John W. McConnell, Jr , Robert D. Rogers
Idaho National Engineering Laboratory
P.O. Box 1625, Idaho Falls, Idaho 83415

Julie D. Jastrow
Argonne National Laboratory
Argonne, IL 60439

Steven R. Cline
Oak Ridge National Laboratory
Oak Ridge, TN 37831

Terry M. Sullivan
Brookhaven National Laboratory
Upton, NY 11973

Phil Reed
U.S. Nuclear Regulatory Commission
Rockville, MD 20852

The Field Lysimeter Investigations: Low-Level Waste Data Base Development Program is obtaining information on the performance of radioactive waste forms. These experiments were recently shut down and the contents of the lysimeters have been examined in accordance with a detailed waste form and soil sampling plan. Ion-exchange resins from a commercial nuclear power station were solidified into waste forms using portland cement and vinyl ester-styrene. These waste forms were tested to (a) obtain information on performance of waste forms in typical disposal environments, (b) compare field results with bench leach studies, c) develop a low-level waste data base for use in performance assessment source term calculations, and (d) apply the DUST computer code to compare predicted cumulative release to actual field data. The program, funded by the Nuclear Regulatory Commission (NRC), includes observed radionuclide releases from waste forms in field lysimeters at two test sites over 10 years of successful operation. The purpose of this paper is to present the results of the examination of waste forms and soils of the two lysimeter arrays after shut down. During this examination, the waste forms were characterized after removal from the lysimeters and the results compared to the findings of the original characterizations. Vertical soil cores were taken from the soil columns and analyzed with radiochemistry to define movement of radionuclides in the soils after release from the waste forms. A comparison is made of the DUST and BLT code predictions of releases and movement, using recently developed partition coefficients and leachate measurements, to actual radionuclide movement through the soil columns as determined from these core analyses.

This paper presents the experimental results of the examination of the waste forms and soils from the two field lysimeter arrays. While results of this program have been presented at previous WM meetings, this paper completes the study, and includes discussion of the final results of the examination of the soils and waste forms from the lysimeters after experiment shutdown and exhumation. During the examination, the waste forms were characterized, after removal from the lysimeter, and these results are compared to the original characterizations. Vertical soil cores were taken from the soil columns and were analyzed by radio chemistry to determine movement of radionuclides after release from the waste forms. A comparison is made of the DUST and BLT code predicted releases to releases determined from the lysimeter waste form and soil analyses. That comparison uses leachate analysis and distribution coefficients (K_d) recently obtained from laboratory analysis of the lysimeter soils and sand.

The U.S. Nuclear Regulatory Commission (NRC) has enacted regulations that link LLW acceptance criteria to the long-term satisfactory performance of the waste. Under Code of Federal Regulations (CFR) 10, Part 61, "Licensing Requirements for Land Disposal of Radioactive Wastes" (1), commercially generated LLW is classified as Class A, B, C, or Greater Than Class C. Wastes classified as either Class B or Class C must be stabilized for a minimum of 300 years. To verify the 300-year stability of waste forms, the NRC originally specified the use of short-term standardized tests with the intention that such tests would provide information relevant to near-surface disposal performance objectives. Those tests were initially published in the NRC Branch "Technical Position on Waste Form" (2), and have been revised in Revision 1 of the Technical Position (3). Of critical importance in the disposal of LLW is the need for a detailed understanding of the waste form behavior. That is necessary because the radionuclide source is the driving force behind the site performance. A major requirement in any site licensing is also the site performance assessment, which is used to evaluate whether or not a proposed disposal site will meet performance objectives. The objective of the Field Lysimeter Investigations: Low-Level Waste Data Base Development Program is to compare the results of short-term laboratory leach testing, performed earlier by the INEEL, with actual leaching in the field. This program, funded by the NRC, has been operating lysimeters for over 10 years to obtain information on the performance of radioactive waste forms in a disposal environment and to investigate waste form stability per requirements of 10 CFR 61. The experiment measured the releases of radionuclides and chemical species from the waste forms and the subsequent transport through soil columns to sampling locations within the lysimeters. The recent radiochemical analysis of the waste forms and soil samples has added significantly to the information base concerning movement of radionuclides through the soils of these lsyimeters. This study was developed to field test waste forms composed of solidified ion-exchange resin materials from EPICOR-II^a prefilters used in the cleanup of Unit 2 of the Three Mile Island Nuclear Power Station (4). Wastes used in the study are significant because they are nuclear grade ion-exchange media with high loadings of radionuclides.

