Our research is one method of investigation currently underway to understand the mobility of the inorganic species in the subsurface and to evaluate the possibility of creating a chemical precipitate that would block or slow the migration of transuranic contaminants. We use Thin Layer Chromatography (TLC) to evaluate movement of barrier constituents, and Secondary Ion Mass Spectrometry (SIMS) to perform analysis of the matrix after infusion.

The TLC technique is commonly used as a separation tool because it is cost-effective, samples are easily prepared, and many samples can be run simultaneously (Hsu et al. 2003). TLC plates consist of thin layers of adsorbent stationary phase (silica gel, alumina, cellulose, polyamides, or ion exchange resins) attached to glass or plastic backing. Movement of a solution occurs by capillary action, and solute separation results from partitioning between the adsorbent stationary phase and the solutes. Once TLC is performed, the challenge becomes locating and identifying the solutes on the TLC plate in order to monitor the mobility of the salts such as sodium phosphate and calcium nitrate. Some conventions for detection include natural color, fluorescence, ultraviolet absorption, radioactive labeling, or the use of chemical or biological detection reagents (Sherma 2000).

Recently, methods have been reported in the literature for analysis of TLC plates using laser desorption mass spectrometry (Chen 1999; Wasch et al. 1998; Hilaire et al. 1998; Chen et al. 2000; Mowthorpe et al. 1999). SIMS has been used less frequently, but also has been shown to be effective (Busch et al. 1992; Busch 1992). SIMS is a highly sensitive technique useful for determining both the atomic and molecular composition of solid surfaces. In SIMS, the sample surface is bombarded with a high-energy (in this case 5 keV) primary ion beam. The primary ions impact the surface, “sputtering” secondary ions into the gas phase. The secondary ions are then drawn into a spectrometer for mass analysis. In our experiments, the mass spectrum generated is used to track the mobility and determine the kind of chemical reactions that occurred. The use of SIMS to analyze small sections of the TLC plates is advantageous because the sample can be studied in the solid state, as a two-dimensional analog to what would occur in the three-dimensional subsurface instrumentation at the INEEL provides simultaneous analysis of anionic and cationic chemical species on the sample surface.

Secondary objectives of the research were to gain an understanding of how well SIMS can detect certain ions off the thin layer chromatograms, and to observe the migration and behavior of these ions. Specifically, ions of interest for our experiments are the silica-derived ion SiO₂^-, and cation and anion species of mono- di- and tri-basic sodium phosphates (Na₃HPO₄, Na₂HPO₄, NaH₂PO₄). This preliminary research will set the stage for more specific experiments designed to emulate the conditions at the INEEL site and further evaluate the viability of using a precipitate to halt the migration of radioactive or heavy metal contaminants. Once immobilized, clean-up could be facilitated or obviated.

The mass spectrometer used in this TLC investigation is a custom, in-house built Triple Quadrupole Secondary Ion Mass Spectrometer (TQ-SIMS). A schematic diagram of this instrument has previously been published (Groenewold et al. 1997). The instrument uses a perrhenate (ReO^4-) primary ion beam to bombard the surface of the sample. Impact causes fragments of chemicals on the sample surface to be desorbed as gas-phase ions, which are sorted by mass and detected by the mass spectrometer, generating a mass spectrum. A unique aspect of the TQ-SIMS is that it can analyze positive and negative ions simultaneously. The ReO^4- primary ion can be readily produced, and generates more abundant molecular secondary ions compared with atomic primary ions (Delmore et al. 1995). This primary ion gun is inside a 12-in diameter spherical vacuum chamber, and operates at ion energy of 5.0 keV, between 190 and 230 pA of primary ion current. The pressure in the instrument is approximately 1 x 10^-6 Torr. A wire flattened at one end was used as a sample holder for the sample to be analyzed. This approach was used for both the chromatographic matrix samples and pure salts that were analyzed to establish benchmark spectral behavior.

