The proposed resin for HB-Line Phase II, Reillexä HPQ, was tested in the laboratory under expected plant conditions and was found to load a maximum of 117 grams of plutonium per liter of resin. With a plutonium feed concentration of no more than 5 grams per liter, the 20-liter resin column in HB-Line should not be capable of retaining more than 2.4 kilograms of plutonium at any time during its life cycle. A batch size of 1 kilogram of plutonium should be within the operating range of the process with minimal raffinate losses as long as a full 20-liter charge of resin is loaded into the column.

The new HB-Line facility was designed and built in the early to mid 1980’s. Phase II of HB-Line is currently being prepared for startup to stabilize Pu solutions. This facility was designed to receive Pu (or Np) nitrate solutions from H-Canyon and convert them into oxides for storage or shipment. After receipt of Pu solution, anion exchange columns will both purify and concentrate the Pu nitrate solution, after which it will be converted to an oxide via oxalate precipitation, filtration and calcination. The existing tanks and interconnecting piping associated with anion exchange are shown in Figure 1.

Figure 1.  Process Flow Diagram of Anion Exchange Process in HB-Line

 

Pu solution will be received from H-Canyon as a dilute Pu nitrate solution (2-3 g Pu/L, 3-5 M HNO₃). Valence and acid concentration adjustment will be performed in a feed adjustment tank via the addition of ferrous sulfamate (FS), sodium nitrite and 64% HNO₃. In the past, Pu valence adjustment was performed by adding sufficient FS to make the solution 0.05 M FS and then oxidizing the Fe²⁺ and Pu³⁺ to Fe³⁺ and Pu⁴⁺ via the addition of sodium nitrite. The FS reduces all Pu ^4+,5+,6+ to Pu ³⁺ and the nitrite re-oxidizes the Pu³⁺ back to Pu ⁴⁺. Pu ⁴⁺ is the only oxidation state of Pu that forms the anionic nitrate complex that loads onto anion resin. The high nitrate concentration and radiolysis will produce sufficient HNO₂ to oxidize both the Fe²⁺ and Pu³⁺ after several days, but the nitrite addition allows the material to be processed rapidly. A "heat-kill" has been used, instead of a nitrite treatment, to accelerate the natural re-oxidation of Fe²⁺ and Pu³⁺ to a period of less than one hour. The valence adjustment may be performed either before or after the acid adjustment, but it is preferable to valence adjust first since the acid adjustment normally doubles the solution volume and thus requires almost twice the quantity of FS to be used. This affects the amount of Fe and sulfate in the waste stream. Once properly adjusted, weapons grade Pu should remain stable as Pu⁴⁺ for weeks to months due to the effect of nitrate complexation.

The anion column will be prepared for a Pu run by "reconditioning" the resin bed with a quantity of 8 M HNO₃ passed through the piping and the bed to flush dilute acid from the system. HB-Line will pump the adjusted feed from the receipt/feed adjustment tanks to a column feed tank. The feed solution will be pumped "up-flow" through the column, absorbing the anionic Pu(NO₃)₆^2- complex as it passes through the resin bed. The column raffinate (which contains the cationic metal impurities and is normally waste) will pass by a NaI detector along its path to an H-Canyon tank. This detector should indicate unexpectedly large Pu losses during loading or decontamination. After a full batch of Pu feed solution has been loaded onto the column (probably using one-half to two-thirds of the resin bed’s physical capacity), the bed will be washed with 8 M HNO₃ to remove residual impure solution from the equipment and decontaminate the Pu loaded on the resin. This step will also be performed "up-flow". The Pu will then be eluted with 0.35 M HNO₃ by gravity feed from a head tank "down-flow" through the column. The elution stream passes through a sight-glass that is instrumented with fiber optics to a colorimeter to determine the point to start collecting Pu into the product concentrate tank. The initial effluent from the column collected during elution is commonly referred to as the "heads cut". The high Pu concentration solution collected next is referred to as the "hearts" or product cut. Any dilute acid concentration, dilute Pu concentration solution collected near the end of the elution is referred to as the "tails" cut. HB-Line is not planning on collecting a tails cut on a routine basis. At the end of a column run, the column is left in a dilute acid concentration state with little or no Pu heel. This is considered to be a safe condition for storage until the next run. A NaI detector will monitor the Pu inventory on the column throughout the column run to ensure that excessive Pu inventory is not unintentionally loaded on the column.

