Environmental Issues and Waste Management Technologies in the Ceramic and Nuclear Industries VIII
Related titles published by The American Ceramic Society: Environmental Issues and Waste Management Technologies in the Ceramic and Nuclear Industries VII (Ceramic Transactions,Volume I 32)
Edited by Gary L. Smith, S.K. Sundaram, and Dane R. Spearing
02002, ISBN I-57498- 146-3 Environmental Issues and Waste Management Technologies in the Ceramic and Nuclear Industries VII (Ceramic TransactionsVolume I 19) Edited by Dane R. Spearing, Gary L. Smith, and Robert L. Putnam 0200 I, ISBN 1-57498- I 16- I The Magic of Ceramics
By David W. Richerson
02000, ISBN 1-57498-050-5 Environmental Issues and Waste Management Technologies in the Ceramic and Nuclear Industries V (Ceramic TransactionsVolume I 0 7)
Edited by GregoryT. Chandler and Xiangdong Feng
02000, ISBN 1-57498-097- I Ceramic Innovations in the 20th Century Edited by John B.Wachtman Jr. 0 I 999, ISBN I-57498-093-9 Environmental and Waste Management Technologies in the Ceramic and Nuclear Industries IV (Ceramic TransactionsVolume 93)
Edited by J. C. Marra and G.T. Chandler 0 1999, ISBN I-57498-057-2 Environmental and Waste Management Technologies in the Ceramic and Nuclear Industries 111 (Ceramic TransactionsVolume 8 7) Edited by D. Peeler and J. C. Marra 0 1998, ISBN I-57498-035- I Environmental and Waste Management Technologies in the Ceramic and Nuclear Industries I1 (Ceramic TransactionsVolume 72) EditedV. Jain and D. Peeler 0 1996, ISBN: I-57498-023-8 Environmental and Waste Management Technologies in the Ceramic and Nuclear Industries (Ceramic TransactionsVolume 6 I ) Edited by! Jain and R. Palmer 0 1995, ISBN I-57498-004- I Environmental and Waste Management Issues in the Ceramic Industry I1 (Ceramic TransactionsVolume 45) Edited by D. Bickford, S.Bates,V. Jain, and G. Smith 0 I 994, ISBN 0-944904-79-3 Environmental and Waste Management Issues in the Ceramic Industry I (Ceramic TransactionsVolume 39) Edited by G. B. Mellinger 0 I994, ISBN I-944904-7I-8 For information on ordering titles published by The American Ceramic Society, or to request a publications catalog, please contact our Customer Service Department at 6 14-794-5890 (phone), 6 14-794-5892 (fax),
[email protected] (e-mail), or write to Customer Service Department, 735 Ceramic Place, Westerville, OH 4308 I , USA. Visit o u r on-line book catalog at www.ceramics.org.
Environmental Issues and Waste Management Technologies in the Ceramic and Nuclear Industries VIII Proceedings of the Science and ~ e c h ~ ino A~dressing ~ o ~ ~ y Ceramic €nvironmen~alIssues in the Ceramic l n ~ u s and Science and Technology for the Nuclear Industry symposia held at the /04* An~uu/Meeting ofThe American Ceramic Sociely 2002 in St Louis, Missouri A ~ r i28-30, /
€ ~ i by~ e ~ S.K. Sundaram
Pacific Northwest National laboratory
Dane R. Spearing
Los Alamos National Laboratory
john D.Vienna Pacific Northwest National Laboratory
Published by The American Ceramic Society 735 Ceramic Place Westerville, Ohio 4308 I ~.ceramics.or~
Proceedings of the Science and Technology in Addressing Environmental Issues in the Ceramic Industry and Ceramic Science and Technology for the Nuclear Industry symposia held at the I04* Annual Meeting ofThe American Ceramic Societ).:April 28-30,2002 in S t Louis, Missouri
Copyright 2003,The American Ceramic Society. All rights reserved. Statements of fact and opinion are the responsibility of the authors alone and do not imply an opinion on the part of the officers,staff,or members ofThe American Ceramic SocietyThe American Ceramic Society assumes no responsibility for the statements and opinions advanced by the contributors to its publications or by the speakers at its programs. Registered names and trademarks, etc., used in this publication, even without specific indication thereof, are not to be considered unprotected by the law. No part of this book may be reproduced,stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical,photocopying, microfilming, recording, or otherwise, without written permission from the publisher:
Authorization t o photocopy for internal or personal use beyond the limits of Sections I07 and I08 of the US. Copyright Law is granted by the American Ceramic Society, ISSN 1040- I I 22 provided that the appropriate fee is paid directly t o the Copyright Clearance Center; Inc.,222 Rosewood Drive, Danvers, MA 0 I923 USA, www.copyright.com. Prior t o photocopying items for educational classroom use, please contact Copyright Clearance Center; Inc. This consent does not extend t o copying items for general distribution or for advertising or promotional purposes or to republishing items in whole or in part in any work in any format. Please direct republication or special copying permission requests to the Senior Director; Publications,The American Ceramic Society, PO Box 6 136,Westerville,Ohio 43086-6 136, USA.
Cover photo: Scanning electron micrograph o f a crushed glass sample is courtesy ofTAkai, D. Chen, Y Yamamoto,7: Shirakami, K. Urabe, K Kuraoka, and IYazawa, and appears as Figure I in their paper “Sodium Extraction from Waste Glass by Acid Leaching to Obtain Silica Source for Construction Materials,” which begins on page 39.
For information on ordering titles published byThe American Ceramic Society, or t o requert a publications catalog, please call 6 I 4-794-5890. Printed in the United States of America.
4 3 2 1-05 04 03 02 ISSN 1042-1 122 ISBN 1-57498-159-5
Preface
............................................
xi
Recycling of Ceramics and Glasses Industrial Applications for Spent Refractory Materials jJ? Bennett and I(.-S. Kwong
......... 3
Ceramic-Based Magnetic Extractants for Removal of Organics from Water .................................
IS
A. Apblett, S.M. AI-Fadul, and T.M.Trad
Investigation on a Recycling Process ofwaste Colored Glass .............................. D, Chen, H. Masui,TAkai, and TYazawa
Use of Mid-Delaware River Dredge Sediment as a Raw Material in Ceramic Processing ..................... K, Hill and R.A. Haber
1
.23
3I
Sodium Extraction from Waste Glass by Acid Leaching t o Obtain a Silica Source for Construction Materials . . . . . . . . . 39 T. Akai, D. Chen,YYamamoto,%Shirakami, K.Urabe, K. Kuraoka,and IYazawa
Emissions in Glass and Ceramic Industries Analysis of Emissions from Nitrate Containing Glasses S, Luo and L.E,jones
. . . . . . . 49
l Characterizing Particulate Emissions using MicrometernScale X-Ray Fluorescence
................... 59
J.F. Shackelford, FIB, Kelly, S.S. Cliff, M. Jimenez-Cruz,and TA. Cahill
1
Dilatometry and Mass Spectrometry Study of the Decomposition and Sintering of Calcium Carbonate K, Feng and S.J. Lombardo
~
V
.........67
Lead-Free Electronics: Current and Pending Legislation J.M.Schoenung
. . . . . . 75
First Delisting Petition Approval by the US EPA for a Vitrified Mixed Waste ...............................
.83
J.B.Pickett, C.M.Jantzen,and L.C. Martin
Characterization of Defense Nuclear Waste using Hazardous Waste Guidance: lnsights on the Process at Hanford . . . . . . . . . 95 M. Lerchen, L. Huffman,W. Hamel, and K. Wiemers
Effect of TransitionIN on-Transition Metal Modification on the Activity of Ga2O3-AI2O3Catalyst for NOx Reduction by Hydrocarbon under Oxygen-Rich Conditions . . . . . . . . . . . I05 M.H.Zahir; S. Katayama, K.Maeda, and M.Awano
Vitrification Technology and Melter Disassembly COGEMA Experience in Operating and Dismantling HLW Melter .............................
I13
R. Do-Quang, J.L.Desvaux, I? Mougnard,A. Jouan,and C. Ladirat
Conceptual Methods for Disposal of a DWPF Melter and Components .............................
123
M.E. Smith, D.F.Bickford, F.M. Heckendorn, and E.M. Kriikku
Evaluation of Crystallinity Constraint for HLW Glass Processing ...............................
133
I? Hrma,J. Maty65, and D.-S. Kim
Ruthenium - Spine1 Interaction in a Model High-LevelWaste (HLW) Glass ........................
I 4I
TM.Willwater,J.V.Crum, S.M. Goodwin, and S.K. Sundaram
Glass Formulation and Testing Interim Models Developed to Predict Key Hanford Waste Glass Properties using Composition . . . . . . . . . . . . . . . I 5 I J.D.Vienna,D-S. Kim, and I? Hrma
Relationship between Liquidus Temperature and Solubility I? Hrma and J.D.Vienna
vi
. . . . I59
Glass Formulation for INEEL Sodium Bearingwaste J.D.Vienna,D.-S. Kim, and D.K.Peeler
Vitrification of Korean Low-Level Waste
. . . . . . . . 169
. . . . . . . . . . . . . . . . . I77
L.O. Nelson, I? Kong, G.Anderson, K. Choi, C.-W. Kim, and S.-W. Shin
Phase Equilibria7Viscosity7 Durability, and Raman Spectra in the System for Idaho Nuclear Waste Forms . . . . . . . . . . . . . I85 S.V. Raman, B.A. Scholes,A. Erickson, and A.A. Zareba
Measurement of SimulatedWaste Glass Viscosity R.F. Schumacher;TB. Edwards, D.K. Peeler; and A.G. Blum
. . . . . . . . . . I99
Hanford Tank Waste Treatment Hanford Low-LevelWaste Form Performance for Meeting Land Disposal Requirements . . . . . . . . . . . . . . . . . . .2 R.F. Schumacher;C.L. Crawford, N.E. Bibler; D.M. Ferrara, H.D. Smith, G.L. Smith,J.D.Vienna,D.B.Blumenkranz, D.J.Swanberg, I.L. Pegg, and IS. Muller
Leaching Mechanism of Borosilicate Glasses under TCLP Conditions ..............................
2I5
H. Gan and I.L. P e g
Electrochemical Studies of SuIfate-Containing Waste Glass Melts .................................
.225
LVidensky,H. Gan,A.C. Buechele, and I.L. Pegg
Durability Testing and Modeling Modeling High-Level Waste Glass Degradation in Performance Assessment Calculations . . . . . . . . . . . . . . . . . . .235 W.L. Ebert
Waste Glass Corrosion: Some Open Questions I?Hrma,J.D.Vienna,and J.D.Yeager
. . . . . . . . . . . 245
Vapor Phase Hydration of Glasses in H 2 0and D 2 0
. . . . . . . . 253
TR. Schatz,A.C. Buechele, C.F. Mooers, R.Wysoczanski, and I.L. Pegg
vii
Modeling Fluid Chemistry Inside a Waste Package Due t o Waste Form and Waste Package Corrosion . . . . . . . . . . . . . 263 V. Jain and N. Sridhar
Leaching Full-Scale Fractured Glass Blocks Y Minet and N. Godon
. . . . . . . . . . . . . . .275
Development of Sensors for Waste Package Testing and Monitoring in the Long-Term Repository Environments V. lain, S. Brossia, D. Dunn, and L.Yang
Corrosion of Partially Crystallized Glasses I? Hrma, B.J.Riley, and ].D.Vienna
. . . . . 283
. . . . . . . . . . . . . . . 29 I
Ceramic and Alternative Waste Forms Development of Titanate Ceramic Wasteforms and Crystal Chemistry of Incorporated Uranium and Plutonium . . . . . . . . 301 E.R.Vance
Substitution of Zr, Mg,AI, Fe, Mn, CO,and Ni in Zirconolite, CaZrTi20, ..............................
3I3
E.R.Vance,J.V.Hanna, B.A. Hunter, B.D. Beg, D.S. Perera, H. Li, and Z.-M. Zhang
Effects of Sub-Surface Damage Induced by Mechanical Polishing on LeachTesting of Cesium-Bearing Hollandite M.L. Carter; E.R.Vance,DJAttard, and D.R.G.Mitchell
. . . . . 321
Iron Phosphate Glasses for Vitrifying Sodium BearingWaste C.-W. Kim, D. Zhu, D.E. Day, and D. Gombert
. . 329
Phosphate Glasses for Vitrification of Waste with High Sulfur Content.. ...............................
337
Solubility of High-Chrome Nuclear Wastes in Iron Phosphate Glasses ..................................
347
D.-S.Kim, ].D.Vienna, I? Hrma, and N. Cassingham
W. Huang, C.-W. Kim, C.S. Ray, and D.E.Day
Development of a Sampling Method for Qualification of a Ceramic High-LevelWaste Form . . . . . . . . . . . . . . . . . . . 355 TI? O’Holleran and K.J.Bateman
...
Vlll
Microwave Heating for Production of a Glass Bonded Ceramic High-LevelWaste Form ............... , 3 6 3 TF?O’Holleran
Morphology and Composition of Simulant Waste Loaded Polymer Composite - Phase Inversion, Encapsulation, and Durability .......................... H.D.Smith, G.L. Smith, G. Xia, and B.j.j. Zelinski 93NbMAS NMR of Niobium Containing Silicotitanate Exchange Materials ....................... B.R. Cherry, M. Nyman, and TM. Alam
37 I
377
Selective Absorption of Heavy Metals and Radionuclides from Water in a Direct-to-Ceramic Process .............. 385 B.F? Kiran,A , ~Apblett, , and M. Ch~hboun;
ix
This page intentionally left blank
In 2002,The American Ceramic Society hosted several symposia focusing on eight broad topics at its Annual Meeting. Two key symposia, Ceramic Science and Technology for the Nuclear Industry and Science and Technology in Addressing ~nvironmentalissues in the Ceramic ~ndustry, clearly illustrate the delicate balance that exists among the environment, the processes/technologiesthat have been used in glass and ceramic industries, as well as the wastes both nuclear and nonunuclear (hazardous) that have been generated. Ceramics and glasses play a critical role in the nuclear industry. Nuclear fuels and waste forms for low-level and high-level radioactive, mixed, and hazardous wastes are primari~y either ceramic or glass. With increasingly stringent environmental regulations and demands that are placed on our limited natural resources, it is critical to identify and adequately address environmen~lissues in the ceramic and glass industry to ensure longevity and success. In ceramic/glass manufacturing, companies are beginning to focus on green ceramics, performing life cycle analyses, and adopting environmental stewardship to manufacture environmen~llyfriendly products. Effective and responsible environmental stewardship is becoming increasingly more important in the world. The symposia and subsequent proceedings help foster continued scientific understanding, techno~ogicalgrowth, and environmental stewards~ip within the fields of ceramics, glass, and envjronmen~l/nuclearengineering. This proceedings combines key papers that were presented at the above mentioned symposia duringThe American Ceramic Society I 04th Annua~ Meeting & Exposition held April 28-30,2002 in St. Louis, Missouri. This is the fourteenth volume published by The American Ceramic Society in the areas of waste management and environmental issues in relation to ceramics. Previous proceedings on nuclear waste management and environmental issues date from I983 and include Advances in Ceramics volumes 8 and 20 and Ceramic Transactions volumes 9,23,39,45,6 I , 72,87, 93, t 07, I 19, and i32.
xi
The editors thank Robert L. Putnam, Boyd Clark, Linda E. Jones,Jeff Kohli, G. L. Smith and Carol M. Jantzen for their help in organizing the symposia and James I? Bennett and James C. Marra, for their contribution in chairing the sessions and keeping the presentations running smoothly. The editors appreciate the support from John Marra,Vijay Jain, Greg Chandler, and the Nuclear and EnvironmentalTechnology Division and Glass and Optical Materials Division of The American Ceramic Society. The editors acknowledge most importantly, the authors and reviewers. Without them, a high quality proceedings volume and timely publication would not be possible. The editors also appreciate all the support from Kevin G. Ewsuk and Chris Schnitzer. Lastly, the editors thankTeresa Schott of the Pacific Northwest National Laboratory and the book publishing team at The American Ceramic Society: Mary Cassells, Sarah Godby, Greg Geiger, and Mark Mecklenborg. Their support and contributions were instrumental in the publication of this volume. S.K. Sundaram Dane R. Spearing John D.Vienna
xii
Recycling of Ceramics and Glasses
This page intentionally left blank
INDUSTRIAL APPLICATIONS FOR SPENT REFRACTORY MATERIALS James P. Bennett, Kyei-Sing Kwong Albany Research Center - USDOE 1450 Queen Ave. SW Albany, OR 97321 Phone: 541-967-5983 FAX: 541-967-5845 E-Mail: Bennett@,alrc.doe.gov ABSTRACT The reusehecycling of spent refractory materials by industry is limited to a few companies because of the lack of economic and legislative driving forces. For most users of refractories in the United States, it is more economical to landfill spent refractory materials than to reuse/recycle the material. Where recycling is successful, applications for the spent refractory materials are primarily as a refractory raw material and as a slag conditioner. Applications for spent refractory materials in steel, aluminum, glass, and brass industries will be evaluated, with emphasis on what cormnon elements make up these programs. INTRODUCTION The reuse of refractory materials after removal from service is not commonly practiced by industry because of economics and a lack of legislated driving forces, with most spent refractory material being land filled after removal. Structured programs for the recyclingheuse of spent refractories do not exist. Interest in recycling has been cyclic, driven in part by legislation, environmental concerns, product stewardship, company environmental policy, high landfill costs, or a lack of landfill space. Marketplace forces have also started to play a role in recycling through companies seeking IS0 14000 certification. Historical recycling of spent refractory materials has been practiced by refractory companies, with off specification material, floor sweepings, old refractory formulations, excess material, grinding, and some spent material often added back to refractory batch formulations. Where used in these formulations, the spent refractory acts as a grog, reducing drying and firing shrinkage, giving volume stability, and reducing energy consumption during firing. In the United To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
3
States, large scale recycling of spent refractory materials is limited to a few industries, such as the glass, where the driving force is environmental regulations due to the use of chrome oxide as a raw material in the refractory. Chrome oxide changes valence from the trivalent to hexivalant state under certain use conditions, causing it to be classified as a hazardous waste. Refractory companies have been reluctant to take spent material back from users for a number of reasons. Top among concerns include the amount and frequency of spent refractory available and concerns over the performance of materials that contain spent refractory. Other concerns center on refractory contamination and its beneficiation costs, shipping distances to a refractory user, andor limited demand for the processed materials. These factors can make beneficiating and recycling material uneconomical when compared to virgin, mined material costs, consistency, and availability. Attempts to recycle in the United States have tended to follow economic cycles, with some companies interested in recycling during good economic times, but focusing on core business activities during a recession. Mergers and consolidation in the refractory industry have not helped recycling programs, with efforts at recycling decreasing or being eliminated as consolidation in the refractory industry has occurred. Established programs with refractory users in the glass industry have survived because of the economic costs associated with hazardous waste disposal. More emphasis has been placed on recycling in Europe because of a lack of landfill space, because the distance between refractory users and producers is generally small, and because regulations have been legislated that impact waste disposal. Some steel producers in Europe, for example, mandate the return of spent refractory material as part of their contract with the refractory producer. In Japan, recyclers have placed emphasis on recyclingfreusing spent refractory materials because of decreasing landfill space, focusing on basic materials as slag conditioners and on high value slide gate valves
’.
Refractory users who want spent refractory reusedrecycled must keep in mind that spent refractory must be utilized in a product for some application, and that there are refractory users who do not want refractories containing spent material because of concerns over decreased service life. Other refractory users do not share the performance concerns with the reuse of recycled materials, but don’t have the storage space for spent material or don’t have refractory producers or recyclers in close proximity to process the spent material economically into a product. Ideally, a refractory user would like a material with the longest life that is the most cost effective to use in an application. Most recycling decisions are based on economics - does it make economic sense to recycle or to landfill spent
4
Environmental Issues and Waste Management TechnologiesVIII
material? Potential applications for spent refractory materials are listed in Table I. Table I - Possible applications for went refractory materials Roofing granules Refractory component Tile body component Insulating powder Raw material for glass Component in cement, aggregate for Highway road aggregatehubbase concrete Ferro alloy (high chrome containing Building component materials) Abrasive Fuel source (SIC, C containing materials) Soil conditioner Filler for bulk items Carbon, silicon source Landscape material Grog in ceramic materials Soil stabilization Sulphur removal in the ladle steels Slag conditioner Waste neutralizatiodtreatment (acids, pathogens) Contamination of refractory materials is an area of great concern in recycling. It can range from process infiltration into the refractory to the mixing of zoned linings on tearout from the furnace. Beneficiation of contaminated material makes up a high portion of refractory recycling costs, underscoring the importance of keeping the material clean on tearout ’. Not only does clean material have more potential applications, it also has lower beneficiation costs. Recycled refractory material uses can range from low to high volume applications with varying material values.
A number of companies recycle spent refractory materials. These companies have established markets or applications for the spent material and take in no more material than they can market. They possess the specialized knowledge and equipment for processing the spent refractory material. Limitations exist on processing the material, which center around contamination and beneficiation costs, shipping distances andor limited demand for the processed materials, factors that can make the recycled material uneconomical when compared to virgin, mined material costs, constancy, and availability. Regardless of who does the beneficiation of the refractory material, flow sheets tend to follow the setup as shown in figure 1. Materials are typically sorted on removal, especially if the refractory lining is zoned or if heavy contamination of the refractory has occurred. This is done by hand at a few recyclers. The spent refractory is next crushed, dried, and screened. Other beneficiation circuits can be added to this flow sheet, but a simple flow sheet is typically used. This type of
Environmental Issues and Waste Management Technologies VIII
5
beneficiation circuit is typically followed by industries such as glass, steel, alwninum, and brass. Common beneficiation flow sheets and applications for beneficiated material at a number of industries in these areas will be discussed. On sitelln place refractory beneficiationlremoval
*
rl Primary crushing
Drying
Figure 1 - General beneficiation process for refractory materials GLASS INDUSTRY RECYCLING Recycling in the textile and fiberglass industry of the United States has been brought about by Federal regulations impacting the disposal of spent refractory materials containing chrome oxide, which in the presence of alkali/alkaline earth at elevated temperatures forms hexivalent chrome, a hazardous material. Regulations governing the amount of hexivalent chrome oxide in a material evolved from the Resource Conservation and Recovery Act of 1976, which was used as the basis for establishing the Toxicity Characteristics Leaching Procedure (TCLP). A leachable limit of 5 mg/L was established for chrome under this act. Because of this legislation, many steel producers have worked to eliminate any chrome containing materials from their plant. Chrome oxide containing refractories have many good properties that cannot be reproduced by other refractories in the glass industry, which has brought about the continued usage of this material in this field. Refractory producers have traditionally placed emphasis on performance, not recycleability. Little thought or effort has been made towards developing easily recycled refractory materials. With the disposal restrictions and costs placed on chrome oxide containing refractories, recycling programs for many chrome oxide-
6
Environmental Issues and Waste Management Technologies VIII
containing products from the glass industry were developed over a period of years. In the U.S., magnesia-chrome spent refractory material from the reheat checkers of glass furnaces are recycled back as a refractory raw material after treatment by a proprietary water leaching process that removes soluble sulfates and chromates '. Fused material from glass furnaces is crushed and added back as a raw material feed for new refractories 5 , however, upper limits have been found to exist on the quantity of material that may be added before the performance of the material is adversely affected below acceptable levels. The performance issues are due to impurities present in the spent refractory.
STEEL INDUSTRY RECYCLING A number of refractory recycling programs have been successfully developed by both integrated and electric arc furnace (EM) steel producers, although these programs are not extensively used. More potential applications for internal reuse of spent refractory materials exist at integrated shops than EAF shops, if the mills are willing andor able to develop these applications. Flow sheets representative of spent refractory reuse as a slag conditioner and as a refractory raw material are shown in figure 2 (a-b). Magnetic material is typically removed from the refractory materials for remelt in the steelmaking furnaces, although this is not shown in both flow sheets. Mill service personnel, who perform other services at the steel mills, typically do magnetic material removal. Reuse of the spent refractory material as slag conditioners is typically limited to basic materials for EAF and basic oxygen furnace applications. Hydration and the speed of hydration limit basic material reuse. It is important to note that when basic materials are reused as slag conditioners, CaO and MgO levels in the sla has a large impact on the slag viscosity, slag foaming, and refractory wear . Additions of basic materials must be made with careful attention paid to the slag chemistry (Si02, CaO, MgO, FeO, and Al2O3), temperature, CaO/(Si02 + A1203) ratio, and the A1203/(Si02 + Al2O3) ratio.
B
Work done at MEFOS in Sweden has indicated spent alumina-silicate materials can be used for s u l k removal from ladle slags. Alumina raises slag viscosity, so one must be careful how much material is added. It is of interest to note that one of the flow sheets in figure 2 shows all spent refractory material being reused as a refractory raw material while the other shows 55 pct being returned to the refractory manufacturer (probably for reuse as a refractory raw material). Also note that some material (5 pct) is disposed of in a landfill in one of the steel mills. No applications could be found for this material because of contamination fi-om the process and because of zoned materials. Other application for materials removed from steel mills depend on material chemistry,
Environmental Issues and Waste Management Technologies VIII
7
but typically involves applications as blast furnace flux,sinter plant additives, sludge thickeners, road and concrete aggregate, and sand blast materials.
AbO,
+-
t
Mill Service
$ . SiO,
MgOlC
*’
Segregated on Removal
Refractory Manufacturer
* Sinter Plant * Ladle Slag Conditioner
2 (b)
Figure 2 (a-b) - Process flow sheets used in the steel industry.
ALUMINUM INDUSTRY RECYCLING Recycling of spent refi-actory materials in the aluminum industry has been limited primarily to carbon bake furnace brick *. Refiactories removed fiom the primary andor secondary melting furnaces are typically monolithic materials and have issues associated with impurities, types of material, size of material removed from furnaces, variable bond strength, and possible anchors in the monolithic. It may be possible to utilize the high alumina materials removed the primary and secondary furnaces to satisfjr the alumina requirements of cement if local raw materials utilized by the cement producer are deficient in alumina. Alumina substitution research was conducted by the Univ. of Missouri at Rolla and
8
Environmental Issues and Waste Management TechnologiesVIII
indicated this as a possible application '. A flow sheet from a company associated with reuse of carbon bake furnace brick from primary aluminum production is shown in figure 3. Furnace Tearout
(1.4 %)
I
v
Leave intact Reuse
I Storage
I
Segregate *iscard
Clean I Reuse Dense Firebricks
-1 1 I
IFB's
pfFqq Castable
Castable
Figure 3 - Process flow sheet used to recycle spent refractory materials from a carbon bake furnace of an aluminum Droducer.
Note that the process flow sheet shows that spent refractory material removed from the furnace was reused, recycled, or discarded, with over 98 pct of the refractory material being either recycled or reused. The bulk of the recycled refractory was reused in refractory castables and was used to rebuild the carbon bake furnace. The original refractory was approximately 50 pct A1203. Other applications for the dense alumina-silicate refractory were as a decorative aggregate and as a roadway aggregate. This recycling program had the support and commitment of management when enacted. The driving forces were environmental (reduced landfill space) and economic (reduced furnace rebuilding costs). Management endorsed and supported the plan. BRASS INDUSTRY RECYCLING Recycling by the brass and bronze industry is being driven by environmental regulations. Brass oftentimes contains lead, an element that slowly builds up in the brass furnace lining through vapor penetration or through direct contact with the metal and slag l0. The limit of lead in the furnace lining is found to be above the 5 mg/L allowable limit by TCLP testing. A process flow sheet for a brass and
Environmental Issues and Waste Management Technologies VIII
9
bronze producer indicating how the spent refractory is disposedrecycled is shown in figure 4. Material segregation on removal from the furnace
I I I Crushing
Screening
I
-
t
I
Reuse slag conditioner hazardous waste landfill
Figure 4 - Process General beneficiation process for refractory materials
Refractory material that comes from brass and bronze finaces are about 70pct alumina. Because it is considered a hazardous waste and must be disposed of in a hazardous waste landfill, the brass and bronze recyclers have started to crush the refractory and reuse it in the furnace to saturated the slag with alumina, reducing refractory wear through corrosion. COMMON ELEMENTS OF RECYCLING PROGRAMS Companies that recycle have done much work to ensure their programs remain successful. Management typically has appointed an individual to head the program and he has the commitment and backing of management. An evaluation of the types and quantities of refractory materials coming to the plant, areas of usage within the plant, current “disposal” practices, and an assignment of costs associated with the spent refractory handling/disposal (estimate both current and hture costs) should be made. An assessment should also be made as to possible reuse or disposal options for the spent materials and consideration given to production process changes that would result in a refractory waste reduction. This assessment should be made according to the priorities listed in Table I1 l l . Table I1 - Priorities for spent refractory waste reuse/disposal l1 1. Reduction in the process 2. Internal reuse 3. External reuse 4. Treatment 5. Landfill
10
Environmental Issues and Waste Management Technologies VIII
Reductions in the amount of refractory waste generated through an evaluation process can significantly reduce the amount of material to recycle. Technological advances, for example, have played a significant role in reducing both the amount and the frequency of spent refractory material generation. Examples are improved ladle and furnace design, maintenance and repair schedules; better slag and metallurgy control; slag splashing l 2 and improved refractory materials and installation practices. Improved refractory materials and installation practices include better customer service by refractory producers; hot patching; monolithic materials; zoning; gunning; subcontracted refractory installation, tearout and maintenance; the practice of the endless lining concept 13; and refractory wear sensing using laser wear profiling. After considering ways to reduce refractory wastes generated by the process, consideration should be given to internal utilization of the spent refractory material. An example of internal reuse would be the use of basic refractory materials as slag conditioners in electric arc furnaces or the reuse of spent MgO/C materials on the sidewalls of basic oxygen furnaces to freeze slag (slag splashing). Active recyclingheuse programs are necessary to reuse or recycle what refractory wastes are generated and to deal with the changing nature of refractory materials and the process. When initiating any recycling program, it is best to fxst recycle those materials that are the easiest to recycle or which have an established marketlapplication first, leaving a positive recycling experience in the plant. The examples of refractory recycling shown in figures 2-4 are of materials in applications easily accomplished by a refractory user. More difficult materials or materials with limited demand should be evaluated and processed at a later time, or can continue to be disposed of in a landfill if an economic justification or driving force does not exist. A common problem in any recycling program involves the crushing operation where dust can be generated. Mill services in the U.S. have attempted to beneficiate spent refractory material for steel mills but have encountered problems with dust suppression, causing most to discontinue processing the spent refractory material for reusehecycling. Where the basic materials are crushed for reuse as a slag conditioner, care must be taken to process the spent refractory into particle sizes that will dissolve in the slag during the heat (below 2.5 cm in size), yet will not be so fine as to be exhausted into the dust collector, altering the chemistry and lowering the market value of recoverable Zn in the EAF dust.
-
Regardless of how refractory recycling/reuse is approached, the commitment and backing of management is essential to develop and implement a successful
Environmental Issues and Waste Management TechnologiesVIII
11
program. Management should appoint one individual to head the effort, providing him with strong support. External utilization or processing by companies specializing in recycling spent refractory material should also be considered, an option that would allow the refractory user to concentrate on their core business. The economics of any recycling program should be compared with purchased materials before commitments are made. The last option for handling spent refractory material should be refractory waste treatment andor disposal in a landfill. CONCLUSIONS Spent refractories can be reusedhecycled as a raw material source in a number of products or processes ranging from low to high volume applications and with varying material values. Material cleanliness, beneficiation costs, and material consistency will play a large role in determining these applications. Successful recycling efforts are utilized in a number of industries, including glass, steel, aluminum, and brass. Applications for the spent refractories have typically been internal rather than external. At a minimum, processing of the spent refractory material typically involves segregation into the different types of refractories, than crushing and magnetic separation. The most common applications for refractory reusehecycling are as a refractory raw material or as a slag conditioner. Recyclers are often a viable option for refractory reusehecycling. Barriers to material reuse are oftentimes based more on an unwillingness to change prior practices and on a lack of driving forces than on barriers caused by technical issues. Refractories removed from service are typically landfilled. A number of driving forces to encourage recycling have recently emerged. These include growing concerns over legislation, the environment, and future liability. REFERENCES 1. Takahashi, H., M. Tsuno, and M. Hayaishi, “Recycling of Used Refractories in an Electric Steelmaking Shop,” Journal of the Tech. Assoc. of Refractories, Japan, 20(4), 2000, pp. 249-253. 2. Oxnard, R.T., “Refractory Recycling,” ACerS, 73( 10), Oct. 1994, pp 46-49. 3. U.S. Code of Federal Regulations, Title 40-Protection of the Environment, Part 26 1--Identification and Listing of Hazardous Waste, July 1, 1999. 4. Noga, J., “Refractory Recycling Developments,’’ Ceram. Eng. Sci. Proc., 15(2), 1994, pp 73-77.
12
Environmental Issues and Waste Management Technologies VIII
5. Webber, R.A., “Recycling at Corhart - A 30 Year Success Story,” Ceram. Eng. Sci. Pro., 16(1), 1995, pp 214-215. 6. Kwong, J., and J.P. Bennett, “Achieving MgO Saturated Foamy Slags in the EM,” Proceeding of the 5gthEAF Conference and lgth Process Technology Conf. Proc., Nov. 11-14, 2001, Phoenix AZ,published by the ISS, pp 277285. 7. Viklund-White, C., H.Johansson, and R. Ponkala, “Utilization of Spent Refractories as Slag Formers in Steelmaking,” Proc. of the 6th Int’l Conf. on Molten Slags, Fluzes, and Salts, ed. by S. Seetharaman and D. Sichen, meeting held in Stockholm, Sweden and Helsinki, Finland, June 12-17, 2000, pub. on CD by Division of Metallurgy, KTH, Sweden, paper is 13 pages. 8. Holmes, L., N.S. Schubert, A. Mooney, J. Bennett, and KS. Kwong, “Recycling of Spent Refractory Material from Carbon Baking Fwnaces,” Proc. of the UNITECR 5thBiennial Worldwide Congress held in New Orleans, LA, USA, Nov. 4-7, 1997, pp 477-486.
9. Smith, J.D. and K.D. Peaslee, “Spent Refractory Waste Recycling from NonFerrous Metals Manufacturers in Missouri,” Proc. of the 4th Int’l Symposium on Recycling of Metals an Engineered Materials, edited by D.L Stewart, J.C Daley, and R.L. Stephens, meeting held in Pittsburgh, PA on Oct. 22-25, 2000, published by TMS, pp 1385-1394. 10. Kwong, K.S., J.P. Bennett, K.W. Collins, and A.E. Wynne 111, “The Recycling of a 70 % A1203 Spent Refractory,” Proc. of the UNITECR 5th Biennial Worldwide Congress held in New Orleans, LA, USA, Nov. 4-7, 1997, pp 487-496. 11. Abrino, D.E., “Waste Minimization in Industries Using Refractory Materials,” Proc. of the UNITECR 5th Biennial Worldwide Congress held in New Orleans, LA, USA, Nov. 4-7, 1997, pp.465-471. 12. Goodson, K.M., N. Donagly, and R.O. Russell, “Furnace Refractory Maintenance and Slag Splashing,” Iron and Steelmaker, ISS, 22(6) 1995, pp 3 1-34. 13. Alasarela, E. and W. Eitel, “How Infinite is Endless Linings of Ladles,” Paper in Proc. of the UNITECR, San Paulo, Brazil, Oct 3 1-Nov. 3, 1993, pp 12671278.
Environmental Issues and Waste Management TechnologiesVIII
13
This page intentionally left blank
CERAMIC-BASED MAGNETIC EXTRACTANTS FOR REMOVAL OF ORGANICS FROM WATER Allen Apblett, Solaiman M. Al-Fadul, and Tarek M. Trad Department of Chemistry Oklahoma State University Stillwater, OK 74078-3071 ABSTRACT Magnetic filtration can provide rapid, efficient removal of magnetic materials from a waste stream. However, since most pollutants are non-magnetic, it is necessary to use magnetically-active “extractants” to bind to pollutants and allow their separation by a magnetic filter. Excellent candidates for extractants are particles of magnetic ceramic oxides such as ferrites and magnetite whose surfaces have been derivitized to provide binding sites for toxic metal ions or organic pollutants. Such materials can be used to separate dyes or petrochemicals from water and break oil in water emulsions. INTRODUCTION The application of efficient magnetic filtration to decontamination and waste treatment operations is attractive because it can provide very rapid separation of pollutants from aqueous waste streams. This coupled with the ability to switch the filter on and off electronically (avoiding any need for mechanical contact) allows the minimization of exposure of workers to harmful agents. However, since many environmental contaminants are not magnetic, magnetic filtering aids must be developed that bind the materials and allow their magnetic separation. This problem has been addressed in coal beneficiation by use of magnetic fluids that are composed of magnetic particles, a suspending agent, and a carrier solvent that selectively wets the contaminant particles (oxide minerals in the case of coal) [ 11. Several approaches have been previously developed for separation of oil in water mixtures. The simplest method was to mix an extremely large excess of magnetite (ratio of 40 Fe304 1 oil (fatty acid) by weight) so that the oil absorbed onto the surface of the magnetic powder [2]. Subsequent magnetic filtration reduced the oil from 500 ppm to 2 ppm. The success of the absorption method can be attributed to the oil fEst being emulsified in an ionic form-an approach that is not applicable To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society, Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
15
to normal hydrocarbons. When the same approach is applied to a 3000 pprn Bheavy oil in water emulsion, the final concentration of oil was only reduced to 57 ppm [3]. Another method that utilized a magnetic suspension and an acidic coagulant merely achieved a final concentration of 103 ppm [4]. There is also a successful application of the magnetic-filler-in-polymer technology that has been used for oil slick removal. In this case the polymer was polystyrene and the filler was iron oxide. The extractant was sprayed onto an oil slick by a watercraft travelling through the spill [51. The polymer/oil sludge was then collected on rotating magnetic disks. Using this approach, a 99% recovery of spilt oil was achieved with 20 volumes of oil being collected per unit volume of polymer. The objective of the research reported herein is the development of single component systems for use as magnetic filtration aids i.e. magnetic materials that can absorb pollutants and allow their separation from water via magnetic filtration (Figure 1).
Figure 1. Cartoon of Magnetic Extraction Separation of oil/water emulsions might also be feasible using magnetic extraction. The outer shells of the extractants will have a strong affinity for both the oil and the hydrophobic tails of any surfactants that might be present. Therefore, it should be possible to have the surfactant/oil micelles bind strongly to the extractant and be influenced to separate from the water via a magnetic field. This potential alternative to other methods of breaking emulsions such as
16
Environmental Issues and Waste Management Technologies VIII
coalescing filters could provide considerabletimeand money savings to the petroleum industry. EXPERIMENTAL Magnetite was purchased from Strem Chemicals while Sibrid polyimide/siloxane co-polymer was purchased from Geleste. All other reagents were ACS grade and were purchased from Aldrich. All of these chemicals were used as received. Water was purified by reverse osmosis and deionization. Toluene and xylene were HPLC grade and were used without further purification. X-ray powder diffraction (XRD) patterns were recorded on a Bruker AXS D-8 Advance X-ray powder diffractometer using copper K- radiation. Surface areas were measured by nitrogen adsorption isotherms on a Quantachrome Nova 10 instrument. Magnetic filtration was performed using glass pipets packed loosely with steel wool (#OO fine grade) and taped to the side of an electromagnet. The latter magnet was a 24 V extended-reach-electromagnet with a 170 pound pull and dimensions 3"xl"x1.4". The electromagnet was powered by a 30watt direct current power supply. Preparation of PolydimethylsiloxaneMagnetiteComposites Composites of polydimethylsiloxane (PDMS) and magnetite were prepared by cross-linking a PDMS polymer at moderate temperature with the magnetic substrate. PDMS (15.0 g) was stirred with 20.0 g of magnetite to yield a paste that was heated to 280°C for one hour. Afterwards, the material was washed extensively with toluene in a Soxhlet rextractor to remove any unattached PDMS and was then dried in a vacuum oven at room temperature. Two different PDMS oils were used as starting materials: a low viscosity (10 centistokes) material and a moderate viscosity material (1000 centistokes). These yielded strikingly different materials, a homogenous thinly-coated powder in the first case and a rubbery composite in the latter. Both materials are hydrophobic and float on water. Preparation of Octadecylsilsesquioxane-CoatedMagnetite Magnetite powder (30 g) was treated with a solution of octadecyltrimethoxysilane (1.O g) in toluene (25 g). After 12 hours, the derivitized powders were isolated by filtration, washed with toluene, and were dried in vacuo. Preparation of 3-(Ethylenediaminepropyl)silsesquioxane-Coated Magnetite Magnetite powder (10 g) was saturated with water by placing in an enclosed jar containing water for 12 hours. It was then placed in a hybridization tube along with a solution of trimethoxysilylpropylethylenediamine (1.O g) in toluene (80 ml), The mixture was then heated to 90°C while rotating ion a rotisserie in a hybridization oven. After 24 hours, the derivitized powders were isolated by filtration, washed with toluene, and were dried in vacuo.
Environmental Issues and Waste Management TechnologiesVIII
17
Preparation of SibricUMagnetiteComposite Sibrid (20.0 g of a15 wt% solution in N-methylpyrrolidine) was mixed with 30.0 g of magnetite to yield a sticky paste. The mixture was then placed in an oven for 24 hours at 100°C. Preparation of Magnetic Activated Carbons Magnetic activated carbons were prepared by modifling a literature procedure for activated carbon manufacture. Sawdust (200 g) was saturated with a 10% aqueous solution of sulfiuric acid (50 g) containing iron(I1) sulfate heptahydrate(5 g). The mixture was dried in air to air dry and was then fired to 500°C under a nitrogen atmosphere. ARer cooling to room temperature, the resulting powders were exposed to air at which point they became quite hot due to rapid oxidation of iron(I1) oxide to magnetite. The final result was a magnetite-containing activated carbon. A similar procedure using a nickel sulfate hexahydrate (5 g)/ Fe(S04)*6H20 (10 g) mixture in 10% sulfuric acid (50 ml) yielded a nickelhron impregnated activated carbon. Testing of Magnetic Extractants for Dye Removal from Aqueous Solution Solutions of Congo Red and Bromothymol Blue were prepared with a concentration of 30 ppm. The pH of the latter dye solution was adjusted to 4.0 so that the dye was in its deprotonated, yellow form. Next, 7.0 g of dye solution was mixed with 0.20 g of magnetic extractant and the mixture was shaken for two minutes. The resulting solution was subjected to magnetic filtration and was then analyzed for dye content by UVNisible spectroscopy. The wavelength for the absorption measurements was performed at the maximum visible light absorption for each dye, 497 and 420 nm, respectively. Dye concentrations were calculated using a calibration curve constructed from serial dilutions of the dyes over the range of interest. Testing of Magnetic Extractants for Breaking of an Emulsion A stable emulsion was prepared by diluting a 35:20:45 weight percent paraffin oil/triethanolamine/oleic acid mixture to 1000 ppm in water [6]. This yielded an indefinitely stable white emulsion. A second emulsion was prepared by sonicating 1.0 g of mineral oil and 0.1 f of Brij-35 in 1 L of water. Ten grams of each emulsion was treated with 1.0 g of magnetic extractant by briefly shaking the two materials together in a glass vial for one minute. The mixture was then passed through a magnetic filter. The extent of emulsion removal was then assessed by measuring the solution turbidity using a nephelometer. RESULTS AND DISCUSSION Preparation of Magnetic Extractants A facile method for making magnetically-active activated carbons was developed. These materials were simply prepared by impregnating sawdust with aqueous solutions of iron sulfate or irodnickel sulfate mixtures and then
18
Environmental Issues and Waste Management Technologies VIII
subjecting the treated sawdust to a procedure for preparation of activated carbon [7]. X-ray powder diffraction showed that the resulting materials contained poorly crystalline iron or nickel salts along with traces of calcium sulfate derived from calcium ions naturally present in wood. In the iron-containing activated carbon, the iron-containing phases were amorphous and could not be detected by XRD. The nickelkon derivative displayed broad reflections for y-Fe203 (maghemite) and NiFeS2 (petalite). However, the X-ray diffraction intensity for these phases was weak indicating that the bulk of the metal oxides were dispersed as amorphous small particles. This is beneficial since it prevents the particles from becoming permanently magnetized. Magnetic testing of the powders with a strong electromagnet indicated that the activated carbons were strongly ferromagnetic and no non-magnetic particles were present. Nine of the powders demonstrated any remnant magnetization when the power to the magnet was switched ofc an important property so that the powder will not stick to non-magnetized steel. A recent report indicated that polydimethylsiloxane(PDMS) is a good absorbent for phenanthrene [S] prompted the preparation of composites of PDMS with magnetite. These were synthesized by cross-linking a PDMS polymer at moderate temperature in a paste with the magnetic substrate. This is similar to a procedure reported by Soares et al. for coating alumina, calcium carbonate, and hematite with PDMS [9]. Two different PDMS oils were used as starting materials: a low viscosity (10 centistokes) material and a moderate viscosity material (1000 centistokes). These yielded powders with substantially different thicknesses of PDMS coatings so that the PDMS-1000 product was a rubbery composite while the PDMS-10 product was a loose powder. A different polymer-coated powder was prepared by casting a commercial silioxanehmide co-polymer material from an n-methylpyrollidine solution onto a magnetite powder. The composite thus prepared is stable in water and in nonpolar organic solvents so that they can be used for magnetic filtration of aqueous solutions and then be cleaned for reuse by washing with an organic solvent. The reaction of hydrolyzable organosilicon alkoxides provides another facile method for derivitizing surfaces of metal oxides. In this investigation octadecyl and ethylendiaminegroups were covalently anchored to the surface by treating magnetite powders with octadecyltrimethoxysilaneand N-(trimethoxysilylpropyl)ethylenediamine,respectively. These reagent condensewith surface hydroxyls on the iron oxide surface, leading to a monolayer of pendant octadecyl or ethylendiamine groups grafted to the metal surface via a cross-linked silica layer. Thus, the particle surface becomes coated with a monolayer of polymerized silsequioxanes, (RSiO1.5)~. Testing of Magnetic Extractants The testing of the extractants was performed using 30 ppm aqueous solutions of two dyes, one that was anionic, Congo Red, and one that was neutral,
Environmental Issues and Waste Management TechnologiesVIII
19
Bromothymol Blue in its yellow, sulfone form (see Figure 2). For the latter dye, the pH of the solution was adjusted to 4.0 to ensure it remained in the sulfone form. Each magnetic extractant was assessed for its ability to separate the dye from water via magnetic extraction. The performance of the extractants varied widely depending on the nature of the binding groups attached to the magnetic core (Table I). The anionic dye, Congo red, was efficiently adsorbed by both the untreated magnetite and the ethylenediamine-derivitizedmagnetite. The surface of magnetite is naturally-positively charged while the amine groups can be protonated by water to generate positive charges and this may account for their enhanced adsorption of Congo red. Non-polar coatings on magnetite led to very poor adsorption of Congo red but, with the exception of the octadecylsilsequioxane derivative, they were moderately successful at removing the neutral dye, Bromothymol blue, from water. These results are encouraging since they suggest that selective extractants for different classes of aqueous contaminants might be developed. Magnetite was a much poorer adsorbant for the Bromothymol blue sulfone then for Congo red as might be expected for the interaction of a positively-charged surface with a neutral and a negatively-charged molecule, respectively. However, the ethylenediamine derivative did a very good job with the neutral dye, possibly due to strong hydrogen bonding between the amine and the sulfone.
S03Na
S03Na
Congo Red
Br
CH(CHd2
Bromothymol Blue (Sulfone Form)
Figure 2. Structures of Dyes Used in this Investigation. The magnetic activated carbons with their relatively high surface areas might have been expected to significantly outperform the other extractants but this was only true for the adsorption of Congo red by the nickel iron derivative. While the activated carbons were useful adsorbants for both dyes, they performed best with the anionic dye suggesting that part of their surfaces is occupied by metal particles. This hypothesis is further supported by the enhanced uptake of dyes by the nickel-containing derivative as compared to the one that contained only iron.
20
Environmental Issues and Waste Management Technologies VIII
TABLE I. Results from Treatment of 7.0 g of Aqueous Solutions of 30 ppm Dye Solutions with 0.20 g of Magnetic Extractants. Concentrations in ppm. Extractant Surface Area (m2/g) [C. Red] [B. Blue] Magnetite Octadecylsilsesquioxane PDMS-10 PDMS-1000 Sibrid 3-(Ethy1enediamine)propy1 Activated Carboaagnetite Activated Carbon/ Nickel Iron
6.5 2.0 4.2 2.1 2.8 6.3 273 265
4.8 25.8 25.8 29.4 25.8 6.3 2.9 0.11
22.3 26.6 9.7 21.0 16.7 6.1 9.7 2.5
Several of the extractants were tested for their ability to break emulsions using magnetic filtration. The idea for this process is to have the extractant bear functional groups that can segregate at the oil water interface so that a magnetic field can sweep the oil particle out of an aqueous mixture. Two types of emulsions were tested, one prepared with a neutral surfactant (Brij 35) and the other an anionic surfactant (oleate). The extractants were briefly mixed with these emulsions and then the mixtures were separated by magnetic filtration. The effectiveness of treatment was determined by use of a nephelometer and the results are displayed in Table 11. Surprisingly, magnetite did a very good job of removing the organics from both types of emulsions and performed best with the neutral surfactant. The derivatized magnetites performed less effectively and this may be a reflection of their tendency to be poorly wetted by water and to float on the surface while magnetite disperses well throughout the solution. The results show that magnetic extractants are capable of breaking emulsions and suggest that optimization of the magnetic extractant could result in complete breaking of an oil in water emulsion. The results also indicate that the best derivative would contain a polar headgroup and a hydrophobic tail so that it can partition effectively at the oil-water interface. The relatively successful breaking of the Brij 35/oil aqueous emulsion using magnetite coated with an imiddimethylsiloxane copolymer (Sibrid) suggests that this approach may be effective since the copolymer consists of alternating blocks of hydrophobic and hydrophillic groups. CONCLUSIONS Magnetic extractants based on magnetite or ferromagnetic activated carbons show promise for separation of organic species from water. Variation of the adsorbant material placed on magnetite particles can be used to change the selectivity of the magnetic extractant. Successful separation of oil from aqueous emulsions is also possible via magnetic filtration using magnetite as an extractant.
Environmental Issues and Waste Management TechnologiesVIII
21
TABLE 11. Results from Treatment of 15 g of 100 ppm Emulsion with 1.O g of Magnetic Extractants. Results in Nephelometer Units (NTU). Extractant Brii emulsion Oleate emulsion 56.7 Initial emulsion 59.1 1.2 7.8 Magnetite 30.3 17.2 PDMS-1000 20.4 Sibrid 9.1 ACKNOWLEDGEMNT The Integrated Petroleum Environmental Consortium is gratefully acknowledged for supporting this research. The National Science Foundation, Division of Materials Research, is thanked for Award Number 987 1259 that provided funds for the X-ray powder diffractometer used in this investigation. REFERENCES [I] T. A. Sladek, "Coal Beneficiation with Magnetic Fluids" in Industrial Applications of Magnetic Separation Y. A. Liu, Ed. (Institute of Electrical and Electronics Engineers, New York, 1979). [2] W. F. Lorenc, J. A. Hyde, "Oil Removal from Waste Waters" U.S. Patent, 3161511,1974. [3] E. Nagata, H. Iwamoto, M. Kobayashi, "Separation of Oil and Water" Japan Patent, 16111493, 1977. [4] G. S. Pantelyat, V. G. Sleptsov, 3, 18-19 (1998). "Treatment of Wastewaters Containing Lubricants and Detergents by Magnetic Filtration" Vodosnabzh. Sanit. Tekh. 3, 18-19 (1998). [5] B. A. Bolto, D. R. Dixon, R. J. Eldridge, E. A. Swinton, D. E. Weis, Willis, D., H. A. J. Battaerd, P. H. Young, "The Use of Magnetic Polymers in Water Treatment" J. Polymer Sci. Symp. 49,215-225 (1975). [6] S. H. Shin, D. S. Kim, 35, 3040-47 (2001). "Studies on the Interfacial Characterization of O N Emulsion for the Optimization of Its Treatment" Environ. Sci. Technol. 35,3040-3047 (200 1). [7] A. Yehaskel, Activated Carbon Manufacture and Regeneration (Noyes Data Corporation, New Jersey, 1978). [8] J. Poerschmann, T. Gorecki, F.-D. Kopinke, "Sorption of Very Hydrophobic Organic Compounds onto Poly(dimethylsi1oxane) and Dissolved Humic Organic Matter" Environ. Sci. Technol. 34,3824-3830 (2000). [9] R. F. Soares, C. A. P. Leite, W. J. Botter, F. Galembeck, "Inorganic Particle Coating with Poly(dimethy1siloxane)" J. Appl. Polym. Sci. 60,200 1-2006 (1996).
22
Environmental Issues and Waste Management TechnologiesVIII
INVESTIGATION ON A RECYCLING PROCESS OF WASTE COLORED GLASS Danping Chen and Hirotsugu Masui Conversion and Control by Advanced Chemistry, PRESTO, JST, AIST Kansai, Special Division of Green Life Technology Ecoglass Research Group, 1-8-31 Midorigaoka, Ikeda, Osaka 563-8577, JAPAN
*Tomoko Akai Conversion and Control by Advanced Chemistry, PRESTO, JST and National Institute of Advanced Industrial Science and Technology, AIST Kansai, Special Division of Green Life Technology Ecoglass Research Group, 1-8-31 Midorigaoka, Ikeda, Osaka 563-8577, JAPAN
* *Tetsuo Yazawa National Institute of Advanced Industrial Science and Technology, AIST Kansai, Special Division of Green Life Technology Ecoglass Research Group, 1-8-31 Midorigaoka, Ikeda, Osaka 563-8577, JAPAN ABSTRACT A recycling process for waste colored glasses through phase-separation is newly proposed. The colored soda-lime-silicate glass was leached by an acid solution after being re-melted with B203.It was found that all cations except Si4+ in the soda-lirne-borosilicate glass were leached by the acid. The colored glass was successfully bleached, and highly pure silica powders were obtained using this method. INTRODUCTION A large amount of colored glass waste is produced in high consumption nations. Currently, only a portion of the colored glasses waste is used as raw material to be *Corresponding author **Present address: Himeiji Institute of Technology,2167 Shosha, Himeji, Hyogo, 671-2201,Japan
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
23
re-melted. Most of the remainder is non-recyclable, and is directly put into landfills, because colored glasses is considered impossible to decolorize [1,2]. The practice of putting this glass into landfills has give rise to environmental, social and economic problems, all of which have been increasing in recent years in many municipalities and countries, especially in Japan, the EU, and Taiwan [l-31. Because of national regulations which reinforce environmental protection measures, there is a strong need to utilize waste glasses. There have been proposals to recycle colored glass waste, including using it as part of the coarse aggregate in cement and concrete [3-61, and the extraction of SiO2 from the glass waste by alkali fusion [7]. Due to the reaction between the alkali in glass and the reactive silica in cement, the use of glass as part of the coarse aggregate in concrete does not work well, because of the strength regression and excessive expansion [5-61. Using an alkali fusion, it is possible to separate Si02 and other contents of the glass waste, but it produces a lot of alkali waste. Therefore, it has been necessary to research new methods for re-utilizing colored glasses waste. We noticed that the transition metal ions always concentrate in the B203-rich phase of the phase separated soda-borosilicate glass [8], as shown in Figurel. Furthermore, as such manufacturing process the Vycor glass, the B203-rich phase can be leached out by the hot acid solution, and the remaining highly porous structure can be sintered together to form an almost pure SiO2 glass. Then, the phase-separating property of the 1 glass may be used to extract the alkali 2 ions and decolorize the colored glass we will propose a Figure 1. Schematic representation of In this recycling process for waste colored phase separation in soda- borosilicate glass glasses based On the Property Of glass l.SiOn-ri& phase; 2, B 2 0 3 - ~ c hphase phase separation.
EXPERIMENTAL The base glass used in the present experiment consisted of two types. One type was colored glass waste, the exact compositions of which were unknown. This glass may have contained the elements of Na, Ca, Al, Si, CryFe, Cu, COand Ti [7]. The composition of the other glass used was 15.2Na20*10.2Ca0*73.2SiO21.3~203*0.1c~03(wt%) (the blue base glass) and 15.2Na20*10.2Ca0*73.2Si02* 1.3&03* CrzO3 (wt%) (the green base glass). The base green or blue glass and the H3B03 were mixed (base glass: B203=100:x,x= 15- 60) in a mortar. In order to treat the colored glass waste and obtain the blocks of glass samples, the waste glass and Si02, H3B03, and Na2C03 as well as Al(OH)3 were mixed in this
24
Environmental Issues and Waste Management Technologies VIII
proportion: glass waste:Si0~:B~0~:Na~0:Al~0~ = 100:50:150:6:6. Next, a 50g batch was melted in a platinum crucible at temperatures of 1400 "C for 4 hours, and the melt was then poured on to a graphite plate. Next, the glass obtained was annealed at 560-72OoC for various amounts of time, ranging from 10 to 80 h. The heat-treated and the non-heat-treated powder or blocks of glass were leached in 0.5- 3.ON HNO3for 24-48h. After the acid leaching, the solutions were filtered and the remaining glass was washed with deionized water. The content of Na in the leaching solutions was chemically analyzed by an atomic absorption spectrophotometer (AA-6800, Shimazu Co.). The contents of B, Ca, Cr, CO and AI in the leaching solutions were chemically analyzed by an inductively coupled plasma emission spectrometer (SPS7800,Seiko Instruments). The compositions of the leached glass were chemically analyzed based on JISR 3101 (the Japanese Industrial Standards for chemical analysis of soda-lime-magnesia -silica glass). The blocks of glass samples before and after acid leaching were polished, and the optical absorption spectra were measured in the 200 to 9OOnm wavelength range with a UV/VIS spectrophotometer(UV-240, Shimadzu Co.). To examine the structural changes with change of glass composition, we measured the "B NMR spectra. All "B NMR spectra were measured on a Chemagnetics CMX-200 spectrometer with v0=64.2504 MHz (B,=4.7 T), H3B03 (1M aq.) used as a frequency reference. The free induction decay was acquired using a single pulse of typically 1 ps with a repetition time of 3 s. The signal was typically accumulated 256 times. RESULTS AND DISCUSSION Table I lists the leaching rate of various elements from the blue base glass re-melted with different amounts of B203 (x). Because of the vaporization of some Na2O and B2O3 during the re-melting process, the leaching rates of Na and B were about 10% lower than their actual leaching rate. However, the leaching rates of Ca, CO and Al were a little higher than their actual leaching rates. The leaching rate of Si was about 1.0%, and did not change with an increased B2O3 content. This result shows that the leaching rate of all of the elements, except for Si, increased with an
B
CO
x
Na
Ca
15
48.6
44.7 47.4 21.9
25
88.6 101.7 82.6
35
88.0 104.7 86.6 105.1 31.5
66.1
Al 4.2 12.1
.
Table Leaching rate (%)of various elements of (green base glass:B,O, =lOO:x) glass after acid treatment x INa Ca B Cr Al
20
53.0
56.4 49.1
30
89.6 101.0 90.5
45
85.8
98.6
3.8
3.9
34.8 22.1
88.1 93.0 90.5
Environmental Issues and Waste Management Technologies VIII
25
Tablem. Analytical composition (wt%) of the (blue base glass:B,O, =100:30) glass after heat and acid treatment Treatmentconditions Si02 Na2O CaO A 1 2 0 3 COO B203 Heat treated at 640°C for 65h 98,0 o.05 0.02 1.7 0.01 0.19 and acid treated in 1N HNO,
increasing B2O3 content. Although this glass was not heat-treated, the elements of Nay Ca, CO and B dissolved in the heat acid as the x=35. Above x=35, the blue glass powder was decolorized to be colorlessness. This implies that the micro-phase separation exists in the quenched transparent glass. However, although the leaching rate of Al increased with the increasing B2O3contentYthe maximum leaching rate achieved was only 31.5%; this indicates that the site of the Al element in the glass is different from the site of Nay Ca and Co. We also analyzed the composition of the blue base glass after it was re-melted with B2O3 and leached, and the result is listed in Table U.The Si02 purity of the glass after the acid treatment reached to 98wt%, and the coloring ions of the cobalt were entirely eliminated. However, as mentioned above, the majority of A l 2 0 3 remained, due to the low amount of B2O3 added to the glass. Table II lists the leaching rates of various elements of the green base glass re-melted with different amounts of B2O3 (x). The leaching behavior of Nay Ca and B was the same as that of blue base glass. However, the leaching rate of Cr was similar to the Al element. Most of the Cr and Al could not be leached out, and the glass had some green or yellow color in it up to x= 45%. Since the leaching experiments showed that the Na, Ca, CO and B in soda-lime-borosilicateglass without heat treatment easily dissolved in the hot acid solution, we deduced that there is a micro-phase separation in the quenched glass, and the cations of NayCa and COare enriched in the surrounding B atoms. However, as for the cations of Cr and Al, they were not dissolved in the hot acid solution until x= 45wt%. This may mainly relate to a change in glass structure with the B2O3 content, because the ratio of B03/B04 in the soda-lime -borosilicate glass structure changed with the 400 200 0 -200-400 ratio of ( Na2O+CaO)/ B203. PPm Figure 2 displays the "B NMR spectra of Figure 2. "B NMR spectra of the the green base glass after the addition of B2O3 green base glass after the addition glass. A narrow sharp signal near 0 ppm of different quantities of B203 corresponds to the tetrahedral boron, BO4, and (green base glass:B203=100:x)
26
Environmental Issues and Waste Management Technologies VIII
two overlapping, broad, split signals between 5 and 20 ppm correspond to the trigonal boron, BO3. It can be seen that the peaks corresponding to the trigonal boron BO3 increase with the amount of B2O3. This suggests that the fraction of tetrahedral boron, BO4 in this glass decreaseswith the addition of B2O3. The phase separation of glass results from the selective bonding between various chemical bonds in the glass structure. In addition, the nature of the chemical bonds, such as the covalent bond or ionic bond, results in the selective bonding. It is known that the nature of the chemical bond in oxide is related to the field strength. The B3' cation in BO3 and B04 as well as Si4' cation in Si04, have different field strength [9]. The order of these field strengths is as follows: B3+(in BO+ Si4' (in Si04)>B3+(inB04) The COions may selectively connect with the chemical bond with a lower cation field strength of in the glasses, such as B-0in B04,Na-0 and Ca-0. Therefore, it may easily concentrate in the B203-rich phase, and easily dissolve in the hot acid solution with boron, as shown in Table I . However, the Cr ions in the glass behave differently. As shown in Table II, the Cr leaching rate of the green base glass re-melted with 30%&03 was 34.8%. When the glass was re-melted under a reducing atmosphere, the Cr leaching rate only decreased to 10%. This result reveals that the low valence of Cr ions is distributed in the SiO2-rich phase, and that the Cr ions dissolved in the hot acid solution may be the high valence of Cr. The Cr6' ion may be similar to the COion; it is distributed in the BzOs-rich phase. As for the A13' and C?' ions in the MO6(M=Al, Cr) octahedron, they may selectively connect with the chemical bond of Si-0, which has a higher cation
""
2 50
I
I
Y
c) Q)
2
M El
40
30
. r (
c 20 0
4
10
- 0 0
20
40
0
60
80
Time (hour) Figure 3. Leaching rate of aluminum in blue base glass with 30% B203 added after heat treatment at 650°C for different times.
L
0
20
40
60
80
T i m e (hour)
Figure4. Leaching rates of chromium aluminum in green base glass with 35% B203 added after heat treatment at 650 "Cfor different times.
Environmental Issues and Waste Management TechnologiesVIII
27
field strength than that of B-0 in BO4. Then the ions of A13' and C? mainly concentrate in the SiO2-rich phase. However, the fraction of the three-coordinated BO3, with a field strength slightly higher than that of the Si4' cation, increase with the addition of B2O3; some A13+and Cr3' ions in MO6 (M=Al, Cr) octahedron can bond with the chemical bond of B-0 in three-coordinated BO3.ThenYin the case of high B203 content, A13' and C? concentrate in the interface between the SiO2-rich phase and the B203-rich phase as well as the B203-rich phase. These results suggest that the transition metal ions are not always enriched in the B203-rich phase of the phase separated soda-borosilicate glass [8]. The distribution of the ions having multiple valence state in the phase separated glass varied with the glass composition. The discussion above is based on the hypothesis that there is a micro-phase separation in the quenched glass. When the micro-phase separation is developed by heat-treatment, the distribution of various atoms in the phase-separated glass appears to be altered. Figures 3 and 4 show the effect of heat treatment times on the leaching rate of A13' and Cr3' ions. The leaching experiment showed that the change in the leaching rates of the CO, Na and Ca ions, probably concentrated in the B203 phase, were slight. However, with regards to the A13' and C? ions, the leaching rate was greatly changed with different heat treatment times. The Al leaching rate achieved its maximum value when the blue base glass with 30% B2O3 added was heat-treated at 650 "Cfor 40h. Figure 4 shows a similar result for the A13+and Cr3' ions in the green base glass. When the green base glass re-melted with 35% B2O3 was heat-treated at 650°C for 20h, the Al and Cr leaching rates showed their maximum value. The size of the separated phase changed with the temperature and time of the heat treatment, and may influence the Al and Cr leaching rates, especially when the ions of A13' and Cr3' concentrated in the interface between the SiO2-rich phase and BzOs-rich phase. In addition, the glass structure and valence and site of the Cr may undergo some changes during heat treatment, which could also influence the leaching rate. Tomozawa suggested a change in the glass structure with heat treatment temperatures through an anal sis of the immiscibility controversy of borosilicate glass [lO].High- resolution 71B, 29Siand 27AlNMR revealed temperature dependent structural changes in borate, borosilicate and boroaluminate glass [ll].These changes involved the change in
Glass composition lOOWaste green bottlet45B203 *100Greenbase glasst45B203
28
Na 0.91 1.1 (86)
Ca 0.88 0.84 (104)
Cr 1.92 0.74 (98)
Fe 2.53
B 1.25 1.29 (84)
Al 8.3 7.2 (92)
Si 17.1 3.8 (0.9)
Environmental Issues and Waste Management Technologies VIII
the boron coordination numbers and the transformation of AlO6 or AlOs units into A104 units with the lowering of the cooling rate for transforming the liquids into glass. We also measured the NMR spectra of the glass before and after the heat treatment, and the results supported the above conclusion [12]. Since the transformation of AlO6 or AlOs units into A04 units requires a charge compensation [9], this transformation appears difficult in the SiO2-rich phase with low alkali and alkaline earth ions contents. The transformation of AlO6 or AlOs units into AlO4 units implies that the A13' ions leave the SiO2-rich phase and enter the interface. Consequently, the Al ions are easily leached out in the hot acid solution. The size of the separated phase may increases with the amount of time of the heat treatment when heat treatment is beyond 20h. The increasing of the size results in a decrease in the interface of phase separation. This is an important reason for changing the leaching rate when increasing the heat treatment time, as shown in Figure 4. As for the Cr ions, the Cr6+ ions may distribute in the B203-rich phase and the influence of the heat treatment upon the Cr6' ions is slight. However, the changes in the glass structure and the phase separation resulting from the heat treatment may also influence the Cr3' ions, which means that the relationship between the heat treatment time and the leaching rate for the A13+and Cr3' ions are very similar. We treated the actual green bottle glass waste based on the above method and experimental results. The green bottle glass waste was successfully bleached by the phase separation and acid treatment. The results are shown in Table IV. An 100 approximate composition of the green bottle glass can be known in comparison 8o with the leaching rate of the green base glass listed in Table IV. Therefore, the 55 green bottle waste was converted into , colorless, porous, almost pure Si02 glass. , The porous glass obtained after the phase separation and acid leaching treatments was subsequently sintered at llOO°C, and + 20 = became a compact non-porous glass 2.Qlass 2 after possessing properties similar to that of 200 300 400 500 600 700 800 900 silica glass. Figure 5 shows the optical Wavelength (nm) transmission curves of a block of colored glass waste with a thickness of >lmm (l), Figure S. transmission along With the glass after the Phase curves of waste mlored glass Separation and acid leaching treatments with before and after phase separation a lmm thickness (2). and acid leaching treatments.
Environmental Issues and Waste Management TechnologiesVIII
29
SUMMARY In conclusion, we have demonstrated a new method for recycling colored glass waste through glass phase-separation. Colored soda-lime-silicate glass was re-melted with B2O3, and the soda-lime-borosilicateglass was then heat-treated for phase separating. The non-heat-treated glass and the heat-treated glass were leached with a HNO3 acid solution at 90°C. It was found that all cations except Si4’ in the soda-lime-borosilicate glass could be leached by acid. The powder or blocks of colored glass were successfully bleached, and a silicate glass with a high Si02 purity was obtained. REFERENCES 1. LYasui, “Several Aspects of Glass Recycle”; New Glass, Vol. 16 [2] 9-14 (2001) 2. D. Workman, “Recycling and How to Get the Message Across”. 19th Int. Conger. On Glass. Edinburgh, Scotland, V01.2~12,(2001) 3. N. Su and J.S. Chen, “Engineering Properties of Asphalt Concrete Made with Recycled Glass”, Resources, Conservation and Recycling, Vol. 000 000400 (2002) 4. Y. Shao, T. Lefort, S. Moras and D. Rodriguez, “Studies on Concrete Containing Ground Waste glass”, Cement und Concrete Research , Vol. 30, 91-100 (2000) 5. C.D. Johnson, “ Waste Glass as Coarse Aggregate for Concrete”, J Testing and Evaluation, Vol. 2 [5] 344-350 (1974). 6. K. Asaga, K. Kanai, H.Kuga, S. Hirose and M. Daimon, “Hydration of Portland Cement in the Addition of Waste Bottle Glass Powder”, Inorganic Materials, V01.4~423-430(1997) 7. H. Mori, “ Extraction of Silicon Dioxide from Waste Glasses by Alkali Fusion”, Proc. 19th Int. Conger. On Glass. Edinburgh, Scotland, V01.2, 13-14, (2001) 8. W.Vogel, “Phase Separation in Glass”, J.Non-Cryst. Solid, Vol. 25, 172-215 (1977) 9. H. Scholze, “Glass: Nature, Structure, and Properties”, Springer-Verlag (1990) pp.108-09 and pp.135-138 10.M. Tomozawa, “A Source of the Immiscibility Controversy of Borate and Borosilicate Glass System”,JAm. Ceram.Soc., Vo1.82 [11 206-208 (1999) ll.S.Sen, Z.Xu and J.F. Stebbins, “Temperature Dependent Structural Changes in Borate, Borosilicate and Boroaluminate Liquids: High-resolution llB, *’Si and 27AlNMR’, J.Non-Cryst. Solid, Vol. 226,29-40 (1998) 12. D. Chen, H. Masui, T.Akai and T. Yazawa, to be submitted to J.Non-Cryst. Solid for publication
30
Environmental Issues and Waste Management Technologies VIII
USE OF MID-DELAWARE RIVER DREDGE SEDIMENT AS A RAW MATERIAL IN CERAMIC PROCESSING Kimberly Hill, R. A. Haber, Rutgers University ABSTRACT Traditionally, the millions of tons of sediment dredged from New Jersey's rivers and channels were disposed of in the ocean or at land-based sites. Due to the recent environmental and legislative regulations, ocean disposal is no longer favorable. In addition, capacities of land-based disposal sites are quickly receding. Hence, non-traditional applications and disposal methods must be found for the dredge sediment. This was the first work in which Delaware River dredge sediments were evaluated for ceramic applications. The results show how thesse materials could potentially be introduced into c e d c industries as a new low-cost raw material source. The objective of this work was to characterize dredge material from the mid-Delaware River and to formulate a commercially sound ceramic tile product containing the maximum possible amount of dredge material. A matrix of formulations was prepared with dredge material used alone and also with a low cost local New Jersey clay, other recycled materials and commercially beneficiated materials. The dredge material was a coarse, highly quartz material with little organic content or soluble anions. Characterization results showed that the dredge material needed to be combined with clay and an auxiliary flux to reach the desired water absorption and breaking strength for floor and wall tile applications. I S 0 standards categorize tile by the forming method and water absorption value. Strength for commercial-sized tile must also meet IS0 specifications. The formulation concluded to be most applicable for floor tile consisted of 40% dredge materials, 50% New Jersey clay, 10% limestone. Numerous compositions showed potential for use as wall tile.
BACKGROUND ON DREDGING
Dredging is the practice of excavating material from the bottom of a waterway-rivers, bays, channels and ports-to allow for the safe passage of large vessels. Dredging is necessary to the U. S. economy. One container ship holds the cargo equivalent of 6,000 tractor-trailers or 1500 rail cars.' Thus, ships provide an efficient means of transporting goods. Many waterways through To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
31
which ships travel are naturally only 20 feet deep. However, today’s megaships ride 40-50 feet below the water’s surface. Material must be removed for safe navigation. Traditionally, when material was removed from a waterway, it was placed in ocean disposal sites or in one of two land-based disposal sites. Due to recent legislative and environmental restrictions, disposal in ocean-based sites has become unfavorable. One land-based option was landfills; however, landfills have only limited space for dredge sediment. In addition, tip fees make this type of disposal quite costly. The second option was placing material in confined disposal facilities. These are land sites specifically dedicated to dredge disposal. In New Jersey alone, there are 76 confined disposal facilities (CDFS)? Many of these sites have reached maximum capacity and are no longer operational. Of those that remain open, many are quickly approaching maximum capacity. With a growing number of dredging projects, limited disposal space and increasing environmental regulations for safe material disposal, non-traditional uses and placement of dredge sediment are no longer novel concepts, but are now necessary actions. Non-traditional uses include construction materials, landfill capping, replacement fill and shore protection, to name a few? The objective of this work was to characterize mid-Delaware River dredge sediments with respect to inorganic chemistry, mineralogy, particle characteristics and fired properties. Results of materials characterization were then used to evaluate the applicability of the dredge material for ceramic processes. The goal was to maximize the quantity of dredge used in ceramic bodies, but with minimal beneficiation. Since this was unprecedented work, experimental compositions were designed to explore a wide range of options. The dredge material was used alone, as a blend with other locally available raw materials and as a blend with commercially available raw materials. A secondary objective of this work was to incorporate a locally available New Jersey clay. Experimental compositions for ceramic floor and wall tile were chosen based on properties comparisons with commercially available products. For all compositions, water absorption and breaking stren@h were measured. MID-DELAWARE RIVER DREDGE SEDIMENT By wet sieve and x-ray diffraction analyses, it was found that the midDelaware River dredge sediment was coarse-grained quartzose material. A minor presence of clay minerals was detected in the -200 mesh (74 pm)fiaction of the material, which contributed only 20% of the particle size distribution. The dredge also contained negligible soluble anions, measured by ion chromatography, and less than 1% total organic content, as measured by hydrogen peroxide digestion. The vitrification point of the material was found to be greater than 1500°C. This was determined by firing one-inch diameter dry pressed discs of
32
Environmental Issues and Waste Management Technologies VIII
the material at increasing temperatures fiom 1250°C to 1500"C, exhausting the limits of the kiln in use and the limit of temperatures of interest for using this material in manufacturing. The color of the resulting body was red at lower temperatures and brown at higher temperatures. With such a high vitrification point, it was concluded that employing this material alone in a ceramic body was not feasible. LOCAL, NEW JERSEY CLAY In the mid-l980s, during construction at the Burlington County Solid Waste Facilities Complex in Burlington County, NJ, more than four million tons of clay material were excavated and stockpiled. No commercial uses have been found for large tonnages of this clay; thus, the stockpiles remain onsite at the landfill. Since the New Jersey clay is located in the same county as the midDelaware River dredge material, it would be cost-effective to use these raw materials together in a ceramic body. Estimates from the county place the price of this clay to be $3-8 per ton. X-ray diffraction showed that quartz is the major phase in the local New Jersey clay. Minor phases include illite, smectite, mica and pyrite. Coarse glauconite particles ( > 1 5 0 p ) also appear as a minor phase. The clay contains 10,000 ppm to 16,000 ppm sulfur, depending on the size fraction of the material being analyzed, as determined by Leco carbon and sulfur analysis. Much of the s u l h content appears in mineral form, as pyrite (FeS2). Soluble sulphates measure 30-43 ppdg, according to ion chromatography tests. Due to the pyrite, the clay also contains appreciable amounts of Fe2O3. The decomposition of pyrite causes the material to bloat between 1150°C and 1200°C. With increasing temperatures, the clay fires to orangish red or to brown. The results from materials characterization aided logical development of formulations for ceramic tile, brick and lightweight aggregate applications. Given that the river dredge was a silicious material, it was necessary to combine it with other raw materials to produce a commercially viable product. The New Jersey clay served as one clay component to add to the raw materials matrix. The clay minerals include illite and montmorillonite. The appreciable s u l h and iron contents in the New Jersey clay must be considered when determining firing parameters. Due to the vesicular nature of the New Jersey clay, a white commercial ball clay was used as a replacement in selected compositions. In this case H.C.Spinks', C&C ball clay was chosen To further tailor the frring properties, an auxiliary flux was added to the matrix of raw materials. Both a local New Jersey limestone and Feldspar Corporation's F-4 feldspar were used.
Environmental Issues and Waste Management Technologies VIII
33
CERAMIC TILE Following characterization, the second phase of this project was to formulate compositions including the maximum possible amount of dredge. To evaluate the applicability of the experimental compositions for tile, water absorption was measured for bodies fired to 1150°C and 1200°C. These values were then compared to IS0 standards4for commercial ceramic tile to discriminate between bodies applicable for floor tile or wall tile and bodies not applicable for either. Breaking strength was measured by biaxial flexure. Tile is categorized by the forming method, extruded or pressed, and the water absorption values. Water absorption (E) categories are as follows: E<0.5% (Group Ia); 0.5%<E<3% (Group Ib); 3%<ES6% (Group IIa); 6%<ESlO% (Group I&); and D 1 0 % (Group 111). According to specifications reported by leading American tile manufacturers, water absorption values for floor tile range from 0.3% to 4% and breaking strength must meet a minimum force of 160 kg (350 lbs). Values for wall tile vary by application.
EXPERIMENTAL COMPOSITIONS Experimental compositions are plotted on the triaxial diagram shown in Figure 1. A solid dot represents compositions containing the New Jersey clay. The ring represents compositions containing the commercial ball clay. The auxiliary flux was limestone. Of the two-component formulations, only the 50-50 mix of dredge material and ball clay resulted in a body with a water absorption value below 17% at 1200°C. This body had the highest strength of the twocomponent bodies. Other two component compositions had water absorption values up to 22% and were broken with little force when fired to 1200°C. Six compositions were formulated with 10% limestone, to decrease the water absorption value and to impart added strength. The water absorption and breaking strength values for all compositions are reported in Table I. Any of these six bodies could be used for wall tile when fired to 1200°C. Those with the New Jersey clay were dark red to brown in color. Bodies with ball clay fired to pink or buffcolors. The water absorption values for formulations containing ball clay were consistently higher than for those containing the New Jersey clay. The New Jersey clay contained 10% Fe203 The iron oxide in combination with other alkalies in the clay, such as K20, Na20 and MgO, and the CaO from the limestone fluxed the body, reducing the eutectic temperature at which the body began to melt. The ball clay does not contain appreciable levels of iron or other alkalies; thus, densification will occur at higher temperatures. The 40% dredge-50% New Jersey clay-10% limestone composition produced a body with 2.3% water absorption at 1200°C. This body was the most promising for floor tile applications.
34
Environmental Issues and Waste Management Technologies VIII
Producing tile batched with the river dredge and the New Jersey clay can dramatically reduce raw material costs for tile manufacturers. Estimating raw materials for a typical commercial body to cost $30 per ton, the river dredge and the New Jersey clay to cost $5 per ton each, and the limestone to cost $15 per ton, raw materials for the 40% dredge-50% New Jersey clay-10% limestone would cost only $6 per ton. If ball clay costs $25 per ton, that cost increases to $16 per ton, which is still almost half the typical cost. In place of the limestone, waste glass could be used for the fluxing agent. Estimated cost for the glass is -$5 per ton due to the urgent need to find a use for the mixed color fraction of the crushed glass. This contains clear, amber and green glass. The $6 per ton body now becomes only $4, and the $16 per ton can be reduced to $14. CONCLUSIONS It was concluded that the mid-Delaware River dredge sediment could be used as an alternative raw material in ceramic processing, with the potential to significantly reduce raw material costs for manufacturers. To produce a commercially viable product, the dredge must be used in combination with other raw materials. In this work, the dredge was combined with a local New Jersey clay, a commercial ball clay and limestone, an auxiliary flux. It was shown herein that a composition of 40% dredge-50% New Jersey clay-10% limestone fired to 1200°C produced a floor tile product with a water absorption of 2.3%, applicable for floor tile. Bodies formulated with 40-80% river dredge, 10-40% clay, and 10% limestone could be used for wall tile. The clay component in these bodies may be the New Jersey clay or commercial ball clay. REFERENCES 1. www.nap.usace.anny.mil/dredge/dl.htm. 2. www.state.nj.us/transportation/maritime/DMMP .PDF. 3. U. S. Army Corps of Engineers, New York District. “Dredged Material Management Plan for the Port of New York and New Jersey, Draft of the Implementation Report,” September 1999. 4. IS0 Standard 13006:1998(E), “Ceramic tiles-Definitions, classification, characteristics and marking.”
Environmental Issues and Waste Management Technologies VIII
35
Limestone Base Formulations
\3/
50% River Dredge 50% clay
75% River Dredge 25% clay
River Dredge
Figure 1. Experimental compositions plotted on a triaxial diagram.
36
Environmental Issues and Waste Management Technologies VIII
Table I. Water absorption and breaking strength values for experimental compositions fired to 1200°C.
Environmental Issues and Waste Management Technologies VIII
37
This page intentionally left blank
SODIUM EXTRACTION FROM THE WASTE GLASS BY ACID LEACHING TO OBTAIN SILICA SOURCE FOR CONSTRUCTION MATERIALS
T.h i * , D. Chen,
Y. Yamamoto, T. Shirakami and K
Conversion and Control by Advanced Chemistry, PRESTO, Japan Science and Technology Corporation, Midorigaoka, Ikeda, Osaka, Japan
Urabe Department of Materials Chemistry, Faculty of Science and Technology, Ryukoku Univ. Seta, Otsu, Shiga, 5202194, Japan
K. Kuraoka, T.Yazawa' Special Division for Green Life Technology, National Institute of Advanced Industrial Technology and Science, Midorigaoka, Ikeda, Osaka, Japan
*corresponding author, also Special Division for Green Life Technology,AIST +
Present Address
Himeji Institute of Technology, University, Shosha, Shosha, Hheji, 6712001, Hyogo Japan.
Abstract For the purpose of utilizing waste glass as construction materials, i.e., replacement of sand in cement, we studied the exhraction of sodium fiom the soda-lime-silicate glass powders by acid solution. The soda-lime-glass having similar composition of bottle glass was ground to the size smaller than 6 3 p and leached in nitric acid. The fractions of leached Na and Ca were estimated by analyzing the concentration of those elements in leached solution by atomic
~
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
39
absorption spectroscopy and inductively coupled plasma spectroscopy (ICP). The leaching behavior largely depended on the particle size and treatment temperature. It is found that the elusion of Na can be explained by diffusion equation. The preliminary reaction test for this leached glass and cement is also presented.
Introduction The development for the recycling technology is highly in demand due to the recent trend of forbidding the landfill disposal [I 3. The nonrecyclable colored glass is now becoming a crucial problem in waste Management worldwide. A new technology to recycle colored glasses is being developed recently [2]. Because of the large amount of waste colored glasses (i.e. more than a few hundreds thousand tons per year in Japan), the recycling method should be such that can utilize a large amount of quantity. In this respect, one of the promising methods is to utilize them as construction materials. The usage as silica source in cement suits this purpose, and the properties of the cement mixed with soda-lime-silicate glasses have already been reported [3,4,5,6]. The use fly ash as a silica source is now practically available method [A, but the use of waste glass with high sodium concentration is essentially suffered fiom alkali-silica reaction [3]. Although it is reported that the alkali-silica reaction is not serious as one expects [4], it is impossible to say that the alkali-silica reaction never occurs unless the concentration of sodium is low enough to prevent the alkali silica reaction. It is a well-known fact in glass industry and science that sodium is easily leached compared with calcium fiom silica network when it is immersed to acid solution [8]. By utilizing this phenomena, waste glass can be converted calcium silicates which can be used not only as silica source in cement but also blocks and so on. In this paper, we report a leaching behavior of alkali in Na20-CaO-Si02 -A203 glasses. Two different size of glass powders were leached in acid, and the e x d o n rate of each ion were investigated by analyzing the concentration of ions in acid. The preliminary test of mixing of cement and leached glasses is
40
Environmental Issues and Waste Management Technologies VIII
presented.
Experimental The composition of glass powder used for the study is typical composition of bottle glass (1 5.3Na20-10.2Ca0-73.2SiO2-1.3Al203(wt%)). The starting materids of the glass were Na2C03, CaCO3, Si@ and Al(OH)3. The appropriate amount of mixture were ground and melted in alumina crucible. The composition of the glass analyzed is shown in Table 1. Table I The compositionof the &ss used for the study(wt%).
The
bulk
glass thus obtained NOmhztl Anal@
NazO 15.3 14.3
CaO 10.2 10.5
SiO2 73.2 73.8
A1203
1.3 1.4
was crushed and ground. To study particle size effect, two different size of glasses, 6 2 - 3 8 p (type I) and < 3 8 w (TypeII), were sieved by mesh. The particle size was analyzed by using a HORIBA particle analyzer LA-920. The shapes of the powders were also observed by scanning electron microscope (JEOL JSM 5200) with acceleration voltage of 25kV. The sample was loaded on carbon tape and coated with gold. The glass powder was leached in 3N nitric acid at 90°C, 120°C and 140°C for Ih-48h. The leaching rate of Na fiom the glass was estimated by analyzing the concentration of Na in acid solution by atomic absorption spectroscopy (Shimadzu AA-680). The leaching rate of Ca and Si is determined by analyzing their concentration in the solution by ICP (Seiko SPS 7800). For a preliminary test for the reactivity of leached glass with cement, concretes containing three different types of glasses, the silica glass, leached glass obtained in this study and unleached soda-lime-silicate glass were prepared. 60wto/o of each type of glass and 4owto/o of cement were thoroughly mixed. Water is then added to the mixture. The mixture was cast in a tube mold (diameter 2Omm). The obtained solid was cured at 60°C for 24hours in a thermostatic chamber, and then again cured in an autoclave at 120°C for 24 hours. The obtained concrete was crushed and ground into fine powder. XRD measurement was performed on RIGAKU RINT 2000. Thermal gravhetric measurement (TG)
Environmental Issues and Waste Management Technologies VIII
41
and differentialthermal analysis (DTA) were measured on RIGAKU TAS-200. Results and Discussion Figure 1 displays the SEM micrograph of glass samples used in the study. The ground glasses exhibited angular shape. The size of TYPE I glass has a relatively narrow size distribution. On the other hand, in TYPE II powder, small particles (
Fig. 1 Scanning electron micrograph of TYPE I (above) and TYPE I1 (below).
0.1
I
10
Fig. 2 Size distribution of glass powder TYPE I (left) and I1 bight).
42
Environmental Issues and Waste Management Technologies VIII
A typical example for the amount of eluted ion as a function of leaching time is presented in Fig. 3. Sodiumextraction rate is much larger than calcium and silicon due to the fast diffusion in Table II Average, mode 8od median size glasses. Figure 4 presents the leaching of the glass powder used for the study. behavior of sodium fiom TYPE I and Diameter / pm TYPE I TYPE II
TYPE I1 glass. The leaching behavior average mode median
53.4 48.0 47.4
Fig. 3 Amount of extracted Na, Ca and Si from the type11 glass powder leached in 3N HNOs at 140°C.
19.1 24.4 16.1
t'"
of sodium is greatly dependent on the leaching temperature as well as the particle size. The rate-determining step of Na leaching is the diffusion of sodium from the inside of the glass in the cases of fibers
or
porous
glasses
of
Fin. 4 Amount of extracted Na from the
it is very reasonable to assume that the TYPE I, (b) TYPE II alkali leaching fiom the soda-limesilicate glass powder follows diffusion equation. We have solved the diffusion equation from the spherical particle and found that leaching fraction (LF)of sodium fiom the glass can be expressed as follows [101,
Environmental Issues and Waste Management Technologies VIII
43
where D is diffusion coefficient of Na and d is the diameter of the particle. The
solid lines in Fig. 4 shows the result fitted by eq. (1) using least square method. The obtained parameters, a = 6 are , tabulated in Table III. d
To examine Table I11 The obtained value for a from the fitting the experimental data (Fig. 4)by eq. (1). whether alkali leaching affects the reaction with ~ e m ~ m - l l Din/d I lW5sIn K TPEI T P E I1 cement, we carried out a 9.71 15.2 363 preliminary reaction test 393 19.5 33.O 27.6 61.4 between leached glass with 413
44
Environmental Issues and Waste Management Technologies VIII
0 X
w 0 Q
2
w
-2
'lkmperature('C1 400,.
,
,
. , . ,
,
Temperature ("c) 6 X-ray -action pattern of concrete containing unleaced (top), leached (middle) and silica glass
Fig.
bottom).
Fig. 6 DTA and TG of concrete containing unleaced (top), leached (middle) and silica glass (bottom).
reaction, and therefore more than 90% of sodium should be leached out from the glass powder for the practical use. From the diffusion constant obtained in Table 111, we estimated the particle should be less than 5 p at the leaching temperature of 90°C [101. The study of the leaching behavior from finer glass particles and its alkali-silica reaction test is now in progress.
Environmental Issues and Waste Management Technologies VIII
45
Conclusion We have studied the de-alkali process by acid leaching at 90-140°C fiom the glass powder having different particle size. It is found that alkali leaching rate can be explained by diffiion controlled process. The leaching fkaction of Na achieved was as high as 64%(C38pm) when treated at 140°C. The reaction of the cement and leached glass thus obtained was examined by using XRD and DTA. It was shown that alkali-silica reaction seems to be restrained by the alkali leached. Referencw M. Pelino, “Recycling of zinc-hydrometallurgy waste in glass and glas ceramic materials.”, Waste Manage., 20,561 (2000). For example, Patents, JP2000-191353,H11-060324 (publicationNo.) 3 K Asaga, S. Itoh, A. Hiroshima,K. Wanibuchi, and M. Daimon,‘‘HydrotheIma.l reaction of Portland cement in the addition of waste bottle glass powder”, Inorganic Materials, 2,259,473 (1995) 9.Hirose, N. Ozawa,K. Asaga and M. Daimon, “Propertiesof hardened portland cement containing glass powder.”, JCA proceeding of cement & concrete, 50, 14 (1996). Y Shao, T. Lefort, S. Morsa and D. Rodrifuez, “Studies on concrete containing ground waste glass”, Cement. Con.Res., 30,91 (2000). K. Asaga, K. Kanai, H. Kuga, T. Hirose, M. Daimon, Inorganic Materials, 4,423 (1997). DDL Chung, “Review: Improving cement-based materials by using silica b e ” , J. Mater. Sci., 37,4,673 (2002). A. Paul, “Chemistry of Glasses” in Chapter 6, Chapman and Hall, London-New York (1990). I. R. Beattie, Trans.Farady. Soc. 49,1059 (1953). l0 T. Akai, D. Chen, T. Ymwa, H. Yammoto, T. Shirakami and K. Urabe, J. Am.Ceram. Soc., to be submitted.
’
46
Environmental Issues and Waste Management Technologies VIII
Emissions in Glass and Ceramic Industries
This page intentionally left blank
Analysis of Emissions from Nitrate Containing Glasses S . Luo and L. E. Jones School of Ceramic Engineering and Materials Science Alfkd University Alfied, New York, U.S.A Abstract
Nitrate glass can be used for waste remediation. Nitrate, which is added to glass batch as a fining agent and strong oxidizing agent, has been identified as a source of NOx. The decomposition mechanism of KN03 is studied in UHP Ar using a software package for the chemical thermodynamic calculation, FACT-Sage 5.0, Thermogravimetry (TG), Differential S d g Calorbetry (DSC), and Differential Thermal Analysis @TA) over a temperature range of 25-1300 O C . The TG/DSC/DTA system is coupled with Fourier Transform I&ared (F'TIR) for gas identification and quantification. Theoretical calculations and experiments achieved consistent results. A KNO3 phase change occurs at about 130 "C. KNo3 melts at about 330 "C before decomposition. Liquid K N 0 3 decomposition initiates at 600 "C. It decomposes into K202 (s), KO2 (s), evolving NO (8) and NO2 (g). The initial temperature for the decomposition of nitrate containing tubing glass batch in UHP Ar with the same flow rate is 550 "C. The NOx emission species are identified to be NO (g), NOz (g), and N20 (g). Nitrate decomposition in the tubing glass batch demonstrates higher mass loss rate and higher concentration of NOx emissions than that of pure nitrate decomposition. Introduction
Nitrates are deliberately added in glass batch as fining agents to lower the melting point of the primary constituents of the glass batch, and to decrease the viscosity of the glass melt which facilitates the encapsulation of waste. Nitrates are also strong oxidizing agents in the glass batch. However, even with all of these benefds, NOx emissions produced by &ate decomposition in glass batch are environmental problems. The United States has set strict standards to control NOx emissions', therefore, understanding the chemistry of NOx evolution is critical to the control of these emissions. There are four types of NOx emissions that are formed in glass fktories2: 1) Fuel NOx: The reaction of organically bonded nitrogen in the fuel with the oxygen coning fiom the air; 2) Prompt NOx: Hydrocarbon radicals in the fuel react with nitrogen molecules to form amines and cyano compounds that can react with oxygen; 3) Thermal NOx: The reaction of molecular oxygen and nitrogen at high temperature; and 4) Batch NOx: NOx emission evolved fiom the nitrate fining agent in the glass batch. The mechanisms associated with the formation of the first three types of NOx are related to the combustion process and have been widely studied3-6. However, the chemistry and evolution of the NOx fiom glass batch has not been studied in detail. The purpose of this study is to accurately and sensitively measure the NOx emissions evolved fiom nitrate containing glass batch, to examine the mechanism of nitrate decomposition in glass melts, and to evaluate the To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
49
influence of additional batch constituents on the mechanism of nitrate decomposition. The on-going study addresses the mechanism connected with the decomposition of a common fining agent, mo3. Theoretical Background The earliest investigations of nitrate chemical decomposition focused primarily on the fimction of nitrates as electrolytes in a high temperature battery or as heat transfer fluid in solar thermal-electric power systems. It is widely believed that the decomposition of sodium nitrate and potassium nitrate follows similar reaction paths7-15: Nitrate ( NO; ) decomposes into the nitrite ( NO; ) and evolves oxygen according to reaction (1). Here M = Na' or K ' :
The nitrite undergoes three possible decomposition steps: 2MN@(s) +M20 (s) + 2NO (g) + 1/202(g) 2MN02(~)-+ 2MO2 (s)+ N2 (g) 2MN02(~)+ M202 (s) + 2NO (g) In addition to this argument, nitrate is also believed to decompose directly into oxide, peroxide, or superoxide according to the following reactions15-16:
Both theories argue that the gas emissions h m nitrate decomposition are 0 2 (g), NO (g), and N2(g), and NO2 (g) is not a decomposition product in the proposed reactions. A small amount of NO2 (g) was detected in some earlier research. It was regarded as the result of the reaction of NO with 0 2 that existed in the gas analysis instrument.
