This article was published in an Elsevier journal. The attached copy is furnished to the author for non-commercial research and education use, including for instruction at the author’s institution, sharing with colleagues and providing to institution administration. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright
20
Embed
Author's personal copy - Available Technologiesavailabletechnologies.pnnl.gov/media/183_612200832312.pdf · Author's personal copy ... The PUF column volumetric water content was
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
This article was published in an Elsevier journal. The attached copyis furnished to the author for non-commercial research and
education use, including for instruction at the author’s institution,sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies areencouraged to visit:
Accelerated weathering of high-level andplutonium-bearing lanthanide borosilicate waste glasses
under hydraulically unsaturated conditions
Eric M. Pierce a,*, B.P. McGrail a, P.F. Martin a, J. Marra b,B.W. Arey a, K.N. Geiszler a
a Environmental Technology Directorate, Pacific Northwest National Laboratory, P.O. Box 999, MS: K6-81,
Richland, WA 99352, United Statesb Savannah River National Laboratory, Westinghouse Savannah River Company, Aiken, SC 29808, United States
Received 15 December 2005; accepted 8 March 2007Editorial handling by M. Gascoyne
Available online 10 May 2007
Abstract
The US Department of Energy (DOE) has proposed that a can-in-canister waste package design be used for disposal ofexcess weapons-grade Pu at the proposed mined geologic repository at Yucca Mountain, Nevada. This configuration con-sists of a high-level waste (HLW) canister fitted with a rack that holds mini-canisters containing a Pu-bearing lanthanideborosilicate (LaBS) waste glass and/or titanate-based ceramic (�15% of the total canister volume). The remaining volumeof the HLW canister is then filled with HLW glass (�85% of the total canister volume). A 6-a pressurized unsaturated flow(PUF) test was conducted to investigate waste form–waste form interactions that may occur when water penetrates thecanisters and contacts the waste forms. The PUF column volumetric water content was observed to increase steadily dur-ing the test because of water accumulation associated with alteration phases formed on the surfaces of the glasses. Periodicexcursions in effluent pH, electrical conductivity, and solution chemistry were monitored and correlated with the formationof a clay phase(s) during the test. Geochemical modeling, with the EQ3NR code, of select effluent solution samples sug-gests the dominant secondary reaction product for the surrogate HLW glass, SRL-202, is a smectite di-octahedral clayphase(s), possibly nontronite [Na0.33 Fe2(AlSi)4O10(OH)2 Æ n(H2O)] or beidellite [Na0.33Al2.33Si3.67O10(OH)2]. This clayphase was identified in scanning electron microscope (SEM) images as discrete spherical particles growing out of ahydrated gel layer on reacted SRL-202 glass. Alpha energy analysis (AEA) of aliquots of select effluent samples that werefiltered through a 1.8 nm filter suggest that approximately 80% of the total measurable Pu was in the form of a filterableparticulate, in comparison to unfiltered aliquots of the same sample. These results suggest the filterable particles are>1.8 nm but smaller than the 0.2 lm average diameter openings of the Ti porous plate situated at the base of the column.In this advection-dominated system, Pu appeared to be migrating principally as or in association with colloids after beingreleased from the LaBS glass. Analyses of reacted LaBS glass particles with SEM with energy dispersive X-ray spectros-copy suggest that Pu may have segregated into a discrete disk-like phase, possibly PuO2. Alteration products that containthe neutron absorber Gd have not been positively identified. Separation of the Pu and the neutron absorber Gd during
0883-2927/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.apgeochem.2007.03.056
glass dissolution and transport could be a criticality issue for the proposed repository. However, the translation and inter-pretation of these long-term PUF test results to actual disposed waste packages requires further analysis.� 2007 Elsevier Ltd. All rights reserved.
1. Introduction
Representatives of Russia and the United Statesheld discussions in the 1990s about the dispositionof approximately 50 metric tons (MT) of surplusweapons-grade Pu (Rankin and Gould, 2000). Dur-ing these discussions, a preliminary environmentalimpact statement was completed. In response tothe results presented in the preliminary environmen-tal impact statement, the US Department of Energy(DOE) implemented a program to provide for thesafe and secure storage of surplus weapons-gradePu. As part of this program, DOE decided to build(1) the Pit Disassembly and Conversion Facility(PDCF), (2) the Mixed Oxide Fuel FabricationFacility (MFFF), and (3) the Plutonium Immobili-zation Plant (PIP) at the Savannah River Site inwest-central South Carolina (Rankin and Gould,2000). DOE also decided that the can-in-canisterdesign was the preferred option for the disposal ofan immobilized Pu-bearing waste form at the pro-posed mined geologic repository at Yucca Moun-tain, Nevada (Rankin and Gould, 2000). Thisconfiguration consists of a high-level waste (HLW)canister (approximately 0.6 m outer diameter and3 m tall) fitted with a rack that holds mini-canisters(approximately 0.1 m outer diameter and 0.5 m tall)containing a Pu-bearing ceramic and/or glass(�15% of the total canister volume). The remainingvolume of the HLW canister is then filled withHLW glass (�85% of the total canister volume).The use of this can-in-canister design providessecure storage for Pu, because the engineered bar-rier created by the HLW glass requires a heavilyshield facility to retrieve the Pu which makes retrie-val difficult.
Although a titanate-based ceramic was previ-ously selected as the Pu-immobilization waste form(Cochran et al., 1997; Gray et al., 1997; Myers et al.,1997), the use of glass, namely a lanthanide borosil-icate (LaBS), instead of ceramic has reemerged as acandidate waste form (Marra and Ebert, 2003). Thisoccurred because, unlike the titanate-based ceramic,glass can accommodate the wide variety of impuri-ties that is associated with highly impure weapons-grade Pu currently slated for disposition. Before
DOE can conduct a credible safety analysis, infor-mation on long-term dissolution of these wasteglasses in a can-in-canister configuration must beevaluated.
