Click for next page ( 49


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 48
5 immobilization To reduce the risk of radionuclide transport to the environment, high-level radioactive waste must be immobilized (i.e., converted to a stable solid form) before its disposal in a geological repository. The durabi I ity of the immobi I ized waste form, together with the corrosion- resistant overpack protection agai nst grou ndwater i ntrusion i ncl uded i n the design of the future repository, prevent or minimize migration of radioactive elements to the environment. Because repository space is limited, it is important to achieve the highest waste-volume reduction to minimize the number of containers needed for the immobilized waste. Hence, this would lead to reduced costs by more efficient use of waste immobi I ization, storage and transport faci I ities, and possibly to more efficient use of repository space.3 The current immobilization technique used by DOE for HLW is vitrification in a borosi I icate glass matrix (see Sidebar 5.1 ). The DOE vitrification process involves blending HLW with borosilicate glass frit or glass precursor chemicals, such as oxides and carbonates. The mix- ture is heated in a Joule-heated melter (see Sidebar 5.2) to form a molten glass, which is then poured into stainless steel canisters and allowed to cool. The DOE is currently operating HLW vitrification plants at the SRS and at the WVDP. The current plan at the Hanford Site is to vitrify both HLW and LLW in borosilicate glass.2 Finally, the INEEL will also immobilize its SBW in borosilicate glass and is now considering the option of vitrification for its calcined waste. 'The cost of producing an HLW glass log is approximately $1 million. A 1 per- cent increase in waste loading at the SRS could reduce cleanup costs by $200 mi 11 ion (Hrma et al., 1 998). 2The TPA decision to vitrify the LLW at Hanford is a departure from the strategy used at the SRS and at the WVDP, where LLW is instead immobilized in a cement- based material, referred to as "saltstone" at the SRS (TPA, 1998). H ~ G H - L E V E E W A S T E 48

OCR for page 48
SIDEBAR 5.1 SITE-BY-SITE HLW IMMOBILIZATION BASELINES Immobilization at the Hanford Site Both the HLW and the LLW will be vitrified at Hanford using Joule-heated melters, commencing in 2007.The Hanford tanks contain a wide range of reprocessing chemicals and wastes from the early bis- muth phosphate and REDOX processes, as well as PUREX wastes. During Phase I of waste immobiliza- tion at Hanford, ending in 201 8, approximately 1 0 percent of the HLW is to be retrieved and vitrified. Little or no blending between tanks to smooth any waste composition variability is planned during Phase I because of high costs and limited availability of free tank space. Furthermore, although the Hanford melter will also use Joule-heating to immobilize HLW in borosilicate glass, it is proposed to use a melter feed containing raw chemicals (oxides, carbonates, etc.), rather than frit, to give greater flexibility in glass-batch preparation. Immobilization at INEEL The site has calcified liquid, acidic HLW to produce approximately 3.8 million liters of granular solid waste that is currently being stored, pending a decision on final immobilization and/or disposal.The remaining 5.3 million liters of liquid SEW is to be vitrified (Huntoon, 2000). A previous NRC committee on the INEEL recommended in its report that the calcified material be stored until the repository loca- tion and waste form acceptance criteria have been established (NRC, 1999b).That committee also advocated further investigation of a number of viable candidate waste forms for the calcified wastes, in addition to borosilicate glass,with the main objective being to increase the waste loading.These candidate waste forms include (1) alternative glass compositions, including high-waste-loading glasses prepared in single-use containers; (2) crystalline ceramics prepared by hot uniaxial or isostatic press- ing; (3) glass-ceramic materials; and (4) cement-based waste forms. Immobilization at SRS In the DWPF vitrification facility at the SRS, HLW is immobilized in a borosilicate glass matrix.The DWPF melter, described in Sidebar 5.2, uses a wet feed (approximately 50 percent water) comprised of a slurry of waste and frit.The waste originates from a two-year homogeneous batch where HLW, retrieved from different tanks, is blended. A target "window" of feed compositions, consisting of a ternary diagram based on mixtures of two waste feeds and a glass frit (Figure 5.1) is used to determine the composition of the melter feed.This window has been established on the basis of previous melting trials with slurry feeds of frit plus simulated wastes. In this process control strategy, called Product Composition Control System (PCCS), portions of the two-year batch of waste are fed into a tank where, depending on the waste feed characteristics, the amount of frit (of a given composition) is adjusted so that the frit volume is minimized and the predicted properties of the final glass will fall within the tar- get window. A statistically designed variability study comprised of approximately 30 glass melts is per- formed on every waste batch (about every 2 years) to ensure that waste-frit mixtures are correctly pre- dicted by the PCCS models.The use of this process control strategy and a large homogeneous two-year sludge batch minimizes the number of actual radioactive glasses that need to be analyzed from the canisters produced. Therefore, control of the final radioactive glass composition mainly relies on the PCCS process control strategy by ensuring that the melter feed composition is such that the resulting glass exceeds, with a 95 percent confidence level, the measured durability of the benchmark Environmental Assessment (EA) glass. m m 0 b i I i z a t i 0 n 49