Experimental
This work addresses the NOx emission issue through analyzing the complete sequential
decomposition of potassium nitrate using thermodynamic calculations and differential decomposition experiments. The thermodynamic calculation was performed using F*A*C*T-Sage5.0 (Facility for the Analysis of Chemical Thermodynamics) software pkage17which employs Gibbs fiee energy minimization as the calculation method. To characterize decomposition of pure mo3 and tubing glass batch containhg m03, a
50
Environmental Issues and Waste Management Technologies VIII
NETZSCH 409PC TG/DSC/DTA system was used. This system was coupled with a Fourier Transform Mared (FTIR) for gas identiEication and quantification. The experiments were done in 93.3 kPa ultra high purity (UHP) At at a flow rate of 100 cc/min and a constant heating rate of 20 Wmin. The KN03used in the experiment was 99.999%pure and fiom the Alfa Aesar (A h-hnscmMatthey Company, MA. USA). The sample weight for pure potassium nitrate decomposition experiment was 9.60 mg. The composition of the tubing glass batch is given in Table 118. The pure oxides of Si02, A l 2 0 3 , CaO, MgO, and KNO3 were used to prepare tubing glass batch. The sample weight of glass batch was 56.65 mg, with 17.58 mg KN03. The crucible material used during this reporting period was alumina. Table 1. Composition of a tubing glass batch in weight % of oxides
I I
Composition of oxides
weight %
SiOz
72.1
~
MgO K20
I
I
3.4 17.3
I I
Results and Discussion Figures 1 and 2 are the thermochemical prediction results for the decomposition products and the decomposition behavior of KNO3(s) in UHP Ar as a function of temperature, respectively. There is remarkable s h d a r i t y between the predicted chemistry and that observed experimentally. KN03 in Ar undergoes a crystal structure transition fiom K N 0 3 (s) to mo3 (s2) at 132 "C; this transition is predicted to occur at 130 "C. The transition ends at 143 "C. KNO3 in Ar melts at 335 "C and the melting phenomenon ends at 339 "C. The predicted solid to liquid phase transformation occurs at 337 "C. At 600 "C, the mo3 (s) begins to decompose, and there is a concurrent large endothermic peak. The TG mass loss curve in this region has several linear mass loss regions. Each of these regions is connected with a set of decomposition events. These events are identified in part by thermodynamic calculation and are supported by the gas analysis results. At 630 "C, K202 (s) is a predicted product fiom the decomposition of KNO3 (1). The maximum value of K202(s) remains stable through 1214 "C at which KzO(1) arises via K202(s) decomposition. There is also evidence for KO2 (s) at temperature range between 855 - 868 "C.
Environmental Issues and Waste Management Technologies VIII
51
1.2 1 .o
. 0.6 .
0.8
I
0.4
0.2
0.0 -0.2 0
/ I ..
855°C 1
.
200
.
.
1
.
.
400
.
1
.
.
.
1
.
600
.
.
1
800
868°C .
.
.
1
.
1000
1200
T("C)
Figurel. Equilibriumdecomposition products fiom lmol KNO3 in 93.3 kPa Ar calculated using FACT- Sage5.0
20.0 0.0 n
€R
-20.0
W
U) fn
2
-40.0
U) fn
-60.0 -80.0 -100.0 0
200
400
600
800
1000
1200
TCC)
Figure 2. TG and DSC of pure mo3 in 93.3 kPa UHP Ar at a flow rate of 100 cc/min and a heating rate of 20 Wmin
52
Environmental Issues and Waste Management Technologies VIII
Figure 3. FTLR spectra taken fiom pure KNO3 decomposition in 93.3 kPa UHP Ar flowing at 100 cclmin and a heating rate of 20 Wmin Figure 3 gives a series of FTIR spectra taken fiom the KN03 decomposition in Ar at 5OO0C, 600 "C, 800 "C, and 900 "C, respectively. At 500 "C, no NOx evolution occurs, and no mass loss is observed. However, at 600 "C, the emissions include NO (v2 1876 cm-I), and NO2 (v3 1617cm-'). The concentrations measured are 57.0 ppmV and 14.5 ppmV for NO and N02, respectively. At 800 OC, the gas analysis indicates that NO2 has completely disappeared. This tells us that NO2 evolved only at the onset of the decomposition between 600 "C and 800 "C. However, NO reaches its highest concentration of 684.8 ppmV at 800 "C. At the fiequency range of 1100-1500 an-',there appears to be a significant concentration of N20. Albeit, there is also evidence that this might be a spectrum of a small amount of evaporation of KN03l9. At 900 OC, NO concentration decreased but the peak height at fiequency range of 1100-1500 cm'*increased. Further work is on-going to evaluate this vibration mode. From Figure 3, we can see that NO2 and NO are directly evolved fiom KNo3 decomposition; therefore, the mass loss observed in Figure 2 can be attributed to the following reactions:
Environmental Issues and Waste Management Technologies VIII
53
The decomposition of a tubing glass batch in UHP Ar is given in Figure 4. KNo3 undergoes a crystal structure transition at 124 "C and melts at 327 "C. A moisture mass loss occurs at 390-450 "C and is the result of chemisorlxd water. The decomposition of the glass batch initiates at 550 "C, and is completed at 815 "C. Figure 5 gives a series of lF91R spectra taken fiom the decompositionof the tubing glass batch in Ar at 500 O C , 78OoC, 820 "C, and 950 "C, respectively. At 500 "C, no mass loss is observed, and no NOx is evolved. At 780 "C, the emissions include NO (v2 1876 cm-'), N& (v3 1617 cm-' ) and N20 (v3 2223 cm-'). The concentrations measured for NO (g) and NO2 (g) are 7354 ppmV and 312 ppmV, respectively. At 820 "C, the emission species are still NO (g), N02(g) and N20(g). The measured concentrationsof NO and NO2 decrease to 6379 ppmV and 177 ppmV, respectively. At 950 "C, no apparent NOx emission peaks are observed. The comparison between the decomposition of pure KNO3 and tubing glass batch is given in Table 2. The mass loss is higher for pure KNO3 decomposition, but the concentrationsof NOx emissions fiom pure KNO3 are much less thanthose evolved fiom
0.5
0.0
*
0.4
n
-5.0
0.3
U
*9
5
n
-In In In 0
0.2
3 0.1 8
2 -l0.O
0
- 15.0
-0.1
-20.0 0
-0.2
200
400
600
800
1000
1200
T ("C)
Figure 4. TG and DTA of a tubing glass batch in 93.3 kPa UHP Ar at a flow rate of 100 cc/min and a heating rate of 20 Wmin
54
Environmental Issues and Waste Management Technologies VIII
Figure 5.
FTIR spectrataken fkom the decomposition of the tubing glass batch in 93.3 kPa UHP Ar flowing at 100 cc/min
glass batch. In addition, the mass loss rate of the decomposition of tubing glass batch is higher than that of pure KNo3 decomposition. These data indicate that while NO (8) and NO2 (g) are the primary species evolved fkom glass batch, NO (g) and NO2 (g), as well as other gaseous species (for example, N2,02, and/or NzO) appear to be products fkom the decomposition of pure KNO3. The latter phenomena might explain, at least in part, the slower decomposition rate observed for pure KNo3. The reactions (10) and (11) also descrii the K N 0 3 decomposition bebavior in the glass batch. Table 2. Comparisonbetween the decomposition of pure KNo3 and tubing glass batch
Pure K N 0 3 I I
KN03 mass in sample (mg)
I
9.60
Concentration of NOx
NO
emissions (mg/ -03)
NO2
17.58
I
I
KN03 mass loss %
Tubing glass batch
Rate of decomposition at 100? mass 10SS ( I I l g / I l l g K N o 3 - S )
I
80
58
0.071
0.211
0.001
~
I
I
~-
~~
0.014 ~~
~
0.91x 10 -3
Environmental Issues and Waste Management Technologies VIII
~
1.4 3
~~~
~
10 -3
55
Reference
56
1.
"National Ambient Air Quality Standards" In Environmental Protection Agency. Accessed on: July, 2002. Available at
2.
R. G. C. Beerkens, "The Role of Gases in Glass Melting Processes," Glastech Ber Glass Sci Technol, 68 [12] 369-80 (1995).
3.
E. K. C. and. H. Dehne, et.al "Low- NOx Burner for Glass - Melting Furnaces The Hi- Rad Burner," Ceram. Eng. Sci. Proc., 14 [3-41 126-38 (1993).
4.
P. B. Eleazer and A. G. Slavejkov, "Clean Firing of Glass Furnaces Through the Use of Oxygen," Ceram. Eng. Sci. Proc., 15 [2] 157-74 (1994).
5.
S . Drogue, S. Breininger ,and R. RUiz, "Minimization of NOx Emissions with Improved Oxy-Fuel Combustion: Controlled Pulsated Combustion," Ceram. Eng. Sci. Proc., 15 [2] 147-58 (1994).
6.
D. Norman, "Emissions to air from the glass industry," Am. Ceram. Soc. Bull., 78 [7] 57-62 (1999).
7.
E. S. Freeman, "The Kinetics of the Thermal Decomposition of Sodium Nitrate and of the Reaction between Sodium Nitrate and Oxygen," Journal of Physical Chemistry, 60 1487-93 (1956).
8.
E. S. Freeman, "The Kinetics of the Thermal Decomposition of Potassium Nitrate and of the Reaction Between Potassium Nitrate and Oxygen," JournaZ of American Chemistry Society, 79 838-42 (1 957).
9.
G. D. Sirotkin, "Equilibrium in Melts of the Nitrates and Nitrites of Sodium and Potassium," Journal of Inorganic Chemistry, 4 [1 11 1 180-82 (1959).
10.
R. F. Bartholomew, "A Study of the Equilibrium KNO3(l)+KNO2(l)+1/2 02(g) over the Temperature Range 550-750 OC," Journal of Physical Chemistry, 70 3442-46 (1966).
11.
B. D. Bond and P. W. M. Jacobs, "The Thermal Decomposition of Sodium Nitrate," J Chem. Soc. (A), 9 1265-68 (1966).
12.
R. N. Kust and J. D. Burke, "Thermal Decomposition in Alkali Metal Nitrate Melts," Inorganis Nuclear Chemistry Letters, 6 333-35 (1970).
13.
Y. Hoshino, T. Utsunomiya ,and 0. Abe, "The Thermal Decompsition of Sodium Nitrate and the Effects of Several Oxides on the Decomposition," Bull. Chem. Soc. Jpn., 54 1385-91 (1980).
Environmental Issues and Waste Management Technologies VIII
14.
C. M.Kramer, 2.A. Munir ,and K. H. Stern, "Evaporationof NaN03, KN03, and NaN02,"High TemperatureScience, 16 257-67 (1983).
15.
C. M. Kramer, 2.A. Munir ,and J. V. Volponi, "SimultaneousDynamic Thermogravimetry and Mass Spectrometry of the Evaporation of Alkali Metal Nitrates and Nitrites,"J. Therm. Anal. ,27 401-08 (1983).
16.
H. R Bartos and J. L. Margrave, "Communicationsto the editor: The Thermal Decomposition ofNaN03,"Journal of Physical Chemistry, 60 256 (1956). 17. C. E. Bale, A. D. Pelton, W.T. Thoqson ,and e. d, "FACTSage 5.0" (2001) [Computer Software]. Accessed on: July, 2002. Available at ~www.factsage.com>or xwww. c r c t p o l y m t l c o ~
18.
A. K. Varshneya, Fundamentals of Inorganic Glasses; Ch. chapter1.Academic
19.
D. Smith, D. W.James ,and J. P. Devlin, "VibrationalSpectra of Moleculat Metal Nitrate Monomr and Dimem," J. Chem. Phys., 54 [1014437-42 (197 1).
Press, Boston, 1994.
Environmental Issues and Waste Management TechnologiesVIII
57
This page intentionally left blank
CHARACTERIZING PARTICULATE EMISSIONS USING MICROMETERSCALE X-RAYFLUORESCENCE J.F. Shackelford, P.B. Kelly, S.S. Cliff, M. Jimenez-Cruz, and T.A. Cahill Department of Chemical Engineering and Materials Science and DELTA Group University of California Davis, CA 95616 ABSTRACT Particulate emissions from industrial processes are a primary focus of regulatory standards. Special concern is focused on PM2.5, particulate emissions smaller than 2.5 micrometers in diameter. The use of micrometer-scale x-ray fluorescence (micro-XRF) is especially effective for characterizing the chemical composition of these particulate samples. Our micro-XRF system is beam line 10.3.1 at the Advanced Light Source of the Lawrence Berkeley National Laboratory. The samplers used to collect various size- and time-resolved particles are described. Recent samples from the ruins of the World Trade Center will be given as an example of ceramic particulates. The DELTA (Detection and Evaluation of Long-Range Transport of Aerosols) Group also uses a variety of complementary characterizationtools in addition to micro-XRF. INTRODUCTION Particulate emissions from ceramic industry are central to a variety of environmental, health, and regulatory issues. An analogous and tragic example of ceramic particulates with potential health consequences was the emission of aerosols in and around the site of the World Trade Center (WTC), Ground Zero. We report the evaluation of such emissions during the initial cleanup and recovery
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
59
period October - December 2001. Although there is significant concern regarding the health effects of the smoke and dust resulting from the collapse of the World Trade Center, characterization of the aerosol for exposure assessment remains a difficult task. Determination of the composition of the aerosol as a function of size is critical in providing data relevant to health effects. While particulate matter greater than 10 microns tends to be trapped in the upper airways, smaller particles will penetrate deep into the lungs and is cleared from the body through the cardio-vascular system. Respiratory toxicology analysis based on the observed chemical composition is needed to understand which components of the aerosol drive the pulmonary response and require treatment. The DELTA (Detection and Evaluation of Long-Range Transport of Aerosols) Group obtained additional experience in working with ambient samples during the NSF ACE-Asia experiment (Spring 2001) in which an extensive of array of ground-based aerosol samplers were deployed in Japan, Korea, China, and Taiwan. DELTA Group utilizes a suite of innovative, highly sensitive, state-of-the-art analytical techniques coupled to size-resolved aerosol samples collected with a multistage, rotating drum impactor. The sampler physically separates particles into eight size ranges based upon their aerodynamic diameters (i.e., 10-5.0, 5.02.5, 2.5-1.1, 1.1-0.75, 0.75-0.56, 0.56-0.34, 0.34-0.24, and 0.24-0.09 micrometers). The sampler ran at a constant flow rate of 10.0 liters per minute, allowing conversion of the mass per unit time measurement to be converted to mass per cubic meter. The collection drums rotated at a rate of 4 mm per day, providing a time record for each size fraction as a function of position on the mylar substrate (1,2). Thus analysis of each collection strip provides timeresolved data for each size fraction of aerosol for the ten week period using two sets of aerosol collection strips. As far as we know, these samples are the only highly size-segregated samples collected near Ground Zero between Oct. 2 and mid-December, and are thus a unique archive of physical and chemical information. The time resolved nature of the sampling is critical in determination of which aerosols originated from Ground Zero and which constitute baseline
60
Environmental Issues and Waste Management Technologies VIII
exposure in New York City. Correlation of collection time with wind speed and direction allow separation of WTC derived aerosol from the ambient exposure. The suite of techniques that are used by DELTA Group to analyze such samples include: 1) Scanning Transmission Ion Microscopy (STIM) and Differential Beta Attenuation Mass Monitor (i.e., Beta-Gauge) for total mass exposure determination 2) Synchrotron X-ray Fluorescence (micro-XRF), (quantitative Na through U) for elemental exposure determination 3) Proton Elastic Scattering Analysis (PESA), organic mass determination 4) Laser Desorption Ionization Time-of-Flight Mass Spectrometry (LDITOFMS), for chemical fingerprint exposure determination 5 ) Scanning and transmission electron microscopy including a Field Emission Gun - Scanning Electron Microscope (FEG-SEM) with EDS (energy-dispersive x-ray spectrometry) and the ability to do crystallography by back scattered electron diffraction (EBSD). The analysis of size resolved, time resolved aerosols collected during October to mid-December 200 1, in combination with respiratory toxicology provides a unique opportunity for exposure assessment through a multi-disciplinary team effort. PRELIMINARY RESULTS FROM THE WORLD TRADE CENTER (GROUND ZERO) The US DOE office in NYC requested deployment of a UCD aerosol sampler in September 2001 in response to concerns of air quality in the New York City area following the collapse of the World Trade Center. The UCD Program deployed aerosol sampling equipment on October 2, 2001 at the EPA facilities at the 201 Varick Street, NYC location. Deployment and preliminary analysis has been done on a voluntary basis as a public service without funds from DOE or EPA. The samples were in two sets collected over a ten-week period during which aerosols were being generated from continuing fires and re-entrainment of the original dust plume as a consequence of clean-up efforts. The time and size
Environmental Issues and Waste Management Technologies VIII
61
resolved nature of the sampling method provides a unique opportunity for thorough analysis of the aerosol. Partial analysis of the collected aerosol is reported here. Two independent methods are used to determine mass loading of the aerosol. Scanning transmission ion microscopy was used to provide mass-vs.-time data that reflected the effects of rain, which reduces the total suspended aerosol, and wind, which directs the plume from Ground Zero toward and away from the sampling location. The sampler ran at a constant flow rate of 10.0 liters per minute, allowing conversion of the mass per unit time measurement to be converted to mass per cubic meter. The first set of aerosol strips was scanned by STIM, leaving the second set to be analyzed. Validation of the STIM results by beta gage analysis is in progress. The differential beta attenuation mass monitor determines the total mass of the collected aerosol by measuring the transmission of beta particles through the collection strip. Time resolution of the beta-gage measurements will be 2-hour segments. Micrometer-Scale X-ray Fluorescence Synchrotron X-ray fluorescence provides quantitative determination of elemental composition of the collected aerosol for heavier elements. The range amenable to micro-XRF analysis is sodium through uranium. The analysis is performed at Beamline 10.3.1 of the Advanced Light Source at Lawrence Berkeley National Laboratory through the auspices of the Participating Research Team administered by the UC Davis Department of Applied Science. Time resolved data for the 0.24-0.1 micrometer size fraction showed significant silicon, sulfur, vanadium, chromium, and nickel. Elemental mass exposure yields important data for the consideration of toxic metals, but chemical analysis provides data on the nonmetals in the aerosol. Summation of the elements observed by micro-XRF yields total mineral mass for the aerosol. The advantage that micro-XRF brings to these studies is in three factors, unique to the UCD team at LBNL. Preliminary results on the WTC samples demonstrate the sensitivity of the system for elemental analysis. First, the beam is both white
62
Environmental Issues and Waste Management Technologies VIII
(i.e. continuum between 4-20 KeV) and 100% polarized, greatly reducing background present in standard tube source XRF (and even more so in electron beam excitation such as with SEMs and electron-microprobes). This was the aspect of micro-XRF we exploited in previous Kuwaiti smoke studies (3,4). Second, the beam is highly intense, a function of the ALS’s high radiance, allowing high count rates. Third, it can be focused to beam spots on the order of 2 micrometers, with no loss in trace sensitivity. This factor, when combined with our continuously rotating DRUM samplers, allows for very short time resolution (90 min, corresponding to a 250 pm spot), on substrates 168 mm long (representing 6 week samples). An example of a recent application of all these advantages is illustrated in our analysis of an Asian pollution episode at Cheeka peak, WA (5) in which both gross (soils, sulfates) and trace (zinc, copper, arsenic) species were recorded over a 60 hour impact as a function of particle size. Time resolved data for the 0.24-0.1 micrometer size fraction of the WTC aerosol collected on October 3 found significant silicon, sulfur, vanadium, chromium, and nickel. The sampling site was in direct line with the plume from the WTC on October 3, thus the data reflect the composition of the WTC aerosol on that date. Elemental mass exposure yields important data for the consideration of toxic metals, but chemical analysis provides data on the non-metals in the aerosol. Summation of the elements observed by micro-XRF yields total mineral mass for the aerosol. Figure 1 illustrates how the “mineral” content (representing the Si peak) of the WTC plum drops off with decreasing particle size while the “sulfates” (representing the S peak) rise. This behavior is comparable to a “typical” aerosol from Davis, CA except that the sulfate signal in the WTC sample remains high into the “ultrafhe” range. Complementary CharacterizationTechniques Laser desorption ionization time of flight mass spectrometry (LDITOF-MS) is a powerful method for characterization of sulphates, nitrates, and organic aerosols collected impactor strips. LDITOF-MS has been previously used to examine aerosol samples (6). The analysis for the early hours of October 3, 2001 on the
Environmental Issues and Waste Management Technologies VIII
63
0.24-0.1 micrometer size fraction aerosol showed that the sulfur observed by the micro-XRF is in the form of sulfate. The carbon is observed to be present in graphitic/aromatic form. The presence of PAH’s in the fine-particle fraction suggest high temperature combustion sources (7).
The preliminary SEM results were obtained from a conventional instrument, an IS1 DS-130 with an Oxford EDS system. Using energy-dispersive x-ray spectrometry (EDS) with the SEM, we found that relatively coarse particles (2.5 pm to 12 pm, aerodynamic size) illustrated in Figure 2 have predominantly ceramic or “mineral”-like (calcium alumino-silicate)compositions. As a practical matter, such compositions are routinely found in common atmospheric dust, but it would also correspond to the composition of the massive amount of structural concrete that was pulverized in the WTC disaster. The large concentration of aerosols identified by the STIM measurements would suggest that a significant component of the “mineral” composition was due to the pulverized concrete. Consistent with the micro-XRF results, the mineral-like signal largely drops off for the relatively fine particles (0.09 pm to 0.34 pm, aerodynamic size). In this range, the predominant chemical signal using EDS was sulfur. We could not identify oxygen or hydrogen as their atomic numbers were below the light element limit of our EDS system. Nonetheless, our data are consistent with the mass spectrometry results of LDI-TOFMS that specifically identified sulfates (SO; and HSO;) in the aerosols. In addition to chemical analysis of individual particles, the use of the SEM provides, of course, visual images of the morphology of the aerosol particles. These images also provide validation of the aerodynamic sizing of the collected aerosols.
CONCLUSIONS
The camhinatian of the analytical results for each of the applied tnefhods provides an accurate determination of the suspended aerosol plume in NYC during October to December 2001. The synthesis of the data for sulfur and silicon (the primary mineral component) has been performed for the aerosol segment corresponding to October 3, 2001. The overall picture indicates that the coarse material was
64
Environmental Issues and Waste Management Technologies VIII
1
World Trade Cents Particle Plume
I
Typical Partide Size Distribution
~bw3.2001
Figure 1. Mineral content (from Si peak) and sulphates (from S peak) for World Trade Center plume and typical aerosol from Davis, California.
Figure 2. SEM image of typical, ceramic-like coarse-range aerosol particles (5 pm to 12 pm) from the World Trade Center plume.
Environmental Issues and Waste Management Technologies VIII
65
primarily mineral in nature, while the ultrafine particulate was dominated by sulfate. There is also a smaller, but significant component of the ultrafine particulate that has a signature indicative of fumed silica. Initial group efforts have concentrated on the October 3 segment of the sample that corresponded to the WTC plume. The vast majority of the sample, however, is yet to be fully examined and is the focus of continuing effort. Finally, we submit that these analytical techniques, especially micro-XRF, have substantial benefit for characterizing particulate emissions from various sources in the ceramic and glass industries. REFERENCES
1) O.G. Raabe, D.A. Braaten, R.L. Axelbaum, S.V. Teague and T.A. Cahill. “Calibration Studies of the DRUM Impactor,” Journal of Aerosol Science, 19 [2]183-95 (1988). 2) T.A. Cahill and P. Wakabayashi, “Compositional Analysis of Size-segregated Aerosol Samples”; pp. 21 1-228 in Measurement Challenges in Atmospheric Chemistry, Edited by Leonard New man, Chapter 7, American Chemical Society, Washington, D.C., 1993 3) T.A. Cahill, Kent Wiikinson, and Russ Schnell. “Composition Analyses of Size-resolved Aerosol Samples Taken from Aircraft Downwind of Kuwait, Spring, 1991,” Journal of Geophysical Research, 97 [D13] 14513-20 (1992). 4) J.S. Reid, T.A. Cahill, and M. R. Dunlap, “ Geometric/aerodynamic equivalent Diameter Ratios of Ash Aggregate Aerosols Collected in Burning Kuwaiti Well Fields,” Atmospheric Environment, 28 [131 2227-34 (1994). 5) T.A. Cahill and K.D. Perry, Asian Anthropogenic Influence at Cheeka Peak,” (abstract) AGU Annual Meeting, Washington, D.C., Dec. 1998. 6) P.J. Silva and K.A. Rather, “Online Characterization of Individual Particles from Automobile Emissions,” Environ. Sci. Technol., 31 3074-80 (1997). 7) D.Z. Bezabeh, A.D. Jones, L.L.Ashbaugh, and P.B. Kelly, “Screening of Aerosol Filter Samples for PAHs and Nitro PAHs by Laser Desorption Ionization TOF Mass Spectrometry,” Aero. Sci. Technol., 30 288-99 (1999).
66
Environmental Issues and Waste Management Technologies VIII
DILATOMETRY AND MASS SPECTROMETRYSTUDY OF THE DECOMPOSITION AND SINTERING OF CALCIUM CARBONATE
Kai Feng and Stephen J. Lombardo Department of Chemical Engineering University of Missouri-Columbia Columbia, MO 65211,USA ABSTRACT A combined dilatometer and mass spectrometer apparatus has been used to examine the decomposition and sintering of calcium carbonate. The evolution of carbon dioxide fiom the decomposition reaction occurred fiom 500-1200°C for linear heating rates of 1-15"C/min. The decomposition reaction was correlated with an initial period of shrinkage of 2-4%. A second period of shrinkage of 24% occurred above 1lOO"C, which was attributed to the sintering of nascent calcium oxide produced by the decomposition reaction.
INTRODUCTION During the heating of particulate ceramic bodies to high temperature, a number of kinetic phenomena may occur which include sintering, decomposition or dissociation, volatilization of sintering aids, and removal of native oxide surface films. The latter three phenomena may lead to the appearance of species in the gas. Although a large number of studies in the past 30 years have focused on the characterization of species that are evolved fiom inorganic materials into the gas phase [ 1-51, very little information has appeared in the literature on the relationship between the reactions that occur at high temperature and the rate of sintering. Because both reactions and sintering are activated processes, simultaneous monitoring of reaction products and dimensional changes can allow for a comparison of the activation energies of the different processes. The detection with mass spectrometers [1-51 of gas-phase species evolved fkom inorganic materials has been widely reported and the use of dilatometers [6111 to characterize dimensional changes accompanying sintering is also well known. We have recently combined a dilatometer with a mass spectrometer [121 to study the reactions that occur at high temperature as dimensional changes
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
67
occur. As a test of the feasibility of using the apparatus, we examined as a model system the decomposition [13-151 and sintering [16-201 of calcium carbonate. This system is advantageous from a methods development standpoint because both decomposition and shrinkage occur over the same temperature range. In addition to the experimental advantages afforded by this system, calcium carbonate has technological importance as well. The use of limestone in flue-gas desulfurization is becoming increasingly more wide spread due to environmental concerns related to suliixr emissions. In addition, both calcium carbonate [21] and the decomposition product calcium oxide [18,221 are difficult to sinter, but carbon dioxide has been reported to have a catalytic effect on the densification of calcium oxide [22,23]. In this work, the decomposition and sintering of calcium carbonate are examined with a dilatometer and mass spectrometer system. With these two techniques, the evolution of carbon dioxide, the dimensional changes, and the bulk density can be determined as a function of temperature and time. These data can thus be used to provide insight into the relationships between some of the different kinetic phenomena. EXPERIMENTAL The experimental apparatus [12] consists of a gas handling system, The furnace/dilatometer assembly, and mass spectrometer (MS). fiunace/dilatometer (L70/2000, Linseis, Germany) can be heated to 2000°C and the temperature is measured by a W5%-Re26% (type C) thennocouple placed directly below the sample holder. A slip stream from the effluent of the furnace is introduced into a mass spectrometer (Hewlett Packard 5971). The method used to covert intensity data from the mass spectrometer into number of moles and partial pressures has been described elsewhere [121. Prior to each experiment, the f h a c e is evacuated at room temperature to 6.6 Pa (50 milliton) and then backfilled with Ar (purity299.99%) at room temperature. The fiunace is next evacuated and then heated to 2OOOC for 20 min to remove gases weakly adsorbed by the graphite elements and insulation. The argon gas is then introduced into the furnace at a constant flow rate of 16 literh and 1 atm total pressure. The mass spectra and shrinkage profiles were monitored as a fhction of temperature for different heating cycles. The powder used for the experiments is CaCO3 of purity >99.95% (Sigma-Aldrich, Milwaukee, WI). The sample for each run was 0.7 g of powder with 1% by weight binder 03005, Rohm and Hass,Philadelphia, PA), and the sample was pressed into a cylinder of 1 cm height by 0.6 cm diameter.
68
Environmental Issues and Waste Management Technologies VIII
RESULTS AND DISCUSSION The first results are for an experiment in which a specimen of CaCU3 was heated at P=5OC/min in flowing argon followed by a hold at 120OOC for 1 h. Figure l a shows the intensities of the species recorded by the MS as a function of temperature. From room temperature to 250°C, the baseline intmsities of all the species are flat, and no reactions are occurring. At 275"C, a small peak in the d . (C02) intensity is observed which corresponds to decomposition of the binder used to prepare the samples. As the temperature is M e r increased, CO2 fiom the decomposition of CaCO3 is evolved in large amounts over the range T=550-10OO0C. The maximum intensity, which occurs near P85OoC,conesponds to a partial pressure of 4.9 kPa (37 torr). The intensities of d ~ 1 2 16, , and 28 are seen to increase in a pattern signal and also to have maxima at 85OOC. These similar to that of the dz== latter masses appear in relative concentrations consistent with the hgmentation factors for CO2 [12]. For temperatures above 900°C, the intensities of = ' d and 16 continue to decrease whereas the intensities of d ~ 2 and 8 12 exhibit second smaller maxima at T=950°C. The appearance of these two peaks is attributed to reaction of the decomposition product CO2 with the graphite in the furnace to form CO. In an earlier work [12], we have solved for the equilibrium concatrations for mixtures of C02, C, and CO and demonstrated semiqualitative agreement between the experimental results and the equilibrium calculations. The intensity of the d r =signal can be integrated to detemine the number of moles of CO2 produced by the decomposition reaction. Table I illustrates the level of agreement between the amount of CO2 observed in the gas phase and the measured and stoichiometric weight loss of the samples. In general, the weight loss calculated from the integrated intensity signals is within &10% of the gravimetric and stoichiometric amounts. Table 1 Dimensional change and weight loss as a function of heating rate, f3. The samples of calcium carbonate were heated to 1200°C and then held for 1 h. Total Weight loss of samples (g) and error (%) Heating rate, Dimensional From From From ("C/min) Change (%) Stoichiometry Gravimetry Intensity Error 1
3.91
0.298
0.299
0.328
9.7
5
3.98
0.293
0.294
0.3 13
6.5
10
4.41
0.270
0.270
0.242
-10.4
15
5.64
0.307
0.306
0.283
-7.5
Environmental Issues and Waste Management Technologies VIII
69
20000
1200
n
n
d
1000
16000 Q
800
12000
600
8000
400
Q
= t
a-
m
r c
u)
4000
200
D
0 0
0 200 400 600 800 1000 1200 1400
1.o 0 .o -1.o -2.0 -3.O -4.O -5.0
0
200 400 600 800 1000 1200 1400 Temperature (%)
Fig. 1 a) Mass spectrometer signals and b) dimensional change versus temperature when a sample of calcium carbonate is heated in flowing argon at 5"Clmi.n to 120OOC and then held for 1 h.
70
Environmental Issues and Waste Management Technologies VIII
The dimensional change of the sample as a function of temperature is displayed in Fig. lb. From room temperature to 600°C, the sample undergoes an expansion of approximately 0.5%. When this is converted to a coefficient of thermal expansion, a CTE value of 1 0 . 3 ~ 1 0 ~ results, / ~ C which is consistent with the tabulated value of 9x1O6/"C for CaCO3 [24]. At 6OO0C, shrinkage of the sample begins and two regions of shrinkage are observed. The first region is strongly correlated with the evolution of CO2 in Fig. l a fiom CaCO3 decomposition. The shrinkage continues up to 900°C but then slows as the sample is completely converted to CaO and the evolution of CO2 ceases. As the temperature is M e r raised f?om 900 to 1100°C, the sample expands by 0.2%. This amount of expansion, which corresponds to much less than the tabulated CTE for CaO [24], arises because under the conditions examined here, sintering and expansion are competing phenomena. Above 11OO°C, the sample shrinksby another 1% at the end of the 1 h hold at 1200°C. The decomposition kinetics of CaCO3 were next examined as a h c t i o n of the heating rate. The intensity of the main decomposition product, C02, which is displayed in Fig. 2a, is seen to appear at progressively higher temperatures with increasing heating rate, and the peak shapes all appear similar, thus suggesting that a common kinetic mechanism underlies the decomposition. The shrinkage of the CaCO3 samples as a function of heating rate is displayed in Fig. 2b. For all of the heating rates, the decomposition and initial period of shrinkage occur over the same temperature range. The first 3% of shrinkage for heating rates of 5, 10, and lS°C/min, however, are nearly superimposed, in contrast to the shrinkage oc&g at a heating rate of l"C/min. This superposition at the faster heating rates may arise because the diffiional transport underlying the microstructural rearrangement is rate limiting. The data in Fig. 2 and Table 1 indicate that the total shrinkage observed at 1200°C for 1 h ranges 24% beyond the shrinkage that accompanied the decomposition. The amount of shrinkage observed during the hold period at 120OOC is also proportional to the heating rate. Because at faster heating rates CO2 is evolved over a shorter period of time, the concentration of CO2 in the furnace is greater, and the larger shrinkage at the faster heating rates may arise be the catalytic effect of CO2 [22,23], and possibly CO, on the sintering of CaO. The amount of shrinkage in Fig. 2b accompanying the decomposition reaction is also seen to be proportional to the heating rate. The larger initial shrinkage at the fater heating rates may also possibly be attributed to a catalytic effect of CO2 on the sintering of CaCO3. During the heating of CaCO3 to high temperature, a number of processes are occurring which effect the instantaneous value of the bulk density of the sample. Weight loss fiom CO2 evolution is occurring, dimensional changes fiom
Environmental Issues and Waste Management Technologies VIII
71
25000
d
dII
$
20000
b
5 15000 v)
C
e! 10000 C
I
U
3
. I
5000
I
!i
0 500
0
700
900
1100
1300
Temperature eC) n
8 Q F Q
6
I
Q
c
0
1.o
0.0 -1.0
-2.0 -3.0
. I
-4.0 0
E
i3
-5.0 -6.0
500
700
900
1100
1300
Temperature (%) Fig. 2 a) Normalized intensity of carbon dioxide evolution and b) dimensional change as a function of temperature obtained for the decomposition of calcium carbonate at different heating rates.
72
Environmental Issues and Waste Management Technologies VIII
transformation between the calcite and rock salt crystallographic structures are taking place, porosity is being generated, and bulk-like sinteakg is occurring. To describe these combined effects, we define a two-phase bulk density as 1
where m denotes the mass of calcium carbonate (CC) and calcium oxide (CaO), and both can be obtained at any point in the heating cycle fiom integration of the intensity data from the mass spectrometer. The total volume of the sample, 6,is also time dependent and can be obtained fkom the known initial volume and iiom the shrinkage data. Analyzing the data by Eq. 1 thus provides insight into how the density and porosity vary over the entire heating cycle, and this is shown in Fig. 3 for all of the heating rates. The combination of the decomposition, shrinkage, and density data reported here thus s w e s to provide both qualitative and quantitative insight into the occurrence of the different kinetic phenomena. The analysis of these data by kinetics models and the activation energies for CaCO3 decomposition and shrinkage have been reported elsewhere [25].
2.0 t 1.9 1.8 -m E 1.7 0 3 1.6 E 1.5 2 1.4 U
m-
8
1.3
..
A
.
A
4
0
P
A
.
4
.. . A
A
A
1.2 1.I
1 .o 500 600 700 800 900 100011001200 Temperature ('C) Fig. 3 Density versus temperature obtained for the decomposition of calcium carbonate at different heating rates. The density was calculated by taking into account both weight loss and dimensional change of the sample.
Environmental Issues and Waste Management Technologies VIII
73
CONCLUSIONS The decomposition and sintering of CaCO3 have been measured as a function of heating rate by a mass spectrometer and dilatometer system. The decomposition occurs over a 400-7OO0C temperature range, which depends on the heating rate, whereas shrinkage of the sample occurred over two distinct temperature intervals. The first period of shrinkage was strongly correlated with the evolution of carbon dioxide from the decomposition reaction, whereas the second period of shrinkage corresponds to the sintering of CaO produced from the decomposition reaction.
Acknowledgement Partial support was obtained fkom the University of Missouri Research Board. References
1. J. L. Margrave, The Characterization of High-Temperatwe Vapors, John Wiley & Sons,New York,1967. 2. J. H. Hastie, Ed., Characterization of High Temperature Vapors and Gases, Vol. 1 NBS Special Publication561,U. S.Department of Commerce, 1979. 3. P. Rocabois, C. Chatillon, and C. Bernard, J. Am. Ceram. Soc., 79 (1996)1351. 4. S.-S. Lh, J. Am. C m SOC., 58 (1975)160. 5. H. Nanri, Sh. Ishida, N.Takechi, K.Watanabe, and M. Wakamatsu,J. Soc. Mat.Sci. Japan, 45 (1996)694. 6. C. Boberski, H. Bestgen, and R Hamminger, J. Eur.C m . Soc., 9 (1992)95. 7. J. J u g , Keramische Zeitschrift, 42 (1990)830. 8. 0.Abe and S.Kanzaki, J. Ceram. Soc.Japan, 97 (1989)187. 9. M. L. Mecartney, J. Mat. Sci. L e t , 6 (1987)370. 10. H.-J. Kleebe, W.Braue, and W.Luxem, J. Mat. Sci., 29 (1994)1265. 11. S.Siegel, U He~mann,and G. Putzky, Key Engineering Materials, 89-91(1994)237. 12. K.Feng and S. J. Lombardo, “High-Temperature Reaction Networks in Graphite Furnaces,” J. Mater. Sci, 37 (2002)2747. 13. C. H. Sattafield and F. Feakes, AIChE J., 5 (1959)115. 14. A. W.D.Hills,Chem. Eng. Sci., 23 (1968)297. 15. G.Narsimhan, Chem. Eng. Sci., 16 (1961)7. 16. C. R Milne, G. D. Silcox, S. W.Pershing, and D. A. Kirchgesser, Ind. Eng. Chem. Res., 29 (1990)139. 17. E. K.Powell and A. W.Searcy, Commum. Am. Ceram. Soc., 3 (1982)42. 18. A. B. Fuertes, D.Alvarez, F. Rubiera, J. J. Pis, and G. Marbh, Chem. Eng. Commua, 109 (1991)73. 19. R H. Borgwardt, AIChE J., 31 (1985)103. 20. R H.Borgwardt, Chem. Eng. Sci., 44 (1989)53. 21. F. Tbtard, D.Bemachehsollant, and E. Champion, J. T h m . Anal. Cal., 56 (1999)1461. 22. R H. Borgwardt, hd.Eng. Chem. Res., 28 (1989)493. 23. G.G. Knutsen, Thermodynamics and Kinetics of the Reaction: Calcium Carbonate = Calcium Oxide + Carbon Dioxide. Ph.D thesis. Lawrence Berkeley Lab., Univ. of California, Berkeley, CA, USA,1980. 24. Thennal Expansion of Nonmetallic Solids, Vol 13A. Purdue University Thennophysical Properties Research Center, Lafayette, IL. 25. K.Feng and S.J. Lombardo, to appear in J. of Ceramic Processing Research, (2002).
74
Environmental Issues and Waste Management Technologies VIII
LEAD FREE ELECTRONICS: CURRENT AND PENDING LEGISLATION Julie M. Schoenung Associate Professor Department of Civil and Environmental Engineering 844E Engineering Tower University of California, Irvine Irvine, CA 92697-2175 USA ABSTRACT Lead (Pb) has been the target of environmental regulations for many applications because of its severe effects on public health. Historically, lead has been phased-out as the material of choice as an additive for paint and gasoline. Its use in batteries has led to major recycling efforts. Now it is being targeted within various electronics applications, including cathode ray tubes (CRTs), solders and piezo/ferroelectrics. This paper reviews the extent of lead use in these applications and the legislation, both current and pending, that could restrict continued use. Initiatives at the state level, the federal level, and abroad will be reviewed. INTRODUCTION There has been a steady increase in the production of electronic equipment over the last 40 years, compounded with technological innovations. This has resulted in a substantial increase in waste electrical and electronic equipment (WEEE or e-waste), which represented 6 million tons in 1998 (approximately 4% of the municipal waste stream). This volume, which is expected to increase by 3% to 5% annually, could thus double in the next 12 to 15 years [13. As more than 90% of e-waste is landfilled or incinerated without pre-treatment, a large percentage of pollutants in the municipal waste stream can be traced to e-waste [l]. These include heavy metals, and especially lead, which are known to have adverse health effects. Chronic lead exposure has been linked to neurological, reproductive, renal, and hematological disorders [2]. Children are especially at risk, as early blood levels of lead can adversely affect their development.
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
75
ELECTRONIC PRODUCTS THAT CONTAIN LEAD (Pb) There are three general categories of electronic products that contain lead (Pb): (1) cathode ray tubes, (2) solders and packaging, and (3) functional oxide ceramics. Cathode ray tubes (CRTs) are used in televisions and computer monitors. The lead in these devices is contained within some of the glass components. Solders and device packaging contain lead in metallic form within the low melting, eutectic composition, lead-tin alloy. Functional oxide ceramic devices include capacitors, piezoelectrics and ferroelectrics, which contain lead within their ionically bonded, perovskite structures. Figure 1 is a schematic diagram of a CRT, showing the various components in the system. Lead is contained within the glass in two sections: the funnel section and the panel. It is also used in the low-melting solder frit that is required to connect and seal the two parts of a color CRT [3]. Although historically the lead-containing glass in the panel was needed to prevent the penetration of x-rays, alternative systems without leaded panels are the current standard. The lead within these glasses is tightly bonded, is contained within the glass matrix and is stable and immobile. It is in the form of lead oxide (PbO), with compositions in individual CRT glass components of approximately 29, 23, and 28 wt% in the stem, funnel and neck, respectively. In the frit used as a glass solder or sealant, 75 wt% PbO is typical. The panel glass composition depends on whether the CRT is being used for a color or a monochrome system. For monochrome, 1.7 wt% PbO is typical; for color, 2.2 wt% PbO is typical [4]. The total lead content in the glass components of a complete CRT display system is approximately 8 wt% for no-lead panel CRTs and 10 wt% for leaded panel CRTs
PI.
In electronics, lead-tin solders are primarily used for interconnecting and packaging electronic components and assemblies, such as on printed circuit (or wiring) boards. Approximately 10,900 tons of refined lead was used for soldering in 1998 [6]. This solder is metallic and typically has a composition of 38.1% Pb 61.9% Sn, and a melting point of 183 C. Possible substitutes to lead-based solders contain tin (Sn), silver (Ag), bismuth (Bi), copper (Cu), indium (In), antimony (Sb), zinc (Zn), gold (Au) andor germanium (Ge). Functional oxide ceramic devices are made from materials that exhibit special dielectric and ferroelectric properties. Although barium titanate-based materials are often used, especially for capacitor applications, other lead-based materials are also used. The lead-based oxides include a wide range of compositions forming the general categories of materials called PZTs and PLZTs (lead-zirconate-titanates and lead-lanthanum-zirconate-titanates), in which the lead is contained within the ionically bonded perovskite structure. Typical compositions for lead-containing oxides include Pb(Zr,Ti)03 for PZTs and (Pb0.88 LaO.12) (Zro.70 Tio.30) 0 3 for PLZTS "71.
76
Environmental Issues and Waste Management Technologies VIII
CURRENT AND PENDING LEGISLATION In the United States, government-sponsored programs targeted at reducing lead exposure have mostly focused on lead in paint. However, there is increasing recognition of the need to diverslfy the coverage of protective policies against lead poisoning. At the federal level, the U.S. EPA has recently changed the reporting criteria for lead under the Toxic Release Inventory (TRI) “public right to know” program for facilities that manufacture, process or otherwise use more than 220 kg (100 lbs) of lead annually (down from 5,000 kg) [8]. Current estimates indicate that this new rule could affect more than 9,000 industrial facilities, nationwide. The U.S. EPA is also evaluating their list of “persistent, bioaccumulative and toxic” chemicals (PBTs). Lead is not on the currently approved list but has been added to the pending list that is now under discussion [9]. The purpose of the PBT list is to “focus federal, state, industry, and public attention on actions that reduce the generation of these PBT chemicals in RCRA (Resource Conservation and Recovery Act) hazardous waste by 50 percent by 2005”. The U.S. EPA has also established an e-waste prevention campaign that is targeted at waste prevention, reuse and recycling [lO]. The State of California has a similar campaign that is focused on reducing and recycling packaging materials, as well as computers and components [ll]. Other states including Florida, Massachusetts, Minnesota, South Carolina, and Wisconsin, as well as industry organizations such as the Electronics Industry Alliance, have also initiated ewaste prevention campaigns [ 12,131, These campaigns have resulted in a collection of disparate approaches including curbside collection programs, imposition of recycling fees at time of purchase, and provision of funds for CRT recyclers and dismantlers. The states of Massachusetts and California have gone one step further by implementing legislation that bans the disposal of CRTs in municipal solid waste disposal facilities [ 12,141. These mandates force the collection and recycling of CRTs, allowing for optional disposal in costly hazardous waste landfills. To date, domestic legislation and voluntary initiatives do not currently exist for lead-based solders or lead-based oxide ceramics. The U.S. also does not have recycling legislation of any kind. In response to the diverse approaches being taken by the different states and the need to bring other stakeholders into the dialogue, a nation-wide initiative has been established. This initiative, called the National Electronics Product Stewardship Initiative (NEPSI), is being coordinated by the Center for Clean Products and Clean Technologies at the University of Tennessee [15]. The NEPSI group’s main goal for the dialogue is “the development of a system, which includes a mechanism to maximize the collection, reuse and recycling of used electronics, while considering appropriate incentives to design products that facilitate source reduction, reuse and recycling; reduce toxicity; and increase
Environmental Issues and Waste Management Technologies VIII
77
recycled content.” Stakeholders include federal, state and local governments, manufacturers, retailers, recyclers, and environmental groups. The proposed timeline for completing their final agreement, including the details on a financing model, a collection, reuse and recycling model, and a regulatory model, is September 2002. In contrast to the approach being taken in the U.S., which is focused on voluntary efforts, a number of other countries have already enacted legislation designed to minimize the environmental risks resulting from the disposal of ewaste, especially risks that result fiom the toxic heavy metals contained in this waste. Denmark was the first country to eliminate lead entirely fiom industrial activities. The Netherlands and Sweden followed suit with their own respective legislations. Although these three nations had a common goal, their approaches to the problem vary considerably in scope, driving force, and financing mechanism [16]. In an effort to harmonize these efforts, the European Union (EU) has adopted two proposed directives: (1) The Directive on Treatment of Waste from Electrical and Electronic Equipment (WEEE), and (2) the Directive on the Limitation of Hazardous Substances in Electrical and Electronic Equipment (ROHS). The WEEE Directive focuses on the collection, reuse and recycling of electronic waste, whereas the ROHS Directive focuses on the restriction and use of hazardous substances in electronic equipment [17]. With respect to lead use in electronics, the ROHS Directive is most relevant in that it mandates that by January 1, 2007 all new electrical and electronic equipment put on the market will not contain lead, and five other substances. This legislation has caught the attention of manufacturers throughout the world, as they do not wish to lose their market share in Europe. Even the Japanese, who have their own e-waste legislation, are concerned about the extensive ramifications of an explicit and complete ban on lead [18]. The intent of the ROHS Directive includes all applications of lead, including CRT glass, solders, and piezo/ferroelectrics. Japan is the other country that has been progressive about enacting waste management and recycling legislation. Five key pieces of legislation are listed below [ 16,171: 1. Waste Management and Public Cleansing Law (Waste Management Law), 1970; Amended in 1991, 1992, 1994. 2. The Law for the Promotion of Use of Recyclable Waste (Recycling Promotion Law), 1991. 3. Water Pollution Prevention Law, 1994. 4. Law for the Promotion of Separate Collection and Recycling of Containers and Packaging (Packaging Recycling Law), 1995. 5. Specified Home Appliance Recycling ( S H A R ) Law, 1998. The combined effect of these laws is to encourage reuse and recycling of waste and to control the amount of lead released into the environment. Japanese
78
Environmental Issues and Waste Management Technologies VIII
manufacturers, who have a history of being environmentally conscious even without legislation, have seen the lead-free initiative as a marketing opportunity and have already made major strides toward achieving lead-free products. LEAD IN THE ENVIRONMENT When looking at the potential for environmental impact from the use of lead in electronic devices, it is important to consider the entire life cycle of the product (see Figure 2). The stages in the life cycle include extraction, manufacturing, use, and disposal. Recycling after use is also a stage, with a feedback loop into the manufacturing stage. From each of these stages except the “use” stage, pollution results in the form of waste and emissions that contaminate the air, water and land. The use stage is not considered because lead and other heavy metals are not direct threats while an electronic device is in use. Most of the legislation and initiatives described above have focused on the recycling and disposal stages in the life cycle. These stages present the most obvious environmental threats. During recycling, if a lead-containing electronic device is disassembled or melted for component or material recovery, respectively, there is the potential for exposure to the lead’s toxic effects. During disposal, exposure risks also exist. Simple landfill procedures as practiced in the U.S. might not seem to produce a risk for lead contamination into the groundwater, after all the lead is tightly bonded in the glass and oxide ceramics and encased within the electronic devices in the form of solder. However, U.S. EPA leaching tests require that samples be pulverized prior to testing [3]. Under these circumstances, the lead is much more accessible. In other countries the lead is equally more accessible [161. In Europe it is common to incinerate municipal waste. Lead in the waste then accumulates in the sludge; it also volatilizes into the air. In Japan the concern is the dust created when waste is shredded prior to being disposed of in a landfill. Clearly these concerns have been a primary driving force in the development of existing legislation. Unfortunately, two of the other life cycle stages, namely extraction and manufacturing, have not been considered in developing much of this legislation. Although occupational health is clearly an issue, the extraction stage is of particular concern. Research has shown that if lead is banned and alternative solder materials are used even more severe environmental consequences could result [19]. For instance, there may be insufficient supply of some of the alternative metals or extensive concentration of the raw materials could be required to satisfy demand. Therefore, the mining of these alternatives could create severe environmental impacts on localized regions. Furthermore, the toxicity and leachability of the alternatives has not been comprehensively evaluated. Limited research indicates that bismuth is probably more toxic than lead, yet it was the third most popular alloying element of alternative solders
Environmental Issues and Waste Management Technologies VIII
79
selected by industry, government and professional organizations. Results from another study indicate that silver and antimony alloys would not pass EPA leaching criteria, yet these elements also played prominent roles in alternative alloy selections [20]. CONCLUSIONS Lead is widely used in electronic devices in order to produce unique materials characteristics, such as low melting points, x-ray absorption and dielectric properties. Applications include leaded-glass components in cathode ray tubes, lead-tin solders in device assemblies, and lead-based oxides for piezoelectrics, ferroelectrics and capacitors. Of these three general categories of applications, CRTs have been the most targeted in the U.S. because of their solid waste management implications and potential for groundwater contamination. Lead itself has been most targeted by the EU, where it has been effectively banned starting in 2007. This ban has created tremendous incentive for manufacturers to pursue the development of lead-free alternatives within the electronics sector. In Japan, lead-free has been capitalized upon as a marketing opportunity. ACKNOWLEDGEMENTS I would like to thank several colleagues and students who have assisted with this research: Anna Ku, An Tu Nguyen, Oladele Ogunseitan, Jean-Daniel Saphores, and Andrew Shapiro. I would also like to thank the following programs for financial support: AT&T Foundation’s Industrial Ecology Faculty Fellowship Program and UCI’s Executive Vice Chancellor Program to support multi-disciplinary research. REFERENCES 1. Commission of the European Communities, “Amended proposal for a Directive of the European Parliament and of the council on the restriction on the use of certain hazardous substances in electrical and electronic equipment.” 2000/0159, Brussels, 2001. 2. D.R. Juberg, “Lead and Human Health: An Update.’’ 2ndedition. Prepared for the American Council on Science and Health (ACSH), 2000. http ://ww w.acsh. org/publications/bookletd1ead-u pdate.html 3. T.G. Townsend, S. Mussen, Y-C. Jang, and I-H. Chung, “Characterization of Lead Leachability from Cathode Ray Tubes using the Toxicity Characteristic Leaching Procedure,” Florida Center for Solid and Hazardous Waste Management, Report 99-5, 1999. http://www .enveng.u fl.edu/homepp/townsend/research/CRT/CRTDec99.pdf
80
Environmental Issues and Waste Management Technologies VIII
4. J. H.Connelly and D.J. Lopata, “CRTs and TV Picture Tubes,” Engineered Materials Handbook, Vol. 4, Ceramics and Glasses, ASM Intl, 1991, p. 1040. 5. Electronic Industries Alliance, “Lead in Cathode Ray Tubes (CRTs) Information Sheet,” December 2001. http ://www.eiae.org/whatsnew/attachments/Lead-in-CRTs. pdf 6. G.R. Smith, “Lead.” US.Geological Survey, Minerals Yearbook, Metals and Minerals, 1: 44.1-44.24, 1998. 7. I. Burn, “Ceramic Capacitor Dielectrics,” Engineered Materials Handbook, Vol. 4, Ceramics and Glasses, ASM Intl, 1991, pp. 1112-1118. 8. United States Environmental Protection Agency, “TRI Lead Rule,” 2001. http://www.epa.gov/tri/tri_pb-rule. htm 9. United States Environmental Protection Agency, “Notice of Availability of Draft RCRA Waste Minimization PBT Chemical List,” 1998. http://www.epa.gov/epaoswer/hazwaste/minimize/chemlist/index. htm 10. United States Environmental Protection Agency, “Electronics: A New Opportunity for Waste Prevention, Reuse, and Recycling,” 2001. http ://ww w. epa.gov/epaoswer/osw/elec-fs. pdf 11. California Integrated Waste Management Board, “Electronic Equipment: Reducing Waste at the CIWMB,” 200 1. http://www.ciwmb.ca.gov/electronics/casestudies/CNirMB.htm 12. United States Environmental Protection Agency, “Extended Producer Responsibility: State Activities,” 2001. http://www.epa.gov/epr/elec-leg.htm 13. Electronic Industries AUiance, “EIA Announces Industry Pilot Project for Electronics Recycling,” 200 1. s.ch?press_id=6 http ://ww w. eia.org/communications/press~release/pres 14. California Department of Toxic Substances Control, “Managing Waste Cathode Ray Tubes,” Fact Sheet, August 2001, http://www .dtsc.ca.gov/docs/hwmp/docs/HWM-FS-CRT-EmergencyRegs.pdf 15. National Electronics Product Stewardship Initiative, http://www.nepsi.org. 16. N. Tojo, “Analysis of EPR Policies and Legislation through Comparative Study of Selected EPR Programmes for EEE,” M.S. Thesis, Lund University, Sweden, 1999. 17. P. Le Fevre, “EnvironmentalIssues in Power Electronics (Lead Free),” APEC 2002. http://www .ericsson.COd s ustainability/pd€‘Apec-2002( Final).pdf 18. Japan Business Council in Europe, “ W E E and ROS: Four Key Issues,” http ://www .j%ce.org/files/keyissues-wee-ro s.pdf. 19. B .R.Allenby, Design for the Environment: Implementing Industrial Ecology. Dissertation, 1992. 20. E.B. Smith I11 and L.K. Swanger, “Are Lead-free Solders Really EnvironmentallyFriendly?’ Sugace Mount Technology (March): 64-66, 1999.
Environmental Issues and Waste Management TechnologiesVIII
81
Figure 1. Parts of a Cathode Ray Tube (CRT) [3].
Figure 2. Schematic of the Life Cycle for Lead in Electronic Devices.
82
Environmental Issues and Waste Management Technologies VIn
FIRST DELISTING PETITION APPROVAL BY THE US EPA FOR A VITRIFIED MIXED WASTE
John B. Pickett, Carol M. Jantzen and Lynn C. Martin Westinghouse Savannah River Co. Aiken, SC 29808 ABSTRACT Delisting of a vitrified hazardous and radioactive (mixed) waste stored at the Savannah River Site (SRS) has been approved by Environmental Protection Agency (EPA), Region 4. The original waste was a wastewater treatment sludge fiom electroplating operations; hence it was a “listed” F-006 Resource Conservation and Recovery Act (RCRA) hazardous waste. The plating line waste also contained significant amounts of depleted uranium, making it a “mixed” waste. The final wasteform was generated by VitriQing the 670,000 gallons of stored waste sludge. The delisting petition for the final wasteform was based on bench-scale treatability studies, pilot-scale test results, and analyses of the fmal glass wasteform. This approval allows exclusion of the vitrified waste fkom RCRA regulations. This is the fust approval of a delisting petition for a mixed vitrified waste in the United States. To meet the EPA delisting treatment standards, the waste needed to be stabilized to control the leaching of hazardous constituents fkom the final wasteform. The Westinghouse Savannah River Co. (WSRC) contracted a vendor (GTS Duratek, Columbia, MD) to stabilize the mixed waste, using a temporary Vitrification Treatment Facility (VTF) at the SRS. The fmal vitrified waste is in the form of glass ovoids (gems), which are stored in 71-gallon dnuns. The fmal volume of stored waste is -210,000 gallons, for a volume reduction of approximately 69%. Volume reduction, combined with excluding the waste fiom hazardous waste management regulations, will significantly reduce overall disposal costs, resulting in cost savings of $7 to $15 million. INTRODUCTION The waste for which the delisting petition was requested was a mixed plating line sludge generated at the Savannah River Site, (SRS) near Aiken, South Carolina. The original plating line wastes were generated under the purview of Atomic Energy Act (AEA), and were managed and controlled by the US. To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
83
Department of Energy-Savannah River (DOE-SR). The operating contractor for the DOE at the SRS is the Westinghouse Savannah River Co. The nickel plating line sludge was a hazardous waste (F-006listed waste) and since it contained a significant amount of depleted uranium, it was a “mixed” waste as defined by the Federal Facilities Compliance Act of 1992. In order to comply with the RCRA (Resource Conservation and Recovery Act) laws, the waste had to be treated to meet the Land Disposal Restriction regulations (LDR’s). In addition, since the sludge had been stored in on-site tanks longer than allowed by the LDR’s, the sludge was subject to a 1991 Federal Facilities Compliance Agreement (FFCA) between the US Environmental Protection Agency (EPA), Region 4 and the DOE-Savannah River. The final wasteform needed to be delisted to allow disposal as a Low Level Radioactive Waste (LLRW), rather than as a mixed (radioactive and hazardous) waste. The SRS had no disposal facilities permitted for the disposal of mixed or hazardous wastes, and had no plans to construct any such facility at the Savannah River Site. Disposal of the delisted wasteform is planned for on-site disposal in a DOE managed LLRW disposal facility. The management of the DOE facility, both engineering and administrative controls, combined with the physical properties of the vitrified wasteform, will insure that no health or environmental risks will ensue fiom long term disposal. The final waste form generated from the vitrification process was packaged in Department of Transportation (DOT) approved containers. The treated waste was stored in a RCRA interim status container storage facility while awaiting verification that the analyses are below EPA delisting levels and the approval of the final delisting petition. The treated waste was shown to meet the LDR limits for treated wastes, therefore storage was no longer prohibited. Wastewater Plating line Sludge Generation The plating line sludge was generated in the Reactor Materials Department or 300 M-Area, (M-Area), which is located in the northeast section of the SRS. M-Area manufacturing operations produced various reactor components, which consisted of: aluminum housings; aluminum canned, depleted uranium targets; extruded enriched uranium-aluminum alloy fuel tubes; and lithium-aluminum alloy target tubes and control rods. The majority of the hazardous/radioactive (mixed) plating line sludge was produced during the production of the depleted uranium targets, which were then irradiated in the site’s nuclear material production reactors. Liquid effluents generated from the M-Area manufacturing operations consisted primarily of duminum forming and metal finishing residues, including nickel electroplating solutions and solutions fiom component recovery operations. Process waste effluents comprised rinse waters, acidic
84
Environmental Issues and Waste Management Technologies VIII
stack scrubber effluents, and spent process solutions from the three main production buildings and two analytical support laboratories. The process wastes were treated at the Liquid Effluent Treatment Facility (LETF), which was comprised of three separate wastewater treatment and storage facilities:
1. The Process Waste Interim TreatmenVStorage Facility (PWITISF) was an industrial wastewater tank storage facility. Storage in the tanks generated additional listed F-006 plating line sludge due to separation and settling of residues from the supernatant in the tanks. 2. The Chemical Transfer Facility (CTF),where concentrated spent plating line solutions were neutralized and then transferred to the PWIT/SF storage tanks. These neutralized sludges were a listed F006 mixed waste (wastewater treatment sludges from electroplating operations). 3. The Dilute Effluent Treatment Facility (DETF), which was an industrial wastewater treatment facility. The DETF treated all of the dilute plating line and laboratory effluents generated in M-Area. The DETF utilized the Best Available Technology Economically Achievable (BATEA) for the aluminum forming and metal finishing industry effluents. Treatment of area wastewaters by the DETF resulted in the generation of a listed F006 mixed waste stream (wastewater treatment sludges from electroplating operations). The solids precipitated from the plating line effluents were filtered and the subsequent filtercake was transferred to the PWIT/SF storage tanks, a t h e CTF. SRS Waste Streams Treated The wastes that were treated also included a number of other listed andor characteristically hazardous wastes generated at the Savannah River Site. The plating line wastes were blended and mixed in one of the PWITISF 500,000gallon storage tanks (Tank # 7). The blended plating sludge was then transferred to a 35,000 gallon feed preparation tank. The melter feed batches were mixed with glass forming materials and then vitrified. A number of additional SRS waste streams were identified which would be suitable for vitrification. These waste streams were described in the DOE-SR's Approved Site Treatment Plan, developed pursuant to the Federal Facility Compliance Act of 1992. The waste streams were added to the feed preparation tank,blended with the plating line sludge, and then vitrified. The two main M-Area waste streams were: SR-WOO4, Filtercake from supernate treatment in the DETF (F-006), and SR-WO37, M-Area wastewater sludge from 3 13-M plating operations (F-006).
Environmental Issues and Waste Management Technologies VIII
85
The additional streams were: SR-WOOS, Mark 15 filtercake, a material slightly enriched in U-235 (F-006) SR-WO29, Treatability samples from both vitrification and cementitious treatability studies (F-006) SR-WO31, An enriched uraniurdchromium solution from the site’s Naval Fuels Facility @-007) SR-WO38, Sump material fiom an M-Area process building, with trace amounts of Cr 0-007) SR-WO39, Residual nickel plating solution, with trace lead (D-008) SR-WO48, Soils from SRS spills and sampling programs (suspect characteristicallyhazardous) SR-WO54, Enriched uranium solutions, containing lead, from an M-Area laboratory (D-OOS), and SR-WO82, Soils fiom the excavation of the Chemicals, Metals, and Pesticides (CMP) Pits (F-027). Treatment of an F-027 Mixed Waste Approximately 90 cubic feet of soil contaminated with tritium and dioxind furans were treated in the VTF (waste stream # W082). This waste stream (two B-12 containers) was all that remained of a waste stream (-500 B-12 containers) that had been treated by a commercial hazardous waste incinerator (Aptus, KS). The commercial incinerator was not permitted for the radioactive soil, so there was no known treatment capacity. This waste soil was added directly to the VTF melter, which provided complete destruction (less than analytical destruction limits) of the dioxins and furans. The initial soil and fmal glass results are given in Table I. This waste stream was an F-027 mixed waste and the treated waste was an F-028 mixed waste (F-028 wastes are “residues resulting fiom the incineration or thermal treatment of soil contaminated with F-027 constituents”).
Table I. Soil Concentration and Final Glass Results for Dioxins and Furans LDR Limit Constituent Analytical Results mg/kg mg/kg Soil Final Glass Hexachlorodibenzo-p-dioxin 0.0058 <0.0003 1 0.001 Pentachlorodibenzo-p-dioxin 0.00 17 <0.00030 0.001 Tetrachlorodibenzo-p-dioxin 0.0024 <0.00014 0.001 Hexachlorodibenzohans <0.00006 <0.00028 0.001 Pentachlorodibenzof s 0.00046 <0.00027 0.00 1 Tetrachlorodibenzofs 0.00042
86
Environmental Issues and Waste Management Technologies VIII
This was the first commercial destruction of a dioxin contaminated waste soil in a joule-heated melter, and the first treatment of a dioxin mixed waste. The analytical results for the F-027constituents in the final wastefonn were all less than detection limit, and the EPA approved the request to exclude the fmd glass from hazardous waste disposal regulations as an F-028 waste. Vitrification The F006 mixed waste sludges were treated (stabilized) in a sub-contractor owned and operated Vendor Treatment Facility (VTF). The VTF was intended to be a temporary facility adjacent to the DETF and PWIT/SF tanks. The VTF was designed, built, and operated by GTS Duratek, of Columbia, MD - under a procurement sub-contract to WSRC. GTS Duratek constructed a nominal 5 tons per day melter to vitrify the mixed wastes. GTS Duratek used its own finding to construct the VTF, and was not reimbursed until the waste was treated. This was considered one of the most successll “privatization” efforts ever conducted in the DOE complex.’ Radioactive waste treatment started in October, 1996 and was completed February 24, 1999. The VTF has been deactivated and is shut down.
DELISTING The hazardous waste management regulations contain a listing of hazardous wastes fiom “non-specific sources” (40 CFR 261, subpart D). These wastes are termed “listed” wastes, and must be managed as hazardous wastes, unless they are excluded under 40 CFR 260.20 and 260.22. The EPA recognized that some listed wastes may not be hazardous due to differences in feed stocks or industrial processes. A person may seek to exclude a waste, at a particular generating facility, from the lists in 261 subpart D by petitioning for a regulatory amendment. “In order to be successfid the petitioner must demonstrate to the satisfaction of the Department that the waste produced by a particular generating facility does not meet any of the criteria under which the waste was listed as hazardous”. Upfiont Delisting Petition The EPA may grant an “Upfiont Delisting Petition” for wastes andor waste 2 residues that are yet to be generated, but will be generated in the future . The petition will be evaluated by the agency based on the available information (e. g., pilot-scale data) that demonstratesthat the petitioned waste will most likely meet the delisting criteria. If the “Upfiont Delisting Petition” is approved, the petitioner will be codident that the fmal waste would be non-hazardous, and he can proceed with his facility construction. The upfront exclusion would require testing from the full-scale system to veri@ that the system is operating as
Environmental Issues and Waste Management TechnologiesVIII
87
described in the petition. An upfront petition also provides the petitioner with the regulator’s concurrence on the constituents of concern and the final analytical levels that must be met. The full-scale sampling plan and sampling protocols are also included in the exclusion. An upfiont delisting exclusion provides the petitioner with a great deal of confidence with respect to the conditions that must be met to generate a non-hazardous waste. The Department of Energy - Savannah River (DOE-SR) and WSRC prepared and submitted an “Upfiont Delisting P e t i t i ~ n ” ~to the United States Environmental Protection Agency (USEPA) in September 1996. The exclusion for the yet to be generated vitrified wastes was based on: all of the untreated waste characteristics; process descriptions of the generation of the untreated waste; process description of the proposed wasteform treatment facility and process; laboratory scale treatability study results; and results fiom pilot-scale testing (vitrification) of actual M-Area waste sludges. The bench-top and pilot-vitrification studies demonstrated that the vitrified plating line wastes would meet the delisting criteria. The use of the vitrification process to stabilize hazardous wastes was based on a number of properties associated with the vitrification and the fmal product, i.e.; well developed process technologies based on the high level nuclear waste stabilization program (HLVIT); the powerfid solvating properties of glass melts, particularly borosilicate, and its ability to atomistically bond a wide range and large amounts of hazardous inorganic and radioactive components; the complete destruction of organic compounds at the high vitrification temperatures; a stable, homogeneous wasteform that is highly resistant to aqueous corrosion; and a high-density wasteform resulting in large volume reduction fiom the original wastes. The “Upfront Delisting Petition” included a compilation of the hazardous constituents of concern that would be present in the final wasteform. The constituents were selected based on those hazardous metallic constituents that were shown to be present above analytical detection limit in glasses from the pilot-scale testing (with actual M-Area plating line wastes). The constituents of concern included in the “Upfiont Delisting Petition were Ag, As, Ba,Be, Cd, Cr, Pb, Hg, Ni, and Se. No organic constituents of concern were included in the
88
Environmental Issues and Waste Management Technologies VIII
Upfiont Petition, nor the final petition, since none were detected. The petition also requested concurrence with the final delisting limits, which were based on the EPA’s Composite Model for Landfills (EPACML). In the past, the EPA has used the EPACML fate and transport model, modified for delisting, as one approach for determining the delisting levels for petitioned waste. See 56 FR 32993-33012, July 18, 1991, for details on the use of the EPACML, model to determine the concentrations of constituents in a waste that will not result in groundwater contamination. As noted above, treatment of the radioactive mixed waste sludge started in October 1996. As the ccUpfront Delisting Petition” was submitted only a few weeks before the start of treatment, the EPA did not have an opportunity to approve the “Upfiont Delisting Petition” prior to the start of treatment. However, the EPA, Region 4, did conduct a preliminary review, and used the EPACML calculation to determine that the constituents of concern identified in the Upfront Petition appeared to be justified, and that the DAF would be appropriate for the final volume of glass. In addition, WSRC requested the EPA to determine whether any radioactive constituents (i. e., depleted uranium) would be considered a constituent of concern. The EPA replied that “as long as the radioactive wastes are subject to regulation under the Atomic Energy Act, Region 4 will not consider the radioactive isotopes as constituents of concern in evaluating delisting petitions. Therefore, a radioactive mixed waste that is delisted, based on meeting delisting criteria for RCRA hazardous constituents of concern, may be disposed as a rad-only waste”. This guidance by the EPA was important to the treatment of the M-Area plating line sludges. A final waste glass with greater than 5-6 wt. % uranium could have a TCLP leachant greater than the DAF x the preliminary drinking water limit for uranium (i. e.; 100 x 0.020 mg/L = 2.0 mg/L TCLP leachant). The guidance by EPA allowed GTS Duratek to maximize the uranium waste loading, and therefore minimize the final glass volume. Final Delisting Petition The EPA, Region IV proposed to grant the delisting petition on March 15, 4 5 2002 . Final approval was granted August 21, 2002 . The following is a quote fiom the final approval. “The Environmental Protection Agency (EPA or Agency) today is granting a petition submitted by the United States Department of Energy Savannah River Operations O8ce (DOE-SR) to exclude (or “delist”) a certain hazardous waste fiom the lists of hazardous wastes under the Resource Conservation and Recovery Act (RCRA). DOE-SR generated the petitioned waste by treating wastes fiom various activities at the Savannah River Site (SRS). The petitioned waste meets the deJinitions of listed RCRA hazardous wastes F006 and F028. DOE-SR petitioned EPA to grant a onetime, generator-specijk delistingfor its F006 and F028 waste, because DOE-
Environmental Issues and Waste Management TechnologiesVIII
89
SR believes that its waste does not meet the criteria for which these types of wastes were listed. The waste is a radioactive mixed waste ( R W because it is both a RCRA hazardous waste and a radioactive waste. EPA reviewed all of the waste-specijk information provided by DOE-SR, performed calculations, and determined that the waste, which has a low level of radioactivity, could be disposed in a landfill for low-level radioactive waste without harming human health and the environment. Thepetition isfor a one-time delisting, because the petitioned waste has been generated, will be completely disposed of at one time, and will not be generated again. Today's final rule grant 's DOE-SR 's petition to delist its F006 and F028 waste. No public comments on the proposed rule were received Today%finalaction means that DOE-SR 's petitioned waste will no longer be classijied as F006 and F028, and will not be subject to regulation as a hazardous waste under Subtitle C of RCRA, provided that it is disposed in a low-level radioactive waste landfill, The waste will still be subject to the Atomic Energy Act and local, State, and Federal regulations for low-level radioactive solid wastes that are not RCRA hazardous wastes". In support of its petition, the delisting petition included: 1. descriptions ofthe waste streams that contributed to the petitioned waste, the areas where the contributing waste streams were generated, and the vitrification treatment process that generated the petitioned waste; 2. Material Safety Data Sheets (MSDSs) for all chemicals used in processes tha1 generated the waste streams from which the petitioned waste was derived and in the vitrification process that generated the petitioned waste; 3. the total volume of petitioned waste generated; 4. results of analysis of untreated waste and the petitioned waste for all constituents in Appendix VIII of 40 CFR part 261 or Appendix IX of part 264; 5. results of the analysis of leachate obtained by means of the Toxicity Characteristic Leaching Procedure ((TCLP), SW-846 Method 13 1 l), from the petitioned waste and historical results obtained by the Extraction Procedure Toxicity leaching method (EPTox), SW-846 Method 1 3 10); 6. results of the determinations for the hazardous characteristics of ignitability, corrosivity, and reactivity, in these wastes; and 7. results of the Multiple Extraction Procedure SW-846 Method 1320 on the glass. The EPA, Region 4, used the Delisting Risk Assessment S o h a r e ORAS), developed by EPA, Region 6, to evaluate this delisting petition. The DRAS uses a 1996 model, which is called the EPA Composite Model for Leachate Migration with Transformation Products (EPACMTP). The EPACMTP improves on the older EPACML model (simple DAF multiplier) in several
90
Environmental Issues and Waste Management Technologies VIII
ways. EPA used the DRAS to calculate delisting levels and to evaluate the impact of DOE-SR‘s petitioned waste on human health and the environment. Delisting levels are the maximum allowable concentrations for hazardous constituents in the waste, so that disposal in a landfill will not harm human health and the environment by contaminating groundwater, surface water, or air. The DR4S performs a risk assessment for petitioned wastes that are disposed of in the two waste management units of concern: surface impoundments for liquid wastes and landfills for non-liquid wastes. DOE-SR’s petitioned waste is solid, not liquid, and will be disposed in a landfill; therefore, only the application of DRAS to landfills was discussed in EPA’s proposed exclusion. DRAS calculates releases from solid-phase wastes in a landfill, with the following assumptions: (1) the wastes are disposed in a Subtitle D landfill and covered with a 2-foot-thick native soil layer; (2) the landfill is unlined or effectively unlined due to a liner that will eventually completely fail. The two parameters used to characterize landfills are (1) area and (2) depth (the thickness of the waste layer). Data to characterize landfills were obtained fiom a nationwide survey of industrial Subtitle D landfills. Parameters and assumptions used to estimate infiltration of leachate fiom a landfill are provided in the EPACMTP Background Document and User’s Guide, Office of Solid Waste, U. S. EPA, Washington, D.C., September 1996. The details of the DRAS calculations conducted by the EPA are given in Reference 4. The results of the DRAS calculations are compared to the EPACML model, to the RCRA characteristically hazardous limits, to the LDR Universal Treatment Standards (UTS), and the actual M-Area glass results in Table 11. The EPA selected the hazardous constituents of concern (CoC’s) based on the final confirmation analyses conducted on grab and composite samples of the vitrified M-Area waste. Two of the CoC’s proposed in the initial Upfront Delisting Petition were not detected in the fmal glass (Hg, and Se), and thus were not included in the final exclusion. No organics were detected in the final glass analyses and no organics were listed as constituents of concern. The final M-Area glass results were all below the delisting limits, as calculated by the DRAS methodology, or the characteristicallyhazardous TCLP limit. All of the calculated DRAS limits exceeded the DAF = 100 approach, except for nickel and arsenic. The delisting limits for Ni and As were significantly lower via the DRAS calculation (vs. the DAF model). Although the Universal Treatment Standards (UTS) were not applied to the M-Area glasses, those limits would have been the strictest, if applicable, for Ba, Cd, Cr, Pb, and Ag. Although the EPA did not apply the UTS, an individual state with delisting authority may chose to apply those limits, which in many instances would be the most restrictive.
Environmental Issues and Waste Management Technologies VIII
91
Table 11. Calculated Delisting Levels for the M-Area Constituents of Concern, vs. DAF, Characteristic and UTS Limits, and Actual M-Area Glass
*Since the calculated delisting level (by DRAS) for these constituents was higher than the characteristically hazardous limit, the latter (lower) limit applies. ** EPACML limit = 100x Drinking water MCL COST SAVINGS The cost savings afforded by the combination of vitrification and delisting were significant. The major cost saving was associated with the lower disposal costs due to the volume reduction for vitrification vs. cementitious stabilization. Cement wasteforms are almost always no less volume than the original waste, with volume increases of 2X, or greater, being common. The disposal costs for a cementitious or vitrified waste from the M-Area sludge were estimated based on the project cost to construct an on-site Hazardous Wastehiixed Waste (HWMW) disposal vault at the SRS (1991 estimate). When alternative treatment and disposal paths for the largest volume SRS waste streams (the M-Area sludges and the CIF blowdown) were defined in the early ‘go’s, the HWMW vault project was canceled. The cost estimates for the various alternatives are summarized below in Table 111. The cost estimates for the HWMW vault do not include operational, closure, or post closure costs, so they are conservatively low.
92
Environmental Issues and Waste Management Technologies VIII
Table 111. Cost Savings for Treated M-Area Waste Disposal Final Waste Form Type Cement
Glass
Waste Volume Treated (Gallons) Final Waste Volume (Gals)* Disposal Cost ($millions)** Disposal if Delisted ($millions)#
670,000 2 10,000 2.6 1.1
670,000 670,000 or 1,340,000 8.4 16.8
.
-
Cost Savings; Volume: Vitrification vs. Cementitious, $5.8 to 14.2 million Cost Saving, Delisted vs. Not Delisted, $1.5 million Total Cost Saving for M-Area Waste, $7.3 to 15.7 million *Cementitious stabilization usually results in at least a 2x volume increase. The frnal volume for the M-Area glass, in containers, was 210,000 gallons. ** Disposal cost for the treated waste was based on a cost of $8 million for construction of a Hazardous Wastemixed Waste disposal vault divided by its capacity, which gives -$12.50/gallon (~$100per cu. R.) #Disposal cost at SRS low level radioactive trench (-$40/cu. ft, -$5/gal) The total cost saving due to the combination of vitrification and delisting is at least $7 million, with a more probable cost saving of >$I 5 million. The delisting portion of the cost savings for the vitrified wasteform, and subsequent disposal as a low-level radioactive waste rather than a mixed waste, saved $1 to 1.5 million (included the petition preparation costs).
CONCLUSIONS The EPA, Region 4 granted the delisting of the vitrified M-Area plating line wastes, and other SRS wastes that were included in the treatment process, on August 2 1,2002. This is the first delisting petition approval for a vitrified mixed waste in the US. During the 45-day comment period allowed in the proposed delisting exclusion, the EPA received neither comments nor requests for public hearings. The authors strongly recommend that a petitioner for mixed waste prepare an “Upfiont Delisting Petition”, if at all feasible. Such a petition needs to include the delisting limits calculated by the DRAS technique, rather than the older EPACML method, as the delisting limits may be stricter for certain constituents. The petition should also clearly determine whether the regulators would use the UTS limits, since these are the strictest limits in some cases. It is imperative to know what delisting limits will apply when designing a waste treatment process and to maximize waste loading in the final wasteform.
Environmental Issues and Waste Management Technologies VIII
93
AUTHOR’S NOTE
This proposed approval by the EPA had been issued when this paper was
submitted for publication at the 2002 St. Louis meeting. Since the final approval was granted in August, prior to publication in the fmal Proceedings, the paper has been updated to reflect the final approval. ACKNOWLEDGMENT The proposed approval of the delisting petition was prepared by Dr. Judy Sophianopoulos, of the US Environmental Protection Agency, Region 4, of the South RCRA Enforcement and Compliance Section, Atlanta, GA 30303. Dr. Sophianopoulos conducted the risk modeling analysis using the DRAS software, and included the modeling results in the delisting approval. Dr. Sophianopoulos 4 prepared the notification of the proposed approval in the Federal Register 5 (March 15,2002), and also the final approval (August 21,2002). Although the delisting petition approval by Dr. Sophianopoulos was based on the information 6 provided in WSRC’s Delisting Petition , the petition would not have been successful Without Dr. Sophianopoulos’ expertise and diligence in preparing the approval. REFERENCES
1. J. B. PICKETT, S . W.NORFORD, J. C. MSJSALL, and D. G. BILLS, “Vitrification and Privatization Success,))Proceedings of the Environmental Issues and Waste Management Technologiesin the Ceramic and Nuclear Industries Vx St. Louis, MO, Mi, 2001, Ceramic Transactions, Vol. 119, page number 219. The American Ceramic Society, Westerville, OH 43018 (2001). 2. EPA (Delisting Section, Office of Solid Waste, U. S . Environmental Protection Agency), Petitions to Delist Hazardous Wastes,A Guidance Manual, Second Edition, EPM530-R-93-007 (March 1993). 3. J. B. PICKETT, “Upfint Delisting Petition for Vitrijied M-Area Plating Line Wastes”, WSRC-TR-96-0244, Westinghouse Savannah River Co., Aiken, SC 29808 (1996). 4. EPA (v. S . Environmental Protection Agency), Hmardous Waste Management System; Identification and Listing of Hazardous Waste: Proposed Exclusion; Proposed Rule, 67 Federal Register 11639, (March 15,2002). 5. EPA (U. S . Environmental Protection Agency), Hazardous Waste Management System; Identijication and Listing of Hazardous Waste: Proposed Exclusion; Final Rule, 67 Federal Register 54124, (August 21,2002). 6. J. B. PICKETT, “Delisting Petition for Vitrified M-Area Plating Line Wastes”, WSRC-TR-96-0244, Rev. 2, Westinghouse Savannah River Co., Aiken, SC 29808 (2000).
94
Environmental Issues and Waste Management Technologies VIII
CHARACTERIZATION OF DEFENSE NUCLEAR WASTE USING HAZARDOUS WASTE GUIDANCE. INSIGHTS ON THE PROCESS AT HANFORD Megan Lerchen Pacific Northwest National Laboratory 902 Battelle Boulevard P.O. Box 999 Richland, WA 99352 William Hamel U. S. Department of Energy, Office of River Protection 2440 Stevens Center PO BOX450, MSIN H6-60 Richland, WA 99352
Lori Huffman U. S. Department of Energy, Ofice of River Protection 2440 Stevens Center PO Box 450, MSIN H6-60 Richland, WA 99352 Karyn Wiemers DMJMH+N 3250 Port of Benton Blvd Richland, WA 99352
ABSTRACT Federal hazardous waste regulations were developed for management of industrial waste. These same regulations are now applicable for much of the nation’s defense nuclear wastes. At the U.S. Department of Energy’s Hanford Site in southeast Washington State, one of the nation’s largest inventories of nuclear waste remains in storage in large underground tanks. The waste’s regulatory designation and its composition and form constrain acceptable treatment and disposal options. Obtaining detailed knowledge of the tank waste composition presents many challenges. Early insights fi-om a performance-based approach to demonstrating achievable quality standards will be discussed in the context of environmental guidance, permitting, and compliance under the hazardous waste regulations. INTRODUCTION The U.S. Department of Energy (DOE) is required to store, treat, and dispose of high-level waste at DOE’S Hanford Site in southeast Washington. Quality data supporting the project’s regulatory and engineering needs must be available. Over the last few years, DOE has made significant progress in defming and putting into effect characterization requirements for currently stored Hanford To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
95
radioactive tank wastes to meet data needs for treatment and final disposition. This effort has relied on a cooperative, teaming approach between DOE, the regulators, and the implementing contractors in order to tailor the characterization effort to obtaining acceptable, usefid data. Hanford Tank Waste Hanford has 53 million gallons of high-level waste, containing 190 million curies of radioactivity, stored in 177 underground tanks. Accumulation of the waste began in 1944 with the inception of the H d o r d defense production mission as part of the Manhattan Project. Current operations consist of waste receipts from activities such as deactivation and decommissioning work, analytical and processing laboratories, ongoing tank waste management operations, and early efforts for tank closure demonstrations. The underground tanks are within ten miles of the Columbia River, the largest river in the Pacific Northwest. Many of the tanks are past their design life, and 67 of the older tanks are known or suspected to have leaked. In addition, the newer tanks are quickly nearing their capacity. The only permanent solution is to treat and immobilize the tank waste into an inert waste form.
Tank Waste Treatment It is planned that the dangerous waste and radioactive constituents in Hanford's high-level tank waste will be initially separated, through pretreatment if necessary, into lower and higher activity fractions followed by fmal treatment to make disposable waste forms. Pretreatment will partition constituents between low-activity and high-level fractions to meet disposal requirements and minimize product volume. All of the higher activity fiaction will be made into durable, disposable glass waste forms through vitrification at the future waste treatment plant. The low activity fraction will likewise be made into durable, disposable glass waste forms or, if applicable requirements are met, solidification by supplemental waste treatment. A significant challenge presented by tank waste is the overall uncertainty in the detailed characterization knowledge as opposed to the bulk waste constituents such as sodium, aluminosilicates, nitrate, and hydroxide. This adds to the difficulty in planning and designing for tank waste treatment facilities because in lieu of certain characterization knowledge, the project has been using bounding or other conservative estimates where needed. RCRA REGULATION OF TANK WASTE Because of past Hanford-specific practices and its location in Washington State, Resource Conservation and Recovery Act (RCRA) requirements and their applicability to Hanford waste differ in some important aspects from other DOE
96
Environmental Issues and Waste Management Technologies VIII
sites. Under the Washington State RCRA program, the tank waste is designated for multiple RCRA waste codes. Each of these codes drives a requirement to use particular treatment technologies andor meet particular numeric performance standards in order to meet the RCRA treatment requirements for disposal. Tank waste characterization data will be used to support upcoming petitions to the regulators for a new treatment variance and delisting. In addition, characterization data are being used to support RCRA permitting for tank waste treatment and may be used to meet other tank waste management data needs related to regulatory compliance. DATA QUALITY OBJECTIVES FOR REGULATED COMPOUNDS In 1998, DOE and the Washington State Department of Ecology (Ecology) agreed upon the jointly prepared Data Quality Objectives (DQO) document referred to as the Regulatory DQO.’ The outcome of the Regulatory DQO was a prioritized list of 173 compounds for analysis that was selected fiom an initial list of nearly 1000 regulated compounds. The selection process involved a systematic review of each compound by a team of tank waste chemistry experts, including representatives from DOE and The evaluation focused on the plausibility of the regulated compounds’ existence in the tank waste matrix and a prioritization based on relative toxicity (Figure l).5 Target U. S. Environmental Protection Agency (EPA) methods (EPA publication SW-846, Test Methods for Evaluating Solid Waste, PhysicalKhemical Methods, EPA 600/4-79-020)or other equivdent environmental methods were identified for characterizing each of these prioritized compounds.6
Figure 1. Logic diagram for analyte selection and prioritization process used in the Regulatory DQO
Environmental Issues and Waste Management Technologies VIII
97
This DQO is focused, in part, on land disposal restrictions and permitting for waste treatment. At the time the DQO was conducted, the permitting and design efforts for the waste treatment plant currently under construction were in their infancy. Therefore, the parties agreed that the DQO implementation would be conducted in a step-wise fashion. This was intended to allow for refining the characterization needs as data are collected and the waste treatment plant permitting and design also move forward. A timeline for Step 1 of the DQO and the baseline schedule for the Waste Treatment Plant are shown in Figure 2. The timeline shows completion of Step 1 in the near future, before commissioning of the waste treatment plant.
Figure 2. Data quality objectives implementation timeline. Note that subsequent to The American Ceramic Society presentation, the milestone for reporting was delayed to first quarter FY2003. RCRA CHARACTERIZATIONMETHODS VALIDATION Methods for nine groups of compounds were investigated for their applicability to the tank waste solid and liquid matrices. The nine groups were metals, anions and organic acids, mercury, ammonia, cyanide, volatiles, semivolatiles, polar volatiles, and polychlorinated biphenylsl pesticides. The status of these analyses as of April 2002 is shown in Table 1.
98
Environmental Issues and Waste Management Technologies VIII
I I
Table 1. Data quality objectives implementation status - April 2002' Analysis
I I
Preparative Method
I I
Analysis Method
I I
status
Liquids: SW-846 3005A Solids: SW-846 3050B ASTM 4503 (modified)
Metals
Reporting in progress
SW-846 9056
I EPA350.32 I
I
I
Ammonia
NIA
Mercury
NIA
Liquids: SW-846 7470A Solids: SW-846 7471A
Reporting in progress
Cyanide
NIA
SW-846 9012A
Reporting in progress
Volatiles
Liquids: SW-846 5030B Solids: SW-846 5035
SW-846 8260B
Water and sand testing in progress
Semivolatiles Liquids: SW-846 351OC Solids: SW-846 3550B
SW-846 8270C
Reporting in progress
Polar VOlatileS SW-846 502 1
SW-846 8260B
Reporting in progress
SW-846 8081A (Pesticides) SW-846 8082 (PCBs)
Waste testing in progress
Pesticides/ PCBs
Liquids: SW-846 351OC Solids: SW-846 3550B
ASTM- American Society of Testing and Materials EPA-U.S. Environmental ProtectionAgency IC-ion chromatography ICPIAES- inductively coupled plasmdatomic emission spectrometry ICPMS- inductively coupled plasmdmass spectrometry NIA-not applicable PCBs-polychlorinated biphenyls
Reporting in progress
I
'All analytical methods fiom Test Methods for Evaluating Solid Waste, PhysicaVChemical Methods SW-846, unless otherwise noted. 'Methods for Chemical Analysis of Water and Wastes, EPA 600/4-79-020, March 1983.
The application of EPA's SW-846 guidance document, Test Methods for Evaluating Solid Waste, PhysicaVChernical Methods: to the unique radioactive waste matrix necessitated that a strategy be developed to support analytical method validation. This strategy was developed as part of the Regulatory DQO process. The initial step in this strategy has three main parts: 1. Determining method detection limits (MDLs) and estimated quantitation limits (EQLs) in water and sand to demonstrate the laboratories' ability to pefiorm the method with standard matrices; 2. Determining MDLs and EQLs in tank waste liquids and solids to demonstrate the ability to apply the methods and establish the MDLs and EQLs for liquid and solid radioactive waste matrices; and
Environmental Issues and Waste Management Technologies VIII
99
3. Conducting a holding time and storage condition study to understand the effect of radioactive waste sample management on characterization data measurements. Parts 1 and 2 started with the preparation of an implementing plan* for the Regulatory DQO followed by preparation of detailed laboratory test plans for each method. The results of the analyses were compared with quality control parameters documented in the project’s Quality Assurance Plan. These performance parameters were generally consistent with the SW-846 methods, specific American Society of Testing and Materials (ASTM) methods, or alternative methods approved by Washington State (WAC 173-303-1lO).’ Detection limits were also compared with available (albeit often preliminary) regulatory thresholds and were determined to be generally adequate for decisionmaking purposes. Final results are expected to be published in 2002. The data evaluation for Parts 1 and 2 involved an interactive process between the performing lab, their contractor, DOE, and Ecology. A jointly agreed-to format and protocol for interim data transmittal provided faster access to decisioncritical data (Figure 3) and a fairly thorough verification of the data quality. Decision-makers real-time involvement at key hold points yielded ownership of the results. A strong commitment to the process and the schedule by the contractors, DOE, and Ecology facilitated exceptional communication, minimizing delays.
Figure 3. Data quality objectives data verification and stakeholder review process
100
Environmental Issues and Waste Management Technologies VIII
The third part of the Regulatory DQO Step 1 is the performance of a holding time and storage condition study. This study will evaluate the effects of sample storage upon sample integrity, analyte degradation and loses, and analytical data measurements. While a number of statistical approaches have been evaluated for this study, hture implementation pathways have not been finalized. The DQO Step 1 concludes with a critical decision point. The applicability of methods for RCRA compliance will have been demonstrated for a solid waste matrix and a liquid waste matrix. In parallel permitting, testing of treatment efficiencies, plant design, and alternate treatment and disposal strategies will have advanced. This information will be assimilated to provide a basis for future characterization needs supporting the acquisition of decision-quality data.
