METHOD 6200
FIELD PORTABLE X-RAY FLUORESCENCE SPECTROMETRY FOR THE
DETERMINATION OF ELEMENTAL CONCENTRATIONS IN SOIL AND SEDIMENT
1.0 SCOPE AND APPLICATION
1.1	This method applies only to radioisotope source instruments and is
applicable to the in situ and intrusive analysis of the 26 analytes listed
in Table 1 for soil and sediment samples. Some common elements are not
listed in Table 1 because they are considered "light" elements that cannot
be detected by field portable x-ray fluorescence (FPXRF). They are:
lithium, beryllium, sodium, magnesium, aluminum, silicon, and phosphorus.
Most of the analytes listed in Table 1 are of environmental concern, while
a few others have interference effects or change the elemental composition
of the matrix, affecting quantitation of the analytes of interest.
Generally elements of atomic number 16 or greater can be detected and
quantitated by FPXRF.
1.2	Detection limits depend on several factors, the analyte of interest, the
type of detector used, the type of excitation source, the strength of the
excitation source, count times used to irradiate the sample, physical
matrix effects, chemical matrix effects, and interelement spectral
interferences. General instrument detection limits for analytes of
interest in environmental applications are shown in Table 1. These
detection limits apply to a clean matrix of quartz sand (silicon dioxide)
free of interelement spectral interferences using long (600-second) count
times. These detection limits are given for guidance only and will vary
depending on the sample matrix, which instrument is used, and operating
conditions. A discussion of field performance-based detection limits is
presented in Section 13.4 of this method. The clean matrix and field
performance-based detection limits should be used for general planning
purposes, and a third detection limit discussed, based on the standard
deviation around single measurements, should be used in assessing data
quality. This detection limit is discussed in Sections 9.7 and 11.3.
1.3	Use of this method is restricted to personnel either trained and
knowledgeable in the operation of an XRF instrument or under the
supervision of a trained and knowledgeable individual. This method is a
screening method to be used with confirmatory analysis using EPA-approved
methods. This method's main strength is as a rapid field screening
procedure. The method detection limits (MDL) of FPXRF are above the
toxicity characteristic regulatory level for most RCRA analytes. If the
precision, accuracy, and detection limits of FPXRF meet the data quality
DRAFT
6200 - 1
Revision 0
July 1996

-------
objectives (DQO) of your project, then XRF is a fast, powerful, cost
effective technology for site characterization.
2.0 SUMMARY OF METHOD
2.1 The FPXRF technologies described in this method use sealed radioisotope
sources to irradiate samples with x-rays. When a sample is irradiated
with x-rays, the source x-rays may undergo either scattering or absorption
by sample atoms. This later process is known as the photoelectric effect.
When an atom absorbs the source x-rays, the incident radiation dislodges
electrons from the innermost shells of the atom, creating vacancies. The
electron vacancies are filled by electrons cascading in from outer
electron shells. Electrons in outer shells have higher energy states than
inner shell electrons, and the outer shell electrons give off energy as
they cascade down into the inner shell vacancies. This rearrangement of
electrons results in emission of x-rays characteristic of the given atom.
The emission of x-rays, in this manner, is termed x-ray fluorescence.
Three electron shells are generally involved in emission of x-rays during
FPXRF analysis of environmental samples: the K, L, and M shells. A
typical emission pattern, also called an emission spectrum, for a given
metal has multiple intensity peaks generated from the emission of K, L, or
M shell electrons. The most commonly measured x-ray emissions are from
the K and L shells; only metals with an atomic number greater than 57 have
measurable M shell emissions.
Each characteristic x-ray line is defined with the letter K, L, or M,
which signifies which shell had the original vacancy and by a subscript
alpha (a) or beta (B), which indicates the higher shell from which
electrons fell to fill the vacancy and produce the x-ray. For example, a
K„ line is produced by a vacancy in the K shell filled by an L shell
electron, whereas a K0 line is produced by a vacancy in the K shell filled
by an M shell electron. The K,, transition is on average 6 to 7 times more
probable than the Ka transition; therefore, the Ka line is approximately
7 times more intense than the K6 line for a given element, making the K.
line the choice for quantitation purposes.
The K lines for a given element are the most energetic lines and are the
preferred lines for analysis. For a given atom, the x-rays emitted from
L transitions are always less energetic than those emitted from K
transitions. Unlike the K lines, the main L emission lines (La and L6) for
an element are of nearly equal intensity. The choice of one or the other
depends on what interfering element lines might be present. The L
DRAFT
6200 - 2
Revision 0
July 1996

-------
emission lines are useful for analyses involving elements of atomic number
(Z) 58 (cerium) through 92 (uranium).
An x-ray source can excite characteristic x-rays from an element only if
the source energy is greater than the absorption edge energy for the
particular line group of the element that is, the K absorption edge, L
absorption edge, or M absorption edge energy. The absorption edge energy
is somewhat greater than the corresponding line energy. Actually, the K
absorption edge energy is approximately the sum of the K, L, and M line
energies of the particular element, and the L absorption edge energy is
approximately the sum of the L and M line energies. FPXRF is more
sensitive to an element with an absorption edge energy close to but less
than the excitation energy of the source. For example, when using a
cadmium-109 source, which has an excitation energy of 22.1 kiloelectron
volts (keV), FPXRF would exhibit better sensitivity for zirconium which
has a K line energy of 15.7 keV than to chromium, which has a K line
energy of 5.41 keV.
2.2 Under this method, inorganic analytes of interest are identified and
quantitated using a field portable energy-dispersive x-ray fluorescence
spectrometer. Radiation from one or more radioisotope sources is used to
generate characteristic x-ray emissions from elements in a sample. Up to
three sources may be used to irradiate a sample. Each source emits a
specific set of primary x-rays that excite a corresponding range of
elements in a sample. When more than one source can excite the element of
interest, the source 1s selected according to Its excitation efficiency
for the element of interest.
For measurement, the sample is positioned in front of the probe window.
This can be done in two manners using FPXRF instruments: in situ or
intrusive. If operated in the in situ mode, the probe window is placed in
direct contact with the soil surface to be analyzed. When an FPXRF
Instrument is operated in the intrusive mode, a soil or sediment sample
must be collected, prepared, and placed in a sample cup. The sample cup
is then placed on top of the window inside a protective cover for
analysis.
Sample analysis is then initiated by exposing the sample to primary
radiation from the source. Fluorescent and backscattered x-rays from the
sample enter through the detector window and are converted into electric
pulses in the detector. The detector in FPXRF instruments is usually
either a solid-state detector or a gas-filled proportional counter.
Within the detector, energies of the characteristic x-rays are converted
into a train of electric pulses, the amplitudes of which are linearly
DRAFT
6200 - 3
Revision 0
July 1996

-------
proportional to the energy of the x-rays. An electronic multichannel
analyzer (MCA) measures the pulse amplitudes, which is the basis of
qualitative x-ray analysis. The number of counts at a given energy per
unit of time is representative of the element concentration in a sample
and is the basis for quantitative analysis. Most FPXRF instruments are
menu-driven from software built into the units or from personal computers
(PC).
The measurement time of each source is user-selectable. Shorter source
measurement times (30 seconds) are generally used for initial screening
and hot spot delineation, and longer measurement times (up to 300 seconds)
are typically used to meet higher precision and accuracy requirements.
FPXRF Instruments can be calibrated using the following methods:
internally using fundamental parameters determined by the manufacturer,
empirically based on site-specific calibration standards (SSCS), or based
on Compton peak ratios. The Compton peak is produced by backscattering of
the source radiation. Some FPXRF instruments can be calibrated using
multiple methods.
3.0 DEFINITIONS
3.1	FPXRF: Field Portable X-Ray Fluorescence.
3.2	MCA: Multichannel Analyzer for measuring pulse amplitude.
3.3	SSCS: Site Specific Calibration Standard.
3.4	FP: Fundamental Parameter.
3.5	ROI: Region of Interest.
3.6	SRM: Standard Reference Material. A standard containing certified
amounts of metals in soil or sediment.
3.7	eV: Electron Volt. A unit of energy equivalent to the amount of energy
gained by an electron passing through a potential difference of one volt.
3.8	Refer to Chapter One and Chapter Three for additional definitions.
4.0 INTERFERENCES
4.1 The total method error for FPXRF analysis is defined as the square root of
the sum of squares of both instrument precision and user- or application-
related error. Generally, instrument precision is the least significant
DRAFT
6200 - 4
Revision 0
July 1996

-------
source of error in FPXRF analysis. User- or application-related error is
generally more significant and varies with each site and method used.
Some sources of interference can be minimized or controlled by the
instrument operator, but others cannot. Common sources of user- or
application-related error are discussed below.
4.2	Physical matrix effects result from variations 1n the physical character
of the sample. These variations may include such parameters as particle
size, uniformity, homogeneity, and surface condition. For example, if any
analyte exists in the form of very fine particles in a coarser-grained
matrix, the analyte's concentration measured by the FPXRF will vary
depending on how fine particles are distributed within the coarser-grained
matrix. If the fine particles "settle" to the bottom of the sample cup,
the analyte concentration measurement will be higher than if the fine
particles are not mixed in well and stay on top of the coarser-grained
particles in the sample cup. One way to reduce such error is to grind and
sieve all soil samples to a uniform particle size thus reducing sample-to-
sample particle size variability. Homogeneity is always a concern when
dealing with soil samples. Every effort should be made to thoroughly mix
and homogenize soil samples before analysis. Field studies have shown
heterogeneity of the sample generally has the largest impact on
comparability with confirmatory samples.
4.3	Moisture content may affect the accuracy of analysis of soil and sediment
sample analyses. When the moisture content is between 5 and 20 percent,
the overall error from moisture may be minimal. However, moisture content
may be a major source of error when analyzing samples of surface soil or
sediment that are saturated with water. This error can be minimized by
drying the samples in a convection or toaster oven. Microwave drying is
not recommended because.field studies have shown that microwave drying can
increase variability between FPXRF data and confirmatory analysis and
because metal fragments in the sample can cause arcing to occur in a
microwave.
4.4	Inconsistent positioning of samples in front of the probe window is a
potential source of error because the x-ray signal decreases as the
distance from the radioactive source increases. This error is minimized
by maintaining the same distance between the window and each sample. For
the best results, the window of the probe should be in direct contact with
the sample, which means that the sample should be flat and smooth to
provide a good contact surface.
DRAFT
6200 - 5
Revision 0
July 1996

-------
4.5	Chemical matrix effects result from differences in the concentrations of
interfering elements. These effects occur as either spectral
interferences (peak overlaps) or as x-ray absorption and enhancement
phenomena. Both effects are common in soils contaminated with heavy
metals. As examples of absorption and enhancement effects; iron (Fe)
tends to absorb copper (Cu) x-rays, reducing the intensity of the Cu
measured by the detector, while chromium (Cr) will be enhanced at the
expense of Fe because the absorption edge of Cr is slightly lower in
energy than the fluorescent peak of iron. The effects can be corrected
mathematically through the use of fundamental parameter (FP) coefficients.
The effects also can be compensated for using SSCS, which contain all the
elements present on site that can interfere with one another.
4.6	When present in a sample, certain x-ray lines from different elements can
be very close in energy and, therefore, can cause interference by
producing a severely overlapped spectrum. The degree to which a detector
can resolve the two different peaks depends on the energy resolution of
the detector. If the energy difference between the two peaks in electron
volts is less than the resolution of the detector in electron volts, then
the detector will not be able to fully resolve the peaks.
The most common spectrum overlaps involve the KB line of element Z-l with
the Kc line of element Z. This is called the Kc/Ku interference. Because
the KqIKb intensity ratio for a given element usually is about 7:1, the
interfering element, Z-l, must be present at large concentrations to cause
a problem. Two examples of this type of spectral interference involve the
presence of large concentrations of vanadium (V) when attempting to
measure Cr or the presence of large concentrations of Fe when attempting
to measure cobalt (Co). The V K„ and energies are 4.95 and 5.43 keV,
respectively, and the Cr K* energy is 5.41 keV. The Fe K,, and KB energies
are 6.40 and 7.06 keV, respectively, and the Co K,, energy is 6.92 keV. The
difference between the V Kb and Cr K,, energies is 20 eV, and the difference
between the Fe KB and the Co K,, energies is 140 eV. The resolution of the
highest-resolution detectors in FPXRF instruments is 170 eV. Therefore,
large amounts of V and Fe will interfere with quantitation of Cr or Co,
respectively. The presence of Fe is a frequent problem because it is
often found in soils at tens of thousands of parts per million (ppm).
4.7	Other interferences can arise from K/L, K/M, and L/M line overlaps.
Although these overlaps are less common. Examples of such overlap involve
arsenic (As) Ko/lead (Pb) U and sulfur (S) Ko/Pb Ma. In the As/Pb case,
Pb can be measured from the Pb L6 line, and As can be measured from either
the As Ka or the As Kfi line; in this way the interference can be corrected.
If the As Kfl line is used, sensitivity will be decreased by a factor of two
DRAFT
6200 - 6
Revision 0
July 1996

-------
to five times because it is a less intense line than the As Ka line. If
the As Ka line is used in the presence of Pb, mathematical corrections
within the instrument software can be used to subtract out the Pb
interference. However, because of the limits of mathematical corrections,
As concentrations cannot be efficiently calculated for samples with Pb:As
ratios of 10:1 or more. This high ratio of Pb to As may result in no As
being reported regardless of the actual concentration present.
No instrument can fully compensate for this interference. It is important
for an operator to understand this limitation of FPXRF instruments and
consult with the manufacturer of the FPXRF instrument to evaluate options
to minimize this limitation. The operator's decision will be based on
action levels for metals in soil established for the site, matrix effects,
capabilities of the instrument, data quality objectives, and the ratio of
lead to arsenic known to be present at the site. If a site is encountered
that contains lead at concentrations greater than ten times the
concentration of arsenic it is advisable that all critical soil samples be
sent off site for confirmatory analysis by an EPA-approved method.
4.8	If SSCS are used to calibrate an FPXRF instrument, the samples collected
must be representative of the site under investigation. Representative
soil sampling ensures that a sample or group of samples accurately
reflects the concentrations of the contaminants of concern at a given time
and location. Analytical results for representative samples reflect
variations in the presence and concentration ranges of contaminants
throughout a site. Variables affecting sample representativeness include
differences in soil type, contaminant concentration variability, sample
collection and preparation variability, and analytical variability, all of
which should be minimized as much as possible.
4.9	Soil physical and chemical effects may be corrected using SSCS that have
been analyzed by inductively coupled plasma (ICP) or atomic absorption
(AA) methods. However, a major source of error can be introduced if these
samples are not representative of the site or if the analytical error is
large. Another concern is the type of digestion procedure used to prepare
the soil samples for the reference analysis. Analytical results for the
confirmatory method will vary depending on whether a partial digestion
procedure, such as SW-846 Method 3050A, or a total digestion procedure,
such as Method 3052 is used. It is known that depending on the nature of
the soil or sediment, Method 3050 will achieve differing extraction
efficiencies for different analytes of interest. The confirmatory method
should meet the project data quality objectives.
DRAFT
6200 - 7
Revision 0
July 1996

-------
XRF measures the total concentration of an element therefore to achieve
the greatest comparability of this method with the reference method
(reduced bias), a total digestion procedure should be used for sample
preparation. However, in the study used to generate the performance data
for this method, the confirmatory method used was Method 3050, and the
FPXRF data compared very well with regression correlation coefficients (r2
often exceeding 0.95, except for barium and chromium. (See Table 8 in
Section 13.0). The critical factor is that the digestion procedure and
analytical reference method used should meet the data quality objectives
(DQO) of the project and match the method used for confirmation analysis.
4.10 Ambient temperature changes can affect the gain of the amplifiers
producing instrument drift. Gain or drift is primarily a function of the
electronics (amplifier or preamplifier) and not the detector as most
instrument detectors are cooled to a constant temperature. Most FPXRF
instruments have a built-in automatic gain control. If the automatic gain
control is allowed to make periodic adjustments, the instrument will
compensate for the influence of temperature changes on its energy scale.
If the FPXRF instrument has an automatic gain control function, the
operator will not have to adjust the instrument's gain unless an error
message appears. If an error message appears, the operator should follow
the manufacturer's procedures for troubleshooting the problem. Often this
involves performing a new energy calibration. The performance of an
energy calibration check to assess drift is a quality control measure
discussed in Section 9.3
If the operator is instructed by the manufacturer to manually conduct a
gain check because of increasing or decreasing ambient temperature, it is
standard to perform a gain check after every 10 to 20 sample measurements
or once an hour whichever is more frequent.- It is also suggested that a
gain check be performed if the temperature fluctuates more than 10 to
20°F. The operator should follow the manufacturer's recommendations for
gain check frequency.
5.0 SAFETY
5.1 Proper training for the safe operation of the instrument and radiation
training should be completed by the analyst prior to analysis. Radiation
safety for each specific instrument can be found in the operators manual.
Protective shielding should never be removed by the analyst or any
personnel other than the manufacturer. The analyst should be aware of the
local state and national regulations that pertain to the use of radiation-
producing equipment and radioactive materials with which compliance is
required. Licenses for radioactive materials are of two types; (1)
general license which is usually provided by the manufacturer for
DRAFT
6200 - 8
Revision 0
July 1996

-------
receiving, acquiring, owning, possessing, using, and transferring
radioactive material incorporated in a device or equipment, and (2)
specific license which is issued to named persons for the operation of
radioactive instruments as required by local state agencies. There should
be a person appointed within the organization that is solely responsible
for properly instructing all personnel, maintaining inspection records,
and monitoring x-ray equipment at regular intervals. A copy of the
radioactive material licenses and leak tests should be present with the
instrument at all times and available to local and national authorities
upon request.
5.2	Radiation monitoring equipment should be used with the handling of the
instrument. The operator and the surrounding environment should be
monitored continually for analyst exposure to radiation. Thermal
luminescent detectors (TLD) in the form of badges and rings are used to
monitor operator radiation exposure. The TLDs should be worn in the area
of most frequent exposure. The maximum permissible whole-body dose from
occupational exposure is 5 Roentgen Equivalent Man (REM) per year.
Possible exposure pathways for radiation to enter the body are ingestion,
inhaling, and absorption. The best precaution to prevent radiation
exposure is distance and shielding.
5.3	Refer to Section 3.1.4 of Chapter Three for guidance on some proper safety
protocols.
6.0 EQUIPMENT AND SUPPLIES
6.1 FPXRF Spectrometer; An FPXRF spectrometer consists of four major
components: (1) a source that provides x-rays; (2) a sample presentation
device; (3) a detector that converts.x-ray-generated photons emitted from
the sample into measurable electronic signals; and (4) a data processing
unit that contains an emission or fluorescence energy analyzer, such as an
MCA that processes the signals into an x-ray energy spectrum from which
elemental concentrations in the sample may be calculated, and a data
display and storage system. These components and additional, optional
items, are discussed below.
6.1.1 Excitation Sources: All FPXRF instruments use sealed radioisotope
sources to produce x-rays in order to irradiate samples. The FPXRF
instrument may contain between one and three radioisotope sources.
Common radioisotope sources used for analysis for metals in soils
are iron (Fe)-55, cadmium (Cd)-109, americium (Am)-241, and curium
(Cm)-244. These sources may be contained in a probe along with a
window and the detector; the probe is connected to a data reduction
DRAFT
6200 - 9
Revision 0
July 1996

-------
and handling system by means of a flexible cable. Alternatively,
the sources, window, and detector may be included in the same unit
as the data reduction and handling system.
The relative strength of the radioisotope sources is measured in
units of millicuries (mCi). All other components of the FPXRF
system being equal, the stronger the source, the greater the
sensitivity and precision of a given instrument. Radioisotope
sources undergo constant decay. In fact, it is this decay process
that emits the primary x-rays used to excite samples for FPXRF
analysis. The decay of radioisotopes is measured in "half-lives."
The half-life of a radioisotope is defined as the length of time
required to reduce the radioisotopes strength or activity by half.
Developers of FPXRF technologies recommend source replacement at
regular intervals based on the source's half-life. The
characteristic x-rays emitted from each of the different sources
have energies capable of exciting a certain range of analytes in a
sample. Table 2 summarizes the characteristics of four common
radioisotope sources.
6.1.2	Sample Presentation Device: FPXRF instruments can be operated in two
modes: in situ and intrusive. If operated in the in situ mode, the
probe window is placed in direct contact with the soil surface to be
analyzed. When an FPXRF instrument is operated in the intrusive
mode, a soil or sediment sample must be collected, prepared, and
placed in a sample cup. For most FPXRF instruments operated in the
intrusive mode, the probe is rotated so that the window faces
upward. A protective sample cover is placed over the window, and
the sample cup is placed on top of the window inside the protective
sample cover for analysis.
6.1.3	Detectors: The detectors in the FPXRF instruments can be either
solid-state detectors or gas-filled, proportional counter detectors.
Common solid-state detectors include mercuric iodide (Hgl2), silicon
pin diode and lithium-drifted silicon Si(li). The Hgl2 detector is
operated at a moderately subambient temperature controlled by a low
power thermoelectric cooler. The silicon pin diode detector also is
cooled via the thermoelectric Peltier effect. The Si(Li) detector
must be cooled to at least -90 eC either with liquid nitrogen or by
thermoelectric cooling via the Peltier effect. Instruments with a
Si(Li) detector have an internal liquid nitrogen dewar with a
capacity of 0.5 to 1.0 liter. Proportional counter detectors are
rugged and lightweight, which are important features of a field
portable detector. However, the resolution of a proportional
DRAFT
6200 - 10
Revision 0
July 1996

-------
counter detector is not as good as that of a solid-state detector.
The energy resolution of a detector for characteristic x-rays is
usually expressed in terms of full width at half-maximum (FWHM)
height of the manganese Ka peak at 5.89 keV. The typical
resolutions of the above mentioned detectors are as follows: HgI2-
270 eV; silicon pin diode-700 eV; Si(Li)—170 eV; and gas-filled,
proportional counter-750 eV.
During operation of a solid-state detector, an x-ray photon strikes
a biased, solid-state crystal and loses energy in the crystal by
producing electron-hole pairs. The electric charge produced is
collected and provides a current pulse that is directly proportional
to the energy of the x-ray photon absorbed by the crystal of the
detector. A gas-filled, proportional counter detector is an
ionization chamber filled with a mixture of noble and other gases.
An x-ray photon entering the chamber ionizes the gas atoms. The
electric charge produced is collected and provides an electric
signal that is directly proportional to the energy of the x-ray
photon absorbed by the gas in the detector.
6.1.4 Data Processing Units: The key component in the data processing unit
of an FPXRF instrument is the MCA. The MCA receives pulses from the
detector and sorts them by their amptitudes (energy level). The MCA
counts pulses per second to determine the height of the peak in a
spectrum, which is indicative of the target analyte's concentration.
The spectrum of element peaks are built on the MCA. The MCAs in
FPXRF instruments have from 256 to 2,048 channels. The
concentrations of target analytes are usually shown in parts per
million on a liquid crystal display (LCD) in the instrument. FPXRF
instruments can store both spectra and from 100 to 500 sets of
numerical analytical results. Most FPXRF instruments are menu-
driven from software built into the units or from PCs. Once the
data-storage memory of an FPXRF unit is full, data can be downloaded
by means of an RS-232 port and cable to a PC.
6.2	Spare battery chargers
6.3	Polyethylene sample cups: 31 millimeters (mm) to 40 mm in diameter with
collar, or equivalent (appropriate for FPXRF instrument)
6.4	X-ray fluorescent window film: Mylar, Kapton, Spectrolene, polypropylene,
or equivalent; 2.5 or 6.0 micrometers (um) thick
DRAFT
6200 - 11
Revision 0
July 1996

-------
6.5	Mortar and pestle: glass, agate, or aluminum oxide; for grinding soil and
sediment samples
6.6	Containers: glass or plastic to store samples
6.7	Sieves: 60-mesh (0.25 mm), stainless-steel, Nylon, or equivalent for
preparing soil and sediment samples
6.8	Trowels: for smoothing soil surfaces and collecting soil samples
6.9	Plastic bags: used for collection and homogenization of soil samples
6.10	Drying oven: standard convection or toaster oven, for soil and sediment
samples that require drying
7.0 REAGENTS AND STANDARDS
7.1	PURE ELEMENT STANDARDS: Each pure, single-element standard is intended to
produce strong characteristic x-ray peaks of the element of interest only.
Other elements present must not contribute to the fluorescence spectrum.
A set of pure element standards for commonly sought analytes is supplied
by the instrument manufacturer, if required for the instrument; not all
instruments require the pure element standards. The standards are used to
set the region of interest (ROI) for each element. They also can be used
as energy calibration and resolution check samples.
7.2	SITE-SPECIFIC CALIBRATION STANDARDS: Instruments that employ fundamental
parameters (FP) or similar mathematical models in minimizing matrix
effects may not require SSCS. If the FP calibration model is to be
optimized or if empirical calibration is necessary, then SSCSs must be
collected, prepared, and analyzed.
7.2.1	The SSCS must be representative of the matrix to be analyzed by
FPXRF. These samples must be well homogenized. A minimum of ten
samples spanning the concentration ranges of the analytes of
interest and of the interfering elements must be obtained from the
site. A sample size of 4 to 8 ounces is recommended, and standard
glass sampling jars should be used.
7.2.2	Each sample should be oven-dried for 2 to 4 hours at a temperature
of less than 150°C. If mercury is to be analyzed, a separate sample
portion must remain undried, as heating may volatilize the mercury.
When the sample is dry, all large, organic debris and
nonrepresentative material, such as twigs, leaves, roots, insects,
DRAFT
6200 - 12
Revision 0
July 1996

-------
asphalt, and rock should be removed. The sample should be ground
with a mortar and pestle and passed through a 60-mesh sieve. Only
the coarse rock fraction should remain on the screen.
7.2.3 The sample should be homogenized by using a riffle splitter or by
placing 150 to 200 grams of the dried, sieved sample on a piece of
kraft or butcher paper about 1.5 by 1.5 feet in size. Each corner
of the paper should be lifted alternately, rolling the soil over on
itself and toward the opposite corner. The soil should be rolled on
itself 20 times. Approximately 5 grams of the sample should then be
removed and placed in a sample cup for FPXRF analysis. The rest of
the prepared sample should be sent off site for ICP or AA analysis.
The method use for confirmatory analysis should meet the data
quality objectives of the project.
7.3	BLANK SAMPLES: The blank samples should be from a "clean" quartz or
silicon dioxide matrix that is free of any analytes at concentrations
above the method detection limits. These samples are used to monitor for
cross-contamination and laboratory-induced contaminants or interferences.
7.4	STANDARD REFERENCE MATERIALS: Standard reference materials (SRM) are
standards containing certified amounts of metals in soil or sediment.
These standards are used for accuracy and performance checks of FPXRF
analyses. SRMs can be obtained from the National Institute of Standards
and Technology (NIST), the U.S. Geological Survey (USGS), the Canadian
National Research Council, and the national bureau of standards in foreign
nations. Pertinent NIST SRMs for FPXRF analysis include 2704, Buffalo
River Sediment; 2709, San Joaquin Soil; and 2710 and 2711, Montana Soil.
These SRMs contain soil or sediment from actual sites that has been
analyzed using independent inorganic analytical methods by many different
laboratories.
8.0 SAMPLE COLLECTION, PRESERVATION, AND STORAGE
Sample handling and preservation procedures used in FPXRF analyses should follow the
guidelines in Chapter Three, Metallic Analytes, Section 3.1.3.
9.0 QUALITY CONTROL
9.1	Refer to Chapter One for additional guidance on quality assurance
protocols. All field data sheets and quality control data should be
maintained for reference or inspection.
9.2	Energy Calibration Check: To determine whether an FPXRF instrument is
operating within resolution and stability tolerances, an energy
DRAFT
6200 - 13
Revision 0
July 1996

-------
calibration check should be run. The energy calibration check determines
whether the characteristic x-ray lines are shifting, which would indicate
drift within the instrument. As discussed in Section 4.10, this check
also serves as a gain check in the event that ambient temperatures are
fluctuating greatly (> 10 to 20°F).
The energy calibration check should be run at a frequency consistent with
manufacturers recommendations. Generally, this would be at the beginning
of each working day, after the batteries are changed or the instrument is
shut off, at the end of each working day, and at any other time when the
instrument operator believes that drift is occurring during analysis. A
pure element such as iron, manganese, copper, or lead is often used for
the energy calibration check. A manufacturer-recommended count time per
source should be used for the check.
9.2.1 The instrument manufacturer's manual specifies the channel or
kiloelectron volt level at which a pure element peak should appear
and the expected intensity of the peak. The intensity and channel
number of the pure element as measured using the radioactive source
should be checked and compared to the manufacturer's recommendation.
If the energy calibration check does not meet the manufacturer's
criteria, then the pure element sample should be repositioned and
reanalyzed. If the criteria are still not met, then an energy
calibration should be performed as described in the manufacturer's
manual. With some FPXRF instruments, once a spectrum is acquired
from the energy calibration check, the peak can be optimized and
realigned to the manufacturer's specifications using their software.
9.3 Blank Samples: Two types of blank samples should be analyzed for FPXRF
analysis: instrument blanks and method blanks. An instrument blank is
used to verify that no contamination exists in the spectrometer or on the
probe window.
9.3.1 The instrument blank can be silicon dioxide, a Teflon block, a
quartz block, "clean" sand, or lithium carbonate. This instrument
blank should be analyzed on each working day before and after
analyses are conducted and once per every twenty samples. An
instrument blank should also be analyzed whenever contamination is
suspected by the analyst. The frequency of analysis will vary with
the data quality objectives of the project. A manufacturer-
recommended count time per source should be used for the blank
analysis. No element concentrations above the method detection
limits should be found in the instrument blank. If concentrations
exceed these limits, then the probe window and the check sample
DRAFT
6200 - 14
Revision 0
July 1996

-------
should be checked for contamination. If contamination is not a
problem, then the instrument must be "zeroed" by following the
manufacturer's instructions.
9.3.2 A method blank is used to monitor for laboratory-induced
contaminants or interferences. The method blank can be "clean"
silica sand or lithium carbonate that undergoes the same preparation
procedure as the samples. A method blank must be analyzed at least
daily. The frequency of analysis will depend on the data quality
objectives of the project. To be acceptable, a method blank must
not contain any analyte at a concentration above its method
detection limit. If an analyte's concentration exceeds its method
detection limit, the cause of the problem must be identified, and
all samples analyzed with the method blank must be reanalyzed.
9.4	Calibration Verification Checks: A calibration verification check sample
is used to check the accuracy of the instrument and to assess the
stability and consistency of the analysis for the analytes of interest.
A check sample should be analyzed at the beginning of each working day,
during active sample analyses, and at the end of each working day. The
frequency of calibration checks during active analysis will depend on the
data quality objectives of the project. The check sample should be a well
characterized soil sample from the site that is representative of site
samples in terms of particle size and degree of homogeneity and that
contains contaminants at concentrations near the action levels. If a
site-specific sample is not available, then an NIST or other SRM that
contains the analytes of interest can be used to verify the accuracy of
the instrument. The measured value for each target analyte should be
within ±20 percent (%D) of the true value for the calibration verification
check to be acceptable. If a measured value falls outside this range,
then the check sample should be reanalyzed. If the value continues to
fall outside the acceptance range, the instrument should be recalibrated,
and the batch of samples analyzed before the unacceptable calibration
verification check must be reanalyzed.
9.5	Precision Measurements: The precision of the method is monitored by
analyzing a sample with low, moderate, or high concentrations of target
analytes. The frequency of precision measurements will depend on the data
quality objectives for the data. A minimum of one precision sample should
be run per day. Each precision sample should be analyzed 7 times in
replicate. It is recommended that precision measurements be obtained for
samples with varying concentration ranges to assess the effect of
concentration on method precision. Determining method precision for
analytes at concentrations near the site action levels can be extremely
DRAFT
6200 - 15
Revision 0
July 1996

-------
important if the FPXRF results are to be used in an enforcement action;
therefore selection of at least one sample with target analyte
concentrations at or near the site action levels or levels of concern is
therefore recommended. A precision sample is analyzed by the instrument
for the same field analysis time as used for other project samples. The
relative standard deviation (RSD) of the sample mean is used to assess
method precision. For FPXRF data to be considered adequately precise, the
RSD should not be greater than 20 percent with the exception of chromium.
RSD values for chromium should not be greater than 30 percent.
The equation for calculating RSD is as follows:
RSD ¦ (SD/Mean Concentration) x 100
where:
RSD = Relative standard deviation for the precision measurement for
the analyte
SD - Standard deviation of the concentration for the analyte
Mean Concentration ¦= Mean concentration for the analyte
The precision or reproducibility of a measurement will improve with
increasing count time, however, increasing the count time by a factor of
4 will provide only 2 times better precision, so there is a point of
diminishing return. Increasing the count time also decreases the
detection limit, and decreases sample throughput.
9.6 Detection Limits: Results for replicate analyses of a low-concentration
sample, SSCS, or SRM can be used to generate an average site-specific
method detection and quantitation limits. In this case, the method
detection limit is defined as 3 times the standard deviation of the
results for the low-concentration samples and the method quantitation
limit is defined as 10 times the standard deviation of the same results.
Another means of determining method detection and quantitation limits
involves use of counting statistics. In FPXRF analysis, the standard
deviation from counting statistics is defined as SD = (N)1*, where SD is the
standard deviation for a target analyte peak and N is the gross counts for
the peak of the analyte of interest. Three times this standard deviation
would be the method detection limit and 10 times this standard deviation
would be the method quantitation limit. If both of the abovementioned
approaches are used to calculate method detection limits, the larger of
the standard deviations should be used to provide the more conservative
detection limits.
This SD based detection limit criteria must be used by the operator to
evaluate each measurement for its useability. A measurement above the
DRAFT
6200 - 16
Revision 0
July 1996

