson, S. C., L. B. Koutny, R. Hergenr¨oder, A. W. Moore, Jr., and J. M. Ramsey. Microchip capillary electrophoresis with an integrated postcolumn reactor. Anal. Chem. 66, 3472–3476 (1994a). Jacobson, S. C., R. Hergenr¨oder, L. B. Koutny, and J. M. Ramsey. High-speed separations on a microchip. Anal. Chem. 66, 1114–1118 (1994b). Jacobson, S. C., R. Hergenr¨oder, L. B. Koutny, R. J. Warmack, and J. M. Ramsey. Effects of injection schemes and column geometry on the performance of microchip electrophoresis devices. Anal. Chem. 66, 1107–1113 (1994c). Jacobson, S. C., R. Hergenroder, L. B. Joutny, and J. M. Ramsey. Open channel electrochromatography on a microchip. Anal. Chem. 66, 2369–2373 (1994d). Jares-Erijman, E. A. and T. M. Jovin, FRET imaging. Nature Biotech. 21, 1387–1395 (2003). Jenkins, A. L., S. Y. Bae, J. M. Lochner, and J. L. Jensen, in Proc. 2nd Joint Conf. on Point Detection for Chem. Bio. Def., 2004, Williamsburg, VA, 2004.

340

BIBLIOGRAPHY

Jenkins, T. F., M. E. Walsh, P. H. Miyares, J. Kopzynski, T. Ranney, V. George, J. Pennington, and T. Berry. Analysis of Explosive-Related Chemical Signatures in Soil Samples Collected Near Buried Landmines. U.S. Army Engineer Research and Development Center—Cold Regions Research and Engineering Laboratory, ERDCCRREL, Report ERDC TR-00-5, Hanover, NH, March 2000a. Jenkins, T. F., M. E. Walsh, P. H. Miyares, J. A. Kopczynski, T. A. Ranney, V. George, J. C. Pennington, and T. E. Berry. Analysis of Explosives-Related Signature Chemicals in Soil Samples Collected Near Buried Landmines, ERDC Technical Report, Cold Regions Research and Engineering Laboratory, Hanover, NH, 2000b. Jia, Z-J., Q. Fang, and Z-L. Fang. Bonding of glass microﬂuidic chips at room temperatures. Anal. Chem. 76, 5597–5602 (2004). Johnston, J. M., M. Williams, L. P. Waggoner, C. C. Edge, R. E. Dugan, and S. F. Hallowell. Canine detection odor signatures for mine-related explosives. Proc. SPIE, Detection and Remediation Technologies for Mines and Minelike Targets III 3392(1), 490–501 (1998). Jones, C. D. On the structure of instantaneous plumes in the atmosphere. J. Hazard. Mater. 7, 87–112 (1983). Joynt, V. The Mechem explosive and drug detection system (MEDDS), in I. G. McLean, Ed. Mine Detection Dogs: Training, Operations and Odour Detection. Geneva International Centre for Humanitarian Demining (GICHD), Geneva, 2003, pp. 165–174. Kahn S. M., Ed. Proc. 1st Proc. Int. Symp. Explosives Detection Technology, FAA, Atlantic City, 1992. Keller, T. A., I. Powell, and M. J. Weissburg. Role of olfactory appendages in chemically mediated orientation of blue crabs. Mar. Ecol. Prog. Ser. 261, 217–231 (2003). Kenna, B. T. and F. J. Conrad. Studies of the Adsorption/Desorption Behavior of ExplosiveLike Molecules. Sandia Report SAND86-0141, Sandia National Laboratories, Albuquerque, NM, 1986. King, C. Ed. Jane’s Mines and Mine Clearance, 4th ed., Janes Information Group, 1999–2000. Kirk-Othmer Encyclopedia of Chemical Technology, 3rd ed., Vol. 9. Wiley, New York, 1980. Klank, H., J. P. Kutter, and O. Geschke. CO2 -laser micromachining and back-end processing for rapid production of PMMA-based microﬂuidic systems. Lab Chip 2, 242–246 (2002). Klostermeier, D. and D. P. Millar. Time-resolved ﬂuorescence resonance energy transfer: A versatile tool for the analysis of nucleic acids. Biopolymers 61, 159–179 (2002). Koehl, M. A. R., J. R. Koseff, J. P. Crimaldi, M. G. McCay, T. Cooper, M. B. Wiley, and P. A. Moore. Lobster snifﬁng: Antennule design and hydrodynamic ﬁltering of information in an odor plume. Science 294, 1948–1951 (2001). A. N. Kolmogorov. The local structure of turbulence in incompressible viscous ﬂuid for very large Reynolds numbers. Dolk. Akad. Nauk SSSR 30, 301 (1941). Koyuncu, H., E. Seven, and A C¸alimli. Examination of some organic explosives by ion mobility spectrometry. Turk. J. Chem. 29, 255–264 (2005). Kricka, L. J., P. Fortina, N. J. Panaro, P. Wilding, G. Alonso-Amigo, and H. Becker. Fabrication of plastic microchips by hot embossing. Lab Chip 2, 1–4 (2002). Krishnan, R. V., R. Varma, and S. Mayor. Fluorescence methods to probe nanometerscale organization of molecules in living cell membranes. J. Fluorescence 11, 211–226 (2001).

BIBLIOGRAPHY

341

Kutter, J. P., S. C. Jacobson, and J. M. Ramsey. Solid phase extraction on microﬂuidic devices. J. Microcolumn Sep. 12(2), 93–97 (2000). la Grone, M., C. Cumming, M. Fisher, and E. Towers. Investigation of an area reduction method for suspected mineﬁelds using an ultra-sensitive chemical vapor detector, in Proceedings of Aerosense 2002, Volume 4742, International Society for Optical Engineering, Orlando, FL, April, 2002. la Grone, M., C. Cumming, C. Aker, M. Fisher, M. Fox, D. Reust, M. Rockley, T. Swager, E. Towers, and V. Williams. Developments in the use of an ampliﬁed ﬂuorescent polymer-based sensor for the detection of landmines, in Proceedings of Aerosense 2001, Volume 4394-111, International Society for Optical Engineering, Orlando, FL, April, 2001. la Grone, M., C. Cumming, M. Fisher, D. Reust, and R. Taylor. Landmine detection by chemical signature: Detection of vapors of nitroaromatic compounds by ﬂuorescence quenching of novel polymer materials, in A. C. Dubey, J. F. Harvey, J. T. Broach, and R. E. Dugan, Eds. Detection and Remediation Technologies for Mines and Minelike Targets 1V, Proceedings of SPIE, Vol. 3710, pp. 409–420, April 5–9, 1999, Orlando, FL. Lakowicz, J. R. Principles of Fluorescence Spectroscopy, Kluwer Academic: New York, 1999. Lavrick, V., M. J. Sepaniak, and P. G. Datskos. Rev. Sci. Instrum. 75, 2229 (2004). Laramee J. A., C. A. Kocher, and M. L. Deinzer. Anal. Chem. 64, 2316 (1992). Leatherdale, C. A., W. K. Woo, F. V. Mikulec, and M. G. Bawendi. On the absorption cross section of CdSe nanocrystal quantum dots. J. Phys. Chem. B. 106, 7619–7622 (2002). Lee H. G., E. D. Lee, and M. L. Lee. Proc. Int. Symp. Explosives Detection Technol., S. M. Khan, Ed. FAA, Atlantic City, 1992, p. 619. Leggett, D. C., J. H. Cragin, T. F. Jenkins, and T. A. Ranney. Release of ExplosiveRelated Vapors from Landmines. U.S. Army Engineer Research and Development Center—Cold Regions Research and Engineering Laboratory, ERDC-CRREL Technical Report TR-00-2, Hanover, NH, February 2001. Leggett, D. C., T. F. Jenkins, A. Hogan, T. A. Ranney, and P. H. Miyares. External Contamination on Landmines by Organic Nitro-Compounds. U.S. Army Engineer Research and Development Center—Cold Regions Research and Engineering Laboratory, ERDC-CRREL Technical Report TR-00-2, Hanover, NH, March 2000. Leggett, D. C. T. F. Jenkins, and R. P. Murmann. Composition of Vapors Evolved from Military TNT as Inﬂuenced by Temperature, Solid Composition, Age, and Source. SR 77-16/AD A040632, U.S. Army Corps of Engineers, Cold Regions Research and Engineering Laboratory. Hanover, NH, 1977. Li, J. The Cyranose Chemical Vapor Sensor. Sensors Magazine Online, August 2000; visited 9/19/05 http://www.sensorsmag.com/articles/0800/56/main.shtml. Lu, Q., G. E. Collins, M. Smith, and J. Wang. Sensitive capillary electrophoresis microchip determination of trinitroaromatic explosives in nonaqueous electrolyte following solid phase extraction. Anal. Chim. Acta 469, 253–260 (2002). MacDonald, J., J. R. Lockwood, J. McFee, T. Altshuler, T. Broach, L. Carin, R. Harmon, C. Rappaport, W. Scott, and R. Weaver. Alternatives for Landmine Detection, RAND Science and Technology Policy Institute, Santa Monica, 2003. Mafra-Neto, A. and R. T. Card´e. Fine-scale structure of pheromone plumes modulates upwind orientation of ﬂying moths. Nature 369, 142–144 (1994).

