IMPORTANT COPYRIGHT INFORMATION

Transcription

1 IMPORTANT COPYRIGHT INFORMATION The following PDF article was originally published in the Journal of the Air & Waste Management Association and is fully protected under the copyright laws of the United States of America. The author of this article alone has been granted permission to copy and distribute this PDF. Additional uses of the PDF/article by the author(s) or recipients, including posting it on a Web site, are prohibited without the express consent of the Air & Waste Management Association. If you are interested in reusing, redistributing, or posting online all or parts of the enclosed article, please contact the offices of the Journal of the Air & Waste Management Association at Phone: , ext journal@awma.org Web: You may also contact the Copyright Clearance Center for all permissions related to the Journal of the Air & Waste Management Association: Copyright 2006 Air & Waste Management Association

2 TECHNICAL PAPER ISSN J. Air & Waste Manage. Assoc. 56: Copyright 2006 Air & Waste Management Association Validation of Three New Methods for Determination of Metal Emissions Using a Modified Environmental Protection Agency Method 301 Catherine A. Yanca, Douglas C. Barth, Krag A. Petterson, Michael P. Nakanishi, and John A. Cooper Cooper Environmental Services, LLC, Portland, OR Bruce E. Johnsen SAIF Co., Portland, OR Richard H. Lambert Engineering Tech Center, Eli Lilly and Co., Indianapolis, IN Daniel G. Bivins U.S. Environmental Protection Agency, Research Triangle Park, NC ABSTRACT Three new methods applicable to the determination of hazardous metal concentrations in stationary source emissions were developed and evaluated for use in U.S. Environmental Protection Agency (EPA) compliance applications. Two of the three independent methods, a continuous emissions monitor-based method (Xact) and an X-ray-based filter method (XFM), are used to measure metal emissions. The third method involves a quantitative aerosol generator (QAG), which produces a reference aerosol used to evaluate the measurement methods. A modification of EPA Method 301 was used to validate the three methods for As, Cd, Cr, Pb, and Hg, representing three hazardous waste combustor Maximum Achievable Control Technology (MACT) metal categories (low volatile, semivolatile, and volatile). The modified procedure tested the methods using more stringent criteria than EPA Method 301; these criteria included accuracy, precision, and linearity. The aerosol generation method was evaluated in the laboratory by comparing actual with theoretical aerosol concentrations. The measurement methods were evaluated at a hazardous waste combustor (HWC) by comparing measured with reference aerosol concentrations. The QAG, Xact, and XFM met the modified Method 301 validation criteria. All three of the methods demonstrated precisions and accuracies on the order of 5%. In addition, correlation coefficients for each method were on the order of 0.99, confirming the methods linear response and high precision over a wide range of concentrations. The measurement methods should be applicable to emissions from a wide range of sources, and the reference aerosol generator should be applicable to additional analytes. EPA recently approved an alternative monitoring petition for an HWC at Eli Lilly s Tippecanoe site in Lafayette, IN, in which the Xact is used for demonstrating compliance with the HWC MACT metal emissions (low volatile, semivolatile, and volatile). The QAG reference aerosol generator was approved as a method for providing a quantitative reference aerosol, which is required for certification and continuing quality assurance of the Xact. IMPLICATIONS The two measurement methods for determining metal emissions validated in this study (the Xact and the XFM) include significant improvements in measurement technology. With these two new methods, plant operators can make more precise, accurate, and frequent measurements, leading to a better understanding of emissions. With the new reference aerosol generator validated in this study (the QAG), development and evaluation of new measurement methods are now more practical. The measurement methods are applicable to a wide range of elements and may be applicable to other sources. The aerosol generator may be applicable to both organic and inorganic analytes. INTRODUCTION Air toxic metals have been associated with a wide range of environmental and human health effects, including respiratory disorders, pulmonary disorders, and cancer. 1,2 EPA regulates the emissions of toxic metals from stationary sources under the National Emission Standards for Hazardous Air Pollutants (NESHAP) or Maximum Achievable Control Technology rule. 3 Currently, emissions are estimated based on control efficiencies established during performance tests and estimated feed rates during normal operations. When actual emissions are measured, EPA Reference Method 29 (Method 29) is used. 4 However, Method 29 is based on 30-yr-old Volume 56 December 2006 Journal of the Air & Waste Management Association 1733