-------
average calculated or manufacturer's detection limit, but smaller than
three times its associated SD, should not be used as a quantitative
measurement. Conversely, if the measurement is below the average
calculated or manufacturer's detection limit, but greater than three times
its associated SO. It should be coded as an estimated value.
9.7 Confirmatory Samples: The comparability of the FPXRF analysis is
determined by submitting FPXRF-analyzed samples for analysis at a
laboratory. The method of confirmatory analysis must meet the project and
XRF measurement data quality objectives. The confirmatory samples must be
splits of the well homogenized sample material. In some cases the
prepared sample cups can be submitted. A minimum of I sample for each 20
FPXRF-analyzed samples should be submitted for confirmatory analysis.
This frequency will depend on data quality objectives. The confirmatory
analyses can also be used to verify the quality of the FPXRF data. The
confirmatory samples should be selected from the lower, middle, and upper
range of concentrations measured by the FPXRF. They should also include
samples with analyte concentrations at or near the site action levels.
The results of the confirmatory analysis and FPXRF analyses should be
evaluated with a least squares linear regression analysis. If the
measured concentrations span more than one order of magnitude, the data
should be log-transformed to standardize variance which is proportional to
the magnitude of measurement. The correlation coefficient (r2) for the
results should be 0.7 or greater for the FPXRF data to be considered
screening level data. If the r2 is 0.9 or greater and inferential
statistics indicate the FPXRF data and the confirmatory data are
statistically equivalent at a 99 percent confidence level, the data could
potentially meet definitive level data criteria.
10.0 CALIBRATION AND STANDARDIZATION
10.1	Instrument Calibration: Instrument calibration procedures vary among FPXRF
instruments. Users of this method should follow the calibration
procedures outlined in the operator's manual for each specific FPXRF
instrument. Generally, however, three types of calibration procedures
exist for FPXRF instruments: FP calibration, empirical calibration, and
the Compton peak ratio or normalization method. These three types of
calibration are discussed below.
10.2	Fundamental Parameters Calibration: FP calibration procedures are
extremely variable. An FP calibration provides the analyst with a
"standardless" calibration. The advantages of FP calibrations over
empirical calibrations include the following:
DRAFT
6200 - 17
Revision 0
July 1996

-------
•	No previously collected site-specific samples are required,
although site-specific samples with confirmed and validated
analytical results for all elements present could be used.
•	Cost is reduced because no confirmatory laboratory results or
calibration standards are required.
However, the analyst should be aware of the limitations imposed on FP
calibration by particle size and matrix effects. These limitations can be
minimized by adhering to the preparation procedure described in Section
7.2. The two FP calibration processes discussed below are based on an
effective energy FP routine and a back scatter with FP (BFP) routine.
Each FPXRF FP calibration process is based on a different iterative
algorithmic method. The calibration procedure for each routine is
explained in detail in the manufacturer's user manual for each FPXRF
instrument; in addition, training courses are offered for each
instrument.
10.2.1 Effective Energy FP Calibration: The effective energy FP
calibration is performed by the manufacturer before an
instrument is sent to the analyst. Although SSCS can be used,
the calibration relies on pure element standards or SRMs such
as those obtained from NIST for the FP calibration. The
effective energy routine relies on the spectrometer response
to pure elements and FP iterative algorithms to compensate for
various matrix effects.
Alpha coefficients are calculated using a variation of the
Sherman equation, which calculates theoretical intensities
from the measurement of pure element samples. These
coefficients indicate the quantitative effect of each matrix
element on an analyte's measured x-ray intensity. Next, the
Lachance Traill algorithm is solved as a set of simultaneous
equations based on the theoretical intensities. The alpha
coefficients are then downloaded into the specific instrument.
The working effective energy FP calibration curve must be
verified before sample analysis begins on each working day,
after every 20 samples are analyzed, and at the end of
sampling. This verification is performed by analyzing either
an NIST SRM or an SSCS that is representative of the site-
specific samples. This SRM or SSCS serves as a calibration
check. A manufacturer-recommended count time per source
should be used for the calibration check. The analyst must
DRAFT
6200 - 18
Revision 0
July 1996

-------
then adjust the y-intercept and slope of the calibration curve
to best fit the known concentrations of target analytes in the
SRM or SSCS.
A percent difference (%D) is then calculated for each target
analyte. The %D should be within ±20 percent of the certified
value for each analyte. If the %D falls outside this
acceptance range, then the calibration curve should be
adjusted by varying the slope of the line or the y-intercept
value for the analyte. The SRM or SSCS is reanalyzed until
the %D falls within ±20 percent. The group of 20 samples
analyzed before an out-of-control calibration check should be
reanalyzed.
The equation to calibrate %D is as follows:
%D - ((Cs - Ck) / CJ x 100
where:
%D = Percent difference
Ck = Certified concentration of standard sample
C3 = Measured concentration of standard sample
10.2.2 BFP Calibration: BFP calibration relies on the ability of the
liquid nitrogen-cooled, Si(Li) solid-state detector to
separate the coherent (Compton) and incoherent (Rayleigh)
backscatter peaks of primary radiation. These peak
intensities are known to be a function of sample composition,
and the ratio of the Compton to Rayleigh peak is a function of
the mass absorption of the sample. The calibration procedure
is explained in detail in the instrument manufacturer's
manual. Following is a general description of the BFP
calibration procedure.
The concentrations of all detected and quantified elements are
entered into the computer software system. Certified element
results for an NIST SRM or confirmed and validated results for
an SSCS can be used. In addition, the concentrations of
oxygen and silicon must be entered; these two concentrations
are not found in standard metals analyses. The manufacturer
provides silicon and oxygen concentrations for typical soil
types. Pure element standards are then analyzed using a
manufacturer-recommended count time per source. The results
are used to calculate correction factors in order to adjust
for spectrum overlap of elements.
DRAFT
6200 - 19
Revision 0
July 1996

-------
The working BFP calibration curve must be verified before
sample analysis begins on each working day, after every 20
samples are analyzed, and at the end of the analysis. This
verification is performed by analyzing either an NIST SRM or
an SSCS that is representative of the site-specific samples.
This SRM or SSCS serves as a calibration check. The standard
sample is analyzed using a manufacturer-recommended count time
per source to check the calibration curve. The analyst must
then adjust the y-intercept and slope of the calibration curve
to best fit the known concentrations of target analytes in the
SRM or SSCS.
A %D is then calculated for each target analyte. The %D
should fall within ±20 percent of the certified value for each
analyte. If the %D falls outside this acceptance range, then
the calibration curve should be adjusted by varying the slope
of the line the y-intercept value for the analyte. The
standard sample is reanalyzed until the %D falls within ±20
percent. The group of 20 samples analyzed before an
out-of-control calibration check should be reanalyzed.
10.3 EMPIRICAL CALIBRATION: An empirical calibration can be performed with
SSCS, site-typical standards, or standards prepared from metal oxides. A
discussion of SSCS is included in Section 7.2; if no previously
characterized samples exist for a specific site, site-typical standards
can be used. Site-typical standards may be selected from commercially
available characterized soils or from SSCS prepared for another site. The
site-typical standards should closely approximate the site's soil matrix
with respect to particle size distribution, mineralogy, and contaminant
analytes. If neither SSGS nor site-typical standards are available, it is
possible to make gravimetric standards by adding metal oxides to a "clean"
sand or silicon dioxide matrix that simulates soil. Metal oxides can be
purchased from various chemical vendors. If standards are made on site,
a balance capable of weighing items to at least two decimal places is
required. Concentrated ICP or AA standard solutions can also be used to
make standards. These solutions are available in concentrations of 10,000
parts per million, thus only small volumes have to be added to the soil.
An empirical calibration using SSCS involves analysis of SSCS by the FPXRF
instrument and by a conventional analytical method such as ICP or AA. A
total acid digestion procedure should be used by the laboratory for sample
preparation. Generally, a minimum of 10 and a maximum of 30 well
characterized SSCS, site-typical standards, or prepared metal oxide
standards are required to perform an adequate empirical calibration. The
DRAFT
6200 - 20
Revision 0
July 1996

-------
number of required standards depends on the number of analytes of interest
and interfering elements. Theoretically, an empirical calibration with
SSCS should provide the most accurate data for a site because the
calibration compensates for site-specific matrix effects.
The first step in an empirical calibration is to analyze the pure element
standards for the elements of interest. This enables the instrument to
set channel limits for each element for spectral deconvolution. Next the
SSCS, site-typical standards, or prepared metal oxide standards are
analyzed using a count time of 200 seconds per source or a count time
recommended by the manufacturer. This will produce a spectrum and net
intensity of each analyte in each standard. The analyte concentrations
for each standard are then entered into the instrument software; these
concentrations are those obtained from the laboratory, the certified
results, or the gravimetrically determined concentrations of the prepared
standards. This gives the instrument analyte values to regress against
corresponding intensities during the modeling stage. The regression
equation correlates the concentrations of an analyte with its net
intensity.
The calibration equation is developed using a least squares fit regression
analysis. After the regression terms to be used in the equation are
defined, a mathematical equation can be developed to calculate the analyte
concentration in an unknown sample. In some FPXRF instruments, the
software of the instrument calculates the regression equation. The
software uses calculated intercept and slope values to form a multiterm
equation. In conjunction with the software in the instrument, the
operator can adjust the multiterm equation to minimize interelement
interferences and optimize the intensity calibration curve.
It is possible to define up to six linear or nonlinear terms in the
regression equation. Terms can be added and deleted to optimize the
equation. The goal is to produce an equation with the smallest regression
error and the highest correlation coefficient. These values are
automatically computed by the software as the regression terms are added,
deleted, or modified. It is also possible to delete data points from the
regression line if these points are significant outliers or if they are
heavily weighing the data. Once the regression equation has been selected
for an analyte, the equation can be entered into the software for
quantitation of analytes in subsequent samples. For an empirical
calibration to be acceptable, the regression equation for a specific
analyte should have a correlation coefficient of 0.98 or greater.
DRAFT
6200 - 21
Revision 0
July 1996

-------
In an empirical calibration, one must apply the DQOs of the project and
ascertain critical or action levels for the analytes of interest. It is
within these concentration ranges or around these action levels that the
FPXRF instrument should be calibrated most accurately. It may not be
possible to develop a good regression equation over several orders of
analyte concentration.
10.4 COMPTON NORMALIZATION METHOD: The Compton normalization method is based
on analysis of a single, certified standard and normalization for the
Compton peak. The Compton peak is produced from incoherent backscattering
of x-ray radiation from the excitation source and is present in the
spectrum of every sample. The Compton peak intensity changes with
differing matrices. Generally, matrices dominated by lighter elements
produce a larger Compton peak, and those dominated by heavier elements
produce a smaller Compton peak. Normalizing to the Compton peak can
reduce problems with varying matrix effects among samples. Compton
normalization is similar to the use of internal standards in organics
analysis. The Compton normalization method may not be effective when
analyte concentrations exceed a few percent.
The certified standard used for this type of calibration could be an NIST
SRM such as 2710 or 2711. The SRM must be a matrix similar to the samples
and must contain the analytes of interests at concentrations near those
expected in the samples. First, a response factor has to be determined
for each analyte. This factor is calculated by dividing the net peak
intensity by the analyte concentration. The net peak intensity is gross
intensity corrected for baseline interference. Concentrations of analytes
in samples are then determined by multiplying the baseline corrected
analyte signal intensity by the normalization factor and by the response
factor. The normalization- factor is the quotient of the baseline
corrected Compton Ko peak intensity of the SRM divided by that of the
samples. Depending on the FPXRF instrument used, these calculations may
be done manually or by the instrument software.
.0 PROCEDURE
11.1 Operation of the various FPXRF instruments will vary according to the
manufacturers' protocols. Before operating any FPXRF instrument, one
should consult the manufacturer's manual. Most manufacturers recommend
that their instruments be allowed to warm up for 15 to 30 minutes before
analysis of samples. This will help alleviate drift or energy calibration
problems later on in analysis.
DRAFT
6200 - 22
Revision 0
July 1996

-------
11.2	Each FPXRF instrument should be operated according to the manufacturer's
recommendations. There are two modes in which FPXRF instruments can be
operated: in situ and intrusive. The in situ mode involves analysis of
an undisturbed soil sediment or sample. Intrusive analysis involves
collection and preparation of a soil or sediment sample before analysis.
Some FPXRF instruments can operate in both modes of analysis, while others
are designed to operate in only one mode. The two modes of analysis are
discussed below.
11.3	For in situ analysis, one requirement is that any large or
nonrepresentative debris be removed from the soil surface before analysis.
This debris includes rocks, pebbles, leaves, vegetation, roots, and
concrete. Another requirement is that the soil surface be as smooth as
possible so that the probe window will have good contact with the surface.
This may require some leveling of the surface with a stainless-steel
trowel. During the study conducted to provide data for this method, this
modest amount of sample preparation was found to take less than 5 minutes
per sample location. The last requirement is that the soil or sediment
not be saturated with water. Manufacturers state that their FPXRF
instruments will perform adequately for soils with moisture contents of 5
to 20 percent but will not perform well for saturated soils, especially if
ponded water exists on the surface. Another recommended technique for in
situ analysis is to tamp the soil to increase soil density and compactness
for better repeatability and representativeness. This condition is
especially important for heavy element analysis, such as barium. Source
count times for in situ analysis usually range from 30 to 120 seconds, but
source count times will vary among instruments and depending on required
detection limits.
11.4	For intrusive analysis of surface or sediment, it is recommended that a
sample be collected from a 4- by 4-inch square that is 1 inch deep. This
will produce a soil sample of approximately 375 grams or 250 cm1, which is
enough soil to fill an 8-ounce jar. The sample should be homogenized,
dried, and ground before analysis. The sample can be homogenized before
or after drying. The homogenization technique to be used after drying is
discussed in Section 5.2. If the sample is homogenized before drying, it
should be thoroughly mixed in a beaker or similar container, or if the
sample is moist and has a high clay content, it can be kneaded in a
plastic bag. One way to monitor homogenization when the sample is kneaded
in a plastic bag is to add sodium fluorescein dye to the sample. After
the moist sample has been homogenized, it is examined under an ultraviolet
light to assess the distribution of sodium fluorescein throughout the
sample. If the fluorescent dye is evenly distributed in the sample,
homogenization is considered complete; if the dye is not evenly
DRAFT
6200 - 23
Revision 0
July 1996

-------
distributed, mixing should continue until the sample has been thoroughly
homogenized. During the study conducted to provide data for this method,
the homogenization procedure using the fluorescein dye required 3 to 5
minutes per sample.
11.5	Once the soil or sediment sample has been homogenized, it should be dried.
This can be accomplished with a toaster oven or convection oven. A small
aliquot of the sample (20 to 50 grams) is placed in a suitable container
for drying. The sample should be dried for 2 to 4 hours in the convection
or toaster oven at a temperature not greater than 150°C. Microwave drying
is not a recommended procedure. Field studies have shown that microwave
drying can increase variability between the FPXRF data and confirmatory
analysis. High levels of metals in a sample can cause arcing in the
microwave oven, and sometimes slag forms in the sample. Microwave oven
drying can also melt plastic containers used to hold the sample.
11.6	The homogenized dried sample material should be ground with a mortar and
pestle and passed through a 60-mesh sieve to achieve a uniform particle
size. Sample grinding should continue until at least 90 percent of the
original sample passes through the sieve. The grinding step normally
takes an average of 10 minutes per sample. An aliquot of the sieved
sample should then be placed in a 31.0-mm polyethylene sample cup (or
equivalent) for analysis. The sample cup should be one-half to three-
quarters full at a minimum. The sample cup should be covered with a 2.5
um Mylar (or equivalent) film for analysis. The rest of the soil sample
should be placed in a jar, labeled, and archived for possible confirmation
analysis. All equipment including the mortar, pestle, and sieves must be
thoroughly cleaned so that any cross-contamination is below the MDLs of
the procedure or DQOs of the analysis (Reference 16.18).
12.0 DATA ANALYSIS AND CALCULATIONS
Most FPXRF instruments have software capable of storing all analytical results and
spectra. The results are displayed in parts per million and can be downloaded to a PC,
which can provide a hard copy printout. Individual measurements that are smaller than
three times their associated SD should not be used for quantitation.
13.0 METHOD PERFORMANCE
13.1 This section discusses four performance factors, field-based method
detection limits, precision, accuracy, and comparability to EPA-approved
methods. The numbers presented in the six tables in this section were
generated from data obtained from six FPXRF instruments. The soil samples
analyzed by the six FPXRF instruments were collected from two sites in the
DRAFT
6200 - 24
Revision 0
July 1996

-------
United States. The soil samples contained several of the target analytes
at concentrations ranging from nondetect to tens of thousands of rag/kg.
13.2	The six FPXRF instruments included the TN 9000 and TN Lead Analyzer
manufactured by TN Spectrace; the X-MET 920 with a Si Li detector and X-MET
920 with a gas-filled proportional detector manufactured by Metorex, Inc.;
the XL Spectrum Analyzer manufactured by Niton; and the MAP Spectrum
Analyzer manufactured by Scitec. The TN 9000 and TN Lead Analyzer both
have a Hgl2 detector. The TN 9000 utilized an Fe-55, Cd-109, and Am-241
source. The TN Lead Analyzer had only a Cd-109 source. The X-Met 920
with the Sill detector had a Cd-109 and Am-241 source. The X-MET 920 with
the gas-filled proportional detector had only a Cd-109 source. The XL
Spectrum Analyzer utilized a silicon pin-diode detector and a Cd-109
source. The MAP Spectrum Analyzer utilized a solid-state silicon detector
and a Cd-109 source.
13.3	All data presented in Tables 3 through 8 were generated using the
following calibrations and source count times. The TN 9000 and TN Lead
Analyzer were calibrated using fundamental parameters using NIST SRM 2710
as a calibration check sample. The TN 9000 was operated using 100, 60,
and 60 second count times for the Cd-109, Fe-55, and Am-241 sources,
respectively. The TN Lead analyzer was operated using a 60 second count
time for the Cd-109 source. The X-MET 920 with the Si(Li) detector was
calibrated using fundamental parameters and one well characterized site-
specific soil standard as a calibration check. It used 140 and 100 second
count times for the Cd-109 and Am-241 sources, respectively. The X-MET
920 with the gas-filled proportional detector was calibrated empirically
using between 10 and 20 well characterized site-specific soil standards.
It used 120 second times for the Cd-109 source. The XL Spectrum Analyzer
utilized NIST SRM 2710 for calibration and the Compton peak normalization
procedure for quantitation based on 60 second count times for the Cd-109
source. The MAP Spectrum Analyzer was internally calibrated by the
manufacturer. The calibration was checked using a well-characterized
site-specific soil standard. It used 240 second times for the Cd-109
source.
13.4	Field-Based Method Detection Limits: The field based method detection
limits are presented in Table 3. The field-based method detection limits
were determined by collecting ten replicate measurements on site-specific
soil samples with metals concentrations 2 to 5 times the expected method
detection limits. Based on these ten replicate measurements, a standard
deviation on the replicate analysis was calculated. The method detection
limits presented in Table 3 are defined as 3 times the standard deviation
for each analyte.
DRAFT
6200 - 25
Revision 0
July 1995

-------
The field-based method detection limits were generated by using the count
times discussed earlier in this section. All the field-based method
detection limits were calculated for soil samples that had been dried and
ground and placed in a sample cup with the exception of the MAP Spectrum
Analyzer. This instrument can only be operated in the in situ mode,
meaning the samples were moist and not ground.
Some of the analytes such as cadmium, mercury, silver, selenium, and
thorium were not detected or only detected at very low concentrations such
that a field-based method detection limit could not be determined. These
analytes are not presented in Table 3. Other analytes such as calcium,
iron, potassium, and titanium were only found at high concentrations
(thousands of mg/kg) so that reasonable method detection limits could not
be calculated. These analytes also are not presented in Table 3.
13.5 Precision Measurements: The precision data is presented in Table 4. Each
of the six FPXRF instruments performed 10 replicate measurements on 12
soil samples that had analyte concentrations ranging from nondetects to
thousands of mg/kg. Each of the 12 soil samples underwent 4 different
preparation techniques from in situ (no preparation) to dried and ground
in a sample cup. Therefore, there were 48 precision data points for five
of the instruments and 24 precision points for the MAP Spectrum Analyzer.
The replicate measurements were taken using the source count times
discussed at the beginning of this section.
For each detectable analyte in each precision sample a mean concentration,
standard deviation, and RSD was calculated for each analyte. The data
presented in Table 4 is an average RSD for the precision samples that had
analyte concentrations at 5 to 10 times the MDL for that analyte for each
instrument. Some analytes such-as mercury, selenium, silver, and thorium
were not detected in any of the precision samples so these analytes are
not listed in Table 4. Some analytes such as cadmium, nickel, and tin
were only detected at concentrations near the MDLs so that an RSD value
calculated at 5 to 10 times the MDL was not possible.
One FPXRF instrument collected replicate measurements on an additional
nine soil samples to provide a better assessment of the effect of sample
preparation on precision. Table 5 shows these results. The additional
nine soil samples were comprised of three from each texture and had
analyte concentrations ranging from near the detection limit of the FPXRF
analyzer to thousands of mg/kg. The FPXRF analyzer only collected
replicate measurements from three of the preparation methods; no
measurements were collected from the in situ homogenized samples. The
FPXRF analyzer conducted five replicate measurements of the in situ field
DRAFT
6200 - 26
Revision 0
July 1996

-------
samples by taking measurements at five different points within the 4-inch
by 4-inch sample square. Ten replicate measurements were collected for
both the intrusive undried and unground and intrusive dried and ground
samples contained in cups. The cups were shaken between each replicate
measurement.
Table 5 shows that the precision dramatically improved from the in situ to
the intrusive measurements. In general there was a slight improvement in
precision when the sample was dried and ground. Two factors caused the
precision for the in situ measurements to be poorer. The major factor is
soil heterogeneity. By moving the probe within the 4-inch by 4-inch
square, measurements of different soil samples were actually taking place
within the square. Table 5 illustrates the dominant effect of soil
heterogeneity. It overwhelmed instrument precision when the FPXRF
analyzer was used 1n this mode. The second factor that caused the RSD
values to be higher for the in situ measurements is the fact that only
five versus ten replicates were taken. A lesser number of measurements
caused the standard deviation to be larger which in turn elevated the RSD
values.
13.6 Accuracy Measurements: Five of the FPXRF instruments (not including the
MAP Spectrum Analyzer) analyzed 18 SRMs using the source count times and
calibration methods given at the beginning of this section. The 18 SRMs
included 9 soil SRMs, 4 stream or river sediment SRMs, 2 sludge SRMs, and
3 ash SRMs. Each of the SRMs contained known concentrations of certain
target analytes. A percent recovery was calculated for each analyte in
each SRM for each FPXRF instrument. Table 6 presents a summary of this
data. With the exception of cadmium, chromium, and nickel, the values
presented in Table 6 were generated from the 13 soil and sediment SRMs
only. The 2 sludge and 3 ash SRMs were included for cadmium, chromium,
and nickel because of the low or nondetectable concentrations of these
three analytes in the soil and sediment SRMs.
Only 11 analytes are presented in Table 6. These are the analytes that
are of environmental concern and provided a significant number of
detections in the SRMs for an accuracy assessment. No data is presented
for the X-MET 920 with the gas-filled proportional detector. This FPXRF
instrument was calibrated empirically using site-specific soil samples.
The percent recovery values from this instrument were very sporadic and
the data did not lend itself to presentation in Table 6.
Table 7 provides a more detailed summary of accuracy data for one FPXRF
instrument (TN 9000) for the 9 soil SRMs and 4 sediment SRMs. Table 7
shows the certified value, measured value, and percent recovery for five
DRAFT
6200 - 27
Revision 0
July 1996

-------
analytes. These analytes were chosen because they are of environmental
concern and were most prevalently certified for in the SRM and detected
by the FPXRF instrument. The first nine SRMs are soil and the last 4 SRMs
are sediment. Percent recoveries for the four NIST SRMs were often
between 90 and 110 percent for all analytes.
13.7 Comparability: Comparability refers to the confidence with which one data
set can be compared to another. In this case, FPXRF data generated from
a large study of six FPXRF instruments was compared to EPA SW-846 Methods
3050A and 6010A which are the standard soil extraction for metals and
analysis by inductively coupled plasma. An evaluation of comparability
was conducted by using linear regression analysis. Three factors were
determined using the linear regression. These factors were the y-
intercept, the slope of the line, and the coefficient of determination
(r2).
As part of the comparability assessment, the effects of soil type and
preparation methods were studied. Three soil types (textures) and four
preparation methods were examined during the study. The preparation
methods evaluated the cumulative effect of particle size, moisture, and
homogenization on comparability. Due to the large volume of data produced
during this study, linear regression data for six analytes from only one
FPXRF instrument is presented in Table 6. Similar trends in the data were
seen for all instruments.
Table 8 shows the regression parameters for the whole data set, broken out
by soil type, and by preparation method. The soil types are as follows:
soil l--sand; soil 2--loam; and soil 3--si1ty clay. The preparation
methods are as follows: preparation 1--in situ in the field; preparation
2--in situ, sample collected and homogenized; preparation 3--intrusive,
with sample in a sample cup but sample still wet and not ground; and
preparation 4--sample dried, ground, passed through a 40-mesh sieve, and
placed in sample cup.
For arsenic, copper, lead, and zinc, the comparability to the confirmatory
laboratory was excellent with r2 values ranging from 0.80 to 0.99 for all
six FPXRF instruments. The slopes of the regression lines for arsenic,
copper, lead, and zinc, were generally between 0.90 and 1.00 indicating
the data would need to be corrected very little or not at all to match the
confirmatory laboratory data. The r2 values and slopes of the regression
lines for barium and chromium were not as good as for the other for
analytes, indicating the data would have to be corrected to match the
confirmatory laboratory.
DRAFT
6200 - 28
Revision 0
July 1996

-------
Table 8 demonstrates that there was little effect of soil type on the
regression parameters for any of the six analytes. The only exceptions
were for barium in soil 1 and copper in soil 3. In both of these cases,
however, it is actually a concentration effect and not a soil effect
causing the poorer comparability. All barium and copper concentrations in
soil 1 and 3, respectively, were less than 350 mg/kg.
Table 8 shows there was a preparation effect on the regression parameters
for all six analytes. With the exception of chromium, the regression
parameters were primarily improved going from preparation 1 to preparation
2. In this step, the sample was removed from the soil surface, all large
debris was removed, and the sample was thoroughly homogenized. The
additional two preparation methods did little to improve the regression
parameters. This data indicates that homogenization is the most critical
factor when comparing the results. It is essential that the sample sent
to the confirmatory laboratory match the FPXRF sample as closely as
possible.
Section 11.0 of this method discusses the time necessary for each of the
sample preparation techniques. Based on the data quality objectives for
the project, an analyst must decide if it is worth the extra time required
to dry and grind the sample for small improvements in comparability.
Homogenization requires 3 to 5 minutes. Drying the sample requires one to
two hours. Grinding and sieving requires another 10 to 15 minutes per
sample. Lastly, when grinding and sieving 1s conducted, time must be
allotted to decontaminate the mortars, pestles, and sieves. Drying and
grinding the samples and decontamination procedures will often dictate
that an extra person be on site so that the analyst can keep up with the
sample collection crew. The cost of requiring an extra person on site to
prepare samples must be balanced with the gain in data quality and sample
throughput.
14.0 POLLUTION PREVENTION
14.1 Pollution prevention encompasses any technique that reduces or eliminates
the quantity and/or toxicity of waste at the point of generation.
Numerous opportunities for pollution prevention exist in laboratory
operation. The EPA has established a preferred hierarchy of environmental
management techniques that places pollution prevention as the management
option of first choice. Whenever feasible, laboratory personnel should
use pollution prevention techniques to address their waste generation.
When wastes cannot be feasibly reduced at the source, the Agency
recommends recycling as the next best option.
DRAFT
6200 - 29
Revision 0
July 1996

-------
14.2 For information about pollution prevention that may be applicable to
laboratories and research institutions consult Less is Better: Laboratory
Chemical management for Haste Reduction available from the American
Chemical Society's Department of Government Relations and Science Policy,
1155 16th Street N.W., Washington D.C. 20036, (202) 872-4477.
15.0 HASTE MANAGEMENT
The Environmental Protection Agency requires that laboratory waste management
practices be conducted consistent with all applicable rules and regulations. The
Agency urges laboratories to protect the air, water, and land by minimizing and
controlling all releases from hoods and bench operations, complying with the
letter and spirit of any sewer discharge permits and regulations, and by
complying with all solid and hazardous waste regulations, particularly the
hazardous waste identification rules and land disposal restrictions. For further
information on waste management, consult The Haste Management Manual for
Laboratory Personnel available from the American Chemical Society at the address
listed in Sec. 14.2.
16.0 REFERENCES
1.	Driscoll, J.N., and others. 1991. "A Multifunctional Portable X-Ray
Fluorescence Instrument for Measurement of Heavy Metals and Radioactivity at
Mixed Waste Sites." American Laboratory. Pages 25-36.
2.	Hewitt, A.D. 1994. "Screening for Metals by X-ray Fluorescence
Spectrometry/Response Factor/Compton K,, Peak Normalization Analysis." American
Environmental Laboratory. Pages 24-32.
3.	Hewitt, A.D. 1995. "Screening for Metals by Portable XRF Using Fundamental
Parameter Analysis and Single Reference Standard Calibration." Fourth
International Symposium on Field Screening Methods for Hazardous Waste and Toxic
Chemicals. Las Vegas, Nevada. February 22-24, 1995.
4.	Hewitt, A.D. 1995. "Rapid Screening of Metals Using Portable High-Resolution
X-Ray Fluorescence Spectrometers." U.S. Army Cold Regions Research and
Engineering Laboratory, Special Report 95-14.
5.	HNU Systems, Inc. 1990. SEFA-P XRF Analyzer Operator's Manual, Version 1.0.
June.
6.	Kane, J.S., S.A. Wilson, J. Lipinski, and L. Butler. 1993. "Leaching
Procedures: A Brief Review of Their Varied Uses and Their Application to
Selected Standard Reference Materials." American Environmental Laboratory. Pages
14-15. June.
7.	Metorex. X-MET 920 User's Manual.
8.	National Institute of Standards and Technology. 1993. "First Preliminary Draft
Recommendation on Portable and Transportable X-Ray Fluorescence Spectrometers for
Field Measurements of Hazardous Elemental Pollutants." March.
DRAFT	6200 - 30	Revision 0
July 1996

-------
9.	Piorek, S. 1994. "Modern, PC Based, High Resolution Portable EDXRF Analyzer
Offers Laboratory Performance for Field, In Situ Analysis of Environmental
Contaminants." Nuclear Instruments and Methods in Physics Research. Volume
A353. Pages 528-533.
10.	Piorek, S., and J.R. Pasmore. 1993. "Standardless, In Situ Analysis of Metallic
Contaminants in the Natural Environment With a PC-Based, High Resolution Portable
X-Ray Analyzer." Third International Symposium on Field Screening Methods for
Hazardous Haste and Toxic Chemicals. Las Vegas, Nevada. February 24-26, 1993.
Volume 2, Pages 1135-1151.
11.	PRC Environmental Management, Inc. (PRC). 1995. Final Demonstration Plan for
the Evaluation of Field Portable X-Ray Fluorescence Technologies.
12.	Shefsky, S. 1995. "Lead in Soil Analysis Using the NITON XL." Fourth
International Symposium on Field Screening Methods for Hazardous Waste and Toxic
Chemicals. Las Vegas, Nevada. February 22-24, 1995.
13.	Spectrace Instruments. 1994. Energy Dispersive X-ray Fluorescence Spectrometry:
An Introduction.
14.	TN Spectrace. Spectrace 9000 Field Portable/Benchtop XRF Training and
Applications Manual.
15.	EPA. 1993. "An X-Ray Fluorescence Survey of Lead Contaminated Residential Soils
in Leadville, Colorado: A Case Study." Environmental Monitoring Systems
Laboratory Office of Research and Development. Las Vegas, Nevada. EPA/600-R-
93/073. March.
16.	EPA Environmental Response Team. 1991. "Field-Portable X-Ray Fluorescence.
Quality Technical Information Bulletin. Volume 1, Number 4. May.
17.	EPA Environmental Response Team. 1992. Spectrace 9000 Field Portable X-Ray
Fluorescence Operating Procedure. SOP 1713.
18.	"Guidance for the Data Quality Objective Process," EPA QA/G-4, September 1994.
DRAFT	6200 - 31	Revision 0
July 1996

-------
17.0 TABLES. DIAGRAMS. FLOWCHARTS. AND VALIDATION DATA
TABLE 1
INTERFERENCE FREE DETECTION LIMITS
Analyte
Chemical
Abstract
Series Number
Detection Limit in
Quartz Sand
(milligrams per
kilogram)
Antimony (Sb)
7440-36-0
40
Arsenic (As)
7440-38-0
40
Barium (Ba)
7440-39-3
20
Cadmium (Cd)
7440-43-9
100
Calcium (Ca)
7440-70-2
70
Chromium (Cr)
7440-47-3
150
Cobalt (Co)
7440-48-4
60
Copper (Cu)
7440-50-8
50
Iron (Fe)
7439-89-6
60
Lead (Pb)
7439-92-1
20
Manganese (Mn)
7439-96-5
70
Mercury (Hg)
7439-97-6
30
Molybdenum (Mo)
7439-93-7
10
Nickel (Ni)
7440-02-0
50
Potassium (K)
7440-09-7
200
Rubidium (Rb)
7440-17-7
10
Selenium (Se)
7782-49-2
40
Silver (Ag)
7440-22-4
70
Strontium (Sr)
7440-24-6
10
Thallium (Tl)
7440-28-0
20
Thorium (Th)
7440-29-1
10
Tin (Sn)
7440-31-5
60
Titanium (Ti)
7440-32-6
50
Vanadium (V)
7440-62-2
50
Zinc (Zn)
7440-66-6
50
Zirconium (Zr)
7440-67-7
10
DRAFT
6200 - 32
Revision 0
July 1996

-------
TABLE 2
SUMMARY OF RADIOISOTOPE SOURCE CHARACTERISTICS
Source
Activity
(mCi)
Half-
Life
(Years)
Excitation
Energy
(keV)
Elemental Analysis Range
Fe-55
20 - 50
2.7
5.9
Sulfur to chromium K Lines
Molybdenum to barium L Lines
Cd-109
5 - 30
1.3
22.1 and 87.9
Calcium to rhodium K Lines
Tantalum to lead K Lines
Barium to uranium L Lines
Am-241
5 - 30
458
26.4 and 59.6
Copper to thulium K Lines
Tungsten to uranium L Lines
Cm-244
60 - 100
17.8
14.2
Titanium to selenium K Lines
Lanthanum to lead L Lines
DRAFT
6200 - 33
Revision 0
July 1996

-------
TABLE 3
FIELD-BASED METHOD DETECTION LIMITS (mg/kg)a
Analyte
Instrument
TN
9000
TN Lead
Analyzer
X-MET 920
(SiLi
Detector)
X-MET 920
(Gas-
Filled
Detector)
XL
Spectrum
Analyzer
MAP Spectrum
Analyzer
Antimony
55
NR
NR
NR
NR
NR
Arsenic
60
50
55
50
110
225
Barium
60
NR
30
400
NR
NR
Chromium
200
460
210
110
900
NR
Cobalt
330
NR
NR
NR
NR
NR
Copper
85
115
75
100
125
525
Lead
45
40
45
100
75
165
Manganese
240
340
NR
NR
NR
NR
Molybdenum
25
NR
NR
NR
30
NR
Nickel
100
NR
NA
NA
NA
NR
Rubidium
30
NR
NR
NR
45
NR
Strontium
35
NR
NR
NR
40
NR
Tin
85
NR
NR
NR
NR
NR
Zinc
80
95
70
NA
110
NA
Zirconium
40
NR
NR
NR
25
NR
Notes:
a MDLs are related to the total number of counts taken. See section 13.3
for count times used to generate this table.
NR Not reported by the FPXRF instrument.
NA Not applicable; analyte was reported by the FPXRF instrument but was not
at high enough concentrations for method detection limit to be
determined.
DRAFT
6200 - 34
Revision 0
July 1996

