ANALYSIS OF TCLP EXTRACTS BY X-RAY FLUORESCENCE

Transcription

1 257 ANALYSIS OF TCLP EXTRACTS BY X-RAY FLUORESCENCE Douglas S. Kendall, Benjamin A. Burns, Jennifer A. Suggs and John J. Mackey National Enforcement Investigations Center, U.S. Environmental Protection Agency, Building 25, Denver Federal Center, Denver, CO ABSTRACT The toxicity characteristic leaching procedure (TCLP) is one test used by the U.S. Environmental Protection Agency (EPA) to determine if a waste material is hazardous; the TCLP test is widely used by the environmental testing community. The elements of concern for toxicity are arsenic, barium, cadmium, chromium, lead, mercury, selenium, and silver. As an environmental forensic laboratory, the National Enforcement Investigations Center (NEIC) of EPA often conducts TCLP testing. It is our practice to confirm TCLP test results that may indicate a violation of a regulation with an alternate technique. This project explored the use of X-ray fluorescence (XRF) as a confirmation technique for the determination of inorganic elements in TCLP extracts. The sample preparation approach investigated was the microliter droplet on filter paper technique. The sample carriers were analyzed under vacuum with a conventional wavelength dispersive XRF spectrometer. Results indicated that TCLP extract analysis by XRF using filters compares well to analysis by inductively coupled plasma-optical emission spectroscopy (ICP-OES). In real-world samples, correlations of 0.98 and 0.94 were achieved for lead and cadmium analyses by XRF and ICP- OES, supporting our intent to use XRF as a confirmation technique. INTRODUCTION XRF measurements are not performed as often as they could be in environmental chemistry laboratories. Part of this is tradition that favors ICP-OES or ICP-mass spectrometry (ICP-MS). A second factor favoring ICP is sensitivity. XRF detection limits are in the low parts per million (ppm) range; ICP techniques can achieve sensitivity in the part per billion (ppb) range and even considerably lower. Part of this ICP advantage can be lost since ICP techniques often require considerable dilution. XRF can analyze samples without dilution. A third factor supporting the use of ICP is easily-obtainable standards, which are simply acid solutions of various elements. Therefore, matrix matching is minimal in ICP analyses. Preparing or obtaining matrix-matched standards for XRF analyses of environmental samples can be challenging, but it is crucial to obtaining the best results. XRF does have advantages that should cause it to be utilized more often for environmental measurements. In fact, field portable XRF spectrometers have become widely used for surveying study sites and informing sampling operations. This paper focuses on laboratory XRF spectrometers and their possible use for environmental measurements. Sample preparation in XRF can be much easier and more straightforward than sample preparation for ICP. Analyzing

2 This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD). This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website ICDD Website -

3 258 samples directly, or with minimal preparation such as drying and grinding, is preferable to performing acid or other types of digestions, and removes issues of insolubility and recovery. Once calibrated, XRF spectrometers are remarkably stable, and since the testing is usually nondestructive, the standards can be used multiple times. This work focuses on one particular application of XRF to environmental measurements, the analysis of TCLP extracts, but there are many other possibilities. THE TOXICITY CHARACTERISTIC LEACHING PROCEDURE Wastes are hazardous if they appear on one of several EPA lists or if they exhibit any one of four characteristics: ignitability, reactivity, corrosivity and toxicity. The most widely used test for toxicity is the TCLP. For this test, samples are separated into liquids and solids. After filtration, the liquids are analyzed as is. The solids are extracted with a twentyfold weight of extraction fluid that is 0.1 molar in acetic acid/acetate and at either ph 2.73 or 4.93 depending on buffer capacity. After filtration, the extract is analyzed for 40 analytes, seven of which are elements well suited for XRF determination. The elements of concern for toxicity are arsenic, barium, cadmium, chromium, lead, mercury, selenium, and silver. Except for mercury, which has the lowest regulatory limit and is best determined using specialized instruments, the elements of interest in TCLP extracts can be determined by XRF. Regulatory limits for these elements vary. The regulatory limit is 0.2 milligrams per liter (mg/l) for mercury, 1.0 mg/l for cadmium and selenium, 100 mg/l for barium, and 5.0 mg/l for the others. Except for barium, these limits are at the lower end of the range susceptible to XRF measurements. However, as we have found, the microliter droplet method makes these limits practical. Why measure these elements in a TCLP extract by XRF? The first answer is that our laboratory is an EPA enforcement laboratory and we confirm regulatory violations if at all possible. Confirmation by XRF is desirable because it is quite different and independent from ICP. Secondly, XRF has the advantage of stability and excellent precision. In a high throughput situation, XRF can be very productive. THE MICROLITER DROPLET METHOD The analysis of small amounts of sample deposited on filter material by X-ray fluorescence spectrometry has been an accepted practice for many years (Sulkowski et al., 1996). The advantages include reduced background scattering and linear calibrations. Sulkowski et al. found that with 0.3-millimeter (mm)-thick polypropylene filters, analysis could be performed from either side of the filter, and that backing the filter with potassium chloride significantly reduced the background. They analyzed solid samples deposited as fine powders, and prepared standards from ICP solution standards. A microliter droplet method was developed for the analysis of elements preconcentrated from solution by liquid liquid extraction (Igarashi et al., 2000). A liquid phase with the