The long-term dissolution of nuclear wasteglasses, including HLW, low-activity waste(LAW), and Pu-bearing LaBS and Alkali-Tin-Sili-cates (ATS) glasses, has been studied for more thantwo decades (Bates et al., 1982, 1995, 1996; Bibleret al., 1996; Ebert and Bates, 1993; Ebert andTam, 1997; Gin et al., 2001; McGrail et al., 2001;Ramsey et al., 1995; Vernaz et al., 2001). Althoughthese results are useful, some of the test methodsused for these studies yielded data with limitedapplicability to the development of models thatdescribe the long-term behavior of glass underrepository-relevant conditions. The majority of theavailable results on the accelerated weathering ofthese waste forms have been produced with statictest methods, such as the vapor hydration test(VHT) (Bates et al., 1982) and product consistencytest (PCT) (ASTM, 1994; Jantzen and Bibler,1987). Both VHT and PCT are closed-system testsconducted under conditions that are quite differentfrom the open-system conditions expected in a geo-logic repository. Specifically, VHT and PCT do notallow for mass transport processes to occur, such asthe transport of water and/or atmospheric gases.The VHT is an unsaturated test method for whicha specimen is suspended in a container with a spec-ified volume of water required to obtain a targetwater vapor saturation. This test method providescritical information about the secondary phaseparagenesis of the waste form, but no informationis obtained about the solution chemistry in contactwith the glass, which is required for long-term per-formance assessment. Contrary to VHT, PCT is awater-saturated static test. For a PCT, dissolvedmaterial is allowed to accumulate in the aqueousphase, thus altering the chemistry of the solutionin contact with the glass. Although informationabout the solution chemistry is obtained, thesechanges may not be representative of the solutionchemistry that is expected in an open-system dis-posal facility. For a more detailed discussion onthe dissolution of HLW glasses see Bates et al.
(1994a,b) and of LAW glasses see McGrail et al.(2001) and Pierce et al. (2004) as well as the refer-ences contained therein.
A lanthanide borosilicate glass, based upon acommercial glass formulation created in the 1930s(Loffler, 1932), was developed at the SavannahRiver Site as a high-temperature glass compositionthat can be used to incorporate Pu (Plodinec,1979; Ramsey et al., 1995). Unlike typical HLWand LAW glasses that contain as much as 20 mass%Na2O, LaBS glass is devoid of alkalis. This lack ofalkalis is believed to be the cause of the decreaseddissolution rate observed in PCTs conducted withLaBS, approximately 20–50 times lower than ratesmeasured in PCTs conducted with typical HLWglass (Mertz et al., 1998). Although a few research-ers have examined the dissolution of LaBS glasswith the VHT, PCT, and SPFT methods (Bateset al., 1995; Bibler et al., 1996; Fortner et al.,2000; Mertz et al., 1998; Ramsey et al., 1995; Stra-chan et al., 1998), none of these studies addressedthe long-term dissolution in a can-in-canister wastepackage configuration under repository-relevantconditions.
In the can-in-canister configuration, several sce-narios of water contact are possible. From a hydro-dynamic viewpoint, after water has breached thecontainer, the most credible contact mode is theslow percolation of water through the waste pack-age under conditions of partial hydraulic saturation.Under these open-flow and transport conditions,water flow through the container is expected to becontrolled by gravity and capillary forces. Thegreater proportion of HLW glass in comparison toLaBS glass suggests that, on average, the dissolu-tion of the HLW glass will dominate the solutionchemistry of any water percolating through a can-in-canister waste package. The objectives for thisinvestigation were to (1) understand how theHLW glass-water interaction impacts the releaseand transport of Pu, (2) evaluate the paragenesisof secondary phases, and (3) evaluate the releaseof neutron absorbers. In addition to the aboveobjectives, the main objective for this study was toaccurately assess the dissolution of HLW and LaBSglass in this can-in-canister configuration with anunsaturated flow-through test method in whichrepository conditions are closely mimicked. As aresult of the above objectives, the canister metalswere not included in this test and should be consid-ered for future experiments. In this paper, theresults from a 6-a waste form-waste form interac-
tion experiment conducted in a sandwich configura-tion (i.e., representative of the can-in-canisterconfiguration) with the PUF test method are wesummarized. The term waste form-waste form, inthis document refers to the relationship betweenthe two glasses used in the experiment, LaBS glassand a simulated HLW borosilicate glass, SRL-202.
2. Glass chemistry and material preparation
The SRL-202 glass and LaBS frit (without PuO2)were prepared by mixing measured amounts ofdried reagent-grade chemicals in a ceramic ball mill.The mixtures were melted at 1450 �C for 1 h in a Pt(90%) Rh (10%) crucible, and the molten glass waspoured onto a cool stainless steel plate. After beingquenched, the LaBS frit was ground into a powderwith a ceramic ball mill and mixed with PuO2 in aratio of 4.5 g of PuO2 to 35.5 g of LaBS frit. TheLaBS frit and PuO2 mixture was then melted at
Table 1Chemical composition (mass%), particle density, geometric andBET surface area, and surface roughness for SRL-202 and LaBSglasses
1450 �C and periodically stirred to produce 40.0 g ofPu-bearing LaBS glass. For the remainder of thisdocument the Pu-bearing LaBS glass will bereferred to as LaBS. The composition of each glass(see Table 1) was confirmed by use of inductivelycoupled plasma optical emission spectroscopy(ICP-OES) and mass spectrometry (ICP-MS) analy-ses of Na2O2 and LiBO2 fusions of glass samples.
The SRL-202 glass specimen contains 19 compo-nents with the concentration of Al2O3, B2O3, CaO,Fe2O3, K2O, Li2O, MgO, MnO, Na2O and SiO2
comprising of 96.2 mass% of the glass with minoramounts of other oxides such as Cr2O3, CuO,Nd2O3, NiO, PbO, TiO2, ZnO and ZrO2. Unlikethe SRL-202 glass specimen, the LaBS glass speci-men is devoid of alkalis (Na2O, K2O, and Li2O)and contains as much as 11.4 mass% PuO2. Approx-imately 30 mass% of the LaBS glass is composed ofrare earths (Gd2O3, La2O3, Nd2O3) with the remain-ing amount consisting of Al2O3, B2O3, PuO2, SiO2,SrO and ZrO2. The rare earth Gd, 7.6 mass%Gd2O3, was added to the LaBS as a neutronabsorber.
The SRL-202 and LaBS glass specimens used inthese experiments were prepared by crushing thesamples in a ceramic ball mill. The crushed glasswas then sieved into the desired size fraction, <250to >170 lm (<20 to >70 mesh), with ASTM stan-dard sieves (ASTM, 2001). After being sized, theglass specimens were washed in 18 MX ultra puredeionized water (DIW), washed again with 18 MXDIW in an ultrasonic bath, rinsed in ethanol, anddried in an oven at 90 �C (±2 �C). Each glass sam-ple, before and after testing, was stored at roomtemperature (�23 �C) in a seal container dessicatorthat contained CaSO4.
