Development of TRU Waste Mobile Analysis Methods for RCRA-Regulated Metals

This is the final report of a one-year, Laboratory D&ted Research and Development (LDRD) project at Los Alamos National Laboratory (LANL). Glow-discharge mass spectrometry (GD-MS), laser-induced breakdown spectroscopy (LIBS), dc-arc atomicemission spectroscopy (DC-AFXAES), laser-ablation inductively-coupled-plasma mass spectrometry (LAICP-MS), and energy-dispersive x-ray fluorescence (EDXRF) were identified as potential solid-sample analytical techniques for mobile characterization of TRU waste. Each technology developer was provided with surrogate TRU waste samples in order to develop an analytical method. Following successful development of the analytical method, five performance evaluation samples were distributed to each of the researchers in a blind round-robin format. Results of the round robin were compared to known values and Transuranic Waste Characterization Program (TWCP) data quality objectives. Only two techniques, DC-ARC-AES and EDXRF, were able to complete the entire project. Methods development for GD-MS and LA-ICP-MS was halted due to the stand-down at the CMR facility. Results of the round-robin analysis are given for the EDXRF and DCARCAES techniques. While DC-ARC-AES met several of the data quality objectives, the performance of the EDXRF technique by far surpassed the DC-ARC-AEiS technique. EDXRF is a simple, rugged, field portable instrument that appears to hold great promise for mobile characterization of TRU waste. The performance of this technique needs to be tested on real TRU samples in order to assess interferences from actinide constituents. In addition, mercury and beryllium analysis will require another analytical technique because the EDXRF method failed to meet the TWCP data quality objectives. Mercury analysis is easily accomplished on solid samples by cold vapor atomic fluorescence (CVAFS). Beryllium can be analyzed by any of a variety of emission techniques. Background and Research Objectives Transuranic (TRU) waste characterization is imminent and a critical compliance activity required for all TRU waste destined for treatment and/or disposal. There is an immediate need for mobile laboratory analytical methods to characterize TRU waste for *Principal Investigator, e-mail: cmahan@lanl.gov


DISCLAIMER
Portions of this document may be illegible in electronic image products.Images are produced from the best available original document.
ICP-MS), and energy-dispersive x-ray fluorescence (EDXRF) were identified as potential solid-sample analytical techniques for mobile characterization of TRU waste.Each technology developer was provided with surrogate TRU waste samples in order to develop an analytical method.Following successful development of the analytical method, five performance evaluation samples were distributed to each of the researchers in a blind round-robin format.Results of the round robin were compared to known values and Transuranic Waste Characterization Program (TWCP)  data quality objectives.Only two techniques, DC-ARC-AES and EDXRF, were able to complete the entire project.Methods development for GD-MS and LA-ICP-MS was halted due to the stand-down at the CMR facility.Results of the round-robin analysis are given for the EDXRF and DCARC-AES techniques.While DC-ARC-AES met several of the data quality objectives, the performance of the EDXRF technique by far surpassed the DC-ARC-AEiS technique.EDXRF is a simple, rugged, field portable instrument that appears to hold great promise for mobile characterization of TRU waste.The performance of this technique needs to be tested on real TRU samples in order to assess interferences from actinide constituents.In addition, mercury and beryllium analysis will require another analytical technique because the EDXRF method failed to meet the TWCP data quality objectives.Mercury analysis is easily accomplished on solid samples by cold vapor atomic fluorescence (CVAFS).Beryllium can be analyzed by any of a variety of emission techniques.