OCR for page 48
Immobilization at WVDP Immobilization operations at the WVDP involve a Joule-heated melter similar to that of the DWPF. However, the process control strategy is different from that used at the SRS. The WVDP uses glass form- ing chemicals rather than premelted frit. During production, numerous analyses are made on large vol- umes of wastes, to allow the feed composition to be adjusted to fit within the target composition win- dow. Numerous samples on every canister of the final glass product are also required to characterize glass-waste properties. ·eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee Frit A .8 .6 .4 .. - .~ / .- ~,%....~ .2 / ., .. . . ',: , ~ .. ...\\ .6 . ... - / . . it\ ~' '. . - . . . .. . - \ Waste 1 \ .8 .6 .4 .2 Waste 2 FIGURE 5. 1 The PCCS window for the DWPF melter feed, based on a mixture of two types of HLWstreams, Waste 1 and Waste2, and glass frit. The solid white area in the diagram, called "Process Acceptable Region (PA R),"is defined by the mechanistic models in PCCS that bounds the range of allowable waste compositions and frit blends. The "target"point indicates maximum waste loading with the minimum amount of frit for a given batch of waste being fed to the melter. Waste-frit blends that fall within the PAR yield a final glass product that meets the appropriate processing and durability criteria. This PCCS allows minimal sampling on the final glass product while maintaining at least 95 percent confidence that the glass product falls within the qualified glass region. The window is defined in terms of relevant glass properties, including such constraints as melt viscosity, product durability and homogeneity, and the temperature for incipient crys- tallization (the liquidus). The viscosity and liquidus must be low enough to process the feed in the melter and pour it into the stainless steel canisters. The PCCS models that define the PAR target window have been established on the basis of H ~ G H - L E V E E W A S T E 50

OCR for page 48
data from 300 to 400 non-radioactive and radioactive laboratory melts and pilot-scale melting trials. To verily/ the PCCS model and the DWPF process composition window, an additional 485 validation glasses, including 237 full-scale canister glass samples, taken during DWPF non-radioactive startup, were analyzed (Jantzen et al., 1998). The PCCS models were developed from glasses whose compositions cover the range of waste streams expected to be processed over the lifetime of the DWPF (Postles and Brown, 1991). The cost of establishing the DWPF process control system was approximately $5 million (Janzlen, personal communication). SOURCE: DOE-Savannah River Site. SIDEBAR 5.2 OPERATION OF THE DWPF JOULE-HEATED MELTER The DWPF facility started non-radioactive operations in 1994 and was used to test a wide range of sim- ulated wastes that covered the range of all wastes anticipated to be processed during the DWPF life- time. Radioactive vitrification operations began at the DWPF in 1996.The Joule-heated melter in use at the SRS, shown below, is operated as follows. An initial charge of dry glass-forming ("batch") materials is fed to the melter and preheated by natural gas burners or electric heaters above the glass pool ("overhead plenum heaters").The batch becomes sufficiently electrically conductive between 600 °C and 700 °C to allow further heating by electric current.The batch is then heated resistively to approxi- mately 1150 °C by passage of an alternating current through the submerged melter electrodes (only 2 of the 4 electrodes are shown).The molten glass produced flows through a narrow region and is removed continuously from a side channel in the melter.The melter is replenished by feeding addition- al batch material onto the melt surface, where it forms a thick crust of abreacted material (the "cold cap") that serves to trap condensable volatile emissions and recycle them into the melt.The molten glass is then poured through the pour spout into stainless steel canisters and allowed to cool. Successful, continuous, long-term glass production normally requires a constant batch composition, uniform feed rate, and steady-state operating conditions. SOURCE: DOE-Savannah River Site. m m 0 b i I i z a t i 0 n 51