Normally an anion exchange column would only be partially loaded, leaving a large amount of excess capacity. This excess capacity results in minimal losses to the raffinate stream. The capacity of the resin must be known under normal process conditions to determine the batch size. If loading is continued, the concentration of Pu in the raffinate stream will gradually rise until visual "break-through" occurs. After visual break-through, the resin continues to load Pu, but an ever-increasing fraction of the Pu in the feed solution is not absorbed. Ultimately the Pu concentration in the raffinate will reach the concentration in the feed and the resin bed will absorb no additional Pu.

The anion exchange resin proposed for use in HB-Line is Reillexä HPQ. Fred Marsh of Los Alamos National Lab (LANL) and Reilly Industries jointly developed this resin based on another polypyridine-based resin, Permutitt SK, which was used in Pu processes in the late 1950’s. Better resistance to radiolytic and chemical damage is attributed to these resins due to the use of the pyridine ring "N" functional group. Marsh also found this resin to be attractive due to its relatively high loading for Pu and its excellent elution behavior (2). In the 1996-7 timeframe, Reilly Industries modified its manufacturing process to increase the anion exchange sites yield by ~20%. This change cast some doubt on the applicability of past data on the original version of this resin for future work with newly manufactured batches of resin. The current work examined the Pu loading characteristics of the "improved" resin. The first objective of this work was to measure the resin capacity under HB-Line process conditions to determine the point at which visual Pu break-through would occur. A second objective was to determine the ultimate capacity of the resin column if loading continued well past the point of visual break-through.

Plant scale anion exchange equipment is typically 100 to 1000 times as large as that used in the laboratory. Normally the process is scaled based on the linear velocity (Q/A, mL/min/cm^{2 º} cm/min) through the resin bed (which is related to residence time in the bed) and the loading profile of the resin. If a laboratory column contains resin at the same depth as the plant equipment, then the scaling problem is primarily reduced to one of linear velocity. However, higher Pu concentrations in the feed solution will produce a higher Pu resin loading. For these experiments, a resin volume of 36 cm³ of settled resin was used in a 19 mm ID glass column, resulting in a cross sectional area of 2.835 cm². To determine a conservatively high value for the Pu loading, a high-end concentration value of roughly 5 g Pu/L feed was used in all runs. The targeted flowrate of 9 mL/min @ 5 g Pu/L for a 2.835 cm² laboratory column was based on a cross-sectional area for the HB-Line. The slower flowrate of 3 mL/min @ 5 g Pu/L (for the lab column) corresponds to a lower Pu loading rate (0.005 g Pu/min/cm²) and the flow velocity is one-third of the baseline condition. This rate serves to test for any significant increase in resin loading, caused by a slow loading flowrate. Table 1 shows the current SRTC test conditions.

Resin Pretreatment

Three different resin pre-treatment methods were used: 1) convert to nitrate form, 2) convert to nitrate form then irradiate, and 3) convert to nitrate form then digest to remove low temperature exotherm. All resin that was tested came from the same original manufacturer’s lot (#80302MA). All resin was initially converted from the chloride form (as-shipped) to the nitrate form by washing with 1 M NaNO₃ (10 bed volumes (BV) in a column was the preferred method, but other methods were acceptable). A large sample of nitrate form resin was irradiated for 100 hr in a ⁶⁰Co irradiation source with a field of 1x10⁶ rad/hr for a total dose of 1x10⁸ rad in July 1998 for this work and several related studies(1). The third sample tested was treated by exposure to hot HNO₃ (8 M, 85°C, 45 min) in a laboratory oven. NO_x fumes were evolved from the sample, but after the treatment, the low temperature exotherm was no longer detectable by RSSTä testing (1).

Column Loading

A sufficient quantity of resin must be converted into the nitrate form and pretreated, if required, prior to loading the column. The resin is generally loaded either by pouring dry resin beads into the column and then wetted with water or dilute nitric acid or by slurrying the resin into the column with water. The resin bed is settled by running water/dilute nitric acid downflow through the resin bed to fill the excess void spaces until all apparent gaps are filled. The final resin bed volume is adjusted by adding a small amount of resin or removing excess resin with a slurry pipette. Once the resin is loaded and settled into the column, every effort is made to not allow the liquid head above the resin to drain below the top of the resin bed. Air bubbles trapped within the moist bed are often very difficult to remove and will cause channeling of the flow through the bed.