342

BIBLIOGRAPHY

Makky W. H. Ed. Proc. 2nd Explosive Detection Technology Symp. & Aviation Security Technology Conf. FAA, Atlantic City, 1996. March R. E. and J. F. J. Todd. Practical Aspects of Ion Trap Mass Spectrometry, Vol. 1, CRC Press, New York, 1995. Mark, H. F. and Kroschwitz J. Encyclopedia of Polymer Science and Technology. Wiley, New York, 1985. Mattoussi, H., J. M. Mauro, E. R. Goldman, G. P. Anderson, V. C. Sundar, F. V. Mikulec, and M. G. Bawendi. Self-assembly of CdSe-ZnS quantum dot bioconjugates using an engineered recombinant protein. J. Am. Chem. Soc. 122, 12142–12150 (2000). McDonald, J. C., D. C. Duffy, J. R. Anderson, D. T. Chiu, H. Wu, O. J. A. Schueller, and G. M. Whitesides. Fabrication of microﬂuidic systems in poly(dimethylsiolxane). Electrophoresis, 21, 27–40 (2000). McLean, I. G., Ed. Mine Detection Dogs: Training, Operations and Odour Detection. Geneva International Centre for Humanitarian Demining (GICHD), Geneva, 2003. ˚ McLean, I. G., H. Bach, R. Fjellanger, and C. Akerblom. Bringing the mineﬁeld to the detector: Updating the REST concept, in Proceedings of EUDEM2-SCOT—2003, Vrije Universiteit, Brussels, Vol. 1, pp. 156–161 (2003). McLuckey, S. A., G. J. Van Berkel, D. E. Goeringer, and G. L. Glish. Anal. Chem. 66, 689A (1994). 737A. McLuckey, S. A., G. L. Glish, K. G. Asano, and B. C. Grant. Anal. Chem. 60, 2220 (1988). Mead, K. S., M. B., Wiley, M. A. R., Koehl, and J. R. Koseff. Fine-scale patterns of odor encounter by the antennules of mantis shrimp tracking turbulent plumes in waveaffected and unidirectional ﬂow. J. Exp. Biol. 206, 181–193 (2003). Medintz, I. L., H. T. Uyeda, E. R. Goldman, and H. Mattoussi. Quantum dot bioconjugates for imaging, labeling, and sensing. Nat. Mater. 4, 35–446 (2005). Medintz, I. L., S. A. Trammell, H. Mattoussi, and J. M. Mauro. Reversible modulation of quantum dot photoluminescence using a protein-bound photochromic ﬂuorescence resonance energy transfer acceptor. J. Am. Chem. Soc. 126, 30–31 (2004). Medintz, I. L., A. R. Clapp, H. Mattoussi, E. R. Goldman, B. R. Fisher, and J. M. Mauro. Self-assembled nanoscale biosensors based on quantum dot FRET donors. Nat. Mater. 2, 630–638 (2003). Metz, S., R. Holzer, and P. Renaud. Polyimide-based microﬂuidic devices. Lab Chip 1, 29–34 (2001). Meyer, R. Explosives, 3rd ed. VCH, New York, 1987. Michael, S. M., B. M. Chien, and D. M. Lubman. Anal. Chem. 65, 2614 (1993). Michael, S. M., B. M. Chien, and D. M. Lubman. Rev. Sci. Instrum. 63, 4277 (1992). Michalet, X. F., F. F. Pinaud, L. A. Bentolila, J. M. Tsay, S. Doose, J. J. Li, G. Sundaresan, A. M. Wu, S. S. Gambhir, and S. Weiss. Quantum dots for live cells, in vivo imaging, and diagnostics. Science 307, 538–544 (2005). Mine Facts CD-ROM, Vol. 1.2, U.S. Department of Defense, Washington, DC, 1998. MISTRAL Detection Israel, www.mistralgroup.com. Miyawaki, A., A. Sawano, and T. Kogure. Lighting up cells: Labeling proteins with ﬂuorophores. Nature Cell Biol. 5, S1–S7 (2003a). Miyawaki, A. Visualization of the spatial and temporal dynamics of intracellular signaling. Dev. Cell 4, 295–305 (2003b). Hollas, J. M., Ed. Modern Spectroscopy, 4th ed. Wiley, Chichester, West Sussex, UK, 2004.

BIBLIOGRAPHY

343

Moore, D. S. Instrumentation for trace detection of high explosives. Rev. Sci. Instrum. 75, 2499–2512 (2004). Moore, P. A. and J. Atema. Spatial information contained in three-dimensional ﬁne structure of an aquatic odor plume. Biol. Bull. 181, 408–418 (1991). Moore, P. A. and D. M. E. Lepper. Role of chemical signals in the orientation behavior of the sea star Asterias forbesi. Biol. Bull. 192, 410–417 (1997). Moore, P. A., N., Scholz, and J. Atema. Chemical orientation of lobsters, Homarus americanus, in turbulent odor plumes. J. Chem. Ecol. 17, 1293–1307. Mourzina, Y., A. Steffen, D. Kalyagin, R. Carius, and A. Offenh¨ausser. Capillary zone electrophoresis of amino acids on a hybrid poly(dimethylsiloxane)-glass chip, Electrophoresis 26, 1849–1860 (2005). Murlis, J. The structure of odour plumes, in T. L., Payne, M. C., Birch, and C. E. J. Kennedy, Eds. Mechanisms in Insect Olfaction. Clarendon Press, Oxford, 1986, pp. 27–38. Murlis, J. and C. D. Jones. Fine-scale structure of odour plumes in relation to insect orientation to distant pheromone and other attractant sources. Physiol. Entomol. 6, 71–86 (1981). Murphy, C. J. Optical sensing with quantum dots. Anal. Chem. 74, 520A–526A (2002). Murray, C. B. and C. R. Kagan. Synthesis and characterization of monodisperse nanocrystals and close-packed nanocrystal assemblies. Ann. Rev. Mater. Sci. 30, 545–610 (2000). Mylne, K. R. The vertical proﬁle of concentration ﬂuctuations in near-surface plumes. Bound.-Lay. Meteorol. 65, 111–136 (1993). Mylne, K. R. Concentration ﬂuctuation measurements in a plume dispersing in a stable surface layer. Bound.-Lay. Meteorol. 60, 15–48 (1992). National Atmospheric Release Advisory Center, http://www.nrac.llnl.gov. National Research Council. Opportunities to Improve Airport Passenger Screening with Mass Spectrometry. National Academies Press, Washington, DC, 2003. Niemeyer, C. M. Functional hybrid devices of proteins and inorganic nanoparticles. Angew. Chem. Int. Ed. 42, 5796–5800 (2003). Niemeyer, C. M. Nanoparticles, proteins, and nucleic acids: Biotechnology meets materials science. Angew. Chem. Int. Ed. 40, 4128–4158 (2001). Nolan, R. V. and D. L. Gravitte. Mine Detecting Canines, Report 2217, U.S. Army Mobility Research and Development Command, Ft Belvoir, VA 1977. Oda, R. P. and J. P. Landers, in J. P. Landers, Ed. Handbook of Capillary Electrophoresis, 2nd ed. CRC Press, New York, 1997, Chapter 1. Oden, P. I., G. Y. Chen, R. A. Steele, R. J. Warmack, and T. Thundat. Appl. Phys. Lett. 68, 1465 (1996). Oleschuk, R. D., L. L. Shultz-Lockyear, Y. Ning, and D. J. Harrison. Trapping of bead-based reagents within microﬂuidic systems: On-chip solid-phase extraction and electrochromatography. Anal. Chem. 72, 585–590 (2000). Oxley, J. C., J. L. Smith, J. Moran, and K. Shinde. Determination of the vapor density of triacetone triperoxide (TATP) using a gas chromatography headspace technique. Propellants, Explosives, Pyrotechnics 30(2), 127–130 (2005). Oxley, J. C., J. L. Smith, and H. Chen. Decomposition of multi-peroxidic compound: Triacetone triperoxides (TATP). Propellants, Explosives and Pyrotechnics 27, 209–216 (2002a).

344

BIBLIOGRAPHY

Oxley, J. C., J. L., Smith, H. Chen, and E. Ciofﬁ. Decomposition of multi-peroxidic compounds: Part II: Hexamethylene triperoxide diamine (HMTD). Thermochem. Acta 388(1–2), 215–225 (2002b). Parak, W. J., D. Gerion, T. Pellegrino, D. Zanchet, C. Michael, S. C. Williams, R. Boudreau, M. A. L. Gros, C. A. Larabell, and A. P. Alivisatos. Biological applications of colloidal nanocrystals. Nanotech. 14, R15–R27 (2003). Pavia, D., G. Lampman, and G. Kriz. Introduction to Spectroscopy. Harcourt College, Fort Worth, 2001. Pawliszyn, J. Solid Phase Microextraction—Theory and Practice. Wiley, New York, 1997. Pella, P. A. Measurement of the vapor pressures of pressures of TNT, 2,4-DNT, 2,6-DNT and EGDN. J. Chem. Thermodynam. 9, 301–305 (1977). Phelan, J. M. and J. L. Barnett. Chemical Sensing Thresholds for Mine Detection Dogs. SAND2002-0692C, Sandia Laboratories Report, Albuquerque, NM, 2002. Phelan, J. M. and J. L. Barnett. Phase Partitioning of TNT and DNT in Soils. Sandia National Laboratories Report SAND2001-0310, Albuquerque, NM, February 2001a. Phelan, J. M. and J. L. Barnett. Solubility of 2,4-dinitrotoluene and 2,4,6-trinitrotoluene in water. J. Chem Eng. Data. Mar/Apr (2001b). Phelan, J. M. and S. W. Webb. Chemical Sensing for Buried Landmines—Fundamental Processes Inﬂuencing Trace Chemical Detection. SAND2002-0909, Sandia National Laboratories, Albuquerque, NM, 2002. Phelan, J. M., P. J. Rodacy, and J. L. Barnett. Explosive Chemical Signatures from Military Ordnance. Sandia National Laboratories Report SAND2001-0755, Albuquerque, NM, April 2001. Phelan, J., P. Rodacy, and J. Barnett. Explosive Chemical Signatures from Military Ordnance. SAND2001-0755. Sandia National Laboratories, Albuquerque, NM, April 2001. Phelan, J. M., M. Gozdor, S. W. Webb, and M. Cal. Laboratory data and model comparisons of the transport of chemical signatures from buried landmines/UX, A. C. Dubey, J. F. Harvey, J. T. Broach, R. E. Dugan, Eds. in Proceedings of the SPIE 14th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, Detection and Remediation Technologies for Mines and Minelike Targets IV, Orlando, FL, April 5–9, 1999. Phelan, J. M. and S. W. Webb. Environmental Fate and Transport of Chemical Signatures from Buried Landmines—Screening Model Formulation and Initial Simulations. Sandia National Laboratories Report, SAND97-1426, Albuquerque, NM, June 1997. Pinnaduwage, L. A., H. F. Ji, and T. Thundat. IEEE Sensors J. 5, 774 (2005). Pinnaduwage, L. A., A. Wig, D. L. Hedden, A. Gehl, D. Yi, T. Thundat, and R. T. Lareau, J. Appl. Phys. 95, 5871 (2004). Pinnaduwage, L. A., V. Boiadjiev, J. E. Hawk, and T. Thundat. Appl. Phys. Lett. 83, 1471 (2003a). Pinnnaduwage, L. A., A. Gehl, D. L. Hedden, G. Muralidharan, T. Thundat, and R. Lareau, T. Sulcheck, L. Manning, B. Rogers, M. Jones, and J. D. Adams. Nature, 425, 6957 (2003b). Pumera, M., J. Wang, F. Opekar, I. Jel´ınek, J. Feldman, H. L¨owe, and S. Hardt. Contactless conductivity detector for microchip capillary electrophoresis. Anal. Chem. 74, 1968–1971 (2002). Qi, S., X. Liu, S. Ford, J. Barrows, G. Thomas, K. Kelly, A. McCandless, K. Lian, J. Goettert, and S. A. Soper. Microﬂuidic devices fabricated in poly(methyl methacrylate) using hot-embossing with integrated sampling capillary and ﬁber optics for ﬂuorescence detection. Lab Chip 2, 88–95 (2002).