3 impinger technology with accuracies and precisions on the order of 15%. 5 7 The method also involves hazardous chemicals, and it requires long turnaround times. It is clear that there are opportunities for improvement in the current monitoring technology. There have been attempts over the past decade to improve metals measurement technology by developing multimetal continuous emissions monitoring systems (CEMS) None of the proposed technologies have been approved for compliance demonstration other than those discussed in this paper. The new technology presented here includes three new methods applicable to the determination of hazardous metal concentrations in emissions: a multimetal CEMS-based method (Xact), a sampling and analysis measurement method (X-ray-based filter method [XFM]), and a National Institute for Standards and Technology (NIST)- traceable reference aerosol generating method (QAG). The methods and the instruments used to perform the methods were developed to provide more precise, accurate, and frequent measurement capabilities than are offered by Method 29. The Xact method uses a CEMS based on X-ray fluorescence (XRF) analysis of aerosol deposits (particulate and vapor phase) on reactive filter tape. 11,12 XFM is comparable to Method 29 but uses reactive filters (solid sorbants) 11,12 instead of impingers to trap both particulate and vapor phase metals, followed by nondestructive XRF analysis of the filters. The quantitative aerosol generator (QAG) is based on nebulization of a metal-containing solution and produces a reference aerosol that can be used to validate measurement methods, such as the Xact and XFM. To evaluate the performance of the new methods, a test plan 13 based on the requirements of EPA Method was developed. Method 301 is a defined procedure for validating new pollutant measurement methods that could potentially be used to demonstrate compliance with permits or other EPA applications. A more stringent modification of this procedure was used to validate the three methods based on accuracy, precision, and linearity criteria. This paper presents: (1) an overview of the three methods, (2) the procedure used for validating these methods, and (3) the results of the validation tests. Details about these instruments and methods, performance specifications, and standard operating procedures are presented elsewhere EXPERIMENTAL WORK Quantitative Reference Aerosol Generation The method for generating a quantitative reference aerosol involves the operation of the QAG instrument. 15,21 Using the QAG, an aerosol of analyte solution droplets is created using a nebulizer. 21,22 The droplets are then passed into a settling chamber where the large droplets are removed and liquid and vapor phases are equilibrated. Droplets 30 m in diameter are allowed to pass into a drying chamber where they are dried before being emitted. The aerosol exiting the drying chamber consists of the transport gas, water and metal vapor, and dried salt particles. The aerosol emission concentration (C a ) is calculated from the analyte concentration in the nebulized solution (C s ), the solution emission rate (R m ) corrected for vapor loss (R v ), and the total volume of nebulizing, drying, and dilution gas used to create the aerosol (F t ). C a C s F t R m R v (1) Quantitation of the emitted aerosol is made possible by: (1) using a large reservoir of NIST-traceable analyte solution to minimize the impact of evaporation on solution concentration, (2) continuously measuring the solution emission rate (R m ) with a NIST-traceable balance, (3) precise control of nebulizer conditions, and (4) stable control of the maximum droplet size introduced into the generator s drying chamber such that only aerosols with high transport efficiencies are generated. The reference aerosol produced by the QAG is process NIST traceable and can be used in addition to its other applications to challenge and evaluate the accuracy, precision, and linearity of the measurement methods. It is applicable to both metal and nonmetal species. Metal Concentration Measurement Xact. The Xact method of metals measurement uses a CEMS, called the Xact CEMS, that collects and analyzes a sample of emissions using filter-based technology and nondestructive XRF. 15,23 During sample collection, the Xact CEMS directs a continuous flow of stack gas through a stilling chamber where its velocity is reduced (i.e., stilled). A continuous subsample is taken from the stilling chamber and drawn through chemically-reactive filter tape, which traps both particulate and vaporphase metals. 11,12,23 The volume of the sample flow is measured, and the filter tape deposits are advanced to a position where they are analyzed for metal mass using XRF analytical procedures similar to those listed in EPA s Compendium Method IO The Xact CEMS then divides the XRF-determined mass per sample by the volume per sample and simultaneously reports concentrations for 20 analyte metals every 15 min. During analysis of the current sample, the next sample is automatically collected. In the Xact CEMS, sampling and analysis is performed continuously in the same instrument except for the time required to advance the tape ( 3 sec) and the time required for automated quality assurance checks ( 1 hr, once per day). Detection limits for the Xact CEMS are listed in Table 1 and range from 0.1 g/dry standard cubic meter (dscm) to 6 g/dscm with 15-min sampling times. These limits can be lowered significantly by varying sampling and analysis conditions. Both the XFM and Xact methods are generally applicable to elements with atomic numbers ranging from 13 (Al) to 92 (U). XFM. Like the Xact, the XFM uses filter-based technology to collect a representative sample of emissions followed by nondestructive XRF analysis of the sample. 15,23 For sample collection, the XFM sampling train connects to a stilling chamber, such as the one used for the Xact CEMS. A representative sample of stack gas is continuously directed into the stilling chamber (at L/min), and 1734 Journal of the Air & Waste Management Association Volume 56 December 2006