Feedstock Preparation and Valence Adjustment

Pu solution is prepared by adjusting the [HNO₃] to 7.5 to 8.5 M and performing a valence adjustment with ferrous sulfamate (FS). Normally a valence adjustment is performed by adding sufficient FS to make the solution 0.05 M FS and then followed by a "heat kill" to oxidize the Fe²⁺ to Fe³⁺. The FS reduces all Pu ^4+,5+,6+ to Pu ³⁺ whereas the heat kill re-oxidizes the Pu³⁺ back to Pu ⁴⁺. The heat kill must be performed after the HNO₃ has been adjusted to 8 M and consists of gently heating the feed solution to 50° C for 30 minutes. The high nitrate concentration and heat produces sufficient HNO₂ to oxidize both the Fe²⁺ and Pu³⁺. In 8 M HNO₃ at 50° C the half life for Fe²⁺ is on the order of 10 minutes compared with 1 hr at 35° C and 10 hr at 25° C. (9). The valence adjustment was performed either before or after the acid adjustment; it is better done first because the acid adjustment normally doubles the solution volume and thus requires twice the quantity of FS. The plant plans to perform a nitrite addition to "kill-off" the FS. This approach has historically been an alternative approach to the heat kill for restoring Pu³⁺ solution to Pu⁴⁺ in both F-Area and H-Area and should be equally effective. However, this method results in higher [SO₄^2-] in the feed and may increase the raffinate losses due to sulfate complexation with Pu⁴⁺.

Lab Equipment

The column used in this work is shown in Figure 2. This column utilized #7 teflonÔ bushings for connecting ¼ inch polypropylene tubing to the column. The column consists of a 19 mm ID glass body to retain the resin bed and a headpiece. The headpiece attached to the column body with a RodavissÔ joint to allow the column to retain a larger pressure head than that allowed by a ground glass joint. As a safety precaution, the head also had a Ace glass pressure relief valve and a pressure gauge to monitor the pressure in case the frit at the bottom of the column plugged. An additional arm with a stopcock and funnel allowed the column to be vented. A sketch of the experimental setup is shown as Figure 3. A standard FMI piston pump was used to pump feed, wash, or elution acid through the column. A ½ inch SwaglockÔ cross and ½ inch optic lens was used to fabricate a flowcell with a 2.54 cm pathlength. A pair of fiber optic lines previously installed through the ceiling of the glovebox allowed a light signal to be brought into the glovebox, passed through the flowcell and carried out to a Zeiss spectrometer controlled by NT-based computer. A detailed parts list for the complete spectrophotometer system used is given in Table 2.

Pu Loading

Sufficient Pu feedstock was prepared for one or more break-through column runs. A sample was taken and analyzed for total acid/free acid, total alpha by radscreen and gamma scan. Because of the importance of the Pu concentration feed values, the radscreen and gamma scans for the feed were normally performed in triplicate. At the beginning of the column run, several column volumes of 8 M HNO₃ were fed to the column to displace the dilute acid the resin was stored in from the previous run. At this point in time the spectrophotometer was checked for proper operation and a new "zero" spectrum was stored (background with no Pu in the flow cell). The Pu feed was then pumped downflow through the column at approximately the desired flowrate, with the raffinate passing through the flow cell and being collected and measured in one of several graduated cylinders. The amount of Pu in the raffinate was monitored by both visual inspection and by periodic spectra taken by the instrumentation. "Grab" samples of the raffinate were also taken on a periodic basis during the course of the loading step and analyzed for Pu content by alpha and gamma counting. The spectra taken were stored along with the time and volume of raffinate collected and the Pu concentrations predicted by the model(s) were matched up to the amount of Pu that had been loaded onto the column. The feed flowrate was periodically checked with a 10mL graduated cylinder and a stopwatch. The flowrates were somewhat variable (sometimes ± 50% of the targeted value), but the average flowrate could generally be regulated within 15% of the desired value. The Pu loaded onto the resin was also visually monitored. Flow abnormalities within the resin bed sometimes caused visible "tailing" of the loaded Pu (e.g. Pu loaded non-uniformly on the resin where the Pu interface is further down the resin bed on one side of the column than the other). As the Pu interface (observed as a green boundary) approached the bottom of the resin bed the levels of Pu in the raffinate rose gradually (to ~0.1 g Pu/L) and the frequency of spectra measurements was increased. Pu "break-through" could be visually detected in the raffinate solution in the range of 0.5 to 1 g Pu/L. For a period of time before and after visual break-through, the Pu concentration in the raffinate rapidly rises from <0.1 to ~2 g Pu/L. As loading continues, the concentration continues to rise, but more slowly as it asymptotically approaches that of the feed concentration. During this time, a significant fraction of the Pu in the feed solution continues to load onto the resin as Pu diffuses deeper into the resin bed and the resin becomes saturated with Pu. The raffinate was collected in two cuts. The first was collected up to the point that visual break-through was detected and the second was collected from that point past the end of the loading phase and through the decontamination wash. This method of collection allowed the early raffinate to be discarded, while recycling the raffinate after break-through.

