3
Physical and Chemical Separations

On a weight basis, most of the calcine is nonradioactive. Therefore, separation of the less radioactive constituents into a low-level waste stream has the potential to reduce the volume of high-level waste (HLW) to be disposed. Various separations objectives1 [e.g., separation of transuranic (TRU) components only, or separation of TRU, cesium (Cs), and strontium (Sr) components from the remainder of the waste] have been proposed in U.S. Department of Energy (DOE) literature. All of the options require calcine dissolution followed by separation of solids from liquids and additional chemical separation steps on the liquid fraction.

This chapter discusses general separations strategies and specific separations methods that could be used to develop waste fractions of high and low activity. Physical (solid-liquid) separations are treated first, followed by chemical separations and detailed discussion of cesium, strontium, and TRU separation plans.

THE SEPARATIONS APPROACH

The calcines stored at Idaho National Engineering and Environmental Laboratory (INEEL) contain large quantifies of nonradioactive constituents from the fuel assemblies2 and from materials added during fuel reprocessing and waste conditioning and calcination. Because a high weight and volume fraction of the calcine is nonradioactive, separation of radionuclides from this bulk is a way to reduce the volume of HLW to be immobilized prior to disposal. If the costs of separations process steps were not excessive and the risks acceptable, this strategy might reduce overall disposal costs and hazards. A full separations approach could remove all of the long-lived actinides, plus the 137Cs and the 90Sr. These are the critical elements whose removal might then allow the remaining material of reduced activity to be delisted from further management as HLW.3 The residues remaining after this partitioning could fall into the

1  

Another important separations objective is the segregation of hazardous chemical constituents [e.g., mercury (Hg) and lead (Pb)] from nonhazardous components. This issue is discussed in a limited way in the context of each separation option of this chapter, and more broadly in Chapters 9 and 10.

2  

Some of these initially nonradioactive fuel assembly materials become radioactive during reactor operations. For example, in the initially nonradioactive zirconium cladding material, Zr-93, a long-lived beta (β) emitter, is generated during exposure to radiation inside nuclear reactors.

3  

This reclassification from HLW to some other category may require regulatory approval, most probably from the USNRC.



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 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory 3 Physical and Chemical Separations On a weight basis, most of the calcine is nonradioactive. Therefore, separation of the less radioactive constituents into a low-level waste stream has the potential to reduce the volume of high-level waste (HLW) to be disposed. Various separations objectives1 [e.g., separation of transuranic (TRU) components only, or separation of TRU, cesium (Cs), and strontium (Sr) components from the remainder of the waste] have been proposed in U.S. Department of Energy (DOE) literature. All of the options require calcine dissolution followed by separation of solids from liquids and additional chemical separation steps on the liquid fraction. This chapter discusses general separations strategies and specific separations methods that could be used to develop waste fractions of high and low activity. Physical (solid-liquid) separations are treated first, followed by chemical separations and detailed discussion of cesium, strontium, and TRU separation plans. THE SEPARATIONS APPROACH The calcines stored at Idaho National Engineering and Environmental Laboratory (INEEL) contain large quantifies of nonradioactive constituents from the fuel assemblies2 and from materials added during fuel reprocessing and waste conditioning and calcination. Because a high weight and volume fraction of the calcine is nonradioactive, separation of radionuclides from this bulk is a way to reduce the volume of HLW to be immobilized prior to disposal. If the costs of separations process steps were not excessive and the risks acceptable, this strategy might reduce overall disposal costs and hazards. A full separations approach could remove all of the long-lived actinides, plus the 137Cs and the 90Sr. These are the critical elements whose removal might then allow the remaining material of reduced activity to be delisted from further management as HLW.3 The residues remaining after this partitioning could fall into the 1   Another important separations objective is the segregation of hazardous chemical constituents [e.g., mercury (Hg) and lead (Pb)] from nonhazardous components. This issue is discussed in a limited way in the context of each separation option of this chapter, and more broadly in Chapters 9 and 10. 2   Some of these initially nonradioactive fuel assembly materials become radioactive during reactor operations. For example, in the initially nonradioactive zirconium cladding material, Zr-93, a long-lived beta (β) emitter, is generated during exposure to radiation inside nuclear reactors. 3   This reclassification from HLW to some other category may require regulatory approval, most probably from the USNRC.