BIBLIOGRAPHY

345

Rahman, S. Effect of bed roughness on scalar mixing in turbulent boundary layers. Ph.D. Thesis, Georgia Institute of Technology, 2002. Rahman, S. and D. R. Webster. The effect of bed roughness on scalar ﬂuctuations in turbulent boundary layers, Exp. Fluids 38, 372–384 (2005). Rains, G. C., Tomberlin, J.K., M. D’Alessandro and W. J. Lewis, Limits of Volatile Chemical Detection of a Parasitoid Wasp, Microplitis croceipes, and an Electronic Nose: A Comparative Study, Transactions of the American Society of Agricultural Engineers, Vol. 47(6): 2145–2152, 2004. Rains, G. C., S. L. Utley and W. J. Lewis, Behavioral Monitoring of Trained Insects for Chemical Detection, Biotechnol. Prog. 2006, 22, 2–8. Ramsey, J. D. and G. E. Collins. Integrated microﬂuidic device for solid-phase extraction coupled to micellar electrokinetic chromatography separation. Anal. Chem. 77, 6664–6670 (2005). Ramsey, J. D., S. C. Jacobson, C. T. Culbertson, and J. M. Ramsey, High-efﬁciency, two-dimensional separations of protein digests on microﬂuidic devices. Anal. Chem. 75, 3758–3764 (2003). Reeder, P. B. and B. W. Ache. Chemotaxis in the Florida spiny lobster, Panulirus argus, Anim. Behav. 28, 831–839. Rhykerd, C. L., D. W. Hannum, D. W. Murray, and J. E. Parmeter. Guide for the Selection of Commercial Explosives Detection Systems for Law Enforcement Applications. NIJ Guide 100-99, NCJ 178913. National Institute of N, Ofﬁce of Science and Technology, Washington, DC, 1999. Rodacy, P. J., P. K. Walker, S. D. Reber, J. Phelan, and J. V Andre. Chemical Sensing of Explosive Targets in the Bedford Basin, Halifax. Nova Scotia SAND2001-3569, Sandia National Laboratories Report, Albuquerque, NM, November 20, 2001. Rodacy, P. J., P. K. Walker, S. D. Reber, J. Phelan, and J. V Andre, Explosive Detection in the Marine Environment and on Land Using Ion Mobility Spectrometry, A Summary of Field Tests. SAND2000-0921, Sandia National Laboratories Report, Albuquerque, NM, Mexico 2000 Roehl, J. E. Environmental and process applications for ion mobility spectrometry. Appl. Spectrosc. Rev. 26(1&2), 1–57 (1991). Rohrlich, M. and W. Sauermilch. Zeitschrift Gesamte Schiess Sprengstoffwesen 38, 97 (1943). Rose, A., Z. G. Zhu, C. F. Madigan, T. M. Swager, and V. Bulovic. Nature 434, 876 (2005). Salapaka, M.V., S. Bergh, J. Lai, and A. Majumdar, J. Appl. Phys. 81, 2480 (1997). St. Louis, R. H. and H. H. Hill, Jr. Ion mobility spectrometry in analytical chemistry. Crit. Rev. Anal. Chem. 62(23), 1201–1209 (1990). Sapsford, K. E., I. L. Medintz, J. P. Golden, J. R. Deschamps, H. T. Uyeda, and H. Mattoussi. Surface-immobilized self-assembled protein-based quantum dot nanoassemblies. Langmuir 20, 7720–7728 (2004). Sarid, D. Scanning Force Microscopy. Oxford University Press, New York, 1991. Seiler, K., D. J. Harrison, and A. Manz. Planar glass chips for capillary electrophoresis: Repetitive sample injection, quantitation, and separation efﬁciency. Anal. Chem. 65, 1481–1488 (1993). Sensor Technology Assessment for Ordnance and Explosive Waste Detection and Location, JPL D-11367, rev B. Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, March 1995.

346

BIBLIOGRAPHY

Settles, G. S., H. C. Ferree. M. D. Tronosky, Z. M. Moyer, and W. J. McGann. 3rd International Symposium on Explosive Detection and Aviation Security, FAA, Atlantic City, 2001. Shuttleworth, R. Pro. Phys. Soc. (London) 63A, 444 (1950). Sleeman, R., S. L. Richards, W. R. Stott, W. R. Davidson, J. G. Luke, B. J. Keely, I. Fletcher, and A. Burton. Proc. of the 50th ASMS Conf. on Mass Spectrom. and Allied Topics, Orlando, FL, 2000. Smith, M., G. E. Collins, and J. Wang, Microscale solid-phase extraction system for explosives. J. Chromatogr. A, 991, 159–167 (2003). Spangler, G. E. and R. A. Miller. Int. J. Mass Spectrom. 214, 95–104 (2002). Spencer, W. F., M. M. Cliath, and S. R. Yates. Soil-pesticide interactions and their impact on the volatilization process, in Environmental Impact of Soil Component Interactions—Natural and Anthropogenic Organics, Vol. 1, CRC Press, Boca Raton, FL, 1995, pp. 371–381. Stacey, M. T., E. A. Cowen, T. M. Powell, E. Dobbins, S. G. Monismith, and J. R. Koseff. Plume dispersion in a stratiﬁed, near-coastal ﬂow: Measurements and modeling. Cont. Shelf Res. 20, 637–663 (2000). J. I. Steinfeld and J. Wormhoudt. Explosives detection: A challenge for physical chemistry. Annu. Rev. Phys. Chem. 49, 203–232 (1998). Stoney, G. G. Proc. R. Soc. London Ser. A 82, 172 (1909). Stott, W. R., W. R. Davidson, and R. Sleeman. Proc. SPIE 2092, 53 (1994). Subcommittee on Consequence Assessment and Protective Action (SCAPA) of the Emergency Management Advisory Committee of the U.S. Department of Energy. http://www.orau.gov/emi/scapa/. Sutherland, A. J. Quantum dots as luminescent probes in biological systems. Curr. Op. Sol. St. Mat. Sci. 6, 365–370 (2002). J. A. Syage, and M. A. Evans. Spectroscopy 16, 14 (2001). J. A, Syage, K. A. Hanold, and M. A. Hanning-Lee. Mass spectrometry based personnel screening system. Proc. INMM—42nd Annual Meeting, Palm Springs, CA, 2001a. Syage, J. A., B. J. Nies, M. D. Evans, and K. A. Hanold. J. Am. Soc. Mass Spectrom. 12, 648 (2001b). Tanyanyiwa, J., E. Abad Villar, M. T. Fernandez Abedul, A. Costa Garcia, W. Hoffma, A. Guber, D. Herrmann, A. Gerlach, N. Gottschlich, and P. Hauser. High-voltage contactless conductivity-detection for lab-on-chip devices using external electrodes on the holder. Analyst 128, 1019–1022 (2003). Taylor and Rickenbach, Army Ordnance, 5, 463 (1924). Theisen, L., D. W. Hannum, D. W. Murray, and J. E. Parmeter. Survey of Commercially Available Explosives Detection Technologies and Equipment 2004, Document No. 208861. National Law Enforcement and Correction Technology Center, a Program of the National Institute of Justice, U.S. Department of Justice, Washington, DC, 2005. Thompkins, B. A. Explosives analysis in the environment, in R. A. Meyers, Ed. Encyclopedia of Analytical Chemistry. Wiley, Chichester, 2000, pp. 2402–2441. Thompson, P. L., L.A. Raner, and J. L. Schnoor. Uptake and transformation of TNT by hybrid poplar trees. Environ. Sci. Technol. 32(7), 975–980, (1998). Thundat, T. and A. Majumdar, in Barth, Humphrey, and Secomb, Eds., Sensors and Sensing in Biology and Engineering. Springer, NewYork, 2004. Thundat, T., P. I. Oden, and R. J. Warmack. Microscale Thermophys. Eng. 1, 185 (1997).