OCR for page 48
The committee identified the following issues related to the choice of borosilicate glass and Joule-heated melters for the immobilization of HEW: · limitations of borosilicate glasses as immobilization medium; · crystal content of the borosi I icate glass matrix; · Iong-term leaching properties of the borosilicate glass matrix; · use of unreacted glass-forming chemicals versus premelted glass frit; foami ng i n Jou [e-heated melters; precipitation of noble metals and crystalline phases in Joule- heated melters; limitations of Joule-heated melters in achieving higher process- ing temperatures; and · alternative immobilization processes to Joule-heated melting. Immobilization Issues The issues listed above are described in the following section, along with research activities that could contribute to their resolution. The overall objective of the long-term basic research recommended for immobilization is to provide the scientific basis to develop robust, high-loading3 immobilization methods and materials that could provide enhancements or alternatives to the current immobi- lization strategy. [imitations of BorosiIicate Glasses as Immobilization Medium Under the current waste acceptance guidelines for the future geo- logical repository (see Sidebar 5.3) the borosilicate glass waste form must meet certain performance requirements, developed on a case- by-case basis. Borosilicate glasses are appropriate immobilization media for many DOE HEW streams at the present level of waste load- ing (currently approximately 28 weight percent at DWPF on a dry cal- cine basis including Na2O and SiO2~. However, it may not be possi- ble to achieve substantial increases in waste loading using borosili- cate glasses. Furthermore, these glasses may not be the optimum media for some problematic wastes such as the INEEL calcines, as 3Waste loading is the fraction of waste contained in a glass log or other waste form. H ~ G H - L E V E E W A S T E 52

OCR for page 48
SIDEBAR 5.3 WASTE ACCEPTANCE CRITERIA The Waste Acceptance System Requirements Document~determines the conditions necessary to be met by spent nuclear fuel and HLW, in order for DOE to be able to accept it for disposal" in a geological repository (DOE-OCRWM, 1 999).The Yucca Mountain site in Nevada is currently under consideration for that repository.The DOE-OCRWM,which is responsible for the development of the HLW repository project, while the USNRC regulates the repository site, is currently developing the waste acceptance criteria. Before 1977, the United States expected to reprocess all spent fuel from commercial reactors. It was intended that all HLW from reprocessing of commercial fuel would be immobilized by incorporation in a borosilicate glass matrix, prior to disposal in a geological repository. Glass-technology programs were initiated to identify suitable waste-glass compositions that would be resistant to leaching under many repository conditions.The glass waste form would thus constitute a primary barrier against release of radionuclides to the environment. A similar search for a practical glass waste form was underway in France and the United Kingdom, two countries that intended to reprocess their spent reactor fuel. When the United States decided, at the beginning of 1977, to forego reprocessing of commercial reactor fuel, the primary form of HLW became spent fuel, and the plans and regulations governing repository disposal were changed accordingly. Later, the DOE was authorized to dispose of HLW from the defense program sites in the first HLW repository.The defense wastes are expected to constitute only about 10 percent of the HLW in the first repository, with the balance of about 90 per- cent being spent fuel from commercial nuclear reactors.The initial defense HLW designated for the first repository will be borosilicate glass from the SRS and from the WVDP. At the present time, the DOE wastes are scheduled for disposal along with spent commercial fuel in a common repository, where the performance acceptance criteria are based on spent fuel characteristics. The underlying objective for the DOE wastes is that disposal performance is predictable and at least as good as spent fuel. As part of the current SRS criteria for waste form acceptance, the borosilicate glass waste form must meet the following requirements (Janzten, 1 993a; 1 993b; 1999): 1. The glass must have a leach resistance greater than the EA glass, the benchmark waste glass identified in the DWPF EA. 2. The glass must exhibit no evident glass-in-glass phase separation. 3. The glass must be essentially free of crystal content. These criteria are SRS guidelines for their current glass waste form, and do not apply to waste forms to be produced at the other DOE sites. The other sites might consider alternative waste forms, and would have to establish their own criteria to achieve comparable performance and acceptance. m m 0 b i I i z a t i 0 n 53