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory U.S. Nuclear Regulatory Commission (USNRC) waste Class A category, possibly enabling shallow land disposal of the materials in a suitable land disposal facility. Unfortunately, as described in reports cited in this chapter, chemicals present in the current calcines interfere with the extraction processes proposed for partitioning. This complicates the chemistry of proposed processes and may require additional "head-end" treatment to remove the interfering elements. The extra chemical treatment could expand facilities needed to perform this work and significantly increase overall disposal costs. The merit of a separations approach is affected by these considerations, which are treated in more detail below. Comments are offered on cesium ion exchange to separate cesium, the strontium extraction (SREX) process to separate strontium, and the transuranium extraction (TRUEX) process to separate actinides, or TRU elements. Application of one or more of these separations processes is under consideration for the INEEL HLW. SOLID-LIQUID SEPARATIONS An efficient solid-liquid separation (SLS) process is necessary to assure process operability and to allow high decontamination factors (DFs)4 to be achieved. Many events can complicate downstream operation without completely preventing it; however, a few complications, like flow blockage caused by accumulation of solids resulting from inadequate SLS, or leaks due to corrosion or equipment failures, can result in unplanned plant shutdowns. The SLS process is therefore one of the critical, if not the most critical, operations. Radiochemical processing experience provides adequate examples (e.g., plugged transfer lines) in which solids in process streams caused a facility to be inoperable or a process to perform unacceptably. The committee believes the importance of SLS is not sufficiently recognized; therefore SLS has received inadequate consideration in both design and development work. To reduce the technical risk, it is recommended that serious development studies be conducted with at least two different SLS approaches using dissolver solution produced from actual aged calcine (i.e., solution derived by leaching calcine with nitric acid, as discussed in Chapter 2, to dissolve calcine components into nitrate forms). Work with simulants is not appropriate for this purpose. The most significant SLS problem is likely to occur in clarifying the feed solution following calcine leaching. Unusually good feed clarification is especially important when large DFs are required, as in the case of USNRC Class A separations. Because the feed solution is almost certainly saturated with respect to critical chemical constituents of the heel, and chemical conditions are changed during the several subsequent operations, there is a strong probability that new solids will be generated during processing (as noted in some tests). Thus, more than one SLS operation may be needed. Two important reasons for requiring an effective SLS to deal with this feed clarification problem follow: The residual solids are enriched in Sr and TRU, the limiting components in decontamination operations to produce Class A waste. Inadequate solids removal has an effect analogous to that of an incomplete separation process. Small-particle solids (in the sub-micron and perhaps colloidal range) are the most difficult to remove and commonly carry TRU preferentially. Therefore, efficient removal of solids down to small-particle size is required to achieve a high DF. An additional crucial consideration that apparently has not been adequately addressed is the role of colloids, which are notorious in plutonium chemistry. Because colloids are un- 4   The decontamination factor, defined in Appendix E, expresses the degree of separations.

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory likely to be removed by the proposed filtration systems and can cause significant degradation of downstream separations operations, they potentially can negate the benefit of otherwise outstanding aqueous separations systems. Potential problems derived from colloid behavior in calcine processing will be exceedingly difficult, if not impossible, to evaluate without testing with an adequate range of actual aged calcine. One report (Brewer et al., 1995: Appendix D) addressed the radioactivity limits in zirconia calcine undissolved solids (UDS) with respect to USNRC Class A requirements. This report concluded that 97.8 to 99.9 percent solids removal is necessary for transuranics in solid (grout) waste prepared from clarified supernate, and 99.9 to 99.99 percent UDS removal is needed for 90Sr. Moreover, because of the ''sum of the fractions" role (see Appendix E), the usual variations in process performance, and additional contributions to Class A limits from soluble constituents (e.g., Tc and residual Cs, Sr, and TRU), the specification for solids removal should be about a hundred times larger than that required for the most limiting contributor to meet the Class A limit. That is, clarified feed should have solids removed by a factor greater than 105. It appears that the significance of this requirement has not been recognized. Physical operability of downstream treatment equipment, such as ion exchange columns or solvent extraction banks, generally requires highly clarified feed. Ion exchange columns retain most solids, and this will block flow if enough solids are present. Ion exchange columns are sometimes successfully used as, in effect, deep-bed filters if the combination of throughput and feed solids content is small enough. This can be advantageous in a once-through process, which may be required in any case. However, the design studies propose to operate ion exchange media through multiple elution and loading cycles; and experience has shown that contaminated solids associated with the ion exchange bed result in degraded decontamination in subsequent cycles. When multiple ion exchange load-elute cycles are used, the presence of even small amounts of solids in the feed generally causes lower raffinate DFs, lower column capacity, and increased pressure drop or dower flow, all of which degrade performance. Although some solvent extraction contactor equipment will operate physically with solids present in the feed, other contactors that may accumulate solids are more sensitive and may become inoperable. The degree of chemical separation of radionuclides (e.g., TRU and Sr isotopes) from these solids is generally degraded seriously. Solids deposited in process equipment become a subsequent source of radioactivity leaching into downstream processes. A more serious problem results from accumulation of solids at the phase interface, which not only decreases DFs, but also can seriously interfere with phase separation and system operability. The DF requirements of some of the process steps will be difficult to achieve at best; therefore, anything that might degrade performance is not acceptable. Current Status As stated above, the information provided to the committee indicates that SLS has not been adequately addressed. There is little quantitative data other than on total UDS and none on LIDS particle size and other characteristics critical to SLS. For the most part, settling followed by some sort of filtration seems to be assumed in the information provided to the committee. For example, a design study (Fluor Daniel, Inc., 1997) uses the following standard description in the two applications—SBW and calcine leach solution—Where SLS is considered:

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory "Ultimate filter sizing and type selection should be based on actual performance testing. It is anticipated that the mean particle size is less than 0.5 micron, and polishing [final] filtration is required to reduce particulate loading prior to separations processing of the SBW [or dissolved calcine]). Evaluation of cartridge filters to effect this polishing filtration should be made with an additional selection criterion being ultimate disposal in vitrification." This description says essentially nothing about the requirements or the method for SLS. The same report assigns "filtration" testing a medium priority, with less than a two per-son-year effort, which the committee believes is inadequate. Recent reports on TRUEX processing propose to change the dissolution procedure by reducing the initial and final nitric acid concentration (Herbst et al., 1998; Brewer et al., 1997a). The reason relates to improvements in both the operability and performance of the TRUEX process (see below for further discussion). However, this change would probably affect the extent of dissolution and the nature of the LIDS, and it is even more likely to increase the fraction of Sr and TRU that does not dissolve. These variables are important to SLS. A change such as this affects many unit operations (in fact, the entire flowsheet), not just the one that instigated the change. The interdependence of the unit operations on changes in conditions for any one operation must not be neglected. In various publications, the filter type discussed appears to be either a cartridge or a back-washable filter, or both, but is not described clearly. In the approaches examined, there is high throughput with a significant fraction of solids that must be recovered. An estimate (a non-conservative one, in the committee's view) of 63.5 to 635 m3 is given in Table XVIII of Brewer et al. (1995); at the same time, the acceptable UDS content in the filtrate is extremely low. Meeting both of these conditions will not be easy. In the committee's experience, treating waste requiring high DFs suggests that such straightforward clarification schemes are often inadequate. A result has been the use, in practice, either of large deep-bed filters (which is disadvantageous because the deep-bed material is routed to HLW and thus increases the solids to be immobilized as HLW) or of cross-flow membrane filtration with very small pore-sized media, such as ultrafiltration (which is disadvantageous because of increased operational complexity and cost). Clarification is perhaps as much an art as a science, but if feed variability is anticipated, membrane filtration is most likely to be successful. During the course of this review, a report on cross-flow filtration testing was provided (Mann and Todd, 1998) that indicates an interest in this approach. The testing did achieve moderately low LIDS; however, the study was limited to a single membrane filter (0.5 µm pore size), and results were not fully encouraging because of rapid flow reduction, probably the results of particles being trapped inside the pores. A smaller pore size will likely be required for cross-flow filtration because there is little or no filter cake to protect the filter media from being clogged by small particles. The committee recommends that increased priority be given to SLS, with evaluation of at least two different types of filtration, one of which should be cross-flow filtration. The use of simulants is of little if any value for such a study, since the actual solids in a variety of real aged wastes must be addressed. CESIUM ION EXCHANGE SEPARATION There are only a few options for recovering Cs from acid solutions. Most options involve either coprecipitation or inorganic ion exchange. A multistage continuous process using ion exchange columns is advantageous because it can result in the required extent of sepa-

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory ration with considerably less sorbent than would be needed for one Or a few batch separation stages, for a given sorbent (and distribution coefficient). Although an ion exchange material in pure powder form gives the best exchange properties, it usually cannot be used in a column because of significant flow resistance. In principle, the powder can be granulated with a binder to yield larger particles suitable for column use. The performance of granulated particles, however, is generally degraded because of the longer diffusion paths, blocking of pores, and changes in properties (with silica-based binders) caused by treatment at elevated temperatures to set the binder. The most selective sorbent materials for acid-side Cs separations are not widely available in the durable granular form required for column use. Some granular sorbents are currently available in research and small production quantities, but there is no assurance that they will be produced commercially over the timeframe of interest. Sorbents can usually be prepared easily in a powder (slurry) form. If the distribution coefficient is large enough (i.e., in excess of 104), the sorbent can be used in a batch mode with one or possibly a few batch separation stages. As the distribution coefficient increases, less sorbent is required. Such batch operation (which is not as dependent on specific material availability but would generate more waste sorbent) has apparently not been considered seriously. Many sorbents can be easily prepared in a slurry or powder form and used in a batch mode. Limited information was provided for the following three granular Cs sorbents considered in this program: FS-2 (a Russian-prepared potassium copper hexacyanoferrate granulated with a silica binder), AMP-PAN (ammonium molybdophosphate granulated with polyacrylonitrile and prepared by the Czech Technical University, Prague), and IONSIV IE-911 (a crystalline silicotitanate developed at the Sandia National Laboratory and commercialized in a granular form by Universal Oil Products). Only the last sorbent is clearly available in quantity. The first two have limited performance data. The literature base for nongranulated sorbents is much more extensive than for these granulated forms, although it is still rather small for acid systems. Experimental Basis The FS-2 evaluation was based on limited testing, first in Russia and later at INEEL. Batch distribution measurements and one column test in Russia, using SBW simulant, gave promising results that led to a decision in 1997 to switch the reference process from AMP-PAN to FS-2 (Olson, 1997). Additional (but still quite limited) tests at INEEL, using both dissolved calcine simulants and real SBW, gave highly varied results (Todd et al., in press; Brewer et al., 1996). Loading results differed with different feeds, possibly because of interference of some constituents (e.g., mercury). Column breakthrough occurred much earlier with real SBW feed than with simulants in column tests. These results led to a recent decision to switch the sorbent choice back to AMP-PAN. In principle, the FS-2 sorbent can be eluted with strong nitric acid and regenerated for repeated loading cycles, thereby generating relatively little waste as spent sorbent. Although the available data suggests that substantial elution is possible, the required high degree of elution for Class A separations will be difficult to achieve. Thus the committee believes that, in practice, the required large DF in the treated column effluent probably cannot be reliably obtained after the first loading cycle, because residual Cs on the column after elution will bleed into the product during subsequent loading cycles due to a reversal of the ion exchange absorption reaction, a reversible equilibrium. Therefore, the committee considers the assumption of adequate elution of this sorbent to be non-conservative. The Fluor Daniel, Inc. (1997) design study assumed the use of FS-2 for repeated cycles with a life of 1 year. Such a long lifetime is undocumented, and the commercial availability of the material is uncertain. This

