Appendix Q to Part 50 - Reference Method for the Determination of Lead in Particulate Matter as PM10 Collected From Ambient Air
40:2.0.1.1.1.0.1.20.18 : Appendix Q
Appendix Q to Part 50 - Reference Method for the Determination of
Lead in Particulate Matter as PM10 Collected From Ambient Air
This Federal Reference Method (FRM) draws heavily from the
specific analytical protocols used by the U.S. EPA.
1. Applicability and Principle
1.1 This method provides for the measurement of the lead (Pb)
concentration in particulate matter that is 10 micrometers or less
(PM10) in ambient air. PM10 is collected on an acceptable (see
section 6.1.2) 46.2 mm diameter polytetrafluoroethylene (PTFE)
filter for 24 hours using active sampling at local conditions with
a low-volume air sampler. The low-volume sampler has an average
flow rate of 16.7 liters per minute (Lpm) and total sampled volume
of 24 cubic meters (m 3) of air. The analysis of Pb in PM10 is
performed on each individual 24-hour sample. Gravimetric mass
analysis of PM10c filters is not required for Pb analysis. For the
purpose of this method, PM10 is defined as particulate matter
having an aerodynamic diameter in the nominal range of 10
micrometers (10 µm) or less.
1.2 For this reference method, PM10 shall be collected with the
PM10c federal reference method (FRM) sampler as described in
appendix O to Part 50 using the same sample period, measurement
procedures, and requirements specified in appendix L of Part 50.
The PM10c sampler is also being used for measurement of PM10−2.5
mass by difference and as such, the PM10c sampler must also meet
all of the performance requirements specified for PM2.5 in appendix
L. The concentration of Pb in the atmosphere is determined in the
total volume of air sampled and expressed in micrograms per cubic
meter (µg/m 3) at local temperature and pressure conditions.
1.3 The FRM will serve as the basis for approving Federal
Equivalent Methods (FEMs) as specified in 40 CFR Part 53 (Reference
and Equivalent Methods). This FRM specifically applies to the
analysis of Pb in PM10 filters collected with the PM10c sampler. If
these filters are analyzed for elements other than Pb, then refer
to the guidance provided in the EPA Inorganic Compendium Method
IO-3.3 (Reference 1 of section 8) for multi-element analysis.
1.4 The PM10c air sampler draws ambient air at a constant
volumetric flow rate into a specially shaped inlet and through an
inertial particle size separator, where the suspended particulate
matter in the PM10 size range is separated for collection on a PTFE
filter over the specified sampling period. The Pb content of the
PM10 sample is analyzed by energy-dispersive X-ray fluorescence
spectrometry (EDXRF). Energy-dispersive X-ray fluorescence
spectrometry provides a means for identification of an element by
measurement of its characteristic X-ray emission energy. The method
allows for quantification of the element by measuring the intensity
of X-rays emitted at the characteristic photon energy and then
relating this intensity to the elemental concentration. The number
or intensity of X-rays produced at a given energy provides a
measure of the amount of the element present by comparisons with
calibration standards. The X-rays are detected and the spectral
signals are acquired and processed with a personal computer. EDXRF
is commonly used as a non-destructive method for quantifying trace
elements in PM. A detailed explanation of quantitative X-ray
spectrometry is described in references 2, 3 and 4.
1.5 Quality assurance (QA) procedures for the collection of
monitoring data are contained in Part 58, appendix A.
2. PM10Pb Measurement Range and Detection Limit.
The values given below in section 2.1 and 2.2 are typical of the
method capabilities. Absolute values will vary for individual
situations depending on the instrument, detector age, and operating
conditions used. Data are typically reported in ng/m 3 for ambient
air samples; however, for this reference method, data will be
reported in µg/m 3 at local temperature and pressure
conditions.
2.1 EDXRF Pb Measurement Range. The typical ambient air
measurement range is 0.001 to 30 µg Pb/m 3, assuming an upper range
calibration standard of about 60 µg Pb per square centimeter (cm
2), a filter deposit area of 11.86 cm 2, and an air volume of 24 m
3. The top range of the EDXRF instrument is much greater than what
is stated here. The top measurement range of quantification is
defined by the level of the high concentration calibration standard
used and can be increased to expand the measurement range as
needed.