Pu Washing and Elution

When the concentration in the raffinate was approximately 90% of that in the feed (or when the prepared feedstock was exhausted), the loading was stopped. Allowances then had to be made to displace the residual Pu feed solution from the resin bed and feed line. There was unloaded Pu in the system due to void space in the column (~50 volume percent or 18 mL), the head-liquid volume above the resin (10 to 40 mL) and finally the liquid volume still in the feed-line (<10mL). Normally this was eliminated by pumping the feed line dry and allowing the head liquid to drain (without letting the level drop below the top of the resin) to a minimal volume above the resin bed (0 to 5 mL). A short decontamination wash of the resin column with 50 mL of 8 M HNO₃ was then performed to remove most of the FS residue and displace the residual Pu feed solution. This wash was included with the second raffinate cut for material balance purposes and futur e recovery of the Pu. The column was then completely eluted with 200 to 400 mL of 0.35 M HNO₃ at 6 mL/min downflow. Although the spectrophotometer was not calibrated for Pu in dilute acid, it still proved useful in detecting trace Pu in the effluent during the latter stages of elution. Because several runs were made with each sample of resin loaded into the column, it was essential that all of the Pu be eluted from the column prior to the next run. After the visible Pu had been removed from the resin, the run could then be safely interrupted and the remainder of the elution could be continued the next day.

Analytical

Samples were taken of feed solutions and composite raffinate and product solutions, as well as "grab" samples of the raffinate stream. These samples were routinely analyzed by radscreen and gamma scan analyses to determine the total alpha activity and the ²⁴¹Am activity to obtain a value for the Pu alpha activity. A specific activity of 1.58x10¹¹ dpm/g Pu was used to convert Pu alpha activity to Pu mass. This conversion factor (specific activity) was known from the history of the Pu used in this work. Radscreen values were corrected from the default 90% efficiency of the method to a 97% efficiency that is specific to Pu/Am materials to give a more accurate result.

Calibration of Spectrophotometer

The spectrophotometer was calibrated for this work by measuring spectra of the feed solution prepared for runs CR198/9 (analyzed for Pu concentration) as well as the spectra of volumetric dilutions of that known Pu solution with 8 M HNO₃ (see Table 3). The Pu for the dilutions was pipeted into a weighed volume of 8 M HNO₃ that would give the correct dilution volume based on a density of 1.25 g solution/cm³. The spectra of four samples and deionized (DI) water were measured in duplicate from 191 to 1024 nm in November 1998 in 1-cm plastic cuvettes. The same equipment was used for all column runs described in this report.

These spectra were mathematically processed using the SRTC-ADS developed SRLMVA program. A PLS (partial least squares) model was built on duplicate spectra from standards A through E (Table 3). Second derivative preprocessing was performed on each spectrum and the modeled wavelength range was limited to 450 to 850 nm to avoid noise and interferences. Calibration sample spectra are shown in Figures 4 and 5. Several development versions of the SRTC-ADS ZMMIS program were used over the course of this work to acquire data from and control the multiplexer and spectrometer. This program displayed the spectra in near real time and used the PLS model (from SRLMVA) to calculate a Pu concentration during the column run. The spectra taken during the column run were saved for future data analysis.

A series of column runs was performed with the current version of Reillexä HPQ. For each experiment, composite samples of the feed, raffinate, and product solutions were analyzed (along with selected "grab" samples) and submitted for analysis. The results from those analyses and the volume of each solution were used to calculate a material balance for each experiment. The amount of Pu absorbed onto the resin was calculated as the difference between the cumulative amount fed and the amount found in the raffinate solutions. This method was used to calculate the visual break-through loading of the resin. The saturated loading was also calculated in this same fashion and was checked by measuring the total amount of Pu found in the eluate (product) solutions. A sample set of results for one of the column runs and some sample calculations for that run are shown in Table 4. Additional results for the other runs are given in the appendix.

