Environ. Sci. Technol. 2008, 42, 4940–4945
Pretreatment of Hanford Medium-Curie Wastes by Fractional Crystallization LAURENT NASSIF, GEORGE DUMONT, HATEM ALYSOURI,† AND RONALD W. ROUSSEAU* School of Chemical & Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332-0100 DON GENEISSE Arena NC Incorporated, 1425 Stevens Center Place, Richland, Washington 99354
Received December 19, 2007. Revised manuscript received April 3, 2008. Accepted April 4, 2008.
Acceleration of the schedule for decontamination of the Hanford site using bulk vitrification requires implementation of a pretreatment operation. Medium-curie waste must be separated into two fractions: one is to go to a waste treatment and immobilization plant, and a second, which is low-activity waste, is to be processed by bulk vitrification. The work described here reports research on using fractional crystallization for that pretreatment. Sodium salts are crystallized by evaporation of water from solutions simulating those removed from singleshell tanks, while leaving cesium in solution. The crystalline products are then recovered and qualified as low-activity waste, which is suitable upon redissolution for processing by bulk vitrification. The experimental program used semibatch operation in which a feed solution was continuously added to maintain a constant level in the crystallizer while evaporating water. The slurry recovered at the end of a run was filtered to recover product crystals, which were then analyzed to determine their composition. The results demonstrated that targets on cesium separation from the solids, fractional recovery of sodium salts, and sulfate content of the recovered salts can be achieved by the process tested.
Introduction There are approximately 53 million gallons of aqueous radioactive waste stored in 177 underground tankss149 single-shell tanks (SST) and 28 double-shell tanks (DST)sat Hanford, Washington. The waste was generated mainly from production of nuclear weapons material during the Cold War and, in accordance with the Hanford Federal Facility Agreement and Consent Order, all underground wastes must be treated by 2028 (1). Original plans called for remediation of all radioactive waste at a waste treatment and immobilization plant (WTP), but, to accelerate disposal, plans have been formulated to send low-activity waste (LAW) that would otherwise have gone to the WTP to supplemental treatment by what is referred to as bulk vitrification. To increase the amount of waste that qualifies for low-activity characterization, a pretreatment process that converts medium-curie * Corresponding author e-mail:
[email protected]. † Present address: Industrial Chemistry Center, Royal Scientific Society, P.O. Box 1438, Aljubaiha, Amman 11941, Jordan. 4940
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 13, 2008
waste to LAW by evaporative fractional crystallization has been proposed (2). The objective of the pretreatment process is to split wastes having an intermediate level of radioactivity into two streams: one in which radioactive species have been preferentially accumulated and a second that can be categorized as LAW and therefore suitable for supplemental treatment by bulk vitrification. The strategy in pretreatment by fractional crystallization is to process a feed solution that contains relatively small concentrations of radioactive species and much higher concentrations of sodium salts by evaporating sufficient solvent (water) to cause crystallization of sodium salts. The crystals are then separated from the residual liquid, which has retained cesium, technetium, and other radionuclides in solution. Three requirements (targets) to be achieved by the process are given in two categories: a minimum value for the process to be considered successful and a desired value that is more challenging (1): • Target 1. Maintain the 137Cs content of the recovered product crystals low enough so that when the solids are redissolved to form a 5 M sodium salt solution, the 137Cs concentration will result in less than 0.05 Ci/L and preferably less than 0.0012 Ci/L. Achieving this target is essential to converting medium-curie waste to a stream that qualifies as low-level waste. • Target 2. Recover at least 50% and preferably 90% of the sodium from the feed solution as product crystals. Removal of sodium salts from the medium-curie waste reduces the volume of the stream that has been enriched in radioactive species and, thus, the final volume of high-level waste. • Target 3. Reduce the sulfate content in the residual solutions so that the mole ratio of sulfate to sodium is less than 0.01 and preferably less