OCR for page 48
stated by a previous N RC committee (N RC, 1 999b). Alternative waste form materials, such as other glasses, glass-ceramics, and polyphase ceramics, may be better suited for these types of waste. Off-gas particulates4 from slurry volatilization represent another potential problem with the use of borosilicate glasses because the particulate compositions will likely differ significantly from that of the original waste feed. Thus, they may have to be blended into the waste feed at relatively low concentrations because they would be highly enriched in volatile species, such as technetium, mercury, iodine, ruthenium, cesium, boron, sodium, and possibly molybdenum. In particular, the high volatility of boron from borosilicate melts is known to induce significant losses of cesium, volatized as a cesium borate (Vance et al., 1 988~. Although cesi u m volati I ization does not seem to create problems for the DWPF,5 it may become a significant factor with future cesium-rich wastes at the Hanford Site or at the SRS, particularly if an improved method for cesium removal from the salt fraction is adopted (NRC, 2000b). Another potential problem caused by off-gas emissions in noted in Sidebar 5.4. Precipitated effluents in the form of fine powders or nanocrystalline conden- sates. Measurements in the initial waste and in the glass indicated that greater than 90 percent of the cesium-137 is incorporated in the product glass (Bibler et al., 2000). SIDEBAR 5.4 VOLATILE EMISSIONS FROM THE DWPF MELTER LINKED TO EVAPORATOR SHUTDOWN? One example of a problem potentially exacerbated by the carryover of off-gas emissions is the severe sodium aluminosilicate fouling of the 2H evaporators at the SRS,where at least 1,100 kilograms of alu- minosilicate have been deposited in the evaporator, to depths of over a meter in some locations. Evaporators are used to concentrate supernatant liquid waste, thus conserving storage space.The car- ryover of condensate from the DWPF evaporator melter is also recycled back to the tank farm 2H evap- orators. It is possible that entrainment or volatilization of sodium, silica, and aluminum species from the DWPF melter off-gas was in part responsible for 2H evaporator fouling. In fact, the 2H evaporator has historically processed streams high in aluminum, and in the past, small quantities of aluminosili- cate buildup have been observed. Since the DWPF began operating and recycling condensate to the evaporator, the concentration of silicon in the latter has increased dramatically. However, a recent study has shown that only half of the aluminosilicate in the 2H evaporator originated from the frit car- ryover from the DWPF melter; the remaining half originated from a different evaporator and from lab- oratory analysis (Jantzen and Laurinat, in press). The EMSP is collaborating with the TFA, the Savannah River Technology Center, and the Oak Ridge National Laboratory to study the formation of aluminosili- cates under conditions similar to those in the 2H evaporator. H ~ G H - L E V E L VV A 5 T E ~ A ~ 54

OCR for page 48
A further issue relates to possible pre-blending of wastes from vari- ous tanks before feeding them to the melter. This method has been used at the SRS to increase the glass waste loading by smoothing the concentration of critical waste components. However, the proposed strategy for Phase I at Hanford is to empty and vitrify the wastes on a tank-by-tank basis, with little or no inter-tank blending of the wastes. Furthermore, the Hanford tank-to-tank composition variability is gen- erally much greater than that at SRS. Thus, this strategy will require separate composition windows for waste plus glass-forming materials to be developed for each tank and may ultimately increase the overall number of glass logs. For low-sodium wastes, the maximum waste loading is usually dictated by the concentrationts) in the waste feed of species with lim- ited solubility in borosilicate glass, such as halides, sulfates, phos- phates, chromium, and bismuth. Glass-in-glass phase separation and/or crystallization of possibly undesirable phases will occur if the solubility limits of these species are exceeded, producing a waste form that may not meet current acceptance criteria. A good example of this limitation is the sulfate content in the waste feeds, which is determined by the various pretreatment stages (see Sidebar 5.51. The solubility of sulfate in borosilicate glass is low, so that a high sulfate content in either the HLW or the LLW streams will lead to separation of a molten sulfur-rich phase within the melter causing foaming problems. Many waste ions, including cesium, are known to partition preferentially into the sulfate phase in coexisting sulfate-silicate melts.6 Furthermore, if the solubility limit of sulfate in 6p.J. Hayward, unpubl ished work, 2000. SIDEBAR 5.5 IMMOBILIZATION OF HANFORD LLW Consideration of LLW immobilization is not part of the task of this committee. Nevertheless, the quan- tity and composition of LLW will be dictated by pretreatment of the HLW feeds. In 1994, an amendment of the TPA between the State of Washington, the EPA, and DOE established that the LLW at Hanford will be immobilized by vitrification in borosilicate glass, using essentially the same technology proposed for HLW immobilization (TPA, 1 998).The Hanford LLW stream will consist predominantly of soluble salts (nitrates, sulfates, phosphates, and carbonates) of sodium, aluminum, and potassium, together with traces of fission products and transuranic elements.The limiting factor in determining waste load- ing in the LLW glass will likely be the sulfate content.This limit will produce an estimated 20 percent increase in the required glass volume for LLW immobilization, compared to the volume that would be required if the LLW sodium content were the limiting factor (Pegg, 2000). m m 0 b i I i z a t i 0 n 55