Background and Research Objectives
Transuranic (TRU) waste characterization is imminent and a critical compliance activity required for all TRU waste destined for treatment and/or disposal.There is an immediate need for mobile laboratory analytical methods to characterize TRU waste for *Principal Investigator, e-mail: cmahan@lanl.gov 1 RCRA-listed metals (Le., Sb, As, Ba, Be, Cd, Cr, Hg, Pb, Ni, Se, Ag, T1, V, and Zn).Current methods of analysis used in the TRU Waste Characterization Program (TWCP) are laboratory-based, costly, have low sample throughput, and require considerable analytical resources to implement the associated Quality Assurance (QA) program.In addition, these methods are not readily amenable for mobile deployment.
Glow-discharge mass spectrometry (GD-MS), laser-induced breakdown spectroscopy (LIBS), dc-arc atomic-emission spectroscopy (DC-ARC-AES) using solidstate integrating detectors, laser-ablation inductively-coupled-plasma mass spectrometry (LA-ICP-MS), and energy-dispersive x-ray fluorescence ( E D m are all direct chemical analysis techniques and were targeted by this research for evaluation as potential analytical methodologies for TRU waste characterization.All these technologies are suitable for mobile deployment.Furthermore, a major benefit of most of the technologies is the potential to analyze the entire suite of target analytes.This would reduce the current protocol requiring three separate analytical methods to a single instrumental technique.Hence, sample throughput is increased and analytical costs are lowered.The objective of this research was to identify and demonstrate the existence of rugged, field-laboratory-based analytical techniques that can meet the TWCP quality assurance objectives and provide full TRU waste analyses at reduced cost.

Importance to LANL's Science and Technology Base and National R&D Needs
Waste Isolation Pilot Plant (WIPP) site.This waste is located at various DOE sites throughout the nation and, generally speaking, the required analytical capabilities for characterization do not exist at the majority of storage sites.Thus, the waste will require transport to a facility with the requisite capabilities.Transportation of TRU waste is tightly regulated, difficult, and costly.By developing mobile analpcal laboratories, these problems are obviated.In addition, if a single technique can replace the current suite of laboratory-based instrumental techniques (ICP-MS, ICP-AES, CVAFS), then approximately $150K can be saved for every 40 drums analyzed.The estimated savings extrapolated to include the anticipated analysis of TRU waste stored throughout the DOE complex could result in a cost reduction of nearly one hundred million dollars.DOE plans to dispose of approximately 175,600 cubic meters of TRU waste at the 2

Scientific Approach and Accomplishments
Each technology developer was provided with surrogate Portland-cernented waste samples to use in the analytical method development process.These surrogates, obtained from Idaho National Engineering and Environmental Laboratory (INEEL), are actual sampIes used in the Performance Demonstration Program (PDP).This program is designed to evaluate the participating laboratories' analytical performance for TRU waste characterization.The analytical results are subjected to an evaluation and the laboratory scored.The results of the scoring determine whether &e laboratory qualifies its TRU analytical characterization methods.preparation procedures, instrument parameter optimization (e.g., analyte wavelength selection and laser power settings) and determination of analytical figures of merit.Following successful development of the analytical method, another set of performance evaluation samples was distributed to each of the researchers in a blind round-robin format.Thus the results of this test also indicate whether the technology meets the performance specification on real evaluation samples.Results of the round robin were compared to known values and TWCP data quality objectives.The data quality objectives are discussed below.Final results of the study, plus estimates of sample throughput, ease of use, and cost were evaluated for the technologies that were able to complete the project goals.
are described below.The method development phase for the techniques LA-ICP-MS and GD-ICP-MS were progressing on schedule but were abruptly halted due to the standdown in the Chemistry and Materials Research (CMR) facility.
The method development phase of this study involved the development of sample DC-ARC-AES and EDXEW were able to complete the project goals.The results

Quality Assurance Program for TRU Waste Disposal
The Quality Assurance Program Plan (QAPP) for the TWCP describes the performance requirements for TRU waste characterization.The data quality objectives (DQOs) are shown in Table 1.Analytical results for PDP samples with concentrations greater than the Program Required Quantitation Limit (PRQL) must result in precisions of less than or equal to 30% and recoveries must be within the range 80-120%.In addition, Table 1 shows the Program Required Detection Limits (PRDL) for each of the target analytes.PDP sample analysis results are scored on these major criteria to determine whether the analytical technique passes or fails, and the laboratory demonstratives effective performance.Additional scoring criteria are made on the "blank" surrogate results.If the laboratory identifies the presence of a target analyte above 50% of the PRQL, then points are subtracted from the overall score.It should also be noted that there are several run-time quality controI (QC) criteria that also must be satisfied for this program.The criteria generally follow SW846 requirements, but are not evaluated in this project., 1993).Our recent work with EDXRF for soil characterization indicated that it was relatively accurate and sensitive for most RCRA metals, with typical biases of less than _+lo% and detection limits as low as a few ppm (Goldstein et al., 1996).Hence, a major objective was to develop and evaluate an analogous EDXRF method for analysis of Portland-cement waste samples over a range of elements and concentrations suitable for RCRA metal analysis.