2.1. Specific surface area measurement and
calculation
The specific surface area of each specimen wascalculated with a geometric formula (McGrailet al., 1997b) Eq. (1),
Sgeo ¼3
qrð1Þ
where Sgeo is the surface area (m2/kg), q is the glassdensity (kg/m3), and r is average radius (m). Theglass density, measured with a He gas pycnometer(Micromeritics, Norcross, Georgia), was determinedto be 2.71 · 103 kg/m3 for SRL-202 glass and3.56 · 103 kg/m3 for LaBS glass, with a correspond-
ing geometric surface area of 4.2 ± 0.8 m2/kg (i.e.,0.0042 ± 0.0008 m2/g) and 3.2 ± 0.6 m2/kg (i.e.,0.0032 ± 0.0006 m2/g), respectively. Eq. (1) assumesthe crushed particles are spherical with no surfaceflaws or porosity and the size distributions of thegrains are normally distributed. For comparison,Kr-adsorption BET measurements were determinedfor the SRL-202 glass (Brunauer et al., 1938) andwas determined to be 19.0 ± 0.2 m2/kg (i.e.,0.019 ± 0.0002 m2/g), which is approximately 4.5times higher than the calculated geometric surfacearea. Because these glasses have very little, if any,porosity as confirmed by SEM, it is believed thehigher value yielded by the Kr-BET analysis arethe result of finer-grained particles (higher surfacearea particles) that have adhered to the material ofthe desired size fraction. The results are consistentwith the results from experiments with low-activitywaste glass monoliths (McGrail et al., 2000b),non-porous natural glasses (Wolff-Boenisch et al.,2004), as well as other borosilicate glasses (Papeliset al., 2003), which suggest the geometric surfacearea best represents the overall glass surface area.Therefore, the rates reported in this study were cal-culated with the geometric surface area. Anotherfactor that complicates the estimate of specific sur-face area is the change each sample undergoes overthe duration of the experiment. Therefore, the equa-tion developed by McGrail et al. (1997b), which al-lows for the change in the sample mass over theduration of the experiment to be computed, wasused to compensate for this effect. Changes in spec-imen masses were calculated from the background-corrected Li concentrations in the effluent solutionsfor SRL-202 and corresponded to a mass loss of�12.4%; whereas the BLaBS effluent concentrationwas used for LaBS glass and corresponded to amass loss of �2.7% after 6a of testing. The BLaBS
concentration was calculated by difference, addi-tional details on extracting the B contribution fromLaBS glass is discussed later in this paper.
3. Pressurized unsaturated flow (PUF) test method
The PUF apparatus (Fig. 1) allows for acceler-ated weathering experiments to be conducted underhydraulically unsaturated conditions, thereby mim-icking the open-flow and transport properties ofthe disposal system environment while allowingthe dissolving glass to achieve a final reaction state.The final reaction state, commonly referred to asstage III in the weathering process of glasses, is a
point reached during long-term weathering thatconsists of the formation of secondary phases, whilestage I and II mechanisms (e.g., network hydrolysis,ion exchange, and network dissolution) occur simul-taneously. The PUF apparatus provides the capabil-ity to vary the volumetric water content fromsaturation to 20% or less, minimize the flow rateto increase liquid residence time, and operate at amaximum temperature of 90 �C. The PUF columnoperates under a hydraulically unsaturated condi-tion by creating a steady-state vertical water flow,while maintaining uniform water content through-out the column; by using gravity to assist in drain-age; and by maintaining a constant pressurethroughout the column. Constant pressure is main-tained with a porous Ti plate and gas pressure.
The PUF system and test procedure have beendescribed previously by McGrail et al. (1996,1999, 2000a) and Pierce et al. (2004, 2006), and onlya general description is provided in this paper. ThePUF system contains a 0.0762-m long and 0.0191-m diameter column fabricated from a chemicallyinert material, polyetheretherketone (PEEK), so
that dissolution reactions are not influenced byinteraction with the column material. A porous Tiplate with a nominal pore size of 0.2 lm is sealedin the bottom of the column to provide an adequatepressure differential for the conducting of fluid whileoperating under unsaturated conditions (Wierengaet al., 1993). Titanium was chosen because it ishighly resistant to dissolution and has excellent7wetting properties. Once the porous Ti plate iswater-saturated, water but not air is allowed toflow-through the 0.2 lm pores, as long as theapplied pressure differential does not exceed theair entry relief pressure, referred to as the bubblepressure, of the Ti plate. If the pressure differentialis exceeded, air will escape through the plate andcompromise the ability to maintain unsaturatedflow conditions in the column. The PUF test com-puter control system runs LabVIEWTM (NationalInstruments Corporation) software for logging testdata from several thermocouples, pressure sensors,inline sensors that measure effluent pH and conduc-tivity, and from an electronic strain gauge that mea-sures column weight to accurately track water mass
Fig. 1. Schematic of the second generation pressurized unsaturated flow (PUF) apparatus, which has the ability to conduct twosimultaneous tests. The PUF apparatus consist of an influent reservoir, syringe pump, insulation wrapped column, electronic balance,pressure line and reservoir (PUF port), influent and effluent solution 1/16th Teflon lines, thermocouples (type J and type T), in line pHprobe and electrical conductivity meter, collection vial, and a computer that records column weight, pH and electrical conductivity. Thecolumn was heated by applying an electrical current to a heat taped wrapped Al sleeve. A 0.2 lm Ti porous plate, situated at the base ofthe column, constant gas pressure, constant water flow, and gravity assisted drainage allows for the PUF column to operate under ahydraulically unsaturated condition.
balance and saturation level. The column alsoincludes a PUF port, which is an electronically actu-ated valve that periodically vents the column gases.The purpose of column venting is to prevent reduc-tion in the partial pressure of important gases, espe-cially O2 and CO2, which may be consumed in avariety of chemical reactions.
The PUF column for the waste form-waste forminteraction experiment was packed first withapproximately one-half the total required SRL-202glass (12.87 g), then with the LaBS glass (8.64 g),and finally with the remaining SRL-202 glass(13.65 g) (Fig. 2). This resulted in a packed PUFcolumn that contained approximately 80% SRL-202 glass and 20% LaBS glass. The mass differencebetween the full and empty column was used to cal-culate the initial porosity of approximately0.44 ± 0.03 (unitless). Mass change and bed volumewere also tracked while packing each layer to com-pute the porosity of each layer. Individual bedporosity was within the experimental uncertaintyof the measurement reported above. After packing,the column was vacuum saturated with 18 MX DIWat ambient temperature. A temperature controllerwas then programmed to heat the column to90 �C(±2 �C) in approximately 1 h (1 �C/min). Thecolumn initially was allowed to desaturate by grav-ity drainage during heating and was also ventedperiodically to maintain an internal pressure lessthan the bubble pressure of the porous plate. Afterreaching 90 �C(±2 �C), the DIW influent valve wasopened and influent was set to a flow rate of1 mL/day. The influent solution, 18 MX DIW, was
stored in a sealed 50 mL influent reservoir whichwas periodically refilled during the experiment. Col-umn venting was set to occur once per hour, so thatthe partial pressure of O2 and CO2 could remain rel-atively constant.
