Developments in Surface Contamination and Cleaning
This page intentionally left blank
Developments in Surface Contamination and Cleaning Volume Three Methods for Removal of Particle Contaminants
Edited by
Rajiv Kohli and K.L. Mittal
AMSTERDAM l BOSTON l HEIDELBERG l LONDON NEW YORK l OXFORD l PARIS l SAN DIEGO SAN FRANCISCO l SINGAPORE l SYDNEY l TOKYO William Andrew is an imprint of Elsevier
William Andrew is an imprint of Elsevier The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK 30 Corporate Drive, Suite 400, Burlington, MA 01803, USA First edition 2011 Copyright Ó 2011 Elsevier Inc. All rights reserved. Cover image (far left): Courtesy of Tav Tech Ltd., Yehud, Israel. 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 or otherwise without the prior written permission of the publisher Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email:
[email protected]. Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material Notice No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress ISBNe13: 978-1-4377-7885-4 For information on all William Andrew publications visit our web site at books.elsevier.com Printed and bound in the UK 11 12 13 14 15 10 9 8 7 6 5 4 3 2 1
Contents
Preface About the Editors List of Contributors
Chapter 1
Supersonic Nano-Particle Beam Technique for Removing Nano-Sized Contaminant Particles from Surfaces
vii xi xiii
1
Jin W. Lee
Chapter 2
Megasonic Cleaning
31
R. Nagarajan, S. Awad and K.R. Gopi
Chapter 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles and Film Contamination 63 M.D. Murthy Peri, Ivin Varghese and Cetin Cetinkaya
Chapter 4
Non-Aqueous Interior Surface Cleaning Using Projectiles
123
Rajiv Kohli
Chapter 5
Electrostatic Removal of Particles and its Applications to Self-Cleaning Solar Panels and Solar Concentrators
149
M.K. Mazumder, R. Sharma, A.S. Biris, M.N. Horenstein, J. Zhang, H. Ishihara, J.W. Stark, S. Blumenthal and O. Sadder
Chapter 6
Alternate Semi-Aqueous Precision Cleaning Techniques: Steam Cleaning and Supersonic Gas/Liquid Cleaning Systems
201
Rajiv Kohli Index
239 v
This page intentionally left blank
Preface
The purpose of this book series on Developments in Surface Contamination and Cleaning is to provide a continuous state-of-the-art critical look at the current knowledge of the behavior of both film-type and particulate surface contaminants. The first two volumes, published in 2008 and 2010, respectively, covered various topics dealing with the fundamental nature of contaminants, their measurement and characterization, and different techniques for their removal. The present book is the third volume in the series. The individual contributions in the present book provide state-of-the-art reviews by subject matter experts on removal of solid contaminants from surfaces. Conventional techniques of removing contaminant particles from wafer surface are not successful in removing particles smaller than 50 nm, and new techniques based on laser ablation or cryogenic aerosols are also limited to a similar level of cleaning. In the supersonic nano-particle beam technique, described by Jin-Won Lee in his contribution, contaminant particles are removed by one-to-one collisions with nano-sized bullet particles, and successful cleaning has been demonstrated for contaminant particle sizes down to 20 nm. Small size and high velocity of the bullet particles are two key factors contributing to the success. Bullet particles used in this technique are smaller by a factor of 10 or more than those used in existing cryogenic aerosol cleaning, and the velocity is in the supersonic regime, 3e5 times as high as that in the existing cryogenic aerosol technique. The supersonic beam of nano-sized bullet particles with required size and velocity can be generated in two different ways e supersonic nozzle expansion and electrospray. Technical requirements for removing nano-contaminants are assessed first, and then techniques for generating a nano-particle beam and experimental results on cleaning performance follow sequentially. A brief introduction to the supersonic particle beam technique based on electrospray is included. It is well known that micrometer- and sub-micrometer-sized particulate contaminants cause defects in microelectronic devices, resulting in yield loss and reliability degradation, and that acoustic fields can greatly enhance removal of such contaminants from product and component surfaces. The chapter by Ramamurty Nagarajan, Sami Awad and K.R. Gopi focuses on megasonic cleaning. Megasonic cleaning uses higher frequencies at and above 1000 kHz; it produces controlled cavitation. An important distinction between ultrasonic and megasonic cleaning is that the higher megasonic frequencies do not cause the violent cavitation effects found with ultrasonic frequencies. This significantly vii
viii
Preface
reduces or eliminates cavitation erosion and the likelihood of surface damage to the product being cleaned. Parts that would be damaged by ultrasonic frequencies or cavitation effects can often be cleaned without damage in a megasonic bath using the same solution. With ultrasonics, cavitation occurs throughout the tank, and all sides of submerged parts are cleaned. With conventional megasonics, only the side of the part that is facing the transducer(s) is cleaned. Megasonic cleaning is widely used for removing particles from wafer surfaces, as well as from critical component surfaces in other high-technology products. If ultrasonics is the ‘workhorse’ of the parts cleaning industry, megasonics performs that role in precision cleaning. With further study and optimization, it has the potential to extend its applicability to the nano-regime and below. M.D. Murthy Peri, Ivin Varghese and Cetin Cetinkaya discuss laser cleaning for removal of nano/micro-scale particles and film contamination in their contribution. They describe a laser cleaning method introduced in recent years that utilizes shockwaves generated using the supersonic expansion of a laser-induced plasma (LIP) core. In this cleaning technique, the direct interaction of the laser beam with the substrate is avoided. The authors provide a detailed description of the technique, discuss the assumptions of the theory of blast wave/supersonic expansion, present the results of particle removal experiments and damage effects, characterize the LIP cleaning technique, and report recent advancements in this technique. Contamination of the interiors of hoses, pipes, and tubes is a critical problem in many industries because it often leads to corrosion and service breakdowns, requiring extensive repairs at substantial financial and health costs. Particulate and hydrocarbon contamination in lines used in high-pressure liquid and gaseous oxygen systems can also be a fire hazard. Rajiv Kohli describes a non-aqueous projectile cleaning method to clean the internal surfaces of tubular components that overcomes the shortcomings of traditional tube cleaning techniques. A pneumatic launcher shoots the projectile, sized slightly larger than the internal diameter of the tube, into the tube. As the projectile travels through the tube, it removes the contamination deposited on the internal tube surface and forces it out of the tube. Cleaning can be accomplished very effectively in seconds. In their contribution, Malay Mazumder and his co-authors have presented the basic principles and operation of the electrodynamic screen for removing dust particles from solar panels. By applying a three-phase high-voltage alternating-current electric field to the electrodes, the resulting electrodynamic field repels and removes dust particles from the screens, regardless of whether the dust particles are initially charged or uncharged. Self-cleaning solar panels can be manufactured incorporating electrodynamic screens that derive their low-power output of approximately 10 W/m2 from the solar panels. Under normal atmospheric conditions in desert locations, only a few minutes of cleaning will be needed per day. More frequent cleaning of the panels will be needed during dust storms.
Preface
ix
Precision steam cleaning and supersonic gas-liquid cleaning are two alternate semi-aqueous cleaning methods for removal of solid contaminants in a wide variety of applications. Steam cleaning employs dry superheated steam to remove surface contaminants. It is a low-cost, effective method for precision cleaning and for decontamination of microbially contaminated surfaces. Supersonic gas-liquid cleaning is based on accelerating the cleaning liquid, suspended as droplets in a gas stream, to supersonic velocities through a convergingediverging nozzle. The gas-liquid mixture has the kinetic energy to very effectively remove surface contaminants. This method can also be used for surface cleanliness verification. Both methods use very low volumes of aqueous liquids and are viable alternatives to solvent cleaning in many applications. In his second chapter, Rajiv Kohli discusses the principle of each cleaning technique and provides an overview of available equipment and operating considerations, as well as some of the applications of these cleaning methods. The contributions in this book provide a valuable source of information on the current status and recent developments in the respective topics on surface contamination and cleaning. The book will be of value to government, academic, and industry personnel involved in research and development, manufacturing, process and quality control, and procurement specifications in microelectronics, aerospace, optics, xerography, joining (adhesive bonding), and other industries. We would like to express our heartfelt thanks to all the authors in this book for their contributions, enthusiasm, and cooperation. Our sincere appreciation goes to Matthew Deans, our publisher, who has strongly supported publication of this book and the future volumes in this series. The editorial staff at Elsevier has been instrumental in seeing the book through to publication. Rajiv Kohli would also like to thank Jody Mantell for her tireless efforts in locating obscure and difficult-to-access reference materials. The web companion site can be found at http://www.elsevierdirect.com/ companions/9781437778854. Rajiv Kohli, Houston, Texas
Kash L. Mittal, Hopewell Junction, New York
This page intentionally left blank
About the Editors
Dr. Rajiv Kohli is a leading expert with The Aerospace Corporation in contaminant particle behavior, surface cleaning, and contamination control. At the NASA Johnson Space Center in Houston, Texas, he provides technical support for contamination control related to ground-based and manned spaceflight hardware for the Space Shuttle and the International Space Station, as well as for unmanned spacecraft. Dr. Kohli was involved in developing solvent-based cleaning applications for use in the nuclear industry and he also developed an innovative microabrasive system for a wide variety of precision cleaning and micro-processing applications in the commercial industry. He is the principal editor of the new book series ‘Developments in Surface Contamination and Cleaning’; the first two volumes in the series were published in 2008 and 2010, respectively, and the present book is the third volume in the series. Previously, Dr. Kohli co-authored the book Commercial Utilization of Space: An International Comparison of Framework Conditions, and he has published more than 200 technical papers, articles, and reports on precision cleaning, advanced materials, chemical thermodynamics, environmental degradation of materials, and technical and economic assessment of emerging technologies. Dr. Kohli was recently recognized for his contributions to NASA’s Space Shuttle Return to Flight effort with the Public Service Medal, one of the agency’s highest awards.
xi
xii
About the Editors
Dr. Kashmiri Lal ‘Kash’ Mittal was associated with IBM from 1972 to 1994. Currently, he is teaching and consulting in the areas of surface contamination and cleaning and in adhesion science and technology. He is the Editor-in-Chief of the Journal of Adhesion Science and Technology and is the editor of 100 published books, many of them dealing with surface contamination and cleaning. Dr. Mittal was recognized for his contributions and accomplishments by the worldwide adhesion community which organized in his honor on his 50th birthday the 1st International Congress on Adhesion Science and Technology in Amsterdam in 1995. The Kash Mittal Award was inaugurated in his honor for his extensive efforts and significant contributions in the field of colloid and interface chemistry. Among his numerous awards, Dr. Mittal was awarded the title of doctor honoris causa by the Maria CurieSklodowska University in Lublin, Poland, in 2003. More recently, he was honored in Boston by the international adhesion community on the occasion of publication of his 100th edited book.
Contributors
S.B. Awad, Crest Ultrasonics Corporation, P.O. Box 7266, Trenton, NJ 08628, USA A.S. Biris, Department of Applied Physics, University of Arkansas at Little Rock, Little Rock, AR 72204, USA S. Blumenthal, Department of Electrical and Computer Engineering, Boston University, 8 St. Mary’s Street, Boston, MA 02215, USA C. Cetinkaya, Department of Mechanical and Aeronautical Engineering, Center for Advanced Materials Processing, Wallace H. Coulter School of Engineering, Clarkson University, Potsdam, NY 13699-5725, USA K.R. Gopi, Advanced Ceramic Technologies, Plot 121, Jalan Perusahaan, Bukit Tengah Industrial Park, 14000 Bukit Mertajam, Penang, Malaysia M.N. Horenstein, Department of Electrical and Computer Engineering, Photonics Bldg. Room 527, Boston University, 8 St. Mary’s Street, Boston, MA 02215, USA H. Ishihara, Department of Applied Physics, University of Arkansas at Little Rock, Little Rock, AR 72204, USA R. Kohli, The Aerospace Corporation, 2525 Bay Area Boulevard, Suite 600, Houston, TX 77058-1556, USA J.W. Lee, Department of Mechanical Engineering, Pohang University of Science and Technology (POSTECH), Hyoja 31, Pohang, Kyungbuk, 790-784, South Korea M.K. Mazumder, Department of Electrical and Computer Engineering, Photonics Bldg. Room 532, Boston University, 8 St. Mary’s Street, Boston, MA 02215, USA R. Nagarajan, Department of Chemical Engineering, IIT Madras, Chennai 600036, India M.D. Murthy Peri, Surface Conditioning Division, FSI International Inc., 3455 Lyman Blvd, Chaska, MN 55318, USA O. Sadder, Department of Electrical and Computer Engineering, Boston University, 8 St. Mary’s Street, Boston, MA 02215, USA R. Sharma, Renewable Energy Technology Program, Arkansas State University, Jonesboro, AR 72467, USA J.W. Stark, Department of Electrical and Computer Engineering, Boston University, 8 St. Mary’s Street, Boston, MA 02215, USA I. Varghese, Eco-Snow Systems, Rave N.P., Inc., 4935A Southfront Road, Livermore, CA 94551, USA J. Zhang, Department of Applied Physics, University of Arkansas at Little Rock, Little Rock, AR 72204, USA
xiii
This page intentionally left blank
Chapter 1
Supersonic Nano-Particle Beam Technique for Removing Nano-Sized Contaminant Particles from Surfaces Jin W. Lee Department of Mechanical Engineering, Pohang University of Science and Technology (POSTECH), Pohang, South Korea
Chapter Outline
1. Introduction 2. Theoretical Background 3. Supersonic Nozzle Beam Technique
1 3 11
4. Electrospray Technique 5. Summary References
24 27 28
1. INTRODUCTION Particulate contamination seriously affects the manufacturing yield of micrometer- and sub-micrometer-scale devices. Semiconductor device features are expected to decrease continuously, reaching 25 nm by 2015 for the dynamic random access memory (DRAM)/flash memory devices, and as a result the critical defect size is expected to decrease to 20 nm by 2011 and 12.5 nm by 2015 [1]. Various nanotechnology-based devices with feature dimensions in the nanometer size range will also be marketed in time, which may accelerate the size level for contamination control to decrease. Although the theoretical adhesion force for the contaminant particles in the nanometer range is linearly proportional to the particle size, the fluid-dynamic drag force is proportional to the second power of particle size, so the use of drag force for cleaning becomes less efficient as the contaminant size is decreased. When it is recalled that the boundary layer thickness for typical wet processes is in the micrometer range, the removal performance of wet cleaning processes Developments in Surface Contamination and Cleaning. Copyright Ó 2011 Elsevier Inc. All rights reserved.
1
2
Developments in Surface Contamination and Cleaning
will get worse for nano-size contaminants. It is generally agreed that conventional techniques work poorly for sub-micrometer particles [2,3], and other disadvantages with wet chemistry cleaning become more apparent at smaller scales [4,5]. State-of-the-art cleaning technology based on conventional techniques is limited to removing particles larger than 50 nm [6e9]. One promising technique applicable in the sub-micrometer or nano-size range is the so-called cryogenic aerosol technique, where the contaminated surface is bombarded by fine particles of volatile material moving at a high velocity. It is easily inferred that contaminant particles adhered on a surface can be removed when the transferred energy from the bullet particles is sufficient to overcome the adhesion energy between the contaminant and the substrate. CO2 snow cleaning has long been used for cleaning large particles from optical devices or mechanical components [10], and was proved recently to be effective for cleaning particles down to the 30 nm size range [11]. McDermott et al. [12] showed experimentally that contaminant particles of 0.1e30 mm diameter could be removed efficiently using the cryogenic argon aerosols, and a number of studies reported the applicability of argon aerosol technique to nano-particle cleaning [11]. Argon, nitrogen and carbon dioxide are the most common cleaning agents used, and each offers advantages and disadvantages over the others. Aerosol cleaning is a promising alternative to the classical cleaning methods. This technique has matured in industry for large particles in the micrometer range, but not very well yet for nanometer particles. In this technique a condensable gas or gas mixture is pre-cooled close to liquid nitrogen temperature and then expanded through a simple nozzle like a cylindrical hole. During cooling, part of the gas becomes liquid, which gets atomized into fine droplets and then solidifies while expanding through a nozzle. Typical aerosol particle size is in the range of 0.5e50 mm and the velocity is about 100 m/s. State-of-the-art cleaning size remains around 30 nm [11], but cleaning efficiency drops very rapidly for contaminant particles smaller than 50 nm [13e15]. In another research study, Ar/N2 snow could remove particles down to 90 nm, but efficiency dropped rapidly below 100 nm [17,18]. Up to now it has been generally accepted that the kinetic energy or momentum of the aerosol particles is the key factor in determining the removal of contaminant particles. Yi et al. [19] showed by molecular dynamics (MD) simulation that removal efficiency for nano-sized contaminant particles was dependent on the velocity of the bullet particles closer to wV3, much more than V2, and concluded that even for the same kinetic energy condition smaller aerosol particles moving at a higher velocity should give better removal performance. It was also shown that if the bullet particle was too large compared to the target contaminant by a factor of 10 or more, the fragmented atoms/molecules of the bullet particle after collision may even surround the contaminant particle, preventing it from leaving the surface. This implies that there may exist a maximum allowable size of the bullet particle for removing
Chapter | 1
Supersonic Nano-Particle Beam Technique
3
contaminants in the nanometer range. When the bullet particle is excessively large in size, only a negligibly small fraction of its total kinetic energy is transferred to the contaminant particle and the rest is transferred to the substrate or the patterns nearby, causing undesirable damages. The use of smaller bullet particles moving at higher velocities is expected to have extra advantages in cleaning narrow trenches and in reducing the damage potential. The ratio of damage on the substrate or patterns to cleaning efficiency is expected to become smaller if cleaning is done with smaller aerosol particles moving at a higher velocity. Besides the unique cleaning performance for nano-contaminants which is nearly impossible with other techniques, the nano-particle beam technique has a number of practical advantages over other techniques. 1. It is a gas phase process and is also a perfectly dry process; so fewer chemicals are consumed and no post-processing is needed. 2. The basic mechanism is similar to that of wafer deposition or etching, so it can be incorporated into the wafer processing process with ease. 3. The system is compatible with in situ vacuum wafer processing; thus, there is no need to expose the wafer to the atmosphere between processing and cleaning. 4. Since it is vacuum- and fabrication-compatible, in situ monitoring or measurement is possible with SEM and other relevant techniques. 5. The cleaning head contains no moving parts. 6. It cleans by a one-to-one collision, so cleaning is very fast. There are shortcomings too. The biggest concern with the supersonic particle beam technique is the potential damage to the patterns, but the use of the nanoparticle beam reduces the damage potential drastically. Another disadvantage results from the fact that it is a line-of-sight process, so shielded parts cannot be cleaned at all. A multiple-path processing operation is needed.
2. THEORETICAL BACKGROUND When a particle adhered on a surface is removed by a momentum or an energy transfer from colliding particles, removal efficiency is usually formulated in terms of the amount of transferred quantities e force, moment, momentum or energy. The action of force and moment is instantaneous and not accumulated, but momentum and energy can be transferred over time and their effect can be accumulated. However, it is not yet well established which parameter is the proper criterion for the removal of nano-sized particles.
2.1. Adhesion Force No matter what criterion is used for removal, the starting point is always the adhesion force between a particle and a substrate. For a nano-sized particle, the
4
Developments in Surface Contamination and Cleaning
short-range van der Waals (VdW) force is the dominant one in the absence of electrostatic interaction or surface tension. There are three different interactions between molecules comprising the particle and the substrate: Keesom interaction between two permanent dipoles; Debye interaction between a permanent and an induced dipole; and London interaction between two induced dipoles. All three interactions are inversely proportional to the sixth power of the separation distance [20]. A simple formula for the VdW force for two spherical particles of diameter d1 and d2 can be written as eq. (1.1), which reduces to eq. (1.2) for a particle of diameter d on an infinite flat surface. A d2 Fs ¼ (1.1) d1 ðd1 þ d2 Þ 12z2 Fs ¼
Ad 12z2
(1.2)
Here A is the Hamaker constant which is a material property proportional to the product of the molecular number densities of the materials involved, and z is the minimum contact distance between particle and substrate, which is usually assumed to be 0.4 nm. The VdW force changes linearly with particle diameter, and for the case of A w 8.0 eV (¼1.28 10e18 J) which is typical of ceramic materials, the adhesion force becomes Fs (N) w 1.6 10e7 d (mm). This equation does not consider the many-body force or the retarding force, so it is strictly valid for large separation >5 nm, but it is widely used for adhered particles as a first approximation. The actual adhesion force can be very different from these simple equations due to asperity effects and deformation. Every surface has a surface roughness, and the contact force decreases as the number and height of asperities increase. Usually any height less than 0.5 nm is considered as molecularly smooth. The adhesion force always generates deformation on particles and surfaces, resulting in an increased contact area and contact force. After a long residence time the contact force can increase up to 100-fold for polymers and 20-fold for metals and oxides.
2.2. Force for Removal The adhesion force always acts normal to the surface. If a particle is pulled normal to a surface, the force required to detach the particle from the surface should be greater than the adhesion force. In most cleaning processes, however, particles are not removed by pulling against a surface, and the force on the particle is predominantly parallel to the surface. The force required to make the particle move, leaving the adhered spot by sliding or rolling, is much smaller, 1/10e1/100, than the adhesion force due to the rolling and asperity effect.
Chapter | 1
Supersonic Nano-Particle Beam Technique
5
Though there is no reliable theory for the removal force yet, a simple expression for it is usually written in the following way. If the interaction force between the bullet particle and the target particle is constant during the interaction period, the average removal force acting on the target particle can be written as the change of momentum of the bullet particle divided by the interaction time. Deceleration of the bullet particle is then constant, and the interaction time can be written as dt w db/v. Now the impact force generated by a collision can be written as follows. Fb ¼
dðmb vÞ mb v2 wdb v2 wðdb vÞ2 w db dt
(1.3)
Equation (1.3) predicts that the removal force becomes higher for the smaller particle size if kinetic energy or momentum is the same. If particle size and velocity are controlled independently, removal will be determined by the relative dominance between the magnitudes of Fs and Fb given in eqs. (1.2) and (1.3), respectively. If particle size and velocity are coupled, as in the electrospray technique, another equation relating size and velocity has to be incorporated. The above equations will hold well with reasonable accuracy for bullet and target particles of similar size. If two particles are of very different size, however, the interaction time cannot be simplified as described above, and the interaction force will not be constant during the interaction period. In any case, there does not exist a reliable theory for the removal force, particularly for nano-particles.
2.3. Fundamental Mechanism of Nano-Particle Removal 2.3.1. MD Simulation It is almost impossible to visualize the detailed process of nano-particle removal by experimental means, but characteristics of particle collision and removal in the nanometer scale can be well simulated by the MD technique. Yi et al. [19] successfully simulated the collision of a soft/volatile/fragile nanosized particle with a rigid surface or a hard particle on a surface, elucidating the effects of various factors on the particle removal characteristics. A standard MD algorithm was used, where the bullet particle, the contaminant particle and the substrate were modeled by clusters of molecules interacting via pair-wise potentials. An argon particle which is usually used as the bullet particle was simulated with the basic Lennard-Jones (LJ) potential, and the contaminant particle was simulated as an LJ particle with variable density (r) and binding energy (3). One example result is shown in Figure 1.1 for an LJ solid particle with 3 ¼ 10.0, which is slightly softer than an Al or a Cu particle, shot by an argon bullet particle at an angle of q ¼ 45 . The left figures show the states before collision. When Vx ¼ Vz ¼ V0 ¼ 2.0(3/m)1/2 (316 m/s) with total kinetic energy
6
Developments in Surface Contamination and Cleaning
FIGURE 1.1 Snapshots from MD simulation of the collision between a bullet particle (dark blob) and a contaminant particle (light blob) at three different times for two different bullet velocities: (a) Vx ¼ Vz ¼ 2.0(3/m)1/2 (316 m/s); (b) Vx ¼ Vz ¼ 3.0(3/m)1/2 (474 m/s)
of 1.38 10e19 J as in Figure 1.1a, the argon particle is partially disintegrated, but the contaminant particle just slides and rolls in the þx direction, without leaving the surface. When the particle velocity is increased by 50% with total kinetic energy of 3.10 10e19 J as in Figure 1.1b, the argon particle is completely disintegrated, and the contaminant particle becomes detached from the surface after collision. It is clearly shown that a higher kinetic energy of the bullet particle is more effective for removal.
2.3.2. Fundamental Factor for Removal A fundamental question from the viewpoint of the particle beam technique is whether the determining factor for particle removal is kinetic energy, momentum or force. An MD simulation with different combinations of mass and velocity at the same kinetic energy condition could answer the question [19], where the kinetic energy was twice as high as in Figure 1.1a but with different mass and velocity combinations: (a) 2m and V0; (b) m and O2V0; (c) m/2 and 2V0 (Fig. 1.2). At the high-mass and low-velocity condition as in Figure 1.2a, bullet particle disintegration is not complete, and the contaminant particle moves with sliding and rolling but stops after some distance. At the intermediate mass and velocity condition as in Figure 1.2b, bullet particle disintegration is almost complete, and the contaminant particle becomes detached after moving some distance with sliding and rolling. Part of the molecules comprising the contaminant particle gets disintegrated from the
Chapter | 1
Supersonic Nano-Particle Beam Technique
7
FIGURE 1.2 Snapshots from MD simulation of the collision between a bullet particle (dark blob) and a contaminant particle (light blob) at three different times, when the bullet kinetic energy is twice as high as in Figure 1.1 but at three different combinations of mass and velocity: (a) 2m and V0; (b) m and O2V0; (c) m/2 and 2V0
main body of the detached particle during the collision process. At the lowmass and high-velocity condition as in Figure 1.2c, bullet particle disintegration is complete, and the contaminant particle gets detached from the surface with a high velocity soon after collision. Of the three cases of Figure 1.2, case (a) has twice as high and case (b) O2 times as high a momentum as case (c). The simulation results lead us to conclude that momentum (wmv) cannot be an indicator for particle removal, and neither is kinetic energy (wmv2) a proper indicator. Particle removal seems to be determined by a new parameter with a much stronger dependence on velocity than mass. It also follows that an increased velocity is more effective for particle removal than an increased mass.
8
Developments in Surface Contamination and Cleaning
The behavior of the contaminant particle after collision can be classified into three modes: (1) the particle just oscillates about a fixed point but does not move at all; (2) the particle keeps moving with rolling and sliding but does not detach from the surface; and (3) the particle gets detached. The post-collision behaviors observed at various velocities and shooting angles from MD simulation show that a higher velocity gives a better removal and shooting angles between 15 and 45 give optimum performance.
2.3.3. Kinetic Energy vs. Binding Energy Volatile bullet particles disintegrate when colliding on the substrate or with the contaminant particle, and some extra energy can be released from the broken bonds in addition to the center-of-mass kinetic energy. MD simulation on the relative dominance on removal performance between the kinetic energy due to center-of-mass velocity and the internal energy due to binding of the bullet molecules shows that the kinetic energy is the determining factor. When the velocity of the bullet particle is high enough, the contaminant particle can be removed by the burst (binding energy) alone, but using the burst effect alone is not an efficient way for particle removal [19]. Another important finding from the simulation is that the kinetic energy of the bullet particle causing particle removal is almost ten times as high as the adhesion energy of the target LJ particle. It has long been believed that particle removal becomes effective once the kinetic energy of the bullet particle is higher than the adhesion energy [12], but the MD results show that almost tenfold higher kinetic energy is required for particle removal. This seeming discrepancy between the bullet particle energy and adhesion energy can be attributed to various reasons. First of all, in MD simulations removal is defined as the apparent detachment after some time, but in real situations another mechanism, such as the thermophoresis or the carrier gas flow, can induce or accelerate particle motion. Contaminant particles can then leave the surface even when impacted by a bullet particle of much lower kinetic energy than is required for detachment in the absence of any other force than collision. Another factor, which seems more fundamental and critical, is that because of bullet particle fragmentation on collision, the kinetic energy of the bullet particle is only partially transferred to the target particle. The effect of bullet particle fragmentation on energy transfer to the target contaminant could be simulated using a hard bullet particle which does not fragment on collision. Figure 1.3a shows the time variation of the kinetic energy of the bullet and target particle during a short period after collision. When the argon particle is completely fragmented after collision, energy transfer to the contaminant particle is not efficient (black solid lines). On the other hand, when the argon particle does not fragment, more of the kinetic energy is transferred to the contaminant particle (dotted lines). Figure 1.3b shows the velocity change of the contaminant particle. The x-directional
Chapter | 1
Supersonic Nano-Particle Beam Technique
9
FIGURE 1.3 Time variation of kinetic energy of bullet and contaminant particles (a) and the center of mass velocity of the contaminant particle before and after collision (b); q ¼ 45 , Vx ¼ 3.0 (474 m/s) and Vz ¼ e3.0. 1s ¼ 2.15 ps. ‘CP’ refers to contaminant particle, and in (a) the solid black line is for a fragile bullet and the dotted line is for an artificial hard bullet
velocity abruptly increases after collision, implying that the contaminant particle begins to move, reaches a peak velocity after about 60s, and then slows down. The z-directional velocity begins to change after some time lag, which means that the initial motion is a sort of slip along the surface. When the bullet particle does not fragment, the z-directional velocity of the contaminant particle is negative immediately after collision, but restores to zero with time. This implies that the target particle gets compressed upon collision and recovers
10
Developments in Surface Contamination and Cleaning
later. The interaction time (acceleration of the contaminant particle and deceleration of the bullet particle) is much longer for the fragile bullet particle, about 10 ps, and the maximum x-directional velocity of the contaminant particle impacted by a hard bullet particle is about 25% higher than that with a fragile bullet particle.
2.3.4. Other Considerations Deformation also affects the removal behavior due to the increased adhesion in the presence of deformation. Post-collision behavior of the strongly bound target particle is much different from that of a weakly bound particle without deformation. When the bullet particle velocity is low and at low impact angle (close to horizontal), the contaminant particle starts to move but stops very soon, but at high impact angles (close to vertical) it does not move but gets deformed by collision. When the bullet particle velocity is high, the weakly deformed particle gets detached if shot at low angle, but it fragments without moving if shot at a high angle. The strongly deformed particle fragments, partially or fully, irrespective of the impact angle, because the strong adhesion force prevents the contaminant particle from moving. Since the velocity of bullet particles cannot be much higher than 1000 m/s, particle bombardment may not be an effective means of removing deformed particles. When Hill reported the CO2 snow cleaning mechanism in 1994 [21], she proposed that physical contact was not enough to dislodge the contaminant particle. Instead, as the cold snow flake approaches the surface, it is exposed to a large, much warmer surface, the solid snow rapidly changes phase, and a high-velocity burst is created, which is attributed as the main force for dislodging of the particles. For bullet particles of argon, the same mechanism may come into play. However, when the time required for an argon particle to be completely evaporated is compared with the time-of-flight to collision, it becomes evident that the collision is much faster than evaporation. Long et al. [22] simulated the evaporation of a liquid drop, and the time required for a 1.7 nm diameter droplet to be fully evaporated was calculated as more than 1.8 ns when the droplet and surrounding temperatures were 78 K and 120 K, respectively. The expected vaporization time for the conditions considered in this discussion is much longer than the collision time, and the particle fragmentation is not due to evaporation but due to physical collision. Only when the incoming velocity of the bullet particle is very low, about 10 m/s, it does become probable that the bullet argon particle does not fragment after collision but gets vaporized through contact with a warm surface. The effect of density and hardness of the contaminant particle on the removal behavior was also studied. Because the Hamaker constant is proportional to density, the adhesion energy is increased due to the higher density, and a higher energy is required to dislodge the target particle. Simulation results show that the removal efficiency for particles of higher density and hardness is
Chapter | 1
Supersonic Nano-Particle Beam Technique
11
generally lower, particularly for the conditions of low velocities and small collision angles. At an intermediate collision angle of 45 the removal performance stays almost the same, and at higher angles of 60 and 75 removal is even enhanced at high velocities. This removal enhancement at high collision angles and velocities can be attributed to the elastic repercussion of the elastic target particle against the hard substrate. And from a practical viewpoint it can be safely assumed that the removal characteristics are dominated by the particleesubstrate adhesion energy, not by the intra-particle binding energy.
2.4. Consideration of the Bullet Particle Size Effect Though particle movement is the starting point in the whole removal process, removal is completed only when the particle acquires a sufficiently high velocity to take off from the surface. As can be inferred from Figure 1.1, fragments of the bullet particle may pile up around the departing contaminant particle, preventing it from gaining a high enough velocity. A smaller-size bullet particle is advantageous in this sense. A small bullet particle ensures a short interaction time and, thus, a high interaction force. Also, if the colliding (bullet) particle is fragile, not all the kinetic energy of the bullet particle but only the kinetic energy of the bullet molecules close to the location of contact between the bullet and contaminant particle will act on the contaminant particle, and the rest will be transferred to the fragmented and/or evaporated bullet molecules or substrate. The larger the bullet particle the smaller the fraction of the total kinetic energy of the bullet that will be transferred to the contaminant particle. It thus follows that small bullet particles are favored for removal if only the total kinetic energy is high enough.
3. SUPERSONIC NOZZLE BEAM TECHNIQUE In order to prevent the potential blocking of the moving contaminant by the fragments of bullet particles, bullet size needs to be reduced and velocity increased instead. In the current cryogenic aerosol technique, bullet particles are generated by atomization, and the final size of bullet particles is in the micrometer range and the velocity w100 m/s. For a contaminant size of 10 nm, a 1 mm size bullet is too large. The bullet size needs to be reduced below 100 nm, and velocity increased to higher than 500 m/s. Atomization cannot generate particles with these properties effectively, and new techniques have to be sought. One way of generating extremely tiny particles is the electrospray technique, where an electrostatic repulsive force is used to disintegrate a liquid into extremely tiny droplets. This technique will be briefly introduced in the next section. Another process working in the opposite direction of size enlargement is the homogeneous nucleation and growth during supersonic expansion through a Laval nozzle. In this technique particle size and velocity can be controlled by the nozzle contour, gas composition, and stagnation pressure and temperature.
12
Developments in Surface Contamination and Cleaning
3.1. Homogeneous Nucleation and Growth One efficient way of generating a high-velocity beam of particles with sizes below 100 nm and high number density is homogeneous nucleation and growth during a supersonic expansion. Particle generation by supersonic expansion has been investigated experimentally and numerically for several decades, using a pure gas or a mixture of a condensable gas and an inert carrier gas such as helium [23e29]. In general, particles or droplets generated by homogeneous nucleation and growth are very small in size, usually less than 10 nm. Particle size can be increased if higher pressure and lower temperature are used, but there appears to be a practical limit due to the pressure ceiling imposed by the saturation pressure and the temperature range required for nucleation in the supersonic section to avoid clogging of the nozzle throat.
3.1.1. General Features of Homogeneous Nucleation and Growth In heterogeneous condensation occurring on a foreign surface, condensation starts and continues indefinitely so long as the vapor pressure remains higher than or equal to the saturation pressure at the surface temperature. On the other hand, in homogeneous nucleation where condensation nuclei are formed amid a gas environment, a much higher gas/vapor pressure is required than the saturation pressure in order to generate condensed nuclei that can remain stable in the gas environment and even grow in size. The super-saturation ratio (S), the ratio of the vapor pressure to the saturation pressure, required for stable nucleation varies depending on the cooling speed. In supersonic expansion through a micro nozzle, the cooling rate can be as high as 106 K/s and S may reach values as high as 100 (Fig. 1.4). When the expansion path of a gas crosses the saturated vapor line on a PeT diagram, the gas becomes supersaturated. Though various sizes of FIGURE 1.4 Saturation curves (lines) and nucleation onset points (symbols) for argon (dark dots) and nitrogen (light dots). Vertical lines are the solidification lines, and the curved lines are the sublimation and vaporization lines
Chapter | 1
Supersonic Nano-Particle Beam Technique
13
nuclei are formed when S is greater than 1, only those nuclei larger than the so-called critical size (r*) can grow stably. The super-saturation ratio required for a sufficiently high nuclei formation rate is called the critical super-saturation ratio, S*, but the criterion for S* is rather arbitrary. One way of defining S* is to use the point of departure from the isentropic expansion path, which results from the latent heat released by the nuclei and condensed molecules. The turning point is usually called the nucleation onset point (Fig. 1.5a). Beyond this point, the nuclei keep growing in size along the flow path as long as S > 1.0 (Fig. 1.5c). Particle size can be increased if higher pressure and lower temperature are used, but there still appears to be a practical limit due to the pressure ceiling of the saturation pressure and the temperature range required for nucleation only in the supersonic section to avoid clogging of the nozzle throat. As will be shown later, the typical particle size that can be reached by homogeneous nucleation and growth with a nozzle of a few cm length is limited to 30 nm diameter for Ar and 50 nm for N2 at 3000 Torr and 120 K. The simplest model for the homogeneous nucleation is that of Volmer and Weber [30], where the critical size (r*) and the nucleation rate (J) are modeled by the following equations, and J is calculated for each constituent gas species separately when a gas mixture is considered. Here An is the Avogadro’s number, and DGðrÞ the Gibbs energy change for forming a critical nucleus. r ¼
2s rl RT ln S
DGðr Þ ¼
16ps3
3ðRT ln SÞ2 2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P 1 2sA3n DGðr Þ exp Jðr Þclassical ¼ 3 RT rl pMwt kT
(1.4) (1.5)
(1.6)
Particles or droplets, once nucleated, continue to grow due to impingement by the surrounding gas molecules. The growth process can be modeled in various ways, and the model of Hill [31] is one of the simplest, where the net condensation rate is determined by a balance between collisional condensation and spontaneous evaporation due to the higher particle temperature resulting from the latent heat release. dr x P PD pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (1.7) ¼ dt rc 2pRTD 2pRT P and T are for the impinging gas phase, PD is the saturation pressure at the equilibrium particle temperature TD, and rc is the condensate density. x is called the mass accommodation coefficient, introduced to correct for the discrepancy between the model prediction and experimental results, and is very
14
Developments in Surface Contamination and Cleaning
(a)
P[bar]
3
Saturation line, Solid-Liquid Saturation line, Vapor-Solid and Vapor-Liquid Critical onset points Expansion path
2
1
0 40
Nucleation rate [cm-3 sec-1]
(b)
60
80
T [K]
100
2.0E+19 N2
1.5E+19 1.0E+19 5.0E+18 0.0E+00 0.2
0.4 0.6 X [cm]
0.8
1
100
Particle Diameter [nm]
(c)
80 60 40 20 0
0
0.2
0.4
0.6
0.8
1
X [cm] FIGURE 1.5 Particle generation by pure nitrogen expansion starting at P0, N2 ¼ 3000 Torr and T0 ¼ 110 K: (a) the expansion path plotted on the PeT diagram; (b) distribution of nucleation rate; (c) the growth paths of the nuclei generated at corresponding axial locations of (b)
Chapter | 1
Supersonic Nano-Particle Beam Technique
15
close to 1.0 for most cases [31]. When there is more than one condensable species in the gas phase, particle growth is the sum of condensational growths by each condensable species. Microscopic details of condensational growth can be found in Yi and Lee [32].
3.1.2. Example of Generating N2 Particles by Nucleation and Growth When pure N2 gas is expanded through a nozzle starting from P0 ¼ 3000 Torr and T0 ¼ 110 K, the characteristics of nucleation and growth are as shown in Figure 1.5. Nucleation starts when the expansion path arrives at the onset line (Fig. 1.5a), but terminates within 4 ms of nucleation inception (Fig. 1.5b). Also the nucleation rate is extremely high, in the range of 1025/m3/s. Nuclei generated at various locations along the nozzle will grow in size toward the nozzle exit (Fig. 1.5c). The growth rate is high just after nucleation or within the nucleating zone, but is quite low after nucleation is terminated or downstream of the nucleating zone. The high growth rate in the nucleating zone results from condensation on the nuclei, whose number or surface area increases explosively within the nucleating zone (Fig. 1.5b). The total number density of particles at the exit is about 5 1018/m3. Continued condensation past the nucleation zone makes the particle size increase steadily, but at a much lower rate. Particle size continues to grow as long as supersonic expansion is maintained, but most particle growth is accomplished within a short distance from the nucleation zone, and the mean particle size at the exit is about 50 nm. Due to the extremely high nucleation rate and rapid growth, particle size distribution is very narrow, with a geometric standard deviation of about 1.27. The maximum particle size obtainable by homogeneous nucleation and growth at 3000 Torr and 120 K is 30 nm for Ar and 50 nm for N2 [33].
3.2. Bullet Particle Generation Experimental results for removing 20-nm particles with nano-particle beam generated with a supersonic nozzle are well summarized by Lee et al. [34,35]. The cleaning facility consists of a nozzle for particle generation, an extra nozzle for purge gas, a wafer motion stage and various monitors (Fig. 1.6). The whole system is contained in a vacuum chamber with an ultimate chamber pressure below 1 Torr. In order to generate bullet particles a condensable gas, such as Ar, N2, CO2 or a mixture of these gases, is pre-cooled close to the triple point of the condensable gas, and then expanded through a supersonic Laval nozzle in a vacuum environment. During supersonic expansion through the nozzle, small condensation nuclei are formed and grow in size, where the final size can be controlled by the stagnation pressure and temperature, back pressure of the vacuum chamber, and the nozzle contour. Velocity of the particle beam is
16
Developments in Surface Contamination and Cleaning
FIGURE 1.6 Schematic of the setup for the supersonic nozzle for aerosol cleaning
controlled by the nozzle contour, gas composition and stagnation temperature. When a smaller particle size at a higher population or velocity is required, a mixture of the condensable gas and a light carrier gas such as He is used. When the partial pressure of the condensable species is higher than the saturation pressure, part of the condensable gas will change to liquid, and particles are formed by atomization. On the other hand, when the condensable partial pressure is lower than the saturation pressure nuclei will be formed through homogeneous nucleation. When a larger particle is needed, the starting pressure is increased or a longer nozzle is used. In order to estimate the size of the bullet particles, a wafer coated with a photoresist (PR) film was exposed to the particle beam for 3 seconds and then scanning electron microscope (SEM) images were taken. When Ar particles were generated by liquid atomization at a pressure of 4000 Torr, higher than the saturation pressure as in the conventional cryogenic particle cleaning, dents of 1e10 mm diameter were formed (Fig. 1.7b). On the other hand, when Ar particles were generated by homogeneous nucleation at a pressure of 3000 Torr, lower than the saturation pressure, dents of much smaller diameter of 50e80 nm were formed (Fig. 1.7a). It is clearly shown that bullet particles generated by gas-phase nucleation are smaller by a factor of 10e20 than those generated by liquid breakup and the particles also have a narrower size distribution.
3.3. Cleaning 20 nm Particles with Argon Bullets Particle size and velocity can be varied by varying the nozzle contour, gas composition, carrier gas, and starting pressure and temperature. Particle size is increased when started at high pressure and low temperature and when a long
Chapter | 1
Supersonic Nano-Particle Beam Technique
17
FIGURE 1.7 SEM images of the dents on a wafer surface coated with photoresist film: (a) by the particle beam generated by homogeneous nucleation and growth through the long contoured nozzle (tiny black dots), and (b) by liquid atomization through the short nozzle
nozzle is used. Particle velocity is increased when started at high temperature and when a light carrier gas is mixed. Proper combination of conditions has to be sought in order to get optimum bullet particles. In any case, the Mach number at the nozzle exit should be higher than 5.0 in order to achieve satisfactory cleaning. In the following discussions, cleaning results obtained with Ar particles generated at various conditions are given. Bullet particles are generated with two different gas mixtures, pure argon or 1:1 Ar/He mixture, pre-cooled to 90e120 K. Two Laval nozzles with different expansion angles and lengths are considered, where the long nozzle is three times as long as the short nozzle. Wafers coated with various contaminant particles e Al2O3, SiO2, SiC, TiO2 and Cu e are exposed to the Ar particle beam for 2 minutes for cleaning.
3.3.1. Ar Particles Generated from Pure Argon The discussion will start with bullets generated from pure Ar, because it is closer to the conventional cryogenic aerosol technique. When an Si wafer coated with Al2O3 particles is bombarded by a particle beam generated with pure argon at 1500 Torr through the short nozzle, the cleaning effect is almost zero, but when the pressure is raised to 1850 Torr, all the particles are completely removed, irrespective of the size (Fig. 1.8). Since the saturation pressure of Ar at 95 K is about 1600 Torr, Ar particles generated at 1500 Torr are formed via homogeneous nucleation, but the size is too small to remove the contaminant particles. At 1850 Torr, instead, large enough particles are generated through atomization to remove the contaminant particles. Argon particles generated at 1850 Torr make dents of diameters 1e10 mm on a PR-coated film (Fig. 1.7b), while particles generated at 1500 Torr do not make dents of visible size. When cleaning is carried out at different background pressures, cleaning performance degrades with increase in chamber pressure
18
Developments in Surface Contamination and Cleaning
FIGURE 1.8 SEM images before (left) and after (right) cleaning. Al2O3 particles on Si surface and argon stagnation pressure of 1500 Torr (top) and 1850 Torr (bottom)
(Fig. 1.9). From this observation, it can be concluded that the much improved removal performance of the Ar particles formed by atomization in a Laval nozzle can be attributed to the effect of increased supersonic velocity in contrast to subsonic velocity in conventional aerosol techniques. Insufficient expansion or acceleration at high chamber pressure conditions lowers the velocity of the generated particles. Cleaning characteristics for other ceramic particles such as SiO2, TiO2 (Fig. 1.10) and SiC are almost the same as for Al2O3. Particles deposited in a trench with 1 mm width and 1 mm depth are seen to be completely removed except in the shadow area (Fig. 1.11). Particles left in the shadow area can be removed easily by a second cleaning by injecting the beam in the opposite direction.
3.3.2. Argon Particles Generated from Ar/He Mixture The increase in flow velocity by use of a light carrier gas should result in improved removal, but at the same time results in reduced time for particle growth, which is the flow time through the nozzle. Thus, the advantage of using a carrier gas can be assured only when the reduced growth time can be compensated for by using a longer nozzle. Argon particles are generated with 1:1 Ar/He mixture through two different nozzles e the short nozzle and the long nozzle e starting from the same conditions of 3000 Torr and 95 K, and shot at a flat Si wafer coated with 20 nm Al2O3 particles. When the short
Chapter | 1
Supersonic Nano-Particle Beam Technique
19
FIGURE 1.9 Cleaning of TiO2 particles on Si surface with argon particles generated at 1850 Torr and 95 K at four different chamber pressures: (a) 150 Torr, (b) 100 Torr, (c) 50 Torr, and (d) 10 Torr. Image before cleaning is similar to the left figure of Figure 1.10
nozzle is used, contaminant particles are not removed at all (Fig. 1.12), but when the long nozzle is used, the surface is perfectly cleaned, irrespective of the size of the contaminants (Fig. 1.13). Since the saturation pressure of argon at 95 K is about 1600 Torr, the pressure cannot be increased any further in order to generate bullet particles by means of homogeneous nucleation. It is certain that bullet particles are formed through gas-phase nucleation and growth, and the velocity of the particle beam is higher than 300 m/s if the chamber pressure is controlled low enough to ensure a true supersonic expansion inside and downstream of the nozzle to the wafer surface. It is successfully demonstrated in Figure 1.13 that argon bullet particles generated by homogeneous nucleation through a Laval nozzle can remove contaminants on a flat Si surface down to 20 nm. The only difference between Figures 1.12 and 1.13 lies in a different nozzle length or expansion angle. Roughly speaking, the size of the bullet particles from the long nozzle is three times as large as those from the short nozzle, because the rate of growth in particle diameter is almost linearly proportional to growth time or residence time, and absolute flow speed does not change sensitively to Mach number at
20 Developments in Surface Contamination and Cleaning
FIGURE 1.10 SEM images for TiO2 particles on Si surface before cleaning (left) and after cleaning with argon particles generated at 1500 Torr (middle) and 1850 Torr (right)
Chapter | 1
Supersonic Nano-Particle Beam Technique
21
FIGURE 1.11 SEM images before (left) and after (right) cleaning of Al2O3 particles in a 1 1 mm trench using argon particles generated at 1850 Torr
FIGURE 1.12 SEM image (a) before and (b) after cleaning with the short nozzle. Al2O3 particles on Si surface and 1:1 Ar/He mixture at 3000 Torr and 95 K
FIGURE 1.13 SEM image (a) before and (b) after cleaning with the long nozzle. Al2O3 particles on Si surface and 1:1 Ar/He mixture at 3000 Torr and 95 K
22
Developments in Surface Contamination and Cleaning
high Mach number conditions [31]. Since a factor of 3 in diameter is equivalent to a factor of 27 in mass, the insufficient cleaning with the short nozzle can be attributed to insufficient kinetic energy of the bullet particles generated with the short nozzle. If a lower pressure of 2000 Torr is tried with the long nozzle, more than 90% of the contaminants are removed (Fig. 1.14). Considering the low partial pressure (1000 Torr) of Ar at this condition, it is verified that the bullet particles generated through a well-contoured Laval nozzle are very effective in removing contaminant particles down to 20 nm range. However, no cleaning is achieved with pure Ar at 1500 Torr (Fig. 1.15). Comparison of the two cases in Figure 1.13 (1:1 Ar/He mixture at 3000 Torr) and in Figure 1.15 (pure Ar at 1500 Torr) shows that the use of light carrier gas has the effect of increasing particle velocity through increased sonic speed, and also enhancing particle growth by removing condensation heat through collisions on growing bullets. Both effects should result in enhanced removal performance [36]. The increase in flow velocity by use of a carrier gas will result in reduced flow time through
FIGURE 1.14 SEM images (a) before and (b) after cleaning with the long nozzle. Al2O3 particles on Si surface and 1:1 Ar/He mixture at 2000 Torr and 95 K
FIGURE 1.15 SEM images (a) before and (b) after cleaning with the long nozzle. Al2O3 particles on Si surface and pure Ar at 1500 Torr and 95 K
Chapter | 1
Supersonic Nano-Particle Beam Technique
23
FIGURE 1.16 SEM images before (left) and after (right) cleaning of Cu particles on a Si surface using 1:1 Ar/He gas mixture expanded through a long nozzle at 4000 Torr
the nozzle, or reduced growth time. Thus, the advantage of using a carrier gas appears only when the reduced growth time can be compensated for by enhanced heat removal or growth. Use of a long nozzle is an effective way of taking advantage of the heat removal effect without the adverse effect of reduced growth time. When the short nozzle is used (Fig. 1.12), the typical bullet particle size is thought to be in the range of 15e25 nm, which will not be enough to remove contaminants of 20e80 nm size. Here again, cleaning characteristics for other ceramic particles such as SiO2, TiO2, and SiC are almost the same, and also narrow trenches can be cleaned very effectively due to the small size of the bullets. Also, Cu particles are completely cleaned when a 1:1 Ar/He mixture is expanded at 4000 Torr and 95 K through the long nozzle (Fig. 1.16). Based on the above findings, it is now acceptable that even smaller contaminants than 20 nm can be cleaned effectively, if only the bullet particle size is well controlled such that it is close to or a little larger than the contaminant size.
3.4. CO2 Snow Cleaning As was mentioned in the Introduction, the particle beam technique was first developed with CO2. CO2 from the pressure vessel is ejected through a simple nozzle starting from a liquid or liquid/gas mixture. Thus, CO2 particles of large size are generated, and are visible like snow. CO2 snow has long been used for cleaning glass/optic products and mechanical parts. Lately, it has begun to be used for wafer cleaning, but its use was limited to contaminants larger than 100 nm. Very recently, van der Donck et al. [11] reported successful removal of 30 nm polystyrene latex (PSL) particles on flat surfaces and also in trenches. It is not possible to understand details of the technique since the generation conditions or the particle properties were not clearly shown, but it is introduced here for information.
24
Developments in Surface Contamination and Cleaning
CO2 snow was generated by expanding CO2 gas or liquid through a variable orifice unit with a ball valve (Model K5-10s, Applied Surface Technologies, New Providence, NJ, USA). The pressure conditions or nozzle parameters were not specified. Two different contaminants, PSL and silica, were tested on, or in, three different substrates e flat, trench, and gap between sharp lines. The width and depth/height of the line or trench were 100/150/200/ 250/300 nm, and 100/250 nm, respectively. Trench/line patterns were e-beam written on an Si wafer in a 10 10 array, and etched on Si or CVD deposited on Al. PSL particles of 30 and 50 nm were spin-coated, and 30e100 nm silica particles were deposited by a low-pressure impactor. The nozzleesubstrate distance and the angle were varied over 20/30/40 mm and 30/60/90 degrees, respectively, and cleaning speed was 4 cm2/s with liquid CO2 and 1 cm2/s with gas CO2. Silica particles were almost perfectly cleaned even in a trench, and even when cleaned 7 months after deposition. For cleaning 50 nm PSL particles in a trench or on a line with liquid CO2, the cleaning efficiency was 60% for a trench (250 nm depth) and 70e80% between lines (100 nm height). The angle was very important, with the efficiency at 60 angle (vertical) being just onethird of that at a 30 angle (75%). For PSL particles on a flat surface, removal efficiency varied a lot depending on the distance and whether the starting medium was liquid or gas. With liquid, 100% removal of 30 nm PSL was observed at the smallest distance (20 mm), but the best performance with gas was 96% for 30 nm PSL at 30 mm distance. It was not explained why the smaller distance of 20 mm was not tried. Damage to the patterns is a serious concern with the particle beam technique. When gas was used, there was no damage at all. When liquid was used, thin lines were damaged, but the trenches were not damaged. Damage was severe when cleaning was done sideways.
4. ELECTROSPRAY TECHNIQUE Another method of generating tiny particles moving at a high speed is the electrospray technique. The electrospray technique has long been developed for applications in the field of space propulsion, coating, and particle formation. Mahoney and his colleagues have applied this technique to nano-particle removal for over 20 years, and called it the Micro Cluster Beam Technique. Only a brief introduction is given here, and details of the technique can be found in their publications [4,37,38].
4.1. Electrospray When a conducting liquid is fed into a metal capillary tube and positive voltage is applied to the tube, positive ions migrate toward the open surface and negative ions are attracted to the tube wall. If the electrostatic stress on the
Chapter | 1
Supersonic Nano-Particle Beam Technique
25
interface/surface is higher than the surface tension, the liquid meniscus formed at an opening is disrupted and tiny droplets are ejected. The electrostatic stress or the electric field near the nozzle tip is increased when a small-diameter nozzle and/or a high voltage are used [39,40]. Particles generated in this technique are charged liquid droplets, not neutral solid particles. Usually an electric voltage is maintained between the nozzle and the surrounding emitter electrode, and another electric voltage is applied between the emitter electrode and the wafer surface for particle acceleration. The velocity or kinetic energy at collision is controlled by the accelerating voltage. There appear several different modes of spraying depending on the electric voltage and liquid flow rate. At low-voltage conditions, the liquid meniscus is shaped like a single sharp tip, and spraying occurs at the tip in a single array, which is called the cone jet mode. If the voltage is increased over a certain threshold value, spraying occurs at multiple spots along the rim of the nozzle exit, which is called the crown or rim jet mode. When a high generation rate is needed as in particle removal, the rim jet mode is usually used. In the rim jet mode, the size of the generated droplets is non-uniform, but the mean size increases with flow rate at a given electric field strength (E) and decreases with E at a given flow rate. Electrospraying is possible in both vacuum and atmospheric conditions, but a higher maximum voltage before breakdown is needed in vacuum. Since the generated droplets are charged, fast removal of the charges accumulating on the substrate is an important practical problem. Beam half angle is a function of the acceleration voltage, increasing with decreased voltage, and typical values are a few degrees at 10e20 kV. Droplets are usually much smaller than 1 mm, and the energy per cluster is O (1 MeV), which corresponds to w1 eV per molecule of the cluster. Particles are accelerated to supersonic velocity in order to clean surface films and particles.
4.2. Particle Removal Though this technique uses different means of generating and accelerating bullet particles, it is similar to the supersonic nozzle technique in the removal mechanism. Removal is based on the one-to-one collision between bullet and contaminant, so the kinetic energy of the droplets is the source of particle removal, and size matching is important. Films are removed by a micro-shock induced by a collision of the bullets on a substrate. Removal performance is determined by the kinetic energy of the colliding droplets, which is again determined by the accelerating voltage. The formula for the removal force was derived by combining theoretical and experimental equations. When the Rayleigh criterion, eq. (1.8), of the maximum charge sustainable by a liquid sphere of surface tension g and diameter d [41] is incorporated into an experimental relationship between the
26
Developments in Surface Contamination and Cleaning
electric voltage Va and current I, eq. (1.9), formulas for the charge per particle mass (q/m), particle diameter (d), and velocity (v) can be obtained as eqs. (1.10)e(1.12). The final result for the impact force, eq. (1.13), shows that the removal force is proportional to d1/2, varying much more slowly with particle size than for the supersonic nozzle technique. q w g1=2 d3=2
(1.8)
I w Va2
(1.9)
q w g1=2 d3=2 m 2=3 Q g1=3 d w g1=3 w 4=3 w Va4=3 I Va h q i1=2 w Va3=2 Va v ¼ 2 m Fb ¼
dðmvÞ mv2 w g1=2 d 1=2 Va w Va1=3 w d dt
(1.10) (1.11) (1.12) (1.13)
For a glycerol solution with surface tension g ¼ 0.005 N/m, these equations predict particle size, velocity and generation rate as 40e135 nm, 0.63e2.73 km/s and 0.1e3.7 1010/s for voltage variations in the range 5e15 kV. High voltage gives small size, high velocity and large number of particles. The impact force is estimated as 1.5e2.5 (10e5 N/particle). In case the cluster and the contaminant are of the same size, the removal force calculated by eq. (1.13) is 1e2 orders of magnitude larger than the adhesion force for contaminant sizes of 10 nme1.0 mm, with the relative factor larger for the smaller contaminant sizes.