4 Table 1. Xact and XFM detection limits. Xact a ( g/dscm) XFM b ( g/dscm) Cr As Cd Hg Pb Mn Co Ni Se Ag Sb Cu Fe Zn Br Sr Tl Notes: a 95% Confidence, interference free, 15-min sample. b 95% confidence, interference free, 30-min sample. the XFM pulls a subsample of 1 L/min into its sampling train. The subsample is diluted and cooled before being drawn through a stacked filter cassette containing a polytetrafluoroethylene filter followed by a chemically reactive filter. 11,12,23 Particulate-phase metals are trapped on the nonreactive polytetrafluoroethylene filter, whereas vapor-phase metals, including Hg, are collected on the chemically reactive filter or solid sorbent. After sample collection, the filter cassettes are removed from the sampling train, and the resulting filter deposits are analyzed for metal mass using nondestructive XRF procedures similar to those listed in EPA s Compendium Method IO The filters do not require handling during any part of the XFM procedure, and they can be readily shipped to a laboratory for analysis or analyzed in the field. The emissions concentration is calculated by dividing the XRF-determined metal mass by the volume of emissions that passed through the filters. Typical sample times are 30 min, and analysis turnaround times can be on the order of a few hours with a field-portable XRF analyzer. Detection limits typically range from 0.06 g/dscm to 1.6 g/dscm with 30-min sample times (Table 1). Substantially lower detection limits can be attained by varying sampling and analysis conditions. Method Independence The QAG-generated aerosol concentration is independent of the aerosol concentrations measured by the Xact and XFM. Method independence is confirmed using NIST traceability and additional procedures. The concentration of the aerosol produced by the QAG is procedurally traceable to NIST through nebulized solution standards and QAG operating parameters. The XFM- and Xact-measured concentrations are both analytically NIST traceable through thin film standards and NIST Standard Reference Materials 1832 and In addition to the independent Figure 1. Experimental arrangement for laboratory and field tests. NIST traceability of the generation and measurement methods, inductively coupled plasma (ICP) analysis conducted at an independent laboratory (Chester LabNet, Tigard, OR) confirmed the standard solution concentrations used in the QAG and the filter deposit measurements made by the XFM method. Validation Methodology The procedures used to validate the QAG, XFM, and Xact were based on EPA Method 301. This EPA method is required when new methods of measuring analytes are to be used on a source in lieu of the standard reference method. Method 301 establishes accuracy and precision capabilities of a new method by comparing its determined concentrations with those of an accepted reference method. The resulting bias between concentrations from the two methods is used as an indicator of accuracy. Simultaneous method measurements are used to determine precision. Following normal Method 301 procedures, the Xact and XFM measurement methods would be tested against Reference Method 29. However, the precision and accuracy of Method 29, which are on the order of 15%, 5 7 were insufficient for verifying the precisions and accuracies believed to be attainable with the Xact and the XFM (i.e., potential precision 5% and accuracy 5%). The validation procedure presented in this section is a modified Method 301 in which Method 29 comparisons were replaced with QAG reference aerosol comparisons. 13 This modified method was more stringent than Method 301, because it included more replicate runs, sequential measurements instead of simultaneous measurements, additional validation criteria (e.g., linearity), and independent NIST traceability of results. The first step in the modified Method 301 procedure required laboratory validation of the QAG emissions, and the second step was a field validation of the XFM and Xact measurements. The general experimental arrangement used for both laboratory and field tests is schematically illustrated in Figure 1. First, the QAG generated an aerosol, which was blended with laboratory air for laboratory tests and stack gas for field tests. The aerosol flow was then directed into the Xact CEMS stilling chamber 15,17,19 from which the Xact CEMS and XFM simultaneously withdrew representative samples. After sampling, the XFM filters were removed and analyzed, whereas the Xact automatically reported metal concentrations every 15 min for the duration of the test. It should be noted that an Xact CEMS located at the Cooper Environmental Services facility was Volume 56 December 2006 Journal of the Air & Waste Management Association 1735