than 0.0022. This ratio must be controlled in order not to exceed the solubility limits of the salts in glass (see for example Manara et al. (3)). The objective of the work described here was to explore the feasibility of using fractional crystallization to meet the above targets. Results were obtained using a laboratory semibatch unit fed with simulated waste under conditions approximating those that would be used in a full-scale operation. The feed solutions were considered typical of wastes recovered from single-shell tanks (SST). The contents of these tanks are to be recovered by pumping out supernatant liquid and adding additional water, which dissolves the salt cake and perhaps entrains insoluble material as it is pumped from the vessel. Accordingly, recovering the SST wastes produces a feed to the pretreatment operation whose composition depends upon whether the solution is obtained early or late in the evacuation of a tank. These variations were considered in the present work by using simulant feed solutions whose compositions were either (a) more like what would be drawn off early in the tank evacuation and referred to as SST Early Feed, or (b) representative of those solutions from late in the process and referred to as SST Late Feed. The experimental program was guided by simulations generated from thermodynamic modeling using ESP (Environmental Simulation Program) software with the MSE (Mixed Solvent Electrolyte) supplement, which was obtained from OLI Systems, Inc. (Morris Plains, NJ). Modeling provides predictions of the equilibrium behavior of the complex SST solutions, and each crystallized species has unique equilibrium behavior that defines the relationship between solubility and the solution temperature and presence of other solutes. Of course these variables also affect nucleation and growth kinetics, which have major roles in determining crystal 10.1021/es7031696 CCC: $40.75
2008 American Chemical Society
Published on Web 05/28/2008
TABLE 1. Compositions of Simulant Solutions for SST Early Feed and SST Late Feed chemical
MW
early feed (M)
late feed (M)
NaAlO2 · 2H2O NaOH Na2CO3 Na2C2O4 KNO3 NaNO3 NaNO2 Na2SO4 Na3PO4 · 12H2O NaCl NaF Na2Cr2O7 · 2H2O CsNO3
118.0 40.0 106.0 134.0 101.1 85.0 69.0 142.0 380.1 58.4 42.0 298.0 194.9
0.27 0.62 0.61 0.006 0.018 3.26 0.51 0.13 0.005 0.07 0.01 0.02 0.005 g/L
0.036 0.1 0.1 0.054 0.003 0.53 0.071 0.021 0.025 0.013 0.056 0.003 0.0008g/L
size distribution and morphology, but which cannot be predicted using thermodynamic modeling. Experimental System. The species utilized to prepare the two SST feed solutions and their compositions are given in Table 1 (4). It should be noted that the compositions in Table 1 differ slightly from those used in earlier work on this project, but none of the results presented here appear to be materially altered by these modest variations. The crystallizer operation was designed to produce a maximum solids content in the range of 30–40% of the slurry mass. Early experimentation and the need to maintain good mixing led to utilization of constant-volume, semibatch operation of the crystallizer. Two stages of operation were required to achieve the desired yield of sodium crystalline species when the system was fed with SST Early Feed. Only one stage was used for SST Late Feed. A schematic of the operating strategy is shown in Figure 1. The feed to each stage was added in two ways; the desired initial volume of solution was introduced to the crystallizer, and subsequently feed solution was added intermittently throughout a run so as to maintain a constant slurry volume. The feed to Stage 1 was composed of the waste solution, while the feed to Stage 2 was obtained from the solution recovered upon filtration of the slurry from Stage 1. After collection of the filtrate, additional water was added to keep all species in solution during handling. The dilution did not affect the behavior of Stage 2, although it was necessary to adjust the target condensate-to-feed ratio (mass of condensate collected divided by the mass of feed solution introduced to the crystallizer) to account for the additional water. Complete details of the experimental equipment and procedures are given elsewhere (5, 6), while sufficient information is provided here to understand the nature of the experiments. The crystallizers used in the study had nominal volumes of 300 and 100 mL, which facilitated multiple-stage operations in which the volume of feed available for the second stage was significantly reduced because of evapora-
FIGURE 1. Operating strategy for two-stage semibatch fractional crystallization of SST feed solutions.