’
SUMMARY DISCUSSION With the approaching end of Step 1 of the Regulatory DQO, we are heading into a consciously built-in decision point where the parties to the DQO committed upfront to reexamine whether changes are warranted in Step 2 implementation based on Step 1 results. In this evaluation, it is intended that continued tank waste characterization efforts will be tailored to obtaining effective data based on a better understanding of tank waste characterization capabilities and a current look at what data needs may be met. The validation of preparation and analysis methods for regulated compounds in tank waste matrices was intended to be the first step in providing technically defensible data for continued tank waste storage, preparation of the waste treatment plant permit, preparation of the treatment variance and delisting petitions, and to possibly meet other tank waste management data needs related to regulatory compliance. As the end of Step 1 approaches, deliberate cooperative efforts among regulators, stakeholders, waste management contractors, and the DOE should continue to focus on collecting data that are effective in serving their decision-making needs. Progress toward this end will be best supported by an attitude reflected by Nancy Wentworth, director of Quality staff for the EPA Office of Environmental Information (2002), “Get the right data, Get the data right, and Keep the data right.” ACKNOWLEDGEMENTS The authors would like to acknowledge the contributions of Gertrude Patello and her project staff, Battelle, Pacific Northwest Division; David Blwnenkranz, Bechtel National Inc; Nancy Welliver, DMJM H+N; and Jerry Yokel, Department of Ecology.
Environmental Issues and Waste Management Technologies VIII
101
REFERENCES I KD Wiemers, ME Lerchen, M Miller, and K Meier, “Regulatory Data Quality Objectives Supporting Tank Waste Remediation System. Privatization Project,” USDOE Report PNNL-12040, Rev. 0, Pacific Northwest National Laboratory, Richland, Washington, 1998. * KD Wiemers, RT Hallen, H Babad, LK Jagoda, and K Meier, “A Compilation of Regulated Organic ConstituentsNot Associated with the Hanford Site, Richland Washington,” USDOE Report PNNL-11927, Pacific Northwest National Laboratory, Richland, Washington, 1998. KD Wiemers, P Daling, and K Meier, “Rationale for Selection of Pesticides, Herbicides, and Related Compounds from the Hanford SSTDST Waste Considered for Analysis in Support of the Regulatory DQO (Privatization),” USDOE Report PNNL-12039, Pacific Northwest National Laboratory, Richland, W a s ~ g t o n 1998. , KD Wiemers, H Babad, RT Hallen, LP Jackson, and ME Lerchen, “An Assessment of the Stability and the Potential for In-Situ Synthesis of Regulated Organic Compounds in High Level Radioactive Waste Stored at Hanford, Richland, Was~ngton,”USDOE Report PNNL- 11943, Pacific Northwest National Laboratory, Richland, Washington, 1998. KD Wiemers, ME Lerchen, MS Miller, and NC Welliver, “Logical Selection of Analytes for H d o r d TWRS Privatization Waste Feeds,” 53rdNorthwest Regional Meeting of the American Chemical Society, Richland, W ~ ~ ~ o n , 1998. KD Wiemers, ME Lerchen, and M Miller, “An Approach for the Analysis of Regulatory Analytes in High Level Radioactive Waste Stored at H d o r d , Richland, Washington,” USDOE Report PNNL-11942, Pacific Northwest National Laboratory, Richland, W a s ~ n ~ o1998. n, EPA, “Test Methods for Evaluation Solid Waste PhysicaVChemical Methods,” SW-846,3rd Edition, as amended by Updates I (July, 1992), IIA (August, 19931, IIB (January, 1995), and 111, US. Environmental Protection Agency, Washington, D.C., 1997. GK Patello, TL Almeida, JA Campbell, OT Farmer; EW Hoppe; CZ Soderquiest, RG Swoboda, MW Urie; and JJ Wagner, ‘‘Regulatory DQO Test Plan for Determining Method Detection Limits, Estimated Quantitation Limits, and Quality Assurance Criteria for Specified Analytes,” USDOE Report PNNL13429, Pacific Northwest National Laboratory, Richland, Washington, 2001.
’
102
Environmental Issues and Waste Management Technologies VIII
Chapter 173-303 WAC. “Dangerous Waste Regulations.” Washington Administrative Code, as amended, http://www.ecy.wa.gov/ laws-rules/lawsetc.html. l0 Crumbling, Deana M. et. al “Managing Uncertainty in Environmental Decisions,” Environmental Science and Technology (October 1,2001).
Environmental Issues and Waste Management Technologies VIII
103
This page intentionally left blank
EFFECT OF T ~ S I T I O N ~ O N - T ~ S I T IMETAL O N MODI~CATION ON THE ACTIVITY OF GazO3-Al203 CATALYST FOR NOx REDUCTION BY HYDROCARBON UNDER OXYGEN-RICH CONDITIONS
Md. Hasan Zahir, Singo Katayama and Kunihiro Maeda Synergy Ceramics Laboratory, FCRA, Shimo-Shidami, Moriyama-ku, Nag0ya 463-8687, Japan
Masanobu Awano Synergy Materials Research Center, National Institute of Advanced Industrial Science and Technology (AIST), S ~ 0 - ~ oS~ y ~a m a~ - ~ ku, Nagoya 463-8687, Japan
ABSTRACT
The effect of additives on the catalytic performance of Ga203-Al203 has been studied for the selective reduction of NO with ethylene in oxygen-rich atmosphere. Ga203-Al203 with additions of Zn, Ni, Cr, Mn, Fe and La metal oxides were prepared by a co-precipitation method Among the catalysts tested, Zn-Ga203-Al203 with 15% Zn content exhibited the highest activity over a wide range of temperature. The addition of NiO and ZnO enhanced further the activity of Ga203-Al203 in the lower temperature region for the case of Ni doped and higher temperature region (450-600 'C) for the case of Zn-Ga203-Al203catalysts. Zn- Ga~O3-Al203catalysts are active for the re~uctionof NO with both CH4 and CtH4 in the presence of oxygen. The fact that no deactivation behavior was observed during continuous reaction conditions as well as its excellent reproducibility character is remarkable. The high activity and selectivity of nontransition metal doped-Ga203-~203,were att~butedto the spine1 type structure, containing highly dispersed transition and non- transition metal cations in A 1 2 0 3 matrix. INTRODUCTION The selective catalytic reduction (SCR) of NOx by hydrocarbons is under investigation worldwide as the most attractive technique for NO reduction to Nz in exhaust gas streams. In this regard Ga203-Alr03catalyst has the potential for a breakthrough in the catalytic reduction of nitric oxide by hydrocarbon under excess oxygen conditions [I]. The capability to retain high catalytic activity in the presence of considerable excess of oxygen is one of the unique properties of this ~
_
-
_
To the extent a u t h o ~ e under d the laws of the United States of America, a11 c o p ~ g hinterests t in this pub~cationare the property of The American Ceramic Society. Any dup~cation,reproduction, or republicationof this p ~ ~ c a t i or o nany part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright ClearanceCenter, is prohibited.
Environ~~ntal Issues and Waste ~ a n a g Tec~nologies ~ ~ ~ n ~VIII
105
catalyst. In addition, Ga203-A1203showed higher tolerance against water, and SO2 than Ga-ZSM-5, which indicates the possibility of designing de-NOx catalysts with high durability by using a nonzeolitic matrix [2]. However, the effective temperature window of the catalyst for NO reduction was relatively narrow and thermal stability is questionable. Some SCR catalysts containing aluminate-like phase were reported to be effective for NO reduction. For example, Hamada et al. [3], Kung et al. [4] studied Co/Al2O3, and showed that as the calcination temperature increased up to 1073 K, supported cobalt species changed from Co304 to C o A l 2 0 4 and the activity and selectivity increased. Okazaki et al. reported the similar results on Ni/Al203, and suggested that active metal species of these catalysts are the corresponding metal aluminates [5]. However, previous studies mentioned that zinc-alumina spinels work well for many applications because of their strong acidbase properties [6]. Although Zn is sometimes regarded as member of the 3d transition metals, ZnO is actually not a transition metal oxide because the 3d orbital of Zn2" is filled [7]. Usage of non-transition metal oxide for NO reduction has effective influence because the high selectivity of Ga203-Al203 was originated fiom non-transition metal (gallium) i.e., non-reducing nature of gallium ion resulting in the low oxidation activity [13. Therefore, we assumed that Zn-Ga2O3-Al203 i.e., structurally isomorphous ZnAl2O4 and ZnGazO4 or ZnAlGa04 system might be interesting for the reduction of NO by hydrocarbon. In this study, we investigated the effect of metal oxide additives namely Zn, Ni, Cr, Mn, Fe and La oxides on Ga203-Al203 with spine1 structure for the selective reduction of NO by CH4 and C2H4. EXPERIMENTAL All metals doped Ga203-&03 catalysts were prepared by co-precipitation method. Metal nitrate was used as metal sources, while ammonium carbonate were used as bases. The loading of metal oxide and Ga203 was fixed at 15 and 30mol %, respectively. Appropriate amounts of starting nitrates were dissolved in distilled water and aqueous ammonium carbonate ((NH4)2C03) solution (2.0 mol/L) was added to the solution to coprecipitate metallic ions. In this procedure, the pH of the mixed solution was kept at ca. 9.0 and then the solution was vigorously stirred for 24 h. The precipitate thus obtained was washed with distilled water three times, followed by drying at 110 OC and calcinations at 800 0 C for 5h in air. The crystalline phases were identified by X-ray diffractometry. Morphology of the product particles was examined using scanning electron microscopy. Particles size and surface area of the powders were also determined
106
Environmental Issues and Waste Management Technologies VIII
by laser diffraction method and Brunauer-Emmett-Teller nitrogen adsorption technique. Catalytic activity of the resulting powder was measured using a fixed bed flow reactor. The sample powders of 200 mg were placed between quartz wool plugs in the reactor. The reactant gas mixture consisted of 0.1 % NO, 1.510%0 2 and 0.2 % C2H4 balanced by helium at a space velocity (SV)of 12,000 h-'. Flow rate of each reactant gas was controlled by a mass flow controller. The total gas flow rate was fixed 50 cm3min-' (WE= 0.24 g s ~ m - ~Decomposition ). of the nitric oxide and total NOx in the reactor effluent was detected using a chemiluminescent NO-NOx gas analyzer. All other reactants were analyzed by on-line gas chromatograph. Before the catalytic reaction, the catalysts were pretreated in air at 800 OC for 5h. NO conversion to NZ and C2H4 conversion to CO2 calculations are based on the following expressions: NO conversion to N2 (%) = {2[N2]/[NO]'"}x 100, C2H4 conversion to CO2 (%) = {(1/2)[CO~]/[C~H4]'"}XlOO where [NOIin or [C2H4Iia are the inlet NO or C2H4 concentration, respectively, and [Nz] or [CO,] are the concentration of N2, or CO2 in the reactor effluent gas. RESULTS AND DISCUSSION Fig.1 shows XRD patterns of ZnO and Ni- GatO3-AlzO3 diffiaction lines due to a spinel phase. Graphs for the X-ray spectra of these powders did not show any excess ZnO, NiO, A1203, or Ga203that was present as unreacted material. On the other hand, as for the catalysts containing La, Mn and Cr, gave the diffraction peaks assigned to each metal oxide additives, indicating the formation of phase pure spinel or y-Al203 might not be possible for this present reaction system. Lines due to ZnO were not detected even on 30 mol.% Zn-Gaz03-Al203, while those due to spinel phase became intense, probably because its composition is close to mzo4 or ZnGazO4 (stoichiometric spinel). It is well known that zn&o4 and ZnGazO4 are structurally isomorphous, where the A13+and Ga3+ cations are interchangeable, and a compound that contains both aluminum and gallium i.e., zinc aluminogallate, can be synthesized [8]. Additive effect of metal oxides: Fig.2 shows the activity of Ni-, Zn-, Mn-, Cr-,Fe-, and La-doped Gaz03N203 for NO reduction by CzH4 in the presence of oxygen. Although Ga203M2O3 catalyzed effectively the NO reduction by CzH1 with high NO conversion more than 98%, the addition of NiO and ZnO enhanced further the activity of Ga203-Al203 in the lower temperature region (350-530 OC) for the case of Ni
Environmental Issues and Waste Management TechnologiesVIII
107
doped Ga203-Al203 and higher temperature region for the case of Zn-Ga203-Al203 at wide range of temperature 450-600 'C. Usage of non-transit ion metal oxide for NO reduction is important because the high selectivity of Gaz03A1203 was originated fiom non-transit ion metal (gallium) i.e., G a,O,-Al,Os-C r non-reducing nature of gallium ion resulting in the low o~dation activity f2]. Therefore it might assume that two 20 30 40 50 60 70 non-transition metal Dlffroctlon angle, 28 / * ( C u K a ) Fig.1. XRD patterns of various trensitlon metal doped G a 2 0 3 - A k 0 3 catalysts. i O n S (Ga3' and Zn2+) in the inert tetrahedral position enhance the activity at wide range of temperature. That is to say, ZnO on the Gaz03-Al203 exhibit a cooperative effect with Ga3" for the removal of NOx with C2H4, especially when the temperature is between 400 to 600 'C. The ~ a x i NO ~ ~conversion m on Zn- and Ni-Gaz03-Al203, was identical in compare with its non-doped Ga203-AlzO3. This is a remarkable result, because reported results of Inz03-, SnOz-, COO-, CuO- and &-doped Ga203-Al203 shows the maximum NO conversion was much less than that on Gaz03-A1203[2]. This is because h y ~ o c oxidation ~ ~ n by oxygen, which is a side reaction consuming hydrocarbon, proceeds predominantly because of the too high oxidation activity of CuO and Ag. In the case of In203 and SnO2, NO reduction was also decreased, although their propene oxidation activity was not so high. On the other hand for Mn, Cr, and La-modified catalysts, the enhancement in activity was very poor probably because the weak formation ability of y-Al2O3 phase or tiny spine1 structure. The activity of FeO-Ga203-AlzO3 was also low and it can be assume that the presence of large iron oxide particles catalyzing C2H4 oxidation with dioxygen. It is worth noting that on open supports, such as alumina, zirconia and sulfated zirconia, the activity and selectivity of CO,Pd, Ag, Ni, Mn, and Fe are limited by structural
108
E n v ~ o n ~ ~ nIssues t a l and Waste ~ a ~ a g e ~~ee nc ~ t n o ~ oVIII gi~s
effects, especially low dispersion, which renders the catalyst more active for CH4 combustion than for the SCR of NO [2]. RecentIy it has been shown that Ga203/Al203 system exhibits high catalytic activity not only by higher hydrocarbon but also by lower hydrocarbon. In particular, SCR by CH4 is most challenging since CH4 is relatively inert and generally reacts with 0 2 much faster than with NO thermally and over most catalysts [9]. In this study we observed that Zn-Gaz03/A1203 and Ni-Ga203/Alz03 catalysts retained their high catalytic activity even if CH4 concentration becomes 1000 ppm. Effect of 0 2 concentration The capability to retain high catalytic activity in the presence of considerable excess of oxygen is one of the unique properties of this catalyst. At present, the more widespread and well-supported point of view is that the catalytic activity of gallia containing systems for NO reduction in the presence of excess oxygen is due to the presence of gallium coordinately unsaturated cations (cus) on the surface (Lewis acid sites and Gazis)[lO]. In the presence of 1-10%0 2 , the NO conversion did not bend over, achieving almost complete reduction at 500 0 C. In the absence of oxygen, the NO conversion was quite low over Ga203Al2O3 and Zn-Ga203-Al203. The NO reduction to N2 increased up to 98% with increasing oxygen concentration. The results shown that the presence of oxygen is necessary for NO reduction and improves the NO removal activities by accelerating NO oxidation. Lifetime and reproducibility Another remarkable feature of Zn-Gaz03-Alz03 is, as expected, high thermal stability and stable catalytic performance, which is one important criterion for commercial application. The effects of recycling on the activity of the Zn-Ga203-Al203 catalyst are another point worth of discussion. For Zn-Ga203A I 2 0 3 catalysts, a steady-state conversion (96%) was obtained immediately at the
Environmental Issues and Waste Management Technologies VIII
109
beginning of the run at 5OO0C, and the conversion did not decrease even after 24h of continuous reaction. CONCLUSIONS
Transitiodnon-transition metal oxides-aluminogallate, exhibited spinel-type structure, show high activity and thermal stability. Highly effective catalysts for SCR by hydrocarbons (CH4 and C2H4) were obtained without the E M - 5 framework. The activityhtability of Zn-Ga’03-Al203 catalyst remained high under demanding reaction conditions; namely in 10% 0 2 , 1000 ppm NO, 1000 ppm hydrocarbon and high space velocities. No deactivation behavior was observed during continuous reaction conditions as well as its excellent reproducibility character is remarkable. These results clearly demonstrate a strategy to design de-NOx catalysts with high activity and durability by using non-zeolitic matrix. REFERENCES ‘M. Haneda, Y. Kintaichi, T. Mizushima, N. Kakuta, H. Hamada, Applied Catalysis B, 31 81-92 (2001). ’K. Shimizu, A. Satsuma, T. Hattori, Applied Catalysis B, 16 319-326 (1998). 3H. Hamada, Catalysis Today, 22 21-40 (1994). 4JY. Yan, MC. Kung, WMH. Sachtler and HH. Kung, Journal of Catalysis, 172 [l]178-186 (1997). ’N. Okazaki, Y. Katoh, Y.Shiina, A. Tada and M. Iwamoto, Chemistry Letter, 889-890 (1997). 6 R. Roesky, J. Weiguny, H. Bestgen, U. Dingerdissen,Applied Catalysis A, 176 213-220 (1999). 7 T. Tsubota, M. Ohtaki, K. Eguchi and H. Arai, Journal of Materials, Chemistry, 7 [l] 85-90 (1997). 8 SK. Sampath and JF. Cordaro, Journal of the American Ceramic Society, 81 31 649-654 (1998). ‘Jd. Armor, Catalysis Today 29 43-45 (1996). 10 Yu. N. Pushkar, A. Sinitsky, 00. Parenago, AN. Kharlanov, EV. Lunina Applied Surface Science, 167 69 (2000).
110
Environmental Issues and Waste Management Technologies VIII
Vitrification Technology and Melter Disassembly
This page intentionally left blank
COGEMA EXPERIENCE IN OPERATING AND DISMANTLINGHLW MELTER
R DO-QUANG (1) ,JL.DESVAllX(2), P. MOUGNARD(2) A. JOUAN (3), C. LADLRAT (3) (1) COGEMA, 2 rue Paul Dautier, BP 4,78141 V & y Cedex, FRANCE (2) COGEMA, 5044 Beaumont La Hague cedex (3) CEA, Valrh6 / Marcoule, BP 171,30207 Bagnols-sur-Ct2z.e Cedex, FRANCE ABSTRACT
The vitrificationof high-level liquid waste produced from nuclear fuel reprocessing has been carried out industrially for over 20 years by COGEMA, with two main objectives : containment of the long lived fission products and reductionof the final volume of waste. Research performed by the CEA (the French Atomic Energy Commission) in the 1950's led to the commissioning of the Marcoule Vitrification Facility (AVM) in 1978. In this plant, vitrified waste is obtained by first evaporating and calcining the nitric acid fed solutioncontainingfission products. The calcine is then fed together with glass fiit into an induction-heated metallic melter. Based on the industrial experience gained in the Marcoule Vitrification Facility, the vitrification process was implemented at an even larger scale in the late 1980's in the R7 and T7 vitrification facilities of the La Hague reprocessing plants.
So far, COGEMA's vitrification facilities have produced more than 11000 high-level glass canisters, representing more than 4500 tons of glass and 4500 million curies immobilized in glass. More than a technical success, in-line vitrificationof HLW produced by o p t i n g reprocessingplants has become a commercial reality. The basic principles leading to the choice of the two-step vitrification process with hot induction metallicmelter are : The separation of the functions, to have simpler and more compact equipment and to limit the size of the melter allowed for complete in-cell assembly and dissasembly with moderate size overhead cranes, master-slave manipulators and remote controlled tools. Easy remote maintenance of the process equipments with an optimization of solid wastes generated during operation The fact that the heating system is outside the metallic melter and thus independent from the melting pot, allow it to be not contaminated by HAL glass, to be not sensitive to the glass melting (no wear, no corrosion, no shorting and easy to start even if the metallic melter is full of glass), to be stop and to be maintain easily .
In parallel consistent and long term R&D programs have enabled continuous improvement of the process. The average melter lifetime now exceeds the design basis value (5000 hours instead of 2000 hours) and less than one week is necessary to stop a vitrification line, to change the melter and to restart the vitrificationoperation. The melting pot replacement is based on preventive maintenance to change equipment or subequipment before the failure in order to minimize the number of component or subcomponent to be change and the level of component's contamination The volume of secondary wastes fiom maintenance operations is thus minimized. Pieces of worn equipment are generally of small size, can be easily splitted for conditioning in accordance with COGEMA solid waste management strategy.
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
113
1) COGEMA industrial exDerience in Vitritlcation The vitrification of high-level liquid waste produced from nuclear fuel reprocessing has been camed out industrially for over 20 years by COGEMA, with two main objectives : containment of fission products and reduction of the final volume of waste. Research performed by the CEA (the French Atomic Energy Commission) in the 1950's led to the selection of borosilicate glass as the most suitable containment matrix for waste from spent nuclear fbel. This led to the commissioningof the MarcouleVitrification Facility (AVM) in 1978. Based on the industrial experience gained in the Marcoule Vitrification Facility, the A W vitrification process was implemented at larger scale in the late 1980's in the R7 and T7 facilities in order to operate its in line with the UP2 and UP3 reprocessing plants. Both vitrification facilities are equipped with three vitrification lines having each a glass production capacity of 25 kg/h. With two line in operation and one in stand by, each vitrification facility of La Hague reprocessing plants was designed to have a glass throughput of 50 kg/h and to meet the production requirements with sufficient flexibility of operation .
So far, COGEMA's R7 and T7 facilities at la Hague have produced (by the end of 2001) more than 8700 high-level glass canisters (CSD-V), representing more than 3500 tons of glass and 3,6 billion curies immobilized in glass. More than a technical success, in-line vitrification of HLW produced by operating reprocessing plants has become a commercial reality that led, in 1995, to the first return of glass canisters to COGEMA customers. 2)The R7/T7 Process 2.1) The La Hame two steo vitrification orocess
The French Atomic Energy Commission (CEA) began research on the immobilization of HLW in 1957 and led to the choice of a two-step vitrification process implemented first in the Marcoule Vitrification facility (AVM-1978) and extrapolated it in term of capacity and design for La Hague Vitrification Facility (R7-1989 and T7-1992).
114
Environmental Issues and Waste Management Technologies VIII
In the two step process, the nitric acid solution containing the concentrated fission products solution coming from reprocessing operation is adjusted in a stirred vessel to make its chemical composition compatible with specificationsfor the glass product. The solution is then fed to a rotary calciner where it is heated up to 600°C by four distinct zones.
The calcine is heated under air and most of the nitrates are transformed into oxides. Aluminum nitrate is added to the feed prior to calcination in order to avoid sticking in the calciner (melting of NaN03). Sugar is also added to the feed prior to calcination to reduce some of the nitrates and to limit Ruthenium volatility. At the outlet of the calciner, the feed is still in the oxidized state, with significant amounts of nitrates left. The calcine falls directly into the melting pot along with the glass frit which is fed separately. The melting pot is fed continuously but is batch poured. The melting pot is made of base nickel alloys; The glass in the bottom of the melter is maintained heated to a temperature of llOO°C and is fully oxidized. A glass batch is poured into a canister every 8 to 12 hours depending on the composition of the HLW solutions being treated. The canister (CSD-V), which has a volume of 150 liters is filled with two batches of 200 kg each. The maximum activity at the time of vitrification is of 760,000 Ci per canister The maximum contact dose rate of the canister at the time of production can be greater than lO5rad/h Off-gas treatment comprises a hot wet scrubber with tilted baffles, a water vapor condenser, an absorption column, a washing column, a iodine filter and three HEPA filters. The most active gas washing solutions are recycled from the wet scrubber to the calciner. 2.2) Glass Prodact Oaalitv
The R 7 m glass was designed to hold, at the maximum, 18.5 % of radioactive waste oxides (fission products, actinides, noble metals and Zr fines), or equivalently an overall maximum waste loading ratio of 28 %. This limit was in fact set to avoid excessive heating of the glass during interim storage. The glass product has a very high activity content (predominantly 13’Cs, %r) and significant amounts of noble metals (3 wt % max). During the qualification process for the R7 and T7 facilities, waste homogeneity has been demonstrated through grab samples during pouring and destructive examination of canisters. Homogeneity of the product was satisfactory and no undissolved feed was observed. Satisfactory quality of the glass has also been demonstrated through the examination of production samples obtained in both the R7 and T7 facilities. The glasses were homogeneous with no undissolved feed and their characteristicswere in full agreement with the expected values. The residence time of the glass in the melter is in the range of a few hours, which is enough for complete glass elaboration, provided that the temperature is sufficient. The R 7 m formulation is known worldwide to have an outstanding durability, especially in the long term. Normalized releases using a powder test very similar to the 7-day PCT are less than 1/10 of the US acceptability criteria.
Environmental Issues and Waste Management Technologies VIII
115
The basic principles leading to the choice of the two-step vitrification process and the multi lines design of La Hague vitrificationfacilities were : The separation of the vitrification functions, Easy remote maintenance of the process equipment. The separation of the process functions (calcinatiodvitrification) led to simpler and more compact equipment which is always desirable in a highly radioactive environment; Easy remote maintenance of all process equipment allowed for complete in-cell assembly and dissasembly with moderate size overhead cranes, master-slave manipulators and remote controlled tools. From a more general point of view, main process equipment of each vitrification line are located in a separate cell, while pouring and cooling cells are common to the three lines. All of these cells are equipped with cranes, master slaves manipulators and shielded windows for remote maintenance. They are associated with parking cells allowing crane maintenance as well as introduction of new replacement equipment. In the process cells, layout is optimized in order to facilitate access, modifications, and even addition of new equipment. The process equipment considered to be the least reliable are designed to be modular (i.e the calciner), so that their main subcomponentsare relatively compact and easy to replace remotely. The volume of secondary waste generated by maintenance operations is thus minimized. Pieces of worn equipment are generally small in size and can easily be splitted for conditioning in glass type canisters or cemented drum. As a result, maintenance operations are fully integrated into the process design and method of operations, which is of utmost importance to minimize downtime, volume and activity of secondary waste as well as to increase availability for production. 4) The Induction Heated Melter
4.1) General Drincfole The first work on vitrification of radioactive waste began in France in 1957 at the Saclay nuclear center. Techniques developed during this period to produce glass early used an induction-heated metal pot. The melter consisting of copper coil inductors embeded in a concrete structure, is designed to have a very long life time and can be remotely dismantled. The melter surround the melting pot which is the only really consumable item. The obvious advantage of this solution is that the heating system is outside the metallic melter (melting pot) and thus
0
Independent fiom the melting pot Not contaminated by HAL glass Not sensitiveto the glass melting (no wear, no corrosion, no shorting) Easy to start (even if the metallic melter is full or empty), to stop, to maintain or to replace.
Another major advantage of induction heating system is the simplicity to heat by Joule effect a metallic melter by using electric inductors.
116
Environmental Issues and Waste Management Technologies VIII
From PIVER (~O'S), the first industrial-scale prototype unit in the world intended for vitrification of concentrated fission product solutions with induction melter technology (batch direct liquid feeding, 90 kg glass batch pour, 12,5 metric ton of glass for 5,25 million Ci), through AVM (70's) and to R7 and T7 (~O'S), induction heating have been always successllly industrially operated by CEA and COGEMA. 4.2) R7R7 Induction Melter Description
The separation of the calcination function fiom the melting function allows to limit the size melter with regard to the design capacity : 0
The evaporative capacity of the R7R7 calciner is of about 80 Vh. The power delivered to the melting pot is used to melt solid product (calcine and fiit glass) and not to evaporate liquid.
Thus, the melting pot is ovoid (long axis 1 m, short axis 0,35 m, total height 1,40 m, weight around 400 kg), and is made of base nickel alloys.
.M
The melter is supplied with 4000 Hz power by a 200 kVA generator at a voltage adjustable up to 600V through four superposed copper inductors cooled by water and connected to the cooling circuit by flexiblepiping. The generator output voltage allows overall adjustment, and each inductor is supplied to also allow individual power adjustment L
I I
t-
L._L.-.-. L
R7/T7 melter 4.2.1) Glass Donring
The melting pot is equipped with two bottom pouring nozzle, also heated by induction, to fill glass canisters : The first is used for nominal melter operation to periodically fill the canister with 200 kg glass charge. The second ensures a complete emptying of the melting pot at the end of each vitrification campaign or for maintenance intervention, and thus participates to reduce the levels of activity of the secondary waste. 4.2.2) Glass stirring
Initially, the R7 / T7 melters were only equipped with bubble stirring devices which were efficient to produce the glass for nominal values of the noble metal content. However it was demonstrated by CEA that glass viscosity and frit glass - calcine reactivity are influenced by noble metals.
Environmental Issues and Waste Management TechnologiesVIII
117
As a consequence, COGEMA has undertaken the development of a mechanical stirrer in 1994 and
has deployed first these mechanical stkers on the T7 facility in 1996. The main objectives associated to the deployment of this new technology was to increase the noble metal content in the feed to the maximum specified noble metal content (3%) value and at the same time maintain the throughput capacity of the vitrification lines. The mechanical stirring was developed in addition to the bubbling device.
Melting pot are thus now equipped with both mechanical and bubbling devices; stirring devices have been designed to be compliant with the melting pot life time. A particular R&D program has supported materials selection used for the mechanical stirrer. Mechanical stirring proved to be successful in achieving the objectives since the noble metals content in the glass was increased h m an initial 1.5 wt. % to 3 wt. % (the maximum value) without any pouring problems or any metal accumulation in the bottom of the melter, with respect to the expected glass throughput. In fact, mechanical stirring also participate to the melting pot life time increase by reducing the melter wall temperature dispersion (see 4.2.4). 4.23) Control and monitoring
The instrumentation installed in R7fT7 melting pot enablesmeasurements of 0 0
The level and temperature of the molten glass The temperature of the melting pot’s wall
Since 1995, the level of glass in the R7 and T7 vitrification facility melting pots has been directly measured. There is a perfect correlation between this level measurement and the material balances determined with the crucible feed equipment. Factory calibration of the melter enables conversion of the level measurement into mass of glass It is very important to measure the temperature of molten glass and of the melting pot wall, as the temperature both determines the quality of the glass produced and governs proper operation of the process. Hot crucibles in the COGEMA vitrification facilities are equipped with thermocouples to measure:
0
The temperature of the molten glass, as this ensures that glass production conditions are adequate (a minimum average temperature of the molten glass is required as one of the parameters on which the quality of the glass produced depends), Wall temperatures of the melting pots, as these control heating of the melting pots (1 17OOC maximum under control).
The type of thermocouples used in induction heated crucible facilities at La Hague have been the subject of careful engineering and constant improvement.
A thermocouple characterization and evaluation program including: 0
Long duration endurance tests at constant temperature (1 15OOC) and with numerous temperature cycles (more than 2000), Integration of substantial doses of irradiation,
have been canied out and have validated the type of thermocouples currently used in the La Hague vitrification facility melting pots.
118
Environmental Issues and Waste Management Technologies VIII
A continuous monitoring of the melting pot temperatures and of the electrical induction parameters are used 0 0
To operate the melting pot To follow the melting pot ageing and wear and to prevent failure. 4.2.4) Melter life time
One of the major on-line developments undertaken on the melter has been the work performed to extend the melter's lifetime. At the start of operations in the R7 facility, the lifetime of the oval-shaped metallic melter was less than the design basis value (2000 hours) due to the combined effects of thermal, and mechanical stresses as well as corrosion in the gaseous phase. An important R&D program was launched. Comprehensive studies were performed in order to better understand the electromagnetic, thermal and mechanical behavior of the melting pot at the different stages of operation as well as the corrosion mechanisms at play. In particular, the power transfer fiom the inductor coils to the melting pot wall was analyzed in detail.
These studies helped to determine which species were responsible for corrosion. They also showed that the thermal power released in the melting pot's wall and therefore the temperature gradients in the melting pot could lead to high levels of stress as well as condensation of certain corrosive species. As a result of these studies, the design, material of the melter and the method of controlling the temperatures were modified. These changes led to a sharp increase in the lifetime of the melters. At present, after ten years of operation, the lifetime of the standard melting pot exceeds the design basis wlues by more than a factor of two. The average melting pot lifetime is of about 5000 hours, with an award at 6400 hours corresponding to more than 200 hundred glass canisters produced with a single melter. Today one melting pot by year and by vitrification line is used in R7/M vitrification facilities,that has a great impact on reducing process downtime and secondary wastes. 5) Melter redscement and dismantling
The R7 and T7 are mature vitrification facilities where operation and maintenance principles have been optimized with two main objectives : 0 0
Maximize production availability Minimize volume of secondary wastes
Even if efforts have been made to avoid the need for maintenance (e.g. by improving the reliability of equipment), some equipment, because of their nature or complexity, need to be periodically maintain or replace. A specific effort has been done to facilitate their maintenance and minimize intervention duration.
The R7R7 melter, is thus designed to be small, compact, easy to replace remotely, and as consequence cheap. As a result, melter's maintenance operations are l l l y inkgrated into the process design and method of operations.
Environmental Issues and Waste Management Technologies VIII
119
Remote maintenance is performed incell with cranes and remotely operated tools, using master-slave manipulatorsor servo manipulators. The melting pot replacement is based on preventive maintenance to change equipment before failure in order to minimize. 0
0
The number of componentor subcomponentsto be changed (meltingpot, melters.. .) The level of component’scontamination (limit contact with glass)
The unique experiencegained by COGEMA’svitrificationoperators, and the continuouscontrol of the melting pot operation (temperature, electrical induction parameters, analysis of small drift) allows us to anticipate and to change equipment before fatal wear ; this management is a mix between ‘periodical exchange’and ‘management value’ The main source of preventive intervention are metallic pot wear (corrosion, thermal and mechanical constraints ...) When the decision to change a melting is taken, the melter is emptied fiom its molten glass and the vitrificationline is progressivelystopped. When the melting pot is cold, it is disconnected fiom the calciner, and the melter which is on a mobile cart, is moved back. It is then possible to get the melting pot out of the melter, to transfer it to the dismantling cell (located above the vitrificationcell) and to replace it by a new one.
To have range of magnitude, ten days are routinely necessary to stop and to restart a vitrification line with a new melting pot.
Melting pot transfmji-omthe melter to the dismantling cell 5.2) Melting Dot dismantling Over the whole COGEMA complex, and especially at the L a Hague facilities, effort have been
focused for several year on the minimization and rationalization of the volume of conditioned solid waste coming from maintenance operation (contaminated tools, replaced components...) in standardized waste package. For example the Vitrified residue (CSD-V) and compacted residue (CSDC) have the same external design allowing 0
120
To standardizedhandling operation (same devicesto handle either CSD-C or CSD-V canisters)
Environmental Issues and Waste Management Technologies VIII
0
To simplify transport operation (same shape of transportation cask) To rationalize interim and long term storage if needed (additional possibility of mixed storage CSD-C/CSD-V)
These efforts encompassed all the aspects of waste generation and management, fiom plant and equipment design, to the development of specific repair techniques or to the implementation of rigorous sorting and decontamination of the waste at the source especially for materials in contact with
HLW. Vitrification dismantling cell are equipped with dismantling and decontamination tools, counting cell, and waste preconditioning unit. After dismantling and decontamination operation, pieces of worn equipment are generally of small size and sorted according to their activity level and material nature to be 0
Recycled inside the process, and conditioned in CSD-V type glass canister Send to La Hague waste treatment facility (AD2) to be cemented as technological waste, or, at soon compacted in the new La Hague Compaction facility (ACC) according to their level of activity (LLW / ILW).
For the melting pot, the dismantling operation consists to immobilize the melter in a vice and to cut it in small pieces of 300 x300 rnm by means of a sawing machine. For the melting pot, the dismantling operation consists to immobilize the melter in a vice and to cut it in small pieces of 300 x300 mm by using a sawing machine. Generally several blades (three to five) are necessary to completely dismantle the melting pot in small pieces.
Sawing machine
The mechanical saw alternative movement participate also to separate possible residual cold glass fiom metallic pieces and contribute do downgrade the waste. A complemenm mechanical cleaning of metallic piece could be completed by pneumatic drill, operated by master slave manipulator.
To have range of magnitude, six days of continuous operation are routinely necessary to dismantle a melting pot. Metallic pieces are decontaminated in high temperature nitric acid tank, counted in a measurement cell, sorted and preconditioned in EW basket (650 mm diameter, 1 m height). Three ILW baskets are necessary to precondition a melting pot and associated worn saw’sblades.
To finish the waste conditioning, ILW technological baskets are routinely sent by mean of mobile shielded cask to the A D 2 facility to be cemented and conditioned as technological waste in
Environmental Issues and Waste Management TechnologiesVIII
121
CBF-C2 residue (1200 liters volume, 4 metric ton max in weight). Generally three 500 Ci CBFC2 are necessary to condition a melting pot and its auxilaries (saw’s blades, current maintenance tools). Soon, a second way for ILW basket conditioning will be to send it (according to their activity) to the new Compaction Facility (ACC) to reduce by factor 4 the final waste volume by using universal standard canisters (CSD-C). 0
The recovered glass coming from dismantling melting pot operation are conditioned in HLW basket (380mm diameter, 1 m height) to be recycled in a specific standard vitrified canister (CSDv).
Melting pots are dismantled in line with the process in La Hague vitrification facilities according to the La Hague waste management strategy, allowing an optimization and standardization of the final waste volume. Because of the little number of melting pots to be dismantle by year, the treatment of melting pot in dismantling cells is a routine and short time operation l l l y integrated in the ordinary vitrification facility life. 6)Conclnsion
COGEMA has been operating industrial HLW vitrification facilities for over twenty years. The feedback fiom hot operations and the long-term R&D programs conducted with the CEA have helped to continuously improve the process in all of its aspects (glass formulation, process, associated technologies, operations and maintenance). The R7 and T7 vitrification facilities operating in-line with COGEMA’s two major 800 ton capacity commercial reprocessing plants have had outstanding records of operation, not only fiom the standpoint of total glass production and plant availability but also with respect to safety, remote incell maintenability, and secondary waste generated, demonstrating the maturity of the French vitrification process. With respect with the COGEMA layout and maintenance concepts, the melter induction technology used in R7/T7 vitrification has been designed to be small, compact, reliable and easy to replace remotely. As a result, melter’s maintenance operations are fully integrated into the process design and method of operations, and leads to optimiz,e maintenance time intervention and volume of secondary waste.
To go a step further in the Induction heating technology development, CEA and COGEMA are
committed in the development of the Cold crucible Melter (CCM) technology. The use of the Cold crucible Melter technology will lead to a virtually unlimited equipment service life and great flexibility in dealing with different types of waste. The high specific power directly transferred to the melt will allow high operating temperatures and high waste loading factor without any impact on the process equipment. And at last, the CCM technology remains always compact, simple and modular taking benefit from COGEMA’s melters operation experience.
References : Major breakfhroughsin high level waste vitrification G. MEHLMAN, R.DO QUANG, A. JOUAN, Waste Management ’99, Tucson
122
Environmental Issues and Waste Management TechnologiesVIII
CONCEPTUAL METHODS FOR DISPOSAL OF A DWPF MELTER AND COMPONENTS *Michael E. Smith, Dennis F. Bickford, Frank M. Heckendorn, Eric M. Kriikku Westinghouse Savannah River Company Savannah River Technology Center "Building 773-43A Aiken, SC 29808 ABSTRACT The Defense Waste Processing Facility (DWPF) has processed over 1.8 million kilograms of high level waste (HLW) glass since radioactive startup in 1996. The DWPF Melter is the heart of the vitrification process. The current plan is to store failed HLW equipment like the melter in Failed Equipment Storage Vaults. While this storage is acceptable in the short term, technology must be developed for proper long-term storage of these melters and other HLW equipment. Potential methods, including dismantlement sequences, for the disassembly, size reduction, and decontamination of a failed DWPF Melter will be discussed. INTRODUCTION The Defense Waste Processing Facility (DWPF) Melter was started up in 1994. It began processing radioactive feeds in April 1996. To date over 1000 canisters (1/6 of the projected canisters) have been filled with more than 1.8 million kilograms of radioactive glass. During the processing of this high level waste (HLW) at DWPF a need will arise to develop remote andor robotic systems to disassemble contaminated HLW processing equipment. This includes failed melters, vessels, and equipment. The current approach is to store this equipment in Failed Equipment Storage Vaults (FESV). While this storage is acceptable in the short term, technology must be developed to properly dispose of this equipment. This should include dismantlinghize reduction of the equipment, decontamination, disposal of the majority of the material as low level waste (LLW), and disposal of the remaining fraction as HLW materials. The DWPF melter will probably be the most difficult DWPF HLW equipment to disassemble and decontaminate. A single DWPF melter can hold up to 6000 kilograms of HLW glass. If an approach can be developed to dismantle and dispose of the melter, then similar techniques could be used with other HLW processing equipment. The design life of the melter was two years. This was based on a cited corrosion rate of the melter Monofrax K-3 refractories of 5.4 mils per day.' The actual expected corrosion rate should be much less and is based on pilot scale melter work and the operating history of the first DWPF melter (Melter 1). This melter is still operating in the DWPF after seven years. Although the melter design has proven robust, the life of future DWPF melters may not be as long due to the processing of feeds with higher levels of noble metals.* During the DWPF operations time of 25 to 30 years, several melters will most probably fail and be temporarily stored in FESV's. There is no facility at SRS that is currently setup to perform the dismantlement work on these melters. One concept would return failed melters and other large equipment to the DWPF Canyon after waste processing is completed. The equipment would then be disassembled and size reduced in the canyon. The required manipulators, tools, etc. are not currently in place or specified. In addition, there is no existing To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
123
plan on the dismantlement of a DWPF Melter. The need to remotely disassemble and size reduce HLW equipment is a DOE complex wide problem. Making the task more difficult is the fact that the various sites have different melters and facilities available to do the work. This paper identifies potential methods for the disassembly, size reduction, and decontamination of the DWPF Melter. DWPF MELTER DESCRIPTION The DWPF Melter, shown in Figure 1, is a refractory-lined, sealed, stainless steel vessel with a 1.8-meter internal diameter and approximately 2.1 meters of internal height. Figure 2 shows DWPF Melter 1 assembly with the frame and lifting yoke. The 3.8-centimeter thick vessel shell is 304L stainless steel with an outer diameter of 2.56 meters. The shell contains no remotely removable parts. All shell components have at least a two-year design life.
Figure 1 - DWPF Melter Cross-Section Primary glass containment is achieved by use of 30.5-centimeter thick Monofrax K-3 fused cast refractory brick. This refractory is very dense and hard. Korundal XD refractory was used in the vapor space. Zirmul refractory was chosen for the refractory placed underneath the Monofrax K-3. The glass pool is maintained between 1050 - 1170°C. The DWPF Melter electrodes are four uncooled plates fabricated from Inconel 690 of sufficient thickness to last greater than two years. The DWPF Melter vessel shell is penetrated in the vapor space by four pairs of horizontal resistance-type Inconel 690 dome heater tubes. Each tube is 8.3 centimeters in diameter. The DWPF riser heater consists of an Inconel 690 serpentine heater that surrounds the ten-centimeter inside diameter Inconel 690 riser channel. The pour spout heater also consists of an Inconel 690 serpentine heater that surrounds the five-centimeter diameter of the pour spout channel. These strip heaters keep the glass flowing through the riser and pour spout channels at a temperature between 1050-1100°C. The DWPF Melter is composed of various components that are quite bulky and heavy. Table I gives the weights of the major components (total is 67,600 kilogram^).^
124
Environmental Issues and Waste Management Technologies VIII
Figure 2 - DWPF Melter Assembly Table 1 - Weights of DWPF Melter Major Components Component Weight (kilograms) Vessel 17,600 Frame 12,900 Refractory 25,200 Components 9,800 Piping 1,200 Nozzle Material 900 Not only is the DWPF Melter large and heavy, it was not designed to be remotely disassembled. For example, the melter lid is attached to the rest of the melter with 56 bolts that are tightened and then tack welded in place. This lack of design for remote disassembly is also seen in that most of the major melter parts (like the lid) where not designed for remote removal. PREVIOUS MELTER DISMANTLEMENT EXPERIENCE There has been limited remote experience with the disassembly of HLW melters. There have been several pilot-scale or full scale melters that have either been inspected or disassembled after being used in non-radioactive testing. This section discusses these activities. Scale Glass Melter (SGM) Inspection The Scale Glass Melter was a two-thirds scale non-radioactive pilot melter of the DWPF Melter that was operated at SRS between 1985 and 1988. Before the melter was shut down a sequence of “high-risk‘’ tests which thermally cycled the melter were performed. After these tests and additional feed tests, the SGM was drained and shut down, the melter lid removed, and the refractories inspected. The main finding of the inspection of the SGM after all of this thermal cycling was that the K-3 refractory was basically intact with some minimal wear at the melt line. Integrated DWPF Melter System (IDMS) The IDMS was a one-ninth scale, non-radioactive DWPF pilot system. The melter was still operational at the time of its 1995 shutdown, although deposits of noble metals on the melter floor
Environmental Issues and Waste Management Technologies VIII
125
were requiring higher current from the lower electrodes to keep the lower melt pool in the operational temperature range. Findings pertinent to melter disassembly work are as follows4: 0 The melt pool refractory (Monofrax K-3) experienced some thinning and spalling at the sides in the melt zone (area between glass and feed). Overall the K-3 was in excellent condition. 0 The Mullfrax 202 vapor space refractory (similar to Korundal XD) was in excellent shape. The lower electrodes showed wear at their bottom faces. This was due to the channeling of a portion of the lower electrode current flow (due to noble metals) through this area. West Valley Functional and Checkout Testing of Systems (FACTS) Melter Disassembly At the West Valley Demonstration Project (WVDP), non-radioactive vitrification processing operations were conducted from 1984 to 1989 using a slurry-fed ceramic melter similar to the one now in radioactive operation at the WVDP. After the tests were completed, the melter was removed and manually disassembled. The melter was then inspected to determine the impact on the melter internals due to five years of glass processing. The following gives the sequence of steps used to disassemble the melter5: The seal weld between the melter shell and lid were manually plasma arc cut. The lid was then lifted and inverted to inspect the refractory. The refractory was removed from the melter lid by use of jackhammers and pry bars. The process required hammering through the refractory to the fiberboard backing and then working the hammers and pry bars under the refractory and around the nozzles. The size of the refractory waste pieces could not be controlled. 0 The melter cavity was disassembled top down in layers. The vapor space refractory was first removed using the same manual techniques as were used on the lid. Glass penetrations 0.64 to 1.3 centimeters were noted in the refractory cracks below the melt line. The refractories were somewhat difficult to handle even manually. 0 Once the innermost layer of glass contact refractory was removed, the remainder was removed cleanly. Pamela Vitrification Plant Radioactive Melter The Pamela plant in Dessel, Belgium was designed to vitrify high-level liquid reprocessing waste. The plant is somewhat unique in that is was designed to handle the remote dismantling of large HLW equipment such as the melter. Operations were started in 1985. After the initial vitrification program was completed in 1991, two melters and other large scale equipment had to be dismantled so that the plant could be reused for a second vitrification program that was planned to be started in 1999. From October 1991 till March 1994 one melter and three other large pieces of equipment were dismantled and dismantling waste was conditioned. In total 187 drums (200 liters each) containing cemented medium-level waste were transferred to the ap ropriate on-site facilities. Some five tons of low-level dismantling waste were transferred as well.
B
Although the plant had a dismantling cell, it was decided to perform the dismantling operations in the melter cell. Both the melter cell and the dismantling cells are equipped with a heavy-duty manipulator. The melter cell also has a 2-ton overhead crane while the dismantling cell has a 20ton overhead crane. The melter cell has three lead glass windows at ground level with masterslave manipulator pairs at each window. Three more windows are located on the second level. Two of these higher windows have master slave manipulators as well. The design of the melter cell and the dismantling cell took into account the ability to dismantle the melter and process the dismantling waste. This design included having these two cells close to each other and having an adjacent intervention cell that allowed for the maintenance of the cell
126
Environmental Issues and Waste Management Technologies VIII
cranes and heavy-duty manipulators. The dismantling was done by using existing manipulation systems and by adapting traditional tools like grinding discs, impact wrenches, vacuum cleaners, hydraulic jacks, hammer drills and grab tools for remote work. The work was completed within 2.5 years with about 25,000 man-hours. Lessons learned from this task are summarized below. 0 Dismantling techniques should be considered when designing a vitrification plant (access to and visual control in shielded cells must be easy) 0 Manipulators must allow for operation of multiple types of tools Large drums should be used for packaging of waste to reduce size reduction time 0 Personnel doing the work must be experienced remote manipulator operators 0 Grinding discs were better at cutting stainless steel than diamond encrusted saw blades (faster and less expensive) Spare parts used in the dismantlement work should be in stock to minimize time losses RELEVANT DOE D&D ACTIVITIES There have been several DOE funded decontamination and decommissioning (D&D) activities that are relevant to the dismantling of large radioactive equipment. A TFA report from O N 3 summarizes these various activities. These activities include the D&D Chicago-Pile No. 5 (CP-5) research reactor, the size reduction of the Tokamak Fusion Test Reactor Vacuum Vessel, and D&D work at the INEEL South Tank Farm in January 2000. Technologies used included the Dual Arm Work Platform (DAWP) at CP-5 (DAWP developed by ORNL), diamond wire cutting by Bluegrass Concrete Cutting, Inc., and the Modified Brokk Demolition Machine. A jointly written report from SRS and ORNL gives an outline for large-scale system operations and D&D work.7 Finally, an SRS report describing the conversion of a section of one of the separations buildings at SRS from a remote-crane-operated facility into a master-slave-manipulator-operated facility will give some insight into what it might take to convert an existing radioactive facility into a remote dismantlement facility.’ This work includes equipment removal and building decontamination. It also includes the installation of new service and support equipment. CURRENT DWPF MELTER STORAGEhIISPOSALPLAN During the design of the DWPF, special cells and equipment were considered for the dismantlement of failed DWPF Melters but were deemed too expensive. Therefore the current long-term storage/disposalplan for failed DWPF Melters is as follows. Disconnect the melter assembly from the Melt Cell Move the melter assembly to the Remote Equipment Decontamination Cell (REDC) for external decontamination Place the melter assembly into a carbon steel sealed storage box Transfer it via rail car to an underground Failed Equipment Storage Vault (FESV) There are now two existing FESV’s at DWPF. Each FESV is a reinforced concrete structure designed as a Category 1 nuclear facility. One melter assembly can be stored in each vault. Permanent storage of these melters in the FESV’s is probably not a viable option as this would amount to using the Savannah River Site as a HLW repository. Therefore, it is assumed that at the end of the life of the DWPF the failed melters will have to be size reduced and then the parts segregated based on waste classification. This may not have to be done if their radioactive content is low enough to meet incidental waste rules. This may be accomplished by moving the melter assemblies (one at a time) back into the DWPF Canyon to be D&D’d. There is no formal plan at this time as to how to do this D&D work. Because the DWPF was not specifically designed for large KLW equipment D&D, another existing facility or a new facility designed solely for this purpose may instead be used. Finally,
Environmental Issues and Waste Management Technologies VIII
127
there is the option of shipping these failed melters to the HLW repository. The size and weight of the melter assembly along with the potential amount of HLW glass in the melter will probably make this transportation option impossible. Also, the federal HLW repository design has no provision for accepting and storing a package of this size and weight. DWPF MELTER DISMANTLEMENT This section describes the basic tasks that will be required to dismantle the DWPF Melter. The exact details cannot be addressed because the location of the dismantlement has not been determined. In addition, the details of how the melter assembly is disconnected and moved are not discussed here as these will be covered in DWPF procedures. There are basically three different approaches to DWPF Melter dismantlement. The first approach would be to remotely remove enough glass from the melter so that the melter would be considered non-HLW. The second approach would be to dismantle the melter from the outside in. The last approach would be the dismantlement of the melter from the inside out. Each of these options is discussed below. Many of the steps are similar for the various approaches and therefore the tooling required for the tasks are discussed in a later section. Glass Removal Approach The chances of removing enough glass from an intact melter to have the melter disposed of as non-HLW are fairly low. This is because the glass would have to be mined out remotely through the various 10-centimeter diameter top head nozzles. The glass cannot be accessed via the 20centimeter diameter off-gas nozzles because the dome heaters are directly below these ports. It may, however, be conceivable to take off the melter lid and then remove enough of the glass to store the melter as non-HLW. With regards to the maximum radiation levels for the melter to be disposed of as non-HLW, these numbers and rules change with time. Therefore, the specifics are not discussed here as dismantlement work will not occur for probably another 10 to 20 years. This timing could of course be accelerated if multiple melter failures occur due to problems such as the settling of noble metals. The basic steps to perform this dismantlement are as follows. Remove all melter top head components (the components would probably be removed if possible before the melter was shutdown so that residual melt pool glass would not trap these components in the glass) Remove all melter jumpers Cut and grind off the 56 melter lid bolts Remove the melter lid Cut away and remove the melter dome heaters sections on the inside of the melter Breakup/remove the glass fiom inside of the melter and melter lid and place in DWPF canisters or other appropriate disposal vessels Decontaminate the melter assembly and outside of the melter shell as much as possible (may require covering of melter with cover plate) without getting water, etc. inside of melter Determine radiation field of melter lid and main melter, estimate remaining HLW inventory Reinstall the melter lid back onto the melter Place melter assembly in storage box and either return to FESV or bury as non-HLW Place glass removed from melter in DWPF canisters or other appropriate disposal vessels Size reduce top head components and segregate based on activity Ievel (ensure HLW pieces can fit into DWPF canisters or other appropriate disposal vessels) Place HLW top head component pieces into same DWPF canisters or other appropriate disposal vessels that contains the glass removed from the melter and seal canister (decontaminate canister/vessel surface if needed)
I28
Environmental Issues and Waste Management Technologies VIII
The main advantage of this approach is that it is the simplest and least expensive of the three alternatives. Minimal amounts of size reduction of melter materials would be required. The effort to remove the glass from the melter would be about the same as the other two approaches. The main disadvantage is the uncertainty as to whether or not the removal of the glass would then allow for the disposal of the melter as non-HLW. Also, the details of how to reinstall the melter lid back on the melter are uncertain at this time. If repository and waste regulations allow for this approach at the time of disposal, then it is recommended that it be used for the DWPF Melters. Inside-Out Melter Dismantlement Approach This approach is the same basic type of method cited in the West Valley dismantlement report.5 Remove all melter top head components Remove all jumpers Cut and grind off the 56 melter lid bolts Remove the melter lid Removehreakup the melter lid refractories (may be possible to first remove HLW contaminants from surface of these refractories to allow disposal of most of it as non-HLW) Cut away and remove the melter dome heaters sections on the inside of the melter Removehreakup the melter refractories above the glass level (may be possible to first remove HLW contaminants from refractory surfaces to allow disposal of most of it as non-HLW) Breakuph-emoveglass from inside of melter Removehreakup the remaining glass contact refractories (may be possible to first remove HLW contaminants from refractory surfaces to allow disposal of most of it as non-HLW) Remove and decontaminate melter electrodes Cut, remove the riserlpour spout section Cut, decontaminate, and remove the melter shell Cut, decontaminate, and remove the melter frame assembly (melter shell and frame may not need to be cut up or decontaminatedas they may be able to be disposed intact as LLW) Place glass removed from melter in DWPF canisters or other appropriate disposal vessels Size reduce top head components and segregate based on activity level Place HLW top head component pieces into same DWPF canisters or other appropriate disposal vessels that contains the glass removed from the melter This approach allows the melter shell to provide containment for the refractory pieces, therefore heIping to minimize the amount of contamination in the dismantling facility. It also allows for the movement of the melter by the use of the lifting yoke until the melter shell is cut. This approach is similar to the one that was used on the Pamela Melter as previously discussed.6 Outside-In Melter Dismantlement Approach In this approach, the melter shell is removed before the removal of the refractory. This approach is similar to the second disassembly alternative cited in the West Valley melter dismantlement report5 The less contaminated refractory (located at the higher melter elevations) is taken first and then the glass-contaminated refractory is removed. This allows for the bulk of the frame and shell to be disposed of as small pieces of LLW. The basic steps to accomplish this dismantlement are as follows. Perform the first six steps of the Inside-Out Melter Approach previously discussed Cut away, decontaminate, chop up, and remove the top portion of the melter frame assembly Cut, decontaminate, and remove the melter shell (not the bottom portion of the shell)
Environmental Issues and Waste Management TechnologiesVIII
129
Removehreakup the melter refractories above the glass (may be possible to first remove HLW contaminants from refractory surfaces to allow disposal of most of it as non-HLW) Pryhemovehreakup the melter refractories below the glass (may be possible to first remove HLW contaminants from refractory surfaces to allow disposal of most of it as non-HLW) Breakuph-emoveglass from inside of melter Cut, decontaminate, and remove the remaining bottom portion of the melter shell Cut away, decontaminate, chop up, and remove the remaining bottom portion of the melter frame assembly Place glass removed from melter in DWPF canisters or other appropriate disposal vessels Chop up top head components and segregate based on activity level Place HLW top head component pieces into same DWPF canisters or other appropriate disposal vessels that contains the glass removed from the melter. The easier to reacWless contaminated refractories are removed first. Unfortunately, there is a greater chance of spreading contamination in the dismantling facility as there is no containment provided by the melter shell. In addition, the glass contact refractories will not be supported when being broken apart. This approach could, however, make the handling of the refractories easier. CANDIDATE DISMANTLING TOOLS In considering tools for the disassembly of a melter or other large HLW equipment there are several characteristics of the tools that must be considered. Obviously the tool must be able to remotely perform the task required. Because they will be used in a remote radioactive environment, other criteria should be considered as well. One of these is reliability. The speed at which the tool accomplishes the work along with the ease that is can be done must also be factored. The ease of repair or change out of the tooling is still another factor. Tooling should also be chosen based on whether or not it can be used with the existing manipulators, etc. that are in the dismantlement facility. Another consideration is the amount and type of secondary waste that will be generated by the various tools being considered. Finally, the use of proven technology is a prudent approach when choosing tools for this type of work. The West Valley and Oak Ridge reports have excellent summaries of tooling that could be used.”* The various tasks are given below with tooling that should be considered for each step. Before the melter lid can be removed, the 56 top head bolts that hold the lid to the melter vessel must be cut. This includes the tack welds from the nuts to the bolts and the nuts to the flange. The best choice for this would be a grinder tool with grinder discs. Grinding discs were successfully used during the Pamela Melter dismantlement.6Lifting the melter lid remotely was not considered in the design of the melter. There are four lifting holes located on the top of the melter lid that were used to lift the lid during installation. By using manipulators, chains may be able to be reinstalled at these four points to remove the lid. If not, the lid could be raised by accessing the various nozzles. This technique, however, could cause problems by breaking up the melter lid refractories during movement. A lid lifting jig or fixture could also be fabricated. This suggests that a welder to install lifting and handling tabs may be required. The breakup of the various melter refractories could be accomplished by the use of various tools. These include the following. Hydraulic or mechanically actuated wedge (to move refractory stuck in place) Ram (free falling weighted chisel) Hydraulic or mechanical spreader Jackhammer 0 Diamond wire saw
130
Environmental Issues and Waste Management Technologies VIII
La Bounty Crusher Rotary tool with face grinding wheel (for refractory surface cleaning) Needle gun (for refractory surface cleaning) Abrasive water jet Grapple device (for picking up refractory pieces) Vacuum (to grab smaller pieces) The most likely scenario would be to grind or cut the face of the refractories to remove as much contamination as possible. A vacuum system could be used to collect the waste dusting from the grinding. A jackhammer with the proper force and stroke would then be used to break up the refractory. This was the tool of choice for the Pamela Melter work.6A heavy-use remote arm with “grabbing fingers” and possibly a modified grapple device (attached to an overhead crane) currently used at West Valley would be used to grab and move the broken up refractory pieces. The cutting of the Inconel 690 dome heaters (8.3-centimeter outer diarneted5.7-centimeter inner diameter) will have to be done with the heaters in place. The following tools may be used. Plasma torch Grinding discs CircuIar saw Abrasive water jet Shears Needle gun Grapple device If grinding discs are used, a vacuum system could be used to collect the waste dusting. The top head components are also made out of Incone1690 and have a 7.6-centimeter outer diameter. The grinding disk or a needle gun can be used to remove glass from these components before they are size reduced. Per past experience with the SGM, there should be about 2.5 centimeters of glass in the DWPF Melter if it is drained completely via the drain valve.” A melter that was not drained could have as much as 6000 kilograms of glass. The following tools may be used to remove the glass. Abrasive water jet Vacuum Needlegun 0 Jack hammer Pneumatic chisel Ram (weight chisel) Grapple device The cutting of the Melter Frame Assembly and the Melter Shell could use the following tools. 0 Shears Plasma torch Circular saw Abrasive water jet Grinding discs SUMMARY/RECOMMENDATIONS The current plan is to store failed DWPF Melters in FESV’s located at the DWPF. Because these failed melters may hold up to 6000 kilograms of radioactive HLW glass, they will eventually need to be D&D’d. At this time there is no facility at SRS specifically designed for the D&D of large HLW equipment. It may be possible to take the lid off of failed DWPF Melters and remove enough glass to then classify the melters as non-HLW. This would then allow the melters to be disposed onsite without a full scale D&D effort. If for technical or regulatory reasons this is not possible, two disassembly options are given. With the need for D&D work on HLW equipment recognized by the DOE complex, a DWPF Melter glass removal remote demonstration overseen by SRS is currently planned for FY03 at ORNL’s robotics facility using the Scale Glass Melter. In addition, West Valley is performing D&D tests on HLW vitrification equipment. In closing, the following recommendations are given with regards to this DWPF Melter D&D task.
Environmental Issues and Waste Management Technologies VIII
131
If possible, remove the lid from the failed DWPF melters, remove as much glass as possible, replace the lid, and then store the melters as non-HLW glass. Pick tooling that has been proven to work in other D&D activities. Equipment reliability must be considered as well as ease of operation. Cleaning contaminants from the parts of the melter (refractory, etc) should be done to minimize the amount of HLW to store. The amount of secondary waste that will be produced must be considered when choosing the D&D sequence and tooling involved. If the DWPF Melter cannot be drained and it is still operational (yet deemed to be shut down due to the number of years run or reduced performance), then the flushing of the melter with at least three melter volumes of non-radioactive glass should be considered. Flushing of a drained melter should be considered as well. The design of future DWPF Melters and new HLW vitrification plants such as Hanford and Idaho should consider adding features that will make the D&D of these melters easier: 1. Design a way to remotely disengage and pick up the melter lid. 2. Add several larger nozzles (possibly for the feed tube nozzles) to allow easier glass clean out of the melter (especially noble metals at the melter floor) without removing the lid. 3. Consider making some of the failure mode parts of the melter (for example the riser/pour spout) remotely replaceable. 4. Consider a new melter design that is easily replaced and D&D’d. This design should weigh much less and minimize the amount of glass contact refiactory. Future HLW vitrification sites should consider designing the vitrification facility or a special celllfacility for the remote D&D of large HLW processing equipment. REFERENCES 1. Basic Data Report, Defense Waste Processing Vitrification Facility Sludge Plant, USDOE Report WSRC-RP-92-1186, July 1992. 2. D. F. Bickford, M. E. Smith, “The Behavior and Effects of the Noble Metals in the DWPF Melter System (U)”, USDOE Report WSRC-TR-97-0370, Savannah River Technology Center, November 30,1997. 3. B. S. Richardson, “Melter Glass Removal and Dismantlement”, USDOE Report ORNL/TM2000/324, Oak Ridge National Laboratory, October 2000. 4. C. M. Jantzen, D. P. Lambert, “Inspection and Analysis of the Integrated DWPF Melter System (IDMS) after Seven Years of Operation (U)”, USDOE Report WSRC-RP-96-575, Savannah River Technology Center, February 6, 1997. 5. “Recommended Methods for Decontamination and Decommissioning, Size Reduction, and Disposal of Melter and Components - Evaluation Report”,West Valley Nuclear Services, Co., February 28,2001 (no author or documentnumber cited). 6. P. Luycx, M.Demonie, M. Snoeclat, L. Baeten, (Belgoprocess, Gravenstraat 73, B-2480 Dessel (Belgium)). 1996. “Experience gained with the dismantlingof large components of the Pamela Vitrification Plant”, Proceedings of the International Topical Meeting on Nuclear and Hazardous Waste Management Spectrum ‘96. August 18-23, 1996 at Seattle, WA (U.S.A.), American Nuclear Society, Inc., La Grange Park, IL (U.S.A.). pp. 1717-1727. 7. F. Heckendom and R Kress, “Outline for Large-scale System Operations and D&D Report”, USDOE Report WSRC-TR-2000-00364, Savannah River Technology Center and Oak Ridge National Laboratory, September 2000. 8. R. D. Kelsch, A. J. Lethco, and J. B. Mellon, “Multipurpose Processing Facility to Separate Actinides”, Proceedings of 20* Conference on Remote Systems Technology, 1972.
132
Environmental Issues and Waste Management Technologies VIII
EVALUATION OF CRYSTALLINITY CONSTRAINT FOR HLW GLASS PROCESSWG Pave1 Hrma, Josef Maty69; and Dong-Sang Kim Pacific Northwest National Laboratory P.O. BOX999, MS:K6-24 Richland, WA 99352 ABSTRACT It has been a commonly held assumption that constraining liquidus temperature (7') prevents the accumulation of crystalline phases in the high-level waste (HLW) glass melter because crystals, if they form at all, should dissolve easily in the melt at temperatures above liquidus. This, as the model calculation showed, is not the case in melters with fast circulation flow. If the melt circulates rapidly between cool and hot regions, crystals do not have a sufficient time to dissolve while in the hot zone. As a result, a steady-state size and concentration of crystals is established throughout most of the melter during normal operation. A consequence of this result is that the rate of crystal accumulation in the melter only slightly increases with increasing , 'T but strongly increases with increasing crystal size. For the melter simulated by the model, the TL could be 100°C above the accepted constraint without a serious impact on melter performance. Nucleation agents that keep crystals small abound in most HLWs but are often absent in simulated wastes for experimental melter runs. The weak impact of TL on melter performance is an important finding because without the current TL constraint, the HLW glass volume at H d o r d can significantly decrease. INTRODUCTION One main risk for the continuous operation of high-level waste (HLW) glass melters is the accumulation of solid phases, such as noble metals, spinel, eskolaite, or zirconiumcontaining minerals. To lower this risk, HLW glass is formulated with a constrained liquidus temperature (TA).'An unfortunate consequence of this constraint is to limit the waste loading in the glass, leading to a high volume of waste glass and, in turn,to high capital, production, and disposal costs. Optimizing glass composition to achieve maximum waste loading compatible with glass property constraints' has shown that without the TLconstraint, the waste-glass volume could decrease by 12 to 16%, which for Hanford represents substantial saving. The TAconstraint is based on the assumption that if the estimated glass temperature at the bottom of the melter is higher than TL,spinel or other crystals (except noble metals) would not be present within the melt, and thus the only problem to deal with would be the settling of noble metals. However, the melt temperature is as low as 850°C in the melting Current address: Institute of Inorganic Chemistry, Academy of Sciences of the Czech Republic, V HoleSoviEkhch 41, 18000 Prague 8, Czech Republic.
a
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
133
zone below the cold cap where spinel or other crystalline phases are likely to exist. It is the commonly held (tacit) assumption that these crystals dissolve as soon as the melt reaches a temperature above the TL. This assumption was checked with the Glass Service-Glass Furnace Model (GSGFM), a mathematical model originally designed for commercial glass making? This model was augmented by adding to it an algorithm for predicting the spinel distribution and accumulationrate in the melter?-5 The algorithm used to augment the model is based on extensive experimental studies6-12conducted fiom 1998 to 2001 to measure the parameters that describe spinel formation and settling in molten glass. This paper summarizes the outcome of the spinel-distributionmodeling and discusses in greater detail the consequences for HLW glass vitrification. EXPERIMENTAL STUDIES Experimental studies described in detail in previously published papers6-12were all conducted with MS-7 glass (Table I), an 11-componet generic HLW glass that formed spinel as the primary crystalline phase. The physical properties of the glass that were needed for mathematical modeling (density, viscosity, and electrical conductivity) were measured as functions of temperature6(Table 11). To measure the density of spinel, spinel crystals were isolated fiom the glass by digesting the glass phase in an acid. The methods used to obtain phase equilibria, the kinetics of nucleation, growth and dissolution of spinel crystals, the settling rate, and the properties of the sludge layer of spinel in MS-7 glass are briefly described below. Oxide A1203 B2O3 Cr2O3 Fe203
Mass Fraction 0.0800 0.0700 0.0030 0.1150
Oxide Li20 MgO MnO Na2O
Mass Fraction 0.0454 0.0060 0.0050 0.1530
Oxide NiO SiO2 Zr02
Mass Fraction 0.0095 0.4531 0.0600
Table 11. Properties of MS-7 Glada) Property Equation or Value Viscosity exp(-12.3 +19723/T) Electrical Conductivio exp(6.97-2914/(T-466)) SpecificHeat Melt Density Spinel Density
Pg
kg-m"
2722.65-0.2077T
kg.mm3
5140
Liquidus Temperature Spinel Equilibrium 0.04334{ l-exp[-5110.7(l/T-l/T')]~ Volume Fraction Mass Transfer Coefficient !Data validity range is 1i.om 850°C to 1200°C;(b)Tis in K
I
134
Environmental Issues and Waste Management Technologies VIII
The equilibrium fraction of spinel in MS-7glass was measured using quantitative Xray diffraction (XRD) as a function of temperature, glass composition (adding to or removing components from the MS-7baseline), and the partial pressure of oxygen (p02)?-~ The model has so far been applied only to data obtained from glass equilibrated with air. Thepoz field will be added to the model in the fbture because it affects both foaming and spinel precipitation in the melter. The main focus of the experimental study was on kinetics, that is, the measurements of the rates of nucleation, crystal growth, and crystal di~solution.6~~ It was discovered early on that spinel nucleation proceeds nearly instantaneou~ly.'~Moreover, the feed melting studies revealed that spinel crystals precipitate already from the primary (nitrate) melt.* These low-chromium crystals dissolved as the conversion progressed, but could also survive as seeds for crystals that form in the glass. The rate of nucleation was measured by counting the number of crystals per unit volume of glass. For the MS-7 baseline glass, the number of nuclei was low at the glassprocessing-temperature range, and strongly increased with decreasing temperature. The number density at the glass-processing temperature interval increased up to 4 orders of magnitude when nucleation agents were added to the glass as minor components. The most effective agents were noble metals! The addition of noble metals resulted in at least a tenfold decrease in the crystal size, and hence at least a hundredfold decrease in the rate of crystal settling. The rate of crystal growth and dissolution was determined by measuring crystal size in thin sections of quenched glass samples that were isothermally heat treated? A special methodology was developed for the settling-rate study to eliminate all possible sources of convection, such as bubbles and surface forces." The measured Stokes coefficient was only slightly different from that for hindered settling" of cubic particles in room-temperature liquids. Finally, the density of spinel sludge was determined both in laboratory crucibles and in a sample taken from a pilot-scale melter.12 For the purpose of mathematical modeling, it was necessary to formulate constitutive equations for spinel-glass-mixture equilibrium and kinetic behavior as h c t i o n s of melt temperature. The rate of spinel crystal growth and dissolution was represented by the Hixson-Crowell equation7in the form
da -=kH(Co -C) dt where a is the crystal size, COis the equilibrium solid-spine1 mass fraction, C is the solidspinel mass fraction, and kH is the mass-transfer coefficient. The mass-transfer coefficient varied with temperature according to the Arrhenius relationship shown in Table 11. MATHEMATICAL MODELING The GS-GFM code was applied to a HLW glass melter (the West Valley type) (Figure 1). The impact of the growing sludge layer on the macroscopic melt flow was omitted for simplicity. The distributions of glass velocities and temperatures in the melter were calculated using a 3-D mathematical model of glass flow, heat transfer, and Joulean heat generation in the melter space coupled through temperature- and compositiondependent properties of the glass. The procedure of control volumes was used. The model
Environmental Issues and Waste Management Technologies VIII
135
involved the kinetics of the growth and dissolution of spinel crystals, the class representation of spinel crystal-size distribution, and the algorithms for the moving and settling of spinel crystals inside the melter.
Figure 1. Stationary temperature and velocity fields in the cross section of a HLW melter Crystals were allowed to grow or dissolve according to Equation (1) and settle following the modified Stokes law. To represent the distribution of crystal sizes in each control volume, 20 classes of crystal sizes were used in calculations, described by their center (the mean size of the i-th size class, ai) and by their equivalent width (wi); thus,
C;=,wi=a,,.
x:=l
The volume fraction of crystals in the control volume was
C= Nv,a,?.The mass balance of the crystalline phase was used for each size class in the form:
where R(Nv,) is the interchange among classes due to size change, si is the volumetric generation of spinel crystals by nucleation, and w, = 0.2O5g(ps - pg)azrj’ is the Stokes velocity of spinel crystals in glass; here g is the gravity acceleration, TJ is the dynamic viscosity of the melt, and p, and pgare the densities of spinel crystals and glass. The thickness of the sludge layer on the melter bottom and the slant melter walls was calculated from the mass balance of spinel crystals in the regions adjacent to the bottom and from the measured concentration of spinel in the sludge. The thin-layer approximation was used. In each control volume above the bottom, the i-th size height, hi, was calculated for the time step At as hi = A t w s i . All crystals within hi fall fast enough to settle on the bottom during At. The settled fraction of crystals in the control volume is si = h i / k where Az is the height of the control volume. The height of the sludge layer is n
h =( k / ~ ) ~ s i N V , a , ? i=l
(3)
where V, is the volume fraction of spinel crystals in the sludge.
136
Environmental Issues and Waste Management Technologies VIII
Monodispersed spinel crystals entered the melt from the cold cap. In the reference case, the input crystals concentration was Cjpj = 110 kg/m3 and their size was aifl= 1 pm. The nucleation density of spinel was ns = exp(13.6224.01823T). The number of spinel crystals moving from one control volume to another in the region with T .c Tt was increased to match ns if the next control volume was cooler than ''7 ,and the current number of particles was lower than n,. New nuclei were generated in the size class of the smallest crystals. For each time step, the model performed the following sequence of operations: 1) Load the temperature and velocity fields from the GS-GFM. 2) Start crystal nucleation. 3) Obtain w,for each size class. 4) Account for growth and dissolution: da dt
[CO
= k(T)
-tN,.,a; i=l
1
(4)
5 ) Redistribute crystals into unified size classes. 6) Calculate crystal settling in control volumes above the bottom and slant walls. 7) Repeat Steps 2 through 6 to reach a stationary state.
RESULTS AND DISCUSSION Figure 1 displays the stationary temperature and velocity fields in the cross section through the melting space calculated by the GS-GFM. The melter produced a rather vigorous circulation flow, assuring a good mixing and homogenizing and a nearly uniform temperature field. The average temperature of the molten glass was 1104°C. The average glass velocity was approximately one order of magnitude higher compared to standard industrial glass fiunaces.
Figure 2. Spine1 concentration distribution
Environmental Issues and Waste Management Technologies VIII
137
The distribution of spinel crystals in the longitudinal and cross section for the calculated reference case is displayed in Figure 2. The fast flow of the glass melt resulted in a nearly uniform distribution of crystals in the melting space? Hence, the melter behaved as a fast (nearly ideal) mixer. The rate of growth of the sludge layer was computed as a function of the position in the melter? The sludge-layer thickness profile at the melter bottom after 30 h of melter performance is shown for the reference case in Figure 3. Note the increased layer thickness towards the melter outflow. The sludge layer grew slowly, only 0.04 &year. Sludge that thin is not expected to interfere with melter operation. Figure 4 presents the result of a parametric study conducted to investigate the influence of the input crystal size and the 'T on the sludge-layer growth. The dependence of the sludge-layer growth rate on the input size of spinel crystals has a power-law form with the exponent of 2.5 (Figure 4, left plot), which is higher than that which the Stokes equation would suggest. When the initial crystal size increased from 1 pm to 5 pm, the sludge-layer Figure 3. The thickness profile of deposited spinel Crystals rate increased from (the sludge layer) at the bottom after 30 h of HLW melter 0.04 to 2 performance (reference case) (for TL = 1078°C). Crystals larger than 10 pm produced a sludge layer several cm thick after one year of melter performance. With continuous operation, > 10-pm crystals would gradually obstruct the melter outflow. The right plot in Figure 4 shows that the sludge-layer thickness mildly grew with increasing . ' T All calculated data with the crystal size interval from 1 to 100 pm and'T interval from 950 to 1200°C can be approximated as
where v h = &/dt is the sludge-layer thickness growth rate, vho = 27.2 d y e a r , a0 is the initial crystal size, a0 = 100 pm, p = 6.30, and q = -2.96~10"I?. As this equation suggests (see also Figure 4), the Stokes exponent is a fimction of 7 ' . This function is approximated as linear in Equation (9,but is slightly nonlinear as the lack of fit of the 10pm line (in Figure 4) reveals. When a0 = 100 pm, vh is independent of . '7 The Stokes exponent increases as the TLdecreases, reaching the value of 2 at TL= 1179°C and 2.5 at TL = 1040°C. These results suggest that within broad ranges of variation, the 'T should not cause technological problems by its impact on spinel accumulation if the melter is a fast
138
Environmental Issues and Waste Management Technologies VIII
mixer. The most important parameter for the sludge-layer growth control is the size of spinel crystals. There are a number of additional applications for this model. The model can potenI00 tially be extended to incorporate the oxi10 dation-reduction equilibrium and the concentration and I sizes (including rates of growth) of gas bubbles. Melter 0.1 idling can also be simulated? The 0.01 model can be used to test different I 10 100 melter designs, the input size of spinel crystals [pm] effect of bubbling, melter operation parameters, and the impact of glass composition on spinel settling. Thus, melter design, melter operation, and glass formulation can be optimized. CONCLUSIONS The mathematical model presented m = enables calculating 0.01 I ; , , . , , , , 950 loo0 1050 I100 1150 I200 the distribution of spinel crystal conT' PCI centration within 7 = 1078°C the melt and the Figure 4. The effects of the initial size of crystals at ' (above) and of 7'~ with input crystal size as the parameter (below) evolution of the sludge-layer thick- on the growth rate of the sludge layer of deposited spinel crystals at the HLW melter bottom ness on the HLW glass melter bottom. It shows that fast-mixing melters keep a nearly uniform concentration and size of crystals during normal operation. Hence, keeping TL> 1050°C may not prevent accumulation of crystals in the melter whereas crystals may not settle in the I
Environmental Issues and Waste Management Technologies VIII
I
139
melter even if TL< 1050°C. The accumulation of solids in the melter is primarily determined by the initial crystal size. Nucleation agents (noble metals) keep crystals small in most HLWs, but are often absent in simulated wastes for experimental melter runs, thus exacerbating the potentially false TLconstraints. Without the seemingly unnecessary current TLconstraint, the HLW glass volume at Hanford can significantly decrease, with correspondingly significant cost savings. This model and the conclusions drawn from modeling should be verifiedhalidated with actual melter operation data. ACKNOWLEDGMENT Pacific Northwest National Laboratory is operated for the U.S. Department of Energy (DOE) by Battelle under Contract DE-AC06-76RLO 1830. This work was funded by the DOE through the Environmental Management Science Program. REFERENCES (‘>D.S.Kim and J.D. Vienna, “Influence of Glass property Restrictions on Hanford HLW Glass Volume,” Ceram. Tram 132, 105-115 (2002). (*)P. Schill, Calculation of Three-Dimensional Steady Flows and Temperature Using Multigrid Method,” Proceedings of the International Congress on Glass, Vol 29, 336343, Leningrad, 1989. (3)J. Matya, J. KlouEek, L. NEmec, and M. Trochta, “Spinel settling in HLW melters,” The gh International Conference Proceedings (ICEM’OI), Bruges, Belgium, 2001. (4)J. MatyhS, Description of the Behavior of Multitude Particles in Non-isothermal Convective Melting Space, PhD. Thesis, Laboratory of Inorganic Materials, Prague, Czech Republic, 2001. Schill and M. Trochta, “Advanced Mathematical Modeling of Special Glass Furnaces,” Proceedings of the 2002 GLASS ODYSSEY, 6‘h EGS Conference, Montpelier, 2002. (@P.Hrma and J. Alton, “Dissolution and Growth of Spinel Crystals in a High-Level Waste Glass,” The gh International Conference Proceedings (ICEM’OI), Bruges, Belgium, 2001. (’)J. Alton, T.J. Plaisted, and P. Hrma, ‘‘Spine1 Nucleation and Growth of Spinel Crystals in a Borosilicate Glass” accepted in . I Non-Cryst. Solids. (‘’P. Izak, P. Hrma, B.W. Arey, and T.J. Plaisted, “Effect of Batch Melting, Temperature History, and Minor Component Addition on Spinel Crystallization in High-Level Waste Glass,” J. Non-Cryst. Solids 289,17-29 (2001). (’)P. Hrma, P. Izak, J.D. Vienna, G.M. Irwin and M-L. Thomas, “Partial Molar Liquidus Temperatures of Multivalent Elements in Multicomponent Borosilicate Glass,”Phys. Chem. Glasses 43 (2) 128-136 (2002).
( O)J. Klouzek, J. Alton, T.J. Plaisted, and P. Hrma, “Crucible Study of Spinel Settling in Hip-Level Waste Glass,” Ceram. Tram. 119,301-308 (2001). (l ’E. Barnea, and J. Mizrahi, “A Generalized Approach to the Fluid Dynamics of Particulate Systems, Part 1,” Chem. Eng. J. 5, 171-189 (1973). (12)M.Jiricka and P. Hrma, “Chemical and Mechanical Properties of Spinel Sludge in High-Level Waste Glass,” Ceramic-Silikaty 46 (1), 1-7 (2002) (13)J.G.Reynolds and P. Hrma, “The Kinetics of Spinel Crystallization from a HighLevel Waste Glass,” Mat. Res. Soc. Symp.Proc. 465,261-268 (1997).
140
Environmental Issues and Waste Management Technologies VIII
RUTHENIUM - SPINEL INTERACTION IN A MODEL HIGH-LEVEL WASTE (HLW) GLASS T. M. Willwater', J. V. Crum, S. M. Goodwin, and S. K. Sundaram Pacific Northwest National Laboratory Richland, WA 99352 ABSTRACT Noble metals (for example ruthenium) act as nucleation sites for the precipitation of spinel (crystalline) phases. The noble metals along with the spinel phases will settle to the bottom of the melter causing local viscosity increase, power fluctuations, and even potentially shorting of electrodes leading to premature melter failure. We studied the partitioning of ruthenium in a model high-level waste glass. Ruthenium oxide was chosen as this was predominantly found in melter tests with feeds containing noble metals at the bottom of the melter. A doping of 10 wt % of ruthenium oxide was selected to simulate somewhat the conditions at the bottom of the melter where noble metals accumulate. The heat-treatment conditions (temperature and duration) were chosen from reported literature, such that large crystals of trevorite (NiFe204) were formed in the glass. The spinel-glass interface was characterized using scanning electron microscopy (SEM) and microprobe characterization. SEM results showed the crystals distributed in the glass matrix. Microprobe measured the ruthenium concentration across and around the spinel-glass interfaces. The results did not show significant partition of ruthenium in the spinel. INTRODUCTION Precipitation and settling of noble metals (for example, rhodium, ruthenium, and palladium) in high-level waste (HLW) glass melts processed in joule-heated melters can lead to operational difficulties. In addition, the noble metals act as nucleation sites for the precipitation and growth of spinel (crystalline) phases, which in turn will settle to the bottom of the melter and cause the viscosity of the melt to increase in that region. Summer Intern,Pre-Service Teacher (PST) program, Pima Community College, AZ To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management TechnologiesVIII
141
Spine1 forms during the batch-melting reactions [l] in Fe- and Ni-containing HLW glass batch, dissolves the batch melts, precipitates from the HLW glass again as the melt temperature cools below the liquidus temperature, TL [2-41. Once crystals are formed, they tend to settle down in a melter, which could potentially lead to power fluctuations, current excursions, enhanced electrode corrosion, and even shorting of electrodes causing premature melter failure. It is important to understand the solubility of noble metals as well as their partitioning between the crystal phase formed and the glass. Hrma and coworkers [5-81 have extensively studied the formation, dissolution, and growth of spinel crystals in a model borosilicate glass system (MS-7) (base composition as well as compositions with trace amounts of noble metals). These works have used the Hixson-Corwell equation (based on Fick’s Law) to determine mass-transfer coefficients for dissolution and growth and found that these coefficients were found to fit one Arrehenius function of temperature. Three major melter test campaigns testing noble metals have been completed in the past: 1) PNNL test, 2) German melter test, and 3) Integrated DWPF (Defense Waste Processing Facility) Melter System (IDMS). Noble metals have been included in glass development studies since some of the earliest waste solidification and vitrification work at PNNL [9]. The insolubility of noble metals in glasses was observed at those early stages and was also known from the literature; however, the effect this insolubility could have on melter operation was not known. Early works in 1970s included crucible and laboratory-scale tests. Since then, five major studies, gradient furnace testing (GFT), research scale melter (RSM) testing, engineering scale melter (ESM) testing, modeling, and engineering analysis, were completed at PNNL. German melter tests (1980s and 1990s) showed that the accumulation of noble metals could be greatly decreased by increasing the slope of the melter floor. The IDMS was designed as a pilotscale test facility for the DWPF. Before testing with the IDMS, two short-term noble metals campaigns with a 1/1OOth scale mini-melter revealed a need for extended noble metals testing. Numerous test runs with the IDMS melter addressed the designs of the DWPF feed preparation system, offgas system, and the melter itself. The IDMS engineering-scale melter is prototypic of the DWPF melter, designed with a melt surface area of 0.29 m2 (approximately 1/9th of the DWPF surface area), and a melt volume of 0.20 m3. The IDMS has conducted a total of 16 noble metal-related runs with four different types of wastes sludges containing various amounts of noble metals [lO-121. All these melter tests results clearly indicate that the most commonly found species is high concentrations of Ru02 in the melter and Ru has always been found in association with Ru02 and other noble metals and spinels. Interaction of noble metals and spinel crystal at high noble metal concentrations has not been systematically studied. The objective of the present study is to generate preliminary data addressing this issue.
142
Environmental Issues and Waste Management Technologies VIII
EXPERIMENTAL METHODS AND MATERIALS A model borosilicate glass (MS-7) is chosen as this is the most studied composition used for investigation of kinetics of growth of spinel crystals. Trevorite (NiFezO4) is only dominant crystalline phase formed in this composition that is shown in Table I. Table I. Composition of MS-7 glass Oxide Glass Comp, Source wt% chemical
Na20 NiO Si02
8.00 7.00 0.30 1 1.50 4.50 0.60 0.50 15.30 1 .oo 45.30
Total
100.00
A1203 B203 Cr203 Fe203 Li20 MgO
MnO
zro2
6.00
Total
A1203 H3B03 Cr203 Fe203 Li2C03 MgO
8.00 12.43 0.30 1 1.50 11.13
MnO
0.50 41.59 1 .oo 45.30 6.00 138.35
Na2C03 NiO Si02
zroz
0.60
By following the standard glass melting procedure, a 500g sample of the base MS-7 glass was first prepared then milled in an agate mill for 6 minutes and eventually put in the furnace at 1250" C where it was maintained for one hour. The melt was cooled and the glass was milled again in a tungsten carbide mill for 6 minutes. It was then re-melted again at 1250" C for an hour. The cooled glass was ground into a fine powder with the tungsten carbide mill for 4 minutes. At this point, small batches of 10 grams each were prepared. 10 wt% of Ru02 was now added to each of the samples. The samples were then placed in 1 x 1 x 1 cm3 platinum-gold crucibles. Thee samples were then put heat-treated at 1200°C for 30 minutes and the temperature was then decreased to 800°C. The heat treatment conditions were chosen from reported literature [5]. The conditions corresponded to the highest linear spinel growth rate (21.9 pnh) reported in the base MS-7. The samples were then sectioned and polished for further characterization. Xray diffraction 0 ) was used to confirm the primary crystalline phase was trevorite. Optical microscopy showed significant amount of settling of large trevorite crystals at the bottom of the crucible. The spinel-glass interfaces in the sample near the bottom of the crucible region were characterized using scanning
Environmental Issues and Waste Management TechnologiesVIII
143
electron microscopy (SEM - JEOL JSM-5900) and microprobe (JEOL JXA- 8600 Superprobe).
RESULTS AND DISCUSSION The largest crystals (about 50 pn) were found near the bottom of where the crucible. Figure 1 shows the secondary electron micrograph of a representative glass-spine1 interface in the sample with 1 0 WWOof Ru02 heat-treated at 800°C for 7 hours. Prominent trevorite crystals are surrounded by Ru/RuO2-rich particulates, confinning glass saturated with excess Ru. Figure 2 shows the microprobe data. The inset shows the back-scattered electron image of the same location shown in Figure 1. The burning marks shown in the image are the points where the Ru concentration was measured. The spots are 1 0 ym apart so they do not interfere with the neighboring spots significantly. The Ru concentration is not measurable from points 1 to 5. Then, it starts to increase steadily. The point 11 is close to a crystal as seen in Figure 2. The increase can not be attributed to the crystal as the points 6 - 1 0 are not on the crystal. Additionally, the Z-contrast shows a spongy white Ru-containing phase in this region.
Figure 1. Secondary Electron Micrograph of the Spine1- Glass Interface (1 0 wt% RuO~,800°C,7 hours)
144
Environmental Issues and Waste Management Technologies VIII
Figure 2. Ruthenium Concentration at the Points Marked on the Back-scattered Electron Micrograph (Inset) of the Region shown in Figure 1 Figure 3 shows the secondary electron micrograph of a representative glassspinel interface in the sample with 10 wt% of Ru02 heat-treated at 800°C for 24 hours. The spinel-glass interfaces features similar to the 7 hours sample. The size of crystals has not changed significantly. Figure 4 shows the microprobe data. The inset shows the back-scattered electron image of the same location shown in Figure 3. The Ru concentration is not measurable from points 1 to 3. Then, it starts to increase through points 4-10 with points 5-7 showing not measurable concentration in the crystal seen in the Figure 2. The point 11 is at the other end of the crystal that is close to the region populated by Ru-rich particdates region. The increase can be attributed to dissolution of Ru into the spinel crystal. A systematic evaluation is proposed to determine distribution of Ru as a function of glass chemistry to establish the mechanism of partition of Ru in spinel crystals.
Environmental Issues and Waste Management Technologies VIII
145
Figure 3. Secondary Electron Micrograph of the Spine1- Glass Interface (10 wt% RuO~,800°C, 24 hours)
Figure 4. Ruthenium Concentration at the Points Marked on the Back-scattered Electron Micrograph (Inset) of the Region shown in Figure 3
146
Environmental Issues and Waste Management Technologies VIII
CONCLUSION Preliminary data indicate increasing Ru concentration in spine1 crytal formed at 800°C for 24 hours in MS-7 with 10 wt% of Ru02. Further data generation and analysis will be needed to establish a partitioning mechanism. ACKNOWLEDGEMENTS We acknowledge United States Department of Energy (DOE) - Office of Science for support to TMW under the Community College Initiative (CCI) Program and Royace Aikin, Science Education Specialist and CCI manager, and his Assistant Dale Johns, for all their support. PNNL is operated by Battelle Memorial Institute for the U.S. Department of Energy under contract DE-AC0676RtO 1830. REFERENCES 1. P. I&, P. Hrma, M. J. Schweiger in Nuclear Site Remediation, Editors: P. G. Eller and W. R. Heneman, ACS Symposium Series, 778, p. 314, American Chemical Society, Washington, DC, 2000. 2. P. Hrma and J. D. Vienna in Proceedings of Waste Management 00, Tucson, AS, 2000, CD-ROM. 3. M. Mika, M. J. Schweiger, P. Hrma in ScientiJc Basis for Nuclear Waste Managementz, Editor: 1. R. Triay, 465, p. 71, Materials Research Society, Warrendale, PA, USA, 1997. 4. P. Hrma, J. D. Vienna, J. V. Crum, G. F. Piepel, M. Mike in Scientijk Basis for Nuclear Waste Management ZXI.1, Editors: R. W. Smith and D. W. Shoesmith, 608, p. 67 1, Materials Research Society, Warrendale, PA, USA, 2000. 5. J. Alton, T. Plaisted, P. Hrma, Dissolution and Growth of Spinel Crystals in a Borosilicate Glass, J. Non-Crystal. Soli&, 311,24-35,2002. 6. J. Klouiek, J. Alton, P. Hrma, T. Plaisted in Ceramic Transactions, Editors: D. R. Spearing, G. L. Smith, and R. L. Putnam, 119, p. 301, American Ceramic Society, Westerville, OH, USA, 2001. 7. T. Plaisted, J. Alton, B. Wilson, P. Hrma in Ceramic Transactions, Editors: D. R. Spearing, G. L,Smith, and R. L. Putnam, 119, p. 291, American Ceramic Society, Westerville, OH, USA, 2001. 8. T. Plaisted, F. MO,C. Young, P. h a in Ceramic Transactions, Editors: D. R. Spearing, G. L. Smith, and R. L. Putnam, 119, p. 3 17, American Ceramic Society, Westerville, OH, USA, 2001. 9. S. IS.Sundaram and J. M. Perez, Noble Metals and Spinel Settling in High Level Waste Glass Melters, PNNL-13347, September 2000.
Environmental Issues and Waste Management Technologies VIII
147
10. N. D. Hutson, J. R. Zamecnik, M. E. Smith, D. H. Miller, and J.A. Ritter in Integrated D WPF Melter System (IDMS) Campaign Report: The First Two Noble Metals Operations (U). WSRC-TR-9 1-400, Defense Waste Processing Technology, Savannah River Laboratory, Aiken, SC, 1991. 11. N. D. Hutson in Integrated D WPF Melter System (IDMS) Campaign Report: Hanford Waste Vitrijkation Plant (HWVP) Process Demonstration (v). WSRC-TR-92-0403, Rev. 1,Westinghouse Savannah River Company, Savannah River Technology Center, Aiken, SC, USA, 1992; N.D. Hutson and M. E. Smith, The Behavior and Effects of the Noble Metals in the DWPF Melter System in Proceedings of the High Level Radioactive Waste Management Conference,American Nuclear Society, La Grange Park, Illinois. 1:541-548, 1992. 12. N. D. Hutson in IDMS Task Summary Report Part 1: The Behavior and Efects of the Noble Metals in the DWPF Melter System. WSRC-TR-93-0458, Savannah River Technology Center, Aiken, SC, USA, 1993.