5 used in the laboratory tests, and an Xact CEMS located at Eli Lilly and Company s Tippecanoe facility in Lafayette, IN, was used in the field. QAG Laboratory Validation. The QAG was evaluated in a series of laboratory tests 15 in which a QAG-generated aerosol was combined with ambient laboratory air and simultaneously sampled by the Xact CEMS and XFM sampling train. The theoretical QAG concentration calculated using eq 1 was then compared with the actual concentrations measured with the Xact and XFM to determine the accuracy, precision, and linearity of the QAG method. All of the QAG-generated aerosols contained As, Cd, Cr, Pb, and Hg, representing the volatile, semivolatile, and lowvolatility metal categories as defined in the NESHAP HWC rule. 3 Tests were conducted at four concentration levels (high, mid, low, and blank) ranging from 0 to 130 g/dscm. In these laboratory tests, the QAG aerosol concentrations were compared with results from a total of 22 XFM and 135 Xact runs. Only Xact runs that corresponded with simultaneous XFM runs were used for accuracy and precision determinations. All of the runs for both methods were used in the linear regression analysis. The XFM sample time was 30 min per run and the Xact was 15 min; concentrations from two Xact runs were averaged to represent a 30-min sample period for comparison with the XFM. All of the methods were performed according to specifications set forth by their test method and standard operating procedures Xact and XFM Field Validation. The Xact and the XFM were evaluated in a series of field tests conducted at a hazardous waste incinerator, where an Xact CEMS was in operation. 15 For each field test, the validated QAG was used to generate a reference aerosol of known concentration, which was spiked into the hazardous waste incinerator flue gas. This dynamically-spiked flue gas was transported to the Xact CEMS stilling chamber where it was simultaneously sampled by the Xact CEMS and XFM sampling train. Measured concentrations from each method were compared with the QAG-spiked stack gas concentration to determine the accuracy and precision of the Xact and XFM methods. All of the QAG-generated aerosols contained As, Cd, Cr, Pb, and Hg, and tests were conducted at four concentration levels (high, mid, low, and blank) ranging from 0 to 130 g/dscm. During these field tests, a total of 24 XFM and 192 Xact runs were compared with the QAG aerosol concentrations. Again, only corresponding Xact/XFM runs were used for accuracy and precision determinations, but all of the runs were used for linear regression analysis. During these tests, the background concentrations of the stack gas were negligible relative to the spiked stack gas concentrations. All of the methods were performed according to specifications set forth by their test method and standard operating procedures Validation Criteria Criteria for method validation were based on Method 301 bias and precision requirements and were either equal to those listed in Method 301 or were more stringent. A general criterion for all of the validation tests was that 9 runs had to be used to demonstrate acceptable values for accuracy, precision, and linearity. 25 Accuracy, precision, and linearity criteria are defined and discussed below. Accuracy. Accuracy for the ith element (A i ) is defined as bias (B i, g/dscm) divided by the theoretical or reference concentration of the QAG generated aerosol (C i QAG, g/ dscm), expressed as a percent. A i B i C i QAG 100 (2) Bias is the difference between the mean measured aerosol concentration (C i M, g/dscm) and the theoretical or reference concentration (C i QAG, g/dscm). B i C i M C i QAG (3) For the QAG, accuracy values 20% were defined as acceptable for validation. For the measurement methods, this same criterion was required for accuracy values determined at concentrations at or near the emission limit as required by Method 301. In addition to meeting the above accuracy criterion, the bias was required to be insignificant relative to the uncertainty in the bias. In this case, the uncertainty in the bias for the ith element (U i B )is the square root of the sum of the uncertainty in the QAG aerosol concentration (U i QAG ) squared and the standard deviation of the mean of the measurement method concentration (SDM i M ) squared, as shown in following equation: U i B U i QAG 2 SDM i M 2 (4) where the QAG propagated uncertainty was estimated to be 5% during the laboratory and field tests. 15 Bias was considered significant at the 95% confidence level if B i 2U i B.IfB i was found to be significant, a correction factor was required. The correction factor (CF i ) is calculated by dividing the concentration of the QAG aerosol (C i QAG )by the concentration of the measurement method (C i M ). CF i C QAG i M (5) C i Precision. Total precision for the ith element (P i t )isdefined as percent relative standard deviation (PRSD i ), or the standard deviation (SD i ) divided by the mean concentration (C i ) of a given method determined from sequential measurements. P i t SD i C i 100 (6) Because there is random variability associated with both the QAG aerosol concentration, as well as the measured aerosol concentration, the total precision (P i t )isanindicator of the combined variability introduced by all 1736 Journal of the Air & Waste Management Association Volume 56 December 2006