4.3. Typical System Configuration and Performance In order to control conductivity of the liquid solution, ammonium acetate was added to the following solutions to 1.0 M concentration: 50/50 H2O/ N-methyl-2-pyrrolidone (NMP), conductive glycerol solution, pure glycerol, H2O/methanol and H2O/isopropyl alcohol. Typical system configuration and operating conditions used by Mahoney et al. [4,37,38] were as follows. l l
l l l l
Chamber pressure: 10e3e10e5 Torr Fused silica glass capillary: inner diameter (ID) 50 mm, outer diameter (OD) 375 mm Liquid flow rate: 0.5e2 mL/min Emitteretarget distance: 125 mm Voltage and current: 15 kV, 0.5 mA Cleaning time: 5 minutes.
Chapter | 1
Supersonic Nano-Particle Beam Technique
27
The cleaning performance obtained with the above-mentioned conditions was (1) particles in the size range of 50 nme5 mm were completely removed, but particles larger than 5 mm remained on the surface and could not be removed [38]; (2) 1 mm PSL particles were removed even in a crevice; and (3) one emitter can clean 20 cm2 in 2 minutes. This technique has several advantages for removing nano-contaminants. l l l l l l
Easy and gentle velocity control Extremely low liquid consumption rate, on the order of wmL/wafer Wide area coverage using a linear array of nozzles and scanning Reduced impact damage Low power consumption No consumption of expensive or hazardous liquids.
Potential cleaning application areas are wide ranging, such as wafer backside cleaning, flat panel displays, photo lithography masks, post chemical mechanical planarization (CMP) cleaning, disk drives, microelectromechanical systems (MEMS) devices, optical gyroscopes, and space optics.
5. SUMMARY The new aerosol cleaning technique based on nano-sized particles/droplets, instead of micrometer-sized particles/droplets used in the conventional aerosol cleaning techniques, can overcome the 50 nm particle size barrier and remove contaminant particles of ceramics, polymers, and metallic copper down to 20 nm with nearly 100% efficiency. Key to the successful cleaning of nanosized contaminant particles is the combination of smaller bullet size and a higher velocity. Bullet particles satisfying these requirements can be generated in two different ways e supersonic expansion and electrospray. In the supersonic expansion technique, solid bullet particles are generated by homogeneous nucleation and growth, where particle size and velocity are controlled by the gas composition, nozzle contour, and stagnation conditions e pressure and temperature. In the electrospray technique, liquid droplets are generated by electrostatic atomization, where droplet size and velocity are controlled by the solution composition and the accelerating voltage. Both techniques have advantages and disadvantages relative to each other, but nanosized bullets impacting at supersonic speed common to both techniques. Smaller size and higher velocity of the bullet particles than in the conventional aerosol techniques reduces the damage potential, in addition to achieving improved cleaning performance.
ACKNOWLEDGMENT This work was supported by the Korea Science and Engineering Foundation (KOSEF) grants funded by the Korea government (MOST) (No. ROA-2008-000-20045-0).
28
Developments in Surface Contamination and Cleaning
REFERENCES [1] SEMATECH, International Technology Roadmap for Semiconductors (2007). [2] D.S. Rimai, D.J. Quesnel, Fundamentals of Particle Adhesion, Global Press, Srbija. Presently available through The Adhesion Society, Blacksburg, VA, USA, 2001. [3] W. Zapka, W. Ziemlich, A.C. Tam, Efficient Pulsed Laser Removal of 0.2 mm Sized Particles from a Solid Surface, Appl. Phys. Lett. 58 (1991) 2217. [4] J.F. Mahoney, P. Julius, S. Carl, J.C. Andersen, Removal of Particulate and Film Contaminants by Impacting Surfaces with Microcluster Beams, in: K.L. Mittal (Ed.), Particles on Surfaces 5&6: Detection, Adhesion and Removal, VSP, Utrecht, The Netherlands, 1999, pp. 311e325. [5] Y. Momonoi, K. Yokogawa, M. Izawa, Dry Cleaning Technique for Particle Removal Based on Gas-Flow and Down-Flow Plasma, J. Vacuum Sci. Technol. B 22 (2004) 268. [6] K. Bakhtari, R.O. Guldiken, P. Makaram, A.A. Busnaina, J.G. Park, Experimental and Numerical Investigation of Nanoparticle Removal Using Acoustic Streaming and the Effect of Time, J. Electrochem. Soc. 153, G846 (2006). [7] S.I. Kudryashov, S.D. Allen, S.D. Shukla, Experimental and Theoretical Studies of Laser Cleaning Mechanisms for Submicrometer Particulates on Si Surfaces, Particulate Sci. Technol. 24 (2006) 281. [8] H.K. Lim, D.D. Jang, D.S. Kim, J.W. Lee, Correlation between Particle Removal and Shock-wave Dynamics in the Laser Shock Cleaning Process, J. Appl. Phys. 97 (2005) 054903e1. [9] R. Vanderwood, C. Cetinkaya, Nanoparticle Removal from Trenches and Pinholes with Pulsed-Laser Induced Plasma and Shock Waves, J. Adhesion Sci. Technol. 17 (2003) 129. [10] R. Sherman, Carbon Dioxide Snow Cleaning, in: K.L. Mittal (Ed.), Particles on Surfaces 5&6: Detection, Adhesion and Removal, VSP, Utrecht, The Netherlands, 1999, pp. 221e237. [11] J.C.J. van der Donck, R. Schmits, R.E. van Vliet, A.G.T.M. Bastein, Removal of Sub-100-nm Particles from Structured Substrates with CO2 Snow, in: K.L. Mittal (Ed.), Particles on Surfaces 9: Detection, Adhesion and Removal, VSP/Brill, Leiden, The Netherlands, 2006, pp. 291e302. [12] W.T. McDermott, R.C. Ockovic, J.J. Wu, R.J. Miller, Removing Submicron Surface Particles Using a Cryogenic Argon-Aerosol Technique, Microcontamination (October 1991) 33e36. [13] J.M. Lauerhaas, J.F. Weygand, G.P. Thomes, Advanced Cryogenic Aerosol Cleaning: Application to Damage-Free Cleaning of Sensitive Structured Wafers, Proc. IEEE/SEMI Advanced Semiconductor Manufacturing Conference (2005) 11e16. [14] H. Lin, K. Chioujones, J. Lauerhaas, T. Freebern, C. Yu, Damage-free Cryogenic Aerosol Clean Processes, IEEE Trans. Semiconductor Manufacturing 20 (2007) 101. [15] T.J. Wagener, K. Kawaguchi, Improved Yields for the Nano-Technology Era Using Cryogenic Aerosols, Proc. IEEE/SEMI Advanced Semiconductor Manufacturing Conference (2004) 467e471. [16] N. Narayanswami, J.F. Weygand, P. Reuther, K.K. Christenson, J.W. Butterbaugh, S.H. Yoo, et al., Evaluation of Particle Removal Efficiency in Wafer Cleaning Processes, Semiconductor Intl (June 2000). [17] M. Okada, S. Kwada, Y. Sonoda, Stencil Reticle Cleaning Using an Ar Aerosol Cleaning Technique, J. Vac. Sci. Technol. B 20 (2002) 71.
Chapter | 1
Supersonic Nano-Particle Beam Technique
29
[18] P.G. Clark, J.W. Butterbaugh, G.P. Thomes, J.F. Weygand, T.J. Wagner, D.S. Becker, Compatibility of a Cryogenic Aerosol Process on SiLK@ and Porous MSQ, Proc. IEEE Intl. Symp. on Semiconductor Manufacturing (2003) 479e482. [19] M.Y. Yi, D.S. Kim, J.W. Lee, J. Koplik, Molecular Dynamics (MD) Simulation on the Collision of a Nano-Sized Particle onto Another Nano-Sized Particle Adhered on a Flat Substrate, J. Aerosol Sci. 36 (2005) 1427. [20] H. Krupp, Particle Adhesion: Theory and Experiment, Adv. Colloid Interf. Sci. 1 (1967) 111. [21] E.A. Hill, Carbon Dioxide Snow Examination and Experimentation, Precision Cleaning 36 (February 1994). [22] L.N. Long, M.M. Micci, B.C. Wong, Molecular Dynamics Simulations of Droplet Evaporation, Comput. Phys. Comm. 96 (1996) 167. [23] M.R. Hoare, P. Pal, P.P. Wegener, Argon Clusters and Homogeneous Nucleation: Comparison of Experiment and Theory, J. Colloid Interf. Sci. 75 (1980) 126. [24] N.G. Garcia, J.M.S. Torroja, Monte Carlo Calculation of Argon Clusters in Homogeneous Nucleation, Phys. Rev. Lett. 47 (1981) 186. [25] G. Koppenwallner, C. Dankert, Homogeneous Condensation in Nitrogen, Argon, and Water Vapor Free Jets, J. Phys. Chem. 91 (1987) 2482. [26] Y. Okada, K. Sunouchi, H. Ryu, A. Patra, K. Ashimine, K. Takeuchi, Measurement of Condensation Onset in Steady Supersonic Laval Nozzle Flow for the Molecular Laser Isotope Separation Process, J. Nucl. Sci. Technol. 35 (1998) 158. [27] D.W. Oxtoby, Homogeneous Nucleation: Theory and Experiment, J. Phys. Cond. Matter 4 (1992) 7627. [28] P.P. Wegener, Nucleation of Nitrogen: Experiment and Theory, J. Phys. Chem. 91 (1987) 2479. [29] B.J.C. Wu, P.P. Wegener, G.D. Stein, Homogeneous Nucleation of Argon Carried in Helium in Supersonic Nozzle Flow, J. Chem. Phys. 69 (1978) 776. [30] M. Volmer, A. Weber, Nuclei Formation in Supersaturated States, Z. Phys. Chem. 119 (1926) 277. [31] P.G. Hill, Condensation of Water Vapor During Supersonic Expansion in Nozzles, J. Fluid Mech. 25 (1966) 593. [32] M.Y. Yi, J.W. Lee, Condensation and Evaporation of a Nano-sized Particle Moving in a Fluid Environment, J. Aerosol Sci. 38 (2007) 764. [33] H. Bae, I. Kim, E. Kim, J.W. Lee, Generation of Nano-Sized Ar-N2 Compound Particles by Homogeneous Nucleation and Heterogeneous Growth in a Supersonic Expansion, J. Aerosol Sci. 41 (2010) 243. [34] J.W. Lee, K.S. Kang, K.H. Lee, M.Y. Yi, M.J. Lee, Removing 20 nm Particles Using a Supersonic Argon Particle Beam Generated with a Contoured Laval Nozzle, J. Adhesion Sci. Technol. 23 (2009) 769. [35] K.S. Hwang, M.J. Lee, M.Y. Yi, J.W. Lee, Removing 20 nm Ceramic Particles Using a Supersonic Particle Beam from a Contoured Laval Nozzle, Thin Solid Films 517 (2009) 3866. [36] D. Kane, S.P. Fisenko, M. Rusyniak, M.S. El-Shall, The Effect of Carrier Gas Pressure on Vapor Phase Nucleation Experiments Using a Thermal Diffusion Cloud Chamber, J. Chem. Phys. 111 (1999) 8496. [37] J.F. Mahoney, Microcluster-Surface Interactions: A New Method for Surface Cleaning, Int. J. Mass Spectrom. Ion Proc. 174 (1998) 253. [38] J. Perel, J. Mahoney, P. Kopalidis, R. Becker, Particle Removal by Collisions with Energetic Clusters, in: K.L. Mittal (Ed.), Particles on Surfaces 8: Detection, Adhesion and Removal, VSP, Utrecht, The Netherlands, 2003, pp. 345e352.
30
Developments in Surface Contamination and Cleaning
[39] A.T. Blades, M.G. Ikonomou, P. Kebarle, Mechanism of Electrospray Mass Spectrometry. Electrospray as an Electrolysis Cell, Anal. Chem. 63 (1991) 2109. [40] G.J. Van Berkel, F. Zhou, Characterization of an Electrospray Ion Source as a ControlledCurrent Electrolytic Cell, Anal. Chem. 67 (1995) 2916. [41] Lord Rayleigh, On the Equilibrium of Liquid Conducting Masses Charged with Electricity, Phil. Mag. 14 (1882) 184.
Chapter 2
Megasonic Cleaning R. Nagarajan,1 S. Awad2 and K.R. Gopi3 Department of Chemical Engineering, IIT Madras, Chennai, India, 2Crest Ultrasonics Corporation, Trenton, NJ, USA, 3Advanced Ceramic Technologies, Penang, Malaysia 1
Chapter Outline
1. Introduction 2. Cleaning Mechanism 3. Theory of Megasonic Cleaning 4. Surface Cleanliness Measurement
31 34 41 45
5. Megasonic System Evaluation in the Laboratory and in Industry 6. Industry Case Studies 7. Concluding Remarks References
48
58 59 60
1. INTRODUCTION Wikipedia, the online encyclopedia, defines megasonic cleaning as ‘a type of acoustic cleaning, related to ultrasonic cleaning’. It is a gentler cleaning mechanism, less likely to cause damage, and is currently used extensively in wafer cleaning in semiconductor manufacturing. It is well known that micrometer and sub-mm-sized particulate contaminants cause defects in microelectronic devices, resulting in yield loss and reliability degradation [1e11], and that acoustic fields can greatly enhance removal of such contaminants from product and component surfaces. Similar to ultrasonic cleaning, megasonics utilizes a transducer, usually composed of piezoelectric crystals to create megasonic energy (Fig. 2.1) [12]. Megasonic energy is of a higher frequency (800e2000 kHz) than typical ultrasonic cleaners (<100 kHz). As a result, the cavitation that occurs is gentler and on a much smaller scale [13]. The difference between ultrasonic cleaning [14] and megasonic cleaning lies in the frequency that is used to generate the acoustic waves. Ultrasonic cleaning uses lower frequencies; it produces random cavitation. Megasonic cleaning uses higher frequencies at and above 1000 kHz; it produces controlled cavitation. An important distinction between the two methods is that the higher Developments in Surface Contamination and Cleaning. Copyright Ó 2011 Elsevier Inc. All rights reserved.
31
32
Developments in Surface Contamination and Cleaning
FIGURE 2.1 Finite element simulation of propagation of megasonic waves in a tank [12]
megasonic frequencies do not cause the violent cavitation effects found with ultrasonic frequencies. This significantly reduces or eliminates cavitation erosion and the likelihood of surface damage to the product being cleaned. Parts that would be damaged by ultrasonic frequencies or cavitation effects can often be cleaned without damage in a megasonic bath using the same solution. With ultrasonics, cavitation occurs throughout the tank, and all sides of the submerged parts are cleaned. With conventional megasonics, only the side of the part that is facing the transducer(s) is cleaned (Fig. 2.2) [15]. Megasonic cleaning is widely used for removing particles from wafer surfaces, as well as from critical component surfaces in other high-technology products, but its underlying mechanism is not clearly understood even by many of the practitioners. Lester [16] indicates that, in general, there are two principal classes of mechanisms through which megasonic cleaners could accomplish cleaning: through direct action of the sound field with the particle; and through indirect action, via acoustic cavitation. A single-wafer megasonic cleaner has been described by Wu et al. [17] where control of sound distribution is achieved by varying the operating conditions of the transducer assembly. This cleaner is designed for integration with process tools in both front-end and back-end applications. Kim et al. [18]
Chapter | 2
Megasonic Cleaning
33
FIGURE 2.2 Micro-streaming in a megasonic field [15]
present a comparison between the design and operation of typical ultrasonic and megasonic cleaners, sketched in Figure 2.3. They have theoretically analyzed and experimentally characterized both types of cleaners. The transmission of the ultrasonic wave is shown to be affected not only by the geometric and material parameters of the plate, but also by the mass density and wave speed in the adjacent liquid of the incident wave region and the transmitted wave region. The mass density and wave speed in the liquid are found to be dependent on the temperature and the chemical composition of the solution. For megasonic cleaners, it has been shown that the transmission of the sound power through the bottom plate of the inner container is affected by the plate thickness and inclination as well as by the liquid temperature, and the optimal condition for the maximum cleaning performance has been found in terms of several relevant parameters. Berg et al. [19] have demonstrated a ‘reciprocity principle’ in megasonic cleaning, i.e. decreasing exposure time to a megasonic field while increasing field intensity by the same ratio results in equivalent cleaning performance. Thus, cleaning performance depends only on the ‘acoustic exposure’ (average
34
Developments in Surface Contamination and Cleaning
FIGURE 2.3 Cross-sectional diagrams of (a) ultrasonic and (b) megasonic cleaners [18]
power times exposure time), rather than on time or power separately. This implies that megasonic cleaning equipment can best be improved by providing higher average power transducers.
2. CLEANING MECHANISM Sonic cleaners work by two principal mechanisms: l l
Cavitation Acoustic streaming.
Cavitation is a process where the constructive interference of sonic energy causes the formation of rarified bubbles in the cleaning fluid. When these microscopic bubbles implode (due to the passage of the rarefaction energy, the moving sound wave), they produce microscopic jets of liquid that can impinge on the surface of parts to be cleaned. These high-velocity jets remove particles
Chapter | 2
35
Megasonic Cleaning
FIGURE 2.4 Ultrasonic cavitations and cleaning
from surfaces and convey cleaning chemicals to organic and inorganic chemical contamination on the surface. This mechanism has been described in detail by Awad and Nagarajan [20]. When acoustic waves constructively combine, the resulting decrease in pressure creates a localized bubble. In properly degassed solutions, this bubble consists almost entirely of solvent vapors. When the ultrasonic pressure wave goes through a compression cycle, the localized pressure drops and the bubble collapses. When this occurs, a microscopic jet of liquid is formed, jetting from the bubble wall into the volume of the bubble. This high-velocity jet scours the surface of parts it comes in contact with, knocking loose material from the surface (Figs 2.4 and 2.5). Cavitations are generated on the order of microseconds. At the 20 kHz frequency, it is estimated that the pressure is about 35e70 kPa and the transient localized temperatures are about 5273 K, with the velocity of microstreaming around 400 km/h. In acoustic streaming, bulk movement of fluid occurs. Contaminants that get removed from the surface are carried away by acoustic streaming, and hence, are FIGURE 2.5 Growth and collapse (implosion) of a cavitation bubble
36
Developments in Surface Contamination and Cleaning
prevented from re-attaching to the surface. Acoustic streaming can penetrate through the boundary layer of motionless fluid that surrounds all of the surfaces in the ultrasonic tank. Particles dislodged from the surface by cavitation action are not swept away from the surface and may get re-attached. At high frequencies (>200 kHz) the acoustic streaming is highly directional, and so the orientation of the part to be cleaned becomes critical. At low ultrasonic frequencies, the acoustic streaming is randomized and not highly directional. Gouk et al. [21] have obtained sound pressure fields in cleaning tanks using a ‘ray acoustic’ model that vectorially adds rays of acoustic energy emitted isotropically from a transducer, including various interference and barrier effects. Microstreaming near a wafer surface is simulated and particle removal forces are estimated, with the results being compared against laboratory measurements. Cavitation and acoustic streaming work together in all forms of ultrasonic cleaning, but the relative contribution of each is a function of frequency. At low ultrasonic frequencies, cavitation is very strong and dominates the cleaning process. At high ultrasonic frequencies, characteristic of megasonic cleaning, cavitation bubbles are very small, but acoustic streaming velocities can be very high. Thus at high frequencies, acoustic streaming dominates the cleaning process and less of the cleaning occurs due to cavitation. Figure 2.6 shows a plot of cavitation strength versus frequency. We note that the cavitation strength decreases rapidly with increasing frequency. Also, cavitation abundance (bubble density) with frequency is pictured in Figure 2.7. (Note the decrease in bubble size with increase in bubble density.) High-intensity ultrasonic waves create fine bubbles in the liquid medium, which grow to maximum sizes proportional to the applied ultrasonic frequency
FIGURE 2.6 Cavitation strength as a function of frequency
Chapter | 2
Megasonic Cleaning
37
FIGURE 2.7 Cavitation abundance varies with frequency
and then implode, releasing their energies. The higher the frequency, the smaller is the cavitation bubble size. Thus, in the megasonic regime, cavitation intensity is relatively weak, but the number density of bubbles is large, and the bubble size is very small. Minsier and Proost [22] point out that cavitation effects can still be significant in the megasonic range. Their calculations for different gas equations of state inside the bubble show that the van der Waals law predicts a slightly higher liquid velocity at the shock-front than when considering a perfect gas law. Also, decreasing the value of the interfacial tension at the bubbleeliquid interface results in an increase in the liquid velocity at the shock front. Their calculations indicate that the strength of the shock waves emitted upon spherical bubble collapse can cause delamination of typical device structures used in microelectronics. The boundary layer effect is very significant in case of surface cleaning. During this process, the cleaning solution rushes past the substrate being cleaned, forcing chemicals onto contaminant particles, removing them from the surface, and carrying them away. On a microscopic scale, the fluid friction at the surface causes a very thin layer of solution to move more slowly than the bulk solution. This layer of slow-moving fluid at the substrate surface is called the boundary layer (Fig. 2.8). The boundary layer effectively shields the substrate surface from fresh chemistry and shields contaminants from the removal forces of the bulk fluid. As frequency increases, the momentumtransfer boundary layer thickness decreases as square of the frequency. Hence, in the megasonic regime, even sub-mm particles are exposed to the cleaning fluid and chemistry.
38
Developments in Surface Contamination and Cleaning
FIGURE 2.8 Change in boundary layer thickness (relative to contaminant size) from ultrasonic to megasonic regime
The thickness of the boundary layer surrounding the parts is a function of the ultrasonic frequency in the tank. The higher the ultrasonic frequency, the thinner is the boundary layer. This is illustrated in Figure 2.9, where the boundary layer thickness is plotted as a function of frequency. The boundary layer, next to the substrate surface where the sound does not penetrate, is essentially motionless. At 40 kHz, it is fairly thick at 2.8 mm where smaller particles can be trapped. As frequency increases, the boundary layer is reduced, permitting the fluid to get closer to the surface and, therefore, the contaminants. For example, at 400 kHz, the boundary layer is reduced to 0.98 mm. In the >1 MHz range, the boundary layer is thin enough to expose
FIGURE 2.9 The relationship between frequency and boundary layer thickness for room temperature water (theoretical simulation)
Chapter | 2
Megasonic Cleaning
39
FIGURE 2.10 Effect of acoustic field parameters on streaming velocity [23]
even nanometer-sized particles to shear forces. Bakhtari et al. [23] have presented the effect of acoustic frequency and amplitude (represented as intensity) on streaming velocity in Figure 2.10. They have also developed a model for particle removal based on ratio of removal moment to adhesion-resisting moment (MR) (Fig. 2.11), and compared their theoretical prediction with experimental data (Fig. 2.12).
FIGURE 2.11 Moment ratio model for particle removal from a surface. MR is the removal moment; MA is the adhesion resisting moment; Fdrag is the drag force; FAdhesion is the adhesion force; Felec double layer is the electric double layer force; R is the particle radius; a is the contact radius between the deformed particle and the surface [23]
40
Developments in Surface Contamination and Cleaning
FIGURE 2.12 Experimental and theoretical particle removal efficiencies. Si-cap refers to a 4 nm silicon capping layer on the wafer [23]
At very high frequencies approaching 1 MHz, cavitation becomes a secondary phenomenon compared to ‘acoustic streaming’, which is the timeindependent fluid motion generated by a sound field [24]. The associated streamlines are shown in Figure 2.13. Principally, this flow is categorized into two main types [24e28]: (1) streaming caused due to spatial attenuation of the wave in free space; and (2)
FIGURE 2.13
Fluid motion generated by a sound field [24]
Chapter | 2
Megasonic Cleaning
41
streaming caused by friction between the vibrating medium and the solid wall. The second mechanism is further classified as inner streaming that is induced within the acoustic boundary layer, and outer streaming which is the steady vortex flow developed outside the acoustic boundary layer. In particle removal, streaming activity close to a surface plays the key role. The so-called Schlichting streaming occurs in a viscous boundary layer in a sound field [29]. This streaming produces vortices of a scale much smaller than the acoustic wave length. Velocity gradients are large, and transport is enhanced due to this streaming.
3. THEORY OF MEGASONIC CLEANING Deymier et al. [30] have characterized the forces of adhesion and removal as a function of size for spherical particles on a silicon wafer surface (Fig. 2.14). The adhesion forces represented as dotted lines in the figure are for three different separation distances between the particle and the silicon surface, ˚ . The dashed-dotted line stands for the removal force in namely 5, 7, and 10 A the case of the silicon/water system for an excited Rayleigh wave. The dashed line refers to the water/silicon/water system and excitation of a bulk wave. They found that the component of the streaming force parallel to the solidefluid interface can lead to particle removal by a rolling and tugging mechanism. They have shown that subjecting a silicon wafer to a grazing incident acoustic wave, as is traditionally done in megasonic cleaning tanks, may not lead to an optimum cleaning efficiency. Normal and parallel components of the streaming
FIGURE 2.14 Removal and adhesion forces on a spherical particle [25]
42
Developments in Surface Contamination and Cleaning
force are strongly dependent on the incidence angle of the incoming wave. Their results suggest that cleaning efficiency may be improved by subjecting a wafer to incident acoustic waves sampling a wide range of incidence angles, between e45 and 45 . Kim et al. [31] have proposed that particles near radially oscillating bubbles can be detached from a solid surface by two mechanisms. The liquidegas interface is passed through by a particle located within the oscillation range during the bubble’s radial motion. Then the torque due to Laplace pressure is exerted on the particle. On the other hand, particles located outside the range swept by the meniscus are affected by the short-range dynamic pressure gradient. In summary, the movies obtained through their work reveal that megasonic cleaning is achieved by local fluid motion induced by oscillating bubbles close to resonant size. The estimation of the exerted torque by the two proposed mechanisms provides a possible explanation. Globally generated acoustic pressure gradient and acoustic streaming have only minor direct effects on particle removal. However, they may play secondary roles in cleaning. For the cleaning of an entire wafer, multiple bubbles should move around so that their paths can cover the entire surface. This bubble movement is primarily caused by the Bjerknes force [32], which is the result of the acoustic pressure gradient. Furthermore, the particles lifted off the solid surface must be driven away to prevent reattachment, which could be made possible by acoustic streaming. They also note that the mechanism of megasonic cleaning is different from that of surface cleaning by millimeter-sized bubbles which can be generated by laser irradiation, and which collapse to generate a high-speed liquid jet onto a solid surface. The general theory of ultrasonic cleaning has been presented in detail by Awad and Nagarajan [20]. It may be reasonably inferred that bubble collapse pressure, PC, directly affects the extent of surface impact by cavitation forces, and thereby influences any induced erosive effects. The instantaneous cluster wall velocity at the time of cluster collapse, SC (m/s), is calculated by Kanthale et al. [33] as: SC ¼ ½PC =ðrbð1 bÞÞ1=2
(2.1)
where r is the density of the medium, and b is the void fraction in the cluster. A stated limitation of this model is that ‘acoustic streaming’ e i.e. medium streaming induced by pressure gradient along the ultrasonic beam e has been neglected (in order to simplify the model by neglecting frictional forces between the liquid and cavity cluster). This aspect needs to be dealt with in generalizing the model for megasonic cleaning.
3.1. Contribution of Acoustic Streaming to Particle Removal from Immersed Surfaces Acoustic waves that propagate in liquids observe the general laws of hydrodynamics. Nyborg [24] solved the Navier-Stokes equation for a Newtonian
Chapter | 2
Megasonic Cleaning
43
liquid using second-order approximations. Markham [25] showed that streaming was due to sound absorption and relaxation processes. In principle, constant streaming occurs in all acoustic radiation fields, and increases until the intensity is reduced by beam divergence and/or attenuation in the medium [28]. Tjotta [26] introduced a simple formula in which the streaming velocity, SAc, was proportional to the absorption coefficient a, the beam width 2a, and the acoustic intensity I, and was inversely proportional to the viscosity of the medium h, and the sound velocity in the medium c: SAc ¼ 8aIa2 =ðhcÞ
(2.2)
The absorption coefficient, a, varies as the square of the frequency, f, according to the following expression: a ¼ 2hf 2 =ð3rc3 Þ
(2.3)
Combining eqs (2.2) and (2.3) eliminates the dependence of acoustic streaming velocity on fluid viscosity: SAc ¼ 16Ia2 f 2 =ð3rc4 Þ
(2.4)
This acoustic streaming velocity is a significant contributor to the shearing force that acts to dislodge particles from exposed surfaces; the total shearing velocity that the particle experiences may be written as the sum of the contributions to particle shear from cavitational cluster collapse and acoustic streaming [34]: S ¼ SC þ SAc
(2.5)
The corresponding total acoustically induced tangential particle removal shear, stan, may then also be written as a sum of the contributions due to cluster collapse, stan,C, and due to acoustic streaming, stan,Ac: stan ¼ ðh=x0 Þ : ðSÞ
(2.6)
where x0 is the thickness of the boundary layer. It is important to note that, unlike the shearing stress due to a collapsing cavity cluster, the shear imparted by acoustic streaming is non-erosive, and only serves to enhance particle depletion from surfaces. Therefore, the optimal cleaning stress is now augmented by the contribution of streaming shear, per eq. (2.5). Acoustic streaming, which gains in importance as cavitation intensity is reduced, has few detrimental effects on the immersed surface. Thus, its incorporation in the surface cleaning/erosion model does not alter the setting of optimum cavitation pressure, but does enhance the particle removal velocity at the surface. Since the streaming velocity depends on the square of the frequency, it is particularly significant in high-frequency ultrasonic fields
44
Developments in Surface Contamination and Cleaning
FIGURE 2.15 Variation of streaming velocity with frequency
(170 kHz), and even more so in the megasonic range (1 MHz). However, it should be borne in mind that in the case of particles that are strongly adhered to surfaces, acoustic streaming alone cannot be relied upon to dislodge and remove the particles. Typically, a staged cleaning system e with lowerfrequency ultrasonics at the front end to loosen the particles, followed by higher-frequency ultrasonics and megasonics to flush the loosened particles away from the surface e would be required in order to optimize surface cleaning. Experimental acoustic streaming velocities of 4 m/s have been reported in water at a frequency of 850 kHz. At such high frequencies, it can be assumed that all the ultrasonic power is used up in generating streaming motion, i.e. absorbed by the liquid medium. Cavitation phenomena can be mostly neglected. Assuming the power level remains constant, the streaming velocity can be obtained at other frequencies by using the square law dependence of streaming velocity on frequency. The graph in Figure 2.15 illustrates the calculated variation of streaming velocity with frequency. During the streaming process, a linear momentum, whose magnitude is characterized by the frequency of sound applied, is imparted to the fluid medium, and is the source of the shear force that dislodges particles from immersed surfaces. The streaming velocity at 470 kHz is calculated to be approximately 1.22 m/s and, from bubble-collapse velocity calculations, the associated cavitational velocity is found to be of the order of 10e4 m/s (which is relatively very small). Kuehn et al. [35,36] have studied particle removal from semiconductor wafers by megasonic cleaning from a fundamental perspective. They have modeled the sound pressure field and acoustic velocity field in the tank, and compared them to measured data, using a hydrophone and optical flow
Chapter | 2
Megasonic Cleaning
45
visualization, respectively. They have identified large shear forces associated with microstreaming as the primary mechanism of particle removal.
4. SURFACE CLEANLINESS MEASUREMENT While unpatterned wafers and other highly polished surfaces, such as hard disks, can be directly inspected for contamination levels, rough and complex substrates do not lend themselves to such procedures. Various instruments used for indirect (extractive) measurements of surface cleanliness have been extensively described by Awad and Nagarajan [20]. These are briefly summarized here.
4.1. Liquid Particle Counters Liquid particle counters (LPCs) are used to determine the size and number of particles suspended in liquids. This instrument utilizes the principle of ‘near angle light scatter’, shown schematically in Figure 2.16, and consists of a basic light source, a laser diode (wavelength 670.8 nm). The beam from this laser is spatially filtered and focused by a lens assembly to form a small and well-defined illuminated volume within the liquid being inspected. As the illuminated volume moves across a particle suspended in the liquid, some light from the beam will be scattered. Much of this scattered light is in the near-forward direction and is collected by the optical system of the photo-detector assembly. The amplitude and width of this pulse is a function of the size of the particles. The amount of light scattered by a particle in the sensitive zone of the optical system is a function of the scattering angle and the relative index of refraction of the
FIGURE 2.16
Optical principle of the liquid particle counter
46
Developments in Surface Contamination and Cleaning
particle. This instrument collects and averages light that has been scattered in a near-forward direction over a solid angle ranging from 4o to 19o. Variation of collected light is of the order of 15% in a single reading of one particle count. This instrument typically detects particles from 0.5 to 100 mm. When particles are extracted ultrasonically from immersed component surfaces into a liquid medium, the LPC can be used to indirectly quantify surface contamination levels by counting and sizing the extracted particles suspended in the liquid. In the case of complex substrates, such as hard disk drive components, this may be the only practical option to measure surface cleanliness.
4.2. Precision Turbidity Meter The American Public Health Association (APHA) defines turbidity as an ‘expression of the optical property that causes light to be scattered and absorbed rather than transmitted in straight lines through the sample’. Turbidity can be interpreted as a measure of the relative clarity of water. Turbidity is not a direct measure of suspended particles in water; instead, it is a measure of the scattering effect such particles have on light. The optical property expressed as turbidity is the interaction between light and suspended particles in water. A directed beam of light remains relatively undisturbed when transmitted through absolutely pure water, but even the molecules in a pure fluid will scatter light to a certain degree. Therefore, no solution will have zero turbidity. While most earlier turbidimetric methods measured the transmitted light, turbidity measurement standards changed in the 1970s, when the nephelometric turbidimeter was developed, which determines turbidity by the light scattered at an angle of 90 from the incident beam (Fig. 2.17). A 90 detection angle is considered to be the least sensitive to variations in particle size. Nephelometry is a preferred means for measuring turbidity because of the method’s
FIGURE 2.17 Schematic diagram of a nephelometer
Chapter | 2
Megasonic Cleaning
47
sensitivity, precision, and applicability over a wide range of particle sizes and concentrations. The preferred expression of turbidity is nephelometric turbidity units (NTU). As in the case of surface cleanliness measurement using extraction followed by LPC, here, too, extraction followed by turbidimetry can provide quantification of surface particulate levels.
4.3. Microbalance A microbalance may be used to quantify mass loss from a coupon by surface cleaning, or by cavitation erosion. The Cahn C-34/C-35Ô is one such sensitive weight and force measurement instrument. It is designed for weights and forces up to 3.5 grams and is sensitive to changes as small as 0.1 microgram. The balance may be described as a force-to-current converter. The current necessary to produce the required torque motor force is a direct measure of the force on the beam. The process of calibration allows this current to be measured in units of weight (grams). In order to measure mass loss due to ultrasonic cleaning, the material coupon is first cleaned with pure water and dried in the oven so that it loses all its moisture and is weighed using the Cahn Microbalance before immersing it in the beaker full of water. This beaker is then suspended at the center of the tank using a fixture to hold it. The ultrasonic generator is switched on and the power level is adjusted. Experiments may be done to simulate even mild cavitation conditions (i.e. megasonic) and low power inputs. After a certain time (say, 2 minutes), the generator is turned off. The specimen is then taken out and dried in the oven. When the specimen is dry, it is taken out and reweighed using the microbalance. This step is repeated in a multiple-extraction mode.
4.4. Cavitation Meter The ultrasonic cavitation meter (ppb-500Ô) is an instrument used to measure the energy density (in watts per unit area) of cavitation in liquids (Fig. 2.18). It is not a sound meter or hydrophone. The main difference is that it measures cavitation or the collapse of water bubbles as they implode on a surface, instead of sound waves produced by a pressure transducer. The ultrasonic cavitation meter is simple and easy to use, yet it contains sophisticated electronics and options for data storage, retrieval, and analysis. The meter measures the instantaneous energy in a given direction. The probe is a 50-cm (20-inch) long stainless steel tube with an ethylene propylene diene monomer (EPDM) half-sphere at one end and a cable at the other end. The black half-sphere is made of elastic material to isolate the filter lens mounted on it from the holding rod. The lens is a thick quartz crystal. Cavitation generated by the sound pressure waves is produced in the form of bubbles that grow and implode with micro-streaming water jets hitting the filter
48
Developments in Surface Contamination and Cleaning
FIGURE 2.18 Cavitation meter
surface. The sensor mounted behind the lens detects these impacts and the signal is sent via cable to the electronic case.
5. MEGASONIC SYSTEM EVALUATION IN THE LABORATORY AND IN INDUSTRY Cavitation intensity as a function of acoustic field frequency and position in the tank has been studied extensively at ACT Laboratories in Penang, Malaysia. The measured data at 27 locations (nine at each level e top, bottom, and middle) are shown in Figure 2.19. Figure 2.20 presents a comparison of mean cavitation intensity for frequencies extending down to 192 kHz. It is evident that there is a significant reduction in cavitation intensity as frequency increases, and that there is very little variability with respect to position in tank. Vereecke et al. [37] have used a particle measurement method known as ‘haze mapping’ to study the influence of process parameters on megasonic particle removal efficiency (PRE). In particular, the effects of dissolved gas and
Chapter | 2
Megasonic Cleaning
49
FIGURE 2.19 Cavitation intensity as a function of frequency and position within the cleaning tank
FIGURE 2.20 Mean cavitation intensity as a function of frequency (192e960 kHz)
solution chemistry are shown in Figures 2.21 and 2.22. In contrast to ultrasonic cleaning, degassing actually reduces megasonic PRE. Particle removal efficiency was studied by Busnaina et al. [29] at frequencies of 40, 65, 80, and 850 kHz. The 850 kHz experiments were performed using a commercially available megasonic cleaning system (nominal frequency 862 kHz and maximum power input 150 W), while commercial ultrasonic tanks and generators (40, 65, 80, and 100 kHz) were used in the remaining experiments. Silicon wafers used were 125 mm p-type (100). The wafers were cleaned prior to deposition of the particles using an RCA Standard
50
Developments in Surface Contamination and Cleaning
FIGURE 2.21 Influence of dissolved oxygen on the particle removal efficiency (PRE) of 34-nm SiO2 particles [30]
FIGURE 2.22 Particle removal efficiency for 78- and 34-nm SiO2 particles in aerated deionized (DI) water and ammonia/peroxide mixture (APM) solutions (a) with megasonic agitation and (b) without megasonic agitation [37]
Clean (SC1) (1 NH4OH: 1 H2O2: 5 H2O) solution and scanned by a laser surface scanner (with 0.1 mm resolution) to establish a background particle count. The background count of particles of unknown origin was subtracted from the total number of particles on the wafer before and after sonic cleaning. Particles used were polystyrene latex (PSL) spheres, SiO2 (silica) spheres, and non-spherical Si3N4 (silicon nitride). Mean particle diameters of 0.3, 0.4, 0.5, 0.6, 0.7, and 1.0 mm were employed. These particles, originally in a concentrated high-purity aqueous solution, were mixed with isopropanol to
Chapter | 2
51
Megasonic Cleaning
form a dilute solution, eliminating the problem of particle agglomeration. The resulting suspension was atomized using a nebulizer with 0.1 mm filtered air, and deposited onto the wafers. Approximately 150e300 particles were deposited on each wafer, resulting in relatively low pre-cleaning particle counts (1.1e2.4 particles/cm2). Wafers were then loaded into a 25-wafer perfluoroalkoxy (PFA) Teflon cassette, which was inserted vertically into the tank. After the required immersion time in deionized (DI) water, the cassette was removed from the tank. Wafers were rinsed in DI water, dried, and re-scanned to obtain the post-cleaning particle count. Removal efficiency l was computed as follows: lð%Þ ¼ f½Nbefore Nafter =Nbefore g 100
(2.7)
where Nbefore is the number of particles deposited on the water surface prior to sonic cleaning, and Nafter is the number of particles remaining on the surface after cleaning. For any particular operating condition, ten experiments were run, ten removal efficiency values were measured, and their average was calculated. Table 2.1 shows removal efficiencies for various PSL sphere sizes in DI water at each frequency tested [29]. Efficiency was expected to increase with frequency, as indeed it did with the exception of 80 and 100 kHz. In all cases, removal was observed to increase with time until a maximum efficiency was reached, after which there was either no improvement or even a slight degradation in cleaning. Removal efficiency was also observed to decrease with decreasing particle size, consistent with the fact that the ratio of particle adhesion force to removal force scales approximately as particle size. A megasonic cleaner operates more effectively at high electrical power. Yet, there are times when the system is not cleaning effectively and power is increased beyond the saturation level, which continues to reduce cleaning effectiveness. Results from experiments demonstrate that, in fact, higher power (in excess of 300 W) does not deliver effective cleaning [16]. The researchers found that this reduced effectiveness at high power is not caused by limitations
TABLE 2.1 Particle Removal Efficiency (%) at Various Frequencies [29] Particle Diameter, mm
40 kHz
65 kHz
80 kHz
100 kHz
862 kHz
1.0
88
95
83
87
95
0.7
84
87
75
75
90
0.5
75
84
70
70
85
0.3
70
72
65
57
N/A
Note: Each number represents an average of three data points. Immersion time was 20 minutes.
52
Developments in Surface Contamination and Cleaning
in the electrical or transduction systems, but is a cavitating fluid effect. Specifically, if there is too much cavitation, the increased volume fraction of gas in the liquid prevents the acoustic field from propagating. Another experiment [16] examined the sonoluminescence flux within the cleaner e pre-existing gas-filled nuclei necessary for cavitation inception. When the liquid remained undisturbed for 10 minutes, and then engaged, the level of cavitation activity was quite low after an initial spike, and slowly showed decreasing activity over a 2-minute period. If nucleation sites were continually supplied by bubbling air through the liquid both before and during cleaner operation, similar behavior was observed; however, the level of cavitation activity was more than three times higher. In the last experiment [16], the system remained undisturbed for several minutes, and then both the cleaner and the bubbler were engaged. Results showed that the level of cavitation activity gradually increased from the lower value, observed without the bubbler, to the higher value observed when the bubbler was active e indicating that it would be better to bubble gas through the megasonic cleaner while it is in a cleaning mode. Finally, it is believed that in sequential operation, there is no need to move the silicon wafers in the tank because cavitation activity is homogeneously distributed over the entire cleaning region. Again, results indicate a different scenario [16]. When the cleaning region was examined during sequential operation, the transducers generated a band of cavitation activity, with regions between the transducers generating very little cavitation. This explains the existence of uncleaned bands on wafers. Megasonic cleaning, although in widespread use in the semiconductor industry [38], continues to be viewed warily because of damage [39] from the inconsistent surging or fountain effect of traditional megasonic systems (Fig. 2.23). This phenomenon is experienced when one of a group of megasonic lead zirconate titanate (PZT) transducers is set at the resonant frequency of the generator, or almost on the resonant point. It is not uncommon for the liquid to rise 5 centimeters above the liquid level in this surge zone. This surge in power is predictable because the frequency of the generator is set at the average of the frequencies of the PZT array. If the power is reduced to eliminate the surge, the other PZTs underperform. Lack of uniformity and/or damage from the surge minimizes the use of megasonics in silicon wafer cleaning and processes. ‘Megasonic sweeping’ e a novel technology introduced by Megasonic Sweeping Inc. (Trenton, New Jersey, USA) e eliminates traditional power surges and establishes uniform activity in a megasonic cleaning vessel (Figs 2.24 and 2.25). Uniform cleaning and rinsing without damage is possible by sweeping the array of megasonic PZT transducers. This process allocates exactly the same amount of time to each PZT, and each PZT operates at its optimum resonant frequency. The cleanability and erodibility of an acoustic field may be evaluated via the multiple-extraction procedure outlined previously [20]. Alternatively, cleaning
Chapter | 2
Megasonic Cleaning
FIGURE 2.23 Traditional non-sweeping megasonic power distribution in a cleaning vessel
FIGURE 2.24
Sweep megasonic power distribution in a cleaning vessel
53
54
Developments in Surface Contamination and Cleaning
FIGURE 2.25 470 kHz tank with sweeping (courtesy of Crest Ultrasonics Corporation, USA)
efficiency may be evaluated by quantifying the number of particles on a surface prior to and after a cleaning step. For this study [40], two trans-sonic systems (i.e. operating at frequencies below 1 MHz, but exhibiting all mechanistic characteristics of megasonic fields) were used, one operating at 470 kHz and the other at 430 kHz. The 470 kHz system, supplied by Megasonic Sweeping Inc., incorporates sweeping, whereas the 430 kHz system does not. The difference in energy uniformity between the two is obvious as soon as they are switched on. The 430 kHz system only shows activity along the normal direction from the centrally placed transducers, whereas the 470 kHz system shows a very uniform level of agitation over the entire tank. Both units operate with an input power of 600 W. This qualitative comparison was later quantitatively substantiated by measuring the prevalent cavitation intensity using a ppbÒ Cavitation Meter in the two systems. The data are shown in Figure 2.26. At the center, both systems show robust intensity levels, although the 470 kHz system has energy levels nearly double those of the 430 kHz. But the contrast is vivid at the sides and
Chapter | 2
Megasonic Cleaning
55
FIGURE 2.26 Cavitation intensities in center and corners of sweeping megasonics (470 kHz from MSI, Inc., USA) and traditional megasonics (430 kHz)
corners. The 430 kHz system has negligible activity at these off-center locations, whereas the 470 kHz system, with sweep megasonics, continues to show significant energy levels. This is clearly indicative of the fact that power is distributed uniformly to the entire tank when sweep is applied, in stark contrast to conventional megasonic systems. There is less peaking of power at the center. Cleanability and erodibility studies were performed with uncontaminated silicon and glass coupons, as well as coupons intentionally contaminated with PSL particles of sizes ranging from 0.3 mm to 5 mm. Extraction was performed in purified, filtered water at room temperature. At the center, the 430 kHz system shows a ‘fountain’ effect, where the normal energy transmission is so concentrated that it causes a plume of water to shoot up. As expected, this results in 1.5e2 higher surface erodibility at the center of the 430 kHz tank for a clean (uncontaminated) substrate (Fig. 2.27). It is to be
FIGURE 2.27 Erodibility comparison at center of tank (based on turbidity measurement on extract solution)
56
Developments in Surface Contamination and Cleaning
FIGURE 2.28 Cleanability comparison at the center of the tank based on turbidity measurement of the extract solution
noted that every data point in this and the subsequent figures is a mean of at least three replicate runs, and the standard deviation in every case was less than 10e15% of the mean. Cleanability was compared at the corners using PSL-contaminated glass disks. The 470 kHz system produces a multiple-extraction curve with a significantly steeper (by about 50%) initial slope, indicating superior cleaning efficiency (Fig. 2.28). The higher asymptote in this case is indicative of higher acoustic power at the corners of the 470 kHz tank compared to the 430 kHz tank. Also of interest is the mean size of residual particles after various cleaning cycles. This parameter is compared for the two systems in Figure 2.29. It is clear that the 470 kHz system is able to remove finer particles from the surface compared to the 430 kHz system. After five extraction stages, the mean surface-residual particle size in the 470 kHz system is three times smaller than
FIGURE 2.29 Mean size of surface residual particles after multiple cleaning stages
Chapter | 2
Megasonic Cleaning
57
the 430 kHz system. This could have important ramifications in precision cleaning of microelectronic components. Another comparison of the surface cleaning effectiveness of the 430 and 470 kHz systems was performed using a Surface Particle Detector QIIIþÒ (supplied by Pentagon Technologies Inc., USA). This instrument uses air blowing and vacuum suction to entrain particles from the surface into an air stream which is then sampled using a laser particle counter. Residual particle levels on a glass disk after cleaning with the two systems are compared in Figure 2.30. It is evident that the disks cleaned by the 470 kHz sweep megasonics have fewer residual particles (by 2e3) for sizes ranging from 0.3 mm to 1 mm [40]. The presence of ‘sweep’ renders another dimension to megasonic cleaning. The most prominent limitation of megasonics, namely its non-uniformity, has been overcome, to a large extent, by the sweep feature. Instead of the acoustic power being confined to the center of the tank, it is now widely and uniformly distributed to every location in the tank, including the edges and corners. The acoustic intensity, cleanability and erodibility data reported here confirm that 470 kHz ‘swept megasonics’ is superior to conventional 430 kHz cleaning in every aspect. This is a very positive development in the context of semiconductor wafer cleaning, as it will lead to a reduction in contamination-related defects. The resultant yield improvement is projected to be quite significant.
FIGURE 2.30 Comparison of cleaning efficiency of swept (470 kHz) and traditional (430 kHz) megasonic cleaning
58
Developments in Surface Contamination and Cleaning
6. INDUSTRY CASE STUDIES Chang et al. [41] have discussed the use of 0.8e1 MHz frequency acoustic waves to enhance cleaning efficiency for post-etch polymer removal in the Cu-low-k dual damascene process. They demonstrated effective usage of megasonic cleaning without structural damage (though the average Cu root mean square (RMS) roughness, measured by an atomic force microscope (AFM), did increase from 8.5 nm to 9.5 nm at maximum input power), and yields >90%. Processing time was greatly reduced with the megasonic enhancement. Keswani et al. [42] have investigated the feasibility of removal of particles from silicon wafers in electrolyte solutions of different ionic strengths irradiated with megasonic waves, using KCl as model electrolyte and silica as model particles. They have measured the effect of ionic strength on acoustic pressure in solutions using a hydrophone, and found that sound wave pressure amplitude can be increased in electrolyte solutions of ionic strength greater than 0.01 M. The key result from this work was that the removal of particles can be achieved at lower acoustic power densities through the use of simple electrolyte solutions. Kim et al. [43] have developed a megasonic system for nano-pattern cleaning that does not cause damage. An L-type (named for the shape) waveguide made from quartz was found to give best cleaning results. The maximum values and standard deviations of acoustic pressure were decreased by 17% and 14%, respectively. They conclude, therefore, that the L-type would have higher particle removal efficiency and would be less likely to cause pattern damage.
FIGURE 2.31 Defect number after various cleaning processes [36]
Chapter | 2
Megasonic Cleaning
FIGURE 2.32
59
Particle residue by type after various cleaning processes [36]
Huang et al. [44] have compared various cleaning processes used to remove post-CMP (chemical mechanical polishing) residue in hard disk substrate manufacturing. The results are summarized in Figure 2.31. Megasonic cleaning is clearly superior to ultrasonics and scrubbing in defect reduction. This is reinforced for various particle types in Figure 2.32. The authors conclude that brush scrubbing can remove 99% of the contamination, but megasonics is needed to remove sub-mm particles.
7. CONCLUDING REMARKS The theory, experimentation and practice of megasonic cleaning lead to the conclusion that it has inherent advantages vis-a`-vis ultrasonic cleaning in minimization of surface damage and ability to remove fine particles. However, the technique does have its limitations, which must be clearly understood by the practitioner. In particular, the uni-directionality of megasonic fields is a major concern. This has now been addressed effectively by sweep-megasonics. Another issue with megasonic cleaning is its inability to dislodge strongly adhered particles from surfaces. While this is yet to be fully resolved, precleaning of substrates at a lower trans-sonic frequency of 400e600 kHz may be a good strategy to leverage megasonics primarily as a rinsing mechanism for loosened particles on the surface. Unlike ultrasonics which can function very effectively as a ‘physical’ cleaner, with water and surfactant only, a megasonic cleaner typically relies upon strong chemistry to optimize cleaning. From an environmental point of view, this is not entirely desirable, but it must be borne in mind that deploying megasonics very likely results in a reduction in chemical usage and in processing time, both highly desirable outcomes. If ultrasonics is the ‘workhorse’ of the parts cleaning industry, megasonics performs that role in precision cleaning. With further study and optimization, it has the potential to extend its applicability to the nano-regime and below.
60
Developments in Surface Contamination and Cleaning
REFERENCES [1] D.W. Cooper, Particulate Contamination and Microelectronics Manufacturing: An Introduction, Aerosol Sci. Technol. 5 (1986) 287. [2] V.B. Menon, Particle Adhesion to Surfaces: Theory of Cleaning, in: R.P. Donovan (Ed.), Particle Control for Semiconductor Manufacturing, Marcel Dekker, New York, NY, 1990, pp. 359e382. [3] M.A. Mendicino, P.K. Vasudev, P. Maillot, C. Hoener, J. Baylis, J. Bennett, et al., Silicon-onInsulator Material Qualification for Low-Power Complementary Metal-Oxide Semiconductor Application, Thin Solid Films 270 (1995) 578. [4] R. Nagarajan, Survey of Cleaning and Cleanliness Measurement in Disk Drive Manufacture, Precision Cleaning Magazine, pp. 13e22 (February 1997). [5] L. Nebenzahl, R. Nagarajan, L. Volpe, J.S. Wong, O. Melroy, Chemical Integration and Contamination Control in Hard Disk Drive Manufacturing, J. Inst. Environ. Sci. Technol. 41 (1998) 31. [6] G.S. Selwyn, C.A. Weiss, F. Sequeda, C. Huang, In-Situ Analysis of Particle Contamination in Magnetron Sputtering Processes, Thin Solid Films 317 (1998) 85. [7] W.G. Fisher, Particle Interaction with Integrated Circuits, in: R.P. Donovan (Ed.), Particle Control for Semiconductor Manufacturing, Marcel Dekker, New York, NY, 2000, pp. 1e8. [8] S. Huth, O. Breitenstein, A. Huber, D. Dantz, U. Lambert, Localization and Detailed Investigation of Gate Oxide Integrity Defects in Silicon MOS Structures, Microelectronic Eng. 59 (2001) 109. [9] E.-S. Yoon, B. Bhushan, Effect of Particulate Concentration, Materials and Size on the Friction and Wear of a Negative-Pressure Picoslider Flying on a Laser-Textured Disk, Wear 247 (2001) 180. [10] W. Kern (Ed.), Handbook of Semiconductor Wafer Cleaning Technology: Science, Technology, and Applications, Noyes Publications, Park Ridge, NJ, 1993. [11] K.A. Reinhardt, W. Kern (Eds.), Handbook of Semiconductor Wafer Cleaning Technology: Science, Technology, and Applications, second ed., William Andrew, Norwich, NY, 2008. [12] Techsonic Website, http://www.techsonic.fr/megatheory.htm (2010). [13] S.B. Awad, Particle Removal with Ultrasonics and Megasonics, in: K.L. Mittal (Ed.), Particles on Surfaces 7: Detection, Adhesion and Removal, VSP, Utrecht, The Netherlands, 2002, pp. 341e354. [14] I.I. Kashkoush, A.A. Busnaina, F.W. Kern, R.F. Kunesh, Ultrasonic Cleaning of Surfaces: An Overview, in: K.L. Mittal (Ed.), Particles on Surfaces 3: Detection, Adhesion and Removal, Plenum Press, New York, NY, 1991, pp. 217e237. [15] D. Swanson, High Power Megasonic Cleaning in the CMP Application, CMPUG_04_2002, American Vacuum Society CMP User Group (April 2002). [16] M.A. Lester, A Glimpse into Megasonic Cleaning, Semiconductor International 26 (2003) 38. [17] Y. Wu, C. Franklin, M. Bran, B. Fraser, Acoustic property characterization of a single wafer megasonic cleaner, in: R.E. Novak, J. Ruzyllo, T. Hattori (Eds.), Cleaning Technology in Semiconductor Device Manufacturing VI, The Electrochemical Society, Pennington, NJ, 2000, pp. 360e367. [18] J.O. Kim, S. Choi, J.H. Kim, Vibroacoustic Characteristics of Ultrasonic Cleaners, Appl. Acoustics 58 (1999) 211.