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory information demonstrates that the existing database is inadequate to support selection of FS-2 for Cs removal. Other hexacyanoferrates may be practical for Cs removal. In particular, potassium cobalt hexacyanoferrate has been used in Finland on a significant scale for treating neutral and alkaline reactor waste (Harjula et al., 1994). This material can be made in a granular form directly (no binder and granulation process), and was once marketed in the U.S. by BIO-RAD. It has been shown to be useful also in acid solutions. AMP-PAN was evaluated in tests at INEEL using both simulants and dissolved calcine (Miller et al., 1997). The results were interpreted favorably because of large distribution coefficient values (ca. 3,000 ml/g) measured for actual SBW and dissolved Zr calcine. Unfortunately, similar experiments yielded distribution coefficients about sevenfold smaller with simulants of these two wastes, and even smaller values with both actual and simulated Al calcine. Because simulants should, by design, faithfully represent the requisite properties of actual aged waste, these data must be considered troublesome unless the discrepancy can be explained and reproducible data can be demonstrated. The smaller range of distribution coefficients would degrade performance with respect to column capacity, resulting in increased generation of HLW solids. Two column tests were conducted with AMP-PAN, one with actual alumina type calcine and one with a calcine simulant (Todd et al., in press). Each column was eluted with ammonium nitrate solution and loaded for a second cycle. In both cases, the second loading cycle showed a higher Cs concentration in the initial effluent and an earlier breakthrough. It is unlikely that adequate decontamination can be achieved after elution of AMP (as is also the case for FS-2), in which case a once-through process would be required; this results in increased solid HLW because the loaded sorbent becomes HLW. An optimistic estimate based on loading 3,000 bed volumes in a once-through mode suggests that at least 27,000 kg of sorbent would be required. In Todd et al. (in press), literature on the use of AMP in acidic systems is not adequately referenced. AMP-PAN particles are agglomerated with a polyacrylonitrile organic binder (the FS-2 and IONSIV IE-911 sorbents use inorganic binders). The radiolytic stability of this material, with respect to both aggregate degradation and generation of flammable and toxic gases, apparently has not been demonstrated. Thus, even if elution should prove to be practical, the lifetime of the sorbent is not known. IONSIV IE-911 was also tested (Todd et al., in press). The Cs loading capacity was smaller than with FS-2 or AMP-PAN by a factor of two or more. It was found that, for all practical purposes, the IONSIV IE-911 sorbent could not be eluted. This would imply that the loaded sorbent would be part of the HLW fraction, thereby increasing the solid HLW volume. The committee suspects that inadequate elution is a feature common to all the inorganic sorbents as long as raffinate decontamination to well below the USNRC Class A limit is required. IE-911 has one distinct advantage in that it is the only one that is definitely commercially available. Current Status The three sorbents discussed above (FS-2, AMP-PAN, and IONSIV IE-911) have undergone limited comparative testing with radioactive calcine samples (TFA, 1998). With FS-2, the presence of mercury significantly reduced the sorbent's capacity for Cs, and therefore FS-2 was dropped from further consideration. AMP-PAN demonstrated excellent selectivity and capacity and currently is selected as the baseline technology for the full-treatment option. IONSIV IE-911 currently is selected for the limited treatment option in which Cs is

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory stored indefinitely on the sorbent. AMP-PAN was not selected for the latter option because of the organic binder, which could be unstable during storage. The committee concludes that because the Cs sorbent of choice has changed from AMP-PAN to FS-2 and back to AMP-PAN within the last two years, significant uncertainties exist regarding sorbent selection. Much more definitive research is required with a more representative range of feed materials and a wider range of experimental parameters. The current database5 is not adequate to justify a selection. Certain key criteria (such as the required DF for the aqueous product, practicality of adequate elution to permit multiple cycles, sorbent life, quantity of spent sorbent destined for the HLW fraction, and compatibility of the sorbent with the immobilization process) are not known well enough to permit an adequate analysis of options, definition of a flowsheet, or evaluation of a process. Several other Cs separation options were not considered in the detailed written information provided to the committee. One is a cobalt dicarbolide sorbent, which apparently is being applied on a large scale in Russia. Another is the use of a few (e.g., three) stages of simple batch contactors/settlers using inorganic sorbent powders (such as several hexacyanoferrates or AMP), which are easily prepared, followed by good SLS (as in Campbell and Lee, 1991; Campbell, et al., 1991). Batch contactors would generate more solid sorbent to add to the HLW, compared to column operation, but this approach avoids dependence on the foreign commercial availability of expensive granular sorbents for which the long-term supply is questionable. STRONTIUM SEPARATION Based on the information prodded (e.g., Garcia, 1997), calcine presently contains, on average, approximately, 5 Ci of 90Sr per kg of calcine. Assuming a density of 1.2 to 1.6 g/cm3 for calcine (Garcia, 1997: Table 3f), this yields 6,000 to 8,000 Ci/m3. The Class A and C limits of the USNRC's 10 CFR 61 for 90Sr can be met with radioactivity concentrations of 0.04 Ci/m3 and 7,000 Ci/m3, respectively (Fluor Daniel, Inc., 1997: Table 3.2-3). Therefore, no 90Sr separations are required to meet Class C limits if the "sum of the fractions" rule is ignored. 6 However, extraordinarily high separation factors for 90Sr are required to achieve Class A levels. In the briefings received by the committee, INEEL staff cited 90Sr separation factors of 14,000 to 45,000 to meet Class A limits. Much lower Class A separation factors of 1,300 to 4,000 are required for the less radioactive SBW. According to Raytheon Engineers & Constructors (1994: p. 2-12), a DF of 105 is required for converting dissolved calcine to Class A waste, using the assumption that this DF can be achieved with two strontium extraction (SREX) process cycles. Although these references do not contain the complete derivation of these DF values, the large values indicate high separations requirements that in practice can be very difficult to achieve, particularly when there is significant uncertainty in the variability of the feed composition. A number of prior review groups (Murphy et al., 1995: Appendix G) have concluded that the SREX is the preferred method for removing 90Sr from the acidic SBW or dissolved calcine solutions. SREX uses a derivative of 18-crown-6 ether dissolved in a hydrocarbon solvent as extractant to remove strontium from acidic solution. Tributyl phosphate (TBP) is added as a phase modifier. General characteristics of SREX are relatively low single-stage 5   Thompson (no date given) provides a recent update of the current state of development of several cesium separation methods. 6   In support of this statement, although the projected average 90Sr activity is close to the Class C limit in this calculation, this activity would diminish over time due to radioactive decay, and blending will tend to reduce any areas of localized activity greater than the average.