BIBLIOGRAPHY

347

Tomkins, B. A., Explosives Analysis in the Environment, in Encyclopedia of Analytical Chemistry, R. A. Meyers (Ed). pp 2402–2441, John Wiley & Sons, Chichester, 2000. Tomberlin, J. K., M. Tertuliano, G. Rains and W. J. Lewis, Conditioned Microplitis croceipes Cresson (Hymenopteria: Braconidae) Detect and Respond to 2,4 DNT: Development of a Biological Sensor, J. Forensic Sci., Sept. 2005, Vol. 50, No. 5, 5 pages. Paper ID JPS2005014, Available online at: www.astm.org Tortonese, M., R.C. Barrett, and C. F. Quate. Appl. Phys. Lett. 62, 834 (1993). Townsend, J. Pigs, a demining tool of the future? J. Mine Action 7(3), 43–46 (2003). Tufte, O. N. and E. L. Stelzer. J. Appl. Phys. 34, 323 (1963). U.S. Department of State. Hidden Killers: The Global Demining Crisis. U.S. Department of State Publication 190575, Washington, DC, 1998. Verhagen, R., C. Cox, R. Machangu, B. Weetjens, and M. Billet. Preliminary results on the use of Cricetomys rats as indicators of buried explosives in ﬁeld conditions, in I. G. McLean, Ed. Mine Detection Dogs: Training, Operations and Odour Detection. Geneva International Centre for Humanitarian Demining (GICHD), Geneva, 2003, pp. 175–193. Vickers, N. J. and T. C. Baker. Reiterative responses to single strands of odor promote sustained upwind ﬂight and odor source location by moths. Proc. Natl. Acad. Sci. USA 91, 5756–5760 (1994). Wallace, W. FMX the Revised Black Book. Paladin Press, Boulder, CO, 1995. Wallenborg, S. R. and C. G. Bailey. Separation and detection of explosives on a microchip using micellar electrokinetic chromatography and indirect laser-induced ﬂuorescence. Anal. Chem. 72, 1872–1878 (2000). Walsh, M. E. and T. F. Jenkins. Identiﬁcation of TNT Transformation Products in Soil, SR92-16. US Army Corps of Engineers, Cold Regions Research and Engineering Laboratory, Hanover, NH, 1992. Wang, J. Electrochemical detection for microscale analytical systems: A review. Talanta, 56, 223–231 (2002). Wang, J. and M. Pumera. Dual conductivity/amperometric detection system for microchip capillary electrophoresis. Anal. Chem. 74, 5919–5923 (2002). Wang, J., G. Chen, and A. Muck, Jr. Movable contactless conductivity detector for microchip capillary electrophoresis. Anal. Chem. 75, 4475–4479 (2003a). Wang, J., G. Chen, M. Chatrathi, K. Shin, and A. Fujishima. Microchip capillary electrophoresis coupled with a boron-doped diamond electrode-based electrochemical detector. Anal. Chem. 75, 935–939 (2003b). Wang, J., G. Chen, A. Muck, Jr., and G. E. Collins. Electrophoretic microchip with dual-opposite injection for simultaneous measurements of anions and cations, Electrophoresis 24, 3728–3734 (2003c). Wang, J., M. Pumera, and G. Collins. A chip-based capillary electrophoresis-contactless conductivity microsystem for fast measurements of low-explosive ionic components. Analyst 127, 719–723 (2002a). Wang, J., M. Pumera, M. P. Chatrathi, A. Escarpa, R. Konrad, A. Griebel, W. D¨orner, and H. L¨owe. Towards disposable lab-on-a-chip: Poly(methylmethacrylate) microchip electrophoresis device with electrochemical detection. Electrophoresis, 23, 596–601 (2002b). Wang, J., M. Pumera, M. P. Chatrathi, A. Escarpa, M. Musameh, G. Collins, A. Mulchandani, Y. Lin, and K. Olsen. Single-channel microchip for fast screening and detailed identiﬁcation of nitroaromatic explosives or organophosphate nerve agents. Anal. Chem. 74, 1187–1191 (2002c).

348

BIBLIOGRAPHY

Wang, J., B. Tian, and E. Sahlin. Micromachined electrophoresis chips with thick-ﬁlm electrochemical detectors. Anal. Chem. 71, 5436–5440 (1999). Webb, S. W., K. Preuss, J. M. Phelan, and S. Finsterle. Development of a mechanistic model for the movement of chemical signatures from buried landmines/UXO, A. C. Dubey, J. F. Harvey, J. T. Broach, and R. E. Dugan, Eds. in Proceedings of the SPIE 14th Annual International Symposium on Aerospace/Defense Sensing, Simulation and Controls, Detection and Remediation Technologies for Mines and Minelike Targets IV, April 5–9, 1999. Webster, D. R. and M. J. Weissburg. Chemosensory guidance cues in a turbulent chemical odor plume. Limnol. Oceanogr. 46(5), 1034–1047 (2001). Webster, D. R., S. Rahman, and L. P. Dasi. Laser-induced ﬂuorescence measurements of a turbulent plume. J. Engng. Mech. 129, 1130–1137 (2003). Webster, D. R., S. Rahman, and L. P. Dasi. On the usefulness of bilateral comparison to tracking turbulent chemical odor plumes. Limnol. Oceanogr. 46, 1048–1053 (2001). Weissberg, M. J., C. P. James, D. L. Smee, and D. R. Webster, Fluid mechanics produces conﬂicting constraints during olfactory navigation of blue crabs, Callinectes sapidus. J. Exp. Biol. 206, 171–180 (2003). Weissburg, M. J. The ﬂuid dynamical context of chemosensory mediated behavior. Biol. Bull. 198, 188–202 (2000). Weissburg, M. J. and D. B. Dusenbery. Behavioral observations and computer simulations of blue crab movement to a chemical source in a controlled turbulent ﬂow. J. Exp. Biol. 205, 3387–3398 (2002). Weissburg, M. J. and R. K. Zimmer-Faust. Odor plumes and how blue crabs use them in ﬁnding prey. J. Exp. Biol. 197, 349–375 (1994). Weissburg, M. J. and R. K. Zimmer-Faust. Life and death in moving ﬂuids: Hydrodynamic effects on chemosensory-mediated predation. Ecology 74, 1428–1443 (1993). Williams, V. and T. M. Swager. Iptycene-containing poly(aryleneethynylene)s. Macromolecules 33, 4069–4073 (2000). Williams, V. E., J. S. Yang, C. G. Lugmair, Y. J. Miao, and T. M. Swager. Design of novel iptycene-containing ﬂuorescent polymers for the detection of TNT. Proc. SPIE 3710, 402–408 (1999). Willis, C. M. Olfactory detection of human bladder cancer by dogs: Proof of principle study. Br. Med. J. 329(7468), 712–714 (2004). Wolfenstein, R. Ber. Dtsch. Chem. Ges. 28, 2265 (1895). Yang, J. S. and T. M. Swager. Fluorescent porous polymer ﬁlms as TNT chemosensors: Electronic and structural effects. J. Am. Chem. Soc. 120, 11864–11873 (1998a). Yang, J. S. and T. M. Swager. Porous shape persistent ﬂuorescent polymer ﬁlms: An approach to TNT sensory materials. J. Am. Chem. Soc. 120, 5321–5322 (1998b). Yee, E., R. Chan, P. R. Kosteniuk, G. M. Chandler, C. A. Biltoft, and J. F. Bowers. The vertical structure of concentration ﬂuctuation statistics in plumes dispersion in the atmospheric surface layer. Bound.-Lay. Meteorol. 76, 41–67 (1995). Yee, E., R. Chan, P. R. Kosteniuk, G. M. Chandler, C. A. Biltoft, and J. F. Bowers. Experimental measurements of concentration ﬂuctuations and scales in a dispersing plume in the atmospheric surface layer obtained using a very fast response concentration detector. J. Appl. Meteorol. 33, 996–1016 (1994). Yee, E., D. J. Wilson, and B. W. Zelt. Probability-distributions of concentration ﬂuctuations of a weakly diffusive passive plume in turbulent boundary-layer. Bound.-Lay. Meteorol. 64, 321–354 (1993). Yinon, J. Forensic and Environmental Detection of Explosives. Wiley, New York, 1999.

BIBLIOGRAPHY

349

Yinon, J. and S. Zitrin. Modern Methods and Applications in Analysis of Explosives. Wiley, New York, 1993. Yinon, J. J. Mass Spectrom. Rev. 10, 179 (1991). Yoffe, A. D. Semiconductor quantum dots and related systems: Electronic, optical, luminescence and related properties of low dimensional systems. Adv. Phys. 50, 1–208 (2001). Zhang, C-X. and A. Manz. Narrow sample channel injectors for capillary electrophoresis on microchips. Anal. Chem. 73, 2656–2662 (2001). Zhou, Q. and T. Swager. Methodology for enhancing the sensitivity of ﬂuorescent chemosensors: Energy migration in conjugated polymers. J. Am. Chem Soc. 117, 7017–7018 (1995). Ziegler, C. Anal. Bioanal. Chem. 379, 946 (2004). Zimmer-Faust, R. K., C. M. Finelli, N. D. Pentcheff, and D. S. Wethey. Odor plumes and animal navigation in turbulent water ﬂow: A ﬁeld study. Biol. Bull. 188, 111–116 (1995a) Zimmer-Faust, R. K., C. M. Finelli, N. D. Pentcheff, and D. Wethey. Odor plumes and animal navigation in turbulent water ﬂows: A ﬁeld study. Biol. Bull. 188, 111–116 (1995b). Zhou, Q. and T. Swager. Methodology for enhancing the sensitivity of ﬂuorescent chemosensors: Energy migration in conjugated polymers. J. Am. Chem Soc. 117, 7017–7018 (1995).

INDEX

α particles, 212 β particles, 212 γ rays, 212 γ -based sensors, 4 μ Faraday plate, 215–217 π to π ∗ transitions, 305 241 Am, 212 63 Ni, 212, 229 Abkhazia, 315 Acetone, 55, 56 Acoustic, 158 Active Sensors, 4 Aden, 43 ADNT, 134, 137, 181, 162, 163, 186, 232, 274, 331 Adsorb, 77 Adsorbed Explosive Molecules, 259 Adsorption, 245, 247, 248, 250 Adsorption Filter, 186 Adsorption-Induced Cantilever Bending, 245 Adsorption-Induced Forces, 246 Advection, 109, 110, 125, 127 Afghanistan, 187, 314, 315, 323 AFP, 12, 27–31, 134, 137, 142, 149, 195–199, 201, 204–206, 208, 325 Africa, 314 African giant pouched rats, 189, 315 Agriculture, 80

Al Qaeda, 49, 53, 55, 63 Alcohol, Tobacco and Firearms, 259 Aluminum, 39, 48, 66 Amatol, 39 Americium 212, 243 Amino-Dinitrobenzene, 187 Amino-DNT, 134, 137, 162, 163, 181, 186, 232, 274 Ammonium, 40 Ammonium nitrate, 37, 39, 44, 51, 52, 59, 220 Ammonium nitrate icing sugar, 54 Ammonium nitrate, powdered, 53, 54 Amperometric detection, 267, 269, 279 Ampliﬁcation, 196, 198 Ampliﬁer, 17 Amplifying ﬂuorescent polymer, 12, 28, 134, 195, 196 AN, 37, 39, 52, 54 Analysis cycle time, 191 Anarchist literature, 49, 50 Anarchist press, 51 ANFO, 44, 52–54, 220, 233, 307, 308, 331 Angola, 182, 190, 314–316 Animal noses, 177 Animal sensors, 178 ANIS, 54, 325 Antibody, 286 Antigen-antibody, 272, 273