Table 5 summarizes the Pu loading results for the 12 column runs that used the 1998 lot of Reillexä HPQ. Some of these runs were made to recover Pu and did not involve loading to Pu break-through. Flowrates and feed concentrations varied somewhat and are also shown in Table 5. Visual break-through occurs in the range of 59 to 88 g Pu/L resin. The determination of visual break-through is somewhat subjective making it difficult to determine if experimental changes gave a significantly different loading. The results of these tests indicate that there could be a 15 to 30% increase in loading caused by lowering the flowrate from 3 down to 1 mL/min/cm². Some increase is expected, but this shows it to be at most a relatively modest effect. The effects of the various pretreatment methods on visual break-through are less clear. No large change is loading was observed by any of the pretreatment methods. Marsh (3) claimed to observe increases in resin capacity by a boiling acid treatment due to an increase in surface area. The acid treatment in this study was much less severe on the resin structure than the one performed by Marsh. Crooks (1) observed a significant breakdown in resin structure after boiling resin in concentrated HNO₃, but a sample of this "boiled" resin was not tested for Pu loading because the breakdown of the resin was expected to result in an unacceptable pressure drop across a column. Marsh (4) also observed that alpha irradiation did not greatly reduce the Pu capacity. The results of the current work show a small increase in capacity (on the order of 15%), but this result is probably not statistically significant.

A number of these runs were continued well past the point of visual break-through and approached the point of saturating the resin with Pu. Saturated Pu loading in the range of 80 to 117 g Pu/L resin was calculated (based on the amount of Pu eluted). Typically 20 to 30% more Pu was loaded after visual break-through was detected. In several instances 50 to 60% more was loaded, but those runs had relatively low values for break-through loading. The run with the highest loading, CR210, had a material balance discrepancy probably due to a significant error in an analytical result. Two results are reported for that run: 90 and 117 g Pu/L, the first calculated directly and the second calculated from the material balance. Based on all the data from that run, it is more likely that 117 g Pu/L is the correct result, but that value is calculated differently than for all the other runs.

Up to this point all results that have been discussed are based on analytical measurement of "composite" samples from the column runs. The spectrophotometer system provided direct on-line measurements of the Pu concentration in the effluent during the column loading. Utilizing the individual spectra time stamp, the Pu concentrations predicted by those spectra (from the saved spectra), and the experimental records, the Pu concentration in the raffinate was reconstructed as a function of loading time and feed volume. Figure 6 shows a plot of the Pu concentration in the raffinate for several column runs. There are some gaps in the data, but the data generally show that the column initially loads nearly all the Pu from the feed, leaving a very low Pu concentration in the raffinate. Eventually a point is reached where the Pu concentration rapidly rises to 30 to 40% of the feed concentration. Loading continues beyond that point, but an increasing fraction of the Pu in the feed does not get absorbed by the resin bed and passes through the column in the raffinate.

Figure 7 shows the cumulative amount of Pu both loaded on the column and collected in the raffinate bottle during the course of the same column run. The upper family of curves represents the Pu loaded on the resin, while the lower family of curves represents the Pu that passed through the column as part of the raffinate and was collected in the effluent bottle as a waste or recycle stream. As can be seen by looking at the right hand axis, 50 g Pu/L resin can be loaded with minimal losses, but in the range of 80 g Pu/L resin Pu losses started to become significant in many of the runs. The resin loading values that can be determined from Figures 6 and 7 are measured independently from the values shown in Table 5 because they come from the spectrophotometric measurements rather than from sample analysis. The general agreement is fairly good. The slow change in the concentration profiles shown in Figure 6 support a maximum loading of up to ~120 g Pu/L for the current batch of Reillexä HPQ.

All runs made in this laboratory were performed with downflow loading, where the hydraulic pressure tends to compress the resin bed. The actual plant column will be loaded up-flow. If the column is not full of resin and the resin bed is not compressed by the column "top-hat", the resin beads may fluidize and the efficiency of mass transfer may be reduced. This situation should not occur as long as the resin bed fills the column body. Future laboratory work on elution will use a column design that will allow up-flow loading and any additional complications may be recognized at that time.