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory distribution coefficients for Sr (with Kd values of 3 to 5) and significant extraction of Pb, Hg, TRU, Ca, Na, and K. Pb and Hg [both being metals whose disposal is covered by the Resource Conservation and Recovery Act (RCRA)] are especially troublesome for SREX because it is difficult to strip them from the organic extractant. Stripping difficulties can seriously threaten the ability to achieve the high separations required for Class A product. RCRA requirements regarding Pb and Hg can significantly impact waste disposal options. Na, K, and Ca are potential interferences because of mass action effects (Ca concentration can be 106 times the Sr concentration). As a result of these factors, and the relatively low SREX distribution coefficients for Sr, high Sr separations factors can only be achieved with multistage separations and careful control of operating conditions. A feasibility study report (Fluor Daniel, Inc., 1997: Section 4.5.8) contains a concise description of the SREX process as applied to SBW and dissolved calcine. To minimize complications from undesired coextracted cations (notably mercury and TRU); SREX would follow Cs ion exchange and TRUEX. Countercurrent solvent extraction processes using up to 24 stages of centrifugal contactors have been proposed. Strontium (and notably Pb) is stripped with 0.01-M aqueous ammonium nitrate solution. This strip solution would be added to the HLW stream, and the solvent would be chemically purified and recycled to the contactors. Chapter 4 and Table I-13.1 of the feasibility study (Fluor Daniel, Inc. 1997) describe anticipated operating parameters for SREX. This study lists "apparent separation factors"7 as 8.44 × 10-10 (Sr) and 2.54 × 10-4 (Pb) for SBW and 1.42 × 10-9 (Sr). These factors are most likely derived from a theoretical calculation of separations achieved across many successive stages. In practice, the single-stage separation factors in multistage processes decrease drastically through the cascade to essentially one as the target species becomes more dilute. This effect usually prevents the achievement of the theoretical value that is based on a given number of stages. Therefore, a 10-10 "apparent separation factor" is, in the committee's judgment, an unrealistically optimistic estimate for an integrated process, even with two cycles as proposed. The basis for calculating the "apparent separation factor" is not given in the Fluor Daniel, Inc. (1997) study, which does not explicitly include all the input parameters and assumptions used (Kimmel, 1999b). These calculations depend on normalizations that arise from the fact that other solution inputs (e.g., scrub flows) involve feed and decontaminated raffinate solutions that differ in volume, providing dilution of Sr irrespective of any separations achieved in the extraction stages. Nevertheless, as with the DF values quoted at the beginning of this section, these ''apparent separation factors" represent stringent separations requirements. Adequacy of Existing Information INEEL technical reports published in the past few years form much of the technical basis for the site's selection of SREX for 90Sr removal. The major findings of these reports are briefly summarized below. According to Raytheon Engineers & Constructors (1994: p. 4-23), an overall DF of 4,500 was obtained in three successive batch treatments of "actual nuclear waste solution." The report adds that the SREX process had been tested on a bench-scale with simulants, but had not been demonstrated with solutions using actual aged calcine waste. Law et al. (1996) describes centrifugal contactor studies on simulated SBW containing radioactive Sr and Pb tracers. This study indicated that 99.98 percent Sr removal was achieved. Inadequate nitric acid stripping of Pb was observed, leading to concentration of Pb 7   In the ensuing discussion, the term "separations factor," defined in Appendix E, is used to express the degree of separations. Differences in the usage of this term exist among practitioners; in the feasibility study of Fluor-Daniel, Inc. (1997), the term "apparent separation factors" is used.