-------
TABLE 4
PRECISION
Analyte
Average Relative Standard Deviation for Each Instrument
at 5 to 10 Times the MDL
TN
9000
TN Lead
Analyzer
X-MET 920
(SiLi
Detector)
X-MET 920
(Gas-
Filled
Detector)
XL
Spectrum
Analyzer
MAP
Spectrum
Analyzer
Antimony
6.54
NR
NR
NR
NR
NR
Arsenic
5.33
4.11
3.23
1.91
12.47
6.68
Barium
4.02
NR
3.31
5.91
NR
NR
Cadmium
29.84"
NR
24.80"
NR
NR
NR
Calcium
2.16
NR
NR
NR
NR
NR
Chromium
22.25
25.78
22.72
3.91
30.25
NR
Cobalt
33.90
NR
NR
NR
NR
NR
Copper
7.03
9.11
8.49
9.12
12.77
14.86
Iron
1.78
1.67
1.55
NR
2.30
NR
Lead
6.45
5.93
5.05
7.56
6.97
12.16
Manganese
27.04
24.75
NR
NR
NR
NR
Molybdenum
6.95
NR
NR
NR
12.60
NR
Nickel
30.85"
NR
24.92"
20.92"
NA
NR
Potassium
3.90
NR
NR
NR
NR
NR
Rubidium
13.06
NR
NR
NR
32.69"
NR
Strontium
4.28
NR
NR
NR
8.86
NR
Tin
24.32"
NR
NR
NR
NR
NR
Titanium
4.87
NR
NR
NR
NR
NR
Zinc
7.27
7.48
4.26
2.28
10.95
0.83
Zirconium
3.58
NR
NR
NR
6.49
NR
Notes:
a These values are biased high because the concentration of these analytes
in the soil samples was near the detection limit for that particular
FPXRF instrument.
NR Not reported by the FPXRF instrument.
NA Not applicable; analyte was reported by the FPXRF instrument but was
below the method detection limit.
DRAFT
6200 - 35
Revision 0
July 1996

-------
TABLE 5
PRECISION AS AFFECTED BY SAMPLE PREPARATION
Analyte
Average Relative Standard Deviation for Each
Preparation Method
In Situ-Field
Intrusive—
Undried and
Unground
Intrusive-
Dried and
Ground
Antimony
30.1
15.0
14 . 4
Arsenic
22.5
5.36
3.76
Barium
17.3
3.38
2.90
Cadmium*
41.2
30.8
28.3
Calcium
17.5
1.68
1.24
Chromium
17.6
28.5
21.9
Cobalt
28.4
31.1
28.4
Copper
26.4
10.2
7.90
Iron
10.3
1.67
1.57
Lead
25.1
8.55
6.03
Manganese
40.5
12.3
13.0
Mercury
ND
ND
ND
Molybdenum
21.6
20.1
19.2
Nickel"
29.8
20.4
18.2
Potassium
18.6
3.04
2.57
Rubidium
29.8
16.2
18.9
Selenium
ND
20.2
19.5
Silver*
31.9
31.0
29.2
Strontium
15.2
3.38
3.98
Thallium
39.0
16.0
19.5
Thorium
NR
NR
NR
Tin
ND
14.1
15.3
Titanium
13.3
4.15
3.74
Vanadium
NR
NR
NR
Zinc
26.6
13.3
11.1
Zirconium
20.2
5.63
5.18
Notes:
*	These values may be biased high because the concentration of these analytes in
the soil samples was near the detection limit.
ND Not detected.
NR Not reported.
DRAFT
6200 - 36
Revision 0
July 1996

-------
TABLE 6
ACCURACY
Analyte
Instrument
TN 9000
TN Lead Analyzer
X-MET 920 (SiLi Detector)
XL Spectrum Analyzer
n
Range of
% Rec.
Mean
% Rec.
SD
n
Range of
% Rec.
Mean %
Rec.
SD
n
Range of
% Rec.
Mean
% Rec
SD
n
Range of
% Rec.
Mean */»
Rec.
SD
Sb
2
100-149
124.3
NA
	



_


_


_
_
As
5
68-115
92.8
17.3
5
44-105
83.4
23.2
4
9.7-91
47.7
39.7
5
38-535
189.8
206
Ba
9
98-198
135.3
36.9
	
_


9
18-848
168.2
262
_

_
_
Cd
2
99-129
114.3
NA
	
_

	
6
81-202
110.5
45.7
_


..
Cr
2
99-178
138.4
NA
__
_
_
	
7
22-273
143.1
93 8
3
98-625
279.2
300
Cu
8
61-140
95.0
28.8
6
38-107
79.1
27.0
11
10-210
111.8
72.1
8
95-480
203.0
147
Fe
6
78-155
103.7
26.1
6
89-159
102.3
28.6
6
48-94
80.4
16 2
6
26-187
108.6
52.9
Pb
11
66-138
98.9
19.2
11
68-131
97.4
18.4
12
23-94
72.7
20.9
13
80-234
107.3
39.9
Mn
4
81-104
93.1
9.70
3
92-152
113.1
33.8
_

__
_
_



Ni
3
99-122
109.8
12.0
__



_



3
57-123
87.5
33.5
Sr
8
110-178
132.6
23.8




_

__

7
86-209
125.1
39.5
Zn
11
41-130
94.3
24.0
10
81-133
100.0
19.7
12
46-181
106.6
34.7
11
31-199
94.6
42.5
Notes:
n	Number of samples that contained a certified value for the analyte and produced a detectable concentration from the FPXRF instrument.
SD	Standard deviation.
NA	Not applicable; only two data points, therefore, a SO was not calculated.
-	No data.
DRAFT
6200 - 37
Revision 0
July 1996

-------
TABLE 7
ACCURACY FOR TN 9000*
Standard
Reference
Material
Arsenic
Barium
Copper
Lead
Zinc
Cert.
Cone.
Meas.
Cone.
%Rec.
Cert.
Cone.
Meas.
Cone.
%Rec.
Cert.
Cone.
Meas.
Cone.
%Rec.
Cert.
Cone.
Meas.
Cone.
%Rec.
Cert.
Cone.
Meas.
Cone.
%Rec.
RTC CRM-
0 21
24.8
ND
MA
586
1135
193.5
4792
2908
60.7
144742
149947
103.6
546
224
40.9
RTC CRM-
020
397
429
92.5
22.3
ND
NA
753
583
77.4
5195
3444
66.3
3022
3916
129.6
BCR CRM
143R






131
105
80.5
180
206
114.8
1055
1043
99.0
BCR CRM
141






32.6
ND
NA
29.4
ND
NA
81.3
ND
NA
USGS GXR-2
25.0
ND
NA
2240
2946
131.5
76.0
106
140.2
690
742
107.6
530
596
112.4
VSGS GXR-6
330
294
88.9
1300
2581
198.5
66.0
ND
NA
101
80.9
80.1
118
ND
NA
NIST 2711
105
104
99.3
726
801
110.3
114
ND
NA
1162
1172
100.9
350
333
94.9
NIST 2710
626
722
115.4
707
782
110.6
2950
2834
96.1
5532
5420
98.0
6952
6476
93.2
NIST 2709
17.7
ND
NA
968
950
98.1
34. 6
NO
NA
18.9
ND
NA
106
98.5
93.0
NIST 2704
23.4
NO
NA
414
443
107.0
98.6
105
106.2
161
167
103.5
438
427
97.4
CNRC PACS-
1
211
143
67.7
—
77 2
NA
452
302
66.9
404
332
82.3
824
611
74.2
SARM-51
--
—
—
335
466
139.1
268
373
139.2
5200
7199
138.4
2200
2676
121.6
SARM-52
—
—
—
410
527
128.5
219
193
88.1
1200
1107
92.2
264
215
81.4
Notes:
*	All concentrations in milligrams per kilogram.
tRec. Percent recovery.
NO Not detected.
NA Not applicable.
No data.
DRAFT
6200 - 38
Revision 0
July 1996

-------
REGRESS10
Ml Data
Soil]
Soil 2
Soil 3
Prep 1
Prep 2
Prep 3
Prep 4
Arsenic
824
368
453
207
208
204
205
0.94
0.96
0.94
0.87
0.97
0.96
0.96
Int.
1.62
1.41
1.51
2.69
1.38
1.20
1.45
Slope
0.94
0.95
0.96
0.85
0.95
0.99
0.98
TABLE 8
» PARAMETERS FOR COMPARABILITY1
Barium
1255
393
462
400
312
315
315
313
0.71
0.05
0.56
0.85
0.64
0.67
0.78
0.81
Int.
60.3
42.6
30.2
44.7
53.7
64.6
64.6
58.9
Copper
Slope
0.54
0.11
0.66
0.59
0.55
0.52
0.53
0.55
984
385
463
136
256
246
236
246
0.93
0.94
0.92
0.46
0.87
0.96
0.97
0.96
Int
2.19
1.26
2.09
16.60
3.89
2.04
1.45
1.99

Lead
Zinc
Chromium

n
r1
Int.
Slope
n
r1
Int.
Slope
n
r1
Int.
Slope
Ml Data
1205
0.92
1.66
0.95
1103
0.89
1.86
0.95
280
0.70
64.6
0.42
Soil 1
357
0.94
1.41
0.96
329
0.93
1.78
0.93
—
—
—
—
Soil 2
451
0.93
1.62
0.97
423
0.85
2.57
0.90
—
—
—
—
Soil 3
397
0.90
2.40
0.90
351
0.90
1.70
0.98
186
0.66
38.9
0.50
Prep 1
305
0.80
2 88
0.86
286
0.79
3.16
0.87
105
0 80
66.1
0.43
Prep 2
298
0.97
1.41
0.96
272
0.95
1.86
0.93
77
0.51
813
0.36
Prep 3
302
0.98
1 26
0 99
274
0.93
1.32
1.00
49
0 73
53 7
0.45
Prep 4
300
0.96
1.38
1.00
271
0.94
1.41
1.01
49
0.75
31.6 j
0.56
Notes:
1	Log-transformed data
n	Number of data points
H	Coefficient of determination
Int.	Y-intercept
—	No applicable data
DRAFT
6200 - 39
Revision 0
July 1996

-------
Table of Contents
Section	ESflfi
Notice	jj
Foreword 	
Abstract 	jv
List of Figures	'*
List of Tables 	 x
List of Abbreviations and Acronyms	xi
Acknowledgments 	 xiii
1	Executive Summary	 1
2	Introduction 	 3
Demonstration Background, Purpose, and Objectives	 3
Principles of FPXRF Analysis	 3
Reference Methods 	 4
Site Selection		 6
Predemonstration Sampling	 8
Experimental Design 	 8
Qualitative Factors	 10
Quantitative Factors	 10
Evaluation of Analyzer Performance	 12
Deviations to the Demonstration Plan	 18
Sample Homogenization	 19
3	Reference Laboratory Results	 20
Reference Laboratory Methods		20
Reference Laboratory Quality Control		20
Quality Control Review of Reference Laboratory Data		22
Reference Laboratory Sample Receipt, Handling, and Storage Procedures 		22
Sample Holding Times		22
Initial and Continuing Calibrations 		22
Detection Limits 		23
Method Blank Samples 		23
Laboratory Control Samples		23
Predigestion Matrix Spike Samples		23
Postdigestion Matrix Spike Samples 		24
Predigestion Laboratory Duplicate Samples 		24
Postdigestion Laboratory
Duplicate Samples		24
Performance Evaluation Samples 		25
Standard Reference Material Samples 		25
draft

-------
Table of Contents (Continued)
Section	P-a9e
Data Review, Validation, and Reporting		25
Quality Assessment of Reference Laboratory Data 		26
Precision			26
Accuracy		26
Representativeness		30
Completeness 		30
Comparability		31
Use of Qualified Data for Statistical Analysis 			31
4	TN Lead Analyzer 	 35
Background		35
Operational Characteristics		35
Equipment and Accessories		35
Operation of the Analyzer		36
Description of the Technology Operator 		37
Training		37
Reliability		37
Health and Safety 		38
Cost		38
Performance Factors		38
Detection Limits 		38
Throughput		40
Drift		41
Intramethod Assessment		41
Blanks		41
Completeness 			41
Precision		41
Accuracy		42
Comparability 		44
Intermethod Assessment		48
5	TN 9000 	 56
Background		56
Operational Characteristics		56
Equipment and Accessories		56
Operation of the Analyzer		58
Description of the Technology Operator 		58
Training		58
Reliability		58
Health and Safety		59
Cost		59
Performance Factors		59
Detection Limits		60
Throughput		61
Drift		61
Intramethod Assessment		62
Blanks		62
Completeness 		62
Precision		62
Accuracy		63
Comparability		65
Intermethod Assessment		69
DRAFT

-------
Table of Contents (Continued)
6	Applications Assessment and Considerations		79
7	Developer's Comments	 		84
8	References		91
draft

-------
List of Figures
Figure	Page
2-1. Principle of Source Excited X-Ray Fluorescence	 5
2-2.	Sample Preparation and Analysis 	 9
2-3.	Linear and Log-Log Data Plots 		14
3-1.	Pre-and Postdigestion Duplicate Samples 		27
3-2. Reference Method PE and CRM Results		28
3-3.	Reference Method SRM Results		33
4-1.	Critical Zone for the Determination of a Field-Based Method Detection Limit for Copper		41
4-2. Drift Summary—TN Lead Analyzer		42
4-4.	Precision vs. Concentration—TN Lead Analyzer		44
4-3.	Precision Summary—TN Lead Analyzer		44
4-5.	SRM Results—TN Lead Analyzer		46
4-6.	Site-Specific PE Sample Results—TN Lead Analzyer		47
4-7.	PE and CRM Results—TN Lead Analyzer		50
4-8.	Sample Preparation Effect on Lead and Arsenic Results 		52
4-9.	Highest Degree of Data Quality at the Lowest Degree of Sample Preparation—TN Lead Analyzer	53
5-1.	Critical Zone for the Determination of a Field-Based Method Detection Limit for Copper		62
5-2.	Drift Summary—TN 9000 		63
5-3.	Precision vs. Concentration for Lead and Copper—TN 9000 		65
5-4.	Site-Specific PE Sample Results—TN9000 	 		67
5-5.	SRM Results—TN 9000 		68
5-6.	PE and CRM Sample Results—TN 9000 		70
5-7.	Sample Preparation Effect on Arsenic and Lead Results 		73
5-8.	Highest Degree of Data Quality at the Lowest Degree of Sample Preparation—TN 9000 		74
DPAs-T

-------
List of Tables
Table	E3SS
2-1. Performance and Comparability Variables Evaluated 		11
2-2.	Criteria for Characterizing Data Quality		15
3-1.	Reference Laboratory Quality Control Parameters 		21
3-2. SW-846 Method 601 OA LRLs for Target Analytes		23
3-3. Reference Laboratory Accuracy Data for Target Analytes 		30
3-4. SRM Performance Data for Target Analytes		32
3-5.	Leach Percent Recoveries for Select NIST SRMs		34
4-1.	Analyzer Instrument Specifications—TN Lead Analyzer		36
4-2. Incidental Items and Costs —TN Lead Analyzer		39
4-3. Relative Analytical Costs—TN Lead Analyzer Analysis		39
4-4. Method Detection Limits—TN Lead Analyzer 		40
4-5. Precision Summary—TN Lead Analyzer		43
4-6. Accuracy Summary for Site-Specific PE and SRM Results—TN Lead Analyzer 		45
4-7. Accuracy Summary for PE and CRM Results—TN Lead Analyzer		49
4-8. Regression Parameters by Variable—TN Lead Analyzer 	 		51
4-9. Regression Parameters by the Sample Preparation Variable and Soil Type—TN Lead Analyzer .	54
4-10.	Regression Parameters by the Sample Preparation Variable and Site Type—TN Lead Analyzer .	55
5-1.	Analyzer Instrument Specifications—TN 9000		57
5-2. Relative Analytical Costs—TN 9000 		60
5-3. Method Detection Limits—TN 9000 		61
5-4. Precision Summary—TN 9000		64
5-5. Accuracy Summary for Site-Specific PE and SRM Results—TN 9000 		66
5-6. PE and CRM Results—TN 9000 		69
5-7. Regression Parameters by Variable—TN 9000 		72
5-8. Regression Parameters for the Sample Preparation Variable and Soil Type—TN 9000 		75
5-9.	Regression Parameters for the Sample Preparation Variable and Site Type—TN 9000		77
6-1.	Advantages and Limitations—TN Lead Analyzer		80
6-2. Advantages and Limitations—TN 9000 		82
x

-------
List of Abbreviations and Acronyms
a
alpha
P
beta
AC
alternating current
Am241
americium-241
CaMP
Characterization and Monitoring Program
CCB
continuing calibration blank
CCV
continuing calibration verification
Cd109
cadmium-109
CI
confidence interval
CLP
Contract Laboratory Program
cm
centimeter
cm2
square centimeter
cm3
cubic centimeter
CRM
certified r- ference material
eV
electron volt
EPA
Environmental Protection Agency
ERA
Environmental Resource Associates
Fe55
iron-55
FP
fundamental parameter
FPXRF
field portable X-ray fluorescence
Hgi2
mercuric iodide
ICP-AES
inductively coupled plasma-atomic emission spectroscopy
ICAL
initial calibration
ICB
initial calibration blank
ICS
interference check standard
ICV
initial calibration verification
IDL
instrument detection limit
IDW
investigation-derived waste
ITER
innovative technology evaluation report
keV
kiloelectron volt
MS
microgram
LCS
laboratory control samples
I°9i0
base 10 logarithm
LRL
lower reporting limit
MCA
multichannel analyzer
mCi
milliCurie
MDL
method detection limit
xi
draft

-------
List of Abbreviations and Acronyms (Continued)
MRI
Midwest Research Institute
mg/kg
milligram per kilogram
mL
milliliter
mm
millimeter
mrem/hr
millirems per hour
NERL-CRD
National Exposure Research Laboratory—Characterization Research Laboratory
NIST
National Institute of Standards and Technology
NRMRL
National Risk Management Research Laboratory
OSW
Office of Solid Waste
PAL
performance acceptance limit
PARCC
precision, accuracy, representativeness, completeness, and comparability
PC
personal computer
PE
performance evaluation
PI
prediction interval
ppm
part per million
PRC
PRC Environmental Management, Inc.
QA
quality assurance
QAPP
quality assurance project plan
QC
quality control
r
correlation coefficient
r2
coefficient of determination
RCRA
Resource Conservation and Recovery Act
RPD
relative percent difference
RSD
relative standard deviation
RTC
Resource Technology Corporation
SD
standard deviation
SiOj
silicon dioxide
SITE
Superfund Innovative Technology Evaluation
SOP
standard operating procedure
SRM
standard reference material
TC
toxicity characteristic
pm
micrometer
USGS
United States Geological Survey
XRF
X-ray fluorescence
xii
D55 A £T
¦»' U i""L & &

-------
Acknowledgments
The U.S. Environmental Protection Agency (EPA) wishes to acknowledge the support of all those who helped plan
and conduct this demonstration, interpret data, and prepare this report. In particular, for demonstration site access
and relevant background information, Tom Aldridge (ASARCO) and Harold Abdo (RV Hopkins); for turn-key
implementation of this demonstration, Eric Hess, Patrick Splichal, and Harry Ellis (PRC Environmental Management,
Inc.) (913/281-2277); for editorial and publication support, Suzanne Ladish (PRC Environmental Management, Inc.);
for technical and peer review, Paula Hirtz and Alan Byrnes (PRC Environmental Management, Inc.); for analyzer
operation, Bryce Smith, TN 9000, and Robert Beilfuss, TN Lead Analyzer (PRC Environmental Management, Inc.);
for sample preparation, Scott Schulte, Keith Brown, and Curt Enos (PRC Environmental Management, Inc.); and for
EPA project management, Stephen Billets, National Exposure Research Laboratory-Characterization Research
Division (NERL-CRD). In addition, we gratefully acknowledge the participation of Oliver Fordham, EPA Office
of Solid Waste; Piper Peterson, EPA Region 10; Brian Mitchell, EPA Region 7; and Todd Rhea, Peter Berry, and
Raj Natarajan, TN Spectrace.

-------
Section 1
Executive Summary
In April 1995, a demonstration of field portable
X-ray fluorescence (FPXRF) analyzers took place to
evaluate how well they identify and quantify
concentrations of select environmentally toxic metals in
soils. The performance of each analyzer was evaluated
by comparing each analyzer's results to the results
obtained using reference methods considered to be
standards in the environmental industry. In addition,
standard reference materials (SRM) and performance
evaluation (PE) samples were also used to assess each
analyzer's accuracy and comparability.
The primary objectives of this demonstration were
to evaluate FPXRF analyzers for: (1) their accuracy and
precision relative to conventional analytical methods,
(2) the influence of sample matrix variations (texture,
moisture, heterogeneity, and chemical composition) on
their performance, (3) the logistical and economic
resources necessary to operate these technologies in the
field, and (4) create and validate an SW-846 draft
method for FPXRF analysis. Secondary objectives for
this demonstration were to evaluate FPXRF analyzers
for their (1) reliability, ruggedness, cost, and range of
usefulness, (2) data quality, and (3) ease of operation.
This demonstration was intended to provide potential
users with a known reference for comparison purposes
and as a guide for the application of this technology.
The reference methods for evaluating the comparability
of data were SW-846 Methods 3050A and 6010A, "Acid
Digestion of Sediments, Sludges, and Soils" and
"Inductively Coupled Plasma-Atomic Emission Spectro-
scopy (ICP-AES)."
The demonstration was conducted by PRC
Environmental Management, Inc. (PRC), for the
Environmental Protection Agency's (EPA) Superfund
Innovative Technology Evaluation (SITE) Program. The
demonstration was managed by the National Exposure
Research Laboratory-Characterization Research Division
(NERL-CRD) under the Characterization and
Monitoring Program (CaMP), Las Vegas, Nevada.
The FPXRF analyzers were designed to provide
rapid, real-time analysis of metals concentrations in soil
samples. This information will allow investigation and
remediation decisions to be made on site more
efficiently, and will reduce the number of samples that
need to be submined for costly confirmatory analysis.
Of the seven commercially available analyzers evaluated,
two are manufactured by TN Spectrace (the TN 9000
and TN Lead Analyzer); one is manufactured by Niton
Corporation (the Niton XL Spectrum Analyzer); two are
manufactured by Metorex Inc. (the X-MET 920-P
Analyzer and the X-MET 920-MP Analyzer); one is
manufactured by HNU Systems, Inc. (the SEFA-P
Analyzer); and one is manufactured by Scitec
Corporation (the MAP Spectrum Analyzer). The X-
MET 940, a prototype FPXRF analyzer developed by
Metorex, was given special consideration- and replaced
the X-MET 920-P for pan of the RV Hopkins sample
analyses. Instrument problems prevented the X-MET
940 from analyzing all the RV Hopkins site samples.
The X-MET 920-P and X-MET 940 are essentially the
same instrument, only their physical characteristics
differ. This ITER presents information relative to the
TN 9000 and TN Lead Analyzer. Separate ITERs will
be published for the other analyzers demonstrated.
The target analytes for this demonstration were
selected from the Resource Conservation and Recovery
Act's (RCRA) Toxicity Characteristic (TC) list, analytes
known to have a high aquatic toxicity, and analytes
likely to produce interferences for the FPXRF analyzers.
The primary analytes for these comparisons were
arsenic, barium, chromium, copper, lead, and zinc;
nickel, iron, cadmium, and antimony were secondary
analytes. Target analytes arsenic, barium, cadmium,
chromium, and lead were selected from the TC list.
Metals on the TC list but not included for analysis during
this demonstration were mercury, selenium, and silver.
These metals were not included because a demonstration
site could not be found or accessed that exhibited those
metals in a wide range of concentrations along with the
other TC list metals. The other target analytes copper,
zinc, nickel, iron, and antimony were selected due to
1
DP/5

-------
either their environmental toxicity or iheir potential to
cause interference for either the reference method of the
FPXRF analyzers.
To demonstrate the analyzers, hazardous waste sites
in Iowa (the RV Hopkins site) and in Washington (the
ASARCO site) were selected. The sites were selected
because they exhibit a wide range of concentrations for
most of the target metals, they are located in different
climaiological regions of the United States, and
combined, they exhibit three distinct soil textures.
Results of the demonstration axe summarized in
individual innovative technology evaluation reports
(ITER) for each anaiyzer. The purpose of this
specific ITER is to review the demonstration of both the
TN Lead Analyzer and the TN 9000, Iheir capabilities,
associated equipment and accessories, and to evaluate
how closely the results obtained using these two
analyzers compare to the results obtained using
conventional analytical methods.
This demonstration found that both TN Spectrace
analyzers were generally simple to operate in (he field.
The developer provided a training course for the
technology operators that was similar to that provided to
a purchaser of the equipment. The training encompassed
enough FPXRF theory and hands-on analyzer use to
allow operators to manipulate the data collection
software and to adjust instrument parameters such as
count times and target analytes. None of these
adjustments were made by the operators since they were
operating the analyzers under conditions determined by
the developer. Neither analyzer experienced an
operational failure resulting in project down time or data
loss during the analysis of over 1,260 soil samples. The
analyzers were portable, and they could operate
continuously over a 12-hour work day with appropriate
battery changes. The almost continuous rain en-
countered at one of the sites caused no operational
problems with either analyzer. Downloading of data to
both paper and electronic format was accomplished
without difficulty. All radiation levels detected while the
radioactive sources were exposed were far below
permissible levels for both analyzers.
The TN Lead Analyzer reports results for fewer
analytes than the TN 9000, Of the target analytes for
this demonstration, the TN Lead Analyzer reported lead,
arsenic, copper, chromium, zinc, and iron. Barium,
nickel, cadmium and antimony are the four target
analytes not reported by the TN Lead Analyzer. The
TN 9000 reported all of the target analytes for this
demonstration. While reponing fewer anaJytes. the TN
Lead Analyzer uses a single radioactive source and the
TN 9000 can use up to three unique radioactive sources.
The fact that the TN Lead Analyzer used only one
source and used shorter count times resulted in a four- to
five-fold increase in sample throughput for the TN Lead
Analyzer relative to the TN 9000. Both analyzers are
most cost effective when applied at sites where more
than 80 samples will be analyzed. Analytical costs for
these sized sites are almost 33 percent less than
conventional analysts and this difference increases
dramatically with increased sample size.
The TN Lead Anaiyzer produced data meeting Level
3 (equivalent to reference data) criteria for lead, arsenic,
and copper, but produced Level 2 (not equivalent to
reference data, but correctable with confirmatory sample
analysis) quality data for chromium, zinc, and still
exhibiting strong linear correlation to the reference data.
The TN 9000 provided Level 3 quality data for arsenic,
copper, lead, and nickel; and data of Level 2 quality for
barium, chromium, zinc, iron, antimony, and cadmium.
Both analyzers exhibited precision similar to the
reference methods at the 5 to 10 times the precision-
based method detection limit (MDL) concentration level.
The chromium data generally showed the lowest
precision of the primary analytes. With the exception of
antimony and cadmium, the field-based MDLs for these
analyzers were generally 2 to 3 times higher than
precision-based or developer-provided MDLs. Of the
four sample preparation steps evaluated, the initial
sample homogenization had the greatest impact on data
comparability for the FPXRFs. Site and soil texture did
not appear to affect data comparability. The process of
microwave drying of samples increased the relative
percent differences (RPD) between duplicates (for the
reference laboratory), relative to nonmicrowaved
samples. The increased RPD reduced data compar-
ability.
Based on the performance of both TN Spectrace
analyzers, this demonstration found them to be effective
tools for characterizing the concentration of select metals
in environmental soil samples.
2
DP 4

-------
Section 2
Introduction
The purpose of this ITER is to present information
on the demonstration of both the TN Lead Analyzer and
the TN 9000. These analyzers were developed by TN
Spectrace to perform elemental analyses (metals
quantitation) in the petroleum and petrochemical
industry, the mining and minerals industry, and the
environmental field. These FPXRF analyzers use a
mercuric iodide (Hgl2) detector with radioactive sources
to quantitate metals concentrations. Both analyzers can
be operated in either an in situ or intrusive mode. The
in situ mode is commonly called a "point-and-shoot"
mode. In this mode of operation, the point of
measurement on the soil surface is cleared of loose
debris and organic matter, the analyzer's probe is placed
in direct contact with the soil surface and a measurement
is taken. In the intrusive mode of operation, a soil
sample is physically collected, possibly dried or sieved,
and then placed into an FPXRF sample cup. The sample
cup is placed into an analysis chamber on the probe and
a measurement is taken.
This section summarizes general information about
the demonstration such as the purpose, objectives, and
design. Section 3 presents and discusses the quality of
the data produced by the reference methods against
which both analyzers were evaluated. Section 4
discusses the TN Lead Analyzer, its capabilities,
equipment, accessories, accuracy, precision, and its
comparability to reference methods. Section 5 discusses
the TN 9000, its capabilities, equipment, accessories,
accuracy, precision, and its comparability to reference
methods. Section 6 discusses the potential applications
of both analyzers and a suggested framework for a
standard operating procedure (SOP). Section 7 lists
references cited in this text. Section 8 presents
developer comments, EPA responses to developer
comments, and a developer update on the technology.
Demonstration Background, Purpose,
and Objectives
The demonstration was conducted under CaMP, a
component of the SITE Program. CaMP is managed by
NERL-CRD, Las Vegas, Nevada. The goal of the
CaMP is to identify and demonstrate new, innovative,
and commercially available technologies that can sample,
identify, quantify, or monitor changes in contaminants at
hazardous waste sites or that can be used to determine
the physical characteristics of a site more economically,
efficiently, and safely than conventional technologies.
The SITE Program is administered by the National'Risk
Management Laboratory (NRML), Cincinnati, Ohio.
The purpose of this demonstration was to provide
the information needed to fairly and thoroughly evaluate
how well FPXRF analyzers identify and quantify
concentrations of metals in soils. The primary objectives
of this demonstration were to evaluate FPXRF analyzers
in the following areas: (1) their accuracy and precision
relative to conventional analytical methods, (2) the
influence of sample matrix variations (texture, moisture,
heterogeneity, and chemical composition) on their
performances, (3) the logistical and economical
resources necessary to operate these analyzers, and (4)
to create and validate an SW-846 draft method for
FPXRF analysis.
Secondary objectives for this demonstration were to
evaluate FPXRF analyzers for their (1) reliability,
ruggedness, cost, and range of usefulness, (2) data
quality, and (3) ease of operation. The performance of
the FPXRF analyzers were not compared against each
other. Their performances were independently
compared to the performance of conventional analytical
methods commonly used in regulatory enforcement or
compliance activities. In addition, the analyzer's
performance was assessed relative to measurements of
SRMs, PE samples, and other quality control (QC)
samples.
Principles of FPXRF Analysis
FPXRF analyzers operate on the principle of energy
dispersive X-ray fluorescence (XRF) spectrometry
whereby the characteristic energy components of the
excited X.-ray spectrum are analyzed directly via their
3
JRAH

-------
energy proportional response in the X-ray detector.
Compared to the traditional methods of wavelength
diffraction, energy dispersion affords highly efficient full
spectrum measurement which enables the use of low
intensity excitation sources (such as radioisotopes) and
compact design battery-powered field portable
instruments. Many FPXRF instrument designs based on
various energy dispersive detecioT technologies are now
widely used for composition analysis in the industrial
and environmental arena. Such applications can also
make the best use of the essentially nondestructive nature
of the XRF measurement technique. Mainly it is
sufficient that the sample be homogeneous or at least
prepared to the extent that it is effectively homogeneous
on the scale of the X ray penetration. Typical X ray
penetration depths might range from about 0.1 to 1
millimeter (mm) for the X rays of most targeted metal
contaminants in the environmental samples.
Fluorescent X rays are produced by exposing a
sample to an X-ray source having an excitation energy
close to, but greater than, the binding energy of the inner
shell electrons of the elements in a sample, inner shell
electrons are displaced to higher energy orbitals. The
electron vacancies that result are filled by electrons
cascading in from outer electron shells. Electrons in
outer shells have higher energy states than inner shell
electrons, and to fill the vacancies, the outer shell
electrons must give off energy as they cascade into the
inner shell vacancies (Figure 2-1). This release of
energy results in an emission of X rays that are
characteristic to each element. This emission of X rays
is termed X-ray fluorescence.
Because of different electron shell configurations,.
each element emits a unique X ray at set wavelengths
called "characteristic" X rays. The energy of the X ray
is measured in electron volts (eV). By measuring the
peak energies of X rays emitted by a sample, it is
possible to identify and quantify the elemental
composition of a sample. A qualitative analysis of the
samples can be made by identifying the characteristic X
rays produced by the sample. The intensity of each
characteristic X ray emitted is proportional to the
concentration of a given element, and can be measured
to quantitate element concentrations.
Three electron shells are generally involved in
emission of characteristic X rays during FPXRF analysis
of environmental samples: the K, L, and M shells. A
typical emission pattern, also called an emission
spectrum, for a given element has multiple peaks
generated from the emission X rays by the K, L, or M
shell electrons. The most commonly measured X-ray
emissions are from the K and L shells; only elements
with an atomic number of 58 (cerium) or greater have
measurable M shell emissions.
Each characteristic X-ray peak or line, as u is often
called, is defined with the letter K, L, or M. which
signifies which shell had the original vacancy and by a
subscript alpha (a) or beta (Pi), which indicates the
higher shell from which electrons fell to fill the vacancy
and produce the X ray. For example, a Ka-line is
produced by a vacancy in the K shell filled by an L shell
electron, whereas a K^-line is produced by a vacancy in
the K shell filled by an M shell electron. The K0
transition is 10 times more probable than the Kc
transition; therefore^ the Ka-line is approximately
10 times more intense than the K^-lme for a given
analytes, making the Ks-line analysis the preferred
choice for quantitation purposes.
The K-lines for a given analyte are the most
energetic lines and are the preferred lines for analysis.
For a given atom, the X rays emitted from L transitions
are always less energetic than those emitted from K
transitions. Unlike the K-lines, the L-lines (Ln and Ln)
for an analyte are of nearly equal intensity. The choice
of which one to use for analysis depends on the presence
of interfering lines from other analytes. The L-lines are
useful for analyses involving analytes of atomic number
58 (cerium) through 92 (uranium).
An X-ray source can excite characteristic X rays
from an analyte only if it's energy is greater than the
electron binding energies for the target analyte.
The electron binding energy is also known as the
absorption edge energy. The absorption edge energy
represents the amount of energy an electron can absorb
before it is elevated to a higher orbital shell. The
absorption edge energy is somewhat greater than the
corresponding iine energy. Actually, the K-absorption
edge energy is approximately the sum of the K„, L^, and
Ma line energies of the particular element, and the L-
absorption edge energy is approximately the sum of the
L- and M-line energies. FPXRF analytical methods are
more sensitive to analytes with absorption edge energies
close to, but less than, the excitation energy of the
source. For example, when using a cadmium-109
(Cd109) source, which has an excitation energy of 22.1
kiloelectron volts (keV), an FPXRF analyzer would be
more sensitive to zirconium, which has a K-line
absorption edge energy of 15.7 keV. than to chromium,
which has a K-line absorption edge energy of s 4lJr$V
Reference Methods
To assess the performance of the FPXRF analyzers,
FPXRF data was compared to reference data. The
reference methods used for this assessment were
Methods 3050A/6010A, which are considered the
standard for metals analysis in soil for environmental
applications. The term "reference" was substituted
17.49