3.1. Effluent solution analyses
All effluent solutions were monitored for pHand electrical conductivity with in-line sensors.Prior to starting the experiments, the in-line pHprobe was calibrated with NBS buffers (pH 7.00,10.00, or 12.00 at 25 �C). Precision of pH measure-ment was ±0.02 pH units. The in-line PharmaciaBiotech electrical conductivity sensor was cali-brated with a freshly made solution of 1.0 M NaCl.The 1.0 M NaCl solution was prepared by adding11.67 g of analytical grade NaCl powder to200 mL of 18 MX DIW. Concentrations of Gd,La, Nd and Pu in effluent solution samples weredetermined with ICP-MS methods; whereas con-centrations of Al, B, Cr, Fe, K, Li, Na, Si, Tiand Zr were determined with ICP-OES methods.After passing through the 0.2 lm Ti porous plateand the inline sensors, aliquots of the effluentsolutions were acidified with ultra high-purity con-centrated HNO3 and analyzed by ICP-MS andICP-OES methods.
3.2. Post-test solid phase analyses
After 2158 days (approximately 6-a), the PUFexperiment was terminated. Upon termination, thecolumn was vertically split into two halves, andthe reacted solids were sub-sampled as found (looseand moist particles) as a function of depth (2–3 mmintervals). The sub-samples were placed in glassvials, dried at room temperature in a sealed con-tainer with CaSO4 desiccant, and analyzed for sec-ondary reaction products with X-ray diffraction(XRD), scanning electron microscope (SEM), andalpha energy analysis (AEA).
Powder XRD patterns were recorded in a Scin-tag� automated powder diffractometer (Model3520) with Cu Ka radiation X-ray tube(k = 1.54 A). Data were collected in the 2h range:2–65�, with a scanning step size of 0.02� 2h and adwell time of 2 s. Before mounting, a representativesample of the bulk material was ground with anagate mortar and pestle and sealed in a specializedXRD holder (Strachan et al., 2003). The data wereanalyzed with the computer program JADE
Fig. 2. Schematic of the column setup for PUF waste form/wasteform interaction experiment.
(MDI, Livermore, California) combined with theJoint Committee on Powder Diffraction Standards(JCPDS) International Center for Diffraction Data(ICDD) (Newtown Square, Pennsylvania) database.
A JEOL JSM-840 SEM was used to determineparticle morphology and size. The system isequipped with an Oxford Links ISIS 300 energy dis-persive X-ray analysis spectroscopy (EDS) that wasused for qualitative elemental analysis. Operatingconditions were 20 keV for SEM imaging, and 100live seconds with 20–30% dead time for the EDSanalyses. The EDS analyses of particles are limitedto elements with atomic weights heavier than B.Photomicrographs of high-resolution secondaryelectron images were obtained as digital imagesand stored in electronic format. The SEM-EDSmounts consisted of double-sided C tape attachedto a standard Al planchet. The sample mounts werethen coated via vacuum sputtering to improve theconductivity of the samples, and thus, the qualityof the SEM images and EDS signals.
The distribution of 239Pu and 241Am as a func-tion of column depth was determined by analyzingsamples of reacted particles with an OxfordOasisTM Alfa Energy Analysis (AEA) system thatcontains eight individual alpha detectors. This dis-tribution was used to evaluate how far, if at all, Puhas migrated into the bottom half of the SRL-202glass bed. All detectors were energy and efficiencycalibrated for a distinct geometry using standardswith known activities of analyzed isotopes. Identi-fication of 239Pu and 241Am spectra was performedwith the 5.156 MeV peak and 5.486 MeV peak,respectively.
4. Quantification of the elemental release rates andglass dissolution rates
The elemental release rates from the PUF col-umn, based on the concentration of elements mea-sured in the effluent solution samples, weredetermined by the following formula:
Eri;j ¼4eqðci;j;L � ci;j;bÞ
hSgeo;jð1� eÞqpd2Lð2Þ
where ri is the elemental release rate of the ith ele-ment [mol/(m2 s)], ci,jL is the effluent concentrationof the ith element released from the jth glass (mol/L), cijb is the background concentration of the ithelement released from the jth glass (mol/L), d isthe column diameter (m), L is the column length(m), q is the volumetric flow rate (L/s), Sgeo,j is
the specific surface area for the jth glass, e is theporosity (unitless), q is the glass density (kg/m3),and h is the volumetric water content (unitless).Although the majority of the elements exiting thePUF column are present in only one glass, five ele-ments (Al, B, Nd, Si and Zr) are contained in bothglasses. For these five elements, an average particledensity (q = 3.13 · 103 kg/m3) and geometric sur-face area ðSgeo = 3.7 m2/kg) was used to computethe elemental release rates. The volumetric watercontent is calculated based on the mass of a vol-ume of water in a fixed column volume, accountingfor changes in the solution density resulting fromtemperature changes. The background concentra-tion for most elements is typically below the esti-mated quantification limit (EQL) for therespective analysis. The EQL is defined as the low-est calibration standard that can be determinedreproducibly during an analytical run within 10%of the certified value multiplied by the sample dilu-tion factor. The lowest EQLs were for La (from0.2 to 1 lg/L), Gd (from 0.2 to 1 lg/L), Nd(5 lg/L) and Pu (from 0.002 to 0.2 lg/L). HigherEQLs were found for Cr (62.5 lg/L) and Ti(50 lg/L), with the EQLs being the highest for K(1000 lg/L), Na (1000 lg/L), Li (10,000 lg/L)and Si (10,000 lg/L). In cases where the analyteis below the EQL, the background concentrationof the element is set at the value of the EQL.The elemental release rate allows for a comparisonof the release behavior of each element exiting thePUF column, without normalizing the elementsbased on the composition of the glass.
Unlike the elemental release rate, glass dissolu-tion rates, based on the concentration of elementsin the effluent, were normalized to the amount ofthe element present in the glass specimens and deter-mined by
ri;j ¼Eri;j Mi86400
fi;jð3Þ
where ri is the glass dissolution rate based on the ithelement released from the jth glass [g/(m2 d)], Eri;j isthe elemental release rate [mol/(m2 s)], fi is the massfraction of the ith element released from the jth glass(unitless), and Mi is the molecular weight of the ithelement (g/mol).
An estimate of the 2r experimental uncertaintyfor the elemental release and glass dissolution rateswere determined with error propagation. For addi-tional details on the error propagation equationfor PUF experiments see Pierce et al. (2006).