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory in the organic solvent and precipitation of a Pb compound. Aqueous ammonium citrate was shown to effectively stop both Sr and Pb and eliminate Pb precipitation, Stripping efficiencies of 95 percent for Hg, 63 percent for Zr, 8 percent for Al, 9 percent for Ca, and 11 percent for Na were achieved. Law et al. (1997) describes centrifugal contactor studies with simulated and actual SBW. Twenty-four stages of separation using 2-cm contactors led to >99.995 percent 90Sr removal and >94 percent TRU removal, stated to be sufficient to achieve Class A levels. In addition, >94 percent Pb and 83 percent Hg were extracted, but only 5 percent of the Hg was stripped from the organic phase with 3 M nitric acid, meaning that essentially all of the Hg remained in the stripped SREX solvent stream. In practice, high Hg retention would not be acceptable since solvent reuse would not be possible in this process system. In Wood et al. (1997), batch extraction experiments with simulated and actual SBW and dissolved actual calcine are described. Pb, Hg, Na, K, U, and Pu were effectively extracted as well as Sr. The actinide extraction was attributed to the TBP rather than the crown ether. Because Sr back extraction is only slightly acid dependent, interferences from alkali metals appear to be manageable. While Hg was found to be efficiently extracted, stripping from the organic phase was not successful, and the report recommended additional work on Hg stripping. Precipitates proposed as crown:Pb:nitrate complexes were observed at the organic/aqueous interface. Critical Testing Needs The committee believes that the demanding 90Sr separations required to achieve Class A criteria, combined with difficulties indicated by the limited amount of testing carded out so far, represent high programmatic risk. The committee concurs emphatically with the technical reports that repeatedly make the case for additional development. For example, in Chapter 4 of Fluor Daniel, Inc. (1997) it is stated that "Prior to the actual design of a full-scale plant, additional engineering data and complete performance testing of a recycling solvent system will be required." As a second example, the Raytheon Engineers & Constructors (1994) study states that "Laboratory and pilot plant testing with simulated and actual dissolved calcine is necessary for complete development of a SREX flowsheet" (p. 4-23). Elsewhere in Fluor Daniel, Inc. (1997), it is recommended that the same type of additional engineering and operations data is required for both TRUEX and SREX. Specifically, complete subsystems including extraction, scrub, strip, wash, and acid rinse contactors need to be constructed and subjected to sustained tests. Critical process vulnerabilities to be investigated include solvent/extractant recycle and degradation, impurity buildup in the organic phase, temperature effects, and formation of precipitates and emulsions. The behavior of the RCRA-regulated constituents Pb and Hg are particular concerns warranting additional investigation with actual aged calcines because of the difficulty in stripping them from the organic phase, the possible formation of precipitates, and uncertainties in calcine properties. Hg also is of special concern because removal of Hg in the upstream TRUEX operation is intended (and this step, too, is problematic, as noted below in the TRUEX section). Nowhere is the need for excellent upstream SLS more critical than in Class A SREX separations, where high decontamination factors are paramount. Whether expressed as "apparent separation factors" or as DF values, the numbers cited previously for Sr removal indicate stringent separations requirements to meet Class A levels. Although the committee concurs with previous review groups that SREX is a promising approach for removing 90Sr from SBW and dissolved calcine, the committee believes that the current technical status of SREX in application to INEEL calcine, and to a lesser extent

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory SBW, is far too immature to risk the multibillion dollar investment that would be required to build an integrated Class A separations plant. Additional SREX testing with actual aged calcine clearly is required before committing to this action. Alternatives Other alternatives such as inorganic ion exchange may be feasible. It is unlikely, however, that the high DFs required to meet Class A limits could be achieved without generating an unacceptable quantity of additional HLW in the form of spent sorbent. Other activities in pursuit of alternative separations techniques, such as solvent extraction based on chlorinated cobalt dicarbollide (TFA, 1999), were not presented in sufficient detail to the committee to form a position on these methods. TRUEX SEPARATIONS PROCESS The TRUEX chemical process for extracting TRU elements (i.e., actinides) from an aqueous media into an organic media uses a strong chelating agent dissolved in an aqueousimmiscible organic diluent. In TRUEX, the primary chelating agent is octyl (phenyl)-N, N-di-isobutylcarbamoyl methyl phosphine oxide (CMPO) blended with TBP, both dissolved in an inert aliphatic diluent similar to kerosene. The TBP is included to prevent third-phase formation when the chelating agent is heavily loaded with extractable ions. Although the system is similar to PUREX (TBP alone in an immiscible organic solvent which is used to extract only U, Np, and Pu from fission products in nitric acid solutions), TRUEX will readily extract the +3 and +4 actinides as well as the uranyl, neptunyl, and plutonyl ions. A disadvantage is that it also extracts +3 lanthanides and several other common +3 and +4 elements (Fe, Zr, etc.) as well as the Hg present in INEEL wastes. The utilization of TRUEX and PUREX processing systems has been enhanced by the advent of compact, high-throughput, centrifugal contactors. These devices are highly efficient when used as single-stage extractors and are specifically designed to link into extended cascades. This capability permits low distribution coefficient extractions to be linked into cascades of many stages to produce high separation factors for chemical purification. Unfortunately these devices, which use high-speed rotors to generate the interphase contact required, are not tolerant of precipitates formed during separations. Therefore, feedstocks and process reagents must be chosen to prevent formation of interstage precipitates daring operation. This requirement becomes critical when dealing with very complex feedstocks such as the dissolved INEEL calcines that contain large mounts of fluorides, aluminum salts, and zirconium compounds. The idea of simply stripping the actinides from dissolved calcine by feeding it to a TRUEX contactor cascade is an inviting concept, but the realities of making such a process work over the wide range of chemical compositions in these complex feeds will probably lead to major difficulties, as discussed below. Technical Problems As discussed in Calcine Dissolution in Chapter 2, the dissolver product must be clarified prior to chemical processing. This LIDS residue will be enriched considerably in actinides and 90Sr in a complex mixture of chemical compounds that would include Al2O3, CaF2, and ZrO2, all three of which are notorious carriers of actinide elements. If the clarification