6 Table 2. QAG accuracy determined during laboratory tests. Measurement Method ( 65 g/dscm) ( 130 g/dscm) N Accuracy A i % SD N Accuracy A i % SD N Accuracy A i % SD N Accuracy A i % SD Xact XFM Cr As Hg Pb Cd Cr As Hg Pb Cd Notes: SD standard deviation. components of the generation-measurement system; that is, QAG and measurement method. For this series of tests, the magnitude of the QAG precision is considered to be the same as the measurement method. Therefore, the individual precision of the QAG or measurement method (P i ) is defined as the combined, or total, precision (P i t ) divided by the square root of 2. P i PRSD i P t i 2 (7) A precision 10% was required for the method to satisfy established validation requirements. Linearity. Evaluation of linearity was based on parameters derived from a linear least squares regression analysis of the theoretical or reference concentrations and the measured concentrations. Although slope and intercept of the regression line were considered, the key linearity criterion was the correlation coefficient (r). The QAG and both measurement methods must have a correlation coefficient (r) that is 0.85 for method validation. RESULTS AND DISCUSSION QAG Validation Accuracy. QAG accuracy results for each of the five metals tested are listed in Table 2. This table includes results from a comparison of theoretical QAG concentrations with 40 Xact and 20 corresponding XFM measurements. Concentrations ranged from a low of 20 g/dscm to a high of 130 g/dscm. The QAG demonstrated accuracies 7% for all of the elements at each concentration level, well below the 20% validation threshold. These results show that average QAG accuracy for all runs ranged from a low of 1.3% for Hg as measured with the Xact to a high of 4.7% for Hg as measured with the XFM. The overall average accuracy including all five elements, all three concentration levels, and both methods is %. None of the biases were determined to be significant. In addition, independent ICP analysis of a nebulized solution and Table 3. QAG precision determined during laboratory tests. Measurement Method ( 65 g/dscm) ( 130 g/dscm) N Precision P i % N Precision P i % N Precision P i % n Precision P i % Xact XFM Cr As Hg Pb Cd Cr As Hg Pb Cd Volume 56 December 2006 Journal of the Air & Waste Management Association 1737

7 Table 4. QAG linearity determined during laboratory tests. Method n R Slope SD Intercept SD XFM Xact Cr As Hg Pb Cd Cr As Hg Pb Cd Notes: SD standard deviation. selected filter samples were supportive of the XFM measured concentrations. Clearly, these results demonstrate that the QAG is an accurate method for generating a reference aerosol containing metal concentrations in the range from 20 to 130 g/dscm. Precision. Precision for the QAG and the measurement methods tested in the field was determined from sequential measurements taken during a period of 4 hr. The results from these tests are presented in Table 3. The QAG demonstrated precisions well below the 10% validation threshold; average precisions were better than 3% for all of the elements at each concentration level. The average QAG precision for all of the runs ranged from a low of 1% for Cr as measured with the Xact to a high of 2.4% for Hg as measured with XFM. Including all five of the elements, all three of the concentration levels, and both measurement methods, the average QAG precision was 2.2%. These results show that the QAG is a precise method for generating aerosols in the concentration range from 20 to 130 g/dscm. Linearity. Linearity of the QAG was analyzed with linear least-squares regressions of the measured aerosol concentrations compared with the theoretical concentrations over all of the concentration levels. Correlation coefficients (r), slopes, and intercepts from these regressions are presented for each element in Table 4. The correlation coefficients for all of the elements as determined with each measurement method were 0.99, well above the acceptability requirement of Slopes ranged from a low of 0.98 to a high of 1.05, all within 5% of unity (i.e., 1.00). Intercepts ranged from 0.01 to 1.13 and were all 1% of the concentration range. Plots of the regression analyses using all 5 of the metals combined are presented for the Xact in Figure 2 and for the XFM in Figure 3. The QAG again demonstrated correlation coefficients of 0.99, and slopes were within 0.4% of Intercepts were 0.3% of the concentration range. A final regression analysis of all of the measured versus theoretical QAG data yielded a correlation coefficient, a slope within 0.2% of unity ( ), and an intercept 0.1% of Figure 2. QAG linearity as determined by the Xact using all five elements. the concentration range ( ). This strong linear relationship between the theoretical QAG aerosol concentrations and the independently measured concentrations demonstrates the ability of the QAG to produce aerosols quantitatively over a wide range of concentrations. Xact Method Validation Accuracy. The Xact met the 20% validation criterion for