Wastes used in the experiment include a mixture of highly loaded, nuclear-grade, synthetic, organic ion-exchange resins from EPICOR-II prefilter PF-7 and a mixture of organic-exchange resins and an inorganic zeolite from prefilter PF-24. Solidification agents employed to produce the 4.8 x 7.6-cm (1.8x3-in) cylindrical waste forms used in the study were portland type I-II cement and DOW vinyl ester-styrene (VES). Seven of the waste forms were stacked end-to-end and inserted into each lysimeter to provide a 1-L volume. The PF-7 waste contained 89% of the radionuclide activity as Cs-137, while PF-24 contained 94% Cs-137. The PF-7 waste also contained 5% Sr-90, and PF-24 contained 1% Sr-90. There were also measurable amounts of Cs-134, Co-60, and Sb-125 found in those wastes. Details on waste form descriptions, formulations, and initial waste form characterization are given in References (4) and (5). A listing of lysimeter waste form and fill material types are given in Reference (6).

Ten lysimeters were used in this study: five at Oak Ridge National Laboratory (ORNL) in Tennessee and five at Argonne National Laboratory-East (ANL-E) in Illinois. The lysimeters were designed to be self-contained units that will be disposed at the termination of the study. Each lysimeter consisted of a 0.91x3.12-m (3x10-ft) right-circular cylinder divided into an upper compartment that contained fill material, waste forms, and instrumentation, and a lower compartment for collecting leachate as shown in Figure 1. Four lysimeters at each site were filled with soil; a fifth, used as a control, was filled with inert silica oxide sand. The lysimeters at ANL-E contain soil indigenous to the site, while the ORNL lysimeters contained soil taken from Savannah River Laboratory in South Carolina. The soil columns were 2.21-m (7-ft) deep.

Instrumentation in each lysimeter included moisture cup soil-water samplers and soil moisture/temperature probes. The probes, located at three elevations, were connected to an onsite data acquisition system (DAS), which also collected data from a field meteorological station located at each site. Porous cup soil-water samplers and the leachate collection compartment comprised the water sampling components of each lysimeter (Fig. 1). Incoming precipitation moved downward through the soil column to the waste form, then on to cups 3 and 1, and finally to the leachate collector at the bottom. Radial movement of waste form releases were detected in cups 5, 4, and 2, while vertical release was observed by cups 3 and 1. Samples of moisture were withdrawn from the cups and the collector. Lysimeter design, installation, instrumentation, and data acquisition are explained in Reference (6). Monitoring of the lysimeters at ANL-E and ORNL began with the collection of liquid samples in September 1985 and continued through shutdown in October 1995 with sample collection on approximately a quarterly basis. Samples of liquids were taken from locations nearest the waste forms and from the leachate collectors to track the migration of radionuclides, primarily Sr-90 and Cs-137. Each month, data stored on a cassette tape in the DAS were retrieved and translated into an IBM PC-compatible disk file.

Precipitation, air temperature, wind speed, and relative humidity were recorded continuously by the ANL-E and ORNL DAS during the experiment. The cumulative volume of leachate from the lysimeters since the initiation of field work, and examples of the lysimeter soil temperature and moisture data from ANL-E and ORNL sites are found in Reference (7). Data recorded in FY-95 indicate that the lysimeter soil columns at both sites have remained moist during the last reporting period.

Data show that not all nuclides consistently appeared in the water obtained from the moisture cups or the leachate collectors. The nuclide that appeared with the most regularity at both sites was Sr-90. Table I contains a comparison of the percent of inventory release of Sr-90 and Cs-137 found in the moisture cups and leachate water. Gamma-producing nuclide releases from the waste forms to the cups occurred with regularity at both sites (Table I). However, only waste forms at ORNL released detectable amounts of Cs-137 to the leachate collectors (Table I).