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory process is not totally complete, these materials will ride through extraction processes as inter-facial residues8 and settle out in process fractions, significantly lowering the DF throughout the entire process. Removal of these materials is a difficult and tedious task and must not be underestimated. Such insoluble material should be routed directly to HLW fractions to get it away from subsequent process streams. The supernatant solution from calcine dissolution will contain large amounts of F, Ca, Al, Zr, Na, B, and Fe ions, many of which interfere with and greatly complicate TRUEX processing. Several of these interfering ions can, and will, precipitate with the purified fractions as they are stripped from the organic stream. Zr and Fe will load the CMPO extractant phase and compete directly with actinide ion extraction; fluoride ions will cause precipitation with the stripped actinide and lanthanide fractions; and Zr will precipitate with phosphate ions resulting from degradation of the 1-hydroxyethane 1, 1-diphosphonic acid (HEDPA) stripping agent. These are serious problems inherent to TRUEX extraction technology and must be addressed during process development. There are potential cures for some of these problems, but they, too, cause additional downstream difficulties. It is necessary to develop the entire process sequence and not simply address specific problems in any single unit operation as they arise. One approach, in addition to the adaptation of the TRUEX process to a feed stream derived from INEEL calcine, is to consider suitable front-end processes to reduce the amount of key interfering species. 9 One specific problem that a front-end or alternative process could eliminate is the high concentration of species such as fluorides and phosphates. These species cause precipitation of other elements that can mechanically damage the solvent extraction contactors during the back-extraction phase, thereby degrading separations. The committee strongly recommends a careful trade-off study to technically evaluate front-end options. Current Laboratory Experimentation INEEL chemistry personnel are aware of some of the problems outlined above, after having performed recent trial extractions on SBW solutions (Law et al., 1996) that produced interstage precipitates and unreasonable process material balances. Further, it appears that much of the analytical data received for the process fractions may be in error by as much as a factor of two (Law et al., 1998: Table 5, page 13). For example, the quoted distribution coef- 8   These interfacial residues are semi-stable emulsions that contain sparingly soluble, colloidal-size particles that ride on liquid interfaces by surface tension effects. 9   These front-end processes would be used to concentrate actinides apart from other species that could interfere with the TRUEX separations chemistry. An example of such a front-end process of the kind that might be con-sidereal would be precipitation of a hydroxide with ammonia, leaving a basic supernate containing elements such as Na, K, Ca, Ba, St, and Cs. SLS would remove the precipitate. Boiling off excess ammonia would create an acidic ammonium nitrate solution for Cs ion exchange and SREX. Hydroxide precipitate dissolution in dilute nitric acid followed by fluoride precipitation would isolate the actinide and lanthanide elements as a precipitate. The supernate would contain species such as Al, Zr, and Fe but not Cs, St, or TRU elements, and therefore would be a candidate for a low-level waste (LLW) stream. After redissolution of the fluoride precipitate in dilute nitric acid containing borate, the standard TRUEX chemistry then can be applied to extract actinides. Other variations are (1) fluoride precipitation—the addition of fluoride will likely precipitate actinides along with rare earth elements; (2) sulfate precipitation— relatively less information is known about sulfate complexes, formed by the addition of sulfate, but certainly the calcium will precipitate as CaSO4; (3) LaF3 scavenging (see SBW Option 5 in Chapter 12) could be used on calcine redissolved in fluoride solution to separate a TRU fraction (also containing lanthanides) as a solid precipitate. Most of these processes are based on metathesis, to precipitate highly insoluble actinide and rare earth fluorides from more sparingly soluble species (e.g., CaF2) in an environment with an excess of fluoride. The committee cautions that these remarks are speculative and offered only as illustrations of the kinds of techniques that could be considered if TRUEX proves to be too complex and costly to implement.