accuracies demonstrated at the high concentration level (Table 5). For the measurement methods, this level is expected to be most representative of concentrations near the emission limit as required by Method 301. For the 24 high-concentration runs, the Xact accuracies ranged from 1.8% for As to a high of 13% for Hg. The Xact average accuracies for all 40 runs ranged from a low of 4.9% for As to a high of 16% for Hg. The average Xact accuracy calculated using all five of the elements was % for the high concentration and % for all three of the concentration levels. Bias was insignificant for four of the five metals but was significant for Hg at all of the concentration levels. A post-test investigation of this bias revealed a 6% calibration error, although these runs were retained in the dataset without correction. Assuming correct calibration, the Hg accuracies would have been on the same order as the accuracies of the other four elements; for example, the Hg accuracy at the high concentration level would be 7% (i.e., Figure 3. QAG linearity as determined by the XFM using all five elements Journal of the Air & Waste Management Association Volume 56 December 2006

8 Table 5. Xact accuracy determined during field tests. ( 45 g/dscm) ( 110 g/dscm) N Accuracy A i % SD N Accuracy A i % SD N Accuracy A i % SD N Accuracy A i % SD Cr As Hg 8 14 a a a a 5.2 Pb Cd Notes: SD standard deviation; a Bias was determined to be significant for this value. 6% of the 13% accuracy can be attributed to the calibration error). This identified error can be readily corrected and future bias, thus, eliminated. As a result, a correction factor was not required. In addition to the calibration error, the average accuracies for both the Xact and XFM (discussed below) were somewhat impacted by biases in the middle concentration runs. These biases appear to be the result of an increased difficulty in maintaining proper QAG operating conditions in the field compared with the laboratory. The biases were not significant, and these runs were retained in the dataset. The results presented in Table 5 demonstrate a high accuracy for the Xact measurement method even with the Hg calibration error and midconcentration biases. Furthermore, these results suggest even better accuracies with correct Hg calibrations. Precision. The Xact demonstrated better than 4% for all of the concentration levels, well below the 10% validation threshold (Table 6). The average precision for all of the Xact runs ranged from a low of 1.4% for Hg to a high of 1.9% for Cd. Over all five of the elements and all three of the concentration levels, the total average precision for the Xact was calculated as 1.6%. These results demonstrate the high precision of the Xact over a wide range of concentrations. Linearity. Xact linearity was evaluated with linear leastsquares regression analysis of the Xact-measured concentrations compared with the QAG-spiked stack gas reference concentrations over all of the concentration levels. Regression parameters for each element are presented in Table 7. Xact correlation coefficients for all of the elements were 0.99, easily surpassing the 0.85 requirement. Slopes for all of the metals ranged from 0.87 for Hg to a high of 1.02 for Pb. Intercepts ranged from 0.79 to 1.09, all 1% of the concentration range. A plotted regression analysis for all 5 of the metals combined (Figure 4) produced a correlation coefficient of 0.995, a slope within 5% of unity ( ), and an intercept 2% of the concentration range ( ). The strong linear relationship between the Xact concentrations and the reference concentrations confirms the Xact linear response and high precision over a wide range of concentrations and demonstrates its ability to accurately measure metals over this range. XFM Validation Accuracy. The XFM accuracy results for each element are presented in Table 8. At the high concentration level, the XFM accuracies ranged from 3.5% for Hg to 5.8% for Pb, all well below the 20% validation requirement. None of the biases were significant relative to their uncertainty. The average XFM accuracy for all of the runs ranged from a low of 6% for Cd and As to a high of 7.9% for Pb. Combining data from all five of the elements, the average XFM accuracy was calculated as % for the high concentration level and % for all three of the concentration levels. These results clearly demonstrate the high accuracy of the XFM measurement method. Precision. The XFM precisions (Table 9) all met the 10% validation criterion and were better than 7% at all of the concentration levels. The average XFM precision for all of Table 6. Xact precision determined during field tests. ( 45 g/dscm) ( 110 g/dscm) N Precision P i % N Precision P i % N Precision P i % N Precision P i % Cr As Hg Pb Cd Volume 56 December 2006 Journal of the Air & Waste Management Association 1739