Chapter | 2
Megasonic Cleaning
61
[19] D.M. Berg, T. Grimsley, P. Hammond, C.T. Sorensen, New Sonic Cleaning Technology for Particle Removal from Semiconductor Surfaces, in: K.L. Mittal (Ed.), Particles on Surfaces 2: Detection, Adhesion and Removal, 1989, pp. 307e317. [20] S. Awad, R. Nagarajan, Ultrasonic Cleaning, in: R. Kohli, K.L. Mittal (Eds.), Developments in Surface Contamination and Cleaning, Vol. 2, Elsevier, Oxford, UK, 2010, pp. 226e278. [21] R. Gouk, D.B. Kittelson, T.H. Kuehn, Y. Wul, Measurement and Modelling of Megasonic Cleaning Processes, in: K.L. Mittal (Ed.), Particles on Surfaces 5&6: Detection, Adhesion and Removal, VSP, Utrecht, The Netherlands, 1999, pp. 191e201. [22] V. Minsier, J. Proost, Shockwave upon Spherical Bubble Collapse during Cavitation-Induced Megasonic Surface Cleaning, Ultrasonics Sonochem. 15 (2008) 598. [23] K. Bakhtari, R.O. Guldiken, P. Makaram, A.A. Busnaina, J.-G. Park, Experimental and Numerical Investigation of Nano-Particle Removal Using Acoustic Streaming and the Effect of Time, J. Electrochem. Soc. 153, G846 (2006). [24] W.L. Nyborg, Acoustic Streaming, in: W.P. Mason (Ed.), Physical Acoustics II, Academic Press, New York, NY, 1965, pp. 265e331. [25] J.J. Markham, Second Order Acoustic Field: Streaming and Viscosity and Relaxation, Phys. Rev. 86 (1952) 497. [26] S. Tjotta, On Some Non-Linear Effects in Sound Fields, with Special Emphasis on the Generation of Vorticity and the Formation of Streaming Patterns, Arch. Math. Naturvidensk 55 (1959) 1. [27] C.E. Brennen, Cavitation, Bubble Dynamics, Chapter 4, Oxford University Press, Oxford, UK, 1995. [28] A. Nowicki, W. Secomski, J. Wojcik, Acoustic Streaming: Comparison of Low-Amplitude Linear Model with Streaming Velocities Measured by 32-MHz Doppler, Ultrasound in Med. & Biol. 23 (1997) 783. [29] A.A. Busnaina, G.W. Gale, I.I. Kashkoush, Ultrasonic and Megasonic Theory and Experimentation, Parts Cleaning, pp. 13e19 (April 1994). [30] P.A. Deymier, J.O. Vasseur, A. Khelif, B. Djafari-Rouhani, L. Dobrzynski, S. Raghavan, Streaming and Removal Forces due to Second-Order Sound Field During Megasonic Cleaning of Silicon Wafers, J. Appl. Phys. 88 (2000) 6821. [31] W. Kim, T.-H. Kim, J. Choi, H.Y. Kim, Mechanism of Particle Removal by Megasonic Waves, Appl. Phys. Lett. 94 (081908) (2009). [32] I. Akhatov, R. Mettin, C.D. Ohl, U. Parlitz, W. Lauterborn, Bjerknes Force Threshold for Stable Single Bubble Sonoluminescence, Phys. Rev. E 55 (1997) 3747. [33] P.M. Kanthale, P.R. Gogate, A.B. Pandit, A.M. Wilhelm, Cavity Cluster Approach for Quantification of Cavitational Intensity in Sonochemical Reactors, Ultrasonics Sonochem. 10 (2003) 181. [34] R. Nagarajan, Use of Ultrasonic Cavitation in Surface Cleaning: A Mathematical Model to Relate Cleaning Efficiency and Surface Erosion Rate, J. Inst. Environ. Sci. Technol. 49 (2006) 40. [35] T.H. Kuehn, D.B. Kittelson, Y. Wu, R. Gouk, Particle Removal from Semiconductor Surfaces by Megasonic Cleaning, J. Aerosol Sci. 27, S427 (1996). [36] T.H. Kuehn, C.H. Yang, D.B. Kittelson, Influence of Temperature and Dissolved Air on Megasonic Particle Removal, in: K.L. Mittal (Ed.), Particles on Surfaces 7: Detection, Adhesion and Removal, VSP, Utrecht, The Netherlands, 2002, pp. 355e376. [37] G. Vereecke, E. Parton, F. Holsteyns, K. Xu, R. Vos, P.W. Mertens, M.O. Schmidt, T. Bauer, Investigating the Role of Gas Cavitation in Megasonic Nanoparticle Removal, Microcontamination 22 (2004) 57.
62
Developments in Surface Contamination and Cleaning
[38] S.L. Wicks, M.S. Lucey, J.J. Rosato, Effects of Megasonics Coupled with SC-1 Process Parameters on Particle Removal on 300-mm Silicon Wafers, in: K.L. Mittal (Ed.), Particles on Surfaces 8: Detection, Adhesion and Removal, VSP, Utrecht, The Netherlands, 2003, pp. 315e322. [39] R. Nagarajan, Cavitation Erosion of Substrates in Disk Drive Component Cleaning: An Exploratory Study, Wear 152 (1992) 75. [40] M. Goodson, R. Nagarajan, Megasonic Sweeping and Silicon Wafer Cleaning, Solid State Phenomena 145e146 (2009) 27. [41] C.K. Chang, T.H. Foo, M. Mukherjee-Roy, V.N. Bliznetov, H.Y. Li, Enhancing the Efficiency of Post-Etch Polymer Removal Using Megasonic Wet Clean for 0.13-mm Damascene Interconnect Process, Thin Solid Films 462e463 (2004) 292. [42] M. Keswani, S. Raghavan, P. Deymier, S. Verhaverbeke, Megasonic Cleaning of Wafers in Electrolyte Solutions: Possible Role of Electro-Acoustic and Cavitation Effects, Microelectronic Eng. 86 (2009) 132. [43] H. Kim, Y. Lee, E. Lim, Design and Fabrication of an L-Type Waveguide Megasonic System for Cleaning of Nano-Scale Patterns, Current Appl. Phys. 9, e189 (2009). [44] Y. Huang, X. Lu, G. Pan, B. Lee, J. Luo, Particles Detection and Analysis of Hard Disk Substrate after Cleaning of Post Chemical Mechanical Polishing, Appl. Surf. Sci. 255 (2009) 9100.
Chapter 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles and Film Contamination M.D. Murthy Peri,1 Ivin Varghese2 and Cetin Cetinkaya3 Surface Conditioning Division, FSI International Inc., 3455 Lyman Blvd, Chaska, MN 55318, USA, 2Eco-Snow Systems, Rave N.P., Inc., 4935A Southfront Road, Livermore, CA 94551, USA, 3 Department of Mechanical and Aeronautical Engineering, Center for Advanced Materials Processing, Wallace H. Coulter School of Engineering, Clarkson University, Potsdam, NY 13699-5725, USA 1
Chapter Outline
1. Introduction 63 2. Dry Particle Removal 65 Technique Requirements 3. Laser Cleaning Techniques 67
4. Future Directions in Laser Particle Removal Research 5. Conclusions and Remarks References
113 114 117
1. INTRODUCTION A technique for generation of extremely monochromatic radiation in the infra-red optical region of the spectrum using potassium vapor as the active medium was proposed by Schawlow and Townes [1]. Javan [2] and Sanders [3] discussed proposals involving electron-excited gaseous systems. Finally in 1960, Theodore Maiman successfully applied an optical pumping technique to a fluorescent solid resulting in the attainment of negative temperatures and ˚ utilizing ruby as the stimulated optical emission at a wavelength of 6943 A active material, thus demonstrating the first working laser [4,5]. Since then lasers have been applied to several key technological applications, such as micromachining, materials processing, welding, non-destructive testing, cutting, precision metrology, and medical treatments [6]. Due to some critical limitations of conventional cleaning techniques, especially in Developments in Surface Contamination and Cleaning. Copyright Ó 2011 Elsevier Inc. All rights reserved.
63
64
Developments in Surface Contamination and Cleaning
specialized applications such as selective removal of sub-100 nm particles from delicate surfaces, the use of lasers has been extended to particle removal in the semiconductor manufacturing, microelectromechanical systems (MEMS), optics, photonics, and other industries. This chapter is mainly focused on the laser particle removal applications in the semiconductor industry. As the minimum size of the particle that has to be removed from wafers and other starting materials, and the minimum defect size that can be tolerated on substrates have been decreasing with each technological node in semiconductor manufacturing, the critical diameter of the particle that can be tolerated has shrunk to sub-100 nm levels. According to SEMATECH’s 2009 International Technology Roadmap for Semiconductors (ITRS 2009) [7], for front end processes (FEP), the starting materials technology requirements based on front surface particle size in polystyrene latex (PSL) sphere equivalent (A) that needs to be removed is 45 nm (application technology is known) for 2010, 32 nm (application technology is unknown) for 2013 and 22 nm (technology unknown) for 2016 [7]. This implies that for wafer cleaning of sub-32 nm particles an effective technology is unknown. For lithography, the optical mask requirements based on defect size (N) is 36 nm (technology optimized) for 2010, 32 nm (technology known) for 2011 and 29 nm (technology unknown) for 2012 [7]. This implies that there is no effective technology for mask cleaning of sub-29 nm particles. For Extreme Ultraviolet Lithography (EUVL) masks the defect sizes (I) are the same as those for optical masks, and the substrate defect size (L) is 39 nm for 2010, for which it is noted that there exists known technology for this challenge, and 32 nm for 2014 (technology unknown) [7]. This implies that for EUVL masks there is no effective technology for cleaning of sub-32 nm particles. In the case of particles with nanometer-scale characteristic diameters, the intermolecular forces dominate many other forces, especially those that are proportional to the volume (e.g. inertia and gravity) and surface (e.g. hydrodynamic and electrostatic forces) of the particle. This is because the adhesion force is proportional to the diameter (d) of the particle, whereas, the other forces are typically proportional to volume (d3) or area (d2) of the particle, and eventually, as d decreases the other forces diminish faster than the adhesion force due to geometry. As the feature size in nano-manufacturing is continuing to shrink and the number of particles on a surface to be removed is decreasing, the re-deposition-free particle removal requirements are inevitably becoming more stringent as yield remains a critical concern in nano-manufacturing, such as in the semiconductor industry. Consequently, parallel to Moore’s ‘law’ of increasing computational power, in recent years, a strong need for novel sub-100 nm particle removal techniques has emerged with this trend for mechanically delicate structures. Advances in nanotechnology products have also been driving nano-particle removal research and development.
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
65
2. DRY PARTICLE REMOVAL TECHNIQUE REQUIREMENTS The most common cleaning techniques [8,9] employed in various hightechnology industries include brush scrubbing [10e12], ultrasonic and megasonic cleaning [13e15], centrifugal spray cleaning, vapor phase cleaning [16], fluid jet cleaning, and cryogenic cleaning [17e20]. These techniques are often effectively employed for removal of large numbers of micro/nano-scale particles when particle re-deposition is not an issue. The efficiency of the brush scrubbing process is dependent on the mechanical contact of the particle and the brush whiskers, thus generally making it unviable for nanoparticle removal [21,22]. In wet cleaning techniques, the fluid used should overcome the viscous effects and capillary force to generate enough local flow rates to remove the nanoparticles, and ultrasonic excitation of the fluid to increase the efficiency of the process might lead to cavitation and subsequent damage [23]. The wet cleaning techniques have to be followed by a heating process to dry the surface, which is known to result in damage and stains of undesired chemicals on the surfaces [21,24]. In sub-100 nm range, the force required to remove nanoparticles with the conventional techniques could pose a realistic risk of substrate damage for films and multilayers as required removal forces per unit area have become high compared to yield strength of materials and features involved. Over the past two decades or so, several laser-based dry non-contact techniques have been introduced for cleaning contaminants from substrates. A summary of the current techniques used in the industry is detailed in Table 3.1 and the effects of interaction forces are summarized in Table 3.2. The major concerns of the current wet cleaning techniques with respect to semiconductor substrates are three-fold: (i) material loss; (ii) contamination issues due to chemical residues (e.g. haze defects); and (iii) possibility of damage to features present elsewhere from the particle defects on the substrate, since conventional cleaning processes target the entire substrate; in other words, precision cleaning is not possible. More specifically, for example, the wet cleaning techniques cannot be used in some of the BEOL (Back End of the Line) processes as the metal surface exposed to the wet chemicals could cause undesired corrosion, resulting in change of electrical properties of materials and devices. Consequently, dry, chemical-free cleaning techniques such as laser cleaning and cryogenic cleaning would have an edge over conventional wet cleaning techniques in many aspects when: (i) material loss is a concern and high particle removal efficiency is desired; (ii) local/precision cleaning of micrometer/sub-micrometer and sub-100 nm particles (classified as surface and small particles respectively), in crucial processes is desired; (iii) applications involve chemically sensitive features; and (iv) cleaning of optical and photonic substrates (e.g. photomasks) is desired since the chemical residue/haze could change the optical properties of the substrate. Additionally, in the near term the industry projects the utilization of the EUVL technology for the feature sizes 22 nm and below. At this length scale, any contaminant particles adhered to the
Cleaning Method
Advantages
Wet-chemical etching l Centrifugal Spray Cleaning l Ultrasonic (18 to 250 kHz) (cavitation-based) l Megasonic (0.8 to 4.0 MHz)
l l l
Particle size range for removal: 10 to 0.3 mm in diameter High removal efficiency Mature technology
Disadvantages and Limitations l l l l l l
l l l
More effective than wet chemical etching for advanced devices Dry Ongoing developments
l l l l l
Mechanical cleaning l Brush scrubbing l Fluid jet (rinsing, spinning) l Cryogenic cleaning
l l
Very effective for particles up to 1 mm in diameter Mature technology
l l l l
Laser cleaning l Steam laser l Dry laser
l l l l l l l
Sub-mm particle removal (up to 0.1 mm) Dry process Rapid process Noncontact Easy to integrate in process line (cluster tools) No chemicals required Environmental benefits: water conservation and less chemical pollutant generation
l l l l
Difficulties in integration into the process line (cluster tools) Blocking the high aspect ratio surface features Particle control and prevention of re-attachment High cost of chemicals used Expensive chemicals disposal Environmental pollution Low cleaning efficiency Complicated process Waste disposal High cost of high-purity gas Safety issues High maintenance requirements for brushes High potential damage to surfaces Dislodging and dispersing particles due to bubble formation Lower efficiency for sub-mm particles Expensive High initial cost Surface damage possible Still under development, not optimized
Developments in Surface Contamination and Cleaning
Vapor-phase cleaning
66
TABLE 3.1 A Summary of Leading Industrial Cleaning Techniques and Comparison of their Main Characteristics, Advantages and Disadvantages
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
67
TABLE 3.2 A General Grouping of Adhesion Forces for Their Interaction Ranges and Effects Group I
Group II
Group III
Range
Long-range interactions
Short-range interactions
Very-short-range interactions
Interaction
Solidesolid
Solidesolid, solideliquid
Solidesolid, solideliquid
Effects
l l
Electrostatic forces Magnetic forces
l l
l l l
van der Waals forces Sintering effects (diffusion and condensation) Diffusive mixing Mutual dissolution and alloying Capillary forces
l l
Chemical bonds Intermediate bonds
photomask during exposure would result in defects in the masks and, subsequently, on chips on the wafer. As a result, there has been an increasing necessity for using a cleaning technique that is ‘in-tool’ during, or immediately before, the start of the exposure of the next-generation lithography photomasks. Conventional cleaning techniques are often unsuitable for the ‘in-tool’ applications. In this respect, the laser-based technique is considered as a potential ‘in-tool’ cleaning technique [13].
3. LASER CLEANING TECHNIQUES In this class of particle removal techniques, a pulsed laser source is used to generate the particle removal force. Depending on their removal mechanisms, the most common laser-based cleaning techniques are (i) dry laser cleaning (DLC), (ii) wet/steam laser cleaning (WLC), (iii) laser induced plasma (LIP) cleaning, and (iv) advanced laser cleaning (ALC). The effectiveness of these laser-based techniques for removal of sub-100 nm particles has been reported in the literature. For example, the removal of 60 nm PSL particles from silicon (Si) wafers in a damage-free manner using the Laser Induced Plasma is reported in [25]. Some breakthroughs in 40e50 nm particle removal were listed in the literature review in chapter 3 in [26]. The WLC technique was utilized for removal of 51 nm PSL particles from Si wafers submerged in a liquid medium consisting of a mixture of water and ethanol near the meniscus creating a ‘mini-tsunami’ effect [27,28]. The two uncommon laser-based cleaning techniques that are a subset of the DLC technique are Matrix Laser Cleaning and Plasmon Resonance Technique. Matrix Laser
68
Developments in Surface Contamination and Cleaning
Cleaning reported in [29] was successful in removing PSL particles down to 50 nm from Si wafers in a damage-free manner, by using an additional layer of solid CO2 (quench condensed CO2) deposited on the Si surface prior to the laser pulse [26]. Plasmon Resonance Technique [30] was successful in removing 40 nm gold particles from Si wafers by localized excitation of the surface plasmon by incident radiation with a wavelength of 532 nm. The most recent breakthrough for removal of 10e40 nm PSL particles from Si wafers was reported in [31]. The main particle detachment modes are rolling, sliding, and lifting, or a combination of these modes [32].
3.1. Dry Laser Cleaning Method The DLC approach was the first method employed to remove particles from substrates using lasers. Various academic and industrial research groups have reported the experimental, characterization and computational results of this cleaning method in the past [33e46]. In the dry laser method of cleaning, a short-pulsed laser beam is directed on the substrate that has to be cleaned. This laser pulse excitation on the substrate results in rapid thermal expansion and thermomechanical wave propagation and out-of-plane acceleration, thereby exciting the substrate and/or the particles that have to be removed. This high-frequency (nanosecond) acceleration and wave propagation phenomenon generates an inertial force that can shake off the particle adhered to the substrate, provided the generated inertial force exceeds the total adhesion force consisting of several individual forces, such as van der Waals, electrostatic, and capillary forces. The main principle of this mode of sub-micrometer particle removal is the substrate acceleration induced by the thermoelastic field generated by the irradiation of the short-pulsed laser. The substrate attains its maximum acceleration value due to the substrate thermal expansion during the irradiation of the short-pulsed laser in the out-of-plane direction until the peak fluence of the beam is reached. The resulting positive acceleration presses the particle down and increases the contact diameter and the strain energy stored in the deformed particle. Removal can occur only after the surface begins to decelerate. The magnitude of surface acceleration is proportional to the level of fluence due to linearity assumption of thermoelastic effects and is inversely proportional to the duration of the laser pulse squared. However, above a certain level of laser fluence, thermal and/or mechanical damage on the surface could occur [46]. It is also noteworthy that during this process the surface is subjected to high levels of electromagnetic radiation as well. In the dry laser cleaning technique, to understand the modes of substrate damage, both thermal and mechanical (stress) damage thresholds must be well understood and accurately modeled. A key complication in such analysis is that the material properties vary substantially with strain rates and temperature changes, and it is often not well understood how certain materials yield under high thermoelastic strain rate excitations. These thresholds could then be used to
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
69
avoid excessive heat deposition and/or stress levels. The complexity of the thermoelastic process requires that a detailed analysis be carried out for the determination of optimal removal efficiencies. The onset of the material damage due to thermal and mechanical fields in this technique was modeled for the first time in [46], for silicon particles on copper and silicon substrates. It was reported that for the DLC method the required laser fluence for removing a particular size particle depends on the coefficient of reflectivity of the substrate and the particle. Based on the computational model, the critical limit of DLC was determined, in terms of the minimum diameter of the silicon particles that could be removed from silicon and copper substrates without any damage. Our research group at the Photo-Acoustics Research Laboratory at Clarkson University has reported that the type of damage initiation (thermal or mechanical fracture) depends on the particle size that has to be removed and particle-substrate system material properties, and the damage risk for sub-100 nm particle removal with DLC is found to be very high [46]. The thermal damage threshold is the melting temperature, and mechanical damage occurs above the yield stress of the substrate material. Thus, the thermo-elastic simulations from [46] indicate that when silicon particles smaller than 600 nm are removed from a silicon substrate, and when silicon particles smaller than 630 nm are removed from copper substrate using DLC, it would result in thermal and mechanical damage, respectively. It should be noted that the only effects considered in the reported study were linear thermal and mechanical fields, and no other types of damage such as optical and electromagnetic damage mechanisms were considered. Delicate structures on substrates and processing techniques used to machine them sometimes substantially lower the local material yield properties, and such structures can amplify the laser beam on the textured surface due to diffraction, thus their cleaning requires substantially greater care than flat substrates.
3.2. Liquid-Based Laser Cleaning In the liquid-based laser cleaning technique and its several variants, a liquid medium is used to enhance forces applied to the particle and thus improve laser cleaning efficiency. Key effects of this technique include, but are not limited to, the high-speed ablation of the liquid film, explosive boiling of the liquid film, and utilization of the higher density of the liquid film (e.g. the density of water is three orders of magnitude higher than that of air). Laser irradiation of a thin liquid film present on the substrate surface, resulting in explosive vaporization of the liquid layer that provides the momentum transfer for particle removal, is the principle behind steam laser cleaning (SLC) [25,27,35,47e54]. For example, as reported in [35], particle removal was achieved by the explosive vaporization of water (transparent liquid required) with a 248 nm pulsed laser radiation. Gold particles as small as 200 nm were effectively removed, by depositing a liquid film of thickness on the order of a micrometer on the surface, and irradiating the surface with a pulsed laser at a wavelength that is
70
Developments in Surface Contamination and Cleaning
strongly absorbed by the surface and short pulse duration to avoid substrate damage [33]. It was also reported that alumina particles of 100 nm diameter were efficiently removed from a silicon surface with KrF excimer laser radiation of 16 ns pulse length and 120 mJ/cm2 energy density [49]. The method utilizes the process of bubble nucleation in the liquid, and can remove at least 100-nm-sized particles from a solid surface [30,35,50,54]. By depositing a small deionized (DI) water droplet (approximately 100 mm diameter) on the Si wafer sample, 404 nm PSL particles were removed by using laser-induced plasma (LIP) generated 1.4 mm above the substrate surface [55]. This liquidbased laser cleaning technique is different from SLC and WLC because in this technique the plasma is not created in the liquid film, thus, in principle, no cavitation bubble formation or ablation is observed.
3.3. Laser-Induced Plasma Cleaning A laser cleaning method introduced in recent years utilizes shockwaves generated using the supersonic expansion of an LIP core [13,25,31,55e73]. Unlike the DLC technique, in the LIP cleaning technique the direct interaction of the laser beam with the substrate is avoided. Here we provide a detailed description of the technique, blast wave/supersonic expansion theory assumptions, particle removal experiments, damage effects, characterization of the LIP cleaning technique, and the recent advancements in this technique. In the LIP particle cleaning approach, a short pulse (typically in the range of only a few nanoseconds) high-energy infrared laser beam (e.g. at 1064 nm) is focused to a spot in air using a convex lens as depicted in Figure 3.1. LIP is formed above a silicon wafer surface, as shown in Figure 3.1c. This focusing results in the initiation of dielectric breakdown of air at the focal point due to the steep elevation in electromagnetic energy density and, consequently, the local temperature. The breakdown of air leads to formation of a rapidly expanding plasma core. The size of the plasma core expands initially to certain extent and then tends to saturate. As the supersonic plasma expansion saturates, the compressed air surrounding the plasma core emerges as a strong shockwave front. The growth and decay of plasma and the shockwave are illustrated in Figure 3.2 by the shadowgraphs acquired using high-speed photography by Villagra´n-Muniz et al. [74]. In the LIP cleaning technique, the propagating shockwave front is directed onto the substrate to exert a transient pressure field and if this exerted force is greater than the critical adhesion force between the particle and the substrate, the particle will be moved and/or detached. For damage-free removal, the interaction of the plasma core with the substrate is avoided, or at least minimized.
3.3.1. Propagation of LIP Blast Wave Laser-induced breakdown in air, first discovered in 1963 [75], is realized in four successive stages [76]: (i) multi-photon collision with the gas molecules
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
71
FIGURE 3.1 Laser-induced plasma (LIP) nanoparticle removal technique: (a) schematic (not to scale); (b) photograph of the experimental set-up showing the LIP being measured by a pressure transducer (inset); (c) LIP formed above a silicon wafer
72
Developments in Surface Contamination and Cleaning
FIGURE 3.2 A sequence of shadowgrams for the evolution of laser-generated plasma, shock wavefront and hot air core. The original height and width of each frame are 27 mm. The shock wavefronts are labeled as SW. Courtesy of M. Villagra´n-Muniz et al. [74]
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
73
resulting in the initial release of electrons; (ii) the cascade release of electrons initiates ionization of the gases in the focal region and plasma formation; (iii) absorption of laser energy by the opaque plasma leading to formation of the blast wave; and (iv) blast wave propagation into the surrounding gas. Taylor [77] and von Neumann [78] proposed similarity laws to accurately describe the initial motion and pressure of the blast wave during its propagation when it remains strong. Taylor’s similarity laws were independently derived by Sedov [79]. The key assumptions of the blast wave theory are the following. 1. The explosion results in sudden release of a certain amount of energy (E) concentrated at a point. 2. The pressures and velocities produced in the resultant flow fully dominate the initial pressure and sound speed of the ambient air with density r0. The energy E from the explosion and the density r0 of the ambient air are the dimensional parameters that will be utilized to derive the blast wave theory relations [77,79].
3.3.2. Blast Wave Theory Relations Modeled as waves generated by a point explosion, the LIP shockwave is expected to follow the blast wave theory (BWT). The LIP shockwave radius, velocity, and pressure can be approximated by utilizing the BWT relations. Similarity solutions describe the propagation of blast waves. The change of shockwave front radius R (see Fig. 3.3) over time is given by eq. (3.1) [77,79]: 1 E 52 RðtÞ ¼ k t5 (3.1) r0 Assuming the pressure and thermal field are uniformly distributed within the LIP plume, the time-dependent pressure (P) and velocity (u) fields behind the shock are given by eqs (3.2) and (3.3) [80]: 2 8 k 2 r0 E 5 6 t 5 (3.2) PðtÞ ¼ 25 g þ 1 r0 1 4 k E 5 3 t 5 (3.3) uðtÞ ¼ 5 g þ 1 r0 Here R is the shockwave radius, t is the shockwave propagation time, E is the net released energy, r0 is the density of the ambient air, k is a calibration constant, and g is the adiabatic index of air. An experimental comparison of shockwave propagation (radius of the shockwave as a function of time) was reported in [81], as depicted in Figure 3.4. The radius of the shockwave was extracted from the shadowgraphs of the LIP obtained by Villagra´n-Muniz
74
Developments in Surface Contamination and Cleaning
FIGURE 3.3 The incident spherical shockwave of radius R (t) originated and reflected from the substrate (not to scale). P, T and r are the pressure, temperature and the density, while the indices 1, 2, and 5 correspond to properties of the ambient air, behind the incident shockwave and behind the reflected shockwave, respectively
et al. [74], Dors and Parigger [81], and Jiang et al. [82], and was compared with the BWT [77]. Further, the arrival time data from experimental LIP transient pressure measurements at different firing distances was utilized to extract the shockwave propagation [72] for comparison in Figure 3.4. It is evident from Figure 3.4 that the curve obtained in [72] deviates from the BWT approximately after the first 10 microseconds and tends to follow the shadowgraphs in [82]. The shockwave generated typically has a velocity of approximately 1000 m/s during the first few microseconds and decays nearly to 700 m/s at 10 microseconds. Thus, the shockwave generated has a high momentum which could be used to break the adhesion bond between the particle and the substrate. Particle removal also depends on the distance between the plasma core and the substrate which governs the distance the shockwave front has to propagate (as the shockwave front momentum reduces with time as it propagates) and the location of the particle from the point of impact of the shockwave on the substrate. To minimize the risk of radiation damage, it is also necessary to avoid direct interaction of the plasma core with the substrate. The LIP removal technique is mainly employed to selectively remove particles that are in the sub-100 nm range since the LIP core can be generated at an arbitrary point above the substrate surface with the help of an optical beam steering system. The curvature of the shockwave front is at the millimeter scale while the particles to be removed are in sub-micrometer to nanometer size
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
75
FIGURE 3.4 Comparison of experimental shock data reported in the literature with the Blast Wave Theory (BWT). The curve representing LIP experiment represents the arrival time data extracted from the pressure measurements carried out in the present study. The plot depicts the expansion of the shockwave radius (r) as a function of a time (t)
range. Considering at least three orders of magnitude difference in the size, the effects of curvature of the shockwave can be neglected when the shockwave front interacts with particles. Thus, it is appropriate to neglect the curvature effects of the shockwave and consider the shockwaveeparticle interaction due to a plane wave impingement. Based on this assumption, the LIP shockwave loading can be modeled using the one-dimensional relations for normal incident shocks (eqs 3.4 and 3.5). The temperature behind the shock (T2), based on gas dynamics pressureetemperature relationships [83] of eq. (3.4), gþ1 P2 g1 þ P1 T2 ¼ (3.4) gþ1 T1 þ P1 g1
P2
is determined by utilizing the ambient conditions 101.3 kPa for pressure P1 and 298 K for temperature T1, and the pressure behind the shock (P2) based on the BWT is calculated from eq. (3.2). The pressure behind the reflected shockwave (P5) based on normal shock assumption, as shown in Figure 3.3, can thus be described by eq. (3.5): ! 2g P5 P1 g1 h i ¼ 1þ (3.5) P2 P1 1 þ gþ1 P1 g1
P2
76
Developments in Surface Contamination and Cleaning
The corresponding temperature behind the reflected shockwave (T5) is obtained from gas dynamic pressureetemperature relationships [80] and is given as: gþ1 P5 g1 þ P2 T5 ¼ (3.6) gþ1 P2 T2 g1 þ P5
3.3.3. Nanoparticle Detachment with LIP Shockwave A schematic of the experimental set-up for LIP particle removal is depicted in Figure 3.1a. The shock propagation in air and the reflection of the shock front at the substrate surface are shown in Figure 3.1b. A proper understanding of the action of the shock field on a nanoparticle is essential to efficiently utilize the LIP-induced force for effective removal of particles. The particle detachment and removal can be explained by the following considerations: (i) moment balance criterion, and (ii) rocking motion of the particle when subjected to a dynamic pressure/force field. 3.3.3.1. Moment Balance Criterion The detachment force of adhesion (FA) between a spherical particle and a substrate is defined by the Johnson-Kendall-Roberts (JKR) model as: FA ¼
3 pWA D 4
(3.7)
Here WA is the work of adhesion between a spherical particle with diameter D and the substrate [84], assuming that the particle is relatively soft and the force of adhesion between the particle and substrate is predominantly of van der Waals type at static equilibrium. The work of adhesion between a PSL particle and silicon substrate is given as WA ¼ 23:5 mJ=m2 . When a pressure field (P) (due to the shockwave) acting on a nanoparticle is known, the forces exerted on the particle can be determined (Fig. 3.5) [68]. The shockwave front could initiate rolling and/or sliding of particles if the associated critical pressure magnitudes due to the shockwave exceed the force of adhesion due to the substrate. As the force required to roll a particle is less than for sliding and for lifting, assuming that the particle removal takes place due to rolling mode, simple moment balance at the point O (Fig. 3.6) provides an approximate relation for critical pressure Pc required for detachment of a particle in rolling mode: Pc ¼
2aðFA þ mgÞ As ðDjcos qj 2a sin qÞ
(3.8)
Here As is the effective area normal to the applied LIP pressure, m is the mass of the particle, g is the acceleration due to gravity (mg is negligible in submicrometer length-scale), and q is the angle between the applied force and the
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
77
FIGURE 3.5 Elementary particle removal mechanism based on moment balance criterion. a0 in the figure denotes the contact depth, and r, q and z represent the cylindrical coordinate axes
plane parallel to the substrate surface. The contact radius (a) between the spherical particle and the substrate surface is determined by considering elasticity as: 1=3 3pWA D2 a ¼ (3.9) 8K ð1y2 Þ
ð1y2 Þ
where K ¼ 43½ E1 1 þ E2 2 1 and n1, E1, n2 and E2 are the Poisson’s ratios and the Young’s moduli of the particle and substrate materials, respectively.
FIGURE 3.6 Schematic of nanoparticle detachment mechanism based on rocking motion criterion
78
Developments in Surface Contamination and Cleaning
The critical pressure for a 60 nm PSL particle on a silicon substrate is calculated to be approximately 45.35 kPa when q approaches 0 or p. 3.3.3.2. Rocking Motion of Particle Subjected to a Transient Force Field An alternative approach for determining this critical pressure level for removal under transient loading conditions is to take rolling resistance and vibrational motion of the particle into consideration. According to the rolling resistance moment theory, due to bond stiffness, rocking motion of the particle could be excited by a transient pressure field and, consequently, is a potential mechanism for nanoparticle detachment. When no external moment is exerted on a spherical particle, the pressure distribution for a spherical particle on a flat substrate (according to the JKR adhesion model) has to be cylindrically symmetric and hence the moment of resistance in case of symmetric pressure distribution is given by: ZZ xpðx; yÞdxdy ¼ 0 (3.10) My ¼ Here p(x, y) is the pressure distribution in the contact area [85]. Based on the formulation reported in [85], the rolling resistance moment for a particle with radius r on a flat substrate in static equilibrium as a function of the rolling angle x is approximated by: My z6pWA rx
(3.11)
Assuming that the pull-off force for a spherical particle in contact with a flat surface is given by eq. (3.7) and using the equation ::of motion of a spherical particle in free-rotational oscillation on a flat surface I q þ 6pWA xr ¼ 0; where xzrq is the shift in contact area due to the asymmetric pressure, the resonance frequency un of the rocking motion is determined by eq. (3.12) [86]: sffiffiffiffiffiffiffiffiffiffiffiffiffi 1 45 WA un ¼ 3=2 (3.12) 4 r r Here r is the mass density of the particle material. The experimental evidence for the existence of the rolling resistance moment and rocking motion has been demonstrated and reported in [86e88]. The schematic of nanoparticle removal mechanism based on rocking motion criterion is depicted in Figure 3.6. From eq. (3.12), it is evidently difficult to remove small particles, while the larger particles can be removed relatively easily with smaller critical angles (qcrit). For instance, assuming qcrit ¼ (1 e 5) 10e9/r, the rolling resistance moment required for a 100 nm PSL particle on a silicon substrate (assumed to be a cylinder with the length equal to its diameter) is given by Mcrit ¼ 22.1 e 110.7 nN-nm for WA ¼ 23.5 mJ/m2. The resonance frequencies of a 100 nm PSL particle on a silicon substrate for axial and rocking motions are
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
79
determined to be 960 MHz and 227 MHz, respectively [72]. Note that the rocking resonance frequency is much lower than the axial frequency. The pressure field exerted due to the shockwave on a particle is a statistical distribution of the gas moleculeeparticle interaction according to the kinetic gas theory. For a particle located away from the critical distance, if the fluctuations due to the gas moleculeeparticle interactions approach the resonance frequency of rocking motion of the adhesion bond given by eq. (3.12), the amplitude of rocking motion increases until the magnitude of q reaches qcrit and the particle detaches.
3.3.4. Gas MoleculeeNanoparticle Interactions To understand the particle removal mechanism in the LIP technique, a detailed investigation of shockwaveenanoparticle interaction is necessary. Due to the characteristic length-scale of interactions, such an investigation requires molecular level simulation rather than continuum models such as the NavierStokes equations. The transition between the molecular and continuum models for the flow is governed by the Knudsen number defined as: Kn ¼ l=L
(3.13)
Here l is the mean free path and L is a characteristic length of the system. A Knudsen number of 0.1 is generally considered the upper limit for the continuum approach. The Knudsen number for a 100 nm particle under shock conditions in air is estimated as Kn ¼ 0.323 and the mean free path is l ¼ 32 nm. The gas moleculeeparticle interaction was simulated using the Direct Simulation Monte Carlo (DSMC) method according to the kinetic theory of gases, as it is more appropriate in the nanoscale than the Navier-Stokes equations [89]. A computational study using the DSMC has been reported by our group in [70e72]. It was observed in these simulations that when a shockwave front arrives at the particle, the moment exerted on the particle increases rapidly until it reaches a critical maximum and then it reduces by a fraction of maximum moment, termed as mean-moment. The mean-moment exerted on the particle will be constant, until the entire shockwave passes the particle. For instance, Figure 3.7a shows the moment exerted on the particle due to a planar shockwave front obtained from the DS2V software simulations (two-dimensional DSMC) [70]. The maximum and the mean moments obtained from 2D simulations (i) for a 100 nm particle were Mmax ¼ 500 nN-nm and Mmean ¼ 393 nN-nm, [72] and (ii) for a 60 nm particle as Mmax ¼ 60.5 nNnm and Mmean ¼ 41.7 nN-nm [71]. Note that the moment exerted on the particle in 2D simulations is approximately equal to the moment exerted on a cylindrical particle with a length equal to the diameter of the particle in 3D simulations, rather than the regular sphere itself. The critical moments calculated for PSL particles with diameters of 100 nm and 60 nm on a silicon substrate using the rolling resistance moment theory are Mcrit ¼ 22.1e110.7 nN-nm and
80
Developments in Surface Contamination and Cleaning
FIGURE 3.7 Transient moment exerted on a 100 nm particle due to gas moleculeseparticle surface interactions excited by the shockwave (a), and close-up of the frequency spectrum (b) shown in (a). The dot-dash line represents the predicted rocking resonance frequency of 100 nm PSL particle on a silicon substrate
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
81
Mcrit ¼ 13.3e66.4 nN-nm. This implies that the 100 nm PSL particles and sub100 nm particles (60 nm PSL particles) can be more easily removed from the substrates by rolling mode of detachment than other methods of detachment. Furthermore, the transient/jitter part (due to the gas molecules in the shockwave front of the transient moment plot (after 2.5 ns time in Figure 3.7b) for a 100 nm particle) is extracted and the frequency spectrum of the moment plot is analyzed for the rocking frequency. From the frequency spectra it has been deduced that the gas moleculeeparticle interactions would excite the particleesubstrate bond in the rocking motion mode, indicating that this effect is a possible sub-100 nm particle removal mechanism. At low Kn, according to this removal mechanism, the gas molecules oscillating in the shockwave front would bombard the nanoparticle with a range of frequencies and, if this frequency range covers the natural (resonance) rocking frequency of the adhesion bond between the nanoparticle and the substrate, the particle would rock until it reaches critical amplitude breaking the adhesion bond depending upon the thickness of the shockwave. The expected rocking resonance frequencies for a 100 nm PSL and a 60 nm PSL particle on a silicon substrate were approximated as 227 MHz [72] and 488 MHz [71], respectively. The expected rocking frequency for 100 nm PSL particle is represented in Figure 3.7b as a dot-dash line. Thus, from the rolling mode and the rocking motion of particle, a zone for particle removal limits can be determined for a given firing distance d, i.e. the distance from the center of the plasma core to the substrate surface. A summary of the values of the particle rolling moment and the particle removal zone based on the location of the particle from the arrival of the shockwave front is reported in [70]. Thus, the critical rolling moment and the rocking frequency due to the transient pressure field were identified as the two possible mechanisms responsible for particle removal in the LIP technique.
3.3.5. LIP Nanoparticle Removal Experiments Several experiments have been performed to investigate the damage-free particle removal capability of the LIP technique [25,31,69]. The laser typically employed in these experiments was an Nd:YAG pulsed laser with fundamental wavelength of 1064 nm, repetition rate of 10 Hz, pulse width of 5 ns, a beam diameter of 5 mm, and a pulse energy of 370 mJ. Commercially available lasers have variable pulse energies and pulse widths and the selection of the laser depends on the type of application, particles, and the substrate that has to be cleaned. A convex lens of 100 mm focal length with an antireflective coating was used to converge the laser beam and this could be varied from smaller to larger focal lengths, up to a length beyond which optical aberrations might play a role in convergence of the beam to form plasma. The substrates used for the experiments were bare 125 mm, n-type doped [111] silicon wafers with approximately 1 mm thick thermal oxide layers and a bare quartz photomask substrate. Figure 3.8 shows the pre- and the post-scan images acquired using a surface analysis system (SAS) on a 152 mm wafer when
82
Developments in Surface Contamination and Cleaning
FIGURE 3.8 Surface analyzer system (SAS) analysis of the silicon substrate. (a) Before LIP exposure with the 8 8 mm cleaning grid in the inset and (b) after LIP exposure. The white square in (b) indicates the cleaning area grid
the center of the wafer was cleaned. In Figure 3.8b the cleaned area is indicated by a square. The silicon substrates (Figs 3.9e3.14) were cut into small samples of approximately 1.5 cm 1.5 cm area due to scanning electron microscopy (SEM) constraints. The sample wafer was washed with DI water and methanol to remove the initial contamination. In order to locate the test site on the sample during SEM
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
83
FIGURE 3.9 SEM images of ceria CMP slurry particles deposited on a silicon wafer. (a), (b) at 6000, (c), (d), (e) and (f) at 3000 and (g), (h), (i) and (j) at 1000 magnification are pairs of images before and after LIP exposure. Approximate size of some of the individual particles is identified on all the before images. The dash-dot lines indicate the boundaries of the markers and the cleaning area
84
Developments in Surface Contamination and Cleaning
FIGURE 3.9 (continued)
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
FIGURE 3.9 (continued)
85
86
Developments in Surface Contamination and Cleaning
FIGURE 3.9 (continued)
imaging, a diamond-shaped reference mark with an approximate area of 2 mm2 was engraved on the sample. To remove the residual particles from the sample due to this marking, the sample was subjected to the LIP particle removal technique by firing five shots at a firing distance of d ¼ 1.4 mm. The firing distance is the gap between the center of the plasma and the substrate. The particles in the experiments were ceria CMP slurry particles with a size range of 30e400 nm (Fig. 3.9), 60 nm PSL particles (Figs 3.10 and 3.11), 10e40 nm PSL particles (Figs 3.12 and 3.13), and 60 nm gold particles (Fig. 3.14). The particles on the photomask were unknown (Fig. 3.15). The ceria and PSL particle suspensions were diluted in methanol in order to acquire good distribution of the particles in the resultant suspension, and then the resultant suspension was excited to prevent agglomeration of the particles. The drop-agitation technique was employed to deposit the particle suspension on the sample. This process ensured uniform distribution of the particles, avoiding agglomeration. The gold particles were deposited on the silicon sample by adding 1 wt% HF and KAu (CN)2. After the silicon native oxide dissolves in HF, galvanic deposition of Au occurs with simultaneous Au(CN)2e reduction and silicon oxidation and dissolution, forming SiF2e 6 . The surface of the wafer sample was analyzed in a JEOL scanning electron microscope (SEM) and before LIP images of the marked area were acquired. The sample was then mounted on a micrometercontrolled xyz stage. The firing distance was controlled by the z-axis. This firing distance was set to d ¼1.4 mm to achieve good particle removal without substrate damage. The sample was adjusted in the xy directions such that the plasma was formed directly above the marked area and the adjustment was made with the aid of a diode laser. The pulsed laser was triggered for LIP removal and ten shots were fired. The sample was again analyzed using SEM to obtain after LIP images. The before and after images were compared to determine the particle removal efficiency (Figs 3.9e3.15).
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
87
FIGURE 3.10 Before LIP exposure (a) and after LIP exposure (b) SEM images at 5000 magnification. PSL particles with 60 nm diameter are identified in the before and after images. The dashed lines indicate the boundaries between the cleaning zones and location markings
88
Developments in Surface Contamination and Cleaning
FIGURE 3.11 SEM images at 5000 magnification before LIP cleaning (a, c) and after LIP cleaning (b, d) of 60 nm PSL particles at another location. The dashed lines indicate the boundaries between the cleaning zones and location markings. Removal of large agglomerates is illustrated
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
FIGURE 3.11 (continued)
89
90
Developments in Surface Contamination and Cleaning
FIGURE 3.12 SEM images of the test area pre-LIP (a) and post-LIP (b) cleaning at the marked location (dotted line) at 30 000 magnification.
It was observed that if the firing distance was too small, the plasma core would interact directly with the substrate surface and damage the substrate. This damage could be attributed to the thermal field generated in the plasma rather than the mechanical field generated by the shockwave. It should be noted that the photomask sample was a 152 152 mm square sample and the scan of the mask was obtained before and after application of LIP at a firing distance of 2 mm.
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
91
The SEM images of the substrate before and after application of LIP at different locations with 30e400 nm (at 6000 magnification) ceria particles, 60 nm PSL particles (at 5000 magnification), 10e40 nm PSL particles (at 30 000 magnification), 60 nm gold particles (at 10 000 magnification) and
FIGURE 3.13 SEM images of the test area pre-LIP (a) and post-LIP (b) cleaning PSL particles with 10e40 nm diameter at the marked location (dotted line) at 30 000 magnification. (c) and (d) are the higher-magnification images of the test area shown in images (a) and (b), respectively
92
Developments in Surface Contamination and Cleaning
FIGURE 3.13 (continued)
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
93
FIGURE 3.14 SEM images of pre-LIP (a) and post-LIP (b) exposure of the test area with gold particles of 60 nm diameter at the marked location (dashed line) at 10 000 magnification
FIGURE 3.15 Pre-LIP (a) and post-LIP (b) cleaning images of a bare photomask substrate. The line in (b) indicates an added surface particle after the pre-scan
94
Developments in Surface Contamination and Cleaning
unknown particles on the photomasks are shown in Figures 3.9 (Ceria on Silicon), 3.10, 3.11, 3.12 and 3.13 (PSL on Silicon), 3.14 (Au particles on Silicon) and 3.15 (unknown particles on photomask), respectively. These SEM images clearly indicate that the majority of the particles in the pre-scan images have been removed in the after LIP images. The particle removal results are summarized in Table 3.3.
3.3.6. Optimization of LIP Process Parameters The LIP technique has potential for damage-free nanoparticle removal from semiconductor substrates as demonstrated from the references cited above. As predicted in the DSMC simulations [71,72], the moment exerted on a 60 nm particle is lower by approximately a factor of 8.2 when compared to that of a 100 nm particle. This reduction is due to the fact that the available surface area of the particle is substantially less. Thus, for removal of particles smaller than 60 nm, the LIP technique has to be optimized for maximizing the transient pressure, the distance which the shockwave has to travel after it is detached from the plasma and before it interacts with the substrate has to be decreased, and the pulse energy of the laser has to be reduced to avoid larger plasma cores yet obtain the same shock pressure. The main factors that influence the damage threshold of the substrate are yield strength of the material, reflectivity of the material for broad wavelength of irradiation exhibited by the plasma core (especially for photomask), and the thickness of the nanofilm for each layer (for an EUVL mask) on the substrate since mismatch of thermal coefficients would result in thermomechanical damage. Thus, the three parameters to be optimized in the LIP process are: (i) the firing distance; (ii) the pulse energy; and (iii) the number of laser shots to be triggered based on the substrate from which the particles have to be removed. 3.3.6.1. Transient Pressure Measurements In order to determine the transient pressure available for particle removal due to the interaction of the shockwave at different firing distances, accurate transient TABLE 3.3 Summary of LIP Particle Removal Experiments Substrate
Particle
Size Range
Firing Distance
Number of Pulses
Figure Numbers
Silicon
Ceria Slurry
30e400 nm
1.4
10
3.9
Silicon
PSL Spheres
60 nm
1.4
10
3.10, 3.11
Silicon
PSL Spheres
10e40 nm
1.4
12
3.12, 3.13
Silicon
Gold
60 nm
1.4
10
3.14
Quartz
Unknown
50e150 nm
2
10
3.15
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
95
pressure measurements are required. In [70,90], the pressure exerted by the LIP shockwave front is obtained from experiments conducted using a 450 mJ pulsed Nd:YAG laser and a pressure transducer (Kistler, 603B1) with a central frequency of 500 kHz. The selection of the transducer frequency bandwidth was based on our previous experiments with a custom-made broadband polyvinylidene fluoride (PVDF) line transducer reported in [68]. Initially the firing distance was set to 2 mm as there could be interaction of plasma core with the transducer at lower firing distances. The laser pulse was triggered and the transient waveform obtained from the oscilloscope through a charge-to-voltage amplifier was saved. This procedure was repeated up to 15 mm at an increment of 1 mm and the transient pressure waveforms were recorded. The maximum pressure at a firing distance of 2 mm was 155 kPa and the maximum pressure at a firing distance of 15 mm was 12 kPa. The transient pressure profiles are shown in Figure 3.16. It should be noted that the pressure decreases exponentially with the firing distance. Further, the maximum pressure obtained was used as an initial condition in the reported DSMC simulations [71,72]. As mentioned earlier, the control of pulse energy is necessary for substrates which are sensitive to radiation and have high absorbance. It was determined that at the end of the laser pulse, the plasma dimensions are
FIGURE 3.16 Transient pressure (P) of the shockwave measured at the various LIP firing distances d ranging from 2e15 mm at an incremental distance of 1 mm. The highest peak for the pressure corresponds to the firing distance of 2 mm and the lowest peak to 15 mm
96
Developments in Surface Contamination and Cleaning
proportional to the fifth root of the deposited energy [73]. Laser energy can be attenuated either by electrical power supplied to the laser unit or by an optical method. The attenuator used in the reported experiments was a commercially available high-energy optical attenuator. The experimental set-up for LIP attenuation is shown in the Figure 3.17. The attenuator can be set at various dial values which results in certain level of energy attenuation, and hence can be calibrated to get specific laser energies. A medium-power volume absorber power/energy meter (used in [91]) was utilized for measuring the single shot laser pulse energy during the experiments. The medium power meter has a spectral bandwidth that ranges from 190 nm to 3 mm except in the range 625e900 nm. An attenuator setting of zero implies that there is no attenuation and the laser energy recorded on the display unit of the power meter was 315 mJ (specification on the laser was 370 mJ), as some energy was lost in the spectral hole of the power meter (625e900 nm). Measurements at various attenuator settings were recorded and are reported in Table 3.4 and in Figure 3.18. The percent reduction in laser pulse energy is also plotted in Figure 3.19 and reported in Table 3.5. Based on these attenuator experiments, the required energy levels can be obtained. For example, if there is some damage caused by LIP application at a specific firing distance (d), then the attenuator can be utilized to reduce the laser pulse energy without changing d. Thus, the laser pulse energy could be reduced by attenuation and could be optimized for the substrate that is to be cleaned.
FIGURE 3.17 Photograph (top) and instrumentation diagram (bottom) of the experimental setup for the laser attenuation experiments
Chapter | 3
97
Laser Cleaning for Removal of Nano/Micro-Scale Particles
TABLE 3.4 Maximum and Average Single-Shot Laser Pulse Energies at Certain Attenuations Single-Shot Laser Energy (mJ) Attenuator Dial Value
Average
Maximum
12.00
296
301
12.25
296
299
12.50
286
290
12.75
281
283
13.00
269
272
13.25
255
267
13.50
232
234
13.75
205
212
14.00
178
182
14.25
140
143
14.50
101
107
14.75
60
63
Single Shot Energy (mJ)
300
Average Maximum
250
200
150
100
50 12.0 FIGURE 3.18
12.5
13.0 13.5 14.0 Attenuator Dial Value
14.5
15.0
Maximum and average single shot laser pulse energies at certain attenuations
98
Developments in Surface Contamination and Cleaning
90 80
% Reduction in Energy
70 60 50 40 30 20 10 0 12.0
14.0 13.0 13.5 Attenuator Dial Value
12.5
14.5
15.0
FIGURE 3.19 Percent reduction in maximum laser pulse energies obtained for different attenuator settings
TABLE 3.5 Maximum Single-Shot Laser Pulse Energies and Corresponding Percent Reduction in Pulse Energies as Obtained for Different Attenuator Settings Percent Reduction in Energy
Attenuator Dial Value
Single-Shot Laser Energy Maximum (mJ)
0.00
315
0.0
12.00
301
4.4
12.50
290
7.9
13.00
272
13.6
13.50
234
25.7
14.00
182
42.2
14.25
143
54.6
14.50
107
66.0
14.75
63
80.0
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
99
3.3.7. Thermomechanical Analysis of LIP For any cleaning application, the highest achievable particle removal efficiency is required while assuring no damage to the sample being cleaned. Samples with different material compositions would have varied damage thresholds, thereby requiring different process conditions. The possible damage effect which LIP application could inflict has been systematically studied for Cr films on quartz substrates by experiments and computationally on (i) a photomask e 100 nm Cr film on a low CTE (coefficient of thermal expansion) quartz substrate; (ii) an EUVL mask e 2.5 nm Ru film on 105 nm Mo/Si multi-layers (MLs) (modeled as a substrate with average material properties of Mo and Si); and (iii) an EUVL mask e 2.5 nm Ru film on 280 nm Mo/Si MLs (modeled as a substrate with average material properties of Mo and Si) [90]. In the LIP application, it has been identified that the two key interdependent and critical process parameters include the firing distance (d) and the laser pulse energy (E) available for nanoparticle removal. These two parameters govern the initiation and form of damage on a substrate. If d and/or E are above their critical thresholds, substrate damage could occur. Yet the optimum values for d and E would still vary, depending on the damage threshold of the sample being subjected to LIP. Provided there is no direct contact of the LIP core with the sample surface (nanofilms/substrate surface in the case of lithography masks) the two possible sources of damage are the thermomechanical loading from the LIP shockwaves due to gas convection heating, and the thermoelastic radiation intensity heating from the plasma core. The timescale for radiation heating is in nanoseconds, while that for LIP shockwaves is in microseconds, resulting in laser radiation heating effects appearing earlier than those due to the LIP shockwaves on the sample surface [90e93]. The removal of smaller particles would require higher pressure levels attainable either at a lower d for constant E, or with higher E. It is desirable to increase the available pressure without damage risk and determine the thresholds for LIP particle removal. The pressure (P) and temperature (T) fields generated on the substrate [90e93], obtained from LIP at low firing distances d, are to be mitigated to ensure damage-free particle removal in the sub-100 nm range. An estimation of the transient temperature [91] experienced by the lithography masks due to LIP is thus needed to predict damage risk. For practical implementation of the LIP technique, a design space (a set of values for the optimized parameters) for damage-free nanoparticle removal needs to be determined. Transient LIP shockwave pressure measurements were conducted utilizing a piezoelectric transducer (Fig. 3.1b) to characterize the transient pressure field of an LIP shockwave front as reported in [94]. Optimized parameters for 60 nm PSL particle removal on silicon wafers were determined to be the firing distance of d ¼ 1.4 mm and ten laser pulses for the Nd:YAG pulsed-laser with the pulse energy of 370 mJ at the wavelength of 1064 nm [25]. Silicon wafer is a bulk material and crystalline in nature and is
100
Developments in Surface Contamination and Cleaning
mechanically stronger than the delicate nanofilms like Cr and Ru deposited on EUVL masks and photomasks. These nanofilms are generally weakly bonded to the substrates and have much larger surface area to volume ratio. Silicon wafers should, therefore, be expected to be less susceptible to material alteration and/or damage due to thermomechanical loading compared to lithography masks for given process conditions [93]. 3.3.7.1. LIP Shockwave Loading of Prototype Lithography Masks In [90e93] a set of transient waveforms was acquired to characterize the shockwave front for its pressure, while its temperature was approximated utilizing gas dynamics relations. Detailed computational investigations [90e93] have been conducted to determine the dynamic responses experienced by different prototypes due to LIP shockwave thermomechanical loading: (i) a photomask e 100 nm Cr film on a low CTE quartz substrate; (ii) an EUVL mask e 2.5 nm Ru film on 105 nm Mo/Si MLs (modeled as a substrate with average material properties of Mo and Si); and (iii) an EUVL mask e 2.5 nm Ru film on 280 nm Mo/Si MLs (modeled as a substrate with average material properties of Mo and Si). The effects of the shockwave pressure alone (mechanical loading) and shockwave temperature alone (thermal loading) were obtained. It was observed that the radial, circumferential, and shear stresses are largely due to thermal expansion (shockwave temperature) of the film, while axial stresses were generated by the mechanical loading (shockwave pressure) for all the three cases considered. The radial stress (srr), axial stress (szz), and shear stress (srz) components induced by the application of only the LIP shockwave thermomechanical loading were determined for each case: (i) for the prototype photomaske649 MPa, 195 kPa, and 172 kPa, respectively; (ii) for the prototype EUVL mask (105 nm Mo/Si) e 2.25 GPa, 305 kPa, and 20 MPa; respectively; and (iii) for the prototype EUVL mask (280 nm Mo/Si) e 595.7 MPa, 1.08 MPa, and 5.13 MPa, respectively [90,94]. The chief observation from the reported studies was that the temperature rise on the film was of no damage concern due to melting, yet the radial stress field induced in the film by this thermal field is apparently high enough to be considered as a potential source of damage. Further, it was observed that the radial and circumferential transient stress amplitudes were significantly higher than the (static) yield strengths of the film materials in each case. For a static loading case, the obtained stress levels would easily result in damage; however, the dynamic strength of materials is known to be strain ratedependent and thereby a great deal higher than their static counterparts, even though such relations are currently not readily available even for many practical materials. The dynamic yield strength for LIP application on chromium is, however, estimated to be higher than the static yield strength at least by a factor of 2 [94]. The shear stresses calculated at the interface were not of concern for material alteration for the photomask, i.e. no delamination due to interfacial shear was predicted, and the axial stress at the interface was low (comparable to
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
101
shockwave-front pressure) for the structure/model under consideration. For the EUVL mask, on the other hand, the radial and shear stress components were much higher, implying increased damage risk. Gas heating and pressure loading of the substrate due to LIP shockwaves, generated utilizing this pulsed laser, on a photomask or an EUVL mask, at a firing distance of d ¼ 2 mm for a high strain rate gain factor of w4 is not expected to be of concern for damage to the substrate [90]. 3.3.7.2. LIP Radiation Intensity Loading of Prototype Lithography Masks Radiation energy (Erad) measurements are required. Such data help approximate the magnitude of the radiation intensity required as load for conducting computational studies to determine transient thermoelastic responses of substrates. The thermoelastic responses for prototype lithography mask substrates were reported in [91e93]. It was found that for damage concerns, the radial stress component srr was the most critical stress component on the film surface for both the thermomechanical LIP shockwave excitation as well as the LIP radiation exposure. The damage threshold for LIP application on the prototype lithography mask (a 100 nm Cr film on a quartz substrate) was assumed to be at the firing distance of d ¼ 2.5 mm (as slight damage was detected at d ¼ 2 mm, but not at dcr ¼ 2.5 mm) based on experimental observations. It was observed that the Cr film surface responses (i.e. temperature rise and stress levels) due to laser radiation intensity level heating dominated the thermomechanical load due to the LIP shockwaves. Therefore, the corresponding radial stress component srr of the Cr film at very short firing distances (d < dcr) is critical for damage concern. For the Cr film, the actual dynamic yield stress is approximated to be at least four times the material yield stress sy for bulk Cr (362 MPa) due to the high strain rate experienced in the LIP radiation excitation. If the radial stress component srr exceeds the actual dynamic yield stress of Cr film and/or the surface temperature rise leads to melting, then material alteration/damage of the Cr film would occur. For 0% and 20% reflectivity of the Cr film, the dynamic yield stresses were required to be at least 5.32 sy and 4.25 sy, respectively, to prevent inception of material alteration/ damage. At the experimental damage threshold firing distance of dcr ¼ 2.5 mm, the damage limit for maximum radial stress component amplitude is srr,max ¼ 1.93 GPa for no reflectivity of the Cr film, whereas it is 1.54 GPa for the case of 20% reflectivity. A safe firing distance of d ¼ 3.6 mm (as slight damage was detected at d ¼ 3.55 mm, but none at dcr ¼ 3.6 mm) was obtained, based on experimental observations for LIP application on an EUVL mask with a 2.5 nm Ru film on 105 nm Mo/Si MLs on a quartz substrate. Radial stress (srr) and the surface temperature rise (DT) of the Ru film due to laser radiation intensity (wnine times the intensity due to LIP shockwave thermomechanical load) at very small firing distances (d < dcr) are sources of LIP damage concern. Therefore, if the surface temperature rise leads to melting and/or the radial stress (srr) exceeds the actual yield stress (dynamic) of Ru film material alteration/
102
Developments in Surface Contamination and Cleaning
damage of the Ru film would occur. The damage limit for radial stress amplitude (srr) is 21 GPa for no reflectivity case of the Ru film at the critical firing distance of 3.6 mm (dcr). Thermal loading leading to material mechanical failure is determined to be of damage concern for nanofilms. Reflectivity would decrease the film surface responses of the film. The temperature profiles obtained due to both LIP shockwave and radiation heating are lower than the melting points of Cr (photomask) and Ru (EUVL mask) nanofilms, and hence melting is not a potential damage mode in this particular case. It is concluded that the level of radial stress alone is of damage concern. As a potential solution for minimizing damage risk on these nanofilms on lithography masks, introducing residual radial tension in the film is suggested as a means to extend the damage threshold of the Cr and Ru films on photomasks and EUVL masks, respectively. Since compressive stresses are applied due to LIP excitation, this radial stress would lower the resultant stresses that will be experienced on the nanofilms, as a result, increasing the damage-free process window [90,93].