Trace Chemical Sensing of Explosives, Edited by Ronald L. Woodﬁn c 2007 John Wiley & Sons, Inc. Copyright

351

352

INDEX

Antipersonnel mine, 158, 161, 164 Anti-tank mine, 72, 161, 164, 167 AP mine, 158, 161, 164 APCI, 242, 325 API, 230 API/TOF MS, 29–31, 231, 325 APOPO, 178, 185, 186, 189, 190, 315, 325 Aqueous state, 96 Area coverage rate, 190 Area reduction, 167, 168, 184 Arms caches, 323 Armstrong’s mixture, 62 Army RDECOM CERDEC NVESD, 317 Army Research Laboratory, 310 Aroma, 6 Aroma sensing, 177 ASGDI, 230, 325 ASGDI MS/MS, 29–31, 326 Asmara, 313 Aspirin, 51 AT mine, 72, 161, 164, 167, 326 ATF, 259 Atmospheric pressure ionization, 230 Atmospheric pressure ionization time-of-ﬂight mass spectrometer, 29 Atmospheric sampling glow discharge ionization, 230 Atmospheric sampling-glow discharge-ion trap-tandem mass spectrometer, 29 Attogram, 19, 163, 217, 329 Australia, 316 Autonomous underwater vehicle, 139 Autonomous vehicle, 104 AUV, 139, 140, 147, 148, 201, 204, 326 Backscatter, 240 Bali, 65, 66 Baratol, 39 Barium nitrate, 39 Batch, 23 Batchelor scale, 110, 114 BB1, 50 Bees, 9, 178 Benson, Ragnar, 51 Bilateral comparison, 121, 125, 127 Billion, 19, 20 Bioluminescence detector, 29–31 Biosensors, 6 BL, 29–31, 326 Black powder, 220 Blasters, 56 Blasting cap, 49, 63 Blind test, 164 Blue crab, 121, 124, 125

Blue/UV light, 305, 309 Bomb makers, 49–67 Bomber, 242 Bombs, 49, 65 Booster, 54 Bosnia, 164, 313, 314, 316 Bougainville, 316 Boundary layer, 89–91, 93–95, 110, 152, 155 Boundary layer, convective, 240 Boundary layer, turbulent, 112 Buffalo vehicle, 323 Bulk detector, 240 Bulk ﬂow, 125 Bulk sensors, 4 Buried explosives, 95 Buried munitions, 103 Buried source, 99 By-products, 93 C-4, 50, 154, 220, 307, 308, 331 Cambodia, 190, 314, 315 Camera, 309 Camouﬂage pattern, 183, 184, 186 Canals, 96 Cancer detection, 7, 184, 185 Canines, 159, 164–167, 170, 172, 202 Cantilever, 245, 249, 251, 255–259 piezoresistive, 252, 256 Cap, 46, 48, 49 Capacitive transimpedance ampliﬁer, 215 Capillary electrophoresis, 261, 264, 273, 276 Carbon disulﬁde, 46 Casspir vehicle, 323 Castor oil, 47 CCD, 305, 309, 326 C-CD, 270, 326 CCRDC, 69, 326 CE, 261, 326 CE-EC microchips, 266, 268 CE-microchips, 267–269, 271, 278 CGTVA, 188, 189 Charge, minimum, 48 Charge-coupled device, 305 Chechnya, 314 Cheddites, 47 Chemical boundary layer, 91 Chemical klinotaxis, 119 Chemical plumes, 98, 109, 111 Chemical Sensing in the Marine Environment Program, 133, 201 Chemical sensors, 221 Chemical signatures, 101, 159 Chemical vapor sensor, 196 Chemical warfare agents, 281

INDEX

Chemiluminescence, 39, 219 Chemiluminesence detector, 28 Chemiresistors, 12, 27 Chemometrics, 18 Chemosensory tracking, 119 China, 66, 315 Chlorate-based explosives, 51, 65 Chlorates, 44, 46, 48, 51, 66 Chromatography, electrokinetic, 279 Chromophores, 197, 200 Citric acid, 51, 61 Clandestine chemist, 40 Coastal Systems Station, U. S. Navy, 138, 318 Cold Regions Research and Engineering Center, 69 Cold Regions Research and Engineering Laboratory, 72 Comparisons between dogs and chemical methods, 164, 181, 202 Comp-B, 39, 154, 159, 331 Comp-B2, 39, 331 Complex aroma, 190 Complex dielectric constant, 303 Composite explosives, 53 Composition-B, 39, 154, 159, 331 Computer code, 102 Computer models, 102 Concentrating, 17 Concentration, in environment, 18, 25, 30, 81, 88, 89, 91, 93–95, 101, 102, 104 ﬁeld experience, 152, 156, 168, 171, 201 instrument sensing, 223, 228, 266 minimum possible, 20, 23 in plume, 110, 112–115, 120, 125 Concentration ﬁelds, plumes in, 115, 116 Concentration gap, 26, 27 Concentration gradient, plumes in, 114 Concentration maps, underwater plumes, 137 Concentrator, 240 Contact-less, 303 Contactless conductivity detectors, 270 Contactless conductivity microchip, 269 Contaminated, 160 Contamination, 167 Convection, 77, 78, 96, 109 Convective boundary layer, 240 dispersion, 154 transport, 135 CO-OP, 50, 65, 331 Correlation function, 121–123 Council for Scientiﬁc and Industrial Research (South Africa), 178, 180, 184

353

Crack cocaine, 59 Crawler, 139 Crawler robot, 143 Critical diameter, 36 Croatia, 181–184, 190 CROMAC, 181, 326 CRRDC, 72, 73, 326 CSIR, 178, 180, 184, 326 CSME, 133–136, 201, 208, 326 CSS, 138, 143, 326 CTIA, 215, 216 Current, 96, 134 Curve recognition program, 309 Cyranose, 9 DADP, 55, 56, 331 DARPA, 127, 136, 151, 153, 159, 164, 179, 201, 208, 300, 326 DDNP, 40, 331 Decomposition, 59, 78 Decomposition products, 180, 186, 188 Defence Science and Technology Laboratories (UK), 208 Defense Advanced Research Projects Agency, 136, 151, 172 Deﬂagraton, 250, 253, 257–259 Degradation, 92 Degradation products, 162, 181, 276 Dehydration, 55 Demining, 189, 313–316 humanitarian, 168, 179, 190, 313–316 DENEL, 178, 184 Department of Energy, 201 Department of Homeland Security, 259, 281, 317 Department of Transportation, 36 Desert Publications, 49 Detasheet, 220 Detection dogs, 313–316 Detection limits, 217 minimum, 156 Detector, electrochemical, 28, 30, 268 Detector, electron-capture, 145, 219, 221 Detonable, 48 Detonate, 36 Detonation front, 36 Detonation velocity, 38, 53, 54 Detonation wave, 36 Detonator, 46, 63 homemade, 51 Dew, 93 Diatomaceous earth, 46 Dichlorates, 44 Diesel fuel, 52–54

354

INDEX

Differential reﬂection spectroscopy, 5, 303 Differential reﬂectogram, 306–308 Differential reﬂectometry, 303 Differential x ray, 219 Diffusion, 71, 72, 75, 77, 78, 84, 97, 110, 162 Diffusive, 104 Diffusive ﬂux, 75 Diffusivity, 73, 85, 111 eddy, 117 molecular, 126 Discrimination algorithms, 207 Diurnal concentration predictions, 103 Diurnal cycle, 93 Diver-deployed, 141, 146 DMNB, 220, 232, 233 DNB, 20, 74, 92, 160, 161, 267, 270, 272, 274, 331 DNT, 20, 71, 80, 81, 83, 86–88, 92, 94, 134, 137, 157, 160–163, 168, 206, 207, 231, 232, 267, 274, 278, 279, 331 DoD, 201 Dog’s nose, 182, 183 Dog’s Nose Program, 136, 151, 153, 201 Dogs, 7, 94, 159, 164, 170–172, 177, 190, 191, 323 performance of, 202 Doped silicon, 251 DoT, 36, 326 DR, 303, 306, 309, 326 Dragon Runner(robot), 203 Drift region, 213 Drift time, 213, 214 Drophammer, 49, 58, 59 DRS, 5, 303, 326 Drying, 86 Dual x-ray tomography, 220 Dust, 182, 183, 190 Dynamite, 46, 66, 220, 331 E50 , 61 ECD, 145, 221, 326 ECNIMS, 29–31, 231, 326 Eddies, 109, 110 Eddy diffusivity, 117 Edge of the plume, 123 EF&T, 70, 99, 102, 326 EGDN, 220, 232 Electrochemical, 265, 266 Electrochemical detector, 28, 30, 268 Electrokinetic, 261 Electrokinetic chromatography, 279 Electron capture detection, 221 detectors, 145, 219, 221

negative ion mass spectrometry, 29 negative ion MS, 231 Electronic ionizing techniques, 212 Electronic sensor technology, 221 Electroosmosis, 263 Electroosmotic ﬂow, 263 Electropherogram, 267, 268, 272–274, 276 Electrophoresis, 263, 264 Electrostatic discharge, 58 Elephants, 178, 323 EMRTC, 321 Energetic material, 35 Energetic Material Research and Testing Center, 321 Energy budget, 18 Environmental conditions, 166 Environmental Fate and Transport, 70, 99, 102 Environmental Protection Agency, 201 EOD, 313, 316, 318 EOF, 263, 264, 276, 326 Equilibrium vapor concentration, 155 ERC, 159, 160, 162, 163, 165, 166, 169, 171 Eritrea, 313, 315 ERW, 15, 70, 76, 96, 315, 316, 326 ESD, 58, 59, 61, 62, 326 ETD, 219, 220, 222, 242, 326 Evapotranspiration, 77, 162, 165 EVD, 8, 179, 327 Explosive(s), 35 charge, 69, 71 compounds, 199 detection, 195 detection systems, 220 device, improvised, 12, 15, 51, 64, 66, 70, 76, 95, 98, 165, 201, 317 formulations, 39 homemade, 43 improvised, 43–67 ordnance disposal, 313–316 plume, 148 trace detection, 219 vapor, 190, 245 Explosive Remnants of War, 15, 70, 76, 96, 315, 316 Explosive-related compounds, 159