Results from the computer-monitored test met-rics, volumetric water content (h), pH, and electricalconductivity (X), are shown in Fig. 3. The sensordata were smoothed using a bi-square weightingmethod where the smoothed data point, ys, is givenby ys = (1 � x2)2. The parameter x is a weightingcoefficient calculated from a window surroundingthe smoothing location in the set of the independentvariables. A low-order polynomial regression (order2 in this case) is used to compute x for eachsmoothed value. The smoothed data are providedas lines and were used to make qualitative assess-ments of the results Fig. 3.
The volumetric water content (h) results illustratethat, although there were minor excursions during
this experiment, the volumetric water content (h)was relatively steady until day 1500. From day1500 until the test was terminated, the volumetricwater content increased steadily, going from anaverage of 0.26 ± 0.02 to 0.34 ± 0.01. This is dueto changes in the columns hydraulic propertiescaused by an increase in secondary reaction prod-ucts, as well as an increase in the waters of hydra-tion associated with these phases. The formationof fine-grained phases causes the water-retentionto increase as well as the hydraulic conductivityand rate of water flow through the column todecrease, thereby causing the water content toincrease as seen in Fig. 3a. This trend has also beenobserved in the early stages of a PUF test with aless-durable glass formulation, LD6-5412, (McGrailet al., 1999). McGrail et al. (1999) measured thehydraulic conductivity as a function of PUF testduration with the unsaturated flow apparatus
Fig. 3. Computer-monitored test metrics for the volumetric water content (a), pH (b) and electrical conductivity (c). The grey lines are thebi-squared smoothed fit of the raw data and are provided as a guide to the eye.
(UFA) (Conca and Wright, 1992; Gamerdinger andKaplan, 2000). The results from McGrail et al.(1999) illustrated that changes in the water-retentioncharacteristics of the hydrated glass are manifestedas an increase in the volumetric water content dur-ing a PUF test and this increase is related to the ini-tial precipitation of the zeolitic alteration phases,phillipsite (KCaAl3Si5O16) and gobbinisite(Na4CaAl6Si10O32 Æ 12H2O).
Unlike the volumetric water content, pH andelectrical conductivity (X) excursions occur muchmore frequently (Fig. 3b and c). Note the pH probewas lost after day 1400; consequently, data beyondthis point are not available. A direct comparisonof available pH, electrical conductivity, and solutionchemistry data suggest that these data excursionsare correlated to chemical changes occurring at theglass-water interface. These changes can cause theglass reaction rate to increase by orders of magni-tude. For example, it has been shown in flow-through experiments that an increase in pH, from9 to 10, causes an order of magnitude increase inthe reaction rate of aluminoborosilicate glasses(McGrail et al., 1997b; Pierce et al., 2004).
5.2. Effluent solution chemistry
Results from the analyses of effluent samples areprovided in Fig. 4 and represent the release of ele-ments from the PUF column as a whole and notfrom an individual glass specimen. Release of ele-ments from the column illustrates a general trendof decreasing concentration with increasing reactiontime during the early stages of the test (first 200days). The concentrations of B, K, Li, Na and Siare as much as 1 · 105 times greater than Al, Cr,Fe, Nd, Ti and Zr (Fig. 4a, b and c). Fig. 4c alsoillustrates that release of Gd, La and Pu from theLaBS glass ranged from as little as 1 · 103 to asmuch as 1 · 108 times lower than the other elements.Under these conditions, B, K, Li and Na are moresoluble than Al, Cr, Fe, Nd, Ti and Zr and muchmore soluble than Gd, La, and Pu. The high con-centration of Pu observed in these effluent samplesis the result of colloidal particles exiting the column.Results from alpha energy analysis (AEA) of ali-quots of select effluent samples that were filteredthrough a 1.8 nm filter suggest that approximately80% of the total measurable Pu was in the form of
Time, days0 500 1000 1500 2000
log 10
(C
once
ntra
tion)
, (m
mol
/L)
-5
-4
-3
-2
-1
0
1
2
AlBCrFeK
Time, days0 500 1000 1500 2000
log 10
(C
once
ntra
tion
), (
mm
ol/L
)
-5
-4
-3
-2
-1
0
1
2
LiNaSiTiZr
Time, days0 500 1000 1500 2000
log 10
(C
once
ntra
tion
), (
mm
ol/L
)
-10
-9
-8
-7
-6
-5
-4
-3
-2
-1
LaGdNd239Pu
PuO2 Solubility @ pH(23˚C) = 9.0
a b
c
Fig. 4. Log10 concentration of elements, in mmol/L, released from the PUF column measured in effluent solutions as a function of time, indays. The Pu concentration in equilibrium with PuO2 at pH (23 �C) = 9.0 is shown as a dashed horizontal line in (c). Five of the elements;Al, B, Nd, Si and Zr, shown in the above plots are being released from both glasses.
a filterable particulate, in comparison to unfilteredaliquots of the same sample. These results suggestthe filterable particles are >1.8 nm but smaller thanthe 0.2 lm average diameter openings of the Ti por-ous plate situated at the base of the column. This isconsistent with PUF (McGrail et al., 2000a) andPCT (Ebert and Bates, 1993) tests conducted onPu-bearing ceramic and glass, respectively. Ebertand Bates (1993) observed Pu-bearing colloidal par-ticles that were approximately 1.8 nm in size. Trans-port of colloids in a PUF test is to be expected,because the system is advection dominated. Thiscan be demonstrated by calculating a Peclet (Pe)number for this experiment with Eq. (4);
P e ¼qhA
� �x
D; P e > 4 ðadvection-dominated systemÞ
ð4Þwhere q is the flow rate (1.2 · 10�11 m/s), h is theaverage volumetric water content, which changedfrom 0.26 ± 0.02 to 0.34 ± 0.01, A is the columnarea (2.85 · 10�4 m2), x is the column distance(0.0762 m), and D is the molecular diffusion coeffi-cient for water (assumed to be 1.0 · 10�9 m2/s). Thiscalculation resulted in an average Pe number of10.5 ± 2. This value clearly shows that this PUF testis advection-dominated.
Despite the highly advection-dominant system,the steep 239Pu and 241Am concentration gradientdownstream from the LaBS glass indicates a strongattenuation of Pu at the lower LaBS/SRL-202 glassinterface, evident by the 5-order-of-magnitude dropin the measured activity over a distance of 10 mm(Fig. 5). Retention of Pu may be occurring by anynumber of or a combination of factors such as: (1)adsorption from solution onto colloidal particles,(2) colloid filtration (e.g., entrapment of Pu-bearing
colloids in the lower SRL-202 glass beads), and/or(3) formation of Pu-bearing secondary phases.Although it is impossible to distinguish these mech-anisms from the AEA data alone, most of the Puexiting the column was colloidal and suggests col-loid filtration may be the retention mechanism caus-ing the dramatic drop in 239Pu activity.