2.2 Detection Limit (DL). A typical estimate of the
one-sigma detection limit (DL) is about 2 ng Pb/cm 2 or 0.001 µg
Pb/m 3, assuming a filter size of 46.2 mm (filter deposit area of
11.86 cm 2) and a sample air volume of 24 m 3. The DL is an
estimate of the lowest amount of Pb that can be reliably
distinguished from a blank filter. The one-sigma detection limit
for Pb is calculated as the average overall uncertainty or
propagated error for Pb, determined from measurements on a series
of blank filters from the filter lot(s) in use. Detection limits
must be determined for each filter lot in use. If a new filter lot
is used, then a new DL must be determined. The sources of random
error which are considered are calibration uncertainty; system
stability; peak and background counting statistics; uncertainty in
attenuation corrections; and uncertainty in peak overlap
corrections, but the dominating source by far is peak and
background counting statistics. At a minimum, laboratories are to
determine annual estimates of the DL using the guidance provided in
Reference 5.
3. Factors Affecting Bias and Precision of Lead Determination
by EDXRF
3.1 Filter Deposit. X-ray spectra are subject to
distortion if unusually heavy deposits are analyzed. This is the
result of internal absorption of both primary and secondary X-rays
within the sample; however, this is not an issue for Pb due to the
energetic X-rays used to fluoresce Pb and the energetic
characteristic X-rays emitted by Pb. The optimum mass filter
loading for multi-elemental EDXRF analyis is about 100 µg/cm 2 or
1.2 mg/filter for a 46.2-mm filter. Too little deposit material can
also be problematic due to low counting statistics and signal
noise. The particle mass deposit should minimally be 15 µg/cm 2.
The maximum PM10 filter loading or upper concentration limit of
mass expected to be collected by the PM10c sampler is 200 µg/m 3
(Appendix O to Part 50, Section 3.2). This equates to a mass
loading of about 400 µg/cm 2 and is the maximum expected loading
for PM10c filters. This maximum loading is acceptable for the
analysis of Pb and other high-Z elements with very energetic
characteristic X-rays. A properly collected sample will have a
uniform deposit over the entire collection area. Samples with
physical deformities (including a visually non-uniform deposit
area) should not be quantitatively analyzed. Tests on the
uniformity of particle deposition on PM10C filters showed that the
non-uniformity of the filter deposit represents a small fraction of
the overall uncertainty in ambient Pb concentration measurement.
The analysis beam of the XRF analyzer does not cover the entire
filter collection area. The minimum allowable beam size is 10
mm.
3.2 Spectral Interferences and Spectral Overlap. Spectral
interference occurs when the entirety of the analyte spectral lines
of two species are nearly 100% overlapped. The presence of arsenic
(As) is a problematic interference for EDXRF systems which use the
Pb Lα line exclusively to quantify the Pb concentration. This is
because the Pb Lα line and the As Kα lines severely overlap. The
use of multiple Pb lines, including the Lβ and/or the Lγ lines for
quantification must be used to reduce the uncertainty in the Pb
determination in the presence of As. There can be instances when
lines partially overlap the Pb spectral lines, but with the energy
resolution of most detectors these overlaps are typically
de-convoluted using standard spectral de-convolution software
provided by the instrument vendor. An EDXRF protocol for Pb must
define which Pb lines are used for quantification and where
spectral overlaps occur. A de-convolution protocol must be used to
separate all the lines which overlap with Pb.
3.3 Particle Size Effects and Attenuation Correction
Factors. X-ray attenuation is dependent on the X-ray energy,
mass sample loading, composition, and particle size. In some cases,
the excitation and fluorescent X-rays are attenuated as they pass
through the sample. In order to relate the measured intensity of
the X-rays to the thin-film calibration standards used, the
magnitude of any attenuation present must be corrected for. See
references 6, 7, and 8 for more discussion on this issue.
Essentially no attenuation corrections are necessary for Pb in
PM10: Both the incoming excitation X-rays used for analyzing lead
and the fluoresced Pb X-rays are sufficiently energetic that for
particles in this size range and for normal filter loadings, the Pb
X-ray yield is not significantly impacted by attenuation.