TLC

TLC was performed using plastic-backed silica plates, which were cut into strips 1 cm in width and approximately 15-20 cm in length. The plates were spotted with a salt solution to generate a spot with a diameter of about 0.5 cm located about 2.5 cm from the bottom of the plate. The TLC plates were run by placing the strips in a 15 x 100 mm glass test tube containing the solvent at a depth of approximately 1 cm affixed to a ruler (Figure 1). Flow rates were profiled for all experiments to evaluate mobility. This was done by taking measurements every 30 seconds to 1 minute of the height of the solvent front. The solvent front flowed to a height of 10-15 cm up the chromatogram. Once the TLC had dried, samples were taken directly from the backing by impressing a sample holder covered with double-sided tape onto the chromatogram (Figure 2). The matrix adhered to the tape, which was then analyzed directly using the TQ-SIMS. Samples were taken below the solvent level (position -2), between the solvent level and the spot (position -1), at the spot (position 0), and up the paper at intervals of 1 cm (positions 1 and up). These samples were then analyzed in the TQ-SIMS to track how far certain ions, particularly the sodium and phosphate ions, migrated up the TLC strip. The chemicals used included Na₃HPO₄^!12H₂O (J.T. Baker Co.), Na₂HPO₄ (Fisher Scientific), NaH₂PO₄^!H₂O (J.T. Baker Co.), and Ca(NO₃)₂ (Fisher Scientific).

Figure 1 . TLC set up, with test tube affixed to ruler.

To begin the TLC phase of the investigation, Baker-flex plates (J.T. Baker Co.) plastic-coated with cellulose-F were used. In order to determine benchmark mobility and mass spectral behavior, chromatograms were run with water as the solvent, but contained no salt spot. After the chromatogram dried, it was sampled and a secondary ion mass spectrum was generated. Next, two separate chromatograms were spotted with Na₃HPO₄, and then Ca(NO₃)₂, using water as the mobile phase. After the trials using the cellulose paper, the same experiments were conducted using TLC plates that were plastic and coated with silica gel 60, 200 µ (Selecto Scientific). The silica gel plates served as a more accurate model for the subsurface than the cellulose (although cellulose components may exist with the contaminants). In addition, the TLC containing the spot of Na₃HPO₄ was run twice using water as the mobile phase. Dibasic and monobasic forms of sodium phosphate, Na₂HPO₄ and NaH₂PO₄respectively, were also chromatographed using water as the mobile phase to see if the different forms affected mobility. With the silica plates a study was conducted spotting the plates with a 5 µL spot of 0.1 M solutions of the tribasic, dibasic, and monobasic forms of sodium phosphate.

Figure 2. Sample preparation on flattened wire (scale approximately 2x).

Experiments were conducted to determine the minimum concentration of Na₃HPO₄ on the silica gel plates that could be detected by the TQ-SIMS. To do this a Pasteur pipette was used to deposit 1-5 drops of the 0.1 M solution (up to 0.5 M concentration) on the TLC paper, allowing the drops to dry before adding more. Serial dilutions of the 0.1 M solution were performed down to a concentration of 0.038 mM, and single drops were placed on the silica paper. The samples of the solution on the silica were then placed in the TQ-SIMS to see if the ions of interest were observed to give an indication of the minimum detection limit.

Our first set of results served as background information and established standards to use throughout the rest of the experiment. The mass spectrum obtained as a standard from wetted silica paper is shown in figures 3 (cation) and 4 (anion). The predominant peaks in the MS1 cation spectrum represented Si+ (m/z 28) and SiOH+ (m/z 45), while O- (m/z 16), OH- (m/z 17), the silicon oxyanions (m/z 60, 61, 76, 77, 137) were predominant in the anion spectrum. When silica was spiked with one drop of 0.1 M Na₃HPO₄ the appearance of the spectrum changed significantly. The Na⁺ ion was very prominent in the cation spectrum (Figure 5), and sodium oxycations, carbonates, and phosphates were observed as prominent cluster ions (m/z 62, 63, 85, 113, 129, 135, 149, 157, 165, 187). The anion spectrum of Na₃HPO₄-silica (Figure 6) was dominated by O- and OH-, but also contained prominent oxalate (C₂O₄^-) bearing ions, consistent with the observation in the cation spectrum. These ions constituted the signature for Na₃HPO₄ that was sought for identification in the flow experiments.

Figure 3. Cation spectrum of wetted silica TLC paper unmodified.