A comparison of Figs. 3 and 4 illustrates goodcorrelation between the observed excursions in efflu-ent pH and electrical conductivity with peak con-centrations of constituents released from the PUFtest. These results suggest the SRL-202 glass, andperhaps the LaBS glass, have undergone periodicexcursions in reaction rates. However, a majorityof the elements contained in the LaBS glass aresparingly soluble under these test conditions, so itis highly unlikely that a dissolution rate accelerationof the LaBS glass could cause the transients in pHand electrical conductivity. Periodic excursions inthe effluent pH and electrical conductivity have beenobserved in previous PUF tests with the SRL-202glass, as well as with other glass formulations, andhave been correlated with the formation of second-ary phases, especially Na–Ca aluminosilicate zeo-lites (McGrail et al., 1999, 2000a; Pierce et al.,2006).
5.3. SRL-202 and LaBS glass elemental release rate
A comparison of the elemental release rates, Eq.(2), for the major and a few minor components inSRL-202 and LaBS glasses is shown in Fig. 6. Sim-ilar to the concentration data, the rate of elementrelease from the column decreases as the reactiontime increases, especially during the early stages oftesting (Fig. 6). The most significant decrease isobserved during the first year of testing, followedby periodic increases in the elemental release rate.The decrease in the rate of element release suggeststhe value of the ion activity product is approachingequilibrium with respect to a rate-limiting secondaryphase. Over the 6-a testing period, the overallrelease of elements from the PUF column hasdecreased and reached an average steady-staterelease rate, based on B release, of (2.87 ± 0.27) ·10�10 mol/(m2 s), excluding the periodic excursionsin reaction rate. The results in Fig. 6 were also usedto determine the average rate of SRL-202 glass dis-solution, based on Li release, with Eq. (3). The cal-culated rate for the SRL-202 glass was estimated tobe (1.22 ± 0.18) · 10�2 g/(m2 d) (Fig. 7). This rate is8.3 times greater than a rate previously reported by
McGrail et al. (1997a) after a 70-day PUF test withSRL-202 glass, (1.47 ± 0.13) · 10�3 g/(m2 d). Thepresent results suggest that the presence of LaBSglass may be causing the SRL-202 glass to dissolvefaster than previously observed and may be theresult of changes in the solution chemistry thatoccur as the water migrates first through SRL-202
glass bed, to the LaBS glass bed, and finally throughanother SRL-202 glass bed during this waste form-waste form interaction PUF column.
Although the release of alkali elements (i.e., K,Li, and Na) from glass can occur via two mecha-nisms, matrix dissolution and alkali-H+ ionexchange (Bates et al., 1994a,b). It has been illus-trated in single-pass flow-through test (SPFT) andPUF test with LAW glass, (conducted at 90 �Cand a pH (23 �C) > 9.0) that matrix dissolution isthe dominant reaction mechanism, and the processof alkali-H+ ion exchange is negligible (McGrailet al., 2001; Pierce et al., 2004, 2006).
5.4. LaBS glass dissolution rate
An estimate of the LaBS glass dissolution rate inthis waste form-waste form interaction PUF testwas determined by subtracting the SRL-202 glassaverage normalized Li release rate (rLi,SRL-202) fromthe average normalized B release rate [measured bytotal B release, (1.24 ± 0.05) · 10�2 g/(m2 d)] andmultiplying by the ratio of LaBS to SRL-202 massfractions for B [Eq.].
Time, days0 500 1000 1500 2000lo
g 10 (
Ele
men
tal R
elea
se R
ate)
, [m
ol/(
m2 s
)]
-15
-14
-13
-12
-11
-10
-9
-8
AlBCrFeK
Time, days
0 500 1000 1500 2000
log 10
(E
lem
enta
l Rel
ease
Rat
e), [
mol
/(m
2 s)]
-15
-14
-13
-12
-11
-10
-9
-8
LiNaSiTiZr
Time, days0 500 1000 1500 2000
log 10
(E
lem
enta
l Rel
ease
Rat
e), [
mol
/(m
2 s)]
-19
-18
-17
-16
-15
-14
-13
-12
-11
GdLaNd239Pu
ba
c
Fig. 6. Element release rates, in mol/(m2 s), from both SRL-202 and LaBS glasses as a function of time is shown in (a–c). Five of theelements; Al, B, Nd, Si and Zr, shown in the above plots are being released from both glasses.
Time, days
0 500 1000 1500 2000
log 10
(N
orm
. Dis
solu
tion
Rat
e), [
g/(m
2 d)]
-5
-4
-3
-2
-1
0
BLaBS
KLiNa
Fig. 7. Glass dissolution rate, in g/(m2 d), as a function of time isshown for BLaBS, Li, K and Na. The dissolution of SRL-202 glassis represented by the normalized release of the alkali elements (K,Li and Na) whereas, LaBS glass dissolution was estimated bydifference and is represented by BLaBS.
ð5ÞWith this approach, the average normalized B re-lease rate for the LaBS glass (BLaBS) was estimatedto be (1.22 ± 0.14) · 10�3 g/(m2 d). The measuredrate for LaBS glass is 10 times lower than forSRL-202 glass; similar results have been observedwhen comparing rates from static tests (Fortneret al., 2000; Mertz et al., 1998). A comparison ofthe rates as a function of time for SRL-202 glass;indicated by the release of K [(2.37 ± 0.18) ·10�2 g/(m2 d)], Li [(1.22 ± 0.18) ·10�2 g/(m2 d)],and Na [(1.08 ± 0.11) · 10�2 g/(m2 d)], and LaBSglass, indicated by BLaBS release, is shown inFig. 7. A comparison of the average LaBS glass dis-solution rate based on BLaBS, suggests that Gd[(1.87 ± 0.17) · 10�7 g/(m2 d)], La [(8.27 ± 0.95) ·10�8 g/(m2 d)], and Pu [(5.03 ± 1.39) · 10�9 g/(m2 d)]are being retained in the column, either as a discretesecondary phase(s) or via adsorption and/or absorp-tion with secondary phases formed as a result of theglass-water reaction. Since Gd, La and Pu are spar-ingly soluble the reported dissolution rates based onthese elements, are used to compare the relativemagnitude of release from the PUF column in com-parison to BLaBS and is not indicative of the dissolu-tion rate for LaBS glass.