Table 1. Data
Background: EDXRF is one of several methods of direct solid analysis that can Method: The methodology for sample preparation and analysis in EDXRF is relatively straightforward, largely following methods previously developed for soil analysis (Watson et al., 1989;Goldstein et al., 1996).Samples are dried under a heat lamp overnight, mixed and milled in a Spex ball-& for 5 minutes, sieved to 4 0 0 micron size, and then -0.5 g placed as a powder in a microcell for analysis.This simple method of sample preparation takes <20 minutes per sample, but provides a physically homogeneous sample needed for reproducible XRF analysis.It is also simple and fast enough to be amenable to completion in a mobile analytical laboratory.Sample size is reduced by a factor of 10-20 relative to conventional powder X R F through use of a sample microcell and beam collimator.
All analyses were performed using a commercial EDXRF spectrometer, a Spectrace 5000.This instrument has an x-ray tube source with variable source current and voltage up to 1 mA and 50 kV, which permits optimization of excitation conditions for the element of interest.It also uses a high-resolution, electrically cooled Si(Li) detector, which permits simultaneous collection of x-rays of variable energy with minimal spectral interference.The instrument has a fundamental-parameter data-reduction package available for rapid multi-element standardization and quanMication of the acquired spectra.
A nearly identical field-transportable instrument is also available, and so p e r f o m c e of this method should be similar under field conditions.
Analytical conditions were optimized to increase method sensitivity by adjusting the excitation x-ray tube voltage and current.Filters of various thickness and composition were also used to remove low-energy noise from the acquired spectra.Based on tests with standards, four separate excitation conditions were used 1) lowest -Z elements Ca, V, Cr using 12-kV tube voltage and 0.13-mm-thick Al filter, 2) Fe, Ni, Cu, and Zn using 20-kV tube voltage and 0.05-mm-thick Pd filter, 3) As, Se, Hg, T1, Pb, U, and Th using 35-kV tube voltage and 0.127-mm-thick Pd filter, and 4) Ag, Cd, Sb, and Ba using 50-kV tube voltage and 0.63-mm-thick Cu filter.Acquisition time (livetime) was 800 s per condition, hence analysis takes approximately 90 minutes per sample.This long acquisition time was needed to achieve detection limits at the desired levels utilizing the relatively small sample configuration of the microcell geometry.Samples were analyzed utilizing an auto-sampling turret in automated mode overnight, so throughput is presently -10-15 samples/day.Spectral interferences are generally absent under the conditions above.
Standardization of the major elements was accomplished by cement standards spanning a range in composition obtained from the National Institute of Standards and Technology (NIST).A dual approach was used to standardize the trace metals.Trace 5 element standards containing the metals of interest were obtained from prior TRU intercomparison studies.Because these standards spanned a limited range in concentration, additional standards were prepared at LANL by pipetting NIST-traceable multi-element solution standards onto a "blank" cement matrix, followed by drying and homogenization.A total of 6 multi-element standards typically ranging from 10-2000 ppm were used for standardization.
Results: Detection limits for the conditions above are compared to the TRU waste program required detection limits in Table 2.The EDXRF technique meets detection limit requirements for 11 of the 14 metals.Exceptions are V, Hg, and Be, the latter of which is not detected by EDXRF.The detection limit for vanadium is only a factor of two above the required limit, and so it is likely that the required limit could be met by increasing data acquisition time by a factor of 4 or by increasing sample size.However, the required detection limit for mercury is a factor of 40 too low and apparently unattainable by direct EDXRF techniques.Additional field-based techniques, which can more sensitively measure both mercury and beryllium, are required for waste characterization of RCRA metals.Results of the blind round-robin analysis are shown in Tables 3 and 4. Cells that are shaded show results that would have failed the TWCP criteria, and therefore reduce the analytical scoring, if the results would have been submitted to INEEL under the PDP cycle 4 round-robin test.Beryllium cannot be detected by ED-.
The majority of results 6 meet the TWCP data quality objectives for percent recovery and relative percent difference (RPD).Only chromium and mercury deviated slightly from the required recovery for the TRU4 and TRUS blind samples, with recoveries of 78% and 79%, respectively.While V recoveries for samples TRU2,4 and 5 were low, the expected values were less than the PRQL (100 ppm), therefore the low recovery does not reduce the analytical score.TRU3 was the performance demonstration 'blank' sample, that is the sample is comprised of just the Portland cement matrix and is not spiked with any of the RCRA elements.Two false positives were found for Ba and Cr.However, it should be noted that a T-test is applied by the scorer using all submitted analysis from the participating laboratories and often times the laboratories indicate the presence of contamination.