FAA, 220, 243, 259 FAIMS, 221, 327 False alarm rate, 164, 202 False alarms, 158 False negative, 13, 219 False positive, 13, 219, 221, 238, 240, 304 Fate and transport, 152

INDEX

FBI, 208, 317 Federal Aviation Administration, 220, 243, 259 Femtogram, 19, 163, 179, 196, 201, 217, 231, 329 Fertilizer, 52 Fertilizer-based explosives, 51–55 Fertilizer-grade AN, 52 FGAN, 52, 53 Fick’s law, 110 Fido, 134, 145, 151–153, 156, 160, 162–164, 167–170, 172, 179, 181, 182, 195, 196, 199–201, 203, 204, 205, 206, 208, 323 Fido XT, 203, 204, 208 Field experience, 177, 190 practice, 184 test, 164 Field asymmetric ion mobility spectrometry, 221 Field ion spectrometer, 29 Filament, plume, 101, 110, 112, 114, 119, 125, 135, 144, 155–157 Filamentary, 104 Filamentation, 101 Filter materials, 186 Filters, 171, 323 FIS, 29–31, 327 Flash powder, 48, 51, 66 Fluoresce, 197 Fluorescence, 136, 195, 197, 198, 200, 265, 274, 279 Fluorescence images, 295, 299 Fluorescence quenching, 199 Fluorescence resonance energy transfer, 287 Fluorescent chromophore, 198 detector, 29, 30 polymers, 195 Fluorophores, 136 Flux, 73–76, 91 Flux rate, 23, 87 Flux rate nomenclature, 25 FMX, 50, 51, 327 FMX Revised Black Book, 50 FOI, 178, 186–188, 327 Following plumes, 99 F¨orster energy transfer, 199 F¨orster resonance energy transfer, 286–300 Fort A. P. Hill, 169 Fort Belvoir, 167, 317 Fort Leonard Wood, 164, 169, 201 Fort Ord, 15 Foster-Miller, 138, 139, 143, 145, 203

355

Foster-Miller Talon, 138, 203 Fourier transform, 225 Fourier transform, ion cyclotron resonance mass spectrometry, 318 Fourier transform mass spectrometer, 225 Freezing, 92, 95 FRET, 286–300, 327 Friction, 46, 58, 61, 62 Friction sensitivity, 66 FTMS, 225, 226, 327 Fulminate, 46, 60 GA, 50, 51, 327 Gamma-based sensors, 4 Gas chromatography, 167, 219, 221 /chemiluminesence, 28 /differential ion mobility spectrometer, 28 /electron capture detector, 29 /ion mobility spectrometer, 29 /mass spectrometer, 28, 29 /surface acoustic wave, 28 Gaussian, 155 Gaussian distribution, 135 Gaussian proﬁle, 115 Gaza Strip, 63 GC, 145, 219, 221 GC/DMS, 28–31, 327 GC/ECD, 29–31, 327 GC/IMS, 29–31, 327 GC/MS, 29–31, 327 GC/SAW, 29–31, 327 GC-CL, 28–31, 327 GC-ECD, 145 Geneva International Centre for Humanitarian Demining, 178, 322 GICHD, 178, 322, 327 Global Training Academy, Somerset, Texas, 179 GPR, 5, 158, 169 Gradient, 114, 119 Grifﬁn Analytical, 221 Ground AN, 54 Ground penetrating radar, 5, 158, 169 Guerrilla’s Arsenal, 50 Gunpowder Plot, 43 H50 , 58, 61 Half-life, 92, 162 Halifax, 134, 148 Halo Trust, 315 HD, 179 Helhofﬁte, 46, 50 Helicopter, 181 Henry’s Law, 80, 81

356

INDEX

Hexamethylene triperoxide diamine, 37 Hexamine, 51, 61 Hidden Markov methods, 104 High performance liquid chromatography, 28 Hitachi, 221 HMM, 104, 327 HMTD, 37, 39, 48, 49, 51, 55, 60–63, 154, 220 HMX, 37, 39, 220, 231, 232, 307, 308 Homemade detonator, 51 explosives, 43 initiator, 63 Homogeneity, 101 HOMO-LUMO, 305 HPLC, 28–31, 267, 278 HPLC-UV, 188 Humanitarian demining, 168, 179, 190, 313–316 Humidity, 184 Hybrid MS, 226 Hydrogen peroxide, 51, 56, 61 Hypochlorites, 44 ICAO, 220, 327 ICBL, 314 Icing sugar, 51, 54 ICRC, 158 Idaho National Laboratory, 252 IE, 43–67, 317, 327 IED, 12, 15, 51, 64, 66, 70, 76, 96, 98, 165, 201, 317, 327 Impact, 58, 61 Improvised explosive device, 12, 15, 51, 64, 66, 70, 76, 95, 98, 165, 201, 317 explosives, 43–67 initiator, 63 Improvised Munitions Black Book, Vol I, 50 Impurities, 58 IMS, 27–31, 211, 212, 214–216, 219–221, 228–230, 236, 240, 242, 327 IMS manufacturers, 216 India, 314 Indonesia, 65, 66 Infrared, 158 Infrared spectrometer, 29 Ingesting, 212 Initiation, explosives, 48, 59 Initiator, homemade, 63 Inorganic oxidizers, 39 Insects, 9, 94 Institute of Soldier Nanotechnologies, 208 Insult, 40, 48

Integral length scale, 122, 123 Interferent, 171, 172, 206, 258 International Civil Aviation Organization, 220 International Committee of the Red Cross, 158 Internet, 49, 50 Iodates, 44 Ion cyclotron resonance, 225 Ion mobility spectrometer, 27–31, 211, 212, 214–216, 219–221, 228–230, 236, 240, 242 Ion trap ion mobility spectrometer, 28 Ion trap/TOF MS, 226 Ionization, 228 Ionizing element, 212 IR, 158, 323, 327 IR spectrometry, 29–31, 221 Iraq, 314–316, 323 iRobot, 203 Irrigated ﬁelds, 96 IRS, 29–31, 221, 327 Israel, 55, 64, 316 ITIMS, 28–31, 327 ITMS, 224, 227 James Madison University, Mine Action Information Center, 322 Jane’s Information Group, Jane’s Mines and Mine Countermeasures, 322 Jordan, 55, 60, 63 Kahoolawe, 15 K´arm´an vortex street, 127 Kd , 81, 83 Kd ’ , 82, 94 KH , 80, 81 Kieselguhr, 46 Kinepac, 54 Kolmogorov scale, 110, 114 Ks/a , 94 Kuwait, 314 Lab-on-a-chip, 28, 261, 263–267 LabVIEW, 307, 309 Lakes, 96 Laminar ﬂow, 98, 119 Landmine, 31, 73–77, 85–98, 157–159, 162, 164, 165, 169, 188, 313–316 Landmine signature chemicals, 92 Landmine soil probe, 168–170 Laos, 315 Laser-induced breakdown spectroscopy, 158 Laser-induced ﬂuorescence, 112, 271, 280 Lead azide, 40

INDEX

Leakage, 73, 183 LED, 200, 327 LIBS, 158, 327 LIF, 112, 271, 280, 327 Limit of detection, 8, 23, 27, 31, 179, 181, 231, 233, 234, 254, 258, 277 Liquid phase, 80 Lithium ion battery, 215 LOC/HPLC, 28–31, 327 Locating the point source, 118 LOD, 8, 23, 27, 31, 179, 181, 231, 233, 234, 254, 258, 277, 327 Low-metal-content mines, 158, 159 Luminescent semiconductor nanocrystals, 285 Maltose binding protein, 286 Manus, 316 Mapping a plume, 104 Marine environment, 148 Mass ﬂux, 152 Mass fraction, 83, 84 Mass production, 246 Mass Spec Analytical, 221 Mass Spectrometry, 211, 219, 220 Massachusetts Institute of Technology, 134, 154, 195 Mathematical models, 104 MBP, 286–299, 328 MDD, 178–180, 183, 184, 186, 189, 191, 328 Meander, 126, 127 Mechem, 8 MECHEM, 8, 165, 167, 170, 171, 178, 181, 182, 184, 185, 190–323, 315, 323 MECHEM Explosive and Drug Detection System, 165, 178 MEDDS, 8, 95, 165, 170, 178–182, 184–186, 190, 191, 328, 323 Memorial Institute for the Prevention of Terrorism, 281 MEMS, 205, 208, 328 Menchen gegen Minen, 314 Mercury fulminate, 40, 48, 49, 60 Metal detector, 14, 158, 159 Meteorological conditions, 100 MgM, 314 Microbial degradation, 77, 78 Microbiological processes, 92 Microcantilever, 245–259 piezoresistive, 253 Microchip, 261, 268, 276 Microfabricated, 250, 261 Microﬂuidic, 261

357

Microgram, 19, 329 Micromachined, 246 Milligram, 19, 329 Mine Action Information Center, James Madison University, 322 Mine(s), 73–76, 164, 167, 168, 183, 189, 191, 323, 323 clearance, 190, 313–316 clearers, 323 detection, 186 detection dog, 164, 167, 178, 313–316 low-metal-content 158, 159 signature, 165, 166 warfare, 318 Mine Detection Rats, 315 Mine protected vehicle, 323 Mine resistant vehicle, 191, 323 Mine Warfare Association, xxv, 322 Mineﬁeld, 158, 164, 177, 190, 314, 323 Min´elites, 47, 51 Minimum charge, 48 Minimum detection limit, 156 Minimum possible concentration, 21, 24 Ministry of Defence, Australia, 318 MINWARA, xxv, 322 MIP, 221, 328 MIPT, 281 MIT, 134, 154, 172, 179, 195, 208, 328 MNT, 232, 332 Model(s), computer, 102 mathematical, 103 plume search, 104 tracking, 104 Molar density, 19 extinction coefﬁcient, 285 volume, 19 Moldability, 51 Mole, 20 Molecular diffusion, 110, 114 diffusivity, 126 weight, 20, 21, 229 Molecularly imprinted polymer, 221 Monochromatic light, 304 Monolayer ﬁlms, 248 Monomolecular explosives, 53 Morocco, 55, 64 Morogoro, 188 Mozambique, 164, 182, 188, 189, 315 MPV, 323 MRV, 191, 323, 328