Figure 4. Anion spectrum of wetted silica TLC paper unmodified.

Figure 5. Cation spectrum of tribasic sodium phosphate on unwetted silica.

Figure 6. Anion spectrum of tribasic sodium phosphate on unwetted silica.

Temporal Flow Profiles

The rate at which the solvent front develops was influenced by the presence of salt in the matrix. When temporal flow profiles were compared for cellulose, Ca(NO₃)₂ spotted cellulose, and Na₃HPO₄ spotted cellulose (Figure 7), the evolution of the solvent front was retarded when the solvent encountered the salt spots. After that, the rates of solvent front evolution were nearly identical.

Figure 7. Temporal flow profile of water moving through an unmodified cellulose TLC plate, a plate spotted with Ca(NO3)2.

A similar experiment was performed in which temporal flow profiles were compared for silica, Ca(NO₃)₂ spotted silica, and Na₃HPO₄ spotted silica. As in the case of cellulose, the rate of solvent front evolution slowed significantly when the salt spot was encountered (Figure 8). However, the rate of flow through the silica remained slower throughout the course of the experiment (Figure 9).

Figure 7. Temporal flow profile of water moving through an unmodified cellulose TLC plate, a plate spotted with Ca(NO3)2.	Figure 8. Temporal flow profile of water moving through an unmodified silica TLC plate, a plate spotted with Ca(NO3)2, and a plate spotted with Na3PO4.

Alteration of the flow rate by the presence of the salts was surprising. The salts no doubt result in increased ionic strength, which may enhance interactions between the adsorbent stationary phase and the mobile phases. Altered pH may also play a role: the basic Na₃HPO₄ was retarded more than the neutral Ca(NO₃)₂.

The temporal evolution of the solvent front was also compared for silica chromatograms spotted with three different sodium phosphates. The quantity and molarity was held constant in all three experiments. It was found that the extent of retardation decreased in the order Na₃HPO₄ > Na₂HPO₄ > NaH₂PO₄. The solvent was impeded immediately following the point at which it encountered the salt. Beyond this point, rates of solvent front evolution were very similar.

Figure 9. Temporal flow profile of water moving through silica TLC plates spotted with NaH2PO4 (monobasic), Na2HPO4 (dibasic), and Na3HPO4 (tribasic).

Spatial Ion Profiles

Figures 10, 11, and 12 track ions of interest as they moved up the chromatographic paper. Based on the benchmark mass spectra that were collected for the salts of interest, we selected Na⁺, PO₂^-, and PO₃^- for spatial tracking. Both PO₃^- and PO₂^- were considered to be signatures for phosphate (intact PO4 species were in general not observed). Because of run-to-run changes in the morphology of the SIMS spectrum, absolute ion abundances can change significantly; however, this problem was overcome by normalizing the abundance of the ion of interest to the abundance of an ion that is consistently produced from the matrix. Hence, Na⁺ was normalized to Si+ and both PO₂^-, and PO₃^- were normalized to SiO₂^-. These normalized abundances were plotted versus position on the developed chromatogram.

Polymodal behavior was observed for Na₃HPO₄ (Figure 10). The phosphate, which produced PO₂^-, and PO₃^- , was observed to be distributed between 5 and 12 cm. There are two local maxima in the PO₃^- plot, a broad feature between 5 and 8 cm, and a second smaller feature maximizing at ~11 cm. PO₂^- displays similar behavior. The Na⁺ ion displays three local maxima, a very large signal at 2 cm, a broad, lower abundance signal at 5-8 cm, and a smaller signal at 11 cm. The results suggest that a monosodium phosphate solute is traveling with the solvent front, having an RF value of close to 1.0. A higher Na⁺/ PO₃^- ratio was observed at an RF ≈ 0.58, which suggests the presence of a disodium phosphate species. The very abundant peak at RF ≈ 0.17 indicates that most of the sodium moved very little, and suggested that the Na⁺ cation was exchanging with surface hydroxyls on the silica surface.

Figure 10. Spatial ion profiles of normalized Na+, PO2-, and PO3- transported from a Na3HPO4 spot through a silica matrix with water as the mobile phase.