9 Table 7. Xact linearity determined during field tests. N R Slope SD Intercept SD Cr As Hg Pb Cd Notes: SD standard deviation. the runs ranged from a low of 1.8% for Cr and As and to a high of 3.7% for Pb. The total average XFM precision was calculated as 2.6% over all five of the elements and all three of the concentration levels. These results demonstrate that the XFM met this acceptance criterion and is precise over a wide range of concentrations. Figure 4. Xact linearity using all five elements. Table 10. XFM linearity determined during field tests. N R Slope SD Intercept SD Cr As Hg Pb Cd Notes: SD standard deviation. Linearity. The results of the XFM linearity evaluation are presented on Table 10. These parameters were determined with linear least-squares regression analysis of the XFMmeasured concentrations compared with the QAG-spiked stack gas concentrations over all of the concentration levels. XFM correlation coefficients for all elements were 0.99, well above the 0.85 acceptance value. Slopes for all of the metals ranged from 1.04 to 1.07, all within 7% of unity. Intercepts were all 3% of the linearity test concentration range and ranged from 1.61 to A plotted regression analysis for all 5 of the metals combined is presented for the XFM in Figure 5. The combined metal regression yielded a correlation coefficient of 0.996, a slope within 5% of 1.00 ( ), and an intercept 2% of the full concentration range ( ). The strong correlation of the XFM concentrations with the reference concentrations demonstrates the ability of the XFM to accurately measure a wide range of concentrations and confirms its precision over this range. Summary A summary of each method s performance in reference to the validation criteria is presented in Table 11. This table Table 8. XFM accuracy determined during field tests. ( 45 g/dscm) ( 110 g/dscm) N Accuracy A i % SD N Accuracy A i % SD N Accuracy A i % SD N Accuracy A i % SD Cr As Hg Pb Cd Notes: SD standard deviation. Table 9. XFM precision determined during field tests. ( 45 g/dscm) ( 110 g/dscm) N Precision P i % N Precision P i % N Precision P i % N Precision P i % Cr As Hg Pb Cd Journal of the Air & Waste Management Association Volume 56 December 2006

10 CONCLUSIONS The three independent methods discussed in this paper met EPA Method 301 requirements for compliance applications and are precise, accurate, and linear tools applicable to the determination of hazardous metal concentrations in emissions. The QAG and the Xact technologies have currently been approved by EPA for compliance demonstration at a HWC. 26 In addition to the five elements measured with these tests, the methods are expected to be applicable to most hazardous elements in stack and fugitive emissions from stationary sources, as well as in the ambient environment. The QAG reference aerosol generator could, thus, be used as a generally applicable reference aerosol generator for research, method certification, and audits. The XFM measurement methods could be used for performance testing, initial certification, and continuing quality assurance audits of multimetal CEMS. Finally, the Xact could be used to demonstrate ongoing compliance through its use as a CEMS, and its technology could be extended to an instrumental analysis procedure with applications similar to the XFM. In addition to the results presented in this paper, all three methods have demonstrated high precisions and accuracies in other field and laboratory tests The measurement methods have also consistently demonstrated frequent sampling times, rapid analysis turnaround times (15 min for the Xact and 2 hr for the XFM if a field-portable analyzer is used), and low detection limits. Analysis is nondestructive, and the methods can be independently audited with the QAG. In addition to metals, the QAG can potentially generate aerosols of both organic and inorganic analytes. Figure 5. XFM linearity using all five elements. clearly shows that the three methods easily passed all of the required criteria for all five of the metals. Table 11. Summary of validation results. Parameters Criteria QAG Method XFM Method Xact Method No. of runs (n) Accuracy (A) 20% 5% 6% a 7% a,b Precision (P) 10% 3% 4% 4% Linearity (r) Notes: a Results from high concentration level; b Assuming correct calibration of Hg. REFERENCES 1. Prieditis, H.; Adamson, I.Y. Comparative Pulmonary Toxicity of Various Soluble Metals Found in Urban Particulate Dust; Exp. Lung. Res. 2002, 28, Monn, C.; Becker, S. Cytotoxicity and Induction of Proinflammatory Cytokines from Human Monocytes Exposed to Fine (PM2.5) and Coarse Particles (PM10 2.5) in Outdoor and Indoor Air; Toxicol. Appl. Pharmacol. 1999, 155, U.S. Environmental Protection Agency. National Emissions Standards for Hazardous Air Pollutants for Source Categories. In Code of Federal Regulations, 40 CFR 63; U.S. Government Printing Office: Washington, DC, U.S. Environmental Protection Agency. Method 29, Determination of Metals Emissions from Stationary Sources. In Code of Federal Regulations, 40 CFR 60; U.S. Government Printing Office: Washington, DC, U.S. Environmental Protection Agency. Evaluation of Two Methods for the Measurement of Mercury Emissions in Exhaust Gases from a Municipal Waste Combustor; EPA 450/ U.S. Government Printing Office: Washington, DC, Rigo, H.G.; Chandler A.J. Quantitation Limits for Reference Methods 23, 26, and 29; J. Air & Waste Manage. Assoc. 1999, 49, American Society of Mechanical Engineers, Research Committee on Industrial and Municipal Waste. Reference Method Accuracy and Precision (ReMAP): Phase 1, Precision of Manual Stack Emission Measurements; Report No. CRTD-60; American Society of Mechanical Engineers International: Washington, DC, Zhang, H.; Yueh, F.Y.; Singh, J.P. Evaluation of the Potential of Laser- Induced Breakdown Spectroscopy for Detection of Trace in Liquid; J. Air & Waste Manage. Assoc. 