3.3.8. Onset of Material Alterations on Lithography Masks During LIP Exposure For any cleaning application, the ideal requirement is complete particle removal, with no new adders and no damage to the sample. Depending on the industrial application, different samples would have varied structures and features (e.g. trenches) that are very delicate, thus having diverse damage thresholds. It is noteworthy that, as with many semiconductor manufacturing applications, for damage-free lithography mask cleaning, the determination of the onset of material alterations and/or damage of the nanofilms deposited on lithography masks is critical, rather than the extent of damage. In [90e93], experiments were conducted to investigate the onset of material alterations in nanofilms on lithography masks (EUVL masks and photomasks) due to LIP exposure. Thermomechanical loading from LIP shockwaves, i.e. the combination of the thermal (shockwave temperature) and the mechanical (shockwave pressure) effects, as well as the thermal loading due to the laser radiation intensity (I0) heating from the LIP core, were considered for material alteration studies on the nanofilms and are described in Section 3.3.7, and reported in [90e93]. The critical firing distance dcr, i.e. the minimum safe LIP firing distance beyond which there will be no material alterations, needs to be identified for a given laser pulse energy. This critical firing distance would be different depending on the damage thresholds of various samples. Damage and/or material alterations would occur if the nanofilms are not able to sustain the induced film surface stresses and temperature rise due to the combination of radiation heating and LIP shockwave loading during LIP application on the samples. If the firing distance d < dcr for the specific sample, then the load experienced by the nanofilm on the EUVL masks and/or photomasks would result in either film surface temperature higher than the melting point or induced stresses larger than the yield and/or rupture stress,
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
103
resulting in material alterations such as melting, cracking, channeling, peeling, film stripping, and discoloration [90e93]. Investigation of the onset of material alterations on EUVL masks and photomasks due to LIP exposure, and determining experimental damage thresholds for estimating safe firing distances for damage-free nanoparticle removal were reported previously [90e93]. Possible material alterations that occur due to LIP application on these masks were determined in order to prevent onset of the least of these material alterations, and thus ensure damage-free removal. LIP was demonstrated as a damage-free sub100 nm particle removal technique for EUVL masks and photomasks. The damage threshold for LIP application on a photomask with a 100 nm Cr film on a quartz substrate was assumed to be at the firing distance of d ¼ 2.5 mm based on experimental observations. Extreme cases of intentional material alterations, such as surface cracks, channeling, peeling, film stripping, melting and discoloration, are observed in the optical microscope (Fig. 3.20) and scanning electron microscope (SEM) images (Figs 3.21aed) on the photomask when deliberately subjected to 2 minutes of continuous LIP exposure (~1200 shots) at a firing distance of d ~ 0 mm (plasma hitting the film surface) [74]. Surface cracks and channeling are the first experimentally observed material alteration modes on the Cr nanofilm (when d < dcr) at a firing distance of d ¼ 2 mm, as seen in the optical microscope images in Figure 3.22. The minimum size of PSL particle (soft particle) that can be removed from the Cr film, utilizing LIP at the safe firing distance of d ¼ 2.5 mm with the 138 kPa available shockwave pressure, is predicted as 46 nm, based on the JKR model for rolling detachment of spherical particles from a flat substrate [92]. The safe firing distance of d ¼ 3.6 mm, based on experimental observations for LIP application on EUVL mask with 2.5 nm Ru film on 105 nm of Mo/Si MLs on a quartz substrate, was obtained as described in Section 3.3.7. The PSL particle size of 59 nm is the smallest particle that can be removed with LIP at the safe firing distance of d ¼ 3.6 mm from the Ru film with the 93.46 kPa available shockwave pressure (based on the JKR model for rolling detachment of spherical particles from a flat substrate). The predicted order (increasing d or decreasing level of excitation) of material alterations in EUVL masks due to LIP exposure was found to be material melting, film/MLs stripping, cracks, peeling and discoloration (smallest material alteration). The level of cracking in photomasks is observed to be substantially higher than in EUVL masks [90]. As reported in [90], EUVL masks are determined to be weaker (larger dcr ¼ 3.6 mm) compared to the photomasks (dcr ¼ 2.5 mm) for onset of material alterations due to LIP exposure, therefore greater caution is required for LIP application on EUVL masks during nanoparticle removal.
3.4. Advanced LIP Cleaning Technique In [90] it has been identified that there is a threshold for the smallest-sized particle that can be removed in a damage-free manner utilizing the LIP
104
Developments in Surface Contamination and Cleaning
FIGURE 3.20 Optical microscope images at 100 magnification depicting extreme cases of material alterations (surface cracks, channeling, discoloration, peeling, film stripping) of 100 nm Cr film on a quartz substrate intentionally created at LIP firing distance of d w 0 mm (plasma in contact with nanofilm surface) [92]
technique in air from lithography masks. If particles smaller than this limit need to be removed, some LIP pressure amplification technique is required, since more pressure needs to be applied to break the adhesion bond for smaller-sized particles. If LIP is utilized in air at a firing distance d < dcr, material alteration/damage would occur due to radiation heating and LIP shockwaves, resulting in either surface temperatures greater than the melting point and/or induced stresses exceeding the dynamic yield stress of the material under consideration. The removal of 60 nm [31] as well as 10e40 nm [25] PSL particles from silicon substrates has been successfully demonstrated utilizing the LIP removal technique in air. The instrumentation
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
105
FIGURE 3.21 Scanning electron microscope (SEM) images at (a) 500 (b) 1000, (c) and (d) 5000 magnifications depicting extreme cases of intentional LIP generated material alterations (surface cracks, peeling, film stripping, melting and channeling) of 100 nm Cr film on a quartz substrate at firing distance of d w 0 mm (plasma in contact with nanofilm surface) [92]
diagram for the pressure measurement set-up for LIP in air is depicted in Figure 3.23a. The maximum pressure obtained from LIP in air for a 370 mJ, 1064 nm Q-switched Nd:YAG pulsed laser was 156 kPa at a firing distance d ¼ 2 mm. Damage was reported to occur on 100 nm Cr film on a quartz substrate at a firing distance d ¼ 2 mm [92]. From the analyses of the radiation heating and LIP thermomechanical shockwave loading (Section 3.3.7), it is found that the induced film surface radial stress and temperature rise are critical for damage concerns. Radiation heating from the plasma core is determined to be the more serious LIP damage source as the film surface responses dominate those due to LIP shockwave loading. As reported in [90], various pressure amplification techniques, such as LIP in a pressurized chamber, shock tubes in air, wet-LIP, and submerged shock tubes, have been investigated. Pressure calibration of LIP at a firing distance of d above a pressure transducer on a prototype lithography mask is shown in Figure 3.23b.
106
Developments in Surface Contamination and Cleaning
FIGURE 3.22 Optical microscope image at 50 magnification depicting slight material alterations (surface cracks and channeling) of 100 nm Cr film on a quartz substrate at LIP firing distance of d ¼ 2 mm for a single laser shot [92]
3.4.1. LIP in Pressurized Chambers As demonstrated and analyzed in [90], the shockwave pressure can be increased by increasing the chamber pressure. LIP was created in a pressurized chamber at various ambient static pressure levels to determine its effect on the LIP pressure obtained because pressure amplification was predicted. The experimental set-up for LIP pressure measurements used in [90] is shown in Figure 3.24. The components of the experimental set-up include a pressure chamber built of Plexiglas (inner diameter of 25.4 mm, outer diameter of 37.9 mm, and chamber length of 63.5 mm), a 25.4 mm diameter and 50 mm focal length plano-convex lens, high-energy laser wedged windows specific to the 1064 nm pulsed laser with diameter of 25.4 mm and thickness of 6.1 mm, a tank that can sustain up to 0.56 MPa pressure (along with a pressure gauge), a regulator to keep constant pressure inside the chamber (also with a pressure gauge), a Kistler transient pressure transducer, a charge amplifier, and a digitizing oscilloscope. The LIP peak pressures obtained utilizing the pressure transducer for various levels of additional chamber pressures (dP ¼ 0e0.4 MPa) are depicted in Figure 3.25 and reported in Table 3.6 [90]. The LIP transient pressure waveforms recorded by the pressure transducer are shown in Figure 3.26, for the different pressure levels that were added to the pressurized chamber. The inset in Figure 3.26 shows the zoomed-in images of the peaks of the different waveforms. It is observed from Table 3.6 as
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
107
FIGURE 3.23 (a) Instrumentation diagram for LIP measurement in air, and (b) transient pressure calibration recorded by an embedded pressure transducer on a prototype lithography mask by forming LIP at a firing distance d. SMIF is the acronym for standard mechanical interface
well as Figures 3.25 and 3.26 that the LIP pressure increases with increase in pressure supplied in the chamber. The LIP pressure obtained almost doubled on adding the initial 0.56 MPa pressure, but further increase in pressure resulted in only small increases in LIP pressure, which can be attributed to the cracking of the laser window due to laser exposure during the experiments. Thus, LIP pressure is observed to increase with added pressure in the chamber, with peak pressure of 217 kPa, and could be used for nanoparticle removal.
3.4.2. In-Air Shock Tubes Another method for amplifying shockwave pressure is to use confining cavity structures, called shock tubes [95e97], to limit the propagation directions of
108
Developments in Surface Contamination and Cleaning
FIGURE 3.24 Experimental set-up for LIP creation in a pressure chamber. The photograph above shows the experimental set-up and the schematic shows the pressure chamber arrangement
pressure waves, and hence increase dynamic pressure in the direction of propagation. It was observed that LIP resulted in a peak pressure of 156 kPa in air, and by adding 0.4 MPa of air pressure, while utilizing a pressure chamber, up to 217 kPa was obtained [90]. For obtaining even higher pressures some other pressure amplification technique is required. Further, it will be an added advantage if higher pressures can be achieved at larger firing distances in order to prevent the thermal effects due to radiation heating from the plasma core, determined as a crucial LIP damage source (Section 3.3.7), that could result in material alterations while removing still smaller particles. Therefore, a novel method for the amplification of the pressure (P) obtained from the shockwaves at farther distance from the LIP was introduced in [95]. This amplification can be achieved by constraining the volume and direction available for the expansion of the LIP by focusing the laser beam inside a cylindrical shock tube [95]. Shock tubes can also be utilized to reduce the laser pulse energy (E) while obtaining the same pressure levels, thereby resulting in reduced temperatures
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
109
FIGURE 3.25 Peak LIP pressures in the pressure chamber for different added pressure levels
(T) on the surface and less damage concern. The instrumentation diagram for shock tubes in air is depicted in Figure 3.27a. The objective is to experimentally optimize a shock tube for its two-fold potential to either amplify the pressure field (for better particle removal) or mitigate the temperature (T) experienced on the surface (to reduce material alteration/damage concern). As the distance from the LIP core to the substrate (firing distance d) decreases, the thermal loading on the substrate increases. A pressure decrease of an order of magnitude per 5 mm is observed for LIP shockwaves in air. Higher pressures at distances significantly farther from the core of LIP are obtained with the shock tube technique. Shock tube effectiveness is quantified by its pressure
TABLE 3.6 LIP Pressures in Pressure Chamber for Added Pressures in Chamber Added Pressures in Chamber (kPa) 0
LIP Pressure from Transducer (kPa) 91
101.33
168
202.65
187
303.98
193
405.30
217
110
Developments in Surface Contamination and Cleaning
FIGURE 3.26 LIP transient pressures obtained in the pressure chamber for different added pressure levels
amplification factor. A pressure amplification factor of 11 was obtained at the firing distance d ¼ 10 mm since the shock tube generated a transient pressure of 523 kPa, while in air the LIP transient pressure was 47.5 kPa [95].
3.4.3. Wet-LIP Cleaning The generation and utilization of LIP in liquids (i.e. wet-LIP cleaning) [96] is another pressure amplification technique, in which LIP is created inside a liquid medium (e.g. water) in an immersion tank, resulting in higher pressure levels from the shockwaves compared to pressure levels obtained in air, due mainly to the higher density of water (a factor of ~775 [96]). Another advantage in wetLIP is that the thermal effects of the LIP become mitigated to an extent because of heat loss to water. Furthermore, LIP is formed as a long streak (compared to the elliptical plume in air) in water medium, thus the applied thermal load at the target is restricted, and only the mechanical effect of the shockwave is utilized. The threshold irradiance for laser-induced breakdown for 1064 nm laser in pure water for a 5 ns pulse is approximately 70 GW/cm2 [98]. The instrumentation diagram for wet-LIP is shown in Figure 3.27b. A pressure amplification factor of 5 was observed, from 110 kPa to 550 kPa, when the LIP was created in water as opposed to air, at the same firing distance d ¼ 4 mm. A peak pressure of 1030 kPa was obtained by wet-LIP at a firing distance d ¼ 0.5 mm. As reported in [96], wet-LIP is a potential pressure amplification technique and gives higher pressures (1030 kPa) than shock tubes in air (523 kPa).
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
111
FIGURE 3.27 Instrumentation diagrams for (a) LIP utilizing a shock tube in air; (b) wet-LIP; and (c) submerged shock tube (in water)
112
Developments in Surface Contamination and Cleaning
FIGURE 3.27 (continued)
3.4.4. Submerged Shock Tubes Underwater LIP transient pressure amplification utilizing shock tubes (submerged shock tubes) to assist non-contact particle removal was investigated in [96,97]. This amplification approach could also reduce radiation exposure of the substrate, which was identified as the leading cause of LIP damage during nanoparticle removal (Section 3.3.7). The instrumentation diagram used in [90] for submerged shock tubes (in water) is shown in Figure 3.27c. With the aid of a submerged shock tube the maximum pressure amplitude of 6.48 MPa is observed, demonstrating significant LIP shockwave front pressure amplification and, as a result, particles with smaller sizes and/or stronger adhesion bonds can be removed. It is predicted that with wet-LIP one could remove particles down to 10 nm size, while with submerged shock tubes, removal of PSL particles much smaller than 10 nm from a silicon substrate is theoretically possible. Thus, these LIP pressure amplification techniques have the potential to be utilized for damage-free particle removal from substrates such as patterned silicon wafers, EUVL masks, and photomasks [95]. The experimental LIP pressures obtained at various firing distances (d) utilizing LIP in air, shock tubes in air, wet-LIP, and submerged shock tubes in water are shown in Figure 3.28. It is observed that maximum LIP pressure amplification is obtained with submerged shock tubes, the best being submerged shock tube 3 with transient pressure of 6.48 MPa.
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
113
FIGURE 3.28 LIP pressures available at various firing distances d in air, for shock tubes in air, wet-LIP in water, and for submerged shock tubes in water
4. FUTURE DIRECTIONS IN LASER PARTICLE REMOVAL RESEARCH The utility and capability of the laser-based LIP technique have been studied and demonstrated by removing various sub-100 nm particles from silicon substrates [25,31,69]. Selective removal of nanoparticles is a key attribute of this laser technique. Due to the potential portability, it is also being considered for developing as an ‘in-tool’ technique for EUVL tools. As discussed in this chapter, it is a fast, dry and non-contact technique in which damage risk can be reduced. This technique has critical advantages when (i) material loss is a concern and high particle removal efficiency is desired; (ii) local and precision cleaning of small and surface particles in crucial processes is desired; (iii) applications involve chemically sensitive features; (iv) photomask cleaning is desired as the chemical residue/haze would change the optical properties of the substrate; and (v) hydrophobic substrates are used. The typical applications in the industry include, but are not limited to, post-copper CMP cleaning, postdeposition cleaning, and cleaning steps that include exposed metal and that are susceptible to chemical attack. The capability of this technique for removing nanoparticles from patterned structures is yet to be investigated and demonstrated. The potential complications in the application of LIP to surfaces with delicate features stem from two sources: (i) in general, the mechanical strength of features on a substrate is weaker than that of the substrate materials, and (ii) features can amplify the optical, thermal, and mechanical effects on the substrate, thus leading to increased substrate damage risk. For example, it is known that the thermomechanical damage could result in damage of high-aspect ratio structures in
114
Developments in Surface Contamination and Cleaning
which heat transfer is enhanced due to increased surface area. However, the advanced LIP techniques could be used to mitigate the damage concerns by careful selection of the method and tuning the process based on the sensitivity of the structure. This design space needs to be further explored for specific nanoparticle removal applications. Combination of techniques to improve particle removal efficiency, while being able to minimize damage concern, is one of the logical next steps in this area of research. For example, irradiation with an ultraviolet (UV) laser prior to laser (Nd:YAG) shockwave cleaning was observed to increase the particle removal efficiency over 95%, while it was 50% with UV laser alone and less than 25% with laser shockwave cleaning alone, when attempted on organic PSL particles on Si capping layer of an EUVL mask [62]. In [99], an integrated laser-based wet cleaning tool, which utilizes the physical forces from laser shockwave cleaning to enhance the cleaning chemistries, has been utilized to show effective particle removal down to 30 nm from EUVL blanks. It is also possible to combine localized cleaning by the LIP method with full substrate cleaning using another technique, such as cryogenic cleaning. Such a combined approach would still offer a completely dry cleaning technique and hence could be utilized in applications where conventional wet cleaning methods cannot be used. It is evident that further research in the field is required to fully utilize laser cleaning techniques for future technology challenges in several fields. Certain aspects of LIP are still not well understood, especially at the nano-second time scale. Further research is needed in radiation heating, pressure waves, nanoparticleegas molecule interactions at the nano-scale, laser stability, and the control of LIP cores to better define the design space of the LIP method for advanced special cleaning applications. Additional research is also needed in the areas of the plasma coreesubstrate interactions and the effects of near-field radiation for optimizing the advanced LIP techniques for sub-100 nm particle removal. As the feature sizes in semiconductor manufacturing decrease further, it is reasonable to predict that the needs for such applications will also become more apparent.
5. CONCLUSIONS AND REMARKS In selective nanoparticle removal, dry, chemical-free cleaning techniques have an edge over conventional wet cleaning techniques in many aspects: (i) when material loss is a concern and high particle removal efficiency is desired; (ii) local and precision cleaning of surface and small (sub-micrometer and nano) particles in crucial processes is desired; (iii) applications involve chemically sensitive features; and (iv) photomask cleaning because the chemical residue/ haze would alter the optical properties of the substrate. This chapter has presented a comprehensive summary of recent progress in laser-based cleaning techniques. In the laser techniques, a laser beam with short
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
115
pulse duration is used as a source of energy for nanoparticle detachment. In DLC the pulsed laser beam is directed on the substrate. The irradiation results in acceleration of the substrate due to the thermomechanical expansion and wave propagation. The main reported shortcoming of the DLC technique is the probability of substrate damage for nanoparticle removal. The computational work in the literature indicates that when silicon particles smaller than 600 nm are removed from a silicon substrate, and when silicon particles smaller than 630 nm are removed from a copper substrate using DLC, it would result in thermal and mechanical damages. Liquid-based laser cleaning techniques have also been developed in order to enhance the laser cleaning efficiency and to reduce the substrate damage risk for nanoparticles. Some of the effects include the ablation of the liquid film, explosive boiling of the liquid film and utilization of the higher density of the liquid film (e.g. density of water is three orders of magnitude higher than that of air). In addition, an extensive summary of the LIP technique is provided. One of the key improvements in the LIP technique over DLC is that direct contact of the laser beam or plasma core with the substrate directly is avoided. The propagation and decay of the shockwave pressure and thermal fields and the plasma core are illustrated with the aid of shadowgraphs from the literature. The expansion/propagation of the shockwave after it detaches from plasma core is described based on the blast wave theory (BWT). It is also shown from the transient pressure experiments that the shockwave propagation in the LIP technique deviates from the BWT after the first 10 microseconds of the process. The particle removal mechanisms developed, namely, critical moment and rocking motion of particle based on simple moment balance and a twodimensional adhesion theory (the stresses in both x and y direction are considered in obtaining this formulation), were also described [100]. Further, the use of molecular dynamics simulations based on DSMC were discussed and some key results are presented to shed light on the shockwaveenanoparticle interactions. The necessity for molecular level simulations based on kinetic theory of gases rather than simulations based on continuum equations for modeling shockwaveenanoparticle interaction is explained based on a Knudsen number argument. If the moment exerted on the nanoparticle is greater than the rolling resistance moment, it would result in straining of the adhesion bond and, consequently, removal of the particle. The removal of the nanoparticles which did not attain the critical moment due to the location of the particle away from the impact of the shockwave on the substrate, but still would be removed as a result of the gas moleculeenanoparticle interactions, was explained based on the rocking frequency criterion. According to this criterion, it is possible that the individual gas molecules would oscillate and bombard the nanoparticle with a certain frequency band and, if this frequency covers the rocking resonance frequency of the adhesion bond of the particleesubstrate system, the particle would rock and result in removal after the adhesion bond is ruptured. The capability of the LIP technique has been
116
Developments in Surface Contamination and Cleaning
demonstrated by the particle removal experiments from substrates. Removal of 30e400 nm ceria CMP slurry particles, 10e60 nm PSL particles, and 60 nm gold particles from silicon substrates is demonstrated. The removal of unknown particles from blank photomask is also reported. One of the key advantages of the LIP technique is the selective removal of nanoparticles from substrate rather than targeting the surface of the entire substrate. The SEM images published in the literature clearly indicate that 10e60 nm PSL particles can be removed effectively with this non-contact, dry and fast technique without any substrate damage. Thus, the potential of an efficient sub-100 nm particle removal technique, which is essential in the semiconductor industry and for nanotechnology applications, is demonstrated with LIP. Optimization of the main LIP process parameters, laser energy and firing distance, is also discussed in the literature. To assess damage risk, thermalmechanical analyses have been conducted for LIP shockwave loading and for LIP radiation intensity loading. It was concluded that gas heating and pressure loading of the substrate due to LIP shockwaves, on a photomask and an EUVL mask, generated utilizing the pulsed laser at a firing distance of d ¼ 2 mm for a high strain rate gain factor of w4, may not be of concern for surface damage. The damage threshold for LIP application on the prototype lithography mask, a 100 nm Cr film on a quartz substrate, was assumed to be at the firing distance of d ¼ 2.5 mm (as slight damage was detected at d ¼ 2 mm, but not at dcr ¼ 2.5 mm) based on experimental observations. The timescale for radiation heating is in nanoseconds while it is in microseconds for LIP shockwaves, resulting in laser radiation heating effects showing up earlier than those due to the LIP shockwaves on the sample surface. It was observed that the nanofilm surface responses (temperature rise and stresses) due to laser radiation intensity level heating dominated the thermomechanical load due to the LIP shockwaves. For a photomask with 100 nm Cr film on quartz, at the experimental damage threshold firing distance of dcr ¼ 2.5 mm, the damage limit for maximum radial stress component amplitude is srr,max ¼ 1.93 GPa for no reflectivity of the nanofilm, whereas it is 1.54 GPa for 20% reflectivity. A safe firing distance of d ¼ 3.6 mm (slight damage was detected at d ¼ 3.55 mm, but not at dcr ¼ 3.6 mm), based on experimental observations for LIP application on an EUVL mask with a 2.5 nm Ru film on 105 nm Mo/Si MLs on a quartz substrate, was obtained. Radial stress (srr) and the surface temperature rise (DT) of the Ru film due to laser radiation intensity (w9 times that due to LIP shockwave thermomechanical load) at very small firing distances (d < dcr) are sources for LIP damage concern. If the surface temperature rise leads to melting and/or the radial stress (srr) exceeds the actual yield stress (dynamic) of Ru film material, alteration/damage of the Ru film would occur. It is reported that determination of the onset of material alterations and/or damage of the nanofilms deposited on lithography masks, rather than the extent of damage, is critical to ensure damage-free lithography mask cleaning. Surface cracks and channeling are the first experimentally observed material
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
117
alteration modes on the Cr nanofilm at a firing distance of d ¼ 2 mm. The predicted order (increasing d or decreasing level of excitation) of material alterations in EUVL masks due to LIP exposure is melting, film/MLs stripping, cracks, peeling and discoloration (smallest material alteration). The level of cracking in photomasks is observed to be much higher than that for EUVL masks. In summary, EUVL masks are determined to be weaker (larger dcr ¼ 3.6 mm) compared to the photomasks (dcr ¼ 2.5 mm) for onset of material alterations due to LIP exposure, therefore more caution is required for LIP application on EUVL masks during nanoparticle removal. A set of advanced LIP techniques, namely shock tubes in air, wet-LIP and submerged shock tubes, are discussed for removal of sub-100 nm particles without damage concern. Some future research directions for laser-based nanoparticle cleaning are also identified and discussed.
ACKNOWLEDGMENTS Funds from the National Science Foundation, Intel, International SEMATECH, Praxair Electronics, NYSERDA, NYSTAR, CAMP, and Clarkson University are acknowledged. The authors would also like to thank our previous colleagues from the Photo-Acoustic Research Laboratory, L. Chen, R. Vanderwood, T. Hooper, J. Wu, J. Lin, V.K. Devarapalli, T. Dunbar, D. Zhou, A.J. Kadaksham, B. Maynard and D.A. Thomas, for their valuable contributions in better understanding and maturing the LIP technology.
REFERENCES [1] A.L. Schawlow, C.H. Townes, Infrared and Optical Masers, Phys. Rev. 112 (1940) (1958). [2] A. Javan, Possibility of Production of Negative Temperature in Gas Discharges, Phys. Rev. Lett. 3 (1959) 87. [3] J.H. Sanders, Optical Maser Design, Phys. Rev. Lett. 3 (86) (1959). [4] T.H. Maiman, Optical and Microwave-Optical Experiments in Ruby, Phys. Rev. Lett. 4 (564) (1960). [5] T.H. Maiman, Stimulated Optical Radiation in Ruby, Nature 187 (493) (1960). [6] H. Kogelnik, T. Li, Laser Beams and Resonators, Appl. Opt. 5 (1550) (1966). [7] The International Technology Roadmap for Semiconductors (ITRS) 2009 Edition, SEMATECH International, San Jose, CA (2009). [8] K.L. Mittal (Ed.), Particles on Surfaces: Detection, Adhesion and Removal, Vol. 8, VSP, Utrecht, The Netherlands, 2003. [9] K.L. Mittal (Ed.), Particles on Surfaces: Detection, Adhesion and Removal, Vol. 9, VSP, Brill, Leiden, The Netherlands, 2006. [10] A.A. Busnaina, H. Lin, N. Moumen, J.W. Feng, J. Taylor, Particle Adhesion and Removal Mechanisms in Post-CMP Cleaning Processes, IEEE Trans. Semicond. Manuf. 15 (374) (2002). [11] K. Xu, R. Vos, G. Vereecke, G. Doumen, W. Fyen, P.W. Mertens, et al., Fundamental Study of the Removal Mechanisms of Nano-Sized Particles Using Brush Scrubber Cleaning, J. Vac. Sci. Technol. B 23 (2160) (2005).
118
Developments in Surface Contamination and Cleaning
[12] K. Xu, R. Vos, G. Vereecke, G. Doumen, W. Fyen, P.W. Mertens, et al., Particle Adhesion and Removal Mechanisms During Brush Scrubber Cleaning, J. Vac. Sci. Technol. B 22 (2844) (2004). [13] N.A. Lammers, A. Bleeker, Laser Shockwave Cleaning of EUV Reticles, Proc. SPIE Volume 6730 (67304P) (2007). [14] M. Olim, A Theoretical Evaluation of Megasonic Cleaning for Submicron Particles, J. Electrochem. Soc. 144 (3657) (1997). [15] G. Vereecke, F. Holsteyns, S. Arnauts, S. Beckx, P. Jaenen, K. Kenis, et al., Evaluation of Megasonic Cleaning for Sub-90nm Technologies, Solid State Phenomena 103e104 (141) (2005). [16] W. Kern, Overview and Evolution of Silicon Wafer Cleaning Technology, in: K.A. Reinhardt, W. Kern (Eds.), Handbook of Silicon Wafer Cleaning Technology, second ed., William Andrew Publishing, Norwich, NY, 2007. Ch. 1. [17] C. Bowers, I. Varghese, M. Balooch, J. Rodriguez, Advances in CO2 Cryogenic Aerosol Technology for Photomask Post AFM Repair, in: Proc. SPIE, Volume 7488 (74881R) (2009). [18] C. Bowers, I. Varghese, M. Balooch, W. Brandt, Advances in Post AFM Repair Cleaning of Photomask with CO2 Cryogenic Aerosol Technology, Proc. SPIE Volume 7379 (73791I) (2009). [19] C. Toscano, G. Ahmadi, Particle Removal Mechanisms in Cryogenic Surface Cleaning, J. Adhesion 79 (175) (2003). [20] J.M. Lauerhaas, C. Morote, J. Ple, Cryogenic Aerosol for Zero-Damage Cleaning, Semicond. Intl. 28 (36) (2005). [21] S. Wolf, R.N. Tauber, Silicon Processing for the VLSI Era. Volume 1: Process Technology, second ed., Lattice Press, Sunset Beach, California, USA, 1999. [22] J.W. Butterbaugh, C.K. Gould, Wafer Cleaning Techniques for Meeting the Challenges of Advanced Semiconductor Manufacturing, Semiconductor Manuf. 7 (33) (2006). [23] P.W. Mertens, E. Parton, Sub-100 nm Technologies Drive Single-Wafer Wet Cleaning, Solid State Technol. 45 (51) (2002). [24] H. Aoki, S. Yamasaki, N. Aoto, Cleaning for Deep-Submicron Structures, Electrochem. Soc. Proc. 99 (102) (1999). [25] I. Varghese, C. Cetinkaya, Non-contact Removal of 60-nm Latex Particles from Silicon Wafers with Laser-Induced Plasma, J. Adhesion Sci. Technol. 18 (795) (2004). [26] B.S. Luk’yanchuk, W.D. Song, Z.B. Wang, M.H. Hong, T.C. Chong, J. Graf, et al., New Methods for Laser Cleaning of Nanoparticles, in: C.R. Phipps (Ed.), Laser Ablation and its Applications, Springer, New York, NY, 2007. Ch. 3. [27] W.D. Song, M.H. Hong, B.S. Luk’yanchuk, T.C. Chong, Laser-Induced Cavitation Bubbles for Cleaning of Solid Surfaces, J. Appl. Phys. 95 (2952) (2004). [28] W.D. Song, M.H. Hong, B.S. Luk’yanchuk, A Method and Apparatus for Cleaning Surfaces, U.S. Patent 6,777,642 B2 (2004). [29] J. Graf, B.S. Luk’yanchuk, M. Mosbacher, M.H. Hong, T.C. Chong, P. Leiderer, Optimization of the Energy Transfer Medium in Laser Cleaning: A New Strategy, Abstracts of 4th International Workshop on Laser Cleaning, Sydney, Australia, 2004. [30] B.S. Luk’yanchuk, Z.B. Wang, Y. Zhou, M.H. Hong, W.D. Song, T.C. Chong, Particle on Surface: About Possible Acoustic and Plasmonics Effects in Dry Laser Cleaning, in: D.M. Kane (Ed.), Laser Cleaning II, World Scientific, Singapore, 2006, pp. 81e117. [31] M.D. Murthy Peri, V.K. Devarapalli, C. Cetinkaya, Selective Removal of 10-40-nm Particles from Silicon Wafers Using Laser-Induced Plasma Shockwaves, J. Adhesion Sci. Technol. 21 (331) (2007).
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
119
[32] M. Soltani, G. Ahmadi, Particle Removal Mechanisms under Substrate Acceleration, J. Adhesion 44 (161) (1994). [33] W. Zapka, W. Ziemlich, A.C. Tam, Efficient Pulsed Laser Removal of 0.2 mm Sized Particles from a Solid Surface, Appl. Phys. Lett. 58 (2217) (1991). [34] K. Imen, J. Lee, S.D. Allen, Laser-Assisted Micron Scale Particle Removal, Appl. Phys. Lett. 58 (203) (1991). [35] A.C. Tam, W.P. Leung, W. Zapka, W. Ziemlich, Laser-Cleaning Techniques for Removal of Surface Particles, J. Appl. Phys. 71 (3515) (1992). [36] X. Wu, Excimer Laser-Induced Removal of Particles from Silicon Surfaces, Ph.D. Dissertation, University of Montreal, Montreal, Canada, 1999. [37] D.R. Halfpenny, D.M. Kane, A Quantitative Analysis of Single Pulse Ultraviolet Dry Laser Cleaning, J. Appl. Phys. 86 (6641) (1999). [38] C. Cetinkaya, J. Lin, R. Vanderwood, C. Li, Nanoparticle Removal with a Pulsed Laser: Thermoelastic Transient Surface Acceleration Simulations for Analyzing Edge Effects, J. Adhesion Sci. Technol. 16 (1979) (2002). [39] D. Grojo, M. Boyomo-Onana, A. Cros, Ph. Delaporte, Influence of Laser Pulse Shape on Dry Laser Cleaning, Appl. Surf. Sci. 252 (4786) (2006). [40] G. Vereecke, E. Rohr, M.M. Heyns, Laser-Assisted Removal of Particles on Silicon Wafers, J. Appl. Phys. 85 (3837) (1999). [41] P. Leiderer, J. Boneberg, V. Dobler, M. Mosbacher, H.-J. Mu¨nzer, N. Chaoui, et al., LaserInduced Particle Removal from Silicon Wafers, Proc. SPIE Volume 4065 (249) (2000). [42] G. Schrems, M. Delamare, N. Arnold, P. Leiderer, D. Ba¨uerle, Influence of Storage Time on Laser Cleaning of SiO2 on Si, Appl. Phys. A 76 (847) (2003). [43] Y.W. Zheng, B.S. Luk’yanchuk, Y.F. Lu, W.D. Song, Z.H. Mai, Dry Laser Cleaning of Particles from Solid Substrates: Experiments and Theory, J. Appl. Phys. 90 (2135) (2001). [44] D.J. Hwang, N. Misra, C.P. Grigoropoulos, A.M. Minor, S.S. Mao, Situ Monitoring of Laser Cleaning by Coupling a Pulsed Laser Beam with a Scanning Electron Microscope, Appl. Phys. A 91 (219) (2008). [45] C. Cetinkaya, C. Wu, C. Li, Laser-Based Transient Surface Acceleration of Thermoelastic Layers, J. Sound Vib 231 (195) (2000). [46] J. Lin, C. Cetinkaya, Potential Thermo-Mechanical Substrate Damage in Nanoparticle Removal with Pulsed Lasers, J. Adhesion Sci. Technol. 17 (91) (2003). [47] E.Y. Assendel’ft, V.I. Beklemyshev, I.I. Makhonin, Yu. N. Petrov, A.M. Prokhorov, V.I. Pustovoy, Optoacoustic Effect on the Desorption of Microscopic Particles from a Solid Surface into a Liquid, Sov. Tech. Phys. Lett. 14 (444) (1988). [48] Yu. N. Petrov, Laser Cleaning of Semiconductor Surface, Proc. SPIE Volume 1352 (266) (1990). [49] W. Zapka, A.C. Tam, G. Ayers, W. Ziemlich, Liquid Film Enhanced Laser Cleaning, Microelectron. Eng. 17 (473) (1992). [50] M. Mosbacher, V. Dobler, J. Boneberg, P. Leiderer, Optical Field Enhancement Effects in Laser-Assisted Particle Removal, Appl. Phys. A 72 (41) (2001). [51] P. Leiderer, M. Mosbacher, V. Dobler, A. Schilling, O. Yavas, B.S. Luk’yanchuk, et al., Steam Laser Cleaning of Silicon Surfaces: Laser-Induced Gas Bubble Nucleation and Efficiency Measurements, in: B.S. Luk’yanchuk (Ed.), Laser Cleaning, World Scientific, Singapore, 2002, pp. 255e310. Ch. 6. [52] W. Zapka, The Road to Steam Laser Cleaning, in: B.S. Luk’yanchuk (Ed.), Laser Cleaning, World Scientific, Singapore, 2002, pp. 23e48. Ch. 1.
120
Developments in Surface Contamination and Cleaning
[53] F. Lang, S. Georgiou, P. Leiderer, Phase Transition Dynamics Measurements in Superheated Liquids by Monitoring the Ejection of Nanometer-Thick Films, Appl. Phys. Lett. 85 (2759) (2004). [54] F. Lang, M. Mosbacher, P. Leiderer, Near Field Induced Defects and Influence of the Liquid Layer Thickness in Steam Laser Cleaning of Silicon Wafers, Appl. Phys. A 77 (117) (2003). [55] V.K. Devarapalli, M.D. Murthy Peri, C. Cetinkaya, Particle Removal with Liquid-FilmEnhanced Laser-Induced Plasma, J. Adhesion Sci. Technol. 20 (233) (2006). [56] J.M. Lee, K.G. Watkins, W.M. Steen, Angular Laser Cleaning for Effective Removal of Particles from a Solid Surface, Appl. Phys. A 71 (671) (2000). [57] J.M. Lee, C. Curran, K.G. Watkins, Laser Removal of Copper Particles from Silicon Wafers Using UV, Visible and IR Radiation, Appl. Phys. A 73 (219) (2001). [58] J.M. Lee, K.G. Watkins, Removal of Small Particles on Silicon Wafer by Laser-Induced Airborne Plasma Shock Waves, J. Appl. Phys. 89 (6496) (2001). [59] J.M. Lee, K.G. Watkins, W.M. Steen, Surface Cleaning of Silicon Wafer by Laser Sparking, J. Laser Applic. 13 (154) (2001). [60] S.H. Lee, J.G. Park, J.M. Lee, S.H. Cho, H.K. Cho, Si Wafer Surface Cleaning Using LaserInduced Shock Wave: A New Dry Cleaning Methodology, Surf. Coat. Technol. 169e170 (178) (2003). [61] H. Lim, D. Kim, Optical Diagnostics for Particle-Cleaning Process Utilizing Laser-Induced Shockwave, Appl. Phys. A 79 (965) (2004). [62] S. Lee, Y. Kang, J. Park, A.A. Busnaina, J. Lee, T. Kim, et al., Laser Shock Removal of Nanoparticles from Si Capping Layer of Extreme Ultraviolet Lithography Masks, Jpn. J. Appl. Phys. 44 (5560) (2005). [63] B. Oh, J.W. Lee, J.M. Lee, D. Kim, Numerical Simulation of Laser Shock Cleaning Process for Micro-Scale Particle Removal, J. Adhesion Sci. Technol. 22 (635) (2008). [64] D. Jang, J. Lee, J.M. Lee, D. Kim, Visualization of Particle Trajectories in the Laser Shock Cleaning Process, J. Appl. Phys. 93 (147) (2008). [65] P. Zhang, B. Bian, Z. Li, Particle Saltation Removal in Laser-Induced Plasma Shockwave Cleaning, Appl. Surf. Sci. 254 (1444) (2007). [66] I. Varghese, M.D. Murthy Peri, T.J. Dunbar, B. Maynard, D.A. Thomas, C. Cetinkaya, Removal of Nanoparticles with Laser Induced Plasma, J. Adhesion Sci. Technol. 22 (651) (2008). [67] C. Cetinkaya, T.R. Hooper, Efficiency Studies of Particle Removal with Pulsed-Laser Induced Plasma, J. Adhesion Sci. Technol. 17 (751) (2003). [68] C. Cetinkaya, M.D. Murthy Peri, Non-Contact Nanoparticle Removal with Laser Induced Plasma Pulses, Nanotechnology 15 (435) (2004). [69] V.K. Devarapalli, Y. Li, C. Cetinkaya, Post-Chemical Mechanical Polishing Cleaning of Silicon Wafers with Laser-Induced Plasma, J. Adhesion Sci. Technol. 18 (779) (2004). [70] D. Zhou, A.T.J. Kadaksham, M.D. Murthy Peri, I. Varghese, C. Cetinkaya, Nanoparticle Detachment Using Shockwaves, Proc. Inst. Mech. Engr., Part N: J. Nanoeng. Nanosystems 219 (91) (2005). [71] D. Zhou, C. Cetinkaya, Molecular-Level Mechanisms of Nanoparticle Detachment in LaserInduced Plasma Shockwaves, Appl. Phys. Lett. 88 (173109) (2006). [72] M.D. Murthy Peri, I. Varghese, D. Zhou, A. John, C. Li, C. Cetinkaya, Nanoparticle Removal Using Laser-Induced Plasma Shockwaves, Particulate Sci. Technol. 25 (91) (2007). [73] F. Ferioli, Experimental Characterization of Laser-Induced Plasmas and Application to Gas Composition Measurements, Ph.D. Dissertation, University of Maryland, College Park, MD, 2005.
Chapter | 3
Laser Cleaning for Removal of Nano/Micro-Scale Particles
121
[74] M. Villagra´n-Muniz, H. Sobral, E. Camps, Shadowgraphy and Interferometry Using a CW Laser and a CCD of a Laser-Induced Plasma in Atmospheric Air, IEEE Trans. Plasma Sci. 29 (613) (2001). [75] P.D. Maker, R.W. Terhune, C.M. Savage, Optical Third Harmonic Generation, in: P. Grivet, N. Bloembergen (Eds.), Proc. 3rd Intl. Conf. Quantum Electronics, Columbia University Press, New York, NY, 1964, p. 1559. [76] H. Yan, R. Adelgren, M. Boguzko, G. Elliots, D. Knight, Laser Energy Deposition in Quiescent Air, AIAA J 41 (1988) (2003). [77] G.I. Taylor, The Formation of a Blast Wave by a Very Intense Explosion: I. Theoretical Discussion, Proc. Royal Soc. London A 201 (159) (1950). [78] J. von Neumann, Collected Works of J. von Neumann. Volume VI, Pergamon Press, New York, NY, 1963. [79] L.I. Sedov, Similarity and Dimensional Methods in Mechanics, Academic Press, New York, NY, 1959. [80] G.B. Whitham, Linear and Nonlinear Waves, John Wiley and Sons, New York, NY, 1974. [81] I.G. Dors, C.G. Parigger, Computational Fluid-Dynamic Model of Laser-Induced Breakdown in Air, Appl. Opt. 42 (5978) (2003). [82] Z. Jiang, K. Takayama, K.P.B. Moosad, O. Onodera, M. Sun, Numerical and Experimental Study of a Micro-Blast Wave Generated by Pulsed-Laser Beam Focusing, Shock Waves 8 (337) (1998). [83] J.D. Anderson, Modern Compressible Flow with Historical Perspective, second ed., McGraw-Hill, New York, NY, 1990. [84] K.L. Johnson, K. Kendall, A.D. Roberts, Surface Energy and the Contact of Elastic Solids, Proc. Royal Soc. London 324 (301) (1971). [85] C. Dominik, A.G.G.M. Tielens, Resistance to Rolling in the Adhesive Contact of Two Elastic Spheres, Phil. Mag. A. 72 (783) (1995). [86] M.D. Murthy Peri, C. Cetinkaya, Rolling Resistance Moment of Microspheres on Surfaces, Phil. Mag. 85 (1347) (2005). [87] W. Ding, A.J. Howard, M.D. Murthy Peri, C. Cetinkaya, Rolling Resistance Moment of Microspheres on Surfaces: Contact Measurements, Phil. Mag. 87 (5685) (2007). [88] W. Ding, H. Zhang, C. Cetinkaya, Rolling Resistance Moment-Based Adhesion Characterization of Microspheres, J. Adhesion 84 (996) (2008). [89] G.A. Bird, Molecular Gas Dynamics and the Direct Simulation of Gas Flows, Oxford University Press, Oxford, U.K., 1994. [90] I. Varghese, Onset of Material Alterations and Damage-free Nanoparticle Removal Utilizing Laser Induced Plasma for Nanofilms on Lithography Photomasks, Ph.D. Dissertation, Clarkson University, Potsdam, NY, 2008. [91] M.D. Murthy Peri, D. Zhou, I. Varghese, C. Cetinkaya, Transient Thermoelastic Response of Nanofilms Under Radiation Heating From Pulsed Laser-Induced Plasma, IEEE Trans. Semicond. Manuf. 21 (116) (2008). [92] I. Varghese, D. Zhou, M.D. Murthy Peri, C. Cetinkaya, Onset of Material Alterations Due to Laser-Induced Plasma Exposure in Nanofilms Deposited on Photomasks, IEEE Trans. Semicond. Manuf. 22 (579) (2009). [93] I. Varghese, C. Cetinkaya, Laser Induced Plasma Exposure on Extreme Ultra-Violet Lithography Masks, to be published in IEEE Trans. Semicond, Manuf (2010). [94] I. Varghese, D. Zhou, M.D. Murthy Peri, C. Cetinkaya, Thermal Loading of Laser Induced Plasma Shockwaves on Thin Films in Nanoparticle Removal, J. Appl. Phys. 101 (113106) (2007).
122
Developments in Surface Contamination and Cleaning
[95] T.J. Dunbar, B. Maynard, D.A. Thomas, M.D. Murthy Peri, I. Varghese, C. Cetinkaya, Pressure Amplification of Laser Induced Plasma Shockwaves with Shock Tubes for Nanoparticle Removal, J. Adhesion Sci. Technol. 21 (67) (2007). [96] T.J. Dunbar, M.D. Murthy Peri, I. Varghese, C. Cetinkaya, Submerged Laser-Induced Plasma Amplification of Shockwaves Using Shock Tubes for Nanoparticle Removal, J. Adhesion Sci. Technol. 21 (1425) (2007). [97] T.J. Dunbar, C. Cetinkaya, Underwater Pressure Amplification of Laser-Induced Plasma Shock Waves for Particle Removal Applications, Appl. Phys. Lett. 91 (051912) (2007). [98] J. Noack, A. Vogel, Laser-Induced Plasma Formation in Water at Nanosecond to Femtosecond Time Scales: Calculation of Thresholds, Absorption Coefficients, and Energy Density, IEEE J. of Quantum Electronics 35 (1156) (1999). [99] A. Johnkadaksham, A. Rastegar, Nanoparticle Removal from EUV Mask Blanks Using Wet and Dry Laser Shock Cleaning, Proc. SPIE Photomask Technology 7122 (2008) 7122e7181. [100] M.D. Murthy Peri, Nanoparticle Removal Using Laser Induced Plasma (LIP) Technique and Study of Detachment Modes Based on Molecular Dynamics Simulations, Ph.D. Dissertation, Clarkson University, Potsdam, New York, 2008.
Chapter 4
Non-Aqueous Interior Surface Cleaning Using Projectiles Rajiv Kohli, The Aerospace Corporation, NASA Johnson Space Center, 2525 Bay Area Blvd, Suite 600, Houston, TX 77058, USA
Chapter Outline 1. 2. 3. 4. 5.
Introduction Types of Contamination Effects of Contamination Fluid Cleanliness Levels Tube Cleaning Methods
123 124 125 126 132
6. Non-Aqueous Projectile Cleaning Method 7. Summary Disclaimer References
134 143 143 144
1. INTRODUCTION Contamination of the interior of hoses, pipes, and tubes is a critical problem in many industries because it often leads to corrosion and service breakdowns, requiring extensive repairs at substantial financial and health costs [1]. Particulate and hydrocarbon contamination in lines used in high-pressure liquid and gaseous oxygen systems can also be a fire hazard. Fires have occurred in space, such as on the Mir Space Station, and in aircraft life support, medical applications, aerospace applications, construction materials, and oxygen production [2e5]. And if the contamination is hazardous in nature, any breakdown in the system can lead to major environmental problems. Fluid lines must meet minimum cleanliness levels to prevent the lines from becoming contaminated, thereby reducing the availability and life of the delivery systems. At the other extreme, contamination of carbon nanotubes degrades their properties and presents a different set of challenges for cleaning. Recently, several methods have been developed and successfully applied to clean carbon nanotubes using wet chemical techniques, laser cleaning, plasma, oxygen radicals, or a chemical sleeve [6e14]. Plasma cleaning and steam cleaning have Developments in Surface Contamination and Cleaning. Copyright Ó 2011 Elsevier Inc. All rights reserved.
123
124
Developments in Surface Contamination and Cleaning
been shown to be very effective in cleaning fine-bore surgical instruments [15,16]. Palladium alloy membranes are used in hydrogen purification tubes [17,18]. This chapter describes a non-aqueous cleaning technique for removing solid contaminants from the interior of large-bore tubular components.
2. TYPES OF CONTAMINATION In industrial fluid systems contamination is generally present in the form of solid particles, hydrocarbon films, microbiological materials, or dissolved ions in solution. These contaminants lead to progressive fouling of the interior surfaces, thereby reducing the efficiency of the delivery system. The principal modes and types of fouling are described below [1,19e22]. l
l
l
l
Biological fouling. Micro-organisms (such as bacteria) present in natural waters form an organic film on the tube surface, which tend to grow due to the growth of micro-organism populations. In addition, the microorganisms are a source of nutrients for macro-organisms (such as algae, mussels, seaweed, and other organic fibrous organisms) which can attach to the surface. Growth of the macro-organism films can exacerbate fouling. Sulfate-reducing bacteria are known to produce corrosive by-products, while iron-oxidizing bacteria may actually consume base metal resulting in, for example, manganese pitting of stainless steel. Also, the filtering effect of the porous biofilms tends to entrap fine particulates, further exacerbating fouling problems. Corrosion fouling. Corrosion layers can build up on the surface if the basic, very thin, non-porous protective oxide film is disrupted due to changes in operating conditions and the water chemistry. If a porous oxide is formed, it can also accelerate pitting and corrosion. Deposition or particulate fouling. Suspended fine-particulate debris such as clay, silt, and biogrowth, or precipitated crystalline solids, deposit on tube surfaces in any orientation. Larger particles can settle on horizontal surfaces due to gravitational setting under low-flow conditions. Deposition can promote other fouling mechanisms such as microbial corrosion due to microbes present in the deposit. Scaling or crystallization fouling. Scale occurs when dissolved salts precipitate and deposit on the surface. This can occur due to evaporation of the solvent, or when the solubility limit of the salt is exceeded due to heating (inverse solubility of salts such as CaCO3, Ca3(PO4)2, MgSiO3 and Li2CO3) or cooling (normal solubility). Other causes are pH variations and mixing of fluid streams of different compositions. Non-crystalline solid fouling can also occur if the fluid stream is cooled below the solidification temperature of a component in the stream, for example, wax in crude oil.
Chapter | 4
l
Non-Aqueous Interior Surface Cleaning Using Projectiles
125
Chemical reaction fouling. Chemical reactions that produce a solid phase at or near the surface can cause fouling. The surface itself does not participate, but it can cause thermal degradation of one of the components of the fluid stream, or it may lead to polymerization with the formation of a plastic or rubber-like surface deposit.
Most commonly, several fouling processes occur simultaneously. For example, fine particle deposit increases as such particles are trapped in biofilms formed on the surface. It is critical to fluid delivery system performance that such tubular components are thoroughly cleaned and maintained clean.
3. EFFECTS OF CONTAMINATION Research worldwide has found at least 75% of all hydraulic and pneumatic systems degrade and fail due to fluid contamination [1,23e34]. Contamination causes degradation of the fluids and the performance of the hydraulic system, and, ultimately, its failure due to material degradation (e.g. corrosion, fatigue, wear), increased internal leakage, jamming from accumulated sludge or silt, or excessive heat generated due to loss of control of flow and pressure. Even when care is taken during production and assembly of fluid systems and the system is thoroughly flushed with a cleaning fluid, some hydraulic hoses, tubes, and pipes have been found to contain contaminant particles as large as 800e1200 mm. Smaller contaminant particles, which can be observed with the unaided eye (>40 mm), are also invariably present together with the larger particles. For a fluid system installation which could contain 100e200 meters of hose and tube combinations, this would be approximately 6e10 grams of contamination produced during assembly of the system. Examples of contaminants present in hydraulic systems include core sand, weld spatter, machine swarf (debris or waste resulting from metalworking operations), pipe scale and rust, fibrous material, packaging residue, paint flakes, rubber particles from the hoses and seals, and oil oxidation products. Most hydraulic system contamination failures are caused by solid particles that chemically react with the fluid, or by fouling the system through accumulation. Particle size range of most concern is 5e20 mm, but even 0.5 mm particles can be harmful to most systems due to their increased propensity for silting. With the increased demand for higher system pressures and faster cycle times, manufacturers have been imposing tighter tolerances and clearances between moving surfaces, which, in turn, demands cleaner fluids. For example, a tolerance of 2e5 mm gives a dynamic clearance of 1e2.5 mm between two mating surfaces. The damaging effects of solid particle contamination are influenced by the composition, size, shape, and abrasiveness of the particles. Metal particles tend to catalyze oil oxidation and contribute to corrosion. High concentrations of small hard metal or metal oxide particles (10 mm), which are smaller than the clearance between the two mating surfaces, form silt which erodes the interior
126
Developments in Surface Contamination and Cleaning
mating surfaces of valves, causing loss of control characteristics and efficiency of the component. Jamming of the moving parts can also occur, leading to component malfunction and erratic machine operation. Contaminating particles larger than the clearance gap block ports and orifices, and can cause erosion and leakage of a valve if trapped in the working clearance between the poppet and its seat. The larger particles, in turn, can break into smaller particles. High flow velocities or operating pressures can exacerbate the problem. Contaminating solid particles that are about the same size as the clearance between two moving surfaces can cause both jamming and rapid wear of the mating surfaces. The abrasive action of the original particles due to metal-tometal contact produces new particles and a chain reaction of abrasion and contamination is initiated. Increased dynamic clearance between the mating surfaces causes increased system leakage, loss of system efficiency and control, and localized heat (and increased maintenance costs). The other wear mechanisms that result in self-generating contaminants include: adhesive, abrasive, erosion, fatigue, delamination, corrosive, electro-corrosive, fretting corrosion, cavitation, electrical discharge, and polishing wear. Each of these types of wear categories has its own mechanism and symptoms; however, in practice they may occur individually, combined, or in sequence. Contaminants, such as water and air or other gases in oil, can degrade the performance of hydraulic systems and component failure. Free water, which forms when the saturation point of the oil is exceeded, can adversely impact the fluid chemistry by reacting with oxidation products and purposely added chemicals (additives) to form organic acid compounds and sludge. Free water can also be present as emulsified droplets suspended throughout the fluid. Other detrimental effects of free water include accelerated corrosion and abrasive wear, metal fatigue, reduced bearing life, jamming due to ice crystals formed at low temperatures, loss of dielectric strength, and bacterial problems. Air and other gases in oils may cause foaming, slow system response, higher temperatures, pump cavitation, lack of system pressure, and accelerated oil oxidation. Contamination presents the potential for critical failure of manned and unmanned space missions which could result in loss of mission or loss of crew or both. For example, particle contamination in a flow control valve can cause blockage of fuel delivery to the main engine of a manned vehicle or to the propulsion system of a satellite, causing the system to malfunction or fail entirely; or, worse, it can ignite the fuel with disastrous consequences. Clean fluid is the only way to achieve optimum performance from a hydraulic or pneumatic system.