148
Environmental Issues and Waste Management Technologies VIII
Glass Formulation and Testing
This page intentionally left blank
INTERIM MODELS DEVELOPED TO PREDICT KEY HANFORD WASTE GLASS PROPERTIES USING COMPOSITION John D. Vienna, Dong-Sang Kim, and Pave1 Hrma Pacific Northwest National Laboratory, Richland, WA 99352 ABSTRACT Over the past several years the amount of waste glass property data available in the open literature has increased markedly. We have compiled the data from over 2000 glass compositions, evaluated the data for consistency, and fit glass property models to portions of this database [13. The properties modeled include normalized releases of boron (rg), sodium (r& and lithium (rLi) from glass exposed to the product consistency test (PCT) [2], liquidus temperature (TL)of glasses in the spine1 and zircon primary phase field, viscosity (q) at 1150°C (q1150) and as a function of temperature (q& and molar volume (V). These models were compared to some of the previously available models and were found to predict the properties of glasses not used in model fitting better and covered broader glass composition regions than the previous ones. This paper summarizes the data collected and the models that resulted from this effort. INTRODUCTION Efforts are being made to increase the efficiency and decrease the cost of vitrifying radioactive waste stored in tanks at U.S. Department of Energy(D0E) waste sites. The compositions of acceptable and processable higklevel waste (HLW) and low-activity waste (LAW) glasses need to be optimized to minimize the wastexorm volume and, hence, save cost. Glass composition and associated properties from glasses tested at Pacific Northwest National Laboratory, West Valley Demonstration Project, Savannah River Technology Center, Vitreous State Laboratory at Catholic University of America, Idaho National Engineering and Environmental Laboratory, and several other institutions were reviewed and compiled into a single database. This database, although not comprehensive, represents a large fraction of data on waste-glass compositions and properties that were available at this tine. The compositions of glasses in this database were converted to mole fiactions of oxides (and elements in the case of TLmodels) using standard techniques. The compositions were screened for applicability to immobilization of Hanford HLW andor LAW. These data were then used to fit the composition parameters or coefficients in glass property models. The models were validated using subsets of the data not used in their development and the validation results and composition ranges of validity were compared to a number of previously reported glass property models including those reported by Hrma et al. [3,4] along with others. Due to space limitations, the
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management TechnologiesVIII
151
composition region of validity for these models were not included in the paper but can be found in the detailed report [13. RESULTS Product Consistency Test Response Normalized ByLi, and Na releases in the PCT are calculated with the formula: N
h [ r j (glrn2>]= C r j i x i i=l
where,j is the element released (B, Li, and Na), i is oxide component, N is number of N
components, xi is the i-th component mole fraction where
xi = 1,and rji are the
i=l
coefficients listed in Table 1. The basis for this model form has been frequently published (see, for example [S]). Also listed in the table are the summary statistics of the model fits including R2 (the fraction of variation of ln[q] accounted for by the model), R2,aj (adjusted for the number of coefficients), R 2 d (the R2that would be calculated for each glass if it were removed from the model, the model fit to remaining glasses), s (root mean square error), number of glasses, and minimum, maximum, and mean of response. Generally, the R2 values are lower than those from models reported earlier. However, these models were better able to predict PCT responses of glasses not used to fit the model than any of the previous models compared. It is not surprisingthat the R2values are low, as, ln[ri] is not generally linear with composition except over small composition regions. Table 1. Coefficients for PCT Response
* For those components not listed and those listed with “---” as a coefficient, the ‘cothers”coefficient should be used.
152
Environmental Issues and Waste Management Technologies VIII
Viscosity The qllso is calculated according to the equation:
where, hi is the i-th component coefficient listed in Table 2. This, linear, model form has been shown to be highly successful in modeling of q with composition. As can be seen by the R2,97% of the variation in ln[qllso]is accounted for by this simple linear approximation. The generally accepted relationship between q and absolute temperature (T) is given by: ln[q] = C+D/(T-To). However, over narrow ranges of T, ln[q] is nearly linear with 1IT (e.g., ln[rlj = A+B/T) and the coefficients describing these temperature effects are known to vary linearly with composition. Since a majority of the qT data is over a suficiently narrow temperature range to be easily approximated with the linear relationship and that relationship contains only two parameters that must be fit to composition, q~is represented by Equation (3) in this study, where Ai and Bi are the coefficients listed in Table 2. This model describes nearly 98% of the variation in data.
Table 2. Coefficients for Viscosity
* LN203 is the combined lanthanides and yttrium oxides.
I'
I
Liquidus Temperature The TL models were developed based on the work of Vienna et al. [q. As a thermodynamic quantity, TL can be related to the state functions according to:
Environmental Issues and Waste Management Technologies VIII
153
-AG~ TL =RlnK
(4)
where A d is the free energy of formation of the crystalline phase from the melt, and K is the reaction constant. In simple systems, such as crystallization of X Y from solution, K is taken as the inverse of the product of X and Y concentrations in solution. However, for crystallization of crystals that are generally solid solutions from multicomponent waste glass melts, the value of K is a significantly more complicated function of composition which includes activities of components in the crystal solid solution and their activities in the melt. As A d i s also a function of composition, the quantity on the right hand side of Equation (4) has been empirically fit to composition. Previous studies [3,4,6] have shown this factor to be linear with composition, having coefficients (3;) similar to ci,hi,Aj, and Bj, discussed above. A TL model which accounts for the effect of component concentrations in the melt on the activities of spinel components in the melt using ion potential (Pi)was developed [q. Using these relationships, models for TL of waste glasses in the spinel ([Fe,Mn,Ni][Fe,Cr,Mn]~O~) and zircon (ZrSiO,) primary phase fields were fit to the appropriate subsets of experimental data. The 7"'s of melts in the spinel primary phase field are calculated with the formula:
where, i represents the electropositive-element components, Pi is the ratio of the i-th component valence to crystal radius reported by Shannon [7J, and Z,tjony @ion, tcov, and Ocov are coefficients reported in Table 3. In this model, components are broken into three groups represented by the three terms in Equation (5). The first group includes the major spinel components minus Fe, the second group includes the alkali and alkaline-earth components, and the last group includes all components not in the first two groups. This model fits the data very well with roughly 90% of the variation in TL explained by the model. Although the previously published model [qshowed slightly better summary statistics, the advantage of this model is a better estimate of Mn effects on TLand a significantly broader composition region of model applicability. Likewise this model was fitted to TL data in the zircon primary phase field according to:
where the only component in the first term is Zr. The coefficients and summary statistics for the TL model in the zircon primary phase field are also listed in Table 3. This model also fits the data well, explaining roughly 87% of the variation in TLdata. Insufficient data was available to fit TL - composition models in other primary phase fields of interest to waste immobilization.
154
Environmental Issues and Waste Management Technologies VIII
Table 3. Coefficients for TL
Molar Volume The database includes density (p) data on glasses at room temperature. In an ideal mixture, the volume of the mixture is given by the sum of partial volumes of the mixture constituents. Clearly glass is not an ideal mixture, however, a model based on volume is more likely to be linear than one based on p. Therefore, the molar volume ( y) is fit to data according to:
v = Cv,x, N
i=l
(7)
where yl. is the partial molar volume of the i-th component in glass. The yl. values are listed in Table 4. Density (p)is then calculated according to:
where Mi is the molecular mass of the i-th component. The experimental data was suEcient to estimate 6 for 18 glass components. However, it is also possible to estimate these volumes using standard ionic radii. Through the use of Shannon’s crystal radii [7], the of all 56 components found in the database were estimated according to:
zi + br; vi = aro3 2
Environmental Issues and Waste Management Technologies VIII
(9)
155
where vi is the apparent yl. per cation in the component, ro is the radius of oxygen, ri is the cation radius, Zj is the cation valence, and a and b are empirically fit parameters. Table 4 lists the resulting yl. values for the 56 component model along with summary statistics of both models. Table 4. Partial Molar Volumes (cm3) . 56-Comp. Vi 1 Component I 18-Comp. V, I 56-Comp. yi 46.149 I 120.000 30.048 18.866 15.214 7.526
ItLi20 Na20
NiO
SiOz
SrO Ti02 zro2 ZnO BeO Bi203
19.943 19.834 12.668 25.316 17.611 17.964 27.081 15.069
CdO
coo
cszo 40.000 156.000 122.250 1 43.OOO ' 47.000 l 28.000 128.800 52.000 3 1.700 i 35.000
R'dj
V'
Rz p' R'adi
pw
365 0.949 0.946 0.921 0.918
365 0.946 0.937 0.9 17 0.902
* - The R' and RLadjvalues were calculated
on both the molar volume (V) and density (p)
bases.
~
CONCLUSIONS A series of models were developed and validated for use in predicting key waste glass properties as functions of composition. These models included models forre, rN,, rLi, q1150, q ~ V/p, , and''7 in the spine1and zircon primary phase fields. The fit statistics of
156
Environmental Issues and Waste Management Technologies VIII
these models suggest that they are roughly as good as previously published models, however, these models cover broader composition regions and were able to better estimate data not used in model fitting. We recommend that these models be applied (within their appropriate composition regions of validity) for the purpose of rough property estimation over relatively broad composition regions. For more precise property estimation in relatively small composition regions, new models should be fitted to data specifically developed in those composition regions. ACKNOWLEDGEMENTS The authors are grateful to Carol Jantzen (SRTC) and Ian Pegg (CUA) for supplying their data for inclusion in the database; Scott Cooley (PNNL), Steve Lambert (NHC), David Peeler (SRTC), Greg Piepel (PNNL), and Jacob Reynolds (WGI-WTP) for careful review of this work and helpful comments; and Bill Holtzscheiter (SRTC) and Ken Gasper (CHG) for programmatic guidance and support. This work was funded by the DOE Office of Science and Technology under the Tanks Focus Area Immobilization Program. Pacific Northwest National Laboratory is operated for the U. S. Department of Energy by Battelle Memorial Institute under Contract DE-AC06-76RLO 1830. REFERENCES [11 JD Vienna, DS Kim, and P Hrma, Database and Interim Glass Property Models for Hanford HL W and LA W Glasses, PNNL- 14060, Pacific Northwest National Laboratory, Richland, WA (2002). [2] ASTM International, Standard Test Methodsfor Determining Chemical Durabiliq of Nuclear, Hazardous, and Mixed Waste Glasses and Multiphase Glass Ceramics: The Product Consistency Test (PClJ,ASTM C 1285-02, West Conshohoken, PA (2002). [3] P Hrma, GF Piepel, JD Vienna, SK Cooley, DS Kim, €URussell, Database and Interim Glass Property Modelsfor Hanford HL W Glasses, PNNL 13573, Pacific Northwest National Laboratory, Richland, WA (200 1). [4] P Hrma, GF Piepel, MJ Schweiger, DE Smith, DS Kim, PE Redgate, JD Vienna, CA LoPresti, DB Simpson, DK Peeler, and MH Langowski, Property/Composition Relationshipsfor Hanford High-Level Waste Glasses Melting at I 15OoC,PNL- 10359, Pacific Northwest Laboratory, Richland, WA (1 994). [5] CM Jantzen, ‘‘ThermodynamicApproach to Glass Corrosion,” in Corrosion of Glass, Ceramics, and Ceramic Superconductors,eds., DE Clark and BK Zoitos, Noyes Publications, Park Ridge, NJ (1992). [6] JD Vienna, P Hrma, JV Crum, and M Mika, “Liquidus Temperature Composition Model for Multi-Component Glasses in the Fe, CryNi, and Mn Spine1 Primary Phase Field,” . I Non-Cryst. Sol.,292: 1-24 (200 1). [7] RD Shannon, “Revised Effective Ionic Radii and Systematic Study of Interatomic Distances in Halides and Chalcogenides,”Acta Cryst. A32:75 1-767 (1976).
Environmental Issues and Waste Management Technologies VIII
157
This page intentionally left blank
RELATIONSHIP BETWEEN LIQUDUS TEMPERATURE AND SOLUBILITY Pave1 Hrma and John D.Vienna Pacific Northwest National Laboratory P.O. BOX999, MS: K6-24 Richland, WA 99352 ABSTRACT The literature on high-level waste glass crystallization uses three basic ways of organizing data: 1) solubilities of sparsely soluble glass components are plotted as functions of temperature; 2) liquidus temperature (T') of glass is expressed as a b c t i o n of glass composition; and 3) fractions of crystalline phases at equilibrium with glass are measured as a function of temperature. To make the results mathematically tractable, the response functions are constructed by fitting simple mathematical expressions to data. The relationship between solubility-based and T'-based formulae is discussed. INTRODUCTION It is common in materials science that material properties, such as viscosity or heat conductivity, are represented as functions of thermodynamic state variables, i.e., temperature (9, pressure, and composition. These functional relationships are called response functions. A special class of properties, such as liquidus temperature (TL), describes the state of the material at equilibrium. The purpose of this contribution is to review basic concepts used to characterize the phase behavior of high-level waste (HLW) glasses, including solubility limits and solubility products, and to discuss their relationship to .'7 SOLUBILITY LIMIT The solubility limit of an oxide (component A) in a HLW glass melt at a given temperature is commonly determined by adding component A to the glass until a solid phase appears at equilibrium. However, caution is needed when using this term. If the solid phase that forms on adding A to the mixture is a compound AB of A with component Bywe should more correctly speak about the solubility limit of AB, not A. For example, when Cr2O3 concentration is systematically increased in a HLW glass containing NiO and FezO3, the solid phase that first appears can be eskolaite (Cr2O3) or spinel, a solid solution of chromite (FeCr204) and nichromite (NiCr204) with magnetite (Fe304). When the primary phase is eskolaite, the solubility limit is that of Cr2O3 in that particular glass. If the primary phase is spinel, we can still talk about Cr203 solubility, but the solubility limit is that of spinel. Chromium, which exists in HLW glass in two dominant valences, as Cr(II1) and Cr(VI), brings an additional level of complexity to the phase behavior of HLW glass. As To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
159
Figure 1 shows, in a high-basicity HLW glass with 1.50 mass% Fe203 and 0.15 mass% NiO, the primary phase at temperatures above 1000°C is eskolaite.' Below 1000°C, a liquid chromate precipitates as a separate phase. Spinel forms only at temperatures below 850°C. In the eskolaite region, the Cr203 solubility limit is lowest at 1200°C (2.3 mass%) and increases at temperatures higher or lower than 1200°C. Around lOOO"C, the Cr2O3 fraction dissolved in this glass could be as high as 3.5 mass %. It is advantageous to express 1500 glass composition in terms of ,400 associate species. The concept of glass structure composed of ~ 1 3 ~ associate species, usually identical g12W 2 to crystalline phases, was proved useful in understanding phase k 1 1 ~ behavio? and has been ,!?,m experimentally evidenced by spectroscopic studies? For example, postulating the existence 8oo of dissolved nichromite, we can 0 0.5 1 cr203tAisaCtion &ass%f5 3.5 write Figure 1. Equilibrium phase diagram' for the Cr2O3 in a HLW glass with low content of NiO Cr203(2)+NiO(Z) * NiCr204(Z) (1) and Fe203 where the symbol ( I ) indicates a species dissolved in glass. Precipitation of solid nichromite can be described as a reaction
70-
8 a c1
cu
60 ;,
4-
O 3oIt
s .2 20-
9
160
10;,
*Fe "i '0 o
-
I
~
A oNi+S+ik l ~
precipitate from glass are seen in Figure 2. Spinelforming oxides also participate in a number of other associate species, including acmite (NaFeSi206) that forms
Environmental Issues and Waste Management Technologies VIII
a solid solution, or a segregated liquid. Accepting such hard limits for HLW glass formulation leads to unnecessary low waste loadings. LIQUIDUS TEMPERATURE The TL versus composition function has generally a tractable form within a single primary phase field.' As any mixture property, TLcan be expressed in the form of partial properties, i.e., 6 N
T, = zTLixi
(3)
i=l
where TLj is the i* component partial molar TLand xi is the i* component mole fraction. The T'is are generally functions of composition. Fortunately, the NiO MgO ranges of concentrations of individual components in HLW glasses are usually sufficiently narrow to allow us to approximate Q 1050. TLjS as constants. Thus, the TL hypersurface within each primary phase field is approximated as a flat 950 t hyperplane. An example of the 4.w -0,OZ 0 am 0-04 4ptW=-l3 multiple slopes of such a hyperplane is shown in Figure 3 that displays the effect of a number of glass F i p e 3 . SpinelprimaryphaseTLaSafunction of addition (&j) of oxides to a baseline glass7 components added to or removed fiom MS-7 glass7 (containing, in mass%, 0.3 Cr203, 11.5 Fe2O3, and 0.95 NiO). t
I
i
1
SOLUBILITY PRODUCT Jantzen* suggested estimating TL using free energies of formation of selected crystalline phases. Plodinec' proposed a solubility product model for TL when the primary phase was either an associate species, such as nichromite, or an end-component, such as Gd2O3. Expressing the equilibrium constants of the corresponding reactions, such as (1) and (2), and approximating activities as concentrations, one obtains the relationship xCr,&xNio= A exp(-B / T)
(4)
where A and B are constants for a given glass. At T = TL, xCrzq and xNi0are equal to Crz03 and NiO fractions in the original glass. To compare Equation (4) with (3), we linearize Equation (4) by resolving exp(-B/TL) around a baseline temperature, TB. On neglecting small, higher-order terms (assuming that TL- TB<< TBand TdB << l), we get
Environmental Issues and Waste Management Technologies VIII
161
where Tz= (T; /AB)exp(B/T,). For a fair comparison, we rewrite Equation (3) as
where TRrepresents the rest of the sum. In Equation (5), the effects of Cr203 and NiO are represented by a single cross-product term, whereas in Equation (6) by two linear terms. 1350 Plodinec' succeeded in fitting Equation (4) to data from a 1300 narrow composition region. As 1250 Figure 4 demonstrates, Equation 1200 (4) with A and B independent of 1150 composition cannot be used for B 1100 large HLW composition regions, -a 1050 such as that at Hanford, because it ignores the effects of key glass 3 loo0 components. Solubility product is 950 a useful concept for solutions 900 with a single solvent, such as 850 H20, but not for HLW glass 850 950 1050 1150 1250 135 melts, where the solvent varies Measured TL("C) widely in composition. To illustrate this point, Figure 4. Calculated versus measured TLfor spinelconsider MS-7 glass with TL = precipitating Hanford glasses using E uation (4) 1078°C. Using the solubility with optimized coefficients19 language, this means that the Cr2O3 solubility at 1078°C equals the Cr2O3 fraction in this glass (0.30 mass%). At 1050°C, the solubility of Cr203 drops to 0.16 mass%. The remaining Cr203 precipitates in the form of spinel. If the melter does not accept glass that forms crystals at T 2 1050"C, the waste loading in MS-7 would have to be decreased by 47% based on the solubility product model. Fortunately, 0.3 mass% Cr203 can be dissolved in MS-7 at 1050°C if approximately 1 mass% Na20 is added to the glass. Equation (3) allows an optimized formulation of glass composition that increases the dissolved fraction of Cr2O3 to its maximum level. In the spinel primary phase field, the spinel-forming reaction significantly decreases Cr2O3 solubility in HLW glass because of the low solubility limit of spinel. A more complex behavior, such as that shown in Figure 1, can be described by Equation (3) combined with an equation representing the oxidation-reduction equilibrium.
5 h
/
Y
EQUILIBRIUM CRYSTALLINITY AT TEMPERATURES BELOW TL The linear form of Equation (3) simplifies the optimization of HLW glass for processing technologies that constrain TL. However, some advanced melters can handle glass containing up to several per cent crystals, especially when the crystals are small (1 to 5 pm). The glass processing constraint is no longer expressed in terms of TLfor such melters, but in terms of allowable fraction and size of crystals in the melt. These parameters depend on the crystallization kinetics, mainly the rate of nucleation, growth
162
Environmental Issues and Waste Management Technologies VIII
and dissolution, though the equilibrium fraction of crystals below TLis an essential aspect of the constraint. To generate data on the fraction of crystalline phase below TL requires precise measurements, using image analysis or quantitative X-ray difiaction anal~sis.~ The sample must be at phase equilibrium at the heat-treatment temperature and in case of multivalent species also at redox equilibrium with the appropriate atmosphere. Crystal settling and melt volatilization must be prevented fiom influencing the results. For a quick estimate, the fraction of crystals below TLcan be calculated from TLdata, especially when the primary phase is the only solid that precipitates." In this case, a temperature below TLis the liquidus temperature of the residual glass (TLJ at equilibrium with the primary phase. Thus, T = T L ~and, , by Equation (l),
By the i-the component mass balance,
where Xr,i is the ?' component mole fraction in residual glass, x, is the crystalline phase mole fraction in the mixture, and xc,i is the i* component mole fraction in the crystalline phase. Joining Equations (7) and (8),
where T, = X Z I T L i x c j (for pure trevorite, T, = 4435°C). The RHS approximation is permissible if x, << 1, which is usually the case. Thus, x,
=-AT
Tc
where AT= TL- T L=~'T - T is the temperature difference below TL. It is important to bear in mind that composition of the crystalline phase, such as spinel, often changes with temperature and thus the relationship between x, and T is generally nonlinear. The value of T, is extremely sensitive to chromium and nickel content in spinel because the TLi values are large for these components. Unfortunately, the usefulness of Equation (10) is limited by the difficulty in determining spinel composition with a sufficient accuracy. A different approach is based on Equation (2), which can be written for any primary phase. At equilibrium (taking the solid as the standard state), lna = const. - AH,/RT, where a is the activity of dissolved primary phase at equilibrium with the solid (e.g., spinel), and AH is the dissolution enthalpy (which is assumed to be independent of T for small
Environmental Issues and Waste Management Technologies VIII
163
undercoolings). Since a = atotat TL, where atotis the value of a when all crystals are dissolved, we have ln(a/afol)= - B@T- l/T'), where BL = M / R . Assuming that activity coefficients change little with the primary-phase concentration in the melt (both solid and dissolved), we can Write a/atot -x,)/xc,tot. Expressing x,, we finally obtain
Here xc,rorcorresponds to the total fraction of the primary phase in the 3500 melt. Equation (11) is an 0 3000 approximation applicable to a narrow @ 2500 composition region at which this =I equation is fit to data (usually a few 5 2000 Q) mass% of crystals). Hence, AHis not 1500 a spinel heat of fusion and the E 1000 temperature corresponding in 500 Equation (11) to x, = 1 is not a 0 melting point of spinel. This is -,.-. 0 0.2 0.4 0.6 0.8 1 schematically illustrated in Figure 5, Spine1 Mass Fraction representing melt G as a binary Figure 5 A schematic of glass-spine1 binary mixture of spinel and the remaining components. Line 1,which represents Equation (1 l), and line 2, which approximates the real ?'L, have a common segment on a narrow interval of spinel fraction. Only on this interval, Equation (11) describes the actual behavior. For MS-7 glass, xc,rof= 8.25 mass%! It is interesting to see whether this fraction of spinel is compatible with component mass balance. If all Cr is in nichromite, all remaining Ni in trevorite, and all remaining Fe in magnetite, then MS-7 would have, in mass%, 0.45 NiCr204, 2.52 NiFezO4, and 9.48 Fe304. If spinel composition in MS-7 is similar to that in Figure 2, then, at 950°C, spinel would take from the glass, in mass%, 0.21 NiCr204, 1.41 NiFezO4, and 1.07 Fe304; the rest would remain dissolved. If all Cr and all Ni from MS-7glass eventually goes to spinel to make up the total of xc,tot= 8.25 mass%, then only 5.29 mass % Fe304 is needed; the rest of Fe304 (4.19 mass%) would remain dissolved or precipitate in other phases. Accordingly, there is enough material in the glass to produce up to 8.25 mass% spinel. Figure 6 compares measured d a d 2 with calculations based on a modified Equations (10) and (1 1): 4000
0
N
y, =
(ai + biT)xj
i=l
and
164
Environmental Issues and Waste Management Technologies VIII
where yc is the spinel crystal volume fraction, at, bi, Aiy Bi, and Ti are ?’ component coefficients, and xi is ?’ component mass fraction. The component coefficients are listed in Table I. Note: Equation (12) follows from (10) with TL/Tc = c a p , and
c-’ -Er,
= bixi . Equation (13) follows from (1 1) when each composition-dependent coefficient is written in terms of partial coefficients, as in Equation (3). 0.14
Table I. Component Coefficients in Equations (12) and (1 3)
1
I Al&
-0.02
0.02
0.06
t
1
i
M;;o
-0.02
I NiO L
-
y
-0.02
0.02
0.06
I
I
1.06
I
I
0.41
1
I
-
2.62
I
1.80 1.20 -1.40 0.02 32.21 1 7.01 -0.16 -3.47 1.26 -0.24
1
~~
2.10 I
0.1
Na70
0.14
-5.82
0.1
PzOs Si07
I
I
I I
-11.88 -1.83 I -37.31 I -6.84 1.84 I
1.96 -0.05 6.74 0.44 -0.27
i
I
I
1.14 -0.12 I 1.63 I -0.17 -0.15 1
-
1
Figure 6. Calculated versus measured spinel fractions in HLW glasses based on Equations (1 Z), top, and (1 3), bottom
basured ( m s hchn)
A NOTE ON RESPONSE FUNCTIONS It is often objected that phenomenological response functions, such as Equation (3) are constructed as if “nothing [were] known about the solubility of a given component in g l a ~ s ” .In~ fact, mechanistic and structural models for glass properties abound in the literature, but “stars” that can predict the response of a property to wide ranges of state variables with a satisfactory accuracy are rare. For example, a recent model that
Environmental Issues and Waste Management Technologies VIII
165
correlates TLi)swith ionic potentials’o predicts component effects also for components whose effects have not been measured, resulting in higher prediction accuracy. CONCLUSIONS 1. Although certain glass components have significantly bigger impacts on the solidliquid phase equilibria in molten glass (and on glass properties in general), for glass formulation purposes, it is desirable to consider all major and key minor components. 2. Partial molar or specific properties, such as TLi (within each primary phase field), express the effects of individual components on the key properties and are useful for the development of glass formulation schemes. 3. Simple concepts, such as solubility limit and solubility product, though useful for aqueous solutions, are of little or limited use when applied to highly multicomponent glass-fonning melts. 4. Equilibrium fractions of crystals below TL can be determined either as a linear function of temperature or using an equation that is identical in form to the dilute ideal solution equation. 5. The glass melting process usually proceeds far from equilibrium. Therefore, kinetic aspects (crystals nucleation and growth) are of equal importance. ACKNOWLEDGMENTS This study was fimded by the U.S. Department of Energy’s Office of Science and Technology (through the Tanks Focus Area). Pacific Northwest National Laboratory is operated for the US. Department of Energy by Battelle under Contract DE-ACO676RL01830. The authors wish to thank Dong Kim for careful review of the paper and Wayne Cosby for editing. REFERENCES ‘B.K. Wilson, T.J. Plaisted, P. Hrma, and J.D. Vienna, “The Effect of Chromium Concentration on High-Level Waste Glass Properties within the Eskolaite Primary Phase Region,” in preparation. 2K.E. Spear, T.M. Besmann, and E.C. Beahm, “Thermochemical Modeling of Glass: Application to High-Level Nuclear Waste Glass,” MRS Bull. 24 (4) 37-44 (1999). 3H. Li, Y. Su, M. Qian, P. Hrma, J.D. Vienna, and D.E. Smith, “Chemical and Physical Correlation for Nepheline in Borosilicate Glasses: Effects Of A1203, B203, Na20, and Si02,” (in preparation). T.J. Plaisted, F. MO,B.K. Wilson, C. Young, and P. Hrma, “Surface Crystallization and Composition of Spinel and Acmite in High-Level Waste Glass,” Ceram. Tram. 119, 317-325 (2001). 5S.S. Kim and T.H. Sanders, “thermodynamic Modeling of Phase Diagrams in Binary Alkali Silicate Systems,” Jr., J Amer. Ceram. Soc. 74 (8) 1833-1 840 (1991). 6P. Hrma, P. Izak, J.D. Vienna, G.M. Irwin and M-L. Thomas, “Partial Molar Liquidus Temperatures of Multivalent Elements in Multicomponent Borosilicate Glass,” Phys. Chem. Glasses 43 (2) 128-136 (2002). 7J.D. Vienna, P. Hrma, J.V. Crum, and M. Mika, “Liquidus TemperatureComposition Model for Multicomponent Glasses in the Fe, Cr, Ni, and Mn Spinel Primary Phase Field,” J Non-Cryst. Solids 292, 1-24 (2001).
166
Environmental Issues and Waste Management Technologies VIII
‘C.M. Jantzen, “First-Principles Process Product Models for the Vitrification of Nuclear Waste: Relationship of Glass Composition to Glass Viscosity, Receptivity, Liquidus Temperature, and Durability,” Ceram. Trans. 23,37-52 (1991). ’M. J. Plodinec, “Solubility Approach for Modeling Waste Glass Liquidus,” Mat. Res. Soc. Symp. Proc. 556,223-230 (1999). ”J.D. Vienna, The EHect of Temperature and Composition on the Solubility of Chromium in Multi-component Alkali-borosilicate Glasses, Ph.D. Thesis, Washington State University, 2002. “M.W. Stachnik, P. Hrma, and H. Li, “Effects of High-level Waste Glass Corn osition on Spine1 Precipitation,” Ceram. Trans. 107,123-130 (2000). ‘5.D. Vienna, D.-S. Kim, P. Hrma, and G.F. Piepel, Database and Interim Glass Property Models for Hanford HL W and LA W Glasses, PNNL-14060, Pacific Northwest National Laboratory, Richland, Washington, 2002.
Environmental Issues and Waste Management Technologies VIII
167
This page intentionally left blank
GLASS FORMULATION FOR MEEL SODIUM BEARING WASTE
John D. Vienna and Dong-Sang Kim Pacific Northwest National Laboratory, Richland, WA 99352 David K. Peeler Westinghouse Savannah River Company, Aiken, SC 29808 ABSTRACT Studies were performed to develop and test a glass formulation for immobilization of sodium-bearing waste (SBW), which is a high soda, acidic, high-activity waste stored at the Idaho National Engineering and Environmental Laboratory (INEEL) in 10 underground tanks. It was determined in previous studies that SBW’s sulfur content dictates its loading in borosilicate glasses. If the s u l h content (which is -4.5 mass% SO3 on a non-volatile oxide basis in SBW) of the melter feed is too high, then a molten, alkali-sulfate-containing salt phase accumulates on the melt surface. The avoidance of salt accumulation during the melter process and the maximization of sulfur incorporation into the glass melt were the main focus of this development work. A glass (SBW-22-20) was developed for 20 mass% SBW (on a non-volatile oxide basis), which contained 0.91 mass% SO3,that met all the processing and product-quality constraints determined for SBW vitrification at a planned INEEL treatment plant. This paper summarizes the formulation efforts and presents the data developed on a series of glasses with simulated SBW. INTRODUCTION This study represents the third phase of baseline glass formulation development for INEEL SBW. In the fmt phase a glass was formulated to demonstrate the feasibility of direct vitrification of SBW using current melter technologies [11. The second phase was aimed at developing a glass to demonstrate the direct vitrification process and determine the range of expected waste loadings, assuming sulfur lost to the offgas would be grouted [2,3]. This third phase of the study was aimed at formulation of a baseline glass composition to be used in development of engineering data for vitrification plant design. The overriding assumption during the third phase was that salt accumulation in the melter would not be tolerated and that nearly all of the sulfur lost to the offgas would be recycled back to the melter feed. Different SBW compositions were assumed for each phase of the study - 1998 SBW,2000 WM-180, and 2001 WM-180 in phases 1,2, and 3, respectively. Theses waste compositions were generally similar with high concentrations of Na, Al, K, S, and Ca. However, the concentration of S in the 2001 W 1 8 0 was roughly 25%higher than in the previous two compositions. The details of the waste compositions are given in Vienna et al. [4] and are summarized in Table 1. _____
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
169
Table 1. Concentrations of Major For the purposes of this study, the Components in SBW Compositions melter technologies considered as used in the Three Phases of Glass possibilities for treatment of SBW included Formulation (in mass% of non-volatile the Joule-heated ceramic-melter (JHCM) or oxides) the induction-heated cold-crucible melter (ICCM) and the final disposal site for the 1998 2000 2001 SBW glass was the U. S . Federal Geologic Oxide SBW WM-180 WM-180 Repository. These technologies and A1203 27.34 27.96 27.52 disposal choices result in a set of glass CaO 2.23 2.22 2.15 property restrictions that must be considered during glass formulation. To meet the c1 1.04 0.88 0.87 current waste acceptance criteria for F 0.98 0.57 0.73 disposal of the glass, it was assumed that the Fe203 1.55 1.43 1.41 waste form be a borosilicate glass with K20 7.92 7.62 7.53 releases of Na, B, and Li, normalized to MnO 0.78 0.82 0.81 their concentrations (rNa, rB, and rLi; Na2O 50.05 52-54 5 1.9 1 respectively) in glass at least two standard 1.19 p205 0.80 0.79 deviations below those of the Defense Waste Processing Facility Environmental so3 3.73 3.57 4.55 Assessment (EA) glass when exposed to the Zr02 1.oo 0.0 1 0.0 1 product consistency test (PCT) IS]. Total 97.81 98.42 98.29 Although the mean rg, rNa, and rLi of the EA glass &e 8.35 g/rn2,6.67 g/m2, and 4.78 g/m2;respectively [6];more conservative releases of 1 g/m2 were targeted during formulation. The glass processing related restrictions are dependent upon which of the two melter technologies are selected. It was decided to focus on a single glass formulation that could be processed in either JHCM or ICCM, with processing related properties acceptable for both melters, but not ideal for either. The processing related properties considered were: processing temperature (TM) of 115OoC,viscosity (q) at 1150°C (qllso) between 5 and 10 Pa.s,(a) electrical conductivity ( E ) at 1150°C ( ~ ~ between ~ ~ 0 ) 10 and 100 S/m, Iiquidus temperature (TL) below 1050°C, and most importantly no salt accumulation. Although important to a successful process, no explicit constraints on corrosivity, volatility, processing rate were used during formulation. Existing glass property composition models were used to estimate q1150,E1150, rB, rNa, and rLi during glass formulation [7,8]. It was assumed, and later confirmed, that the TL constraint would not be a problem for most test glasses. The experimental study then focused on determining the effects of waste loading and additive composition on the formation of a molten salt and on the retention of s u l k in glass. Unfortunately, the behavior of s u l h in a full scale glass melter is difficult to simulate in laboratory scale crucible melts. Therefore, a host of experimental methods were used to evaluate s u l k behavior in melter feeds with systematically varied composition. (a) During initial development efforts (ICCM-11 through - 19), a ql150 range of 2 to 10 Pas was considered.
170
Environmental Issues and Waste Management Technologies VIII
EXPERIMENTAL A series of different tests were performed to determine the impact of SBW loading and additive composition on sulhr retention and salt formation. Details of the experimental approach and procedures are discussed in [4] and summarized in the following subsections.
Additive Compositions A series of additive compositions were developed to systematically vary concentrations of B2O3, CaO, Fe2O3, K20, Li20, MgO, Na20, Si02, TiQ, V205, and Zr02; within the confines of the SBW composition and the property constraints discussed above. A total of 27 sets of additives were formulated over the three phases of the study, identified as SBW-1 through -27. Figure 1 shows the pair-wise component concentrations of additives in a scatterplot matrix while Table 2 lists the compositions.
f"
4.
ia
Me0
I 00
$( 0 I0
0
0,xIw0
0 810 14
0 51015051015
02468
00
0
Moo0
I
I
I
I0
to
I
0
~~
O D O
I
I
lm
I0
XMom M O c 0
no2
3 4 5 8 0.511.5202468W6570 0 1 2 3
0
a
0
I
I
loo
I
I
zfo2
I
024680.51 2
Figure 1. Scatterplot matrix of additive component concentrations, in mass% oxides. The symbols X, Y, and Z represent the selected compositions from phase 1 (SBW-I), 2 (SBW-9), and 3 (SBW-22); respectively.
Environmental Issues and Waste Management Technologies VIII
171
Table 2. Composition of Additives (in mass% of oxides)
SBW-1 through -3 contain 3.08 mass% TiOz and SBW-10 contains 2 mass% BaO
Oxide Crucible Melts Oxides, carbonates, simple salts and boric acid were mixed in the appropriate proportions to fabricate glasses of the nominal composition of the phase 3 glasses SBW1 1 through SBW-27 with between 15 and 30 mass% of 2001 W 1 8 0 . The mixtures were melted in covered Pt/Rh cruciblesat 1 15OoCfor 2 hours. The glasses were quenched on a steel plate and samples were heat treated according to a simulated canister
172
Environmental Issues and Waste Management Technologies VIII
centerline cooling (CCC) schedule. The PCT of both quenched and CCC glass samples were measured according to standard procedures.[5] The compositions of these glasses and sulfur loss were measured by Na2Oz and KOH fusions followed by acid digestion and inductively coupled plasma atomic emission spectroscopy (ICM-AES). Melt viscosities and conductivities were measured as functions of temperature using a spindle and apposing plate methods, respectively. TL of selected glasses were bounded by 24 h heat treatment at 1050°C and optical microscopy of the resulting samples. Simulant Melts Simulated SBW, with the composition of 2001 WM-180, was fabricated fiom salts (primarily nitrates) and acids to achieve the appropriate composition including 1.OlM H',5.27 M NO3-,122.6 g oxideh,, and 2.06 M Na+.[4] The simulant was mixed with appropriate amounts of additives in the forms of single metal oxides, hydroxides, boric acid, plus sucrose as a reductant. The simulated melter feed was loaded into crucibles, dried on a hotplate, covered, and ramp heated to 1150°C and held at that temperature for 30 minutes. Due to the relatively high solids loading and impact of additives on feed chemistry (e.g., gelation due to pH changes); additional HN03 was added to many of the feeds to maintain adequate feed rheology. Centimeter Scale Melter Tests The centheter scaled melter (CSM) tests were performed to more closely match the process responsible for sulfur partitioning between glass, off-gas, and salt. CSM tests were performed using simulated melter feed described above (in addition to feeds fabricated using prefritted additives) and were conducted at a feed rate appropriate to maintain a cold-cap on the melt surface. In a previous phase of the study (phase2), feed rates were held constant (systematically varied in only 4 experiments). Sufficient feed was fabricated to obtain between 40 and 90 grams of glass. Glass samples were analyzed by fusion and ICP-AES for composition and by a wet colorhetric method to determine redox. Other Testing and Characterization Additional tests were performed with ultimate additive composition (SBW-22) in slurry-fed melt rate furnace (SMRF) 191, research scale Joule-heated melter (RSM) [101, pilot scale JHCM (PSM) [113 and a small scale ICCM [123. These results are reported elsewhere but will be briefly summarized in this report.
RESULTS AND DISCUSSION Glass Composition Efects on Sul& Behavior Crucible melts with oxide and carbonate batches showed a clear relationship between the fraction of sulfur captured in the melt and glass composition as shown in Figure 2. The distinguishing difference between the compositions of SBW-25 and the other two glasses is the concentrations of Fez03 (which is higher in SBW-25) and total alkali (R20=Li20+Na20+K20)(which is lower in SBW-25). The same general trends were found in CSM tests which typically showed higher fraction of sulfur retained in the melt with lower sulfur and Fe203content in the feed and higher concentrations of alkali and
Environmental Issues and Waste Management Technologies VIII
173
alkaline earth components. These trends were confirmed in melter tests with SBW-9 [13,141 that contained higher Fez03 and lower R20 and alkaline earth (AO=BaO+ CaO+MgO) in which a large fraction of the sulfur went to the off-gas, while in melter tests with SBW-22 nearly all the s u l k remained in the glass. Salt formation was also found to be a strong fbnction of composition. In crucible melts with oxides and carbonate feeds. there was a clear trend between
1.200
:1.100
=
3
0"
1.OOo
* s
0.900
E
0.600
1 0.800 3 0.700 0.500
0.500 0.600 0.700 0.800 0.900 1.OOo T8rget Mass% S
1.100 1.200
q in G l u r
tests the same general trend was found with the exception that one glass which had an intermediate R2WAO concentration did form a salt. This glass had the lowest fraction of R20 from Li20 and the lowest fraction of R20+AO from R20. From the results of these tests, we conclude that the concentration of R2OtAO strongly affects the ability of a glass to incorporate the sulfur from the feed into the melt. The higher the concentration of these components (CaO, MgO, BaO, K20, Na20, and Li2O) the larger fraction of the sulfur found in the glass melt and the less likely a salt layer will form. Additionally, the most benefit can be obtained by the addition of a good mix of these components rather than any single component. We investigated the possible connection between salt formation and the product of Na2O and SO3 concentrations which has been suggested recently. We found no relationship between the formation of salt and [Na20]x[S03] (in mass%) in our CSM tests with systematic variation in glass composition or in melter tests with SBW-9 and SBW-22 formulationsdescribed in the following section. The glasses that did not form a salt in these tests spanned the range of 8.81[Na2OJx[SO&24 while tests that did form a salt in these tests spanned the range of 181[Na20]x[SO3]121 and no significant correlation was observed. Baseline Glass Recommendation and Results Testing showed that SBW-22 additives (6 B203,S CaO, 1.5 Fe203,6.1 LizO, 1.8 MgO, 4.3 Na20,68 S i O ~ ~ 4V205,2.4 .9 mass% 21-02) with 20 mass% 2001 WM-180 waste composition (SBW-22-20) meet all the objectives of the formulation effort. This composition showed one of the highest s u l k retentions of any composition tested, was able to be melted in all tests with significantly higher sulfur concentrationswithout the formation of a segregated molten salt, and met all the property restrictions described above (q1150 = 6.5 Pas, ~ 1 1 5 0= 42.8 S/m, rB(Q) = 0.39(Q) rB(CCC) = 0.30, rNa(Q) = 0.61, rNa(CCC) = 0.52, rLi(Q) = 0.62, rLi(CCC) = 0.55, and no crystallinity after CCC or 1050°C heat treatments). The recommended base glass (SBW-22) was successfully processed in the SMRF at 18.2,20, and 22 mass% 2001 WM-180 simulant without the formation of a segregated
174
Environmental Issues and Waste Management Technologies VIII
molten salt and with fidl retention of sulfbr (within experimental uncertainty).[9] Tests with the SBW-22-20 simulant in the RSM were also successful, showing a capacity for higher sulfur content that the targeted 0.91 mass% SQ.[ 101 The reductant type and concentrations changed broadly in the RSM test and on average the glass retained 94.2% of the sulfur with 3.4% found in the off-gas and 2.4% unaccounted for. With success in the SMRF and RSM, a PSM test was performed with SBW-22-20 simulant.[ 11J Although the PSM test was too short to generate conclusive data on the process, no signs of salt accumulation were obtained and >97% of the sulfur was retained in glass. An ICCM test was also successfully performed using the SBW-22-20 and higher waste loadings to demonstrate feasibility.[ 121 These results conclude that the laboratory scale tests used to develop a glass for SBW vitrification adequately simulated the melter process of sulfur partitioning. A glass was developed and successfully tested to serve as the process baseline should direct vitrification of SBW be pursued. Finally, SO3 concentrations as high as 0.91 mass% are obtainable in high soda glass with nearly full sulfur retention and without the accumulation of a molten salt. ACKNOWLEDGEMENTS The authors are grateful to Denny Bickford (SRTC), Alex Cozzi (SCTC), Ronald Goles (PNNL), Pave1 Hrma (PNNL), Harry Smith (PNNL), and Douglas Witt (SRTC) for their valuable comments and suggestions; Bill Holtzscheiter (SRTC), Arlin Olson (INEEL), and Keith Perry (INEEL) for management support and guidance; Bill Buchmiller (PNNL), Jarrod Crwn (PNNL), Brett MacIsaac (PNNL), Irene Reamer (SRTC), Mike Schweiger (PNNL), and Phyllis Workman (SRTC) for laboratory assistance; and David Best (SRTC), Erick Frickey (SRTC), and Ron Sanders (KLM Analytical) for analytical support. This work was funded by the U.S. Department of Energy (DOE) OEce of Science and Technology under the Tanks Focus Area Immobilization Program. Pacific Northwest National Laboratory is operated for the DOE by Battelle under Contract DE-AC06-76RLO 1830. REFERENCES [11 JD Vienna, MJ Schweiger, DE Smith, HD Smith, JV Cam, DK Peeler, IA Reamer, CA Musick, and RD Tillotson, Glass Formulation Developmentfor INEEL Sodium-Bearing Wmte, PNNL-12234, Pacific Northwest National Laboratory, Richland, WA (1999). [23 DK Peeler, TB Edwards, IA Reamer, RJ Workman, JD Vienna, JV Crum, MJ Schweiger, Glass Formulation Development For INEEL Sodium-Bearing Waste (WM180) (U, WSRC-TR-2001-00295, Westinghouse Savannah River Company, Aiken, SC (2001). [3J JG Darab, DD Graham, BD MacIsaac, RL Russell, DK Peeler, HD Smith, and JD Vienna, Sulfur Partitioning During Vitrifxation of INEEL Sodium Bearing Wmte: Status Report, PNNL- 13588, Pacific Northwest National Laboratory, Richland, WA (200 1). [4] JD Vienna, WC Buchmiller, JV Crum, DD Graham, DS Kim, BD MacIsaac, MJ Schweiger, DK Peeler, TB Edwards, IA Reamer, RJ Workman, Glass Formulation
Environmental Issues and Waste Management Technologies VIII
175
Development for INEEL Sodium-Bearing Waste, PNNL- 14050, Pacific Northwest National Laboratory, Richland, WA (2002). [5] American Society for Testing and Materials (ASTM), “Standard Test Method for Determining Chemical Durability of Nuclear Waste Glasses, The Product Consistency Test (PCT).” ASTM-C-1285-97, in Annual Book OfASTMStandards, Vol. 12.01, Philadelphia, PA (1998). [6] CM Jantzen, NE Bibler, DC Beam, CL Crawford, and MA Pickett, Characterization of the Defense Waste Processing Facility (DWPF)Environmental Assessment (EA)Glass Standard Reference Material (U), WSRC-TR-92-346, Rev. 1, Westinghouse Savannah River Company, Aiken, SC ( 1993). [7] P Hrma, GF Piepel, JD Vienna, SK Cooley, DS Kim, and RL Russell, Database and Interim Glass Property Models for Hanford HL W Glmses, PNNL-13573, Pacific Northwest National Laboratory, Richland, WA (200 1). [8] P Hrma,GF Piepel, MJ Schweiger, DE Smith, DS Kim, PE Redgate, JD Vienna, CA LoPresti, DB Simpson, DK Peeler, and MH Langowski, Property/Composition Relationships for Hanford High-Level Waste Glasses Melting at I ISOOC,PNL-10359, Vol. 1 and 2, Pacific Northwest Laboratory, Richland, WA (1994). [9] AD Cozzi, DF Bickford, and ME Stone, Slurry Fed Melt Rate Furnace Runs to Support Glass Formulation Development for INEEL Sodium-Bearing Wmte, WSRC-TR2002-00 192, Westinghouse Savannah River Company, Aiken, SC (2002). [ 101 RW Goles, JA Del Debbio, RJ Kirkham, BD MacIsaac, JA McCray, DD Siemer, and NR Soelberg, Test Summary Report INEEL Sodium-Bearing Waste Vitrification Demonstration RSM-OI-2, PNNL- 13869, Pacific Northwest National Laboratory, Richland, WA (2002). [113 Personal communication from Keith Peny from Idaho National Engineering and Environmental Laboratory, Idaho Falls, ID. [12] Personal communications from Albert Aloy from Klopin Radium Institute, St. Petersberg, Russia and Sergey Stefanovski from Radon Institute, MOSCOW, Russia. [131 RW Goles, JM Perez, BD MacIsaac, DD Siemer, and JA McCray, Test summary Report INEEL Sodium-Bearing Waste Vitri@ation Demonstration RSM-01-01, PNNL- 13522, Pacific Northwest Laboratory, Richland, WA (2001). [141 KJ Perry, RR Kimmitt, NR Soelberg, RD Tillotson, and AN Olson, Test Results fiom sB W-FY9I-PS-OIVitrification Demonstrution of Sodium Bearing Waste Simulant using WM-I80 Surrogate, INEEL/EXT-01-01073, Idaho National Engineering and Environmental Laboratory, Idaho Fall, ID (2001).
176
Environmental Issues and Waste Management Technologies VIII
VITRIFICATION OF KOREAN LOW-LEVEL WASTE Lee 0.Nelson, Peter Kong, and Gary Anderson INEEL Waste Management Technologies P.O.Box 1625, MS 3921 Idaho Falls, ID 83415
Kwansik Choi, Cheon-Woo Kim, and Sang-Woon Shin KEPCOmErnC South Korea
ABSTRACT Idaho National Engineering and Environmental Laboratory (INEEL) and Korean Electric Power Company (KEPCO) researchers collaborated to develop waste glass formulations to vitrify LLW from Korean nuclear power plants. Modified glass property models were used to predict glass properties of candidate formulations. Several glass formulations were developed that met the KEPCO glass processing and property constraints. Initial material characterizations showed good comparisons between the measured and predicted properties. Based on the results of this work, existing glass property models may be used to develop acceptable Korean LLW glass formulations. INTRODUCTION Idaho National Engineering and Environmental Laboratory (INEEL) and Korean Electric Power Corporation (KEPCO) researchers collaborated to develop preliminary borosilicate glass fonnulations for low-level radioactive waste (LLW) generated at commercial nuclear power plants operated by KEPCO. The composition of the KEPCO LLW, on an organic-free basis, is shown in Table I. Two waste types are involved: dry active waste (DAW) comprising blotter paper, packing material, contaminated clothing, disposable shoe covers, and plastic sheeting, and resin waste, mainly spent ion exchange resins from plant demineralizers. The work presented here tested three waste blends containing 67%, 80%, and 93% DAW. The LLW product will likely be vitrified in a cold-crucible melter at approximately 115OoC. Both processing and regulatory constraints were considered in developing and screening candidate glass formulations (Table II).
~
~~
~
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
177
GLASS FORMULATION MODELING For this work, the INEEL modified linear property models originally developed at Pacific Northwest National Laboratories (P"L),"2 which are accurate for the specific glass composition range shown in Table III. In KEPCO waste glass formulations, concentrations of one major chemical component (calcium oxide) and six minor components (titanium dioxide, potassium oxide, nickel oxide, manganese dioxide, phosphate, and sulfate) fall outside the bounds of the glass property model and may adversely affect its accuracy. Table I. Chemical composition by percent for KEPCO LLW streams and three LLW blends. Component
sio,
DAW 15.1
18.9
B203
NazO Li20 CaO MgO Fe203 A1203
TiO2 KZO NiO Mn02 p205
so3
Resin
1.6 16.8 51.1 6.4 0.6 1.4 15.7 3.7 0.2 2.4 1.8
6.3
13.1 0.3
44.5
93% DAW 14.0 1.3 1.5 1.2 47.5 6.0 1.0 1.3 14.6 3.4 0.9 0.2 2.2 4.8
80%DAW 12.1 3.8 1.3 3.4 40.9 5.1 1.7 1.1 12.6 3.0 2.6 0.2 1.9 10.3
67% DAW 10.1 6.3 1.o 5.6 34.1 4.3 2.5 0.9 10.5 2.5 4.4 0.2 1.6 16.0
Table II.KEPCO glass property constraints for candidate glass formulations. Glass ProDertv Constraint Waste Loading 2Owt% minimum Electrical Conductivity 0.2 to 0.6 S/cm Viscosity Between 10 and 100 Poise PCT Leach Response 95%(1.5 orders of magnitude) lower than EA glass Processing Temperature Less than 1200°C Table III. Composition range of glass property models (by percent). Model Bounds SiOz B2O3 Na20 Li20 CaO MgO Fe203 A1203 Lower 37 5 5 1 1 0 6 0 Upper 57 20 20 7 2 2 10 15
178
2102
3 5
Others 5 8
Environmental Issues and Waste Management Technologies VIII
The linear property model for viscosity, electrical conductivity, and elemental leach rate as measured by the PCT is expressed as
where Ma = property of interest, bd = linear property coefficient, g~ = mass fraction of chemical component. The linear property model for the pH of the PCT leachate is written as
i=l
where La = property of interest. The original PNNL glass property model coefficients are shown in Table IV. Coefficients for components that were not present in the original property model were added to the model by the INEEL, based on their expected behavior in a borosilicate glass (Table V).
MEASUREMENT OF MODELS To verify glass-modeling results, laboratory measurements were completed for selected glass properties, including molten glass viscosity, molten glass electrical conductivity, conductivity of PCT leachate solutions, and PCT elemental leach response (see Table VI). Table JY.Glass property model coefficients (mass basis).lP2
NazO LizO CaO MgO F e 0 3 A1203 ZIoz Others 10.7 19.7 -0.61 2.93 -4.23 -17.34 -10.81 -0.73 -6.04 B 12.00 17.6 22.6 -8.71 10.90 -3.20 -25.41 -10.56 0.16 Li 10.2 14 18.4 -5.35 7.12 -4.51 -22.31 -10.06 0.62 9.41 19.4 Na 19.1 -1.96 11.80 -4.10 -25.43 -11.42 -0.66 pH 3.33 23.6 31.2 17.2 15.30 8.59 5.36 7.61 9.27 Thermal properties, Td glass transition temperature (K), In qllso= viscoSity at 1150°C (Pas), In qlm= electrical conductivity at 1150°C (Skm) T,i(K) 896 858 402 -298 895 767 700 817 1003 637 -6.2 -11.0 -34.2 -7.5 7.4 -0.2 -2.8 0.0 11.3 Inqllm 9.0 2.3 11.0 23.5 1.4 1.1 2.6 1.3 1.1 3.5 InEllso 0.9
Si
SiO,
-2.97 -4.32 -3.23 -4.41 8.19
Bz@
7
Environmental Issues and Waste Management Technologies VIII
179
Table V. Glass property model coefficients added for KEPCO waste components. Ti02 K20 NiO Mn02 pZo5 PCT Leach Response Si -2.97 10.7 -6.04 -2.97 -2.97 B -4.32 17.6 -8.71 -4.32 -4.32 Li -3.23 -3.23 -3.23 14 -5.35 Na -4.4 1 19.4 - 1.96 -4.41 -4.4 1 pH 8.19 23.6 17.2 8.19 8.19 Thermal properties, In q1150 = viscosity at 1150°C (Pa.s), In ~ 1 1 5 = 0 electrical conductivity at 115OOC (S/cm) lnq1150 9.0 -11.0 -7.5 9.0 9.0 In ~1150 0.9 11.0 1.4 0.9 0.9
Table VI. Properties measured in the laboratory. Property Molten glass viscosity
Molten glass electrical conductivity Scanningelectron microscopy X-ray diffraction Elemental leach response Leachate conductivity
Instrumentation Brookfield Digital Viscometer,Model DV-111 rotating spindle viscometer. The 1.4-cm diameter Pt-20wt%Rh spindle was submerged in molten glass contained in a small (50-mm diameter) Pt crucible to complete the viscosity measurements. Pt-20%Rh probe with 7 x 38-mm blades set 9 mm apart, llcHz frequency AC current, and J3P 4262A LCR meter Phillips Model XL30ESEM Siemens D5000 equipped with Bruker DefracB software Inductively coupled plasma with mass spectrometer (ICP/MS) Orion Model 126 conductivity probe
The PCT procedure4 was followed to determine the leach resistance of the waste glasses. Before carrying out the elemental analyses of the PCT leachates, the leachate conductivity was measured via an INEEL-developed procedure5 to yield a preliminary indication of the glasses’ leach resistance. If the relationship between the leachate conductivity and total mass loss is assumed to be linear, then the total mass loss of candidate glass formulas can be estimated as follows:
LC Estimated Total Mass Loss = LCm
(3)
where LC is the leachate conductivity of the candidate formulation and LCsm is the leachate conductivity of environmental assessment @A) glass. The INEEL has used the leachate conductivity test for several years to compare leach response of glass waste forms. The method has proven to be quick, reliable, and inexpensive. The estimated total mass loss becomes more accurate as the composition of the reference glass approaches the composition of interest.
180
Environmental Issues and Waste Management Technologies VIII
RESULTS Twenty glass formulations were developed 11 met processing requirements and could be characterized in the laboratory (Table Vn) while 9 were rejected because of phase separation, a property not encompassed by the models. Table VII. Compositions of candidate KEPCO glasses that were developed using glass property models. Shading indicates compositions outside the bounds of the original property model.
Tables VIII, IX,and X show predicted and measured viscosity and electrical conductivity of the glass compositions studied, the measured total mass loss rate as approximated by the leachate conductivity, and predicted and measured normalized PCT elemental release values. For the PNNL models, accuracy has been reported in terms of correlation coefficients and R2values. For formulations KllA, K12A, K12B, and K12E, which have iron concentrations outside the model’s lower bound, R2 was calculated separately. Glass samples also were analyzed by X-ray diffraction (XRD) and a scanning electron microscope (SEM) equipped with an electron dispersive spectrograph. Both the XRD and SEM results show that the glass formulations developed were amorphous pure glasses, with no crystalline phase observed in the glass matrix.
DISCUSSION Of the 11 processable glasses developed using existing glass property models, one (K11A) met all property constraints identified for KEPCO LLW. Three (K12A, K12B, and K12E) met all except the durability constraint for Li. Although all 11 formulations met the glass processing constraint for viscosity, the modeling results did not provide a very good fit to the laboratory data. The R2 value for viscosity of 0.38 is much lower than the literature-reported value of 0.94. This is caused by the relatively high calcium oxide concentrations in
Environmental Issues and Waste Management Technologies VIII
181
KEPCO waste (approximately 10%at a waste loading of 20 wt%) compared to the bounds of the glass property model (1-2%). Since calcium oxide is a waste component, its concentration cannot be reduced unless another waste that does not contain calcium oxide is added or the waste loading is decreased below 20 wt%. To improve the model's accuracy, a dedicated composition variation study is required. Table VIII. Predicted and measured viscosities and thermal conductivities for KEPCO LLW glass formulations. Boldface indicates the glass met the processing constraint. Viscosity
Glass
K1
K2 K3 K4 K5 K6 K7 K11A K12A K12B K12E
Predicted
54 51 53 50 42 46 76 56 48 51 53
(PI
Measured 49 40
RZ 0.38
36 15 21 33 47
0.2
35 38 40 40
Electrical Conductivity (S/cm) Measured Predicted
0.26 0.26 0.29 0.50 0.42 0.29 0.26 0.23 0.17 0.15 0.17
R2 0.99
0.28 0.48
-
0.26 0.28
ND
-
I
Table IX.PCT leachate conductivity and pH values for EA and candidate KEPCO glass compositions. Boldface indicates the glass met the durability constraint. PCT Leachate Glass Formulation EA
K1 K2
K3 K4
K5 K6
K7 K11A
Conductivity' (Wm)
Estimated Total Mass Loss (compared to EA Glass)
3970 (3940) 790 (830) 830 (815) 575 (585) 1891 (1866) 1144 887 (877) 401 (413)
1.o 0.20 0.21 0.15 0.47 0.29 0.22 0.10
P
pH PCT Leachate
10.9 11.7 11.4 11.3 10.7 10.1 9.6 9.3 9.6
M
R2
11.4
-
ND -
10.8 11.4 11.0 11.1 10.5 10.1 9.1 9.0 9.0
0.85
0.85 K12A 217 (209) 0.05 K12B 289 (298) 0.07 K12E 184.8 (187.1) 0.05 Key: P = Predicted leach response; M = Measured leach response; R2= Square of the correlation coefficient 'Values in parenthesis are duplicate samples.
182
Environmental Issues and Waste Management Technologies VIII
Table X. Predicted and measured normalized PCT elemental release values. Boldface indicates values that met the Drocessing constraint.
K4
0.18
K5
0.09
K6 K7 KllA K12A K12B K12E Const.
0.07 0.04 0.04 0.05 0.10 0.04
-
0.32 0.17 0.20 0.12 0.07
0.066
0.103 0.064
0.11
0.25 0.09
0.07 0.02 0.95 0.06 0.14 0.41 0.11
-
-
0.68 037 0.48 0.24 0.15 0.26 0.44 0.22 0.48
0.31 0.13 0.10 0.04 0.97 0.11 0.21 0.58 0.17
-
-
1.39 0.78 0.92 0.54 0.25 0.39 0.58 0.36 0.25
0.33 0.12 0.10 0.03 0.93 0.07 0.14 0.42 0.12
-
-
1.10 0.57 0.83 0.29 0.18 0.20
0.98
0.34
0.18 0.37
-
Key: P = Predicted leach response;M = Measured leach response; R2= Square of the correlation coefficient; EA = Environmental Assessment Glass;Const. = KEPCO Constraint on leach resistance (95%lower than EA Glass).
Electrical conductivity was not measured for all glasses, since glasses that meet the viscosity constraint are expected to also meet the electrical conductivity constraint. The R2 value of 0.99 for comparison of the predicted and measured glass electrical conductivity shows that the glass property models were able to predict this property fairly well. Hence, the model for electrical conductivity appears accurate for the glass compositions studied. Leach resistance was measured by two methods: PCT elemental release and PCT leachate conductivity. The PCT elemental release constraint was met by the K11A glass for all elements and by the K12A, K12B, and K12E glasses for all elements except Li. Several of the other formulations met the constraint for some elements. Leachate conductivity test results showed that only the K12A and K12E glasses met this property constraint. (A technical problem precluded leachate conductivity measurement for glass K11A). Importantly, both the leachate conductivity measurements and the PCT elemental leach response predicted the same glasses to be the most durable. Since the leachate conductivity test is faster and much less expensive than measuring the PCT elemental leach response, its use during scoping glass formulation studies can reduce costs and still identify formulations that are leach resistant. The measured leach resistance of the formulations matched the predicted values fairly well. For example, the R2 value associated with the elemental leach resistance and pH of the PCT leachate was 0.78 to 0.98,comparing favorably to the literature-reported value of 0.74 to 0.9 1.
Environmental Issues and Waste Management Technologies VIII
183
CONCLUSIONS AND RECOMMENDATIONS Glass formulation K11A meets all the KEPCO glass property constraints. Use of this glass formulation on the pilot scale is recommended. Glass formulations K12A, K12B, and K12E meet nearly all of the processing constraints and may be suitable for additional testing. Linear glass property models may be used to predict the properties of KEPCO LLW as a function of composition. The observed results indicate that the models for electrical conductivity, PCT elemental release, and pH of PCT leachate are able to accurately predict these properties. However, the model for viscosity was able to provide only qualitative results. To develop a more accurate model for KEPCO wastes, a composition variation study is recommended. This study and associated model development would improve the ability of a model to accurately predict candidate KEPCO LLW glass properties. The model could be extended to support operations after a waste blend and base waste-glass formula are developed. REFERENCES ‘P. Hrma et al., “Prediction of Nuclear Waste Glass Dissolution as a Function of Composition,” Ceramic Transactions, 61 497-504 (1995). 2P. Hrma et al., “Prediction of Processing Properties for Nuclear Waste Glasses,” Ceramic Transactions, 61 505-13 (1995). 3G. Piepel et al., “Statistical Experimental Design of a Waste Glass Study,” Ceramic Transactions, 61,489-96 (1995). 4C. M. Jantzen et al., Nuclear Waste Glass Product Consistency Test (PCT) Version 5.0 (U), WSRC-TR-90-539 Rev. 2. DOE Savannah River, Jan. 1992. k.Vinjamuri, “Solidification of High Concentration Sodium-Bearing Wastes by Vitrification,” 1997 DOE Low Level Waste Management Conference, Salt Lake City, UT. 6Personalcommunicationswith K. S. Choi, by email, Aug. 2,2001.
184
Environmental Issues and Waste Management Technologies VIII
PHASE EQUILIBRIA, VISCOSITY, DURABILITY, AND RAMAN SPECTRA IN THE SYSTEM FOR IDAHO NUCLEAR WASTE FORMS
S. V. Raman, B.A. Scholes, A. Erickson Idaho National Engineering and Environmental Laboratory Idaho Falls, ID. 83415 A. A. Zareba Department of Chemistry University of Houston Houston, TX. 77204 ABSTRACT In an effort to immobilize the high level and mixed waste components of Idaho, an integrated study of phase relations, liquid flow and quenched liquid structure was conducted using simulated non radioactive calcine composition as a model example. The solubility of waste components in a boroaluminosilicate liquid is here expressed in the isothermal composite phase equilibria projection and its compositional cross sections. These formulation diagrams predict with certainty the chemical variations needed to immobilize waste streams in waste forms containing predictable proportions of crystalline and glass phases as a function of processing temperature and rheology. The dependence of viscous flow on temperatures above the liquidus is explained in light of changes in chemical bonding and molecular symmetry as revealed by Rmm spectra of glasses selected from specific compositional regions of these diagrams. A higher durability for glass monoliths is predicted in the direction of increasing covalency and intermediate glass structure. INTRODUCTION The investigation of phase equilibria, liquid flow and glass structure in a stems fiom the multicomponent system Si02-AlzO~-B~03-CaO-CaF~-Na20-ZrO2 complex chemical composition of the sodium bearing waste [I] and granular calcines [2,3] that are currently stored in the tanks and bin sets at the Idaho Nuclear Technology and Engineering Center (INTEC). On the basis of
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management TechnologiesVIII
185
abundance, the waste stream components can be subdivided into major, minor and trace groups. A typical grouping for the Idaho calcines is shown in Table I. Table I. An average categorization of calcine components into major, minor and trace groups Major weight% Minor weight% Trace weight% 40 CdO 2 cs20
186
Environmental Issues and Waste Management Technologies VIII
one more component Ti02 as this is not a part of the original waste stream. It must also be recognized that Si02 is an unavoidable solvent, which cannot be replaced by Ti02 because of the need for liquid phase to mobilize the solutes and to flow the product into containment from the melter. Further, the nucleation tendency of TiO2 in a calcine containing borosilicate liquid is likely to lead to formation of many phases, such as dendritic titanite (CaTiSiOs) [ 5 ] , spherulitic ulvospinel (TiFe204) [5], cadmium orthoperovskite ( C ~ O . ~ C Q . ~IS], T ~prismatic O~) acicular rutile (Ti02) [ 5 ] , and zirconium and calcium oxide depleted alkali boroaluminosilicate glass [2, 51, leading to increasing the durability degrading features in a glass-ceramic microstructure. Thus immobilizing the Idaho waste streams in titanate based waste forms 16, 71 is not a suitable processing path, when all the components of Table I are considered. It must also be emphasized that zircon is far more abundant in natural rocks than zirconalite, while both are comparably durable with minor differences in susceptibility to irradiation damage [8,9]. The retention of parental and daughter products like uranium, thorium and lead in zircon from Archaean times is clearly documented, leading to the use of this mineral as an important geochronologic indicator following correction for helium loss [IO]. Another important waste load limiting component in the calcine is calcium fluoride, which readily crystallizes to form fluorite (CaF2) glass-ceramic. The formation of glass waste form compositional regime for the waste stream in Table I is bound by the solubility limits of zirconia and calcium fluoride. The two components also differ in bonding characteristics, while 2 1 0 2 with 50% ionicity [I 11 contributes to partial covalent bonding, CaF2 is 80% ionic [I 13. Yet other important modifying components with higher ionicity [ll] are CaO (61% ionic) and Na2O (65% ionic) in the major group of calcine waste stream in Table I that are more soluble than CaF2 or 21-02 in the silicate liquid. The chemical bonding characteristics and component solubility in the boroaluminosilicate liquid and equilibrated crystals evidently determine the durability, waste loading and compositional expanse of waste forms in the system of interest, which is made of major calcine components (Table I) and silica. These components compose the main frame of the waste form, in which minor and trace components shown in Table I are expected to occur as interstitial or substituted ionic inclusions by possibly imparting minor compositional shifts to phase boundaries. In this paper, the first part of the experimental endeavor, that is the interactions among the major components are examined by melting experiments under ambient conditions from which phase equilibria is evolved that forms the basis for determining the compositional expanse of glass and glass-zircon-fluorite waste forms. The formability and durability of these waste forms is judged from liquid flow and the structure of liquid as quenched glass. These two properties are interrelated. They are influenced by the relative interactions among the
Environmental Issues and Waste Management Technologies VIII
187
components in a silicate network. Complementing relations among these properties are examined by Raman spectra of quenched glasses and viscosity of corresponding liquids to gain insight into the role of molecular characteristics involved in the formation of mixed solutions and their flow behavior.
EXPERIMENTAL PROCEDURE Reagent grade components of SiOz, Al2O3, H3B03, CaO, CaF2, Na2C03, and Zr02 were mixed in varying proportions to form 100 gram batches. The batches were heat treated at 850°C to drive off moisture and carbonate, and melted at 1200°C in alumina crucibles under ambient atmosphere. Each compositional melt was poured into a cylindrical aluminum mold and annealed at 450°C for 12 hours. The crucible walls and melts were visually examined for possible phase separation, melt homogeneity and crystallization during pouring. The same features in annealed glasses were examined in a petrographic microscope. Crystals were identified by optical properties [12] and x-ray powder diffraction. For determining liquidus phase boundaries, the compositions of melts were changed across the boundaries to observe the appearance or disappearance of crystallites without altering the temperature of 1200°C. Euhedral outlines of the crystals were inferred to indicate equilibration with the melt. The solidus boundary was determined by compositions that contained liquid phase as little as a glaze in the solidified crystalline mass. The mass was broken apart from the crucible, crushed to powder and characterized by x-ray powder diffraction for sharp crystalline peaks and minimal presence of a broad amorphous background or impurities. For X-ray powder diffraction Bruker D8Advance model was used at the Idaho Research and Center (IRC) and the diffraction set up was of thetatheta type. For viscosity measurements, the Theta high temperature rotating viscometer with Brookfield DV-I11 model Rheometer was used. The apparatus was calibrated using NBS sample No. 710 for soda-lime-silica glass [131. A calculated volume of crushed glass needed to completely immerse the IOmm diameter platinum spindle was loaded into a 50 ml cylindrical platinum crucible and melted to 1250°C. The sample temperature was measured by attaching the thermocouple close to the crucible base. A separate thermocouple was used for controlling the h a c e temperature and the torque was measured over the range of 850 to 1600°C. Viscosity measurements were taken fust at the melt temperature after the spindle and torque readings had stabilized. The glass was held at each set temperature for 15 minutes. The viscosity measurements at each set point are based on an arithmetic mean of torque readings taken every 15 seconds on the spindle rotating in the glass melt over the last 5 minutes of the 15 minute hold time.
188
Environmental Issues and Waste Management Technologies VIII
For Raman spectroscopy, the samples contained in the platinum crucibles following viscosity measurements were remelted and poured into separate aluminum molds to form lcm x lcm x 2cm rectangular billets. The Raman spectroscopy was conducted at the University of Houston, using a conventional scanning Raman instrument equipped with a SPEX 1403 double monochromator. Both the entrance and exit spectral slit widths were maintained at 5cm". The sample was mounted onto a sample holder and subjected to 514.5 nm coherent 90-6 Ar' ion laser excitation at 45" to the vertical surface of the rectangular billet. The scattered radiation was analyzed at 90" to the sample surface, such that the scattered and incident radiations formed an angle of 45". Insertion of analyzer and polarization scrambler in the scattered beam path enabled collection of Raman spectra in parallel and perpendicular scattering modes. This scattering geometry was tested by using fused silica as standard and no differences were noted between the present 45" collection angle and the 90" one reported in the literature [14]. The spectra were corrected for background by subtracting the Rayleigh tail fiom the actual spectra using GRAMS32 software of Thermo Galactic Inc. The Raman spectrometer set up and experimental method are described in detail elsewhere [151.
RESULTS AND DISCUSSION Phase relations in 70wt% (Base glass)-30 wt% (CaO+CaF2+NazO+Zr02) The seven component variables Si02-A1203-B203-CaO-CaF2-Na20-ZrO2 were successively decreased to evolve the system 7Owt% (Base gEass)-30 wt%(CaO+CaF2+Na20+ZrOr) so that the major calcine components (CaO+CaF2+Na2O+ZrO2) remain as the only independent variables. The other two major components B2O3 and A1203 of Table I are combined with Si02 to yield a fixed molar ratio of 26.5B203:4.5A1203:69Si02for the base glass. The choice of this composition is based on the phase relations described elsewhere [16] so that a homogeneous glass could be made below 1600°C.A1203addition usually increases the melting temperature and hence its concentration was kept at a low level and was also added to minimize the possibility of liquid phase separation that is otherwise prominent in the subliquidus region of Si02-Bz03 system [16]. A 2 0 3 addition is here considered necessary, for it is present in varying concentrations in all the waste streams of Idaho. Its concentration on the order of 4.5 mol % raises the melting temperature to 1350°C according to the phase diagram in reference [17]. The melt viscosity measures 523 Pa.s at 1400°C and quenches to homogeneous transparent glass when poured from a temperature of 1650°C where the viscosity as extrapolated fiom lower temperature measurements using Vogel-Tamman-Fulcher model [13] is 38 Pa.s. Addition of calcine components in the mass ratio of 70 wtY0 base glass and 30 wt% calcine components produces homogeneous liquids, the forming temperatures of which
Environmental Issues and Waste Management Technologies VIII
189
are a function of the relative ratios of the calcine components. The relative mass ratios of CaO:CaF2:Na20:ZrO;! leading to a single liquid and thus to a homogeneous quenched glass at a forming temperature of 1200°C are shown as liquidglass region in the unfolded isothermal quaternary projection Figure 1.
c'ao
Figure 1. A 1200°C isothermal projection of the open faceg*{mternary diagram from boroaluminosilicatebase glass apex. L+Q =liquid + quartz. These glass modifLing components lead to a large drop in the pouring temperature (that is the temperature at which the viscosity is 40 Pa.s) from 1650°C for base glass to 1200°C to 900°C range for the waste glass. With the furnace temperature held constantly at 1200"C, variations in the relative proportions of the calcine components lead to nucleation of crystallites, defining the compositional expanse of the liquid + crystal field and the phase boundaries in Figure 1. The coexisting transparent liquid with crystallite inclusions turns from opalescent to white mass with compositional change towards the solidus. The solidus is marked as the compositional point at which the crucible contains a glazed white mass at 1200 OC and defines the compositional boundary of terminal crystallization of the liquidus crystals fluorite, zircon and quartz. While the viscosity near the liquidus composition in the fluorite + liquid field is on the order of 1 Pa.s, it is about 40 Pa.s in the zircon + liquid field and sharply rises with increase in zirconia content. Solution Behavior within the Quaternary Projection (CaO+CaF2+Na20+Zr02) A folded geometry turns Figure 1 into a quaternary projection, with CaO apex and Na~O-CaF2-Zr02triangular base in Figure 2. The phase relations in the base triangle of Figure 1 arise in the absence of calcium oxide. Since the system covers the major components of the Idaho wastes, the compositions of the various
190
Environmental Issues and Waste Management TechnologiesVIII
calcines [2.3] and sodium bearing waste (SBW) [l] are expected to occur within the quaternary in different compositional directions.
Figure 2. Cross sections of Figure 1 at (a) 3 wtY0 CaO and (b) 24 wtY0 CaO The application of this diagram to the sodium bearing waste is based on the assumption that Na20 and Al203, which are a part of this quaternary, are likely to form the predominant particle residue resulting from evaporation and condensation by steam reforming of the SBW liquid. The solution behavior within the quaternary is here examined with calcium oxide as the variable and two cross sections of the quaternary were considered at 3 wt% CaO and 24 wt% CaO, since these concentrations approximate the upper and lower limiting bounds of CaO in the calcines. The cross sections indicated at 0.3 x 10 wt% and 0.3 x 80 wt% CaO in Figure 1 are shown as projections on Na20-CaF2-2102 plane in Figures 2a and 2b. In both figures the experimental liquidus is shown to depart fiom the linear mixing along the ~ 5 x and 4 y7-y8 compositional joins. The concave and convex compositional deviations are induced by changes in Na20 for positioning the liquidus boundary at the junction of liquid and liquid + crystalline fields. The phase boundaries and the phases of the fields were identified by means of X-ray powder diffraction and visual examination of the products in the crucible for changes from homogeneous to heterogeneous glasses. A typical X-ray powder diffraction is shown in Figure 3. 712
X i 06 zircon & fluorite
1 ; 7
F
10
20
Figure 3. X-ray powder diffraction showing coexistence of zircon (Z), fluorite (F) and liquid in the corresponding phase field of Figure 2a at composition point X 1OB.
30
40
50
60
70
80
2 theta (degrees)
Environmental Issues and Waste Management Technologies VIII
191
The X-ray peaks for fluorite (F) and zircon (Z) are noted to occur superimposed on the amorphous background, suggesting that the field is made of crystalline and liquid phases. Similarly, other fields were revealed by X-ray powder diffraction, enabling demarcation of the cotectic joins partitioning the fields bounded by solidus and liquidus into liquid+zircon, liquid+fluorite, liquid+zircon+fluorite, liquid+baddeleyite+zircon+quartz, and liquid+zircon+fluorite+quartz in Figure 2a at 3 wt% CaO cross section of the quaternary diagram (Fig. 1). In Figure 2a the prcportions of major calcine components (Table 1) occur in the liquid+zircon+fluorite+quartz field, suggesting the phases that would compose the waste form if the calcine were to be processed in this region. Alternatively the diagram can be used to select proportions of additive components using the lever rule [18] to position the calcine in glass or glass-ceramic fields with predictable compositions for the phases in the waste forms. The change in composition to 24 wt% CaO content shows presence of only quartz as the liquidus phase in the equilibrium diagram of Figure 2b. This phase was noted to form a floating cover with a homogeneous liquid beneath it in the crucible.
Raman Spectra of Glasses From comparison of Raman spectral bands in Figures 4 and 5, the sequence of changes in the glass structure become apparent with addition of boric and aluminum oxides to fused silica, and subsequent addition of the major calcine components (Table I) to the resulting boroaluminosilicate glass.
I
500 1000 1500 Raman frequency shifts (anm1)
500 1000 Raman frequency shifts (an-')
1500
Figure 4. Raman spectra of fused silica G19 and boroaluminosilicateglass G20. (a) and (b) under parallel (c) and (d) under perpendicular analyzing conditions
192
Environmental Issues and Waste Management TechnologiesVIII
6000
3ooo
5000
2500
F
4000
P
2000
4
1500
9
I
C
3000 2000
IOW
I000 0
0
400 800 1200 1600 Raman Frequency Shifts (em")
400 800 1200 Raman Frequency Shks (cm
0 1500
-')
Figure 5. Raman spectra under parallel (P) and perpendicular (N) analyzing conditions of compostions x5 and x4 shown in Figure 2a All the typical bands that have been previously reported in the literature [ 19221 are observed in Figures 4 and 5 and only the changes as a function of composition are considered here in the four prominent bands ranging in wave numbers of (1) 467 to 499 crn-', (2) 770 to 800 cm-', (3) 971 to1059 cm-' and (4) 1366 to1441 cm-'. Commencing from fused silica in Figure 4% it will be noted that the Si-0-Si rocking and stretching vibrations [19-211 in the 440 and 498 cm-' wave numbers couple to become a single 477 cm-' band in the boroaluminosilicate glass of Figure 4b. Other important changes involve decrease in the 604 cm-' defect band [21] (Fig. 4b), shift in the frequency of Si-0-Si bond bending and stretching doublet [19-211 at 1058 and 1198 cm" (Fig. 4a) to lower values of 1024 and 1109 cm-' (Fig. 4b) and a nearly total loss of anisotropy in the perpendicularly scattered Raman spectra of boroaluminosilicate glass (Fig. 4d). The prominent anisotropic doublet of 789-833 cm-' in Figure 4a and 4c, supposed to result from silicon motion in the tetrahedral cage [19-21J is masked by metaborate and tetraborate bands [22] at 770 and 798 cm-' in Figure 4b. In addition, a broad and weak borate stretching originates around 1433 cm-' (Fig. 4b). The bridged boroaluminosilicate network of Figure 4b is significantly modified with the dissolution of calcine components in Figures 5a and 5b. Along with the presence of intense Si-0-Si stretching modes in the 467 and 499 cm-', there is a pronounced transformation of transverse stretching vibration of fused silica at 1058 cm" in Figure 4a into intense bands around 973 and 1057 cm-' wave numbers in Figures 5a and 5b that have been previously assigned to the silicon-non-bridging oxygen stretching [20]. Contemporaneous to this change is
Environmental Issues and Waste Management Technologies VIII
193
the narrowin of the width with increased intensity for the borate band in the 1392 to 1441 cm- range in Figure 5a and 5b relative to the same band in the pure boroaluminosilicate glass of Figure 4b at 1433 cm-'. The compositions x4 and x5 to which the Raman spectra of Figure 5a and 5b belong, occur respectively along zirconia- sodium oxide and calcium fluoride-sodium oxide joins of Figure 2a. The differences in the Raman spectra then are attributed to occurrence of zirconia, sodium oxide and calcium fluoride that differ from one another in the nature of bonding, with covalency in the order SO%, 35% and 20% [l 13. Ionic bonds make little contribution to Raman intensity [23] as a result the differences in Raman bands may be attributed to Zr-0 and Na-0 linkages. This difference is evident in the non-bridging region, where zirconia containing glass XS (Fig. 2a) has an intense and strongly anisotropic band at 971 cm-l, while the 1059 cm-' nonbridging band is located on its shoulder (Fig. 5a). In contrast, the latter band, with a weaker anisotropy at 1047 cm" (Fig. 5b) composes the non-bridging region of the x4 glass positioned on the CaF2-Na20 join in Figure 2a. The notable decrease in intensity of the 1047 cm" (Fig. 5b) relative to the intensity of 971 cm-' (Fig. 5a) in the perpendicular scattering mode, is related to decrease in covalency caused by ionicity of Na-0 linkage in the 1047 cm" band. In the presence of Zr-0 linkage, the 1047 cm-' band appears as a shoulder of the intense 971 cm" band in Figure 5a. Another change of interest between glasses x5 and x4 is the occurrence of 359 cm-' band on the shoulder of the 467 cm-' bridging band in the glass x5, and a distinct absence of the same feature in calcium fluoride containing glass x4. Also a cascading decrease in the intensity for 626 and 740 cm-' bands are noted in glass x4, whereas in glass x5 the comparable bands at 623 and 756 cm-' are of nearly equal intensity. These spectral complexities are currently not explained in this paper.
F
Viscosity of Liquids The temperature dependence of viscosity for pure and modifying components containing boroaluminosilicate liquids is shown in Figures 6a to 6c. For all compositions the trends are Arrhenian [24] above the liquidus temperature. In all these compositions the boric oxide, aluminum oxide and silica proportions are maintained invariant as specified in the composition of the boroaluminosilicate glass 26.5B203:4.5A1203:69Si02. The activation energy for viscous flow of this liquid calculates to 282 kJ/mol (Fig. 6a) and is nearly 10 kJ/mol Iess than the framework liquid of germanium oxide [25]. Addition of modifying components leads to a decrease in activation energy. The values range from 193 to 101 kJ/mol (Fig. 6b) for compositions x5, x4 and xl (Fig. 2a) and from 180 to 164 kJ/mol (Fig. 5c) for compositions y7, y8 and y4 (Fig. 2b). In proceeding from x5 to x4 and y7 to y8 viscosity decreases with decreasing ZrOz/CaFz proportion and so do their activation energies.
194
Environmental Issues and Waste Management Technologies VIII
50 -
Q -c
48-
4.8
.
44
-
F
4.2 -
T-' x 1 e (K-1)
Figure 6 . Temperature dependence of viscosity and activation energy for flow of homogeneous liquids above the liquidus, (a) boroaluminoslicate liquid, (b) x5,x4, xl liquid compositions shown in Figure 2a, and (c), y7,y8 and y4 liquid compositions shown in Figure 2b.
0.5' 6.0
62
6.4
T-' x 10-4 (K-1)
6B
I
hE
The median compositions x l and y4 (Fig. 2) occur near the concave and convex liquidus and their activation energies depart from the linear trend in response to sodium oxide content. Thus the activation energy for x 1 (1 5 1 kJ/mol) is higher than the corresponding composition on the linear x5-x4 join (Fig.2a). Opposing effect is noted for the median composition y4 because of the increase in sodium oxide content relative to its mixed composition on the y7-y8 join of Figure 2b. The viscous flow of y4 liquid has an activation energy (164 kJ/mol) that is lower than the mixed composition along the linear join y7-y8. A higher activation energy for x5 (193 kJ/mol) than for y7 (1 SO kJ/mol) along the zirconiasodium oxide join (Fig. 2a and 2b) is attributed to higher ZrOz/NazO ratio and lower CaO (3 wt %) in the former composition. Similarly, a lower activation energy for x4 (101 kj/mol) than for y8 (168 kj/mol) along the calcium fluoridesodium oxide join (Fig. 2a and 2b) results fiom higher CaF2Ma20 and lower CaO (3 wt %) in the former liquid. Considering that the activation energies here vary
Environmental Issues and Waste Management TechnologiesVIII
195
in response to relative proportions of covalency and ionicity contributing components of ZrO2 (50 % covalent) and Na20 (65% ionic) and CaF2 (80% ionic) [ 113, it is of interest to explain the variations in viscous flow in light of changes in molecular symmetry. Although, the exact nature of molecular symmetry is not expected to be evident fkom the present Raman spectral studies, it may be suggested that the strongly anisotropic 971 cm-' band in composition x5 (Fig. 2a and 6a), indicates non-totally symmetric [23, 261 oscillatory modes that could arise from symmetric or antisymmetric molecular species [27]. This anisotropy is noted to decrease as is evident from the relatively decreased intensity of the nontotally symmetric 1047 cm-' band in ionic component dominated composition x4 (Fig. 2a and 6b). From this comparison it may be concluded that greater steric hindrance to viscous flow is imparted by the molecular structure in x5 than in x4, and perhaps the key difference is in the molecular symmetry, with possible presence of antisymmetric species in x5. It is also of interest to note the higher durability of x5 type than the x4 type structures as judged from the concentration of the leachable elements sodium and boron in the leachates, arising from testing of homogeneous glass monoliths that are comparable to x5 and x4 in composition and Raman spectra [27,28]. CONCLUSIONS For the model calcine composition selected in this study, the phase equilibrium approach predicts the possible waste loading in glass, and glassceramic waste forms. The two waste dissolution limiting components, zirconia and calcium fluoride restrict waste solubility in glass, and enhance the waste concentration in the glass-ceramic by promoting crystallization of zircon and fluorite. The opposing influence of zirconia and calcium fluoride on viscosity is revealed by large changes in the activation energy for liquid flow. While calcium fluoride decreases the activation energy for flow and increases devitrification, zirconia enhances the activation energy for flow and induces crystallization in the melt. The synergism among these properties is noted in the neighborhood of the triple point where liquid, fluorite and zircon meet in the 120OOC isothermal cross section of the phase diagram. Raman spectroscopic results indicate the dependence of durability [27,28] and viscosity on chemical bonding, molecular symmetry and atomic ordering in glass. The molecular symmetry that is dominated by covalency, seems to increase the activation energy for viscous flow and also enhance the durability of quenched glass. In the same direction, the atomic ordering in glass appears to depart towards intermediate order from the typical random network [27]. We acknowledge the support provided by the US Department of EnergyLaboratory Directed Research and Development program (DOE-LDRD).
196
Environmental Issues and Waste Management TechnologiesVIII
REFERENCES 1. C.M. Barnes, “Feed Composition for the Sodium Bearing Waste Treatment Process,” pp. 77, Idaho National Engineering and Environmental laboratory Bechtel BWXT Idaho, LLC, Report No. INEELEXT-2000-01378(2001). 2. S.V. Raman, “Hot Isostatically Pressed Aluminosilicate Glasss-Ceramic with Natural Crystalline Analogues for Immobilizing the Calcined High Level Nuclear Waste at the Idaho Chemical Processing Plant,” pp. 1 19, Westinghouse Idaho Nuclear Company Report No. WINCO-1173,(1 993). 3. S.V. Raman, “Advanced Formulation for Homogenization of Refractory Oxides in Glass and Ceramic Phases: Implications for Dispositioning Idaho High Level Waste (HLW) Calcine and Sodium Bearing Waste (SBW) Liquid,” pp. 47, Idaho National Engineering and Environmental laboratory (2001). Bechtel BWXT Idaho, LLC, Report No. INEEL/EXT-O1-01021 4. S.V. Raman, “Microstructures and Leach Rates of Glass-Ceramic Nuclear Waste Forms Developed by Partial Vitrification in a Hot Isostatic Press,” J. Materials Science, 33, [7]1679-1957(1998). 5 . K. Vinjamuri, S.V. Raman, D.A. Knecht and J.D. Herzog, “Waste Form Development for Immobilization of High Level Waste Calcine at the Idaho Chemical Processing Plant,” in High Level Radioactive Waste Management Proceedings of the Third International Conference, Las Vegas, Nevada, Vol.
2, 1261-1271(1992). 6. E.R. Vance, “Synroc: A Suitable Waste Form for Actinides,” MRS Bulletin, 19,[I2J 29-32(1 994). 7. A.E. Ringwood, S.E. Kesson, N.G. Ware, W. Hibberson and A. Major, “Immobilization of High Level Nuclear Reactor Wastes in SYNROC,” Nature, 278 [151 219-223 (1 979). 8. W. Sinclair and A.E. Ringwood, “Alpha-Recoil Damage in Natural Zirconalite and Perovskite,” Geochemical Journal, 15,229-243(1984). 9. W.J. Weber, “Radiation Induced Defects and Amorphization in Zircon,” J. Mater. Res., 5, [113 2687-2697(1 990). 10.P.A. Reiners, ‘‘(U-Th)/He Chronometry Experiences a Renaissance,” EOS, Transactions, AGU, 83,[3]21 -26(2002). 11.R.C. Evans, An Introduction to Crystal Chemistry; pp: 67-70,Cambridge University Press, Cambridge, UK, 1966. 12. W.A. Deer, R.A. Howie and J. Zussman, An Introduction to the Rock Forming Minerals; pp. 528, Longmans, Green and Co. LTD,London, UK,1967. 1 3.National Bureau of Standards, Soda-Lime-Silica Glass, Standard Sample No. 710,(1962). 14.R. Shuker and R.W. Gamon, “Raman Scattering in Amorphous Materials,” in Light Scattering in Solids, edited by M. Balkanski, F l m a r i o n Sciences, Paris, 334-338,1971.
Environmental Issues and Waste Management Technologies VIII
197
15. R.S. Czernuszewicz, “Resonance Raman Spectroscopy of Metalloproteins Using CW Laser Excitation,” in Methods in Molecular Biology, edited by C. Jones, B. Mulloy and A.H. Thomas,Humana press, Totowa, NY. 17, 345-374 (1 993). 16. R.J. Charles and F.E. Wagstaff, “Metastable Immiscibility in the B203-Si02 System,” J. Am. Ceram. Soc., 51, [l] 16-20 (1968). 17. E.M. Levin, C.R. Robbins and H.F. McMurdie, Phase Diagrams for Cerarnists, p. 260, The American Ceramic Society, Inc., Columbus, Ohio, USA, 1964. 18. EM. Levin, C.R. Robbins and H.F. McMurdie, Phase Diagrams for Cerarnists,p . 7, The American Ceramic Society, Inc., Columbus, Ohio, USA, 1964. 19. M. Hass, ‘‘Ramanspectra of vitreous silica, germania and sodium silicate glasses,”J. Phys. Chem. Solids, 3 1,415-422 (1970). 20. P. McMillan, “Structural studies of silicate glasses and melts - applications and limitations of Raman spectroscopy,” American Mineralogist, 69, 622-644 (1984). 21. R.J. Bell and P. Dean, “Atomic vibrations in vitreous silica,” Discussions of the Faraday Society, 50, 55-61 (1970).). 22. W.L. Konijnendijk and J.M. Stevels, “Structure of borate and borosilicate glasses by Raman spectroscopy,” in Borate Glasses, edited by L.D. Pye, V.D. Freshette and N.J. Kreidl, Materials Science Research, Plenum Press, 12,259279 (1977). 23. J. Tang and A.C. Albrecht, “Studies in Raman intensity theory,” Journal of Chemical Physics, 49,1144- 1154 (1965). 24. W.L. McCauley and D. Apelian, ‘(Temperaturedependence of the viscosity of liquids,” in Second International Symposium on Metallurgical Slags and Fluxes Proceedings, edited by H.A. Pine and D.R. Gaskell, 925-947 (1 984). 25. G. Urbain, Y. Bottinga and P. Richet, “Viscosity of liquid silica, silicates and alumino-silicates,” Geochimica et Cosmochimica Acta, 46, 1061- 1072 (1982). 26. D.A. Long, Raman Spectroscopy; pp 276, McGraw-Hill International Book Company, NY., USA, 1977. 27. S.V. Raman,“Raman spectra, structural units and durability of nuclear waste glasses with variations in composition and crystallization: implications for intermediate order in the glass network,” Philosophical Magazine A, 82, 3055-3085 (2002). 28. S.V. Raman,“Analysis of the hydrated zone in nuclear waste glass forms by electron microprobe, Raman spectroscopy and difision models,” Physics and Chemistry of Glasses, 42,27-41 (2001).
198
Environmental Issues and Waste Management Technologies VIII
Measurement of Simulated Waste Glass Viscosity R.F. Schurnacher, T.B. Edwards, and D.K. Peeler Westinghouse Savannah River Company Savannah River Technology Center Aiken, SC 29808 A.G. Blum Orton Ceramic Foundation Westerville, OH (614) 895-2663 ABSTRACT A small, high-temperature, glass viscometer instrument was established and evaluated using a simulated waste glass in a comparative test with eight other laboratories using their standard viscometers. Results from the comparison with standard viscometers indicated excellent accuracy and repeatabilitywith the smaller unit. This small unit exhibited distinct advantages in physical size, the amount of glass required for testing, and the simplicity of operation. These advantages can be particularly important for work with radioactive materials.