The primary objective of the recently completed waste form and soil sampling was to obtain the waste forms from all of the five lysimeters at each site in cylindrical cores. Secondary objectives were to extract soil cores, soil microbial samples, selected moisture cups, filter fabric samples, and filter support stone samples. Seven soil cores were to be taken per lysimeter, one for the waste form and six for soils. Four soil grab samples, one filter cloth cut sample, and one filter support rock grab sample per lysimeter were planned. Moisture cup numbers 1 and 3 from each lysimeter also were to be collected. All waste form cores were successfully taken at both sites and were transported to the INEEL for detailed examination. Nearly all soil cores were taken at ANL-E, while only cores number 1 and sample 1 of cores number 2 were taken at ORNL. Samples of filter cloth, filter support rock, and moisture cups 1 and 3 were obtained only from the ANL-E control lysimeter.

A diagram of the sample locations and sizes is shown in Fig. 1, and Table II lists the cores and samples. The waste form cores (Number 2) were 7.5-cm (3-in) in diameter and 58.5-cm (23-in)long. That length contained all seven waste form samples. The soil cores and microbial soil samples were 3.3-cm (1.5-in) in diameter and various lengths. All were taken with coring tools made up of multiple 25-cm (10-in) long segments. The microbial soil samples were also taken with the 3.3-cm (1.5-in) diameter tool using one segment. All cores were contained in plastic, cylindrical-core tool liners that were closed with plastic end caps. Radial and vertical position of the coring tools was controlled during coring operations by use of special guide plates and bushings. The coring tools and tips were specifically designed for this task by Art's Manufacturing and Supply of American Falls, Idaho. The INEEL designed the special guide system.

Radiochemical characterizations, including full-length gamma scanning of each seven-sample waste form and radiochemical analysis of segments of selected samples from each waste form, were designed to provide information on the remaining waste form radionuclide inventory. Waste form physical condition was determined by visual examination, weighing, and compressive testing. Soil cores were radiochemically analyzed for nuclide content, as were the filter cloth, filter support stone, and moisture cup. These data can then be used to determine radionuclide material balance within each lysimeter, radionuclide pathways through the soil columns, and radionuclide holdup factors of the various components of each lysimeter system. Other soil samples are being examined for microbial activity, which may then be related to waste form physical condition indirectly or more directly by microbial examination of waste form surface swipe samples. The analysis of microbial samples has not been completed but viable growth has been observed in all waste form surface swipes and soil samples.

The waste forms, comprised of 7 samples each 5-cm (1.8-in) in diameter by 7.5-cm (3-in) long, were examined in detail to determine the effects of over 10 years of disposal site burial. Full length gamma scanning was performed on the waste form cores before the plastic liners used to collect them were opened. All waste forms showed reasonably uniform doses from sample to sample and good agreement in magnitude of dose between like waste forms. One waste form exhibited a significant dose decrease at the lower end. Later examination of the core (ANL3) found that the bottom sample was missing. It was later found, still in the soil column at ANL-E. This set of measurements allowed confirmation that the individual sample inventories were within acceptable limits.

Photographs were taken of the waste form cores after they had been opened and the samples exposed. Examination indicated that a majority of the samples were in good to excellent condition. There was some damage (cracks and edge chips) caused during installation in 1985, as indicated by soiled fractures. In other cases the fracture surfaces were clean, indicating damage caused during coring operations. Most samples had soil adhering to all surfaces. Those from ANL-E were estimated to have about twice as much adhering soil. The ANL-E soil also adhered more tenaciously.

All waste forms were weighed and those weights were compared to the original as-cast weights. In general, the cement samples, which had more adhering soil, weighed more after field test. The VES samples weighed less after field test. It is thought that all samples lost weight due to dry-out during hot cell storage (5 months) but the adhering soil caused the cement waste forms to weigh more than as-cast. Microbial action could also have resulted in a weight loss but that process has not as yet been quantified by this work.