4
DR£*~T

-------
Excitation X ray tram the
FPXRF Source
K Shea Electrons
An excited electron is displaced. creating an
electron vacancy.
Nucleus
An outer electron then electron cascades to the mer electron
shell to (ill the vacancy. As this electron cascades. d releases
energy m the lortn ol an X ray.
L Shell Electrons
Characteristic X ray
FIGURE 2-1. PRINCIPLE OF SOURCE EXCITED X-R-
source excited X-ray fluorescence.
for "confirmatory" since the data was used as a baseline
for comparison. Midwest Research Institute (MRI) was
awarded the subcontract to analyze soil samples using
the reference methods in accordance with Federal
Acquisition Requirements (FAR). The award was made
based on MRI's costs, ability to meet the
demonstration's quality assurance project plan (QAPP),
and its position as the only commercial laboratory
identified that could perform all the required analyses.
A special request was made by Mr. Oliver
Fordham, technical advisor, EPA OSW, for MRI to
analyze some of the soil samples to validate the
performance of draft Method 3052 "Microwave Assisted
Acid Digestion of Ash and Other Siliceous Wastes."
Thirty percent of the soil samples were extracted using
draft Method 3052, and then analyzed by Method
6010A. The data generated from the draft Method 3052
and Method 6010A analysis were not used for
comparative purposes to the FPXRF data in this
demonstration.
Method 3050A is the standard acid extraction
method used for determining metals concentrations in
environmental soil samples. It is not a total digestion
method, and potentially does not extract all the metals in
a soil sample. Method 6010A is the standard method
used to analyze Method 3050A extracts.
FLUORESCENCE: This figure illustrates the dynamics of
High quality, well documented reference laboratory
results were essential for meeting the purposes and
objectives of the demonstration. For a true and accurate
assessment, the reference methods had to provide a
known level of data quality. For all measurement and
monitoring activities conducted by EPA, the Agency
requires that data quality parameters be established based
on the end uses of the data. Data quality parameters
include five indicators of data quality referred to as the
PARCC parameters: precision, accuracy, represen-
tativeness, completeness, and comparability. In ad-
dition, MDLs are often used to assess data quality.
Reference methods were evaluated using the
PARCC parameters to establish the quality of data
generated and to ensure that the comparison of FPXRF
analyzers to reference methods was true and accurate.
The following paragraphs provide definitions of each of
the PARCC parameters used to evaluate the reference
data.
Precision refers to the degree of mutual agreement
among duplicate measurements and provides an estimate
of random error. Precision for the reference methods
were expressed in terms of RPDs of laboratory duplicate
samples.
Accuracy refers to the difference between a sample
result and the reference or true value. Bias, a measure
5
3RAFT

-------
of the departure from the complete accuracy, can be
estimated from the reference or true value. Accuracy
and bias for the reference laboratory were assessed
through an evaluation of calibration standard linearity
and method blank results: and percent recoveries of
matrix spike samples, laboratory control samples (LCS),
and PE samples.
Representativeness refers to the degree to which
data accurately and precisely measures the conditions
and characteristics of the parameter of interest
represented by the data. Representativeness for the
reference laboratory was ensured by executing consistent
sample collection procedures including sample locations,
sampling procedures, sample storage, sample packaging,
sample shipping, sampling equipment decontamination,
and proper laboratory sample handling procedures.
Representativeness was ensured by using each reference
method at its optimum capability to provide results that
represented the most accurate and precise measurement
it was capable of achieving. The combination of the
existing method requirements supplemented by the
demonstration QAPP provided the guidance to assure
optimum performance of the method.
Representativeness was assessed through an evaluation
of QC samples including: calibration standards, method
blank samples, duplicate samples, and PE samples.
Completeness refers to the amount of data collected
from a measurement process compared to the amount
that was expected to be obtained. For the reference
data, completeness referred to the proportion of valid,
acceptable data generated.
Comparability refers to the confidence with which
one data set can be compared to another. Data generated
from the reference methods should provide comparable
data to any other laboratory performing analysis of the
same samples with the same analytical methods.
Comparability for the reference methods was achieved
through the use of SOPs, EPA-published analytical
methods, and the demonstration QAPP. QC samples
that were used to evaluate comparability include:
calibration standards, method blank samples, matrix
spike samples, duplicate samples, LCSs, and PE
samples.
Site Selection
PRC conducted a search for suitable demonstration
sites between September and November 1994. The
following criteria were used to select appropriate sites:
• The site owner had to agree to allow access for the
demonstration.
•	The site had to have soil contaminated with some or
all of the target heavy metals. (Slag, ash. and other
mineralized metals deposits would not be assessed
during the demonstration.)
•	The site had to be accessible to two-wheel drive
vehicles.
•	The site had to exhibit one or more of the following
soil texrures: sand, clay, or loam.
•	The site had to exhibit surface soil contamination.
•	The sites had to be situated in different climaco-
logical environments.
PRC contacted NERL, regional EPA offices, state
environmental agencies, and metals fabrication and
smelting contacts to create an initial list of potential
demonstration sites. PRC received considerable
assistance from EPA RCRA and Superfund Branches in
Regions 4, 6, 7, 8, 9, and 10. PRC also contacted the
Montana Department of Health and Environment, the
Nevada Bureau of Mines and Geology, Oklahoma
Department of Environmental Quality, the Arizona
Department of Environmental Quality, the Missouri
Department of Natural Resources, the Arizona Bureau of
Geology, and the New Mexico Bureau of Mines and
Mineral Resources. PRC surveyed its offices in Kansas
City, Kansas; Atlanta, Georgia; Denver, Colorado;
Dallas, Texas; Albuquerque, New Mexico; Helena.
Montana; Chicago, Illinois; Seattle, Washington; and
San Francisco, California, for information regarding
potential sites. These PRC offices have existing RCRA,
Superfund, or Navy environmental contracts that allow
access to regional, state and federal site information.
PRC also used the Record of Decision Scan (Morgan,
Lewis, and Bockius 1993) data base to search for
appropriate sites. Through these contacts, 46 individual
hazardous waste sites were identified.
Screening of these 46 candidate sites was based on
the site-selection criteria listed above and the assistance
of the various state and federal agencies and others listed
above. Based on this screening, it was determined that
the RV Hopkins and ASARCO sites met the most site
selection criteria, and would be the sites used for the
demonstration.
The ASARCO site is the location of a former lead
and copper smelter situated on the shore of
Commencement Bay in Tacoma, Washington. Prior to
1890, sawmills were operated at the location of the
ASARCO site. Lead smelting and refining operations
began at this site in 1890 and continued to 1912. In
1905, ASARCO purchased the property and continued
6

-------
the smelting operations. In 1912, ASARCO converted
the operation to a copper refining and smelting
operation. ASARCO further refined the by-products of
the smelting operation into arsenic, sulfuric acid, liquid
sulfur dioxide, and slag. AH smehing-related operations
were discontinued at this site in 1985.
The ASARCO site is pan of the Commencement
Bay Nearshore/Tideflais Superfund site in Tacoma,
Washington. The Commencement Bay site was placed
on the National Priorities list (NPL) in September 1983.
In September 1986, ASARCO and EPA Region 10
entered inio an Administrative Order on Consent in
which ASARCO agreed to conduct a remedial
investigation and feasibility study. Tbe remedial
investigation was completed in 1992; the feasibility study
was completed in 1994.
The ASARCO site is located on the Point of
Defiance peninsula in the municipalities of Ruston
and Tacoma, Washington. The site consists of 67 acres
of land adjacent to Commencement Bay. The site is
marked by steep slopes leading into the bay, slag fill that
was used to extend the original shoreline, a cooling
water pond, and the various buildings associated with the
smelting process. Partial facility demolition was
conducted in 1987. Most of the buildings were
demolished between 1993 and 1994. The only buildings
remaining are the Fine Ore Building, the Administrative
Building, and a Maintenance Garage.
Past soil sampling targeted four general areas of
metals contamination at the site: the plant administration
area, the former cooling pond, the 1987 demolition area,
and the off-site residential areas adjacent to the smelter
stack. Previous sampling has shown surficial soils to be
more contaminated than subsurface soils. Arsenic,
copper, and lead are the predominant contaminants in the
local soils. The highest arsenic conceiurations were
found in the soils around the former arsenic kitchen,
along with cadmium and mercury. The soils around the
former cooling pond contained the highest copper
concentrations, and high levels of silver, selenium,
barium, and chromium. Lead concentrations are highest
northeast of the arsenic plant.
Much of the site is covered with artificial fill
material of varying thickness and composition. Two
general types of fill are found on site: granular fill and
massive slag fill. The composition of the granular fill
material ranges from sand to silty, sandy gravel with
demolition debris and slag debris intermixed throughout.
The massive slag fill is a solid, fractured media
restricted to the plant site. The surface soil in the plant
administration area has a layer of slag particles on top,
ranging from 1 to 3 inches thick. Surficial material in
the parking lot area and southwest of the stack is mostly
of glacial origin and composed of various mixtures of
sand, gravel, and cobbles. The soils around the former
cooling pond are fine-grained lacustrine silts and clays.
Alluvium in the drainage upgradient of the former
cooling pond has been almost entirety covered with
granular fill material. Generally, soils in the arsenic
kitchen and stack hilt areas are sand mixed with gravel
or sandy clay mixed with cobbles. No slag was analyzed
as part of this demonstration.
The RV Hopkins site is located in the west end of
Davenport, Iowa. The facility occupies approximately
6.68 acres in a heavy industrial/commercial zoned area.
Industrial activities in the area of the RV Hopkins
property included the manufacture of railroad locomotive
engines during the mid-1800s. The RV Hopkins
property was a rock quarry during the late 1800s.
Aerial surveys beginning in 1929 show that the rock
quarry occupied the majority of the site initially,
gradually decreasing until being completely filled by
1982. It was reported that the site was used to dispose
of demolition debris, automotive, and scrap metal. The
site also has been used by a company that recycled lead
acid batteries.
RV Hopkins began operating as a drum
reconditioner in 1951 across the street from its current
location. In 1964, the site owner reportedly covered the
former quarry area of the site with foundry sand. No
foundry sand was analyzed as part of this demonstration.
RV Hopkins receives between 400 to 600 drums per day
for reconditioning, accepting only drums which meet the
definition of "empty" according to 40 Code of Federal
Regulations 261.7. (2). Most of the drums received at
the facility come from the paint, oil, and chemical
industries.
The area is reported to be underlain by Devonian-
aged Wapsipinicon Limestone, and grey-green shale,
lime mud, and sand stringers dating back to the
Pennsylvanian age.
The RV Hopkins property is composed of five
buildings: the office and warehouse, a warehouse used
to store drums of hazardous waste and a waste pile, a
manufacturing building, a drum reclamation furnace, and
a cutting shed. The office and the warehouse are located
on the southwest corner of the site. Areas investigated
on site include the furnace area, the old and new
baghouses, the former drum storage area on ihe north
end of the facility, the former landfill, and a drainage
ditch. Major contaminants include barium, lead,
chromium, and zinc, as well as lesser concentrations of
other metals, such as copper and nickel, pesticides, and
volatile organic compounds.
7
nRAF7

-------
Based on past sampling data, the most highly
concentrated contaminants in the furnace area are
chromium, lead, and zinc. The highest concentrations
of chromium, lead, and zinc are at the furnace entrance,
as opposed to the furnace exit. The concentrations of
lead are higher in the old baghouse than in the new,
while the new baghouse exhibits a
higher concentration of chromium, as well as high iron,
lead, and barium concentrations. The former landfill has
concentrations of barium, chromium, lead, nickel, and
zinc greater than 1,000 milligrams per kilogram
(mg/kg). Lead is the most prevalent contaminant in the
former drum storage area with lesser concentrations of
barium, chromium, and zinc.
Predemonstration Sampling
Predemonstration sampling was conducted at both
sites between December 5 and 14, 1994. This
predemonstration sampling had the following objectives:
•	To provide data on or verify the extent of surface
contamination at each site and to locate optimum
sampling areas for the demonstration.
•	To allow the developers to analyze samples from the
demonstration sites in advance of the demonstration,
and if necessary, refine and recalibrate their
technologies and revise their operating instructions.
This allowed the developers to optimize their
analyzers for the demonstration.
•	To evaluate samples for the presence of any un-
anticipated matrix effects or interferences that might
occur during the demonstration.
•	To check the QA/QC procedures of the reference
laboratory.
One hundred soil samples were analyzed by FPXRF
on site during the predemonstration sampling activities.
The samples represented a wide range in metals
concentrations and soil textures. Thirty-nine samples
were submitted for reference method analysis using
Methods 3050A/6010A. Twenty-nine of these samples
were split and sent to the developers. Nine field
duplicates were collected and submitted for reference
method analysis to assess proposed sample
homogenizaiion procedures. One purchased PE sample
also was submitted to the reference laboratory to provide
an initial check of its accuracy.
Additionally, three predemonstration samples were
collected from each site, representing low, medium, and
high concentrations. These samples were dried and
ground and then analyzed by six independent laboratories
before the demonstration to create site-specific PE
samples. These samples were analyzed with lab-
oratory-grade XRF analyzers.
Experimental Design
The experimental design of this demonstration was
developed to meet the primary and secondary objectives
stated at the beginning of this section, and was approved
by all demonstration participants prior to the start of the
demonstration. The design is detailed in the
demonstration plan (PRC 1995). This design is
summarized below.
Approximately 100 soil samples were collected from
each of three target soil textures: clay, loam, and sand.
During predemonstration sampling activities, sandy and
loamy textured soils were identified at the ASARCO site
and silty clay soils were identified at the RV Hopkins
site.
The two TN Spectrace analyzers demonstrated can
be operated in either an in situ or intrusive mode.
During the demonstration, the two modes of FPXRF
analysis, in situ and intrusive, involved slightly different
measurement and sampling procedures (Figure 2-2).
Each sampling and analysis procedure was designed to
reflect common applications of FPXRF analyzers. For
in situ analysis, an area 4 inches by 4 inches square was
cleared of all vegetation, debris, and gravel larger than
2 mm in diameter. The FPXRF in situ analyzers took
one measurement in each sample area. This data
represented FPXRF in situ technology measurements for
unprepared soils (in situ-unprepared). Replicate
measurements were taken at 4 percent of these locations
to assess analyzer precision (Figure 2-2).
After the in situ-unprepared analysis was complete
at a given location, the soil within the 4-inch by 4-inch
square was removed to a depth of 1 inch and
homogenized in a plastic bag. This produced a soil
sample of approximately 375 grams or 250 cubic
centimeters (cm3). Sample homogenization was
monitored by adding 1 to 2 grams of sodium fluorescein
salt which fluoresces when it was exposed to ultraviolet
light to the sample homogenization bag. During the
predemonstration, it was determined that sodium
fluorescein did not affect the FPXRF or reference
method analysis. Sample homogenization took place by
kneading the sample and sodium fluorescein salt in a
plastic bag for 2 minutes. After 2 minutes, the sample
preparation technician examined the sample under
ultraviolet light to assess the distribution of sodium
fluorescein throughout the sample. If the sodium
fluorescein salt was not evenly distributed throughout the
sample, the homogenization and checking process was
repeated until the sodium fluorescein was evenly
draft

-------
YES
IN SITU MEASUREMENT
{4X4 INCH GRID)
NO
YES
NO
YES
YES
NO
PACKAGE JO
CJUMS FOR
CONFIRMATORY
ANALYSIS
NO
YES
NO
JS SAMPLE A
FIELD DUPLICATE
s. 20*? .
^ WAS SAMPLE N
PREVIOUSLY USED
FOR PRECISION
DETERMINATION?
/ CONDUCT \
PRECISION MEASUREMENT
(4* OF
\ LOCATIONS) v'
' WAS SAMPLE N
PREV10ULSY USED
FOR PRECISION
DETERMINATION?
x WAS SAMPLE N
PREVIOUSLY USED
FOR PRECISION
DETERMINATION?
END
CHOOSE
SAMPLE LOCATION
PASS SOIL THROUGH
NO. JO SIEVE
CONDUCT 10 REPLICATE
MEASUREMENTS
SPREAD OUT SOIL IN
•INCH DEEP PETRI DISH
CONDUCT
IN SITU
MEASUREMENT
COLLECT SOIL FROM
GRID TO DEPTH OF I INCH
HOMOGENIZE
WITH
FLUORESCEIN DYE
CONDUCT 10
REPLICATE
MEASUREMENTS
LABEL 30% FOR BOTH
3052/601QA ANALYSIS
AND 3030A/60I0A
DRY SAMPLE IN
CONVECTION OVEN
2 HOURS AT 150"C
SAMPLE FOR
CONFIRMATORY
ANALYSIS
GRIND AND PASS
THROUGH NO 40
SIEVE
COLLECT 10-20 GRAMS
FOR MOISTURE
CONTENT DETERMINATION
COLLECT 10-20 GRAM
SUBSAMPLE FOR
WATER CONTENT
DETERMINATION
LABEL 30* FOR
BOTH 3052/6010
AND 3QS0AAQI0A
ANALYSIS
COLLECT
10 REPLICATE
MEASUREMENTS
WITHOUT
MOVING PROBE
SPLIT AND PACKAGE
TWO 20-GRAM AUQUOTS
FOR REFERENCE METHOD
ANALYSIS
PREPARE
10* SAMPLES DRY
20 GRAMS WITH
MICROWAVE OVEN
CONDUCT
INTRUSIVE INSTRUMENT
MEASUREMENTS
(NO PREPARATION)
COLLECT
10 REPLICATE
MEASUREMENTS
WITHOUT
MOVING PROBE
CONDUCT
INTRUSIVE INSTRUMENT
MEASUREMENT OF
MATERIAL PASSING
NO. 40 SIEVE
FIGURE 2-2. SAMPLE PREPARATION AND ANALYSIS: This flowchart depicts the handling procedures for each
sample taken during the demonstration.

-------
distributed throughout the sample. This monitoring
process assumes that even distribution of sodium
fluorescein is indicative of good sample homogenization.
The effectiveness of this homogenization is discussed
later in this section.
The homogenized sample was then spread out inside
a 1-inch-deep petri dish. Each FPXRF analyzer took
one measurement from this homogenized materia]. This
represented the homogenized sample analysis for the in
situ analyzers (in situ-prepared). This approximated the
common practice of sample homogenization in a plastic
bag and subsequent sample measurement through the
bag. Replicate measurements were also collected from
4 percent of these samples to assess analyzer precision.
These replicate measurements were made on the same
soils as the unprepared precision measurements.
Following the in situ-prepared analysis, the
homogenized sample material was passed through a No.
10 mesh sieve (2-mm openings) and approximately 10
grams of this material was placed in a sample cup for
analysis by t ¦ FPXRFs in an intrusive mode. The same
sample cup was used for each FPXRF analyzer.
Replicate measurements were collected from 4 percent
of these samples to assess analyzer precision. These
replicate measurements were made on the same soils as
the in situ-prepared precision measurements. This data
represented FPXRF intrusive mode measurements on
soils with no sample preparation (intrusive-unprepared).
Sample material from this preparation step was collected
and submitted to the reference laboratory for reference
method analysis.
Following the intrusive-unprepared analysis an
aliquot of the soil sample was dried in a convection oven
at 110 °C for one hour and ground with a mortar and
pestle until it passed through a No. 40 stainless-steel
sieve (0.425-mm openings). The sample was then
analyzed by the FPXRF analyzers in an intrusive mode.
Four percent of these samples underwent replicate
measurements to evaluate analyzer precision. These
replicate measurements were made on the same soils as
the intrusive-unprepared precision measurements. This
data represented FPXRF intrusive measurements on
prepared soils (intrusive-prepared).
These four sample preparation steps allowed the
evaluation of the effects of sample preparation on
FPXRF comparability to reference data.
A PRC senior QA reviewer independent of the
demonstration performed an audit of procedures used
during the field work at the ASARCO site. This QC
audit evaluated the implementation of the approved
demonstration plan. The audit report concluded that the
demonstration plan was being adhered to and that data
collection was proceeding as defined in the
demonstration plan (PRC 1995).
Qualitative Factors
There are a number of factors which are important
to data collection but are difficult to quantify and must
be evaluated qualitatively. One such factor is the ease of
learning to use a given FPXRF analyzer and using it in
the field. To assess this factor, PRC personnel were
trained by a developer representative on how to operate
each FPXRF analyzer. All operators met or exceeded
the developer's minimum requirements for education and
previous experience. Demonstration procedures were
designed to simulate routine field conditions as closely as
possible. For this reason, operators were trained just
prior to the demonstration, and did not have prior
experience on the specific FPXRF analyzer that they
used in the demonstration. The developers trained the
operators using their respective operator training
manuals. Based on this training and field experience,
the operators prepared a subjective evaluation assessing
the training and technology operation during the
demonstration (Sections 4 and 5).
Many analytical methods exhibit significant
"operator effects," in which individual differences in
sample preparation or operator technique result in a sig-
nificant effect on the numerical results. To reduce the
possible influence of operator effects, a single operator
was used to operate each FPXRF analyzer. While this
reduced some potential error from the evaluation, it did
not allow evaluation of the analyzers for their
susceptibility to operator-induced error. Sample prepar-
ation variation effects were minimized in the field by
using the same personnel to prepare samples. To elimi-
nate the influence on the reference method analysis, only
one reference laboratory was used to analyze the
samples.
Other important factors that could not be easily
quantified include the portability of the analyzers and
their susceptibility to catastrophic failure. Operators
recorded notes on qualitative factors for each analyzer in
a field logbook (Sections 4 and S).
Quantitative Factors
Many factors in this demonstration were quantifiable
by various means. Examples of quantitative factors
evaluated included analyzer performance near regulatory
action levels, the effect of sample preparation, effects of
microwave sample drying, count times, health and safety
considerations, costs, and interferences.
10
DRAF

-------
The data developed by the FPXRF analyzers were
compared to reference data foT the following primary
analytes: arsenic, barium, chromium, copper, lead, and
zinc; and for the following secondary analytes: nickel,
iron, cadmium, and antimony.
Evaluations of analyzer data comparability involved
examining the effects of site, soil type (three distinct
textures), and preparation on performance and
comparability (Table 2-1). Two sites were sampled for
this demonstration: and therefore, two site variables
were examined (RV Hopkins and ASARCO sites).
These sites produced samples from three distinct soil
types, and therefore, three soil variables were examined
(clays, sands, and loams}. Four sample preparation
steps were used: (1) in situ-unprepared, (2) in situ-
prepared, (3) intrusive-unprepared, and (4) intrusive-
prepared. These variables were nested as follows: site
was divided into RV Hopkins and ASARCO data sets;
the RV Hopkins data represented the clay soil type, and
the ASARCO data was divided into sand and loam soil
types; and each soil type was subdivided by the four soil
preparations. These variables allowed the examination
of panicle size and homogenization effects on data
comparability. These effects were hypothesized to have
the greatest potential impact on data comparability.
Of greatest interest to potential users is analyzer
performance near action levels. For this reason,
samples were approximately distributed as follows: 25
percent in the 0 to 100 parts per million (ppm) range, 50
percent in the 100 to 1,000 ppm range, and 25 percent
in the greater than 1,000 ppm range. The lower range
tests analyzer performance near MDLs; the middle range
tests analyzer performance in the range of many action
levels for inorganic contaminants; and the last range tests
analyzer performance on grossly contaminated soils. All
samples collected for the demonstration were split
between the FPXRF analyzers and reference laboratory
for analysis. Metal concentrations measured using the
reference methods were considered to represent the
"true" concentrations in each sample. "Where duplicate
samples existed, concentrations for (he duplicates were
averaged and the average concentration was considered
to represent the true value for the sample pair. This was
specified in the demonstration plan (PRC 1995). The
reference methods reported measurable concentrations of
target analytes for all of the demonstration samples
analyzed.
In addition to the quantitative factors discussed
above, the common FPXRF sample preparation
technique of microwave drying samples was evaluated.
Sample temperatures during this procedure can be high
enough to melt some mineral fractions in the sample or
combust organic matter. Several metals which present
environmental hazards can volatilize at elevated
TABLE 2-1. PERFORMANCE AND
COMPARABILITY VARIABLES EVALUATED
Variables	
Site (315) Soil (315) Preparation Step
ASARCO
(215)
Sand (100)
in situ-unprepared {100]
in situ-prepared [100]
intrusive-unprepared [100]
intrusive-prepared [100]

Loam (115)
in situ-unprepared [115]
in situ-prepared [115]
intrusive-unprepared [115]
intrusive-prepared [115]
RV Hopkins
(100)
Clay (100)
in situ-unprepared [100]
in situ-prepared [100]
intrusive-unprepared [100]
intrusive-prepared [100]
Notes.
( ) Total number of sample points.
I j Total number of measurements taken.
temperatures. Arsenic sublimes at 188 "C, within the
potential temperature range achieved during microwave
drying of samples. To assess this potential effect, 10
percent of the homogenized, crushed, oven-dried and
sieved (No. 40) samples were split and heated in a
microwave oven on high for 3 minutes. This time was
chosen to approximate the common microwave drying
times used in the field. These split samples were then
submitted for reference analysts. The reference data for
these samples were compared to the corresponding
reference data produced from the convection oven-dried
sample. These data showed the effects of the microwave
drying variable on analyte concentration. The effect of
microwave drying was evaluated for the reference
laboratory in an attempt to identify any potential effect
on data comparability.
Another quantitative variable evaluated was the
count time used to acquire data. During che formal
sample quantitation and precision measurement phase of
the demonstration, the count times were set by the
developers and remained constant throughout the
demonstration. Count times can be tailored to produce
the best results for specific target analytes, however, the
developers selected count times which produced the best
results for the entire suite of target analytes. To allow
a preliminary assessment of the effect of count times,
select soil samples were analyzed in replicate using twice
as long count times as those set by the developers for the
demonstration. This allowed the evaluation of the
effects of couot times on analyzer performance.
An important health and safety issue during the
demonstration was the effectiveness of radioactivity

-------
shielding of each FPXRF analyzer. This effectiveness
was quantitatively measured. Occasional radiation
readings were made with a gamma ray detector near
each analyzer to assess, the potential for exposure to
radiation.
The cost of using each FPXRF analyzer was another
important quantitative factor. Cost includes analyzer
purchase or rental, expendable supplies, such as liquid
nitrogen and sample cups, and nonexpendable costs, such
as labor, licensing agreements for the radioactive
.<¦ iurces, operator training costs, and disposal of
l.jvestigation-derived waste (IDW). These costs and
sample throughput were tracked during the
demonstration to assess the cost per sample. Since
sample throughput can be affected by adjusting count
times, operators used only the developer-specified count
times throughout the demonstration.
Factors that could have effected the quantitative
evaluations include interference effects and matrix
effects. Some of these effects and the procedures used
to evaluate their influence during this demonstration are
summarized below:
•	Heterogeneity: For in situ-unprepared measure-
ments, heterogeneity was partially controlled by
restricting measurements to within a 4-by-
4-inch area. For measurements after the initial
point-and-shoot preparation, heterogeneity was
minimized by sample homogenization. This ef-fect
was evaluated through the sample prep-aration data.
•	Particle Size: The effect of particle size was
evaluated with the two intrusive sample prep-
arations. Theoretically, precision and accuracy
should increase as particle size decreases and
becomes uniform.
•	Moisture Content: It has been suggested that major
shifts in sample moisture content can affect a
sample's relative fluorescence. This effect could
not be evaluated as thoroughly as planned because
of the small difference in sample moisture content
observed at the two sites. This effect was partially
examined in the comparison of analyzer
performance between intrusive-unprepared and
intrusive-prepared analyses. This step in sample
preparation involved drying and grinding.
•	Overlapping Spectra of Elements: Interferences
result from overlapping spectra of metals that emit
X rays with similar energy levels. The reference
method analysis provided data on the concentration
of potential interferants in each sample.
Evaluation of Analyzer Performance
Metals concentrations measured by the different
analyzers were compared to replicate sample
measurement results, corresponding reference laboratory
data, and to other QA/QC sample results. These
comparisons were conducted independently for each
target analyte. These measurements were used to
determine an analyzer's accuracy, data quality level,
method precision, and comparability to reference
methods. Performance on PE samples and SRM
analysis was used to assess analyzer accuracy. Relative
standard deviations (RSD) on replicate measurements
were used to determine analyzer precision. These data
were also used to help determine the data quality of each
FPXRF analyzer's output. The data comparability and
quality determination was primarily based on a
comparison of the analyzer's data and the reference
method's data. Linear regression and a matched pairs t-
test were the statistical tools used to assess comparability
and data quality.
A principal goal of this demonstration was the
comparison of FPXRF data and the reference data.
Methods 3050A/6010A were selected as the reference
methods because they represent the regulatory standard
against which FPXRF data is generally compared before
they are accepted for site characterization. In comparing
the FPXRF data and conventional data it is important to
recognize that while they obtain similar data, i.e., metals
concentrations, the nature of the sample on which the
data are obtained is not identical. There is significant
overlap in the nature of the samples being measured, but
there are also potentially major differences. These
differences or "perspectives," allow the ability to
characterize the same sample in slightly different ways.
Both have a role in site characterization and remediation.
It is important to consider these differences and the
measurement error intrinsic to each method when
comparing the FPXRF against a conventional analytical
method.
The reference methods involve wet chemical
analysis and partial digestion of approximately O.S grams
of sample (approximately 0.25 cm3, depending on
sample bulk density). The digestion process extracts the
most acid soluble portion of the sample, which
represents the material from most surfaces, and clay and
carbonate minerals. Since the digestion is not complete,
the less acid soluble components are not digested and are
not included in the analysis. These components may
include the coarser-grained quartz, feldspar, lithic
components, and certain metal complexes. FPXRF
analyzers generally produce X-ray excitation in an area
of approximately 3 centimeters squared (cm2) to a depth
12
draft

-------
of approximately 2.5 centimeters (cm). This equates to
a sample volume of approximately 7.5 cm3. X rays
returning to the detector are derived from all matrix
material including the larger grained quartz, feldspar,
lithic minerals, metal complexes, and organics. Because
the FPXRF method analyzes all material, it represents a
total analysis in contrast to the reference methods, which
represent a select or partial analysis. This difference can
result in FPXRF concentrations that are higher than cor-
responding reference data when metals are contained
within nonacid soluble complexes or constituents. In
comparison of the two methods, it is important to note
that if metals are contained in nonacid soluble
complexes, a difference between the FPXRF analyzers
and the reference methods is not due to error in the
FPXRF method, but is due to the inherent differences in
the two types of analytical methods.
The comparison of FPXRF data and the reference
data used linear regression as the primary statistical tool.
Linear regression analysis intrinsically contains
assumptions and conditions that must be valid for the
data set for proper analysis. Three important
assumptions involve: (1) linearity of the relationship,
(2)	confidence intervals and constant error variance, and
(3)	insignificant measurement error for the independent
variable (reference data). The assumption of linearity
requires that the independent variable (reference data)
and the dependent variable (FPXRF data) are linearly
related and are not related by some curvilinear or more
complex relationship. Figure 2-3 illustrates that the two
methods are, in fact, related linearly and that this
assumption is correct. Assumptions concerning
confidence intervals and constant error variance require
that the error be normally distributed, sum to equal zero,
be independent, and exhibit constant error variance for
the data set. Figure 2-3 also illustrates that for linear
concentration values this assumption is not correct (at
higher concentrations the scatter around the regression
line increases), but that for the logarithmic transform
(shown as a log-log plot) of the data this assumption is
valid (the scatter around the regression line is relatively
uniform over the entire concentration range). The
change in error distribution (scatter) evident in the
untransformed data results in the disproportionate
influence of large compared with small data values on
the regression analysis. Since all correlations were
performed using log10 transformed concentrations these
assumptions were satisfied.
The last assumption, requiring an insignificant
measurement error for the independent variable
(reference data) is not absolutely true for all elements.
The consequences of measurement error vary depending
on whether the error is caused by the reference methods
or the FPXRF method. If the error is random or if the
error for the reference methods is small compared to the
total regression error, then conventional regression
analysis can be performed and the error becomes a part
of the random error term of the regression model. This
error is shown in the regression summary tables in
Sections 4 and 5 as the "standard error." In this case,
deviations from perfect comparability can be tied to an
analyzer's performance. If the error for the reference
methods is large compared to the total error for the
correlation of the FPXRF and the reference data, then
deviations from perfect comparability might be due in
pan to measurement error in the reference methods.
It is a reasonable assumption that any measurement
errors in either the reference methods or the FPXRF
method are independent of each other. Given this
assumption, then the total regression error is
approximately the sum of the measurement error
associated with the reference methods and the
measurement error associated with the FPXRF method.
The reference methods' precision is a measurement of
the independent variable error, and the mean square
error expressed in the regression analysis is a relative
measure of the total regression error that was
determined during the regression analysis. Precision
data for the reference methods, obtained from RPD
analyses on the duplicate samples from each site, for
each analyte, indicated the error for the reference
methods was less than 10 percent of the total regression
error for all target analytes. Subsequently, 90 percent of
the total measurement error can be attributed to
measurement error associated with the analyzers. Based
on this data, the reference data did allow unambiguous
resolution of data quality determination.
The comparison of the reference data to the FPXRF
data is referred to as intermethod comparison. All
reference and QA/QC data were generated using an
EPA-approved Level 3 analytical method. If the data
obtained by an analyzer were statistically similar to the
reference methods, the analyzer was considered capable
of producing Level 3 data. As the statistical significance
of the comparability decreased, an analyzer was
considered to produce data of a correspondingly lower
quality. Table 2-2 defines the criteria that must be met
for a technology to be considered Level 1, 2, or 3 for
this demonstration. The control limits presented in
Table 2-2 were defined through consultation with EPA
NERL-CRD and EPA Office of Solid Waste (OSW).
Data from this demonstration was used to place
analyzer data into one of three data quality levels. These
data quality levels are defined by EPA (1993) and in the
demonstration plan (PRC 1995). The three data quality
levels are as follows: (1) qualitative, (2) screening, and
(3) definitive. The last two data quality levels are
defined in the "Data Quality Objectives Process for
13
draft