INTRODUCTION
The measurement of glass viscosity is very importantto the process of vitrifying radioactive waste streams. The viscosity is an important parameter in controlling the batch melting, refractory corrosion, and the ability to pour the glass [11. In radioactive waste vitrification programs, it is normal to measure the viscosity of a wide compositional range of non-radioactive glasses prior to the initiation of radioactive operations. The resulting information is then used to develop a viscosity-composition model that will predict the viscosity of the radioactive glasses [2]. The viscosity measurements must be both accurate and reproducible in order to obtain the most realistic and efficient model. In order to obtain accurate results, it is normal to frequently
test the viscometer unit with glass reference materials. These viscosity standards can be obtained from the National Institute of Standards and Technology (NIST)* or another controlled source of glass. These glass standards are relatively expensive and the viscometer may require hundreds of grams of glass for each calibration test. Simplicity of viscometer operation also improves the chance of reproducible results. In some cases the radioactivity of the glass may be low enough to permit measurement in a hood or a glove box and a small viscometer size is important for both insertion and storage in the containment area. The minimal amount of glass required for this viscosity measurement reduces the radiation exposure to the operator. *National Institute of Standards and Technology, Bldg. 202, Room 204, Gaithersburg,MD 20899
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
199
EXPERIMENTAL The Orton High-Temperature Viscometer meets the requirements of ASTM-C-965, Method A and has a small size with a foot-print of approximately 16 x 15 inches and a height of 50 inches. The unit is presented schematically in Figure 1 and in a photograph of the unit as shown in Figure 2.
Figure 1.
200
Schematic of Orton High Temperature Viscometer.
Environmental Issues and Waste Management Technologies VIII
Figure 2.
Photograph of Orton High Temperature Viscometer
Environmental Issues and Waste Management TechnologiesVIII
201
The platinum wound, vertical tube finnace is moved down to expose the heavy platinum crucible. A programmable controller with digital temperatureread-outs is used to control the furnace temperature. The viscosity is determined by measuring the percent torque on the platinum spindle with a digital read-out Brookfield RVDV-II+ viscometer. The glass temperature is measured by a thermocouple touching the under side of the platinum crucible and the furnace temperature is controlled by a separate thermocouple inside the fiunace. Vertical alignment of the unit, including the spindle to crucible alignment, is very critical to insure proper measurements. In addition, the unit should be set-up to provide a specific separation distance between the crucible bottom and the bottom of the spindle. The unit was initially standardized with the NIST glasses 717a (borosilicateglass), 7 10a (sodalime glass) and a limited amount of 71 1 (lead-silica glass). The 71 1 material had a viscosity range similar to most high-level waste glasses, but additional samples of this glass were not obtainable from NIST. The temperature range for these determinations was between 900 to 1500°C. The density of all the glasses was determined using the procedures in ASTM-C-693. After the density was determined, 2.60 cc of glass were weighed out and placed into the platinum crucible (6-7 grams). Using the published viscosity for the NIST glasses, a spindle constant was developed for the Orton High-Temperature Viscometer resulting in Equation 1. K = [NIST Viscosity (Poise)] x [RotationalSpeed(rpm)] / (% Torque)
1.
Where K is the calculated spindle constant and the rotational speed and % torque were obtained from the Brookfield Viscometer. It was found that the spindle “constant” varied as a function of temperature and a linear equation could be developed to describe this relationship. Further thermal probes of the furnace indicated that much of the spindle constant variation could be attributed to a slight thermal gradient in the vicinity of the crucible as the h a c e was heated to increasing temperature. It was concluded that by employing a linear relation for the spindle constant with temperature, the corrections for the thermal gradient and other errors could be readily corrected and the viscosity of the d o w n samples calculated. The spindle constant equation obtained fiom the NIST glasses is presented as Equation 2.
K(T) = 175.19 - 0.0493 x T
2.
Where K is the calculated spindle constant and T is the temperature in “C. The viscosity of the glass was then calculated from Equation 3. Viscosity(Poise) = K(T) x (% Torque)/ Rotational Speed (rpm)
3.
The viscometer is routinely calibrated by measuring a standard glass several times and determiningthe spindle constant - temperature relationship.
ROUND ROBIN EVALUATION The Pacific Northwest National Laboratory established a comparison of viscosity measurement techniques at various laboratories using a sample of Defense Waste Processing Facility (DWPF) Start-up Frit. This glass is an alkali-alumino-borosilicateglass roughly representative of glasses produced at the U.S. DOE high-level vitrification sites and was selected for its availability and thorough documentation [3]. The composition of this glass is presented in Table 1.
202
Environmental Issues and Waste Management Technologies VIII
Table 1. Chemical Composition of DWPF Start-up Frit in Weight YO* Oxide A1203 B203 BaO CaO Cr203 Fe203
--
Li20 MgO Mn02 Na20 NiO SiO, Ti02
1.2 0.1 13.5 2.6 3.5 0.7 2.3 11.5 1.1 49.0 1.o
Total
100.0
K20
m-
*
Tarpet (Wt?h) 4.7 8.8
--
Measured (Wt??) 4.60 8.5 1 0.10 1.47 0.09 14.20 2.70 3.25 0.84 2.37 11.53 1.1 1 47.90 1.18 0.11 99.96
Manufactured by Ferro Corporation Lot No. 10-27-87, SG-565 - Cleveland, Ohio
The large sample of glass was ground to a fine powder, blended, and split into individual lots of 1000 grams. The viscosity of samples was determined at eight different laboratories using their standard methodology and viscometers. The present investigationreceived lots 57, 58, and 59 for testing. Portions of each lot of glass powder were removed and melted in covered platinum crucibles for two hours at 1150 "C. The density of the unannealed glass was determined twice for each of the three samples and was found to vary between 2.680 to 2.690 g/cc. Based on these measurements, 2.60 cc of glass were weighed out and added to the Orton viscometer crucible. The crucible was placed in the viscometer furnace and heated to 120OOC for one hour. AAer the hold at 1200"C, the spindle torque, spindle rpm, and temperature were recorded. The temperature was then reduced in 50°C intervals to 950°C with 30 minute holds at each measurement temperature. The typical time for the complete determination at 6 hold temperatures was approximately 7 hours. The DWPF Start-up Frit was measured 11 times over a three-month period by the same technician. The viscosity was calculated from the measured temperature, the percent torque, and the spindle speed using Equation 3. Each set of viscosity measurements for the DWPF, Start-up Frit was fit to a Fulcher equation as shown in Equation 4. In (Viscosity) = A + B / (T - C )
4.
In this equation, In (Viscosity) represents the natural logarithm of the calculated viscosity (Poise), and A, B, and C represent the parameters of the Fulcher Equation. The temperature in ("C) is represented as T.
Environmental Issues and Waste Management Technologies VIII
203
The Fulcher parameters and viscosities calculated for the temperatures 1200, 1150,1100, 1050, 1000, and 950 are presented as Table 2. Examination of Table 2 shows very good agreement between the calculated viscosity values with a coefficient of variability ranging between approximately 1.O to 2.0 %. The slightly higher variability at the lower temperatures may be inherent in this instrument or may be due to the measurements being made in the vicinity of the liquidus temperature. The liquidus temperature for this glass was determined to be between 1025 to 105OOC [3].
Table 2.
Calculated Fulcher Parameters and Viscosities for DWPF Start-up Frit
Glass Date
Fulcher Parameters Calculated Viscositv (Poise) at TemDerature
(“C)
No.
Measured
57
9/20/00
-3.924 7471
175.8 306.9
170.9
57
9/22/00
-4.022 7675
156.1 282.9
159.6 96.0
57
10/17/00
-3.811 7304
185.3
311.1
58
9/19/00
-3.757 7261
182.9
58
9/25/00
-3.472 6795
206.7
58
10/18/00
-3.937 7583
164.9 305.6
58
12/13/00
-4.051 7743
155.0
59
9/21/00
-4.196 7996
143.5
59
9/26/00
59 59
B
1050
1100
1150
1200
101.8
64.1
42.3
29.1
60.9
40.4
27.9
173.1
103.1 65.0
43.0
29.6
301.4
168.9
101.2 64.1
42.6
29.4
290.1
163.0 98.1
62.5
41.8
29.0
171.4
102.6 64.9
43.0
29.6
295.6
166.1
99.6
63.0
41.7
28.8
297.3
166.9
99.8
63.0
41.6
28.6
-3.845 7300
181.3 284.5
159.3
95.4
60.4
40.1
27.7
10/19/00
-3.872 7428
175.2
303.3
169.6
101.4 64.1
42.4
29.2
12./15/00
-4.030 7698
159.8
302.4
169.4
101.2 63.9
42.3
29.1
Average Viscosity (Poise)
298.3
167.1
100.0 63.2
41.9
28.9
Coefficient of Variability (%)
2.07
1.60
1.32
1.16
1.20
A
C
950
1000
1.18
The resulting viscosity data has been plotted and presented in Figure 3 along with the equation and final compiled curve from the “Round Robin” comparisonand the eight independent Version 4.0 from SAS laboratory detenninations. The nonlinear fitting platform of the .IMPTM Institute, Inc. [4] was used to determine a Fulcher equation for all of the available round robin viscosity data from the eight contributing laboratories(Equation 5).
204
Environmental Issues and Waste Management Technologies VIII
Ln (Viscosity)= -5.2892 + 5831.114 / (T -280.1372)
5.
In this equation the viscosity is in Pascal seconds (Pa.s) and T is the temperature in “C [5]. To convert Pascal seconds to Poise multiply the resulting Pascal seconds by a factor of ten. The temperature range for this model was from 450 to 1250°C. 700 600
500 n
Q) v)
-5 a. 400
.-E 8 300
U
U)
U)
5
200 100
0 800
850
900
950
I000 1050 1100 1150 1200 1250 1300 Temperture (Celsius)
Figure 3.
A Plot of the Measured Viscosities over the Temperature Range between 900 and 1250°C Compared to the Final Calculated Viscosity Curve from the Round Robin Evaluations.
CONCLUSIONS The Orton High Temperature Viscometer was shown to provide accurate and precise viscosity determinationsover a temperaturerange between 950 to 1200°C. The repeatability of the determinations was very good and within 1 to 2% coefficient of variability. The accuracy of the results compared to a “Round Robin” evaluationbetween the derived Fulcher Equation was demonstrated graphically.
Environmental Issues and Waste Management Technologies VIII
205
The instrument is relatively easy to establish and operate. It requires a minimal amount of glass, between 6 and 7 grams of material, and the small sample size may be critical in cases where the glass is radioactive. The unit’s small size and foot print would permit operation in controlled areas such as glove boxes or hoods.
1. 2. 3.
4. 5.
Bickford, D.M., Applewhite-Ramsey,A., Jantzen, C.M., and Brown, K.G., “Control of Radioactive Waste Glass M e l t e d , Preliminary General Limits at Savannah River,: J. Am. C-. SOC.73 [lo], 2896-2902, (1990). Jantzen, C.M., First Principles Process-ProductModels for Vitrification of Nuclear Waste: Relationship of Glass Compositionto Glass Viscosity, Resistivity, Liquidus Temperature, and Durability,” Ceramic Transactions, 23 37-5 1, (1992). Jantzen, C.M., “Characterization of the Defense Waste Processing Facility ( D W F ) Startup Frit-0,” WSRC-RP-89-18,1989. SAS Institute, Inc. JMPTM Statisticsand Graphics Guide: JMP Version 4,SAS Institute, Inc. Cary NC, 2000. Results of Round Robin Comparison - To be published at a future date.
ACKNOWEGEMENTS It is important to recognize the contribution of R.J. Workman, fiom the Savannah River Technology Center, who performed all of the viscosity determinations with this unit and J.D. Vienna, fiom the Pacific Northwest National Laboratory, who established and coordinated the Round Robin comparison. Preliminary establishment of the viscometer was carried out by the Clemson Environmental Technologies Laboratory, Clemson University. Funding was provided by the U.S. Department of Energy under Contract No. DE-AC09-96SR18500.
206
Environmental Issues and Waste Management TechnologiesVIII
Hanford Tank Waste Treatment
This page intentionally left blank
HANFORD LOW-LEVEL WASTE FORM PERFORMANCE FOR MEETING LAND DISPOSAL REQUIREMENTS R. F. Schumacher, C. L. Crawford, N. E. Bibler and D. M. Fenara, Savannah River Technology Center, Savannah River Site Building 773-4 1A Aiken, SC 29808
H. D. Smith, G. L. Smith and J. D Vienna, Pacific Northwest National Lab P.O. BOX999 / MS K6-24 Richland, WA 99352
I. L. Pegg and I. S. Muller Catholic University of America 620 Michigan Ave. N.E. Washington, D.C. 20064
D. B. Blwnenkranz and D. J. Swanberg, Bechtel National, Inc. 3350 George Washington Way Richland, WA 99352 ABSTRACT Immobilized Low-activity waste (ILAW) from the Hanford site will be disposed of in near-surface burial grounds and must be processed into a chemically durable waste form to prevent release of hazardous constituents to the environment. To meet this goal, the LAW will be immobilized in borosilicate glass. The DOE Office of River Protection and the River Protection ProjectWaste Treatment Plant (RPP-WTP) project have agreed on testing requirements that the immobilized LAW glass must meet to demonstrate chemical durability. Two of the tests are the Product Consistency Test (PCT)' and Environmental Protection Agency's (EPA) Toxicity Characteristic Leaching Procedure (TCLP)? This paper provides results of RPP-WTP PCT and TCLP testing on both actual radioactive and non-radioactive simulant LAW glasses to show they meet the associated land disposal requirements. INTRODUCTION The LAW glass produced from the planned RPP-WTP is to be stored in a near surface vault repository on the Hanford site. Detailed descriptions of existing disposal vaults as well as conceptual designs for new ILAW disposal ~
~
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
209
facilities at the Hanford site have been previously pre~ented.~ Borosilicate glass is the selected Tri-party agreement waste form for immobilization of Hanford LAW. The borosilicate glass waste form has proven stability and durability. Per the Department of Energy (DOE) Office of River Protection and Bechtel National, Inc. contract: the ILAW glass normalized mass loss of sodium, silicon, and boron shall be less than 2.0 grams/m2 using a seven-day ASTM- C1285-97 procedure’ ‘Product Consistency Test’ run at 90 OC. The ILAW product shall also be acceptable for land disposal under Resource Conservation and Recovery Act (RCRA) Land Disposal Restrictions (LDR) 40CFR268: and thus must show compliance via the EPA’s TCLP. BACKGROUND AND OBJECTIVES Four Hanford tank supernatant samples were processed through representative unit operations for High-Level Waste (HLW) pretreatment inchding entrained solids removal, strontium/transuranic (Sr/TRU) precipitation and filtration, and cesium and technetium removal via ion exchange. These processes remove most of the waste radioactivity. The resulting decontaminated supernatant samples were converted into LAW glass by consideration of their respective analytical characterizations. The LAW pretreated waste supernatants were blended with certain mineral glass formers to produce crucible-scale melter feeds. Vitreous State Laboratory (VSL) personnel of Catholic University of America (CUA) provided the target glass compositions and also fabricated simulated LAW glasses that were similar, but not exactly the same composition as the radioactive waste glasses. Two different Hanford tank wastes of the Envelope C group (Tanks 241-AN-107 & AN-102) and two different tank wastes of the Envelope A group (Tanks241-AW-101 and AN-103) were used in this study. The overall objective of this work was to show compliance with the RPP-WTP contractual durability requirements via the PCT and TCLP testing of crucible melt glass produced from actual radioactive Hadord tank samples. LAW GLASS FABRICATION AND CHARACTERIZATION The radioactive LAW glasses produced for this study were made from melter feed slurries that were dried, calcined and melted in platinum crucibles at 1150 *C. The mineral glass formers used consisted of kyanite (Al2Si03), orthoboric acid (H3BO3), wollastonite (CaSi03), red iron oxide pigment (Fe203), olivine (Mg2Si04), silica sand (SO*), rutile ore (TiOz), zinc oxide (ZnO), zircon sand (ZrSiO4), lithium carbonate (LiCO3). Sugar (C12H22011) was added to control glass oxidation state. Glass formulations considered sodium as the main waste component loading indicator with a minimum of 10 wt% Na20 for the Env. C glasses and a minimum of 14 wt% Na20 for the Env. A glasses. Further details describing the pretreatment of waste streams, fabrication of glasses and associated analytical characterization of the product glasses can be found in technical references.
210
Environmental Issues and Waste Management Technologies VIII
PRODUCT CONSISTENCY TEST The glass durability test known as the PCT is an ASTM procedure (ASTM C1285-97) that tests the durability of crushed glasses over a seven-day test at 90 "C. The test uses glass particles sized between 74 and 149 microns, or 100 to 200 mesh size. The sample size is typically > 1 g, with leachate volume to sample mass equal to IOX. Tests are performed in unsensitized Type 304L stainless steel. The initial and final pH values of the leachate solutions are measured and the filtered, acidified final product leachates are characterized for soluble components leached fiom the crushed glass. Results are reported as normalized elemental mass releases. Normalization considers the measured concentration of the element in the leachate and the concentration (wt%) of the same element in the glass. The PCT results fiom durability tests with glasses used in this study are presented in Table I. Tables I shows normalized release data for the radioactive glasses and their respective surrogate glasses. Data is also shown for a nonradioactive reference glass, the Low-Activity Reference Material (LRM) glass.7 The data in Table I clearly show that normalized mass release for all analytes are well below the contract specified maximum of 2 g/m2 for the seven-day PCT at 90 OC. Comparison of the surrogate glass leach data to their respective radioactive glasses also indicates good agreement even though the surrogate and radioactive compositions were not exactly the same.
-
TOXICITY CHARACTERISTICLEACHING PROCEDURE (TCLP) The TCLP is specified by EPA SW-846 Method 1311 and uses glass particles that are capable of passing through a 9.5-mm (0.375-in.) standard sieve. The glass sample is placed into an extractor vessel with an extraction fluid mass equal to 20 times the mass of the glass sample. The extraction fluid used for these tests was TCLP extraction fluid #1 consisting of 5.7 mL glacial acetic acid, 64.3 mL of 1N NaOH in one liter of ASTM water with resulting pH of 4.93. The extractor vessel containing the sample is rotated end over end at 30 rpm for 18 hours at room temperature of 22 OC. The resulting liquid is then separated from the glass particles by filtration and the leachate is analyzed for the analytes of concern. Tables I1 and I11 present the TCLP test results for the four different radioactive glasses tested in this work. The four radioactive glasses were produced fiom pretreated alkaline supernatants that are generally very low in concentrations of RCRA metals. The data shown in these tables demonstrate that actual waste glasses easily meet Universal Treatment Standards (UTS).' Similar data has been collected in tests with simulated ILAW glasses spiked with elevated concentrations of RCRA metals.8
Environmental Issues and Waste Management Technologies VIII
21 1
Table 11. TCLP Test Results for Tanks AW- 101 and AN- 107
1
UTS (40CFR268) TCLPLevels
I
AW-101 Leachate
I
Glass
AN-1 07 Leachate ug/mL 0.0011 0.008 0.04 <0.03 0.0 13 0.29 0.33 0.028 <0.0007
I Glass ug/g <0.0003 <0.006
240 <0.03 40 235 430 38 <0.0007
<0.013
X0.013
<0.01
c0.01
<0.001
<0.001
* Indicates vitrification is the LDR treatment standard for this metal - UTS shown for information only. # Se has a UTS (5.7 mg/L TCLP) that is greater than the toxicity characteristic, therefore the Se toxicity characteristic level of Washington State Acceptance Criteria (WAC) 173-303-090 is shown.
212
Environmental Issues and Waste Management TechnologiesVIII
Table 111. TCLP Test Results for Tanks AN-1 03 and AN-1 02
I
Analyte Ag* As* Ba* Be Cd* Cr* Ni Pb* Sb Se # T1 Hg*
I I
UTS (40CFR268) TCLP Levels ug/d 0.14 5.0 21 1.22 0.11 0.60 11 0.75 1.15 1 .o 0.20 0.025
AN- 102
AN-103 Leachate udmL <0.04 1 <0.005 2.9
Glass 100
-a
<0.002 <0.012 <0.025
I I
~0.064 <0.3 <0.2 0.015 <0.06 <0.0001 I I
37 <0.9 26 230 180 150 <77
I
<5
eoo 0.44
I I
Leachate WmL <0.0046 <0.02 2.46 <0.0009 <0.0013 0.0084 0.0094 <0.033 <0.02 1 4 . 0 16 <0.067 <0.0005
Glass udg -400 <350 65 X0.45 X2.4
300 140 <85
I I
<38 <30 420 <0.1
1
* Indicates vitrification is the LDR treatment standard for this metal - UTS shown for information only. # Se has a UTS (5.7 mgL TCLP) that is greater than the toxicity characteristic, therefore the Se toxicity characteristic level of Washington State Acceptance Criteria (WAC) 173-303-090 is shown. CONCLUSIONS Data presented in this work address the overall objective aimed at showing compliance with the RPP-WTP contractual durability requirements via the PCT and TCLP testing of crucible melt glasses produced from actual radioactive Hanford tank samples. Comparison of normalized release rates from PCT’s on both actual and surrogate LAW glasses to the contract specified maximum of 2 g/m2 indicates that all of these glasses readily meet this durability standard. The ILAW glasses also effectively immobilize RCRA metals in actual waste samples.
REFERENCES ASTM C 1285-97 (1998). SfandardTest Methodsfor Determining Chemical Durability of Nuclear Waste Glasses: The Product Consistency Test (PCT). American Society for Testing and Materials, Easton, Maryland. SW-846, Test Methods for Evaluating Solid Waste, PhysicaVChemical Methods. U.S. Environmental Protection Agency, Office of Solid Waste and Emergency Response, Washington, D.C.
Environmental Issues and Waste Management Technologies VIII
213
R. Puigh, “Appendix I, Disposal Facility Data for the Hanford Immobilized Low-Activity Tank Waste Performance Assessment,” HNF-5 636, Rev. 0, HNF-4950, Rev. 1, Fluor Federal Services, Richland, Washington, (December 1999). River Protection Project - Waste Treatment Plant (RPP-WTP) Contract, Bechtel National, Inc.-- Design, Construction, and Commissioning of the Hanford Tank Waste Treatment and Immobilization Plant: Contract No. DE-AC27OlRV 14136, http://www.hanford.gov/orp/, (October 2002). 40CFR268. Land Disposal Restrictions. Code of Federal Regulations. U.S. Environmental Protection Agency, Washington, D.C. G. L. Smith et al., “Vitrification and Product Testing of AW-101 and AN-107 Retreated Wastes,” PNNL-13372, WTP-RPT-003, Rev. 0 (October 2000); C. L. Crawford et al., “Crucible-Scale Active Vitrification Testing Envelope A, Tank 241-AN-103,” WSRC-TR-2000-00322, SRT-RPP-200000021, Rev. 1, (June 2001); C. L. Crawford et al., “Crucible-Scale Active Vitrification Testing Envelope C, Tank 241-AN-102,” WSRC-TR-2000-00371, SRT-FWP-2000-00022, Rev. 0 (June 2001). W. L. Ebert and S. L. Wolf, “Dissolution Test for Low-Activity Waste Product Acceptance,” Argonne National Laboratory, Proceedings of Spectrum ’98, Denver, CO, Sept 13-18, 1998, pp. 724-73 1. I. S. Muller, A. C. Buechele and I. L. Pegg, “Glass Formulation and Testing with RPP-WTP LAW Simulants,” VSL-OlR3560-2, Rev. 0, (February 23, 2001).
’
214
Environmental Issues and Waste Management Technologies VIII
LEACHING MECHANISM OF BOROSILICATE GLASSES UNDER TCLP CONDITIONS Hao Gan and Ian L. Pegg Vitreous State Laboratory The Catholic University of America Washington, DC 20064
ABSTRACT The toxicity characteristic leaching procedure (TCLP) is a prescribed regulatory test that involves contacting the sample with a sodium acetate buffer at room temperature for 18 hours. Our previous work has investigated the effects of glass composition on TCLP release and, in particular, showed that the normalized elemental releases fall into three groups with approximately congruent leaching within each group but significant rate differences between the three groups. In this study, we have investigated the time dependence of the release from waste glasses under TCLP conditions from less than one hour to nearly two months. Tests were also performed with and without the prescribed constant end-over-end agitation. The leachates were analyzed and the leached glass samples were subjected to microscopic characterizationof the altered surface layers that formed. The results demonstrate that agitation plays an important role in the TCLP test and results in significantly increased rates of release and thinner surface layers due to abraqive removal of the surface layers. This process also underlies the previously observed congruent leaching within distinct element groups. Results from two-step leach tests under TCLP conditions but without the end-over-end agitation show that the surface layers that are formed can act as effective mass transport barriers, their presence producing significant reductions in release rates. INTRODUCTION In the United States, the Toxicity Characteristic Leaching Procedure (TCLP) promulgated by the Environmental Protection Agency' is the basis for determination of hazardous characteristics of wastes and treated wastes. In particular, glasses that are being developed for vitrification of tank wastes at the Hanford site must meet this requirement with respect to Land Disposal Restrictions (LDR),as well as delisting in the case of vitrified high-level wastes
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management TechnologiesVIII
215
(HLW). Given the significance of the TCLP, it is desirable to understand the corrosion behavior of glasses under TCLP conditions as well as the effects of glass composition on TCLP performance. To this end we have previously reported24 quantitative relationships between glass composition and TCLP response based on the observations that the normalized element releases fall into groups that follow rather regular proportionalities. Our previous work2" examined the leaching behavior of nearly 30 elements (including not only the standard RCRA elements but also all of the major glass constituents) over wide ranges of glass composition. It was found that the alkalis, alkaline earths, divalent transition mebls, boron, silver, and uranium were released nearly congruently ("advanced" elements). In a similar way, the P-block elements, Si, T1, Se, Sb, and Pb, were released at a similar but significantly lower rates than elements in the first group ("retarded elements"). Finally, Al, Fe, Zr, Cr, and As were released most slowly and exhibited comparatively irregular trends ("slow or irregular elements"). While to some extent, these trends are consistent with expectations based on gross solubility considerations: particularly in the case of the latter group, these striking elemental groupings cannot be explained on that basis alone. An improved understanding of the features of the reactions that give rise to such phenomena as well as the mechanism of glass corrosion under TCLP conditions is clearly desirable from the perspective of the development of improved glass Table 1. Target glass compositions (wfoh) of selected HLW glasses. formulations and the control of the TCLP response of such formulations during Glass HLW98-31 HLW99-98 HLW99-58 production. Consequently, the present A1203 work investigated the leaching behavior of representative HLW glasses under standard TCLP conditions as well as under several non-standard variants of the CdO 2.001 2.00 10.39 TCLP that examined the time Fe203 1 .ool 1.00 0.17 dependence, the effects of agitation, and K20 6.00 3 .OO 0.00 the effects of altered surface layers that Li20 0.06 1.25 0.44 MgO form on the test glasses. WO.,
EXPERIMENTAL Three representative HLW glasses with TCLP results ranging from excellent to mediocre to oor were selected from a large data set!T ' he compositions of the three selected glasses are listed in Table 1. The standard TCLP consists, in summary, of extracting crushed glass
216
Na20
NiO p205
Sio2
SrO TiO? ZnO zr0.2
3.03 6.59 0.54 0.13 45.53 2.32 0.06 2.00 3.56
6.39
10.00
1.88 0.25 30.00 9.81 0.63 2.00 3 .OO
0.20 20.00 0.66
0.00 30.00 0.20 0.22 2.00 0.00
Environmental Issues and Waste Management Technologies VIII
(100 grams, <3/8") in a sodium acetate buffer solution (2 liters, pH = 4.93) for 18 hours at room temperature with constant end-over-end agitation. The surfacearea-to-volume ratio is about 20 m-', The leachate concentrations were measured by direct current plasma atomic emission spectroscopy (DCP-AES). The overall uncertainty associated with this test is estimated to be about k15 %. To investigate the time evolution of TCLP leachate concentrations, the tests were continued to much longer times, during which a small volume of solution (about 20-40 ml fiom a total of 2000 ml)was removed at various intervals. The negligible effect of the interruptions and the decrease in total solution volume due to sampling was confirmed by conducting uninterrupted parallel tests simultaneously for the same time interval; no significant differences were observed. In addition to variation of the test duration, non-standard "static tests" (as opposed to the standard "dynamic test") were performed under otherwise identical TCLP conditions to investigate the impact of the mechanical agitation. For these tests, a magnetic stirrer was used to provide agitation of the TCLP leachant without the abrasion of the glass pieces that occurs in the standard TCLP as a result of the constant end-over-end agitation. In each of these tests, the leached glass pieces were also sampled to permit characterization of the surface layers. Finally, two-stage tests were performed in which glass samples that had been previously leached under TCLP conditions were subjected to leaching in a fresh TCLP leachant solution. Glass samples that were removed fiom the TCLP leachates were sectioned and examined by scanning electron microscopy coupled with energy dispersive xray spectroscopy ( S E W D S ) in order to characterize the surface layers that were formed in the leaching process.
RESULTS AND DISCUSSION Surface Layer Development in Dynamic TCLP Tests SEM micrographs of HLW9958 glass sampled after 1 hour, 8 and 24 hours during TCLP tests showed the effects of agitation on surface
Figure 1. SEM images of the cross section of HLW9958 glass after dynamic TCLP test for 1 hour; 8 hours; and 24 hours. An altered layer (the grey area) has developed on the surface of the reacted glass.
Environmental Issues and Waste Management Technologies VIII
217
layer damage (Figure 1). A surface layer of about 2-3 microns thick developed aRer only one hour of dynamic TCLP testing (Figure 1). However, the thickness of the layer had not grown significantly after 8 hours of continuous tumbled leaching and was almost completely removed after 24 hours, leaving altered material only in pockets, as shown in Figure 1. Apparently, the continuous end-over-end tumbling that is mandated in the standard (dynamic) TCLP test results in considerable abrasion of the colliding glass pieces, which gradually removes any altered surface layers that form. To confirm that this was the case and to further investigate the effects of agitation, tests were performed without the end-over-end agitation (static tests). Surface Layer Development in Static TCLP Tests Static TCLP tests were 2. SEM images of the cross section of in the end-over- Figure HLW99-58 glass after static TCLP test for 1 end a@tation was by hour; 2 hours; 4 hours; 8 hours; and 24 hours. simple COn~UOUSstirring of the An altered layer (the grey area) has developed leachant. SEM micrographs of the on the surface of the reacted glass. same glass (HLW99-58)sampled after 1 hour, 2 hours, 4 hours, 8 hours, and 24 hours during static TCLP testing show that an altered surface layer grew continuously up to approximately 12 microns in 24 hours (Figure 2). The cracks perpendicular to the layer have most likely developed during the drying process after the TCLP tests. There is little evidence of damage to or spalling of the layers in the static tests. This observation confirms that the end-to-end tumbling in the standard (dynamic) TCLP test is effective in removing the altered surface layer as it develops. Such significant differences in surface layer formation would be expected to be reflected in TCLP leachate concentrations, as discussed below. Formation and removal of the surface layer during the dynamic test may be responsible for many characteristic features of glass leaching in the TCLP test.
218
Environmental Issues and Waste Management Technologies VIII
This process would be expected to accelerate the leaching process by continuously exposing fresh and more reactive glass surface. This effect likely is at least partially responsible for the fact that normalized glass leach rates under TCLP condition at room temperature are considerably higher than those in the Product Consistency Test (PCT, ASTM C 1285), despite the higher temperature (90°C), SN, and pH of the latter. Another feature that is unique to the TCLP is the surprisingly uniform element release rates of chemically dis- Figure 3. SEM/EDS line scans of selected elements in the parate elements leached layer on HLW99-58,a) Ca, Na, K, Fe and 0;b) Si and after simple Al. The arrow indicates the location of the interface between the leached layer and the unaltered glass. The horizontal axis normalization by marks the distance from the surface toward the interior of the their concentra- cross- sectioned glass sample in microns. tions in the glass.3~~ It would seem unlikely that a purely chemical process could be responsible for the observed grouped leaching behavior and, in fact, in the absence of mechanical abrasion (static tests) these groupings largely disappear. SEM-EDS line scans (Figure 3) of selected elements perpendicular to the surface layers of altered glasses after static TCLP tests reveal that the grouped leaching rates of the ''advanced'', "retarded" and "slow or irregular" are consistent with their affrnity for the altered surface layer. The advanced elements, such as NayK, Ca, etc., are depleted fi-om the surface layers developed on HLW99-58 (Figure 3) and HLW99-98 (Figure 4). In contrast, the retarded element Si and slow elements A1 and Fe are substantially enriched in the altered surface layer. For both HLW99-58 and HLW99-98, a concentration gradient is clearly evident for each of the elements analyzed. Impact of Suflace Layers on TCLP Leachate Concentrations Figure 5 shows that the thickness of the surface layers grows linearly with the
Environmental Issues and Waste Management TechnologiesVIII
219
square root of time, which suggests the layer development is a diffbsioncontrolled process. It is also important to note that the layer growth rate varies drastically from glass to glass. As shown in Figure 5 , the layer formed on the least durable of the three glasses, HLW99-58, was 20 times as thick as that on the modestly durable glass, HLW99-98, after 24-hour static tests; the most durable glass, HLW9831, showed no sign of alteration after a 50-day static test (Figure 6). Also shown in Figure 5 (insert) are the normalized cadmium TCLP Figure 4. SEM/EDS line scans of selected elements in releases together with the the leached layer on HLW99-98. The arrow indicates surface layer thicknesses for the location of the interface between the Ieached layer the three glasses studied; and the unaltered glass. The horizontal axis marks the distance from the surface toward the interior of the cadmium is representative of cross-sectionedglass sample in microns. the "advanced" group of elements. Glasses with thicker altered surface layers released cadmium faster than glasses with thinner surface layers. The normalized standard TCLP cadmium release for HLW99-58 is over a factor of 100 times that of HLW98-3 1.
8 g
$2 'O
9 $ '
*
m ~ ~ l v w p u * * ~
Figure 5. Surface layer growth in static TCLP tests on mwgg-58, ~ ~ ~ 9 9 - and 9 8 HLW98-31. Insert shows surface layer formation rates and Cd releases after 18 hours dynamic TCLP tests normalized by Cd mass fractions in the corresponding glass.
220
Figure 6. SEM h a g e of the cross section of HLW98-31 glass after static TCLP test for 50 days- No surface layer is dkCernable-
Environmental Issues and Waste Management Technologies VIII
one of cause and effect. Thus, it could be argued that the layer forms simply as a consequence of the leaching process under TCLP conditions and that the thickness of the layer is merely a manifestation of the degree to which a glass has been leached. Conversely, a growing surface layer could serve as a mass transport barrier, which could then directly influence the reaction progress and, therefore, the observed TCLP leachate concentrations. The latter possibility appears likely in light of the concentration gradients observed in the leached layers. Accordingly, M e r experiments were performed to investigate this hypothesis. In previous studies,8-" two-step (or sometimes multi-step) leaching experiments have been performed in efforts to determine whether the typically observed diminishing of the reaction rate with time is due to decreasing reaction affinity ("saturation" effects) or due to the growth of a surface layer that acts as a mass transport barrier. In such experiments, a glass sample is leached for a certain period of time and then transferred to fresh leachant solution. If saturation effects are dominant, the kinetics in the second step should simply replay those observed in the first step; conversely, if an effective mass transport barrier developed in the first step, the rate should be little changed by replacing the leachant solution, and the kinetics in the second step should essentially continue from those of the first step. In such experiments, Chick et concluded that saturation effects were dominant, while Xing et al.' concluded that surface layer effects were significant and even dominant, depending on the glass composition; more recent experiments of this kind" are consistent with the latter conclusion but arguments to the contrary have also been made." In 150 the present work, a similar aa, experimental protocol was employed to investigate glass leaching under il TCLP conditions. HLW99-58glass was selected for the comparison experiments since a relatively thick layer had been Filled symbolsare from fresh glass samples. observed to form on this glass in a. 1E Q Open symbols are from preleached glass sample. n: relatively short period of time. Figure 7 shows the boron TCLP release versus time in dynamic tests on HLW99-58in four independent sets of Figure 7. B TCLP release from HLW99-58 experiments; boron is representative of glasses during dynamic TCLP tests. The the "advanced" group of elements. As filled symbols are from three independent TCLP tests. The open circles are from a described above, three sets of data dynamic TCLP test using an HLW99-58 relate to the leaching of pristine glass, glass sample that had been previously while the fourth set relates to the leached under dynamic test conditions. leaching of a sample of glass that had
@'
I
i
Environmental Issues and Waste Management Technologies VIII
221
previously undergone 24 hours of dynamic testing. No significant difference is observed between the glasses with or without prior leaching. This result is not surprising since we have shown earlier that the end-over-end tumbling during the dynamic TCLP would have effectively removed any surface layer that formed. Consequently, similar .tests were also performed using the static TCLP in which the surface layers are not removed. Figure 8 shows the results from static 8. B and Si TCLP release from TCLP tests; one set of data relates to the Figure fresh and pre-leached HLW99-58 glasses leaching of pristine HLW99-58 glass, during static TCLP tests. The filled while the second set relates to the symbols are results from a fresh glass, leaching of a sample of glass that had static TCLP test. The open circles are previously undergone 24 hours of static results from a static TCLP test using an glass that had been previously testing. Both boron ("advanced") and HLW99-58 leached under static test conditions. silicon ("retarded") releases are drastically impacted by the presence of the surface layer, both exhibiting rate reductions of about a factor of two. These results suggest that surface layers that develop on the glass during TCLP testing can indeed act as mass transport barriers that serve to reduce the release rate. Unfortunately, similar tests with the very durable glass HLW98-31 are confounded by the very low leach rates for that glass. However, it is likely that HLW98-31 is an "intrinsically" durable glass that naturally forms very thin altered surface layers, which then may or may not provide significant mass transport barriers. Conversely, in the case of the poorly durable HLW99-58, relatively thick surface layers form quickly, which then provide mass transport barriers that have a clear and significant impacts on the release rates.
CONCLUSIONS Surface layer formation and removal is an important factor in TCLP release. Without the prescribed end-over-end agitation, the surface layer grows parabolically with time and the layer formation rate varies greatly depending on the glass composition. The results from the present work suggest that glasses that exhibit faster layer growth show higher TCLP leachate concentrations. The results of two-step leach tests indicate that surface layers can act as mass transport barriers to glass corrosion under TCLP conditions in the absence of mechanical removal of the layers. This conclusion is also supported by the observed concentration gradients within the surface layers.
222
Environmental Issues and Waste Management Technologies VIII
The composition of the surface layers that form in the TCLP is generally consistent with the selective leaching process reflected in the TCLP leachate compositions. The "advanced" elements are depleted while the "retarded" and "slow" elements are enriched in the layers developed and preserved during static TCLP experiments. The constant end-over-end agitation prescribed in the standard TCLP results in mechanical removal of the surface layers by abrasion between the glass pieces that constitute the test sample. This process ensures nearly congruent TCLP dissolution for elements within distinctive groups with each group differing by their relative normalized release rates; consequently, this process is essential to the simple ratios reported previo~sly?~~ The resulting mechanical removal of the surface layers leads to significantly increased TCLP leachate concentrations as compared to tests performed without the end-over-end agitation. ACKNOWLEDGMENTS The authors would like to thank the staff of the VSL, particularly Chu-Fen Feng, for support with glass preparation, testing, and analytical work and M.C. Paul for manuscript preparation. REFERENCES 1.
2.
3. 4. 5. 6. 7.
8. 9.
10. 11.
US Environmental Protection Agency, SW-846, Method 13 1 1, Toxicity Characteristic Leaching Procedure. S.S. Fu and I.L. Pegg, Ceramic Transactions, Eds. G.T. Chandler and X. Feng, vol. 107, pp. 261, (1999). H. Gan and I.L. Pegg, Ceramic Transactions, Eds. G.L. Smith, S.K. Sundaram and D.R.Spearing, vol. 132,335, (2002). H. Gan and I.L. Pegg, Development of Property Composition Models for RPPFJTP HL W Glasses, VSL Final Report, VSL-OlR3600-1, July, 2001. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, National Association of Corrosion Engineers, Houston, Texas, USA, (1974). W.K. Kot and I.L. Pegg, Glass Formulation and Testing with RPP-WTP HLW Simulants, Final Report, VSL-OlR2540-2,February 16,2001. S.S. Fu and I.L. Pegg, Glass Formulation and Testing with TWRS HLW Simulants, VSL Final Report, January 1998. L.A. Chick and L.R. Pederson, Mat. Res. Soc. Symp. Proc., vol. 26,635, (1984). S.-B. Xing, A.C. Buechele, and I.L.Pegg, Mat. Res. Soc. S'mp. Proc., vol. 333, 541, (1994). A. Gauthier, P. Le Coushuner, and J.-H. Thornassin, Mat. Res. Soc. Symp., Proc., vol. 713,555, (2002). B.P. McGrail, J.P. Icenhower, and E.A. Rodriguez, Mat. Res. S'mp. Soc. Proc., vol. 713,537, (2002).
Environmental Issues and Waste Management Technologies VIII
223
This page intentionally left blank
ELECTROCHEMICAL STUDIES OF SULFATE-CONTAININGWASTE GLASS MELTS Igor Vidensky, Hao Gan, Andrew C. Buechele and Ian L. Pegg Vitreous State Laboratory, The Catholic University of America, Washington, DC ABSTRACT The sulfate content of wastes that are scheduled for treatment by vitrification can be the waste-loading-limiting factor as a result of the limited solubility of sulhr in silicate melts. Excess s u l k results in the formation of an undesirable molten salt phase on the surface of the glass melt. Consequently, on-line measurement of the sulfur content of the glass melt and the detection of a separate salt phase in off-normal conditions is of practical importance. In the present work, square wave voltammetry was shown to be very sensitive to the sulfate concentration in a representative molten glass. A differential current signal derived from the measurements correlated linearly with the independently measured so3 content of the glass melt over the range from about 0.2 to over 1 wt%. In addition, the onset of a separate molten sulfate phase was clearly evident in the measured voltammograms as both a characteristic step and the appearance of fluctuations at potentials more reducing than about -600 mV. These results suggest the possibility of relatively simple in situ real-time measurements that could determine both the sulfur content of the glass melt and the onset of an undesirable molten sulfate phase. Such a capability would provide a valuable process control element for waste glass melters that are processing feeds with high s u l h contents. INTRODUCTION Many waste streams that are scheduled for vitrification, including the Low Activity Waste (LAW) fraction fiom the Hanford tanks, contain significant amounts of sulfate. Since s u l k has a relatively low solubility in silicate melts it can become the waste-loading-limiting constituent in the waste'-3. This is especially true for wastes that are also high in sodium content, as is the case for typical LAW streams. Excess sulfate can lead to the separation of a molten sulfate phase during vitrification as a result of both solubility factors and kinetic constraints on the rate of sulfur uptake by the melt'-3. Formation of such a phase is undesirable fiom many perspectives, since the phase is generally more corrosive, more electrically conductive, more fluid, and lower melting than the underlying glass'". Typically, these issues are addressed entirely by control of the chemical To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
225
composition of the feed to the vitrification facility. However, for higher-sulfate waste streams, it would also be desirable to be able to monitor the sulfate content of the glass melt and to be able to detect the onset of the formation of a molten salt phase under off normal conditions. Ideally, these determinations would be made in real time from in situ measurements. The present work investigated the potential application of electrochemicaltechniques to this problem. Pulse voltammetry has long been used for investigation of the electrochemical behavior of aqueous solutions and, over the past decade, also of molten glass systems4y5. Now employing computer-controlled modem electronics, square-wave voltammetry has gained wider application to redox reactions in molten glass systems as a result of its excellent sensitivity and high resolution6-'6.In particular, the redox equilibria of sulhr species have been investigated in simple silicate melt systems'7718.However, we know of no previous work that has been done to develop a systematic electrochemicalmethod to quantitatively monitor the sulfate content in complex borosilicate waste glass melts and to detect the presence of a molten salt layer in situ. S u l h can be incorporated into silicate systems as many different chemical forms and valence state^'^*^^. The change of oxidation state of sulfur species has long been used in the glass industry for fining of commercial glass. Consequently, it is possible that the suIfur content in a waste glass melt could be quantified by its characteristic reduction peak in a voltammetric measurement as sulfur species change from one oxidation state to another. As a separate molten salt layer develops on the top of the glass pool, changes in the electrochemical properties of the phase-separated system would also be expected, which it may be possible to detect in the same voltammetric measurement for use as an indicator of salt formation. The purpose of this study was therefore to investigate electrochemical methods to monitor the sulfate content in molten waste glasses and the feasibility of an early sulfate layer detection technique that may ultimately be implementable in a waste glass melter.
EXPERIMENTALMETHOD The experimental approach involved the measurement of voltammograms of molten glass samples with a range of so3 contents, which provided the basis for quantitative calibration of the differential current signal against the independently measured SO3 content. The SO3 content was gradually increased through the point at which a separate sulfate salt phase formed, which allowed comparison of the electrochemical characteristics with the data collected fiom the salt-free samples. The base glass used in the present work was a borosilicate glass containing about 49 wt% Si02, 10 wt% B2O3, 9 wt% CaO, 6.5 wt% A1203, together with lesser amounts of Na.20, Li2O,ZrOz, ZnO, MgO, TiOz, K20, Rez0-1,and Cr203. The glass was made by fusing appropriate amounts of oxides and carbonates at 1200°C for 2 hours with continuous stirring, followed by quenching of the melt in a graphite mold. The chemical composition of the base glass was confirmed by Xray fluorescence spectroscopy (XRF).
226
Environmental Issues and Waste Management Technologies VIII
The electrochemical measurements were conducted at 1130°C. The glass sample (200 grams) was placed in a 200 ml platinum crucible that was located in the center of a tube h a c e ; the sample was allowed to equilibrate for 4-6 hours before the voltammetry measurements were made. S u b r was introduced into the glass either as part of the batched chemicals (as Na~S04)before glass melting or introduced as gaseous SO3 that was bubbled through the molten glass during the equilibration period at the measurement temperature. The partial pressure of SO3 gas was controlled by metering the flow rates of SO2 and 0 2 , which react to produce the equilibrium concentration of so3 at the measurement temperature. Samples of the molten glass were removed after each equilibration period for measurement of the dissolved SO3 content by XRF. The onset point of phase separation (formation of a separate molten sulfate salt layer on the glass surface) was determined by both visual inspection of melt surface during the gas bubbling process and the presence of solidified alkali sulfate salts on the surface of the glass after the test sample was cooled to room temperature. Table 1 summarizes the SO3 partial pressures, the corresponding dissolved SO3 contents in the glass samples, and the presence or absence of a separate salt phase for each test condition. A three-electrode cell q[mV/s] system was used for voltammetric measurements, 1 consisting of three platinum electrodes (2.5 rnm diameter, 99.9% pure) and a platinum crucible filled to two-thirds of its height with molten glass. All potentials reported in this work are referenced to Timeone of the platinum electrodes. A computercontrolled potentiostat Figure 1. Square wave potential waveform (Radiometer Analytical, PGZ applied in voltammetric experiment. AE is square 402) was used for the wave amplitude, R is ramp rate. Solid dots (a) electrochemical measure- and (b) indicate time when the electric current is ments. The parameters of the measured.
Ji
Environmental Issues and Waste Management Technologies VIII
227
square-wave voltarnmetry (SWV) Potential [mVl measurements were selected to gjve 6oo 4,, -200 0 a staircase ramp superimposed on a rectangular wave of controllable 0-5amplitude and frequency, as -15. illustrated in Figure 1. The three principal control parameters and their selected values were the 9-5. dmoltensatt fiequencyf(20 Hz),the square wave 4.5 amplitude AE (25 mV), and the linear ramp rate R for the staircase Figure 2. Square wave voltammograms of (*' mV'sec for the "fast" ramp and homogeneous glass melts and molten mV/sec for the these LiNaS04. The test numbers refer to are for Table 1. Test parameters are p 2 0 H z , measurements'*. For each cycle, the h ~ = 2 5 ~=80 ~ ,m~/sec. forward and reverse current was measured at the end of the corresponding half wave (points a and b in Figure 1); the difference of the two measurements was then normalized by the surface area of the working electrode and reported as the differential current density, d i (mA/cm2). Because the current difference is measured before and after a rather small potential step, the signal is differential in nature, which allows a much higher sensitivity for redox reactions of species at much lower concentrations.
s-2.5.
mica'
RESULTS AND DISCUSSION
Voftammogramsof Homogeneous Sulfir-Containing Glass Melts Representative voltammograms of homogeneous melts with different s u l k contents are shown in Figure 2, where the background differential current density at -120 mV has been subtracted from all of the curves; the results showed good reproducibility. The differential current density for the S--ee melt shows a trend of increasing reduction current from about -200 mV to -400 mV, which flattens out thereafter producing a monotonic curve from -120mV to -800 mV. The voltammograms of the glass melts are similar between about -100 mV and -500mV. However, it is important to note that the minima of the reduction envelope featured in the S-containing melts at around -530 mV show an increasing trend with increasing sulfate content in the molten glasses. Based on this observation, the difference of the differential current at that point fiom the corresponding value in the sulfir free glass (the "peak current") was investigated as a measure of the sulhr content in the glass. Figure 3 shows that the peak current correlates linearly with the SO3 content measured by XRF. The correlation suggests good sensitivity despite the very low s u l k levels in these glass melts. The results indicate that addition of SO3 to the test melt changed either the electrical characteristics of the cell or the bulk electrochemical properties of the glass melt. However, increases in the SO3 concentration in the glass melts produced decreases in the reduction current difference in the test system.
228
Environmental Issues and Waste Management Technologies VIII
Conversely, if indeed sulfur 1.2 was being reduced in the potential scan, the opposite trend would be expected. A reduction peak at around -300 mV (using a ZrO2/air reference electrode) in a different glass system was previousIy attributed to the reduction of SO3 to S0217918. 0 0.2 0.4 0.6 0.8 1 1.2 However, as shown in SOS concentration in glass (XRF) (wW.) Figure 2, the square-wave voltamnogram of molten Figure 3. Linear proportionality between peak LiNaS04 at the same current at -530 mV from SWV and concentrations temperature does not show an of SO3 in the corresponding glass samples evident reduction peak within measured by XRF. Sulfur was added as SO, gas that potential range. (closed circles) or as Na2S04(open circle). Moreover, the curve for the molten LiNaS04 (Figure 2) shows the Potential [my 800 400 400 -200 0 smallest reduction current difference, 0 which is consistent with the trend defined by the S-containing glass 1 results but contrary to the hypothesis of sulfate reduction. Further studies are 2 E in progress to elucidate the origin of 3-3 the observed effects and to identifjl the d mechanism responsible for the 4 observed linear correlation in Figure 3. One avenue of investigation considers b: Test 3 (nosalt); c: Test 6 (onset); the possible adsorption of the gas I V_..- - -- - d g w a L w L____ phase onto the electrode surface during the voltamnogram measurements. The Figure 4. Square wave voltammograms gas reaction S03=S02+ 1 / 0 2 is of phase-separated melts (glass melt and known to be catalyzed by platinum. If sulfate layer). Fluctuation of the current at potentials less than -6OOmV signals this reaction indeed occurs on the the formation of the molten sulfate layer. electrode surface, part of the working SWV parameters arep2OH2, AE=25mV, electrode surface could be covered R = 5 mV/sec. with a thin film of the adsorbed gases. It is then conceivable that the changes in the measured differential current densities actually reflect changes in the extent of occlusion of the electrode surface as a result of this reaction. However, this still leaves open the question of the nature of the reduction process itself Despite these open questions, from a practical perspective, the present results clearly demonstrate the potential for measurement of the sulfbr content in molten glasses by square wave voltammetry
Environmental Issues and Waste Management Technologies VIII
229
over the range of about 0.2 to over 1 wt% and it is likely that further improvements sensitivity and precision can be made.
Voltamrnograms of Glass Melt with a Anode Cathode Separate Sulfate Layer The voltarnmograms of the test glass melts become distinctively different once a separate sulfate layer develops on the surface of the molten glass. Figure 4 compares the ----voltammograms (collected using the slower staircase ramp) of a salt-free molten glass, molten salt, and two representative tests in which a molten salt phase was confirmed to be present on the surface of the molten glass. The differential current density curves show a large step (greater than 100%) in the reduction current below about -600 mV when the molten 5. diagram salt layer is present on the glass surface. More of gas bubble importantly, the signals become very noisy in generation and intemption of that region, and display large fluctuations. current flow in fie thin molten Repeated measurements have shown clearly salt layer on the surface of fie that both the step in the reduction current and glass melt. It is proposed that the appearance of noise in the signal are SO2 and O2 form as a result of associated with the onset of the sulfate layer the reduction of so3 in the molten sulfate layer. Thickness on the surface of the molten glass. A reduction envelope at around -600 mV of salt layer is exaggeratedis expected for the reduction of S6+to S4+ or more reduced species. However, this alone would not explain the onset of large fluctuations in the square wave voltammograms. A likely cause of these fluctuations is the generation of gas bubbles fiom the reduction of SO3 to SO2 and 0 2 . In contrast to the hypothesized gas film adsorbed on the electrode surface proposed above, the generation of macroscopic gas bubbles in the thin sulfate layer is suggested. The non-conductive and transient gas bubbles would dramatically alter the current path through the thin layer. Because the electrical conductivity of molten alkaline sulfate is at least ten times higher than that of the glass melt, the intemption of the current through the sulfate layer due to the gas bubbles would be strongly reflected in the voltamograms. Figure 5 shows a schematic illustration of this process. It is not surprising that similar signal fluctuations are absent in measurements made on molten LiNaS04 in the absence of the underlying glass melt because the volume occupied by the gas bubbles would amount to only a small fraction of the total volume of molten salt. Consequently, the perturbations in the measured signal would be correspondingly smaller, as observed. This mechanism would indicate that a highly conductive
230
Environmental Issues and Waste Management Technologies VIII
thin layer in parallel (electrically) to a less conductive volume is necessary to cause the observed signal fluctuations.
CONCLUSIONS Square wave voltammetry was shown to be very sensitive to the sulfate concentration in the glass studied in the present work. The signal correlated linearly with the independently measured SO3 content of the glass melt over the range from about 0.2 to over 1 wt% and it is likely that -er improvements sensitivity and precision can be made. In addition, the onset of a separate molten sulfate phase was clearly evident in the measured voltammograms as both a characteristic step and the appearance of fluctuations at potentials more reducing than about -600 mV. These results suggest the possibility of relatively simple in situ real-time measurements that could determine both the s u l k content of the glass melt and the onset of an undesirable molten sulfate phase. Such a capability would provide a valuable process control element for waste glass melters that are processing feeds with high sulfbr contents. ACKNOWLEDG.MENTS The authors would like to thank M.C. Paul for help with manuscript preparation. REFERENCES W.K. Kot, H. Gan, and I.L. Pegg, "Sulfiu incorporation in waste glass 1. melts of various compositions," Ceramic Transactions, vol. 107, pp. 4 4 1 , Eds. G.T. Chandler and X. Feng, American Ceramic Society, 2000, 2. I.L. Pegg, H. Gan, I.S. Muller, D.A. McKeown and K.S. Matlack, "Summary of Preliminary Results on Enhanced Sulfate Incorporation during Vitrification of LAW Feeds," Interim Report, VSL-OOR3630-1, Rev. 1, April, 2000. 3. K.S.Matlack, S.P. Morgan, and I.L. Pegg, "Melter Tests with LAW Envelope B Simulants to Support Enhanced Sulfate Incorporation," Final Report, VSL-00R3501- 1, Rev. 0, November, 2000. 4. K. Takahashi, Y. Miura, "Electrochemical studies on ionic behavior in molten glasses," J. Non-Crystalline Solids, 80, 11-19, (1 986). M. Mark, M.P. Bruigs and M. Skylls-Kazacos, "Anodic voltammetric 5. behavior of a platinum electrode in molten sodium disilicate glass containing Fe2O3," J. Non-Crystalline Solids, 105, 7-16, (1988). C. Russel, E. Freude, "Voltammetric studies of the redox behavior of 6. various multivalent ions in soda-lime-silica glass melts," Physics and Chemistry of Glasses, 30,2,62-68, (1989). 7. J.A. Duffy, F.G. Bauke, "Redox equilibria and corrosion in molten silicates: Relationship with electrode potentials in aqueous solution," J. Phys. Chem., 99,9189-9193, (1995).
Environmental Issues and Waste Management Technologies VIII
231
8.
9. 10.
11. 12. 13. 14. 15. 16. 17. 18.
19. 20.
232
J.-Y. Tilquin, P. Duveiller, J. Glibert, P. Claes, "Effect of basicity on redox equilibria in sodium silicate melts: An in situ electrochemical investigation," J. Non-Crystalline Solids, 211,95-104, (1997). D. Lizarazu, P. Steinmetz, J.L. Bernard, "Corrosion of nickel-chromium alloys by molten glass at 1100°C: An electrochemical study," Materials Science Forum, 215-254,709-720, (1997). P. Steinmetz, C. Rapin, "Corrosion of high temperature alloys by molten glass," Materials Science Forum, 369-372,865-872, (2001). R. Pascova, C. Russel, "Thermodynamic and difhsion of iron in homogeneous and phase separated sodium borosilicate melts," J. NonCrystalline Solids, 208,237-246, ( 1996). J. DeStryker, S . Gerlach, G. von der Gonna, C. Russel, "Voltammetric studies of Fe2+/Fe3' -redox equlibria in some Na20 CaO A1203 liquids," J. Non-Crystalline Solids, 272, 131 - 138, (2000). "Electrochemistry of Glasses and glass melts, including glass electrodes," Eds. H. Bach, F. Baucke, D.Krause, Springer-Verlag, Berlin, Heidelberg New York, (2001). H.D. Schreiber, A.L. Hockman, "Redox chemistry in candidate glasses for nuclear waste immobilization," J. Am. Ceram. Soc., 70,591-594, (1987). M. Medlin, K. Sienerth, H. Schreiber, "Electrochemical determination of reduction potentials in glass-forming melts," J. Non-Crystalline Solids, 240, 193-201, (1998). J. DeStrycker, P. Westbroek, E. Temmerman, "Electrochemical behavior and detection of Co(I1) in molten glass by cyclic and square wave voltammetry," Electrochemistry Communications, 4,4146, (2002). J.-Y. Tilquin, P. Duveiller, J. Gilbert, P. Claes, "Electrochemical behavior of sulfate in sodium silicates at 1000°C." Electrochimica Acta, 42, 15 2339-2346, (1997). M. Kraub, R. Neuhaus, "Square-wave and alternating current voltammetry measurements in glass melts containing sulphur and partly iron as redox species," Proc. International Congress on Glass, vol. 2, Extended Abstracts, Edinburg, Scotland, July 2001, p.p.747-748. D.A. McKeown, IS. Muller, H. Gan, I.L. Pegg, and C.A. Kendziora, "Raman studies of sulfur in borosilicate waste glasses: sulfate environments," J. Non-Crystalline Solids, 288, 1 9 1- 1 99 (2001). D.A. McKeown, I.S. Muller, H. Gan, I.L. Pegg, and W.C. Stolte, "Determination of sulfur environments in borosilicate waste glasses using x-ray absorption near-edge spectroscopy," J. Non-Crystalline Solids, submitted, 2002.
Environmental Issues and Waste Management TechnologiesVIII
Durability Testing and Modeling
This page intentionally left blank
MODELING HIGH-LEVEL WASTE GLASS DEGRADATION IN PERFORMANCE ASSESSMENT CALCULATIONS William L. Ebert Argonne National Laboratory 9700 S. Cass Ave. Argonne, Illinois 60439
ABSTRACT A rate expression and model parameter values that provide an upper bound for the dissolution rates of high-level waste glasses were determined for use in performance assessment calculations to evaluate the suitability of the Yucca Mountain site for use as a high-level radioactive waste repository. The effects of temperature and solution pH on the glass dissolution rate were modeled explicitly, whereas the effects of glass composition, solute feedback, and alteration phase formation were bounded. The range and distribution of model parameter values are being redefined to provide realistic glass dissolution rates for total system performance assessment calculations for the Yucca Mountain license application. The results of MCC-1 tests, product consistency tests, vapor hydration tests, and unsaturated (drip) tests are being used to determine model parameter values. This paper describes the model and how test results are used to develop model parameter values. INTRODUCTION In total system performance assessment (TSPA) calculations for the proposed Yucca Mountain disposal system, the release rates of radionulcides fiom highlevel waste (HLW) glasses are calculated as the product of the inventory and dissolution rate of the glass. The model used to calculate the glass dissolution rate' is based on the mechanistic model that has been developed for borosilicate waste glass dissolution over the past two decades, but simplified for use as a submodel in the overall TSPA model. The mechanistic glass dissolution model includes parameters that account for the effects of glass composition, pH, temperature, and activities of solute species (primarily orthosilicic acid, H4SiO4) on the dissolutionrate. The rate expression can be written as rate = ko 10 qopH exp(-EJRT) (1-Q/K)
(1)
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
235
where ko is the intrinsic rate constant, which depends on the glass composition; q is the pH dependence factor; E, is the activation energy; R is the gas constant; T is absolute temperature; Q is the ion activity product; and K is the psuedosolubility product for the glass. The term (l-QK) is referred to as the affinity term. In the simplified model developed for TSPA calculations to support the Yucca Mountain site recommendation (TSPA-SR), a single parameter was defined to account for the effects of the glass and solution compositions on the dissolution-namely, the effective rate coefficient hff. The effective rate coefficient represents the product of the intrinsic rate constant and the a t y term, and the rate expression is written as rate = hff10
IIOPH
exp(-EJRT)
(2)
Although the value of the affinity term varies with the solution chemistry, the effective rate coefficient is assigned a constant value for TSPA calculations. This simplification was made because the solution chemistry of water contacting waste glass in the simulations is not tracked, except for pH. The amount of a given radionuclide released in TSPA calculations was calculated by multiplying the rate from Equation2 by the surface area of glass that was contacted by water, the contact time, and the mass fraction of that radionuclide in the glass. q, and E, to provide an The ranges of values were selected for parameters upper bound to glass dissolution rates for TSPA-SR calculations. Tests in which glass was immersed in water were used to determine parameter values. Values of q and E, were determined fkom the results of single-pass flow-through tests2 and the value of hffwas determined from the results of 7-day Product Consistency Tests (PCTs)? Degradation due to corrosion in humid air and dripping water was bounded by the aqueous corrosion rate. New parameter ranges are being determined for use in calculations to support the Yucca Mountain license application (LA). These are intended to provide a more realistic calculation of glass degradation under disposal conditions. The purpose of this paper is to describe the development of the TSPA-LA glass degradation model and how it will differ from the TSPA-SR glass degradation model. Values for the parameters q and E, in Equation 2 for the TSPA-SR model were determined by evaluating literature data. Because the dissolution rates of borosilicate glasses are known to have different temperature and pH dependencies in acidic and basic solutions: separate values of q and E, were determined for dissolution in acidic and basic solutions from tests in which the value of the affinity term remained near 1. The values used for TSPA-SR calculationsare: acidic solutions: alkaline solutions:
q= q=
-0.6k0.1 0.4 k 0.1
E,=80klOkJ/mol E, = 80 k 10 kTlmol
The range of the parameter values spans the range of values that have been measured for different borosilicate glasses. 236
Environmental Issues and Waste Management Technologies VIII
Both the contributions of ko and the affinity term to the glass dissolution rate were evaluated to select the range of values of bfffor TSPA-SR. Series of MCC-1 static leach tests4were conducted to determine the sensitivity of the value of k0 to glass composition. The dissolution rates measured in short-term tests in which the affmity term remained near 1 were used to calculate the value of k~ using Equation 1 and the mean values of q and E,. The value of was found to be insensitive to the glass compo~ition.~ The value of the affinity term changes over time due to dissolution of the glass, which results in an increase in the value of Q. This results in a decrease in the value of the affinity term and a decrease in the dissolution rate. The dissolution rate can decrease by several orders of magnitude as the glass dissolves. However, the dissolution rates of some reference waste glasses have been seen to increase by more than 1OX coincidentally with the formation of certain alteration phases. Although the mechanism by which alteration phase formation affects the glass dissolution rate is not l l l y understood, it is probably due to changes in the solution chemistry that occur as a result of the precipitation of silicon-bearing alteration phases. The dissolution model used for TSPA-SR was intended to bound the dissolution rates that have been measured in laboratory tests. Although it cannot be predicted when or if rate-increasing alteration phases will form in the disposal environment, the parameters used in the TSPA-LA model were selected to bound the experimentally measured rates. The small number of glasses for which long-term dissolution rates have been measured was deemed insufficient to identify a rate that was likely to bound the rates of all waste glasses. It was found that the results of 7-day Product Consistency Tests-Method A (PCT-A) with these glasses bounded the rates after alteration phases formed. The PCT-A, which is the ASTM International test method C1285: is called for in the DOE Waste Acceptance System Requirements document (WASRD)6 to monitor the consistency of the chemical durability of HLW glasses. To be accepted for disposal, the PCT-A response of all HLW glasses must be reported by direct measurement or by modeling, and must be lower than the response of the benchmark SRL EA glass. Because PCT-A results are available for a wide range of waste glass compositions, they provide an excellent database for determining values of parameters that are sensitive to glass composition. In addition, the chemical durabilities of waste glasses that are developed in the future (e.g., to immobilize wastes at the Hadord and Idaho sites) can be related to the results of TSPA calculations conducted for the Yucca Mountain repository license application. This comparison will provide insight into whether the impact of disposing those glasses on the system performance is bounded by the TSPA-LA calculations. The Yucca Mountain site is hydrologically unsaturated and waste glass will likely be contacted by water vapor or dripping water prior to being submerged in water. The model developed for aqueous corrosion is intended to also provide an upper bound to corrosion under unsaturated conditions. Other tests have been
Environmental Issues and Waste Management Technologies VIII
237
conducted to evaluate corrosion of glasses exposed to humid air and dripping water. Those test results will be used to determine the lower limits for the range of parameter values used for TSPA-LA calculations. The distribution of parameter values between the upper bound defined by the response in PCT-A tests and the lower bound defined by the response in tests in humid air and dripping water will be selected to reflect the anticipated environmental conditions. Because it is expected that waste glass will be exposed primarily to humid air over the 10,000-year regulated service life of the disposal system and contact by significant volumes of groundwater will be rare, the distribution of parameter values will likely be skewed toward values giving lower dissolution rates. EXPERIMENTAL Four sets of experimental data are being used to determine model parameter values for TSPA-LA calculations. In the first series, the sensitivity of k~ to glass composition over the likely range of borosilicate waste glasses to be disposed was evduated by conducting short-term MCC-1 tests following the ASTM International test method C1220.4 These tests provided a measure of the dissolution rates of nine relevant glass compositions under conditions in which the value of the affinity term remained nearly 1. Tests were conducted with monolithic glass samples cut as right circular cylinders approximately 1 cm in diameter and 2 mm thick. Sample faces were polished to 600-grit finishes. Tests were conducted in demineralized water in Type 304L stainless steel vessels at 90°C for durations of 2 to 32 days. Six of the glasses selected for testing are representative of likely high-level waste glasses. Three other glasses that had higher or lower alumina contents than the reference waste glasses were also tested to expand the composition range that was evaluated. Additional MCC-1 tests are being conducted to measure the effect of dissolved iron and iron corrosion products on the glass dissolution rate. These results will be used to determine if an additional term is needed in the rate equation to account for the effects of iron. The second series of tests were 7-day PCTs conducted with the same suite of glasses used in the MCC-1 tests. The PCTs were conducted with crushed glass in the -100 +200 mesh size fraction at a mass ratio of 1 g of glass per 10 g of demineralized water. Tests were conducted at 90°C for 7 days. The concentrations of B, Na, and Si in the test solutions were used to determine the average glass dissolution rate over the 7-day test duration. These data were used to determine parameter values for the TSPA-SR model. Additional tests are being conducted to supplement this database for the TSPA-LA model. The third series of tests were vapor hydration tests (VHTs) that were conducted with three glasses representative of DWPF glasses. Tests were conducted by suspending two monolithic samples in a test vessel with a small amount of water and reacting for times between a few hours and several years. Reaction occurs as the water vaporizes then condenses on the glass samples. A variety of VHTs were conducted to evaluate the effect of glass composition,
238
Environmental Issues and Waste Management Technologies VIII
relative humidity, and temperature on glass corrosion. The thickness of a surface alteration layer was used as a measure of the extent of reaction. Some VHTs were also conducted with the nine glasses used in the fust two series of experiments. These tests results will be used to determine parameter values representing the lower limit of the glass dissolution rate in the TSPA-LA model. The fourth set of experiments were unsaturated (drip) tests conducted with glasses representative of DWPF and WVDP waste glasses. These tests were conducted by periodically dripping tuff groundwater on a monolithic glass sample, then collecting and analyzing the solutions that dripped off the samples. Tests were conducted at 90°C and solutions have been sampled approximately semi-annually for about 13 years. The solution concentrations of glass components and radionuclides were used to calculate the glass dissolution rates and radionuclide release rates. Solution pH values were measured at room temperature with a combination electrode, and the solution compositions were determined with inductively coupled plasma-atomic emission spectroscopy or inductively coupled plasma-mass spectrometry. Some radionuclides (Np, Pu,and Am) were analyzed with alpha spectroscopy. These tests results will also be used to determine parameter values representing the lower limit of the glass dissolution rate in the TSPA-LA model and to verify that the glass degradation rate limits radionuclide release. 111. RESULTS AND DISCUSSION The results of the MCC-1 tests with the nine glasses are summarized in Table I as the normalized dissolution rates based on boron. Equation 1 was used to extract values of the intrinsic rate constants (16) using the dissolution rates measured in the tests and the values q = 0.4, E, = 80 kJ/mol, and (1-QK)= 1. Note that the dissolution rates calculated from the solution results give the rates at 90°C, whereas the solution pH values were measured at room temperature. The effect of temperature on the pH was assumed to be dominated by the temperature sensitivity of the ion product of water, which has a value about 1.6 units lower at 9OoC than at room temperature (23'C). Therefore, the pH at 9OoC was estimated for use in Equation 1 by subtracting 0.8 fiom the pH measured at room temperature for all analyses. Analysis of the MCC-1 test results indicates that the glass dissolution rates are not sensitive to the amount of aluminum in the glass under these conditions. Therefore, the values of were be assumed to be similar for all waste glasses, the product l c ~ (1-Q/K) was expressed as a single term, hs. Equation 2 can be written as or
rate = hff10 wPHexp(-EJRT)
(3a)
log rate = log hff+ q.pH - EJ2.303RT
Environmental Issues and Waste Management Technologies VIII
239
PNL 7668 I 0.0 9.2 mean k s a Average pH measured at room temperature. Mass % Al in glass. Reference high-level waste (HLW) glass.
1.1 0.95 k 0.40
8.19 7.91 k 0.16
Values of bffwere determined from the results of PCT-A conducted with the same nine glasses used in the MCC-1 tests discussed above. The PCT-A rates and values of bfffor other glasses were calculated fiom data in the literature. Values of bffwere determined at 90°C by using Equation 3 with q = 0.4 and E, = 80 kJ/mol. The measured rates and pH values and the calculated values of log bffare summarized in Table 11. (The measured pH values were adjusted to 90°C by subtracting 0.8.) Values of determined from PCT-A results were compared with values of bffdetermined for glasses for which long-term data were available. It was found that the value of hffdetermined fiom the PCT-A rate of a glass did not bound the value of bffdetermined fiom the long-term rate of the same glass in every case. However, the value of bfffiom the PCT-A rate of the Environmental Assessment (SRL EA) glass does bound the values of hff calculated fiom the long-term rates of the glasses known to be affected by alteration phase formation. The values in Table 11 were used to determine the range of values of bffto bound all waste glasses in TSPA-SR calculations. The range was selected to be log10 bff= 6.9 k 0.5, where bffhas units g/(m2.d). The low end of this range, which is 6.4 g/(m2.d), is the average of the values in Table 11 for all glasses except SFU EA glass. The high end of the range, which is 7.4 g/(m2.d), is the mean plus one standard deviation and bounds the values for all glasses that were evaluated (including SRL EA glass). The rate expression used in the TSPA-SR glass degradation model for dissolution in alkaline solutions is log10 (bounding dissolution rate in alkaline solutions) = 6.9 k 0.5 g/(m2.d) + (0.4 k O.l).pH + log10 {exp(-80 k 10 kJ/mol/RT)}
240
(4)
Environmental Issues and Waste Management Technologies VIII
Table II. Rates" and pHbfrom PCT -Aand Calculated Values of lon,.Lc Glass PCT-A ratea log10kff pH 1.2 SRL E A ~ 7.15 11.85 7.02 11.63 SRL131U I 0.69 6.29 10.42 I SRL202U I 0.043 I 6.57 I 10.65 6.65 9.8 5.42 11.20 6.15 10.66 SRL 165U 0.044 10.31 6.35 0.039 9.98 6.42 WV ref 6 PNL 76-68 0.18 9.43 7.30 0.052 10.67 6.27 Hanford-D ,
I
I
- .-
aRate= NL(B)/7 for PCT-A at 9OoC,in g/(mz*d). bpHmeasured at room temperature. in g/(m2*d). 'For dValuescalculated from literature data; see Reference 1. A similar analysis could not be conducted to determine the parameter values for dissolution in acidic solutions because the dissolution of waste glass generates alkaline solutions. Processes that could result in a decrease in the pH of the groundwater contacting disposed waste glass include radiolysis of moist air, microbial effects, and corrosion of the steel containers. Therefore, the results of tests conducted in acidic pH buffers under conditions in which the value of the affinity term remained near 1 were used to determine parameter values that bound the dissolution rates (based on boron) in acidic solutions.' The rate expression for dissolution in acidic solutions is: log10 (bounding dissolution rate in acidic solutions) = 14 f 1 g/(m2*d)+ (-0.6 f O.l)*pH + log10 (exp(-80 f 10 kJ/moI/RT))
(5)
Because the value of kffin acidic solutions, which is 14 f 1 g/(m2md), does not take into account the possible slowing effects of dissolved components, the rates calculated by using Equation 5 provide a highly conservative upper bound to waste glass dissolution in acidic solutions. The glass dissolution rate used in TSPA-SR was the higher of the rates calculated using Equations 4 and 5 at a particular pH and temperature. The rates calculated at four temperatures using the mean parameter values are plotted in Figure 1. A minimum occurs at pH 7.1 at all temperatures since the value of E, is the same for the acid and base legs. The rates calculated using Equations 3 and 4 will likely be retained as upper bounds for glass dissolution in the TSPA-LA glass degradation model. The results of VHTs and unsaturated (drip) tests are being evaluated to provide
Environmental Issues and Waste Management Technologies VIII
241
realistic dissolution rates for glass exposed to humid air and dripping water. It is expected that the rates extracted fiom those test results will be used directly, because neither test method provides quantitative information regarding the solution pH, and the unsaturated (drip) tests are being conducted only at 90°C. The approach that is currently being evaluated is to account for uncertainty through the bffterm. Uncertainty, in this context, includes variance due to the range of glass compositions and exposure conditions. In this approach, the value of the kffterm would be extracted from the rates measured in VHT and unsaturated (drip) test using the value of E, from immersion tests. This is done to determine the range of values of bffto be used in TSPA-LA calculations. Deconvolution of the pH effects on the dissolution rate is artificial in the sense that it is the calculated dissolution rate that is compared with the experimental rate. The uncertainty in approximating the solution pH in these tests is “undone” when the rate is calculated. In other words, the distinction between the and lOqoPH terms in Equations 1, 2, and 3 is convenient for relating test results to the model, although the values of &E and q could be combined into a single term. This will be done to incorporated the results of VHT and unsaturated (drip) tests.
-5
L *
0
..
I
2
’
‘
I
4
’ ’
’
’
6
’
.
*
I
’ ’ ’
I
’ ’ ’
’ 4
’ ’
*
8 1 0 1 2 1 4
PH
Figure 2. Plot of the Rates fiom Equations 3 and 4 vs. pH at 25, 50, 75, and 1OOOC. IV. SUMMARY The mechanistic rate expression for borosilicate waste glass dissolution that was abstracted to provide bounding rates for TSPA-SR calculations is being modified to provide rates for TSPA-LA calculations. The model includes and parameters for the product of the intrinsic rate constant and affinity term (bff) for pH and temperature dependencies. Model parameter values were determined by evaluation of literature data (for the pH and temperature dependencies) and by specific laboratory tests using relevant glass compositions. The value of the 242
Environmental Issues and Waste Management Technologies VIII
affinity term relevant to long-term disposal under alkaline conditions was estimated from PCT-A results for use in TSPA-SR calculations. Because of the absence of data to evaluate the chemical affinity in acid solutions, the value of the affinity term was conservatively modeled to be one under acidic conditions. The keff term is being reevaluated for TSPA-LA calculations. It is likely that the range of values for that pararneter will be used to account for glass degradation rates in humid air and in dripping water. It is expected that the bounding values determined for the TSPA-SR model will be retained as upper limits in the TSPALA model, and that values that provide for lower dissolution rates will be determined from on-going experiments. ACKNOWLEDGMENTS This work was supported by the U.S. Department of Energy, Ofice of Civilian Radioactive Waste Management. Work at Argonne National Laboratory for the U.S. Department of Energy is conducted under contract W-31-109-Eng-38. REFERENCES ‘U.S. Department of Energy, Office of Civilian Radioactive Waste Management: “Defense HLW glass degradation,” OCRWM Report ANL-EBSMD-000016 Rev. 00 ICNO1, Las Vegas, NV (2001).
2K.G. Knauss, W. L. Bourcier, K. D. McKeegan, C. I. Merzbacher, S. N. Nguyen, F. J. Ryerson, D. K. Smith, and H. C. Weed: “Dissolution kinetics of a simple analogue nuclear waste glass as a function of pH, time, and temperature,” Scientific Basis for Nuclear Waste Management XIII, Materials Research Society, Pittsburgh, Pennsylvania, 371-381 (1990). 3ASTM International, ASTM C1285-02, “Standard test methods for determining chemical durability of nuclear, hazardous, and mixed waste glasses, and Multiphase Glass Ceramics: The Product Consistency Test (PCT),” West Conshohocken, Pennsylvania (2002). 4ASTM International, ASTM C1220-98, “Standard test method for static leaching of monolithic waste forms for disposal of radioactive waste,” West Conshohocken, Pennsylvania (2002). ’W. L. Ebert, V. N. Zyryanov, and J. C. Cunnane, “Estimating Model Parameter Values for Total System Performance Assessment,” Mat. Res. Soc. Symp. Proc. 608,751-758 (2000). ‘U.S. Department of Energy, Office of Civilian Radioactive Waste Management: “Waste Acceptance System Requirements Document,” DOERW0351 Rev. 04, Washington, D.C. (2001).
Environmental Issues and Waste Management Technologies VIII
243
This page intentionally left blank
WASTE GLASS CORROSION: SOME OPEN QUESTIONS Pave1 Hrma, John D. Vienna, and John D. Yeager Pacific Northwest National Laboratory P.O.BOX999, MS:K6-24 Richland, WA 99352 ABSTRACT An equation for time evolution of glass corrosion in a closed system is proposed. Examples of fitting this equation to vapor-hydration test (VHT) and product consistency test data are shown. It is argued that the stage of accelerated corrosion of waste glass is a temporary spike caused by a transition to a different mechanism (not associated solely with high-alumina content in glass) and followed by slower steady corrosion. The effect of temperature and glass composition on the VHT rate of corrosion is evaluated. Results of different corrosion tests are compared. Progress towards a frame-indifferent rate equation is outlined. INTRODUCTION For glasses reacting with water in a closed system, the alteration process may be divided into four stages. In Stage I, glass reacts with the dilute solution. In Stage 11, the corrosion rate decreases with an increasing concentration of orthosilicic acid”* and other species3y4until the solution becomes supersaturated with respect to a number of solid phases (typically zeo1ites5’6) that nucleate and precipitate in Stage 111. In Stage IVYa balance between dissolution and precipitation is established, and corrosion proceeds with a roughly constant rate. Stages I to I11 were observed in experiments conducted under static conditions with water in contact with the glass.59798Jiricka et al? determined the existence of Stage IV in a vapor hydration test (VHT). The change of the corrosion mechanism in Stage I11 proceeds abruptly, like a sudden precipitation of ice in super-cookd water. The sudden increase in the glass alteration rate exceeds the initial dissolution rate of the glass in a solution of identical pH. This indicates that a portion of glass that has been penetrated by water during Stage II is rapidly converted to alteration products. Two major goals of glass corrosion studies are to compute the long-term corrosion behavior of glass in the repository and to develop a waste-glass formulation optimized for the best repository performance. The first step toward this goal is describing the effects of temperature and glass composition on glass corrosion with a small number of physically meaningful parameters. Geochemical codes have been applied to glass corrosion analysis by Grambod and others by computing phase equilibria between minerals and water. These codes, combined with a
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
245
transition-statetheory based rate law, are rather complex and unsuitable for describing the glasscorrosion kinetics in which nucleation and growth of secondary phases proceeds far from equilibrium, especially in Stage 111. Also, up to Stage 111, the corrosion products are mostly amorphous, and the mineraIs precipitate in Stages I11 and IV from the solution on the outer surface of the gel layer">" while the gel layer aging is a slower process. At a first approximation, the product consistency test (PCT) release data (Stages I and 11) can be fit with a power-law function of t h ~ e . ' ~ 'A ' ~linear model was used to fit the Stage IV results from the VHT;14 Stage I11 was represented by a power-law function'5 and a modified Avrami model.16 Ultimately, rate equations are needed that can be applied to more general conditions. In this paper, we attempt to describe the time evolution of glass-corrosion in experiments, in Stages I to IV, conducted in closed vessels under isothermal conditions. The effects of temperature, glass surface-to-solution volume ratio, and glass compositionare modeled for selected corrosion parameters. MATHEMATICAL REPRESENTATION The following 7-parameter function can reasonably describe glass-corrosion as a fbnction of time under isothermal conditions in a closed vessel:
;J 1+[; 1
m = r o - 1+tl" n
-1
-+-arctan2 t -t 3t
1
(r,f+m,)-m,
Here m is the altered glass mass per unit surface area; t is the corrosion time; n, ro, rm,tl, t2, t3, and n~ are temperature and composition-dependent (time-independent) coefficients; and mo = [1/2 - (l/n)ar~tan(r~/r~)]m~. The first term represents Stages I and 11; here tl is the Stage I to I1 transition time, n is the corrosion rate decrease exponent (0 < n < l), and ro is the initial rate of corrosion. The second term stands for Stages I11 and IV, which are related through nucleation and growth of aluminosilicate phases; t2 is the Stage I11 time, t3 is the Stage I11 duration, r, is the final rate of corrosion, and m4 represents the corrosion extent before Stage IV.Stage I11 cannot be exactly described by a deterministic formula; the arctangent h c t i o n is a suitable bridge function. Stage I11 is short (t3 << t 2 ) and is triggered by the nucleation and growth of solid alteration products from a supersaturated solution. It often involves cracking of glass that has been weakened by water penetration and a rapid conversion to gel. It may consist of several spikes as far-from-equilibrium processes cascade at different time intervals, as has been observed in the pressurized unsaturated flow-through (PUF) test." In Stage I, t << tl and Equation (1) reduces to m = rd. During most of Stage 11, tr << t << t2; thus, m = ml(t/tl)", where ml = rotl/n; accordingly, a power-law model describes Stage I1 corrosion. Stage I plus I1 equation (0 < t << t2)can be extended to include the glass surface-tosolution volume ratio (4:
246
Environmental Issues and Waste Management Technologies VIII
where 00 is a scale factor, and q is the glass-water interface exponent. At t >> t2, the second term reduces to m = rn4 + r d . The maximum rate of corrosion in Stage I11 is r,, = dmldt at t = t2, i.e., r,, = n-'(r& + rn&. Assuming that Stage I11 sets in when the solution concentration, c = mo, reaches a certain level of saturation, say c2, at which key alteration products (e.g., aluminosilicate minerals) begin to precipitate, we can estimate the unknown value of 0 for VHT using the formula
where t2.w and t 2 . p ~are ~ the transition times obtained at VHT and PCT carried at the same temperature. This expression follows from Equation (2) written for t = t2 >> tl and c 2 . m = C2,PCT. Material parameters, such as n, q, ro, and r,, are functions of temperature (T) and glass composition that can be controlled by the waste-product manufacturing and storage. For r,, we can estimate the temperature effects as: lnr, = A , + E , I R T
(4)
where A , is a constant, R is the gas constant, and E, is the apparent activation energy (the dissolution of even a simple mineral consists of several reaction steps'*). E, can be temperaturedependent. The effect of glass composition on material parameters, such as r,, is usually described as19
' i component mole fiaction, vi is the i* where N is the number of glass components, xi is the component coefficient, vg is ij-the second-order coefficient, and xi0 is the i* component reference mole fraction. RESULTS Figure 1 displays PCT 90°C normalized B release d a d 2 (Stages I111) and VHT 200°C d a d 4 (Stages IIIV) for four glasses, The resulting coefficients are in Table 1. As Figure 1 shows, t2.w z 200 days and f 2 3000 days, both at 90°C and PCT with opc~= 2x104 m-'. By Equation (3), we obtain am = 5 . 6 ~ 1 m-'. 0 ~ The dashed line in the Hanford low-activity waste
Table I. Fitted corrosion coefficients Glass
~
ro g/mzd
4 n h PCT 90°C
4@'
r, m4 E,'"' g/m2d g/mz kJ/mol VHT 200OC
~
@')=L80 to 102 kJ/mol(92.5 kl/mol when fit to all data) for 73 HLP glasses at temperature from 125 to 300OC.
Environmental Issues and Waste Management Technologies VIII
247
(LAW) product acceptance (HLP)-46 plot represents the correspondinghypothetical m vs. t line for VHT at 90°C based on Equation (1) with coefficients listed in Table 1. loo00
HLP-02
; E
loo. 10-
s
10
1.
0.1
'
-
1
0.01 0.1
1
I
1
175°C
A
200°C
10
0.1
~250% I
loo
la
I
0.01 0.01
0.1
1
10
100
loo0
IOOM
I
N 9 10 +20 m-1 9 2 0 0 In-1 E2000 m-1 x20m m-1 i l : L 10 ~
1
0.1 0.01
0.1
I 0.1
1
10
Tme (days)
100
0.01
0.1
1
0.1
0.01
, 1
I
10
Time (days)
100
100
Figure 1. VHT and PCT with fitted lines To obtain the values of the coefficients, it was assumed that the normalized B releases are equivalent to the glass mass altered per unit area. In reality, corrosion rates based on B release can be underestimated and even negative because of the absorption of B in the gel layer.20It was also assumed that the change in the VHT coupon thickness represents the glass volume loss, thus ignoring the volume expansion of glass due to water diffusion. Though B absorption and water diffusion may somewhat affect quantitative comparison between PCT and VHT SiOz -6.29 corrosion data, it is obvious that less than 100 pm of glass 10.25 A1203 dissolved in Stages I and I1 as the scale on the right side of -14.51 IB203 the HLP-52 plot (Figure 1) indicates. On the other hand, -22.10 Fe203 -5 1.39 Stage IV begins only when several hundred micrometers of ,Ti02 -123.O 1 glass have been dissolved. Zl-02 'R20 36.81 Using Equation ( 3 , r, (200°C) was fit as a function of A0 10.28 glass composition for 73 HLP glasses. Table I1 summarizes (B70-rO.08807)' 328.54 coefficient values (R = "a, K, Li] and A = [Ca, Mg, Zn], and all other components [Others]). Contrary to its effect on PCT releases, A1203 increases rmbecause of the precipitation of aluminosilicates.6Components that tie up alkali ions and
248
Environmental Issues and Waste Management Technologies VIII
cross-link glass networks decrease the r,. Contrary to Jantzen’s finding21that glass corrosion is enhanced by ferrous irons scavenging silica from the solution, VHT and PCT responses were virtually unaffected by redox. DISCUSSION Comparison with Other Data VHT data for LAW glasses were in good agreement with data reported by other^?^-^^ As a function of glass composition, 7-day PCT release was described by numerous Lu et al.16 fit Stage I11 data to time and composition using a 41-parameter model (their model yields r, = 0 and therefore cannot describe the Stage IV data). Figure 2 displays the correlation between the alteration rates from VHT at 200°C and the PUF test” that exposes glass (250 to 4 2 0 - p particles) to an unsaturated environment at 99”P
The E, values reported for glass c o r r o ~ i o n ~ ~range ’ ” ~ from 32 kJ/mol (for soda-lime and low Si02 glasses) to 100 kJ/mol (for R7T7 glass:’ E, = 60 f 5 kJ/mol within 25°C and 300°C). cautions against using T as an accelerating parameter until the ratedetermining mechanism is understood. For PNL 76 to 68, E, changed outside the interval of 50 to 150”C,29 possibly because of a change from Stage I <50°C to Stage I1 at 50 to 150°C (53 kJ/mol) to Stages I11 and N >150”C. It is uncertain if VHT-measured E, would be unchanged at T 425°C. Time to Dissolve The estimated repository temperature for Hanford LAW glass is 15°C. We can extrapolate the time to dissolve a 1-mm layer of glass under the VHT conditions to this temperature. As Figure 3 shows, the time to dissolve considerably shortens when E, decreases linearly with T from the value listed in Table I at 200°C to 30 kJ/mol at 15°C. Based on the activation energy for Figure 3. Effect of E, (constant or variable) on the t2 of mp-46, estimated as 14 k ~ / ~ ~ time to dissolve t;?=: 1 . 6 ~ 1 0years ~ at 15°C. This time shortens to 250 years if EQ decreases linearly with T from 114 kJ/mol at 200°C to 30 kJ/mol at 15°C.
Environmental Issues and Waste Management Technologies VIII
249
Towards the Rate Equation A rate equation is necessary for applications under a broad range of conditions. It must be frame-indifferent, i.e., the rate cannot be an explicate function of time. The time derivative of Equation ( 1 ) is not frame-indifferent. Assuming that r, << ro, then for Stages I and 11, when t << t2, r,, = r,(l+bm,)P , where b and p are material coefficients (p < 0). A comparison with Equation ( 1 ) shows that n = l/(l-p)and tl = n/bro. Equation (2) can be easily modified to include open systems. However, it is still unsatisfactory because it has the corrosion progress as a variable, whereas the ccrrosion rate in Stages I and I1 responds to the solution composition, which depends on the boundary conditions (closed boundaries or flow through). For Stage XII (within the interval of t2 - t3 < t < t2 + t3), the rate can be approximated as rIl1
=
rmax , where the coefficients are fitting parameters whose physical meaning is 1 + [(t- t,)/t,12
limited to a closed system. For Stage IV, when t >> t2, we have r,, = r, , resembling the final corrosion rate proposed by Grambod. CONCLUSIONS This paper shows that the glass-to-alterationproduct reaction in a closed system (Stages I and 11) can be represented by a power-law function of time and the interface area-to-solutionvolume ratio. When glass corrosion reaches the stage thought to be controlled by the precipitation of key alteration products (e.g., zeolites) (Stage IV), the rate of corrosion is constant. Several important questions need to be resolved before the laboratory test results can be used for predicting glass corrosion under repository conditions. How long will it take for the glass to reach Stage III? What will trigger Stage 111under repository conditions? Will the glass reach Stage IV at all? ACKNOWLEDGMENTS This study was funded by the U.S. Department of Energy’s Offices of Science and Technology (through the Tanks Focus Area) and River Protection. Pacific Northwest National Laboratory is operated for the U.S. Department of Energy by Battelle under Contract DE-AC0676RL01830. The authors wish to thank Dong Kim for careful review of the paper and Wayne Cosby for assistance in editing it. REFERENCES ‘T. Advocat, J.L. Crovisier, B. Fritz, and E. Vernaz, “Thermokinetic Model of Borosilicate Glass Dissolution: Contextural Affinity,” Mat. Res. Soc. S’mp. Proc. 176,241-248 (1990). 2B. Grambow, Nuclear Waste Glass Dissolution: Mechanism, Model, and Application, JSS Project Report 87-02, Swedish Nuclear Fuel and Waste Management Co., Stockholm, 1987. 3T. Advocat, J.L. Chouchan, J.L. Crovisier, C. Guy, V. Daux, C. Jegou, S . Gin, and E. Vernaz, “BorosilicateNuclear Waste Glass Alteration Kinetics: Chemical Inhibition and Affinity Control,” Mad. Res. Soc. S’p. 506, p 63-70 (1998). 4P.K. Abraitis, D. J. Vaugham, F.R. Livens, J. Monteith, D.P. Tivedi, and J.S. Small, “Dissolution of a Complex Borosilicate Glass at 60°C: The Influence of pH and Proton
250
Environmental Issues and Waste Management Technologies VIII
Adsorption on the Congruence of Short-termLeaching.” In: Material Research Society Sym osium Proceedings. 506, p 47-54 (1998). Van Iseghem and B. Grambow, “The Long-Term Corrosion and Modeling of Two Simulated Belgian Reference High-Level Waste Glasses,” Matl. Res. Soc. Symp. 112, p 63 1-639 (1988). 6D.M. Strachan and T.L. Croak, “Compositional effects on long-term dissolution of borosilicate glass,” J. Non-cryst. Sol., 272, p 22-33 (2000). ’J.C. .Cunnane (ed.), High-Level Waste Borosilicate Glass: A Compendium of Corrosion Characteristics. DOE-EM-0177, U.S. Department of Energy, Washington, D.C., 1994. 8K.B. Harvey, F.A.B. Larocque, S. Watson, and D.C. Doern, “The Dissolution of a Simple Specimen of a Glass,” Ceram. Trans. 23, p 123-133 (1991). ’A. Jiricka, J.D. Vienna, P. Hrma,and D.M. Strachan, “The Effect of Experimental Conditions and Evaluation Techniques on the Alteration of Low-Activity Waste Glasses by Vapor Hydration,” J. Non-cryst. Sol., 292, p 25-43 (2001). “T.A. Abrajano and J.K. Bates, “Analytical Electron Microscopy of Leached Nuclear Waste Glasses,” Ceram. Trans. 9, p 2 1 1-228 (1990). “S. Gin, I. Ribet, and M. Couillard, “Role and Properties of the Gel Formed during Nuclear Glass Alteration: Importance of Gel Formation Conditions,”J. Nucl. Matl. 298, p. 1-10 (2001). 12J.D. Vienna, A. Jiricka, B.P. McGrail, B.M. Jorgensen, D.E. Smith, B.R. Allen, J.C. Marra, D.K. Peeler, K.G. Brown, I.A. Reamer, and W.L. Ebert, Hanford Immobilized LA W Product Acceptance: Initial Tank Focus Area Testing Data Package, PNNL-13 101, Pacific Northwest National Laboratory, Richland, Washington, 2000. 13R.L. Schulz, T.H. Lorier, D.K. Peeler, K.G. Brown, I.A. Reamer, J.D. Vienna, A. Jiricka, B.M. Jorgenson, and D.E. Smith, Hanford Immobilized LA W Product Acceptance: Tanks Focus Area Testing Data Package II, PNNL-13344, Pacific Northwest National Laboratory, Richland, Washington, 2000. 14J.D.Vienna, P. Hnna, A. Jiricka, D.E. Smith, T.H. Loner, I.A. Reamer, and R.L. Schulz, Hanford Immobilized LA W Product Acceptance Testing: Tanks Focus Area Results, PNNL13744, Pacific Northwest National Laboratory, Richland, Washington, 2001. ”T.A. Abrajano, J.K. Bates, and J.J. Mazer, “Aqueous Corrosion of Natural and Nuclear I Non-Cryst. Waste Glasses: 11. Mechanism of Vapor Hydration of Nuclear Waste Glasses,” . Solids, 108, p 269-288 (1989). ’%.Lu, F. Perez-Cardenas, H. Gan, A.C. Buechele, I.L. Pegg, “Kinetics of Alteration in Vapor Phase Hydration Tests on High Sodium Waste Glasses,” in Ceramic Transactions,Vol. 132,31 1-322, American Ceramic Society, Westerville, Ohio,2002. ”B.P. McGrail, J.P. Icenhower, D.H. Bacon, K.P. Saripalli, H.T. Schaef, P.F. Martin, E.A. Rodriguez, and J.L. Steele, Low-Activity Waste Glass Studies: FY2001 Summary Report. PNNL13671, Pacific Northwest National Laboratory, Richland, Washington, 200 1. 18A.C. Lasaga, “FundamentalApproaches in Describing Mineral Dissolution and PrecipitationRates.” In: Chemical WeatheringRates of Silicate Minerals,31, eds. A. F. White and S. L. Bratley. Mineralogical Society of America, p. 22-86, 1995. 19 P. Hrma, G.F.Piepel, J.D. Vienna, S.K. Cooley, D.S. Kim, R.L. Russell, Database and Interim Glass Property Modelsfor Hanford HL W Glasses, PNNL-13573, Pacific Northwest National Laboratory, Richland, Washington, 200 1.