4. Precision
4.1 Measurement system precision is assessed according to the
procedures set forth in appendix A to part 58. Measurement method
precision is assessed from collocated sampling and analysis. The
goal for acceptable measurement uncertainty, as precision, is
defined as an upper 90 percent confidence limit for the coefficient
of variation (CV) of 20 percent.
5. Bias
5.1 Measurement system bias for monitoring data is assessed
according to the procedures set forth in appendix A of part 58. The
bias is assessed through an audit using spiked filters. The goal
for measurement bias is defined as an upper 95 percent confidence
limit for the absolute bias of 15 percent.
6. Measurement of PTFE Filters by EDXRF
6.1 Sampling
6.1.1 Low-Volume PM10cSampler. The low-volume
PM10c sampler shall be used for PM10 sample collection and operated
in accordance with the performance specifications described in part
50, appendix L.
6.1.2 PTFE Filters and Filter Acceptance Testing. The
PTFE filters used for PM10c sample collection shall meet the
specifications provided in part 50, appendix L. The following
requirements are similar to those currently specified for the
acceptance of PM2.5 filters that are tested for trace elements by
EDXRF. For large filter lots (greater than 500 filters) randomly
select 20 filters from a given lot. For small lots (less than 500
filters) a lesser number of filters may be taken. Analyze each
blank filter separately and calculate the average lead
concentration in ng/cm 2. Ninety percent, or 18 of the 20 filters,
must have an average lead concentration that is less than 4.8 ng
Pb/cm 2.
6.1.2.1 Filter Blanks. Field blank filters shall be
collected along with routine samples. Field blank filters will be
collected that are transported to the sampling site and placed in
the sampler for the duration of sampling without sampling.
Laboratory blank filters from each filter lot used shall be
analyzed with each batch of routine sample filters analyzed.
Laboratory blank filters are used in background subtraction as
discussed below in Section 6.2.4.
6.2 Analysis. The four main categories of random and
systematic error encountered in X-ray fluorescence analysis include
errors from sample collection, the X-ray source, the counting
process, and inter-element effects. These errors are addressed
through the calibration process and mathematical corrections in the
instrument software. Spectral processing methods are well
established and most commercial analyzers have software that can
implement the most common approaches (references 9-11) to
background subtraction, peak overlap correction, counting and
deadtime corrections.
6.2.1 EDXRF Analysis Instrument. An energy-dispersive XRF
system is used. Energy-dispersive XRF systems are available from a
number of commercial vendors. Examples include Thermo
(www.thermo.com), Spectro (http://www.spectro.com),
Xenemetrix (http://www.xenemetrix.com) and PANalytical
(http://www.panalytical.com). 1 The analysis is performed at
room temperature in either vacuum or in a helium atmosphere. The
specific details of the corrections and calibration algorithms are
typically included in commercial analytical instrument software
routines for automated spectral acquisition and processing and vary
by manufacturer. It is important for the analyst to understand the
correction procedures and algorithms of the particular system used,
to ensure that the necessary corrections are applied.
1 These are examples of available systems and is not an all
inclusive list. The mention of commercial products does not imply
endorsement by the U.S. Environmental Protection Agency.
6.2.2 Thin film standards. Thin film standards are used
for calibration because they most closely resemble the layer of
particles on a filter. Thin films standards are typically deposited
on Nuclepore substrates. The preparation of thin film standards is
discussed in reference 8, and 10. The NIST SRM 2783 (Air
Particulate on Filter Media) is currently available on
polycarbonate filters and contains a certified concentration for
Pb. Thin film standards at 15 and 50 µg/cm 2 are commercially
available from MicroMatter Inc. (Arlington, WA).
6.2.3 Filter Preparation. Filters used for sample
collection are 46.2-mm PTFE filters with a pore size of 2 microns
and filter deposit area 11.86 cm 2. Cold storage is not a
requirement for filters analyzed for Pb; however, if filters
scheduled for XRF analysis were stored cold, they must be allowed
to reach room temperature prior to analysis. All filter samples
received for analysis are checked for any holes, tears, or a
non-uniform deposit which would prevent quantitative analysis.