Three samples were selected from each waste form and subjected to compressive testing. All samples exhibited excellent strength with exception of one cement sample loaded with organic/inorganic resins. That test piece failed at 3.25E+6-Pa (472-psi) apparently due to an unnoticed flaw. Those results are presented in Table III. The average compressive strengths when plotted against estimated self irradiation dose (listed in Table III) compare very closely with the results of aging tests given in Reference 9.

a) C= portland cement, D=vinyl ester styrene, 1= organic ion exchange resin, 2= organic/inorganic ion exchange resin.
b) soil - averaged compressive strengths include data of soil from both sites sand - averaged compressive strengths include data from same sample type from both sand and soil.

One sample was selected from each waste form for radiochemical analysis. Those samples were sectioned, pulverized, homogenized, and 2 to 5 gm subsamples were extracted. Those subsamples were analyzed for beta and gamma emitters with the results shown in Table IV. Cesium-137 and Strontium, the primary radionuclides present, will be the nuclides of discussion. Cesium-134 and Cobalt-60 were found randomly in trace amounts. The after-field-test waste form inventories are compared to as-cast inventories in the table. The average release of Cs-137 was 37% compared to the average release of Sr-90 of 65%. A majority of the waste forms exhibited near average releases of both nuclides with the exception of the VES waste forms of ANL3 and ANL4 which showed releases of both nuclides significantly above the averages, Cs-137 at 61 and 50 percent and Sr-90 at 92 and 86 percent respectively. ANL2 and ORNL2 with similar cement waste forms exhibited releases of both nuclides in similar amounts. The similar waste forms in ANL5 and ORNL1 also performed very much alike although ANL5 contained silica sand versus soil in ORNL1.

NA - Data not usable.
a) All activity measurements have been decay corrected to 12/1/96.

Soil and sand samples were taken directly with coring tools and also as part of the waste form cores. Each waste form core held about 1.7-cm (0.75-in) thick shell along side the waste forms and about 5-cm (2-in) of soil/sand in the lower tip. Table V lists the calculated total radionuclide content in the side and tip soils/sands of each waste form core as well as the percent of waste form as-cast inventory for both Cs-137 and Sr-90. While the percent of nuclide inventory retained in the 1.7-cm (0.75-in) thick by 53.5-cm (21-in) long cylindrical side soils/sands were significant, ranging from 0.4 to 8.5 percent for Cs-137 and 0.1 to 10.3 percent for Sr-90, the percent retained in the 5-cm (2-in) long by 7.5-cm (3-in) diameter core tip soils/sands was insignificant at 0.001 to 0.1 percent for Cs-137 and 0.002 to 0.4 percent for Sr-90. The highest percent Cs capture was in ANL3 while the highest percent Sr retention was in ANL1. The ANL soils retained more of both radionuclides than did the ORNL soils. It also appears that the waste forms screened the soil/sand in the tips from downward moving nuclides compared to the soil/sand along side the waste forms.

a) All sample weights were corrected to dry weights.
b) All activity measurements have been decay corrected to 12/1/96.

Detailed radiochemical analysis of the directly cored soil/sand resulted in data sets too extensive to present here, at 151 Cs and 158 Sr data points. However, those data were used to plot isograms of specific activities for both nuclides for the ANL-E lysimeters. The limited results from ORNL cores #1 and #2 make isograms of that data meaningless. The isograms of ANL-E for both Cs and Sr data in lysimeter 4 are included in this paper as Figures 2 and 3.

The data shows that there was radial dispersion of Cs-137 in all the soil Lysimeters to 12.5-cm (5-in) radius and to 25-cm (10-in) radius in the sand lysimeter. That radial movement was seen from above the waste forms to the bottom of the soil column in all but ANL1 where none was measured near the top of the waste form or bottom of the soil column due to loss of those cores. Similar but less extensive Sr-90 dispersion was also found. Soil units ANL1 and ANL4 contained that nuclide at the 25-cm (10-in) radius from the above the waste form to the bottom of the soil column (Figure 3) while in all others it was detected to 12.5-cm (5-in) radius. Figure 2 indicates that Cs moved horizontally

outward fairly uniformly with only a short distance (3-cm or 1-in) seperating the high concentrations from the leading edge of the transport front. Figure 3 shows Sr movement created a deeper gradation (6-cm or 2-in) between the high concentrations and the leading edge of the front.