6.
.
Environmental Issues and Waste Management Technologies VIII
251
2*. Frugier, I. Ribet, and T. Advocat, “Effects of CompositionVariations on the Alteration Kinetics of the UOXl “Light Water,” The f h International Conference Proceedings (ICEM’OI), Bru es, Belgium, 2001. “C.M. Jantzen, “Methods of Simulating Low Redox Potential (Eh) for a Basalt Repository.” Materials Research Society Symposia Proceedings 26, ed. G.L. McVay, 613-621, Elsevier Science Publishing Co., New York, 1984. 221.S.Muller, A.C. Buechele, and I.L. Pegg, Glass Formulation and Testing with RPP-WP LA W Simulants: Final Report, VSL-OOR3560-2, Rev.0, Vitreous State Laboratory, The Catholic University of America, Washington, D.C., 2001. 23B.P.McGrail, J.P. Icenhower, D.H. Bacon, J.D. Vienna, A. Jiricka, W.L. Ebert,P.F. Martin, H.T. Schaef, M.J. O’Hara, and E.A. Rodriguez, Waste Form Release Data Packagefor the 2001 Immobilized Low-Activity Waste Performance Assessment, PNNL- 13043, Rev. 1 , Pacific Northwest National Laboratory, Richland, Washington, 1 999. 24X.Feng, P.R. Hrma, J.H. Westsik, M.J. Schweiger, H. Li, J.D. Vienna, G. Chen, G.F. Piepel, D.K. Peeler, D.E. Smith, B.P. McGrail, S.E. Palmer, D. Kim, Y. Peng, W.K. Hahn, A.J. Bakel, and W.L. Ebert,Glass Optimizationfor Vitrificationof Hanford Site Low-Level Tank Waste.PNNL-10918, Pacific Northwest National Laboratory, Richland, Washington, 1996. 25G.Leturcq, G. Berger, T. Advocat, C. Fillet, 0. Halgand, and E. Vernaz, “Chemical Durability of AluminosilicateGlasses Containing Low Solubility Chemical Elements.” Mat. Res. Soc. Symp. Proc., 506, p 199-214, Pittsburgh, Pennsylvania, 1998. 26B.P.McGrail, B.P., C.W. Lindenmeier, P.F. Martin, and G.W. Gee, “The Pressurized Unsaturated Flow (PUF) Test: A New Method for Engineered-BarrierMaterials Evaluation,” Ceram. Trans.,72, p 3 17-329 (1997). 27W.Lutze, and R.C. EWing, Radioactive Waste Forms for the Future. North Holland, Amsterdam, 1988. 28M.J.Jercinovic and R.C. Ewing, “Corrosion of Geological and Archeological Glasses.” In: Corrosion of Glass, Ceramics, and Ceramic Superconductors,eds. D. E. Clark and B. K. Zoitos. Noyes Publishing, Park Ridge, New Jersey, 1992. 29J.H.Westsik, Jr. and R.D. Peters, “Time and TemperatureDependence of the Leaching of a Simulated High-Level Waste Glass.” In: Scientific Basis for Nuclear Waste Management, 3, Plenum Press, New York, 1981. 30A.Jiricka and A. Helebrant, “Dissolution of Soda-Lime, Silica, and High-Level Waste Glass by Static and Single-Pass Flow-Through Tests,” Ceram. Trans. 107, p 309-316 (2000). 3’C.M. Jantzm, “Relationshipof Glass Composition to Glass Viscosity, Resistivity, Liquidus Temperature, and Durability: First-Principle Process Product Models for Vitrification of Nuclear Waste.” Ceram. Tram. 23,37-51,American Ceramic Society, Westerville, Ohio, 1991. 32J.J.Mazer, J.J., Temperature Efects on Waste Glass Performance, ANL-91/17. Argonne National Laboratory, Argonne, Illinois, 1991.
252
Environmental Issues and Waste Management Technologies VIII
VAPOR PHASE HYDRATION OF GLASSES IN HzO AND DzO
Timothy R. Schatz,’ Andrew C. Buechele,’ Cavin F. Mooers,’ Richard Wysoczanski,2 and Ian L. Pegg’ ‘Vitreous State Laboratory, The Catholic University of America, Washington, DC 20064 Department of Mineral Sciences, National Museum of Natural History, Smithsonian Institution, Washington, DC 20560 2
ABSTRACT Low-activity waste (LAW) glasses were subjected to the Vapor Hydration Test (VHT) in the presence of both H20 and D20 and the altered glasses were examined by scanning electron microscopy (SEM). The observed layer thicknesses, secondary phases formed, and the time dependence of the alterations, which were followed up to test durations of 206 days, were analyzed relative to the identity of the hydrating agent. Although the characteristic stages of the kinetics of the hydration process were preserved between H20 and D20 for the same glass composition, alteration rates were about 2 to 6 times faster in H20 and nearly 30 times faster during rate excursions. The results were interpreted in terms of a primary kinetic isotope effect, suggesting the implication of the breakage of a bond to hydrogen in the rate determining step. Results from IR microscopy showed no discernable isotope effect on water diffusion profiles and showed that water penetration depths were many times the VHT alteration layer thickness. Significant accelerations in reaction progress were triggered by temperature excursions introduced during the VHT, presumably due to the onset of “reflux.” INTRODUCTION Current treatment plans for tank wastes stored at the Hanford site call for separation of the material into high-level waste HL and low-activity waste (LAW) streams, which will then be vitrified separately. The glass product derived fiom vitrification of the LAW fraction must meet or exceed a variety of durability requirements, one of which is the vapor hydration test (VHT).’ In this test, glass samples are exposed to saturated water vapor at elevated temperature and pressure conditions designed to produce a nondripping film of condensed water on the glass surface. The VHT is very effective in accelerating the progress of the glasdwater reaction to its later stages within relatively
IW)
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
253
short experimental timefknes and, as such, allows for observation and analysis of the secondary hydrous mineral and amorphous phases that are believed to be influential in determining long-term glass behavior. The intent of the non-dripping layer on the coupon is to provide a local environment that is very concentrated in reaction products formed in the glass corrosion process. Such a concentration also leads to an elevation in pH, which W e r enhances the corrosion process. A wide range of glass compositions has been developed at the Vitreous State Laboratory (VSL) for Hanford LAW streams; these formulations meet all of the imposed processabiliy waste loading, and product quality requirements, including VHT performance . In the present work, several representative formulations that span a range of performance on the VHT were selected to investigate several aspects of the glass corrosion process under VHT conditions. Specifically, the present work investigated kinetic isotope effects produced by replacement of H20 by D20; diffusion of water into the glass matrix; and the effects of temperature excursions that could lead to reflux conditions in the VHT. Kinetic isotope effects in general and hydrogeddeuterium effects in particular are a standard tool in the elucidation of reaction mechanisms in organic chemistry?-7 The magnitude of these effects is typically stated in terns of the ratio of the reaction rates with the lighter and the heavier isotopes (RmdR~20in the present case). "Primary" kinetic isotope effects arise fiom quantum mechanical zero-point energy differen~es.3~~~~ The ground-state vibrational energy (the zero-point energy) of a bond decreases when the reduced mass is increased and, consequently, D-X bonds have lower energies in the ground state than the corresponding H-X bonds. Breakage of a bond to deuterium therefore requires a larger activation energy than does the corresponding bond to hydrogen in the same environment. Thus, if the bond is broken in the rate-determining step, the rate is decreased by the substitution of H for D as a result of the primary kinetic isotope effect. Since the rate depends exponentially on the activation energy, primary kinetic isotope effects can be very large and typically range fiom factors 1 to 10 for H-0 and D-0 bonds. Observation of large hydrogen kinetic isotope effects is therefore indicative of the involvement of breakage of a bond to hydrogen in the rate determining step of the reaction mechanism. It is also well known that there can be significant "secondary" solvent isotope effects when H20 is replaced by D20, which give rise to changes in both rate and equilibrium position, even when it is certain that H-X bond breakage is not involved in the rate determining step.Q6This effect arises from changes in solvent hydrogen bond strengths, which again is principally due to zero-point ener shifts. Secondary kinetic isotope % effects are generally smaller than the primary effects. Finally, particularly in the case of glass corrosion, diffision processes are also affected by isotopic substitution, although these effects are generally the smallest. A consequence of the equipartition therorem fiom statistical mechanics is that the relative rate of diffision varies as the square root of the mass ratio of the diffising species, which
254
Environmental Issues and Waste Management Technologies VIII
leads to effects of about 1.41, 1.05,and 1.08,for w‘, €320, and €&Of, respectively, as compared to their corresponding deuterated species. The effects of W D replacement on glass corrosion have been investigated in several studies.8’I6 Frischat et al.” reported no ~ s c e ~ a b effect l e while other studies have reported effects of around 1.40 89~13~14s’6and up 3.60.13 It should be noted that both Pederson, et al.13*14 and McGrail et al.“ erroneously attribute the primary kinetic isotope effect directly to changes in vibrational frequency, v. They therefore argue that since the ratio of the H-0 and D-0vibrational frequencies depend on the square root of the reciprocal of the mass ratio, the ratio of the reaction rates will afso depend 0x1the square root of the mass ratio (i.e.$ that RE&DZO vtr.dVrr0 (Ul)“ 1.41), which was consistent with their ob~enrations.’~”~”~ However, as discussed above, the more important effect is the shift in the zero-point energy (which is caused by the shift in vibrational fkquency) since that changes the activation energy on which the reaction rates are exponentialty dependent. As noted above, an isotope effect of about 1.4 is consistent with diffusional ion exchange involving H?,a process that Dran et al.*’ concluded was important. Thus, while small isotope effects are often inconclusive with respect to reaction m e c ~ (due s ~to the need to exclude diffision or secondary isotope effects as possibilities), large effects are typically indicative of breakage of a bond to hydrogen in the rate determining step. Diffision of water into glass c m be conveniently studied by infrared spectroscopy, which has been widely applied to the analysis of water in natural and synthetic glasses with m increasing emphasis being placed on FTIR microscope techniques.” Largely through such efforts, it is well known that water dissolves in silicate glasses as both hydroxyl groups and molecular water.Ig Furthermore, in conjunction with the spatial resolving power of micro-IR instruments, the well-characterized absorption band ~ s i ~ of ethe ninbred ~ spectra of hydrous silicate glasses allow for the m e ~ e m e n t of concentration (difksion) profiles of molecular €320 and s ~ c ~buund l y _OH?* Although quantitative determination of the concentration of hydrous species by IR spectroscopy requires knowledge of the precise molar absorptivities for the bands and n, concen~tion glass composition of interest, in the absence of such ~ o ~ a t i orelative (diffision) profiles can be established. In the present work, these techniques were applied to glass samples subjected to hydration under VHT conditions. Finally, we report results from experiments that have shown that reducing the VHT temperature for a comparatively short time aRe;r an initial incubation period (presumably so that condensatio~and ~ p p from ~ gthe glass coupon (“reflux”) occurs in the VHT reaction vessel) can trigger an accelerated rate of alteration that continues even after normal test conditions are resumed.
-
-
E
~
E
~
-
~
~
The details of the VI-IT method have been described previouslf’. Briefly, a glass coupon of known dimensions is suspended in a sealed, stainless steel pressure vessel
E ~ v ~ o n ~ e nIssues t a l and Waste Manage~ent~ e ~ h n o l o gVIII ~es
255
containing a suficient volume of water to saturate the atmosphere at the test temperature, 20O0C, and provide for a non-dripping sufface layer on the coupon. After a prescribed duration, the coupon is removed, dried, sectioned, and polished for microscopic characterization. Alteration rates are calculated kom the thickness of the alteration layer, as determined by SEM. The glasses used in this work were EN-5 (-24 wt% Na20), EN-4, (- 20 wt% Na2O) and EN-6 and EN-7 (-18.5 wt% Na20); the latter two glasses also contained more K20 (3.9 wt% as opposed to 0.5 wt%). Other major constituents were A1203, B203, FeaO3, and Si02 (about 6, 9, 7, and 45 wt%, respectively) with smaller amounts of CaO, Cr2O3, MgO, Ti02, ZnO, ZrO2, C1, F, CszO, P~OS, and SO3. The glasses subjected to temperature excursion treatment ("reflux") were EN-4, EN-6 and EN-7. Reflux conditions were established by dropping the temperature of the test to 185°C for a period of 24 hours. In the case of EN-4, the drop occurred 48 days into the test on samples in both H20 and D20; the tests were then continued to total durations of 55, 106, and 160 days. For the EN-6 and EN-7 glass samples, the temperature drop occurred 10 days into a test of 24 days duration and was conducted only with H20. Dissolved -OH and -0Ddiffusion profiles in EN-4 and EN-5 were determined using micro-FTIR spectroscopy. Thin sections of VHT coupons run for 24, 55, 102, and 206 days were produced by embedding the coupons in epoxy and cutting slices from the mounts. These were further thinned and polished to thicknesses of 100,250, and 500 pm, with the final stages being done on an Allied High-Tech precision polishing unit using 3-M Imperial diamond lapping film. Room-temperature infrared spectroscopic measurements were conducted in transmittance mode using a BIO-RAD Excalibur FTS 3000 spectrometerwith a microscope attachment at the Department of Mineral Sciences, National Museum of Natural History, Smithsonian Institution. A 10x100 pm aperture was translated across the sample sections to enable collection of transmittance IR spectra at measurement positions of interest. Dissolved -OH diffusion profiles were determined from the intensities of the broad absorption band at -3450 cm-', which contains the hdamental0-H stretchingvibrations of molecular H20 and -OH structural groups.
RESULTS Figure 1 shows the measured layer thickness against test duration for EN-5 tested in H20 and in D20. The large reduction in layer growth rate when D20 is substituted for H20 is evident, with isotope effects of initially about 2 to 5 rising to nearly 16 during the acceleration and falling to about 1.2 afterwards. Also evident in Figure 1, whether hydration occurs with H20 or D20, are the incubation,fast-rising, and nal stages, which are often exhibited in the VHT kinetics of high-sodium waste glasses.2 F
256
Environmental Issues and Waste Management Technologies VIII
Figure 2 shows comparative 400SEM micrographs of EN-5 specimens after VHT durations of 8.5, 9.0, 9.5, and 10.0 days tested in either H20 or D20. Clearly, F much larger layer thicknesses were 2oo produced with H20 employed as E the hydrating agent than with D20 over the same test intervals for this glass composition. The layers show considerable substructure and 0 variation. In general, there are 0 5 20 25 large crystals of analcime or a l0 DAY3 l5 similar sodhl'l ~lU.mho-silicate Figure 1. EN-5 VHT layer development in growing out of the coupon surface. H 2 0 (top) and D20 (bottom). Some of these crystals exhibit an internal substructure with an enrichment in zirconium near the crystal surface due to inclusions of zirconolite. The modified layer on the glass itself is a gel layer depleted of some glass components and containing zones distinguished by differing levels of porosity and secondary phase crystallization. At the outer extreme of the modified layer is a concentration of a hydrated calcium silicate phase, possibly pectolite or wairakite on the basis of X R D data. All but one of the layers shown in Figure 2 contain obvious inclusions of large crystals of analcime. The one exception, the 9.5-day coupon in D20, has a much thinner layer than the specimens at earlier or later times. This suggests that the nucleation
1 c
'
Figure 2. VHT Layer development on EN-5 in H20 and D20,8.5-10 days.
Environmental Issues and Waste Management Technologies VIII
257
and growth of such large crystals within the modified layer is associated with an acceleration in the rate of alteration of the underlying glass.
I/
I
01;
0
50
loo
150
4OOO 3800 3600 3400 3200 3000 2800 2600 2400
200
mn
Firmre 4- IR absorbance in EN-4
Figure 3 shows the time dependence of the VHT altered layer thickness for EN-4 when tested in H 2 0 ; the lower sodium content of this glass results in thinner layers than for EN-5. Figure 4 shows a representative series of micro-IR spectra from edge to center for a 500 p thick sample of EN-4 after a 55-day VHT. The IR absorption bands of dissolved water species have been assigned for a variety of hydrous silicate glasses, and the band observed at -3450 cm-'is attributed to the fbndamental o-H stretching mode of molecular H 2 0 and structurally bound -OH.As would be expected, the molar absorptivity of this band has been shown to vary significantly with composition.'' Although this property has not been analyzed in detail with regard to the multi-component waste glass systems of interest here, the normalized absorbance maxima of the -3450 cm-'band can nonetheless be used to trace relative diffusion profiles of dissolved -OH, whether it be as H20, structurally bound -OH, or both. Such a treatment is presented in Figure 5 for the 500 p thick sample of EN-4 after its 55-day VHT run. Based on these data and analogous data for EN-4 samples after 24, 102, and 206 day VHT runs, an apparent diffusion coefficient for dissolved -OH, on the order of 10"' cm2s-', can be determined. This value is in reasonable agreement with other data in the literature for the diffusion of water into silica gla~s.2~ It is also worth noting that elevated levels of -OH were detected some 60 pm into the EN-4 VHT coupons, far beyond the measured altered later thickness of about 8 p. The mid-IR spectra of the EN-4 and EN-5 samples subjected to VHT in the presence of D20 display the same absorption peaks as those shown in Figure 4 with the broad band assigned to the fundamental 0-H stretching mode shifted to lower fiequency by -100 cm-'. This absorption band is assumed to correspond to the -0D stretching mode although the shift differs from that expected for a simple harmonic oscillator.'0J4
258
Environmental Issues and Waste Management Technologies VIII
0
2S
50
76
100
1!25
175
1-
DAYS
2o
40
6o
Dhtance from Edge [p]
loo
120
Figure 6. EN4VHTwithreflux at48 days in H2O and D20.
Figure 5. Normalized absorbance in EN-4 at 3450 cm-' (-OHband) vs. depth. Comparison of EN-4 and EN-5 VHT samples in H20 and D20 indicated that the relative diffision profiles of either dissolved -OH or -0D were essentially the same for a given glass composition. EN-5 was observed to be rather more permeable to both dissolved -OH and -0Dthan EN-4,
Before MUX
in accord with the latter's better resistance to ~i~~ 7.L~~~~ on E N in~~~0 (top) It and D20 (bottom) before and after reflux. layer go* under should also be noted that no isotopic frequency shift was observed with regard to the more discrete band at -2650 an-',further supporting its assignment to a combination of B-0vibrati0ns.2~ Figure 6 shows the measured layer thickness against test duration for EN-4 in both H20 and D20 when subjected to a temperature excursion ("reflux") at 48 days. Comparison to Figure 3 (no reflux) shows the dramatic effect of reflux on layer growth in the presence of H20; Figure 6 also suggests a delay in the onset of acceleration after reflux in D20 and a much smaller magnitude. Figure 7 shows SEM micrographs of the VHT samples before (24 days) and after (55 days) reflux. The sample refluxed in the presence of H20 formed much thicker layers than that in D20 under otherwise identical conditions. The layer structure in H20 after reflux is also considerably more complex than that in D20. Figure 8 shows SEM micrographs of two EN-6 specimens after VHT durations of 24 days, one of which was brought to reflux conditions 10 days into the test. The refluxed sample formed a 4-fold thicker layer than the sample treated under normal VHT conditions. The structures of the layers on the EN-6 glass specimens are very similar
Environmental Issues and Waste Management Technologies VIII
259
despite the difference in thickness. Conversely, normal and reflux samples of glass EN-7run under conditions identical to those used for EN-6 developed layers of essentially the same structure and thickness.
24 Days, 2oooC
DISCUSSION AND CONCLUSIONS R d w 10 d Into test Normal Condltlons Layer growth rates were 2 to 6 Figure 8. Effect of reflux on EN-6 VHT times faster in ~~0 than in ~~0and up results. to 30 times faster during acceleration periods, which occurred either naturally or were triggered by reflux excursion. As discussed earlier, transport phenomena alone are unlikely to be responsible for effects of this magnitude. The most likely explanation is in terms of a primary kinetic isotope effect, which would implicate breakage of a bond to hydrogen in the transition state of the rate determining reaction step. Isotope effects of this magnitude have been reported previou~ly'~ for glasses tested under PCT conditions over a range of ratios of glass surface area to leachant volume ( S N ) . Since the effect was also ob~erved'~ to increase with SN, large effects under VHT conditions might be expected in view of the very large effective SN for the W T .As discussed earlier, other possible effects include secondary isotope effects as well as isotope effects on pH (pD) and pK values (e.g., for silicic acid), which can impact reaction rates through their effect on the reaction affinity.15 In addition, interpretation of the rate ratios is complicated because comparisons made at the same point in time do not relate to the same point in the overall alteration process (e.g., in the case of accelerations)as a result of the slower overall rate in D20. The small expected isotope effect on water di&sion rate is consistent with the observation of very similar diffision profiles of either dissolved -OH or -0Dfor the same glass compositions in micro-IR experiments. These measurements also showed that for EN-4, water penetrated many times deeper into the VHT coupons than the thickness of the alteration layer. The introduction of brief temperature excursions under VHT conditions, which are presumed to promote reflux, triggered significant accelerations in the rate of layer growth. This effect was also slowed considerably in the presence of D20 versus H20. The results also indicate that glass composition has a strong influence on the degree to which reflux modifies the corrosion rate. ACKNOWLEDGEMENTS The authors would like to thank M.C. Paul for help in manuscript preparation.
260
Environmental Issues and Waste Management Technologies VIII
REFERENCES Bechtel National, Inc. Design, Construction, and Commissioning of the Hanford 1. Tank Waste Treatment and Immobilization Plant, Contract Number: DE-AC27OIRV14136, U.S.Department of Energy, Office of River Protection (12/11/00). Muller, IS., Buechele, A.C., and Pegg, I.L. (2001) Glass Formulation and 2. Testing with RPP-WP LAW Simulants, Final Report, VSL-OOR3560-2, February 23,2001. Morrison, R.T. and Boyd, R.N, (1974), Organic Chemistry, 3d Ed., Allyn and 3. Bacon, Inc., Boston, MA. Collins, C.J. and Bowman, S., Eds., Isotope Eflects in Chemical Reactions, Van 4. Nostrand Reinhold Co., New York, 1970. Bunton, C.A. and Shiner, V.J. Jr., (1961), J. Am. Chem. Soc., 83,42. 5. Bunton, C.A. and Shiner, V.J. Jr., (1961), J . Am. Chem. Soc., 83,3207. 6. Bunton, C.A. and Shiner, V.J. Jr., (1961), J. Am. Chem. Soc., 83,3214. 7. Scholze, H., (1975), Glass Technol., 16,76. 8. Scholze, H., (1977), J. Am. Cerm. Soc., 60, 186. 9. Kumar, B., (1986),XIV Int. Congress on Glass, New Delhi, ~01.2~370. 10. 11. Frischat, G.H., Richter, T., and Borchardt, G., (1986), XIV Int. Congress on Glass, v01.2,370,New Delhi. 12. Pederson, L.R., Baer, D.R., McVay, G.L. & Engelhard, M.H. (1986). J. NonC y t . Solids 86,369. Pederson, L.R. (1987). Phys. Chem. Glasses 28, 17. 13. Pederson, L.R., Baer, D.R., McVay, G.L., Ferris, K.F. & Engelhard, M.H. (1990). 14. Phys. Chem. Glasses 31 (5), 177. 15. Feng, X., Fu, L., Choudhury, T.K, Pegg, I.L., and Macedo, P.B., (1991), Mat. Res. S’p. Proc., 212,49. McGrail, B.P., Icenhower, J.P., Shuh, D.K., Liu, P., Darab, J.G., Baer, D.R., 16. Thevuthasen, S., Shutthanandan, V., Engelhard, M.H., Booth, C.H. & Nachimuthu, P. (2001).J. Non-Cyst. Solids 296,lO. Dran, J.X., Della Mea, G., Paccagnella, A., Petit, J.-C., and Trotingnon, L., 17. (1988), Phys. Chem. Glasses,29,249. 18. Behrens, H., Romano, C., Nowak, M., Holtz, F. & Dhgwell, D.B. (1996). Chemical Geology 128,41. Mandeville, C.W., Webster, J.D., Rutherford, M.J., Taylor, B.E., Timbal, A. & 19. Faure, K. (2002). American Mineralogist 87,8 13 and references therein. 20. Zhang, Y., Stolper, E.M. & Wasserburg, G.J. (1991). Geochimica et CosmochimicaActa 55,44 1. 21. Buechele, A.C., Lofaj, F., Mooers, C., and Pegg, I.L. (2002). Ceramic Transactions 132,301. Lu, X., Perez-Cardenas, F., Gan, H., and Buechele, A.C. (2002). Ceramic 22. Transactions 132,3 1 1.
Environmental Issues and Waste Management Technologies VIII
261
23.
24. 25.
262
Wakabayashi, H. & Tomozawa, M. (1989).J.Am. Ceram. Soc. 72,1850. Sykes, D. & Kubicki, J.D. (1993). ~ ~ o c ~ i meti ~c oa s ~ o c h i mActa i c ~57,1039. Li, H.& Tomozawa, M., (19951, Ceramic Transactions,61,539.
Environmental Issues and Waste Management TechnologiesVIII
MODELING FLUID CHEMISTRY INSIDE A WASTE PACKAGE DUEi TO WASTE FORM AND WASTE PACKAGE CORROSION Vijay Jain and Narasi Sridhar Center for Nuclear Waste Regulatory Analyses Southwest Research Institute 6220 Culebra Road San Antonio, TX 78238
ABSTRACT A typical codisposal waste package (") designed for the proposed high-level waste (HLW) repository at Yucca Mountain (YM), Nevada, will contain five 304L stainless steel (SS) canisters with HLW glass and one US,Department of Energy (DOE)-owned spent nuclear fuel canister made of 316L SS. Radionuclide release from the WPs is a complex process that depends upon the composition and flux of g~undwatercontacting the waste forms; the dissolution rate of HLW glass and spent nuclear fuel; the corrosion rate of WP components made of Alloy 22, and 316L SS, 304L SS HLW canister, and carbon-steel used as canister guides; the solubility of radionuclides; and the retention of radionuclides in secondary phases. In this study, evolution of the chemical composition of the fluid inside a codisposal W P is simulated using the OLI Systems IeSPICSP Software, which is a t h e ~ ~ y speciation n ~ c and process s~mula~on software. Results g dis~butionof indicate that the WP corrosion products play an important rofe in d e t e ~ n i n the species between the solid and aqueous phases. INTRODUCTION The proposed YM repository will store vitrified HLW from the Hanford site, Savannah River Site, West Valley, and Idaho National Environmen~Engineering Laboratory, together with spent nuclear fuel.' The vitrified HLW will contribute4,667 metric ton of heavy metal (MTHM) equivalent out of the total of 70,000 MTHM planned for the repository. WP designs consist of two concentric cylinders in which the waste forms will be placed. The inner cylinder is made of stainless steel Type 316NG with a nominal thickness of 5 cm. The outer cylinder is made of a corrosion resistant nickel-based alloy (AIloy 22) with a nominal thickness of 2 cm. A typical codisposal WP will contain five 304L stainless steel (SS)canisters with HLW glass and one 316L SS DOE-owned spent nuclear fuel canister. The WP will be emplaced horizontally on pallet in a drift as part of an engineered barrier system. The 'cstp may be breached either by an external event or by corrosion that can provide a mechanism for ground water ingress into the Wp.The corrosion of the WP materials and internals is influenced by the chemistry of the water contacting the waste package, temperature, and radiation. Center for Nuclear Waste Regulatory Analyses (CNWRA) studies on Alloy 22 indicate that within the l0,OOO year performance period, breach of a WP by passive corrosion is highly unlikely? However, WP breach by rock falls, stress corrosion cracking, or localized corrosion are plausible scenarios. The objective of this study is to examine the evolution of groundwater To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without
the express written consent of The AmericanCeramic Society or fee paid to the Copyright ClearanceCenter, is prohibited.
Environmental Issues and Waste Management Technologies VIII
263
Table I. Corrosion products used in simulation of in-package chemistry Uniform Corrosion
Localized Corrosion
Glass Hydrolysis
chemistry inside a breached WP using OLI systems ESPKSP software.
SIMULATIONS Evolution of the chemical FeCl, H4SiOQ Fe(OH), com~sitionof a fluid inside a WP is simulated using the OLI Systems ESPKSP Software, Version 6.5, which is a thermodynamic speciation and process simulation ~oftware.~ The schematic shown in Figure 1 describes a mixing sequence used in evaluating the evolution of chemistry inside a WP. In this paper, WOW), simulations are limited to non-radioactive streams c ~ n t ~ n i n g Table If. HLW glass WF composition and major corrosion products as listed in hydrolysis products Table I. The effect of carbon-steel Hydrolysis cannister guides used inside a WP Weight % Component Mole Fkducts was also excluded. For the iron dissolution product under oxidizing SiO, 50 0.571 H4Si04 conditions, Fe(OH), was assumed as the input species rather than FeOOH 15 0.148 H3B03 B2°3 that is observed in atmospheric 7 0.047 Al(OH), corrosion of iron. However, the A403 calculated results show the presence N%O 15 0.242 NaOH of only FeOOH. S S composition is approximated as 0.7Fe. 0.2 Cr, and 0.052 Fe(OH), 12 0.1 Ni (moles) and corrosion 1 0.017 Mg(OH), products are assumed to dissolve MgO c o n ~ e n ~HLW y . glass WF composition is approximated by a simple six-component system shown in Table II. Also shown are hydrolysis products used as an input to the HLW glass WF stream. Simulations are conducted by assuming one mole of corrosion products per liter of water inside a WP. In these simulations, the contributions of corrosion products from uniform and glass corrosion are maintained constant at one molefliter while the contribution from localized corrosion is increased from 0 to 0.25 molesfliter simulating higher contributions fiom localized corrosion inside the WP. The corrosion products from localized corrosion are assumed to be metal chloride salts. This is consistent with the experimental Figure 1 Schematic of mixing sequence of various corrosion obse*ations in pits streams inside a waste package I)
264
Environmental Issues and Waste Management Technologies VIII
and crevices: Simulationsare also conducted by introducing oxygen to the uniform corrosion stream and varying the molar concentration of HLW glass WF.Simulations were conducted at 25°C and 1 atm pressure. The lower temperature is considered because breach of W P is expected after cooling down of the waste. The thermodynamic speciation information was obtained from the Public and GEOCHEM databases in the code. Selected solid phases such as oxides of Ni, Cr and Fe were removed. While oxides are the thermodynamically stable phases, hydroxides and oxyhydroxides are the only observed corrosion products of these elements in aqueous solutions.s
RESULTS Uniform Corrosion Stream The concentration of the corrosion products resulting from uniform corrosion is kept constant at 1 mole/l during simulations. Speciation calculations indicated that Fe, Cr, and Ni were present as FeOOH, Cr(OH), and Ni(OH),. The stream has a pH of 8.4 (Figure 2). The calculated redox potential for the stream was - 126 mV vs. standard hydrogen electrode (SHE) (Figure 3). No dissolved corrosion products were observed in this stream. Localized Corrosion Stream In this stream, the amount of localized corrosion was changed from no corrosion to corrosion that produced 0.25 mole/l of localized corrosion products. The pH of the localized stream at a total concentration of 0.01 moleA concentration was 3.6 and decreased to 3.3 at 0.25 mole/l concentration. Similarly the redox potential in the same range of concentrations of corrosion products from localized corrosion decreased from 72 mV to 57 mV vs. SHE (Figure 3). Simulations indicate that all corrosion products are soluble with Fe present as Fe2+,Ni as Ni2+,Cr as C?. Small amounts of NiCl+,CrC12+and CrCl; complexes were formed at 0.25 moleA or higher concentration of corrosion products. These calculations are consistent with Raman spectroscopy results on analysis of pit solutions!
12
v
X
cl
10 8
E6 4
2
0
1
0
+-
I
0
1 I
I
I
I
I
I
0.05
0.1
0.15
0.2
0.25
-+- Locdkd +unaorm -+- Combined +Glass WF -AInpackage 0 Unaolm(O2) 0 Combined(02) A Inpack(02) 0.3
L o c M n a o r m Corrosion Ratio Figure 2. Changes in the pH caused by localized corrosion
Environmental Issues and Waste Management Technologies VIII
265
Combined Stream Combined stream refers to a combination of 100 uniform corrosion and +--Localized localized corrosion product E +uniform streams. Figure 2 shows the 3 -100 changes in pH as localized +cocombied corrosion stream is mixed a -300 -X - Glass W with uniform corrosion +InPaCW stream. The pH is plotted as -500 a ratio of localized corrosion (L) . . and uniform corrosion -700 II 1 (U) concentration. In these 0 0.1 0.2 . 0.3 simulations, the pH of the uniform corrosion product Localized/Uniform Corrosion Ratio stream is maintained constant because no Figure 3. Changes in the redox potential caused by localized changes are made to this corrosion stream during simulations. The pH of the combined stream dropped from 8.5 to 6.9 when 0.01 moVl of metal chloride was mixed from the localized corrosion stream. The pH then decreased only slightly from 6.9 at 0.01 moVl to 6.4 at 0.25 moleA concentration of localized corrosion solution. Similarly, the redox potential in the same range of localized concentration for the combined stream, as shown in Figure 3, increased from -260 mV to -237 mV. The initial drop in the redox potential shown in Figure 3 represents the change in the redox potential due to the addition of 0.01 moM of localized corrosion products. Figures 4 and 5 show solid and aqueous species present in the combined uniform and localized corrosion waste streams. The percent phases are shown as a ratio of localized corrosion and uniform corrosion. Dominant solid phases are FeOOH, Cr(OH), and Ni(OH),. The concentration of solid phases decrease as the concentration of the localized corrosion stream increases, The aqueous species are mostly Fe2+and Ni2+ions. However, a small quantity of NiW, CrCl,+,and Cr(OH)'+ complexes are present at 0.25 moleA or higher localized corrosion concentration. The concentration of aqueous species increases as the localized stream concentration increases. 300
1
W
4
4
I
I
HLW Glass WF Stream The concentrations of corrosion products resulting from the HLW glass WF hydrolysis was kept constant at 1 mole/l during simulations.This stream has a pH and redox potential of 11 and 118 mV vs. SHE, respectively (Figures 2 and 3).
266
-
100%
2
80% 60%
LILI=O.OO L/lJ=O.Ol
40% 20%
0% Fe0(OH)
C r ( 0 H)3
Ni(0H)Z
Figure 4.Solid corrosion phases in the combined corrosion stream
Environmental Issues and Waste Management Technologies VIII
Fez+
NO+
Figures 6 and 7 show the solid and aqueous species resulting from glass dissolution. The solid phases formed were talc, quartz, Na4Si03,and low albite, FeOOH, etc. The dominant aqueous species were Na" and B(OH)i with minor amounts of B(OH),, NaB(OH)* NaHSi04,and H,SiO,,-.
Figure 5. Aqueous species in the combined corrosion stream
In-Package Solution Stream The in-package solution stream refers to a further combination of the combined stream from uniform corrosion and localized corrosion with the HLW glass WF streams. Figure 2 shows evolution of the pH as the cumulative stream which is the localized corrosion stream mixed with the uniform corrosion stream. The combined stream is then allowed to react with the hydrolyzed HLW glass WF product resulting in the in-package stream. The pH of the in-package solution stream drops from 11.3 with no localized corrosion to 6.8 at 0.25 mold concentration of localized corrosion solution. Similarly, the redox potential in the same range of localized concentration for the in-package solution, as shown in Figure 3, increased from -439 mV to -302 mV vs. SHE. The initial drop in the redox potential from -439 mV to -539, shown in the Figure 3, represents the change in the redox potential due to the addition of 0.01 moVl of localized corrosion products 100% to the uniform corrosion stream, as exhibited by the 80% combined stream (Figure 3). Figures 6 and 7 60% show aqueous and solid
3
species present in the inpackage solution. The percent phases is plotted as a ratio of localized corrosion and uniform corrosion. The solid phases formed are talc, quartz, low albite, FeO(OH), Cr(OH),, Ni(OH),, with small amounts of Na,Si04 and Fe2Si04.Minor quantities of species disappear at localized corrosion concentrations of
E
8
40% 20%
0%
Figure 6. Distribution of solid phases in the in-package solution
Environmental Issues and Waste Management Technologies VIII
267
10%
80%
3
60%
c a
8
40% 20%
0%
I
I
Figure 7.Distribution of aqueous species in the in-package solution 0.25 moles/l or higher. The quantities of talc, quartz, low albite, and Cr(OH), remain fairly constant while the quantities of Ni(OH), and FeO(0H) decrease as the concentration of localized corrosion stream increases. The quantities of aqueous species such as B(OH),, Na', Fe", Ni2+,MgZ+and formation of NiCl', CrCI2+,CrCI,', and Cr(OH)'+ in small concentrationsincrease while the concentrationsof B(OH)/, NaB(OH),, NaHSiO,, and H,SiO,- decrease as the concentration of the localized corrosion stream increases. In addition, simulationsare conducted by varying the amount of HLW glass WF from 0.50 moles/l to 1.5 molesn and mixed with a combined stream containing 0.1 moles/l of localized stream. The results showed that as the concentration of the HLW glass WF increases, the pH of the in-package solution increases from 7 to 9 while the distribution of solid phases and aqueous species remained the same. SIMULATION WITH OXYGEN STREAM In this simulation, 0.1 mole of oxygen was added to the uniform corrosion stream having a R/U ratio of 0.10. The pH of the uniform stream drops from 8.4 to 6.3, the pH of the combined stream drops from 6.5 to 5.2, and the pH of the in-package stream drops from 8.1 to 6.5. Similarly, the redox potential increased from - 126 mV to 568 mV vs. SHE for the uniform stream, -242 mV to 65 1 vs. SHE for the combined stream, and - 377 mV to 544 mV vs. SHE for the in-package solution. A comparison of the species present in the combined stream for R/U = 0.10 with and without oxygen is shown in Figure 8. The results indicate that in the presence of oxygen a significant drop in the concentrationof solid Cr(OH), and Ni(OH), phases is observed. The decrease is compensated by the formation of aqueous Ni2+and Cr"' species.
268
Environmental Issues and Waste Management Technologies VIII
DISCUSSION Evolution of pH The DOE model for glass dissolution used in the abstraction for the total system performance assessment has a significant influence of pH and ternperat~re.~ The model shows that the dissolution rate can be expressed as a function of pH and temperature by a V-shaped curve with minimum at pH of 7.' Figure 2 shows the evolution of pH as the Figure 8. Changes in the distribution of species in combined localized corrosion stream is mixed stream resulting from the presence of oxygen with uniform corrosion stream, then allowed to react with hydrolyzed HLW glass WF stream. The change in pH is significant and is attributed to two factors: mixing of low and high pH streams, and redistribution of species due to changes in solubility limits. The combination of uniform corrosion and HLW glass dissolution, in the absence of localized corrosion, results in an increase in the pH of in-package solution from 11 to 11.3, thus increasing the corrosion rate of glass slightly. However as the contribution of the localized corrosion process increases, the pH of the in-package solution decreases, reducing the glass dissolution rate, provided the pH does not drop below 7.This reduction in the glass dissolution rate may result in a further decrease in pH. The increase in pH after the addition of uniform corrosion stream to HLW glass WF is attributed to the reduction of Na,SiO, concentration from 0.18 moles to 0.005 moles as observed by reaction shown in Eq. (1). The increase in Na" ions is reflected in Na" data shown in Figure 7. 1
Na4Si04 t 4Ht = 4Nat t Si02(aq) t 2H20
(1)
Other dominant reactions contributing to a shift in pH due to redistribution of species are B(0H)i t Ht = B(OW3(aq) t H20
(2)
+ H20= Na+ + H 3 S i 0 i
(3)
H 3 S i 0 i t Ht = Quartz(ppt) t 2 H 2 0
(4)
NaHSi03
In the presence of oxygen, the pH of the localized stream, combined stream, and in-package stream drops significantly as shown in Figure 2. The drop in pH is associated with the formation of Crh complexes shown by Eq.(5). 2Cr(0W3
+ H 2 0 = 8H+ + Cr20;- + 6e-
Environmental Issues and Waste Management Technologies VIII
(5)
269
Even in the presence of oxygen, stainless steels are observed to dissolve essentially as C P species.' The formation of C p ions typically occurs when the stainless steel undergoes transpassive dissolution. The oxidation of Cf'+ to Crb in the aqueous phase by oxygen is kinetically slow, but may occur over long periods of time. The results show that in a closed system the pH of the in-package solution decreases as localized corrosion increases, hence reducing the HLW glass WF dissolution rate. However in a open system where oxygen is present, pH could drop below 7, resulting in higher dissolution rate for HLW glass WF.The final pH will depend upon the competing effects of localized corrosion, which will tend to drop the pH, and glass dissolution, which will shift the pH higher. Evolution of Redox Potential The corrosion behavior of WP components is determined by their corrosion potential. In the passive condition, where uniform dissolution occurs, the corrosion rate is essentially independent of the corrosion potential over a wide range of potential. However, if the corrosion potential exceeds the repassivation potential (also called protection potential), localized corrosion ensues. For a given W P material, the repassivation potential depends mainly upon temperature, and Cl-, and NO,-concentrations. CNWRA studies' at 95°C on 316L SS indicate that for C1concentrations higher than 0.01 M, repassivation potential is lower than corrosion potential. Therefore, at 95°C localized corrosion could occur in 316L SS at C1- concentration as low as 0.01 M. The corrosion potential of WP materials is partly determined by the redox potential of the environment. The corrosion potential is a mixed potential determined by the kinetics of the reduction (cathodic) and dissolution (anodic) processes (Figure 9) . While a direct correlation between the redox and corrosion potentials cannot be established without knowing the kinetics of the cathodic and anodic reactions, knowledge of redox potentials may provide some information on the trends in the corrosion potential. For a passive metal, changes in corrosion potential are directly proportional to those in redox potential. For conditions a0 where active or localized corrosion is already occurring, changes in Redox redox potential may not have a Passive Anodic Curve significant effect on the corrosion potential. potential The redox potential of the ground water entering the WP depends upon concentration of potential species such as dissolved 0, and Cathodic Curve NO,, and pH. In the presence of oxygen, the calculated redox potential of the in-package solution Current is 377 to 544 mV vs. SHE depending Figure 9. Schematic drawing showing the relationship upon the extent of localized between redox potential and corrosion potentials for corrosion assumed. Such a high passive and localized corrosion conditions redox potential is expected to increase the corrosion potential to
'T I
270
'
Environmental Issues and Waste Management Technologies VIII
values higher than the repassivation potential. Expected ground water chemistries indicate that C1concentration could range from 0.002 M to 3.6 M., Therefore, localized corrosion of 316L SS is likely. As localized corrosion progresses, formation of metal chlorides in pits and crevices increase. FeC1, was selected for the ESP simulation instead of FeCl, because studies have shown that FeCl, is a preferred corrosion product." As a result the pH is likely to be low to near-neutral as discussed before. The simulations indicate that in the absence of oxygen, the redox potentials range from -600 mV to 150 mV, depending on the extent of localized corrosion assumed. Note that the redox potentials shown in Figure 3 are the redox potentials observed after corrosion. In Figure 3, the redox potential for the in-package stream dropped from -439mV to -565 mV after introduction of localized corrosion products in the simulation. The redox potential of the in-package stream continued to increase with an increase in localized corrosion. This at first appears to be counter intuitive because the corrosion potential at the bottom of a corroding pit are significantly more anodic than the external corrosion potential. However, the anodic corrosion potential at the bottom of a growing pit occurs because the anodic kinetics occurs in a much more active manner. The corrosion potential under this condition is the point of intersection of the cathodic curve with the active anodic polarization curve (Figure 9). On the other hand, the potential under conditions of uniform and presumably passive dissolution will be significantly higher. However, the redox potential under these conditions may be entirely different depending on the corrosion products in solution. Species Distribution Distribution of species in solid and aqueous species in the solution at a given temperature depends upon its solubility limit, redox potential and pH. In the uniform corrosion stream all species are present as insoluble hydroxides. The Fe-H,O Eh-pH diagram indicates that FeO(0H) and Fe" as the dominant species.' The formation of Fe3+ion occurs at pH < 1 and redox potential greater than 800 mV (SHE). The speciation products observed in the uniform corrosion stream consist of FeOOH, Cr(OH), and Ni(OH),. However in the localized corrosion stream, speciation results indicate formation of Fe", Ni", Cf" ions along with small quantities of FeCl', NiCI', CrCl", and CrCl; complexes formed in solution with 0.25 moles or higher concentrations of metal chlorides in the localized corrosion stream. The combined stream showed that upon mixing uniform and Iocalized stream, a fraction of hydroxides in the uniform stream dissolved in the aqueous species with proportionate increase in the ionic concentration of metal ions in the localized corrosion stream as shown in Figures 4 and 5. In the presence of oxygen in the uniform corrosion stream, the concentrations of the insoluble species in the combined stream significantly reduced, and the formation of C p species was predicted as discussed earlier. In the HLW glass stream, several solid phases such as talc, quartz, Na,SiO,, and low albite were observed to form due to speciation of hydrolyzed products. Formation of solid phases observed in these simulations are similar to secondary phases observed on the glass surface in experimental investigations on several HLW glass comp~sitions.~ Note that in these simulations, a simple six-component glass composition was used that limits the number of secondary phases that can form. In a 20-25 component HLW glass system, the number of secondary phases will increase. However, the quantity of these phases would be limited by the amounts present in the glass matrix. The solid and aqueous species present in the HLW glass stream underwent changes with a drop in the pH of the in-package solution. Na,SiO, dissolved first, followed by talc. The dominant aqueous species observed in the in-package solution are Na' and B(OH)/' with minor amounts of B(OH),, NaB(OH),, NaHSiO,, and H,SiO,-'. Silicates in the aqueous phase dissociated and
Environmental Issues and Waste Management Technologies VIII
271
B(OH),’- converted to B(OH), (aq) with the drop in the pH of the in-package solution. Fe2Si0, is the only solid phase formed by the interaction of HLW glass WF stream and combined stream. The formation of this phase is shown by Eq. (6). This phase dissociated as the pH of the in-package solution dropped. Pan et al.” showed that the precipitates formed in glass samples leached in the presence of FeCl, and FeCI, solutions contain Fe and Si as major components. 2Fe2+ + Si02 (aq) + 2H20 = Fe2Si04 + 4H+
(6)
Simulations, supported by some experimental studies in the literature, show that a wide range of chemistries can exist within a waste package. The DOE’Sin-package chemistry analysis indicate that the pH of the in-package solution could range between 3 and 10 which is consistent with the range of pH observed in these simulations.” However, one of the main differences in the calculations is the assumption by DOE that the corrosion of WP components result only in oxides and hydroxides. DOE calculations suggest the formation of Cr6 as the main factor in the decrease in pH. Additionally, they do not observe a decrease in pH when HLW glass dissolution is assumed. The formation of Cr6+requires redox potential higher than 600 mV vs. SHE. Formation of C f +by the oxidation of CP‘ may be extremely slow under the anticipated conditions at Yucca Mountain. Simulations are underway to examine speciation of radionuclides in the in-package solutions. CONCLUSIONS Results of this simulation study on the evolution of chemistry in the waste package, using a thermodynamic speciation software, show that WP corrosion products play an important role in the determination of the pH, redox potential, and distribution of solid and aqueous species in the in-package solution. The solid phases observed for uniform, localized, and HLW glass WF are similar to experimentally observed phases. The pH of the solution in the absence of localized corrosion increases slightly ,thus a slight increase in HLW glass WF dissolution is expected. Initiation of localized corrosion reduces the pH of the in-package solution. In a closed system the pH of the in-package solution decreases as localized corrosion increases, hence reducing the HLW glass WF dissolution rate. However in a open system where oxygen is present, pH could drop below 7 resulting in higher dissolution rate for HLW glass WF. The final pH will depend upon the competing effects of the localized corrosion, which will tend to drop the pH and the increase in glass dissolution rate, which will shift the pH higher. REFERENCES ‘U.S. Department of Energy, “Yucca Mountain Science and Engineering Report,” DOERW-0539, Las Vegas, NV, 2001. 2Brossia,C.S., L. Browning, D.S.Dunn, O.C. Moghissi, 0. Pensado, and L. Yang, “Effect of Environment on the Corrosion of Waste Package and Drip Shield Materials,” CNWRA 2001-03. Center for Nuclear Waste Regulatory Analyses, San Antonio, TX, 2001. ,M. Rafal, J.W. Berthold, N.C. Scrivner, S.L. Grise, “Models for Electrolyte Solutions,” pp. 601-670; in Modelsfor Themdynamic and Phase Equilibria Calculations,Edited by S.I. Sandler. M. Dekker, New York, 1995. ,Sridhar, N. and D.S. Dunn, “Zn-SituStudy of Salt Film Stability in Simulated Pits of Nickel by Raman and EIectrochemical Impedance Spectroscopies,”Journal of Electrochemical Society 144 [12],4,243-4,253 (1997).
272
Environmental Issues and Waste Management Technologies VIII
5C. Leygraf and T.E. Graedel, “Atmospheric Corrosion,” Wiley-Interscience, New York, 2000. 6Brossia,C.S., D.S. Dunn, and N. Sridhar, “Critical Factors in Localized Corrosion III,” p. 485, Edited by R.G. Kelly. The ElectrochemicalSociety, Pennington, NJ, 1998. ’Ebert, W.L., J.C. Cunnane, and T.A. Thornton, “An HLW Glass Degradation Model for TSPA-SR,” in Proceeding of the ghInternational High-Level Radioactive Waste Management Conference (ZHLRWM), American Nuclear Society, La Grange Park, IL, 2001. *Dunn,D S.,G.A. Cragnolino, and N. Sridhar, “Localized Corrosion Initiation, propagation, and Repassivation of Corrosion Resistant High-Level Nuclear Waste Container Materials,” Corrosion 96, Paper No. 97, NACE International,Houston, TX, 1996. W.S. Department of Energy, “High-Level Waste Glass A Compendium of Corrosion Characteristics,” Volume 2. U.S.Department of Energy, Washington, DC, 1994. ‘(’Pan, Y. -M., V. Jain, M.B. Bogart, and P.G. Deshpande, “Effect of Iron Chlorides on the Dissolution Behavior of Simulated High-Level Waste Glasses,” in Ceramic Transactions, Vol. 119. Edited by D.R. Spearing, G.L. Smith, and R.L. Putman. American Ceramic Society, Westerville, OH, 2001. “CRWMS M&O, “SupplementalScience and Performance Analysis-Report Volume 1 of 2,” TDR-MGR-MD-000007, Revision 00, CRWMS M&O, Las Vegas, NV, 2001.
Environmental Issues and Waste Management Technologies VIII
273
This page intentionally left blank
LEACHITVG FULL-SCALE FRACTURED GLASS BLOCKS Yves Minet and Nicole Godon Commissariat ? 1’Energie i Atomique, Nuclear Energy Division, Confinement Research and Engineering Department, BP 17171, 30207 Bagnols-sur-Ckze,France ABSTRACT This paper summarizes the results of full-scale leach tests with French nuclear glasses, mainly inactive R7T7-type glass. The objective was to measure the glass surface area altered under “initial rate” (YO) conditions, i.e. more severe than the package alteration expected in a geological repository. The effective fracture ratio under these conditions (the ratio between the altered glass mass of the fractured block and that of the monolithic block), equal to 5 1, is compared with the results of other measurements intended to better understand glass fracturing: leaching of loose glass fragments as well as dry measurementsby y tomography and fragment size grading. INTRODUCTION Conservative models of high-level nuclear waste containment glass package alteration, developed mainly for R7T7-type glass, are based on the initial glass alteration rate, (1-0)‘~’.It is important to determinethe surface area (5‘)to which this rate is applicable, however, and which probably exceeds the geometric surface area of the monolithic glass block because of fracturing. Experiments have been conducted since 1985 to evaluate glass fracturing, mainly with full-scale blocks of inactive R7T7-type (SON68) nuclear glass, either by leach testing under penalizing “initial rate” (YO) conditions[21or by dry measurements: t ~ m o g r a p h y [and ~ ~fragment ~ ~ ~ ] size gradingr2].This paper describes the experiments and reviews the results obtained, in order to determine and define an “effective fracture ratio” under initial rate conditions. This concept could be extended to glass package behavior studies with progressively confiied leaching solutions. GLASS BLOCKS STUDIED Fifteen inactive glass blocks representative of vitrified industrial reprocessing waste from light-water reactor fuel (13 R7T7-type SON68 glasses) or gas-graphite
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
275
reactor fuel (2 AVM-type SAN61 glasses) were fabricated by the CEA. The major components in the composition of SON68 and SAN61 glass are indicated in Table I. Table I. Chemical composition (wt%) of SON68 and SAN61 glass Oxide
SON68
SAN61
Oxide
SON68
SAN61
Si02 B203 Na20
45.1 13.9 10.1 4.9
39.7 16.6 16.6 9.9
CaO MgO Fp surrogates Other
4.0 0 11.3 10.7
0.8 5 .O 6.0
A203
5.4
The objective was to characterize fracturing under conditions of penalizing alteration, but realistic with regard to the glass mechanical strength. The glass blocks were poured under different conditions: some into a perforated basket to simulate partial accessibilityto water, and the others into a container comprisingfour remvable Table 11. Glass blocks tested, with manufacturing and test conditions (see below for column headings) 1
2
3
(lkspaper) Sample1141 (Thispaper) W4OOL2' A/4OOL2] B/400L2' LA 3/200[21 L5 W200'" L6 1/2OOLL' ~L7 2/200[21 L8 (ThisDaDer) . L9 BlockDL3I Ll 0 Block A'" L11 Sample2L41 L12
P1 P2 P3 L1 L2 L3
~~
L
1. 2. 3. 4. 5. 6.
276
I
I
1997 1999 1999 1986 1986 1986 1985 1985 1985 1985 1995 1995 1991 1998 1996
4
5
6
7
S9O"C SON68 400 H D100"C SON68 400 H D100"C SAN61 250 H S 100°C SON68 400 S S 100°C SON68 400 H S 100°C SON68 400 H SON68 200 H S 100°C D 100°C SON68 200 S D 100°C SON68 200 H D100"C SON68 200 H D100"C SON68 400 H D100"C SON68 400 H D1OO"C SON68 400 S D 100°C SON68 400 H D100"C SAN61 385 H Column headings:
8
9
P P P L L L L L L L L L L L L
1152d 3.3d 3.7d 21 d 21 d 21 d 21 d 21 d 21 d 21 d 4.3/1.3d 2.U2.1 d 2.3d 3.3/3.0 d 3.813.9d
10
-
T
-
G
-
G
-
T T T
-
Specimen designation in this paper 7. Block leaching conditions4ynamic (D) Specimen designation in prior reference or static (S)-and temperature Year of glass fabrication 8. Block leaching conditions: perforated basket (P) or loose glass (L) Type of glass: SON68 or SAN61 Weight of block (kg) 9. Leach test duration (days) Heat treatment: L a Hague cooling scenario 10. Dry fracturing examination: tomography (T) or fragment grading (G) (H)or slow cooling at 3°C. h-' (S)
Environmental Issues and Waste Management Technologies VIII
sectors that were detached prior to leaching the loose glass in order to maximize the accessibility to flowing water to the cracks and fracture surfaces. Most of the blocks were cooled according to the standard industrial scenario at La Hague, but a few were cooled very slowly (3°C. h-’) to assess the affects of heat treatment on the degree of fracturing. Some of the blocks were also submitted to dry examination (tomography and fragment size grading) before leaching to estimate the total surface area in cracks and fractures of sufficient size to favor glass alteration under initial rate conditions. The specimen blocks are listed in Table II, together with their fabrication and test conditions. Some of the loose specimens (L8, L9, L l l , L12) were tested in two parts, requiring two leach tests. Only a 108 kg piece of block L10 (400 kg) was tested. METHODS FOR MEASURING THE DEGREE OF FRACTURING Leaching Experiments with Full-scale Glass Blocks The experimental objective was to measure the quantity of glass altered per unit time, compared with the values that would be measured by leaching a specimen with the same reactive surface area as the monolithic block. The specimens were leached either in a perforated basket (P1,P2, P3) to simulatepartial accessibility to water of a block inside a breached canister (Figure l), or as loose fragments to maximize the accessibility to flowing water to the cracks and fracture surfaces.
Figure 1. Glass block poured into a perforated basket simulating partial accessibility to water. From Ref. [4].
Figure 2. ‘Large-block Soxhlet” dynamic leaching test facility
Environmental Issues and Waste Management Technologies VIII
277
The 'Large -block Soxh1et"test facility (Figure 2) is designed to leach glass blocks weighing up to 400 kg, 1.1 m high and about 0.4 m in diameter. It consists of a boiler beneath a water-cooled condenser. The main leaching vessel is 2.1 m high and 0.7 m in diameter, with a total volume of 784 L. The glass block (generally weighing 400 kg) is placed inside a leaching boat suspended above the boiler heated by a steam jacket (180°C at 10 bars). Although the test facility is generally used for dynamic leaching experiments at lOO"C, several experiments were performed under static conditions (90-100°C) using a slightly modified setup without the leaching vessel, reducing the internal volume to 523 L. Samples ranging from about 10 to 100 mL of leachate were taken from the boiler and from the leaching boat at intervals ranging from a few hours to a few days (dynamic tests) or months (static tests). The major glass constituent elements (Si, B, Na, Li, MO,Cs, Sr) were analyzed by ICP-AES (atomic absorption spectroscopy for Cs), and the pH was measured at room temperature or 90°C as each sample was taken. The elemental analysis results were used to measure the altered glass mass ( M a l t ) per unit time, from which the leached fracture ratio (LFR) was calculated based on the initial alteration rate (ro)and the monolithic block surface area (Smonolith) according to the following relation: Malt
LFR =
r0 Smonolith
Dry Fractured Surface Area Measurements Tomography of the fractured suqace area of a glass block in its canister. Tomography is a nondestructivetechnique for measuring the developed surface area of a glass block including the cavities (apical void) and all fractures and cracks thicker than the resolution of about 0.1 mm. The block is irradiated by yphotons produced by a radioactive 6oCosource (for two blocks[31:L9 and L10) or by Bremsstrahlung of an electron beam from a linear accelerator (for two other blocks[41:P2 and L1 1). Several 2D tomographic cross sectional images were obtained (up to 120 for P2 and L1 1). The raw images were first corrected for artifacts arising from the test method (hardening of the electron beam in the case of the linear a~celerator'~]), then converted to grayscale for analysis and conversion into a representation of the network of cracks and cavity boundaries. The total surface area of the fractures and cavities was then estimated for each image by counting the number of intersections between a network of straight lines and the fracture network; the resulting estimated surface area is subject to a relative uncertainty of 20-25%14]. The tomographic fracture ratio ( F R ~ )is then defined as follows: FRT =-
Stotul Smonolith
278
Environmental Issues and Waste Management Technologies VIII
Fragment grading. Test blocks LA and L7 were graded prior to leaching[21by counting and weighing the fragments of the block after removal from the canister. The fragments were sorted into nine unit weight classes from ‘iessthan 10 g”to ‘inorethan 5 kg”. The number of fragments (Ni) and total weight (Mi) was determined for each class; the relative surface area of the range was then calculated by assuming all the fragments were identical cubes:
where p is the glass density (2.75 g. cm-3for SON68 glass). The sum of the surface areas for each class (ZSi).The graded fracture ratio (FRG)is thus defiied as follows: FRG =-
CSi
Smonolith
(4)
RESULTS AND DISCUSSION Leach Tests and Dry Surface Area Measurements The results of the fifteen leach tests P1 to P3 (glass blocks leached in perforated baskets) and L1 to L12 (loose fragments) are shown in Table III together with the dry fracture measurement results for blocks P2 and L11 (tomography by y photons produced by Bremsstrahlung of an electron beam from a linear accelerator), L9 and L10 (tomography by y photons Erom a 6oCosource), L4 and L7 (grading). The alteration tracer elements are indicated as well as the raw experimental results (altered lass mass M a l t calculated from the mean normalized mass losses for the tracer elements$1 and used to calculate the annual altered glass fraction AAGF = Malt/Mblock). Three fracture ratios are given: the value determined by the author of the published leach tests, the LFR value calculated from M a l t and Smono/ith using Q. (1) based on the initial alteration rate ro at the water temperature and at neutral pH, and the dry measured fracture ratio. The value of ro was calculated for the test temperature from laboratory test results of 1.6 g m-2d-’(SON68 glass[3141) and 4.0 g m-2d-1(SAN61 glass) at 100°C, with an activation energy of 90 kJ mol-l. Discussion Comparison of fracture ratios. The results shown in TableIII are generally consistent even when the fracture ratios reported in the literature seem to differ: the differences are due to the choice of tracer elements (Si and B for blocks L1 to L712] although Si is partially retained in the alteration gel, particularly under static conditions), to the method of calculating ro, or to an underestimated value of Smonolith. The perforated basket leach tests P1 to P3, correspondingto alterationunder initial rate conditions with a realistic assessment of flowing water accessibility to the block
Environmental Issues and Waste Management Technologies VIII
279
surface, yield a leached fracture ratio (LFR) of about 5 and an annual altered glass fraction (AAGF) of about 0.01 y-' for SON68 glass in a continuously renewed medium at 100°C. The AAGF value observed for specimen P1 can be attributed to the test temperature, and the value for specimenP3 is due to the glass composition (the SAN61 glass alteration rate is about 2 to 3 times higher than for SON68 glass). Table III. Leach testing of industrial-scale glass blocks in perforated baskets or as loose fragments: altered glass mass per unit time and apparent fracture ratio (see below for column headings)
1
2
3
4
5
6
7
P1 P2 P3a Llht L2 L3 L A LSht L6 L7 L8 L9 LIOht L11 L12a
B B,Na,Li,Cs B,Na,Mo B B B B B B B B.Na.Li , , B,Na,Li B. Na. Li B,Na,Li,Cs B,Na,Li
5.5 (1 d) 12.3 k3.0 25.7 13.7 68.2 61.7 37.6 7.6 41.3 52.4 149 166 35.7 145+25 124
0.005 0.011 0.034 0.012 0.062 0.056 0.068 0.014 0.075 0.095 0.14 0.15 0.12 0.13 0.23
90 99 f l 99 99 99 99 99 99 99 99 99 99 99 99 99
0.72 1.5 3.7 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 3.7
1.7 1.7 1.2 1.7 1.7 1.7 1.0 1.0 1.0 1.0 1.7 1.7 0.67 1.7 1.2
I
,
8
9
10
4.5 5 flL4]4.8 7f2T 5.2 1.64[21 5.3 14.1'" 27 12.1''' 24 10.1['' 25 15.0G 1.02[21 5.1 27 5.7''' 7.3IL1 35 15.5b 58 66'" 65 4.4' 51'"' 36 2.2' 53+9[41 57 15+4T 53 -
Column headings: 1. Test block (P: perf. basket: L: loose glass) 7. Monolithic glass surface area Srnonokth (mL) 2. Alteration tracer elements 8. Fracture ratio from published test results 3. Altered glass mass Malt ( g d-') 9. Leached fracture ratio (LFR) calculated 4. Annual altered glass fraction AAGF (yr-') using Eq. (1) 5. Glass contact temperature T ("C) 10. Dry measured fracture ratio and method 6. Initial alteration rate ro(T) at neutral pH (T: tomography, G: fragment grading) (g. m2d-') Notes: a SAN61 (AVM-type) glass blocks (all ht Heat treatment (slow cooling at 3°C- h-') to minimize fracturing others were SON68 (R7T7-type) glass
The loose glass leach tests L1 to L12 also provide consistent results, with LFR values ranging from 25 to 65 and AAGF values between 0.06 and 0.14 y-' (SON68 glass) except for two of the three slow-cooled blocks for which the LFR values were near 5 . Although fracture ratios of 25 to 65 are indicative of the maximum alterable surface area accessible to water and are thus of scientific interest, they are in no way representative of glass alteration in a geological repository because they assume the
280
Environmental Issues and Waste Management Technologies VIII
block is fragmented, dispersed, and accessible to water flowing at a high rate. These values are therefore not usable for operational simulations. The measured dry fracture ratios are also consistent: about 15 for the two graded measurements[21,and between 4.4 and 15 for the three tomography measurements on blocks that were not submitted to slow cooling. The variability of the tomography measurements is due to the resolution: 0.2 mm for y photons from a 6oCosource[33and 0.1 mm for the Bremsstrahlung y photons[41.The fact that the dry measured values lie between the apparent fracture ratios for the restrained and loose leach test values shows that they provide a suitable estimate of the accessible surface area. Effectivefracture ratio. The results show that leach tests can be considered as an overall glass block alterability measurement under initial rate conditions (with leaching solution renewal) at a given temperature. The annual altered glass fraction (AAGF) or ro- S is first measured. These tests give the effective fracture ratio (EFR), or the ratio of the altered glass mass per unit time M a l t to the value Mmonolith that would be obtained for a monolithic glass block entirely accessible to renewed water under the large-block Soxhlet test conditions, and defined as follows: ro(T,pH)- Smonolith where, strictly speaking, ro is specified for the pH in the largeblock Soxhlet test boat if it contained a monolithic glass block. The pH correction is unnecessary for this type of test, however: tests L8, L9 and L10L3]show a gradual rise in the pH on contact with the glass (from 6 to 8.5) with no increase in the alteration rate. The effective fracture ratio (EFR) can therefore be considered equal to the leached fracture ratio (LFR) as in the following relation:
where roo(T) is the initial alteration rate in pure water at neutral pH measured in the laboratory, and Smonolith is the geometric surface area of the glass block. This fracture ratio is equal to 5 1 under realistic conditions of water accessibility to the glass surface (i.e. the glass restrained in a perforated basket) for which the annual altered glass fraction is about 10-2y-’ at 100°C. CONCLUSION AND OUTLOOK Leach tests conducted since 1985 with industrial-scaleglass blocks provide a basis for a serious assessment of fracturing and its effect on alteration under conditions with leaching solution renewal. A protocol was specified for measuring this effect from the annual altered glass fraction (roe S) measured under realistic conditions of water accessibility to the fractures and cracks: the largeblock Soxhlet protocol applied to glass blocks retained inside a perforated basket. An ‘kffectivefracture ratio”was then defined as the ratio of the altered glass quantity measured using this protocol to the altered glass quantity of a perfectly monolithic glass block under the same conditions.
Environmental Issues and Waste Management Technologies VIII
281
The EFR value of 5 2 1 is confirmed by the order of magnitude of the dry fractureratio measurements (ranging from 4 to 15); this is about an order of magnitude lower than the ratio measured during the loose glass leach tests, which allowed maximum accessibility to flowing water (possibly compounded by crack propagation when the package was opened). The latter cannot be considered representative of package alteration in a repository. Nevertheless, the values measured on inactive glass could overestimatethat of the industrial glass blocks in which the temperature gradient due to the presence of radionuclides tends to diminish fracturing. This methodology is currently being used to estimate the scale effects under progressive confinement of the leaching solution. The development of more realistic glass alteration models such as the r(t) model[51implies that the contribution of fractures and internal cracking to glass alteration must be integrated in order to apply those models to industrial-scale glass blocks. Such contribution is expected to be less sigmticant than under initial rate conditions: that is observed on several laboratory testsL6]and interpretated by the presence of a protective alteration gel in the internal cracks, arising from the high concentrations of elements released from the glass (Si in particular). Long-term static leach tests scheduled to last for several months to several years are now in progress to confirm this trend on industrial-scale glass blocks. REFERENCES
’
T. Advocat, J.L. Crovisier,E. Vernaz, G. Ehret and H. Charpentier, ‘Hydrolysisof R7T7 nuclear waste glass in dilute media: Mechanisms and rate as a function of pH’, MRS Symposium Proceedings,Scientific Basis for Nuclear Waste Management XIV 212,57-64 (199 1). J.P. Moncouyoux, A. Aurk and C. Ladirat, ‘Investigationof full -scale high-level waste containment glass blocks”, European Commission Report EUR 13612 EN (199 1). J. Goebbels, P. Reimers, M. Aouri, P. Jollivet,N. Godon, E. Vernaz, C. Lierse, K. Krebs, E. Kaciniel and W.Stower, ‘Non-destructive examination of nuclear radioactive waste packages by advanced radiometric methods”,European Commission Report EUR 18040 EN (1998). M.R. Senk, M. Bailey, B. Illerhaus, J. Goebbels, 0. Haase, A. Kulish, N. Godon and J.L. Chouchan, ‘Characterizationof accessible surfacearea of HLW glass mo noliths by high energy accelerator tomography and comparison with conventional techniques”, European Commission Report EUR 19119 EN (1999). I. Ribet, S. Gin, Y. Minet, E. Vernaz, P. Chaix and R. Do Quang, ‘Long-TemBehavior of Nuclear Glass: the r(t)Operational Model”, Proceedings of GLOBAL 2001: Intemtional Conference on Future Nuclear Glass, Paris (2001). N. Godon, “Simulations exphrimentales des interactions verres-mathriauxd‘environnement: des essais de laboratoire aux tests in-situ” (Experimental simulations of glassenvironmental materials interactions: from laboratory to in-situ tests), pp. 503-5 13 in Glass: Scientific Research for High-Per$ormance Containment, CEA-Valrh8 Summer Session proceedings, Mhjannes-le-Clap,France, edited by Graphot (1997).
282
Environmental Issues and Waste Management Technologies VIII
DEVELOPMENT OF SENSORS FOR WASTE PACKAGE TESTING AND MONITORING IN THE LONG-TERM REPOSITORY ENVIRONMENTS V. Jain, S. Brossia, D. Dunn, and L. Yang, Center for Nuclear Waste Regulatory Analyses Southwest Research Institute San Antonio, TX 78238 ABSTRACT Performance of the waste package (WP) has been shown by the U S . Department of Energy (DOE) through its total system performance assessment as one of the most important factors affecting the long-term performance of the proposed Yucca Mountain repository for high-level radioactive waste (HLW). To assist the Nuclear Regulatory Commission (NRC) in the review of the DOE’S performance confmation plan and the validation of the laboratory corrosion test results in long-term repository environments, a program to develop and evaluate possible sensors for monitoring corrosion processes was initiated. The initial results are summarized in this paper. INTRODUCTION The current DOE plans call for the emplacement of HLW in WPs constructed of a 5 cm thick Type 316 Nuclear Grade stainless steel (SS) inner container, surrounded by a 2 cm thick outer container of Alloy 22 (57Ni-22Cr-13.5Mo-3W3Fe), and placed in a horizontal drift (tunnel)at the proposed site in Yucca Mountain (YM), Nevada. Although Alloy 22 exhibits a low corrosionrate and a high resistance to localizedcorrosion in many aqueousenvironments,the greatest uncertaintyrelated to long-term material behavior is the nature of the environment to which the containers could be exposed. Therefore, estimation of water chemistry during the service period and determination of its effect on corrosion behavior of the outer container surface are important for detecting any anomalous conditions that may lead to premature corrosion and container failure.1,2 Several factors such as concentrationof anionic species,temperature,pH, and redox potential influence the corrosion mode of a metal in a given environment? To initiate and sustain localized corrosion of Alloy 22, a sufficiently high chloride concentration and a corrosion potential higher than a critical potential (represented To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
283
by the repassivation potential as a lower bound) are nece~sary.~ If the repassivation potential and chloride concentration are known as a function of temperature, the corrosion potential can be used as an indicator for the likelihood of localized corrosion. The redox potential of the aqueous environment, which is a measure of its overall oxidizing or reducing nature, is known to influence the corrosion potential. This study is focused on sensors for measuring chloride concentration, redox potential, and localized corrosion. SENSOR DESIGNS AND EXPERIMENTAL SETUP Potentiometric and non-potentiometric methods are used to assess the possibility or occurrence of corrosion. Potentiometric methods rely on the measurement of a potential difference between a sensing electrode and a stable reference. The potentiometricmethods employed includemonitoringpH and chloride concentration using an oxidized tungsten wire for pH and a chloridized silver wire for chloride. These wires, along with Alloy 22 and Type 3 16L SS wires to measure corrosion potentials, were incorporated into a sensor array cell, shown in Figure 1. The sensor array was designed to capture incoming water percolating through crushed tuff in a laboratory-scale drift heater test. A saturated calomel electrode connected to the array cell through a long salt bridgehggin probe was used as a reference electrode. The non-potentiometric Figure 1. Sensor array for potentiometric methods include measurement of measurements solution conductivity and the use of a galvanic couple sensor to measure corrosivity. Conductivitymeasurements can provide some informationregarding the overall corrosivity of the environment because increasing conductivity generally indicates a more corrosive environment. Solution conductivity was determined through the use of two platinum wires incorporated into the sensor array cell. Figure 2 shows a galvanic couple sensor, developed on the bases of the concept put forth by Shinohara et al.4This sensor consists of an interdigitated array of silver electrodes that is electrically isolated from the substrate.
284
Environmental Issues and Waste Management Technologies VIII
The substrates investigatedthus far have been carbon steel and Type 304L SS. Through the use of two substrates, a relative corrosivity scale can be developed. The carbon steel/Ag system is more sensitive to a low overall corrosivitywhereas, given the increased resistance of Type 304L SS to corrosion, the Type 304L SS/Ag system would respond only in more aggressive environments. The possibility that the Type 304L SS/Ag system may be used to detect the onset of localizedcorrosion is also being investigated. These systems were also incorporated into a laboratory-scale simulated drift heater test (Figure 3) to examine the effects of Figure 2. A schematic of the galvanic thermal refluxing on material couple sensor performance and environmental conditions. The drift is being heated to an air temperature of -105 "C to simulate the thermal load that would result from radioactive decay within the waste containers.Deionized water, equilibratedwith the tuff is added continuously for over 18 mo and is dripping onto the sensor arrays, galvanic couple sensors, metal coupons, and solution catch cups. A picture of several coupled multiple-electrode array sensors is given in Figure 4. The coupling of the electrodes through resistors allows the use of a high resolution voltmeter instead of a zero resistance ammeter for the measurement of the coupled currents, as shown in Figure 5 . The working principle of the sensor, the auto-switching and electrodelocation-mapping instrument have been described previ~usly.~The sensing electrodes were cut from rods or wires of alloy under examination. The wires were polished with 600 grit Sic paper prior to Figure 3. Display of sensors inside a simulated drift during the simulated heater test setup
Environmental Issues and Waste Management TechnologiesVIII
285
FiWe 4. Coupled array sensor designs
Figure 5 . Schematic of a coupled multiple electrode array sensor
the start of the tests. All solutions were prepared using reagent chemicals and deionized water. Experiments were conducted in a 3-L glass cell, filled with 2-L airsaturated solutions. The solution was slowly agitated on a magnetic stir plate during the experiments. RESULTS AND DISCUSSION Measurements of pH and chloride concentration using potentiometric methods were found to be accurate and predictable over a range of 2 to 10 for pH and from lOP3 to 1 M for chloride concentration. When water, equilibrated with crushed tuff from YM for 5 months, was dripped onto the sensor array cell, the chloride concentration was determined to be less than 10-3M, even after repeated dryout and rewetting cycles. The solution pH was initially near 6 when the heater was turned on. Results fiom the drift-scale heater test monitoring chemistry are shown in Figure 6. After the heater was turned on and the solution began to evaporate, the pH was noted to increase to nearly 11, likely fi-om exsolution of CO, as a result of the increased temperature. Sensor response in terms of chloride concentration and pH are in line with the compositions of extracted water alliquots fiom the vicinity of the sensor array (chloride 2-8 mM, pH 8.0-8.6). The solution conductivity was also measured during the bench testing and provides an insight into the corrosivity of the environment. The conductivity was noted to increase markedly as the solution began to dryout, reaching a peak value at the point of dryout when electricalcommunicationwas lost between the two Pt wires.
-
286
-
Environmental Issues and Waste Management Technologies VIII
Conductivity, though useh1 for determining 1 12 the presence of liquid c water, cannot easily be $ 11 converted to a corrosion 2 0.1 rate. One of the primary 80 reasons for this is that 10 5 conductivity does not U capture the effect of the 3 0.0 1 9 8 redox potential that is E known to strongly 0.001 8 influence the corrosion 53424 53484 53544 53604 53664 rate by affecting the Time, min corrosion potential of the metal. Furthermore, Figure 6. Chloride concentration and pH measured conductivity cannot using the sensor array in a laboratory scale simulated distinguish among drift different anions and, consequently, cannot be used to determine the specific effect of the various anions on the corrosion mode. The galvanic couple sensors were evaluated in a humidity chamber after .r(
Y
d
1E+OO8
T: --
A A
A A A p
sensor measured as a function of relative humidity and concentration of a 2.5 mL salt
Environmental Issues and Waste Management TechnologiesVIII
287
1XI 0 '
both changes in chloride concentration and %RH, especiallyat low %RH.At higher %RH, the resistance measured was independent of the i! c chloride concentration. 0 Furthermore, based on the 0 . independently measured measured RH, RH, independently a 88 88 the relative corrosivity of the Type304L 304L Type water dripping onto the sensor Stainless Steel in the laboratory scale heater Ipprn CItest is is comparable comparable to to the the test 0 10 pprn CII00 pprn CIcorrosivity of of 100 100 ppm ppm corrosivity c h l o r i d e s o l u t i o n . c h l o r i d e s o l u t i o n . IInn 1XI o7 comparisonto to the the carbon carbon steel steel -- comparison 20 40 60 80 1uu substrate sensor, the Type 304 Relative Humidity, % Figure 8. Type 304 SS galvanic couple sensor ss showed much response as a function of chloride concentration greater resistance to corrosion and thus required a higher and RH chloride concentration (i.e., more corrosive conditions) to achieve a measurable response (Figure 8). Though not strictly providing information on corrosion mode or rate, this sensor design holds promise for detecting and determining the corrosivity of the environment. Figures 9 shows a direct comparison for the standard deviation signals from four coupled multiple-electrode sensors made of Alloys 22 and 600 and Types 3 16 and 304 SS to the changes of temperature in 0.1M ferric chloride solution. It has been shown from our previous w0rk~9~ that the standard deviation of the sensor currents is an effective indicator for localized corrosion. Both the maximum anodic current (most negative currents) and the standard deviation of each sensor increased with increasing temperature. All sensors, with exception of the Alloy 22 sensor, showed a significant increase in the standard deviation with the change from deionized water to ferric chloride solution. The sensor was able to detect the changes in standard deviation at levels as low as 5 x 10-" A. At room temperature, the sensor is adequately sensitive to detect differences in the corrosion of Types 304 and 3 16 SS, Alloy 600 and Alloy 276 in de-ionized water. Alloy 22 was found to be the most resistant to localized corrosion among the alloys tested.
''
288
EnvironmentalIssues Issues and and Waste Waste Management ManagementTechnologies TechnologiesVIII VIII Environmental
1.OE-3
_~ _ _ _ _ _ _ _ _ _ _ _ _ _ - - - - - - - - - - - -- - - - - - 1.OE-11 _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 4 ,' "
1.OE-l2
I
I
I
I
Figure 9. Comparison of the standard deviation signals of four different multiple electrode array sensors in 0.1 M ferric chloride solution at different temperatures
SUMMARY Feasability of potentiometric and non-potentiometric sensors for examining the WP corrosion behavior was shown. Data collected using a laboratory-scale drift heater test corroborates the bench-scale data. A galvanically coupled multipleelectrode sensor was developed and used to measure the real-time localized corrosion of Fe-Ni-Cr-Mo alloys in chloride solutions. It was demonstrated that the multiple electrode array sensor provided a rapid real-time response to the changes in temperature and salt concentration and has a detection limit of 5 x 10-" A with respect to corrosion currents for the miniature electrodes used in the sensors.
Environmental Issues and Waste Management Technologies VIII
289
REFERENCES ‘CivilianRadioactive Waste Management System,Management& Operating Contractor, L‘PerformanceConfirmationPlan,” TDR-PCS-SE40000 1,Revision 1, TRW Environmental Safety Systems, Inc, Las Vegas, NV, 2000. 2C.S. Brossia, D.S. Dunn, O.C. Moghissi, and N. Sridhar, “Assessment of Methodologies to Confirm Container Performance Model Prediction,” CNWRA 2000-06, Center for Nuclear Waste Regulatory Analyses, San Antonio, TX, 2000. 3G. Cragnolino, D. Dunn, C.S. Brossia, V. Jain, K. Chan, “Assessment of Performance Issues Related to Alternate Engineered Barrier System Materials and Design Options,” CNWRA 99-003, Center for Nuclear Waste Regulatory Analyses, San Antonio, TX, 1999. 4T.Shinohara, S. Tsujikawa, S. Motoda, Y. Suzuki,W. Oshikawa, S. Itomura, T. Fukushima, and S. h m o , “Evaluation of Corrosivity of Marine Atmosphere by ACM (Atmospheric Corrosion Monitor) Type Corrosion Sensor,” International Symposium on Plant Aging and Life Predictions of Corrodible Structures, Japan Society of Corrosion Engineering: Sapporo, Japan, 1995. 5L. Yang, N. Sridhar, 0. Pensado, and D. Dunn, “An In-situ Galvanically Coupled Multielectrode Array Sensor for Localized Corrosion”, submitted to Corrosion,2002. 6L.Yang, and D. Dunn, “Evaluationof CorrosionInhibitors in CoolingWater Systems Using a Coupled Multielectrode Array Sensor” CORROSION/2002, Paper No. 02004, Houston, TX: NACE International, 2002. ACKNOWLEDGMENTS This work was supported by the U.S. Nuclear Regulatory Commission (NRC), Office of Nuclear Material Safety and Safeguards, Division of Waste Management (Contract No. NRC-02-97-009). The work is an independent product of the Center for Nuclear Waste Regulatory Analyses and does not necessarily reflect the views or the regulatory position of the NRC.
290
Environmental Issues and Waste Management Technologies VIII
CORROSION OF PARTIALLY CRYSTALLIZED GLASSES Pave1 h a , Brim J. Riley, and John D. Vienna Pacific Northwest National Laboratory P.O. BOX999, MS: K6-24 Richland, WA 99352
ABSTRACT Using existing data on the corrosion of partially crystallized, simulated, highlevel waste glasses, coefficients were introduced to evaluate the cumulative influence of secondary effects, such as residual stresses or concentration gradients on product consistency test response. As compared to predictions based solely on residual glass-composition effects, the results showed that cristobalite, eucryptite, and nepheline had a higher-than-predicted impact on glass corrosion, while the effects of baddeleyite, hematite, calcium-zirconium silicate, and zircon were close to those predicted. The effects of acmite and lithium silicate were opposite to those expected based on their compositions. The analysis revealed important limitations of the databases currently available. A better understanding of corrosion phenomena will require quantitative composition data, microscopic characterization of pristine and corroded surfaces, and long-term tests with glass coupons or monoliths. INTRODUCTION High-waste loaded high-level waste (HLW) glasses are likely to be partially crystallized. To be acceptable for the repository disposal, HLW glasses must satis@ the HLW glass acceptability constraint, defmed by the Waste Acceptance Product Specifications' in terms of the product consistency test (PCV2 response. Equations have been developed that relate PCT releases to the composition of quenched, crystal-fiee but no such relationships have been formulated for partially crystallized glasses. Previous studies found reasonable agreement between measured PCT data for heat-treated glasses and the estimated values for the composition of the residual glass,6l' indicating that the residual glass composition is the major factor that controls the PCT response. Here we focus on the cumulative influence of secondary effects.
~~
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
291
THEORY The PCT normalized release of B from the residual glass @B) is6-"
where r~ is the PCT normalized B release from quenched (crystal-fiee) glass, R B ~ is the extent to which the k-th crystalline phase impacts B release fiom partially crystallized glass, ck is the hction of k-th crystalline phase in glass, and M i s the number of crystalline phases in glass. According to Equation (l), phases with large negative values of R B combined ~ with large values of ck are likely to significantly increase p ~ . Equation (1) i ores several important factors. The canister-centerline temperature history'T -14 commonly used for HLW glass crystallization limits the time for crystal growth, creating regions around crystals that are depleted of some components and leaving regions of unaltered glass. Therefore, concentration gradients are likely to be present in the residual glass. These gradients and thermal expansion mismatches create residual stresses that impact the corrosion rate. Jantzen and Bickford reported enhanced dissolution around crystals of acmite. Dissolution of crystalline phases can be expected to influence short-term tests, such as 7-day PCT. In the 7-day PCT, the corrosion of durable glasses does not proceed beyond 1 pm,which is less than the size of most of crystals that form in glass during the canister centerline cooling (PCT is not designed to measure the corrosion behavior of the bulk glass with crystals). When crystallized glass is crushed for the PCT, fracture surfaces tend to pass through softer inclusions and tensile-stress areas while avoiding rigid inclusions and compressed areas. Thus, the glass-surface-to-solution-volume ratio of the PCT samples of partly crystallized glasses is difficult to estimate, and the hction of crystal at the corrosion surface may be significantly different from the fraction of crystal in the waste form. We also need to consider that the reported fiactions of crystalline phases in much of the available data are subjected to S O % error and that the actual compositions of crystalline phases are generally different from their ideal stoichiometry,and true crystal compositions were rarely dete~mined.'~'~ Finally, when comparing the calculated and measured release values, we need to be aware that the release values from heat-treated glasses were normalized to the original glass composition (i.e., the composition of quenched, crystal-fiee
'
292
Environmental Issues and Waste Management TechnologiesVIII
glass) because the composition and concentration of crystals were subject to high uncertainty. However, those compositions are different from the residual glass composition that is assumed to be the corroding material. To assess the combined impact of secondary effects, which generally vary with the crystalline phase, we have modified Equation (1) by introducing PBk coefficients for each crystalline phase:
k=l
where PBk is the k-th crystalline-phase coefficient for E3 release. COMPUTATIONS Compositions and PCT data of glasses used in this study are listed in previously published and in the recently issued database? The pk ’ values were obtained fitting Equation (2) to data. An iterative optimization technique was used to maximize I? defined as
j=1
where n is the number of data points and ‘4,- is the average normalized B release fi-omj-th crystal-free glass. Calculations were performed for all data (the full data set) and for data from which outliers (glasses with the highest values of Iln[rdp~]I) were removed (the reduced data set). RESULTS AND DISCUSSION The R B k values’2 and calculated PBk values are listed in Table I. As the last row shows, dissolution data of crystallized glasses are represented fairly well by Equation (2) with optimized P’k coefficients; R2 = 0.73 for all data and 0.89 for the reduced set are not unreasonable when compared to R2 = 0.81 and 0.82 for the quenched-glass PCT fust-order models for B and Na releases, respectively? However, some P’k coefficients (those in parentheses) are indeterminable. Let us assume for convenience that only one phase precipitates ( M =1) in the glass. This reduces Equation (2) to
Environmental Issues and Waste Management Technologies VIII
293
Clearly, a crystalline phase has little impact on PCT if
If h v - ~=: RBk, then, by Equation (3), lnm = b ~and , P B can ~ have an arbitrary value. In our database, the extreme values for re in g/m2were b B ( m j n ) = -2.12 and I n r ~ (=~3.78. ~ ) Values of P’k for phases with -2.12 < RBk < 3.78 are virtually indeterminable. When calculated with an iterative optimization technique, the pk ’ values for these phases are likely to change significantly even if one data point is added to or removed fiom the database. For all phases in Table I that do not meet ’ > 0, precipitation of these phases Condition (9,R B <~ InrB(mjn). Provided that pk results, by Equation (4), in an increased B release fiom the glass. Table I. ODtimized ’p coeficients for crvstallinc 3hases ~ta Set
Formula Crystalline phase RBk PBP) 12 (-9.04) acmite NaFeSi20c -1.9 4 1.oo baddeleyite -14.5 zfl2 1 0.44 calcium zirconium silicate Ca2ZrSi4012 -7 SiOz 2.13 12 -4.1 cristobalite MgSi03 enstatite (-6.25) 4 -0.3 6.76 LiAlSiO4 eucryptite 3 -8.1 Ca2A12Si07 1 9.75 gehlenite -14 0.74 hematite - maghemite 4 -11.9 Fe203 1 (4.68) NaMg2CrSi30I 0 krinovite 0.6 1 lithium magnesium silicate Li2MgSiO4 2.9 (0.Ow lithium silicate Li2Si03 2.6 17 (-36.3) 1.62 NaAlSiO, nepheline 19 -6.2 olivine 1.o Mg2Si04 3 (-1.15) -11.9 0.02 spine1 NiFQO4 18 -9.3 ZrSi04 1.17 zircon 5 0.729 R2 [%even outliers were removed. @ b enumber of heat-treated glasses with the k-th crystalline phase. cc>virtuallyindeterminable values are in parentheses.
E
d Data Set(’)
m@)
11 3 1 11 4 3 1 3 1 1 14 15 3 15 4
(4.19) 1.03 1.11 1.87 (-17.7) 6.76 9.77 1.12 (205.6) (-141-1) (-34.9) 1.68 (-1.85) 0.96 2.07 0.885
Acmite, lithium silicate, olivine, enstatite, krinovite, and lithium-magnesium silicate are virtually indeterminable, but acmite and lithium silicate are
294
Environmental Issues and Waste Management Technologies VIII
represented by large numbers of data, and their RBk values are close enough to the limits of the indeterminability interval to justifjr further discussion. Based on the R B values, ~ one would expect that acmite slightly increases and lithium silicate slightly decreases the B release from most glasses, but the measured effects are exactly opposite, as the negative PBk coefficients indicate. This anomaly can be attributed to acmite being a solid solution of several clinopyroxenes'8 with an overall composition substantially different from the assumed stoichiometric formula. Regarding lithium silicate, we can argue that this rapidly dissolving phase increases solution pH, thus enhancing corrosion. The negative impact of lithium silicate precipitation on glass corrosion was noticed in previous Lithium silicate was a single crystalline phase in two glasses of our database. Their PCT solution pH was above 12, but the corresponding PCT Li releases were substantially smaller (6.6 and 8.8 g/m2) than B releases (12.5 and 20.4).3*'0 The R B k value for spine1 (Table I) is fairly outside the indeterminability interval only because the impact on Ni was ignored for the lack of information about the effect of Ni on PCT release. A recent studgo on glasses in a different composition region brought evidence that Ni may decrease glass-corrosion resistance to the extent at which spinel would meet Condition (4), resulting in an indeterminable PBk. This would agree with the general observation that spinel precipitation does not detectably affect glass dissolution? Gehlenite appeared in only one glass together with cristobalite and acmite. Its large PBk is subjected to a high uncertainty. For baddeleyite, hematite, calciumzirconium silicate, and zircon, at least one of their two PBk values listed in Table I is close to 1, suggesting that different secondary effects impacting the PCT mutually compensate. For cristobalite, the larger-than-one PBk may be associated with liquid-liquid phase separation that probably preceded cristobalite precipitation, leaving crystals embedded in a high-B glass subjected to tensile stress. The large PBk of eucryptite, which occurred in three glasses at low fractions (in two glasses as the single crystalline phase), indicates that eucryptite impacts glass corrosion more than its R B k suggests. Nepheline, another simple alkalialuminosilicate, shows a similar tendency, though in a milder form. Nepheline fractions were reported for a group of glasses by Li et al.: who measured c k by calibrating x-ray diffraction (XRD) with natural nepheline. Though measurements by Menkhaus et a1.16 showed a substantial excess of Si and Fe in nepheline, the corresponding &k was not significantly different from that of stoichiometric nepheline.'Oy''Hence, its PBk > 1 is unlikely to be caused by a deviation from the ideal stoichiometry, or an error in the measured hction in glass.