4. FLUID CLEANLINESS LEVELS Particle contamination can reduce the service life of fluid delivery systems. A correctly maintained fluid can eliminate 75e85% of the hydraulic machine’s
Chapter | 4
Non-Aqueous Interior Surface Cleaning Using Projectiles
127
future failures and increase the life expectancy of its components. Some level of particle contamination is always present in the fluid streams, even in a new fluid. New hydraulic fluids can contain 500 000 particles larger than 5 mm per 100 mL [23,27,29,33], which exceeds the level recommended for hydraulic system normal operation. Operations of processing equipment used for synthesis are sources of particle contamination. Transfer of hydraulic fluids from delivery tanks to storage containers at point of use can make them susceptible to ten times the amount of contamination than if the fluid were retrieved directly from the manufacturer. The level of acceptable cleanliness depends on the type of fluid delivery system [24,25,34].
4.1. Hydraulic Fluids The contamination level in hydraulic and lubricating oils is characterized by an oil cleanliness code. Due to their shape, size tolerance, contamination sensitivity, function, and operating method different components tolerate different contamination levels of operating fluids. These contamination levels of operating liquids are defined by oil cleanliness codes according to various standards.
4.1.1. ISO Standard The most frequently used classification for oil cleanliness codes in fluid power systems is defined by the international standard ISO 4406 (Table 4.1) [35]. The cleanliness level may be measured either by automatic or microscopic particle counting. The classification of the cleanliness level of a hydraulic oil is identified by a three-number code for particles >4 mm, >6 mm, and >14 mm in a 100 mL sample when measured with an automatic particle counter [36e39]. For example, an oil sample from a hydraulic system gave the following results of particle counts measured with an automatic particle counter. Particle Size
Number of Particles
ISO Code
>4 mm >6 mm >14 mm
85 376 15 516 1301
17 14 11
According to Table 4.1, this hydraulic oil has an oil cleanliness of ISO 17/14/11. If an optical microscope is used to count the particles [37,40], only a twodigit code is used to designate the cleanliness of the sample. In this case, only particle sizes larger than 5 mm and 15 mm are considered. As an example, an oil sample with the following particle count distribution measured with an optical microscope would be assigned an oil cleanliness of ISO 17/15.
128
Developments in Surface Contamination and Cleaning
Particle Size
Number of Particles
ISO Code
>5 mm >15 mm
81 412 17 979
17 15
TABLE 4.1 Oil Cleanliness Codes per ISO 4406 [35] No. of Particles > Given Size ISO Code
From
To
7
64
130
8
130
250
9
250
500
10
500
1000
11
1000
2000
12
2000
4000
13
4000
8000
14
8000
16 000
15
16 000
32 000
16
32 000
64 000
17
64 000
130 000
18
130 000
250 000
19
250 000
500 000
20
500 000
1 000 000
21
1 000 000
2 000 000
22
2 000 000
4 000 000
23
4 000 000
8 000 000
4.1.2 SAE Standard Another important standard for oil cleanliness codes is SAE AS4059 [41]. In contrast to ISO 4406, this standard also includes coarser particles >70 mm diameter. The codes are shown in Table 4.2. The rate at which damage occurs is dependent on the internal clearances of the components within the system, the size and quantity of particles present in
Chapter | 4
129
Non-Aqueous Interior Surface Cleaning Using Projectiles
TABLE 4.2 Oil Cleanliness Codes per SAE AS4059 [41] Maximum Number of Particles for Given Particle Size Optical Counting
>1 mm
>5 mm
>15 mm
>25 mm
>50 mm
>100 mm
Automatic Counting
>4 mm
>6 mm
>14 mm
>21 mm
>38 mm
>70 mm
Size Code Class
A
B
C
D
E
F
Code 000
195
76
14
3
1
0
Code 00
390
152
27
5
1
0
Code 0
780
304
54
10
2
0
Code 1
1560
609
109
20
4
1
Code 2
3120
1220
217
39
7
1
Code 3
6520
2430
432
76
13
2
Code 4
12 500
4860
864
152
26
4
Code 5
25 000
9730
1730
306
53
8
Code 6
50 000
19 500
3460
612
106
16
Code 7
100 000
38 900
6920
1220
212
32
Code 8
200 000
77 900
13 900
2450
424
64
Code 9
400 000
156 000
27 700
4900
848
128
Code 10
800 000
311 000
55 400
9800
1700
256
the fluid, and the system pressure. Typical internal clearances of hydraulic components are shown in Table 4.3 [24,42]. The minimum recommended fluid cleanliness levels for different types of hydraulic systems, defined according to ISO and SAE standards, are shown in Table 4.4. Table 4.5 [43,44] recommends conservative target ISO cleanliness codes based on several component manufacturers’ guidelines and extensive field studies for standard industrial operating conditions in systems using petroleumbased fluids.
4.1.3. Military Standard The United States Department of Defense (DoD) has issued a standard MILH-5606G for cleanliness of hydraulic fluids [45]. The number of solid particles in each 100 mL sample of hydraulic fluid cannot exceed the values shown in Table 4.6.
130
Developments in Surface Contamination and Cleaning
TABLE 4.3 Typical Internal Clearances of Hydraulic Components [24,42] Component Type
Typical Internal Clearance, mm
Gear pump
0.5e5
Vane pump
0.5e13
Piston pump
0.5e40
Proportional valve
2.5e40
Servo valve
1.0e63
Control valve
0.5e40
Pressure valve
13e40
Linear actuator
50e250
Bearings
0.5e100
TABLE 4.4 Minimum Fluid Cleanliness Levels for Different Types of Hydraulic Systems [24,43] Minimum Recommended Cleanliness Level Type of Hydraulic System
ISO 4406
SAE 4059
Silt sensitive
15/13/10
1
Servo
16/14/11
2
High pressure (25e40 MPa)
17/15/12
3
Normal pressure (15e25 MPa)
18/16/13
4
Medium pressure (5e15 MPa)
20/18/15
6
Low pressure (< 5 MPa)
20/18/15
e
Large clearance
21/19/16
e
4.2. Non-Hydraulic Fluids For aerospace applications, strict cleanliness requirements have been derived from liquid oxygen (LOX) system compatibility, since contaminants such as particles and hydrocarbon greases and oils can easily ignite in the presence of LOX. These cleanliness requirements are specified for all gaseous and
Chapter | 4
131
Non-Aqueous Interior Surface Cleaning Using Projectiles
TABLE 4.5 Recommended Target ISO Cleanliness Codes for Systems Using Petroleum-Based Fluids per ISO 4406 for Particle Sizes 4 mm/6 mm/14 mm [43,44] Operating Pressure, MPa Hydraulic System
<14
21.2
>21.2
Pumps Fixed gear
20/18/15
19/17/15
e
Fixed piston
19/17/14
18/16/13
17/15/12
Fixed vane
20/18/15
19/17/14
18/16/13
Variable piston
18/16/13
17/15/13
16/14/12
Variable vane
18/16/13
17/15/12
e
Valves Cartridge
18/16/13
17/15/12
17/15/12
Check valve
20/18/15
20/18/15
19/17/14
Directional (solenoid)
20/18/15
19/17/14
18/16/13
Flow control
19/17/14
18/16/13
18/16/13
Pressure control (modulating)
19/17/14
18/16/13
17/15/12
Proportional cartridge valve
17/15/12
17/15/12
16/14/11
Proportional directional
17/15/12
17/15/12
16/14/11
Proportional flow control
17/15/12
17/15/12
16/14/11
Proportional pressure control
17/15/12
17/15/12
16/14/11
Servo valve
16/14/11
16/14/11
15/13/10
Bearings Ball bearing
15/13/10
e
e
Gearbox (industrial)
17/16/13
e
e
Journal bearing (high speed)
17/15/12
e
e
Journal bearing (low speed)
17/15/12
e
e
Roller bearing
16/14/11
e
e
Actuators Cylinders
17/15/12
16/14/11
15/13/10
Vane motors
20/18/15
19/17/14
18/16/13
(Continued)
132
Developments in Surface Contamination and Cleaning
TABLE 4.5 Recommended Target ISO Cleanliness Codes for Systems Using Petroleum-Based Fluids per ISO 4406 for Particle Sizes 4 mm/6 mm/14 mm [43,44]econt’d Operating Pressure, MPa Hydraulic System
<14
21.2
>21.2
Axial piston motors
19/17/14
18/16/13
17/15/12
Gear motors
20/18/14
19/17/13
18/16/13
Radial piston motors
20/18/15
19/17/14
18/16/13
Test stands Test stands
15/13/10
15/13/10
15/13/10
Hydrostatic transmissions
17/15/13
16/14/11
16/14/11
TABLE 4.6 Maximum Number of Solid Particles Allowed According to MIL-H-5606G [45] Particle Size Range (Largest Dimension), mm
Maximum Number of Particles Allowed (Automatic Particle Counting)
5e15
10 000
16e25
1000
26e50
150
51e100
20
>100
5
non-gaseous fluids used in high-pressure oxygen systems. The cleanliness limits (chemical and particle contaminants) for fluids used in the Space Shuttle systems are specified in [46]. Precision cleanliness levels are specified for applications where contamination control limits are necessary to ensure reliability and performance of fluids and parts and components exposed to the fluids [38].
5. TUBE CLEANING METHODS Several different cleaning methods have been developed and have been successfully employed for in-line and off-line cleaning of tubes [1,20,22,47e68].
Chapter | 4
1. 2. 3. 4. 5. 6.
Non-Aqueous Interior Surface Cleaning Using Projectiles
133
Steam cleaning Air blowing High pressure water jet cleaning Mechanical cleaning Projectile cleaning Chemical cleaning (alkali or acid solutions, solvents).
Steam cleaning is best suited for the removal of oil and grease. Chemicals such as Na3PO4 may be added to steam to increase its cleaning effectiveness. Debris such as dirt and rust will also be removed if the velocity of the steam is adequate to blow loose debris out of the tube, or if the pressure is sufficient to break up adherent debris. This is usually not the case with field application of typical steam cleaners. The method is also time consuming. The same velocity problems and limitations of steam cleaning also apply to air blowing. High-pressure water jet cleaning employs water at high pressures to 415 MPa to clean the tube. A mechanical pig may also be forced into the tube to enhance cleaning effectiveness. This method will remove dirt and even scale. Large volumes of water are used and the waste water must be disposed of at high cost. Although the use of high-pressure water can be effective with certain deposits, the jet nozzle must be moved along the tube slowly, and the time required to clean a heat exchanger can be excessive. Mechanical cleaning involves the use of air-driven high-speed rotating tools such as brushes, buffing tools, hones, scrapers, or cutters that clean by abrasive action. One disadvantage of high-speed rotation is the inability to control dwell time on the surface. Low-speed rotary cleaning systems have been developed to overcome this disadvantage. Depending on the type of contamination, the tools can be made from soft or hard plastics, metals, or hard ceramics. Brushes are available with bristle sizes as small as 3 mm for medical applications such as cleaning catheters or endoscopes [67,68]. Often, the cleaners have ports through which water or other liquids can be injected to clean and flush in a single operation. The cleaning method is fast, economical, and safe for straight tubes, and it can clean almost all types of deposits including hard scale. With properly designed and manufactured scrapers, there is minimal loss of base metal of the tube. According to one investigation, it would require nearly 1000 years of annual CuNi condenser tube cleaning to result in a critical reduction (~30 to 50%) in tube wall thickness loss [61]. Brushes have also been effective in cleaning tubes with enhanced internal surfaces (spirally indented, grooved, or finned), or tubes with thin metal inserts or epoxy type coatings. However, the presence of sharp bends in small-diameter tubing makes this cleaning method often inapplicable. Projectile cleaning has traditionally employed compressed air and water or high-pressure water alone to propel a cleaning projectile through the tubes to remove deposits. Projectiles can range from rubber bullets to brushes to hard
134
Developments in Surface Contamination and Cleaning
plastic or various metal or non-metallic abrasive (carbide) scrapers. Most projectiles are only suitable for light deposits such as mud and algae, although carbide scrapers have been used to successfully remove hard calcium carbonate deposits from condenser tubes. One advantage in using water is that the wastewater containing the deposit can be collected for laboratory analysis. Chemical cleaning with alkaline or acid solutions is very effective in removing scale or hard oxide films, as well as other hydrocarbon contaminants and debris. Other liquids, such as solvents and even hydraulic oil, have also been used, but they are less effective in removing strongly adhering deposits. However, the use of hazardous chemicals increases the risks, requires enhanced personnel safety and expertise, and adds significant waste disposal costs to the cleaning operation. The cleaning process is also time consuming. Its use has been declining in recent years.
6. NON-AQUEOUS PROJECTILE CLEANING METHOD Many of the cleaning methods mentioned in Section 5 are only partially effective, are often time consuming and expensive, use hazardous chemicals and solvents, or they use large volumes of water which must also be disposed at high costs. To overcome these shortcomings, a non-aqueous method has been developed that uses projectiles to clean the internal surfaces of tubular components [49e53,55e58]. A pneumatic launcher propels the projectile, sized slightly larger than the internal diameter of the tube, into the tube. As the projectile travels through the tube, it removes the contamination deposited on the internal tube surface and forces it out of the tube. Cleaning can be accomplished very effectively in seconds.
6.1. Principle of the Cleaning Method The principle of the projectile cleaning method is illustrated in Figure 4.1. A pneumatic launcher with an acetal compression nozzle is used to propel a compressible polyurethane projectile pneumatically through the tube to be cleaned. The projectile is 20e30% larger in external diameter than the internal diameter of the tube. Once compressed through the nozzle, the projectile expands against the internal surface and achieves and maintains full 360 contact with the internal surface of the tube. The elasticity of the projectile material and the friction on the surface, combined with the propulsion force from the compressed gas, allow the projectile to remove the contamination on the surface as it travels through the system and eject the removed contamination from the open end of the tube, hose, or pipe. The projectiles travel at approximately 15 m/s, so cleaning can be accomplished very quickly, even with bends, curves, or elbow joints in the tube. The only requirement is a source of compressed gas and an entrance and exit in the tube to be cleaned.
Chapter | 4
Non-Aqueous Interior Surface Cleaning Using Projectiles
135
FIGURE 4.1 Schematic of the non-aqueous projectile tube cleaning system [69]
6.2. Equipment The basic cleaning system comprises a pneumatic launcher, the nozzle, and the cleaning projectile [70e76]. It is available as a manual system or as a semi- or fully automated system that can be integrated in a production line.
6.2.1. Pneumatic Launchers The hand-held launcher (Fig. 4.2a) is ideal for small production shops, mobile hose fabrication, and job site applications because of its size and portability. It is manually operated with single projectile feeding and the simple design incorporates non-fatigue ergonomic features and safety mechanisms that make it easy to operate safely for long periods. The bench-mounted launcher is generally installed and operated as a fixed system with automatic or manual projectile feeding and dispensing (Fig. 4.2b,c). These launchers have a cycle time of less than 2 seconds and are well suited for use in production applications.
6.2.2. Projectiles The projectile cleans by achieving and maintaining pressure against the internal surface of the hose, tube, or pipe. This pressure is achieved because the projectile is approximately 20e30% larger than the internal diameter of the tube to be cleaned. For instance, a 50 mm projectile is recommended for a 38 mm hose. Also, the length of a projectile should be greater than the width so the projectile does not tumble. There are several types of projectiles varying in density, porosity, and surface structure (Fig. 4.3). The projectiles are manufactured from virgin foam or from rebonded foam. Rebonded foam may flake and contaminate the surface
136
Developments in Surface Contamination and Cleaning
FIGURE 4.2 Projectile launchers. (a) Hand-held unit. (b) Bench mount unit with projectile dispenser. (c) Bench mount system with nozzles. Courtesy of Tube Clean GmbH, Hinwil, Switzerland and Ultra Clean Technologies, Bridgeton, NJ
FIGURE 4.3 Available projectiles used for cleaning. (a) Standard projectile. (b) Coupling projectile. (c) Product recovery projectile. (d) Abrasive projectile. (e) Grinding projectile [74,75]. Courtesy of Tube Clean GmbH, Hinwil, Switzerland and Ultra Clean Technologies, Bridgeton, NJ
Chapter | 4
Non-Aqueous Interior Surface Cleaning Using Projectiles
137
as it travels through the tube. The density should be sufficiently high to prevent the compressed gas from passing through the projectile, but not too high to prevent the projectile from being compressed and to prevent it from negotiating curves, bends, and other constrictions. The typical density of the foam is in the range 80e200 kg/m3. A. Standard projectile. This ultraclean projectile removes fine particles of loose contamination and can also be used for product purging. It is used in tube-cleaning applications to remove mandrel lubricants, grease, and oil after the bending process. B. Coupling projectiles. These ultraclean projectiles can be used in nonstraight-line tubes and pipes. The structure and mechanical properties of the coupling projectile provide sufficient flexibility for the projectile to compress through joints, couplings, bends, kinks, and other reductions in the system. C. Product recovery projectiles. The product recovery projectiles are designed to travel through a system without the need for disassembly. The projectiles have a closed cell structure and mechanical properties to achieve compactability up to 60%. This ensures maximum expansion and contact within the tube, allowing for efficient product recovery. These projectiles are used to recover residual product after processing in the food and beverage industry. D. Abrasive projectile. Abrasive projectiles are manufactured with abrasive gauze fitted at the front of the projectile. The gauze acts as an effective scrubber to remove surface rust or scale from straight tubes or tube assemblies. An ultraclean standard or coupling projectile should always be used after an abrasive projectile to insure removal of abrasive gauze debris from the tube. E. Grinding projectile. Grinding projectiles are coated with abrasive medium such as alumina and are used to remove heavy contamination and corrosion layers (scale or rust) in straight lengths of tube or pipe. For a 6 mm diameter tube, a 4e6 mm grinding projectile is recommended. For larger-diameter tubes, the projectile should be the same diameter as the tube. The grinding projectile must be followed by an ultraclean standard projectile to ensure removal of any grinding debris from the tube.
6.2.3. Ultraclean Nozzles Nozzles are available in different sizes and configurations (Fig. 4.4). The nozzles have smooth aerodynamic internal surfaces to easily propel the projectiles. A. Hose nozzle. The nozzle is inserted into the hose (Fig. 4.4a), so the external diameter of the nozzle must be less than the internal diameter of the hose.
138
Developments in Surface Contamination and Cleaning
FIGURE 4.4 Examples of nozzles and configurations for cleaning. (a) Hose nozzle. (b) JIC nozzle and coupling. (c) Tube nozzle [75]. The top row shows the actual nozzles, while the schematics in the bottom row show the nozzle configuration during cleaning. Courtesy of Tube Clean GmbH, Hinwil, Switzerland
B. JIC (Joint Industry Conference) nozzle. The JIC nozzle has a male flare at the top that will mate against the female JIC coupling on hose and tube assemblies. The male flared fitting of the JIC nozzle butts against the female JIC coupling. The use of a JIC nozzle may be required to mate correctly with a flared end of the tubing. C. BSP (British Standard Pipe) coupling fitting. The female flared fitting of the BSP nozzle butts against the male BSP coupling. D. Tube nozzle. The nozzle is inserted over the tube (Fig. 4.4c), therefore, the external diameter of the tube must be less than the internal diameter of the nozzle. There is a stop on the inside of the nozzle that forms an airtight seal when the tube is fully inserted into the nozzle.
6.2.4. Automated Cleaning Systems Automated systems for high-volume, high-rate cleaning in production lines are available [72,75]. These systems are supplied with manual or automatic projectile feeding and dispensing and can be operated as mobile or stationary units (Fig. 4.5). The systems can be integrated into automatic processing lines and can clean pipes, hoses, and tubes with inside diameters of 2e60 mm at rates as high as 3000 articles per hour.
6.3. Operating Considerations The launchers have interchangeable nozzles and projectiles available in a variety of diameters, so all tube sizes from 2 mm to 110 mm can be cleaned to achieve ISO 4406 15/13/10 cleanliness level. The projectiles will travel through a tube that has been kinked or flattened by about 40e60% along the length of the tube, but the constriction cannot occur at the beginning of the tube where the projectile is entering. The smaller the tube, the less tolerance there will be for constrictions. Similar considerations apply to other constricting
Chapter | 4
Non-Aqueous Interior Surface Cleaning Using Projectiles
139
FIGURE 4.5 Automated production level cleaning systems offered by Tube Clean GmbH. (a) Full automated system. (b) Mobile semi-automated system. Courtesy of Tube Clean GmbH, Hinwil, Switzerland
configurations such as knots, coils, or clamped joints. The projectile will go through 90, 180, or 360 degree bends. The standard, coupling and abrasive projectiles will negotiate sweeping bends and the standard ultraclean projectiles will handle tight bends. Branching tube geometries and tees can be cleaned provided one branch is sealed and the projectile enters the tube at the correct entry location. The projectile will go through coupling and ball valves and similar designs, but not through a non-return valve or butterfly valve or one of similar design. As long as there is sufficient volume of gas behind the projectile, it will keep traveling through the tube. Thus, long sections of tubes hundreds of meters in length can be cleaned. The normal operating pressure is 0.59e0.97 MPa. Operating the launcher at the high end of the recommended pressure range can enhance contamination removal. If gas pressure is too low, the correct-sized projectile could possibly become lodged in the tube. The use of a smaller projectile will increase the number of projectiles required and the time for cleaning and reduce the overall effectiveness of the system. If the gas pressure is too high, the valve and trigger in the launcher may not operate properly. The gas source should always be regulated to the correct pressure and filtered to insure contaminant-free dry gas. The benchmounted and automated systems incorporate a suitable filter in the gas line. The standard ultraclean or product recovery projectiles will remove silt and soil, slurry and moisture from the line. Hardened deposits usually require an abrasive or grinding projectile for removal. This is always followed by a standard ultraclean projectile to remove any residual contamination from the abrasive cleaning step. The standard and abrasive projectiles will not damage the surface; however, grinding projectiles may cause some scratching. For
140
Developments in Surface Contamination and Cleaning
certain applications, the projectiles can also be soaked in rust-inhibiting chemicals that will be applied to the entire inner surface as it travels through the tube, accomplishing cleaning and rust prevention in a single step. In other cases, such as paint lines, the projectile may need to be soaked in paint thinners to assist in removing the paint. To maintain the cleanliness of the component or the hardware after cleaning, ultraclean tape is often applied over the open ends. The tape can be easily removed before final assembly of the component or installation of the hardware. Selection of the correct projectile size is important to the cleaning operation. If the projectile is too large it will not leave the nozzle and if it is too small it will not clean effectively. The ejected projectile is often an indicator of the condition of the tube. Damage of the internal tube wall or the presence of a sharp surface protrusion, such as a burr, weld spatter, or a broken wire, can shred the projectile. A projectile collector is frequently used to collect the projectiles to assess the internal condition of the tube and to determine whether additional cleaning is required, as well as to collect the removed contaminants for identification and laboratory analysis. Depending on the cleanliness of the application, the projectiles could be cleaned and reused. Reusing the projectiles in the medical, food and beverage, or hydraulic applications is generally not recommended, but other applications with less stringent cleanliness requirements may benefit from reduced costs by cleaning and reusing the projectiles. The additional resources and costs of cleaning the projectiles may offset the cost advantages of reuse. A projectile control and verification system is often used in high-volume, high-rate cleaning applications to ensure that a projectile is never left inside a hose or tube being cleaned.
6.4. Advantages and Disadvantages 6.4.1. Advantages l l
l l
l
l l
The cleaning system is simple, safe, and easy to operate. High cleanliness levels can be achieved, which can also reduce warranty claims. Hardware can be cleaned very rapidly, saving costs and reducing downtime. Assembled systems do not have to be disassembled for cleaning. The projectile negotiates tee joints, elbows, and 90 bends. If a tube component fails, it can simply be disconnected from the assembly for cleaning. The system can be used as an indicator of the internal condition of the component or assembled hardware. Long lengths are not an obstacle to cleaning. The use of hazardous cleaning chemicals and solvents is reduced or eliminated, thereby saving high disposal costs. Personnel risks from inhalation, contact, and disposal are reduced.
Chapter | 4
l
l
Non-Aqueous Interior Surface Cleaning Using Projectiles
141
It is environmentally friendly. The projectiles are inexpensive and can be readily disposed. Operator morale is improved by the use of an efficient and effective cleaning system.
6.4.2. Disadvantages As with any projectile cleaning method, this method has some disadvantages. l
l
l
l
l
Hard deposits, such as rust and scale, will require multiple cleaning cycles with abrasive and grinding projectiles, followed by a standard projectile. This will increase the operating costs for maintenance of such systems. Also, grinding projectiles do not effectively negotiate bends, so only straight sections with hardened deposits can be cleaned. Both ends of the tube, hose, or pipe must be open for cleaning and recovery of projectiles, residual product, or removed contaminants. Removal of the end caps increases labor requirements. There is a safety risk of the projectile being launched back at the operator when the launcher is removed from a plugged tube that becomes pressurized with the gas. Most systems include a safety feature to prevent this from occurring. The method is not very effective on enhanced surfaces (convoluted, finned, or ridged tube configurations). The tube must have a consistent internal diameter throughout its length. The system will not clean if there are significant reductions and expansions in the tube. It must be disassembled into sections to clean.
6.5. Applications The projectile cleaning system is used in a wide range of commercial and industrial applications that transmit power through fluid transfer as listed in Table 4.7 [74e76]. For example, a steam generator with a thermal output of 1100 MW may have nearly 70 km of tubes flowing water at high temperatures and pressures. These tubes are cleaned regularly as part of a preventive maintenance program. Srimongkolkul [77] evaluated the cleanliness of a spiral hydraulic hose (25.4 mm internal diameter, 610 mm long). The hose had been cut with an abrasive wheel or a mechanical saw and cleaned using standard projectiles. Particle counts were measured after flushing the hose with clean filtered hydraulic oil (ISO 4406 15/13/10) before and after cleaning. The results showed the hose could be cleaned to ISO 4406 15/13/10 cleanliness level. A hygiene and microbiological assessment [78] was recently performed on the cleaning system offered by Tube Clean GmbH. The intent of the assessment was to determine the suitability of the system for cleaning, disinfection, and
Use
Hydraulic and pneumatic lines
Eliminates rubber contamination, metal particles, contaminated oil, moisture contamination introduced through the manufacture and cutting that reduce operating efficiency and cause breakdown and component failure
Steam boilers
Removes most scaling in steam pipes for servicing during regular maintenance
Heat exchangers and condenser tubes
Eliminates contamination that reduces heat transfer resulting in low-level performance
Air conditioning and refrigeration
Eliminates minute particles in copper tubes, coolant lines that affect system performance
Oxygen and gas lines
Eliminates oil, grease, and other contaminants from copper or stainless steel tubing
Oil, gas, and chemical processing
Efficient cleaning and purging of product pipes as part of service maintenance
Earthmoving and mining equipment maintenance
Removal of contaminants in hydraulic assemblies, new equipment, and for repairs to used or failed equipment, thus reducing downtime. Allows transmission of fluid energy efficiently to the working elements on heavy equipment. Reduces flushing time and filter usage
Rubber and plastics
Removal of latex from conveyor pipes. Removal of by-products, plastic fibers, and other deposits from injection molding lines
Automotive servicing
Cleaning of fuel lines, brake lines, air conditioning, and power steering lines prior to assembly and servicing of components
Wave guide
Cleaning of microwave signal transfer lines
Food and beverage product recovery and contamination removal
Retrieval of product (e.g. chocolate, ice cream, syrups, other liquids) from lines during product changeover and general cleaning of the lines, thereby reducing or eliminating solvents or detergents. Eliminates contamination (bacteria, yeast buildup, and other microbiological contaminants) by effectively cleaning the internal surface of TempriteÒ (Temprite, Chicago, IL) coils, superchiller coils, beer, juice, and other carbonated beverage lines
Gun barrels
Removes rust, scale, or powder residue from gun barrels much faster than brushing or swabbing
Developments in Surface Contamination and Cleaning
Application
142
TABLE 4.7 Applications of the Projectile Cleaning System [74e76]
Chapter | 4
Non-Aqueous Interior Surface Cleaning Using Projectiles
143
drying of drinking water hoses (designated ‘S’) for short-term application. A hose with high biofilm and inorganic contamination loading (designated ‘U’) was used for comparison. Drinking water was circulated through the hoses for 8 weeks at ambient temperature. The hoses were then cleaned with standard or product recovery projectiles. The ‘S’ hose required only two projectile shots to clean and dry the hose; the second shot projectile appeared to be clean and dry. By comparison, even the fourth shot projectile from the ‘U’ hose was covered in a slimy fluid together with a large amount of debris, indicating considerable residual contamination in the hose. The bacterial count in the collected wastewater and ejected projectiles gave 60 CFU/mL for the ‘S’ hose compared with >150 CFU/mL for the ‘U’ hose. Finally, tests were performed to disinfect a bacterially contaminated ‘S’ hose by shooting multiple projectiles soaked in a suitable disinfectant through the hose. No evidence of bacterial contamination was found on the projectile or on the inner surface of the hose after the second projectile shot, indicating complete and effective disinfection of the hose. Several manufacturers of hydraulic hose and hydraulic components and other user organizations in different industrial sectors worldwide have successfully employed the projectile cleaning system to achieve cost-effective and efficient high-rate cleaning of hoses, tubes, and pipes.
7. SUMMARY Internal contamination of tubes, hoses, and pipes reduces the operating efficiency of industrial systems that transmit fluids. Traditional cleaning methods, such as water jetting, solvent and chemical cleaning, and mechanical cleaning with liquid flushing, are often time consuming and expensive, use hazardous chemicals or solvents, or they use large volumes of water. The pneumatic projectile cleaning method overcomes these disadvantages. A pneumatic launcher propels the projectile, sized slightly larger than the internal diameter of the tube, into the tube. As the projectile travels through the tube, it removes the contamination deposited on the internal tube surface and forces it out of the tube. Tubes with internal diameters from 2 mm to 60 mm can be cleaned to ISO 4406 15/13/10 cleanliness levels. Cleaning can be accomplished very effectively in seconds. Automated systems can achieve cleaning rates as high as 3000 articles per hour. The method is applicable in a wide variety of industries.
ACKNOWLEDGMENT The author would like to thank Jody Mantell for help with the references.
DISCLAIMER Mention of commercial products in this chapter is for information only and does not imply recommendation or endorsement by The Aerospace
144
Developments in Surface Contamination and Cleaning
Corporation. All trademarks, service marks, and trade names are the property of their respective owners.
REFERENCES [1] H. Mu¨ller-Steinhagen (Ed.), Handbook of Heat Exchanger Fouling. Mitigation and Cleaning Technologies, Publico Publications, Essen, Germany, 2000. [2] NFPA 53, Recommended Practice on Materials, Equipment, and Systems Used in OxygenEnriched Atmospheres, National Fire Protection Association, Quincy, MA, 2004. [3] AIGA 005/04, Fire Hazards of Oxygen and Oxygen Enriched Atmospheres, Asia Industrial Gases Association, Singapore, 2004. [4] H.D. Beeson, S.R. Smith, W.F. Stewart (Eds.), Safe Use of Oxygen and Oxygen Systems: Handbook for Design, Operation, and Maintenance, ASTM Manual 36, second ed., ASTM International, West Conshohocken, PA, 2007. [5] J.M. Stolzfuss, Oxygen Systems, NASA White Sands Test Facility, Las Cruces, NM, 2010. http://www.nasa.gov/centers/wstf/laboratories/oxygen/index.html. [6] I.W. Chiang, B.E. Brinson, R.E. Smalley, J.L. Margrave, R.H. Hauge, Purification and Characterization of Single-Wall Carbon Nanotubes, J. Phys. Chem. B 105 (2001) 1157. [7] J.S. Kim, K.S. Ahn, C.O. Kim, J.P. Honga, Ultraviolet Laser Treatment of Multiwall Carbon Nanotubes Grown at Low Temperature, Appl. Phys. Lett. 82 (2003) 1607. [8] M.T. Martı´nez, M.A. Callejas, A.M. Benito, M. Cochet, T. Seeger, A. Anso´n, et al., Modifications of Single-Wall Carbon Nanotubes upon Oxidative Purification Treatments, Nanotechnology 14 (2003) 691. [9] B. Bundjemil, E. Borowiak-Palen, A. Graff, T. Pichler, M. Guerioune, J. Fink, et al., Elimination of Metal Catalyst and Carbon-Like Impurities from Single-Wall Carbon Nanotube Raw Material, Appl. Phys. A 78 (2004) 311. [10] M.P. Petkov, Non-Destructive Cleaning of Carbon Nanotube Surfaces: Removal of Organic Contaminants and Chemical Residue with Oxygen Radicals, Microsc. Microanal. 11 (Suppl 2) (2005) 1928. [11] N. Grobert, Carbon Nanotubes e Becoming Clean, Mater. Today 10 (2007) 28. [12] K.E. Hurst, A.C. Dillon, D.A. Keenan, J.H. Lehman, Cleaning of Carbon Nanotubes near the p-Plasmon Resonance, Chem. Phys. Lett. 433 (2007) 301. [13] A. Capobianchi, S. Foglia, P. Imperatori, A. Notargiacomo, M. Giammatteo, T. Del Buono, et al., Controlled Filling and External Cleaning of Multi-Wall Carbon Nanotubes Using a Wet Chemical Method, Carbon 45 (2007) 2205. [14] S.Y. Ju, W.P. Kopcha, F. Papadimitrakopoulos, Brightly Fluorescent Single-Walled Carbon Nanotubes via an Oxygen-Excluding Surfactant Organization, Science 323 (2009) 1319. [15] J. Pollak, M. Moisan, D. Ke´roack, J. Se´guin, J. Barbeau, Plasma Sterilisation within Long and Narrow Bore Dielectric Tubes Contaminated with Stacked Bacterial Spores, Plasma Proc. Polym. 5 (2008) 14. [16] R. Kohli, Alternate Semi-Aqueous Precision Cleaning Techniques: Steam Cleaning and Supersonic Gas/Liquid Cleaning Systems, in: R. Kohli, K.L. Mittal (Eds.), Developments in Surface Contamination and Cleaning. Removal of Particulate Contaminants Volume 3, Elsevier, Oxford, U.K., 2011, p. 201. [17] Palladium Membrane Hydrogen Purifiers, Outside-In or Inside-Out Flow e Which Design is Best for Compound Semiconductors? Technical Bulletin 0303, Johnson Matthey Gas Purification Technology, West Chester, PA, 2010. http://pureguard.net/cm/Library/Technical_ Bulletins.html.
Chapter | 4
Non-Aqueous Interior Surface Cleaning Using Projectiles
145
[18] Outside-In vs, Inside-Out: Which Method Produces Cleaner Hydrogen? White Paper, PowerþEnergy, Inc, Bucks County, PA, 2010. http://www.powerandenergy.com/papers/ outsidevsinside.html. [19] T. Kuppan, Heat Exchanger Design Handbook, CRC Press, Boca Raton, FL, 2000, pp. 393-422. [20] P. Watkinson, H. Mu¨ller-Steinhagen, M. Reza Malayeri (Eds.), Heat Exchanger Fouling and Cleaning: Fundamentals and Applications, Berkeley Electronic Press, Berkeley, CA, 2004. [21] A.G. Howell, G.E. Saxon, Condenser Tube Fouling and Failures: Cause and Mitigation, Power Plant Chemistry 7 (2005) 708. [22] H. Mu¨ller-Steinhagen, M. Reza Malayeri, P. Watkinson (Eds.), Heat Exchanger Fouling and Cleaning VII, Berkeley Electronic Press, Berkeley, CA, 2007. [23] B. Battat, W. Babcock, Understanding and Reducing the Effects of Contamination on Hydraulic Fluids and Systems, The AMPTIAC Quarterly 7 (1) (2003) 11. [24] B. Casey, How to Define and Achieve Hydraulic Fluid Cleanliness, Machinery Lubrication, March 2004. http://www.machinerylubrication.com. [25] W. Babcock, B. Battat, Reducing the Effects of Contamination on Hydraulic Fluids and Systems, Practicing Oil Analysis (November 2006). [26] J.B. Mordas, How Clean Does Your System Need To Be? Hydraulics & Pneumatics (June 2006). http://www.hydraulicspneumatics.com/200/Issue/Article/False/21746/Issue. [27] L. Badal, J. Sutherland, Hydraulic System Cleanliness and Equipment Warranty Compliance in the Construction Industry, White Paper, Chevron Products Company, San Ramon, CA, 2007. [28] B. Casey, Defining and Maintaining Fluid Cleanliness for Maximum Hydraulic Component Life, Plant Maintenance Resource Center, October 2007. http://www.plant-maintenance.com/ articles/hydraulic_fluid_cleanliness.pdf. [29] M. Moon, How Clean are Your Lubricants? Trends in Food Sci. Technol. 18 (2007) s74. [30] Taking Lubricant Cleanliness to the Next Level, Machinery Lubrication Magazine, January 2008. http://www.machinerylubrication.com/Read/1291/lubricant-cleanliness. [31] M. Moon, Lubricant Contaminants Limit Gear Life, Gear Solutions, June 2009. http:// www.gearsolutions.com/article/detail/5903/lubricant-contaminants-limit-gear-life. [32] C. Lee, Industrial Lubricants Reduce, Re-Use & Recycle, 2010, 27e33. http://reliabilityweb.com/index.php/articles/industrial_lubricants_reduce_re-use_recycle/. [33] B. Casey, Adding Hydraulic Oil e Without the Dirt, 2010. http://www.insidersecretstohydraulics.com/hydraulic-oil.html. [34] B. Casey, Preventing Hydraulic Failure, Electronic Book, 2010. http://www.preventing hydraulicfailures.com. [35] ISO 4406, Hydraulic Fluid Power e Fluids e Method for Coding Level of Contamination by Solid Particles, International Organization for Standardization, Geneva, Switzerland, 1999. [36] ISO 11171, Hydraulic Fluid Power e Calibration of Automatic Particle Counters for Liquids, International Organization for Standardization, Geneva, Switzerland, 1999. [37] ARP 598, The Determination of Particulate Contamination in Liquids by the Particle Count Method, SAE International, Warrendale, PA, 1986. [38] IEST-STD-CC1246D, Product Cleanliness Levels and Contamination Control Program, Institute of Environmental Science and Technology, Rolling Meadows, IL, 2002. [39] ASTM D6786 e 08, Standard Test Method for Particle Count in Mineral Insulating Oil Using Automatic Optical Particle Counters, ASTM International, West Conshohocken, PA, 2008.
146
Developments in Surface Contamination and Cleaning
[40] ASTM F312 e 08, Standard Test Methods for Microscopical Sizing and Counting Particles from Aerospace Fluids on Membrane Filters, ASTM International, West Conshohocken, PA, 2008. [41] SAE AS4059, Aerospace Fluid Power e Cleanliness Classification for Hydraulic Fluids, SAE International, Warrendale, PA, 2001. [42] Recommended Oil Cleanliness Codes for Components, Bosch Rexroth Filtration Systems GmbH, Lohr, Germany, 2010. www.boschrexroth.nl. [43] Oil Cleanliness Codes, Bosch Rexroth Filtration Systems GmbH, Lohr, Germany, 2010. www.boschrexroth.nl. [44] Selecting Target ISO Cleanliness Codes, Precision Filtration Products, Pennsburg, PA, 2010. www.precisionfiltration.com. [45] MIL-H-5606G, Hydraulic Fluid Petroleum Base: Aircraft, Missile, and Ordnance, U.S. Department of Defense, Wright-Patterson Air Force Base, Dayton, OH, 1994. [46] SE-S-0073, Space Shuttle Fluid Procurement and Use Control, Specification, National Aeronautics and Space Administration, Johnson Space Center, Houston, TX, 1995. [47] S.Ya. Kunin, Machine for Cleaning Pipes, Metallurgia 6 (1970) 395. [48] B.H. Herre, Investigation of Chemical Cleaning Procedures for Replacement Boiler Tubes, Report PPE-222-R, Pennsylvania Power and Light Company, Allentown, PA, September 1986. [49] D.W. Casella, Pneumatic Gun, U.S. Patent 4,974,277 (1990). [50] J.B. Fowler, Pneumatic Gun and Projectiles Therefor, U.S. Patent 5,329,660 (1994). [51] J.B. Fowler, Pneumatic Gun and Projectiles Therefor, U.S. Patent 5,555,585 (1996). [52] E. Schef, A Method and an Apparatus for Internal Cleaning of Pipes or Tubes, International Patent WO/1999/043450 (1999). [53] E. Schef, Method and Apparatus for Internal Cleaning of Pipes or Tubes, U.S. Patent 6,082,378 (2000). [54] P. Courville, M.L. Connell, J.C. Tucker, A.L. Branch, R.S. Tyre, The Development of a Coiled-Tubing Deployed Slow-Rotating Jet Cleaning Tool that Enhances Cleaning and Allows Jet Cutting of Tubulars, Paper 62741-MS, IADC/SPE Asia Pacific Drilling Technology Conference, International Association of Drilling Contractors (IADC), Houston, TX, 2000. [55] D. Menzie, Automatic Pneumatic Projectile Dispensing System, International Patent WO/ 2002/081109 (2002). [56] J.B. Unternaehrer, Automatic Pneumatic Projectile Launching System, International Patent WO/2003/086672 (2003). [57] B. Riley, System for Cleaning Gun Barrels, U.S. Patent 6,668,480 (2003). [58] J. Svenson and R. Axelsson, Applicator for Inside Cleaning of Pipe and Hose, U.S. Patent 6,578,226 (2003). [59] S. Spielmann, An Overview: Chiller Tube Cleaning, 2003.www.contractingbusiness.com. [60] D.J. Smith, Condenser Cleaning Saving $1 Million Annually, Power Engineering, May 2004. [61] G. Saxon, A. Howell, The Practical Application and Innovation of Cleaning Technology for Condensers, Energy-Tech.com, ASME Power Division Special Section, (August 2005) pp. 19e26. [62] A. Pivovarov, Cleaning of Submerged Surfaces by Discharge of Pressurized Cavitating Fluids, U.S. Patent Application 2006/0151634 (2006). [63] J. Laughlin, T. Hansen, Nuclear Plant Efficiently Removes Calcium Carbonate from Condenser Tubes, Power Engineering, July 2007.
Chapter | 4
Non-Aqueous Interior Surface Cleaning Using Projectiles
147
[64] L.A. Martini, Slow Rotating Fluid Jetting Tool for Cleaning a Well Bore, U.S. Patent 7,314,083 (2008). [65] Y. Tabani, M.E. Labib, Method for Cleaning Hollow Tubing and Fibers, U.S. Patent 7,367,346 (2008). [66] D. Feng, C. Huang, K. Zhou, P. Wang, J. Liu, S. Li, Crucial Technology Research on Pipeline Jet Cleaning, in: C. Hong, H. Liu, Y. Huang, Y. Xiong (Eds.), Intelligent Robotics and Applications, Springer Verlag, Berlin and Heidelberg, Germany, 2008, pp. 1137e1144. [67] Disposable Endoscopic Cleaning Brushes, Clinical Choice, Greensboro, NC, 2010. www.clinicalchoice.com. [68] Tube Brushes, Schaefer Brush, Waukesha, WI, 2010. www.schaeferbrush.com. [69] Alka Technical Solutions, Bergamo, Italy, 2010. http://www.alka-srl.com/Home.html. [70] B. Riley, Product Review of the Contamination Eliminator System, in Proc. 24th Mr. Clean Conference, Princeton, NJ, J. Dennis (Ed.) 1998, pp. 300e305. [71] H. Landolt, COMPRI Tube Clean System. The Economical Way to Clean Heat Exchangers, in: H. Mu¨ller-Steinhagen (Ed.), Handbook of Heat Exchanger Fouling. Mitigation and Cleaning Technologies, Publico Publications, Essen, Germany, 2000, pp. 70e74. [72] JetCleaner Cleaning System for Hose and Tube, Eurocomp AB, Avesta, Sweden, 2009. http:// www.eurocomp.se. [73] Hose, Tube & Pipe Cleaning Technology, COMPRI Tube Clean SA, Woodside, South Australia, 2010. http://www.tubecleansa.com.au/ and http://www.compri.com.au/. [74] Hose and Tube Cleaning, Ultraclean Technologies Corporation, Bridgeton, NJ, 2010. http:// www.ultracleantech.com/. [75] Tube Clean GmbH, Hinwil, Switzerland, 2010. http://www.compritubeclean.com/de/ produkte.html. [76] MegaCleanÔ Hose and Tube Cleaning System, Gates Corporation, Denver, CO, 2010. [77] V. Srimongkolkul, Hose Cleanliness Evaluation Report, Oil Pure Technologies, Kansas City, MO, 2001. http://www.ultracleantech.com/. [78] G.-J. Tuschewitzki, C. Schell, Hygienisch-mikrobiologische Bewertung des Rohr- und Schlauchreinigungssystems COMPRI Tube Clean System, Report W-187408-10-SI, Hygiene-Institut des Ruhrgebiets, Gelsenkirchen, Germany, 2010.
This page intentionally left blank
Chapter 5
Electrostatic Removal of Particles and its Applications to Self-Cleaning Solar Panels and Solar Concentrators M.K. Mazumder,1 R. Sharma,2 A.S. Biris,3 M.N. Horenstein,1 J. Zhang,3 H. Ishihara,3 J.W. Stark,1 S. Blumenthal1 and O. Sadder1 Department of Electrical and Computer Engineering, Boston University, Boston, MA 02215, USA, 2Renewable Energy Technology Program, Arkansas State University, Jonesboro, AR 72467, USA, 3Department of Applied Physics, University of Arkansas at Little Rock, Little Rock, AR 72204, USA
1
Chapter Outline
1. Introduction 150 2. Solar Power Potential 151 and the Global Energy Needs 3. Atmospheric Dust and 154 Its Deposition on Solar Panel 4. Loss of PV Output Power 155 Caused by Dust Deposition 5. Electrostatic Charging of 157 Dust Particles 6. Dust Deposition Process: 158 Effects of Size and Charge Distributions 7. Transmission Loss Due to 160 Atmospheric Dust 8. Experimental Studies on 162 Solar Panel Obscuration by Dust Deposition Developments in Surface Contamination and Cleaning. Copyright Ó 2011 Elsevier Inc. All rights reserved.
9. Effect of Microstructural 163 Deposition Pattern: Particle Size, Shape, and Electrostatic Charge Distributions 10. Removal of Dust From 165 Solar Panels Using Low-Power Electrodynamic Screens 11. Trajectories of Charged 170 Particles on the Electrodynamic Screen 12. Dielectrophoretic Force 175 13. Tribocharging of Particles 177 14. Removal of Uncharged 178 Conducting Particles 15. High-Voltage Three-Phase 179 Power Supply for the Electrodynamic Screen 16. Testing of the 184 Electrodynamic Screen 149
150
Developments in Surface Contamination and Cleaning
17. Measurement of 190 Maximum Power Point Operation of the PV System With EDS 18. Testing the Solar Panels 191 Integrated With EDS for Maximum Power Point Operation
19. Results 193 20. Summary and Conclusions 196 References 197
1. INTRODUCTION By the mid-twenty-first century, it is anticipated that photovoltaic (PV) plant installations will be widely used providing electrical power up to 25% of the total global energy needs. These installations will likely be located more in the desert and semi-arid regions not useful for farming or other commercial applications. Nearly one-third of the land areas of the world belong to these regions. It has been estimated that if only 4% of the total usable desert areas are utilized for PV power plants to harvest solar radiation, the current total global energy needs can be completely met with negligibly small CO2 emission compared to the current global warming threat imposed by greenhouse gas emissions [1]. While inexhaustible for the foreseeable future, solar energy is a low power density (z1.0 kilowatt (kW) per square meter) source. Installations of megawatt (MW)-scale solar power plants require large areas of land, in locations where solar irradiance is high throughout the year. For example, a 125 MW solar PV installation will require 2.6 km2 area of land. Desert regions are ideal for large-scale photovoltaic and photothermal (PT) installations where conventional power plants are not built because of the lack of water supply needed for plant operation. PV systems are scalable; their power-generating capacity can range from milliwatt (mW) level (used for hand-held calculators) to MW to gigawatt (GW) levels for industrial grid connected power generation. This is possible since all PV systems are modular in construction. Currently, single crystalline silicon (c-Si) and multi-crystalline silicon (mc-Si) solar panels comprise more than 80% of all solar PV installations [2]. Each solar module consists of approximately 36 c-Si solar cells, each cell providing only about 1.5 W. Each solar panel consists of one or more modules and the number of panels used in a solar PV array depends upon the total power output capacity. The solar arrays can operate independently or they can be banked together to generate power at kW to GW scales. For instance, a 1.0 GW PV system will require approximately 6 million modules mounted on solar panels and framed on the PV arrays; the land area requirement for such an installation is approximately 23 km2.
Chapter | 5
Electrostatic Removal of Particles and its Applications
151
One of the problems of desert operation of large-scale solar PV systems is the airborne dust that often deposits on the panels. Efficient operation of solarPV installations requires clean panels for collecting solar radiation with a high efficiency [3]. Dust deposition on solar panels obscures solar flux, significantly reducing the efficiency of the systems. An integrated electrodynamic screen (EDS) on each solar panel can provide automatic and continuous removal of dust from solar panels without requiring water or any moving parts. The EDS technology was developed under a National Aeronautics and Space Administration (NASA) project for protecting solar panels on the Moon and Mars [4]. Reported experimental studies show that 95% of dust deposited on the panels can be removed by EDS in less than two minutes, drawing less than 2% of the panel power output during the cleaning period. The dust particles are removed by electrostatic and dielectrophoretic forces. The self-cleaning solar panels with EDS are effective for both charged and uncharged dust particles. We present here a brief review of the need for dust mitigation in terrestrial and space applications of solar panels, basic principles of the electrodynamic dust removal process, electrical power supply needed for the EDS operation, test procedures used for evaluation of the solar panels, and a brief outline of the ongoing research for increasing the energy yield of solar panels. Potential applications of self-cleaning solar panels in PV systems, particularly in arid and semi-arid regions, are included, and the economic advantage in payback for the added cost is examined.
2. SOLAR POWER POTENTIAL AND THE GLOBAL ENERGY NEEDS The average total irradiance from the sun received outside the Earth’s atmosphere is approximately 1366 W/m2 (solar constant). Attenuation by the atmosphere is measured by air mass (AM) [2,5]. For example, AM 0 represents solar irradiance received outside the Earth’s atmosphere and AM 1.0 represents the atmospheric attenuation of solar irradiance when solar radiation is incident vertically in a clear sky (at an angle 90 with respect to the horizon). The average solar radiation incident on a horizontal plane on the ground depends upon the angle at which the rays strike and on the latitude of the location. The solar energy received at any location varies throughout the day. The standard terrestrial solar radiation spectrum used for averaging is at AM 1.5 (corresponding to the sun being at an angle 42o). The atmospheric transmission loss at this angle reduces irradiance to about 900 W/m2; however, for convenience the accepted standard terrestrial irradiance is considered as 1000 W/m2 [5,6]. All solar panels are tested at this level of irradiance striking the cells at normal incidence at a temperature 298 K. The average insolation (incident solar radiation) data in different parts of the world are available from satellite measurements such as Landsat. The averaging process takes into account that direct solar radiation is received only
152
Developments in Surface Contamination and Cleaning
during the day. At any given moment half the planet does not receive any solar radiation; the average extraterrestrial flux incident on a unit surface area is onefourth the solar constant, which is approximately 342 W/m2. Average global irradiance received on the surface of the Earth varies from less than 100 W/m2 at high latitudes to more than 300 W/m2 at the sunniest places. The annual insolation on a horizontal plane is maximum in the continental desert areas between latitude 25oN and 25oS of the equator [7], and it falls off toward both the equator and towards the poles. In the equatorial regions, cloud cover contributes considerably to the variation of insolation, whereas seasonal variations cause major fluctuations of irradiance in the northern climates. Considering these variations, the mean annual horizontal global irradiation data in energy units of kWh (1 kWh ¼ 3.6 106 J) are available; for example, it is highest in the Sahara Desert at 2685 kWh/m2/year, followed by the Great Sandy Desert at 2343 kWh/m2/year, the Thar Desert at 2179 kWh/m2/year, and the Gobi Desert at 1701 kWh/m2/year. Globally, harvesting solar radiation using PV and PT systems is the fastest-growing industry in the energy sector; the annual average growth is in excess of 35% beginning in the year 2000, and most of the growth is taking place in Europe, Japan, China, the US, and India. Energy consumption correlates roughly with the gross national product and the climate. Per capita usage of energy in Japan and in Europe is 6 kW; in the US it is 11.4 kW, in China 1.6 kW, and in India 0.7 kW. The available solar power is approximately 120 000 TW (1 terawatt (TW) ¼ 1012 W). The average global power consumption recorded in 2008 was approximately 15 TW and the corresponding energy consumption was 474 exajoules (474 1018 J); about 90% of the energy is currently derived from fossil fuels. Considering a future world population of 10 billion with an average power need per person of 10 kW, the total power need will be 1011 kW or 100 TW. Ideally, solar power can clearly meet this future global power need. To meet this need in its entirety, PV installations, using solar panels with 14% efficiency, will require approximately 3.5 105 km2 of land area located in regions of high insolation. If one-fifth of all seven desert regions of the world are used for PV installations, the total available area will be in excess of 2 106 km2. Thus, only a fraction of the arid and semi-arid areas is needed to meet the foreseeable global power needs. Vast areas of barren deserts remain unused where vegetation cannot grow and solar irradiation level is high throughout the year. Approximately 20e50% of the desert areas are suitable for PV installations not considering the hilly terrains and sand dunes. Most large-scale PV installations are located on hard ground covered with gravel and sand. The ground is least disturbed by the installations except for the mounting supports used for the large number of arrays. The arrays are tilted (often at the latitude angle) with respect to the horizontal plane and are spaced apart from each other to avoid shading. The space factor, as it is called, is usually 50%. Thus, the land area needed is
Chapter | 5
Electrostatic Removal of Particles and its Applications
153
slightly more than twice the area of the PV panels. PV systems are robust, have low operating and maintenance costs, and the solar panels usually last more than 25 years. One of the advantages of the PV system is that it can support irrigation; solar panels are now extensively used for this purpose in India, and in many parts of the world where grid-connected electric power lines are not available. Stand-alone PV systems can also be used for desalination in coastal areas where clean water is scarce. Availability of clean water is considered to be a major health threat in many parts of the world. PV systems in remote semi-arid regions can provide electricity for water irrigation and desalination of sea water by the reverse osmosis process [1]. In the US, the total electrical power production in 2008 was close to 1.0 TW and the total energy consumption in that year was about 4100 TWh. According to the US Department of Energy (DOE) estimate (2010), the total US solar PV potential is 206 TW. For comparison, power production capacities of other renewable sources are: wind 10 TW (on shore and off shore), geothermal 0.04 TW, hydropower 0.014 TW, and biomass 0.08 TW. The installed capacity (in watts) of a PV system is the product of local solar average irradiance (W/m2), total area of the PV panels (m2), PV conversion efficiency (varies from 6% for amorphous Si (a-Si) to 14% for c-Si solar panels), and the plant performance ratio (typically 70%). The Mojave Desert, with an area of approximately 75 000 km2 and less than 25 cm annual rainfall, has a huge potential for solar power plants. Currently, the combined solar-energy-generating systems at Mojave have a capacity of 354 MW. One of these PV systems is the Nellis PV power plant with 14 MW capacity producing energy of 30 106 kWh/year. By 2011, another PV power plant (Mojave Solar Park) with a capacity of 553 MW will become operational. This new plant will be installed over an area of 24 km2. The next large-scale PV system (AV Solar Ranch One), to be commissioned in 2013, will add 230 MW to the solar power plant capacity with an estimated energy delivery of 600 million kWh/year. Once operational, the plant will supply electrical energy to 70 000 homes. The estimated generation cost of electricity of the AV Solar Ranch One is not expected to exceed $0.13 per kWh. Currently, the national average of electricity cost to the customer is approximately $0.10 per kWh. In some States like Massachusetts, the electric utility cost is about $0.19 per kWh. If 25% of the total electric power in the US is to be produced from solar PV systems, and one-half of all US solar power is to come from the Mojave Desert, it will require approximately 6250 km2 for the PV plants with a total generating capacity of 125 GW. The c-Si solar modules are the most efficient, but also the most expensive, at just under $4.00/Wp (peak watt); mc-Si solar modules are slightly less efficient, but are also less expensive, about $2.50/Wp. Currently, the thin-film PV panels are the least expensive, close to $1.00/Wp. Since both the lifetime and the efficiency vary depending upon the type of
154
Developments in Surface Contamination and Cleaning
module used, the area needed for the PV installation, and the cost of the PV solar system, depend upon the selected module. The module price generally represents 50% of the total installation cost. With c-Si modules, the total installation cost for PV plants producing 125 GW (12.5% of the total electric power in the US) will be approximately $500 billion. This is about ten times the cost of the Gulf of Mexico oil leak disaster of 2010. The major goal of the PV manufacturers is to reduce the cost of the solar panel to $1.00/Wp, with high efficiency and longevity. When this goal is achieved, the solar PV system will be highly competitive with conventional fuels, even without considering the environmental benefits of renewable energy systems. Very large-scale PV systems will be sustainable if each installation generates enough revenue to provide resources for installing two plants of the same size during the life span of the first plant. Like all conventional power plants, one of the most important economic factors is the cost per kWh in PV installations. Most PV systems are designed based on the type of solar cells to be used (c-Si, mc-Si, thin film CdTe/CdS, or copper indium selenide (CIS/CdS)), their performance specified under standard testing conditions (STC: irradiance 1000 W/m2, AM 1.5, temperature 298 K), their cost, and their anticipated life time. However, the actual energy yield in a PV plant depends upon (1) the average annual illumination intensity incident on the solar modules in the area where the system will be actually installed, and (2) the operating temperature of the modules. The output power is not linearly proportional to the incident solar irradiation since the photo-conversion efficiency of the cells depends upon the intensity of incident light and the ambient conditions. When there is a loss of illumination intensity caused by obscuration of light by dust layer on the panels, there are three adverse effects: (1) the power output is reduced; (2) there is a decrease in conversion efficiency; and (3) formation of hot spots and dead cells if the modules are partially blocked by dust layer deposits. When some of the cells are covered by dust, the shaded cells do not generate power to match the other cells; rather, they act as a dead load on the working cells. As a result, the temperature of the shaded cells increases, forming hot spots. Unless efficient protection devices are used to prevent the formation of hot spots, the modules can get permanently damaged. In the following sections, we describe the loss of transmittance of the incident solar radiation to solar modules and methods for protecting them.