TRU4/TRUS
Relative standardization uncertainties (%RSD), based on agreement of the six trace-element standards, are given in Table 5.The uncertainty due to standardization is 10% or less for 10 of the 13 detected metals.Exceptions are Hg, Ag, and Ba, which have significantly poorer precision.The higher standardization uncertainties for barium and silver seem to reflect some small bias in the expected values for either the LANL or externally prepared trace-element standards for these elements.It should be noted that when comparing element concentrations between national laboratory PDP results, there is often a higher correlation between the laboratories than to the PDP program 'expected' values.This has been shown in several instances.The high uncertainty for mercury reflects its low concentration, ranging from 7 to 54 ppm in the standards measured.
External precision and accuracy are evaluated from results for three blind quality control samples, given in Table 5.The typical external precision is based on reproducibility of analyses of three replicate aliquots of three different QC samples.For all elements with the exception of Hg, this relative uncertainty is less than 7%.Again, mercury has poorer reproducibility due to its lower concentration in the analyzed samples.Based on these results for samples above the quantitation level, a typical external precision of a few percent can be obtained with this EDXRF method.

Conclusions:
These results indicate that field-based EDXRF techniques can meet requirements for analyses of a large majority of the RCRA-metals in TRU waste samples.While EDXRF appears to be suitable for analyses of 11 of the 14 RCRAmetals, a few metals including mercury, vanadium, and beryllium have detection limits that are significantly above the program requirements.As a result, additional techniques should be investigated for characterization of those metals.In addition, effects of inherent radiation of the samples and matrix variability on method performance also need to be investigated, although these effects are expected to be relatively minor in most cases.

DC-A RC-A ES Project
soIid-state integrating detector to measure the spectral emission intensities produced when a sample is vaporized and excited by a dc arc.The solid-state integrating detector represents relatively new technology, which has the potential to improve analytical performance over the traditional dc arc and conventional spectroscopic techniques.Samples are pulverized, mixed with graphite powder, and burned in the lower of two vertically mounted graphite electrodes.The detector chip is similar to a photographic plate 9 Background: DC-ARC-AES is a bulk-solids analytical technique that uses a ~ Analytlcal Gap: 4mm Excitation: 15 amps Gas Flow Rate: 5 LPM in that it provides for continuous wavelength coverage and hence most elements in the periodic table can be determined if present in sufficient quantity.Potential analytical benefits over conventional spectroscopic methods include full elemental fingerprinting of the sample, the ability to detect weak spectral lines in the midst of strong matrix signals, improved sample throughput, simultaneous background correction, minimal sample preparation, and instrumental ruggedness.
Method: Method development was a multivariate process involving optimization of the graphite-diluent weight to sample weight, wavelength selection, element integration times, calibration strategies, electrode gap distance, gas flow rates, and applied current.Portland cement surrogate samples with known concentrations of RCRA elements were used to investigate and optimize these parameters.Analysis parameters are shown in Tables 6 and 7. Round-robin samples were diluted with graphite powder in a 1-to-5 ratio.Calibration standards were prepared using graphite powdered SPEX G7 standards diluted with a PortIand cement 'blank' for matrix matching purposes.The dc arc technique did not perform as well as expected on the round-robin test.Two major problems exist that may improve the results.First, background correction must be more precise.The 5-fold dilution of the matrix results in concomitant concentrations of interfering elements at very high levels.Over-or undercorrection of these interferences result in biases.Secondly, the technique has poor precision.It is believed this is due to a nonhomogenous sample with varying particle sizes.Better sampling, grinding, and sieving processes could result in improved precision and accuracies.Furthermore, an alternate method with a greater dilution of the sample with graphite powder has been tested and the preliminary results show improvements in the detection limits and recoveries for several of the elements.