Samples with physical deformities are not quantitatively
analyzable. The filters are carefully removed with tweezers from
the Petri dish and securely placed into the instrument-specific
sampler holder for analysis. Care must be taken to protect filters
from contamination prior to analysis. Filters must be kept covered
when not being analyzed. No other preparation of filter samples is
required.
6.2.4 Calibration. In general, calibration determines
each element's sensitivity, i.e., its response in x-ray
counts/sec to each µg/cm 2 of a standard and an interference
coefficient for each element that causes interference with another
one (See section 3.2 above). The sensitivity can be determined by a
linear plot of count rate versus concentration (µg/cm 2) in which
the slope is the instrument's sensitivity for that element. A more
precise way, which requires fewer standards, is to fit sensitivity
versus atomic number. Calibration is a complex task in the
operation of an XRF system. Two major functions accomplished by
calibration are the production of reference spectra which are used
for fitting and the determination of the elemental sensitivities.
Included in the reference spectra (referred to as “shapes”) are
background-subtracted peak shapes of the elements to be analyzed
(as well as interfering elements) and spectral backgrounds. Pure
element thin film standards are used for the element peak shapes
and clean filter blanks from the same lot as routine filter samples
are used for the background. The analysis of Pb in PM filter
deposits is based on the assumption that the thickness of the
deposit is small with respect to the characteristic Pb X-ray
transmission thickness. Therefore, the concentration of Pb in a
sample is determined by first calibrating the spectrometer with
thin film standards to determine the sensitivity factor for Pb and
then analyzing the unknown samples under identical excitation
conditions as used to determine the calibration. Calibration shall
be performed annually or when significant repairs or changes occur
(e.g., a change in fluorescers, X-ray tubes, or detector).
Calibration establishes the elemental sensitivity factors and the
magnitude of interference or overlap coefficients. See
reference 7 for more detailed discussion of calibration and
analysis of shapes standards for background correction, coarse
particle absorption corrections, and spectral overlap.
6.2.4.1 Spectral Peak Fitting. The EPA uses a library of
pure element peak shapes (shape standards) to extract the elemental
background-free peak areas from an unknown spectrum. It is also
possible to fit spectra using peak stripping or analytically
defined functions such as modified Gaussian functions. The EPA
shape standards are generated from pure, mono-elemental thin film
standards. The shape standards are acquired for sufficiently long
times to provide a large number of counts in the peaks of interest.
It is not necessary for the concentration of the standard to be
known. A slight contaminant in the region of interest in a shape
standard can have a significant and serious effect on the ability
of the least squares fitting algorithm to fit the shapes to the
unknown spectrum. It is these elemental peak shapes that are fitted
to the peaks in an unknown sample during spectral processing by the
analyzer. In addition to this library of elemental shapes there is
also a background shape spectrum for the filter type used as
discussed below in section 6.2.4.2 of this section.
6.2.4.2 Background Measurement and Correction. A
background spectrum generated by the filter itself must be
subtracted from the X-ray spectrum prior to extracting peak areas.
Background spectra must be obtained for each filter lot used for
sample collection. The background shape standards which are used
for background fitting are created at the time of calibration. If a
new lot of filters is used, new background spectra must be
obtained. A minimum of 20 clean blank filters from each filter lot
are kept in a sealed container and are used exclusively for
background measurement and correction. The spectra acquired on
individual blank filters are added together to produce a single
spectrum for each of the secondary targets or fluorescers used in
the analysis of lead. Individual blank filter spectra which show
atypical contamination are excluded from the summed spectra. The
summed spectra are fitted to the appropriate background during
spectral processing. Background correction is automatically
included during spectral processing of each sample.
7. Calculation.
7.1 PM10 Pb concentrations. The PM10 Pb
concentration in the atmosphere (µg/m 3) is calculated using the
following equation:
Where,
MPb is the mass per unit volume
for lead in µg/m 3;
CPb is the mass per unit area for lead
in µg/cm 2 as measured by XRF; A is the filter deposit area in cm
2;
VLC is the total volume of air sampled by the PM10c
sampler in actual volume units measured at local conditions of
temperature and pressure, as provided by the sampler in m 3.