3. ATMOSPHERIC DUST AND ITS DEPOSITION ON SOLAR PANEL About 30e50% of the extraterrestrial irradiance is lost by scattering and absorption by the atmosphere before reaching the surface of the Earth. The mean annual energy available on a horizontal plane in different parts of the world has been measured by using satellites and the measured data are
Chapter | 5
Electrostatic Removal of Particles and its Applications
155
published by the International Energy Agency (IEA). Solar energy has regular daily and annual cycles due to the Earth’s rotation and inclination of its axis with the ecliptic plane as the Earth moves around the sun [7]. Deposition of dust on the panels, particularly when these are installed in dusty areas, deserts, or along highways, could severely minimize solar-topower output efficiency. In desert regions, where solar radiation is intense and available almost throughout the year, obscuration of solar radiation, caused by dust and by the shadowing effect due to passing clouds, is a significant problem. Solar panel obscuration by dust is caused by: (1) suspended dust in the atmosphere (aerosol) that is directly in the optical path of incoming radiation to the solar panels, causing extinction of light even when the cells are relatively clean, and (2) extinction of light due to dust deposited on the panel surface. Both conditions occur during high dust concentration in the atmosphere. Dust storms can completely obscure solar radiation; for example, a dust storm in Sydney, Australia, darkened the sky for several days with red dust in 2009. Our analysis and experimental studies reported here are limited to the obscuration by dust deposition on the panel and removal of the deposited dust by electrodynamic screens. Experimental data on the concentration of suspended atmospheric particulate matter (PM) with particle size smaller that 10 mm in diameter (referred to as PM10) in different regions of the world are available from remote measurements made by the US National Oceanic and Atmospheric Administration (NOAA) and by US Environmental Protection Agency (EPA) [8,9]. Airborne particles in this range (PM10) at high concentration are of major concern because of health risks and visibility degradation. Maximum light attenuation comes from the particles in the range 0.3e0.6 mm in diameter since their dimension is comparable to the wavelength of light in the visible region. The dust deposition rates in different regions of the world have been measured under normal atmospheric conditions. The rate of dust deposition on solar panels, particularly in the PM10 range, is useful to predict possible obscuration of light. For example, some of the deserts in regions of the Middle East have an average dust deposition rate of approximately 0.36 g/m2/day; the Negev Desert has an average dust deposition of 0.5 g/m2/day; and the Mojave Desert has a relatively low dust deposition rate of 0.17 g/m2/day. The dust deposition rate varies over a wide range [8,9]. Figure 5.1 shows a dust devil forming in the dry lake area of the Mojave Desert. Such dust devils can obscure sunlight completely and can deposit dust on the panels with high mass concentration.
4. LOSS OF PV OUTPUT POWER CAUSED BY DUST DEPOSITION Atmospheric aerosol particles include mineral dust stirred up from the ground, salt particles from sea spray as the droplets evaporate, anthropogenic particles
156
Developments in Surface Contamination and Cleaning
FIGURE 5.1 Dust devil formation on the dry lake area of the Mojave Desert. Photo courtesy of Creative Commons Corporation, San Francisco, CA. http://www.animalu.com/pics/photos.htm Jeff T. Alu
such as particulate pollutants discharged from the power plants, biological particles, photochemically produced particles of sulfates and nitrates, soot particles from fires and vehicle exhaust, and road dust. As most of these particles are produced at or near ground level, their concentration decreases almost exponentially as a function of height, and most of the particles are within a height of about 1.5 km above ground level. In the arid and semi-arid areas, airborne particles are mostly metal oxides (SiO2, Al2O3, and other oxides) and are likely to be of high electrical resistivity. Both concentration and the size distribution of the atmospheric particles vary widely with respect to the geographical location and time. Particle size distribution varies from nanometers to several micrometers. In the arid and semi-arid areas, there are often dust storms which arise due to wind erosion
Chapter | 5
Electrostatic Removal of Particles and its Applications
157
from a dry surface. Wind gusts pick up dust from dry surfaces of the desert by saltation and transport the dust, often as a thick cloud, from one place to another [10]. Like deserts, multi-year droughts in many parts of the world are the primary sources of dust cloud transport over a large distance. Often, the concentration of these dust clouds can be dense enough to darken the sky over a large area and blanket everything on land with a thick layer of dust. Major dust storm occurrences in many parts of the world since the 1930s have been documented by meteorologists [10,11]. Several places including North America, East Asia, Australia, Hong Kong, Taiwan, Japan, and South Korea were affected. For example, dust storms in 1971, 2004, and 2007 covered several cities of the US. Australian dust storms covered many cities with a blanket of reddish orange dust in 1983, and again in 2009. A large part of Pakistan was severely affected by a dust cyclone and caused many deaths in 2007. Similarly, sand storms were recorded in many parts of the Middle East. Volcanic eruptions, such as the one that occurred in Iceland in 2010, caused deposition of ash in many parts of Europe. Large-scale PV installations can be severely affected by such dust deposition on solar panels. Dust storms in desert areas are frequently followed by light rains which cause a thin layer of mud to cover the exposed panel surface. Rapid cleaning of the dust layer, either during or immediately after the storm, can minimize the cleaning problem. Many PV panels installed in arid and semi-arid areas are cleaned with water; the process is labor intensive and costly, particularly where clean water is scarce and expensive. Most PV panels are constructed with tempered borosilicate glass as the front surface which makes the panels easy to clean with water. However, there is a recent trend to have the front surface of the glass textured to minimize reflection losses and to trap light. Deposition of microscopic dust particles on the textured surface may defeat these advances.
5. ELECTROSTATIC CHARGING OF DUST PARTICLES Wind erosion of dust from a surface under dry conditions is likely to cause triboelectric charging of the particles during the lift-off from the surface. The wind speed in desert areas can reach up to 18 m/s. As the dust particles become airborne from the surface, the inter-particle separation process contributes to bipolar charging of the dust. Windblown dust, in the size range from sub-mm to 50 mm in diameter, rising from desert surfaces, continues to add to the atmospheric dust load. Sedimentation is the primary mechanism of dust deposition on the solar panels. The daily buildup of dust layers on the panels depends on (1) the atmospheric dust concentration, (2) the settling velocity of dust particles, (3) the adhesion of the dust layer on the surface, (4) aggregation properties of dust deposits, and (5) dust removal by wind.
158
Developments in Surface Contamination and Cleaning
The adverse effects of dust on the solar panel performance depend upon: (1) density of the suspended dust cloud (mg/m3) in the atmosphere; (2) size and shape distribution of the particles; (3) optical properties (scattering and absorption) of dust; (4) electrostatic charge; (5) adsorbed moisture on the surface of the particles; and (6) surface mass density (kg/m2) of the dust layer deposited on the solar panel surface. Wind and rain can clean the surface and may provide sufficient cleaning where rainfall is frequent throughout the year. In many parts of the world rainfall is seasonal, such as the monsoon season of India, where atmospheric dust concentration is high during the other seasons of the year. The presence of cloud cover and suspended dust in the atmosphere cause significant obscuration of sunlight even when the glass-covered solar panels are clean. Similarly, a significant transmission loss occurs from the dust layer deposited on the front glass surface of the solar panel even when the sky is clear. In general, both phenomena contribute to the attenuation of incident radiation.
6. DUST DEPOSITION PROCESS: EFFECTS OF SIZE AND CHARGE DISTRIBUTIONS The atmospheric particles vary in size from 0.001 to 100 mm in diameter. The number concentration of the particles decreases rapidly as the diameter increases. Particles smaller than 1.0 mm in aerodynamic diameter remain suspended in the atmosphere much longer compared to the coarser particles, since the settling velocity of the particles varies as the square of the particle diameter. Particle deposition on the solar panel surface is caused by four deposition mechanisms: (1) gravitational settling; (2) deposition by diffusion; (3) electrostatic deposition; and (4) inertial and diffusive deposition from turbulent flow. Of these, gravitational settling plays the primary role. The gravitational settling velocity VTS can be written as [12]: VTS ¼ rp ðdp Þ2 gCc =18 h
(5.1)
where rp is the particle bulk density, h is the viscosity of the gaseous medium, Cc is the Cunningham slip correction factor, and g is the acceleration due to gravity. The correction factor Cc can be taken as 1.0 for particles larger than 1.0 mm. Particle deposition by diffusion is significant for particles smaller than 1.0 mm. The deposition rate J due to particle diffusion per unit area of surface for a time period t can be expressed as: J ¼ N0 ðD=ptÞ1=2
(5.2)
The diffusion coefficient D is given by: D ¼ kTCc =3phd
(5.3)
Chapter | 5
Electrostatic Removal of Particles and its Applications
159
with k being the Boltzmann constant and T is the absolute temperature. The diffusion coefficient D is independent of the density of the particles. While the mass fraction of the fine particles that deposit on solar panels is small compared to that of the coarse particles, the extinction coefficient Qext and the specific surface area are higher for small particles. The deposition rate due to settling increases as d2, while the rate of deposition due to diffusion increases inversely with particle diameter d. When the dust is being carried by high winds or dust storms, diffusion and inertial deposition from turbulent flow can occur on the panels. For particles larger than 1 mm in diameter, their impaction on the panel surface could be one of the dominant mechanisms of deposition. However, the wind shear force is also one of most effective removal processes in a turbulent flow. Both deposition and removal occur during a dust storm. The removal efficiency depends upon the relative strength of the force of adhesion between the particles and the panel surface, and the shear force at the boundary layer. The shear force is proportional to d2, whereas the adhesion force is directly proportional to d. Therefore, as the particle size decreases, particularly for sub-mm particles, the force of adhesion becomes much stronger than the liftoff force of the wind. Electrostatic forces also play an important role in the deposition process for small particles carrying a high charge-to-mass ratio. When a charged particle comes close to a grounded metal or to an insulating surface supported on a grounded conducting substrate, it experiences electrostatic attraction to the surface due to the image force. If the surface has an electrostatic charge, there will be Coulomb forces of attraction or repulsion. These forces vary inversely as the square of the distance between the particle and the surface. Therefore, electrostatic forces are effective only when the distance is small, typically when a charged particle is within a few millimeters from the surface. The Coulomb force arises when the panel surface is electrostatically charged, often caused by the previously deposited charged particles. The electrostatic force becomes dominant when the charged particle reaches the vicinity of the surface driven by other forces, such as gravity, diffusion, and turbulent transport. Once the charged particle is close to the surface, the electrostatic force of attraction or repulsion depends upon the polarity of the charge of the particle approaching the panel and the charge distribution of the deposited dust layer on the panel. The charge distribution of the particles can change the microstructural deposition pattern, and hence the opacity, of the surface dust layer. The charge distribution provides the details of both the polarity and the magnitude of charge of individual particles. For particles with unipolar charge distribution, the electrostatic deposition forces are likely to make the dust layer more uniform on the panel with strong adhesion to the panel surface. As each particle comes close to the surface, it experiences electrostatic attraction to the panel caused by the image force and repulsion by the neighboring
160
Developments in Surface Contamination and Cleaning
particles deposited with charge of the same polarity. This process makes the particle land on the surface where the repulsion force is weak because of the least number of deposited particles with the same polarity. The localized distribution of the electrostatic field causes the dust layer to be relatively uniform resulting in a high packing density of dust on the surface. Even a thin layer of densely packed dust can cause a high obscuration of incident radiation. If the particles are bipolarly charged, the particles aggregate readily on the surface because of the inter-particle attraction. The aggregates are often in the form of dendrites, microscopic tree-like structures, of particles loosely bound to the panel surface. Such a structure often allows a major component of the forward scattered light to enter the solar panel, and, as a result, the transmission loss is reduced. The loosely bound particles can become re-suspended in air and be removed by wind. Thus, the particle size distribution and the packing density of dust layer on the panel have a strong influence on the optical transmission loss.
7. TRANSMISSION LOSS DUE TO ATMOSPHERIC DUST A parallel beam of sunlight passing through an atmosphere containing suspended particles is assumed to be normally incident on a solar panel. If we assume the particles are monodispersed with respect to size, spherical in shape with diameter d, Ap is the projected area of each particle towards the incident beam, Qext is the extinction coefficient (efficiency) of individual particles and L is the atmospheric path length of the incident beam, the ratio of the intensity of the transmitted beam (I) to that of the incident beam (I0) can be written as [12]: ðI=I0 Þ ¼ expðN0 Ap Qext LÞ
(5.4)
where N0 is the number of dust particles per unit volume in the optical path of the incident solar radiation. The extinction efficiency Qext of a particle is the sum of its scattering efficiency Qs and the absorption efficiency Qa. The extinction efficiency varies from 0 to 5, depending on the particle size parameter a (a ¼ pd/ l), the particle shape, and the complex reflective index m (m ¼ m0 e m0ai) with m0 being the real component of the index of refraction, m0a being the imaginary part, and i ¼ (e1)1/2. Based on the diameter d of the particles, and the refractive index m for a given wavelength l, we can calculate the size parameter a, and determine the value of Qs, Qa, and Qext from the Mie scattering theory [13]. Since the particle cloud density N0 is likely to be a function of height L vertically above ground, we can consider a similar concentration of suspended particles as a function of length perpendicular to the surface of the solar panel tilted at an angle. We can express the ratio (I/I0) with respect to the total number of particles suspended in a dust column (in the path of the incoming
Chapter | 5
Electrostatic Removal of Particles and its Applications
161
light over a unit area (1 m2) of the solar panel). Writing N ¼ N0L, eq. (5.4) can be rewritten as: ðI=I0 Þ ¼ expðNAp Qext Þ
(5.5)
where N is the number of particles suspended per unit area (1 m2) normal to the panel in the direction of the incident beam. Since the product NApQext is dimensionless, it is often termed as an ‘optical depth’ (s) and is given by: s ¼ NAp Qext
(5.6)
The ‘optical depth’ represents the loss of light intensity during its transmission through the optical path due to scattering and absorption in the atmosphere. Experimental data on the optical depth, with respect to the vertical path in the atmosphere above ground, are available for many sites. Since the solar panels are usually tilted (southward in the northern hemisphere) with respect to the vertical plane, and are often fixed in that position, the optical depth for a path of radiation, incident at angle q with respect to the normal direction, can be expressed as s0 ¼ (AM)s, where AM is the air mass, defined as AM ¼ 1/cosq, where q is the zenith angle. Thus, the ratio (I/I0) becomes: ðI=I0 Þ ¼ es
1
(5.7)
The optical path s (for q ¼ 90o), under normal atmospheric conditions, has been measured in many areas. In a relatively clear sky with the particle mass median diameter in the range of 1.0 to 2.0 mm, the obscuration by the suspended particles in the atmosphere is very low. However, in desert areas settling of particles on the panel surface is most likely to be a significant factor in the transmission loss for the coarse dust with particle diameter larger than 1.0 mm. If we assume that the average particle diameter is 1.0 mm and if we consider the peak of the solar radiation spectrum is at a wavelength l ¼ 0.52 mm, then a ¼ pd/l ¼ 6.8. For absorbing particles with a equal to or larger than 5.0, Qext approaches its limiting value of 2. For particle diameters in the range 0.2 to 1.0 mm, Qext oscillates around 2.0 with the maximum value approaching 5.0. The extinction coefficient for nonabsorbing spheres with refractive index between 1.33 and 1.5 oscillates around the value of 2.5 as a(m e 1) increases until it reaches its limiting value of 2.5. For irregular particles in the size range between 1 and 4 mm in equivalent diameter, both maximum and minimum values for Qext are close to 2.0. For a given particulate mass concentration in the atmosphere, maximum light extinction occurs for particles with diameters in the range between 0.3 and 0.8 mm, where Qext reaches its maximum value around 5.0. Scattering of light reaches a maximum level when the diameter of the particles is comparable to the wavelength of the radiation. In this size range of mineral dust, the transmission loss is due primarily to the scattering effect. True absorption occurs
162
Developments in Surface Contamination and Cleaning
when the complex refractive index m of the particles, such as soot, has a significant imaginary component m0a where the absorption effect predominates scattering. For a monolayer of dust particles of diameter d deposited on the surface of a solar panel, the extinction of light can be calculated using eq. (5.5), where N is the number of deposited particles per unit area. For polydispersed particles, the product of NAp Qext can be written as: X s ¼ NAp Qext ¼ pðN0 Þi ðdi =2Þ2 ðQext Þi (5.8) i
where (N0)i is the particle number concentration with diameter di having extinction efficiency (Qext)i. For particles larger than 1.0 mm in diameter, we can assume Qext is close to 2.0. When multiple layers of particles deposit on the panel surface, estimation of transmission loss is difficult since both size distribution and orientation of the particles are not known. One approach for estimating the transmission loss in such a case is to treat the loss of individual layers and multiply the fractional losses for each layer. For a high number density of particles, multiple scattering effects will have to be taken into account and polydispersed particles with a high packing density may completely obscure the incoming light.
8. EXPERIMENTAL STUDIES ON SOLAR PANEL OBSCURATION BY DUST DEPOSITION In order to make quantitative measurements on the obscuration of a solar panel due to the deposition of dust layers, experimental studies were performed using volcanic dust samples, containing primarily of SiO2, Al2O3, TiO2, and other mineral oxides. The test dust sample was classified by using a sieve to remove particles larger than 40 mm in diameter. A pneumatic dust dispersion device was constructed and used to disperse the test dust fairly uniformly over a glass cover. The particle size analysis of the test dust was carried out using a particle size analyzer and the measured size distribution is shown in Figure 5.2. The d10, d50, d90 values were found to be 1.22 mm, 9.06 mm, and 38.45 mm, respectively. A test chamber was used for studying obscuration by dust deposition [14]. A single crystal Si solar cell was used with a glass plate that covered the solar cell. Before each test, the dust sample was dried in an oven at a temperature of 373 K for at least 24 hours. For each experimental run, a small amount of dust was dispersed using the dust dispersion device to form a dust cloud within the test chamber. The dust particles were then allowed to settle on the glass plate. A xenon lamp was used to illuminate the solar cell through the glass cover. The spectral radiation of the solar lamp approximately simulated solar radiation for the test purpose. The output power of the solar cell was measured with and without the deposited dust layer. The mass concentration in mg/cm2 deposited
Chapter | 5
Electrostatic Removal of Particles and its Applications
163
FIGURE 5.2 Particle size distribution (shown in dark solid steps) of the test dust as measured by using a MicrotracÒ particle size analyzer. The line connecting the dots shows the cumulative size distribution plotted as a function of particle diameter in mm
on the glass cover was measured with a microbalance. The uniformity of the dust layer was examined by using an optical microscope. The power output was plotted as a function of the mass concentration of the dust layer deposited on the glass cover. The normalized power output of the solar cell, plotted as a function of the surface mass density of dust deposited on the front glass plate placed over the solar cell (Fig. 5.3) shows an exponential decay of power output as surface dust loading increased. The decay curve agrees with the theoretical prediction of eq. (5.5). When the surface mass density of the dust deposit increased to 1.5 mg/cm2, the output power decreased by more than 90%.
9. EFFECT OF MICROSTRUCTURAL DEPOSITION PATTERN: PARTICLE SIZE, SHAPE, AND ELECTROSTATIC CHARGE DISTRIBUTIONS The microstructural characteristics of the deposited dust on the glass surface were examined using an optical microscope. It was found that the dust particles on the surface formed dendrites (tree-like structures of agglomerates). This deposition pattern can be explained based on the bipolar electrostatic charge distributions of the particles of dust samples. The particles became bipolarly charged during the dispersion process used in the experiments. A similar charging process is expected when dust particles becomes airborne from a dry surface by wind shear forces. A dendrite
164
Developments in Surface Contamination and Cleaning
FIGURE 5.3 Normalized output power of a single crystal solar cell as a function of mass concentration dust (mg/cm2) deposited on the front cover glass. A xenon lamp was used to illuminate the solar cell. The mass median diameter of the dust sample was approximately 9 mm
deposition pattern has the advantage of forming a low packing density of the deposited dust on the surface of the solar panels, leaving some clear surface that reduces obscuration. Particles of the same size deposited as a monolayer, where the outer surface of each particle is in contact with the surface of the neighboring particles, will result in a maximum surface density of dust on the glass plate. The maximum surface area coverage of the plate will not, however, exceed 66%, regardless of the diameter of the monodispersed spherical particles. Considering geometrical optics, at least 34% of the light should still be transmitted through the dust layer of single spherical particles. The surface mass density can be calculated for such an ideal case for each layer of particles. However, if the particles are of different diameters, the interstitial spaces can get filled by small particles resulting in higher packing density of dust particles. The obscuration of light increases as the particle size distribution becomes wider. Polydispersed particles of irregular shapes having unipolar charge distribution will deposit on the surface with a high packing density, and can cause maximum obscuration of light for a given surface mass density of dust measured as mg/cm2. Soot mixed with dust will have increased extinction of light. Thus, the actual obscuration of light depends upon several factors: (1) the size, shape, and the charge distributions of particles; (2) the packing density or the microstructural deposition pattern of the particles (which depends upon the particle size, shape, and the electrostatic charge distributions); and (3) surface mass density of the deposited dust layer. In summary, both theoretical analysis and experimental studies show that dust deposition on the solar panels can cause a significant loss of PV output power.
Chapter | 5
Electrostatic Removal of Particles and its Applications
165
10. REMOVAL OF DUST FROM SOLAR PANELS USING LOW-POWER ELECTRODYNAMIC SCREENS Early work on electrodynamic traveling-wave systems and their applications in different fields were reported by several authors [15e19]. Figure 5.4 illustrates the basic principles of an EDS [20]. In this figure a free screen is shown without any encapsulation by a dielectric film. The electrodes can be energized either by using a single-phase or by a three-phase alternating current (AC) drive signal. The single-phase excitation (Fig. 5.4a) produces a standing wave between the electrodes and a polyphase drive (Fig. 5.4b) produces a traveling wave. The latter is preferable since the traveling wave has a strong translational component that rapidly moves the dust particles from one end of the screen to the other [17]. The standing wave produced by single-phase excitation also works for removing the particles from the screen. A standing wave can be considered as the superposition of two traveling waves moving in opposite directions. Thus, any instability in the AC electric field, the presence of harmonics in the applied field, or any air currents on the top surface of the panel, produce a drift velocity to transport the dust particles levitated by the electrodyamic forces from the surface to move away from the screen. Figure 5.4 shows a set of parallel electrodes, insulated from each other and connected to an AC voltage source. In Figure 5.4a, the electric field lines between two adjacent electrodes are shown. The electric field is non-uniform with respect to the spatial coordinates and it varies with time. Charged particles within this electric field experience an oscillatory motion; a particle with a charge þq is shown in Figure 5.4a at an instant of time when it is close to one of the positive electrodes. As the particle moves along the curved field lines, it experiences a centrifugal force. In addition, a dielectric particle in a nonuniform electric field experiences a dielectrophoretic force [21,22]. This force is experienced by dielectric particles, uncharged or charged, in a divergent electric field AC or direct current (DC).
FIGURE 5.4 Schematic of (a) single-phase EDS and (b) three-phase EDS
166
Developments in Surface Contamination and Cleaning
In modeling the performance of an EDS, it is necessary to calculate the repulsive forces that are applied to the particles and the transport velocity when a multiphase drive is applied as shown in Figure 5.4b. A general equation of motion of charged particles repelled and being transported can be represented by [22]: mp
dVp þ 6phrVp ðr; tÞ ¼ qE0 ðrÞcos ut þ Fext ; dt
(5.9)
where mp is the particle mass, Vp ðr; tÞis the particle velocity, h is the viscosity of the gas surrounding the screen, r is the particle radius, q is the electrostatic charge on the particle, and Eo ðrÞcos ut is the applied electric field of angular frequency u, and Fext represents external forces, such as gravity. In the above equation r represents coordinates (x,y,z), whereas r is used to denote particle radius. The electric field E(x,y,z,t) and the motion of the charged particle Vp(x,y,z,t) are related in a complex manner making it difficult to solve the Laplace equations involving the particle motion as a function of mp and q. Masuda, who first introduced the concept of an electric curtain [16], solved the equation numerically with appropriate approximations. His simulation agreed well with his experimental data. Referring to the particle of charge þq shown in Figure 5.4a, it is possible to examine the instantaneous electrostatic force experienced by the particle in the field E(t). The Coulomb force of repulsion will be: Fe ¼ qE
(5.10)
where q ¼ ss4pr2C, and ss in the surface charge density. If we assume that particles are tribocharged to their saturation level, then: qs ¼ 4pr2 2:64 105 C
(5.11)
where the factor 2.64 10e5 C/m2 is taken as the maximum surface charge density [22] of a spherical particle with diameter r > 1.0 mm at the atmosphere pressure of Earth (z0.1 MPa). The charge density limit is set by the onset of corona discharge initiated by the field created by the charged particle (selffield) when the maximum surface charge density is exceeded. Then the maximum repulsive force experienced by a particle with surface charge density ss is: FeðmaxÞ ¼ qmax Emax ¼ Kmax r2 ;
(5.12)
where qmax is proportional to r2, Emax is taken as 5 105 V/m for safe operation of the electrodes below the breakdown electric field of air at atmospheric pressure, and Kmax is a constant at a given ambient condition. At atmospheric pressure, Kmax ¼ 112.9r2. When a small particle with a saturation charge þ qmax
Chapter | 5
Electrostatic Removal of Particles and its Applications
167
approaches the screen and is close to the electrode where the field is Emax, the velocity acquired by the particle can be estimated if we assume that the particle motion obeys Stokes law (eq. (5.9)). The maximum velocity of the particle at steady state will occur when the drag resistance equals the electrostatic driving force: 6phrVpðmaxÞ ¼ 2:64 105 4pr2 5 105
(5.13)
We assume the particle is in air under atmospheric conditions, h is 1.8 10e5 N s/m2, and Emax is 5 105 V/m. Under these conditions, the maximum velocity Vp(max) is proportional to the radius of the particle: VpðmaxÞ ¼ 4:88 105 r m=s
(5.14)
For example, when r ¼ 1 mm, Vmax z 500 cm/s and when r ¼ 100 mm, Vmax z 50 m/s. These estimated values of maximum velocity for highly charged particles exemplify their vigorous motion on an EDS. Charged particles will acquire their steady state velocity within a time period of approximately 3sp, where sp is the aerodynamic relaxation time of the particle [12]. At atmospheric pressure, sp ¼ 13 ms for r ¼ 1 mm, and sp ¼ 30 ms for r ¼ 50 mm. For a 50-mm aerodynamic diameter particle the value of 3sp is 90 ms. The time period T, or the frequency of the electric field of excitation of the EDS electrodes, is determined from the range of values of 3sp for the range of particles to be removed. EDS is operated generally at a frequency range of 4e20 Hz. Figure 5.5 shows an arrangement of a single-phase electrodynamic screen embedded within a dielectric film. The dielectric film protects the electrodes from environmental degradation and provides a dielectric surface for tribocharging the dust particles. A phased voltage is applied to the screen electrodes to lift the particles and move them away from the screen by electrostatic forces. A flexible electrodynamic screen made of transparent conducting electrodes embedded in a transparent dielectric film can be used to remove dust from the surface of the solar panels [20]. Typically electrodes are made of indium tin oxide (ITO) 50 mm in diameter with inter-electrode separation Dust Particle Migration
Solar Cell
EDS
Phased Electrodes
FIGURE 5.5 A cross-sectional view of transparent parallel electrodes embedded in a transparent film or glass panel. The electrodes are energized by phased pulsed voltage for lifting and removing deposited dust particles from the solar panels or mirrors
168
Developments in Surface Contamination and Cleaning
distance of 500 mm. The electrodes are encapsulated in polyurethane (PU) of 200 mm thickness. In eq. (5.9), we have assumed an initial electrostatic charge þq on the particle. However, if a particle with no initial charge (q ¼ 0) deposits on the surface of the screen, it will acquire a charge by triboelectrification as the particle moves on the film surface by dielectrophoretic force (discussed in Section 12). In most cases, particles approaching the surface of the EDS will have a significant electrostatic charge, since dust particles become triboelectrically charged during their lift-off from the ground, as discussed earlier. Figure 5.6 shows arrangements for single- and three-phase electrode configurations of EDS. An arrangement for placing the EDS over the solar panel is shown in Figure 5.7. Here the screen is composed of parallel transparent conducting electrodes embedded in the top surface of a transparent dielectric film that is placed over the solar panels for dust cleaning. The screen can be placed directly over the solar panel or at a short distance over the surface covering the panel. If the electrodes are embedded within a thin dielectric film, the film serves (1) as a physical protection against mechanical impaction and abrasion by the dust particles, particularly during a dust storm, and (2) as an electrodynamic dust shield for minimizing obscuration of incident optical radiation. The dielectric film also serves as a means to charge the uncharged particles, thus making the screen effective against both charged and uncharged dust.
FIGURE 5.6 A schematic layout of (a) single-phase (left diagram) and (b) three-phase EDS (right diagram) electrodes
Chapter | 5
Electrostatic Removal of Particles and its Applications
169
ITO electrodes PU film
Screen SiO2 substrate SnO2, ITO or ZnO front electrode
Solar cell
a-Si:H p-layer, i-layer and n-layer ZnO reflection enhancement Aluminum back
FIGURE 5.7 Transparent EDS embedded in a transparent polyurethane (PU) film is placed over a solar panel. The ITO electrodes are of triangular cross-section which provides a more uniform distribution of the electric field compared with the field distribution produced by electrodes of rectangular cross-section. The figure shows an a-Si solar cell integrated with an EDS
The power requirement for operating the EDS is approximately 10 W/m2 of the solar panels, and the EDS is designed to operate by drawing the required power from the solar panels. The EDS is energized only when cleaning is needed. When the EDS is energized it takes less than two minutes to clean the panels. Under normal atmospheric conditions, dust cleaning may not be required for more than a few minutes per day, while the average power produced per square meter of the solar panels during the peak hours is approximately 100 watts. Experiments were conducted with high dust loading conditions exceeding 10 mg/cm2 when the obscuration reached nearly 100%. When the EDS was energized, more than 90% of the dust was removed. Figure 5.8 shows a three-phase power supply developed for the operation of the EDS. The power supply was designed and constructed to have low weight, small size, and low-power requirements for its operation.
FIGURE 5.8 A three-phase power supply on a circuit board is shown connected to a three-phase EDS screen. The electrodes are embedded in a dielectric film
170
Developments in Surface Contamination and Cleaning
11. TRAJECTORIES OF CHARGED PARTICLES ON THE ELECTRODYNAMIC SCREEN A simplified model of the positions and trajectories of drift-dominated charged particles driven by traveling or standing wave voltages over the EDS is presented here [23]. The analysis assumes that the self-field from the particles is much smaller than the imposed field from the traveling or standing wave voltages (Figs 5.4 and 5.7). The traveling wave is generated by applying a three-phase electrode system (Fig. 5.8) embedded in the screen, whereas a standing wave is formed when a single-phase electrode system (Fig. 5.5) is used. Figure 5.6 shows the electrode layout for single- and three-phase voltage drives. Figure 5.9 shows the two-dimensional geometry of a traveling wave potential at x ¼ d: vðx ¼ d; z; tÞ ¼ V0 cosðut kzÞ
(5.15)
where V0 is the amplitude of particle motion on the surface of the screen that is covered by a dielectric layer of thickness d, permittivity 3, and conductivity s. The dielectric film is considered to have some leakage due to its finite conductivity (s s 0). The region for x > 0 is taken to be free space with permittivity 30. The system is assumed to extend infinitely in the y direction with no field dependence on the y coordinate. In the absence of significant volume charge in the two regions d < x < 0 and x > 0, the governing equation is Laplace’s equation in both regions. The necessary boundary conditions are: Fðx ¼ d; z; tÞ ¼ vðx ¼ d; z; tÞ ¼ V0 cosðut kzÞ Fðx ¼ 0þ ; z; tÞ ¼ F ðx ¼ 0 ; z; tÞ 30
vEx ðx ¼ 0þ ; z; tÞ vEx ðx ¼ 0 ; z; tÞ ¼ 3 þ sEx ðx ¼ 0 ; z; tÞ vt vt
(5.16)
FIGURE 5.9 A traveling wave of potential is applied at x ¼ d. A lossy dielectric layer of thickness d, permittivity 3, and conductivity s prevents charged dust from penetrating into the region x < 0. The x ¼ 0 surface has reduced potential magnitude (V1 < V0) and a lagging phase b to the driving x ¼ d potential
Chapter | 5
Electrostatic Removal of Particles and its Applications
171
It is convenient to introduce complex amplitude notation so that the scalar electric potential can be written as: jðutkzÞ b Fðx; z; tÞ ¼ Re½ FðxÞe
(5.17)
where Re signifies the real part. Then the solution is: ( b FðxÞ ¼
V0 sinh kx Vb1 sinh kðx þ dÞ þ sinh kd sinh kd Vb1 ekx
d x 0
(5.18)
x0
where Vb1 is the complex amplitude potential at x ¼ 0 and is given by: Vb1 ¼
ðs þ ju3ÞV0 sinh kd½ ju30 þ ð ju3 þ sÞcoth kd
(5.19)
The magnitude and phase of Vb1 ¼ V1 ejb are: #1=2 " 2ð32 u2 þ s2 Þ V 1 ¼ V0 2 ð3 320 Þu2 þ s2 þ ðð32 þ 320 Þu2 þ s2 Þcosh 2kd þ 2330 u2 sinh 2kd b ¼ tan1
ðu2 32
þ
u30 ssinh kd 2 s Þcosh kd þ 33
0u
2
sinh kd
(5.20) (5.21)
Note that if s ¼ 0, V1 ¼
3V0 ; ½3 cosh kd þ 30 sinh kd
b ¼ 0
The complex amplitude electric field in each region can then be written as: ( b1 cosh kðxþdÞ k½V0 cosh kx V b dF b d x 0 sinh kd (5.22) Ex ¼ ¼ dx b kx x0 kV 1e b Ebz ¼ jk F Thus, the electric field for x > 0 is: E ¼ Re½ð Ebx ix þ Ebz iz ÞejðutkzÞ ¼ kV1 ½cosðut kz bÞix sinðut kz bÞiz ekx
11.1. Charged Particle Trajectories for x > 0 We now assume that charged particles, each with total charge q and radius r, are initially uniformly distributed for x > 0 and that their self-field is very small
172
Developments in Surface Contamination and Cleaning
compared to the applied field. We also assume that the Coulomb force on the particles gives them a drift velocity opposed by Stokes’ viscous drag where the medium for x > 0 has viscosity h: 6phrv ¼ qE The particle mobility m is written as: m ¼ v=E ¼ q=6phr
(5.23)
Neglecting particle inertia gives the x and z components of particle motion as: dx ¼ mEx ¼ mkV1 cosðut kz bÞekx dt dx vz ¼ ¼ mEz ¼ mkV1 sinðut kz bÞekx dt vx ¼
(5.24) (5.25)
We now replace the traveling wave voltage at x ¼ 0 in Figure 5.9 by a standing wave: vðx ¼ d; z; tÞ ¼ V0 cos ut
cos kz
(5.26)
Then, the solution for the scalar electric potential is of the form: b Fðx; z; tÞ ¼ Re½ FðxÞcos kz ejut
(5.27)
b where FðxÞ is of the same form as eq. (5.18). The boundary conditions are still given by eq. (5.16) with the traveling wave potential at x ¼ d replaced by eq. (5.27). Then, the solutions of eqs (5.19)e(5.22) are valid here and the electric field for x > 0 is given by: E ¼ kV1 cosðut bÞ½cos kz ix þ sin kz iz ekx
(5.28)
The analogous charge transport equations to eqs. (5.27) and (5.28) are: dx ¼ mEx ¼ mkV1 cosðut bÞcos kz ekx dt dz ¼ mEz ¼ mkV1 cosðut bÞsin kz ekx vz ¼ dt
vx ¼
(5.29) (5.30)
11.2. Removal of Charged Particles A particle with diameter 1.0 mm, with a charge þq, and subjected to an excitation voltage waveform, V0 sin ut (or the electric field E0 cos(ut)), will move with a velocity Vp(t) ¼ Vpo cos(ut f), as shown in Figure 5.10. The particle with charge þq undergoes a hopping motion along the curved lines of force between the adjacent electrodes. The figure shows a positively
Chapter | 5
Electrostatic Removal of Particles and its Applications
173
FIGURE 5.10 Cross-section of an EDS made of a flexible polyethylene terephthalate (PET) film of 500 mm thickness on which transparent ITO electrodes of rectangular cross-section (width 10 mm, height 10 mm) are deposited with an inter-electrode spacing of 1000 mm. The electrodes are embedded within a PU film coating with a film thickness of 50 mm. The thickness of the electrodes is varied from 10 to 100 mm and the inter-electrode spacing from 100 mm to 1000 mm for optimization of EDS operation
charged particle experiencing two forces of repulsion, one tangential to the curved field lines and the other normal to the curved path. The normal component is the centrifugal force which arises due to the curvilinear motion of the particles. This normal component provides the lift force. If the particle gains positive charge by multiple contacts with the surface of the screen, the repulsion forces will increase, thereby moving the particles away from the surface. A negatively charged particle will move in the opposite direction. Such motion was observed using large particles with diameters of 60e80 mm. Figure 5.11 shows the distribution of electric field as a function of electrode width and inter-electrode spacing, while the electrode thickness remained constant. As the inter-electrode distance and the width are reduced, the surface density of the divergent electric field on the screen increases, making the screen more effective in removing the dust from the screen. However, increasing the spatial frequency of the electrode will cause additional obscuration of the incoming solar radiation even when the electrodes are relatively transparent. In Figure 5.11, the electric field distributions were modeled as follows: 1250 mm wide rectangular electrodes with 2500 mm spacing (top); 1250 mm wide rectangular electrodes with 1250 mm spacing (middle); and 625 mm wide rectangular electrode with 625 mm spacing (bottom). The model shows an electric field intensity depression occurs over the middle of the rectangular electrodes that indicates dust may collect in these regions. The field distribution
174
Developments in Surface Contamination and Cleaning
FIGURE 5.11 The electric field has been modeled for three electrode configurations. The top section shows electrodes with relatively large dimensions and large inter-electrode spacing. The spatial distribution of the divergent electric field intensity is non-uniform. As the electrode dimensions and the inter-electrode spacing are reduced, more uniform field intensity distributions are achieved (middle and bottom sections)
analysis shows that triangular-shaped electrodes as shown in Figure 5.7 would be most preferred.
11.3. Removal of Charged and Uncharged Particles When the electrodes are embedded within a dielectric film, and are excited with sine waves as shown in Figures 5.4, 5.5, 5.7, 5.10, and 5.12, both initially charged and uncharged particles with different electrical resistivities are efficiently removed from the EDS. The removal process for an initially charged (q) particle is due primarily to the Coulomb force, and for uncharged dielectric particles the removal process is caused by dielectrophoretic force, as explained below and illustrated in Figures 5.13 and 5.14. Figure 5.14 shows the interaction of the uncharged conducting particle in the electrodynamic field, where the primary charging mechanisms are induction and tribocharging. In summary, the electrodynamic mechanisms involved in the dust removal process are: (1) Coulomb force; (2) dielectrophoretic force; (3) triboelectric charging; (4) induction charging; (5) non-uniform electric field distribution on the screen surface; and (6) temporal variation of the phased voltage waveform over the surface of the screen.
Chapter | 5
Electrostatic Removal of Particles and its Applications
175
FIGURE 5.12 A charged (þq) particle of diameter 2 mm, located at A, is subjected to an AC electrical field E0sin(ut) applied between the adjacent electrodes as shown. The frequency of the electric field is 4 Hz
12. DIELECTROPHORETIC FORCE An electric field E produces some displacements of the electrons and ions within a dielectric particle. This process is similar to the polarization of any atom when an external electric field distorts the electron cloud with respect to the nucleus of the atom. In a microscopic particle, when two charges þq and q are separated by a distance d, a dipole of moment qd is formed. The combined effect of each of these elementary dipoles within a microscopic
176
Developments in Surface Contamination and Cleaning
Dipole moment on an initially uncharged particle Dielectric surface
Vo sin( ωt)
Electrodes
PU film
FIGURE 5.13 An uncharged dielectric particle, deposited on the surface of a dielectric film, is experiencing a dielectrophoretic force because of the induced dipole moment on the particle by the applied non-uniform electric field
FIGURE 5.14 Induction charging of conducting and semi-conducting particles deposited on a dielectric screen with embedded electrodes. 31 and 32 are the relative dielectric constants of the screen and the particle, respectively
particle can be considered as a single polarization vector P as shown in Figure 5.13. By virtue of the dipole moment induced on the particle placed in a non-uniform electric field, the particle experiences dielectrophoretic force [21,22], expressed in eq. (5.31), in the field gradient produced by the applied voltage across the electrodes embedded in the screen. Fd ¼ ðPVÞE
(5.31)
where P is the polarization vector (field-induced dipole moment) and is equal to np, where p ¼ qd, the dipole moment for each individual dipole within a single particle, and n is the number of dipoles. The translational dielectrophoretic force Fd is proportional to the product P and grad E. A spherical particle of radius r having a dielectric constant 32, resting on a medium of dielectric constant 31, and experiencing an electric field gradient, will experience a force Fd which can be expressed as [21,22]: F d ¼ 2pr3 31
32 31 VjEj2 32 þ 31
(5.32)
Chapter | 5
Electrostatic Removal of Particles and its Applications
177
The direction of the force Fd (the upper bar is used for vector notation) depends upon the sign of the term (32 e 31). In a non-uniform electric field, the polarization force can induce vigorous motion of a neutral particle suspended in a fluid medium (eq. (5.12)). If the permittivity of the particle is higher than that of the medium, the dielectrophoretic force pulls the particle towards the region of higher field intensity. In an alternating field, an uncharged dielectric particle approaching the EDS will oscillate and roll back and forth. As the particle makes contact with the surface of the screen, the movement of the initially neutral particle will cause electrostatic charging by triboelectrification. The acquired electrostatic charge, positive or negative, will add Coulomb force of repulsion, lifting the particle off the surface.
13. TRIBOCHARGING OF PARTICLES When an uncharged dielectric particle is resting on the surface of the dielectric film (Fig. 5.13) and the electric field is then applied to the electrode system, it will experience the translational dielectrophoretic force causing particle motion over the polymer film. Since the force is proportional to the square of the field gradient and the field is oscillating, the fluctuations of the dielectrophoretic force of attraction on the particle makes it roll or move along the field gradient lines on the dielectric film surface. These particle motions on the dielectric surface of the screen cause the particle to become triboelectrified to a significantly high charge level. Thus, the initially uncharged particles resting on the screen are tribocharged, resulting in a net charge gain on the particle which causes the particles to jump off the screen by the Coulomb force of repulsion. When the particles, charged or uncharged, in contact with a dielectric film oscillate or roll even with a small amplitude of motion due to Coulomb or dielectrophoretic force, the associated friction against the surface of the screen causes the particles to acquire triboelectric charges. The polarity and magnitude of the charge depends upon the work function difference between the two contacting surfaces. The added charge on the particle increases the amplitude of particle motion until the charge level is high enough for its removal from the surface. In the tribocharging process, there is a charge exchange between the particle and the screen surface. Beside the work function difference, the charge level acquired by the particle will also depend upon the force of adhesion between the particle and the screen surface and environmental factors such as relative humidity. As the charged particles are lifted off, the screen surface remains charged with the same magnitude but of opposite polarity, and thus it is necessary to drain the charges from the film surface for its continued charge exchange operation for dust removal. It is, therefore, necessary to use an optimum level of conductance in the dielectric film.
178
Developments in Surface Contamination and Cleaning
Figures 5.12 and 5.13 show the electric field distributions and particle motion on an EDS film. Both the charging process and charged particle motion are shown schematically. In Figure 5.12, we assume the positions of the particle as a function of time are as shown in the three time frames, starting at t1 ¼ 0 (top frame), t2 ¼ 125 ms (middle frame), and at t3 ¼ 500 ms (bottom frame). The hopping motion of the particle Vpo sin(ut f) is shown schematically where f is the phase lag for the particle, given by tane1 usp, where sp is the aerodynamic relaxation time. Figure 5.13 shows the dipole moments induced on an uncharged dielectric particle approaching an EDS film. Since the applied electric field oscillates, the particle experiences a force which is proportional to the gradient of the square of the field, and moves along the field lines on the dielectric surface. In this process of particle motion on the surface of the film, the particle gains a significantly high triboelectric charge, until the particles jump off the screen.
14. REMOVAL OF UNCHARGED CONDUCTING PARTICLES The process of electrostatic charging and removal also works for conducting particles. The particles, initially uncharged and deposited on the dielectric screen, will experience a force due to the induced charge. The charge on a spherical particle can be approximated as: qfE0 r2
(5.33)
And the force Fi is proportional to E20r2. The process is similar to the charging of the dielectric particles. In general, particle charging takes place by both mechanisms combined with contributions of different degrees depending upon the materials involved. Table 5.1 shows the relative contributions as a function of the approximated values of particle resistivity. Some of the limitations of the charging process are also included in the table. The induction charging time constant of a particle resting on a dielectric screen will depend upon the conduction of charge caused by the induced electric field from the particle to the dielectric surface of the screen. In many applications of the EDS, it is assumed that the particles are of mostly dielectric materials with certain conductivity and the surface of the film embedding the screens will also have a significant surface conductivity for draining the charge imparted by the contact charging of the particles. It is necessary to have the necessary amount of surface conductivity to avoid charge accumulation on the surface of the dielectric screen. The requirement of surface conductivity of the screen is a critical factor for dust removal. An excessive surface conductivity will shield the electric field, whereas a very high resistivity will lead to an excessive accumulation of surface charge on the film that will reduce tribocharging and increase particle adhesion
Chapter | 5
179
Electrostatic Removal of Particles and its Applications
TABLE 5.1 Electrostatic Charging Processes for Initially Uncharged Particles Deposited on an Electrodynamic Screen Embedded in an Insulative Polymer Film. Particle Resistivity
Charging Processes
Limitations
r < 10 Um
Induction charging Triboelectrification
Film surface resistivity too high High humidity, r < 0.5 mm, low electric field
108 Um < r < 1011 Um
Induction charging Dielectrophoretic motion Triboelectrification
Same as above
r > 1011 Um
Dielectrophoretic motion Triboelectrification
Same as above, low work function difference between the contacting surfaces
8
to the surface. The time constant for induction charging of the particle will depend upon the product of the particle’s effective surface resistivity rp and the dielectric constant 32 of the particle. If rp is the effective resistivity of the particle on the screen surface, then the charging time constant sc can be approximated as: sc ¼ 32 30 rp
(5.34)
where 30 is the permittivity of free space and is given by 30 ¼ 8:854 1012 F=m: Induction charging takes place for all materials but becomes effective when sc << T; where T is the time period of the applied electric field E0 sinðutÞ: For example, if rp ¼ 108 Um, and 32 ¼ 2.0, the charging time constant will be approximately 2 ms, which is much shorter than 250 ms time period T if the frequency of the applied AC field is 4 Hz. Figure 5.12 shows an example of a 2.0 mm diameter particle trajectory if we consider that the particle has acquired a positive charge q during the half cycle before t ¼ 0þ. For particles with higher resistivity, the charging is due primarily to the combination of dielectrophoresis, induction, and triboelectrification.
15. HIGH-VOLTAGE THREE-PHASE POWER SUPPLY FOR THE ELECTRODYNAMIC SCREEN 15.1. Design Requirements An EDS requires a high-voltage AC power supply to build an electrostatic traveling wave. The power supply design is mainly to meet the requirements for
180
Developments in Surface Contamination and Cleaning
energizing the EDS electrodes, providing three-phase voltages with the features of variable frequency, variable amplitude, and experimental convenience. In the following sections we discuss some fundamentals of EDS power supply and give descriptions of two practical designs. We will also discuss the future design of industry-standard EDS power supply. An EDS power supply provides three-phase voltages with 120 phase shift. For various laboratory tests the required voltage amplitude may change from several hundred volts to several kV. Operating frequency is adjustable from several Hz to several hundred Hz. The power requirement is quite low for an experimental EDS. For example, the required power for a 60-cm2 EDS is about 10 mW which includes the power needed to operate the power supply. Much simpler design is also available for specific applications. In most applications [23e25], the three-phase voltages are assumed to be sinusoidal waveforms as shown in Figure 5.15a. However, in practice, the square-waveform voltages as shown in Figure 5.15b can also be applied to EDS operation for dust removal. The main differences are the power consumption of the power supply, current waveforms, and system costs. An EDS can be assumed as a capacitive load. When three-phase voltages are sinusoidal, the three-phase currents are also sinusoidal and lead the voltages by 90 . In the case of square-waveform of three-phase voltages, the three-phase currents have very large current harmonic contents and high peak values.
FIGURE 5.15 Three-phase voltages (a) in sinusoidal waveforms and (b) in square waveforms
Chapter | 5
Electrostatic Removal of Particles and its Applications
181
FIGURE 5.15 (continued) Three phase square wave forms
However, a square-waveform Power supply has advantages of simple circuit design, low power losses, and low manufacturing costs. As shown in Figure 5.16, an EDS power supply consists of: (1) a DC/DC converter to generate a DC high voltage up to several kV; (2) an inverter to convert DC high voltage to three-phase AC voltage outputs with 120 phase shifts; and (3) control and protection units. The inverter control and circuit protection against over-voltage and over-current faults are essential for EDS operation.
15.2. MOSFET-Based EDS Power Supply The EDS power supply works at high voltage and extremely low current. A high-voltage metal oxide semiconductor field-effect transistor (MOSFET) is suitable. Linear MOSFETs can run up to 1.5 kV, while switching MOSFETs can operate up to 4 kV. Linear MOSFETs have an extended forward-bias safe operating area and can be used to develop sinusoidal EDS power supply. Switching MOSFETs are mainly used for switching operations and can be used for square-waveform EDS power supply. For higher-voltage operation, series connection of MOSFETs is possible.
182
Developments in Surface Contamination and Cleaning
FIGURE 5.16 Block diagram of an EDS power supply
An experimental design of a square-waveform EDS power supply is shown in Figure 5.17. By setting the input voltage from 0 to 12 V, we can control the DC/ DC converter EMCO-F10 (EMCO High Voltage Corporation, Sutter Creek, CA) to generate a DC voltage from 0 to 1 kV. Two photographs of the EDS power supply are shown in Figures 5.18a and 5.18b. The inverter consists of three legs, each consisting of two MOSFETs (1.2 kV/4.5 U). Each leg provides a phase voltage output. By switching on and off two MOSFETs in one leg sequentially, one square-waveform voltage is generated as shown in Figure 5.15b. The digital signal processing (DSP) controller is used to control MOSFETs of three legs to generate three-phase output voltages in square waveforms. The frequency can be set from 0 to 1 kHz. The DSP controller also detects the DC-link voltage and current for measurement and system protection against over-voltage and over-current faults.
FIGURE 5.17 Schematic diagram of an experimental EDS power supply based on 1.2-kV MOSFETs
Chapter | 5
FIGURE 5.18
Electrostatic Removal of Particles and its Applications
183
MOSFET-based EDS power supply (a) in a housing and (b) circuits in a PCB
The power circuit can be modified for higher voltage up to 3 kV by changing the DC/DC converter and power MOSFETs. The highest voltage rating of single power MOSFETs available is 4 kV. For a voltage higher than 3 kV, the circuit must be changed, for example, to connect several power MOSFETs in series for higher voltage operation. The DC/DC converter of the same series can reach up to 14 kV.
184
Developments in Surface Contamination and Cleaning
15.3. High-Voltage Linear Amplifier-Based Sinusoidal Power Supply A simple solution to realize an EDS power supply in sinusoidal waveforms is to use high-voltage linear amplifiers. A high-voltage linear amplifier is able to generate high-voltage output following the input waveform. The gain of the amplifier can be up to 50 dB. A high-voltage linear amplifier can actually generate an arbitrary waveform depending only on the input signal. An experimental EDS sinusoidal power supply is shown in Figure 5.19. The high-voltage linear amplifier is AS3B1 (Matsusada Precision Inc., Shiga, Japan). A Labview workstation (not shown in Figure 5.19a) is applied to generate control signals for three-phase voltages and data acquisition for EDS experiments. The block diagram is shown in Figure 5.19b. The power supply can generate three-phase voltages with the peak values from e3 kV to 3 kV, and frequencies from 0 to 1.5 kHz. The power rating is 3 W. Electrical design of high-voltage power supply is a mature technology in power electronics. Single-phase high-voltage power supplies are commercially available. Such a commercial product is generally designed for special applications and is expensive. For EDS operations, three-phase or other multi-phase high-voltage power supplies are required. The costs and physical sizes are important for the wide applications of EDS technology in the future. Hence, it is necessary to find the optimal design of EDS power supplies. For the industry-standard design of EDS power supplies, there are still many questions to be answered by EDS investigations, such as how to select an optimal operating frequency and voltage magnitude. It is expected that both frequency and voltage magnitude are strongly related to EDS applications: (1) the locations of the EDS application (for instance, on Earth or on Mars); (2) the characteristics of dust particles; and (3) the EDS size and power requirements. The average power required for operating a three-phase screen constructed on a printed circuit board (PCB) with square wave excitation is shown in Figure 5.20. The power drawn by the electrodes (with 1 mm spacing) is shown for three operating frequencies 4, 10 and 15 Hz. Similar power vs. applied voltage relationship for a transparent screen made of a polyethylene terephthalate (PET) screen with ITO electrodes (spacing 1 mm) is shown in Figure 5.21.