-------
FIGURE 2-3. LINEAR AND LOG-LOG DATA PLOTS: These graphs illustrate the linear relationship between the
FPXRF data and the reference data. The linear data plots illustrate the concentration dependence of this relationship
with increased scatter at higher concentrations. The log-log plots eliminate this concentration dependence effect.
Scatter is relatively constant over the entire plot.
01 10
8 .
CTJ
3 f
fc s
5L 3
^	O
CO	-c
C	H
<
Tt
O
Linear Data Rot-Lead
i-;+
4 ++
+ ± +¦
_i_
_i_
_i_
2	4	6	8
Thousands
Reference Concentration (ppm)
10


10 r
Ol

6 -
fc
ifi

CO
C
ra
6 -
a
cn
3
O
4 -

£

Q
K

CD

2 -
?




o i
Linear Data Rot-Lead

+ + +&-*
, +± +
+
_l_
_L
! 4 € 8
Thousands
Reference Data (mg/kg)
10
^ 10000
a>
S 1000
a
v
Nl
ID
¦o
10
ai
100 -
10
Log-Log Data Rot-Lead
10	100	1000	10000
Reference Concentration (ppm)

Log-Log Data Rot-Lead
10000

3


1* 1000


<0


S
+¦

i 100
JfiSPP"

£


10
i i

10 100 1000 10000

Reference Data (mg/kg)
Superfund" (EPA 1993). Data quality Level 1 was
defined in the demonstration plan (PRC 1995) to further
resolve the screening level data as defined by EPA.
Screening level data as defined in the above referenced
guidance document does not distinguish between data
that simply identifies the presence or absence of
contamination and data that is quantitative.
Definitive data is considered to be the highest level
of data quality. This data is generated using rigorous
analytical methods, such as approved EPA methods.
The data is analyte-specific with confirmation of analyte
identity and concentration. In addition, either analytical
or total measurement error must be determined. The
methods used must produce tangible raw data (e.g.,
chromaiograms, spectra, digital values) in the form of
paper printouts or computer-generated electronic files.
Data may be generated at the site or at an off-site
location, as long as the QA/QC requirements are
satisfied.
Screening data is generated by rapid, less precise
methods of analysis with less rigorous sample
preparation. Sample preparation steps may use simpler
procedures, such as dilution with a solvent instead of
more detailed extraction/digestion and cleanup
procedures. Screening data provide unconfirmed analyte
identification and quantification, although the quanti-
fication may be relatively imprecise. At least 10 percent
of the screening data must be confirmed using analytical
methods and QA/QC procedures and criteria associated
with definitive data. Screening data without associated
confirmation data is not considered to be of known
quality.
Qualitative data are generated by rapid, less precise
methods of analysis with little to no sample preparation.
This level of data is considered to indicate the presence
or absence of contaminants in a sample matrix, but does
not provide reliable concentration estimates. The data
may be compound-specific or specific to classes of
14

-------
TABLE 2-2. CRITERIA FOR CHARACTERIZING DATA QUALITY
Data Quality	Level	Statistical Parameter'
Definitive
3
r2 = 0.85 to 1.0. The slope and y-intercept must be statistically similar to 1.0 and 0.0.
respectively. The precision (RSD) must be less than or equal to 10 percent and
inferential statistics indicate the two data sets are statistically similar.
Screening
2
r2 = 0.70 to 1.0. The precision (RSD) must be between 10 and 20 percent, the data
meets developer performance specifications, normal deviate test statistics on the two
regression parameters indicate that one or both of them are not statistically similar to
their ideal values, or in the case where the regression analysis indicates the data is of
Level 3 quality, but the inferential statistics indicate the data sets are statistically
different.
Qualitative Screening
1
r2 = less than 0.70 or the precision (RSD) is greater than 20 percent. The data must
have less than a 10 percent false negative rate when identifying the absence of
contaminants or the data does not meet its developer's performance specifications.
These statistical tests and parameters are discussed later in the "Intermethod Comparison subsections in
Sections 4 and 5.
Coefficient of determination.
Relative standard deviation.
Notes:
a
r2
RSD
contaminants. Generally, confirmatory sampling is not
required if an analyzer's operation is verified with one
or more check samples.
Approved EPA methods for FPXRF do not exist.
PRC created a draft Method 6200 "Field Portable X-Ray
Fluorescence Spectrometry for the Determination of
Elemental Concentrations in Soil and Sediment" at
EPA's request. The draft method has been submitted for
inclusion in Update 4 of SW-846 scheduled for 1997.
For the purpose of this demonstration, the lack of a
current EPA-approved final FPXRF method did not
preclude the analyzers' data from being considered
definitive. The main criteria for data quality level
assignment was based on the comparability of each
analyzer's data to the data produced by the reference
methods, as well as analyzer-specific criteria such as
precision. Table 2-2 lists the criteria used to define the
data quality levels. These criteria are based on past
demonstrations and were defmed in the demonstration
plan (PRC 1995).
The comparability data set for each analyzer
consisted of 1,260 matched pairs of FPXRF and
reference method data for each target element. This data
set was analyzed as a whole and then subdivided and
analyzed with respect to each of the variables listed on
Table 2-1. This nesting of variables allowed the
independent assessment of the potential influence of each
variable on comparability.
For evaluation of the performance of the analyzers,
a total of 315 soil samples were analyzed by the
reference methods. These 315 samples were analyzed
by the FPXRF analyzers for each of the four sample
preparation steps. This produced 1,260 data values for
each analyzer. Seventy of the 315 samples submitted to
the reference laboratory were split and submitted as field
duplicates to assess the sample homogenization process.
Thirty three of the 315 samples were also split and
microwave-dried; then submitted for reference method
analysis to assess the effect of microwave drying. Of
the 315 samples submitted for reference method
analysis, 215 were collected from the ASARCO site and
100 were collected from the RV Hopkins site.
Approximately twice as many samples were collected at
the ASARCO site because two of the target soil textures
(sands and loams) were found there. Only one target
soil texture (clay) was found at the RV Hopkins site.
An evaluation of the influence of sample
preparation, site, and soil type variables was conducted.
The evaluation of the influence of the site and soil
variables was limited to the examination of the lead and
zinc data, exclusively. These were the only primary
analytes that exhibited a wide distribution of
concentrations across all sites and soil textures. The ef-
fects of sample preparation were evaluated for all target
analytes. If the evaluation of the influence of a given
variable did not result in a better correlation, as
exhibited by a higher coefficient of determination (r2)
and smaller standard error of the estimate, then the
influence was considered to be insignificant. However,
if the correlation worsened, the cause was examined and
reported on. If the correlation improved, resulting in an
improved r2 and reduced standard error of the estimate,
then the impact of the variable was considered
significant. For example, if the r2 and standard error of
the estimate for a given target analyte improved when
the data set was divided into the four sample preparation
15
¦> _..v ^ r*

-------
steps, the sample preparation variable was determined to
be significant. Once this was determined, the variables
of site and soil type were evaluated for each of the four
sample preparations. If the site or soil type variable
improved the regression parameters for a given soil
preparation, then that variable was also considered
significant.
After the significant variables were identified, the
potential impact of analyte concentration was examined.
This was accomplished by dividing the most significant
variable's data set into three concentration ranges: 0 to
100 ppm; 100 to 1,000 ppm; and greater than 1,000
ppm. Then linear regression analysis was conducted on
the three data sets. If this did not result in improved A
and reduced standard errors of the estimate, the
relationship between the analyzer's data and the
reference data was considered linear over the entire
range of concentrations encountered during the
demonstration. This would mean that there was no
concentration effect.
Numerous statistical tests have been designed to
evaluate the significance of differences between two
populations. In comparing the performance of the
FPXRF against the reference methods, the straight-line
regression comparison and the paired t-test were
considered the optimal statistical tests. The paired t-iest
provides a classic test for comparing two populations,
but is limited to analysis of the average or mean
difference between those populations. Statistical analysis
using comparison of two straight-line regression
equations provides information not only about how the
two populations compare on average, but also about how
they compare over the range of values exhibited by both
variables and how they differ at the intercept.
Therefore, this statistical analysis provides information
about the structure of the relationship, that is, do they
differ at high or low concentrations or both. It also
indicates whether the FPXRF data is biased or shifted
relative to the reference data or if the two lines are a
coincident. By comparing the regression equation of the
FPXRF data regressed against the reference data with
the "ideal" regression equation of ihe reference data
regressed against itself (exhibiting a slope = 1.0, an
intercept = 0.0, and an r2 = 1.0), it is possible to
provide evaluation criteria for how the FPXRF compares
against the reference methods at all concentrations.
Linear regression provides an equation that
represents a line (Equation 2-1). Five linear regression
parameters were considered when the level of data
quality produced by the FPXRF analyzers was assessed.
These factors were the y-imercepi, the slope of the
regression line, standard error of the estimate, the
correlation coefficient (r) and r2. In linear regression
analysis, the r provides a measure of the degree or
strength of the correlation between the dependent
variable (FPXRF data), and the independent variable
(reference data). The r2 provides a measure of the
fraction of total variation which is accounted for by the
regression relation (Havlick and Crain 1988). That is,
it is a measure of the scatter about a regression line and.
thus, is a measure of the strength of the linear
association.
(2-1)
Y = mX * 4
where
b is the y-mltretpt of Ihe regression lint,
m is the slope of the regression line,
and Y and X are the dependent and
independent variables, respectively.
Values for r vary from a value of 1 to -1, either of
which indicates a perfect positive or negative correlation
between the independent and dependent variables. A
positive correlation coefficient indicates that as the
independent variable increases the dependent variable
also increases. A negative correlation coefficient
indicates an inverse relationship, as the independent
variable increases the dependent variable decreases. An
r2 of 1.0 indicates that the linear equation explains all the
variation between the FPXRF and reference data. As
ihe r2 varies from 1.0, there is more unexplained
variation, due to such influences as lack of perfect
association with the dependent variable (FPXRF data),
or the influence of other independent variables.
The use of least squares linear regression has
limitations. Least squares regression provides a linear
equation which minimizes the squares of the differences
between the dependent variable and the regression line.
For data sets produced in this demonstration, variance
was proportional to the magnitude of the measurements.
That is, a measurement of 100 ppm may exhibit a 10
percent variance of 10 ppm, while a 1,000 ppm
measurement exhibits a 10 percent variance of 100 ppm.
For data sets with a large range in values, the largest
measurements in a data set exert disproportionate
influence on (he regression analysis because the least
squares regression must account for the variance
associated with the higher valued measurements. This
can result in an equation that has minimized error for
high values, but almost neglects error for low values
because their influence in minimizing dependent variable
error is small or negligible. In some cases, the resulting
equations, biased by high-value data, can may lead to
inappropriate conclusions concerning data quality. The
range of the data examined for the analyzers spanned
between two and four orders of magnitude (e.g., 10 to
100,000 ppm) for both the primary and secondary
16
DRAFT

-------
analytes. This wide range in values and the associated
wide range in variance (influenced by concentration)
created the potential for this problem to occur in the
demonstration data set. To provide a correlation that
was equally influenced by both high and low values
logarithms (log^) of the dependent and independent
variables were used thus scaling the concentration
measurements and providing equal weight in the least
squares regression analysis to both small and large
values (Figure 2-3). All statistical evaluations were
carried out on logI0 transformed data.
If the regression correlation exhibited a coefficient
of determination, r2, between 0.85 and 1.0, the FPXRF
data was considered to have met the first requirement for
Level 3 data classification (Table 2-2). At this point, the
regression line's y-intercept and slope were examined.
A slope of 1.0 and a y-intercept of 0.0 would mean that
the results of the FPXRF analyzer matched those of the
reference laboratory perfectly (FPXRF=reference).
Theoretically, the more the slope and y-intercept differ
from the values of 1.0 and 0.0, the less accurate the
FPXRF technology. However, a slope or y-intercept
can differ slightly from these values without that
difference being statistically significant. To determine
whether such differences were statistically significant,
the Z test statistic for parallelism and for common
intercept were used, using a 95 percent confidence level
for the comparison (Equations 2-2 and 2-3) (Kleinbaum
and Kupper 1978).
(2-2)
Slope Test for Significant Differences
JSEm * 0
where
m is the slope of the regression line,
SE is the standard error of the slope,
and Z is the normal deviate test statistic.
(2-3)
Y-intercept Test for Significant Differences
2 . 6
yjSEt - 0
where
b is the y-intercept of the
regression equation.
The matched pairs t-test was also used to evaluate
whether the two sets of data were significantly different.
The paired t-test compares data sets, which are
composed of matched pairs of data. The significance of
the relationship between two matched-pairs sets of data
can be determined by comparing the calculated [-statistic
with the critical t-value determined from a standard
t-distribution table at the desired level of significance and
degrees of freedom. To meet data quality Level 3
requirements, both the slope and y-intercept had to be
statistically the same as their ideal values, as defmed in
the demonstration plan (PRC 1995), and the data had to
be statistically similar as measured by the t-test. Data
meeting these criteria were considered statistically
equivalent to the reference data.
If the r2 was between 0.70 and 1, or the slope or
intercept were not statistically equivalent to their
laboratory, the analyzer was considered to produce Level
2 quality data (Table 2-2). The linear regression was
deemed sufficiently significant that bias could be
identified and corrected. Results in this case could be
mathematically corrected if 10 to 20 percent of the
samples are sent to a reference laboratory. Reference
laboratory analysis results for a percentage of the
samples would provide a basis for determining a
correction factor.
Data placed in the Level 1 category exhibit r2 values
less than 0.70. These data either were not statistically
similar to the reference data based on inferential
statistics or they did not meet the developer performance
specifications. An analyzer producing data at this level
is considered capable of detection of the presence or lack
of contamination, above its detection limit, with at least
a 90 percent accuracy rate, but is not considered suitable
for reporting of concentrations (Table 2-2).
The MDLs of the analyzers were determined in two
ways. One approach followed standard SW-846
protocol. In this approach, standard deviations (SD)
from precision measurements for samples exhibiting
contamination 5 to 10 times the estimated detection
levels of the analyzers were multiplied by 3. The
resultant number represented the lower MDL for the
analyzers.
In a second approach, MDLs were determined by
analysis of the low concentration outliers on the FPXRF
and reference method data cross-plots. These cross-plots
for all analytes characteristically exhibited a region
below the MDL where FPXRF response was constant,
within an error range, for decreasing reference method
concentrations. Above the MDL, the FPXRF

-------
concentrations increased linearly with increasing
reference method values. Effectively, the linear
correlation between the two methods abruptly changes to
no correlation below the MDL. The value of the MDL
was assigned by determining the minimum concentration
below which no FPXRF values were within two standard
deviations of the regression line. This data also
represented a portion of the regression line that if
included resulted in a decrease in the correlation
coefficient rather than an increase. This MDL
represented a field- or performance-based MDL.
Deviations to the Demonstration Plan
Seven deviations were made from the demonstration
plan (PRC 1995) during on-site activities. The first
deviation dealt with determining the moisture content of
samples. The demonstration plan stated that an aliquot
of the original sample would be used for determining
moisture content. Instead, a small aliquot of soil was
collected immediately adjacent to the original sample
location for determining moisture content. This was
done to conserve sample volume needed for the
reference laboratory. The moisture content sample was
not put through the homogenizing and sieving steps prior
to drying.
The second deviation dealt with the sample drying
procedures for moisture content determination. The
demonstration plan (PRC 1995) required that the
moisture content samples would be dried in a convection
oven at 150 "C for 2 hours. Through testing (visual
observations), it was found that the samples were
completely dried in 1 hour while heating samples to only
110 °C. Therefore, to conserve time, and to reduce the
potential volatilization of metals from the samples, the
samples for moisture content determination and the
intrusive-prepared samples were dried in a convection
oven at 110 °C for 1 hour.
The third deviation involved a^sting analyzer drift
due to changes in temperature. The demonstration plan
(PRC 1995) indicated that at each site, each analyzer
would measure the same SRM or PE sample at 2-hour
intervals during at least one day of field operation.
However, since ambient air temperature did not fluctuate
more than 20" F on any day throughout the
demonstration, potential analyzer drift due to changes in
temperature was not assessed.
The fourth deviation involved the drying of samples
with a microwave. Instead of microwaving the samples
on high for 5 minutes, as described in the demonstration
plan (PRC 1995), the .samples were microwaved on high
for only 3 minutes. This modification was made because
the plastic weigh boats, which contained the samples.
were melting and burning when left in the microwave for
5 minutes. In addition, many of the samples were
melting to form a slag. PRC found (through visual
observations) that the samples were completely dry after
only 3 minutes of micro-waving. This 3-minute
microwave drying is still within common microwave
drying times used in the field.
An analysis of the microwaved samples showed that
the microwave drying process had a significant impact
on the analytical results. The mean RPD for the
microwaved and nonmicrowaved data were significantly
different at a 95 percent confidence level. This suggests
that the microwave drying process somehow increases
error " and sample concentration variability.
This difference may be due to the extreme heat and
drying altering the reference method's extraction
efficiency for target compounds. For this evaluation of
the effects of microwave drying, there were 736 matched
pairs of data where both element, measurements were
positive. Four hundred seventy-one of these pairs
exhibited RPDs less than 10 percent. This 10 percent
level is within the acceptable precision limits for the
reference laboratory as defined in the demonstration
QAPP. Two hundred sixty-five of the pairs exhibited
RPDs greater than 10 percent. RPDs greater than 10
percent may have causes other than analysis-induced
error. Of these 265, 96 pairs indicated an increase in
metals concentration with microwaving, and 169 pairs
indicated reductions in metals concentration. The RPDs
for the microwaved samples were 2 to 3 times worse
than the RPDs from the field duplicates. This further
supports the hypothesis that microwave drying increases
variability.
The fifth deviation involved reducing the percentage
of analyzer precision measuring points. The
demonstration plan (PRC 1995) called for 10 percent of
the samples to be used for assessment of analyzer
precision. Due to the lengthy times required to complete
analysis of an analyzer precision sample, only four
percent of the samples were used to assess analyzer
precision. This reduction in samples was approved by
the EPA technical advisor, OSW representatives, and the
field demonstration team leader. This eliminated 720
precision measurements and saved between 24 and 240
hours of analysis time. The final precision
determinations for this demonstration were based on 48
sets of 10 replicate measurements for each analyzer.
The sixth deviation involved method blanks.
Method blanks were to be analyzed each day and consist
of lithium carbonate that had been used in all sample
preparation steps. Each analyzer had its own method
blank samples, provided by the developer. Therefore,
at the ASARCO site, each analyzer used its own method
blank samples. However, at the RV Hopkins site, each
18
DRAF'

-------
analyzer used lithium carbonate method blanks that were
prepared in the field, in addition to its own method blank
samples. Both types of method blank analysis indicated
that method-induced contamination was not a problem.
The seventh deviation involved assessing the
accuracy of each analyzer. Accuracy was to be assessed
through FPXRF analysis of 10 to 12 SRM or PE
samples. Each analyzer measured a total of 28 SRM or
PE samples. In addition, PE samples were used to
evaluate accuracy of the reference method, and SRMs
were used to evaluate the accuracy of the analyzers.
This is because the PE concentrations are based on acid
extractable concentrations while SRM concentrations
represent total metals concentration. SRM data was used
for comparative purposes for the reference methods as
was PE data for the FPXRF data.
Sample Homogenization
A key issue in ensuring the effectiveness of this
demonstration was ensuring that environmental samples
analyzed by the reference laboratory and by each of the
FPXRF analyzers were subsamples from a homogenous
sample. To address this issue, sampling personnel
exercised particular care throughout the field work to
ensure that samples were thoroughly homogenized
before they were split for analysis. Homogenization was
conducted by kneading the soil in a plastic bag for a
minimum of 2 minutes. If after this time the samples
did not appear to be well homogenized, they were
kneaded for an additional 2 minutes. This continued
until the samples appeared to be well homogenized.
Sodium fluorescein was used as an indicator of
thorough homogenization. Approximately one-quarter
teaspoon of dry sodium fluorescein powder was added to
each sample prior to homogenization. After the
homogenization was completed, the sample was
examined under an ultraviolet light to assess the
distribution of sodium fluorescein throughout the sample.
If the fluorescent dye was evenly dispersed in the
sample, homogenization was considered complete. If
the dye was not evenly distributed throughout the
sample, the homogenization mixing was continued and
repeatedly checked until the dye was evenly distributed
throughout the sample.
To evaluate the sample homogenization process used
in this demonstration, 70 field duplicate sample pairs
were analyzed by the reference laboratory. Sample
homogenization was critical to this demonstration; it
assured that the samples measured by the FPXRF
analyzers were as close as possible to being the same as
the samples analyzed by the reference laboratory. This
similarity was essential to the primary objective of this
demonstration, the evaluation of comparability between
an analyzer's results and those of the reference method.
The homogenization process was evaluated by
determining the RPD between paired field duplicate
samples. The RPDs for the field duplicate samples
reflect the total error for the homogenization process and
the analytical method combined (Equation 2-4). When
total error was determined for the entire data set, the
resultant mean RPD total (error) and 95 percent
confidence interval was 9.7 ± 1.4, for all analytes
reported. When only the primary analytes were
considered, the RPD total (error) and 95 percent
confidence interval was 7.6 ± 1.2. Including the
secondary analytes in the RPD calculation produced a
mean RPD total (error) and 95 percent confidence
interval of 9.3 ± 1.6.
(2-4)
Sample Homogenization Error =
^[(Totai Measurement Error)1 - {Laboratory Error)1]
Using internal QAJQC data from 27 analyses, it was
possible to determine the reference laboratory's method
error. The reference analytical method precision, as
measured by the 95 percent confidence interval around
the mean RPDs (laboratory error) of. predigestion
laboratory duplicate analyses was 9.3± 2.9 for the target
analytes.
To determine the error introduced by the sample
homogenization alone, the error estimate for the
reference analytical method was subtracted from the total
precision (Equation 2-5). Based on the data presented
above, the laboratory-induced error was less than or
approximately equal to the total error. This indicates
that the sample homogenization (preparation) process
contributed little or no error to the overall sample
analysis process. Although the possibility for poorly
homogenized samples exists under any homogenization
routine, at the scale of analysis used by this
demonstration, the samples were considered to be almost
completely homogenized.
(2-5)
Total Utanrtmtnt Error ¦
y[(Samp!t Homogtriiation Error)1 * (LaboratoryError)1)
19
draft

-------
Section 3
Reference Laboratory Results
All soil samples from the ASARCO and RV
Hopkins sites collected during this demonstration were
submitted to the reference laboratory for trace metals
analysis. The trace metals analytical data is discussed
below.
Reference Laboratory Methods
Samples collected during this demonstration were
homogenized and split for extraction using
Methods 3050A/6010A.
Method 30S0A uses an acid digestion procedure
where 1 to 2 grams of soil are digested on a hot plate
with nitric acid, followed by hydrogen peroxide, and is
then refluxed with hydrochloric acid. One gram of soil
was used for extraction of the demonstration samples.
The final digestion volume is 100 milliliters (mL). The
soil sample extracts are then analyzed by Method 6010A.
Method 6010A provides analysis of metals using
ICP-AES. A plasma is produced by applying a.radio-.,
frequency field to a quartz tube wrapped by a coil or
solenoid through which argon gas is flowing. The radio-
frequency field creates a changing magnetic field in the
flowing gas inside the coil, inducing a circulating eddy
current on the argon gas that, in turn, heats it. Plasma
is initiated by an ignition source and quickly stabilizes
with a core temperature of 9,000 to 10,000 degrees
Kelvin.
Soil sample extracts are nebulized, and the aerosol
is transferred to the plasma. Individual analytes intro-
duced into the plasma absorb energy from the plasma
and are excited to higher energy states. These higher
energy states have short lifetimes and the individual
elements quickly fall back to their ground energy state
by releasing a photon. The energy of the emitted photon
is defined by the wavelength of electromagnetic radiation
produced. Since many electronic transitions are possible
for each individual element, several discrete wavelengths
of energy are emitted by each element. Method 6010A
provides one recommended wavelength of light to
monitor for each analyte. Due to complex spectra with
similar wavelengths from different elements in
environmental samples, Method 6010A requires that
interference corrections be applied for quantification of
individual analytes.
Normal turnaround times for the analysis of soil
samples by Methods 3050A/6010A range from 21 to 90
days depending on the complexity of the soil samples
and the amount of QC documentation that is required.
Faster turnaround times of 1 day to 14 days can be
obtained, but for an additional cost.
Costs for the analysis of soil samples by Methods
3050A/6010A range from $150 to S350 per sample
depending on turnaround times and the amount of QC
documentation that is required.
Reference Laboratory Quality Control
The reference laboratory holds certifications for
performing target analyte list metals analysis with the
U.S. Army Corps of Engineers-Missouri River Division,
the State of California, and the State of Utah. These
certifications include on-site laboratory audits, data
package review audits, and the analysis of PE samples
supplied by the certifying agency. PE samples are
supplied at least once per year from each of the
certifying agencies. The reference laboratory's results
for the PE samples are compared to true value results
and certifying agency acceptance limits for the PE
samples. Continuation of these certifications hinges
upon acceptable results for the audits and the PE
samples.
The analysis of soil samples by the reference
laboratory was governed by the QC criteria in its SOPs,
Method 6010A, and the demonstration QAPP. Table 3-
1 provides QAPP QC requirements that were monitored
20
DRAFI

-------
TABLE 3-1. REFERENCE LABORATORY QUALITY CONTROL PARAMETERS
Parameter
Frequency
Reference Method
Requirement
QAPP Requirement
Initial Calibration
Verification (ICV)
Standard
With each initial
calibration
± 10 percent of true value
±10 percent of true value
Continuing Calibration
Verification (CCV)
Standard
After analysis of every
10 samples and at the
end of analytical run
± 10 percent of true value
±10 percent of true value
Initial and Continuing
Calibration Blanks
(ICB) and (CCB)
With each continuing
calibration, after
analysis of every
10 samples, and at the
end of analytical run
± 3 standard deviations of
the analyzer background
mean
No target analytes at
concentrations greater than
2 times the lower reporting
limit (LRL)
Interference Check
Standard (ICS)
With every initial
calibration and after
analysis of 20 samples
± 20 percent of true value
± 20 percent of true value
High Level Calibration
Check Standard
With every initial
calibration
± 5 percent of true value
± 10 percent of true value
Method Blanks
With each batch of
samples of a similar
matrix
No QC requirement
specified
No target analytes at
concentrations greater than
2 times the LRL
Laboratory Control
Samples
With each batch of
samples of a similar
matrix
No QC requirement
specified
80 to 120 percent recovery
Predigestion Matrix
Spike Samples
With each batch of
samples of a similar
matrix
80 to 120 percent recovery
80 to 120 percent recovery
Postdigestion Matrix
Spike Samples
With each batch of
samples of a similar
matnx
75 to 125 percent recovery
80 to 120 percent recovery
Performance
Evaluation Samples
As submitted during
demonstration
No QC requirement
specified
80 to 120 percent recovery
within performance
acceptance limits (PAL)
Predigestion
Laboratory Duplicate
Samples
With each batch of
samples of a similar
matrix
20 percent relative percent
difference (RPD)8
20 percent RPD0
Postdigestion
Laboratory Duplicate
Samples
With each batch of
samples of a similar
matnx
No QC requirement
specified
10 percent RPD0
Notes:
RPD control limits only pertain to onginal and laboratory duplicate sample results that were greater than 10
times the instrument detection limit (IDL).
RPO control limits only pertain to onginal and laboratory duplicate sample results that were greater than or
equal to 10 times the LRL
and evaluated for the target analytes. Method 6010A
QC guidelines also are included in Table 3-1. Due to the
complex spectra derived from the analysts of the
demonstration samples, the QAPP QC requirements
were applied only to the primary analytes. The QAPP
QC requirements also were monitored and evaluated for
the secondary analytes and other analytes reported by the
reference laboratory. However, corrective actions were
not required for the secondary analytes and other
analytes reported by the reference laboratory.
21
A •
J

-------
PRC performed three on-site audits of the reference
laboratory during the analysis of predemonstration and
demonstration samples. These audits were conducted to
observe and evaluate the procedures used by the
reference laboratory, and to ensure that these procedures
adhered to the QAPP QC requirements. Audit findings
revealed that the reference laboratory was following the
QAPP QC requirements. During these audits it was
determined that the reference laboratory was having
problems meeting two of the QAPP QC requirements:
method blank results and the high level calibration check
standard's percent recovery. Due to these problems
these two QAPP QC requirements were widened. The
QC requirement for method blank sample results was
changed from no target analytes at concentrations greater
than the lower reporting limit (LRL) to two times the
LRL. The QC requirement for the high level calibration
standard percent recovery was changed from ±5
percent to ±10 percent of the true value. These
changes were approved by NERL-CRD and OSW.
The reference laboratory internally reviewed its data
before releasing it. PRC conducted a QC review on the
data based on the QAPP QC requirements and corrective
actions listed in the demonstration plan (PRC 1995).
Quality Control Review of
Reference Laboratory Data
The QC data review focused upon the compliance of
the data with the QC requirements specified in the
demonstration QAPP. The following sections discuss
results from the QC review of the reference laboratory
data.
Reference Laboratory Sample Receipt,
Handling, and Storage Procedures
Demonstration samples were divided into batches of
no more than 20 samples per batch prior to delivery of
the samples to the reference laboratory. A total of 23
batches containing 315 samples and 70 field duplicate
samples were submitted to the reference laboratory. The
samples were shipped in sealed coolers at ambient
temperature to the reference laboratory under a chain of
custody.
Upon receipt of the demonstration samples, the
reference laboratory assigned each sample a unique
numb?' and logged each into its laboratory tracking
system. The samples were then transferred to the
reference laboratory's sample storage refrigerators to
await sample extraction.
Samples were transferred to the extraction section of
laboratory under an internal chain of custody. Upon
completion of extraction, the remaining samples were
returned to the sample storage refrigerators. Soil sample
extracts were refrigerated in the extraction laboratory
while awaiting sample analysis.
Sample Holding Times
The maximum allowable holding time from the date
of sample collection to the date of extraction and analysis
using Methods 3050A/6010A is 180 days. Maximum
holding times were not exceeded for any samples during
this demonstration.
Initial and Continuing Calibrations
Prior to sample analysis, initial calibrations (ICAL)
were performed. ICALs for Method 6010A consist of
the analysis of three concentrations of each target analyte
and a calibration blank. The low concentration standard
is the concentration used to verify the LRL of the
method. The remaining standards are used to define the
linear range of the 1CP-AES. The ICAL is used to
establish calibration curves for each target analyte.
Method 6010A requires an initial calibration verification
(ICV) standard to be analyzed with each ICAL. The
method control limit for the ICV is ± 10 percent. An
interference check sample (ICS) and a high level
calibration check standard is required to be analyzed
with every ICAL to assess the accuracy of the ICAL.
The control limits for the ICS and high level calibration
check standard were ±20 percent recovery and ± 10
percent of the true value, respectively. All ICALs,
ICVs, and ICSs met the respective QC requirements for
all target analytes.
Continuing calibration verification (CCV) standards
and continuing calibration blanks (CCB) were analyzed
following the analysis of every 10 samples and at the end
of an analytical run. Analysis of the ICS was also
required after every group of 20 sample analyses.
These QC samples were analyzed to check the validity
of the ICAL. The control limits for the CCVs were ±
10 percent of the true value. The control limits for
CCBs were no target analyte detected at concentrations
greater than 2 times the LRL. All CCVs. CCBs, and
ICSs met the QAPP requirements for the target analytes
with the exception of one CCV where the barium
recovery was outside the control limit. Since barium
was a primary analyte, the sample batch associated with
this CCV was reanalyzed and the resultant barium
recovery met the QC control criteria.
draft

-------
Detection Limits
The reference laboratory LRLs for the target
analytes are listed in Table 3-2. These LRLs were
generated through the use of a MDL study of a clean soil
matrix. This clean soil matrix was also used for method
blank samples and LCSs during the analysis of
demonstration samples. The MDL study involved seven
analyses of the clean soil matrix spiked with low
concentrations of the target analytes. The mean and
standard deviation of the response for each target
analyte was calculated. The LRL was defined as the
mean plus three times the standard deviation of the
response for each target analyte included in the method
detection limit study. All LRLs listed in Table 3-2 were
met and maintained throughout the analysis of the
demonstration samples.
TABLE 3-2. SW-846 METHOD 601 OA LRLs FOR
TARGET ANALYTES
Analyte
LRL (mg/kg)
Antimony
6.4
Arsenic*
10.6
Barium*
5.0
Cadmium
0.80
Chromium*
2.0
Copper*
1.2
Iron
600"
Lead*
8.4
Nickel
3.0
Zinc*
2.0
Notes:
3 LRL elevated due to background interference.
Primary analyte.
mg/kg milligram per kilogram.
The reference laboratory reported soil sample results
in units of mg/kg wet weight. All reference laboratory
results referred to in this report are wet-weight sample
results.
Method Blank Samples
Method blanks were prepared using a clean soil
matrix and acid digestion reagents used in the extraction
procedure. A minimum of one method blank sample
was analyzed for each of the 23 batches of demonstration
samples submined for reference laboratory analysis. All
method blanks provided results for target analytes at
concentrations less than 2 times the levels shown in
Table 3-2.
Laboratory Control Samples
All LCSs met the QAPP QC requirements for all
primary and secondary analytes except those discussed
below.
The primary analytes copper and lead were observed
outside the QC limits in one of the 23 batches of samples
analyzed. Reanalysis of the affected batches was not
performed by the reference laboratory. This data was
qualified by the reference laboratory. Copper and lead
data for all samples included in (he affected batches were
rejected and not used for demonstration statistical
comparisons.
Concentrations of secondary analytes antimony,
nickel, and cadmium were observed outside the QC
limits in the LCSs. Antimony LCS recoveries were
continually outside the control limits while nickel and
cadmium LCS recoveries were only occasionally outside
QC limits. Antimony was a problem analyte and
appeared to be affected by acid digestion, which can
cause recoveries to fall outside control limits. Antimony
recoveries ranged from 70 to 80 percent. Since
secondary analytes were not subject to the corrective
actions listed in the demonstration QAPP, no reanalysis
was performed based on the LCS results of the
secondary target analytes. These values were qualified
by the reference laboratory. All other secondary analyte
LCS recoveries fell within the QAPP control limits.
Predigestion Matrix Spike Samples
One predigestion matrix spike sample and duplicate
was prepared by the reference laboratory for each batch
of demonstration samples submitted for analysis. The
predigestion matrix spike duplicate sample was not
required by the QAPP. but is a routine sample prepared
by the reference laboratory. This duplicate sample can
provide data that indicates if out-of-control recoveries
are due to matrix interferences or laboratory errors.
Predigestion spike recovery results for the primary
analytes arsenic, barium, chromium, copper, lead, and
zinc were outside control limits for at least 1 of the 23
sample batches analyzed by the reference method.
These control limit problems were due to either matrix
effects or initial spiking concentrations below native
analyte concentrations.
23