7.2 PM10 Pb Uncertainty Calculations.
The principal contributors to total uncertainty of XRF values
include: field sampling; filter deposit area; XRF calibration;
attenuation or loss of the x-ray signals due to the other
components of the particulate sample; and determination of the Pb
X-ray emission peak area by curve fitting. See reference 12 for a
detailed discussion of how uncertainties are similarly calculated
for the PM2.5 Chemical Speciation program.
The model for calculating total uncertainty is:
δtot = (
δf2 +
δa2 +
δc2 +
δv2)
1/2 Where,
δf = fitting uncertainty (XRF-specific, from 2 to
100 + %)
δa = attenuation uncertainty (XRF-specific,
insignificant for Pb)
δc = calibration uncertainty (combined
lab uncertainty, assumed as 5%)
δv = volume/deposition size
uncertainty (combined field uncertainty, assumed as 5%)
8. References
1. Inorganic Compendium Method IO-3.3; Determination of Metals
in Ambient Particulate Matter Using X-Ray Fluorescence (XRF)
Spectroscopy; U.S. Environmental Protection Agency, Cincinnati, OH
45268. EPA/625/R-96/010a. June 1999.
2. Jenkins, R., Gould, R.W., and Gedcke, D. Quantitative X-ray
Spectrometry: Second Edition. Marcel Dekker, Inc., New York, NY.
1995.
3. Jenkins, R. X-Ray Fluorescence Spectrometry: Second Edition
in Chemical Analysis, a Series of Monographs on Analytical
Chemistry and Its Applications, Volume 152. Editor J.D.Winefordner;
John Wiley & Sons, Inc., New York, NY. 1999.
4. Dzubay, T.G. X-ray Fluorescence Analysis of Environmental
Samples, Ann Arbor Science Publishers Inc., 1977.
5. Code of Federal Regulations (CFR) 40, Part 136, Appendix B;
Definition and Procedure for the Determination of the Method
Detection Limit - Revision 1.1.
6. Drane, E.A, Rickel, D.G., and Courtney, W.J., “Computer Code
for Analysis X-Ray Fluorescence Spectra of Airborne Particulate
Matter,” in Advances in X-Ray Analysis, J.R. Rhodes, Ed., Plenum
Publishing Corporation, New York, NY, p. 23 (1980).
7. Analysis of Energy-Dispersive X-ray Spectra of Ambient
Aerosols with Shapes Optimization, Guidance Document; TR-WDE-06-02;
prepared under contract EP-D-05-065 for the U.S. Environmental
Protection Agency, National Exposure Research Laboratory. March
2006.
8. Billiet, J., Dams, R., and Hoste, J. (1980) Multielement Thin
Film Standards for XRF Analysis, X-Ray Spectrometry, 9(4):
206-211.
9. Bonner, N.A.; Bazan, F.; and Camp, D.C. (1973). Elemental
analysis of air filter samples using x-ray fluorescence. Report No.
UCRL-51388. Prepared for U.S. Atomic Energy Commission, by Univ. of
Calif., Lawrence Livermore Laboratory, Livermore, CA.
10. Dzubay, T.G.; Lamothe, P.J.; and Yoshuda, H. (1977). Polymer
films as calibration standards for X-ray fluorescence analysis.
Adv. X-Ray Anal., 20:411.
11. Giauque, R.D.; Garrett, R.B.; and Goda, L.Y. (1977).
Calibration of energy-dispersive X-ray spectrometers for analysis
of thin environmental samples. In X-Ray Fluorescence Analysis of
Environmental Samples, T.G. Dzubay, Ed., Ann Arbor Science
Publishers, Ann Arbor, MI, pp. 153-181.
12. Harmonization of Interlaboratory X-ray Fluorescence
Measurement Uncertainties, Detailed Discussion Paper; August 4,
2006; prepared for the Office of Air Quality Planning and Standards
under EPA contract 68-D-03-038.
http://www.epa.gov/ttn/amtic/files/ambient/pm25/spec/xrfdet.pdf.
[73 FR 67052, Nov. 12, 2008]