OCR for page 48
H LW or LLW glass is exceeded i Inadvertently, separation and accumu- lation of a molten sulfate or sulfide phase at the melt surface could also cause enhanced corrosion of the upper electrodets) and refracto- ries. The alloy used for electrodes, Inconel-690@, is known to be sus- ceptible to attack from sulfur compounds, particularly under reducing conditions. tong-Term Research Needs Basic long-term research in material sciences is needed to seek alternative waste form materials, such as glass-ceramics and polyphase crystalline ceramics, for producing acceptable immobilized waste with higher concentrations of HLW from variable-composition feeds. Descriptions of many previously developed alternative waste form materials for HLW can be found in the relevant literature (Hayward, 1988; Lutze and Ewing, 1988; Donald et al., 1997~. Further development of some of these materials could allow this goal to be achieved. Further research is needed to identify more economic alternatives to borosilicate glass for immobilizing LLW in Hanford. One example could be to investigate alternative waste form materials (e.g., cement- based materials) that can incorporate higher sulfate concentrations than are possible with borosilicate glass.7 In the event of future immobilization of Hanford LLW in Joule- melted borosilicate glass, research will be needed to study the corro- sion mechanisms) of Inconel-690'~' in glass melts containing high sul- fur concentrations under various redox conditions, with the goal of minimizing corrosion and/or identifying more corrosion-resistant al lays. Crystal Content of the BorosiIicate Glass Matrix The current stipulations for the SRS borosilicate glass waste form state that the waste glass should not contain any significant degree of crystallinity or glass-in-glass phase separation Oantzen et al., 1999~. Crystallization can occur during cooling of molten HLW glass if there is sufficient overlap between the temperature ranges for substantial crystal nucleation and crystal growth. Potential crystalline phases appearing in SRS and Hanford wastes glasses are spinels (A2+LB3+1204, em.. NiFe-OA). clinoovroxenes (orincioallv acmite NaFeSi-O ). alkali L) ' ~ of'' ~ ~ ', , , ~ O', aluminosilicates (principally nepheline NaAISiO4, albite NaAISi3O~, 7The committee is aware of the difficulties of reversing the TPA decision to vit- rify LLW (TPA, 1998). However, the EMSP should investigate alternative waste forms as part of a contingency approach to the current baseline program. H ~ G H - L E V E E W A S T E

OCR for page 48
which can occur in partial solid solution with nepheline, and eucryptite LiAISi3O~), lithium silicates (e.g., Li2SiO3), cristobalite (SiO2), hematite (Fe203), and zircon (ZrSiO41. Some projected precipi- tations in INEEL glass compositions are nepheline, fluorapatite (Ca~0PPO416F2), lithium phosphate (LiPO4), baddeleyite (ZrO2) and alkali aluminosilicate sulfides. Above the glass transition temperature of HEW glasses, canister centerline cooling rates of approximately 0.05°C per second permit rapid-nucleating and rapid-growing phases, such as spinels and nepheline, to precipitate but disfavor difficult-to- nucleate or slow-growing phases like acmite or zircon. Spinels have been identified as major crystalline phases in glass after heat treat- ment at temperatures between 500°C and 900°C in WVDP glass com- positions Gain et al., 1993) and, as indicated earlier, may accumulate in Joule melters and be entrained in glass carried over from the melter. Uncontrolled crystallization or phase separation of certain of these crystalline phases within the glass log during cooling has the potential for reducing the durability of the final waste form. The glass durability is determined by employing the product consistency test (PCT), which measures the release rates for sodium, lithium, silicon, and boron dur- ing water leaching. The deleterious influence on glass durability of certain crystalline phases derives from both chemical and mechanical effects on the surrounding glass: the residual glass composition is altered, and the glass matrix is stressed by the volume mismatch of the crystal and the glass space it replaces. Survey studies (Bailey and Hrma, 1995) have suggested that the residual glass composition is the major factor that controls the PCT response of glasses with durable crystal I i ne phases. Spinels are general Iy conceded (Jantzen and Bickford, 1 985) to have little effect on glass durability, because they do not substantially alter the chemistry of the remaining glass-forming and modifying ele- ments (silicon, aluminum, boron, and sodium). Moreover, spinels are characterized by their cubic crystal symmetry, which leads to near- isotropic~ interface energies and strain distributions. Therefore, it is likely that spinels lead to equiaxed crystalline precipitate with mini- ~lsotropy refers to having equal physical properties (such as refractive index, thermal expansivity, elastic constants) in different crystallographic directions. Cubic crystals are usually more isotropic than other crystal systems with lower symmetry (tetragonal, hexagonal, orthorhombic, monoclinic, or triclinic). However, in some cases, cubic crystals can exhibit properties that are far from isotropic. m m 0 b i I i z a t i 0 n