tion in the first stage. Internals of the 300-mL crystallizer included four equally spaced baffles, but limited volume prevented their use in the 100-mL unit. The agitators used in the crystallizers varied according to the size of the vessel, with that used in the 100-mL unit being slightly smaller. Mixing intensity was controlled by an adjustable-speed motor connected by a rubber tube to the glass shaft turning the impellers. The temperature of the crystallizer contents was measured by a thermocouple and recorded on a computer. Evaporation was driven by heat added to the system by a fluid pumped through the jacket of the crystallizer. The rate of evaporation was manipulated by adjusting the temperature of the heating fluid. Vapor generated in the crystallizer was condensed in a glass heat exchanger with cooling water; condensate flowed from the heat exchanger through a flexible tube to a collection vessel resting on a balance. Readings from the balance were transmitted to and stored on a computer, which allowed real-time estimation of the condensate-to-feed ratio. A vacuum pump reduced the pressure in the system to the desired value, which was set by manual adjustment of a valve. Operating Characteristics and Results. The utility of fractional crystallization can be illustrated by the experiences associated with two out of the more than 60 such runs that have been conducted. These two runs were referred to as Certification Runs, as these were operated according to fixed specifications and subjected to full analyses. Here our purpose in selecting them for discussion is to illustrate the general methodology and resulting outcomes on each of the feed solutions. Most of remaining runs were designed to explore specific variables in the operation of the crystallizer and are not germane to the purpose of the present discussion. A summary of the values of process variables used in the two runs under examination is given in Table 2. Included are feed identification, crystallizer volume, operating temperature, evaporation rate, total collected condensate, and final condensate-to-feed ratio. An illustration of how mass balances around the system were satisfied is given in Figure 2. The labels on streams are self-explanatory except for those identified as accumulation; these reflect recovered masses of crystalline material adhering to the walls and internals of the crystallizers. Such formations reflect the nature of heat transfer to the crystallizers: that is, some of the crystallized salts exhibit a reduction in solubility as temperature is increased, which means they have a tendency to crystallize on a heated surface. Such encrustations are not expected in scaled-up, continuous crystallization because the operations will be run at steady-state with low supersaturations. Moreover, encrustations can be removed in large-scale systems through installation of proper internals. The streams shown by dashed lines are those accounting for measured losses of materials. Closures of total mass balances, which are an important element in validating the results of each run, were within 5% around each processing step. Process Requirements. Chemical analyses of key process streams provided information that allowed determination of whether or not the runs had achieved the process targets. Accordingly, samples were taken of filtrates, spent wash liquor, unwashed and washed crystals, and solids accumulated on the crystallizer walls. The samples were then analyzed for key components, including cesium, sodium, aluminum, chromium, phosphate, and sulfate. All samples were sent for analysis in liquid form so that homogeneity was ensured. The resulting compositions were then used to obtain species balances on the key components around the process as shown in Table 3 . The closures of these balances were influenced by the low solute concentrations in the feed solutions and the dilution steps required in preparing the samples. Cesium activity was calculated directly from the chemical VOL. 42, NO. 13, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4941
TABLE 2. Conditions at Which Runs 38b and 40 Were Conducted run/stage numbers
feed
crystallizer volume (mL)
operating T (°C)
evaporation rate (g/h)
total condensate (g)
condensate-to-feed ratio
38b/1 38b/2 40/1
SST Early filtrate from Stage 1 SST Late
300 100 300
66 40 60
26 39 91
753 505 1808
0.48 0.63 0.88
analyses of the washed crystals obtained from Stages 1 and 2. The process target for cesium was determined based on redissolution of recovered crystals to form a solution that is 5 M in sodium; as stated earlier, a satisfactory process outcome is for such solutions to have a specific cesium activity of less than 0.05 Ci/L, but it is desirable for that value to be less than 0.0012 Ci/L. Since the analytical results were in terms of wt % for major components and for cesium in ppm (mass of species per million mass units of the sample), the approach to process goals had to be estimated using available information on specific activity of 137Cs and the fraction of
the Cs in the radioactive waste that is 137Cs. The approach to Target 1 can be determined in two ways, either by estimating the radioactivity in the recovered product and comparing that value to the explicit process target or by estimating a quantity called the decontamination factor, DF, which is a measure of the Cs removed from the feed solution. To illustrate the first of these methods, assume that a sample of the final product crystals was found to contain 25 wt % sodium (Na) and 0.2 ppm cesium (Cs). The basis of calculation for determining if the product meets the target is to estimate the activity of a solution of this material
FIGURE 2. Mass balances around Stages 1 and 2 of an SST Early Feed run. Solid arrows are the process streams and the dotted arrows represent quantified losses. Feed to Stage 2 includes Filtrate from Stage 1 plus dilution water. Closure on a total mass balance was performed for each dashed box around a process unit.