-------
Barium, copper, and lead predigestion matrix spike
recovery results were outside control limits in 3, 5, and
2 sample batches, respectively. In all of these cases, the
unacceptable recoveries were caused by spiking concen-
trations being much lower than native concentrations of
the analytes. These samples were reprepared, spiked
with higher concentrations of analytes, re-extracted, and
reanalyzed. In all cases the spike recoveries fell within
control limits upon repreparation and reanalysis.
One predigestion matrix spike recovery was outside
control limits for arsenic. The predigestion matrix spike
duplicate sample also was outside of control limits. This
sample exhibited an acceptable RPD for the recovery of
arsenic in the predigestion matrix spike and duplicate,
thus indicating a matrix interference may have been
responsible for the low recovery. This sample was not
reprepared and reanalyzed.
Chromium predigestion matrix spike recoveries
were outside control limits in 7 of the 23 batches of
samples analyzed. Five of these 7 failures exhibited
recoveries ranging from 67 to 78 percent recovery, close
to the low end of the control limits. These recoveries
were similar in the predigestion matrix spike duplicate
samples prepared and analyzed in the same batch. This
indicates that these five failures were due to matrix
interferences. The postdigestion matrix spike duplicate
samples prepared and analyzed along with the remaining
two failures did not agree with the recoveries of the
postdigestion matrix spike samples, indicating that these
two failures may be due to laboratory error, possibly
inaccuracies in sample spiking. These seven
predigestion matrix spike samples were not reprepared
or reanalyzed.
The zinc predigestion matrix spike recovery data
was outside control limits for four batches of samples
analyzed. In three of the spike recovery pairs,
recoveries ranged from 70 to 76 percent, close to the
lower end of the control limits. The fourth recovery was
much less than the lower end of the control limits. All
of the predigestion matrix spike duplicate samples
provided recoveries that agreed with the recoveries for
the predigestion matrix spike sample recoveries
indicating that the low recoveries were due to matrix
effects. These predigestion matrix spikes and associated
samples were not reextracted and reanalyzed.
The secondary analytes cadmium, iron, and nickel
had predigestion spike recoveries outside control limits.
Cadmium spike recoveries were outside control limits
six times. These recoveries ranged from 71 to 85
percent. Iron spike recoveries were outside of control
limits once. Nickel spike recoveries were outside
control limits 4 times. These recoveries ranged from 74
to 83 percent. Antimony spike recoveries were always
outside control limits. No corrective action was taken
for secondary target analytes.
Demonstration sample results for all target analytes
that did not meet the control limits for predigestion
matrix spike recovery were qualified by the reference
laboratory.
Postdigestion Matrix Spike Samples
All postdigestion matrix spike recovery results were
within the control limit of 80 to 120 percent recovery for
the primary analytes.
Secondary analytes antimony and iron were
observed outside the control limits. However, no
corrective action was taken for secondary analytes as
stated in the demonstration QAPP. All postdigestion
spike recoveries for target analytes met the QA/QC
requirements of the QAPP and were considered
acceptable.
Predigestion Laboratory Duplicate Samples
Predigestion laboratory duplicate RPD results were
within the control limit of 20 percent for analyte
concentrations greater than 10 times the LRL except for
the following instances. RPDs for primary analytes
barium, arsenic, lead, chromium, and copper were
observed above the control limit in five predigestion
laboratory duplicate samples. These samples were
reprepared and reanalyzed according to the corrective
actions listed in the QAPP. The reanalysis produced
acceptable RPD results for these primary analytes.
RPD results for the secondary analytes antimony,
nickel, and cadmium were observed outside the control
limit for a number of sample batches. No corrective
action was taken for secondary analytes that exceeded
the RPD control limit.
Postdigestion Laboratory
Duplicate Samples
All primary anaiyte postdigestion laboratory
duplicate RPD results were less than the 10 percent
control limit for analyte concentrations greater than 10
times the LRL.
The RPDs for secondary analytes antimony and iron
were observed above the 10 percent control limit in two
24
OR*

-------
sample batches. No corrective action was taken for
secondary target analytes that exceeded the RPD control
limit.
Performance Evaluation Samples
PE samples were purchased from Environmental
Resource Associates (ERA). . The PE samples are
Priority PollutnT^/Contract Laboratory Program (CLP)
QC standards for inorganics in soil. These types of
samples are used by EPA to verify accuracy and
laboratory performance. Trace metal values are
certified by interlaboratory round robin analyses. ERA
lists performance acceptance limits (PAL) for each
analyte that represent a 95 percent confidence interval
(CI) around the certified value. PALs are generated by
peer laboratories in ERA's InterLaB1" program using the
same samples that the reference laboratory analyzed and
the same analytical methods. The reported value for
each analyte in the PE sample must fall within the PAL
range for the accuracy to be acceptable. Four PE
samples were submitted "double blind" (the reference
laboratory was not notified that the samples were QC
samples or of the certified values for each element) to
the reference laboratory for analysis by Methods
3050A/6010A. Reference laboratory results for all
(arget analytes are discussed later in the Accuracy
subsection.
Four certified reference materials (CRM) purchased
from Resource Technology Corporation (RTC) also were
used as PE samples to verify the accuracy and
performance of the reference laboratory. These four
CRMs were actual samples from contaminated sites.
They consisted of two soil, one sludge, and one ash
CRM. Metal values in the CRMs are certified by round
robin analyses of at least 20 laboratories according to the
requirements specified by the EPA Cooperative Research
and Development Agreement. The certified reference
values were determined by Methods 3050A/6010A.
RTC provides a 95 percent PAL around each reference
value in which measurements should fall 19 of 20 times.
The reponed value from the reference laboratory for
each analyte must fall within this PAL for the accuracy
to be considered acceptable. As with the four PE
samples, the four CRMs were submitted "double blind"
to the reference laboratory for analysis by Methods
3050A/6010A. The reference laboratory results for the
target analytes are discussed later in the Accuracy
subsection.
Standard Reference Material Samples
As stated in the demonstration plan (PRC 1995), PE
samples also consisted of SRMs. The SRMs consisted
of solid matrices such as soil, ash, and sludge. Certified
analyte concentrations for SRMs are determined on an
analyte by analyte basis by multiple analytical methods
including but not limited to ICP-AES, flame atomic
absorption spectroscopy, ICP-mass spectrometry, XRF,
instrumental neutron activation analysis, hydride genera-
tion atomic absorption spectroscopy, and polarography.
These certified values represent total analyte concen-
trations or complete extraction. This is different from
the PE samples, CRM samples, and the reference
methods, which use acid extraction that allows
quantitation of only acid extractable analyte
concentrations.
The reference laboratory analyzed 14 SRMs
supplied by the National Institute of Standards and
Technology (NIST), U.S. Geological Survey (USGS).
National Research Council Canada. South African
Bureau of Standards, and Commission of the European
Communities. The percentage of analyses of SRMs that
were within the QAPP-defined control limits of 80 to
120 percent recovery were calculated for each primary
and secondary analyte.
Analyses of SRMs were not intended to assess the
accuracy of Methods 3050A/6010A as were the ERA PE
or RTC CRM samples. Comparison of 3050A/6010A
acid leach data to SRM data cannot be used to establish
method validity (Kane and others 1993). This is
because SRM values are acquired by analyzing the
samples by methods other than the ICP-AES method. In
addition, these other methods use sample preparation
techniques different than those for Methods
3050A/6010A. This is one reason no PALs are
published with the SRM certified values. Therefore, the
SRMs were not considered an absolute test of the
reference laboratory's accuracy for Methods
3050A/6010A.
The SRM sample results were not used to assess
method accuracy or to validate the reference methods.
This was due to the fact that the reported analyte
concentrations for SRMs represent total analyte
concentrations. The reference methods are not total
metals content analysis methods, rather they target the
teachable concentrations of metals. This is consistent
with the NIST guidance against using SRMs to assess
performance on leaching based analytical methods (Kane
and others 1993).
Data Review, Validation, and Reporting
Demonstration data was internally reviewed and
validated by the reference laboratory. Validation
involved the identifying and qualifying data affected by
QC procedures or samples that did not meet the QC
requirements of the QAPP. Validated sample results
draft

-------
were reported using both hard copy and electronic disk
deliverable formats. QC summary reports were supplied
with the hard copy results. Qualified data was identified
and discussed in the QC summary reports provided by
the reference laboratory.
Demonstration data reported by the reference
laboratory contained three types of data qualifiers: C,
Q, and M. Type C qualifiers included the following:
•	U - the analyte was analyzed for, but not
detected.
•	B - the reported value was obtained from a
reading that was less than the LRL, but
greater than or equal to the IDL.
Type Q qualifiers included the following:
•	N - spiked sample recovery was not within
control limits.
•	* - duplicate analysis was not within
control limits.
Type M qualifiers include the following:
•	P - analysis performed by ICP-AES.
Quality Assessment of
Reference Laboratory Data
An assessment of the reference laboratory data was
performed using the PARCC parameters discussed in
Section 2. PARCC parameters are used as indicators of
data quality and were evaluated using the review of
reference laboratory data discussed above. The
following sections discuss the data quality for each
PARCC parameter.
The quality assessment was limited to an evaluation
of the primary analytes. Secondary and other analytes
reported by the reference laboratory were not required
to meet the QC requirements specified in the QAPP.
Discussion of the secondary analytes is presented in the
precision, accuracy, and comparability sections for
informational purposes only.
Precision
Precision for the reference laboratory data was
assessed through an evaluation of the RPD produced
from the analysis of predigestion laboratory duplicate
samples and postdigestion laboratory duplicate samples.
Predigestion laboratory duplicate samples provide an
indication of the method precision, while postdigestion
laboratory duplicate samples provide an indication of
instrument performance. Figure 3-1 provides a
graphical summary of the reference method precision
data.
The predigestion duplicate RPDs for the primary
and secondary analytes fell within the 2Q percent control
limit, specified in the QAPP, for 17 out of 23 batches of
demonstration samples. These six control limit
exceedances involved only 11 out of the 230 analytes
evaluated for predigesuon duplicate precision in these 23
sample batches (Figure 3-1). This equates to 95 percent
of the predigestion duplicate data meeting the QAPP
control limits. Six of the analytes exceeding control
limits had RPDs less than 30 percent. Three of the
analytes exceeding control limits had RPDs between 30
and 40 percent. Two of the analytes exceeding control
limits had RPDs greater than 60 percent. These control
limit exceedances are possibly due to nonhomogeneity of
the sample or simply due to chance as would be expected
with a normal distribution of precision analyses.
The postdigestion duplicate RPDs for the primary
and secondary analytes fell within the 10 percent control
limit, specified in the QAPP, for 21 out of 23 batches of
demonstration samples. These two control limit
exceedances involved only three out of the 230 analytes
evaluated for postdigestion duplicate precision in these
23 sample batches (Figure 3-1). This equates to 99
percent of the postdigestion duplicate data meeting the
QAPP control limits. The RPDs for these three control
limit exceedances ranged from 11 percent to 14 percent.
Accuracy
Accuracy for the reference laboratory data was
assessed through evaluations of the PE samples
(including the CRMs). LCSs, method blank sample
results, and pre- and postdigestion matrix spike samples.
PE samples were used to assess the absolute accuracy of
the reference laboratory method as a whole, while LCSs,
method blanks, and pre- and postdigestion matrix spike
samples were used to assess the accuracy of each batch
of demonstration samples.
A total of eight PE and CRM samples were analyzed
by the reference laboratory. These included four ERA
PE samples and four RTC CRM samples. One of the
ERA PE samples was submitted to the reference
laboratory in duplicate, thereby producing nine results to
validate accuracy. The accuracy data for all primary
and secondary analytes are presented in Table 3-3 and
displayed on Figure 3-2. Accuracy was assessed over a
wide-concentration range for all 10 analytes with
concentrations for all analytes except iron spanning one
26
DRAFT

-------
FIGURE 3-1. PRE- AND POSTDIGESTION DUPLICATE SAMPLES: The top graph illustrates the reference
laboratory's performance on analyzing predraestion duplicate samples. Twenty percent RPD represents the prediaestion
duplicate control limits defined in the demonstration QAPP. Two points were deleted from this top figure: barium at 65
percent RPD and copper at 138 percent RPD. The bottom graph illustrates the reference laboratory's performance on
analyzing oostdiaestion duplicate samples. Ten percent RPD represents the postdiaestton duplicate control limits
defined in the demonstration QAPP.
a
a.
tr
40
30
a
E 20
a 10
Predigestion Duplicate Samples
	cp	
Antimony Arsenic Bimim Chromium Cadmnim Copper iron
Lead
Nickel
Zinc
o
a.
a:
40
30
- 20
¦2 10
Rostdigestion Duplicate Samples
II
-8-
i_i
			(J	
I!
I #
I—ll
i
f!
Antimony Artemc Barium Chromium Cadmium Copper Iron
Lead
Nickel
Zinc
order of magnitude and some analyte concentrations
spanning two or more orders of magnitude.
Reference laboratory results for all target analytes
in the ERA PE samples fell within the PALs. In the
case of the RTC CRM PE samples, reference laboratory
results for copper in one CRM and zinc in two CRMs
fell outside the published acceptance limits. One of the
two out-of-range zinc results was only slightly above the
upper acceptance limit (811 versus 774 mg/kg). The
other out-of-range zinc result and the out-of-range
copper result were about three times higher than the
certified value and occurred in the same CRM. These
two high results skewed the mean percent recovery for
copper and zinc shown in Table 3-3. Figure 3-2 shows
that the remaining percent recoveries for copper and zinc
were all near 100 percent.
Table 3-3 shows that a total of 83 results were
obtained for the 10 target analytes. Eighty of the 83
results or 96.4 percent fell within the PALs. Only 3
times out of 83 times did the reference method results
fall outside PALs. This occurred once for iron and
twice for zinc. Based on this high percentage of accept-
able results for the ERA and CRM PE samples, the
accuracy of the reference methods was considered
acceptable.
LCS percent recoveries for all the primary analytes
were acceptable in 21 of the 23 sample batches. Lead
recovery was unacceptable in one sample batch and lead
27
w • "

-------
FIGURE 3-2. REFERENCE METHOD PE AND CRM RESULTS: These graphs illustrate the relationship between the
reference data and the true values for the PE or CRM samples. The gray bars represent the percent recovery for the
reference data. Each set of three bars (black, white, and gray) represent a single PE or CRM sample. Based on this
high percentage of acceptable results for the ERA and CRM PE samples, the accuracy of the reference laboratory
method was considered acceptable.
10000
o>
1000
e
¦B 100
CO
10

150
125 ^
O)
>
o
u
0>
100 o:
c
0)
u
75 I
50
Antimony
I Reference Data ~ TrueValue QV> Recovery
800
O)
| 600
S. 400
n
0)
c 200
I
I
1
l
150
125
c
>
o
o
100 &
c
a>
o
75 I
50
Barium
m Reference Data q True Value ~ * Recovery
500
g> 400
~ 300
.S
I 200
C
0)
I ,00
,200
1
[
JE
D
160 ^
O)
>
o
o
120 £
c
0)
u
80 I
40
Arsenic
IReference Data O True Value
Q% Recovery
500
400
300
I
c
£
To
^ 200
0)
u
8 100
i
II
£
100
60
60
40
0)
>
o
u
&
c
a>
u
£
Cadmum
I Reference Data a True Value
Q% Recovery

200
O)
150
I
100
w

o

F
50
<3


0
i

120

100
IT

g
80
$

£
60
S.
40

Chromum
|30S0 Data Q True Values ~% Recovery
results for each sample in the batch were rejected.
Copper recovery was unacceptable in another sample
batch and copper results for each sample in this batch
also were rejected. Percent recoveries of the remaining
primary analytes in each of these two batches were
acceptable. In all, 136 of 138 LCS results or 98.5
percent fell within the control limits.
Method blank samples for all 23 batches of
demonstration samples provided results of less than 2
times the LRL for all primary target analytes. This
method blank control limit was a deviation from the
QAPP, which had originally set the control limit at no
target analytes at concentrations greater than the LRL.
This control limit was widened at the request of the
28
\..

-------
FIGURE 3-2 (Continued). REFERENCE METHOD PE AND CRM RESULTS: These graphs illustrate the relationship
between the reference data and the true values for the PE or CRM samples. The gray bars represent the percent
recovery for the reference data. Each set of three bars (black, white, and gray) represent a single PE or CRM sample.
Based on this high percentage of acceptable results for the ERA and CRM PE samples, the accuracy of the reference
laboratory method was considered acceptable.
100000
¦as
i*
10000
¦B 1000
V
U
100
10
tun
400
300 ^
©
>
o
200 £
c
Q>
u
100
£
Copper
(Reference Data Q True Value Q%Recovery
10000
o>
t 1000
c
.Q
C
0>
o
c
100
125
100 ^
a>
§
u
75 £
c
©
o
50 |
25
Lead
^ReferenceData ~TrueValue ~SReeovery
at
35
25
n a
£ °
C	-C
01 K
15
J
I
120
100 C
v
o
o
O)
CC
c
O)
u
80
60 £
40
Iron
I Reference Data ~ True Value
~ ^Recovery
100000
o> 10000
I
C
£
"c
V
u
1000
100
10
1
llllhll
120
100 -
o>
>
o
u
0)
bo EL
c
0)
u
60
40
£
Nickel
I Reference Data QTnie Value ~ * Recovery
100000
.400
10000
H 1000
c
«
o
100
10
Itao
300
©
>
o
o
flj
.200 &
1:
0)
O
100
4
Zinc
(Reference Data ~ True Value
~%Reco»ery
reference laboratory. A number of batches were
providing method blank results for target anaiytes at
concentrations greater than the LRL, but less than 2
times the LRL. This alteration was allowed because
even at 2 times the LRL, positive results for the method
blank samples were still significantly lower than the
MDLs for each of the FPXRF analyzers. The results
from the method blank samples did not affect the
accuracy of the reference data as it was to be used in the
demonstration statistical evaluation of FPXRF analyzers.
The percent recovery for the predigestion matrix
spike samples fell outside of the 80 to 120 percent
control limit specified in the QAPP in several of the
23 batches of demonstration samples. The predigestion
matrix spike sample results indicate that the accuracy of
29
,* .r*""

-------
TABLE 3-3. REFERENCE LABORATORY ACCURACY DATA FOR TARGET ANALYTES
Analyte
n
Percent Within
Acceptance
Range
Mean
Percent
Recovery
Range of
Percent
Recovery
3D of
Percent
Recovery
Concentration
Range (mg/kg)
Antimony
6
100
104
63 - 125
15
50-4,955
Arsenic
8
1Q0
106
90 - 160
22
25 - 397
Barium
9
100
105
83 - 139
21
19 - 586
Cadmium
9
100
34
63-93
10
1.2-432
Chromium
9
100
91
77-101
8
11 - 187
Copper
9
89
123
90 - 332
79
144 - 4,792
Iron
7
100
98
. 79-113
12
6,481-28,664
Lead
8
100
86
35-108
22
52 - 5,194
Nickel
S
100
95
79 -107
10
13 -13.279
Zinc
9
78
120
79 - 309
72
76 - 3,021
Number of PE or CRM samples with detectable analyle concentrations.
Standard deviation.
Milligrams per kilogram.
Notes:
n
SD
mg/kg
specific target analytes in samples from the affected
batches mrv be suspect. These results were qualified by
the reference laboratory. This data was not excluded
from use for the demonstration statistical comparison.
A discussion of the use of ibis qualified data is included
in the "Use of Qualified Data for Statistical Analysis"
subsection.
The R.PD for the postdigestion matrix spike samples
fell within the 80 to 120 percent control limit specified .
in the QAPP for all 23 batches of demonstration
samples.
The quality assessment of the reference laboratory
data indicated that the absolute accuracy of the method
was acceptable. Only 3.6 percent of the CRM and PE
sample results fell outside of the PALs. However, based
on professional judgement, it was determined that this
small percentage of these outliers did not justify rejection
of any detnonstratiart sample data. The accuracy
assessment also indicated that most of the batches were
acceptable. Two batches were affected by LCS outliers,
and several batches of data were qualified due to
predigestion matrix spike recovery outliers. This data
was rejected or qualified. Rejected data was not used.
Qualified data was used as discussed below.
Representativeness
Representativeness of the analytical data was
evaluated through laboratory audits performed during the
course of sample analysis by the reference laboratory,
and QC sample analyses including; method blank
samples, laboratory duplicate samples, and CRM and PE
samples. These QC samples were determined to provide
acceptable results. From these evaluations, it was
determined that representativeness of the reference data
was acceptable.
Completeness
Results were obtained for all soil samples extracted
and analyzed by Methods 3050A/6010A, Some results
were rejected or qualified. Rejected result? were
deemed incomplete. Qualified results were usable for
certain purposes and were deemed as complete.
To calculate completeness, the number of non-
rejected results was determined. This number was
divided by the total number of results expected, and
multiplied fey 103 so express tompletentss. as. a per-
centage. A total of 385 samples were submitted for
analysis. Of the reported analytes, six were primary
analytes, resulting in an expected 2,310 results. Forty
of these were rejected, resulting in 2,270 complete
results. Reference laboratory completeness was
determined co be 98.3 percent, which exceeded the
objective for this demonstration of 95 percent. The
reference laboratory's completeness was, therefore,
considered acceptable.
30
A tTT
'^^-,3 a

-------
Comparability
Comparability of the reference data was controlled
by following laboratory SOPs written for the
performance of sample analysis using Methods
3050A/6010A. QC criteria defined in the SW-846
methods and the demonstration plan (PRC 1995) were
followed to ensure that reference data would provide
comparable results to any laboratory reporting results
for the same samples.
Reference results indicated that Methods
3050A/6010A did not provide comparable results for
some analytes in the SRM samples. SRM performance
data for target analytes is summarized in Table 3-4 and
displayed on Figure 3-3. As with the PEs, the analyte
concentrations spanned up to three orders of magnitude
in the SRMs. The percentage of acceptable (80 to 120
percent recovery) SRM results and mean percent
recovery were less than 50 percent for the analytes
antimony, barium, chromium, iron, and nickel. The low
recoveries for these five analytes reflect the lesser
tendency for them to be acid-extracted (Kane and others
1993).
Under contract to the EPA, multiple laboratories
analyzed NIST SRMs 2709, 2710, and 2711 by Methods
3050A/6010A. A range, median value, and percent
leach recovery based on the median value for each
detectable element were then published as an addendum
to the SRM certificates. These median values are not
certified, but provide a baseline for comparison to other
laboratories analyzing these SRMs by Methods
3050A/6010A. Table 3-5 presents the published percent
leach recovery for the 10 primary and secondary
analytes and the reference laboratory's results for these
three NIST SRMs. Table 3-5 shows that the results
produced by the reference laboratory were consistent
with the published results indicating good comparability
to other laboratories using the same analytical methods
on the same samples.
The inability of Methods 3050A/6010A to achieve
the predetermined 80 to 120 percent recovery
requirement indicated that the methods used to
determine the certified values for the SRM samples were
not comparable to Methods 3050A/6010A. Differences
in the sample extraction methods and the use of different
analytical instruments and techniques for each method
were the major factors of this noncomparability.
Because of these differences, it was not surprising that
the mean percent recovery was less than 100 percent for
the target analytes. The lack of comparability of
Methods 3050A/6010A to the total metals content in the
SRMs did not effect the quality of the data generated by
the reference laboratory.
The assessment of comparability for the reference
data revealed that it should be comparable to other
laboratories performing analysis of the same samples
using the same extraction and analytical methods, but
may not be comparable to laboratories performing
analysis of the same samples using different extraction
and analytical methods, or by methods producing total
analyte concentration data.
Use of Qualified Data for Statistical Analysis
As noted above, the reference laboratory results
were reported and validated, qualified, or rejected by
approved QC procedures. Data was qualified for
predigestion matrix spike recovery and pre- and
postdigestion laboratory duplicate RPD control limit
outliers. These problems were not considered
sufficiently serious to preclude the use of coded data.
Appropriate corrective action as stated in the
demonstration plan (PRC 1995) was instituted. The
result of the corrective action indicated that the poor
percent recovery and RPD results were due to matrix
effects. Since eliminating the matrix effects would
require additional analysis using different methods: such
as atomic absorption spectrometry using the method of
standard additions for analyte quantification, the matrix
effects were simply noted and were not corrected.
Rejection of a large percentage of data would
increase the apparent variation between the reference
data and the FPXRF data. This apparent variation
would probably be similar to that introduced by using the
data. For these reasons, the qualified data were used.
This action was also supported by the fact that the data
from the sample batches associated with the QC control
limit problems was either not used in the comparability
study, or the data was not identified as anomalous (as
outliers) during the comparability study.
PARCC parameters for the reference laboratory
data were determined to be acceptable. It was expected
that any laboratory performing analysis of these samples
using Methods 3050A/6010A would experience
comparable matrix effects. A primary objective of this
demonstration was to compare sample results from the
FPXRF analyzers to Methods 3050A/6010A, the most
widely used regulatory agency-approved methods for
determining metal concentrations in soil samples for
environmental applications. The comparison of FPXRF
and the reference methods had to take into account
certain limitations of both methods, including matrix
effects. For these reasons, qualified reference data was
used for statistical analysis.
The QC review and QA of the reference data
indicated that over 98 percent of the data either met the
31


-------
TABLE 3-4. SRM PERFORMANCE DATA FOR TARGET ANALYTES
Analyte
n
Percent Within
Acceptance
Range
Mean
Percent
Recovery
Range of
Percent
Recovery
SD of
Percent
Recovery
Concentration
Range (mg/kg)
Antimony
5
0
22
15-37
9
3.8-171
Arsenic
11
72
84
67-106
10
18-626
Banum
8
12
41
21-89
21
414-1,300
Cadmium
10
50
80
43-95
15
2.4 - 72
Chromium
10
0
45
14-67
16
36 - 509
Copper
17
88
82
33-94
17
35 - 2,950
Iron
7
14
62
23-84
25
28,900 - 94,000
Lead
17
82
83
37-99
17
19-5.532
Nickel
16
19
67
25-91
17
14 - 299
Zinc
16
75
81
32-93
14
81 - 6,952
Notes:
n Number of samples with detectable analyte concentrations.
SD Standard deviation,
mg/kg Milligrams per kilogram.
demonstration QAPP objectives or it was QC coded for
reasons not limiting its use in the data evaluation. The
remaining data was found to accurately and precisely
represent the concentrations of target analytes in
demonstration samples. Less than 3 percent of the data
was rejected based on QAPP criteria. Rejected data was
not used for statistical analysis. The reference data
was as good or better, than other laboratory analyses of
the same samples performed using the same extraction
and analytical methods. The reference data met the
definitive data quality criteria (Level 3) and was of
sufficient quality to support regulatory activities. The
reference data was found to be acceptable for
comparative purposes with the FPXRF data.
32
draft

-------
200
o>
<£ 150
f
•I 100
SE
c

£ 50
1
111 jil-H
Antimony
40
30 >*
0)
>
o
o
20 31
C
0)
— 10

IgReferenceData QTmeVatin QKReeovaiy
800
600
¦1 400
re
c
0>
g 200
.
3
u
80 §.
c
0)
(J
£
60
40
Arsenic
I Reference Data ~ TrveValue Q % Recovery
10000
I1
f 1000
c
.9
c
ti
u
c
100
10
Baraim
|Reference Data ~ True Value
~ %Recovery
80
60
2 40
to
W
c
£ 20
100
Cadmum
| Reference Data Q True Value Q St Recovery
600
ra
t 400
C
o
£ 200
U
c
u
i!
M
Chrormim
|Referenee Data oTiueValue ~* Recovery
10000
Reference Data Q True Value n* Recovery
FIGURE 3-3. REFERENCE METHOD SRM RESULTS: These graphs illustrate the relationship between the reference
data and the true values for the SRM samples. The gray bars represent the percent recovery for the reference data.
Each set of three bars (black, white, gray) represent a single SRM sample.

-------
tior
,100
C	to
1— "O
c	c
o	«
M	til
m	a
c	r
©	H*
U
c
90.
70
50.
30
10
.60
.60
I
o
o
£
.40 =
Bill'
.20
.0
*
Iron
1 Reference Data O True Value ~ %Recovery
400

300
C
J 200
<0
i—
c
V
g 100
M
100
60
60 g
« §
u
£
Nickel
BRetermtcoData pT rue Value Q%Recovery
120
10000
100
§
{? 1000
c
O)
u
c
<3
40
too
Lead
^ Refemrce Data QTrue Value Q W>ecovery
10000
100
1000
60
c
a
40
100
c
8
Zinc
0ReferenceData ~ TmeVelue QltRecovefy
FIGURE 3-3 (Continued). REFERENCE METHOD SRM RESULTS; These graphs illustrate the relationship between
the reference data and the true values for the SRM samples. The gray bars represenl the percent recovery for the
reference data. Each set of three bars (black, white, gray) represent a single SRM sample.
TABLE 3-5. LEACH PERCENT RECOVERIES FOR SELECT NIST SRMS

NIST SRM 2709
NIST SRM 2710
NIST SRM 2711

Published
Result
Reference
Laboratory
Result
Published
Result
Reference
Laboratory
Result
Published
Result
Reference
Laboratory
Result
Antimony
-
-
21
-
-
20
Arsenic
-
106
94
87
86
91
Barium
41
37
51
45
28
25
Cadmium
-
-
92
84
96
87
Chromium
61
-
49
-
43
49
Copper
92
85
92
92
88
90
Iron
86
84
B0
78
76
56
Lead
69
87
92
96
95
90
Nickel
89
76
71
69
78
70
Zinc
94
78
85
86
89
85
Notes.
a Published results found in addendum to SRM certificates for NIST SRMs 2709,2710. and 2711.
NIST National Institute of Standards and Technology.
SRM Standard reference materials.
- Anaiyte not present above the method LRL.
7*^ IT** yo
34

-------
Section 4
TN Lead Analyzer
This section provides information on the TN Lead
Analyzer including background information, operational
characteristics, performance factors, a data quality
assessment, and a comparison of its results with those of
the reference laboratory.
Background
Since 1988, the developer has produced field
portable and laboratory-grade XRF technologies for a
broad range of applications. The TN Lead Analyzer was
released in 1993 specifically for analyzing lead in a
variety of matrices such as soil, paint and paint chips,
surface dust, and air filters. Using the "Soils
Application" software supplied with the analyzer, it can
also identify and quantify arsenic, chromium, iron,
copper, zinc, and manganese in soils.
The TN Lead Analyzer uses an Hgl2 semiconductor
detector that achieves a Manganese-Ktt X-ray resolution
of better than 300 eV. The detector is operated at a
moderately subambient temperature controlled by a low
power thermoelectric (Peltier) cooler in the measurement
probe.
To perform either an in situ or intrusive analysis, a
sample is positioned in from of the plastic film probe
measurement window and sample measurement is
initiated. This exposes the sample to primary radiation
from the source. Fluorescent and backscatiered X rays
from the sample re-enter the analyzer through the
window and are counted in the high resolution Hgl2
detector. When analyzing intrusive samples, the probe
is placed upright in a stand and the sample, which is
contained in a thin-windowed plastic cup, is placed over
the probe measurement window and beneath a swing-
down safety shield.
Contaminant concentrations are computed using a
fundamental parameter (FP) calibrated algorithm that is
part of the TN Lead Analyzer's software package. The
TN Lead Analyzer uses FPs to calibrate its detector.
The FPs are based on the physics of the excitation of
target analytes and the emission of X rays. The FP
method does not require site-specific calibration samples;
however, site specific samples can be used to customize
the calibration to a particular site or matrix. The
software package supports multiple XRF calibrations.
Each application is a complete analysis configuration,
including target analytes to be measured, interfering
target analytes in the sample, and a set of FP calibration
coefficients. This information is loaded into the analyzer
at the factory.
Operational Characteristics
This section discusses equipment and accessories,
operation of the analyzer, description of the operator,
training, reliability of the analyzer, health and safety
concerns, and cost to operate the analyzer.
Equipment and Accessories
The TN Lead Analyzer comes with all of the
equipment necessary for both in situ and intrusive
operation (Table 4-1). A hard-shell carrying case
containing the equipment protected by foam inserts is
provided for transportation and storage.
Two main components make up the analytical
system: a probe and an electronics unit. The probe
contains the radioisotope source, Cd109, for sample
excitation and the Hgl2 detector for anaiyte identification
and quantitation. The source is encapsulated and housed
in a metal turret with additional lead shielding inside the
probe. The source exposes the sample to excitation
radiation through a sealed 1-inch-diameter mylar window
in the face of the probe. The X-ray-induced
fluorescence from the sample passes back through the
window and is intercepted by the Hgl2 detector. The
detector identifies and measures the energy of each X
ray and builds a spectrum of anaiyte peaks on a
35


-------
TABLE 4-1. ANALYZER INSTRUMENT SPECIFICATIONS—TN LEAD ANALYZER
Characteristic
Specification
Resolution
< 300 eV (Manganese-Ka)
Source
30 millicuries (mCi) Cd109 (with shim inserts)
Detector
Hgl2-Peltier cooled
Probe Size
12.7 cm x 7.6 cm x 21.6 cm
Probe Weight
1.9 kilograms
Probe Operating Temperature
0 to 49 °C
Electronics Unit Sire
32 cm x 30 cm x 10 cm
Electronics Unit Weight
6.7 Kilograms
Electronics Unit Operating Temperature
0 to 49 "C
Electronics Unit Storage Capacity
600 sets of numerical results and 100 spectra
Power Source
120V or 220V (alternating current) or internal
batteries
Operational Checks
3 NIST SRMs, silicon dioxide (Si02) and Teflon®
blanks, pure element check sample kit
Intrusive Operation
Uniblock probe stand
Computer Interface Operation
RS 232 serial input/output cable, operators
manual, application and results software, and
training video
Contact:
Raj Natarajan
2555 N. interstate Hwy. 35
Round Rock, TX 78664
(800) 736-0801
(512) 388-9200 (FAX)