4 259 preconcentrated analytes at a volume of 100 microliters (µl) was placed on filter paper and dried for XRF analysis. The calibrations were linear and the determinations precise. Measurements were done with gold films and titanium solutions deposited on Rigaku Ultracarry sample holders (Okhrimovskyy et al., 2006). Titanium in the amount of 50.8 µg was deposited as a 50-µL droplet. For the titanium deposit on the Ultracarry filters, the front-side intensity was equaled by the back-side intensity. The angular dependence of the measured intensity was observed. Using specialized equipment and techniques, it is possible to prepare deposits in nanoliter and picoliter amounts (Fittschen and Havrilla, 2010). These small deposits are used in conjunction with the small beam sizes of micro X-ray fluorescence (MXRF). Quantitative measurements can be made with nanogram or picogram amounts. METHOD For this work, we used commercially available filters designed with the filter paper attached in the center of plastic film surrounded by an outer plastic support ring (Rigaku Ultracarry). Standards or samples were deposited in amounts of 5 to 50 µl on the filter paper, then dried at 50 C for one hour. Multiple deposits with drying after each addition were performed in order to increase the amounts on the filters. Standards were prepared from commercial ICP-OES standard solutions. Both single-element and multi-element calibrations were performed and proved to be equivalent. Filters were analyzed on a Rigaku ZSX wavelength dispersive XRF spectrometer with a rhodium tube using the arsenic Kβ line, the barium Lα line, the lead Lβ line, and the Kα lines for the other four elements. With a rhodium tube, cadmium and silver must be measured with a zirconium filter. This reduces the sensitivity for the measurement of these two elements. The arsenic Kβ line must be measured in order to avoid lead interference, and this line is less intense than the Kα line. The sample cups used to hold the filters are made of stainless steel, and the chromium in the steel caused a high intercept in the chromium calibration. Nevertheless, it was possible to make good chromium measurements, as shown by the detection limit for chromium. The limit of detection is an important measure for an analytical method. A useful definition is the smallest amount of substance that can reliably be distinguished from zero. Jenkins (1999) proposed a limit of detection equation based on counting statistics: the standard deviation of a given number of counts is equal to the square root of the counts. 3 Rb/Tb In this equation, m is the slope of the calibration equation. For this work, the units are micrograms per counts per second (µg/cps). If the slope is calculated in units of cps/µg, m should be in the denominator of the equation. Rb is the count rate of the background, in cps, and Tb is the time spent counting the background. The factor of 3 is for the 95% confidence level. Table 1 shows this equation applied to typical calibration data collected for this work.

5 260 Table 1. Detection Limits for TCLP Elements Element Slope (µg/cps) Background Intensity (cps) Background Time (sec) Limit of Detection (µg) Arsenic Barium Cadmium Chromium Lead Selenium Silver One measure of sample preparation is the precision with which preparation can be performed. Five of the filters were prepared in identical ways. To each was added 5 µg of each of the elements in Table 2. This was done by applying 5 µl of a 100-µg/L multi-element solution prepared from ICP standards. The concentration of each element on each filter was determined using a multi-element calibration. With a moderate amount of care, samples can be prepared with an average relative standard deviation of about 4 %. Table 2. Sample Preparation Repeatability Results Element Relative Standard Deviation Arsenic 3.9 % Barium 3.3 % Chromium 3.7 % Lead 3.7 % Selenium 3.4 % Silver 6.0 % To observe the reproducibility of repeat measurements from a filter, ten measurements were made of a single Ultracarry filter containing 5 µg of each of the elements shown in Chart 1. Each measurement was for 14.4 minutes. The intensities of each element increased as the number of measurements increased. The average gain was about 9 %. The instrument drift during this time was about 1 %, as determined from a standard analyzed both before and after the reproducibility measurements. The observed increase in intensity may be due to sagging of the plastic film supporting the filter toward the beam.