The information from the #1 soil cores at both sites shows that some upward movement of both nuclides occurred, that Cs-137 moved upward in larger concentrations than Sr-90, that ORNL had larger concentrations of both nuclides at higher elevations than did ANL-E, and that the ORNL and ANL-E soil and sand columns experienced hot spots at higher elevations. That is similar to what was observed in the ORNL5 lysimeter and thought to be the result of periodic flooding of the lysimeters which inundated the waste forms as discussed in Reference 7.

Other samples were analyzed for Cs-137 and Sr-90 content. those included Moisture cups #1 and #3, cup #1 silica flour, filter cloth, and filter rock all taken from ANL5. The results of those analyses are shown in Table VI and include total activity of the whole component. Also included is amount of total inventory retained by the component. The percent of inventory retained is very small for all samples and will not be an effect in a material balance. However, those percentages also must be compared to the percentage of inventory that was measured in cup #3 and the leachate collector in ANL5 and ORNL5. That can be done by comparing the values in Table I with those in Table VI. The cup #3 retention of Sr-90 was a factor of 45 less than the smallest lysimeter 5 cup measurement (in ANL5) while the rock and filter cloth retention of Sr-90 were 124 times smaller than the smallest leachate collector measurement (in ANL5). Cup retention had no effect on sand or soil Sr-90 measurements but rock retention could have effected Sr-90 measurements of leachate in the soil lysimeters. Cup #3 retention of Cs-137 in lysimeter 5 was 38 times larger than the smallest observed cup measurement (in ORNL5) and about the same as the largest (in ANL5). Rock retention of Cs-137 was 10 times the ORNL leachate measurement (no Cs -137 was measured in the ANL-E leachate). Both cup and rock retention of Cs-137 effected measurements in lysimeters 5 and probably in the soil lysimeters also.

a) These samples were corrected to dry weight.
b) All activity measurements have been decay corrected to 12/1/96.

The Disposal Unit Source Term (DUST) code (10) has been used to model the release of the radionuclides Cs-137 and Sr-90 from the lysimeter waste forms. DUST is a one-dimensional code that accounts for container performance and waste form leaching (including diffusion-controlled release). Transport can be modeled through finite differences or by a multi-cell mixing cascade approach. The finite difference method was used in the simulations reported in this paper because it is more general than the mixing cell approach and permits modeling of dispersive transport. Use of these data in the DUST code was examined in detail in a paper presented at WM '93 (11) and has been presented more recently in References (7) and (12). Releases from the waste forms of the sand filled lysimeters were examined in those references and compared to the measured data from leachate collectors.

The releases of Cs-137 and Sr-90 from portland type I-II cement located in the inert, sand-filled lysimeters 5 at ORNL and ANL-E were chosen because releases from other lysimeters were substantially lower; therefore, the data were not sufficient to model. At ANL-E, lysimeter 5 contained resin waste from PF-7 solidified in portland type I-II cement; at ORNL, lysimeter 5 contained resin waste from PF-24, which was also solidified in cement (Table I). Diffusion coefficient values measured in laboratory testing of these waste forms were 1.7E-9-cm²/s for Sr-90 in portland cement and 6.0E-10-cm²/s for Cs-137 in portland cement.

The best fit to the Sr releases was obtained with a diffusion coefficient of 9.9E-10-cm²/s (representative of measured values for soil) and a Kd value of 36-cm³/g with a dispersion coefficient of 8-cm (3.1-in). The Kd value measured for the sands from the lysimeter in the laboratory was 6.4-cm³/g. Use of this value in the DUST code led to predicted releases from the lysimeter orders of magnitude greater than observed. This indicates the laboratory Kd value may not be appropriate for the field application.

In an attempt to reproduce the concentration profiles within the lysimeter the two-dimensional model, Breach, Leach, and Transport (BLT) (14) was used. This model assumed cylindrical symmetry with radionuclide release parameters consistent with the soil data (e.g. waste form diffusion coefficient of 1.5E-9-cm²/s for Sr from ANL lysimeter 5). Using the model, with a soil Kd of 6.4-cm³/g and a dispersion coefficient of 8-cm (3.1-in), did not reproduce the soil data in the predicted sorbed concentrations for Sr-90 released in ANL lysimeter 5. There was more release predicted from the bottom of the lysimeter, and the spreading throughout the soil column was greater than the soil data would indicate.