TABLE 3. Species Balances for Runs 38b (SST Early Feed) and 40 SST Late Feed.
a
run 38b
run 40
species
input (g)
output (g)
closure (%)
input (g)
output (g)
closure (%)
Al Cr Cs SO4 P Na
11.9 2.35 0.00493 18.6 2.50 343.7
10.8 2.28 0.00349 13.1 1.85 308.3
9.4 3.0 29.3 29.9 26.0 10.3
2.46 0.3490 0.00112 3.39 1.54 94.8
2.68 0.3494 0.00104 3.51 1.55 92.7
-8.75 -0.12 7.18 -3.52 -0.67 2.27
a
4942
Closure values indicate difference from 100%.
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 13, 2008
TABLE 4. Cesium Separation Performance As Determined by Compositions of Crystalline Products from SST Early Feed Solution Run 38b and SST Late Feed Solution Run 40 solids (g) run 38b
run 40
feed stage 1 stage 2 combined feed stage 1
228.7 37.7 266.4 85.2
containing 5 M Na (i.e., 5 mol Na/L). It is estimated that the fraction of total cesium in Hanford waste that is present as 137Cs varies between 20% and 50%, but for present purposes it is assumed to be 50% (the more conservative value). The specific activity of 137 Cs is 86.58 Ci per gram of 137Cs, which means that the volumetric activity associated with the product is given by the equation 5
100 g 0.2 g Cs mol Na 23 g Na × × × × L mol Na 25 g Na 1000000 g 137 Ci 86.58 Ci 1 g Cs ) 0.004 (1) × 2 g Cs L g 137Cs
Such a calculation allows direct comparison of the analytical results to the process objective. In the second approach, a decontamination factor (DF) is defined as the activity of 137Cs in real waste or the total cesium concentration in a simulant feed at 5 M sodium concentration divided by the corresponding activity or concentration in the salt recovered from the fractional crystallization process, also at a 5 M sodium concentration. That is,
DF )
[activity of 137Cs (at 5 M Na)]feed [activity of 137Cs (at 5 M Na)]waste
)
Cs [wtppm% Na ] ppm Cs [ wt % Na ]
feed
waste
(2) where the compositions in the denominator on the farthest right are given by the analyses of crystals produced in the operation and those in the numerator are from the feed. For example, suppose the feed to a process contains 10 wt % sodium and 0.20 ppm Cs, and product crystals contain 28 wt % sodium and 0.04 ppm Cs. Such results correspond to a DF of 14, which has been calculated as 0.20 ppm Cs [ 10 wt% Na ] DF ) 0.04 ppm Cs [ 28 wt% Na ]
feed
) 14
(3)
waste
However, to use DF as a measure of effectiveness, the original targets regarding Cs activity must be translated to this new quantity. For example, the SST Early Feed solution contains 2.3 × 10-5 g Cs/g Na, which can be used to estimate the activity of the solution as follows: 2.3 × 10-5
g Cs mol Na 23 g Na 86.58 Ci ×5 × × 137 × g Na L mol Na g Cs 1 g137Cs ) 0.115 Ci/L (4) 2 g Cs
Taking the cesium in the waste solutions to contain 0.5 g 137Cs/g Cs (as in Equation 4), the two targets for the operation
become (DF)min ) 2.3 and (DF)desired ) 96. The targets for SST Late Feed are determined in a similar fashion to be (DF)min ) 1.9 and (DF)desired ) 79.