2.048-channel multichannel analyzer (MCA), which is
contained in the electronics unit. This spectrum contains
the peak lines for all the metals present in the sample.
Spectral data is communicated from the probe to the
electronics unit through a flexible cable of 6, 12. or 20
feet in length. The standard cable length is 6 feet. X-
ray emission peaks are integrated and metal
concentrations in tng/kg or percentage values are
calculated. The electronics unit will store and display
both numerical results and spectra from a measurement.
A maximum of 600 sets of numerical results and 100
spectra can be stored before downloading to a personal
computer (PC) using a RS-232 cable.
The electronics unit can be operated from a battery
or from an alternating current (AC) electric line via a
plug-in adaptor unit. The TN Lead Analyzer is supplied
with two nickel-cadmium batteries and a battery charger.
The batteries last approximately 8 hours and require a
minimum of 14 hours to fully recharge. For in situ
analysis, the developer provided a water-resistant
carrying case and a strap for easy portability on site.
The canying case has a closable flap on cop that can be
closed to protect the unit from adverse weather
conditions.
Other equipment and supplies that are helpful when
using the TN Lead Analyzer, which are not supplied by
the developer, include a PC for downloading the FPXRF
data, protective gloves, paper towels, and a permanent
marking pen.
Operation of the Analyzer
For this demonstration, the TN Lead Analyzer was
operated on battery power only. The in situ analysis was
performed with the analyzer in the carrying case. The
probe was pointed at the soil surface and analysis was
started by pressing a trigger on the back of the probe.
For intrusive analysis, the probe was placed in the
"uniblock" pointed upward with the safety shield
36
A —

-------
attached. The "uniblock" is a free standing support for
the pTobe. All intrusive analyses at both sites were
performed by setting the analyzer on a table top located
indoors. At the ASARCO site, the room was not heated
or cooled so analysis occurred at ambient outdoor
temperatures. At the RV Hopkins site, the area where
the analyzers were operated was maintained at 25 #C.
Description of the Technology Operator
The PRC operator chosen for analyzing soil samples
using the TN Lead Analyzer had a bachelor's degree in
environmental science. Prior to conducting this work,
this operator worked for a year and a half in a
pharmaceutical laboratory as an analytical chemist, and
a half year as an environmental scientist. The operator
received approximately 8 hours of training by the
developer at the start of the demonstration. The training
covered the theoretical background of XRF technology
and specific operation of the TN Lead Analyzer.
Training
The training included step-by-step instructions on
how to set up and use the TN Lead Analyzer. These
instructions covered connecting the nickel-cadmium
battery, attaching the probe to the electronics unit,
setting up the "Soils Applications" software, operating
the keyboard and analyzer software, modifying the count
times for the Cd109 source, setting the probe in the
"uniblock" and attaching the safety shield for intrusive
analysis, downloading results to a PC, and performing
instrument maintenance, for example, replacing the
probe window.
The TN Lead Analyzer was calibrated prior to the
training using an FP algorithm and fme tuned with site-
specific soil samples supplied from the predemonstration
activities. Part of the training included a discussion of
QC requirements, such as the analysis of a pure iron
energy calibration check, a silicon dioxide (Si02) blank,
and at least one NIST SRM. Possible interferences that
could be encountered and recommended procedures for
preparing both in situ and intrusive soil samples for
analysis were discussed in detail. At the conclusion of
the training, the developer was confident that the
operator was ready to operate the TN Lead Analyzer.
The developer accompanied the PRC operator on site
during the first morning at the ASARCO site and
observed the operator analyzing soil samples. No
problems were encountered and the developer left the
site.
The developer stated that a potential operator
requires at least a high school education to operate the
TN Lead Analyzer. The PRC operator had no prior
experience operating a FPXRF instru-
ment and found the TN Lead Analyzer easy to operate.
The training was considered more than sufficient to
successfully operate the analyzer. The operator felt the
size and weight of the TN Lead Analyzer were two of its
greatest attributes. The operator noted that the analyzer
did not become cumbersome during the long days of
field use, especially with the shoulder strap and field
carrying pack. The shorter connection cable (6 foot)
was preferred for field use because a longer cable often
dragged on the ground and posed a hazard for tripping to
the operator. The operator had only one comment on
the design of the optional field pack. The operator felt
it should have a rubberized bottom for easier
decontamination.
The PRC operator noted that because the TN Lead
Analyzer was calibrated prior to arriving at the first site,
it allowed more time for sample analysis. The analyzer
required little computer or technical background for
operation and the menu driven software was easy to use.
Downloading and storing data on diskettes was easy with
some prior knowledge of disk operating system
commands. No data computation or data reduction was
required to obtain the results. Labeling of data points
was considered cumbersome because three buttons had
to be pressed to enter one character in a sample number.
Reliability
A reliability check of the TN Lead Analyzer was
carried out by measuring a check sample daily. The
reliability check was comprised of a 50-second
measurement of a pure iron sample. This measurement
verified (1) the fluorescent element sensitivity; (2) the
spectrometer energy resolution; and (3) the spectrometer
energy calibration. To be acceptable, the measured
relative X ray intensity of iron had to be greater than
0.95 and the equivalent intensity of manganese and
cobalt had to be less than 0.006. Relative intensity
refers to the new value relative to that obtained at the
time of the initial instrument calibration. If the intensity
conditions were not met, then the pure iron sample was
reanalyzed. If the initial readings occurred in the
reanalysis, an energy calibration was needed. No energy
calibrations were required during the demonstration
based on the pure iron sample results, confirming the
high reliability of the TN Lead Analyzer.
During the demonstration, there were frequent light
to moderate rains while the analyzer was performing the
in situ measurements. The developer recommended that
samples analyzed by the TN Lead Analyzer have less
than 20 percent moisture content by weight. The
samples collected during this demonstration contained up
to 30 percent moisture content by weight. This
37
~ P-* ..v.

-------
increased moisture content did not reduce the analyzer's
data comparability. During the ASARCO site sampling,
there was a period of heavy rain for approximately 1.5
hours. After the rain, it was common for the soil
surface to be saturated. This did not pose an operational
problem for the analyzer in the in situ mode. At the
ASARCO and the RV Hopkins sites, the temperatures
ranged from approximately 5 to 16 °C and 6 to 22 "C,
respectively. Despite the less than ideal weather
conditions, there were no mechanical or electronic
problems experienced with the TN Lead Analyzer during
the course of the demonstration. The only maintenance
required was the replacement of the probe window cover
.once due to contamination and damage from small
pebbles. The replacement of the probe window cover
took approximately 2 jo 3 minutes. A spare probe
window was included with the analyzer.
Health and Safety
The potential for exposure to radiation from the
excitation source was the greatest health and safety
consideration while using the analyzer. Radiation was
monitored with a radiation survey meter. Background
radiation at the two sites was between 0.006 and 0.012
mllirems per hour (mrem/hr). Radiation exposure was
monitored in the in situ and intrusive modes whUe the
probe's source was exposed (during a measurement)
obtaining a worst-case scenario. The radiation was
measured within 5 cm of the probe face while the
analyzer was analyzing a sample. Radiation exposure
was also monitored at a point on the probe where the
operator's hand was located during analysis to provide a
realistic value of operator exposure. The TN Lead
Analyzer is sold under a general license, meaning that
the analyzer is designed and constructed in such a way
thai anybody operating it, as per the instruction manual,
will not be exposed to harmful Tadiation levels according
to the Nuclear Regulatory Commission. Many states
still recommend that radiation from survey instruments
be below a certain level. For example, in the state of
Kansas, the permissible occupational exposure is 5,000
millirems per year, which equates to approximately 2 to
3 mrem/hr assuming constant exposure for an entire
work year.
While taking in situ measurements, radiation values
of 0.40 to 0.45 mrem/hr at the probe face and 0.05 to
0,06 mrem/hr at the probe handle were obtained for the
TN Lead Analyzer with the Cdlw source exposed.
While collecting intrusive measurements with the TN
Lead Analyzer, radiation values of 0.50 to 0.60 mrem/hr
directly above the protective cover and 0.05 to 0.06
mrem/hr 1.0 foot from the protective cover were
obtained with the Cd109 source exposed. AH measured
radiation values were less than the occupational level of
2.0 mrem/hr. The operator noted there was no safety
feature on the analyzer that would prevent a person from
accidental exposure by pushing the tngger on the rear of
the probe to start ail analysis while the probe was pointed
at the operator or another person.
Cost
At the.time of the demonstration, the TN Lead
Analyzer cost 539,500 to purchase. This included all of
the equipment necessary to operate the anaiyzer, Table
4-2 shows the incidental items and costs associated with
the use of the TN Lead Analyzer. The analyzer is
warranted for a full year with an optional extended
warranty. The TN Lead Analyzer can be rented from
the developer for $5,000 per month or $3,000 for 2
weeks. Additional field packs can be purchased for S2Q0
and external batteries, charger, and adapter for $750. A
12-month or 24-month extended warranty can be
purchased for S2.750 or £4,750, respectively. Periodic
maintenance includes replacement and disposal of the
Cd109 source every 2 years at a cost of £3,500 to
$3,800. For optimum performance, the Cd'09 source
must be "deshimmed" every 6 to 10 months at a cost of
SI,500. Destumming. is the process of removing
shielding around the source to keep emissions nearly
constant. Because the TN Lead Analyzer contains a
radioisotope, a wipe test must be conducted once every
6 months at a cost of $40. Table 4-3 provides the
relative costs for TN Lead Analyzer analysis and gives
the cost for reference method analysis for comparison
purposes.
The developer offers a training course at its offices
or on site. The cost of a 2-day training course at the
developer's office is only the cost of travel per student.
The cost of an on-site course is $1,000 per day, plus
travel expenses for the developer's instructor. Costs
associated with the operator vary depending on the
technical knowledge and experience of the operator. As
discussed earlier, the TN Lead Analyzer is designed to
be used by individuals with no more than a high school
education and a minimal amount of technical training,
thereby decreasing costs.
Performance Factors
The following paragraphs describe performance
factors, including detection limits, sample throughput,
and drift.
Detection Limits
MDLs. using SW-846 protocols, were determined
by collecting 10 replicate measurements on site-specific
soil samples with metals concentrations 2 to 5 times the
38


-------
TABLE 4-2. INCIDENTAL ITEMS AND COSTS—TN LEAD ANALYZER
COST PER MODE OF OPERATION*
Item
IN SITU (5)
INTRUSIVE {$)
Sample cups (2 packages of 100)
Not required
100
X-ray window film (1 package of 1,000)
Not required
100
Two SRMs
450
450 .
60-mesh sieve with pan and lidb
Not required
125
Aluminum oxide mortar and pestleb
Not required
too
Drying ovenb
Not required
100
Fluorescein dye
Not required
30
Drying pansb
Not required
25
Paper towels and decontamination supplies
20
20
Rental of laptop computer
m
200
Total:
670
1.250
Notes:
SRM Standard reference materials.
a FPXRF analysis of 100 soil sampies.
b Optional, required for extensive sample preparation and expanded quality control checks.
TABLE 4-3. RELATIVE ANALYTICAL COSTS—TN LEAD ANALYZER

TN LEAD ANALYZER
Reference
Methods
3050A/601 OA
No. of Samples
Purchase® f$)
Lease ($)b
Cost Per
Sample ($)
30 (assume 1 day for
analysis, 1.5 days for
per diem, and 1 day
total for
mobilization/de-
mobilization time)
39,500
750 (extra battery)
400 (operator labor® S5Q/hr)
150 (per diem for operator)
400 (mobilization/demobilization0)
1.250 fsuDolies from Table 4-21
3,000/2 weeks
100 (shipping)
400 (operator labor @ $50/hr)
150 (per diem for operator)
400 (mobilization/demobilization0)
1.250 fsuoolies from Table 4-2)
150/sample
42.450
1,415/sample
5,300
176.67/sample
80 (assume 2 days
for analysis,
additional labor and
per diem charge only
vs. 30-sample
scenario)
42,450 (from 30-sample scenario)
50 (per diem for operator)
400 (ODerator labor  S50/hr>
5,300 (from 30-sample scenano)
50 (per diem for operator)
400 (ooerator labor lS> $50/hrt
1 SO^sample
42,900
536.25/sample
5,750
71.88/sample
300 (assume 6 days
for analyses,
additional labor and
per diem charge only
vs. 80-sample
scenario)
42,600 (from 80-sample scenario)
400 (per diem for operator)
1.600 f ooerator labor ® S50/hri
5,750 (from 80-sample scenario)
400 (per diem for operator)
1.600 ("ooerator labor ffi S50/hri
150/sample
44,900
149.67/sample
7,750
25.83/sample
Note:
This purchase scenario is based on a unit purchase for a single project The cost per sample for this
scenario would significantly decrease if the purchased unit were used on multiple projects.
Training costs not included.
Travel costs not included.
39



-------
expected MDLs. This data was obtained during the
precision study. Based on this precision data, a standard
deviation was calculated and the MDLs were defined as
3 times the SD for each analyte. All the precision-based
MDLs were calculated for soil samples that had been
dried and ground and placed in a sample cup, the highest
degree of sample preparation. The precision-based
MDLs for the TN Lead Analyzer are shown in Table 4-
4.
The developer did not provide MDLs specific to the
TN Lead Analyzer. However, the developer does have
published MDLs for the TN 9000 that were acquired
using a 200-second count lime for the Cd109 source on a
Si02 blank free of any potential interferences, but spiked
with the target analytes. According to XRF counting
statistics, the precision-based MDLs will vary by the
square root of the count time. Therefore, since the
count time used in this demonstration was 60 seconds
whereas the developer MDLs were obtained using a 200-
second count time, the developer-listed MDLs were
multiplied by a factor of 1.82 (square root of 3.33 for
comparison purposes). These corrected developer
MDLs are shown in Table 4-4.
Based on the demonstration data, most of these
precision-based MDLs seemed reasonable except
chromium. No reported chromium data below 1,000
mg/kg met the developer's data acceptance criteria. The
developer recommends that no measurement smaller than
3 times its SD be considered valid. The precision-based
MDLs were similar to about 1.5 times higher from the
developer MDLs. It was expected that the precision-
based MDLs would be slightly higher due to increased
matrix inteferences inherent in environmental samples.
Another method of determining MDLs involved the
direct comparison of the FPXRF data and the reference
data. When these sets of data were plotted against each
other the resultant plots were linear. This method is
discussed in greater detail in the "Intermethod
Assessment" later in this section. As the plotted line
approached zero for either method, there was a point at
which the FPXRF data appeared to respond either
randomly or with the same reading for decreasing
concentrations of the reference data. Figure 4-1 shows
an example data plot for copper for the TN Lead
Analyzer to illustrate this effect. By determining the
mean values of this random or singular FPXRF data and
subsequently two SDs around this mean, it was possible
to determine a field or performance-based MDL for the
analyzer. For the TN Lead Analyzer these field-based
MDLs are shown in Table 4-4.
Except for lead and arsenic these field-based MDLs
are 2 to 3 times greater than the developer and precision-
TABLE 4-4. METHOD DETECTION LIMITS—
TN LEAD ANALYZER
Analyte
Developer-
based MDL"
(mg/kg)
Precision-
based MDL
(mg/kg)
Field-based
MDL
(mg/kg)
Arsenic
46
50
77
Chromium
330
460
2.400
Copper
80
115
216
Lead
25
40
44
Zinc
64
95
168
Note:
a Corrected to reflect 60-second count time for
the Cd109 source.
based MDLs. Lead and arsenic MDLs were similar for
both MDL determination approaches.
The other primary and secondary target analytes
(antimony, barium, cadmium, and nickel) for this
demonstration were not reported by this analyzer.
Throughput
The TN Lead Analyzer used a Cd109 source count
time of 60 seconds. With the additional "dead" time of
the system and the time required to label each sample
and store data between sample measurements; the time
required to analyze one soil sample was 2 to
2.5 minutes. This resulted in a throughput of
approximately 25 samples per hour. The developer
claimed an operator could analyze at least 100 samples
per day. The operator found this to be a conservative
estimate. The minimum number of samples analyzed in
a 10-hour day, during the demonstration, was 155
samples. This was for in situ measurements in the field
at the ASARCO site where the operator sometimes had
to tTaverse large distances (up to 0.5 mile) between
samples. The maximum number of samples analyzed in
a 12-hour day was 330 samples for intrusive
measurements at the ASARCO site.
This throughput took into account the time necessary
to analyze daily an average of three QC samples, such as
one SiOj blank, one pure iron sample calibration check,
and one NIST SRM. These QC sample analyses are
recommended by the developer. The sample analysis
time did not include the time required for sample
handling and preparation, or for data downloading,
printing, and documentation. Considerable time was
spent preparing the in situ homogenized samples and the
intrusive samples. Homogenization required an average
of approximately 5 minutes per sample (in situ-
40


-------
100000
10000
100 1000 10000 100000
Reference Data (mg/kg)
FIGURE 4-1. CRITICAL ZONE FOR THE DETER-
MINATION OF A FIELD-BASED METHOD DETECTION
LIMIT FOR COPPER: Between 100 and 200 mg/kg for the
reference data the linear relationship between the two data
sets changes. This point of change identified the point at
which field-based MDLs for the analyzer were determined.
prepared), 20 minutes per sample were required for No.
10 sieving (instrusive-unprepared), and 10 minutes per
sample were required for grinding and sieving (intrusive-
prepared). Approximately 30 minutes were spent daily
downloading the data to a PC and printing out a hard
copy.
Drift
Drift is a measurement of an analyzer's variability
in quantitating a known amount of a standard over time.
For the TN Lead Analyzer, drift was evaluated by
reviewing results from the analysis of NIST SRM 2710.
This SRM contained quantifiable levels of arsenic,
copper, lead, and zinc, and it was analyzed four times by
the TN Lead Analyzer during the demonstration. This
data was reduced to RSDs for the target analytes and the
percent drift from the mean recovery of the true value.
The percent drift from the mean recovery for each day
is shown on Figure 4-2 for the target analytes reported
by the TN Lead Analyzer. The RSD values for all
analytes were less than 8 percent and the mean percent
recoveries were between 90 and 100 percent. The RSD
values for copper, lead, zinc, and iron were all less than
3 percent and 7 percent for arsenic. These low RSD
values and high percent recoveries indicate that for the
concentrations of analytes found in the SRM, the TN
Lead Analyzer displayed little drift during the
demonstration. The minimal drift that did occur was
less than the 10 percent limit listed in the demonstration
QAPP for Level 3 data. The developer has established
no guidelines for drift.
Intramethod Assessment
lntramethod measures of the technology's
performance included its results on analyzer blanks, the
completeness of its results, its intramethod precision,
intramethod accuracy, and intramethod comparability.
The following paragraphs discuss these five
characteristics.
Blanks
Analyzer blanks for the TN Lead Analyzer consisted
of Si02 blanks. These blanks were routinely analyzed
at the beginning and end of the day or at the beginning
and in the middle of the day. They were used to monitor
for contamination of the probe by material such as
residual soil left on the face of the probe. A total of 20
Si02 blanks were analyzed during the demonstration.
None of the target analytes were detected in any of the
20 blanks. The results of the blanks demonstrated that
there was no problem with cross contamination from
sample to sample or with contamination on the window
of the probe.
Completeness
A total of 3 IS soil samples were analyzed four times
(four preparation steps) resulting in 1,260 sample
results. The TN Lead Analyzer produced results for all
1.260 samples for a completeness of 100 percent, above
the demonstration objective of 95 percent. This
demonstrated the reliability and mggedness of this
analyzer.
Precision
Precision was expressed in terms of the percent
RSD between replicate measurements. The percent RSD
is defined as the SD divided by the mean concentration
times 100. The precision data for the target analytes
detectable by the TN Lead Analyzer are shown in Table
4-5. The precision data reflected in the 5 to 10 times the
MDL range reflects the precision generally referred to
in analytical methods such as SW-846, and represent
general method precision.
The TN Lead Analyzer performed 10 replicate
measurements on 12 soil samples that had analyte
concentrations ranging from less than 50 mg/kg to tens
of thousands of mg/kg. Each of the 12 soil samples
underwent the four different sample preparation steps
described previously in Section 2.0. Therefore, there
were a total of 48 precision points for the TN Lead
Analyzer. The replicate measurements were taken using
the source count times discussed in the previous section
41


-------
FIGURE 4-2. DRIFT SUMMARY—TN LEAD ANALYZER: This graph illustrates the drift experienced by the
analyzer at the two demonstration sites.
g







B

~ ~

o

. n 	
a

~
a

J
I
~ D j

~






Arsenic	Copper	lead	Zinc	Iron
Analyte
of this report. For each detectable analyte in each
precision sample, a mean concentration, SD, and RSD
was calculated.
In this demonstration, the analyzer's precision or
RSD for a given analyte had to be less than or equal to
20 percent to be considered Level 2 data and less than or
equal to 10 percent to be considered Level 3 data. The
analyzer's precision data, reflected by the precision data
in the 5 to 10 times MDL range, were all below the 10
percent RSD required for Level 3 data quality
classification. Chromium data was not represented in
this range. Table 4-5 shows that chromium precision
was greater than 20 percent, placing the chromium
results in the Level 1 data quality classification based on
precision. The lower precision for chromium was
expected because chromium is a problematic analyte for
FPXRF analysis, especially at 60-second count times.
There was no observable effect of sample
preparation on precision. This was expected because the
method used to assess precision during this demon-
stration was measuring analyzer precision, not total
method precision. There was a concentration effect on
the precision data as shown on Figure 4-3. Except for
chromium, precision increased with increasing
concentration. In addition. Figure 4-4 shows an
asymptotic relationship between concentration and
precision. In this figure, precision shows tittle
improvement at concentrations greater than 250 ppm,
however, at concentrations below 250 ppm, precision is
highly concentration dependent. Although lead is shown
on Figure 4-4, this send was true for all primary
analytes. The precision samples were purposely chosen
to span a large concentration range to test the effect of
analyte concentration on precision.
Accuracy
Accuracy was assessed for the TN Lead Analyzer
by using site-specific PE samples and SRMs. Accuracy
was evaluated by comparing percent recoveries for each
target analyte reported by the TN Lead Analyzer. The
TN Lead Analyzer analyzed six site-specific PE samples
and 14 SRMs. The operator knew the samples were PE
samples or SRMs, but did not know the true
concentration or the acceptance range. These PE
samples and SRMs were analyzed the same way as all
other samples.
The six site-specific PE samples consisted of three
from each of the two demonstration sites. These six PE
samples were collected during the predemonstration
activities and sent to six independent laboratories for
analysis by laboratory-grade XRF analyzers. The mean
measurement for each analyte was used as the true value
concentration. The 14 SRMs included seven soil, four
stream or river sediment, two ash, and one sludge SRM.
The SRMs were obtained from NIST, USGS,
Commission of European Communities-Community
Bureau of Reference, National Research Council-
Canada. and the South African Bureau of Standards.
The SRMs contained known certified concentrations of
certain target analytes.
These PEs and SRMs did not have published
acceptance ranges. As specified in the demonstration
plan (PRC 1995), an acceptance range of 80 to 120
percent recovery of the true value was used to evaluate
accuracy for the six site-specific PEs and 14 SRMs.
Table 4-6 summarizes the accuracy data for the target
analytes for the TN Lead Analyzer. Figures 4-5 and 4-6
show the true value, the measured value, and percent
42


-------
TABLE 4-5. PRECISION SUMMARY—TN LEAD
ANALYZER

Mean % RSD Values by
Concentration Range
Analyte
5-10
Times
MDL*
(mg/kg)
50 • 500
(mg/kg)
500-
1,000
(mg/kg)
>1,000
(mg/kg)
Arsenic
4.11 (16)
16.47 (8)
3.47 (12)
2.30 (8)
Chromium
ND
ND
21.73(12)
24.62 (4)
Copper
9.11 (8)
18.00 (24)
5.82 (4)
2.60 (12)
Iron
ND
ND
ND
2.18 (48)
Lead
5.93 (12)
8.93 (12)
5.02 (8)
2.52 (20)
Zinc
7.48 (16)
13.42 (24)
7.12 (16)
ND
Notes:
a The MDLs referred to in this column are the
precision-based MDLs shown in Table 4-4.
mg/kg Milligrams per kilogram.
ND No data.
() Number of samples, including all four
preparation steps, each consisting of 10
replicate analyses. Numbers do not always add
up to 48 precision points because some samples
had analyte concentrations below the analyzer's
MDL.
recovery for the individual SRMs and PEs, respectively.
No figure was presented for chromium because only one
sample produced a detectable concentration of chromium
by the TN Lead Analyzer. True value results from the
site-specific PEs and SRMs with concentrations less than
the precision-based MDLs listed in Table 4-4 were
excluded from the accuracy assessment.
Based on the 80 to 120 percent recovery acceptance
range, the TN Lead Analyzer's accuracy varied from 0
percent for chromium to 100 percent for arsenic and
copper in the site-specific PEs. Overall, the TN Lead
Analyzer produced 20 out of 28 results or 71.4 percent
within the 80 to 120 percent recovery acceptance range
for all analytes in the six site-specific PE samples.
Seven of the eight results falling outside of the
acceptance range were below the lower limit of 80
percent recovery. Only the 129 percent recovery for
chromium in one sample was above the upper limit of
120 percent recovery. For all six site-specific PEs, only
three out of 28 percent recoveries were above 100
percent. Table 4-6 also shows that the mean percent
recoveries for all six analytes in the PEs were less than
100 percent. This indicates that, in general, the TN
Lead Analyzer was producing results that were biased
slightly low. However, with the exception of the two
chromium recoveries of 0 and 129 percent, the TN Lead
Analyzer produced percent recoveries for all analytes
between 70 and 107 percent (Table 4-6), which indicated
good accuracy for these samples. This range of percent
recoveries was slightly below the acceptance range of 80
to 120 percent.
Table 4-6 summarizes the accuracy data for the
SRMs. A more detailed analysis of the SRM data is
presented on Figure 4-5. A graph is not presented for
chromium because no samples produced a detectable
chromium concentration by the TN Lead Analyzer. The
TN Lead Analyzer accuracy for the SRMs varied from
0 percent for chromium (only one result for chromium)
to 100 percent for iron. The iron concentrations in the
SRMs were in the tens of thousands of mg/kg which is
in a concentration range that the TN Lead Analyzer
should perform well. Some analytes such as copper,
lead, and zinc had concentrations spanning one or more
orders of magnitude in the SRMs. Overall, the TN Lead
Analyzer produced 31 out of 42 results within the 80 to
120 percent recovery acceptance range for an accuracy
of 73.8 percent. Of the 11 results that fell outside of the
acceptance range, 6 results were low and S were high.
This nearly equal ratio of high results to low in addition
to the mean percent recoveries shown in Table 4-6
indicates that the TN Lead Analyzer was not showing a
high or low bias for copper, iron, lead, and zinc. Hie
TN Lead Analyzer did appear to show a slightly low bias
for arsenic concentrations. The chromium results were
inconclusive. Except for chromium, the TN Lead
Analyzer produced percent recoveries ranging from 38
percent for copper in one sediment SRM to 151 percent
for zinc in the one sludge SRM.
A more detailed analysis of the SRM data showed
that there was a matrix effect on the TN Lead
Analyzer's accuracy (Table 4-6). The TN Lead
Analyzer produced 16 out of 16 results or 100 percent
within the acceptance range for all target analytes in the
seven soil SRMs. This demonstrated that the TN Lead
Analyzer was more accurate when analyzing SRMs that
closely matched the matrix used to set the fundamental
parameters of the analyzer. The TN Lead Analyzer
showed the lowest comparability to the one sludge SRM
by overestimating all analyte concentrations by a factor
of 1.3 to 1.5. The overall accuracy was 60 percent for
the four sediment SRMs and 75 percent for the two ash
SRMs. Specifically, two sediment, one ash, and the
one sludge SRM accounted for all 11 results that fell
outside of the acceptance ranges. This indicates that
SRMs of a different matrix (sediment, ash, or sludge)
than that of soil may not serve as adequate accuracy
checks when the FP calibration is based on soil SRMs.

-------
o
? 0
(0

u.


1 5
c

0>
1 0
o


0)

CL


0
I
A rse n ic
Copper
A naly te
Lead
5-10 tim es MDL
15 0 - 5 0 0 m g /kg
I—i 500 - 1,000 mg/kg l~~1 > 1.000 m g / kg
FIGURE 4-3. PRECISION SUMMARY—TN LEAD ANALYZER: This graph illustrates the precision of the analyzer
over defined concentration ranges. The precision expressed at 5 to 10 times the MDL is generally used to describe
overall precision.
o
c/d
a.
¦o
CO
0)
40
30 r
20
1 0 _
2	4
Thousands -
Lead Concentration (mg/kg)
FIGURE 4-4. PRECISION VS. CONCENTRATION—TN LEAD ANALYZER: This graph illustrates the TN Lead
Analyzer's precision as a function of analyte concentration.
The TN Lead Analyzer was the least accurate for
chromium when assessing the site-specific PEs and
SRMs. This was not unexpected for two reasons. First,
two of the three samples shown in Table 4-6 had
concentrations less than 2 times the precision-based
MDL for chromium, which may have negatively affected
the results. Second, the developer did not design this
analyzer to analyze for chromium and was not certain
what the TN Lead Analyzer's capabilities for chromium
would be. The overall accuracy for the remaining five
analytes for the PEs and SRMs combined was similar,
ranging from 71 percent for zinc to 83 percent for iron.
The TN Lead Analyzer was expected to perform well for
iron given that the iron concentrations in the PEs and
SRMs were well above MDLs yet in a linear range for
the TN Lead Analyzer.
Comparability
Intramethod comparability for the TN Lead
Analyzer was assessed through the analysis of four ERA
PEs and four CRM PEs. This was done to present
44


-------
Notes:
n	Number cf samptes with detectable anaiytes.
3D	Standard deviation,
mgflcg per Kilogram.
MA	Moi applicable. Star.SarddevisSon notcar,cutaneator h.o artess resJfcs.
TABLE 4-S. ACCURACY SUMMARY FOR SITE-SPEC IRC PE AND 5RM RESULTS—
TN LEAD ANALYZER
Analyte
n
Percent
Within Acceptance
Range
Mean
Percent
Recovery
Range of
Percent
Recovery
SD of
Percent
Recovery
Concentration
Range (mg/kg)
Site-Specific Performance Evaluation Samples
Arsenic
3
100
69
87 -92
2.5
424 - 22.444
Chromium
2
0
65
0-129
NA
939 - 3,800
Copper
5
100
92
S3-107
8.9
300-7,132
Iron
6
67
a?
70-98
12
27.320-70,500
Lead
6
67
87
70- 101
12
292 -14.663
Zinc
6
67
82
70-30
7.0
164 - 4.205
Soil Standard Reference Materials
Arsersic
3
100
a?
B5-97
6.3
105 - 626
Copper
1
too
92
92
NA
2,950
Iron
3
100
94
69-99
4.7
28,900 - 35.000
Lead
5
100
101
67-116
12
101 - 5,532
Zinc
4
loo ;
101
63-118
15
350 - 6,952
Sediment Standard Reference Materials
Arsenic
1
0
44
44
NA
211
Chromium
1
0
0
0
NA
509
Copper
4
75
B5
38-106
32
99 - 452
Iron
1
100
99
99
NA
41,100
Lead
4
50
100
75- 131
23
161 - 5,200
Zinc
4
75
97
81 - 126
21
264-2200
Ash & Sludge Standard Reference Materials
Arsenic
2
50
87
73-101
NA
136 -145
Copper
1
0
141
141
MA
696
Iron
2
100
86
85-86
NA
77.800 - 94,000
Lead
3
67
106
88-133
23
68-286
Zinc
3
33
109
56-151
41
210-2,122
potential users additional information on data
comparability relative to different commercially
available QC samples. The eight PEs were analyzed in
the sane way as al) other samples. As described in
Section 3, these eight PE samples had certified analyte
values determined by Methods 3050A/6G10A.
Therefore, since these methods do not necessarily
determine total metals concentrations in a soil, it was
expected that the analyzer would overestimate analyte
concentrations relative to PALs. The ability of the TN
bead Analyzer to produce results within the PALs and
the percent recovery for each of the analytes were used
to evaluate the TN Lead Analyzer's mtramethod
comparability. True value analyte concentrations in the
45


-------
800
550
.5
n
300
50


	[

1
r
JLr
if
12 0
100
$
eo
60 £
40
Arsenc
(Measured Value p True Value
q Percent Recovery
100
80
.5 a eo
40
20
1
ll
120
100
fr
80
60
40
£
ton
(MeasLrsd Value qTrue Value
~ Percent Recovery
10000,
1000
IS
100
10
150
120
90
60
30

<5
Copper
I Measured Value oTmeMa(ue a Percent Recovery
10000
.5
¦»
1000
100
10
Lead
10000
125 >
MeasiradVaJue o True Value a Percent Recovery
160
*2°
SO
40
4
iMeasimd Value cjTrue Value q Percent Recovery
FIGURE 4-5. SRM RESULTS—TN LEAD ANALYZER: These graphs illustrate the relationship between the TN
Lead Analyzer's data (measured values) and the true values lor the SRMs. The gray bars represent the percent
recovery for the TN Lead Analyzer. Each set of three bars (black, white, and gray) represent a single SRM sample.
46
HP A vr

-------
100000
MeasuHlVttue ~Truexoue
[
I
MeasiredVaue ~ True Value ~ Percent Recovery
10000
1000
125
100
75
50
<5
GCpper
iMeasiredNttus DTrueNfelue
100000
10000
1000
100 —
Lead
[Measured Value OTrue Value
Measijed Value aTrueNraue oPereert Recovwy
~ Percert Recovery
120
100
60
60
40
I
&
~Percent Recovery
FIGURE 4-6. SITE-SPECIFIC PE SAMPLE RESULTS—TN LEAD ANALYZER: These graphs illustrate the
relationship between the TN Lead Analyzer's data (measured values) and the true values for the site-specific PE
samples. The gray bars represent the percent recovery for the TN Lead Analyzer. Each set of three bars (black,
white, and gray) represent a single site-specific PE sample.
47
"A -ya»
'L.. - -