358

INDEX

MS, 211, 219–222, 228, 229, 231, 236, 240–242, 328 MS/MS, 224, 228, 230, 233, 242, 328 MSn , 224, 228, 328 Munition, 71 MW, 21, 229, 329 Myanmar, 314 Nagorno Karabakh, 315 Naminbia, 323 Nanogram, 19, 217, 329 ng/g, 19 ng/L, 19 Naval Explosive Ordnance Disposal, Technology Division, 318 Naval Postgraduate School, xxv Naval Research Laboratory, 317 Naval Surface Warfare Center, Panama City, 318 NB, 45, 46, 50, 53, 274, 332 Negative ions, 212 Neutron-based sensors, 4 New Guinea, 316 New Mexico Institute of Mining and Technology, 321 Newsgroups, 56 NG, 37, 220, 232, 307, 308, 332 Nickel 63, 212, 229 Nitramines, 38 Nitrate(s), 44, 48 ester, 37 groups, 66 Nitric acid, 46, 54 Nitric-acid-based explosives, 46 Nitro esters, 180 Nitro groups, 36 Nitroarenes, 38 Nitrobenzene, 45, 46, 53, 65, Nitrocellulose, 37, 180 Nitrogen dioxide, 46 Nitroglycerin, 37, 220, 232, 307, 308 Nitromethane, 45, 54 Nitronapthalene, 46, 47 Nitrourea, 55 NLDIMS, 28–31, 328 NM, 45, 332 NMR, 4, 328 NO2 , 36 NOKSH, 178, 328 Nomadics, 134, 145, 151–153, 166, 167, 170, 178, 179, 182, 195, 201 Nomenclature, concentration, 18, 20 ﬂux rate24

mass, 18 parts per, 19, 20, 330 Non-linear dependence of ion mobility, 28 Norwegian Peoples Aid, 8, 178, 183 313 NPA, 8, 178, 183, 313, 328 NQR, 4, 179, 328 NRe , 90 NRL, 300 NSF, 127 Nuclear magnetic resonance, 4 Nuclear quadrupole resonance, 4, 179 NVESD, 151, 167, 179, 181, 183, 184, 208, 317, 328 Oak Ridge National Laboratory, 259, 317 OB, 36 Octahydro-1, 3, 5, 7-tetranitro-1, 3, 4, 5-tetrazocine, 37 Octol, 39 Odor of mines, 323 Odor plumes, 97, 99, 103, 182 Odor-gated rheotaxis, 124, 125, 127 Ofﬁce of Naval Research, 96, 133, 134, 149, 201, 208, 281, 300 Oklahoma Center for the Advancement of Science and Technology, 208 Oklahoma City Memorial Institute for the Prevention of Terrorism, 208 Olfactory sensors, 183 ONR, 96, 133, 134, 149, 201, 208, 281, 300, 328 Optical spectrometry, 221 Organic matter, 92 Organic peroxides, 48, 51 ORNL, 259, 317 Orthogonal, 213 Orthogonal signals, 256 Orthogonality, 14 Oxanite, 46 Oxidizers, 46 Oxygen balance, 36 P50 , 61 PA, 46, 48, 51, 328 Package bombs, 51 Packaging, 18 Packbot, 203 Paladin Press, 49 Palestinian, 54, 55, 64 Parafﬁn, 47 Particulate sensing, 182 Partition coefﬁcient, 247 Partitioning, 77–79, 84, 96 Parts per nomenclature, 19, 20, 330

INDEX

Parts per quadrillion, 19, 20, 163, 330 Parts per trillion, 19, 20, 163, 330 Pasquill’s stability classes, 99 Passive sensors, 3 PD , 14, 17, 26, 164, 202, 222, 230, 236–238, 330 PDK, 59, 328 PE-4, 154, 332 Pentiptycene, 196 Pentolite, 39 Pentyerthritol tetranitrate, 37 Perchlorates, 44, 48, 66 Performance, 202, 279 Permanganates, 44, 66 Permeability, 73 Peroxide, 39, 44, 48, 49, 55, 65, 66 Peroxide detection kit, 59 Personnel screening, 240, 241 PETN, 20, 37, 39, 40, 49, 58, 59, 61, 62, 154, 180, 220, 229, 231–234, 253–255, 307, 308, 332 Petroleum, 46 PF A Ed. note: P F A and P F P are equivalent, each preferred by some authors, 14, 17, 164, 330 PF P Ed. note: P F A and P F P are equivalent, each preferred by some authors, 222, 230, 235–239, 330 Phlegmatize, 46, 47, 50 Photoluminesence spectra, 285 Photomultiplier, 137, 200, 304 Photoresist, 261, 262 Picogram, 19, 179, 217, 233, 249, 254, 329 Picric acid, 38, 46, 51, 332 Piezoelectric detector, 29, 30 Piezoelectricity, 250 Piezoresistive, 250 cantilever, 252, 256 microcantilever, 253 Pigs, 7 Pipe bombs, 65 PIRA, 50, 53, 54, 63, 65, 328 PL quantum yield, 288, 328 PL spectra, 285, 299, 328 Plant uptake, 78 Plants, 91, 102 Plastic explosives, 154, 157 Plume(s), 70, 96–100, 104, 112–116, 126, 127, 134–138, 142, 144–148, 155, 157, 182 edge, 123 mapping, 104 search models, 104 stability, 99 tracing, 104

359

transport models, 102 following 99 PMA-1, 75, 160, 161, 164, 328 PMA-2, 75, 161, 328 PMDS, 17, 328 PMDS/DVB, 17, 328 PMT, 137, 200, 305, 328 Polydimethylsiloxane, 17 Polydimethylsiloxane/divinylbenzene copolymer, 17 Polymer, 136, 137 Ponds, 96 Poor Man’s C-4, 50 Pore, 79 Pore saturation, 79–80 Portability, 281 Portal, 240–242 Positive ions, 212 Potassium chlorate, 44, 46–48, 50, 65 perchlorate, 48 Powdered AN, 53, 54 ppb, 19, 222, 329 PPM2, 75 ppq, 163, 179, 182, 201, 329 ppt, 163, 222, 230, 254, 259, 329 Precipitation, 162 Preconcentration, 208, 230, 276, 281 Preconcentrator, 102 Precursors, 51, 52 Pretoria (South Africa), 170, 323 Prilled AN, 53 Prills, 52–54 Primary, 48 Primary explosive, 61 Probability of detection, 14, 17, 26, 164, 202, 222, 230, 236–238 Probability of false alarm (= false positive, equivalent terms), 14, 17, 164 Probability of false positive (= false alarm, equivalent terms), 14, 223, 230, 234–239 Probe, 169, 170 Probing, 159 Propellant, 35 Provisional Irish Republican Army, 50 Puddles, 89, 186 Pyrotechnic, 35, 48, 65 Pyrotechnicians, 56 QA, 182, 316, 328 QD(s), 285–300, 328 bioconjugate, 290 lifetime, 294

360

INDEX

QD(s), (Continued) -based nanosensors, 296 protein assemblies, 295, 299 -protein biconjugates, 287 -protein conjugates, 286, 298 QitToF mass analyzer, 232, 328 QitTof MS, 27–31, 226–228, 230, 231, 233, 328 Quadrillion, 20 Quadrupole ion trap, 224 ion trap –time of ﬂight mass spectrometer, 28 mass analyzer, 222 /TOF MS, 226 Quantum dot(s), 285–300 ﬂuorospores, 286 Quantum Magnetics, 179 Quantum yields, 285 Quench, quenched, quenching, 136, 195, 197, 198, 206, 207 Rack-a-Rock, 46, 47, 50 Ragnar Benson, 51 Ragnar’s Homemade Detonators, 50 Rain, 93 Raman spectrometry, 221 Rapid plume tracking, 125 Rats, 7, 94, 177, 185, 186, 188–190, 323, 315 RDX, 20, 37, 39, 40, 57, 61, 62, 149, 154, 180, 186, 222, 230, 231, 232–234, 253, 307, 308, 332 READ, 135, 231, 328 Receiver operator characteristic, 236 Receptor-based coatings, 247 Receptor-based sensing, 247 Recovery time, 200 Regeneration, 255 Relative humidity, 93 Relative sensitivities of animals and electronic systems, 190 Remote environmental monitoring unit, 140 Remote explosive scent tracing, 165, 178 REMUS, 139–142, 145–149, 204, 205, 328 Resolution, 224–226, 229, 281 Resolving power, 234–236, 238, 239 Resonance frequency, 245–247, 250, 251, 258 Response time, 202 REST, 8, 95, 165–167, 169, 171, 178, 179, 181, 184–186, 188, 189, 328 Reynold’s number, 90, 96, 126

RHD, 50, 328 Rheotaxis, odor-gated, 124, 125, 127 Rhodesia, 323 Ribbons, 101 Rivers, 96 Robotic crawler, 137 ROC, 236–238, 329 RONCO, 315 RP helicopter, 181 Russia, 316 SAM, 252–256, 329 Sample volume, 30, 31 Sampling, 16, 281 Sampling time, 23, 26 Sandia, xxi, 70, 86, 134, 178, 186, 215–217, 240, 243 Sandia National Laboratories, xxi, 70, 86, 134, 178, 186, 215–217, 240, 243 Satellite(s), 178, 323 Saturation, 84 SAW, 11, 27, 246 Scans, 213, 214 ScentPrint, 185 Schmidt number, 111 Scintrex, 221 Seabed, 70, 78 SeaDog, 12, 134, 135, 137–142, 145–147, 149, 201, 204, 205, 329 Seaﬂoor, 97 Seal(s), 73, 78, 79 Seamines, 78, 79, 96 SeaPup, 142, 146, 148, 329 Search pattern, 102 Security screening, 219 Selective layers, 247 Selectivity, 134, 199, 202, 206, 245, 246, 252, 255, 304 Self-assembled monolayers, 252 Self-initiating, 63 Semi-active sensors, 4 Semtex, 154, 220 Sensing, 18 Sensitive, 40 Sensitivity, 18, 23, 26, 134, 146, 177, 179, 201, 202, 222, 225, 234, 246, 249, 255, 258, 265, 275, 281, 303, 309 relative, of animals and electronic systems, 190 shock, 46 threshold, 23 to ESD, 59, 62 to friction, 58, 59, 62, 63, 66 to impact, 58