Environmental Issues and Waste Management Technologies VIII
295
CONCLUSIONS The effect of crystallinity on glass corrosion is rather complex. Although /?& coefficients provide some insight into this complexity, the present analysis also reveals serious limitations of the databases currently available. Progress can be achieved if quantitative XRD and scanning electron microscopy/energydispersive spectroscopy or electron microprobe analyses are systematically performed. Data on dissolution rates of individual crystalline phases and simulated residual glasses would M e r enhance understanding of the dissolution process of partially crystallized glasses. For short-term corrosion tests on crushed glass, a microscopic characterization of both pristine and corroded fractured surfaces is advisable. The PCT seem inadequate for assessing the corrosion resistance of partially crystallized glasses. The bulk glass is neither properly represented by fracture surfaces of the grains, nor is the bulk glass reached by the depths of corrosion. Only sufficiently long-term tests performed with glass coupons or monoliths would remove the bias caused by initial reactions on fractured surfaces. ACKNOWLEDGMENTS This study was h d e d by the U.S. Department of Energy’s Office of Science and Technology (through the Tanks Focus Area). Pacific Northwest National Laboratory is operated for the U.S. Department of Energy by Battelle under Contract DE-AC06-76RLO1830. The authors wish to thank Dong Kim for careful review of the paper and Wayne Cosby for assistance in editing. REFERENCES ‘U.S. Department of Energy, Ofice of Environmental Management, Waste Acceptance Product Specifications for Vitrified High-Level Waste Forms, DOE Document EM-WAPS Rev. 2, Washington, DC, 1996. 2American Society for Testing and Materials (ASTM), “Standard Test Methods for Determining Chemical Durability of Nuclear, Hazardous, and Mixed Waste Glasses: The Product Consistency Test (PCT),” C-1285-97. In 1998 Annual Book of ASTM Standards, Vol. 12.01, ASTM, West Conshohocken, Pennsylvania, 1998. 3P. b a yG.F. Piepel, M.J. Schweiger, D.E. Smith, D-S. Kim, P.E. Redgate, J.D. Vienna, C.A. LoPresti, D.B. Simpson, D.K. Peeler and M.H. Langowski, Property/Composition Relationships for Hanford High-Level Wmte Glasses MeZting at 115OoC, PNL-10359, Vol. 1 and 2, Pacific Northwest Laboratory, Richland, Washington, 1994.
296
Environmental Issues and Waste Management Technologies VIII
4
P. h a , G.F. Piepel, J.D. Vienna, P.E. Redgate, M.J. Schweiger, and D.E. Smith, “Prediction of Nuclear Waste Glass Dissolution as a Function of Composition,” Ceramic Transactions,61, 497-504 (1995). ’P. Hrma, G.F. Piepel, J.D. Vienna, S.K.Cooley, D-S. Kim, and R.L. Russell, Database and Interim Glass Property Modelsfor Hanford HL W Glasses, PNNL13573, Pacific Northwest National Laboratory, Richland, Washington, 2001. 6D-S. Kim, D.K. Peeler, and P. Hrma, “Effects of Crystallization on the Chemical Durability of Nuclear Waste Glasses,” Ceram. Trans., 61, 177-185 (1 995). 7 P. Hrma and A.W. Bailey, “High Level Waste at Hanford: Potential for Waste Loading Maximization,” Proc. 1995 Int. Con$ Nucl. Waste Manag. and Environ. Remediation (ICEMt95),1,447-451 (1995). 8A.W. Bailey and P. Hrma, “Waste Loading Maximization for Vitrified Hanford HLW Blend,” Ceram. Trans.,61, 549-556 (1995). ’H.Li, J.D. Vienua, P. Hrma, D.E. Smith, and M.J. Schweiger, “Nepheline Precipitation in High-Level Waste Glasses - Compositional Effects and Impact on the Waste Form Acceptability,” Mater. Res. Soc. Proc.,465,26 1-268 (1997). ‘%.J. Riley, J.A. Rosario and P. Hrma, Impact of HL W Glass Crystallinity on the PCT Response, PNNL- 13491 , Pacific Northwest National Laboratory, Richland, Washington, 200 1. “B.J. Riley, P. Hrma, J.A. Rosario, and J.D. Vienna, “Effect of Crystallization on High-Level Waste Glass Corrosion,” Ceram. Trans. (in press). 12R.E. Edwards, SGM Run 8 - Canister and Glass Temperature During Filling and Cooldown, DPST-87-801, Savannah River Laboratory, Aiken, South Carolina, 1987. 13L. Lee, Thermal Analysis of DWPF Canister During Pouring and Cooldown, DPST-89-269-TLY Savannah River Laboratory, Aiken, South Carolina, 1989. 14D.G. Casler and P. Hrma, ‘Wonisothermal Kinetics of Spinel Crystallization in a HLW Glass,” ScientiJicBasis for Nuclear Waste Management xx71 (Editors D.J. Wronkiewicz and J.H. Lee), Materials Research Society, 556, 255-262 (1999). ”C.M. Jantzen and D.F. Bickford, “Leaching of Devitrified Glass Containing Simulated Nuclear Waste,” Mater. Res. Soc. Proc., 44, 135-146 (1985). ‘6T.J. Menkhaus, P. Hrma, and H. Li, “Kinetics of Nepheline Crystallization fiom High-Level Waste Glass,” Ceram. Trans.,107,461-468 (2000). 17M. Mika, M.J. Schweiger, and P. Hrma, “Liquidus Temperature of Spinel Precipitating High-Level Waste Glass,” Scientific Basis for Nuclear Waste Management XY (Editors W.J. Gray and I.R.Triay), Materials Research Society, 465,71-78 (1997).
Environmental Issues and Waste Management Technologies VIII
297
'*T.J. Plaisted, F. MO, B.K. Wilson, C. Young, and P. h a , "Surface Crystallization and Composition of Spine1 and Acmite in High-Level Waste Glass," Ceram. Trans.,119,317-325 (2001). "C.A. Cicero, S.L. Mara, and M.K. Andrews, Phase Stability Deteminations of D W F Waste Glasses (U), WSRC-TR-93-227, Westinghouse Savannah River Corn any, Aiken, South Carolina, 1993. 'G.A. Scholes, D.K. Peeler, and J.D. Vienna, The Preparation and Characterization of INTEC Phase 3 Composition Variation Study Glasses, INEELEXT-2000-01566, Idaho National Engineering and Environmental Laboratory, Idaho Falls, Idaho, 2000.
298
Environmental Issues and Waste Management Technologies VIII
Ceramic and Alternative Waste Forms
This page intentionally left blank
DEVELOPMENT OF TITANATE CERAMIC WASTEFORMS AND CRYSTAL CHEMISTRY OF INCORPORATED URANIUM AND PLUTONIUM Eric R. Vance, Materials Division, Australian Nuclear Science and Technology Organisation, New Illawarra Road, Menai, NSW 2234, Australia ABSTRACT The development of titanate ceramics for incorporation of high-level radioactive wastes is first reviewed, tracing the evolution of synroc-C designed for Purex-type waste through glass-ceramics for tank wastes and pyrochlorerich ceramics for surplus impure plutonium. The remainder of the paper deals with scientific aspects of the waste ceramics with focus on actinides. The valences of actinides, notably U and Pu, incorporated in these ceramics is shown to depend on crystal field effects, and the presence of charge compensators, as well as the prevailing redox conditions during equilibration of the ceramics at subsolidus temperatures. Experimental techniques of study include X-ray near-edge absorption, diffuse reflection and X-ray photoelectron spectroscopy, and electron microscopy. The occurrence of U4+and U5+in zirconolite, brannerite and pyrochlore, as well as the multi-phase titanate ceramics themselves, is detailed. Both Pu3+and Pu4+can be accommodated in zirconolite, pyrochlore and perovskite. Other work on U and Pu in rare-earth silicate apatites is summarised.
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management TechnologiesVIII
301
EARLY DEVELOPMENT OF CERAMICS FOR HIGH-LEVEL WASTE IMMOBILISATION Borosilicate glasses for the immobilisation of high level wastes from nuclear fuel reprocessing were developed by the US Atomic Energy Commission in the 1950s and were scaled up in the late 1960s. However Pennsylvania State University workers [1,2] in the mid- 1970s devised ceramics for this purpose, based on crystalline silicates, phosphates and molybdates. These so-called supercalcine ceramics were sintered in air at -1 100°C and had very high loadings of fission products, typically 70 wt%, and the chemistry of the different phases was driven by the fission products as majority components. Typical phases were pollucite, CsAlSiO,; powellite, CaMoO,; and rare earth apatites and phosphates (e.g. monazite, REPO,, where RE = trivalent rare earth). All of these had mineral analogues which were known to be very durable in the hot, wet conditions likely to characterise a deep geological repository for the waste. Following work at Sandia on phase assemblages occurring on heating sol-gel titania particles on which HLW fission products and actinides were sorbed, Ringwood and his co-workers in the late 1970s devised a multi-phase titanate ceramic [3] in which nearly all the fission products and actinides in HLW from nuclear fuel reprocessing were incorporated substitutionally in the various phases. These theoretically-dense materials were made by first mixing inactive precursors with liquid (simulated) HLW, drying and calcining in a H2/N2atmosphere for l h at 750°C. The calcine was then mixed with 2 wt% of powdered Ti metal for redox control and then subjected to uniaxial graphite die hot-pressing or hot isostatic pressing at 1200°C. The precursor composition is (wt% oxide): A1203 (5.4); BaO (5.6); CaO (11.0); Ti02 (71.4); Zr02 (6.6). Since 1984, rather than using oxides, a slurry mixture of Ba and Ca hydroxides and transesterified Al, Ti and Zr alkoxides has been used as the precursor [4]. This provides better solid-state reactivity. The principal advantage of this synroc-C ceramic was that the waste ions were dilutely incorporated in durable titanate mineral phases which were considerably more insoluble in water than the silicates and phosphates etc. The waste loading could be varied between zero and 35 wt% using the same inert additive chemistry without substantially changing the basic zirconolite + perovskite + hollandite + rutile phase assemblage, although the percentages of the different phases varied somewhat. There were minor alumina-rich phases,
-
302
Environmental Issues and Waste Management TechnologiesVIII
plus minor metallic phases arising from the 3d and Pd metal group elements which were reduced to metals under the reducing conditions prevailing during hot-pressing. Table 1 shows the phase constitution of Synroc-C, containing 20 wt% HLW, and which radionuclides are incorporated in the various mineralanalogue phases present. Table 1. Composition and mineralogy of Synroc-C Phase wt% 30 Hollandite, Ba(A1,Ti)2Ti6016 Zirconolite, CaZrTi20, 30 Perovskite, CaTiOs 20 Ti oxides 10 Alloy phases 5 * RE, An = rare earths and actinides respectively
Radionuclides in lattice Cs, Rb RE, An* Sr, RE, An Tc, Pd, Rh, Ru etc.
DEVELOPMENT OF SYNROC DERIVATIVES In the early 1990s, the synroc ceramics were refocused towards the study of zirconolite-rich materials for immobilisation of actinide rich wastes such as Pu or partitioned transuranic elements. The initial work during 1991-4 was directed at the latter application in conjunction with the Japanese Atomic Energy Research Institute. There was a strong focus on radiation damage via the incorporation of the alpha-emitter 244Cm(18 year half-life), as had been done with synroc-C and a Na-doped variant thereof [5,6]. Perovskite was also studied for comparison. The work on surplus Pu immobilisation, with Lawrence Livermore National Laboratory (LLNL) as the lead laboratory for the US Department of Energy (DOE), moved from zirconolite- to pyrochlore-rich ceramics during 1994-7. This was because of solid solution limits in the f i s t instance when the target of the work changed from immobilisation of 10 wt% Pu alone to the additional inclusion of 20 wt% U. Moreover it was realised that anisotropic swelling effects could lead to microcracking problems after self-damage due to the alpha decay flux if a non-cubic matrix was employed. The estimated time for amorphisation to be complete is in the order of 1000 years and the resultant
Environmental Issues and Waste Management TechnologiesVIII
303
volume expansion would be around 6%. In addition these ceramics incorporated an atom each of neutron-absorbingGd and Hf for each atom of Pu to deal with potential criticality problems. These problems were not confined to the sample-near-field aggregation of Pu due to leaching was shown to be not a problem, because the leach rates of Pu were spanned by those of the neutron absorbers. The final baseline (no impurities) version [7] of the pyrochlore-rich ceramics chosen by the US DOE in 1998 contained 95 wt% of a pyrochlorestructured Ca,,,,Gdo~2,Hfo,,Uo.,Puo~22Ti207 phase plus 5 wt% of rutile-structured Tio.,Hfo.102. The form of the ceramic was to be 76 mm diameter pellets weighing -5OOg. 560 such pellets were to be enclosed in a US standard canister of Savannah River DWPF glass to provide a radioactive barrier (gamma field) to diversion. The scientific work underlying this choice principally dealt with binary zirconolite/pyrochlorephase diagrams involving rare earths and actinides, and the effects of impurity additions on these diagrams [8]. The principal phase diagrams studied have been CaAn,Zr(,-,,Ti,O, (An = tetravalent U, Np, Pu) and Ca(,-,,Zr,,~,,Gd2xTi207. Studies were made by sintering in air or argon at temperatures of 1400-1500°C. As the actinide or Gd content increases, the zirconolite endmember first transforms at x 0.2 from the 2M polytype to the 4M polytype, characterised by x 0.4. As the addition is taken further, the amount of 4M polytype increases, forming a pure phase at around x = 0.4. Further addition produces pyrochlore having x 0.7, with pure pyrochlore beyond x 0.7. Since the surplus plutonium at which this work was directed contains a variety of impurities, the effects of all these impurities needed to be detailed. To try to simplify this task, the impurities were categorised by valence and broadly speaking, this was found to be appropriate. However where gross ionic size differences were present this approach was an oversimplification. Moreover, the valences themselves of the transition metal impurities were not always clear because the valences depend on the crystal chemistry of the host material and not simply the redox of the firing atmosphere (see below). But the broad thrust of the results was that impurity cation valences of +1 - +4 promoted zirconolite formation at the expense of pyrochlore and cation valences of +5 and +6 promoted pyrochlore formation [8,9].
-
-
304
-
-
Environmental Issues and Waste Management TechnologiesVIII
In equilibrated materials made by the alkoxide route [8], there were no unreacted actinide oxides or brannerite (UTi,O,). However in materials made by the reference oxide route, small amounts of oxides (- 1 wt%) and considerable amounts (- 10 wt% ) of brannerite were present. Little was known about the detailed crystal chemistry and aqueous durability of brannerite except that natural samples are almost always metamict due to many millions of years of self-irradiation damage and U extraction can be carried out under highly acidic conditions. Hence we studied the crystal chemistry and aqueous dissolution behaviour of actinides incorporated in the brannerite structure over the 1997-2001period. Radiation damage phenomena were also studied. As found in natural samples to a lesser degree, alkaline earths, rare earths and actinides could all be substituted for U, and the solid solution limits and charge balance mechanisms could be understood [101. Both tetravalent and pentavalent U were revealed from electronic and X-ray absorption or photoelectron spectroscopy (see below). 3d group metal ions, Nb and A1 could be substituted in the Ti sites. Radiation damage phenomena were studied by the use of MeV heavy ions, followed by TEM studies, to deduce the amounts of atomic displacements necessary to cause amorphisation [113. The dependence of the aqueous dissolution behaviour over the 2-13 range of pH was measured and some rate constants established [12]. However there is a need to study the Eh dependence of the dissolution behaviour. Between 1994 and 1997 a parallel effort was developed to study glassceramics for immobilisation of Hanford tank wastes. The object of this work was to contain as much as possible of the actinides in synroc phases, principally zirconolite, in a boroaluminate silicate glass matrix. Other crystalline phases such as nepheline, zirconia and CaF, were also present [ 131. These glassceramics had waste loadings of 50-70 wt% and leach rates were often 10-100 lower than those for standard EA glass. ACTINIDE VALENCES IN TITANATE PHASES The roles of charge compensators and crystal field stabilisation of actinide valences in addition to redox conditions in the firing atmosphere were studied at ANSTO from about 1992 onwards. An oft-quoted example of crystalchemical stabilisation is that in an air atmosphere at 12Oo"C, the stable oxide of Ce is the dioxide (Ce4') whereas the stable phosphate is monazite (CePO,;
-
Environmental Issues and Waste Management TechnologiesVIII
305
Ce3'). It was assumed at the outset that the valence assumed by the actinides in perovskite and zirconolite would be strongly dominated by the presence or otherwise of charge compensators. For example Pu can in principle be substituted in the Ca site in the trivalent or tetravalent form if an equimolar amount of A1 or Mg respectively is substituted on the Ti site; the overall zirconolite stoichiometries would be: Ca(l.x)Pu,ZrTi(2-,,( AVMg),07. Another method of promoting the formation of Pu4' substituting in the Ca site of zirconolite is to substitute twice the molar amount of Pu as A1 in the Ti site, These kinds of substitutions with a stoichiometry of C~,,~,QU,Z~T~,~~,,,A~,,O,. were then examined in different firing atmospheres. The expectation was that the maximisation of lattice entropy would be a strong driver for a ceramic to adopt a single phase, and that the formation of cation vacancies would have a very high cost in lattice energy, as compared to changes in electronic configuration. However X-ray absorption studies during 1993-8 showed that often when a single-phase resulted, expectations [ 141 about the actinide valences were not met in several cases, although the presence of charge compensators in the lattices was significant [15-18]. It was deduced that cation vacancies could indeed be present. Positron annihilation lifetime spectroscopy verified the presence of such cation vacancies in Ce-doped zirconolites in which the Ce was serving as an inactive simulant of Pu [19]. It was also observed that the redox behaviour for Pu incorporated in zirconolite was similar to that for perovskite, but this was not so for Np [16]. Current results and ongoing research on the titanate phases zirconolite, perovskite and brannerite for (a) tetravalent and pentavalent U; and (b) trivalent and tetravalent Pu are now summarised. The U valence in the pyrochlore phase of pyrochlore-rich ceramics is also touched upon. Other preliminary data on rare-earth silicate apatites are also given. The experimental techniques include X-ray absorption near-edge spectroscopy (XANES), X-ray photoelectron spectroscopy (XPS), and diffuse reflectance spectroscopy (DRS). Many reviews have been given on these topics. In X-ray absorption spectroscopy, the position of the absorption edge can me measured to fractions of an electron volt using synchrotron sources. The exact energy of a given edge depends on the electronic configuration of the absorbing element, so by the use of suitable valence standards, the valence of the element in the sample of interest can be determined. Similar considerations
306
Environmental Issues and Waste Management TechnologiesVIII
apply to XPS where photoelectrons are generated by low-energy monochromatic X-ray sources. DRS examines the electronic absorptions between the excited states and ground states of paramagnetic ions, and these are observed as peaks in the near ultraviolet, visible and infrared spectral regions. The standard form of zirconolite is CaZrTi,O, and it is stable in air and in lower oxygen partial pressures, down to -104 atmospheres [20]. An obvious way to insert U or Pu is as the tetravalent species substituting for Zr. This can be done using Ar or air atmospheres for U and Pu respectively. Difficulties are experienced for substitutions of more than about 0.7 f.u. of the actinide but whether this is for reasons of kinetics or an intrinsic lack of stability of the endmember Ca(U/Pu)Ti,O, pyrochlore structure is still not entirely clear [21]. A variety of other substitutions can however be made. Tetravalent U or €31can be substituted for Ca via charge compensating A1 or Mg substitutions for Ti. Pentavalent U can be substituted for Zr, again if A1 is substituted for Ti as a charge compensator. When Ca(U/Zr)Ti,O, samples prepared in argon are subsequently oxidised, U5+is formed but the changes to the phase assemblage can be complicated at U contents of 0.4 f.u. It is doubtful if much U5+could be substituted for Ca because of the large ionic size difference. U was observed both in +4 and +5 states by DRS in baseline pyrochlorerich ceramics in which the Pu had been replaced by Ce on a molar basis [22]. These ceramics were fired in different atmospheres and contained different amounts of impurities. Pu was nominally substituted in the Ca site of zirconolite without any explicit charge compensation (C~.9Puo.,HfTi,07)[ 161. The replacement of Zr by Hf in zirconolite has no significant effect as the ionic sizes of Hf and Zr are very similar and Hf was used instead of Zr to facilitate XANES measurements). It was deduced from the overall stoichiometry of the major zirconolite, determined by electron microscopy, to occupy both the Ca and Zr sites upon firing in air. A minor pyrochlore-structuredphase, rich in Pu,was also found in conjunction with minor Pu-free phases. The Pu was found from XANES to be tetravalent. Firing in a reducing atmosphere at 1200°C preserved the phase assemblage and 80% of the Pu was found to be reduced to Pu3+by the reducing heat-treatment. In other work [15], Pu was substituted on the Ca site with enough A1 in the Ti site to promote either Pu3+or Pub (A1 being once or twice respectively the equimolar amount of Pu, in the Ti site). Again however the Pu
-
Environmental Issues and Waste Management TechnologiesVIII
307
was all tetravalent in air firing atmospheres (1400°C) in both cases, but could be reduced to Pu3+in H2/N2at 1300 or 1400°C. U in perovskite is still under study, and only the U4+valence state has been positively identified in as yet unpublished work. Pu was notionally substituted[l6] into the Ca site of perovskite with no charge compensators or However in all cases, Pu in the enough AI in the Ti site to favor Pu3+or h4+. perovskite was found to be tetravalent when fired in air at 1500°C and trivalent when subsequently fired at 1200°Cin H,/N,. This was an example of the Pu valence appearing to be dominated by the atmosphere, much like the example of zirconolite given above. As an aside, the same experiments were run with Np instead of Pu and it was found that all Np in the zirconolite samples was tetravalent, whereas trivalent Np could be obtained under reducing conditions when either no A1 or an equimolar amount of A1 to that of Np was present in the Ti site. However when there was twice as much A1 as Np, the Np was tetravalent under the reducing conditions. Ce also behaved quite differently in perovskite and zirconolite [23]. Pure brannerite can only be formed in reducing conditions with U being tetravalent, but the addition of Ca or rare earths allows the formation of U5' upon firing in either air or argon 1101. A Pu4' version of brannerite could also be made[lOJ. Finally, we have studied U and Pu in rare-earth silicate apatites of Ca,RE8(Si04)602stoichiometry (unpublished work). The solid solubility limits of U4+and Pu4+in the Ca/RE sites in materials formed in argon and air firing atmospheres at -1350°C were quite limited (- 0.4 f.u.), but 8 f.u. of Pu3+could be substituted for the RE when a reducing atmosphere was employed.
-
CONCLUSIONS The development of titanate ceramics since 1978 has been briefly reviewed. The interplay between crystal chemical forces and redox conditions in the firing atmosphere on the valence assumed by U and Pu have been outlined in the actinide-bearing synroc phases, zirconolite, pyrochlore, perovskite and brannerite. ACKNOWLEDGMENTS
308
Environmental Issues and Waste Management TechnologiesVIII
I wish to thank a large number of colleagues at ANSTO for collaboration, particularly B. Begg, A. Day, M. Carter, K. Finnie and 2.Zhang. Much of the work was sponsored in part through the USDOE via the LLNL and the Environmental Management Science Program. REFERENCES ‘G.J. McCarthy, “High-Level Waste Ceramics:Materials Considerations and Product Characterization”,Nuclear Technology, 32,92- 104 (1977). 2G. J. McCarthy, W. B. White and D. E. Pfoertsch, Monazite, Materials Research Bulletin, 13, 1239-45 (1978). 3A. E. Ringwood, S. E. Kesson, N. G. Ware, W. Hibberson and A. Major, “Geological immobilisation of nuclear reactor wastes”, Nature, 278, 2 19-23 (1979). 4A. E. Ringwood, S. E. Kesson, K. D. Reeve, D. M. Levins and E. J. Ramm, “Synroc”; pp. 223-334 in Radioactive Waste Forms for the Future, Edited by W. Lutze and R. C. Ewing. Elsevier, Amsterdam, 1988. ’H. Mitamura, S. Matsumoto, K.P. Hart, T. Miyazaki, E.R. Vance, Y. Tamura, S. Togashi and T.J. White, “Aging Effects on Curium-doped Titanate Ceramic Containing Sodium-bearing High-Level Nuclear Waste”, Journal of the American Ceramic Society, 75,392-400 (1992). 6H.Mitamura, S. Matsumoto, M.W.A. Stewart, T. Tsuboi, M. Hashimoto, E.R. Vance, K.P. Hart, Y. Togashi, H. Kanazawa, C.J. Ball and T.J. White, ”Alpha-Decay Damage Effects in Curium-Doped Titanate Ceramic Containing Sodium-Free High-Level Nuclear Waste”, Journal of the American Ceramic Society, 77,2255-64 (1994). 7 B. B. Ebbinghaus, R. A. VanKonynenburg, F. J. Ryerson, E. R. Vance, M. W. A. Stewart, A. Jostsons, J. S . Allender, T. Rankin and J. Congdon, “Ceramic Formulation for the Immobilization of Plutonium”, Waste Management ’98 (CD-ROM; sess65/65-04), WM Symposia Inc., Tucson, AZ,1998. 8M. W. Stewart, E. R. Vance, A. Jostsons, P. A. Walls, K. P. Hart, S. Moricca, R. A. Day, G . R. Lumpkin, C. J. Ball and D. S. Perera, Australian Nuclear Science and Technology Organisation Report R00mO3 1 to Lawrence Livermore National Laboratory, Nov. 2000 9M.W. A. Stewart, E. R. Vance, R. A. Day, S. Leung. A. Brownscombe and M. L. Carter, “Impurity incorporation in Pyrochlore-rich ceramics”; pp.
Environmental Issues and Waste Management Technologies VIII
309
569-76 in Ceramic Transactions (Environmental Issues and Waste Management Technologies V, Edited by G. T. Chandler and X. Feng. American Ceramic Society, Westerville, OH, USA, 2000. ”‘E. R. Vance, J. N. Watson, M. L. Carter, R. A. Day and B. D. Begg, “Crystal Chemistry and Air Stabilization in Air of Brannerite, UTi,O;’, Journal of the American Ceramic Society, 84, 141-4 (2001). “G. R. Lumpkin, K. L. Smith and M. G. Blackford, “Heavy Ion irradiation studies of colurnbite, brannerite, and pyrochlore structure types”, Journal of Nuclear Materials, 289, 177-87 (2001). ‘*Y.Zhang, K. P. Hart,W. L. Bourcier, R. A. Day, M. Colella, B. Thomas, Z. Aly and A. Jostsons, “Kinetics of Uranium Release from Synroc Phases”, Journal of Nuclear Materials, 289,254-62 (2001). 13E.R. Vance, K. P. Hart, R. A. Day, M. L. Carter, M. G. Blackford and B. D. Begg, “Synroc Derivatives for the Hanford Waste Remediation task”; pp. 341-8 in Scientific Basis for Nuclear Waste Management X X , Edited by W. J. Gray and I.R. Triay, Materials Research Society, Warrendale, PA, USA, 1997. 14E.R. Vance, C. J. Ball, K. L. Smith, M. G. Blackford, B. D. Begg and P. J. Angel, “Actinide and rare earth incorporation into zirconolite”, Journal of Alloys and Compounds, 2134,406-9 (1994). ”B. D.Begg, E. R. Vance, R. A. Day, M. Hambley and S . D. Conradson, “Plutonium and Neptunium Incorporation in Zirconolite”; pp. 325-32 in Scientific Basis for Nuclear Waste Management XX, Edited by W. J. Gray and I.R. Triay. Materials Research Society, Warrendale, PA, 1997. 16 B. D. Begg, E. R. Vance, S. D. Conradson, “The Incorporation of Plutonium and Neptunium in Zirconolite and Perovskite”, Journal of Alloys and Compounds, 271-3, 221-6 (1998). 17E. R. Vance, B. D.Begg, R. A. Day and C. J. Ball, “Zirconolite-rich Ceramics for Actinide Wastes”, pp. 767-74 in Scientific Basis for Nuclear Waste Management XVZZZ ,Edited by T. Murakami and R. C. Ewing, Materials Research Society, Pittsburgh, PA, USA, 1995. ‘*B.D. Begg, R. A. Day and A. J. Brownscornbe, “Structural Effect of Pu Substitutions on the Zr-site in Zirconolite”; pp. 259-66 in Scientific Basis for Nuclear Waste Management XXZV, Edited by K. P. Hart and G. R. Lumpkin, Materials Research Society, Warrendale, PA, USA, 2001.
310
Environmental Issues and Waste Management TechnologiesVIII
19J. H. Hadley, F. H. Hsu, Yong Hu, E. R. Vance and B. D. Begg, “Observation of Vacancies by Positron Trapping in Ce-doped Zirconolites”, Journal of the American Ceramic Society, 82,203-5 (1999). 2oB.D. Begg, E. R. Vance, B. A. Hunter and J. V. Hanna, “The structural effects of reduction on Zirconolite”, Journal of Materials Research, 13,3181-90 (1998). 21E.R. Vance, G. R. Lumpkin, M. L. Carter, D. J. Cassidy, C. J. Ball, R. A. Day and B.D. Begg, “The Incorporation of U in Zirconolite (CaZrTi,O,)”, Journal of the American Ceramic Society, accepted for publication. 22M.W.A. Stewart, E. R. Vance, A. Jostsons, K. Finnie, R. A. Day, andB. B. Ebbinghaus, “Atmosphere Processing Effects on Titanate Ceramics designed for Plutonium Disposition”, accepted for publication in Scient$c Basisfor Nuclear Waste Management XXV, Edited by P. McGrail and G. Cragnolino. Materials Research Society, Warrendale, PA, USA, 2002. 23B.D.Begg, E. R. Vance and G . R. Lumpkin, “Charge compensation and the Incorporation of Cerium in Zirconolite and Perovskite”, pp. 79-86 in Scientific Basis for Nuclear Waste Management XXZ, Edited by I. G. McKinley and C. McCombie. Materials Research Society, Warrendale, PA, USA, 1998.
Environmental Issues and Waste Management TechnologiesVIII
31 1
This page intentionally left blank
SUBSTITUTION OF Zr, Mg, Al, Fe, Mn, CO AND Ni in ZIRCONOLITE, CaZrTi,O,. Eric R. Vance, John V. Hanna, Brett A. Hunter, Bruce D. Begg, Dan S. Perera, Huijin Li and Zhao-ming Zhang, Australian Nuclear Science and Technology Organisation, New Illawarra Road, Menai, NSW 2234, Australia ABSTRACT From X-ray fluorescence analysis via scanning electron microscopy (SEM), there was no evidence of Zr occupation of the Ca site of zirconolite (CaZrTi,O,). Mg can inhabit both the Ca and Ti sites as shown by solid-state magic-angle spinning and high-field static nuclear magnetic resonance measurements. Microstructural work indicated that approximately 0.1 formula units of divalent or trivalent Fe, Mn, CO and Ni can also inhabit Ca sites. From SEM, A1 substituted into zirconolite with appropriate charge compensators being present can inhabit Ti sites but not the Ca sites. A Rietveld analysis of the powder neutron diffraction patterns showed that A1 enters the fivefold Ti site preferentially to the sixfold Ti sites. INTRODUCTION In the original synroc-type ceramics [ 11 for Purex-type waste, zirconolite, CaZrTi,O,, incorporates mainly rare earth fission products and waste actinides. However Purex-type wastes often contain process chemicals, so synroc variants or glass-ceramic derivatives have been devised over the last 10 years with the aim to incorporate more and more of the Periodic Table of elements in the titanate phases. Here we have focused on ionic substitutions in zirconolite. When "Sr in the Ca site decays to "Zr it is of interest to see whether the Zr can inhabit the same site. The chemical design of the experimental samples to investigate this possibility was Ca,,ZrO,,OO,,ZrTi,O,,where 0 = Ca site vacancy. Mg, A1 and 3d group elements often accompany fission product wastes. In synroc-C [ 11, 3d corrosion products formed metal alloys. However, oxidising or neutral conditions may be desirable to process synroc derivatives, so the 3d elements may assume different valences. Although Mg, A1 and divalent 3d metal ions will substitute in the Ti sites of zirconolite [2], an object of the present work was to see whether incorporation of such ions in the Ca site was possible. The To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
313
chemical design for divalent ions to enter the Ca site was C~.8M2+o,2zrTi,07 stoichiometry. For trivalent ions, the stoichiometry used was Cao,,M3'o,2ZrTi,,A10~207 in which the substitution of A1 in the Ti sites would provide charge compensation for the trivalent ions in the Ca site. Finally, it was of interest to see whether A1 can substitute in the split 5-fold Ti site as well as the octahedral Ti sites. EXPERIMENTAL The samples were all made by the alkoxidehitrate route in which ethanolic solutions of Ti isopropoxide were mixed with aqueous nitrate solutions in the desired stoichiometric proportions. The mixtures were stir-dried, calcined in air at 750°C and cold-pressed and sintered in air or argon at 1400-1450°C. 25Mgand 27Alsolid state NMR spectra were acquired at a magnetic field strength of 14.1 T (36.72 MHz for "Mg, 156.34 MHz for 27Al)on a Chemagnetics CMX 600 spectrometer. The "Mg experiments were performed on static samples using a conventional n/2-z-n-z solid echo pulse sequence with an extended phase cycle. The 27Alexperiments employed magic-angle-spinning techniques where the sample was rotated at frequencies of -25 kHz, in conjunction with single pulse and acquire experiments. The 25Mgand 27Alchemical shifts were referenced to MgO(s) and 1M [A1(H20),l3'(aq), respectively. Neutron powder diffraction patterns were obtained using the high-resolution powder diffractometer on the HIFAR reactor at ANSTO. A neutron wavelength of 0.14928 nm was used with a 0.05" step size. The powder sample was contained in a thin-walled vanadium can that was slowly rotated during the measurement. Structural refinements were performed using the Rietveld program LHPM [3]. Scanning electron microscopy (SEM) was carried out with a JEOL 6400 instrument operated at 15 keV, and fitted with a Tracor Northern Voyager IV Xray microanalysis System (EDS). A comprehensive set of standards was used for quantitative work, giving a high degree of accuracy [4], approximately 0.02 formula units (f.u.). Powder X-ray diffraction analyses were conducted with a Siemens D500 diffractometer using COK a radiation. RESULTS AND DISCUSSION Zr No evidence of Zr entry into the Ca site was observed in the samples having Zr notionally substituted on the Ca site, with charge compensating Ca vacancies, insofar as only very slightly Ca-deficient zirconolite and slightly Ti-rich ZrTiO, were observed (see Fig 2(a)). Within experimental error of 0.02 f.u., the chemistry could be described by: Cao,,Zr,,,Ti207 0.85 Ca0.,,ZrTi2O7+ 0.28 Zr o,9Ti,.,04.The apparent failure to incorporate Zr on the Ca site in this instance may however be due to the energetic
-
314
Environmental Issues and Waste Management Technologies VIII
cost of incorporating charge compensating Ca vacancies as Rietveld refinements have previously shown that at least 0.15 f.u. of Zr can be incorporated in the Ca site of a zirconolite that displayed Cao~85Zrl~16Til~671no~3307 stoichiometry [5]. Mg
Samples of Ca,-,Mg,ZrTi207 or Ca,-,,Y 2xZrTi2-,Mg,07stoichiometry were made with Mg targeted towards the Ca or Ti sites, respectively. The sintered samples consisted of zirconolite plus other phases at high x values. SEM of a preparation with x = 0.2 indicated that up to 0.15 f.u. of Mg could inhabit the Ca site. 25Mgbroadline (static) NMR showed that complete occupancy of 0.15 f.u. of Mg in the Ca site was possible, as shown by the spectrum in Fig l(a). This spectrum is dominated by a narrow 25Mgresonance at -0 ppm thus indicating a symmetrical environment is experienced in this Ca position. But when a similar amount of Mg was targeted towards the Ti site only partial occupancy, with clear evidence of some Mg occupation of the Ca site, was observed from Fig. l(b). This spectrum shows a broad featureless resonance at -80 kHz downfield from its narrower counterpart which is assigned to Mg in the Ti position; some Mg must also be occupying the Ca position in this sample as evidenced by a much less intense 0 ppm resonance being observed. However, the partial occupation of Mg in the Ca site could be explained by some of the Y also inhabiting the Zr site, which would lead to spillover of Mg into the Ca site. Although &fold co-ordinated Mg is considerably smaller in ionic radius than Ca (0.103 vs. 0.126 nm[6]), it was interesting that its solid solubility in the Ca site of zirconolite was much larger than the value of < 0.02 f.u. found in the present work in perovskite-rich samples of Cao.8Mgo.2Ti03stoichiometry heated at 15OO0C/4days (see also [7]). The discrepancy probably lies in the 12-fold coordination of the alkaline earth ion in the ideal perovskite structure. Fe, Mn, COand Ni Divalent targets. The Fe, CO and Mn samples were fabricated in an Ar atmosphere to encourage these ions to form as divalent species. 2M zirconolite plus some non-zirconolite phases were present from XRD. SEM (Figs. 2(b) to 2(d)) gave the zirconolite compositions as CaO.,1Feo.14Zr0~90Ti2~0807, Cao.90Coo~08Zro.93Ti2~0807 and Cao,82Mno~,8Zro.,,Ti2.06Alo~0307. The A1 contaminant in the Mn-bearing sample will strongly prefer the Ti site (see below). Because the Ca + 3d group ion always added to approximately one f.u., the 3d group element should be incorporated in the Ca site. However we would not rule out the occupation in the Ti sites of a small fraction of the Fe, in the Fe-bearing sample (Ca + Fe occupancy + 1.05 f.u.). From charge balance, the 3d group ion is divalent. In each case, titanate phases containing significant amounts of 3d group ions were observed, e.g. FeTi205, a phase of approximately CoTi40,
Environmental Issues and Waste Management Technologies VIII
315
stoichiometry, and a phase of approximate CaMnTi,O, stoichiometry . These results indicated that the solid solution limits of the 3d group ions in the Ca site had been exceeded in each case. Ni-bearing samples were fabricated in air to promote Ni2' formation. In a sample of Cao~8Nio~2ZrTi207 stoichiometry, the zirconolite composition from SEM was Cao~83Nio~13Zrl,,Til~,,07. Additional Ti-rich ZrTiO, and rutile were observed, with the excess Ni forming NiTiO, (see Fig 2(e)). Trivalent targets. The Fe, CO and Mn-bearing samples were fired in air to encourage the formation of the trivalent 3d group species. All samples showed evidence of phases additional to major zirconolite by XRD. The zirconolite stoichio me trie s ob served by sEM were : ca,,,Fe,~ 14Zro~99Til~,2A10~1307, Ca0.87C00.09Zr1.05Ti1.88A10.~207 and Ca0.81Mn0.13Zr1.02Ti1.81A10.1307~ The occupancies of the Ca site by the 3d group ions were approximately the same as those above for the targeted divalent 3d group ions. However the occupation in the Ti sites of approximately the same molar amount of trivalent A1 (see also below) as the 3d group ion indicates that the 3d group ions are trivalent. Again it was clear that the solid solubilities of the trivalent 3d metals was < 0.2 f.u. The solid solution data for Mg, Fe, COand Ni in the Ca site of zirconolite are given in Table I. Table I. Limits of solid solubility of Mg, Fe, CO and Ni in the Ca site of zirconolite Ion CO2' co3+ Mg
I
Solubility (f.u.) 0.08 0.09 0.15
1
Ion Fe2' Fe3' Ni2+
I
Solubility (f.u.) 0.14 0.14 0.13
I
Ion Mn2+ Mn3'
1
Solubility (f.u.) 0.15* 0.13
I
In %fold co-ordination, the ionic radius [6] of Ca is 0.126 nm whereas those of Fe, CO,Mn (like Mg) are considerably smaller - 0.106,0.104 and 0.1 10 nm (highspin) respectively. The corresponding size of trivalent Fe is even less (0.092 nm) so it is surprising that the solubilities of divalent and trivalent Fe appear to be similar. While Ref 6 does not give %fold ionic sizes for Co3+and Mn3', the sizes of these trivalent ions in 6-fold co-ordination are similar, so the 8-fold sizes would also be expected to be similar, so explaining the broad similarities in deduced solid solubilities of trivalent Fe,Co and Mn. In 6-fold co-ordination the sizes are Mg and divalent Fe, CO,Mn and Ni (assumed to be high-spin) lie in the band of
316
Environmental Issues and Waste Management Technologies VIII
Environmental Issues and Waste Management Technologies VIII
317
0.083-0.097 nm and are significantly larger than Ti (0.0745 nm), with the discrepancies implying fairly limited solid solution behaviour. A1 Samples with stoichiometries of Cao.95Al,,ZrTi,~,,07were fabricated in air to see if A1 could enter the Ca site, but no such evidence was gleaned from SEM, insofar as the fired materials consisted mainly of zirconolite having a stoichiometry of C~~,,Zrl~lTil~85Alo~~207 plus some minor CaTiO,. The zirconolite stoichiometry did not immediately lend itself to an interpretation of whether the A1 preferred the Ca or Ti sites, although it can be said that, because the 2M zirconolite structure cannot contain more than 1 f.u. of Ca[8], only about half the A1 at most could occupy the Ca site. The 27AlMAS NMR spectrum of Fig. l(c) is dominated by a resonance at 0 ppm which strongly suggests that the A1 in this sample is situated in a 6 coordinate 0 environment as described by the Ti site in this stoichiometry.
-
Another point of interest was to see whether A1 substituted in the Ti sites would only enter the octahedral Ti sites or whether inclusion in the split 5-fold Ti site was also possible. Y substituted in the Ca site acted as a charge compensator. Samples were made with Ca,,,Y,ZrTi~,~,,A1,0, (x = 0.1, 0.2 and 0.3) notional stoichiometry and were found to be very nearly single-phase (> 99%) by backscattered scattered electron microscopy. Rietveld refinement of neutron scattering data indicated that -75% of the A1 was located on the split 5-fold Ti site, with the remaining -25% on the Ti3 octahedral site. It was also clearly evident from the refinement that the split 5-fold Ti site is quite disordered and exhibits a wide range of metal-oxygen distances, compared to the very ordered octahedral Ti sites. 27AlMAS NMR studies of such samples (not shown) give results very similar to that of Fig. 2(c) where the spectrum is dominated by a narrow 0 ppm resonance, with no indication of further resonances attributable to 5-coordinate or highly distorted 6-coordinate sites being observable. It is possible that the resonance for bulk of the A1 in these samples is extremely broad and that the much narrower 27Alresonance for the Ti3 octahedral site is simply sitting above it and much more readily observed. This 5-coordinate A1 site may have to be investigated with static broadline experiments or point-by-point acquisitions, as was the case in the 71Ga NMR and NQR studies of Ga substituted zirconolites[5]. These experiments are being currently being undertaken.
-
318
Environmental Issues and Waste Management Technologies VIII
3
I
'
"
300
'
I
~
200
'
'
'
I
100
"
'
'
0
I
~
"
'
I
-100
~
'
'
'
-200
I
"
~
~
I
-300
"
.
'
'
kHz
I
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
400
300
200
100
0
-100
-200
-300
-1
kHz
I . I . I . , . I . I . I . I . I . I . I . I .
300
250
200
150
100
50
0
-50
-100
-150
-200
-250
-
ppm
Figure 1: (a) 25Mgstatic NMR spectrum of Cao~9Mgo~lZrTi20,. (b) 25Mgstatic (c) 27AlMAS NMR spectrum of NMR spectrum of Cao~,Yo~,ZrTil~,Mgo~207. Ca0.95A10.1ZrTi1.9507'
Environmental Issues and Waste Management Technologies VIII
319
CONCLUSIONS AND FINAL REMARKS In zirconolite prepared by ceramic methods with a firing temperature of 1400°C, the respective solubilities in the Ca site of Zr, Mg, Fe2’, Mn2’, Co2’, Ni2’ and A1 were found as < 0.02,0.15,0.14,0.08,0.15,0.13 and < 0.02 formula units respectively. Solubilities in the Ca site of Fe3’, Mn3’ and Co3’ were 0.14, 0.09, and 0.13 f.u. The apparent similarities of these solubilities for divalent and trivalent species of a given ion need to examined further using X-ray absorption spectroscopy to check the deduced valences. The solubility of A1 in the 5-fold Ti site of zirconolite appeared to be much higher than in the octahedral sites from Rietveld analysis, and this result appeared to contradict the NMR result. Further static NMR experiments are necessary to understand this discrepancy.
-
ACKNOWLEDGEMENT The authors wish to thank Dr. M. E. Smith, Physics Dept., University of Warwick, for access to their Chemagnetics CMX600 NMR spectrometer. REFERENCES ‘A. E. Ringwood, S. E. Kesson, N. G. Ware, W. Hibberson and A. Major, “Geological immobilisation of nuclear reactor wastes”, Nature, 278, 219-23 (1979). 2A. E. Ringwood, S. E. Kesson, K. D. Reeve, D. M. Levins and E. J. Ramm, “Synroc”; pp 223-334 in Radioactive Waste Forms for the Future, Edited by W. Lutze and R. C. Ewing. Elsevier, Amsterdam, 1988. 3B. A. Hunter, “Rietica - A Visual Rietveld Program”, IUCR Powder Diffraction Newsletter, v20, 21 (1998). 4G. R. Lumpkin, K.L. Smith, M.G. Blackford, R. Gier6 and C.T. Williams, “Determination of 25 elements in the complex oxide mineral zirconolite by analytical electron microscopy”, Micron, 25, 58 1-7 (1994). ’B.D. Begg, E.R. Vance, B.A. Hunter, J. V. Hanna, “Zirconolite Transformation Under Reducing Conditions”. J . Mater. Res. 13, 3 181-3190 (1998). 6R. D. Shannon, “Revised Ionic Radii and Systematic Studies of Interatomic Distances in Halides and Chalcogenides”, Acta Crystallographica, A32, 75 1-67 (1976). 7R. S. Roth, T. Negas and L. P. Cook, Phase Diagrams for Ceramists, Vol IV, Edited by G. Smith, National Bureau of Standards. Washington, D. C., 1981. 8H.J. Rossell “Zirconolite - a fluorite-related superstructure”, Nature, 283, 282-83 (1980).
320
Environmental Issues and Waste Management Technologies VIII
EFFECTS OF SUB-SURFACE DAMAGE INDUCED BY MECHANICAL POLISHING ON LEACH TESTING OF CESIUM-BEARING HOLLANDITE
M.L. Carter, E.R. Vance, D.J. Attard and D. R. G. Mitchell Australian Nuclear Science & Technology Organisation, Australian Nuclear Science & Technology Organisation, New Illawarra Rd. Lucas Heights, NSW, Australia ABSTRACT The dissolution in deionised water at 90°C of Cs and Ba fiom mechanically polished Cs-doped Ba hollandite samples is examined and the normalised Cs release rates were < 0.003 g/m2/day after 28 days. Cross-sectional transmission electron microscopy (TEM)of these hollandite samples indicated sub-surface damage originating from the mechanical surface preparation prior to leaching. Optimal grinding and polishing techniques were developed to prepare leach samples with damage-free surfaces. The effect of surface damage on the aqueous durability of hollandite in water at 90°C appeared to be small. INTRODUCTION Barium hollandite, [Ba,Csy][(Ti, A1)3’2,,,Ti4+8-~-,]016,is one of the of four main minerals in the multiphase titanate ceramic, synroc, developed for the immobilisation of Purex-type high level nuclear waste (HLW) from the reprocessing of spent nuclear fuel [ 11. The hollandite immobilises cesium isotopes. 13’Cs and particularly 13’Cs have been identified in performance assessment studies as major contributors to long-term releases from repositories, and are difficult to immobilise because of their volatility at high temperatures and tendency to form water-soluble compounds.
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
321
In conjunction with standard aqueous leach tests, the cross-sectional transmission electron microscopy (TEM)technique has been used to study leached hollandite surfaces to gain a better understanding of the reaction products and the corrosion mechanisms [2]. Sub-surface damage stemming from leach disk preparation (prior to TEM specimen preparation) was also found to be present. Surface damage in ceramics can exert significant effects on the mechanical properties [2-31, but the corrosion behaviour of ceramics as a function of surface damage has not received much attention. In this study we assess the influence of surface damage, due to polishing on the leach behaviour of hollandite.
EXPERIMENTAL
Ba hollandites (nominal composition B ~ C S ~ . ~ ( T ~ , . , , A I , , ) ~ ~ ~ , . ~ O containing 0.5 formula units of extra TiOz in the form of rutile were produced by the alkoxide-route [l]. This method involves mixing the correct molar quantities of aluminum sec-butoxide and titanium (IV)isopropoxide in ethanol with Ba nitrate and Cs nitrate dissolved in water, while continuously stirring. The mixture was heated (-1 10°C) and stirred to drive off the alcohol and water. The dry product was then calcined in air for two hours at 750°C and then wet milled. The sample was hot-uniaxially pressed (HUPed) in a 40 mm diameter graphite die for 10 h / 21 MPa at 1200°C. Disks for leaching (7 x 7 x 1.5 mm) were prepared by fist cutting with a slow speed diamond bladed saw using an oil-based coolant. The disks were polished with subsequently finer grades of silicon carbide using a non-polar lubricant. A total of 300-500 pm of material was removed from each cut face of the leach disk during grinding, ensuring that all saw damage was removed. Polishing with 2-4 pm diamond was followed by a 0-0.5 pm diamond polishing step. In our normal preparation method grinding and polishing was performed with a view to produce a desired surface finish, with little consideration of the surface damage aspects. The optirnised polishing process entailed removal of >200 pm of material by each grinding step using increasingly finer grades of Sic paper: 30 pm, 2 2 p , 14 pn and finally removal of 70 pm with 6 pm paper, all with a nonpolar lubricant. There are quite limited systematic data on the determination of
322
Environmental Issues and Waste Management Technologies VIII
damage depths in ceramics as a function of surface grinding. However, direct TEM measurements of dislocation depth in A120,[S] and indirect observations of mechanical damage and related phenomena [6-71 in Al,O, and Ni-Zn ferrite, would suggest a maximum damage depth due to grinding of =20 pm in ceramics. This indicates that our removal depths during grinding are very conservative. Diamond polishing was canied out using 2-4 pm diamond paste for 10 mins under 1.05 kg load, followed by a 0-0.5 pm diamond polish for 10 mins under 1.05kg load, and finally a colloidal silica polish for 25 mins with no added load removed all scratches fiom grinding, without excessive edge rounding. Leach disks were sealed in evacuated glass tubes, with Ti powder as a getter, to maintain a reducing atmosphere, approximately consistent with the conditions in the hollandite fabrication process. Annealing was carried out at 900°C for 2 h. This temperature had to be well below the hot pressing temperature of 1200°C in order to prevent fracture and bubble development, due to relief of stresses built in fiom gases occluded during hot pressing [8]. A JEOL JSM6400 scanning electron microscope (SEM) equipped with a Noran Voyager energy-dispersive spectroscopy system (EDS) was operated at 15 keV for microstructural work. X-ray diffraction (XRD) was performed with a Siemens DSOO diffractometer and CoKa radiation. TEM specimen preparation was a version of the method of [3]. Ion milling was performed with a GATAN precision ion polishing system (PIPSTM),which employed 5 kV A f ions at an angle of incidence of 6" initially. Final thinning was achieved at 3.5 kV at an angle of incidence of 4" once the specimen became light transparent. Specimens were examined using a JEOL 2000FXII TEM operating at 200 kV. MCC-1 leach tests [9] were carried out in deionised water at 90°C for 7 to 28 days. Triplicates were used in all leach tests. Leachate solutions were analysed using inductively coupled plasma-mass spectrometry (ICP-MS). The leachant was cooled to ambient room temperature, filtered (0.45pm) and the pH measured at the end of each leach period. The normalised release rate was calculated by dividing the cumulative extraction rate by the time leached and the elemental concentration in the solid. As the variability in release rate between samples (under the same conditions) is larger than the uncertainties in
Environmental Issues and Waste Management Technologies VIII
323
measuring each element, the values reported are the average values and errors are one standard deviation from the average.
RESULTS AND DISCUSSION XRD/SEM XRD of polished samples showed sharp peaks corresponding to hollandite and rutile structures. The SEM investigation showed the sample to be homogeneous hollandite with micron sized grains of rutile and Ba titanate throughout (see figure 1).
Figure 1: SEM backscattered electron image. H = hollandite (matrix), R = rutile (dark grey) and B = Ba titanate (light grey). Bar =20 pm.
TEM A typical cross-section of a hollandite after ‘normal’ preparation is shown in figures 2a and 2b. The hollandite phase shows marked striations. This is a consequence of twinning, and is a well known response of this phase to ion milling [lO]. Low magnification examination of the outer surface of the material (the original polished surface) showed a layer of damage of up to 100 nm thick (figure 2a). This layer disrupted the striated appearance of the hollandite substrate. High resolution imaging (figure 2a) along the zone axis
324
Environmental Issues and Waste Management Technologies VIII
showed that the lattice fringes were continuous from the substrate into the damaged layer.
Figure 2: TEM image (a) showed a layer of damage of up to loOnm thick. (b) High resolution image. Insert of interface shows lattice fringes are continuous from the substrate into the damaged layer. A typical cross-section of a hollandite after optimised preparation is shown in figures 3a and 3b. Twinning from the hollandite was found to extend up to the surface with no discrete damage layer in evidence. An attempt was made to remove the damaged surface layer induced during normal mechanical polishing by annealing. Vacuum annealing under the conditions chosen (900°C, 2h) did not remove it although some modification occurred. The sub-surface damage layer prior to annealing had given way to a less defective twinned structure. Higher annealing temperatures would evidently be needed to anneal out the damage and that the change of redox conditions relative to those used in fabrication ( oxygen fugacity unknown in detail) would produce instabilities. Samples prepared by the normal and optimal method, were used in the MCC-1 leach test.
Environmental Issues and Waste Management Technologies VIII
325
Figure 3: TEM cross-section of a hollandite after optimised preparation, (a) Twinning from the hollandite was found to extend up to the surface with no discrete damage layer in evidence. (b) High resolution image. Leach Testing DIW normalised solution release rates for Ba and Cs for normal and optimally prepared samples exposed to DIW at 90°C are presented in Table 1. Release rates for Ti and A1 were below detectable levels. There was little difference in the normalised release rates between the optimally prepared HUPed samples and the normally prepared samples in both leach periods. The Ba release rates were slightly higher than the Cs release rates which reflect the second Ba containing phase in the original samples (see SEM above). The Cs release rates agree fairly well with those reported elsewhere [2,11,12]. Due to the presence of a secondary phase containing Ba, congruent release of Ba and Cs was not observed.
326
Environmental Issues and Waste Management Technologies VIII
Table 1: Normalised release rates for Ba and Cs
Sample
Normal
Normalised Release Rates, 0-7 days, g/mz/day cs Ba 0.028(0.006) 0.038 (0.006)
Optimised
0.025 (0.003)
PRpilLltiOIl
Errors in parentheses
Normalised Release Rates, 7-28 days, g/m2/day cs Ba 0.0027 (0.0002) 0.0064(O.ooo9)
0.036 (0.002) 0.0022 (0.0002) 0.0061 (O.ooo9)
CONCLUSIONS The subsurface damage induced by non-optimal surface preparation did not significantly affect the release rate of Cs and Ba in hot-pressed Ba hollandite samples. 0-7 day Cs release rates were -0.03 g/m2/day for samples prepared both ways in fair agreement with the literature [2, 11, 121. Due to the presence of a secondary phase containing Ba, the release rate of Ba exceeds that of Cs, especially in the 7-28 day tests. REFERENCES
‘A.E. Ringwood, S.E. Kesson, K.D. Reeve, D.M. Levins and E.J. R a m , “Synroc”; pp. 233-334 in Radioactive Waste Forms for the Future, edited by W. Lutze and R.C. Ewing. North-Holland, New York, 1988. 2M.L. Carter, E.R. Vance, D.R.G. Mitchell, J.V. Hanna, 2.Zhang, and E. Loi “Fabrication, Characterisation and Leach Testing of Hollandite, (Ba,Cs)(Al,Ti),T&O,, ”,accepted by J o u m l of Materials Research, (2002). 3H.Frei and G. Grathwohl, “Microstructure and Strength of Advanced Ceramics after Machining”, Ceramics International, 19,93-104 (1993). 4R. Haushater, H. Frei and G. Grathwohl, “TEM studies in SSN Surfaces after Machining”; pp. 116-8, in Euro-Ceramics: Properties of Ceramics, Vol. 2, edited by G. de With, R. A. Terpstra and R. Metselaar. London: Elsevier, 1989. ’B. J Hockey, “Observations by Transmission Electron Microscopy on the Subsurface Damage Produced in Aluminium Oxide by Mechanical Polishing and Grinding”, Proceedings of the British Ceramic Society, 20 95-11 1 (1972).
Environmental Issues and Waste Management Technologies VIII
327
‘S. Chandrasekar, K.Kokini and B.Bhushan, “The Effect of Abrasive Properties on Surface Finishing Damage in Ceramics”; pp. 33-46 in ASME Intersociety Symposium on Machining of Advanced Ceramic Components. Chicago, Illinois, 1988. 7H. H. K. Xu and S.Jahanmir, “Simple Technique for Observing Subsurface Damage in Machining of Ceramics”, Journal of the American Ceramic Society, 77 1388-90 (1994). *E. R Vance, M. L. Carter, S .Moricca, M. W. A. Stewart, “The Importance of Redox State in SYNROC Fabrication”. Proceedings of PacRim2, Edited by P. Walls, C. Sorrell and A. Ruys. Symp. 16, [CD-ROM] 1998. %CC, MCC-1P static Leach Test Method (Rev. l), DOE TIC 11400, Materials Characterisation Center, Battelle Northwest Laboratory, Richland, WA, 1983. 10 J. C. Barry, J. L. Hutchison, and R. L. Segall, “Ion-beam damage in hollandite”, Journal of Materials Science, 18 1421- 25 (1983). ”M. L. Carter and E. R. Vance, Proceedings of PacRim2, Edited by P.Walls, C. Sorrell and A. Ruys. Symp. 16, [CD-ROM] 1998. ‘*K.P Hart, E. R. Vance, R.A. Day, B.D. Begg, P.J. Angel and A. Jostsons, ‘‘Immobilizationof Separated Tc and Cs/Sr in Synroc”; pp. 281- 87, in Scientific Basis for Nuclear Waste Management XIX,edited by W.M. Muphy and D.A. Knecht. Materials Research Society, Pittsburgh, PA, USA, 1996.
328
Environmental Issues and Waste Management Technologies VIII
IRON PHOSPHATE GLASSES FOR VITRIFYING SODIUM BEARING WASTE Cheol-Woon Kim, Dongmei Zhu, and Delbert E. Day Graduate Center for Materials Research, University of Missouri-Rolla, MO 65409 Dirk Gombert Idaho National Engineering and Environmental Laboratory, Idaho Falls, ID 83415
ABSTRACT Iron phosphate glassy waste forms containing up to 48 wt% of a sodium bearing waste (SBW) stored at the Idaho National Engineering and Environmental Laboratory (INEEL) have been investigated. The waste-containing iron phosphate glasses (IPG) were easily melted in alumina crucibles at 950-1000 C . No s u l k rich inclusions or phases were detectable in the IPG when examined by scanning electron microscopy (SEM). From 43 to 100 wt% of the sulfate originally present in the waste was retained in the IPG whose leach resistance was excellent. The chemical durability of IPG containing 40 and 48 wt% SBW (IP40WG and IP48WG, respectively) was evaluated from the bulk dissolution rate (DR), the product consistency test (PCT), (7 days in deionized water at 90 C) and from the vapor hydration test (VHT), (7 days at 200 C). The IP40WG had the best chemical durability. The total normalized quantity of Al, Ca, Cr, Fe, K, Mg, Mn, Na, and P released from the IP40WG during PCT was only 1.3 g/L (0.7 g/m2). The excellent chemical durability of the IP40WG was verified by the VHT results where the measured corrosion rate for the IP40WG and IP48WG was < 0.2 and 190 g/m2/day, respectively. These results indicate that iron phosphate glasses containing up to 40% SBW should satisfy all of the existing requirements for chemical durability and suggest that iron phosphate glasses could reduce the current need to adapt borosilicate glasses to any and all waste constituents while substantially reducing the ultimate waste volumes and disposal cost.
INTRODUCTION Only one material, borosilicate glass, is currently approved by the U.S. Department of Energy as a host matrix for vitrifying high level nuclear waste (HLW). However, HLW streams have complex and diverse chemical To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management TechnologiesVIII
329
compositions and many of them contain components such as phosphates and sulfates that are poorly soluble in borosilicate glasses.lY2Such problematic wastes can be pre-processed andor diluted to compensate for the incompatibility with the borosilicate glass matrix. However, it is desirable to avoid pre-treating or diluting the waste since these alternatives will greatly increase the cost of cleaning up the former nuclear weapons production facilities. Direct vitrification in an alternative glass is preferable, if possible, since it should be less expensive. The Idaho National Engineering and Environmental Laboratory (INEEL) stores approximately 5.7 million liters of sodium bearing waste (SBW) and intends for it to be vitrified from 2007 to 2012.2 The SBW composition is relatively high in sodium, aluminum, and sulfate (Table 1); and waste loading in the baseline borosilicate glasses is currently limited to a maximum of 20 wt% due to sulfate ~olubility.~ The objective of the present study was to investigate the waste loading and chemical durability of iron phosphate glasses (IPG) as an alternative glass for vitrifjring the SBW at INEEL. Table 1. Nominal composition and raw materials used to DreDare the simulated SBW.
1-
Raw Material NaNO? A1203 KNOq CaO Fe203
Oxide
c1 F Total
I
0.9 0.8 lOO.0l
CaF2 Total
wt%
68.3 14.4 8.4 0.6 0.7
0.8 100.0
GLASS PREPARATION The appropriate amounts of the raw materials (Table 1) were ground to pass through a 100 mesh sieve, dry mixed by tumbling in a sealed plastic container and then stored in a sealed container until used. The mixture of simulated SBW, Fe2O3, and P205 was thoroughly dry mixed and then melted in a high purity alumina crucible in an electric furnace. Based on the evaluation of 60 trial melts of varying SBW concentration, two iron phosphate glass compositions containing 40 and 48 wt% of the SBW (IP40WG and IP48WG, respectively) were investigated. A total of 600 grams of the IP40WG was prepared by melting two
330
Environmental Issues and Waste Management Technologies VIII
separate 300 gram melts for 4 hours (h) at 900 C followed by 3 h at 1000 C. Each melt was stirred several times with an alumina rod to aid in homogenizing the melt. Most of the melt was poured onto a steel plate and cooled quickly to room temperature without annealing. A few grams from each melt was poured into steel molds to form rectangular bars that were 1 cm 1 cm 5 cm in size to be used for the dissolution rate tests. These bars were annealed at 420 C for approximately 4 h and slowly cooled overnight in the annealing furnace to room temperature. The IP48WG was produced in the same manner except that it was melted at 1000 C for only 3 h. The chemical composition of each glass was calculated from the batch composition and also measured by inductively coupled plasma-emission spectroscopy (ICP-ES) as well as Leco method for sulfur analysis (Table 2). The calculated and measured compositions were in reasonable agreement. Scanning electron microscopy (SEM) and powder X-ray diffraction (XRD) did not reveal any segregation, nodules, or crystalline phases in either sample, IP4OWG and IP48WG, suggesting that both glasses were chemically homogeneous. Table 2. Batch and analyzed (ICP-ES) compositions (wt%) of iron phosphate glasses containi ig 40 and 48 wt% of SBW (IP40WG and IP4t WG,-respectively).
I
Oxide (wt%) Na20 A1203
K20 CaO Fe207
IP40 WG I Batch ICP-ES 20.5 23.2 10.9 12.7 3.0 3.2 0.9 0.8 10.3 11.5
1
I I
IP48 WG
1i:;I l
L
;
1;:;
I
0.2 Cr2O3
SO?
1.4 0.311 F I 0.31 Total lOO.0l Analyzed by Leco NM = not measured C1
I I
I
0.6* NMII NMI 99.91
I
1
I
I
0.41I 0.41 lOO.0l
NM NM 99.9
PROPERTIES High Temperature Viscosity The viscosity of the melts in their melting range was measured using a Brookfield rotating viscometer. A non-standard alumina spindle was made from a 10 cc straight wall crucible. Three standard viscosity oils (97.2, 98.8, and 965 centipoise) were used to calibrate the viscosity device. The estimated error was
Environmental Issues and Waste Management Technologies VIII
331
less than 5%. Each melt was thermally equilibrated at a selected temperature for 30 min, whereupon, a preheated spindle (that had been just above the melt) was immersed in the melt. Spindle speeds of 10, 20, and 50 rpm were used for each measurement. The viscosity of each melt was measured three times (at each temperature & spindle rpm) and then averaged. Figure 1 shows the viscosity and temperature profile of the IP4OWG and IP48WG melts together with the baseline iron phosphate melt 43Fe203.57P205, wt%. The lower viscosity for the two melts containing SBW is consistent with the higher soda content of these melts. Iron phosphate melts are 10 to 100 times more fluid than most silicate melts as indicated by the viscosity curves for the iron phosphate melts shown in Figure 1. At their pouring temperature (1000 C), the viscosity is approximately 4 to 6 poise so a melting time of only a few hours should be adequate to achieve a chemically homogeneous melt free of all batch materials.
4
4
8
v)
400
v)
300 200 100
800
I
850
.
I
900
'
I
950
'
I
.
I
'
I
.
I
.
I
.
I
'
1000 1050 1100 1150 1200 1250 1:
Temperature ("C)
'
l O Z !
I
.
6.5~10~
,
I
7.0~10~
I
.
7.5~10~
I
.
8.0~10~
I
8.5~10~
Temperature (K)"
Figure 1. Viscosity of melts IP40WG and IP48WG over their melting range. Data for a 43Fe203-57P205glass is shown for comparison. a) viscosity (centipoise) vs. temperature ( C); b) viscosity (centipoise-log scale) vs. temperature (K)-'. Other Measurements The density of each glass was measured at room temperature by the Archimedes' method using water as the suspending medium. The average density of each glass, as measured on three samples, is 2.76 and 2.71 g/cm3 for the IP40WG and IP48WG, respectively (Table 3). The estimated error is 0.01 g/cm3. The liquidus temperature (Table 3) was measured per ASTM C 829-81 in a temperature gradient furnace using a platinum tray in which glass particles were fused to form a thin layer of melt (-3 mm). The platinum tray holding the melt was left in the furnace for 24 h to ensure that equilibrium between the crystal and
332
Environmental Issues and Waste Management Technologies VIII
glassy phases was established. The liquidus temperature of each glass was measured twice and then averaged. Precision of the liquidus temperature between two independent measurements in the same fwnace was within 10 C. Table 3. Density and liquidus temperature of IP40WG and IP48WG.
I
IP4OWG IP48WG
I
I
Density 2.76 gkm3 2.71 g/cm3
Liquidus Temperature 740 C 630 C
1
CHEMICAL DURABILITY Dissolution Rate (DR) Specimens were diamond sawed from the annealed glass bars and then ground progressively using 240, 600, and 800 grit Sic papers. The dimensions of each specimen were measured ( 0.001 inch) and were approximately 1 cm 1 cm 1 cm. The specimens were rinsed with acetone and deionized water (DIW), dried at 90 C, cooled to room temperature, and then weighed ( 0.01 mg). Each specimen was suspended by a thin rayon thread in a PyrexTMflask containing 100 ml of DIW. The flasks containing the specimens were placed in an oven at 90 2 C. Specimens were removed after 8, 16, and 32 days, rinsed with DIW, dried at 90 C, cooled, and weighed ( 0.01 mg). The dissolution rate (DR) was calculated from the measured weight loss, surface area, and time.4 All tests were carried out in triplicate. The pH of the leachate was measured each time the sample was removed for weighing. Samples of the Environmental Assessment (EA) glass (provided by the Defense Waste Processing Facility (DWPF) at Westinghouse Savannah River Co.) were also measured at the same conditions and used as a reference for the DRvalues for the
. Y------
.--E
:
-
2
1E-9:
0"
3
IWOWG
pH.5 9
p 3 72
pH=5 9
-1
1E-10
Time (days)
Figure 2. Dissolution rate (DR= g/cm2/min)in DIW at 90 C for iron phosphate glasses (IP40WG and IP48WG) and EA glass. Initial pH of DIW was 5.8. Note the nearly constant pH of the leachate for the IP40WG sample which is due to the larger buffering action (compared to silicate glasses) of phosphate glasses.
Environmental Issues and Waste Management Technologies VIII
333
iron phosphate glasses. The DR of the iron phosphate glasses (IP40WG and IP48WG) and the EA glass is shown in Figure 2. The IP40WG had a DRranging from a low of -1 10-’ to a high of -2 10-’ g/cm2/min.The DRrange for the IP48WG was a low of -5 l 0-’ to a high of -3 10-*g/cm2/min. The DR of the IP40WG and IP48WG iron phosphate glasses in this work was from -25 to -5 times, respectively, lower than that of the EA glass in DIW at 90 C. Product Consistency Test (PCT) The chemical durability of the iron phosphate glasses was also measured by the product consistency test (PCT). After completion of the PCT, the concentration of ions in the leachate was measured by ICP-ES*. A thorough description of the PCT can be found in ASTM C 1285-97, whose procedures were used for the glasses in this work. All tests including pH measurements were conducted in duplicate and then averaged. The normalized elemental mass release was calculated either as g/L or g/m2 from the concentration of each element in the leachate and the mass fraction of element in the in order to provide easier comparison with other standard glasses. The PCT results for the iron phosphate glasses (IP40WG and IP48WG) and other borosilicate glasses are given in Figure 3.
Figure 3. Normalized elemental mass release (g/L (a) or g/m2(b)) from iron phosphate glasses and standard borosilicateglasses after PCT in DIW at 90 C for 7 days. The initial @Hi)and final (pHf)pH of the leachate is given for each glass. Elements where the mass release was 0.01 g/L or g/m2, i.e., Ca, Cr, Fe, Mg, and Mn are not shown in Figure 3. The total normalized quantity of Al, Ca, Cr, * ICP-ES analysis was conducted by Acme Analytical Laboratories Ltd. at 852 East Hastings St., Vancouver, British Columbia, V6A 1R6, Canada.
334
Environmental Issues and Waste Management Technologies VIII
Fe, K, Mg, Mn, Na, and P released from IP40WG was only 1.3 g/L (0.7 g/m2) which is -35 times less than the quantity (B, Li, Na, and Si) leached from the EA borosilicate glass (see Figure 3a). The quantity of ions released from IP4OWG was also less than the quantity released from the CVS-IS and LD6-54-12 standard borosilicate glasses6produced at the Pacific Northwest National Laboratory. The chemical durability of the IP48WG was not as good as that of the IP40WG, but was still comparable to that of the EA glass. Vapor Hydration Test (VHT) A complete description of the vapor hydration test (VHT) can be found in reference 7, whose procedures were used to test the iron phosphate glasses in this work. Figures 4a and 4b show cross sectional views of the IP40WG and IP48WG, respectively, after 7-day VHT at 200 C. As is evident in Figure 4a, no corroded layer was detectable on the surface of the IP40WG glass. More extensive corrosion layers were clearly seen on the surface of the IP48WG samples as shown in Figure 4b.
Figure 4. Optical photo micrographs of the cross sectioned IP40WG (a) and IP48WG (b) specimens after the VHT at 200 C for seven days. The mass of specimen dissolved per unit surface area (corrosion rate) can be calculated by measuring the initial specimen thickness and the thickness of the glass remaining at the end of the test.7 The corrosion rate for the IP40WG and IP48WG was < 0.2 and 190 g/m2/day, respectively. The VHT corrosion rates for the two iron phosphate glasses and two borosilicate glasses are given in Table 4. These results indicate that IP40WG had an outstanding chemical resistance to the humid conditions at 200 C for seven days. Table 4. VHT corrosion rates (g/m2/day)of iron phosphate glasses (IP40WG and IP48WG) and borosilicate glasses' at 200 C for 7 days. LAW-A33 LD6-54- 12 IP40WG IP48WG 190 -140 -196 8'1 0.2
* Test conducted at 175 C, higher value would be expected when conducted at 200 C.
Environmental Issues and Waste Management TechnologiesVIII
335
SUMMARY AND CONCLUSIONS The present investigation of iron phosphate glass wasteforms has identified several advantages for vitrifying the SBW stored at INEEL. Among these advantages is a high waste loading of the simulated SBW from 40 to 48 wt% depending upon the desired leaching resistance or chemical durability. The chemical durability of the iron phosphate vitrified wasteform, IP40WG, is comparable to, or better than that reported for borosilicate glass wasteforms. Iron phosphate glasses containing from 40 to 48 wt% SBW can be melted as low as 950-1000 C . Because of their high fluidity and rapid chemical homogenization (low viscosity at the melting temperature), melting times can be only a few hours compared to the -48 h needed for borosilicate glasses. The high waste loading, low melting temperature, and rapid furnace throughput (short melting times) of iron phosphate glasses suggest that they offer the potential to significantly reduce the cost of vitrifying the sodium bearing waste stored at INEEL. REFERENCES ‘S.L. Lambert and D.S. Kim, “Tank Waste Remediation System High-Level Waste Feed Processability Assessment Report”, PNNL #WHC-SP-1143, PNNL, Hanford, WA, 1994. 2J.M. Perez, Jr., D.F. Bickford, D.E.Day, D.S. Kim, S.L., Lambert, S.L. Marra, D.K. Peeler, D.M. Strachan, M.B. Triplett, J.D. Vienna, and R.S. Wittman, “High-Level Waste Melter Study Report”, PNNL-13582, PNNL, Hanford, WA, 2001. 3D. Gombert, “Cold Crucible Induction Melter Technology: Results of Laboratory Directed Research and Development”, INEEL/EXT-01-01213, INEEL, Idaho Falls, ID, 2001. 4D.E. Day, Z. Wu, C.S. Ray, and P. Hrma, “Chemically durable iron phosphate glass wasteforms”, J. Non. Cryst. Solids 241 (1998) 1-2. 5C.M. Jantzen, N.E. Bibler, D.C. Beam, C.L. Crawford, and M.A. Pickett, “Characterization of the Defense Waste Processing Facility (DWPF) Environmental Assessment (EA) Glass Standard Reference Material (U)”, WSRC-TR-92-346, Rev. 1, Westinghouse Savannah River Company, Aiken, South Carolina, 1993. 6P.R. Hrma, G.F. Piepel, M.J. Schweiger, D.E. Smith, D.S. Kim, P.E. Redgate, J.D. Vienna, C.A. LoPresti, D.B. Simpson, D.K. Peeler, and M.H. Langowski, “Property Composition Relationships for Hanford High-Level Waste Glasses Melting at 1 150 C”, PNL-10359, Volumes 1 and 2, PNNL, Richland, WA, 1994. 7“Vapor-Phase Hydration Test Procedure”, GDL-VHT, Rev. 1, PNNL, Richland, WA, 2000. *A. Jiricka, J. D. Vienna, P. Hrma, and D. M. Strachan, “The Effect of Experimental Conditions and Evaluation Techniques on the Alteration of Low-Activity Waste Glasses by Vapor Hydration.” J. Non. Cryst. Solids, 292 (2001) 25-43.
336
Environmental Issues and Waste Management Technologies VIII
PHOSPHATE GLASSES FOR VITRIFICATION OF WASTES WITH HIGH SULFUR CONTENT Dong-Sang Kim,John D. Vienna, and Pave1 Hrma Pacific Northwest National Laboratory,' Richland, WA 99352 Nathan Cassingham New York State College of Ceramics at Alfred University, Alfred, NY 14802 ABSTRACT The IOW solubility of sulfate in silicate-based glasses, approximately 1 mass% as SO3, limits the loading of high-level waste (HLW) and low-activity waste (LAW) containing high concentrations of sulfur. Based on crucible melting studies, we have shown that the phosphate glasses may incorporate more than 5 mass% SO3;hence, the waste loading can be increased until another constraint is met, such as glass durability. A high-sulfate HLW glass has been formulated and tested to demonstrate the advantages of phosphate glasses. The effect of waste loading on the chemical durability of quenched and slow-cooled phosphate glasses was determined using the Product Consistency Test. INTRODUCTION Phosphate glasses have been studied as alternative waste forms for vitrification of various radioactive wastes [1-41. The advantages of iron phosphate glass as a waste form are good chemical durability, buffering action of P in the leachate, low melting temperature (so low volatilization of radionuclides), and high solubility of some troublesome components in borosilicate glass, such as SO3 and Cr2O3. However, there are technical issues that must be clarified before the phosphate glasses can be adopted as a waste form: the effect of crystallization on chemical durability and radionuclides partitioning, refractory and electrode corrosion, large-scale process demonstration, wasteform qualification, etc. The current estimated loading of wastes with high sulfur contents in borosilicate glasses is about 20 mass% for Idaho National Environmental and Engineering Laboratory (INEEL) sodium bearing waste (SBW) [5] and 10 to 27 mass% for Hanford low activity waste (LAW) [6]; these loadings correspond to approximately 6.5 to 20 mass% Na20 in glass. Without the sulfate problem, an up to three-fold increase of waste loading would be possible for Hanford LAW and about a two-fold increase for SBW if 20 mass% Na2O loading is assumed. Pacific NorthwestNational Laboratory is operated for the U. S. Department of Energy by Battelle under Contract DE-ACO6-76RLO1830. To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
337
The objectives of this study are to demonstrate, through preliminary screening tests based on the chemical durability of quenched and slow-cooled glasses, a possibility of incorporating high levels of sulfate in phosphate glasses and to formulate high-waste loaded phosphate glasses for SBW. Experimental procedures are described followed by results achieved to date. Preliminary results for a H d o r d LAW are also reported. EXPERIMENTAL Table I shows the composition of SBW used in Table 1. Waste composition this study. The composition of Tank WM-180 used in this studv reported by Christian [7] was simplified by deleting Oxide mass% the components with
338
Environmental Issues and Waste Management Technologies VIII
performed at 90°C for both quenched and CCC-treated glasses according to ASTM Method C 1285-97 [9]. Table 2. Glass compositions (in mass%) SBWOldO-xx SBW01-45-xx SBWOl-50-xx 0 10 20 30 0 10 20 0 10 20 11.01 11.01 11.01 11.01 12.38 12.38 12.38 13.76 13.76 13.76 0.14 0.14 0.14 0.14 0.16 0.16 0.16 0.18 0.18 0.18 0.86 0.86 0.86 0.86 1.08 1.08 1.OS 0.97 0.97 0.97 0.35 0.35 0.35 0.35 0.44 0.44 0.44 0.39 0.39 0.39 0.08 0.08 0.08 0.08 0.09 0.09 0.09 0.11 0.11 0.1 1 0.29 0.29 0.29 0.29 0.33 0.33 0.33 0.37 0.37 0.37 0.56 6.56 12.56 18.56 0.63 6.13 11.63 0.71 5.71 10.71 3.01 3.01 3.01 3.01 3.39 3.39 3.39 3.77 3.77 3.77 0.16 0.16 0.16 0.16 0.18 0.18 0.18 0.20 0.20 0.20 0.32 0.32 0.32 0.32 0.36 0.36 0.36 0.41 0.41 0.41 20.76 20.76 20.76 20.76 23.36 23.36 23.36 25.96 25.96 25.96 60.32 54.32 48.32 42.32 55.36 49.86 44.36 50.40 45.40 40.40 0.31 0.31 0.31 0.31 0.35 0.35 0.35 0.39 0.39 0.39 1.82 1.82 1.82 1.82 2.28 2.28 2.28 2.05 2.05 2.05
E
Oxide A1203
RESULTS AND DISCUSSION Sulfate Retention in a Phosphate Glass Figure 1 shows the effect of varying the amount of sulfate added on the sulfate retained in the glass as analyzed by XRF and ICP, which was performed on two glasses and was used as reference data for XRF in the other two glasses. In Figure 1, the typical solubility in soda-lime and borosilicate glasses is also shown for comparison, which is approximately 1 mass% SO3 [51. For borosilicate glasses, undissolved sulfate usually forms an immiscible liquid salt that typically segregates above the melt. When the glass with the salt layer is crushed, homogenized, and remelted, inclusions of undissolved salt are often observed, indicating that the suffate is present above its solubility limit. In the present studies, no salt layer or sulfate inclusions were observed in phosphate glasses with 3.0 to 8.4 mass% SO3 added to the glass batch. Figure 1 shows that phosphate glasses can dissolve more than 5 mass% so3 without forming a salt layer or inclusions. However, the fate of sulfate strongly depends on melting conditions [5]; the present tests were done using dry raw materials and short melting times. More detailed studies on the mass balance of sulfbr under varying melting time and temperature conditions close to an actual melter, using slurry feed of simulated liquid wastes, are needed to establish high sulfate incorporation in phosphate glasses processed on a large scale. Figure 1 also shows that a fiaction of s u l k was lost during melting. This fiaction increases as the sulfate added to the glass increases.
Environmental Issues and Waste Management TechnologiesVIII
339
Glass Formulation Studies with SB W
Table 3 summarizes the results of XRD analysis of quenched and CCC treated glasses with 40 to 50 mass% waste components. The major crystalline phases identified were sodium-aluminurn and sodium-iron phosphates. SEM/EDS analysis performed on selected samples revealed that these phases are solid solutions with Fe in sodiumaluminum phosphate and A1 in sodium iron phosphate. Other phases observed include aluminum phosphate, sodium phosphate, and hematite (FezO3).
0
2
4 6 SO3 added (mass%)
8
Figure 1. Sulfate retained in the glass versus sulfate added in SBWO 1-40-20 glass
340
Environmental Issues and Waste Management Technologies VIII
As shown in Table 3, crystals formed in three quenched glasses and in eight CCCtreated glasses. The CCC glasses without Fe203 addition precipitated sodium-aluminum phosphate phases only. As the Fe203 content in the additive increased, iron-containing phases (sodium-iron phosphate and hematite) became the major phases. Figure 2 shows the change of total crystal mass% as a function of the Fe203 concentration in the additive at different waste loadings. It appears that a range of Fe203 concentrationexists at which CCC glasses possess minimum crystallinity or become fiee of crystals. 4- 40 mass% waste
z
40 -
+45 mass% waste
It 50 mass% waste
3
2. 30
U
Ccl
0
a 0 0
10
20
30
Fe203 (mass%) in Additive
Figure 2. Effect of Fe203 in additive on the total crystal mass% in glasses with 40 to 50 mass% SBWOl Figure 3 shows the change in density as a function of Fe203concentration and waste loading in quenched glasses. Because some quenched glasses had crystals that could affect the density, the glasses with crystals are marked with circles. With one exception, the density increases as the Fez@ fraction increases and decreases as the waste loading increases. The latter is expected because the major waste components (Na, Al) are lighter
than P.
Figure 4 shows the PCT Na normalized releases as a function of F q 0 3 concentration in the additive and the waste loading in quenched glasses. Glasses with crystals are marked with circles. Also included are the EA-glass Na limit (6.68 g/m2) for HLW borosilicate glasses and the more stringent 2 g/m2 limit, which is used for H d o r d LAW glasses (the PCT limits for B and Li release are not applicable for phosphate glasses in this study). The effects of Fe203 and waste loading on the PCT release is difficult to assess because of limited data and the presence of crystals. Only a rough evaluation is presented below.
Environmental Issues and Waste Management Technologies VIII
341
2.8
2.8
4
n
33 2.7 & v)
b
+ 0 mass%Fe203
+40 mass%
2.6
WL 45 mass% WL -+50 mass% WL
mass%Fe203 *203010 mass% Fe203 mass%Fe203 -I)-
-4-
2.5 0
I
I
10
20
F e 2 0 3 mass% in additive
-0-
30
40
45 Waste Loading (mass%)
50
Figure 3. Density as a function of Fe203 in additive (left) and waste loading (right) in quenched glasses. (Circled points represent glasses with crystals.)
n
a
3
f
10
A
10
E
z
1
-+40 mass% W L
1
+ Omass%Fe203
1
+45 mass% WL 0.1
0.1
0
10 20 mass% in additive
Fe203
30
10 mass%Fe203
*20 mass% -I)-
: 40
Fe203
*30 mass% Fe203 I
45 Waste loading (mass%)
I
1
50
Figure 4. PCT Normalized Na release as a function of Fe203 in additive (left) and waste loading (right) in quenched glasses. (Circled points represent glasses with crystals; solid line is the EA glass limit and dotted line represents 2 g/m2limit.) At 40 mass% waste loading, the normalized Na release decreases with an initial Fe203 addition and increases at further addition of Fe2O3. A similar trend can be expected at 45 mass% waste loading, with a higher Fe?03 concentration at the minimum Na release. At 50 mass% waste loading, the addition of Fe203 does not affect the Na release. It seems that the durability of the glass with 50 mass% waste loading, which contains 26.0 mass% NazO and 3.8 mass% K20, may not be improved by the modification of glass composition. At constant Fe203concentration in the additive, the normalized Na release increases with the increase of waste loading as expected. The very
342
Environmental Issues and Waste Management Technologies VIII
high Na release from SBW01-40-00 glass is an exception, which may be associated with the precipitation of 5 mass% AlP04. Figure 5 compares Na releases from quenched and CCC-treated glasses. The CCC treatment increased or had no effect on the Na release in 7 of 10 glasses and decreased the Na release in three glasses. For borosilicate glasses [10,11], the effect of CCC treatment on PCT durability was explained in terms of the impact of crystallization on the residual glass. An attempt to use similar reasoning in the case of phosphate glass was unsuccessful because of (1) lack of information on the effect of glass components on the PCT response of phosphate glasses and (2) expected nonlinear effects of certain components, e.g., A1203, on glass properties [12]. The strong decrease of Na release from two glasses (SBW01-50-00 and SBW01-50-10) after CCC treatment, indicated by a circle in Figure 5, is counterintuitive. These two compositions formed a white film covering the surfaces of glass particles after a 7-day PCT, indicating a possible precipitation of mineral phases ftom the solution. If the white phase observed at the glass surface contained Na, the actual glass dissolution would be much higher than represented by the Na release obtained from solution analysis. Detailed analysis of these phases was not attempted in this study. 100.0 U U U I
n \
10.0
29
0.1 0.1
1.o
10.0
100.0
PCT Na release (g/m2)- Quenched Figure 5 . PCT Normalized Na releases from quenched versus CCC glasses Based on the conservative limit of 2 g/m2, only one glass (SBW01-40-10) passed the limit on both quenched and CCC glasses. This is one of the glasses that are crystal free in both quenched and CCC forms; it contains 20.8 mass% Na20 and 3.0 mass% KzO. These results indicate that optimization may increase the waste loading of acceptable glass above 40 mass%.
Environmental Issues and Waste Management Technologies VIII
343
LA W glasses
Two glasses were formulated and tested to demonstrate the applicability of phosphate glass as a waste form for Hanford LAW. The glasses were prepared following the same procedure as for the glasses with SBW. The waste composition with the highest SO3 concentration was taken from Muller et al. [6]. Table 4 summarizes the compositions for waste, the additives, and the resulting glass for the glasses with two different additive compositions. The glasses were formulated to have 20 mass% Na2O in a glass that results from a waste loading of 25.8 mass%. As shown in Table 4, the PCT normalized releases of these glasses are well below the 2 g/m2limit. These glasses did not show any indication of sulfate segregation at the melt surface and did not form crystals upon cooling. Table 4. Composition of a Hanford LAW waste used in this study as well as additives, resulting glasses, and results on PCT and density Oxide Pretreated AZ LAPG-T1 LAPG-T2 (mass%) LAW-2 Additive Glass Additive Glass A1203 7.20 1.85 10 9.28 Si02 0.60 0.16 0.16 Cf203 0.81 0.2 1 0.2 1 F 0.9 1 0.23 0.23 Fe203 0.00 30 22.27 20 14.85 K20 2.93 0.75 0.75 Na2O 77.65 20.00 20.00 p205 0.49 70 52.10 70 52.10 so3 9.4 1 2.42 2.42 Sum 100.00 100 100.00 100 100.00 7 day PCT normalized release at 90°C, g/m2 Na 1.281 0.640 JK 0.748 0.320 P 0.776 0.309 Density, g/cm3 2.862 2.783 CONCLUSIONS Phosphate glasses can contain at least 5 mass% SO3 without forming a sulfate layer under laboratory melting conditions using dry raw materials. The preliminary formulation in this study identified SBW01-40-10 as a candidate phosphate glass for vitrification of INEEL SBW, which doubles the waste loading of a typical borosilicate glass, thus resulting in 21 mass% Na2O and 3 mass% K20. This glass did not form crystals when quenched or CCC-treated. Its PCT Na release was less than 2 g/m2 from both quenched and CCC-treated samples. High sulfbr incorporation and good chemical durability were also demonstrated for glasses formulated for Hanford LAW. This study indicates that phosphate glasses can increase the loading of wastes with a high content of sulfur to the point where the waste loading is limited by chemical durability.
344
Environmental Issues and Waste Management Technologies VIII
ACKNOWLEDGEMENTS The authors are grateful to Mike Schweiger and Jarrod Crum for their technical support, Gordon Xia for his critical review of the manuscript, and Wayne Cosby for editorial support. Pacific Northwest National Laboratory is operated for the U. S . Department of Energy by Battelle under Contract DE-AC06-76RLO1830. REFERENCES 113 D.E. Day, 2.Wu, C. S . Ray, and P. Hrma, “Chemically Durable Iron Phosphate Glass Wasteforms,” J Non-Cyst. Solids, 241 1- 12 ( 1998). [2] G. K. Marasinghe, M. Karabulut, C. S . Ray, D. E. Day, D. K. Shuh, P. G. Allen, et. al., “Properties and Structure of Vitrified Iron Phosphate Nuclear Waste forms,” J Non-Cyst. Solids, 263 & 264 146-54 (2000). [3] G. K. Marasinghe, M. Karabulut, X. Fang, C. S . Ray, D. E. Day, D. L. Caulder, et. al., “Vitrified Iron Phosphate Nuclear Waste forms Containing Multiple Waste Components,” Ceramic Transactions, 107 1 15 - 122 (2000). Transactions, American Ceramic Soc. 107, 115-22 (2000). [4] G. K. Marasinghe, M. Karabulut, X. Fang, C. S . Ray, and D. E. Day, “Iron Phosphate Glasses: An Alternative to Borosilicate Glasses for Vitrifying Certain Nuclear Wastes,” Ceramic Transactions, 119 36 1-368 (200 1). [5] J. D. Vienna, W. C. Buchmiller, J. V. Crum, D. D. Graham, D.-S. Kim, B. D. MacIsaac, M. J. Schweiger, D. K. Peeler, T. B. Edwards, I. A. Reamer, and R. J. Workman, “Glass Formulation Development for INEEL Sodium-Bearing Waste,” PNNL-14050, Pacific Northwest National Laboratory, Richland, WA, 200 1. [6] I. S . Muller, A. C. Buechele, and I. L. Pegg, “Glass Formulation and Testing with RPP-WTP LAW Simulants - Final Report,” VSL-0 1R3560-2, Vitreous State Laboratory, The Catholic University of America, Washington D. C., 2001. [7] J. Christian, “Composition and Simulation of Tank WM-180 Sodium-Bearing Waste at the Idaho Nuclear Technology and Engineering Center,” INEELEXT-0 100600, Idaho National Engineering and Environmental Laboratory, Idaho Falls, ID, 2001. [8] S . L. Marra and C. M. Jantzen, “Characterization of Projected DWPF Glasses Heat Treated to Simulate Canister Centerline Cooling (U), WSRC-TR-92-142, Rev. 1, Westinghouse Savannah River Company, Aiken, SC, 1993. [9] American Society for Testing and Materials (ASTM), “Standard Test Method for Determining Chemical Durability of Nuclear Waste Glasses, The Product Consistency Test (PCT).” ASTM-C-1285-97, in Annual Book of ASTM Standards, Vol. 12;01, Philadelphia, PA, 1998. [lO] D.-S. Kim, D. K. Peeler, and P. Hrma, “Effects of Crystallization on the Chemical Durability of Nuclear Waste Glasses,” Ceramic Transactions, 61 177-185 (1 995). [l 13 B. J. Riley, P. Hrma, J. Rosario, and J. D. Vienna, “Effect of Crystallization on High-Level Waste Glass Corrosion,” Ceramic Transactions,132 257-265 (2002). [12] R. K Brow, “Nature of Alumina in Phosphate Glass: I, Properties of Sodium Aluminophosphate Glass,” J Am. Ceram.Soc., 76 [4] 913-18 (1993).
Environmental Issues and Waste Management Technologies VIII
345
This page intentionally left blank
SOLUBILITY OF HIGH C H R O m NUCLEAR WASTES IN IRON PHOSPEUTE GLASSES Wenhai Huang" ,Cheol-Woon Kim , Chandra S. Ray, and Delbert E. Day Materials Research Center and Department of Ceramic Engineering University of Missouri-Rolla 1870 Miner Circle, Rolla, MO 65401 ABSTRACT A simulated waste containing 4 mass % Cr2O3, was designed by combining three highest Cr203-containing HLW at Hanford, WA which contain the highest Cr203 content of the HLW at Hanford. Iron phosphate (P) glasses were investigated as an alternative wasteform for vitriwg this simulated high chrome waste. The solubility limit of Cr2O3 in IP glass was determined to be 2.6 mass % CrzO3. Iron phosphate glasses with a waste loading of up to 75 mass % had an extremely good chemical durability. The total elemental mass release fiom these chemically durable iron phosphate wasteforms (PCT) at 75 mass%waste loading was only 1.86 @m2.
-
INTRODUCTION Some of the high level nuclear waste (HLW) at Hanford WA, whose amount totals about 12.3 million kilogram (Mkg), contains up to 4.25 mass% chromium oxide (Cr203),. The solubility of Cr203 in the alkali alumino-borosilicate(AABS) glasses, currently approved for vitrifying nuclear wastes, is -= 1 mass%. In a study by Kirkbrid, the solubility of Cr2O3 in the M S glasses was estimated to be no more than 0.5 mass%.[21This means that HLW of high chrome content must be diluted by a factor of 4 to 8 for the A A B S glasses, thus, increasing the wasteform volume and vitrification cost. Vitrification of all the high chrome HLW at Hanford in the AABS glasses has been estimated to produce 26.5 to 32.3 Mkg of wasteform, which will require about 9,720 to 12,110 m3 of space (volume) for permanent repository. Reducing the volume of vitrified waste is one means of reducing the overall cost. The vitrification of these high chrome waste in iron phosphate glasses, as an alternative to A A B S glasses, has been investigated in the present study. The maximum waste loading for a simulated waste, blended fiom three high chrome wastes at Hanford, has been determined by optimizing the chemical durability of * On leave fiom: Tongji University, Shanghai 200092, PRC. Email: whhuang@ihw,corn.cn To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
347
the wasteforms. Based on the measured maximum waste loading, the volume of the wasteform that will be produced for disposing of all the high chrome HLW at Hanford in iron phosphate glasses has been estimated.