5.5. Analysis of reacted SRL-202 glass particles
and geochemical modeling
The moisture associated with reacted glass parti-cles removed from the PUF column is shown inFig. 8 as a function of depth. Results suggest thevolumetric water content for each column section
decreases with depth away from the column inletand the most extensive secondary phase formation(and consequent glass degradation) occurred nearthe column inlet, evident by the peak in water con-tent for these particles. Analysis of the reactedgrains at various depths with SEM confirmed thatsamples removed from the top of the column hadthe most extensive alteration. Figs. 9 and 10 illus-trate the presence of a gel layer and numerous par-ticles with a spherical morphology. Here the termgel layer refers to a hydrolyzed layer on the glasssurface that forms as a result of condensation reac-tions that occur at the glass-water interface. TheEDS analyses of the spherical particles indicate thepresence of Al, K, Ca, Fe, Mg, Mn, Ni and Zr;but in comparison to the pristine glass, these spher-ical particles are enriched in Al, Fe, and Si, with asmall amount of Na. The presence of these sphericalparticles is not surprising and has been observed inprevious PUF tests with the SRL-202 glass con-ducted for shorter durations (McGrail et al.,1997a). Further examination of these images sug-gests that some of these particles are embeddedand probably growing out of a gel layer (Fig. 10).
In addition to SEM-EDS analyses, powder XRDand geochemical modeling was used in an attemptto characterize the secondary phase(s) that precipi-tated in the PUF column. The bulk powder XRDresults shown in Fig. 11 suggest the secondary phasepresent in the SEM images may be the smectitedi-octahedral clay mineral nontronite [Na0.33Fe2-(AlSi)4O10(OH)2 Æ n(H2O)], although the 100%reflection (e.g., 9.2 2h) for nontronite was notobserved. Typically, XRD requires the presence of5 wt% or more for detection. To provide additionalsupport for the presence of nontronite, the satura-tion state of the effluent solutions with respectnontronite was evaluated with geochemical model-ing. Applying geochemical modeling, with the ther-modynamic database and reaction code EQ3NR[version 8.0 (Wolery, 1992)], to the measure concen-tration of elements in select effluent samples andallowing the hypothetical solid solutions to precipi-tate, suggests these samples are super-saturated withrespect to two smectite di-octahedral clays, nontro-nite and beidellite [Na0.33Al2.33Si3.67O10(OH) 2].
5.6. Analysis of reacted LaBS glass particles
SEM images (Fig. 12) and X-ray EDS analyses ofreacted LaBS glass confirmed the presence of a dis-crete Pu-bearing secondary phase, probably PuO2,
Distance From Inlet, mm0 20 40 60 80
Vol
umet
ric
Wat
er C
onte
nt
0.00
0.05
0.10
0.15
0.20
0.25Side ASide B
Direction of Flow
LaB
S
#19
#18
SRL
-202
SRL
-202
Ti p
orou
s pl
ate
Fig. 8. Moisture fraction as a function of distance from the PUFcolumn inlet.
which was the predicted secondary phase thatformed when modeling the effluent solution chemis-try with the EQ3NR code (Wolery, 1992). Numer-ous deposits with a plate-like morphology werefound on glass particles removed from a columndepth between 32 and 34 mm. These particles havea high Pu content, do not contain any of the other
major LaBS glass components (i.e., Al, Gd, La,Nd and Si), and appear to have precipitated onthe surface of the LaBS glass. Alteration productsthat contain the neutron absorber, Gd, have notbeen positively identified. Gadolinium is expectedto form insoluble secondary phases after beingreleased from the glass. Therefore, the separation
Fig. 9. SEM photographs of reacted SRL-202 glass particles removed from the top of the PUF column (between 4 and 8 mm from thetop). A smectite-di-octahedral clay phase is shown as spherical particles growing out of a hydrated gel layer on the SRL-202 glass grains.X-ray EDS analyses (eds05) suggest these particles are enriched in Al, Fe and Si, with a small amount of Na; in comparison to the pristineglass; where as the hydrated gel layer (eds06) contains majority of the SRL-202 glass components.
of this neutron absorber from Pu during LaBS glassdissolution and transport could be a criticality issuefor the proposed repository. Previous studies (Bates
et al., 1995; Fortner et al., 2000; Mertz et al., 1998)have shown the formation of individual andadhered Pu-bearing colloids, but none have shown
Fig. 10. SEM photographs of reacted SRL-202 glass particles removed from the top of the PUF column (between 4 and 8 mm from thetop). A large number of particles with a spherical morphology have accumulated on the surface of the glass particles. The EDS analyses(eds04) suggest these particles mainly contain Al, Fe, K and Si, with trace amounts of Ca, Mn, Mg, Ni, Ti and Zr. The hydrated gel layerin the center (eds03) is enriched in Al, Fe, K and Si in comparison to the spherical particles.
the segregation of Pu from the neutron absorbers.For example, a 98-day PCT-B test conducted byFortner et al. (2000) revealed that Pu was associatedwith secondary phases of oxides, oxyhydroxides andsilicates of Gd, La and Nd. It is very difficult tocompare the present results to results obtained inother studies because of differences in the testdesigns and methods used. Currently, it is not clearwhy Pu appears to have segregated from the neu-tron absorber, and additional analysis is requiredto translate these results to actual disposed wastepackages.
Recent unpublished results from a fabricationconducted with a new LaBS glass formulation thatcontains both Gd and Hf at Savannah RiverNational Laboratory (SRNL) suggest the LaBSglass used in the present study may have containedan insoluble Pu phase that was not dissolved duringthe glass fabrication process (James Marra, pers.comm.). Vienna et al. (1996) observed undissolvedPu when the sample was not stirred during the melt-ing process. The stirring technique was used duringfabrication of the LaBS glass in the present study,but similar techniques were not used at SRNL.Based on the unpublished results from SRNL, it ispossible that the disk-like Pu-bearing phase(s)observed in the reacted glass may have been presentin the unreacted glass prior to being used in thePUF test, although the Pu phase was not observedin SEM images of the unreacted LaBS glass.