The calculations were repeated with dispersivity reduced to 0.8-cm (0.3-in). This produced a narrow plume consistent with low predicted concentrations of Sr 12.5-cm (5-in) away from the waste form. However, the releases were much greater than observed and the profile beneath the waste form did not match the data. The model predicted much higher concentrations underneath the waste form as compared to the observed data.

Figure 4 presents the Sr-90 sorbed concentrations for the case with a Kd of 36-cm³/g and a dispersion coefficient of 8-cm (3.1-in) , the best fit value from the one-dimensional case. The shape of the contours are much more consistent with the data. The contours show a two order of magnitude drop in concentration from the waste form to the sampling location at 12.5-cm (5-in), while the data shows a decrease of almost three orders of magnitude. In addition, the data and the predictions both show a substantial drop in concentration below the waste form. Again, the data decreases more rapidly than the predictions. Although the general shape of the contour appears to be consistent with the data, the predicted magnitudes are much greater that the observed soil concentrations. Predicted sorbed concentrations in the waste form region are around 19E+3-Bq/g (5E+5-pCi/g), while the measured concentration is 6.3E+2-Bq/g (1.7E+4-pCi/g). In the waste form region 12.5-cm (5-in) from the centerline, the predicted sorbed concentrations are 1.9E+2-Bq/g (5E+3-pCi/g) while measured concentrations were 7 to 17E-1-Bq/g (18 to 45-pCi/g) at this location. The model predicts slightly higher concentrations beneath the waste form than observed. Model sorbed concentrations beneath the waste form range from 7 to 1850E-1-Bq/g (20 to 5000-pCi/g) while the observed values range from 9 to 78E-1-Bq/g (24 to 212-pCi/g).

The predicted concentrations show a relatively uniform concentration profile at the 12.5-cm (5-in) monitoring locations only dropping off as the bottom of the soil column is reached (Figure 4). This relatively uniform concentration was observed for Sr in all five lysimeters at ANL. This implies that if the predicted concentrations were normalized to the measured soil concentration at the edge of the waste form reasonable agreement could be obtained between predictions and measurements.

The radionuclide that has appeared with most regularity in the leachate at both sites is Sr-90, although Cs-137 was observed regularly in the leachate of all ORNL lysimeters.

Waste form examination showed them to be in good to excellent condition with compressive strengths aligning with aging data according to amount of self irradiation dose.

The average release of Cs-137 from the 10 waste forms was 37% of waste form inventory while the average release of Sr-90 was 65%. These releases are consistent with the 90 day leach test results determined during initial characterization of these waste forms.

The soils and sand taken from the waste form cores adjacent to the waste forms contained from 0.4 to 8.5% of the Cs-137 waste form inventory and from 0.1 to 10.3% of the Sr-90 waste form inventory. The ANL-E soil retained more of both nuclides than did ORNL soil.

Soil core sample data indicate that radionuclides moved radially away from the waste forms in all ANL-E lysimeters. No radial information was obtained from ORNL lysimeters. Upward movement was also found in all lysimeters at both sites with Cs-137 in larger concentrations than Sr-90.

The retention of Cs-137 by the moisture cups affected the measurements. Filter rock retention of both Cs-137 and Sr-90 affected the measurement of those nuclides in the leachate. These retentions must be included in modeling the system.

Modeling of the transport of Sr from ANL lysimeter 5 has been done. Using the inventory obtained from the waste forms leads to much higher predicted concentrations than those measured. However, the general shape of the concentration profiles is consistent with the observed data for Sr in all five ANL lysimeters.

This work was supported by the U.S. Nuclear Regulatory Commission, Office of Nuclear Regulatory Research, under U.S. Department of Energy Idaho Operations Office Contract DE-AC07-94ID13223. Dr. P. R. Reed is the NRC RES Program Manager of this work.