EXPERIMENTAL PROCEDURES Simulated Waste Composition The HLW at H d o r d is divided into 17 different general compositions (called clusters), which contain fiom 0.1 1 to 4.25 mass % Cr203['I. The combined Table 1. Composition, mass %, of the high chrome nuclear waste for clusters #7,8 and 14 at Hanford, WA., the total mass (Mg) of waste in each cluster, and the composition, mass %, of the simulated
mass of all these HLW is estimated to be about 12.3 Mkg. The typical composition of the waste in clusters #7, 8, and 14, which are the three highest CrzO3- containing
348
Environmental Issues and Waste Management Technologies VIII
wastes among the 17 clusters at Hadord is given in Table 1, along with the estimated amount of each waste (bottom line). A simplified waste composition that was a blend of the waste in clusters #7, 8, and 14 was designed. First, the composition, in the column labeled as “Blend” in Table 1 was calculated for each oxide according to the following equation: Mass% oxide content of “X” in the “Blend” (column 5 , Table 1) = [ 1/1334] (Amount of a cluster ‘Y)(Mass% oxide content of “X” in the cluster “i”) where i = 7,8, &14 and 1334 is the combined mass for the three clusters #7,8, and 14. The number of individual (oxide) components in the waste was reduced by combining different oxides that play more or less similar roles in a glass, as shown in the column labeled “combined blend” in Table 1. For example, Na20 and K20 are known to play similar roles in a glass, so the Na20 (25.03) and K20 (0.67) were added together (25.70). This sum was then used as the mass fraction for the major component, in this case Na20, in the combined blend. The composition of the “combined blend” was further simplified by rounding to the closest whole percent, see right hand column in Table 1 (as the “simplified blend (W)”). It is to be noted that CaF2, instead of CaO, was used in the simplified blend to account for the fluorine present in the actual waste. The composition of the simulated waste, W, contains 4 mass% Cr2O3, which is close to the highest Cr203 content (4.25 mass %) waste # 14 at Hanford. Glass Melting Batches that produced 100 g of glass of the general compositions x(W).(100x)P205, mass%, with x ranging from 35 to 80 at 5 % intervals were prepared by Table 2. Composition, mass %, of the simulated high chrome waste, and glassy iron phosphate
mixing appropriate amounts of the oxides listed in the right hand column of Table 2. Note that the P2O5 required was added as sodium phosphate. These batches were
Environmental Issues and Waste Management Technologies VIII
349
melted in fireclay crucibles between 1100 “C (waste content 35%) and 1300 “C (waste content 80%) for 2 h in air. Each melt was stirred two times at 30 min interval with a silica glass rod to aid homogenization. The melts were cast into a steel mold to form rectangular glass bars 1 cm x 1 cm x 5 cm. The glass bars were annealed at 520 to 540 “Cfor 4 h and slowly cooled (overnight) in the annealing furnace to room temperature. Corrosion of the fireclay crucible by the iron phosphate melts was hardly detectable. Each glass was identified as “IP(x)W”,where “IP” stands for iron phosphate and “xYyis the mass% of simulated waste, W. Thus, a designation as IP70W means an iron phosphate glass with a simulated waste (W) loading of 70 mass %. The batch composition for two noteworthy glasses, IP70W and IP75W, are given in Table 2 along with the raw materials used in the batch. The Fe203 content in these phosphate glasses ranged fkom about 3 to 7 mass percent. Powder x-ray diffkaction analysis (XRD) was used to determine the amorphous or crystalline character of the annealed samples.
-
Chemical Durability The chemical durability for these high chrome iron phosphate glasses was measured by three different methods: (1) dissolution rate @R) in de-ionized water (DIW)[?, (2) vapor hydration test (VHT)[41, and (3) product consistency test (PCT)[’. The last two methods, namely, VHT and PCT, are recommended by DOE for determining the chemical durability, although our previous measurements show that the DR value are always in good agreement with the results fkom VHT and PCT. Dissolution Rate(DR) measurements: Specimens 1 cm x 1 cm x 1 cm in size were cut from the annealed glass bars. Their surfaces were polished progressively using 240,600 and 800-grit Sic paper with oil as the cooling agent. After polishing, the specimens were rinsed with acetone and DIW, dried at 90 “C for 30 minutes and cooled to room temperature. Their dimensions and weight (k 0.01 mg) were measured. The rinsing procedure was followed before each sample was weighed for the corrosion test. Each specimen was suspended by a thin rayon thread and immersed in 100 ml of DIW contained in a plastic bottle. The plastic bottle was placed in an oven at 90 “C. After a designated time interval, the sample was removed from the bottle, rinsed with DIW, dried at 90 “C,cooled and weighed(+ 0.Olmg). The dissolution rate(DR) of the bulk sample was calculated fkom the measured weight loss divided by the sample surface area and immersion time. Duplicate samples were measured and the average of the two values is reported herein. The DR values measured after an immersion time of 64 days are shown in Fig. 1 for waste loadings between 35 and 80 mass %. Vapor Hydration Test (WIT): Bulk samples of approximate size 1 cm x 1 cm x 0.15 cm were used for the VHT test. After polishing each surface with 600-grit Si c
350
Environmental Issues and Waste Management Technologies VIII
paper, the sample was washed with flowing DIW, washed dtrsonically in ethanol, and dried at 90 "C.The dimensions and mass of each sample were measured, whereupon, it was suspended using a thin teflon thread inside a stainless steel vessel that contained 0.25 ml DIW. The vessel containing the sample was placed in an oven at 200 "C for 7 days. A cross-sectional view of the IP70W and IP75W glasses, as viewed with an optical microscope after the VHT, is shown in Fig. 2. The corrosion rate of the glass (g/m2/day)was calculated fiom the initial and fmal (uncorroded part) thickness, weight, and size of the sample and the test duration (7days). The thickness of the sample was measured using an optical microscope to an accuracy of kO.002 mm . JaN-mW?*Q!#=
*-a-,
.................. ................................................ @
',
.....................................
I
........................
Fig.1 DissoIution rate of vitrified iron Fig.2 Appearance of IP7OW and IP75W phosphate wasteforms after 64 days in deionized samples after VHT for 7 day at 200 "C water at 90 OC
Product Consistency Test (PCT): The PCT test was conducted on duplicate powder samples as per the procedures in ASTM C-1285-94. The glass powder sample (d = 75 to 150 pm) was ultrasonically washed with DIW three times to remove any small particles that adhere to the surface of the large particles. The sample was then washed twice with ethanol to dissolve any organic materials, and dried at 90 "C overnight. Exactly 1.5 g ( M.0lmg ) of glass powder was mixed with 15 ml of DIW that was contained in a Teflon vessel. The vessel was sealed and placed in an oven at 90 "C for seven days. After completion of the PCT, the leachate was filtered and its pH value was measured.The concentration of ions in the filtered leachate was measured by ICP-ES. The normalized elemental mass release (g/m2) was calculated from the concentration (g/m3) and mass fraction of each element in the glass and the ratio of sample surface area to the volume of DIW (-1.8~10~ m" in the present study). The PCT was measured only for the IP70W and IP75W glasses in the present study. The pH value of the leachate was 9.03 and 9.51 for P70W and IP75W glasses respectively. The normalized total elemental mass release as measured by ICP-ES is given in Table 3. Other Property Measurements:
Environmental Issues and Waste Management Technologies VIII
351
The density(d), glass transition temperature (Tg), and the average thermal expansion coefficient (abetween 100 to 400 "C) for the samples were measured using standard laboratory techniques and given in Table 4. The estimated error of the measurement of d, Tg,and a was < f4 %, f 3 "C and f 5 % respectively.
--
hlormalized IP70W IP75w mass Sample I Sample Sample I Sample
&
I c -
.
c - _ . U
Total (g/m2)
Density ( g/cm31 Tg(C) a(xlO"/"C)
0
-
2
I
0
.
3
I
0
-
4
I
0
.
5
28 (Degree)
I
0
.
B
m o p
70W3OP 65w35p
'
1
1
crp,
7wSp
3
o
.
7
Fig.3 XRD for the Cr203containing asmade iron phosphate wasteforms. Presence of small amounts of crystalline Cr203is evident in the samples with >or = 70 mass % waste loading.
1.35
1.32
1.87
1.85
2.69
2.71
2.75
2.77
2.84
2.84
2.86
2.89
2.92
2.97
496 105
495
495 108
498 109
500 113
478 125
463 126
455 129
461 131
464 135
106
1
0
RESULTS AND DISCUSSION The vitrification of this simulated high chrome waste was achieved by simply adding a source of P2O5 to the waste. No other addition was needed. All the wasteforms containing I 65 mass % waste were glassy. Those containing 2 70 mass % waste contained a small amounts ( 4 . 5 mass % ) of crystalline Cr2O3, see Fig. 3. On the basis that the IP65W composition contains a total of 2.6 mass % Cr203, but is free of crystalline Cr2O3, it is concluded that the solubility limit of Cr2O3 in these iron phosphate melts ( for 2 h melting ) is about 2.6 mass %. This is at least 2.6
352
Environmental Issues and Waste Management Technologies VIII
times larger than the amount of Cr2O3 that can be dissolved in borosilicate melts, (solubility of Cr203 in AABS glass is c 1 mass % 16]) . A quantitative estimate for the amount of crystalline C1.203 present in the samples containing > 65% waste was made by comparing the intensity (cps) of the
Fig. 4 Content of Cri03 crystal (b) calculated fiom the calibration curve in (a) of several as-made iron phosphate wasteforms. The amount of Cr203present in the batch for the respective wasteforms is also shown (b).
most intense XRD peak (28 = 24.5") of Cr2O3 with a previously determined calibration curve. This calibration curve was determined fiom XRD measurements of samples prepared by mixing known amounts (1 to 10 mass %) of crystalline Cr203 with powdered of IP65W glass, which was shown to be amorphous by XRD. The concentration of crystalline Cr2O3 in the mixtures was then plotted as a function of the relative intensity of the XRD peak at 24.5" for the mixture compared to that for pure IP65W glass (L /Ig) to obtain the calibration curve shown in Fig. 4 (a). The concentration of crystalline Cr2O3 detennined by this method in the melts containing = or > 60% waste is shown in Fig. 4 (b). The Cr2O3 content in the as-made batch is also shown for comparison. Clearly, the amount of crystalline Cr2O3 in the samples increased with increasing waste loading above 65%, but was only about 1.3% for the IP8OW sample. As shown in Fig. 1, the dissolution rate in DIW at 90 OC for all the iron phosphate wasteforms containing 55 to 80 mass % waste is upto 50 times smaller than that for soda-lime-silica window glass, even though, those wasteforms containing 70 to 80% waste contain near1 20 mass % Na2O. The IP65W wastefkom, had the lowest dissolution rate (5.9 x 10-IT g/cm2/min)or highest chemical durability in water at 90 "C. The presence of a small amount (up to 1.5 mass%) of crystalline Cr203 in these glasses does not appear to adversely affect their chemical durability to any measurable degree. The excellent chemical durability indicated by the DR measurements (Fig. 1) for these iron phosphate glasses is confinned by the VHT (Fig. 2) and PCT (Table 3)
Environmental Issues and Waste Management Technologies VIII
353
results. As shown in Fig. 2, the LP70W wasteform, which contains about 0.8 mass % crystalline Cr2O3, does not show any visible or detectable corrosion layer on its surface after 7 days in DIW at 200 “C.The average total normalized mass release for IP70W (Table 3) fiom PCT is only 1.33 g/m2.For the IP75W sample, which contains about 1.2 mass% crystalline Cr2O3, a thin corrosion layer only 11 pm thick was observed on the surface after the VHT (Fig. 2). Based on this corrosion layer, the corrosion rate for this glass was calculated to be only 3.3 g/m2/day. Corrosion rates of 140 to 196 g/m2/day have been reported [71 for the LAW-33 and LD6-5412 borosilicate glasses. The average total normalized mass release (1.86 g/m2) for the IP75W sample is a little higher than that of IP70W (Table 3), but is still quite low. CONCLUSIONS The present results clearly show that the high chrome (up to 4.5 mass %) waste at Hanford can be vitrified by simply adding about 30 mass % phosphate to the waste and melting the mixture at 1250 OCfor 2 h. The simulated blend of the three high chrome wastes at Hanford used in the present study are about 10 mass % of the total waste at Hanford. The solubilitylimit of Cr203 in these iron phosphate melts is about 2.6 mass%, compared to < lmass % in common borosilicate glasses. Iron phosphate wastefonns having waste loading of 55 to 75% of the high chrome HLW have an exceptionallyhigh chemical durability (as determined by VHT and PCT). ACKNOWLEDGMENT This work was supported by Department of Energy (DOE) under EMSP grant DOE DE-FG07-96ER45618. REFERECES ‘High- Level Waste Melter Study Report, PNNL-13582, July 2001. 2R.A. Kirkbride, “Tank farm contactor operation and Utilization Plan (TWRSOUT)”, HNF-SD-WM-SP-012, Rev.2, CH2M Hill Hanford Group, Inc., Richland Washington, 2000. 3X. Yu, “Properties and Structure of Sodium-Iron Phosphate Glasses,” Journal of Non-Crystalline Solids, 215 21-3 1 (1997) 4PNNL Technical Document, “Vapor Hydration Test Procedure, GDL-VHT” ’ASTM Standard Test Method for Determining Chemical Durability of Nuclear, Hazardous ,and Mixed Waste Glasses: The Product Consistency Test, C 1285-97 6X.Feng et al, “Glass Optimization for Vitrification of Hanford site Low-Level Tank Waste”, PNNL-10918, Pacific Northwest Laboratory, Richland, WA, 1996 7A. Jiricka et al, “The Effect of Experimental Conditions and Evaluation. Techniques on the Alteration of Activity Glasses by Vapor Hydration,” Journal of Non-Crystalline Solids,292 25-43 (2001).
354
Environmental Issues and Waste Management Technologies VIII
DEVELOPMENT OF A SAMPLING METHOD FOR QUALIFICATION OF A CERAMIC HIGH-LEVEL WASTE FORM T. P. O'Holleran Argonne National Laboratory - West P. 0. Box 2528 Idaho Falls, ID 83403-2528
K. J. Bateman Argonne National Laboratory - West P. 0. Box 2528 Idaho Falls, ID 83403-2528
ABSTRACT A ceramic waste form has been developed to immobilize the salt waste stream from electrometallurgicaltreatment of spent nuclear fuel. The ceramic waste form was originally prepared in a hot isostatic press (HIP). Small HIP capsules called witness tubes were used to obtain representative samples of material for process monitoring, waste form qualification, and archiving. Since installation of a fullscale HIP in existing facilities proved impractical, a new fabrication process was developed. This process fabricates waste forms inside a stainless steel container using a conventional h a c e . Progress in developing a new method of obtaining representative samples is reported. INTRODUCTION Electrometallurgica1 treatment of spent nuclear fuel produces two waste streams: metal fiom cladding hulls and salt from electrorefining. A ceramic waste form has been developed to immobilize the salt waste. A hot isostatic press (HIP) was originally used to prepare the ceramic waste form. Small, easily fabricated HIP capsules called witness tubes were shown to be a practical way to obtain representative samples of cerarnic waste form material for process monitoring, waste form qualification, and archiving.' However, the HIP was found to be impractical for production of full-scale waste forms. A ''pressureless consolidation" process was developed to replace the HIP. This process uses a conventional furnace to fabricate waste forms inside a stainless steel container that becomes part of the waste form. A new method of obtaining representative
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management TechnologiesVIII
355
samples for process monitoring, waste form qualification, and archiving must be developed and integrated into the full-scale production process. The objective of this work is to develop and qualiQ a standardized method for obtaining samples for product consistency testing during production of the pressureless consolidated ceramic waste form. The effort is divided into two phases. During Phase I, candidate sampling methods will be developed using small "laboratoryffscale waste forms to investigate materials interaction issues and develop the methodology. The primary goal of Phase I is to specify the sampling method to be used on full-scale waste forms. A secondary goal is to identiQ a back-up methodology to reduce technical risk. All experiments during Phase I will be performed using non-radioactive materials. During Phase 11, the sampling methodology developed in Phase I will be tested with full-scale production equipment. This testing will be performed in conjunction with process development so that the candidate methodology will emerge from Phase I1 completely integrated with the production process. Phase I1 testing will involve extensive sampling in order to develop the data base necessary to establish the statistical relation between the properties of product consistency samples and the production waste form material. Currently, we are engaged in Phase I of this effort. This paper reports progress to date, and outlines future plans. TECHNICAL APPROACH The Waste Acceptance System Requirements Document (WASRD)2 requires that the Product Consistency Test (PCT)3, process knowledge, or a combination of the two be used to demonstrate that waste forms meet specifications during production. This approach has been adopted for the ceramic waste form produced during electrometallurgical treatment of metallic sodium bonded spent nuclear fuel as described in the Waste Form Compliance Plan4 While waste form qualification during production will rely heavily on process knowledge, some sampling and testing will be conducted on a statistical basis. The samples required are not large (20 - 40 g) however, the size and weight of the production scale waste form (about 0.5 m in diameter and 1 m tall and weighing up to 450 Kg) makes sampling problematic. As the process is currently laid out, there is no room in the hot cell for equipment large enough to obtain samples of waste form material by conventional methods such as cutting or core drilling. Sampling activities could take place at various times during the production process. Basically, the points during the process where sampling activities could occur can be defined based upon whether the waste form material is at processing
356
Environmental Issues and Waste Management Technologies VIII
temperature (hot) or has cooled to ambient temperature (cold). Furthermore, sampling can be defined as a two step process: probing the waste form to isolate a small amount of material, and physically retrieving the material. For example, the Defense Waste Processing Facility obtains samples of high-level waste glass by inserting a small cup into the molten glass pour stream (probing), then removing the cup after collecting enough molten glass for analysis (retrieving).' This sampling activity would be described as hot probing and hot retrieving. On the other hand, the West Valley Demonstration Project obtained glass samples after the glass had cooled by reaching into the canister with remote manipulator (probing) and removing a shard of glass (retrieving)? This sampling activity would be described as cold probing and cold retrieving. Using these concepts, four types of potential sampling activities were defined for the ceramic waste form production process as shown in Table I. Table I. Potential sampling activities defined in terms of the waste form temperature at each of the sampling steps, along with some waste form material properties that could be of interest Probe Retrieve Material Properties of Interest Hot Hot Viscosity, Rheology Hot Cold Viscosity, Rheology, Chemical (interactions), Mechanical Cold Hot Viscosity, Rheology Cold Cold Chemical (interactions), Mechanical EXPERIMENTAL To test sampling methods, experimental waste forms were produced from a 3/1 (by weight) ratio of salt-occluded zeolite A to borosilicate glass frit. The salt was a eutectic mixture of LiCl and KC1, containing simulated (non-radioactive) fission product salts. This mixture of powders was placed into a 500 ml stainless steel beaker. A stainless steel weight slightly smaller in diameter than the inside of the beaker and 4.5 cm thick was placed on top of the powder charge to provide some pressure to assist in consolidation. The beaker was placed into a pot furnace, heated to 915' C, and held for six hours. Three of the potential sampling activities described in Table I have been tested so far. A hot probe - hot retrieve sampling method was devised based on a method for obtaining soil sample^.^ In this method, the hot probe step involves removing the steel weight and inserting a thin-walled stainless steel tube into the waste form material at the end of the heat cycle while still at maximum processing
Environmental Issues and Waste Management TechnologiesVIII
357
temperature. The steel tube had an outside diameter of 1.9 cm, an inside diameter of 1.7 cm, with a 60"taper on the outside of one end. The outside of the tube was coated with boron nitride mold release. The hot retrieve step is removing the stainless steel tube containing a hot sample of material. Two variations of hot probe - cold retrieve methods were attempted. The first was simply a variation of the hot probe - hot retrieve method described above, where the stainless steel tube was not removed until after the waste form had cooled. In the second method, a cavity was drilled into the underside of the stainless steel weight. This was to allow material to flow into the cavity during the heat cycle (the hot probe step). The resulting protrusion of waste form material was to be mechanically removed after the waste form had cooled and the weight was removed (the cold retrieve step). For the cold probe - cold retrieve method, a hole 1.6 cm in diameter was drilled completely through the stainless steel weight and coated with boron nitride mold release. After the stainless steel beaker was filled with starting material and the modified weight placed on top, the hole was filled about half way with additional starting material (the cold probe step). A steel rod the same length as the thickness of the weight and slightly smaller in diameter than the hole, coated with boron nitride mold release, was then inserted into the hole. The purpose of the rod was to apply the same pressure to the potential sample as was applied to the bulk of the waste form material. When the waste form had cooled after the heat cycle, the resulting protrusion was to be mechanically removed (the cold retrieve step).
RESULTS AND DISCUSSION The hot probe - hot retrieve sampling method using a stainless steel tube failed to produce a sample. The tube only penetrated the waste form material about a centimeter, and only with great difficulty. The tube was easily removed, but no waste form material remained in the tube. This method was abandoned after several attempts. The hot probe - cold retrieve method using the stainless steel tube also failed to produce a sample. After cooling, the tube was adhering to the bulk waste form, but it was easily broken free. However, no waste form material was retained in the tube. A curious ring structure remained in the annular depression in the waste form left by the tube. Scanning electron microscopy with energy dispersive X-ray spectroscopy revealed that this ring structure consisted essentially of a layer of oxidized stainless steel containing remnants of the boron nitride mold release (see Figure 1). Iron and chromium were also found to have diffused about ten to
358
Environmental Issues and Waste Management Technologies VIII
twenty microns into the waste form. These results indicate that the boron nitride mold release was ineffective in preventing the steel fiom adhering to the waste form. Separation occurred by mechanical failure within the oxide layer that had formed on the surface of the stainless steel.
Figure 1. Back scattered electron image of the outer portion of the ring structure left behind when the stainless steel tube was removed from the waste form, showing the boron nitride and oxidized stainless steel layers. The second hot probe - cold retrieve method using the modified weight failed to produce a useful sample. Only a small amount of material penetrated into the cavity during the heat cycle. This material broke fiee and remained in the cavity when the weight was removed, but was easily dislodged from the cavity. The waste form material that was retrieved was visibly more porous than the rest of the waste form, suggesting poor consolidation .from lack of pressure in the immediate vicinity. The viscosity of the mixture of molten glass and salt occluded zeolite (or sodalite after the phase transition) at processing temperature is apparently too high to allow the amount of flow needed for this method. The cold probe - cold retrieve method succeeded in producing a sample. When the modified weight was removed from the waste form, the protruding waste form material broke off fiom the bulk waste form, and was retained in the cavity. The sample was removed in one piece by tapping the steel rod to drive the sample out of the cavity.
Environmental Issues and Waste Management TechnologiesVIII
359
In order to be useful for waste form qualification, the sample material must be representative of the bulk waste form material. Properties need not be identical, as long as differences are consistent and preferably small. Measurements of the heights of the weight and the steel rod were made before and after heat treatment to compare the consolidation of the bulk waste form and the sample. If a “consolidation factor” is defined as the ratio of the green height to the fired height, then the bulk waste form achieved a consolidation factor of 1.83, compared to a consolidation factor of 1.77 for the sample. This suggests that the sample material did not achieve quite the fired density of the bulk waste form. Density measured by helium pycnometry (which does not measure open cell porosity) confirmed that the sample material was slightly less dense than the bulk waste form material. The bulk material had a density of 2.25 g/cm3, while the sample had a density of 2.17 g/cm3. Both the consolidation factor and the density of the sample material are only 3% less than the corresponding values measured for the bulk waste form material, which is very near the uncertainty in the measurement and therefore considered acceptable. These slight differences may be attributable to friction between the steel rod and the cavity walls that reduces the effective pressure applied to the sample material by the steel rod. If so, the small differences between the sample material and the bulk waste form material could presumably be eliminated by simply lengthening the steel rod. X-ray powder diffkaction was also performed to compare the phase composition of the sample to the phase composition of the bulk material. The results showed sodalite as the primary crystalline phase, with halite and nepheline as minor phases for both materials. This is the expected phase composition of the ceramic waste form. Most importantly, the phase compositions of the sample and bulk materials are virtually identical as shown in Figure 2.
360
Environmental Issues and Waste Management Technologies VIII
10
20
30 40 Degrees 2 theta
50
60
Figure 2. X-ray powder diffraction patterns of (top) bulk waste form material, and (bottom) sample material. CONCLUSIONS A sampling method has been demonstrated that can be used to obtain waste form qualification samples of ceramic waste form during production. The method is fairly simple, requires no large scale equipment, and should have little impact on the overall process. Samples obtained by this method axe representative of the bulk material as determined by density and phase composition. Thus data obtained from such samples will be acceptable for waste form qualification and process verification. FUTURE PLANS Experiments with the cold probe - cold retrieve sampling method will continue in order to generate additional material for testing and characterization. The Product Consistency Test3will be used to compare leach behavior of test material and bulk material. The use of different materials for the steel weight will also be investigated. Phase I1 activities will begin when full scale production equipment becomes available for testing.
Environmental Issues and Waste Management TechnologiesVIII
361
ACKNOWLEDGEMENTS Argonne National Laboratory is operated for the U. S. Department of Energy by the University of Chicago. This work was supported by the Department of Energy, Nuclear Energy Research and Development Program, under contract no. W-31-109-ENG-38. The authors wish to thank Mr. T. DiSanto, Mr. E. A. Reseigh, and Dr. S. M. Frank for their assistance.
REFERENCES ‘T. P. O’Holleran, S. G . Johnson, and K. J. Bateman, “Ceramic Waste Form Qualification Using Results fiom Witness Tubes,” Radioactive Waste Management and Environmental Restoration, to be published. 2U.S. Department of Energy, “Waste Acceptance System Requirements Document,” DOERW-035 1, Revision 03, DOC ID: E00000000-00S11-170800001 REV 03 (1999). 3American Society for Testing and Materials, “Test Methods for Determining Chemical Durability of Nuclear Waste Glasses: The Product Consistency Test (PCT),” C1285-97, Annual Book of ASTM Standards, 12.01 (1998). 4t’Waste Form Compliance Plan for the Waste Forms Jiom Electrometallurgical Treatment of Spent Nuclear Fuel,” Argonne National Laboratory - West Document No. F0000-0031-ES, REV. 00 (1999). 5N.E. Bibler, J. W. Ray, T. R. Fellinger, 0. B. Hodoh, R. S. Beck, and 0. G. Lien, “Characterizationof the Radioactive Glass Currently Being Produced by the D WPF at Savannah River Site,” Waste Management ’98 Proceedings (1998). 6V. A. DesCamp and C. L. McMahon, “VitrificationFacility at the West Valley Demonstration Project,” Topical Report DoElNE144139-77 (1996). 7American Society for Testing and Materials, “Standard Practice for ThinWalled Tube Sampling of Soils for Geotechnical Purposes,” D 1587-00, Annual Book of ASTM Standards, 04.08 (2001).
362
Environmental Issues and Waste Management TechnologiesVIII
MICROWAVE HEATING FOR PRODUCTION OF A GLASS BONDED CERAMIC HIGH-LEVEL WASTE FORM T. P. O'Holleran Argonne National Laboratory - West P. 0. Box 2528 Idaho Falls, ID 83403-2528 ABSTRACT Argonne National Laboratory has developed a ceramic waste form to immobilize the salt waste fiom electrometallurgical treatment of spent nuclear fuel. The process is being scaled up to produce bodies of 100 Kg or greater. With conventional heating, heat transfer through the starting powder mixture necessitates long process times. Coupling of 2.45 GHz radiation to the starting powders has been demonstrated. The radiation couples most strongly to the salt occluded zeolite powder. The results of these experiments suggest that this ceramic waste form could be produced using microwave heating alone, or by using microwave heating to augment conventional heating. INTRODUCTION During much of the ceramic waste form development effort, a hot isostatic press (HIP) was used to consolidate the powder starting materials. The HIP applies heat and pressure to melt the glass binder and consolidate the powder into a dense solid body. This fabrication route was necessary when the desired end product was a glass-bonded zeolite. Processing temperatures had to be kept relatively low to avoid transforming the zeolite to sodalite, with concurrent release of excess salt. The low processing temperature required the use of pressure to achieve densification. With the selection of glass-bonded sodalite as the frnal waste form (which necessitated a reduction in salt loading), higher processing temperatures could be used for consolidation. A t . the higher processing temperatures used to fabricate the glass-bonded sodalite waste form, densification could be achieved without the application of pressure. The current baseline
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management TechnologiesVIII
363
process for fabricating the glass-bonded sodalite waste form achieves densification using heat only, and is called "pressureless consolidation." As the size of the waste form is scaled up from laboratory scale (on the order of a few centimeters in diameter) to full scale (about 1/2 meter in diameter), heating material in the center of the powder charge becomes more difficult. Thermal conductivity through the loose powder is relatively low, so as the size of the waste form increases, processing time must also be increased to fully densify the material. In the HIP process, this problem was partially alleviated by the application of pressure, since pressure drives densification, and thermal conductivity increases with density. However, since the baseline process for fabricating the glass-bonded sodalite waste form relies solely on temperature to achieve densification, processing times for the full-scale waste form may become excessive. Commercially available microwave ovens operating at 2.45 GHz and at power levels fiom 450 to 850 W have been used to heat zeolites and other aluminosilicates.1s2One advantage of microwave heating is that heat is evolved within the load as microwave energy penetrating the material is absorbed. This results in rapid heating of the load. At 600 W, complete melting of a 10 g, 2.5 cm diameter pellet of Linde 4A was achieved in less than 2 min.' It has been proposed that the initial heating of zeolite 4A (below about 400" C) depends on the degree of hydration, and that dehydrated zeolite could be difficult to heat with microwave radiation alone? The zeolite material used to fabricate the glassbonded sodalite waste form contains essentially no water (< 0.5 wt. %), but does contain approximately 2.5 molecules of occluded chloride salt per pseudo unit cell. The starting material for the glass-bonded sodalite waste form also contains 25 wt. % borosilicate glass. The microwave heating behavior of these materials, alone or in combination, has never been reported in the open literature. The objectives of this work were therefore to determine whether microwave energy would couple sufficiently with the starting material for the glass-bonded sodalite waste form to cause heating, and, if so, to determine if microwave heating could be applied to a production process. EXPERIMENTAL In order to heat materials to high temperatures in a conventional microwave oven, thermal energy generated within the load must not be allowed to escape freely into the microwave cavity. For these experiments, an insulating chamber with internal dimensions 5 cm by 5 cm by 7.5 cm high was constructed from 2.5 cm thick Zircar@ECO-1200B refractory insulating board. When inserted, the 50
364
Environmental Issues and Waste Management Technologies VIII
ml high purity alumina crucibles used for these experiments nearly fill this
chamber. Experiments were performed in two phases. In phase 1, the objective was simply to determine if glass-bonded sodalite starting materials would couple to a microwave field efficiently enough to achieve high temperatures. An uninstrumented commercial microwave oven (CEM model MDS8 1D) with a nominal power output of 850 W at 2.45 GHz was used to qualitatively evaluate the coupling efficiency of the glass-bonded sodalite waste form starting materials In these experiments, separately and as the standard starting mixture. incandescent light escaping through joints in the insulating chamber served as an indicator that the load had reached high temperature. The elapsed time from application of microwave power to observation of incandescent light was used as a relative measure of coupling efficiency. The objective of phase 2 experiments was to quantify the thermal response of the glass-bonded sodalite waste form starting material to a microwave field to allow assessment of potential production applications. For these experiments, a commercial Magic Chef model MCD990B with a nominal power output of 900 W at 2.45 GHz was modified to accept a metal sheathed, ungrounded type K thermocouple. A small hole was drilled through the roof of the oven to allow insertion of the thermocouple, and a corresponding hole was drilled through the roof of the insulating chamber so that the thermocouple could be inserted into the center of the load. The materials used in these experiments were a dehydrated, salt occluded zeolite 4A from UOP (Des Plaines IL),and a borosilicate glass frit from Pemco Corp. (Baltimore MD). Both materials were in powder form, with a nominal particle size of -60+200 mesh. The composition of the glass is given in Table I.
~~
Table I: Composition (as oxides) of the glass frit used to make the glass-bonded sodalite waste form Compound Weight Percent Si02 66.5 B203 19.1 A1203 6.8 Na20 7.1 K?O 0.5 ~~
Salt occluded zeolite 4A was prepared by first drying zeolite 4A at 550" C under vacuum, then loading the dried zeolite with simulated (non-radioactive)
Environmental Issues and Waste Management Technologies VIII
365
electrorefiner salt (8.33/1 zeolite to salt mass ratio) at 500" C in a heated V-mixer. The composition of the salt is shown in Table 11. Table 11: Composition of the salt used to make the salt occluded zeolite used in these experiments Wt. % Salt Salt Wt. % LiCl-KC1 eutectic 69.7 BaC12 1.02 14.9 LaC13 1.22 NaCl KBr 2.3 X 10'2 CeCI3 2.33 RbCl 0.33 PrC13 1.15 src1* 1.01 NdC13 3.89 YC13 0.70 PmCI3 0.11 KI 0.15 SmCI3 0.69 CSCl 2.50 EuC13 4.71 X 10-2 RESULTS AND DISCUSSION Phase 1 Experiments In the first phase 1 experiment, 14.8 g of salt occluded zeolite 4A was loaded into the crucible, filling it about half way. Power was switched on, and incandescence was observed after 225 s. Power was immediately switched off. After cooling, the crucible was removed and examined. Most of the powder appeared unaffected, but a region in the center was cracked and seemed to have begun to sinter. There was a hollow space below this region, at the bottom of which was about a 1 cm piece of material that had melted. The first experiment was repeated with 14.6 g of glass f i t as the load. After three consecutive 10 min runs, the oven was opened and the lid of the insulated enclosure removed to observe the load. The glass powder was quite warm, so another pre-programmed run,this time for 30 min was initiated. Incandescence was observed 503 s into that run. After the crucible had cooled, visual observation showed that approximately half the glass had melted. The same experiment was repeated with 14.8 g of a 311 mixture (by weight) of salt occluded zeolite 4A to glass frit. Incandescence was observed 239 s into the run. After cooling, visual examination revealed that a small portion of the charge had consolidated into a spheroid about 1.5 cm in diameter by 1 cm thick. The results of these experiments are summarized in Table 111. The results of the phase 1 experiments show that both the salt occluded zeolite 4A and the borosilicate glass frit used to make the glass-bonded sodalite waste form can be heated to high temperatures in a microwave field. However, the
366
Environmental Issues and Waste Management Technologies VIII
Table 111. Results of phase 1 microwave heating experiments at a nominal microwave power of 850 W Material Mass of load (g) Time to Incandescence (s) Salt Occluded Zeolite 4A 14.8 225 Borosilicate Glass Frit 14.6 >1800 3/1 (weight) mixture of Salt 14.8 239 Occluded Zeolite to Glass salt occluded zeolite clearly couples more efficiently to the microwave field than the glass. As can be seen from Table 111, the thermal response of the 3/1 (by weight) mixture of salt occluded zeolite to glass used to make the glass-bonded sodalite waste form closely resembles the thermal response of the pure zeolite. This implies that at least initially the salt occluded zeolite component is performing the energy conversion function that heats the entire mixture. The rapid onset of incandescence in both materials is typical of the phenomenon known as thermal runaway, that has been widely reported in the literature?J94This phenomenon can seriously limit the use of microwave heating for production applications. This is especially true for the glass-bonded sodalite waste form, because melting causes the radionuclide-bearing salt to phase separate into halide inclusions that are readily soluble in water. So, while the phase 1 experiments showed that glass-bonded sodalite waste form starting materials can be heated to high temperatures using microwave radiation, the question of whether microwave heating could be used in waste form production remained unanswered. Phase 2 Experiments Some way to control or avoid thermal runaway is necessary to use microwave heating in the glass-bonded sodalite waste form production process. The phase 2 experiments addressed this problem by first quantiQing the thermal response of the starting mixture to identify the onset of thermal runaway, then testing the microwave duty cycle (power setting) as a means of controlling the temperature of the load. A type K thermocouple was inserted approximately into the center of 22.68 g of the salt occluded zeolite 4Nglass baseline mixture for m h g the glass-bonded sodalite waste form. The first experiment was run logging temperatures every 10 s. After 70 s, the temperature jumped from about 400" C to nearly 1200' C, indicating the onset of thermal runaway (see Figure 1). Power was immediately shut off, and the load allowed to cool. While cooling, several power settings above
Environmental Issues and Waste Management Technologies VIII
367
and below 7 were tested. Below 7, the load continued to cool, and above 7 temperature increased to thermal runaway. run - power setting 10 -Re-start
-Initial
-
2
1400 1200 loo0
8
800 600 400
B#
F
- power setting 10
200
0 0
20
40 60 80 Elapsed Time (sec)
100
120
Figure 1: Plot of temperature vs. time for the two full power thermal runaway events encountered during the phase 2 experiments. The load was allowed to cooled to about 100' C, whereupon heating at full power was re-initiated. Thermal runaway was again encountered, although at a slightly longer elapsed time (see Figure 1). Power was switched off, and when the load cooled to about 600' C, heating was re-initiated at a power setting of 7 (70% duty cycle, about 20 s on and 10 s off). At this setting, the temperature quickly rose to about 950' C, then began to oscillate with the duty cycle. Monitoring the temperature for about three minutes indicated that the load was approaching dynamic equilibrium, with a mean temperature around 900' C (see Figure 2). This heating schedule was continued for ten minutes, whereupon the programmed run was automatically terminated. Heating was immediately re-initiated at a power setting of 7 for a programmed time of 30 min, but the oven shut down automatically in response to an overtemperature protection device after 657 s. The experiment was terminated at that point. After cooling, a spheroid approximately 1 cm in diameter was found loosely attached to the thermocouple. The remainder of the powder was apparently unaffected. The solid piece was easily dislodged from the thermocouple, and sectioned for analysis. It appeared to consist of two distinct layers; a fiiable, poorly consolidated layer on the outside, and a well consolidated core on the inside. Optical and scanning electron microscopy revealed that the inner core
368
Environmental Issues and Waste Management Technologies VIII
consisted of a multi-phase outer layer and an amorphous-looking inner core (see Figure 3).
!
1200
1
1000 -
400
G
--
-
~-
200
-
0
0
50
100
150
200
250
Elapsed Time (sec)
F i m e 2: Plot of temperature vs. Time at 70% duty cycle showing approach to dynamic equilibrium at a temperature of about 900" C.
Figure 3: Optical micrograph showing the multiple layers of the solid body formed in the phase 2 heating experiments. Tick marks on the scale at the bottom are one millimeter apart.
Environmental Issues and Waste Management Technologies VIII
369
Except for excess porosity near the inner glassy core, the outer portion of the well-consolidated material resembles normal glass-bonded sodalite waste form material. X-ray powder diffiaction confirmed the phase composition as resembling the conventionally prepared waste form, except with a bit more nepheline (in this case a thermal decomposition by-product of sodalite). The outer, poorly consolidated layer showed less halite (a by-product of glass/sodalite interactions) and a lower amorphous content than the inner portion. These results are consistent with the visual observations evident in figure 3. CONCLUSIONS Glass-bonded sodalite waste form starting materials, particularly the salt occluded zeolite, effectively couple to microwave radiation resulting in heating. While thermal runaway resulting in undesirable melting is possible, simple duty cycle power modulation appears to give sufficient temperature control to make microwave heating for waste form production feasible. Further testing is required to determine how best to apply microwave heating. For example, microwave heating could be used exclusively for producing glass-bonded sodalite waste forms, or it could be used as a boost in conjunction with conventional heating to accelerate heating of the central portion of full-scale waste forms, thereby reducing processing times. Future experiments are planned to address this question. ACKNOWLEDGEMENTS Argonne National Laboratory is operated for the U. S. Department of Energy by the University of Chicago. This work was supported by the Department of Energy, Nuclear Energy Research and Development Program, under contract no. W-3 1-109-ENG-38. The author wishes to thank Ms.M. L. Adamic and Mr. J. R. Krsul for their assistance. REFERENCES S. Komarneni and R. Roy, “Anomalous Microwave Melting of Zeolites,” Materials Letters, 4 [2] 107-1 10 (1986). T. Ohgushi, K. Ishimaru, and S Komarneni, ”Nepheline and Carnegieite Ceramics from A-Type Zeolites by Microwave Heating,” Journal of the American Ceramic Society,84 [2] 321-327 (2001). T. Ohgushi, S. Komarneni, and A. S. Bhalla, “Mechanism of Microwave Heating of Zeolite A,” Journal of Porous Materials, 8 23-35 (2001). 4B. I. Whittington and N. B. Milestone, “The Microwave Heating of Zeolites,” Zeolites, 12 815-8 18 (1992).
’
370
Environmental Issues and Waste Management Technologies VIII
MORPHOLOGY AND COMPOSITION OF SIMULANT WASTE LOADED POLYMER COMPOSITE-PHASE WERSION, ENCAPSULATION, AND DURABILITY Harry D. Smith, Gary L. Smith, Guanguang (Gordon) Xia Pacific Northwest National Laboratory, 902 Battelle Blvd, Richland, Washington, 99352’ Brim J.J. Zelinski Department of Materials Science and Engineering, University of Arizona, Tucson, Arizona, 85721
ABSTRACT Because of their good physical and chemical durability, relatively high salt loading capacity, and low leachability, sol-gel-derived, organic-inorganic hybrid materials (polycerams) show promise as media that could be used to stabilize high salt wastes. Use of this technique has been hindered by the need for highly volatile and flammable organic solvents in the fabrication process. In an effort to overcome this hinderance, we carried out initial development of an alternative production approach based on an aqueous emulsion technology and a “phase inversion” phenomenon that results in encapsulation of the waste form. Our major interests focused on understanding the phenomena and optimizing fabrication methods to produce a final waste form with excellent waste stabilization characteristics. Scanning electron microscopy was used to obtain the microstructures of the waste forms for understanding the migration, distribution, and encapsulation of the salt in the waste forms. The leaching rate of the salt from a waste form was quantified by means of conductivity measurement. INTRODUCTION Over the past 50 years, large amounts of mixed low-level wastes have been generated at U. S . Department of Energy (DOE) sites and other related industries. Salt-containing wastes are always troublesome for treatment due to the high solubility of salts in water and the possible involvement of a broad range of chemical species. Polymexeramic hybrids (polycerams) have been demonstrated to be promising candidates for encapsulating salt wastes (Smith, et al., 1999) in comparison to the other developed technologies such as vitrification and grout. The need to use organic solvents with high volatility and flammability in the fabrication of these polymer and polyceramic materials has offset their advantages. Developing alternative approaches that employ aqueous emulsion systems in waste
Pacific Northwest National Laboratory is operated for the U. S.Department of Energy by Battelle under Contract DE-AC06-76RL01830. To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
371
form fabrication process to avoid using organic solvents were initiated (Liang, et al., 2002) based on discussions with the University of Arizona. For the initial study, efforts were focused on the identification of a good model system for fabricating durable waste forms using aqueous-based systems. This approach is based on the water/oil like phase inversion concept. For our study, a model aqueous emulsion mixture of polystyrene butadiene and epoxy resin was used. The discontinuous emulsified droplets suspended water may first congeal into a continuous, waste-encapsulating phase and then form a tough and durable waste monolith during a curing process. The model waste form itself possesses tough mechanical strength at moderate salt loading with low leachability, and apparently good chemical durability. However, little has been known about the occurrences and outcomes resulted from the “phase inversion” during the waste form fabrication processes. The objectives of this study were to understand the “phase inversion” concept and related processes, which are believed to be crucial to the development of the final waste forms, silica-incorporated polycerams that will meet land disposal requirements. Scanning electron microscopy was used to obtain detailed information about the microstructure of the waste forms for understanding the migration, distribution, and encapsulation of the salt in the waste forms. The leaching rate of the salt from a waste forms was quantified by means of conductivity measurement. MATERIALS AND EXPERIMENTS A commercially available aqueous emulsion, polystyrene-butadiene (PSB) latex (Styronal ND 656, BASF), and epoxy resin (Epo-Kwick Resin, Buehler) were used as the ingredients for fabricating our model polymeric composites. With the aid of a surfactant (sorbitan monooleate, Aldrich), the PSB latex and epoxy resin were emulsified by vigorously stirring. Waste salt surrogate, sodium nitrate, along with a crosslinking agent diethylenetriamine (DETA, Aldrich) was then mixed with the emulsion thoroughly. After the mixture was cured at 80°C in a glass baker for about two days, a robust waste form was produced. The process is depicted in Figure 1. After curing, waste monolith was rinsed with de-ionized water to remove the salt crust which was observed to form on the free surfaces. This salt was collected and quantitatively determined in order to obtain a precise inventory of the salt associated with the sample. The microstructures of broken or cut surfaces of cured samples were characterized using a JEOL 5900 LV Scanning Electron Microscope (SEM) with a built-in EverhardtThronley secondary electron detector and a Robinson series VI scintillation-based backscattered electron detector (BSE). The local salt distribution of unleached and leached samples was analyzed by energy dispersive spectroscopy (EDS). The salt leaching behavior was determined by measuring the conductivity of the leached solution as a function of time. Typically, a sample of 2-5 mm in thickness was sectioned fkom the monolith and immersed in a known volume of dsionized water for salt leaching tests. The conductivity of the solution vs. time was measured with a conductivity meter and recorded. The amount of salt leached out from the sample at a given time therefore was determined.
RESULTS Before curing, the aqueous mixture of polymer precursors and salt solution was a milk-like emulsion. During the system was heated at 80°C in an oven, polymerization was observed while some water droplets were found on the wall of the beaker. The final
372
Environmental Issues and Waste Management Technologies VIII
Polystyrene-butadiene
Epoxy resin and surfactant
NaNOs/Water
1
Mixed and stirred for 30 min
[
1
Cured at 8OoCfor 2 days
1
Figure 1. A Schematic Flowsheet of Aqueous-based Fabrication Process for Producing Polymeric Waste Forms. waste form was a tough solid with some salt residuals on the monolith surfaces. Figure 2 shows photo images that were taken during different fabrication stages.
Figure 2. Photo Tmages of the Fabrication of Waste Form During Curing Process. A) Emulsified mixture of aqueous polymer precursors and salt solution prior to curing. B) and C) Polymerization taking place during curing. D) Final waste form.
Environmental Issues and Waste Management Technologies VIII
313
On a macroscopic scale, most of samples prepared appear to be homogeneous. However, SEM examination revealed that these samples are inhomogeneous at the microscopic scale. In general, as the waste form cures, the salt was distributed throughout the interior with the highest concentration near the free surfaces. The salt that actually migrated to a free surface formed a salt crust there, as shown in Figure 3. In the matrix of the waste form, the salt particles exist in different forms, such as large or small particles dispersed in the polymer matrix and pockets which encased some well crystallized salt particles, as indicated in Figure 4. The salt Figure 3. Backscattered SEM Image of Near retention percentage (S) vs. time for the half a waste Form Showing disk of a waste form (containing 22 wt% the SurfaceNaN03) is shown in Figure 5.
DISCUSSION AND CONCLUSION
Phase Inversion and Salt Distribution
One of the important goals of this study was to evaluate the extent of the “phase inversion” and document its relationship to the leaching rate of the surrogate salt waste (sodium nitrate). Figure 6 gives a schematic representation of phase inversion process. When emulsified epoxy and PSI3 were thoroughly mixed, the resulting emulsion was stable for several days without visible phase separation even with 30wt% salt present in the aqueous phase. Upon water removal from the aqueous mixture during the curing process, the emulsion is believed to transform from the oil-in-water (O/W) type into a water-in-oil (W/O) type. The phase inversion results in the encapsulation of salt particles within the polymeric matrix. Figure 4 may provide us with the evidence of the occurrence of the phase inversion. The salt crystals are observed to be completely entrained in the matrix or encapsulated in voids.
Figure 4. Backscattered SEM Images of the Interior Potion of a Typical Waste Form. On the left is a cut surface of the waste form. Salt particles are embedded in the matrix. On the right is a fracture surface of the same waste form that showing salt crystals trapped in a void.
374
Environmental Issues and Waste Management Technologies VIII
The phase inversion process is considered to be the key step controlling the microstructure of the waste form, which in turn controls the salt distribution, leachability, strength, and chemical durability. The occurrence of the phase inversion arises fiom the attainment of a critical ratio of organic component to water as water is removed during the curing process. Once the phase inversion and polymerization occur, the further
80
= 4
30 20 10
0 0
50
100
150
200
250
300
350
Time (h)
Figure 5. Salt Retention (wt’Y0) vs. Leaching Time for the Sample (22wt% Salt). About 60 wt% of salt leached out from the sample at a very short leaching period, which corresponded to surface salt on the leaching sample. Only a small amount of salt (-2 wt%) was retained in the sample after about 300 hours. evaporation of water from the waste form surface results in the partial migration of the salt from the interior to the free surfaces of the monolith. Note that during the curing process, the waste salt is precipitating out of the aqueous phase. These precipitates will automatically be entrained in the matrix, while any droplets of waste salt solution that are trapped will form spherical voids with salt crystals in them. The water from those voids probably diffuses through the polymer matrix until it reaches open porosity and evaporates. In fact, the salt crystals were always found on the surfaces of the final products. As seen in Figure 4, the microstructure of the central portion of the sample consists of sac-like structures that may be filled with salt. These structuresare imbedded in the matrix phase, appear to completely encapsulate that salt, and are resistant to leaching.
Salt Retention and Leaching As seen in the Figure 5, the salt retention curve of the sample (22 wt% salt) shows different leaching behaviors. At the beginning of the leaching test, the salt retention drops rapidly, which corresponds to the salt crust or interior salt exposed by the sectioning process. This salt can be washed off very quickly and the quantity of the salt can be calculated from the diffusion curve by the extrapolation of the long time leaching behavior to zero time. The leaching rate of the salt fiom the interior of the sample was
Environmental Issues and Waste Management Technologies VIII
375
Figure 6. A Schematic Representation of Phase Inversion Process During Curing for the Fabrication of an Aqueowbased Polymeric Waste Form. On the left, emulsified polymer precursors (filled circles) suspended in water forms the oilin-water ( O N ) type emulsion. Upon water removal from the aqueous system during curing, the emulsion is expected to transfbrm into a water-in-oil (W/O) type (right). The phase inversion results in the encapsulation of salt particles (empty circles on the right) within the polymeric matrix. relatively slow, as shown in the salt retention curve at the later leaching stage. The fact that almost all the salt in the waste form eventually leached out after over 300 hours leaching test indicates open porosity still exists in the polymeric matrix. The formation of the open porosity is believed due to the water evaporation and salt migration towards the surface of the waste form during the curing process. Obviously, the closing of the open porosity by appropriate methods will help to reduce the salt leachability. Post treatment methods, such as reheating or hot-pressing waste forms may enhance the capability of resistance to salt leaching, ACKNOWLEDGEMENTS The authors thank the Laboratory Directed Research and development (LDRD) program supported by Pacific Northwest National Laboratory (PNNL) We also would like to thank Dr. Willam Kuhn and Mr. Jim Buelt for their advises and Dr. Liang Liang for his early work on this project. REFERENCES 1. G a y L. Smith and Brian J.J. Zelinski, Stabilize High Salt Content Waste Using SolGo1 Process, Innovative Technology -DOE/EM-0473, OST Reference #2036, Mixed Waste Focus Area, Prepared for U.S. Department of Energy, Office of Environmental Management, Ofice of Science and Technology, September 1999.
2. Liang, L., Smith, H., Russell, R., Smith, G. & Zelinski, B. J. J. Aqueous Based Polymeric Materials for Waste Form Applications. In G.L. Smith, S.K. Sundaram, and D.R. Spearing (Eds.), Proceedings of the International Symposium on Environmental Issues and Waste Management Technologies in the Ceramic and Nuclear Industries VII, Westerville, Ohio: Ceramic Transactions, The American Ceramic Society (2002).
376
Environmental Issues and Waste Management Technologies VIII
93NbMAS NMR OF NIOBIUM CONTAINING SILICOTITANATE EXCHANGE MATERIALS Brian R. Cherry', May Nyman2,and Todd M. Alam' 'Department of Organic Materials and 2Departmentof Geochemistry, Sandia National Laboratories, Albuquerque, NM 87 185 USA
ABSTRACT Crystalline silicotitanate (CST), HNa3Ti&i2014*4H20, is a highly selective Cs ion exchanger, making it an attractive material for removal of 13'Cs from nuclear waste solutions. The Cs selectivity can be improved further by replacing a fraction of the framework titanium with niobium to form NbCST. High-speed 93Nb magic angle spinning (MAS) nuclear magnetic resonance (NMR) spectroscopy was utilized to characterize framework changes as a function of Cs loading in a series of Nb-CST materials. Based on these 93NbMAS NMR studies it is argued that the niobium octahedra present in Nb-CST have near uniform Nb-0 bond lengths and are slightly distorted from cubic symmetry. INTRODUCTION Technologies that selectively remove radioactive Cs or Sr from nuclear defense wastes are of great interest to the U.S. Department of Energy (DOE) in that the radionuclides 137Csand "Sr are responsible for the majority of the radioactivity in these waste solutions. The nuclear wastes stored at Hanford, the Savannah River Site (SRS), Oak Ridge National Laboratory (ORNL), West Valley, and Idaho National Engineering and Environmental Laboratory (INEEL) present challenges to present Cs removal technologies because these wastes contain very high concentrations of dissolved salts and may be extremely basic (Hanford, SRS) or acidic (INEEL). Further, proposed Cs removal technologies must be able to withstand high radioactive doses without diminished performance. In the early 199O's, Dosch, Anthony, and Gu at Sandia National Laboratories [1, 21 discovered a new silicotitanate inorganic ion exchanger that selectively sorbs 50 ppm Cs from solutions containing -5 M sodium salts. This material, To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
377
called Crystalline Silicotitanate or CST, shows excellent Cs removal capabilities in both highly acidic and highly basic solutions, and is stable at these extreme pHs, as well as in extreme radioactive environments. Further, substitution of 25% of the framework Ti with Nb provides an approximate fourteen-fold increase in the selectivity for Cs, as measured by the distribution coefficient E,ml/g][3] for Cs over Na.[1, 21 In a cooperative agreement with Sandia, Universal Oil Products (UOP LLC) developed the Nb-substituted CST as a product known as IE910, and a granular form known as UOP IONSIV IE91 lTM. Until August 2000, the addition of Nb to the CST framework was a trade secret, protected by a U.S. patent assigned to Sandia National Laboratories.[4] Therefore, the mechanism for increased Cs selectivity with addition of Nb to the CST framework has never been properly investigated. We are currently using a variety of solid-state N M R techniques to determine structural and compositional changes that result from substitution of N b into the CST framework, and how this affects the Cs selectivity of these sorbents. These investigations are being carried out on a series of CST and Nb-CST materials with variable Cs-loadings.
Figure 1: Powder X-ray diffraction spectra of the CST and Nb-CST materials.
A
10
20
30
40
50
60
2-theta
378
Environmental Issues and Waste Management Technologies VIII
Figure 2: CST/Nb-CST framework structure, Ti@b)O6 octahedra (gray), SiO4 tetrahedra (black), framework Na (spheres).
CST Structure and the Nb-site: The 25% Nb-substituted CST has a composition of approximately HNa2Ti3NbSi2014.4H20. The structure is essentially the same as that reported for the Nb-free CST, ~a3Ti4Si2014*4H20.[5]Powder X-ray diffraction spectra comparing the CST and Nb-CST reveal essentially no major crystallographic differences between the two samples (Figure 1). Work in progress on structural investigations on the same series of CST, Nb-CST and Cs-exchanged CST and Nb-CST show that differences in the framework structure between CST and NbCST are minimal.[6] For the sake of discussion, a view of the CST/Nb-CST framework is shown in Figure 2. The TiO6 octahedra (gray clusters) are arranged in cubane clusters of four edge-sharing octahedra. These clusters are cornerlinked to each other in the z-direction through two octahedra per cluster. Overall, the Ti06 octahedra make up double zig-zag chains along the z-direction. These chains are linked in the x- and y- directions by corner-sharing with SiO4 tetrahedra (black clusters). These Si04tetrahedra form chains in the z-direction by alternating with edge-sharing framework sodium sites (spheres). This structural arrangement gives rise to TiO6 octahedra in which all the oxygen atoms are bridging (i.e. no terminal oxygen atoms), and thus very regular Ti-0 bond lengths. The Ti-0 bond lengths reported for Nb-free CST range from 1.89 - 2.07 A. The Nb disordered over 25% of these framework octahedral sites may be either regular like the Ti06 sites, or they may be distorted, as framework NbO6
-
Environmental Issues and Waste Management Technologies VIII
379
octahedra often exhibit axial distortion with one short axial bond (-1.8 A), one long axial bond (- 2.2-2.4 A) and four regular equatorial bonds (-1.9 - 2.1 A).[7, 81 The purpose of this paper is to characterize the niobium coordination environment with 93NbMAS NMR as a function of Cs loading.
EXPERIMENTAL (1) Material Preparation Synthesis of Nb-Substituted C'stalline Silicotitanate (Nb-CST): Titanium isopropoxide (TIPT, 3.43 g, 12 mmol), tetraethylorthosilicate (TEOS, 3.33 g, 16 mmol) and Nb205 (0.54 g, 4 mmol Nb) were added to 50 ml aqueous NaOH (6.6 g, 165 mmol) solution in a 100 ml Teflon liner to an autoclave Parr reactor. The mixture was stirred for 0.5 hr, and then placed in a 200 "C oven for three days. The resulting product, a white microcrystalline powder was collected by filtration (yield 2.2 g of HNa2Ti3NbSi201404H20;86 % yield based on Ti). A small amount of crystalline byproduct was inevitably formed with the major Nb-CST product. Before analyses of the sample, the byproduct was removed by a two step treatment: 1) the Nb-CST with the byproduct was first exposed to a 1 M aqueous HCl wash for three hours at room temperature, and 2) the Nb-CST with byproduct was exposed to a 1 M NaOH wash for three hours at 40 "C. The first step amorphizes the byproduct, and the second step dissolves the resulting amorphous byproduct.
-
Ion Exchange: A series of Cs-exchanged Nb-CST materials were prepared by ion exchange. The maximum amount of Na in Nb-CST that can be readily exchanged for Cs was approximately 25%. For each ion exchange, 3 g of Nb-CST was combined with 50 ml aqueous CsCl solution, containing the appropriate amount of CsCl to obtain a Cs-exchange Nb-CST sample with 3.8, 6.4, 9.0 and 9.6 wt % cesium. The Nb-CST samples were shook with the CsCl solutions at room temperature for 12 hours, and the Cs-exchanged samples were collected by filtration. Inductively Couple Plasma Mass Spectroscopy (ICP MS) was used for compositional analysis of these Cs-exchanged Nb-CST materials. Powder X-ray hffraction was used to examine phase identification, purity and crystallinity, and thermogravimetric analysis (TGA) was used to determine water and OH content. (2) NMR Anal sis The static z3Nb N M R spectrum of the Cs fiee Nb-CST was obtained on a Bruker Avance6OO at 146.72 MHz with a 4mm broadband probe, a Hahn-Echo sequence, a 25 ps inter-pulse delay, and 120K scan averages. All 93Nb MAS N M R spectra were obtained on a Bruker Avance600 at 146.72 MHz with a 2.5 mm broadband probe. Direct polarization MAS spectra were obtained with sample spinning speeds between 31-33 kHz, using a n/12 93Nbpulse ( d 2 = 3 ps),
380
Environmental Issues and Waste Management Technologies VIII
high power TPPM 'H decoupling, 500 ms recycle delay, 1 MHz SW (681 1 ppm) and 128K scan signal averaging. Spectra were processed using linear rediction of the first 8-12 time domain points to reduce acoustic baseline roll. 93Nb spectra were referenced to the secondary standard NbCls in wet acetonitrile, with the sharp resonance, due to [NbC16]-,assigned to 6 = 0.0 ppm.
RESULTS AND DISCUSSION The 93NbStatic NMR spectrum of the Nb-CST material prior to Cs exchange is shown in Figure 3a. The static spectrum results from a complex mixture of quadrupole (CQ) and chemical shift anisotropy (CSA) interactions, with a CQ on the order of 20 MHz. The relationship between the CQand the CSA is the subject of ongoing investigations. Under high speed MAS conditions, o,= 33 kHz, the 93Nbline width is dramatically reduced, (full width half maximum, FWHM 74,000 Hz to 15,000 Hz) Figure 3a. The high rate of spinning speed is necessary to narrow the resonance plus separate the fjrst spinning side bands from the isotropic resonance. The 93NbMAS N M R spectra for the Nb-CST materials as a function of Cs loading are shown in Figure 3b; the frequency shifts and line widths are listed in Table I. The MAS NMR spectra of all the Nb-CST samples show a single, nearly symmetric resonance. Within experimental error, there is no variation in the 93Nbfrequency shift with increasing Cs loading. The line widths span a range from 13000 to 15600 Hz. It should be noted that resonances with large quadrupolar couplings (CQ > 60 MHz) may not be observable using these standard MAS techniques at the given field strength. Interestingly the overall signal intensity does not vary greatly with increasing Cs loading, supporting the argument that the addition of Cs does not produce a new unobservable resonance, with the corresponding loss of the original peak. To date, there have been only a limited number of solid state 93NbMAS investigations.[9-11] The Nb in these Nb-CST materials have an observed 93NbNMR frequency shift (Table I) that is downfield from the 6 = -900 through -1 100 ppm shifts reported for alkali niobate perovskites, lead niobate pyrochlorates, lead magnesium niobate (PMN) and a range of PMNAead titanate (PT) solid solutions (all octahedral bonding configurations, ranging from cubic symmetry to distorted rhombic).[9, 111 The downfield shift of the Nb-CST is consistent with the de-shielding nature of the octahedra participating in the cubane cluster. The niobium sites present in the PMN and PMNPT solid solutions are all corner sharing.[9, 111 The observed line widths for the Nb-CST (Table I) are in the same range observed for slightly distorted octahedral in the PMNPT solids, corresponding to a CQw 17 MHz. In those PMNPT materials, this CQwas assigned to axial (tetragonal) Nb.The line width for the Nb-CST is also much larger than the narrow 2000 Hz resonance observed for the very symmetric cubic Nb site reported in the PMNPT materials.
-
Environmental Issues and Waste Management Technologies VIII
-
381
Figure 3: a) 93Nb Static and MAS NMR spectra of the 0% Cs Nb-CST material. b) 93NbMAS NMR spectra of the Nb-CST materials as a function of Cs loading.
A
a.
b.
A
i\
I
"
'
I
-500
"
"
I
-1000
.
ppm
Table I. 93NbMAS NMR characterization of cesium exchanged Nb-CST materials. wt % C S ,Nb-CST
&so
(ppm)"
FWHM (Hz)
0
-722
15160
3.8
-726
13725
6.4
-726
13155
9.0 9.6
-725 -726
13559 15598
a. Apparent frequency shift, second order quadrupolar effects not determined in these single field strength studies. Estimated error f 5 ppm.
382
Environmental Issues and Waste Management Technologies VIII
In addition to the PMN and PMNPT solid solutions, two other 93NbNMR studies on model compounds helped place bounds on the symmetry of the niobium octahedra present in the Nb-CST materials. The 93Nb MAS NMR line widths in Nb-CST (Table I) are very similar to that found in Sn2Nb207, where the niobium octahedral is characterized by a single Nb-0 bond length of 1.98 A and two 0-Nb-0 bond angles, 90.59' and 89.42".[10] In contrast, the niobium octahedral present in SnNb2O6 (Foordite) is significantly distorted with the Nb-0 bond distances ranging between 1.85 to 2.16 A. The corresponding static 93Nb NMR spectra of Foordite shows a very broad second-ordered broadened yuadrupolar powder pattern with a CQ of -38 MHz.[lO] Based on the observed Nb MAS N M R frequency shifts and line widths observed for the exchanged NbCST (Table I) we conclude that the niobium octahedra present in Nb-CST have near uniform Nb-0 bond lengths and are slightly distorted from cubic symmetry, like the TiO6 sites found in the Nb free CST materials, where the Ti-0 bond length range from 1.89 - 2.07 A. In the Nb-CST materials a small upfield shift of the 93Nbresonance is also observed with the initial Cs exchange. Further Cs uptake has no effect on the observed 93Nbfrequency shift. This observation suggests that as the initial Cs is incorporated into the Nb-CST, the framework adjusts slightly, allowing the exchanged Cs to occupy an optimal binding site in the center of the tunnel. Once this small change in the structural framework has occurred, additional Cs exchange has no further impact on the framework structure. Evidence for this change in the framework structure as Cs is initially loaded has also been observed in the 29Si MAS NMR data.[12] These 93Nb MAS N M R investigations have provided insight into the structural environment of Nb-CST materials, and have demonstrated that no major variation-in the Nb-0 octahedral symmetry occur with incorporation of Cs. This NMR data, along with the ongoing crystallography investigations, suggest the addition of N b to the CST framework does not affect the Cs selectivity by direct interaction of Cs with the framework Nb. We are continuing to investigate the mechanism responsible for the improved Cs selectivity through 23Na,29Si,'H and 133CsNMR experiments. ACKNOWLEDGMENTS Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under Contract DE-AC04-94AL85000. This work was supported by the D.O.E. (Office of Science) Environmental Management Science Program, project #8 1949.
Environmental Issues and Waste Management Technologies VIII
383
REFERENCES 1.
2.
3. 4. 5. 6. 7.
8.
9.
10. 11.
12.
384
Dosch, R.G. and R.G. Anthony, Hydrous Crystalline Sitico-Titanates: New Materialsfor Removal of Radiocesiumfrom Concentrated Salt Solutions with pH's in the 1-14 Range, A Topical Report, 1995, Sandia National Laboratories: Albuquerque, N.M. Gu, D., TAM-5, A Hydrous Crystallline Silicotitatnatefor Removal of Cesiumform Dilute Aqueous Waste, in Kinetics, Catalysis and Reaction Engineering Laboratory, Department of Chemical Engineering. 1995, Texas A&M University: College Station, TX. Zheng, Z., C.V. Philip, R.G. Anthony, J.L. Krumhansl, D.E. Trudell, and J.E. Miller, Ion Exchange of Group I Metals by Hydrous Crystalline Silicotitanates. Ind. Eng. Chem. Res., 1996.35: p. 4246-4256. Anthony, R.G., R.G. Dosch, and C.V. Phillip, U. S. Patent # 6,1lU,378, Sandia National Laboratories: U.S.A. Poojary, D., R. Cahill, and A. Clearfield, Synthesis, Crystal Structure, and Ion-Exchange Properties of a Novel Porous Titanosilicate. Chem. Mater., 1994.6(12): p. 2364-2368. Tripathi, A. and A. Clearfield,personal communication with M Nyman. 2002: Texas A & M University. Jehng, J.M. and I.E. Wachs, Structural Chemistry and Raman-Spectra of Niobium Oxides. Chem. Mater., 1991.3(1): p. 100-107. Nyman, M., T.M. Nenoff, A. Tripathi, J. Parise, and R.S. Maxwell., Sandia Octahedral Molecular Sieves (SOMS): Structural and Property Efects of Charge-Balancing the MV-Substituted (M = Ti, Zr) Niobate Framework. J. Am. Chem. Soc., 2002.124(8): p. 1704-1713. Fitzgerald, J.J., S. Prasad, J. Huang, and J.S. Shore, Solid-state 93NbNMR and 93NbNutation Studies of Polycrystalline Pb(Mgl/3Nb2/3)03and (1x)Pb(Mg1&b2,3)03/xPbTi03 Solid-Solution Relaxor Ferroelectrics. J. Am. Chem. Soc., 2000.122: p. 2556-2566. Cruz, L.P., J.-M. Savariault, J. Rocha, J.-C. Jumas, and J.D.P.d. Jesus, Synthesis and Characterization of Tin Niobates. J. Solid State Chem., 2001. 156: p. 349-354. Prasad, S., P. Zhao, J. Huang, J.J. Fitzgerald, and J.S. Shore, Niobium-93 M Q M S NMR Spectroscopy Study of Alkali and Lead Niobates. Solid State Nuclear Magnetic Resonance, 2001. 19: p. 45-62. Cherry, B.R., M. Nyman, and T.M. Alam, In preparation.
Environmental Issues and Waste Management TechnologiesVIII
SELECTIVE ABSORPTION OF HEAVY METALS AND RADIONUCLIDES FROM WATER IN A DIRECT-TO-CERAMIC PROCESS B.P. Gran, Allen W. Apblett, and Mohamed Chehbouni Department of Chemistry Oklahoma State University Stillwater, OK, 74078. ABSTRACT The ability of molybdenum hydrogen bronze, HMo206to absorb heavy metals and radionuclides from water was investigated. It was found that it could remove substantial amounts of metal ions from water and was selective for those that are chemically-soft or have large radii. The products from uranium, thorium, and neodymium uptake were discovered to be layered metal molybdates while that from lead was wulfenite, PbMo04. In the light of this result, the application molybdenum trioxide for lead adsorption was investigated and it was found to perform similarly to the hydrogen bronze. INTRODUCTION The use of reactive barriers to prevent the spread of pollutants in aquifers is a promising technology that can greatly curtail any environmental endangerment. Furthermore, the reagents used for construction of reactive barriers are generally also amenable to application in pump and treat operations or for treatment of wastewaters. In 1989, the use of granular iron was proposed for in situ remediation of groundwater containing chlorinated organic contaminants. Since that time, the technology has been adopted at numerous sites and has been applied to remediation of other types of organic compounds, inorganic species, and radionuclides [ 1-41.
To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the property of The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, without the express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.
Environmental Issues and Waste Management Technologies VIII
385
Molybdenum hydrogen bronze, HMo206(also called molybdenum blue) is a promising reagent for environmental remediation that has a number of unique properties which suggests it could perform better than other reductants for treatment of contaminated waters and the construction of permeable reactive containment barriers to prevent spread of pollutants within an aquifer. For example, when reductions of inorganic or organic pollutants are performed in a column-type reactor, the color change fiom royal blue to white would greatly facilitate monitoring of the column’s remaining reductive capacity. Unlike other reductants that can be employed in the presence of water and oxygen (such as iron), molybdenum blue has an open layered structure (Figure 1) that allows the entire reductive capacity to be used and enhances the rate of reaction by providing a tremendously enhanced area for the reaction to take place. Since both reduced and oxidized forms of the oxide materials have layered structures through which reactants and products can intercalate, passivation due to build up of oxidized product on the surface does not occur. This is in significant contrast to iron that can form a crust of rust that arrests further reaction of iron particles with contaminant species. Finally, molybdenum blue is easily recycled after use in redox reactions since regeneration only requires treatment with hot isopropanol in the presence of a trace of HCl or with zinc/HCl. In fact, the regeneration process with isopropanol only produces acetone as a by-product and, in actual industrial production, the acetone could be captured and sold as a commodity chemical.
Figure 1. Structure of Molybdenum Blue Molybdenum blue has been demonstrated to be a useful reagent for dechlorination of halocarbons such as carbon tetrachloride[5,6]. In such reactions, the bronze acts as a source of hydride so that, for example, CCl, is reduced to chloroform, CHC13. This may be somewhat surprising since the protons present
386
Environmental Issues and Waste Management Technologies VIII
in the bronzes are attached to oxygen atoms to yield hydroxides that bridge between two molybdenum centers. Nevertheless, the combination of the hydroxide and a molybdenum(V) center provides a water and air-stable source of hydride. The question posed in this investigation is whether or not molybdenum blue can also reduce metal ions and remove them from aqueous solution. Alternatively, the possibility exists that the hydrogen ions can be exchanged with metals allowing their uptake. If so, the metals would be readily released by oxidation of the bronze to Moo3providing a "switchable" ion exchanger amenable to highly concentrating metal ions in a similar fashion to the materials developed by Dorhout et al. [7]. EXPERIMENTAL All reagents were commercial products (ACS Reagent grade or higher) and were used without further purification. Thermogravimetric studies were performed using 10-20 mg samples on a Seiko ExStar 6200 TGA/DTA instrument under a 50 ml/min flow of dry air. The temperature was ramped from 25 to 600°C at a rate of 5"C/min. Bulk pyrolyses at various temperatures were performed in ambient air in a digitally-controlledmuffle furnace using ca. 2 g samples, a ramp of lO"C/min and a hold time of 4 hr .X-ray powder diffraction (XRD) patterns were recorded on a Bruker AXS D-8 Advance X-ray powder diffractometer using copper K,radiation. Crystalline phases were identified using a searcWmatch program and the PDF-2 database of the International Centre for Diffraction Data [8]. Preparation of Molybdenum Blue A round bottom flask was charged with 30.00 g of Moo3, 300 ml of n-butanol and 5 ml of concentrated HC1. The mixture was refluxed for 12 hours at which time it had turned a very dark blue color. At this point, the reaction mixture was cooled to room temperature and was filtered through a fine sintered-glass filter funnel. The dark blue solid was washed with n-butanol and was then dried in a vacuum oven at room temperature. The yield was 28.30 g (94%). The XRD pattern of the product corresponded to that of Mo205(OH)(ICDD #14-0041). Measurement of the Uptake of Metals by Molybdenum Blue Molybdenum blue was tested for the ability to remove Pb2+,Th4', U O F and Nd3+from aqueous solution. HMo206(1.0 g) was reacted with 100 ml of individual approximately 0.lM solutions of Pb2+,Th4', U O F and Nd3+.In all cases, nitrate salts were used with the exception of uranyl where both a nitrate and an acetate salt were tested. After stirring magnetically for a sufficiently long time for complete reaction, as indicated by complete disappearance of the blue color,
Environmental Issues and Waste Management TechnologiesVIII
387
the mixtures were separated by filtration through a 20 ym nylon membrane filter. The solid products were washed copiously with distilled water and then were dried in a vacuum desiccator. They were subsequently characterized by infrared spectroscopy, thermal gravimetric analysis, and X-ray powder diffraction. The uranium and neodymium concentrations in the treated solutions were analyzed using UVNisible spectroscopy (h=415 nm and 521 nm, respectively). Lead was determined gravimetrically as lead chromate [9]. Quantitation of thorium was performed colorimetrically using the blue complex (h=575 nm) formed between thoriwn and carminic acid [101. Selectivity Determination The selectivity of molybdenum blue for actinides was tested by competition experiments with calcium. Thus, the reactions between uranyl nitrate and molybdenum blue were repeated in the presence of 0.5, 1.0, and 5.0 molar equivalents of calcium nitrate per mole of uranyl ion and the uptake of uranium was determined by UVNisible spectroscopy. Reaction of Lead Nitrate with Moo3 Pb(NO& (3.98 g, 12 mmol) was dissolved in 100 ml of water and the resulting solution was stirred with Moo3 (1.44 g, 10 mmol) for 72 hours. The resulting white solid was isolated by filtration was washed with water and dried in a vacuum oven at room temperature for 12 hours. The yield was 1.73 g, corresponding to an uptake of 0.27 g of Pb. XRD analysis showed the solid to be a mixture of PbMo04 (ICDD # 44-1486) and unreacted Moo3. RESULTS AND DISCUSSION Molybdenum blue was tested for its ability to remove Th4+ (as a model for plutonium(IV)}, UO? (of interest in its own right and as a model for PuO?), and Nd3' (as a surrogate for the later transuranics , radioactive lanthanides and Pu3+) from aqueous solution. Also, the uptake of lead as a model heavy metal was also investigated. The experiments that were performed were designed to determine the capacity of the blue reagents for the various metals and attempt to identify the mechanism of metal uptake. Molybdenum blue was reacted with an aqueous solution of each of the metals listed above. The stoichiometry was adjusted so that there was at least a one-fold excess of contaminant metal ions {on the basis of one molar equivalent of metal ion per M(V) site in HMo206}. The experimental conditions and results for the molybdenum blue/metal ion reactions are listed in Table I while the results of the analyses and binding capacity calculations are given in Table 11. The results show that molybdenum
388
Environmental Issues and Waste Management Technologies VIII
blue has a remarkable capacity for absorption of actinides and heavy metals. Molybdenum blue absorbed 122% by weight of uranium, 37% by weight of thorium, 61.6% by weight of neodymium, and 110% by weight of lead. The substitution of acetate ions for nitrate ions has a small, negative effect on the uptake of uranium. These extremely high capacities bode well for the eventual application of these materials in environmental remediation. The uptake of the metals in terms of milliequivalentsper gram of molybdenum blue were 4.27 for neodymium, 5.14 for uranium, to 5.29 for lead. Thus, the moles of metal that can be absorbed by molybdenum blue varies with the metal used. Within the group of doubly-charged metal ions, the moles of metal absorbed are almost equivalent. In the case of the latter metals, the uptake of metals may be expressed as 1.5 moles per mole of HMo206 and is therefore larger in magnitude than the number of Mo(V) centers. This result indicates that the molybdenum(V1) centers in molybdenum blue also play a role in metal binding. The uptake of neodymium was 1.24 moles per mole of molybdenum blue. Table I. Experimental Conditions for Metal Uptake Experiments Weight of Molybdenum Blue (g)
Weight of Solid Product (g)
Color of Solid Product
Uranium acetate
1.04
2.15
Yellow
Uranium nitrate
1.05
2.32
Yellow
Thorium
1.oo
1.40
White
Neodymium
1.10
1.48
Grey
Lead
1.04
2.74
White
Metal
Clearly, the results indicate that the uptake of the metals is not a simple ionexchange reaction since the uptake exceeds the number of exchangeable ions. The color changes observed during the uptake of the metals indicate a redox reaction in which the molybdenum(V) is oxidized to molybdenum (VI) but the final colors also demonstrate that the contaminant metals are not reduced so that the responsibility of a redox process for metal uptake may be ruled out. Presumably, the oxidation of the molybdenum blue is due to reaction with atmospheric oxygen.
Environmental Issues and Waste Management TechnologiesVIII
389
A major concern for the application of molybdenum blue in the field is its selectivity for actinides and heavy metals as opposed to benign cations normally found in natural waters. Therefore, the selectivity of molybdenum blue for uranyl ion over calcium ions was determined. The results are displayed in Table 111 and demonstrate that molybdenum blue is highly selective for uranium. Even a fivefold higher concentration of calcium ions over uranyl ions had little effect on the absorption of uranium. Curiously, the minor effect that calcium does exert on uptake is greatest when it is below equimolar amounts, least when it is present in the same concentration as uranium, and increases thereafter. Table 11. Results of Metal Uptake Experiments Final Concentration
Uptake (mmol)
Metal Capacity (m0Vg)
Metal Capacity (weight %)
0.053 M
4.7
4.5
108%
0.046 M
5.4
5.1
122%
Thorium
0.084 M
1.6
1.6
37.0%
Neodymium
0.047 M
5.3
4.8
69.5%
Lead
0.045 M
5.5
5.3
110%
Uranium acetate Uranium nitrate
Table 111. Results of Competition Experiments Calcium: Uranium Ratio
Final Uranium Concentration
Weight Percent of Uranium Absorbed ~~
0: 1
0.046 M
122%
1:2
0.055 M
100%
1:l
0.049 M
121%
1:5
0.052 M
111%
390
Environmental Issues and Waste Management TechnologiesVIII
Infrared spectral analysis of the various solid products and molybdenum blue was performed in order to gain a better understanding of the nature of the products and to perhaps shed some light on the mechanism of metal uptake. The positions of the molybdenum-oxygenstretches are given in Table IV. Molybdenum blue was found to have a characteristic absorption at 857 cm-' which is different from the bands observed in molybdenum trioxide. In all cases, except for neodymium, this band, attributable to Mo(V)-0 stretching vibrations has disappeared. The neodymium compound is unusual because the molybdenum centers appear to be freely rotating in the solid so that there is rotational structure to the infrared absorptions making it difficult to assign the positions of the vibrations. Nevertheless, the data in Table IV demonstrate that the solid products from reaction of molybdenum blue with uranium, uraniumlcalcium mixtures, and neodymium all contain network polymers based on Moo6 octahedra, as demonstrated by multiple MO-0stretches. By contrast, the lead product had a single strong MO-0absorption at 786 cm-' attributable to a tetrahedral Moo4 center. Table IV. Metal-Oxygen Stretching Frequencies Observed in the Infrared Spectra Nd+ Pb + U+ U+Ca+ Moo3 MoBlue MOBlue MOBlue MOBlue MOBlue
998
999 (w)
998
884
970*
980"
972"
892
90 1
913
857
866 840*
849* 786
569
572
551
533
502
572 498
* U-0 stretches of the uranyl ion In addition to the M-0 stretching bands, the infrared spectra of the solids also contain bands attributable to a small amount of anions that are also absorbed from aqueous solution. The solids from reaction of molybdenum with uranyl, lead,
EnvironmentalIssues and Waste Management TechnologiesVIII
391
and neodymium nitrate all display an infrared absorption at 1384 cm-'. Notably, this band is due to ionic nitrate and not nitrate covalently-bound to the contaminant metals [111. The product from uranyl acetate has weak bands at 1506 and 1436 cm-' that are due to acetate ions - again the positions of these bands do not correspond to acetate bound to uranium (15 14 and 1480 cm-') that was determined from the infrared spectrum of the starting material. The uptake of the anions indicates that when the metals are bound, the charge is not entirely compensated by the negative charge of the molybdate framework. Nevertheless, the absorptions for the extraneous anions are weak indicating a low degree of incorporation into the solid products. This conclusion was supported by the fact that the ceramic yields derived from heating the solids to 600°C in a thermal gravimetric analyzer were quite high (Table V). Indeed, the majority of the weight losses occur between room temperature and 200°C and can be attributed to dehydration and dehydroxylation reactions. Table V. Results of TGA Experiments Metal Salt
Temperature Range of Weight Loss
Ceramic Yield
U acetate
25-436°C
88.2%
U Nitrate
25-469°C
90.2%
Thorium Nitrate
25-502°C
94.9%
Nd Nitrate
25-209"C
96.5%
Pb Nitrate
215-495°C
99.2%
X-ray powder diffraction analysis of the product from lead uptake by HMo206revealed that it consisted mainly of PbMo04 (wulfenite, ICDD # 441486) plus a small amount of molybdite (Moo3, ICDD # 05-0508). The other metals, however, formed crystalline phases that did not match normal molybdate salts. However when heated to 600°C these unidentified phases were converted to a small amount of Moo3 and U02(Mo04),Th(MoO&, or Nd2(Mo04)3 , depending on the metal. These results in combination with the infrared spectral data suggest that the metal ions intercalate between the layers of HMo206and react to give what is likely to be phases that consist of negatively-charged slabs of Moo6 octahedra with the contaminant ions residing between the layers. In the
392
Environmental Issues and Waste Management Technologies VIII
case of lead, however, the interaction between HMo206and Pb” is so strong that the molybdenum oxide layers are destroyed to yield a normal ortho-molybdate salt. Since the uptake of metals does not rely on ion exchange or redox chemistry there is an implication that the reduction of Moo3 to HMo206is not necessary for metal uptake since the parent oxide also consists of layers of Moo6 octahedra. Reaction of aqueous lead nitrate with Moo3 demonstrated that the trioxide could absorb lead ions. In a 72 hour reaction at room temperature, 1.44 g of Moo3was found to absorb 0.27 g of lead from a 1.2 M lead nitrate solution. XRD analysis showed that the reaction was not complete so that unreacted Moo3 was present along with the expected product, PbMo04. CONCLUSION In conclusion, it has been demonstrated that molybdenum blue has an extremely high capacity for absorption of contaminant metals. Considerable information has been collected concerning the mechanism of metal absorption and the results obtained so far suggest intercalation of the metal ions between the layers of HM0206 followed by reaction to yield solids in which the metal ions tare trapped as counterions to the freshly-generated molybdate sites. These reactions are highly selective for heavy metals and suggest considerable promise for application in environmental remediation and as reactive barriers for the prevention of the spread f contaminant plumes. ACKNOWLEDGEMENT Support for this research by Oklahoma State University’s Environmental Center is gratefully acknowledged. The National Science Foundation, Division of Materials Research, is thanked for Award Number 9871259 that provided funds for the X-ray powder diffractometer used in this investigation. REFERENCES 1
D. R. Burris, R. M. Allenking, V. S. Manoranjan, T. J. Campbell, G. A. Loraine, and B. L. Deng, “Chlorinated Ethene Reduction by Cast-Iron: Sorption and Mass Transfer’’J. Environ. Eng., 124, 1012-1019, 1998. L. Charlet, E. Liger, and P. Gerasimo, “Decontamination of TCE-Rich and URich Waters by Granular Iron: Role of Sorbed Fe(I1)”J. Environ. Eng., 125,2530, 1998.
Environmental Issues and Waste Management TechnologiesVIII
393
T. L. Johnson, W. Fish, Y.A. Gorby, and P. G. Tratnyek, “Degradation of Carbon-Tetrachloride by Iron Metal Complexation Effects on the Oxide Surface” J. Contaminant Hydrology, 29,379-398,1998. S. F. Ohannesin and R. W. Gillham, “Long-Term Performance of an In-Situ Iron Wall for Remediation of VOCS” Ground Water, 36, 164-170, 1998.
A.W. Apblett, L.D. Byers, and L.E. Reinhardt, “Dechlorination of Chlorocarbons by Molybdates and Vanadates” in Preprints of Papers Presented at the 213th ACS National Meeting, 37,300-302, 1997. Allen W. Apblett, B.P. Gran, and Katie Oden “ReductiveDechlorination of Chloromethanes Using Tungsten and Molybdenum Hydrogen Bronzes or Sodium Hypophosphite”in Chlorinated Solvents and DNAPLS; Reactive Permeable Barriers and Other Innovations, (ACS Book Series, Washington, DC, 2002), 154164. P.K. Dorhout and S.H. Strauss, “The Design, Synthesis, and Characterization of Redox-Recyclable Materials for Efficient Extraction of Heavy Element Ions from Aqueous Waste Streams”, ACS Symposium Series, 727,53-68, 1999. “Powder Diffraction File (PDF-2)” (International Centre for Diffraction Data, Newtown Square, PA). 9
A. I. Vogel, G. H. Jeffery, J. Bassett, J. Mendham, and R. C. Denney, Vogel’s Textbook of QuantitativeAnalysis, (Longman Scientific and Technical: Burnt Mill, Harlow Essex, UK, 1989), pp. 458-459. l0 F. D.
Snell, C. T. Snell, and C.A. Snell, “Thorim by Carminic Acid” in ColorirnetricMethods of Analysis, Vol. IIA, (D. Van Nostrand Co.: Princeton, N.J., 1959), pp. 518-519. l 1 K. Nakamoto, Infiared
and Raman Spectra of Inorganic and Coordination Compounds, 4th ed. (John Wiley & Sons:, New York, 1986).
394
Environmental Issues and Waste Management TechnologiesVIII
KEYWORD AND AUTHOR INDEX Absorption, 385 Actinides, 301, 314 Akai, T., 23, 39 Alam, T.M., 377 Al-Fadul, S.M., 15 Anderson, G., 177 Apblett, A., 15, 385 Attard, D.J., 321 Awano, M., 105 Barrium hollandite, 23 1 Bateman, K.J., 355 Begg, B.D., 313 Bennett, J.P., 3 Bibler, N.E., 209 Bickford, D.F., 123 Blum, A.G., 199 Blumenkranz, D.B., 209 Borosilicate glass, 169, 209, 215 Brossia, S., 283 Buechele, A.C., 225,253 Cahill, T.A., 59 Calcium carbonate, 67 Carter, M.L., 321 Cassingham, N., 337 Catalyst, 105 Cement, 39 Ceramic waste form, 355, 363 Cesium, 231,377 Chehbouni, M., 385 Chen, D., 23,39 Cherry, B.R., 377 Choi, K., 177 Chromium oxide, 347 Cliff, S.S., 59 COGEMA, 113 Construction material, 39
Corrosion, waste glass, 245, 291 Corrosion, waste package, 263 Crawford, C.L., 209 Crum, J.V., 141 Crystalline silicotitanate, 377 Crystallinity, constraint, 133 Day, D.E., 329,347 Decomposition, 67 Defense Waste Processing Facility (DWPF), 123 Delisting, 83 Desvaux, J.L., 113 Dilatometer, 67 Disposal, 123 Dissolution, 235 DOE, 95 Do-Quang, R., 113 Dredge sediment, 31 Dunn, D., 283 Durability, 185 Ebert, W.L., 235 Edwards, T.B., 199 Electronic applications, 74 Emissions, 49, 67, 105 Emissions, particulate, 59 EPA, 83 Erickson, A., 185 Extractants, magnetic, 15 Feng, K., 67 Ferrara, D.M., 209 Fission products, 3 13 Fluid chemistry, 263 Fracture, glass, 275 Fuel, nuclear, 113
Environmental Issues and Waste Management Technologies VIII
395
Gan, H., 215,225 Glass block, 275 Glass bonded, 363 Glass crystallization, 159 Glass fracture, 275 Glass melter, 123, 133, 141 Glass viscosity, 199 Glass waste, 39 Glass, borosilicate, 169, 209, 2 15 Glass, colored, recycling, 15 Glass, partially crystallized, 29 1 Glass, phase separation, 15 Glass, simulated waste, 199 Glass, iron phosphate, 329, 347 Glass, nitrate containing, 49 Glass, phosphate, 337 Glass, sulfate containing, 225 Godon, N., 275 Gombert, D., 329 Goodwin, S.M., 141 Haber, R.A., 31 Hamel, W., 95 Hanford, 95, 151, 209,225,253, 347 Hanna, J.V., 3 13 Health, 74 Heavy metals, 385 Heckendorn, EM., 123 High-level waste, 123, 133 High-salt waste, 37 1 High-chrome, 347 High-level waste (HLW), 95, 141, 159, 185, 235, 263, 275, 283, 291, 301, 337,347,355,363 High-level waste, liquid, 113 Hill, K., 31 Hollandite, 23 1 Hot isostatic pressing (HIP), 263, 355, 363 Hrma, P., 133, 151, 159, 245, 291, 337 Huang, W., 347 Huffman, L., 95
396
Hunter, B.A., 313 Hydration, 253 Idaho, 185 INEEL, 169, 177, 329 Iron phosphate glass, 329, 347 Jain, V., 263,283 Jantzen, C.M., 83 Jimenez-Cruz, M., 59 Jones, L.E., 49 Jouan, A., 113 Katayama, S., 105 Kelly, P.B., 59 Kim, C.-W., 177,329,347 E m ,D.-S., 133, 151, 169, 337 Kiran, B.P., 385 Kong, P., 177 Kriikku, E.M., 123 Kuraoka, K., 39 Kwong, K.-S., 3 Ladirat, C., 113 Land disposal, 209 Leach testing, 321 Leaching, 39,215,275 Lead, 75 Legislation, 74 Lerchen, M., 95 Li, H., 313 Lombardo, S.J., 67 Low activity waste (LAW), 209, 225, 253 Low-level waste, 177 Luo, s.,49 Maeda, K., 105 Magnetic extractants, 15 Martin, L.C., 83 Mass spectrometer, 67 Masui, H., 23
Environmental Issues and Waste Management TechnologiesVIII
MatyA-, J., 133 Melter, 123, 133, 141 Microwave heating, 363 Micro-XRF, 59 Mid-Delaware River, 31 Minet, Y., 275 Mitchell, D.R.G., 321 Mixed waste, 185 Modeling, 133,235, 263 Models, 151 Molybdenum hydrogen bronze, 385 Monitoring, waste package, 283 Mooers, C.F., 253 Mougnard, P., 113 Muller, I.S., 209
Refractories, 3 Regulations, 95 Repository, 235,263,283 Reuse, 3 Riley, B.J., 291 Ruthenium oxide, 141
Salt wastes, 371 Sampling, 355 Savannah River Site (SRS), 83 Schatz, T.R., 253 Schoenung, J.M., 75 Scholes, B.A., 185 Schumacher, R.F., 199, 209 Sediment, dredge, 3 1 Selective catalytic reduction (SCR), 105 Nelson, L.O., 177 Sensors, 283 Niobium, 377 Shackelford, J.F., 59 Nitrates, 49 Shin, S.-W., 177 NOx, 49, 105 Nuclear fuel, spent, 113,355,263,363 Shirakami, T., 39 Silica, 39 Nyman, M., 377 Silicotitanate, 377 Simulation, 199 O’Holleran, T.P., 355, 263 Sintering, 67 Organics, 15 Sludge, wastewater treatment, 83 Smith, G.L., 209, 371 I?articulate emissions, 59 Smith, H.D., 209,371 1?eeler, D.K., 169,199 Smith, M.E., 123 1?egg, I.L., 209, 215, 225, 253 Sodium bearing waste, 169, 329 I?erera, D.S., 313 Sodium extraction, 39 1?hase equilibria, 185 Solubility, 159 1?hase separation, glass, 15 Sridhar, N., 263 1?ickett, J.B., 83 Strontium, 377 1?lutonium, 301 Sulfate, 225 1?olycerams, 371 Sulfur, 225,337 1?olymer composite, 37 1 Sundaram, S.K., 141 Swanberg, D.J., 209 Radionuclides, 385 Synroc, 231,313 Raman, S.V., 185 Raw material, 31 TCLP, 215 Ray, C.S., 347 Test, 199, 291, 283 Recvclinrz, 3, 15
Environmental Issues and Waste Management Technologies VIII
397
Tile, 31 Titanate ceramics, 23 1,301 Toxicity, 215 Trad, T.M., 15 Urabe, K., 39 Uranium, 301
Zahir, M.H., 105 Zareba, A.A., 185 Zelinski, B.J.J., 371 Zhang, Z., 313 Zhu, D., 329 Zirconolite, 3 13
Vance, E.R., 301, 313, 321 Vapor phase hydration, 253 Vidensky, I., 225 Vienna, J.D., 151, 159, 169, 209, 245, 291,337 Viscometer, 199 Viscosity, 185 Vitrification, 113, 169, 177, 225, 329, 337 Vitrified mixed waste, 83 Waste glass, 39, 151, 159, 225, 235 Waste glass, corrosion, 245 Waste package, 263,283 Waste, high-level, 95, 141, 159, 185, 235, 263, 275, 283, 291, 301, 337, 347,355,363 Waste, sodium bearing, 329 Wastewater treatment sludge, 83 Water, 15, 385 Wiemers, K., 95 Willwater, T.M., 141 World Trade Center, 59 Wysoczanski, R., 253 Xia, G., 371 X-ray fluorescence, 59 Yamamoto, Y., 39 Yang, L., 283 Yazawa, T., 23, 39 Yeager, J.D., 245 Yucca Mountain, 235, 263, 283
398
Environmental Issues and Waste Management Technologies VIII