The dibasic salt Na₂HPO₄ displayed peaks (Figure 11) in the profiles of PO₃^- and PO₂^- at Rf ~ 0.33 (4 cm). The transport of the moderately ionic NaHPO₄^-, and partitioning of this ion among other species may show a slightly elevated PO₃^- and PO₂^- values observed at higher Rf values. Curiously, Na⁺ abundance by SIMS analysis did not clearly track the abundance of the mobile phosphate-derived ions, which suggests competition for the cation between the silica stationary phase and HPO₄^2-. These findings would compel a careful examination of additional benchmark samples that would compare SIMS behavior of the different sodium phosphates.

Figure 11. Spatial ion profiles of normalized Na+, PO2-, and PO3- transported from a Na2HPO4 spot through a silica matrix with water as the mobile phase.

SIMS spectra of the monobasic salt NaH₂PO₄taken at 3 cm and 6 cm showed abundant peaks at m/z 63 and m/z 79 (Rf = 0.25-0.5) representing PO₂^- and PO₃^-ions, respectively. However, the abundance of these peaks was barely above background. The monobasic salts do not easily give up Na⁺ during SIMS analysis and are on average more mobile than the less basic salts.

Figure 12. Spatial ion profiles of normalized Na+, PO2-, and PO3- transported from a NaH2PO4 spot through a silica matrix with water as the mobile phase.

Detection and Sensitivity

The last set of results evaluated the detection limit for the sodium ion and the sensitivity of the approach. Figure 13 shows the normalized counts/second for different intensities of the Na₃HPO₄ solution on silica from the TLC plate. The TQ-SIMS could detect increased Na⁺ at an exposure concentration of 0.038 mM before the peaks became greater than the background. At about 2.5 M exposure concentration, the best signal-to-noise was detected; increasing the concentration beyond this value did not improve the quality of the peaks significantly.

Figure 13. TQ-SIMS detection limits for normalized Na+.

Na₃HPO₄ produced a diagnostic SIMS fingerprint; however, it is difficult to distinguish this when a solution is applied to silica. It may be that the salt partitions into anionic (conjugate base) forms, which are not detectable in the positive ion mode. The Na⁺, PO₃^-, and PO₂^- ions are the best candidates for diagnostic ions for this experiment. Examinations of the spatial ion profiles would compel a rigorous comparison of the SIMS spectra of the three sodium phosphate salts. The observation of carbonate and oxalate cations in the spectrum of silica doped with Na₃HPO₄ at first may seem unusual; however, we have previously shown that basic Na⁺-bearing surfaces fix Co₂, resulting in production of carbonates and oxalates on surfaces (Shaw et al. 2003).

Despite some difficulty interpreting SIMS data, the approach was attractive because of the inherent simplicity in analyzing the results. Utilization of silica plates containing a fluorescent indicator do not always respond to inorganic solutes, and would not be specific for individual species. Extracting the anions and cations from the silica matrix represents another alternative detection approach; however, this was a more labor-intensive approach and also introduces that possibility of inefficient extraction and re-equilibration of the phosphate species, thus skewing results.

Temporal Flow Profiles

The presence of salts on the silica matrix temporarily impedes the evolution of the solvent front. Evidently, encounters with ionic contaminants interfere with the development of the solvent front. The origin of this phenomenon is not known; however, it appears that the solvent requires time to dissolve the salt, and cannot advance forward during this process. More highly ionic solutions clearly retard flow of the solution through the matrix, which would suggest aggressive interactions between the more strongly ionic salts and the adsorbent stationary phase. For example, the solvent front moved most slowly when Na₃HPO₄ was present in the matrix and fastest when NaH₂PO₄was present. This idea is consistent with the notion that ionic attraction is responsible for adsorption of solutes. Since beyond the point at which the solvent encountered the salt, the rates of solvent front evolution were similar, this suggests that all three compounds had equilibrated to the same extent in the pore water of the silica. The slower rate of evolution in the cellulose as compared to the silica may be a consequence of migration of the salt, which may not be occurring to the same extent on the cellulose. The slowing of the solvent front upon contact with the salt spot could be an advantageous phenomenon, since this would allow time for interaction of other solutes in the solvent front to react with the salt, providing opportunity for barrier formation.