6 261 Chart 1. Reproducibility from Ten Measurements of One Ultracarry Filter Reproducibility Results in µg Measurement Sequence As Se Ba Pb Ag Cr Cd As Se RESULTS With acceptable detection limits, repeatability, and reproducibility data to support the utility of the filter paper method, TCLP extracts from real-world samples were analyzed by XRF and compared to data previously generated by ICP-OES analysis. In the first comparison, lead was the analyte of interest. Many truckloads of lead-contaminated soil were scraped from a recycling facility and taken to a landfill. The facility was the site of lead-acid battery recycling. Battery acid containing lead and antimony was dumped on the ground. There were six sampling locations, one of which was collected in triplicate, for a total of eight soil samples. The soils were extracted in duplicate using the TCLP, and the extracts were analyzed by ICP-OES with the results shown in the chart. The extracts were prepared for XRF analysis by placing 50 µl of extract on an Ultracarry filter, then dried at 50 C. The calibration standards were prepared from a lead ICP standard. Chart 2 displays the correlation between the results from the analysis of the extracts by XRF and the analysis of the extracts by ICP-OES. A correlation coefficient of 0.98 was obtained from the comparison of the results, confirming the efficacy of the analytical method for lead in TCLP extracts.

7 262 Chart 2. Comparison of TCLP Lead Analyses by XRF and ICP-OES TCLP Lead Concentrations Lead by XRF (mg/l) y = 0.935x R² = Lead by ICP (mg/l) For the second real-world test, cadmium was the toxic element of interest. Samples of improperly treated waste were extracted using the TCLP procedure. Five of the extracts were over the regulatory limit of 1 mg/l for cadmium as determined by ICP-OES. These samples were prepared for XRF analysis by depositing 250 µl of each sample on Ultracarry filters. This addition was performed in five increments of 50 µl each, with drying between the additions. The samples were analyzed by wavelength dispersive XRF spectrometry using a rhodium tube with a zirconium filter. Cadmium was determined from the Kα peak. Chart 3 shows the comparison of the ICP and XRF results. The use of filtered radiation reduced the sensitivity for cadmium. The correlation of 0.94 indicated that XRF analysis of cadmium in TCLP extracts using the Ultracarry filter method is a viable option for our laboratory.

8 263 Chart 3. Comparison of TCLP Cadmium Analyses by XRF and ICP-OES TCLP Cadmium Concentrations Cadmium by XRF (mg/l) y = x R² = Cadmium by ICP (mg/l) CONCLUSIONS Good correlations between ICP-OES and XRF analysis of toxic elements in TCLP extracts can be achieved using the filter paper method. Advantages such as rapid and easy sample preparation and reasonable element detection limits support the use of the filter paper technique in XRF confirmation analyses. REFERENCES Fittschen, U. A. E. and Havrilla, G. J. (2010). Picoliter droplet deposition using a prototype picoliter pipette: control parameters and application in micro X-ray fluorescence, Anal. Chem. 82, Igarashi, S., Takahashi, A., Ueki, Y., and Yamaguchi, H. (2000). Homogenous liquid-liquid extraction followed by X-ray fluorescence spectrometry of a microdroplet on filter-paper for the simultaneous determination of small amounts of metals, Analyst 125, Jenkins, R. (1999). X-Ray Fluorescence Spectrometry (Wiley-Interscience, New York), 2 nd ed., pp Okhrimovskyy, A., Moriyama, T., and Tsuji, K. (2006). Investigation of experimental arrangements in X-ray fluorescence analysis using ultra-thin sample carrier, Mem. Fac. Eng., Osaka City Univ. 47, Sulkowski, M., Sulkowski, M., and Hirner, A. V. (1996). Determination of trace elements in small amounts of specimen on filter material by wavelength-dispersive X-ray spectrometry, X- ray Spec. 25,