OCR for page 48
mal tetragonal distortion of the surrounding glass matrix. The reported effects of acmite and related clinopyroxenes, which are non-cubic and non-isotropic and comprise one of the most likely crystalline fractions, are controversial Oantzen and Bickford, 1985), but the most recent Pacific Northwest National Laboratories (PNNL) studies con- clude that acmite has virtually no impact on PCT release rates (Riley et al., 2001~. Nepheline precipitation appears to have the most detri- mental impact on dissolution of Hanford and Savannah River HLW glasses and can decrease chemical durability by several orders of magnitude (Ri ley et al., 2001 ). Precipitation of cristobal ite (which is nonetheless cubic) and baddeleyite (which is nearly so) also impact durability negatively. Substantial crystallization of a number of other phases has been shown to have little or even a positive impact on glass durability. Titanium (derived from monosodium titanate or crystalline silicoti- tanate ion exchangers, see Chapter 4), zirconium, phosphorous, and fluorine can all function as effective nucleating agents for crystalliza- tion and are commonly used for nucleating commercial glass-ceram- ics. Thus, it may be necessary to control the concentrations of these potential nucleating agents in the HLW feed to the melter to avoid crystallization of phases that could adversely affect waste form dura- bility. The fact that some crystalline precipitations appear to have little or no adverse impact on chemical durability of HLW glasses suggests that it may be possible to relax the restriction on crystal content in the glass to accommodate a small content of crystalline phase. In turn, this could allow the waste loading in the glass log to be increased. The committee is aware of the research efforts to this end being pursued at the SRS and at PNNL. DOE investigators are currently exploring the potential to increase waste loadings for Hanford, Savannah River, and INEEL by allowing crystalline precipitation upon cool i ng with i n the can ister (Pittman et al ., 2001 ). Other research efforts related to DOE's programs are under way at universities such as The Catholic University of America (Kot and Pegg, 2001 ) and the University of Missouri (Marasinghe et al., 1999, Ray et al., 19991. These research efforts could be effectively complemented by a long- term basic research program within the EMSP to obtain innovative approaches on the effect of crystal precipitation in borosilicate glass. tong-Term Research Need Long-term basic research is needed to broaden the envelope of acceptable borosilicate glass compositions to include a level of crys- tal content that does not adversely affect product durability. H ~ G H - L E V E E W A S T E

OCR for page 48
[ong-Term [caching Properties of BorosiIicate Glass Matrix There has been a considerable amount of experimental research on the chemical durability (leach resistance) of various glass waste forms under a variety of hypothetical repository conditions and on glass corrosion mechanisms, alteration products, and long-term radia- tion-induced degradation.9 However, as noted elsewhere (NRC, 1 996a), the continued development of phenomenological models4° that would allow glass-leaching data to be extrapolated over long time intervals would be advantageous. tong-Term Research Needs Long-term basic research is needed to further develop and verify phenomenological models to predict long-term leachabi I ity of borosilicate glass and other waste forms. Such a model should be developed and verified as a contingency against future waste form disposal issues, such as performance. A similar predictive model would be needed to support the possible future use of alternative waste form materials with higher waste loading, including glasses, glass-ceramics, and polyphase ceramics. The models should also i ncl ude consideration of the i nfl uence on waste form du rabi I ity of such factors as groundwater radiolysis and internal radiation damage (Weber et al., 1997; 19981. Use of Unreacted GIass-Forming Chemicals Versus PremeIted Glass Frit Plans to immobilize HLW in borosilicate glass at Hanford include the option of using a melter feed containing "raw" chemicals (mainly oxides and carbonates), rather than premelted frit. Given the great variability of waste streams in Hanford, this option would avoid the tailoring of premelted frit to the different waste compositions to fall within the target composition window. The rationale for using premelted frits in the melter feeds at DWPF was that residence time in the melter would be reduced because many of the glass-forming reactions would have been performed 9Much of this research is documented in journals, symposium proceedings, (e.g., the annual Materials Research Society "Scientific Basis for Nuclear Waste Management" proceedings) and DOE workshops. 'PA phenomenological model is a multi-component model with predictive power that combines a series of mathematical descriptions of the individual phe- nomena involved. See the glossary in Appendix G. m m 0 b i I i z a t i 0 n 59