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory ficients for 235U and 238U (same element) differ by a factor of approximately 3, and for Pu the variation between two isotopes differs by + 50 percent. Law et al. (1998: Table 4) shows material balances varying from 53 to 130 percent. The committee believes that it is not reasonable to design operational process systems based on so wide a data spread. More detailed and careful development is needed before any actinide separations option can be rationally selected or rejected. Questionable Experimental Process Modifications Herbst et al. (1998) describe a process option using a "low-acid" feed (i.e., approximately 1.1 M H+) to minimize extraction of Zr into the CMPO solvent and using a dilute NH4F scrub solution to remove tramp Zr from the organic phase following the extraction section. While this approach was partially successful in reducing extracted Zr, the follow-on HEDPA strip section apparently precipitated zirconium phosphate. This result indicates that the proposed process is not yet workable. Furthermore, using a low-acid concentration for calcine dissolution will increase the UDS, perhaps significantly. Dilution of standard feed might be a better option, but would dramatically increase process effluent that would then require concentration (presumably by evaporation) to generate an acceptable waste form. Required Technical Demonstration of Proposed Process A pilot-scale demonstration project is necessary to show adequate DFs for actinide extraction. This project needs to be performed on an adequate scale using representative samples of aged calcines from all calcine types. These feed samples may be retrieved from the bins by any of several techniques (e.g., core sampling or coaxial suction-lift sampling). The final demonstration run (a proof test) must use predefined operational parameters and be carried out without parameter changes throughout the run to confirm the suitability of the process to compensate for the known variability in the existing bins. To guarantee the validity of any Class A process, it would be necessary to provide consistent DFs of at least 20,000 for the alpha-emitters to allow for fractional contributions to LLW limits by other species such as Cs, Sr, and Tc. The order of progression of the various unit operations will also be critical since the sequence of steps needed to provide acceptable isolation of Cs, Sr, and TRU will affect each of the chemical operations under consideration. To reduce technical risk (and potential expense) to a reasonable level, the chemical operations needed to perform TRUEX separations should be developed and demonstrated prior to the commitment of resources to construct full-scale plant facilities designed for this purpose. This demonstration should be funded as a research and development effort prior to any commitment to partition actinides from calcine. SEPARATIONS PROCESSING CHALLENGES ASSOCIATED WITH THE COMBINATION OF INDIVIDUAL STEPS Removal of Cs, Sr, and TRU components will require several operations, in addition to the actual separations steps. These operations include retrieval from the tanks or bins, blending, dissolution of most of the solids, solid—liquid separation to provide a highly clarified process feed, preparation of a quantity of suitable and uniform feed for the processes, and finally the separations processes themselves. However, the current database is so limited that

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory there has not been a dear demonstration that some of the methods proposed will, in fact, meet the requirements for the range of feeds that will be encountered. In particular, operating conditions have not been (and cannot be) defined adequately using the limited data that are currently available. Given adequate process development efforts, the committee believes that separations processes can be made to work, but essential parameters in their application to a full-scale system cannot yet be defined. Examples are the mode of operation of the unit operations, flowsheet integration, equipment design, the actual DFs that will be attained, the amounts of HLW and LLW generated, and operating and capital costs. Moreover, some unit operations may achieve the necessary DF, but only at the expense of generation of excessive HLW (and associated disposal cost). It is counterproductive to spend more money carrying out design studies and cost estimates based on the current state of process understanding, rather than spending it on getting useful calcine characterization data that would reduce the uncertainty associated with the process steps discussed in this chapter. To design a processing plant cogently, DOE must know what the starting material is and what requirements the final product must meet. DOE must then develop processes that will address these conditions. In the present program, some key problem areas have received insufficient attention. Examples are (a) blending to generate a feed composition sufficiently stable for the processes to operate with high performance, (b) methods to compensate for chemical interferences arising from existing elements present in the calcine, (c) methods to deal with all the recycle streams, including spent ion exchange sorbents and extraction solvents, and (d) an effective SLS system. The evaluation to date with respect to DF requirements is inadequate. Class A limits appear to have been somewhat arbitrarily assigned to the LLW fraction. These limits can be misinterpreted to mean that the particular isotope under discussion must meet those limits. There are two problems with this concept. First, Class A includes the sum of the fractions role, and isotopes not intended for removal (e.g., Tc) may contribute a significant fraction of that sum. This reduces the limit for those materials that are removed by treatment. Second, processes have variations in performance for many reasons, such as start-up, shutdown, equipment failure, and composition variations. As a result, it is advisable to design for a substantially larger DF than the minimum defined by Class A limits. A prudent practice is to adopt the design basis for each separation step as more than a factor of 10 below the Class A limit. The technical risk is substantially reduced as the criteria for the LLW fraction are relaxed from those of USNRC Class A limits. Thus, for example, achieving the USNRC Class C limit would require little decontamination except for TRU removal. However, this sort of reasoning should not be used to justify setting an arbitrary limit like Class C as a goal. Instead, the actual criteria must be established in concert with the requirements of the disposal site, an issue taken up further in Chapters 8 and 10. The practicality of more lenient criteria than USNRC Class A limits should be examined from the perspective of both the waste treatment processes and the disposal impacts. SUMMARY In summary, there are known treatment methods to accomplish each of the required unit operations to dissolve and process the HLW calcine. The INEEL technical literature shows consideration of many, but not all, reasonable candidates for these treatment methods. Based on the information provided to the committee and the collective experience of its members, the committee believes that it is likely, but not certain, that the objective can be met for each unit operation under well-controlled conditions. It is much less likely that the objective

OCR for page 29
Alternative High-Level Waste Treatments at the Idaho National Engineering and Environmental Laboratory can be met for integrated operations under realistic plant conditions, without encountering undesirably complex operational problems, exorbitant costs, and generation of excessive amounts of secondary wastes. Specifically, the technical risk is substantial if Class A LLW is required. It probably can be done, but at an ultimate cost in both time and money that may be unacceptable. A major development program would be required to reduce the risk and the uncertainty to an acceptable level.