5.7. Analysis of Pu release from LaBS glass
A one-dimensional (1-D) steady-state mass bal-ance equation, Eq. (6), was used to predict the depth
at which the Pu concentration approaches satura-tion with respect to PuO2. The governing equationfor describing 1-D chemical transport assuming aconstant volumetric water content and flow-rate atsteady-state is given by:
Do
2ci
ox2� m
oci
oxþ c ¼ 0 ð6Þ
where D is the dispersion coefficient in m s�1
(1.16 · 10�14 m/s), ci is the concentration of elementi in g/m3, v is the interstitial or pore-water velocity(1.35 · 10�7 m3/s), and c is the source release termin g/(m3 s). The pore-water velocity is equal to theflow rate (q = 1.16 · 10�11 m3/s) divided by theaverage volumetric water content for the entireexperiment (h = 0.30 ± 0.01), which was computedby averaging the observed volumetric water contentfrom day 1 to day 1500 and from day 1500 until testtermination, 0.26 ± 0.02 and 0.34 ± 0.01, respec-tively. The source release term, c, 1.66 · 10�4 g/(m3 s) was calculated for the LaBS glass by multi-plying the average steady-state dissolution rate,based on B release, [rLaBS = (1.22 ± 0.14) · 10�3 g/(m2 d)] times the surface area (SLaBS = 2.77 ·10�2 m2) to bed volume (VLaBS = 4.34 · 10�6 m3).Applying a boundary condition of c(0) = ci,o andocox ðLÞ ¼ 0, the resulting analytical solution for Eq.(6) from van Genuchten and Alves (1982) is:
ciðxÞ ¼ ci;o þcxm
þ cDm2
exp � mLD
� �� exp
ðx� LÞmD
� �� ð7Þ
where ci(x) is the concentration of element i at a setcolumn depth in g m�3, ci,o is the initial concentra-tion of element i in g m�3, and L is the total columnlength extending from the column top through theLaBS glass bed, approximately 15 mm. Using Eq.(7), the predicted Pu concentration was calculatedin 0.04 mm intervals. Because of the observed pHfluctuations (see Fig. 3b) and the fact that PuO2 sol-ubility decreases with increasing pH (Allard andRydberg, 1983; Rai et al., 1999, 2001), a conserva-tive experimental pH value of 9.0 was used to pre-dict the boundary condition for the solubility ofPuO2 using the EQ3NR code (Wolery, 1992).Numerically solving Eq. (7) for x using the pH(23 �C) = 9.0 restriction, the concentration of Pu ispredicted to exceed the PuO2 solubility at approxi-mately 29.0 mm. Although these results suggestthe solubility of PuO2 is exceeded soon after the LaBSglass begins to corrode, in dynamic experiments such
2θ10 20 30 40 50 60
1.2 mm
8.4 mm
unreacted
Nontronite [NaFe4(Si7Al)O20(OH)4]
Fig. 11. Background-corrected XRD pattern of reacted SRL-202glass particles taken between 0 and 9 mm from the top of thePUF column, unreacted glass, and the nontronite PDF patternfor comparison.
Fig. 12. SEM photograph of reacted LaBS glass particles removed from the center of the PUF column (between 32 and 34 mm from thetop). X-ray EDS analyses suggest that the disk-like phase is enriched in Pu (possible PuO2) and devoid of the other LaBS glass components(Al, Gd, La, Nd and Si).
as these, the amorphous analog usually forms ear-lier than the crystalline phase. These thermody-namic calculations are based on the solubilityproduct (Ksp) for the crystalline parent, which is al-ways lower than that of the amorphous analog. Thispredicted depth is within 3 mm of the insoluble Pu-bearing phase(s) present on the LaBS glass reactionproducts removed from a column depth between 32and 34 mm (see Fig. 12). These calculated resultsprovide additional confirmation that the majorityof the Pu released from the corroding glass was re-tained in the region of the LaBS glass bed.
6. Conclusions
In a 6-a long PUF experiment, a simulated HLWglass, SRL-202, and LaBS glass for Pu immobiliza-tion have been allowed to react in a manner similarto the way they might react in a repository. Theresults from this experiment have provided severalimportant insights into the long-term release behav-ior of Pu and neutron absorbers from the LaBSglass. The results show a strong coupling betweenthe chemistry of the water percolating through aporous media of the test materials and the corrosionrate of the LaBS glass. Consequently, should theHLW glass undergo sustained acceleration in its cor-rosion rate due to secondary phase formation, theresulting excursion in pH could significantly impactthe corrosion rate of the LaBS glass. The 2–3 pHunit transient excursions observed in the solutionpH exhibited over the entire course of this PUFexperiment are probably caused by the transientacceleration in the glass corrosion rate from the for-mation of alteration phases. These excursions causeas much as a two order-of-magnitude increase in theB release rate from the PUF column. Comparison ofthe corrosion rate after 70 days of testing, suggeststhe SRL-202 glass is experiencing reaction rate accel-eration in this waste form-waste form interactionPUF test, which is evident by the one to twoorders-of-magnitude increase in the dissolution rate.This reaction rate acceleration is almost certainlydue to the formation of a smectite di-octahedral clay,probably nontronite [Na0.33Fe2(AlSi)4O10(OH)2 Æ(H2O)] or beidellite [Na0.33Al2.33Si3.67O10(OH)2], aswell as changes to the solution chemistry at theSRL-202-LaBS glass interface. As water interactswith the top layer of SRL-202 glass, it becomessaturated with respect to the SRL-202 glass compo-nents and reaches a steady-state pH. In this repre-sentative can-in-canister (i.e., sandwich) column
configuration, this same volume of water is under-saturated with respect to several major componentscontained in the LaBS glass, such as Gd, La, Ndand Pu. As these components are dissolved, the solu-tion chemistry changes and reaches a new steady-state pH. Finally, the solution interacts with thebottom layer of SRL-202 glass, causing an addi-tional change in the solution chemistry. This finalshift in the solution chemistry is expected to be lessdramatic than the changes that occurred in the upperlayer of SRL-202 glass. These shifts in the solutionchemistry have a profound effect on the dissolutionof the SRL-202 and LaBS glasses.
Alpha energy analysis (AEA) of aliquots ofselect effluent samples that were filtered througha 1.8 nm filter suggest that approximately 80% ofthe total measurable Pu was in the form of a filter-able particulate, in comparison to unfiltered ali-quots of the same sample. These results suggestthe filterable particles are >1.8 nm but smaller thanthe 0.2 lm average diameter openings of the Tiporous plate situated at the base of the column.In this advection-dominated system, Pu is migrat-ing principally as or in association with colloidsafter being released from the LaBS glass. Furtheranalyses using the reaction rate of LaBS glassand the 1-D advection-dispersion equation, suggestthat the majority of the Pu that is released will beretained in the region of LaBS glass in a can-in-canister configuration. Also, analyses of reactedLaBS glass particles with SEM-EDS illustrates thatPu may have segregated into a discrete disk-likephase, possibly PuO2. Separation of the Pu andthe neutron absorber Gd during glass dissolutionand transport could be a criticality issue for theproposed repository. However, the translationand interpretation of these long-term PUF testresults to actual disposed waste packages requiresfurther analysis.
Acknowledgements
The authors express gratitude to Todd Schaef andSteven Baum (both of Pacific Northwest NationalLaboratory [PNNL]) for their help in analyzing theXRD results discussed in this paper and analyzingthe hundreds of solution samples we generated,respectively. Helpful comments provided by D.M.Strachan (PNNL), J.P. Icenhower (PNNL), andtwo anonymous reviewers are also appreciated. Thiswork was funded by the US Department of Energyunder Contract DE-AC06-76RLO 1830. PNNL is