16. TESTING OF THE ELECTRODYNAMIC SCREEN 16.1. Experimental Setup Transparent electrodynamic screens were originally developed for their possible applications to NASA’s Mars mission [23e25]. Experiments on testing the screens for their dust removal efficiency (DRE) were carried out with Mars dust simulant. The dust samples were obtained from Johnson Space Center (JSC), Houston, TX, and from Jet Propulsion Laboratory, Pasadena, CA. The
Chapter | 5
Electrostatic Removal of Particles and its Applications
185
FIGURE 5.19 Linear-amplifier-based EDS power supply (a) in a system test setup and (b) block diagram
186
Developments in Surface Contamination and Cleaning
FIGURE 5.20 Power vs. voltage curves for a three-phase PCB EDS with 1-mm electrode spacing
FIGURE 5.21 Power curves for a three-phase ITO electrode-EDS with 1-mm electrode spacing
Chapter | 5
Electrostatic Removal of Particles and its Applications
187
dust was prepared from Hawaii volcanic dust with appropriate size classification and mineral oxide additives to simulate the predicted dust composition of Mars. The primary constituent of the dust is SiO2. The dust samples were sieved so that the upper size limit was approximately 40 mm in diameter. Before each test, the dust sample was dried in an oven at a temperature of 373 K for at least 24 hours. For each experimental run, a small amount of dust was dispersed using a powder pump (a Venturi suction device) for aerodynamic dispersion of the dust inside a test chamber. The dispersion process forms an aerosol containing dust inside the test chamber as the particles were allowed to settle on the screen. Alternatively, a sieve with 40 mm openings was used for dispersing the sample dust.
16.2. Size Classification Test aerosols containing dust particles were classified by using a cyclone classifier that removed the coarse particles with diameters larger than 10 mm. The fine fraction of aerosol was used for the tests. The Count Median Diameter (CMD) of the fine dust ranged from 2 to 5 mm, and the Volume (or Mass) Median diameter (MMD) ranged from 8 to 15 mm. This size range agrees with the anticipated Martian dust cloud, as well as with the size distribution of desert dust.
16.3. Charge Conditioning of Dust Cloud The dust removal efficiency of the EDS was tested against both charged and neutral dust samples. For charging the particles of the test aerosol, corona chargers were used for unipolar charging, either with positive or negative polarity. In most cases, however, the dust particles were tribocharged by tumbling against stainless steel or TeflonÒ beads in a stainless steel or TeflonÒ-lined container [26]. The dust dispersion process creates a cloudcharged particle with both negative and positive polarities. For neutralizing the dust particles, a charge neutralizer was used. The charge neutralization process consists of introducing bipolarly charged ions inside the test chamber. As the dust particles deposit inside the chamber in an atmosphere containing both positive and negative ions, the charge-to-mass ratio decreases as a function of time. Charge neutralization was necessary for testing the screen performance against uncharged particles [27e29].
16.4. Test Procedure A dust cloud chamber with an appropriate air filtration system was used for testing the EDS devices operating with either single- or three-phase AC drive. The test EDS was placed near the bottom of the chamber while the test aerosol was introduced at the top. The size and charge distributions of particles were
188
Developments in Surface Contamination and Cleaning
measured using an Electrical Single Particle Aerodynamic Relaxation Time (ESPART) analyzer [29]. The size distribution of the test dust was also measured using a MicrotracÒ analyzer (Microtrac Inc., Montgomeryville, PA) based on the optical diffraction method. The DRE of the EDS was calculated by measuring the mass of dust on screen before and after the dust removal process. Figures 5.22 and 5.23 show the experimental arrangements for evaluating a transparent EDS against dust loading in laboratory environmental conditions of one atmospheric pressure, temperature of 295 K, and 50% relative
FIGURE 5.22 A 3D view of the test chamber for studying the performance of the electrodynamic screens. The environmental conditions can be adjusted (temperature up to 50 degrees and relative humidity (RH) up to 80% can be achieved) inside the chamber. The solar panels can be tilted at different angles
Chapter | 5
189
Electrostatic Removal of Particles and its Applications
Fan
Screen
Mars dust simulant Brush
Bipolar ion generation AC high voltage Solar lamp/UV radiation
Vibrator
φ1 φ2 φ3
Electrodynamic screen (flexible) Glass support Output Solar panel
3-Phase high voltage (0-2000 V) AC drive (low current)
FIGURE 5.23 An experimental setup for testing transparent electrodynamic screens against mass loading of test dust under normal atmospheric conditions
humidity (RH). Mars simulant dust (JSC-1) was used for testing the panels. The dust simulant was first size classified after grinding the dust with a laboratory milling machine and using a cyclone size classifier to classify the dust to obtain particles with CMD smaller than 10.0 mm. A vibratory dust feeder and a powder pump were used to disperse the dust sample inside the chamber. The dust particles were allowed to settle for about 10 minutes before testing the DRE. The structure and uniformity of the dust layer were examined with a microscope. Measurement of DRE was performed: (1) for positively and negatively charged dust (with corona charging), dust particles that were charged with bipolar charge distribution (tribocharging), and with neutral particles (dust particles with a very low charge); (2) with different dust loadings on the screen (measured in mg/cm2); and (3) continuous and intermittent operation of the EDS. The dust loading on the EDS was measured gravimetrically. The size and charge distributions of the test dust were measured by sampling the particles from the dust cloud inside the chamber. Figure 5.24 shows an experimental arrangement for testing EDS under low atmospheric pressure. In this case, the screen was placed inside a vacuum chamber. The chamber was evacuated to a low pressure of 0.5 kPa and refilled with CO2. The process was repeated several times to reduce the concentration of oxygen and experiments were conducted at 1 kPa pressure in a CO2 environment which approximately simulates the Martian atmosphere.
190
Developments in Surface Contamination and Cleaning
FIGURE 5.24 Experimental arrangement for testing the EDS under Martian atmospheric conditions (0.5 to 1.0 kPa pressure, CO2 atmosphere) to determine the dust removal efficiency as a function of dust loading. For each experiment, the total power requirements for removal of the dust layer were measured as functions of applied voltage pulses and the frequency of operation
17. MEASUREMENT OF MAXIMUM POWER POINT OPERATION OF THE PV SYSTEM WITH EDS The PV systems require external power conditioning electronic circuitry to operate solar cells for delivering maximum power to the load connected to the PV system. Without this control system, the efficiency of the solar panel will be compromised by not operating at the maximum power point (MPP) of the system for which it is designed. Power conditioning and control circuitry is almost always used to measure the current and voltage output of the solar panels, and to adjust the voltage to allow the modules to operate at their MPP. It is therefore necessary to test the solar panel output at its MPP operation. As dust builds up on the panel surface and occludes the light from illuminating the solar cells, the performance of the PV system degrades. To determine if the change in PV output power is caused by a decrease in the insolation due to the apparent motion of the sun or by a cloud cover, or by dust deposition, an external panel surface monitor will have to be used for detecting dust deposition and energizing the EDS system. This monitor detects the surface density of deposited dust on the panel and provides a signal to the power control device to energize the EDS, which will remove dust from the panels as needed and maintain the operation at its maximum efficiency.
Chapter | 5
Electrostatic Removal of Particles and its Applications
191
18. TESTING THE SOLAR PANELS INTEGRATED WITH EDS FOR MAXIMUM POWER POINT OPERATION To test an EDS-integrated solar panel’s performance with external MPP control circuitry, a test chamber was designed for depositing dust on the panel surface under different ambient conditions. Figure 5.23 shows a schematic view of the solar panel test chamber, constructed to control the intensity of light incident on the solar panel, the tilt angle of the panel, relative humidity inside the chamber, and the surface mass concentration of dust dispersed onto the solar panel. The test panels can be evaluated under different conditions while tracking and measuring the maximum power of the solar panel. The light sources used are either a solar simulator (xenon lamp), or a pulse width modulation (PWM)controlled light-emitting diode (LED) light source for lower intensity (10 W/m2) applications. Both light sources can be tested to have precise irradiance for experiments. Dust can be deposited by using a vibrating sieve or a powder pump to control the amount of dust occluding the panel, and a microbalance is used to measure the change in weight of the panel before and after testing. Before dust is deposited on the solar panel, measurements can be taken to determine the maximum power of the clean solar panel. Because the current of the solar panel will be determined by the amount of light on the panel, the power output may not be maximized even under the clean condition. As shown in Figure 5.25, the output voltage can be adjusted to obtain the MPP operation of the solar panel. To adjust the voltage output of the PV modules to be applied to the load for its operation at the MPP, a DCeDC converter is used. In its simplest form, this can be realized by a resistive load that can be varied to change the voltage drop going to the load, or by using a commercial voltage regulator integrated circuit that can have a variable voltage output, such as the LM317 (National Semiconductor, Santa Clara, CA) shown in Figure 5.26.
FIGURE 5.25 Maximum power point operation of a solar panel
192
Developments in Surface Contamination and Cleaning
FIGURE 5.26 Application of an integrated circuit (IC) to control the output voltage for MPP operation
The method most often used for adjusting the voltage is a boost/buck voltage converter, which incorporates an inductive circuit and a MOSFET switch to control the load voltage from the solar panel output. The MOSFET switch can be controlled by a PWM signal, which varies the amount of time the switch is in the open or closed state. This will, in turn, increase or decrease the voltage, which can be used to obtain the MPP. The circuit shown in Figure 5.27 is often used as a converter. With an efficient method of voltage control in place, the circuitry can monitor the input voltage and change the output voltage levels to find the MPP operation. This function is known as maximum power point tracking (MPPT). A microcontroller with an appropriate analog-to-digital converter is used to periodically measure current and voltage from the panel, and an algorithm is implemented to change the PWM signal to switch the MOSFET to keep it operating at the MPP. Commercial electronics are readily available that perform MPPT functions for solar panel arrays.
FIGURE 5.27 Voltage converter for MPP operation
Chapter | 5
Electrostatic Removal of Particles and its Applications
193
Once the MPPT functions have been realized, the maximum power of the clean panel can be measured. After this, the test dust can be deposited onto the panel, and the decreased power output from the solar panel can be measured at its current maximum power output under dust loading. The EDS panel can then be activated to clean the solar panel, and the maximum power is measured once again. To implement a smart EDS system for automatic operation, photodetectors or small cameras can be used to observe the dust layer on the solar panels, and a threshold level can be set to activate the EDS cleaning system using power from the solar panel. At each step in the measurement process, the cleaning efficiency can be monitored by the weight measurements taken before dust deposition, after dust deposition, and once again after EDS cleaning. By integrating weight, irradiance measurements, power measurements, and other environmental measurements into a localized microcontroller, data can be gathered on all aspects of the operation. This can be used to determine the dust concentration per unit area, power output as a function of dust loading, and cleaning efficiency in different environmental conditions.
19. RESULTS The particle size analysis of JSC Mars-1 simulant dust measured using the MicrotracÒ analyzer is shown in Figure 5.28. The d10, d50, d90 values were found to be 1.22 mm, 9.06 mm, and 38.45 mm, respectively. The CMD of simulant dust using an ESPART analyzer was 3.66 mm (standard deviation 0.19, n ¼ 15). The
FIGURE 5.28 Size distribution of Mars dust simulant measured by an ESPART analyzer
194
Developments in Surface Contamination and Cleaning
charge distribution measurements were also performed using an ESPART analyzer and are shown in Figure 5.29. The net charge-to-mass ratio (Q/M) of simulant dust measured with an ESPART analyzer was 2.7 mC/g, as shown in Figure 5.30. This charge distribution was the result of (1) inter-particle charging; (2) tribocharging during grinding; and (3) handling processes. In the dispersion process the dust particles became charged. Some of the experiments were performed by deliberately charging the particles either with a net positive charge or with a net negative charge. When the dust was tribocharged against a TeflonÒ surface, the particles were mostly charged positively and were negatively charged against a stainless steel surface. When the particles were neutralized, most of the particles showed charge close to zero. The performance of the screen was analyzed for dust with a net positive charge and a net negative charge, and for mostly neutral particles (Fig. 5.31). It was found that the efficiency of the screen did not deteriorate for neutral particles [27e29]. A DRE of 85% was obtained for charged Mars dust simulant when excited by single-phase AC. The effect of particle size on DRE was also studied for a three-phase EDS system. A DRE of over 90% was achieved for an EDS with 1.27 mm electrode spacing (Fig. 5.31). Figure 5.32 shows the DRE with and without neutralization of the Mars dust simulant. The EDS system was found to be equally effective for charged and neutral particles.
FIGURE 5.29 Charge distribution of Mars dust simulant as measured by an ESPART analyzer. The particles were bipolarly charged as shown in the figure
Chapter | 5
Electrostatic Removal of Particles and its Applications
195
FIGURE 5.30 The net charge-to-mass ratio in mC/g of Mars dust simulant particles measured using an ESPART analyzer varied depending on the process conditions. The particles became charged primarily during the dispersion process
FIGURE 5.31 Dust removal efficiency of a three-phase EDS operating at 750, 1000, and 1250 volts (electrode spacing ¼ 1.27 mm, electrode width ¼ 0.127 mm, f ¼ 4 Hz, cleaning operation time ¼ 60 s)
196
Developments in Surface Contamination and Cleaning
FIGURE 5.32 Dust removal efficiency of a three-phase EDS with and without charge neutralizer (electrode spacing ¼ 1.27 mm, electrode width ¼ 0.127 mm, peak-to-peak voltage ¼ 1250 V, f ¼ 4 Hz, run time ¼ 60 s, count median (aerodynamic) diameter of the dust particles ¼ 3.66 mm, d10 ¼ 1.22 mm, d50 ¼ 9.06 mm, d90 ¼ 38.45 mm)
20. SUMMARY AND CONCLUSIONS We have presented here the basic principles of EDS operation for removing dust particles from solar panels. In most of our tests three-phase operation produced better dust removal efficiency compared to the single-phase operation. It appears that the traveling wave plays an important role in dust removal. The mathematical analysis provides an outline for the operation of the EDS for charged and uncharged particles driven by traveling or standing wave voltages. The analysis presented here has made simplifying assumptions that environmental conditions are ideal for the EDS operation and that the self-field from the charged particles, including image forces, are negligible compared to the imposed fields from the traveling and standing wave voltages. The test results show that EDS operation remains efficient for all of the three charging conditions: (1) positively charged particles; (2) negatively charged particles; and (3) neutral particles with a net charge close to zero. The design for constructing the EDS has also been empirically optimized. However, most of the tests performed to date have been conducted at atmospheric pressure and at low pressure close to 1 kPa (for Mars applications). A major problem in operating the EDS will be to avoid Paschen breakdown of the electrodynamic field applied across the electrodes. The breakdown voltage, or the sparking potential, according to Paschen’s law is a function of the product pds, where p is the ambient pressure and ds is the distance between the electrodes. The breakdown field is not a linear function of pds; it depends upon the actual value of ds and the number of collisions made by the electron in crossing the gap ds which depends upon the density of the surrounding medium. The electrodynamic screens are made of transparent plastics, such as PET for its ultraviolet (UV) radiation resistance, and a set of parallel conducting
Chapter | 5
Electrostatic Removal of Particles and its Applications
197
electrodes, made of transparent ITO or a conducting polymer, embedded under a thin transparent film on the surface. The transparent plastic film EDS can be efficiently used for protecting solar panels from dust deposition and obscuration of solar radiation. However, for actual terrestrial application, the EDS system needs to be integrated with the solar panel during the manufacturing process. By applying a three-phase high-voltage AC electric field to the electrodes, the resulting electrodynamic field repels and removes dust particles from the screens, regardless of whether the dust particles are initially charged or uncharged. Test results show that self-cleaning solar panels may be manufactured by incorporating electrodynamic screens that derive their power output from the solar panels. For the power output requirements with a three-phase pulsed high-voltage power supply, it is estimated that approximately 10 W may be needed from the solar panel per square meter of the solar panel, a small fraction of power only when cleaning is needed. Under normal atmospheric conditions in desert locations, only a few minutes of cleaning will be needed per day. More frequent cleaning will be needed during dust storms.
ACKNOWLEDGMENTS Both the theoretical and experimental studies were pursued jointly at the Department of Applied Science, University of Arkansas at Little Rock (UALR), Laboratory for Electromagnetic and Electronic Systems, Massachusetts Institute of Technology (MIT), Electrostatics and Surface Physics Laboratory, NASA Kennedy Space Center (KSC), NASA Jet Propulsion Laboratory, and Appalachian State University. The research was supported by a NASA Grant NRA 02-0SS-01 (ROSS-2002), JPL Contract No. 1263202. We are thankful to Rao Surampudi and Chester Chu of JPL for their continued support for EDS development. We are thankful to Sid Clements of ASU, Jim Mantovani of NASA KSC, Praveen Srirama, David Clark and Chris Wyatt of UALR, and Mark Zahn of MIT for their contributions to this study.
REFERENCES [1] K. Kurokawa, K. Komoto, P. van der Vleuten, D. Faiman (Eds.), Energy from the Desert. Practical Proposals for Very Large Scale Photovoltaic Systems, Earthscan, London, UK, 2006. [2] T. Markvart (Ed.), Solar Electricity, second ed., John Wiley & Sons, West Sussex, UK, 2000. [3] M.S. El-Shobokshy, F.M. Hussein, Degradation of Photovoltaic Cell Performance due to Dust Deposition on to Its Surface, Renewable Energy 3 (1993) 585. [4] M.K. Mazumder, A.S. Biris, C.U. Yurteri, R.A. Sims, R. Sharma, C.E. Johnson, et al., Solar Panel Obscuration by Dust in the Martian Atmosphere, in: K.L. Mittal (Ed.), Particles on Surfaces 9: Detection, Adhesion and Removal, VSP, Utrecht, The Netherlands, 2006, pp. 167e195. [5] M.A. Green, Solar Cells: Operating Principles, Technology, and System Applications, Prentice-Hall, Englewood Cliffs, NJ, 1980. [6] G. Landis, Mars Dust Removal Technology, J. Propulsion and Power 14 (1998) 126.
198
Developments in Surface Contamination and Cleaning
[7] D.Y. Goswami, F. Kreith, J.F. Kreider (Eds.), Principles of Solar Engineering, second ed., Taylor and Francis, Philadelphia, PA, 2000. [8] E. Aguado, Effect of Advected Pollutants on Solar Radiation Attenuation: Mojave Desert, California, Atmospheric Env. Part B. Urban Atmosphere 24 (1990) 153. [9] http://interestingenergyfacts.blogspot.com/2008/04/us-solar-energy-map.html [10] D. Mei, L. Xiushan, S. Lin, W. Ping, A Dust Storm Process Dynamic Monitoring with MultiTemporal Modis Data, in Proc. ISPRS Congress Beijing, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, vol. 37 (2008) 965e970. Part B7. [11] H. Chanson, Dust Storm over Brisbane on 23 September 2009 e During and After, Working paper, School of Civil Engineering, University of Queensland, Brisbane, Australia, 2009. See also: A. Ramachandran, Sydney Turns Red: Dust Storm Blankets City, The Sydney Morning Herald, September 23, 2009. http://www.smh.com.au/environment/weather/sydney-turnsred-dust-storm-blankets-city-20090923-g0so.html. [12] W.C. Hinds, Aerosol Technology. Properties, Behavior, and Measurement of Airborne Particles, second ed., John Wiley & Sons, New York, NY, 1999. [13] C.F. Bohren, D.R. Huffman, Absorption and Scattering of Light by Small Particles, Wiley Science Paperback Series, Chichester, UK, 1998. [14] J.R. Robison, C. Wyatt, J. Diaz, J. Zhang, R. Sharma, M.K. Mazumder, Development and Testing of Large Area Electrodynamic Screens for Deployment on Mars Missions, Proc. Electrostatic Society of America Annual Meeting, 2007 pp. 227e233. [15] J.R. Melcher, Traveling-Wave Induced Electroconvection, Phys. Fluids 9 (1966) 1548. [16] S. Masuda, Electric Curtain for Confinement and Transport of Charged Aerosol Particles, Proc. Conference on Electrostatics, June 8e11, Albany, NY, 1971. [17] S. Masuda, K. Fujibayashi, K. Ishida, H. Inaba, Confinement and Transportation of Charged Aerosol Clouds by Electric Curtain, Electronic Eng. Japan 92 (1972) 9. [18] F.M. Moesner, T. Higuchi, Traveling Electric Field Conveyor for Contactless Manipulation of Microparts, Proc. IEEE Indus. Applics. Soc. Mtg 3 (1997) 2004. [19] P. Atten, H.L. Pang, J.L. Reboud, Study of Dust removal by Standing-Wave Electric Curtain for Application to Solar Cells on Mars, IEEE Trans. Industry Applics 45 (2009) 75. [20] M.K. Mazumder, R.A. Sims, J.D. Wilson, Transparent Self-Cleaning Dust Shield, U.S. Patent 6,911,593 (2006). [21] H.A. Pohl, Dielectrophoresis, Cambridge University Press, Cambridge, UK, 1978. [22] J.A. Cross, Electrostatic Principles, Problems and Applications, Taylor and Francis, Adam Hilger Imprint, Bristol, UK, 1987. [23] M.K. Mazumder, R. Sharma, A.S. Biris, J. Zhang, C.I. Calle, M. Zahn, Self-Cleaning Transparent Dust Shields for Protecting Solar Panels and Other Devices, Particulate Sci. Technol. 25 (2007) 5. [24] C.I. Calle, C.R. Buhler, J.G. Mantovani, J.S. Clements, A. Chen, M.K. Mazumder, et al., Electrodynamic Shield to Remove Dust from Solar Panels on Mars, Proc. 41st Space Congress, April 27e30, Cape Canaveral, FL, 2004. [25] A.S. Biris, D. Saini, P.K. Srirama, M.K. Mazumder, R.A. Sims, C.I. Calle, et al., Electrodynamic Removal of Contaminant Particles and its Applications, in Conf. Record 39th IEEE IAS Annual Mtg., Seattle, WA, vol. 2, 2004 pp. 1283e1286. [26] S. Trigwell, Correlation between Surface Structure and Tribocharging of Powders, Ph.D. Dissertation, University of Arkansas at Little Rock, Little Rock, AR, 2003.
Chapter | 5
Electrostatic Removal of Particles and its Applications
199
[27] C.E. Johnson, P.K. Srirama, R. Sharma, K. Pruessner, J. Zhang, M.K. Mazumder, Effect of Particle Size Distribution on the Performance of Electrodynamic Screening Systems, in Conf. Record 40th IEEE IAS Annual Mtg., Hong Kong, vol. 1, 2005, pp. 341e345. [28] C.I. Calle, M.K. Mazumder, C.D. Immer, C.R. Buhler, J.S. Clements, P. Lundeen, et al., Controlled Particle Removal from Surfaces by Electrodynamic Methods for Terrestrial, Lunar, and Martian Environmental Conditions, Electrostatics 2007, J. Physics: Conference Series 142 (2008) 012073. [29] R. Sharma, C.A. Wyatt, J. Zhang, C.I. Calle, N. Mardesich, M.K. Mazumder, Experimental Evaluation and Analysis of Electrodynamic Screen as Dust Mitigation Technology for Future Mars Missions, IEEE Trans. Industry Applics. 45 (2009) 591.
This page intentionally left blank
Chapter 6
Alternate Semi-Aqueous Precision Cleaning Techniques: Steam Cleaning and Supersonic Gas/Liquid Cleaning Systems Rajiv Kohli, The Aerospace Corporation, NASA Johnson Space Center, 2525 Bay Area Blvd, Suite 600, Houston, TX 77058, USA
Chapter Outline
1. Introduction 201 2. Precision Steam Cleaning 202 3. Supersonic Gas-Liquid 212 Cleaning
4. Summary Disclaimer References
231 231 231
1. INTRODUCTION For many decades precision cleaning in industrial applications has involved the use of a variety of solvents, many of which are deemed detrimental to the environment [1,2]. In the late 1990s, many new concerns were raised for environmental protection. Concerns about ozone depletion, global warming, and air pollution led to new regulations and mandates for the reduction of chlorinated solvents, hydrochlorofluorocarbons (HCFCs), trichloroethane, and other ozone-depleting solvents. The search for alternate cleaning methods to replace these solvents has led to the consideration of various alternate cleaning systems [3e5 and references therein]. This chapter describes two alternate semi-aqueous cleaning methods, precision steam cleaning and supersonic gasliquid cleaning.
Developments in Surface Contamination and Cleaning. Copyright Ó 2011 Elsevier Inc. All rights reserved.
201
202
Developments in Surface Contamination and Cleaning
2. PRECISION STEAM CLEANING 2.1. Background Steam can be a practical and environmentally friendly method to clean without the use of solvents. Vapor steam cleaning has been used in a variety of applications from household to industrial to military to medical. Steam can be generated by simply boiling water; however, it is the combination of temperature and pressure that makes steam cleaning effective. A steam cleaner is a highly beneficial piece of equipment for all home, commercial, and industrial applications requiring sanitizing, disinfecting, and cleaning capabilities in a single machine. Most people are familiar with the corner car wash engine steam cleaners that remove oil, grease, and grime from the engine with a pressurized steam spray. Other types of steam systems are used in carpet cleaning and other household applications [6,7], although in many cases, these cleaning systems employ hot water rather than steam for cleaning. The real steam cleaning systems are growing in popularity because of the ability of steam vapor to kill germs and microbes and, in some cases, disinfect without the use of chemical disinfectants [8]. Steam vapor has also been shown to be effective in killing dust mites in carpet, bedding, and upholstery [7,9]. Television infomercials tout the ability of small steam generators to clean household dirt, grime, dust, and germs among other contaminants [10]. These household systems generally operate at low pressure (0.3e0.6 MPa). Large commercial steam systems operating at higher pressure (up to 2 MPa) are utilized in industrial applications (often utilizing boilers and pressure vessels) as degreasers and for cleaning tools and large surfaces [10,11]. One novel application is weed control with steam which has an interesting fringe benefit of sterilizing the soil [10]. Medical applications call for sterilization and disinfection of surfaces such as tools, hospital rooms, and restrooms with steam [8]. In 2005, the University of Washington tested a steam vapor system in restrooms and reported labor savings and hygienic improvements compared with traditional cleaning methods such as mechanical scrubbing with detergent solutions [13]. A recent innovation involves the addition of low zeta potential mineral crystals to the source water to produce an enhanced type of steam vapor that has been shown to kill a broad range of microorganisms after 3e5 seconds’ exposure to steam [14]. Kunze-Concewitz [15] had proposed a method using both water and steam for removing particle residue from chemical mechanical planarization of semiconductor wafers and other applications that require super-clean surfaces, such as manufacture of liquid crystal display products and magnetic disks for hard drives. Steam can also be used for precision cleaning applications where contamination must be removed at a microscopic level. Due to its fluid properties as a vapor (low viscosity), steam has the ability to penetrate inaccessible areas
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
203
where other fluid cleaning methods may be unsuccessful. This is especially important when cleaning printed circuit boards to penetrate blind vias and holes and to penetrate contacts and terminations to remove residual flux that may cause increased electrical resistance and corrosion. Precision steam cleaning is used to remove adhesive, flux, oil, grease, fingerprints, and other contaminants from optics and electronic components, automotive components, various components in the defense industry, jewelry and gems, medical instruments, and dental implants [16e25].
2.2. Principle of Steam Cleaning Plain water is not an effective medium for dissolving hydrocarbons such as soils and greases, but it has the capability to dissolve many inorganic compounds. By using the principle that chemical reaction rates increase with temperature, water heated to create superheated steam becomes a very effective cleaning agent for hydrocarbon compounds. The combination of moisture, heat, and pressure of superheated steam provides the means for immediate removal of contaminants from a surface. The hydrocarbon compounds become less viscous at higher temperatures (some of these compounds may even melt), making it easier to remove them from the surface. For solid contaminants, cleaning is primarily breaking the bond between the contaminant and the surface and the high-pressure steam spray becomes an effective removal agent for solid contaminants. Furthermore, steam also exhibits the fluid properties of a gas which makes it very effective in penetrating the surface boundary layer and accessing tight spaces. The viscosity of steam at 573 K is in the range 20.279 to 20.076 mPa.s for pressures from 0.2 to 2 MPa as compared with the viscosity of liquid water at 373 K in the range 281.74 to 282.22 mPa.s for pressures from 0.2 to 2 MPa [26].
2.3. Description of Steam Cleaning System and Equipment [16e25] A suitable water source is used to pump a metered amount of deionized or otherwise filtered or purified water into the steam generator. A dispersing nozzle or atomizer may be incorporated in the steam generator which enhances steam generation because it disperses the incoming water into small water droplets and directs them onto the heated inner surfaces of the steam generation pot. The water droplets convert instantly to high-pressure superheated steam. The steam which is generated leaves the steam generator by means of a conduit located in the upper half of the generator. The superheated steam is carried to a suitable external nozzle or wand assembly and can be directed as a jet spray onto the part to be cleaned. An operator-activated switch cyclically draws water into the pump to initiate the steam generation process. For smaller
204
Developments in Surface Contamination and Cleaning
single tank models, a short recovery time is required for steam to be generated for each cycle. Moisture content of the steam can range from 5% (dry steam) to 99% (wet steam, a combination of steam and hot water). Dry steam systems use very little water because the vapor is created at a high temperature with very low moisture content between 5 and 6% water. These systems use high pressures but less water for precision and delicate cleaning tasks. A quarter cup of water produces about 1135 liters of steam. Average water use rates range from 3.78 to 19 liters per 8-hour work shift. The steam pressure needed for the systems to obtain the heat for cleaning is produced by a boiler made primarily from stainless steel. Stainless steel offers higher safety margins and is more resistant to scaling, pitting, and corrosion which can contaminate the substrate being cleaned. The part surfaces are often warmed enough during cleaning to dry quickly. However, a drying stage may be needed to prevent oxidation or corrosion of sensitive parts. Flash rusting may be a problem for some materials and preventive measures, such as the use of a rust inhibitor, may be required for such parts. Two types of cleaning units are available. Single tank units have a boiler that is filled directly with water. When the machine runs empty of water, it will need to be cooled before opening the safety pressure cap and refilling with water. The bigger the boiler capacity the longer is the operating time. Modern units also offer a continuous-fill system which consists of a separate nonpressurized, non-heated water tank and the boiler. This allows the machine to be refilled at any time without waiting to cool down the machine. The level of the water in the boiler is constantly detected by a special sensor. This means no downtime because there is always a constant amount of water in the boiler. The flash-heat design typically requires 6e8 minutes’ start-up time, although one model has a start-up time of 1 minute [18]. Sloan and Bellegarde [27] developed a new heating system that has been incorporated into the micro precision steam cleaners offered by VaTran Systems [19]. This new heating system in the boiler cavity employs a smooth heating surface made from material with interconnecting pores that very effectively overcomes the Leidenfrost effect [28]. The Leidenfrost effect is a phenomenon in which a liquid drop impinging on a surface significantly hotter than the boiling point of the liquid immediately forms an insulating vapor layer (approximately 0.06 mm thickness at 433 K [29,30]) which decreases the heat transfer from surface to the liquid and keeps the liquid from boiling rapidly. The pores in the heating surface act as escape passages for the insulating vapor layer, which reduces the thickness of the layer and increases the contact of the droplets with the heated surface. This significantly enhances the conversion of impinging water droplets into steam compared to other systems [31e35].
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
205
The steam cleaners for precision cleaning applications can be provided with continuous steam spray capabilities without the use of boilers and pressure vessels. This is an important safety consideration since pressurized steam storage requires the use of pressure vessels that must meet strict safety requirements. Most models superheat steam for on-demand use of up to 20 minutes, which is long enough for most cleaning applications. Various models are equipped with other safety features such as an audible or visual water-level indicator, automatic check valve to prevent pressure build-up, and automatic shut-off feature. Most models come with a trigger wand assembly and interchangeable nozzles, including single jet for a straight stream; fan jet for a flat spray to clean larger areas; and peripheral nozzles for hole and tube cleaning. The steam delivery system could be hand-held (pistol grip) which affords the operator direct steam control. It allows the user freedom of movement when working on small parts. By contrast, the stationary delivery systems have built-in steam nozzles which allow the operator to grasp the part with two hands for greater control. Precision steam cleaners are lightweight (6e25 kg) and are easily transportable wherever cleaning is required. The waste management system usually includes a cabinet, exhaust fans, spotlights, and cart. Clear vinyl curtains enable the operator to see products being cleaned and avoid splashing. The waste management system has a removable drip pan to capture any contaminant residue. Steam cleaners use very little water to do a large amount of cleaning, which means less waste for disposal. The typical range of specifications of commercial precision cleaning units is given below. Steam cycle time range Voltage Frequency Power requirement Steam pressure Steam temperature Nozzle Tank capacity External dimensions Weight Internal materials of construction System operation
Start-up time
1e20 minutes 100e240 VAC 50e60 Hz 1700e3000 W 0.2e2 MPa 373e573 K Fixed single jet or multiple jets; customizable for given application 3e5 liters 21e38 cm 33e44 cm x 20e33 cm 8e25 kg Stainless steel, brass, TeflonÒ hose. Wetted brass surfaces are suitable for non-contamination-sensitive applications such as jewelry cleaning Single tank for intermittent operation; twin tanks for continuous operation; foot- or hand-operated pedal; pistol-grip or straight nozzle 1e30 minutes
206
Developments in Surface Contamination and Cleaning
Examples of different precision steam cleaners are shown in Figure 6.1. For medical instrument cleaning applications, the instruments frequently need a manual pre-cleaning step to remove contaminants built up in difficult-toaccess locations, primarily due to the design, shape, and construction of the instruments. Often, these contaminants are not removed when the instruments are cleaned in automatic washers before reuse. Recent surveys in several hospitals in Germany indicated that between 5 and 15% of the surgical instruments required a pre-cleaning step to effectively remove all residual contaminants after cleaning/disinfection for reuse of the instruments [36].
FIGURE 6.1 Examples of commercial precision steam cleaners. (a) VaTran micro precision steam cleaner. (b) Elmasteam high-performance steam jet cleaners. (c) Steamshine Lotus model steam cleaners. (d) Reliable Model i700A. (e) Mini-Max cleaner [19e23]. Courtesy of VaTran Systems, Chula Vista, CA; Elma GmbH, Singen, Germany; C.D. Nelson Consulting, Lakemoor, IL; Reliable Corporation, Toronto, Canada; and PDQ Precision, San Diego, CA
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
207
A variety of attachments and adapters (Fig. 6.2) are available for precision steam cleaning of difficult-to-clean medical instruments such as endoscopes, catheters, and orthopedic and dental handpieces [36,37]. A Luer-Lock connector [38,39] allows instruments with the respective coupling device to be connected directly to the handpiece, so that the steam spray can be guided into the rinsing channel of the instrument to clean precisely where it is required without any pressure loss (Fig. 6.3).
2.4. Operating Considerations The first consideration when utilizing vapor steam in precision cleaning applications is the quality of the water used. Tap water contains many contaminants including minerals, silicates, and organic matter, so it is unsuitable for precision cleaning applications. Use of high-purity filtered, deionized, or distilled water is absolutely essential for such applications. Another consideration should be whether the parts being cleaned can withstand the temperature and pressure of the steam spray. Precision steam cleaners can be provided with a variety of outlet pressures and temperatures. Steam saturation quality, supply of wetter or dryer steam, and steam pressure can be adjusted for specific applications for contaminant removal. Cleaning units have steam pressures ranging from 0.14 MPa to over 2 MPa, while temperatures can range from 373 K to 573 K. The temperature decreases very rapidly in the steam vapor plume. Heat generated is minimal for electronic
FIGURE 6.2 Adapters and attachments for precision steam cleaning surgical instruments. (a) Long and short hollow needles with steam vents along the side. (b) Hollow straight needle attachment with steam vent at the tip. (c) Hollow curved needle attachment with steam outlet at the tip. (d) Luer-Lock adapter. (e) Catheter attachments. (f) Hose connection. (g) Mounting pliers. (h) Rack [36]. Courtesy of Elma GmbH, Singen, Germany
208
Developments in Surface Contamination and Cleaning
FIGURE 6.3 Steam cleaning of medical instruments with different adapters. (a) Straight bore needle. (b) Hollow curved bore needle. (c) Direct connection with the Luer-Lock interface [36]. Courtesy of Elma GmbH, Singen, Germany
components such as circuit boards. By controlling the distance from the steam outlet nozzle, temperature-sensitive parts can be cleaned without exposure to unnecessarily high temperatures. Recent testing on aluminum alloy 7075 strips (0.5 mm thick 24.5 cm wide 48.3 mm long) has shown the temperature of the surface was between 311 K and 318 K 10 seconds after removal of steam from the surface [40]. The maximum temperature of the surface was 253 K at 25 mm steam exit distance from the surface to as high as 372 K at 0.64 mm from the surface [40]. Condensed water can penetrate and/or damage joints, seals, and bonded areas, and limits the usefulness of steam vapor cleaning. Preventive measures, such as air blowers, may be required to remove residual moisture. Steam vapor cleaning is optimal when individual components are removed or disassembled from the hardware as part of the normal task. However, disassembly can be time- and labor-intensive. Additives such as detergents, alcohol, or saponifiers are generally not added to the water since most additives will break down at the high temperatures needed to make steam. Often these additives are sprayed on the parts prior to steam cleaning. Caution should be used in operating steam cleaners since hot surfaces, nozzles, and substrates can result from cleaning. Steam pressure is also a safety
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
209
consideration. Personal protective equipment (safety glasses, gloves, appropriate work clothing) should be employed. Operators must be trained properly to use the steam cleaning equipment efficiently and safely. Often, the equipment is employed in an enclosed cabinet to capture the contaminants and to minimize worker exposure to debris from the part being cleaned.
2.5. Advantages and Disadvantages of Precision Steam Cleaning 2.5.1. Advantages l l
l
l l l
l l l l l l l l l l
Eliminates the use of solvents Reduces the amount of hazardous waste and hazardous air emissions generated compared to solvent degreasing Wastewater stream is generally compatible with conventional industrial wastewater plants Low implementation cost utilizing simple equipment Provides solvent cost savings Ideal for removing grease, oil, flux, adhesive, fingerprints, and other contaminants Low water usage Safe, clean design generates superheated steam on demand No high-pressure steam storage Lightweight, portable design for point-of-use applications Auto shut-off feature safeguards heater Steam can reach otherwise inaccessible areas and spaces Elimination and control of the biofilms that resist typical disinfectants Effective technique as a manual pre-cleaning step for medical devices Steam can be used very selectively and precisely Steam cleaning is very handy and fast.
2.5.2. Disadvantages l
l l
Not recommended for temperature- or moisture-sensitive parts. Rusting or surface oxidation may occur Risk of electrostatic discharge (ESD) when cleaning electronic components Damage to joints, seals, and bonds from residual moisture.
2.6. Applications Steam for precision cleaning can be used for a variety of applications. l
l
Removing water-based (no-clean) flux from circuit boards and other soldering applications Cleaning fiber optics and other optical components
210 l l l l l
l l
Developments in Surface Contamination and Cleaning
Cleaning hybrid circuits prior to wire bonding Cleaning medical devices and instruments Cleaning dental implants Cleaning and degreasing aerospace tools and parts Cleaning military hardware such as weapons, automotive parts, electronics, ground support equipment, and other gear Cleaning automotive components to remove grease and grime Cleaning jewelry and gems.
Sloan [41] demonstrated the effectiveness of micro precision steam cleaning on 304 stainless steel substrates. Ten 0.1 m2 plates were pre-cleaned and individually contaminated with approximately 8e16 mg/m2 of dust and drilling lubricant as determined by weighing. The samples were dry steam cleaned for approximately 2e3 minutes and re-weighed. The remaining contamination ranged from 1 to 2 mg/m2; the cleaning efficiencies were between 75 and 93%, with an average cleaning efficiency of 86%. Visual inspection of the cleaned surface showed uniform removal with no evidence of visible contamination on the surface. The prevailing method of cleaning aircraft ejection seats in the US Navy employs an organic solvent such as isopropyl alcohol. This method of cleaning the ejection seat and/or ejection seat components is extremely labor-intensive and generates significant quantities of hazardous waste for disposal. Kwan et al. [42] performed an evaluation of an alternate cleaning method using a wet steam cleaning system. The steam vapor cleaning process is found to be optimal when individual components of the ejection seats were removed or disassembled for cleaning as part of the normal task. This ensured that water was not trapped. But disassembly was time- and labor-intensive. Concern about the deterioration or damage to the O-ring seals was not realized; however, the steam cleaning process did remove the lubricants on the seals and had to be replaced. Panzner et al. [43] investigated cleaning of three gilded objects: fire-gilded copper pipe; a leaf-gilded spire; and a festive bonnet made of fibers wrapped by chemically gilded silver wire. Micro precision steam cleaning was successful in removing surface contamination, although some surface damage was observed. This is not unexpected if the surface coating layer is very thin, as in the case of the gildings which were only 10 mm thick. The effectiveness of precision steam cleaning to decontaminate microbially contaminated surfaces was investigated recently [8]. Dried films of 11 different microorganisms (E. coli, Shigella flexneri, Enterococcus faecalis, Salmonella enterica, Staphylococcus aureus, Pseudomonas aeruginosa, S. aureus, Candida albicans, Asperigillus niger, Bacteriophage MS2, and Clostridium difficile) were deposited on porous clay tiles that are more representative of surfaces found in the health care and food service industries. The initial concentrations of the microorganisms ranged from 1.08 104 to 4.83 106 CFU per test surface (colony-forming unit (CFU) is a unit of measurement for
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
211
viable bacteria numbers). The test surfaces were disinfected with a portable steam cleaner for 0.5, 1.0, 2.0, or 5 seconds. Individual colonies were counted and the results recorded as CFU per test surface. The results showed the high concentration of the diverse array of microorganisms tested was completely destroyed very rapidly. For example, the initial concentration of Salmonella and E. coli of 9.75 104 and 1.14 106 CFU, respectively, was reduced to 3.16 and 5.98 103 CFU after 2 seconds’ exposure and to 0 CFU after 5 seconds’ exposure to steam. Similar studies in Italy showed the efficacy of precision steam cleaning in destroying six bacteria strains (E. coli, Staphylococcus aureus, Streptococcus faecalis, Bacillus cereus, Saccaromyces cerevisiae, and Pseudomonas aeruginosa) deposited on porcelain tiles, TeflonÒ, and stainless steel substrates after 5 seconds’ exposure to steam [44,45]. Industrial and government organizations worldwide involved in precision cleaning are increasingly adopting precision steam cleaning for their precleaning and/or their final cleaning operations for many different applications. The US Department of Defense (DoD) has recommended the adoption of steam cleaning in all service branches as a viable alternative to solvent cleaning of components and parts such as electronics, weapons, printed circuit boards, and other items. Significant cost savings have been realized at different DoD installations by employing precision steam cleaning [16,17]. Steam cleaning of surgical instruments and medical devices has also found increasing acceptance in hospitals worldwide, primarily due to the ability to reach inaccessible locations and the effectiveness of steam cleaning in destroying microorganisms [8,36,44,45]. Recently, ITT designed and built a new facility equipped with modern precision cleaning equipment, including micro precision steam cleaners [46]. The company has incorporated micro precision steam cleaning in the new facility for cleaning mechanical parts and assemblies. ITT cleans parts that can range in size from a single threaded fastener to large composite structures. Materials that are processed include optical glass, ceramics, composites, metals, and various high-performance coatings. First-pass yield of 98% has been achieved in the new facility with annual savings of $1 million, part of which is attributable to the use of micro precision steam cleaning. Duffy et al. [47] have recently developed a novel four-stage automated process to efficiently clean and dry fiber optic endfaces in high-density multiferrule connectors. The system incorporates a micro precision steam cleaning as the first process step, followed by CO2 snow cleaning for final precision cleaning. The parts are dried in flowing heated ionized air. All process steps are carried out with a single multistage programmable cleaning system. It has been used to successfully demonstrate the removal of various contaminants, including fibers, dust, oils, and other residues with better than 90% efficiency. Other examples of applications and case studies of steam cleaning are given by the various commercial vendors of cleaning systems [19e24,48].
212
Developments in Surface Contamination and Cleaning
2.7. Cost Benefits Significant cost savings can be achieved by steam cleaning compared to solvent cleaning. For example, the total annual operational costs to clean 120 guns per day were estimated to be approximately $217 000 using a portable steam cleaner compared with approximately $660 000 for solvent cleaning for estimated savings of $443 000 [16,17]. Similar savings of ~$308 000 were achieved by steam cleaning of aircraft ejection seat frames and components compared with solvent cleaning [42].
3. SUPERSONIC GAS-LIQUID CLEANING 3.1. Background Commercial spray cleaning systems are well established for cleaning various types of parts and surfaces. These systems employ high pressures for cleaning, use large quantities of fluids, and are unsuitable for precision cleaning. If solvents are also used for cleaning, they pose a significant waste disposal problem, even if the solvents are not regulated and are approved for use. As an example, the Kennedy Space Center (KSC) of the National Aeronautics and Space Administration (NASA) was processing approximately 250 000 small and large components (such as valves, pipes, regulators, flexible hose lines, and compressed gas cylinders) through the cleaning facility each year, consuming as much as 27 000 kg of solvent (CFC 113) for cleaning and verification purposes [49,50]. Historically, at KSC precision cleaning to remove surface contamination (particulate and non-volatile residue) was to flow large quantities of water with detergents at high temperature or to flow large volumes of CFC solvents over the components, which tended to be a significant enviornmental problem. The strict cleanliness requirements were derived from liquid oxygen (LOX) system compatibility, since particles and hydrocarbon greases and oils can easily ignite in the presence of LOX. Subsequently, the level of cleanliness was verified according to NASA specification [51] that requires quantitative measurements of particle contamination (number of particles per unit area) and non-volatile residue (NVR as mass per unit area) for space components designed for LOX service and for other fluid systems where high levels of cleanliness are required. This was performed by flushing the defined surface area (0.1 m2) with a CFC solvent. For particle contamination, the sovent was poured through a filter and the particles were counted using an optical microscope. The NVR analysis was performed gravimetrically by evaporating the collected solvent and weighing the remaining solid residue. An effective method was needed at KSC to replace this system of cleaning and verification employing such large volumes of solvents. Other cleaning methods like high-pressure water jets, ultrasonic cleaning, and flushing with approved solvents had several disadvantages. Although
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
213
ultrasonic cleaning is very effective in removing contaminants, the component must be immersed in the cleaning fluid. The size of most components cleaned at KSC made this impractical. Water jet cleaning disadvantages included high volumes of water consumption and the relatively high pressures required for effective contaminant removal. In spite of their excellent solvency potential, cleaning and flushing with even approved solvents does not solve the issue of consumption of large volumes of used solvent which have to be processed, and the tendency for solvent flushing to leave behind insoluble contaminants. To overcome these disadvantages, the supersonic gas-liquid cleaning (SSGLC) concept was developed of accelerating liquid droplets to supersonic velocities (>Mach 3) by using a converging/diverging nozzle and medium pressure (~2 MPa) compressed inert gas or air [52e55]. SS-GLC provided advantages over alternate cleaning methods. Benefits over water jet cleaning included lower operating pressures, less water consumption, and greater surface impingement capability. Benefits over solvent flushing included the elimination of solvent consumption and the removal of insoluble particles by impingement. Using a spraying nozzle meant that large surfaces that could not be immersed in an ultrasonic bath could be cleaned. SS-GLC also provided several advantages over other pressurized cleaning methods. The system did not abrade the surface of the parts being cleaned, and it required much lower levels of pressure while using very little water. These features enabled the system to clean a wide variety of items [56]. In medical applications, removal of solid contaminants from and cleansing of exposed in vivo tissue is necessary during surgical procedures [56]. In addition, such cleansing is necessary in preparation for treatment for dental conditions such as gingivitis, caused by the long-term effects of plaque deposits. Organic matter tends to bond to tissue much more strongly than nonorganic matter, and is generally more difficult to remove than non-organic matter such as fibers, dust, and sand particles. Cleaning with a liquid such as water is often ineffective in removing the particles that are smaller than the thickness of the stagnant laminar boundary layer (flow velocity is zero at the surface) which is formed on the tissue surface [57]. The particles located in the boundary layer have a sufficiently high drag force that cannot be overcome by a liquid stream even with a very high overall velocity. Several cleansing devices have been developed to improve cleaning of a variety of surfaces and systems using one or more of these devices and are available commercially [58e79]. These systems provide a liquid stream with a reduced boundary layer thickness employing liquid and gas nozzle assemblies; a high velocity aerosol of at least partially frozen particles; and pulsed jets of liquid sprayed on to a metal surface to remove small particles. Many of the dermal abrasion and cleansing systems use relatively high liquid flow rates which reduce the cleansing and scouring effect due to the virtually stagnant boundary layer that develops over the surface to be cleaned. High
214
Developments in Surface Contamination and Cleaning
velocity can also cause damage to the surface. Other devices have low flow rate, low-pressure nozzles for mixing fluids, but they employ Venturi tube injection to atomize the liquid, which cannot achieve supersonic velocity in the mixture. In general, these devices provide only a small improvement over non-pulsed spray cleaning systems. The principles of SS-GLC have been applied to develop devices for multiple tissue cleansing and dermal abrasion applications [80e82] that overcome the limitations of other methods. The SS-GLC devices use very small amounts of the cleansing liquid to form a mist of droplets suspended in a highvelocity gas. The small amount of liquid suspended as a mist prevents the formation of a liquid boundary layer which could trap small particles. The gasliquid mixture is accelerated to supersonic velocity and is delivered to the tissue surface, mass, or cavity to be abraded, thereby very effectively scouring and cleansing the surface.
3.2. Principle of Supersonic Gas-Liquid Cleaning SS-GLC is effective in removing particle and non-particle (hydrocarbon film) contaminants. The system mixes gas and liquid from separate pressurized sources; the liquid is suspended as fine droplets in the gas stream. The nozzle has a convergingediverging geometry. Assuming homogeneous and adiabatic uniform velocity with thermal equilibriun (rapid heat transfer) between the gas and the liquid, the compressibility of the gas can accelerate the atomized liquid particles to supersonic velocities [53,54]. Recent work has shown that the assumption of rapid heat transfer between the gas and liquid phases is incorrect; rather, the assumption of no heat transfer between the phases may be more appropriate in describing nozzle flow since liquid droplets have been observed in the jet exiting the nozzle [83]. At the same time, a stagnant liquid boundary layer cannot form on the surface because of the small quantity of liquid used for cleaning. The gas-liquid mixture is ejected at supersonic speeds from one or more nozzles at the end of a handheld wand assembly. At these speeds, the liquid droplets suspended in the gas have the kinetic energy to forcibly dislodge the solid contaminants from the surface, dispersing them into a minimal waste stream. Even small particles that, due to their size may be trapped in a liquid boundary layer, can be removed. The dominant mechanism for hydrocarbon residue removal is due to emulsification upon liquid droplet impact with the target surface [53,84]. The supersonic nozzle tends to emulsify hydrocarbon contaminants, so that the concentration exceeds the contaminant solubility limit in the liquid. The emulsification process is dependent on the size and concentration of the liquid droplets present in the mixture, as well as nozzle design and injection arrangement [84e87].
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
215
3.3. Description of the Method and Equipment 3.3.1. Precision Cleaning Applications For precision cleaning applications, SS-GLC operates by flowing high-pressure air or nitrogen through a throttling valve to the nozzle. Water is injected into the gas flow stream through an inlet orifice upstream of the converging/diverging section of the nozzle. The nozzle design is based on an area ratio (ratio of exit area to the throat area) of 5.44 which gives a Mach number of 3.14 corresponding to a velocity of approximately 1067 m/s (Fig. 6.4) [53]. At this point the rate of change in Mach number with area ratio begins to decrease significantly. More recent work has shown that the measured velocity, 630 m/s, and the computed velocity, 670 m/s, of the gas-liquid mixture are in good agreement, but approximately two-thirds of the expected velocity based on the original design calculations [84]. Two different nozzle designs have been developed and are commonly employed in cleaning applications. In the conventional convergingediverging nozzle, the two-phase jet discharging from the conventional nozzle diverges, thus creating a wider jet with smaller
FIGURE 6.4 Mach number vs. area ratio for the KSC SS-GLC convergingediverging nozzle [53]
216
Developments in Surface Contamination and Cleaning
concentration of liquid droplets at the cleaning surface. By contrast, the jet in the annular nozzle converges at the exit and reaches its greatest concentration a short distance downstream of the nozzle outlet (Fig. 6.5) [86e88]. The jet diameter and intensity are narrower and the concentration of droplets at the target surface is higher than the conventional nozzle at the same pressure and flow rate. The conventional nozzle can cover a larger surface area than the annular nozzle. Supersonic exit velocities can be achieved without an inordinately large exit cone in the nozzle. The mixed gas-liquid flow then enters the converging/diverging nozzle where it is accelerated to supersonic speeds. The supersonic gas-liquid stream is directed onto the surface of the components that require cleaning or cleanliness verification (Fig. 6.6). The velocity imparted to the water by the gas flow gives it sufficient momentum at impact to remove contaminants on the surface of the component being cleaned or verified while simultaneously dissolving the contaminant into the water, which can be captured for cleanliness verification. The flow parameters for the gas-liquid nozzle can be set so virtually any gas and liquid can be used for a desired flow and mixing ratio. In addition, the size and number of nozzles are adjustable. This adjustability makes it possible to create sizes ranging from small hand-held cleaning nozzles to very large multiple-nozzle configurations. For cleanliness verification the cleaning fluid is replaced with water which can be collected for analysis after spraying the surface of the cleaned part (Fig. 6.7). The small volume of cleaning fluid results in reduced solvent usage and the resultant cost of hazardous waste disposal. A commercial SS-GLC system based on this design is shown in Figure 6.8 [89]. The system accommodates the use of distilled or deionized water and the use of compressed breathing air or nitrogen. All wetted parts are fabricated from stainless steel or TeflonÒ. Cleaning and drying functions are controlled
FIGURE 6.5 Convergingediverging nozzle with flow-directing annular insert [86e88]
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
217
FIGURE 6.6 Schematic of the basic supersonic gas-liquid cleaning system for precision cleaning applications [53,54]
from a single trigger. Since the environmentally friendly system requires less than 100 mL of water per minute, there is very little liquid left after cleaning that must be handled as contaminated waste. The system is non-abrasive due to the low mass energy of the atomized water, approximately 0.13 10e6 kg-m/s per 1 mm size water particle, as compared with other spray cleaning methods (see Table 6.1). With a nozzle that can be oriented in any direction, the system is adjustable to allow all sides of a part to be cleaned without reorientation. Designed for operator safety and comfort, the system requires minimal training for operation
FIGURE 6.7 Supersonic gas-liquid cleaning system arranged for precision cleaning and cleanliness verification [54]
218
Developments in Surface Contamination and Cleaning
FIGURE 6.8 Commercial SS-GLC system for precision cleaning applications. Courtesy of VaTran Systems, Chula, Vista, CA
and is easily moved on built-in casters, despite its weight (~200 kg). When operating the SS-GLC adequate hearing protection is required during operation due to the supersonic velocities. Maintenance is minimal with only a few moving parts.