Comparison with Literature Results

Marsh performed the initial Pu work with Reillexä HPQ. Published reports documented the effects of radiation exposure on the version of the resin that was produced in the late 1980s (2,3,4). Although the current work uses a version of the resin with an increased capacity and uses different experimental techniques, some comparison of the data is possible. Marsh performed dynamic batch contact tests in which the amount of Pu loaded onto resin samples exposed to ~5 g Pu/L, 7 M HNO₃ solution was measured over time. After converting the batch contact into units of g Pu/L resin, a rough comparison can be made between the current data and the literature results. The Marsh’s 15-minute batch contact data (2,3,4) has comparable time for mass transfer to that of the 12-minute residence time (H/(Q/A)) of most of the column runs in this work. Marsh’s 6-hour contact data (2,3,4) reflect values that approach equilibrium values. Table 6 tabulates a comparison of Marsh’s data with the current work. The improved capacity resin appears to have an ~10 % increase in Pu loading capacity. Interestingly, Marsh showed a small decrease in Pu capacity as the resin was irradiated with either gamma or alpha dose, but the current work appears to show a small increase. It should be noted that while Marsh found a relatively good radiation stability of Reillexä HPQ compared to the various polystyrene based resins, the resin capacity steadily dropped above 100 Mrad dose. These results suggest that any effort to extend the useful life of Reillexä HPQ much beyond the current limit of 100 Mrad would have limited value.

Uncertainties

As previously stated, instantaneous flowrates occasionally varied widely form the targeted value, but adjustments were quickly made to keep the average flowrate within ~15 % of the targeted value. Since resin loading was found to be only a weak function of flow rate the flowrate uncertainty probably contributes < 1% to the uncertainty in the loading results.

Under the conditions that most column runs were made (9 mL/min, ~5 g Pu/L), ~0.05 g of Pu was fed during each minute of the loading cycle. During the loading time prior to break-through, enough Pu was loaded to increase the resin loading by ~1.4 g Pu/L. Thus the subjective nature of visual break-through detection may make a contribution to the resin loading uncertainty of up to 3 g Pu/L.

Measuring the resin bed depth to determine the volume of resin in the column is relatively accurate measurement, but the resin beds pack into the bed unevenly and settle during the initial use. This uncertainty is estimated at ~2% based on observations of settling as the resin columns are loaded. However additional resin was added during column setup to keep the resin volume at its targeted value. This adjustment should bias the amount of resin loaded towards a higher value and likewise bias the resin loading values upward over what would be observed if such care were not taken. Since several runs were made with the same resin column, this uncertainty does not contribute to the variation in reproducibility between successive runs that use the same resin column.

Analytical measurement uncertainty is generally dominated by dilution errors. Typically, dilution error is estimated as ~3% for this work due to the equipment. Operator errors could easily cause a 30% or more error on an individual sample, but those errors would normally be recognized due to inconsistency and rechecked. An probable error of this type on a CR210 product solution was not recognized until later and was not resolved. The feed solutions were routinely analyzed in triplicate to eliminate these errors because the feed solution concentration has a dominant effect on the loading results.

A material balance was calculated for each run by taking the solution volumes and the analytical results from the feed, product and waste streams. Most column runs had an overall material balance uncertainty of < 4% (many had < 2% uncertainty). Due to a probable dilution error in the analysis of a product sample, Run CR210 could not be resolved better than a +6 to -13% uncertainty (depending what which way the calculation was performed). These results support the viewpoint that routine analytical results have a precision of ~4 %.

The spectrometer itself is a very precise measurement instrument, but accuracy of its results depends on the preparation of the calibration solutions and the modeling of the spectra from those solutions to produce a predictive model. In this case, the standards were prepared by volumetric dilution and thus dilution errors contributed 1 to 2 % to the uncertainty of the resulting model. Because the concentration of the initial solution was based on analytical measurements, the total uncertainty in the concentration of the standards was in the range of 3 to 6%. The uncertainty in the predictive model is difficult to quantify, but is assumed to be considerable less than the uncertainty in the concentration of the standards in this case. Because the more useful measurements were made in the upper end of the concentration range where less dilution error was involved, measurement of column break-through is probably in the low end of this uncertainty range. More accurate calibration standards could have been prepared by gravimetric means, but that effort was not judged to be worthwhile for this particular application. These measurements provide support for the loading results and are included to provide a better understanding of how break-through progresses.

Based on a ~5 g Pu/L feedstock in ~8 M HNO₃, HB-Line can expect to load from 80 g Pu/L up to a maximum of 117 g Pu/L onto Reillexä HPQ anion exchange resin if the columns are run without regard to raffinate losses. If the columns are operated for reasonable loss levels, the resin should load 59 to 88 g Pu/L without significant break-through. This result means that a column containing 20 L of resin may hold 1.2 to 1.8 kg of Pu under nominal conditions, but might hold 2.4 kg of Pu under extreme conditions. These conclusions should hold for resin exposed to a radiation dose of up to 1x10⁸ rad or for resin treated to remove the low-temperature exotherm.