INDEX

Sensor spacing, 123, 124 Sensors, active, 3 gamma-based, 4 neutron-based, 4 olfactory, 183 passive, 3 semi-active, 3 Sequential batch mode, 26 SERDP, 201, 208 Shining Path, 54 Shock sensitivity, 46 Shock wave, 36 Shoot-out, 177 Signal processing, 11, 18 Silicon nitride, 248 Sloping ground, 167 Smell, 323 Smokeless powder, 220 Smuggling, 180 Snifﬁng, 182, 183 SNL, xxi, 70, 86, 134, 178, 186, 215–217, 240, 243 Sodium chlorate, 50, 51, 65 Software, curve recognition, 309 LabVIEW, 307, 309 Soil, 160, 161 moisture, 85, 92, 93 particles, 78, 79, 84, 183 -sampling probe, 169, 170 surface, 95 vapor probe, 169, 170 Solar heating, 100 Solar radiation, 93 Solid phase extraction, 276, 278 Solid phase microextraction, 17, 135, 169, 214 Somalia, 315 Somaliland, 314, 316 Sonar, 133 Sorb, 79, 84, 89, 91 Source, 156 South Africa, 170, 178, 323 South Korea, 316 Soviet mines, 88 Space and Naval Warfare Systems Command, 140, 149 SPAWAR, 140, 149, 329 Speciﬁcity, 17, 234 Spectrometery, 11 Spectroscopic detector, 29, 30 Spectroscopy, 11

361

SPME, 17, 135, 169, 214 Sprengel, 45, 46, 49 Spring constant, 249 Sri Lanka, 314, 315 Stability of air, 100 Stable, 40, 100 Static sensitive, 62, 66 Stealth, 304 Stoney’s formula, 247 Strategic Environmental Research and Development Program, 201 Streams, 96 Sub-zero conditions, 92 Sudan, 315 Sugar, icing, 51, 54 Sulphuric acid, 56 Surface (of sensor) acoustic wave, 247 acoustic wave sensors, 11, 26, 247 -adsorbed molecules, 249 effect sensors, 11 Surface (of soil or munition) concentration, 89 contamination, 71–73, 75 vapor ﬂux, 86 Swipe, 212, 219, 242 Syagen, 221, 243

T2TNT code, 103 Talon (Foster-Miller robot), 203 Tandem MS, 226 Tasmania, 316 TATP, 35, 37, 39, 40, 48, 49, 54, 55, 57–66, 154, 220, 232, 332 TEA, 28, 39 Terrorist, 45–67 Tertiary explosive, 54 Tetrazoladine, 41 Tetryl, 20, 38, 41, 48, 186, 275–277, 307, 308, 332 Thermal convection, 91 Thermal energy analyzer, 28, 39 Thermedics, 221 Thermo-redox, 28, 29 Time averaged, 101, 119, 120 Time-averaged concentration, 115–117 Time-of-ﬂight, 212, 224 TM-62, 72, 88, 169, 329 TMA-5, 75, 164, 329 TMM-1, 72, 75, 161, 329 TNB, 20, 160, 161, 163, 232, 270, 272, 273, 275–279, 293, 296, 297

362

INDEX

TNT, 20, 21, 37, 39, 40, 44, 48, 54, 59, 66, 71, 74, 80, 81, 83, 87, 88, 92, 94, 133, 135, 136, 144–148, 154, 156, 159, 160, 163, 168, 169, 171, 172, 179, 180, 182–184, 186, 187, 190, 196, 201, 206, 207, 220, 222, 229–234, 253, 256, 257, 259, 267, 270, 272–279, 286, 293, 296, 300, 305–308, 323, 332 TNT equivalency, 49 TNT spectrum, 307 TNT vapor strips, 171 ToF, 212, 213, 224, 225 TOFMS, 225, 227, 329 Torpex, 39 TR, 28–31, 329 Trace chemical sensor, 5, 69 Tracking algorithms, 104 Tracking, chemosensory, 119 Tracking problem, 119 Transport, 161 Transportation Security Administration, 208, 220, 242, 243, 259 Triacetone triperoxide, 35 Triazolidine, 41 Trillion, 20 Trinitrotoluene, 37, 332 Triple quadrupole MS, 226 Tritonal, 39 TSA, 208, 220, 242, 243, 259 Tuberculosis, 185, 315 Turbulence, 90, 96, 99, 100, 126, 127 Turbulent, 109, 155 Turbulent boundary layer, 112 Turbulent chemical plume, 118, 120 Turbulent mixing, 109, 110, 113, 122, 125, 127 Turbulent plume, 119 Turbulent water ﬂow, 111 Turkey, 55, 60, 63, 64

U.S. Army, 73 U.S. Army Night Vision and Electronic Sensors Directorate, 151, 172 U.S. Army Engineer Research and Development Center, 69 U.S. Army Humanitarian Demining Program, 167 U.S. Bureau of Alcohol, Tobacco, and Firearms, 220 U.S. Marine Corps, 73, 88, 203 U.S.S. Cole, 43 Ultraviolet, 261 Unburied explosives, 98

Underwater chemical plumes, 133 explosive devices, 133 objects, 95 plumes, 98 unexploded ordnance, 133 Unexploded ordnance, 12, 70 UNi, 54, 55, 332 United Nations Mine Action Service, 313 University of Arizona, 215 University of Western Australia, 322 Unstable, 100 Urea, 40, 54 Urea nitrate, 54, 55, 64, 220 USAF, 208 USMC, 73, 88, 203 USN Coastal Systems Station (Panama City), 149 UUV, 12 UUXO, 133, 135, 136, 148, 149, 329 UV, 200, 261, 262, 303, 307, 329 UXB International, 316 UXO, 12, 15, 70, 73, 74, 76, 95, 96, 98, 134, 149, 189, 313–316, 329 Uzbekistan, 55, 63 Vapor concentrator, 169 detection, 246 detector, 157, 158 ﬂux, 86 ﬂux rate, 184 phase, 78, 79 plume, 152, 156, 157, 183 pressure, 80, 155, 158, 196 sensors, 5 signature, 157, 163, 165 state, 89, 97 Vapor capturing material, 185 Vapor concentration, 162, 180 Vapor concentration, equilibrium, 155 Vaseline, 47, 50 VBIED, 65, 66, 329 Vegetation, 182, 190 Vehicle borne IED, 65, 66 Vietnam, 315 Volatile, 40 VS-50, 75 Vultures, 181 Wasps, 9 Weather, 99 Weed killer, 51, 65 Wetting, 87

INDEX

Wheatstone bridge, 251 WHOI, 140, 149, 329 Wind, 100 Wind tunnel, 213 Woods Hole Oceanographic Institute, 140, 149

x-ray tomography, dual, 220

x-ray, 4, 240 x-ray tomography, 219

Zeptograms, 217 Zimbabwe, 323

Yemen, 314 Yoctograms, 217 Yugoslavia, 161, 314

363

Plate 1 (Figure 1, Chapter 8, p. 183). Diagram of a dog’s snifﬁng passages. Substitute dust with TNT adsorbed for sausages! (Courtesy of Rune Fjellanger, NPA. Used by permission.)

Measured Curve

Wavelength range 225–300 nm (WL1) Left: Enveloping curves (green) and measured curve (white) Right: Difference between enveloping and measured curve (zero: within the enveloping curves) Calculated value for the difference

Warning lights : Product < 10,000 −> red (TNT) Product >10,000 but < 20,000 −> yellow (TNT?) Product > 20,000 −> green (no TNT)

Shows the peaks (location, amplitude, and 2nd derivative (negative=>maximum)

Plate 2 (Figure 6, Chapter 15, p. 309). program.

Result : Product (WL1*WL2*WL2)

Screen shot of the LabVIEW curve recognition

Plate 3 (Figure 4, Chapter 5, p. 114). Fine resolution perspective of a sample instantaneous concentration ﬁeld. H is the channel depth and equals 20 cm. The contour values are normalized by the source concentration.

Plate 4 (Figure 2, Chapter 5, p. 112). Sample instantaneous concentration ﬁeld near the plume source. H is the channel depth and equals 20 cm. The contour values are normalized by the source concentration. (Figure adapted from data in Webster et al. [3].)

Plate 5 (Figure 3, Chapter 5, p. 113). Sample instantaneous concentration ﬁeld. H is the channel depth and equals 20 cm. The contour values are normalized by the source concentration.

Plate 6 (Figure 14, Chapter 6, p. 148). Detection of TNT plume with the SeaPup sensor mounted on the REMUS AUV off the Atlantic Coast of the U.S. in June 2003.

Plate 7 (Figure 11, Chapter 3, p. 65).

Aerial view of Bali bombing scene.

Plate 8 (Figure 5, Chapter 8, p. 189). Searching for mines and UXO in Mozambique using REST procedures.

Plate 9 (Figure 4, Chapter 4, p. 79). Processes affecting molecules released from a sea mine, partially embedded in the seaﬂoor.

Ionizer Drift Tube

Detector

Plate 10 (Figure 5, Chapter 10, p. 216). Handheld, Portable μ-Faraday Differential CTIA detector under development at Sandia National Laboratories in April 2006. (Courtesy Philip J. Rodacy, Sandia National Laboratories.)

Trace Chemical Sensing of Explosives

Explosives and Chemical Weapons Identification

Explosives and Chemical Weapons Identification

Explosives

Explosives

Trace

Trace

Trace

Trace

Chemical Reviews: Modern Topics in Chemical Sensing issue 2

Trace

Trace

Trace

Trace

Trace

Trace

Trace

Ecotoxicology of Explosives

Trace

Trace

Trace

Trace

Trace

Aspects of Explosives Detection

Trace

Trace

Trace

Trace

Trace

Trace

Trace

Trace Chemical Sensing of Explosives

Explosives and Chemical Weapons Identification

Explosives and Chemical Weapons Identification

Explosives

Explosives

Trace

Trace

Trace

Trace

Chemical Reviews: Modern Topics in Chemical Sensing issue 2

Trace

Trace

Trace

Trace

Trace

Trace

Trace

Ecotoxicology of Explosives

Trace

Trace

Trace

Trace

Trace

Aspects of Explosives Detection

Trace

Trace

Trace

Trace

Trace

Trace

Trace

Recommend Documents