Na (wt %)
Cs (ppm)
Cs (Ci/L)
15.0 33.79 28.75 33.07 3.10 29.81
3.2 0.034 0.091 0.042 0.55 0.043
0.105 0.0005 0.0016 0.0006 0.088 0.0007
DF 210 66 166 123.3
TABLE 5. Sodium Recovery from SST Early Feed Run 38b and SST Late Feed Run 40 mass of sodium (g) stream
run 38b
run 40
total input crystals/stage 1 accumulation 1 crystals/stage 2 accumulation 2 total recovered recovery
234.8 76.9 41.7 10.7 19.7 149.0 63.4%
63.6 25.4 22.5 47.9 75.3%
Table 4 shows the analytical results from runs on SST Early and Late Feeds. Given are sodium and cesium contents of washed filter cakes from the two stages of the SST Early Feed run and the single stage of the SST Late Feed run. These values were used to calculate the activities expected and listed in the table for dissolution of each of the filter cakes. As stated earlier, the calculations assume that 50% of the cesium in Hanford waste is in the form of 137Cs and, as shown, the values for the combined solids from the SST Early Feed run are significantly better than the desired target of 0.0012 Ci/L and the decontamination factor associated with blending products from the two stages is well above the desired value of 96. The product from the SST Late Feed run also surpasses the desired targets on activity and DF. Obtaining a complete picture of sodium recovery requires scaling the results from the second stage of a run to estimate what would have occurred if all of the filtrate from Stage 1 had been used in Stage 2. (For an explanation of the fundamentals of scaling calculations, see ref 7.) For example, the actual filtrate in Run 38b from Filtration 1 was 385.9 g, but only 352.6 g were fed to Stage 2. Accordingly, a scale factor of 1.09 was used to adjust the amounts produced in Stage 2. Moreover, it was assumed that all crystalline solids, whether in the recovered filter cakes or the recovered encrustations, contributed to the recovery of sodium. Table 5 provides the results of these estimations for the SST Early Feed Run 38b and the SST Late Feed Run 40. Clearly, the target of recovering at least 50% of the sodium entering the process in the crystalline product has been achieved; however, the recoveries appear well below the optimiztic target of 90%. It should be clear that achieving this target is limited by the solids content of the product slurry and the solubilities of sodium salts. The first of these values is set by what can be handled in the crystallizer, and the second set of values is fixed by the operating temperature. Accordingly, the desirable target may be unachievable. Values of the third of the process criteria, the sulfateto-sodium molar ratio in the filtrate streams from Stages 1 and 2, were calculated using the relative amounts of the two ions in these streams as determined by chemical analysis. The results in Table 6 show that the combined filtrates from SST Early Feed Run 38b and the filtrate from SST Late Feed Run 40 meet the minimum target of 0.01, but only the SST VOL. 42, NO. 13, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4943
TABLE 6. Sulfate-to-Sodium Molar Ratio in Filtrate Streams from SST Early Feed Run 38b and SST Late Feed Run 40
run 38b run 40
filtrate source
wt % Na+
wt % SO4)
molar ratio sulfate:sodium
stage 1 stage 2 combined stage 1
21.76 21.80 21.78 20.87
0.286 0.472 0.336