OCR for page 48
ahead of time during frit manufacture. However, this would not be true if the overall rate of glass production were governed by the rate of waste dissolution in the melt pool or by the rate of heat transfer from the melt pool to the cold cap. There appears to be some uncer- tainty about this issue that should be resolved. The use of unreacted glass farmers at Hanford could conceivably produce unanticipated melter problems (e.g., corrosion, foaming, and precipitation) that have not previously been encountered at the WVDP and at the SRS. For instance, the volume and complexity of the off-gas emissions could be greatly increased from volatilization of the slurry water content and from chemical breakdown of nitrates, nitrites, oxalates, and other organic molecules. These emissions can cause potential entrainment of other volati le species, incl uding technetium, mercury, iodine, ruthenium, cesium, boron, and sodium. tong-Term Research Need Long-term basic research is needed to evaluate the controlling parameters of reaction rates and heat transfer processes in the melter. Results will strengthen the scientific basis for a rational choice between using un reacted glass farmers and using premelted frit in waste feeds for future melter designs (including Hanford melters). Foaming in JouIe-Heated Melters Foaming is the result of redox reactions within the melt and also the breakdown of anions, such as nitrates and carbonates, that gener- ate gas during melting. Excessive foaming can form a physical and thermal barrier between the cold cap and the melt pool and can ulti- mately lead to melter shutdown. Foam formation is generally associat- ed with highly oxidizing conditions in the melter Oain and Pan, 20001. The foam acts as an insulating layer of bubbles between the melt pool and the newly introduced waste slurry feed, eventually forming a cold-cap "bridge" and preventing further waste feed from dissolving in the melt. Foaming also introduces the possibility of enhanced corrosion of the upper electrodeks) and refractories. The problem of foaming in melters has been encountered at differ- ent DOE sites. At the WVDP this phenomenon has been 'accommo- dated' by adding a reducing agent (usually sugar) to the melter feed. The mechanism by which sugar reduces the formation of micro-bub- bles during melting is not well understood. Foaming continues to be an issue at the SRS Oain and Pan, 2000) and it is one of the anticipat- ed problems at the Hanford Site, because of the wider waste compo- sition range. Careful redox and rheology control is required to prevent foaming during the water boil-off stage. A further factor that would probably exacerbate any foaming tendency in Hanford is the pro- H ~ G H - L E V E E VV A 5 T E ~ A ~

OCR for page 48
posed use of oxides and carbonates as glass farmers, rather than frit. Any foaming tendency within the cold cap would be exacerbated by CO2 generation from thermal decomposition of carbonates (Li2CO3 and Na2CO3) in the glass batch. Similar foaming problems were also encountered with the Joule-heated melter used for vitrifying mixed wastes at the Fernald site in Ohio. These problems were attributed to poor red ox control, and also by foaming within immiscible sulfate layers that formed on the melt surface. Further details on foaming and red ox control in melters are described by Jain and Pan (20001. tong-Term Research Needs Long-term basic research is recommended to characterize the behavior of the cold cap formed on the melt surface. Specifically, the sequence of reactions occurring in the cold cap and their influence on foaming tendency do not seem to be well characterized. The items to be eval uated i ncl ude (1 ~ the rates of water removal and breakdown of salts (e.g., nitrates, carbonates, and formates) and of organic addi- tives (e.g., sugar, urea) used to control melt redox conditions; (2) the influence of feed chemistry, including sulfate content; and (3) possible oxygen evolution from red ox reactions occurring within the melt. Thus, it may be possible to minimize or eliminate the potential for foaming using modifications to pretreatment and/or to the physical or chemical properties of the waste stream, (e.g., by pH or red ox adjust- ment, change in the solids-liquid content, or particle size adjustment). Precipitation of Noble Metals and Crystalline Phases in JouIe-Heated Melters Future melting campaigns at the SRS and at the Hanford Site will involve tank wastes with higher concentrations of noble metals (palla- dium, rhodium, and ruthenium), derived from the fission of uranium- 235. Ruthenium is the most abundant noble metal in the Hanford HEW (Jain and Pan, 20001. Noble-metal precipitation within the melter could cause plating-out, short-circuiting, and downward drilling of accumulated metal into the refractory floor, all of which would reduce melter life. Metallic precipitates also have the potential to cause alloying reactions with the Inconel-690@ electrodes. Some or all of these problems are reported to have occurred elsewhere, such "This phenomenon involves an enhanced refractory attack in a glass melter at contact sites between metallic inclusions and refractories.Typically, the attack involves gravity-assisted "drilling" of vertical holes or cavities in the refractories that constitute the melter floor. m m 0 b i I i z a t i 0 n 61