3.3.2. Medical Applications The SS-GLC system developed for medical applications employs a convergingediverging gas nozzle and a liquid discharge nozzle arranged concentrically within the gas discharge nozzle. The gas nozzle configuration is operated such that the inlet gas pressure is at least twice the outlet gas pressure. This pressure drop causes a shockwave in the gas and, depending on the gas pressure, accelerates it to velocities ranging from subsonic to supersonic. At the same time, the liquid flow downstream of the gas discharge nozzle is atomized and forms a mist of liquid droplets (5e100 mm) suspended in the flow of discharged high-velocity gas (Fig. 6.9) [80e82]. Gas (air, oxygen, carbon dioxide, and nitrogen) is supplied from the pressurized gas source at a pressure of 0.28e1 MPa and liquid is supplied from the pressurized liquid source at a pressure in the range 0 to 0.034 MPa. The mist jet
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
219
TABLE 6.1 Typical parameters of different particle removal techniques [90]
Impacting Cleaning particle method size, mm
Impacting Impacting particle particle density, mass, kg kg/m3
Impacting particle velocity, m/s
Momentum transferred per impacting particle, kg-m/s (10e6)
Dry ice pellet blasting
3.18 dia 6.35 long
1562
7.84 10e5
~335
27 652
CO2 particles
<1
1562
8.16 10e7
46
37.33
1562
e7
305
248.87
1.31 10
e6
~0.3
0.41
e9
335
0.45
0.5
CO2 snow 1.48
780
8.16 10
e3
930
1.32 10
Water ice 1.02 dia (large 6.35 long particles)
930
1.9 10e5
162
3071
1000
1.18 10e10
1067
0.13
1067
64.84
Water ice 70 10 (small particles)
SS-GLC
1 10e3 e3
8 10
1000
6.01 10
e8
delivery nozzle arrangement includes at least two gas discharge nozzles or at least two liquid discharge nozzles. A suction conduit is included in the mist jet delivery nozzle arrangement which provides for removal of waste liquid and abraded tissue particles. The device is designed to be used while being held in one hand. A system based on this technology is commercially available for different applications. The JetOxÔ system (Fig. 6.10) is used for wound cleansing and debridement (debridement is the surgical excision of dead, devitalized, or contaminated tissue and removal of foreign matter from a wound [91]) [92e98]. For this application, the system employs medicalgrade oxygen and a sterile cleaning liquid. Saline solution (0.9% sodium chloride) is most commonly employed, although other solutions have been used successfully [99e103]. The spray is precisely calibrated to treat only the affected areas. A unique shield is attached to the delivery nozzle to prevent contamination. The JetPeelÔ system (Fig. 6.11) is employed for dermal applications [104,105]. It contains a control unit and a unique disposable handpiece that
220
Developments in Surface Contamination and Cleaning
FIGURE 6.9 Schematic of the cleaning system for medical applications [80e82]
includes the delivery nozzles. The fine, single-nozzle handpiece (Fig. 6.12a) is suitable for non-invasive mesotherapy, wrinkle and acne treatments, while the triple nozzle handpiece (Fig. 6.12b) is used for most other skin treatments. The system is lightweight and ergonomic and requires only minimal dexterity from the operator. The skin peeling depth can be precisely controlled by controlling the gas pressure and the number of passes, making it possible to individually treat different areas of skin simultaneously without collateral damage. JetPeelÔ is capable of removing the epidermis layer of the skin, thereby increasing tissue oxygenation which contributes to accelerated wound healing [106].
FIGURE 6.10 The JetOxÔ wound cleansing and debridement system [92,93]. Courtesy of TavTech Ltd., Yehud, Israel
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
221
FIGURE 6.11 JetPeelÔ system for dermal applications [104]. Courtesy of TavTech Ltd., Yehud, Israel
222
Developments in Surface Contamination and Cleaning
(a)
(b)
FIGURE 6.12 Handpieces for aesthetic application used with the JetPeelÔ system. (a) Singlenozzle handpiece. (b) Triple-nozzle handpiece [104]. Courtesy of TavTech Ltd., Yehud, Israel
3.4. Advantages and Disadvantages of Supersonic Gas-Liquid Cleaning 3.4.1. Precision Cleaning 3.4.1.1. Advantages l l l l l
l l l l
Uses very little cleaning liquid (<100 mL/min) Reduced hazardous waste volume disposal and associated disposal costs Can be used for cleaning and for cleanliness verification Does not abrade the surface Adaptable design for cleaning large surfaces and parts with complex geometries Portable Few moving parts Operator-friendly system Minimal training required for operation.
3.4.1.2. Disadvantage l
Requires tailoring as cleanliness verification tool for specific contaminants and applications.
3.4.2. Medical Applications 3.4.2.1. Advantages l l l
l
Uses very little cleaning liquid (<2 mL/min) Versatile system that can use many different cleaning liquids Effective in removing very small solid contaminants that could be trapped in a surface boundary layer Enables non-invasive trans-dermal delivery of liquids (nutrient supplements, vitamins, and anti-aging solutions)
Chapter | 6
l l l l l l
l
Alternate Semi-Aqueous Precision Cleaning Techniques
223
Generates minimal waste with reduced associated disposal costs Simple rapid setup for use Operator-friendly, single-handed, safe operation Short learning curve Minimal training required for operation Precision application to desired areas for cleansing with minimal collateral damage Minimal pain and rapid post-treatment healing.
3.4.2.2. Disadvantage l
More time-consuming than traditional debridement and dermal abrasion treatments.
3.5. Applications 3.5.1. Precision Cleaning Validation testing of the cleaning performance of the system developed at KSC was performed by Caimi and Thaxton [53], which required a correlation between the NVR remaining on the surface after cleaning and the total organic carbon (TOC) reading of the water sample collected after cleaning. Stainless steel witness plates with an area of ~0.09 m2 were contaminated with a known quantity (11.1 to 111 mg/m2) of fluorinated greases (such as KrytoxÒ e DuPont, or TribolubeÒ e Aerospace Lubricants Inc.) and other common lubricants used at KSC, and then impinged between 2 and 8 minutes each. The TOC measurements were found to be linearly correlated with the known initial contaminant level (Fig. 6.13) and with the remaining NVR after cleaning (Fig. 6.14). The results also indicated that the nozzle emulsifies the hydrocarbon contaminants well enough not to require another cleaning step. Similar linear correlations were observed in extensive testing with other contaminants and mixtures of contaminants deposited on valve bodies and witness plates from 0.05 to 0.75 m2 [49,50,52,107,108]. SS-GLC was shown to be a consistent technique for cleanliness verification of spaceflight hardware at KSC. The data collected also suggested that the system cleans the components more completely than the previous method using solvent (CFC-113). A similar investigation was carried out to test the effectiveness of the KSC SS-GLC system for cleaning surfaces [89]. The samples consisted of ~0.1 m2 304 stainless steel plates contaminated with a known quantity (7e18 mg) of dust and drilling lubricant, oils, and hydrocarbon and fluorinated greases. The contaminated area on each plate was cleaned by impinging for 2 minutes with the SS-GLC system. SS-GLC tends to emulsify hydrocarbon contamination, so that it exceeded the contaminant solubility limit in water and flows off the surface. The cleanliness verification method measured the TOC in water samples collected from cleaning. Eleven of 20 plates were completely cleaned
224
Developments in Surface Contamination and Cleaning
FIGURE 6.13 TOC vs. the initial contaminant level of the stainless steel plates [53]
with no residue measured in the TOC samples. The TOC samples from the other nine plates measured residue of 0.1 mg or less. The results indicate that the contaminants could be removed with greater than 96% efficiency with the SS-GLC system. Parametric studies have been conducted with a 1.25 cm long convergingediverging nozzle in the SS-GLC system [83]. The pressure upstream of the nozzle was either 2.2 MPa or 2.89 MPa with water flow rates of 0.052 liters/min and 0.056 liters/min, respectively. The mean velocity remains nearly
FIGURE 6.14 Linear correlation between TOC and NVR remaining after cleaning [53]
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
225
constant up to an axial distance of 50 mm. Increasing the air tank pressure increases the mean velocity of the flow at the exit of the nozzle from ~620 m/s to 630 m/s. The optimum tank pressure was around 2 MPa. The best working distance was found to be between 30 and 80 millimeters. The SS-GLC technology has been adapted for cleaning and verifying the cleanliness of the interior surfaces of hollow items, such as small compressed gas cylinders (K-bottles), tanks, and long pipes and tubes (over 15 cm long) [109]. For K-bottle cleaning, the system employs a rotating spray head with three nozzles with only a diverging section for supplying a gas-liquid cleaning mixture to the surface of the parts at supersonic velocity. The diverging nozzle is much smaller than a full convergingediverging nozzle and can fit through the narrow opening (~1.8 cm) of the bottle, although there is a 20% loss in fluid momentum compared with the convergingediverging nozzle. One nozzle is aimed straight down the bottle, while the other nozzles are oriented at 30 and 210 degrees from the bottle axis. The orientation of the three nozzles covers all surfaces of the bottle. The spray head is both rotatable and translatable along its longitudinal axis. No moving parts are exposed to the interior surfaces of the items to be cleaned, thereby reducing the risk of contamination. The system can also be employed for cleanliness verification by simply replacing the cleaning liquid with plain water, and collecting and analyzing the waste water after it has been sprayed onto the item. For large pipes the full convergingediverging nozzle can be employed, so cleaning is more efficient. In this case, a pipe crawler is used to transport the rotating nozzle head through the pipe while pulling a gas-liquid supply hose. Klausner et al. [86,87] investigated the performance of the SS-GLC impingement system developed at KSC as characterized by the rate of contaminant removal. Polished stainless steel substrates were contaminated with grease (8e11.5 mg/m2) and cleaned using both the conventional convergingediverging nozzle and the annular nozzle. The rate of contaminant removal is inversely proportional to the interfacial tension, which, in turn, is inversely proportional to temperature. Thus, the rate of contaminant removal by emulsification should be enhanced by increasing temperature, as was observed. A 15 jet approach angle gave the highest rate of residue removal, consistent with the theoretical consideration that the shearing force is most effective in breaking the cohesive bonds between adjacent contaminant molecules. The contaminant removal rate declined sharply with increasing distance from the cleaning surface due to reduction in droplet concentration and increased viscous drag. As expected, the cleaning performance of the annular nozzle was significantly better than that of the conventional convergingediverging nozzle under the same operating conditions. For the conventional convergingediverging nozzle, the highest removal rate was achieved at a distance of 5 cm from the surface, suggesting that the supersonic jet flow may converge at that distance. Based on these results, optimum performance of the system can be realized by using the annular nozzle with an
226
Developments in Surface Contamination and Cleaning
approach angle of 15 degrees to the cleaning surface at a distance of 2 cm from the surface with as high a mixture temperature as possible without evaporating the water. The cleaning effectiveness of the system could also be increased by the addition of detergents with the lowest dynamic surface tension to the water. Such detergents have been shown to enhance the effectiveness of spray cleaning for soil removal [110]. Kanno et al. [111] have developed a system for cleaning wafers using a novel two-fluid jet nozzle which is capable of removing sub-mm particles. The novel convergingediverging nozzle is capable of accelerating the liquid droplets to high velocity by a gas at relatively low supply pressure. Figure 6.15 shows the liquid droplets accelerated by the two-fluid nozzle reaches sonic velocity at a gas supply pressure of about 3 kgf/cm2 (line A), whereas a conventional low-pressure nozzle requires the gas supply pressure to be 10 kgf/cm2 or above to accelerate the liquid droplets to sonic velocity (line B). In semiconductor manufacturing, the maximum gas pressure used is about 7 kgf/cm2, which is sufficient to achieve supersonic velocity of the jet exiting the two-fluid nozzle. With a conventional nozzle, the jet reaches only subsonic exit velocity. The effectiveness of this new nozzle design in cleaning wafers, represented by the contaminant removal ratio, is compared with conventional low-pressure and high-pressure jetting nozzles for similar operating conditions (Fig. 6.16). The contaminant removal ratio is defined as the ratio of the number of particles removed to the number of particles remaining on the surface. Clearly, cleaning with the new jet nozzle can remove particles smaller than 0.1 mm (curve A) as compared with the other nozzles. Although the low-pressure
FIGURE 6.15 Graph showing the relation between the velocity of the liquid drops and the supply pressure of the gas for different jet nozzles. (a) Two-fluid nozzle. (b) Conventional lowpressure nozzle [111]
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
227
FIGURE 6.16 Plot shows the relation between particle size and contaminant removal ratio with different jet cleaning nozzles. (a) New two-fluid nozzle. (b) A conventional low-pressure nozzle. (c) A conventional high-pressure nozzle [111]
jet nozzle (curve B) is more effective than the high-pressure jet nozzle (curve C), it is unable to remove 0.1 mm particles. Hara [112] has developed a nozzle design that uses a liquid to prevent clogging of powder particles in a jet spray cleaning system for industrial applications. The powder is added to the gas-liquid mixture to enhance cleaning of automobiles, building surfaces, dishes, bottles, and utensils in food and beverage preparation, and other such applications. The powder tends to accumulate in passages of the nozzle where the rate of flow is low, thereby reducing the cleaning efficiency of the nozzle. Water-soluble powders also tend to absorb moisture and adhere to the walls and clog protrusions and stepped portions of the nozzle. The new design employs water as the clogging prevention liquid which is injected into a section of the pressurized gas flow between the powder injection section and the cleaning nozzle. The amount of clogging prevention liquid is smaller than the liquid supplied to the nozzle and is continued to be injected for a given duration after powder injection has stopped. An experiment was conducted to remove graffiti on a concrete wall by using sodium bicarbonate particles as a powder material in a high-velocity gas-liquid mixture (1000 liters/min of air at a pressure of 0.39 MPa, 10 liters/min of water at 13 MPa supplied to the cleaning nozzle). Water was used as the clogging prevention liquid. No accumulation or clogging of the powder was observed in the passages of the cleaning nozzle. The powder was discharged from the nozzle as solid particles and effectively cleaned the concrete surface. The SS-GLC technology is being applied to replace mechanical and chemical cleaning and de-scaling methods currently used by various industries [113,114]. Applied Cryogenic Solutions (ACS) has developed
228
Developments in Surface Contamination and Cleaning
FIGURE 6.17 parts [114]
A multiple nozzle SS-GLC system being used for cleaning heat exchanger
a cleaning system that consists of a spray head containing supersonic convergingediverging nozzles using a source of liquid gas, together with a novel, proprietary pumping system that permits pumping liquid nitrogen, liquid air, or supercritical carbon dioxide to pressures in the range of 138e417 MPa. The size and number of nozzles can be varied so the system can be built in configurations ranging from small hand-held spray heads to large multinozzle cleaners (Fig. 6.17). The system also can be used to verify if a part has been adequately cleaned. Pilot trials on heat exchanger tubing components have shown several benefits: l l
l
l
Superior cleaning in a much shorter period of time Lower energy and labor requirements for cleaning and de-scaling operations Significant reductions in waste volumes by not using water, acidic or basic solutions, organic solvents, or non-volatile solid abrasives as components in the cleaning process Improved energy efficiency in post-cleaning heat exchanger operations.
The ACS system has the flexibility and adaptability for use in existing plants using heat exchangers of various designs and operational configurations. In addition to heat exchanger applications, ACS is adapting the system for other
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
229
applications including: cleaning of coated or contaminated surfaces; air and sea infrastructure applications; mining, natural gas and oil exploration; and other potential uses. 3.5.1.1. Other Applications SS-GLC system can be applied for precision cleaning in the aerospace, automotive, circuit boards, electronics, machinery, metals, plastics, and optics industries. Additional applications include contamination removal in the nuclear, agriculture, food, pharmaceutical, and chemical industries. Commercial applications range from flux removal from printed circuit boards to scouring building exteriors; removal of algae and other organic matter from boat hulls; cleaning dead-ended components; mildew and stain removal; paint removal; removal of salt contamination and surface preparation prior to painting; removal of oil stains and grease spots; and spot cooling [115].
3.5.2. Medical Applications Golan and Hai [116] treated 50 adult volunteers suffering from sun-damaged skin, facial rhytids, skin pigmentation, and post-acne facial scarring with the JetPeelÔ system. The overall duration of the treatment ranged from 5 to 70 minutes. The results were judged to be aesthetically good to excellent by the patients and the medical staff, with a high degree of patient satisfaction. The technology was found to be particularly efficient in treating perioral regions due to its ability to achieve different depths of penetration of the skin. The application is more time-consuming than dermal abrasion or chemical peeling, but the post-treatment healing was found to be smother and quicker which could be attributed to tissue oxygenation with the JetPeelÔ system. One precaution recommended was to avoid the eyelids. In another study, Onesti et al. [117] treated a group of 54 patients with the JetPeelÔ method for skin rejuvenation from scar treatment (surgical, posttraumatic injury, acne), as well as damage from pigmentation and stretch marks. Six treatments of 5e15 minutes were performed. The depth of skin abrasion was a function of time of exposure and the distance of the nozzle from the surface. The results were very satisfactory and the rapid healing led to high patient acceptance of the procedure. Increased absorption and efficiency may be achieved by adding other treatment substances, such as vitamins and homeopathic drugs, to the saline jet stream. Ishikawa [118] employed the JetOx-NDÔ system to effectively treat wounds from infected abrasions, incised or crushed wounds before suturing, and wounds at graft donor sites. Cleaning and debridement was performed with a saline solution and oxygen. No local anesthesia was used. The small volume of cleaning solution (100 mL) required was sufficient for treating a wound 5 cm wide and 1 cm deep and the drainage and foreign matter could be collected
230
Developments in Surface Contamination and Cleaning
easily with gauze and swab. The shield prevented splashing of the cleaning solution. The author reported that the pain caused by the treatment was tolerable without local anesthesia. No cases of post-suturing infection were found. The system was considered to be effective in improving wound care in surgery and emergency situations for cleaning and debridement of wide and deep wounds. Olivares et al. [119] successfully used the JetOxÔ system for cleansing and debridement as part of negative-pressure wound therapy for adult patients over a period of 3 years with injuries of different etiology and evolution resulting in acute, chronic, and complex wounds. The deep chronic wounds represent a very high cost for health systems. Faster healing resulted in enhanced comfort and cost reduction for the patient during the healing process. Mechanical debridement of chronic wounds may be complicated and very painful despite the use of suitable analgesics. Mutumbo et al. [120] conducted a study to evaluate the efficacy and safety after a 1-year experience with the JetOx-HDCÔ system promoting removal of fibrin, without damaging granulation tissues. Ten patients with chronic leg ulcers were subjected to an average of three sessions with the JetOx-HDCÔ system followed by local application of an EMLA (Eutectic Mixture of Local Anesthetics). The results showed a remarkable improvement of wound status. The treatment was found to be less painful and the patients did not need local or general anesthetic. The JetOx-HDCÔ system is considered part of a global therapeutic strategy. As noted earlier, the JetOx-NDÔ system can be used with other non-saline solutions. Brizzio et al. [121] have employed Ringer’s lactate solution (a solution containing sodium chloride, potassium chloride, calcium chloride, and sodium lactate in distilled water [90]) to successfully clean and debride venous leg ulcers (3e50 cm2) in a group of 55 adult patients. Other solutions that have been developed and used successfully for cleansing and debridement include super-oxidized water solutions (oxidized water (99.98%) with reactive species of chlorine and oxygen) and an antimicrobial solution containing dichlorine monoxide (Cl2O) [99e103]. Uribe and Sanchez [101,102] conducted a pilot study to evaluate the epithelialization rates of chronic, uninfected diabetic foot ulcers in 40 adult patients treated topically with a neutral pH super-oxidized water solution. Treatment was conducted with 25 mL of the super-oxidized solution using the JetOxÔ system (O2 pressure 1e1.3 kPa at 1.5 mL/min). The results showed remarkable rates of epithelialization of 0.5e0.7 mm per day with total wound closure observed between 27 and 43 days. In a control group treated with saline solution, wound closure required between 45 and 75 days. Altamirano [122] treated 64 children with various partial- and full-thickness burn injuries. Debridement with the JetOxÔ system was conducted under
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
231
general anesthesia at entry followed by moistening of the wound with a superoxidized water solution. In general, the patients tolerated the daily cleansing and debridement without much pain. The length of hospital stay was shorter, resulting in significant cost savings.
4. SUMMARY Two alternate semi-aqueous cleaning techniques have been developed for various precision cleaning applications. Vapor steam cleaning employs dry superheated steam to remove surface contaminants. It is a low-cost, effective method for precision cleaning and for decontamination of microbially contaminated surfaces. Supersonic gas-liquid cleaning is based on accelerating the cleaning liquid, suspended as droplets in a gas stream, to supersonic velocities through a convergingediverging nozzle. The gas-liquid mixture has the kinetic energy to very effectively remove surface contaminants. This method can also be used for cleanliness verification. Both methods use very low volumes of aqueous liquids and are viable alternatives to solvent cleaning in many applications.
ACKNOWLEDGMENTS The author would like to thank Jim and Jeff Sloan for useful discussions. He is very grateful to Jody Mantell for help with locating obscure reference articles.
DISCLAIMER Mention of commercial products in this chapter is for information only and does not imply recommendation or endorsement by The Aerospace Corporation. All trademarks, service marks, and trade names are the property of their respective owners.
REFERENCES [1] The U.S. Solvent Cleaning Industry and the Transition to Non Ozone Depleting Substances, EPA Report, U.S. Environmental Protection Agency, Washington, DC, 2004. www.epa.gov/ ozone/snap/solvents/EPASolventMarketReport.pdf. [2] J.B. Durkee, Cleaning with Solvents, in: R. Kohli, K.L. Mittal (Eds.), Developments in Surface Contamination and Cleaning, William Andrew Publishing, Norwich, NY, 2008, pp. 759e871. Ch. 15. [3] R. Kohli, Adhesion of Small Particles and Innovative Methods for Their Removal, in: K.L. Mittal (Ed.), Particles on Surfaces 7: Detection, Adhesion and Removal, VSP, Utrecht, The Netherlands, 2002, pp. 113e149. [4] DoD Joint Service Pollution Prevention Opportunity Handbook, Section 8, Department of Defense, Naval Facilities Engineering Service Center (NFESC), Port Hueneme, CA, 2010. [5] R. Kohli, K.L. Mittal (Eds.), Developments in Surface Contamination and Cleaning, William Andrew Publishing, Norwich, NY, 2008.
232
Developments in Surface Contamination and Cleaning
[6] The Ultimate Technology for Clean Room Cleaning e Steam Vapor Cleaning, 2010. www.todayshealthyhome.com/steamvapor/cleaning_uses.html. [7] Steam Vapor the Clear Winner in Fighting Mold in Carpet, International Facility Management Association, Houston, TX, April 2007. www.ifma.org/daily_articles/2007/apr/ 04_09.cfm. [8] B.J. Tanner, Reduction in Infection Risk through Treatment of Microbially Contaminated Surfaces with a Novel, Portable, Saturated Steam Vapor Disinfection System, Am. J. Infection Control 37 (2009) 20. [9] J.C. May, Use ‘Dry’ Steam Vapor Treatment for Healthier Carpet, Upholstery. www.house keepingchannel.com/a 565 (December 2009). [10] Steam Bullet: The Ultimate Cleaning Machine, 2010. www.asseenontv.com/prod-pages/ steam-bullet.html. [11] M. Dietrich, Don’t Run Out of Steam. Parts 3 and 4, Sanitation Canada (Septembere October 2009) 45e47. [12] M.J. Weitz, Method and Apparatus for Steam Cleaning Anilox Inking Rolls, U.S. Patent Application 2003/0167948 (2003). [13] R. Hoverson, Steaming Clean, American School & University Magazine, October 2006. http://asumag.com/Maintenance/university_steaming_clean/. [14] W.J. Bauer, Systems and Methods for Disinfecting and Sterilizing by Applying Steam Vapor Containing Low Zeta Potential Mineral Crystals, U.S. Patent 7,547,413 (2009). [15] H. Kunze-Concewitz, Method of Cleaning Surfaces with Water and Steam, U.S. Patent 5,964,952 (1999). [16] W. Koehler, H. Padilla, Steam Cleaning as a Solvent Alternative, DoD Joint Service Pollution Prevention Opportunity Handbook, Section 8-1-6, Naval Facilities Engineering Service Center (NFESC), Port Hueneme, CA, 2000. [17] R. Vozilla, C. Tittle, Portable Steam Cleaning System (Mini-Max), DoD Joint Service Pollution Prevention Opportunity Handbook, Section 8-1-11, Naval Facilities Engineering Service Center (NFESC), Port Hueneme, CA, 2003. [18] Synopsis of Steam Cleaners, Report Project 06e02, US Air Force Dental Evaluation and Consultation Service, Great Lakes Naval Training Center, Great Lakes, IL 2006. http:// airforcemedicine.afms.mil/idc/groups/public/documents/afms/ctb_109150.pdf. [19] The Micro Precision Steam Cleaner Solves Tough Cleaning Challenges, VaTran Systems, Chula Vista, CA, 2010. www.vatran.com. [20] High Performance Steam Jet Cleaning, Elma GmbH & Co KG, Singen, Germany, 2010. www.elma-ultrasonic.com. [21] Dental and Jewelry Steam Cleaners, Reliable Corporation, Toronto, Ontario, Canada, 2010. www.reliablecorporation.com. [22] Steamshine Steam Cleaning Systems, C.D. Nelson Consulting, Lakemoor, IL, 2010. www.steamshine.com. [23] The Mini-Max Cleaner, PDQ Precision Inc, San Diego, CA, 2010. www.pdqprecision.com. [24] Steam Cleaning, 2010. www.imexserve.it/guide-steam-cleaning.html. [25] W. Swichtenberg, Vapor Steam Cleaning, 2010. http://articles.directorym.com/Vapor_ Steam_Cleaning-a940806.html. [26] E.W. Lemmon, M.O. McLinden, D.G. Friend, Thermophysical Properties of Fluid Systems, in: P.J. Linstrom, W.G. Mallard (Eds.), NIST Chemistry WebBook, NIST Standard Reference Database Number 69, National Institute of Standards and Technology, Gaithersburg, MD, 2010. http://webbook.nist.gov. [27] J.E. Sloan, E.L. Bellegarde, Steam Cleaning System, U.S. Patent 6,299,076 (2001).
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
233
[28] J.G. Leidenfrost, De Aquae Communis Nonnullis Qualitatibus Tractatus, University of Duisburg, Germany (1756). English Translation of Chapter 15: On the Fixation of Water in Diverse Fire, Intl. J. Heat Mass Transfer 9 (1966) 1153. [29] C. Gianino, Leidenfrost Point and Estimate of the Vapour Layer Thickness, Phys. Educ. 43 (2008) 627. [30] A.K. Mozumder, M.R. Ullah, A. Hossain, M.A. Islam, Sessile Drop Evaporation and Leidenfrost Phenomenon, Am. J. Appl. Sci. 7 (2010) 846. [31] M. Friedheim, Steam Jet Cleaning and Sterilizing System, U.S. Patent 4,414,037 (1983). [32] C.D. Nelson, Method and Apparatus for Generating Pressurized Fluid, U.S. Patent 4,878,458 (1989). [33] M. Friedheim, Superheated Vapor Generator and Control System and Method, U.S. Patent 5,471,556 (1995). [34] M. Friedheim, Superheated Vapor Generator System, U.S. Patent 6,006,009 (1999). [35] M. Friedheim, Superheated Vapor Generator and Method of Fabrication Thereof, U.S. Patent 6,181,873 (2001). [36] Elmasteam Steam Cleaning for Medical Engineering, Elma GmbH & Co KG, Singen, Germany, 2010. www.elma-ultrasonic.com. [37] Cleaning Surgical Instruments, SPS Medical, Rush, NY, 2001. www.spsmedical.com/ education/educationarticles.html. [38] J. Weber, B. Beck, Luer-Lock Connector for Medical Devices, U.S. Patent 7,472,932 (2009). [39] G. Guala, Male Luer Lock Connector for Medical Fluid Lines, U.S. Patent 6,893,056 (2005). [40] J. Ham, Steam Exposure Temperature Measurement, GTI Industries, San Diego, CA, September 2002, Test Report 02-339-A, www.pdqprecision.com. [41] Personal communication with Jeff Sloan, VaTran Systems (2009). [42] R.M. Kwan, A.J. Yost, J.V. Santiago, A.C. Herring, M.M. Conlin, Steam Vapor Cleaning Ejection Seat Frames and Components Technical Evaluation, Technical Report NAWCADPAX/TR-2002/244, Naval Air Warfare Center Aircraft Division, Patuxent River, MD, 2003. [43] M. Panzner, G. Weidemann, M. Meier, W. Conrad, A. Kempe, T. Hutsch, Laser Cleaning of Gildings, in: J. Nimmrichter, W. Kautek, M. Schreiner (Eds.), Lasers in the Conservation of Artworks LACONA VI Proceedings, Springer Verlag, Berlin, Germany, 2007, pp. 21e28. [44] Microbiological Tests with Capitani Steam Cleaning Systems, Gialloblu Studio e Controllo Prodotti e Produzioni, Milan, Italy, 2006. www.capitani.it/test_eng.shtml. [45] Anti Bacterial Effect Case Study, Adriatic Union Laboratories, Vicenza, Italy, 2010. www.imexserve.it/guide-case-study.html. [46] S.E. Mackler, Case by Case: Upgrading the Precision Cleaning Process, Process Cleaning Magazine, September/October 2009. [47] J.W. Duffy, D. Hashimoto, J.E. Sloan, E.L. Bellegarde, Development of an Automated Cleaning System for Multi-Ferrule Fiber-Optic Connectors, in: Optical Fiber Communication Conference and Exposition and The National Fiber Optic Engineers Conference, OSA Technical Digest (CD), Paper JThA109, Optical Society of America, 2008. www.opticsinfobase.org/abstract.cfm?URI¼OFC-2008-JThA109. [48] M. Friedheim, Super-Heated, High-Pressure Steam Vapor Cleaning, in: B. Kanegsberg, E. Kanegsberg (Eds.), Handbook for Critical Cleaning, CRC Press, Boca Raton, FL, 2001, pp. 343e348. [49] R.E.B. Caimi, M.D. Littlefield, G.S. Melton, E.A. Thaxton, Cleaning Verification by Air/ Water Impingement, in: Proc. Precision Cleaning ’94 Conference. Witter Publishing, Flemington, NJ, 1994, pp. 31e41.
234
Developments in Surface Contamination and Cleaning
[50] L.L. Jones, M.D. Littlefield, G.S. Melton, R.E.B. Caimi, E.A. Thaxton, Cleaning Verification by Air/Water Impingement, NASA Technical Memorandum NASA-TM-111898, NASA Kennedy Space Center, FL, 1994. [51] Contamination Control Requirements Manual, NASA Document JPR 5322.1, NASA Johnson Space Center, Houston, TX, 2009. [52] W.L. Dearing, L.D. Bales, C.W. Basset, R.E.B. Caimi, G.M. Lafferty, G.S. Melton, et al., Methods for Using Water Impingement in Lieu of Chlorofluorocarbon 113 for Determining the Non-Volatile Residue Level on Precision Cleaned Hardware, in: C.J. Bryan, K. GebertThompson (Eds.), Alternatives to Chlorofluorocarbon Fluids in the Cleaning of Oxygen and Aerospace Systems and Components, ASTM STP-1181. ASTM International, West Conshohocken, PA, 1993, pp. 66e77. [53] R.E.B. Caimi, E.A. Thaxton, Supersonic Gas-Liquid Cleaning System, in: Proc 4th National Technol Transfer Conf, NASA Conference Publication CP-3249 Volume 1 (1993) 232e240. [54] R.E.B. Caimi, F.N. Lin, E.A. Thaxton, Gas-Liquid Supersonic Cleaning and Cleaning Verification Spray System, U.S. Patent 5,730,806 (1998). [55] Super Clean, Super Safe, NASA Spinoff, p. 98 (2002). [56] M.S. Granick, R.L. Gamelli (Eds.), Surgical Wound Healing and Management, Informa Healthcare, New York, NY, 2007. [57] H. Schlichting, K. Gersten, Boundary Layer Theory, eighth ed., Springer Verlag, Heidelberg and Berlin, Germany, 2000. [58] D.R. Spotz, Water Jet Cleaning Appliance, U.S. Patent 3,982,965 (1976). [59] R.A. Hudson, Pulsating Spray Nozzle, U.S. Patent 4,350,158 (1982). [60] D.H. Klosterman, S.M. Laskowski, S.V. Knee, S.-K. Shi, Low Flow Rate-Low Pressure Atomizer Device, U.S. Patent 4,787,404 (1988). [61] C.R. Sperry, A.M. Raff, Method for the Cleansing of Wounds Using an Aerosol Container Having Liquid Wound Cleansing Solution, U.S. Patent 5,059,187 (1991). [62] V.E. Johnson Jr., Enhancing Liquid Jet Erosion, U.S. Patent 4,681,264 (1987). [63] R.C. Lewis Jr., Ultrasonic Wound Cleaning Method and Apparatus, U.S. Patent 4,982,730 (1991). [64] L. Molinari, Adjustable Apparatus for Removing Surface Portions of Human Tissue, U.S. Patent 5,037,432 (1991). [65] D.H. Grulke, D.L. Tyler Sr., W.M. Booth III, Compact Pulsing Pump for Irrigation Handpiece, U.S. Patent 5,046,486 (1991). [66] D.C. Bailey, Method and Apparatus for Cleaning with High Pressure Liquid at Low Flow Rates, U.S. Patent 5,551,909 (1996). [67] S. Palffy, Method and Device for Producing Periodical Impulse Changes in a Fluid Flow, U.S. Patent 5,819,801 (1998). [68] S. Gold, J.V. Mizzi, Liquid Spray Dispenser and Method, U.S. Patent 6,125,843 (2000). [69] P.H. Rose, P. Sferlazzo, R.G. van der Heide, Aerosol Surface Processing, U.S. Patent 6,203,406 (2001). [70] M.E. Labib, C.-Y. Lai, P.A. Materna, G.L. Mahon, Method of Cleaning Passageways Using a Mixed Phase Flow of Gas and a Liquid, U.S. Patent 6,454,871 (2002). [71] D.E. Holt, C.E. Lundberg, D. Austin, M. Foster, Fluid and Air Nozzle and Method for Cleaning Vehicle Lenses, U.S. Patent 6,554,210 (2003). [72] C. Litherland, K. Behzadian, Devices and Methods for Nebulizing Fluids Using Flow Directors, U.S. Patent 6,550,472 (2003). [73] P.J. Weber, L.B. da Silva, A.M. Rubenchik, Tissue Resurfacing Using Biocompatible Materials, U.S. Patent 6,726,693 (2004).
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
235
[74] G. Andersson, Surface Treatment Nozzle, U.S. Patent 6,824,453 (2004). [75] M.A. Hashish, S.J. Craigen, F.M. Sciulli, Y. Baba, Apparatus for Fluid Jet Formation, U.S. Patent 6,945,859 (2005). [76] D.P. Jackson, Carbon Dioxide Snow Apparatus, U.S. Patent 7,293,570 (2007). [77] Y. Tabani, M.E. Labib, Method for Cleaning Hollow Tubing and Fibers, U.S. Patent 7367346 (2008). [78] W.T. McDermott, J.W. Butterbaugh, Cleaning using Argon/Nitrogen Cryogenic Aerosols, in: R. Kohli, K.L. Mittal (Eds.), Developments in Surface Contamination and Cleaning, William Andrew Publishing, Norwich, NY, 2008, pp. 951e986. Ch. 19. [79] R. Sherman, Carbon Dioxide Snow Cleaning, in: R. Kohli, K.L. Mittal (Eds.), Developments in Surface Contamination and Cleaning, William Andrew Publishing, Norwich, NY, 2008, pp. 987e1012. Ch. 20. [80] M. Tavger, Apparatus and Method for Tissue Cleansing, U.S. Patent 6,283,936 (2001). [81] M. Tavger, Dermal Abrasion, U.S. Patent 6,673,081 (2004). [82] M. Tavger, A High Velocity Liquid-Gas Mist Tissue Abrasion Device, Int. Patent WO/2005/ 065032 (2005). [83] F. Kinney, Supersonic Gas-Liquid Cleaning System, NASA Contractor Report CR-97206196 (1996). [84] S.A. Sherif, W.E. Lear, N.S. Winowich, Effect of Slip Velocity and Heat Transfer on the Condensed Phase Momentum Flux of Supersonic Nozzle Flows, Trans. ASME J. Fluids Eng. 122 (2000) 14. [85] W.E. Lear, S.A. Sherif, Aerothermal Considerations for the Design of Two-Phase High Speed Impact Cleansers, Trans. ASME J. Fluids Eng. 122 (2000) 20. [86] J.F. Klausner, R. Mei, S. Near, R. Stith, Performance Enhancement of a High Speed Jet Impingement System for Nonvolatile Residue Removal, NASA Contractor Report CR-97205941, 1997. [87] J.F. Klausner, R. Mei, S. Near, R. Stith, Two-Phase Jet Impingement for NonVolatile Residue Removal, Proc. Inst. Mech. Engrs. Part E: J. Process Mech. Eng. 212 (1998) 271. [88] Flow-Concentrating Supersonic Gas/Liquid Nozzles, Technical Support Package KSC11883, NASA Tech Briefs, 2001. [89] Personal communication with Jim Sloan, VaTran Systems, Chula Vista, CA, 2010. [90] J.E. Sloan, Low Mass Flow, Momentum Cleaning Methods, in: Proc. Precision Cleaning ’97 Conference, pp. 182e185, Witter Publishing, Flemington, NJ, 1997. [91] American Heritage Medical Dictionary, Houghton Mifflin Company, New York, NY, 2007. [92] JetoxÔ - ND and JetoxÔ - HDC Wound Cleansing and Debridement Systems, TavTech Ltd., Yehud, Israel, 2010. www.tav-tech.com. [93] JetoxÔ - ND Jet Lavage Wound Cleansing and Debridement System, DeRoyal, Powell, TN, 2009. www.deroyal.com. [94] B. Noe¨l, Prise en Charge de L’Ulce`re de Jambe D’Origine Veineuse, Revue Me´dicale Suisse, Review No. 16 (April 2005). [95] S. Meaume, L. Te´ot, Step by Step Wound Healing, Jaypee Brothers Medical Publishers, New Delhi, India, 2005. Ch. 2. [96] C. Sussman, B. Bates-Jensen, Wound Care: A Collaborative Practice Manual for Health Professionals, third ed., Lippincott Williams & Wilkins, Philadelphia, PA, 2006. ¨ konomie der Wundheilung, in: T. Wild, J. Aubo¨ck (Eds.), Manual der Wund[97] H. Jenzer, O heilung. Chirurgisch-dermatologischer Leitfaden der modernen Wundbehandlung, Springer Verlag, Vienna, Austria, 2007, pp. 297e305.
236
Developments in Surface Contamination and Cleaning
[98] L. Te´ot, Surgical Debridement, in: M.S. Granick, R.L. Gamelli (Eds.), Surgical Wound Healing and Management, Informa Healthcare, New York, NY, 2007, pp. 91e101. Ch. 10. [99] H. Alimi, A. Guiterrez, Method of Treating Skin Ulcers Using Oxidative Reductive Potential Water Solution, U.S. Patent Application 2006/0235350 (2006). [100] T.A. Wolvos, Advanced Wound Care with Stable, Super-Oxidized Water, Wounds (January 2006 Supplement), pp. 11e13. [101] F.F. Uribe, F.S. Sanchez, Comparative Study of the Use of Jetox and Chloride of Sodium 0.9% versus Jetox and Solutions Superoxidants Electrolyzed, in: Proc. 19th Symp. Advanced Wound Care and Wound Healing Society, San Antonio, Texas, 2006. www. oculusis.com/us/technology/published.php. [102] F.F. Uribe, Effect of a Neutral pH Super-Oxidized Solution in the Healing of Diabetic Foot Ulcers, Poster PW 102 at 2008 WUWHS Congress, World Union of Wound Healing Societies, 2008. www.wuwhs.com/congress2008. [103] R. Northey, Antimicrobial Solutions Containing Dichlorine Monoxide and Methods of Making and Using the Same, U.S. Patent Application 2010/0112092 (2008). [104] JetPeel 3 Multifunction Skin Rejuvenation System, TavTech Ltd, Yehud, Israel, 2010. www.tav-tech.com. [105] R. Kronmeyer, JetPeel Vies to Replace Microdermabrasion, Aesthetic Buyers Guide (September/October 2006) 218e219. [106] E.S. Lindenbaum, M. Tavger, JetPeel e New Aspect for Skin Rejuvenation, unpublished report, 2005. www.tav-tech.com. [107] R.G. Barile, C. Fogarty, C. Cantreil, G.S. Melton, Precision Cleaning Verification of Fluid Components by Air/Water Impingement and Total Carbon Analysis, NASA Technical Memorandum NASA-TM-II742, NASA Kennedy Space Center, FL, 1994. [108] G.S. Melton, R.E.B. Caimi, E.A. Thaxton, Determination of Non-Volatile Residue on Precision Cleaned Oxygen and Aerospace Systems and Components by Means of Water and Total Organic Analysis, in: Proc. 1993 Intl. CFC and Halon Alternatives Conf., 1993, pp. 642e650. [109] R.E.B. Caimi, E.A. Thaxton, Balanced Rotating Spray Tank and Pipe Cleaning and Cleanliness Verification System, U.S. Patent 5.706,842 (1998). [110] N.E. Prieto, W. Lilienthal, P.L. Tortorici, Correlation Between Spray Cleaning Detergency and Dynamic Surface Tension of Nonionic Surfactants, J. Am. Oil Chemists Soc. 73 (1996) 9. [111] I. Kanno, M. Tada, M. Ogawa, Two-Fluid Cleaning Jet Nozzle and Cleaning Apparatus, and Method Utilizing the Same, U.S. Patent 5,918,817 (1999). [112] S. Hara, Cleaning Nozzle and Cleaning Apparatus, U.S. Patent 6,935,576 (2005). [113] Gas-Liquid Supersonic Cleaning Spray System Technology, NASA Innovative Partnership Program, Kennedy Space Center, FL, 2010. [114] Personal Communication with Ken Smith, Applied Cryogenic Solutions, Galveston, TX, 2010. [115] J.B. Gayle, T.S. Jerowski, Developing NASA’s Supersonic Gas-Liquid Cleaning System. Boon to Critical Cleaning, Precision Cleaning Magazine (May 1996) pp. 52e55. [116] J. Golan, N. Hai, JetPeel: A New Technology for Facial Rejuvenation, Annals of Plastic Surgery 54 (2005) 369. [117] M.G. Onesti, G. Curinga, M. Toscani, N. Scuderi, Jet-Peel: New Technique for the Treatment of Skin Imperfections, Dermatologia Clinica 26 (2006) 19. [118] H. Ishikawa, Clinical Use of a New Wound Cleansing and Debridement System, in: Proc. 67th Ann. Mtg. Japan Surgical Association, Tokyo, Japan, 2005, pp. 315-317.
Chapter | 6
Alternate Semi-Aqueous Precision Cleaning Techniques
237
[119] E.R. Olivares, A.V. Martı´nez, C.C. Smith, M.A.Z. Aguirre, Terapia de Presio´n Negativa en el Manejo de Heridas, Cirugia Plastica 18 (2008) 56. [120] M. Mutombo, M. Poiteau, T. Morel, K. Benabdallah, P. Guillain, JETOXÒ System in Management of Leg Ulcer: A Review of One Year’s Experience, Journal des Plaies et Cicatrisations 63 (2008) 25. [121] E. Brizzio, F. Amsler, B. Lun, W. Bla¨ttler, Comparison of Low-Strength Compression Stockings with Bandages for the Treatment of Recalcitrant Venous Ulcers, J. Vascular Surgery 51 (2010) 410. [122] A.M. Altamirano, Reducing Bacterial Infectious Complications from Burn Wounds, Wounds (January 2006 Supplement) 17e19.
This page intentionally left blank
Index
Acoustic streaming cleaning, 35e6, 42e5 particle removal benefits, 43e5 Adhesion forces, general grouping, 67 Aerosol cleaning, and particulate contamination, 2 Argon bullets, for 20 nm particles, 16e23 from AR/He mixture, 18e23 Laval nozzle, 19 long and short nozzles, 18e19, 23 SEM images, 19e22 from pure argon, 17e18 Biological fouling, 124 Bjerknes force, 42 Blast wave theory (BWT) for LIP, 73e6 Boundary layers, with sonic cleaners, 37e9 Bullet particles: fragmentation, 8e10 generation, 15e16 size estimation, 16 Cahn microbalance, 47 Cavitation with megasonic cleaning, 31e2 cavitation meters, 47e8 cleaning process, 34e5 Chemical reaction fouling, 125 Cleaning techniques, advantages/ disadvantages: laser cleaning, 65e6 mechanical cleaning, 65e6 vapour-phase cleaning, 65e6 wet-chemical etching, 65e6 Cleanliness measurement see Surface cleanliness measurement CO2 snow cleaning, 23e4 damage issues, 24 efficiency, 24 and particulate contamination, 2 Contamination see Fluid cleanliness levels; Interior surfaces, contamination; Particulate contamination Corrosion fouling, 124 Cryogenic aerosol cleaning technique, 2 Crystallization/scaling fouling, 124
Debye interaction, 4 Deposition/particulate fouling, 124 Detachment force of adhesion, JKR model, 76e8 Deymier et al characterization of adhesion/ removal forces, 41e2 Dielectrophoretic force, 175e7 Direct Simulation Monte Carlo (DSMC) method, 79 DRAM/flash memory devices, and particulate contamination, 1 Drinking water hose cleaning see Projectiles, cleaning with Dry laser cleaning (DLC), 68e9 damage issues, 68e9 main principle, 68 Dry particle removal, 65e7 Dust removal from solar panels see Electrodynamic screens (EDS); Solar panels, dust removal from Electrodynamic screens (EDS), 165e9 basic principles, 165 dielectrophoretic force, 175e7 high-voltage linear sinusoidal power supply, 184, 185, 186 high-voltage three-phase power supply, 179e84 design requirements, 179e81 MOSFET-based power supply, 181e3 performance modeling, 166e7 power requirements, 169 single-phase EDS arrangement, 167e8 three-phase EDS arrangement, 168 trajectories of charged particles, 170e5 electric field, 170e1 removal of charged particles, 172e6 removal of uncharged particles, 174e6 trajectories, 170e1 tribocharging of particles, 177e8 uncharged conducting particles removal, 178e9 see also Solar panels, dust removal from
239
240 Electrodynamic screens (EDS), dust removal efficiency (DRE) testing, 184e90 charge conditioning of dust cloud, 187 dust removal efficiency, 184e9 experimental setup, 184, 187e8 low pressure testing, 189e90 maximum power point (MPP) for panels, 190e3 adjusting voltage output, 191e2 cleaning efficiency measurement, 193 maximum power point tracking (MMPT), 192e3 test chamber, 189, 191 size classification, 187 test procedure, 187e90 test results, 193e6 net charge-to-mass ratio (Q/M), 194 Electrospray technique, 24e7 about electrospray, 24e5 advantages, 27 basis of particle removal, 25e6 performance, 27 removal force formula, 25e6 typical system, 26e7 Electrostatic charging of particles, 157e8, 159 Fluid cleanliness levels, 126e32 about particulate contamination, 126e7 hydraulic fluids, 127e30 ISO standard, 127e8, 130, 131, 132 military standard, 129e30, 132 SAE standard, 128e9, 130 typical internal clearances, 130 non-hydraulic fluids, 130e2 see also Tube cleaning methods Fouling see Interior surfaces, contamination Gas molecule-nanoparticle interactions with LIP, 79e81 critical rolling moment, 81 Direct Simulation Monte Carlo (DSMC) method, 79 Knudsen number, 79 rocking resonance frequencies, 81 Hamaker constant, 4 Haze mapping, 48e9 High-pressure jet cleaning, 133 Homogeneous nucleation and growth, 12e15 about the nucleation and growth, 12 critical super-saturation ratio, 13 general features, 12e15
Index generation by pure nitrogen expansion example, 14e15 nucleation onset point, 13, 14 Volmer and Weber model, 13e15 Hose cleaning see Projectiles, cleaning with Hydraulic fluids see Fluid cleanliness levels Interior surfaces, contamination: contamination types: biological fouling, 124 chemical reaction fouling, 125 corrosion fouling, 124 deposition or particulate fouling, 124 scaling or crystallization fouling, 124 effects, 125e6 hydraulic/pneumatic systems failure, 125e6 oil degradation with water/gases, 126 in space missions, 126 see also Fluid cleanliness levels; Projectiles, cleaning with; Tube cleaning methods Keesom interaction, 4 Kennedy Space Centre (KSC), and SS-GLC cleaning, 212e13 Knudsen number, 79 Laser cleaning, 63e117 about laser cleaning, 63e4, 67e8, 114e17 advantages/disadvantages summary, 66 dry laser cleaning (DLC), 68e9 future research direction, 113e14 liquid-based laser cleaning, 69e70 see also Laser-induced plasma (LIP) cleaning Laser-induced plasma (LIP) cleaning, 70e103 basic principles, 70e2 blast wave propagation, 70, 73 blast wave theory (BWT), 73e6 shockwave propagation, 73e4 temperature behind shockwave, 76 lithography mask damage and/or alterations issues, 102e6 nanoparticle displacement, 76e9 detachment force of adhesion, JKR model, 76e9 gas molecule-nanoparticle interactions, 79e81 moment balance criterion, 76e8 rocking motion from transient force field, 78e9
241
Index nanoparticle removal experiments, 81e94 results summary, 94 scanning electron microscopy (SEM) images, 83e6, 87e93 surface analysis system (SAS) images, 81e2 process parameters optimization, 94e8 about optimization, 94 attenuator use and setting, 95e8 pulse energy control, 95e6 transient pressure measurement, 94e8 thermomechanical analysis, 99e102 damage issues, 99 radiation intensity loading of masks, 101e2 shockwave loading of masks, 100e1 small particles problems, 99 Laser-induced plasma (LIP) cleaning, advanced techniques, 103e13 about removing smaller particles, 103e5 damage problems with normal techniques, 105 LIP with in-air shock tubes, 107e10 basic principles, 107e8 instrumentation diagram, 111e12 LIP in pressurized chambers, 106e7 experimental set-up, 108 LIP with submerged shock tubes, 112e13 wet-LIP cleaning, 110e11 Laval nozzle, 19 Liquid particle counters (LPCs), 45e6 near angle light scatter, 45e6 Liquid-based laser cleaning, 69e70 steam laser cleaning (SLC), 69e70 London interaction, 4 Mechanical cleaning: advantages/disadvantages, 65e6 tubes, 133 Megasonic cleaning, 31e59 about ultrasonic and megasonic cleaning, 31e4, 59 applications, 32 case studies: Cu-low-k dual damascene process, 58 irradiating wafers in electronic solutions, 58 L-type waveguides, 58 removal of post-CMP residues, 59 cavitation control benefit, 31e2 contribution of acoustic streaming, 42e5 microbalances, 47
reciprocity principle, 33e4 theory, 41e5 Bjerknes force, 42 Deymier et al characterization of adhesion/removal forces, 41e2 Kim et al. on detachment of bubbles, 42 Wu et al. single-wafer cleaner, 32e3 see also Sonic cleaners; Surface cleanliness measurement Megasonic system evaluation, 48e57 damage issues, 52 electrical power value issues, 51e3 examination of sonoluminescence flux, 52 haze mapping, 48e9 mean cavitation intensity with frequency, 48e9 particle removal efficiency (PRE), 48e51 sweep, usage of, 55e7 Microbalances, 47 Mie scattering theory, 160 Molecular dynamics (MD) simulation, 2e3, 5e8 Moment ratio (MR) models, sonic cleaners, 39e40 Nano-particle removal: about the removal, 3 adhesion force, 3e4 bullet particle fragmentation, 8e10 bullet particle size effects, 11 deformation/adhesion issues, 10 force for removal, 4e5 impact force of collision, 5 Hamaker constant, 4 kinetic energy, momentum or force?, 6e8 kinetic energy vs. binding energy, 8e10 Molecular dynamics (MD) simulation, 5e8 particle density/hardness issues, 10e11 particle/substrate interactions: Debye interaction, 4 Keesom interaction, 4 London interaction, 4 van der Waals (VdW) force, 4 Near angle light scatter, 45e6 Nephelometry turbidity units (NTUs), 46e7 Nozzle beam technique see Supersonic nozzle beam technique Particle removal efficiency (PRE), 48e51 Particulate contamination: and aerosol cleaning, 2 and the cryogenic aerosol process, 2
242 Particulate contamination (Continued ) and DRAM/flash memory devices, 1 and the fluid-dynamic drag force, 1 and wet cleaning processes, 1e2 see also Interior surfaces, contamination Photovoltaic (PV) and photothermal (PT) installations, 149e97. see also Electrodynamic screens (EDS); Solar panels, dust removal from; Solar power potential Pipe cleaning see Projectiles, cleaning with Precision steam cleaning, 202e12 about steam cleaning, 202e3 basic principles, 203 commercial examples, 206 cost benefits, 212 delivery systems, 205 equipment, 203e7 medical applications, 206e7, 208 moisture content, 204 spray capabilities, 205 steam sources: continuous fill systems, 204 Leidenfrost effect, 204 single tank units, 204 VaTran Systems, 204 surface drying, 204 typical specification, 205 waste management, 205 Precision steam cleaning, applications, 209e11 aircraft ejection seats, 210 fibre optic endfaces, 211 gilded objects, 210 microbially contaminated surfaces, 210e11 stainless steel substrates, 210 typical examples, 209e10 US Department of Defense (DoD), 211 Precision steam cleaning, operation issues, 207e9 additive use, 208 advantages, 209 disadvantages, 209 safety of operators, 208e9 suitability of objects, 207e8 water quality, 207 Precision turbidity meter, 46e7 Projectiles, cleaning with, 134e43 about cleaning with projectiles, 123e4, 143 advantages/disadvantages, 140e1 applications, 141e3 automated systems, 138e40
Index basic principle, 134 operating issues, 138e40 operating pressures, 139 path constraints, 138e9 for silt, soil, slurry and moisture, 139e40 size considerations, 140 pneumatic launchers, 135 projectiles, 135e7 abrasive projectiles, 136e7 coupling projectiles, 136e7 grinding projectiles, 136e7 product recovery projectiles, 136e7 standard projectiles, 136e7 ultraclean nozzles, 137e8 British Standard Pipe (BSP) nozzles, 138 hose nozzles, 137 Joint Industry Conference (JIC) nozzles, 138 tube nozzles, 138 see also Fluid cleanliness levels; Interior surfaces, contamination; Tube cleaning methods Reciprocity principle, with megasonic cleaning, 33e4 Scaling or crystallization fouling, 124 Schlichting streaming, 41 Solar panels, dust removal from: about the installations, 150e1, 196e7 charge distribution effects, 163e4 deposition rates, 155 power loss, 156e7 problem summary, 151 electrostatic charging of particles, 157e8 charge distributions, size/effect, 158e60 desert areas, 161 optical depth, 161 optical path, 161 microstructural deposition effects, 163e4 size/shape effects, 163e4 suspended dust, 151 test chamber studies, 162e3 transmission loss, 160e2 extinction efficiency, 160e2 Mie scattering theory, 160 see also Electrodynamic screens (EDS); Solar power potential
243
Index Solar power potential, 151e4 available solar power, 151e2 cost/economic issues, 154 dust, adverse effects, 154 global power need, 152e3 PV system benefits, 153 US PV usage and potential, 153e4 see also Photovoltaic (PV) and photothermal (PT) installations Sonic cleaners, 34e41 acoustic streaming process, 35e6 boundary layer effects, 37e9 cavitation with acoustic streaming, 36, 40e1 cavitation process, 34e5 frequency effects, 36e8 moment ratio (MR) models, 39e40 Schlichting streaming, 41 see also Megasonic cleaning Sonoluminescence flux examination, 52 Steam cleaning, pipes, 133. see also Precision steam cleaning Steam laser cleaning (SLC), 69e70 Steam pipe cleaning see Projectiles, cleaning with Supersonic gas-liquid cleaning (SS-GLC), 212e31 about SS-GLC cleaning, 212e14 Kennedy Space Centre (KSC) special needs, 212e13 advantages, 222 basic principles, 214 disadvantages, 222 parameters for different techniques, 219 precision cleaning, 215e18, 223e9 Applied Cryonic Solutions (ACS) systems, 227e8 commercial system example, 216e18 de-scaling, 227e8 gas/water velocity, 216 heat exchanger tubing, 228 hollow items, 225 KSC validation testing, 223 nozzle design, 215e16 parametric studies, 224e5 rate of contamination removal tests, 225e6 semiconductor manufacture, 226e7 wafer cleaning, 226 water-soluble powders, 227
Supersonic gas-liquid cleaning (SS-GLC), medical applications, 218e22, 229e31 advantages, 222e3 burn injuries, 230e1 disadvantages, 223 foot ulcers, 230 gas nozzle configurations, 218e20 infected wounds, 229e30 JetOxTM system, 219e20, 229e30 JetPeelTM system, 219e21, 229 skin damage, 229 Supersonic nano-particle beam technique, 1e27 about the technique, 1e3, 27 advantages, 3 disadvantages, 3 electrospray technique, 24e7 see also Particulate contamination Supersonic nozzle beam technique, 11e24 argon bullets, for cleaning of 20 nm particles, 16e23 bullet particle generation, 15e16 CO2 snow cleaning, 23e4 homogeneous nucleation and growth, 12e15 Surface cleanliness measurement, 45e8 liquid particle counters (LPCs), 45e6 microbalances, 47 nephelometric turbidity units (NTUs), 47 precision turbidity meter, 46e7 ultrasonic cavitation meters, 47e8 Tribocharging of particles, 177e8 Tube cleaning methods, 132e4 chemical cleaning, 134 high-pressure jet cleaning, 133 mechanical cleaning, 133 steam cleaning, 133 see also Fluid cleanliness levels; Interior surfaces, contamination; Projectiles, cleaning with Ultrasonic cavitation meters, 47e8 van der Waals (VdW) force, 4 Vapour-phase cleaning, advantages/ disadvantages summary, 66 Wet cleaning processes: advantages disadvantages summary, 65e6 and particulate contamination, 1e2 wet/steam laser cleaning (WLC), 67
This page intentionally left blank