LA-ICP-MS Participation
We worked on the LA-ICP-MS method for the analysis of cement samples.This involved varying different parameters such as laser wavelength, irradiance, raster speed, and shots per spot.A preliminary method has been developed based upon the results of these experiments.The parameters chosen were those that gave the best short-term signal precision.The figures of merit for the method were in the process of being determined when the laser system failed.Subsequent to restoring the instrument, the CMR building was put in stand-by mode and work could not resume.noted.The performance demonstration samples that were used for method development Issues: During the preliminary work some problems with the standards were 13 are not well characterized, and in some cases there are inconsistencies in the "known" concentrations as determined from the two methods: 1) the amount of analyte added to the cement and 2) the concentration of each analyte as determined from traditional wet chemical analyses.In some cases the calibration curves derived from the different values were considerably different.Some of these problems relate to the fact that the blank contains some of the target analytes (Ba, Cr, V, Zn), which makes determining the method detection limits problematic.
Method: The following preliminary method was devised.The sample is ground using a boron-carbide mortar and pestle.A sample pellet (1-cm diameter) is prepared from a 0.25-g sample.No binder is required.The following conditions are used for sample analysis:

a
530% when sample and duplicate conc are 2 IO X IDL for ICPAES and 2 100 x IDL for ICPMS I b Based on known concentrations c Program required detection limits I OX below the PRQL d Program required quantitation limit Energy Dispersive Xray Project potentially provide improvements in analytical performance (Le.cost, turnaround time, field-based characterization) over traditional wet chemical methods of analysis.Recent developments in field-transportable EDXRF have greatly improved the sensitivity and accuracy of this method for elemental analysis of complex matrices (e.g., Pella et al., 1986; Leyden, 1988; Bilbrey et al., 1988; de Boer, et al.

T
l were off by approximately a factor of two, and we believe can be improved by fmetuning the method.The low detection criteria of Hg preclude analysis by dc arc as well as any other emission technique.The reason for the high detection limits is the interfering 10

2
Laser conditions: third harmonic of a Nd:YAG laser, 355-nm, Q-switched Irradiances at -10 W/cm Scan across the sample using a 2 x 2-mm raster pattern of 100-pm step sizes at a rate of 2 mm per 7 sec Complete one full raster prior to starting MS acquisition After sample acquisition, argon gas is allowed to flow through the laser cell and transfer tubing until signals return to blank levels (-3 minutes).The MS acquisition time is optimized at 60 s with 3 integrations.cases, without the use of an internal standard.Detection limits are in the low-ppm range.Results:Preliminary results indicate a short-term precision of -5% in most 14 .

Table 5
also shows that average accuracy ranges from 85% for Cr to 102% for Ni.Results for vanadium are relatively poor due to its high detection limit and are not included.Given the total propagated uncertainties based on standardization and external

Table 5 ,
none of the other elements show a significant bias at the 95%

Table 5 . EDXRF Method Precision and Accuracy Element
Concentration range for quality control samples used to determine external precision and

Table 7 . Element ChanneIs Results:
Detection limits for the dc arc technique, determined from a Portlandcement-matrix blank sample, are compared to the TWCP PRDLs in Table8.Only 3 elements @a, Be and Ag) were able to meet the stated requirements.Sb, Cd, Cr, Pb, and