OCR for page 48
as in the first melter used at the Pamela vitrification planter (Demonic, 1996). Many of the Hanford wastes, and future wastes to be immobilized at the SRS,43 will also have relatively high iron, aluminum, nickel, manganese, and chromium contents, which, together with chromium oxide sludge from refractory corrosion, will likely cause precipitation of crystalline A2+B3+204-type spinel compounds. These dense insolu- ble phases may accumulate on the melter floor and could conceiv- ably cause throat and/or riser blockage. Furthermore, many iron and chromium-rich spinel compounds exhibit relatively high electrical conductivities at glass-melting temperatures. Thus, their precipitation could lead to possible disruption of the electrical current distribution within the molten glass pool. tong-Term Research Need Modeling efforts, possibly combined with reduced-scale experi- ments, are recommended to study the consequences of precipitation and accumulation of noble metals and spinels on the melter floor. The ultimate goal of this long-term basic research is to increase glass pro- duction rates and prolong the operating life of the melter. [imitations of JouIe-Heated Melters in Achieving Higher Processing Temperatures As noted previously, it may be advantageous to develop alternative glass or glass-ceramic waste form materials to the present generation of borosilicate waste glasses in order to achieve higher waste load- ings, or to immobilize problematic wastes with unusual compositions. For example, this could be the case if the vitrification route is chosen for immobilizing the INEEL high-aluminum and high-zirconium cal- cines, tank heels, and other secondary waste streams from pretreat- ment and vitrification. These alternative waste form materials will like- ly require higher melting temperatures and, if fabricated using Joule- heated melters, may also require new electrode alloys and glass-con- tact refractories with improved corrosion resistance. '2The Pamela vitrification plant consists of two Joule-heated melters operated in Mol. Belgium, from 1985 to 1991. The first melter was shut down after three years as a result of electrical failure from buildup of noble metal sludge on the floor. '3The SRS staff has tested a wide range of glass compositions simulating the dif- ferent types of waste streams projected for the next 25 years (Postles and Brown, 1991). Based on this study, no problems arising from future waste-composition variations are anticipated at the SRS. H ~ G H - L E V E E W A S T E 62

OCR for page 48
tong-Term Research Needs There is a need to study higher-temperature Joule-melting tech- niques as a step toward developing alternative glass or glass-ceramic waste forms with higher waste loadings and as a contingency against difficulties with future problematic wastes. Long-term basic research is required to identify (1 ) improved electrode materials, such as new alloys, ceramics, or cermets; (2) advanced refractories; and (3) alterna- tive electrode-refractory configurations. In all cases, the primary goal is to minimize hi~h-temneratr~re corrosion in the presence of high .. O.. ..., . .. .. . .. ... .. , . .. . ... O.. concentrations of simulated waste under appropriate red ox condi- tions. Alternative Immobilization Processes to JouIe-Heated Melting Alternative immobilization processes may be advantageous as a contingency against unforeseen problems with continuous (Joule- melter) vitrification during the Hanford Phase I and 11 programs. A recent survey of waste immobilization technologies gives examples of some batch-processed alternatives to the continuous melters in cur- rent use at DOE sites, including processes such as induction melting, "cold-crucible" melting, or microwave melting (Jain, 2001 ). In many cases, these alternative processes can avoid some of the inherent problems with a continuous melter, including refractory corrosion and precipitation of noble metals and crystalline phases, although they would likely introduce other technical issues. In general, the use of batch melting would allow greater flexibility in the range of compositions and temperatures for vitrification. This flexibility could be important if a glass or glass-ceramic waste form with a higher melting temperature were selected to immobilize future problematic waste streams or to increase the waste loading. Furthermore, the eventual task of decommissioning and disposal may be simpler with a smaller batch-type melter than with a continuous melter. Some of these alternate immobilization technologies may require drying or pre-calcining of the HEW slurry before it could be blended with other glass or glass-ceramic precursor chemicals and processed. The committee notes that pre-calcination is used at the La Hague, Marcoule (]ouan et al., 1996) and Sellafield (Fairhall and Scales, 1996) vitrification facilities, albeit with an acidic waste stream (i.e., with little or no NaNO3 content) and has also been performed on much of the I N EEL waste. Pre-calcination may offer a number of technical advantages, including prior removal of process off-gases, such as water and nitro- m m 0 b i I i z a t i 0 n

OCR for page 48
gen oxides (NOx), and possible elimination of foaming and associated problems. However, it would add a further step to the overall immo- bi I ization process and wou Id i evolve hand I i ng fi ne powders. Other issues could include exothermic nitrate-organic reactions and forma- tion of viscous sodium nitrate melts. While pre-calcining of the HLW feed does not, in itself, give higher waste loading, it may be a neces- sary step i n any i n novative process to ach ieve th is goal . The I after could include batch-type processes where initial dry blending of the HLW feed with processing additives (e.g., frit or glass-forming chemi- cals) is required, or where it is important to minimize the initial vol- ume of HLW feed. tong-Term Research Needs Long-term basic research is needed to identify and develop alter- native melting techniques, including batch-type processes using con- cepts other than the continuous melters in current use at DOE sites for preparing waste forms with higher waste loadings. This research could include the study of drying or pre-calcining options for the waste feed. H ~ G H - L E V E E W A S T E 64