Spatial Ion Profiles

Phosphates partition between different species, which have differing extents of ionic character, and hence have different mobilities. Data are consistent with fast mobility for species like H2PO4- and NaHPO₄^-, and slower mobility for the more ionic Na₂PO₄-.

Na₃HPO₄ dissociates to Na⁺ and Na₂PO₄- upon solvent intrusion. Na⁺ is largely retained and results in a low Rf in the ion TLC. Ions having low Na/P ratios would be expected to have a lower ability to form dianionic species, and would thus be less well adsorbed, and might move with the solvent front. This would explain the small fraction showing up at 11 cm (Rf = 0.9). More highly ionic species such as Na₂PO₄- might experience slower transport behavior, and show up over a range of Rf values (0.42 to 0.66, 5-8 cm). This explanation is completely consistent with the observation that the rate of migration of the monobasic NaH₂PO₄solution was the fastest recorded amount the salts examined (see Temporal Flow Profiles). Dissociation of the starting salt Na₃HPO₄ would leave a strongly sorbed Na⁺ behind (low Rf = 0.17). The varied distributions of Na⁺ and the phosphate-derived anions are consistent with multiple phosphate species having variable mobility.

Detection and Sensitivity

SIMS was determined to be an effective technique in detecting the presence of the key ions such as Na⁺, PO₃^-, and PO₂^-. Sodium ions in particular are readily detected even at a lower concentration. The ease of sample preparation maintaining the integrity of the sample was also a distinct advantage.

Path Forward

The next steps in this investigation are already underway. TLC experiments are going to be run using combinations of Ca(NO₃)₂ and sodium phosphates, to determine whether a calcium phosphate barrier can be formed. TLC of uranium solutions will reveal interactions with barriers. This will provide more support to the potential development of a precipitated chemical barrier. A trial soil matrix applied to a TLC plate has been created and will be tested in future investigations. The problem of contaminant migration is complicated because of the diversity of the subsurface and the structure and behavior of the aquifer. Extensive background knowledge is a necessary component to understanding mobility and determining the feasibility of a chemical barrier. Our investigation is part of this necessary background leading to more complicated investigations in contaminant mobility.

It is worthwhile noting that the formation of a calcium phosphate barrier using the proposed approach could have unintended outcomes. One of these would be that some of the Ca²⁺ and phosphate species would not precipitate with each other, and might reach the groundwater. Addition of Ca²⁺ would likely be inconsequential in Idaho groundwater because of the prevalence of this element in the local geosphere. In the bench-scale experiments, Ca²⁺ was added as a nitrate salt, because of the high solubility, and low retention of nitrate by mineral surfaces. At field scale, the nitrate could travel all the way through the vadose zone to the groundwater (200 - ~700 ft), where it could serve as a nutrient for bacteria and algae. This possibility suggests that a less reactive anion such as Cl- be substituted for nitrate. Addition of phosphate to the ground water could also be deleterious, for the same reasons as those mentioned for nitrate; however, it is unlikely that significant quantities of phosphate would make it to the aquifer, because phosphate salts formed with indigenous Mg²⁺, Ca²⁺, and Fe³⁺ would be expected to be highly insoluble.

If a barrier were formed, interactions with radionuclides (both actinides and fission products) would be subjects of acute interest. The uranyl dication UO₂₂⁺ may well behave like Ca²⁺, as would ⁹⁰Sr²⁺: stable phosphate precipitates would be expected. Furthermore, apatite (calcium phosphate) phases would also be expected to have strong adsorptive properties for a number of other radionuclides of concern, such as ¹³⁷Cs⁺, ²⁴¹Am³⁺, NpO²⁺, and perhaps for a variety of Pu species. Yet, how these would compete with naturally occurring cations for adsorptive sites on the barrier is unknown. Similarly, the long-term stability of the barrier is not known; the minerals should be relatively stable at the modestly basic pH values typical of local groundwater, however an acidic excursion could cause substantial dissolution. Furthermore, the influence of radiation damage on these minerals is largely unknown. These concerns notwithstanding, the approach has many attractive attributes including simplicity and the potential for long-term stability, and hence merits further study.