2002, 51, Seltzer, M.D. Performance Testing of a Multimetals Continuous Emissions Monitor. J. Air & Waste Manage. Assoc. 2000, 50, Sky, Inc Performance Testing of Multi-Metal Continuous Emissions Monitors; DOE/ID U.S. Department of Energy, Idaho Operations Office: Idaho Falls, ID, Cooper, J.A; Fry, S.C.; Johnsen, B.E. Apparatus and Methods for Removing Mercury from Fluid Stream. U.S. Patent 6,770,250 B2, Cooper, J.A; Fry, S.C.; Johnsen, B.E. Apparatus and Methods for Removing Mercury from Fluid Stream. U.S. Patent 6,872,370 B2, Draft Test Plan for Validating a Reference Procedure for Measuring the Concentration of Metals in Eli Lilly and Company s Hazardous Waste Incinerator Stack Gas; Cooper Environmental Services: Portland, OR, U.S. Environmental Protection Agency. Method 301, Field Validation of Pollution Measurement Methods for Various Media. In Code of Federal Regulations, 40 CFR 56; U.S. Government Printing Office: Washington, DC, Method 301 Evaluation of Three Methods for Multi-Metals Measurement and Validation; Cooper Environmental Services: Portland, OR, Barth, D.C.; Yanca, C.A.; Petterson, K.A., Nakanishi, M.P.; Cooper, J.A.; Johnsen, B.J.; Lambert, R. H. Aerosol Sci. Tech., submitted for publication. 17. Determination of Metal Concentration in CES Xact CEMS Stilling Chamber Using the CES XFM; Cooper Environmental Services: Portland, OR, Determination of Metal Concentrations from Stationary Sources Using the CES XFM; Cooper Environmental Services: Portland, OR, Specifications and Test Procedure for X-Ray Fluorescence Based Multi- Metals Continuous Emissions Monitoring Systems at Stationary Sources; Cooper Environmental Services: Portland, OR, Quality Assurance Requirements for X-Ray Fluorescence Based Multi-Metals Continuous Emissions Monitoring Systems at Stationary Sources; Cooper Environmental Services: Portland, OR, Cooper, J.A.; Barth, D.B.; Nakanishi, M.P.; Petterson, K.A.; Yanca, C.A. A Quantitative, Traceable-to-NIST, Reference Aerosol Generator for Evaluating Aerosol Speciation Monitors and Measurement Methods. In Proceedings of the Air & Waste Management Association s Symposium on Air Quality Measurement Methods and Technology, Durham, NC, May 9 11, 2006; A&WMA: Pittsburgh, PA, in press. 22. May, K.R. The Collison Nebulizer. Description, Performance & Application. J. Aerosol. Sci. 1973, 4, Cooper, J.A.; Barth, D.B.; Nakanishi, M.P.; Petterson, K.A.; Yanca, C.A. New Filter-Based Methods for Precise, Accurate, and Timelier Measurement of Metal Emissions. In Proceedings of the Air & Waste Management Volume 56 December 2006 Journal of the Air & Waste Management Association 1741

11 Association s Symposium on Air Quality Measurement Methods and Technology; Durham, NC, May 9 11, 2006; A&WMA: Pittsburgh, PA, in press. 24. U.S. Environmental Protection Agency. Compendium Method IO-3.3: Determination of Metals in Ambient Particulate Matter Using X-Ray Fluorescence (XRF) Spectroscopy. In Compendium of Methods for the Determination of Inorganic Compounds in Ambient Air; EPA/625/R-96/ 010a. U.S. Government Printing Office: Washington, DC, Watson, J.G.; Lioy, P.J.; Mueller, P.K. The Measurement Process: Precision, Accuracy and Validity. In Air Sampling Instruments for Evaluation of Atmospheric Contaminants, 6th ed.; Lioy, P.J; Lioy, M.J.Y; Eds.; American Conference of Governmental Industrial Hygienists: Cincinnati, OH, 1983; pp Oldham, C.B. Air Measurements and Quality Group, Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC. Written communication with Eli Lilly and Company: Lafayette, IN Hay, K.J.; Boddu, V.M.; Johnsen; B.E.; Cooper, J.A. X-Ray Fluorescence- Based Multi-Metal Continuous Emission Monitor Technology Demonstration; ERDC/CERL-TR-05-3; U.S. Army Corps of Engineers Engineer Research and Development Center: Champaign, IL, Installation and Use of an X-Ray Based Continuous Emissions Monitor for AED Treatability Study; Cooper Environmental Services: Portland, OR, Re-design and Field Testing of an X-Ray Fluorescence Multi-Metals Monitor at the Iowa Army Ammunition Plant, Middleton, Iowa Coal Fired Boiler; Cooper Environmental Services: Portland, OR, Hay, K.J.; Johnsen, B.E.; Ginochio, P.R.; Cooper, J.A. Relative Accuracy Testing of an X-Ray Fluorescence-Based Mercury Monitor at Coal- Fired Boilers; J. Air & Waste Manage. Assoc. 2006, 56, About the Authors Catherine Yanca, Douglas Barth, Krag Petterson, and Michael Nakanishi are environmental scientists at Cooper Environmental Services, LLC. John Cooper is the president of Cooper Environmental Services, LLC. Bruce Johnsen is a claims adjuster at SAIF Co. Richard Lambert is an engineering consultant at the Engineering Tech Center, Eli Lilly and Co. Daniel Bivins is the work assignment manager for the Emission Measurement Center of the Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency. Address correspondence to: John Cooper, Cooper Environmental Services, LLC, SW Nimbus Ave., Suite J6, Portland, OR 97223; phone: ; fax: ; jacooper@cooperenvironmental.com Journal of the Air & Waste Management Association Volume 56 December 2006