-------
ERA and CRM PEs that were below the precision-based
MDLs in Table 4-4 were excluded from the intramethod
comparability assessment.
The TN Lead Analyzer performance data for all
target analytes for the eight CRMs and PEs are
summarized in Table 4-7. The measured values, true
values, and percent recoveries for all detectable analytes
are shown on Figure 4-7. No figure is shown for
chromium because there was only one detect for
chromium. For the ERA PEs, the TN Lead Analyzer
produced 12 out of 18 results or 66.7 percent within the
acceptance range. For the CRMs, the TN Lead
Analyzer produced 8 out of 17 results or 47.0 percent
within the acceptance range. With the ERA and CRM
PEs combined, the TN Lead Analyzer produced 20 out
of 35 results or 57.1 percent within the acceptance
range. Based on the data presented in Table 4-7, the TN
Lead Analyzer's results were more comparable to the
ERA PEs than the CRMs. The better comparability to
the ERA PEs versus the CRMs was unexpected because
the ERA PEs had lower analyte concentrations than the
CRMs. With the exception of iron, the anaJyte
concentrations in the ERA PEs were all less than 350
mg/kp which is less than 5 times the MDL for most of
the analytes.
All six results outside the acceptance limits for the
ERA PEs were above the upper control limit. Four of
the out of range results in the ERA PEs were for iron,
one for copper, and ore for lead. The TN Lead Analyzer
was overestimating iron concentrations in the ERA PEs
by a factor of two shown by the mean percent recovery
in Table 4-7. The one out-of-range result for copper
and one out-of-range result for lead were from samples
with copper and lead concentrations less than 2 times the
MDL for the TN Lead Analyzer.
The TN Lead Analyzer produced only two out of 18
percent recoveries that were less than 100 percent for the
ERA PEs. All mean percent recoveries for the analytes
in the ERA PEs were greater than 100 percent. This
indicates that the TN Lead Analyzer was overestimating
the results compared to the certified values. This is
consistent with the fact that FPXRF is a totals metal
technique whereas Methods 3050A/6010A used to certify
the results in the ERA PEs are not.
Of the nine results outside of the acceptance limits
in the CRMs, two were above the upper control limit
and the other seven were below the lower control limit.
The comparability of the TN Lead Analyzer's results to
the certified values in the CRMs was sample dependent.
For one of the soil CRMs and the ash CRM, the TN
Lead Analyzer produced 7 out of 7 or 100 percent of the
results within the acceptance limits, indicating good
comparability. For the other soil CRM and the sludge
CRM, the TN Lead Analyzer produced only 1 out of 10
or 10 percent of the results within the acceptance limits,
indicating poor comparability. It is possible that
interferences were causing the poor comparability in
these two CRMs. The soil CRM contained nearly 20
percent iron which is much above the .normal 2 to 10
percent iron found in most soil samples. The chromium
content in the sludge CRM was about 16 percent. The
recoveries of copper, iron, lead, and zinc in the sludge
CRM were all about 50 percent. In both CRMs, it is
possible that the FPs may not have been able to
compensate for the high concentrations of iron or
chromium.
Intermethod Assessment
The comparison of the analyzer's results to the
results of the reference methods was conducted using the
statistical methods detailed in Section 2. The purpose of
this statistical evaluation was to determine the
comparability between the FPXRF data and the
reference data. If the FPXRF data were statistically
equivalent to the reference data, they met the Level 3
data quality criteria. If the data did not meet the Level
3 criteria, but could be mathematically corrected to be
equivalent to the reference data, the FPXRF data met the
Level 2 criteria. If the FPXRF data did not meet the
Level 3 criteria, and the statistical evaluation could not
identify a predictable bias in the data, but the analyzer
identified the presence or absence of contamination with
at least a 95 percent accuracy rate, the data were
classified as Level 1 quality.
The TN Lead Analyzer was configured to report
concentrations for five of the six primary analytes, and
one of the secondary analytes. The primary analytes it
reported were arsenic, chromium, copper, lead, and
zinc. Iron was the only secondary analyte reported by
this analyzer.
The regression analysis on the entire data set
indicated that arsenic, copper, lead, and zinc all
exhibited r^s of 0.90 or greater. In all of these cases,
the slopes and y-intercepts were not significantly
different from their ideal values of 1 and 0, respectively.
Additional data evaluation involved the assessment
of the potential influence of the variables site, soil type,
and sample preparation on the regression analysis.
Analysis indicated no apparent impact of the site variable
on the regression. The soil type variable exhibited a
reduced r for zinc in the loam soil by 0.07 r2 units.
The sample preparation variable exhibited the greatest
influence on the regression analysis producing as much
as a three-fold (up to 0.576 r2 units) increase in the r2
48


-------
TABLE 4-7. ACCURACY SUMMARY FOR PE AND CRM RESULTS—TN LEAD ANALYZER
Analyte
n
Percent Within
Acceptance Range
Mean
Percent
Recovery
Range of
Percent
Recovery
SO of
Percent
Recovery
Concentration
Range 
-------
5> 350
Arsenic
HMeasured Value ~ True Value ~ Percent Recovery
1000000
100000
10000
1000
250
Iron
(Measured Value ~ True Value
~Percent Recovery
100000
10000
| 1000
•fa
| 100
10
m
10000
160
120 «
Copper
¦Measured Value ~True Value	~Percent Recovery

1000000
ct>
100000
l!
10000


ID
1000
U

100 _
i


200
>.
150 5
100
50
5
£
I
o
<£
Lead
(Measured Value ~True Value
~ Percent Recovery
150
125
a>
100 S
75
1 50
-25
5
O
6
Zinc
(Measured Value oTrue Value
q Percent Recovery
FIGURE 4-7. PE AND CRM RESULTS—TN LEAD ANALYZER: These graphs illustrate the relationship between
the TN Lead Analyzer's data (measured values) and the true values for the PE and CRM samples. The gray bars
represent the percent recovery for the TN Lead Analyzer. Each set of three bars (black, white, and gray) represent a
single PE or CRM sample.
a*®* .*¦. • %..

-------
TABLE 4-8. REGRESSION PARAMETERS8 BY VARIABLE—TN LEAD ANALYZER
Variable
Arsenic
Chromium
Copper
n
f2
Std. Err.
Y-lnt.
Slope"
n
r*
Std. Err.
Y-lnt.
Slope"
n
r1
Std. Err,
Y-lnl.
Slope"
All Data
SI 5
0.952
0.15
0.20
0 95
136
0.54B
0.16
2.31
0.39
957
0 940
0 17
0 48
0 09
ASARCO Site
80S
0.958
0.14
0.18
0.95
5
0.017
0.07
3.10
0.03
745
0.961
0.14
C.1S
0.88
RV Hopkins Site
8
ND
NO
ND
ND
131
0.585
0 15
2.03
049
145
0.516
0.15
1 44
048
Sand Soil
357
0.966
0.14
o.ts
0.95
2
ND
ND
NO
ND
366
0.946
0.13
0 14
0.98
Loam Soil
449
0.950
0.14
0.20
0.95
3
NO
NO
ND
NO
443
0.951
0.13
0 42
0.92
Clav Soil
8
ND
ND
NO
ND
131
0.585
0.15
2.03
0.49
145
0.516
0 15
1 44
0 4B
In situ-Unprepared
211
0.684
0.22
0.45
0.83
28
0.237
0.17
2.71
0.27
246
0.066
0.26
0.67
0.82
In situ-Prepared
200
0.973
0.11
0,11
097
35
0.554
0.15
2.59
0.29
251
0.942
0.17
0 59
C 85
Intrusive-
Unprepared
204
0.964
0.08
0.10
0.99
40
0.571
0.15
2.13
0.46
242
0.967
0.13
0.38
0.92
Imruswe-Prepared
201
0.9E1
0.09
0.16
0.98
33
0.B07
0.12
1.16
0.77
225
0.975
011
0.21
0.99
TABLE 4-8 (Continued). REGRESSION PARAMETERS8 BY VARIABLE—TN LEAD ANALYZER
Variable
Lead
Zinc
n
r*
Std. Err.
Y-lnt.
Slope"
n
r*
Std. En.
Y-lnt
Slope"
All Data
1165
0.950
0.14
0.30
0.92
1079
0.923
0.12
0.42
0.90
ASARCO Site
780
0.943
0.15
0.22
0.95
732
0.914
0.13
0 43
0.69
RV Hopkins Site
385
0 964
0.11
0 43
0 87
347
0.941
0.11
041
0 90
Sand Soil
347
0.951
0.14
0.21
0.93
322
0.947
0.12
0.33
0.91
Loam Soil
430
0.943
014
0.25
0.95
411
0.673
0.12
0.56
0.86
Clay Soil
385
0964
0.11
0.43
087
347
0941
0.11
0 41
0.90
In situ-Unprepared
296
0.849
0.23
0.48
0.83
233
0.B25
0.18
0.63
0.81
(n situ-Prepared
300
0.960
012
0.36
0.B9
279
0.948
0.10
0.46
0.8S
f ntr usrve-Un preoa red
29B
0.978
0 09
9.23
0 94
27C
0.94 S
0.10
0.35
0.93
Intrusive-Prepared
293
0.975
0 10
0.21
096
250
0.962
0.09
0.25
0.97
Notes:
a	Regression parameters based on log10 transformed data
6	Slope values determined with FPXRF data plotted on x-axis and the reference data plotted on the y-axis.
n	Number of data points.
Int.	Intercept.
S(d. Err.	Standard error.
ND	Analyies not present in significant quantities to provide meaningful regression.
concentration-sorted data sets was evident. This
indicates that the correlation is independent of
concentration for these ranges, and that the regression
analyses associated with the entire data set are
representative of the relationship between the analyzer's
data and the reference data. After examining the
analyzer versus reference data plots, a slight shift in the
slope of the plot was noticed at approximately 2,000
mg/kg (Figure 4-8). When the data was assessed in the
0 to 2,000 mg/kg and greater than 2,000 mg/kg
concentration ranges, a definite concentration effect was
noticed. The regression parameters were better, in all
cases for the data in the 0 to 2,000 mg/kg concentration
range. Lead exhibited the greatest effect, this analyte
did not meet Level 3 data quality criteria in the greater
than 2,000 mg/kg range. Identification of the exact
cause of this concentration effect is beyond the scope of
this project. Possible causes include changes in
51


-------
100000
10000
1000
100
In situ-unprepared—Lead
10



-
Ju v
r +¦



i

i
10	100 1000 10000
Reference Data (mg/kg)
100000
htrusrve-unprepared-Lead
100000
5 10000
1000
100 1000 10000
Reference Data (mg/kg)
100000
In situ-unprepared—Arsenic
100000

10000
1000
100 -
100 1000 10000
Reference Data (mg/kg)
100000
100000
Intrusive-unprepared—Arsenic
o
&>
-2-
"D
n

-------
» 3
3
o
(0
° 1
<
a.
TN Lead A nalyzer
Zinc
Arsenic Chromium Copper Lead
A nalyte
C3 0 ata 0 uality Level	n Sam pie P reparatio n Step
Iron
FIGURE 4-9. HIGHEST DEGREE OF DATA QUALITY AT THE LOWEST DEGREE OF SAMPLE
PREPARATION—TN LEAD ANALYZER: This graph illustrates the highest data quality level that the TN Lead
Analyzer met for each analyte. The small black squares show the lowest degree of sample preparation that
produced the data quality level shown.
reference method accuracy at higher concentrations due
to analyte interferences, and shifts in FPXRF
performance at higher concentrations due to detector
characteristics, or inherent characteristics of the FP
calibration. Whatever the cause, this apparent
concentration effect has a minor effect on overall data
quality.
Iron was the only secondary analyte measured by
this analyzer. Correlation of iron data with the reference
method data indicated thai the data meets the Level 2
criteria after the initial sample homogenization. Sample
preparation did affect this correlation. As noted above,
the greatest increase in correlation was observed after
the initial sample homogenization.
To examine the potential effect of count times on
analyzer comparability, a subset of 26 intrusive-prepared
samples from the RV Hopkins site were analyzed using
doubled count times. This increase in count times
increased the r for both chromium and copper 0.02 and
0.14 units, respectively. None of the" other target
analytes exhibited a count time effect (r^s did not
change) at the count times evaluated.
53
¦7"-i •«. 'six
r' ;j

-------
TABLE 4-9. REGRESSION PARAMETERS8 BY THE SAMPLE PREPARATION VARIABLE AND
SOIL TYPE—TN LEAD ANALYZER
Soli Type
Arsenic
Chromium
Copper
n
i2 (Std. Err.
Y-tnt.
Slope"
n
r2
Std. Err.
Y-lnt.
Slope"
n
r2
Std. Ert.
Y-lnt
Slope"
In Situ-Unprepared
Sand Soil
93
0.920
0.20
0.34
0.86
3
ND
ND
ND
ND
89
0.911
0.16
0.35
0.88
Loam Soil
114
0.876
0.21
0.38
0.87
3
ND
ND
ND
ND
112
0.818
0.27
0.58
0.87
Clay Soil
4
ND
ND
ND
ND
28
0.237
0.17
2.71
0.27
42
0.619
015
1 40
0 53

In Situ-Prepared
Sand Soil
89
0.981
0.10
0.09
0.97
3
ND
ND
ND
ND
90
0.958
0.12
0.09
0.99
Loam Soil
109
0.973
0.10
0.17
0.96
3
ND
ND
ND
ND
113
0.957
0.12
0.54
0.87
Clay Soil
4
ND
ND
ND
ND
31
0.600
0.14
2.1B
0.43
45
0.524
0.11
1.58
0.38

Intrusive-Unprepared
Sand Soil
89
0.987
0.08
0.09
0.98
3
ND
ND
ND
ND
96
0.962
0.11
0.10
1.00
Loam Soil
113
0.987
0.07
0.13
0.99
3
ND
ND
ND
ND
114
0.9B2
0.08
0.36
0.93
Clay Soil
3
ND
ND
ND
ND
39
0.703
0.14
1.94
0.53
35
0 487
0.14
1.51
0 44

Intrusive-Prepared
Sand Soil
88
0.981
0.10
0.15
0.98
3
ND
ND
ND
ND
93
0.969
0.10
0.01
1.04
Loam Soil
113
0.984
0.08
0.17
0.99
3
ND
ND
ND
ND
113
0.980
0.09
0.32
Q.95
Clay Soil
3
ND
ND
ND
ND
33
0.807
0.12
1.16
0.77
22
0.470
0 16
0.24
1.05
TABLE 4-9 (Continued). REGRESSION PARAMETERS3 BY THE SAMPLE PREPARATION
VARIABLE AND SOIL TYPE—TN LEAD ANALYZER
Soil Type
Lead
Zinc
n r2
Std. Err.
Y-lnt.
Slope"
n
r2
Std. Err.
Y-lnt.
Slope"
In Situ-Unprepared
Sand Soil
85
0.871
0.21
0.35
0.84
87
0.910
0.14
0.60
0.78
Loam Soil
110
0.834
0.23
0.49
0.84
101
0.720
0.17
0.78
0.75
Clay Soil
99
0.845
0.22
0.74
0.77
96
0.B44
0.19
0.61
0.85

tn Situ-Prepared
Sand Soil
89
0.959
0.13
0.23
0.93
84
0.966
0.10
0.33
0.92
Loam Soil
111
0.956
0.12
0.39
0.90
107
0.899
0.10
0.66
0.81
Clay Soil
99
0.976
0 09
0.45
0.85
88
0.971
0.08
0.44
0.88

Intrusive-Unprepared
Sand Soil
88
0.980
0.09
0 14
0.97
77
0.972
0 09
0 21
0.97
Loam Soil
109
0983
0.08
0.12
1.00
106
0 910
0.10
0 49
0.89
Clay Soil
100
0.986
0 07
0.37
0.89
86
0.970
0.08
0.36
0.92

Intrusive-Prepared
Sand Soil
89
0.966
0.12
0 18
0.97
76
0.947
0.13
0.22
0.97
Loam Soil
105
0.985
0.07
0.09
1.03
96
0.960
0.07
0.29
0.97
Clay Soil
98
0.988
0.07
0.29
0.92
79
0.983
0.06
0.17
0 99
Notes:
3	Regression parameters based on (og10 transformed data.
0	Slope values determined with FPXRF data plotted on x-axis and the reference data plotted on the y-axis.
n	Number of useable matched pairs of data points.
Int.	Intercept.
Std. Err.	Standard error.
ND	Analyte not present in significant quantities to provide meaningful regression.
54



-------
TABLE 4-10. REGRESSION PARAMETERS" BY THE SAMPLE PREPARATION VARIABLE
AND SITE TYPE—TN LEAD ANALYZER
Site Name
Arsenic
Chromium
Copper
n
r2 fstd. Err.
Y-irn.
Slope"
n
T2
Std. Err. | Y-lnt.
Slope"
n
r2 |std. Err. Y-lnt
Slope"
In Situ-Unprepared
ASARCO
207
0.898
0.21
0.36
0.87
6
ND
ND
ND
ND
200
0 891
0 23
0.24
0,95
RV Hopkins
4
NO
NO
NO
ND
2B
3.237
0.17
2.71
0.27
42
0.619
0.15
1 40
0.53

In Situ-Prepared
ASARCO
199
0.975
0.11
0.11
0.97
6
NO
ND
ND
NO
202
0.963
0.13
0.22
0.96
RV Hopkins
3
ND
ND
ND
ND
31
).600
0.H
2.18
043
45
0.524
0 11
1.58
0 38

Intrusive-Unprepared
ASARCO
202
0.986
0.08
0.10
0.99
5
ND
ND
ND
ND
210
0.978
0.10
0 19
0.98
RV Hopkins
3
ND
NO
ND
ND
39
J.703
OH
1.94
0.53
36 (0.4B7
0 14
1 51
044

Intras ive-Pre pared
ASARCO
201
0.9B1
0.09
0.16
0.93
6
ND
ND
ND
ND
203
0.981
0.09
0.07 [ 1.03
RV Hopkins
3
NO
NO
ND
ND
33
3.807
0.12
1.16
0.77
22
0.470
016
0.24 f 1.05
TABLE 4-10 (Continued). REGRESSION PARAMETERS® BY THE SAMPLE PREPARATION
VARIABLE AND SITE TYPE—TN LEAD ANALYZER

Lead
Zinc

n
r2
Std. Err.
Y-lnt.
Slope"
n
r2
Std. Err.
Y-lnt.
Slope"
Site Name
In Situ-Unprepared
ASARCO
196
0.839
0.229
0440
0.838
188
0.831
0163
0.663
0.776
RV Hopkins
99
0.845
0.217
0.743
0.765
96
0.844
0.188
0.613
0.B49

In Situ-Prepared
ASARCO
200
0953
0.129
0.292
0.918
192
0.932
0.109
0.470
0.B75
RV Hopkins
99
0 976
0.087
0.449
0.848
88
0.971
0.078
0.435
0.879

Intrusive-Unprepared
ASARCO
198
0.978
0.091
0.126
0.988
183
0.935
0.112
0.339
0.937
RV Hopkins
99
0.987
0.066
0.38B
0.881
86
0.970
0.081
0.360
0.917

Intrusive-Prepared


ASARCO
194
0 974
0.100
0.120
1 004
172
0.950
0.102
0.262
0.970
RV Hopkins
98
0.988
0.067
0.293
0.923
79
0 983
0.062
0.170
0.990
Notes:
'	Regression parameters based on log10 transformed data.
6	Slope values determined with FPXRF data plotted on x-axis and the reference data plotted an the y-axis
n	Number of useable matched pairs of data points.
Int.	Intercept.
Std. Err.	Standard error.
NO	Analyte not present m significant quantities to provide meaningful regression.
^ ^ ...
55

-------
Section 5
TN 9000
This section provides information on the TN
9000 including background information, operational
characteristics, performance factors, a data quality
assessment, and a comparison of its results with those of
che reference laboratory.
Background
Since J 988, the developer has produced field
portable and laboratory-grade XRF technologies for a
broad range of applications. The TN 9000 was released
in 1992 for environmental applications.
The TN 9000 uses a Hgl2 semiconductor detector
that achieves a manganese x-ray resolution of
less than 300 eV. The detector is operated by a
moderately subambiem temperature controlled by a low
power thermoelectric (Peltier) cooler in the measurement
probe.
The TN 9000 uses energy dispersive XRF
spectrometry to determine elemental composition of
soils, sludges, aqueous solutions, oils, and other waste
materials, ft uses three radioactive isotopes, iron-55
(Fe55), Cdiw, and Americium-241 (Am241), to produce
excitation X rays, which excite a corresponding range of
metals in a sample. The TN 9000 can identify and
quantify target metals from sulfur through uranium on
the periodic chart of the elements. When more than one
source can excite a given metal, the appropriate source
is selected according to its excitation efficiency for the
target metal. Generally, the source with the excitation
energy closest to, but above, the absorption edge energy
for a given metal is selected for analysis. Interferences
sometimes affect this selection as discussed in
Section 4.
To analyze a sample with the TN 9000, the sample
is positioned in front of the plastic film probe
measurement window and sample measurement is
initiated. The sample is then exposed to primary
radiation from the source. Only one of the three
sources is exposed at a time. If all three sources are
required for a sample's analysis, three source exposures
are sequenced automatically. Fluorescent and
backscartered X rays from the sample re-enter through
the window and are counted by the high resolution Hgl2
detector. The surface probe of the Hgl2 detector
provides for both in situ soil analysis and intrusive soil
analysis. For intrusive analysis, the probe is placed
upright in a stand and the sample, contained in a thin-
windowed plastic cup, is placed over the probe
measurement window and beneath a swing-down safety
shield.
Contaminant concentrations are competed using a
FP calibrated algorithm included in the analyzer's
operations software. The developer uses FPs to calibrate
its FPXRF analyzer. The FPs are based on the physics
of X-ray excitation and emission. The menu-driven
software in the TN 9000 supports multiple XRF
calibrations in an "Soil Applications" software package.
Each application contains a complete analysis
configuration including target metals to be measured,
interfering target metals in the sample, and a set of FP
calibration coefficients. The FP calibration does not
require site-specific calibration samples, however, these
samples can be used to fine tune the calibration.
Operational Characteristics
This section discusses equipment and accessories,
operation of the ajoalyzei, description of the operator,
training, reliability of the analyzer, health and safety
concerns, and cost to operate the analyzer.
Equipment and Accessories
The TN 9000 comes with all of the equipment
necessary for in situ and intrusive operation (Table 5-1).
A hard-shell carrying case containing the equipment
protected by foam inserts is provided for transportation
and storage.
BRA

-------
TABLE 5-1. ANALYZER INSTRUMENT SPECIFICATIONS—TN 9000
Characteristic
Specification
Resolution
< 300 eV {Manganese-Kj,)
Sources
5 mCi Cd109. 50 mCi Fe55, 5 mCi Am241
Detector
Hglj-Peltier cooled
Probe Size
12.7 cm x 7.6 cm x 21.6 cm
Probe Weight
1.9 kilograms
Probe Operating Temperature
0 to 49'C
Electronics Unit Size
32 cm x 30 cm x 10 cm
Electronics Unit Weight
6.7 kilograms
Electronics Unit Operating
0 to 49" C
Electronics Unit Storage Capacity
300 sets of numerical results and 120 spectra
Power Source
120V or 220V (AC) or internal batteries
Operational Checks
3 NIST SRMs, Si02 and Teflon® blanks, pure element check
Intrusive Operation
Unibiock probe stand
Computer Interface Operation
RS 232 serial input/output cable
Contact:
1
Raj Natarajan
2555 N. Interstate Hwy. 35
Round Rock, TX 78664
(800) 736-0801
(512) 388-9200 (FAX)

Two main components make up the analytical
system: a probe and an electronics unit. The probe
contains three radioisotope sources: Fe55 (50 mCi),
Cd109 (5 mCi), and Am241 (5 mCi) for sample excitation
and the Hgl2 detector. The sources are encapsulated and
housed in a metal turret with additional lead shielding
inside the probe. These sources can sequentially expose
the sample to excitation radiation through the sealed
1-inch-diameter polypropylene cover for the mylar
window in the face of the probe. The source-induced
fluorescence from the sample passes back through the
window and is intercepted by the Hgl2 detector. The
detector quantitates ihe energy of each characteristic
emission X ray and builds a spectrum of analyte peaks
on a 2,048-channel MCA, which is contained in the
electronics unit. The probe is 12.7 cm x 7.6 cm x 21.6
cm and weighs 1.9 kilograms. The standard probe
operating temperature is 0 to 49°C and the standard
probe storage temperature is -40 to 43 eC.
Spectral data is communicated to the electronics unit
through a flexible cable of 6. 12. or 20 feet in length.
The standard cable length is 6 feet. X-ray emission
peaks are integrated and concentrations in ppm or
percentage values are calculated. The electronics unit
will store and display both numerical results and spectra
from a measurement. A maximum of' 300 sets of
numerical results and 120 spectra can be stored before
being downloaded to a PC using a RS-232 cable. The
electronics unit is 32 cm x 30 cm x 10 cm and weighs
6.7 kilograms. The standard electronics unit operating
temperature is 0 to 49 °C. The standard electronics unit
storage temperature is -20 to 60 "C.
The electronics unit can be operated from a battery
or from an alternating current electric line via a plug-in
adaptor unit. The TN 9000 is supplied with two nickel-
cadmium batteries and a battery charger. The batteries
last approximately 4 to 5 hours and require a minimum
of 14 hours to fully recharge. For this demonstration,
the developer provided two additional batteries and
chargers so (hat analysis could continue for up to 12
hours per day. For in situ analysis, the developer
provided a water-resistant carrying case and a strap for
easy portability on site. The carrying case has a flap on
top which can be closed to protect the analyzer from the
environment.
Other equipment and supplies that are helpful when
using the TN 9000, which are not supplied by the
developer, include a PC to download data, protective
gloves, paper towels, and a permanent marking pen.
57
rvs * -
» 4 ¦

-------
Operation of the Analyzer
For this demonstration, the TN 9000 was operated
on battery power during the in situ phases of the
demonstration. The in situ analysis was performed with
the analyzer in the carrying case. The probe was placed
in contact with the soil surface and analysis was started
by pressing a trigger on the back of the probe. For
intrusive analysis, the probe was placed in the uniblock
pointed upward with the safety shield attached. All
intrusive analyses at both sites were performed by setting
the analyzer on a table top located indoors. At the
ASARCO site, the room was not heated or cooled so
analysis occurred at ambient outdoor temperatures which
ranged from 5 to 16 °C. At the RV Hopkins site, the
room used for all but the in situ-unprepared analysis was
maintained at approximately 25 °C.
Description of the Technology Operator
The PRC operator chosen for analyzing soil samples
using the TN 9000 has a bachelor's degree in zoology,
which included 30 hours of undergraduate chemistry
study, and a master's degree in environmental
engineering. This operator worked as a gas
chromatography chemist in an environmental analytical
laboratory for 3 years and as an assistant chemist at a
chemical company for 3 years prior to his current job.
His job at PRC, for the past year, has involved
performing on-site analyses, conducting site
investigations, performing risk assessments, and
evaluating remedial design systems.
Training
The PRC operator viewed a 22-minute training
video. The training video described the analyzer,
applications of the analyzer, instructions on the analysis
procedures for in situ and intrusive sample
measurements, and procedures for downloading data
from the analyzer to a PC. The operator then received
approximately 6 hours of training at the start of the
demonstration by the developer. The training covered
the theoretical background of XRF and applications of
the TN 9000 analyzer.
The PRC operator estimated that approximately 80
percent of the training was "hands-on." The training
included step-by-step instructions involving the daily
setup and use of the TN 9000. This involved the
connection of the nickel-cadmium battery, attaching the
probe to the electronics unit, setting up the applications
software, modifying the count times for each source,
setting the probe in the stand bracket and attaching the
safety shield for intrusive analysis, downloading results
to a PC, and maintenance such as the replacement of the
mylar probe window. The developer had calibrated the
TN 9000 prior to the training using a FP algorithm based
on NIST soil SRMs. Pan of the training included a
discussion of QC requirements such as the analysis of a
pure iron energy calibration check, an Si02 blank, and
at least one NIST SRM; possible interferences that could
be encountered: and procedures for preparing both in
situ and intrusive soil samples for analysis. At the
conclusion of the training, the developer was confident
that the operator was ready to operate the TN 9000. The
developer accompanied the operator at the ASARCO site
during the first morning and observed him analyzing soil
samples. No problems were encountered, and the
developer left the site.
The developer stated that a person with at least a
high school education could operate the TN 9000 after
being trained. The PRC operator had no prior
experience operating a FPXRF analyzer and found the
TN 9000 easy to operate. The training video and hands-
on training were more than sufficient to enable the
operator to successfully use the analyzer. The analyzer
required little computer or technical background for
operation and the menu-driven software was user
friendly and simple. No computation or data reduction
was required to obtain the results. The operator noted
that changing the battery, replacing the probe window,
downloading data, and printing data were easy. Two
suggestions for improvement of the TN 9000 were
offered by the operator. He felt it would be helpful to
have a pre warning indicator of battery power failure.
Currently, there is no prompt for a battery change before
analyzer shutdown. The operator also felt it would be
advantageous to the operator when analyzing data to
have a feature that would allow interface with the
analyzer by using a standard PC keyboard. The operator
could then take the resulting spectra files and analyze
them using a "pull-down," menu-driven software on the
PC.
Reliability
A reliability check of the TN Lead Analyzer was
carried out by measuring a check sample daily. This
check comprised a 50-second measurement of a pure
iron sample. By this one measurement a verification
was obtained of (1) the fluorescent element sensitivity;
(2) the spectrometer energy resolution; and (3) the
spectrometer energy calibration. To be acceptable, the
measured relative X ray intensity of iron had to be
greater than 0.95 and the equivalent intensity of
manganese and cobalt had to be less fhan 0.006.
Relative intensity refers to the new value relative to that
obtained at the time of the initial instrument calibration.
If the intensity conditions were not met, then the pure
iron sample was reanalyzed. No energy recalibrations
58

. rk	&

-------
were required during the demonstration based on the
pure iron sample results. This showed that the TN 9000
was reliable during ibis demonstration.
During the demonstration, there were frequent light
to mode rate rains while the FPXRF analyzers -were
performing the in situ measurements. After this rain, it
was common for the soii surface to be saturated. The
developer recommends that samples analyzed by the TN
Lead Analyzer have Jess than 20 percem moisture
content by weight. The -samples collected during this
demonstration contained up to 30 percent moisture
content by weight. This increased moisture content did
not reduce the analyzer's data comparability. At the
ASARCO and RV Hopkins sites, the temperatures
ranged from approximately 5 to 16 °C and 6 to 22 °C,
respectively. Despite the less (ban ideal weather
conditions, there were no mechanical or electronic
problems experienced with the TN 9000 during the
course of the demonstration. The only maintenance
required was the replacement of the probe window cover
twice due to contamination and damage from small
pebbles. The replacement of the probe window cover
took approximately 2 to 3 minutes.
Health and Safety
The potential for exposure to radiation from the
excitation sources was the largest health and safety
consideration while using the FPXRF analyzers.
Radiation was monitored with a radiation survey meter.
Background radiation at the two sites was between 0.006
andO.OL2 tnrem/hi. Radiation exposure was monitored
in both the in situ and intrusive modes white the shutters
of the analyzers were open to obtain a wont-case
scenario. The radiation was measured within 5 cm of
the probe face while the analyzer was analyzing a
sample. Radiation exposure also was monitored at a
point on the probe where the operator's hand was located
during analysis io provide a realistic value of operator
exposure. The TN 9000 is sold under a general license,
meaning that the analyzer is designed and constructed in
such a way that anybody operating it, as per the
instruction manual, wilt not be exposed io harmful
radiation levels according to the Nuclear Regulatory
Commission. Many states still recommend that radiation
from survey instruments be below a certain level. For
example, in the state of Kansas, the permissible
occupational exposure is 5,000 mrem/year, which
equates to approximately 2 to 3 mrem/hr assuming
constant exposure {or an croirc work year.
While taking in situ nutasurenvesKs (probe pointing
down), the fo!lowing radiation values were obtained at
the probe face for the TN 9000: Cdm source, 0.10 to
0.12 mrem/hr; Fess source, 0.025 to 0.035 mrem/hr;
and Aur"', 0.50 to 0.60 mrem/hi. Radiation
background levels were recorded at the probe handle
while the Fe55 and Cd109 sources were exposed. while
0.020 to 0.025 mrem/hr were recorded when the Am~41
source was exposed. While collecting inmisive
measurements with tfce TN 9000. die following
radiation values were obtained on top of the protective
sample cover: CdlW source, 0.09 to 0.10 mrem/hr; Fe55
source, 0.098 to 0.012 mrem/hr; and Am241 source,
0,08 to 0.10 mremte. Aft measured radiation values
were less than the permissible 2.0 mrem/hr, The
operator noted thert was tro safety feature on \hs
analyzer that prevented a person from accidentally
exposing someone by pushing the button on the near of
the probe to scan an analysis while the probe was pointed
at the operator or another person.
Cost
At the time of demonstration, the TN 9000 cost
558,000. This includes all of the equipment necessary
for operation of the analyzer. Table 4-2 shows the
incidental items and costs associated with using the TN
9000. The analyzer has a full-year warranty with an
optional extended warranty. The TN 9000 can be rented
through several companies for $6,000 per month or
£3,500 for 2 weeks. Additional field packs can be
purchased lor 5200 and external batteries, charger, and
adapter for $750, A 12-month or 24-month extended
warranty can be purchased for Si,750 or $4,750,
respectively. Periodic maintenance includes replacement
of the Cd source every 2 years at a cost of S3,500 io
$3,800. The Fe*5 source should be replaced every A to
5 years. The cost of replacement of the Cd109 and Fe51
sources together is 56,800. The Am24' source has a
half-life of 433 years and does not need to be replaced.
Because the TN 9000 contains a radioisotope, a wipe (est
must be performed every 6 months at the cosi of $60.
Table 5-2 provides the relative costs for TN 9000
analysis and gives the cost for reference method analysis
for comparison purposes.
The developer offers a training course at its offices
or on site. The cost of the on-site course is $1,000 per
day, plus travel expenses. Operator costs will vary
depending on the technical knowledge of the operator.
As discussed earlier, the TN 9000 is designed to be used
by individuals with no more than a high school education
and a minimal amount of technical training, thereby
decreasing the cost.
Performance Factors
The following paragraphs describe performance
factors, including detection limits, sample throughput,
and drift.
59


-------
TABLE 5-2. RELATIVE ANALYTICAL COSTS—TN 9000

TN 9000
Reference
Methods
3050A/601OA
No. of Samples
Purchase® ($)
Lease ($)b
Cost Per
Sample (S)
30 (assume 1 day for
analysis, 1.5 days for
per diem, and 1 day
total for
mobilization/de-
mobilization time)
58.000
750 (extra battery)
400 (operator labor @ $50/hr)
150 (per diem for operator)
400 (mobilization/demobilization)0
1.250 fsuDDlies from Table 4-2)
3,500/2 weeks
100 (shipping)
400 (operator labor @ $50/hr)
150 (per diem for operator)
400 (mobilization/demobilization)0
1.250 fsuDDlies from Table 4-2^
150/sample
60,950
2,032/sample
5,800
193/sample
80 (assume 2 days for
analysis, additional
labor and per diem
charge only vs. 30-
sample scenario)
60,950 (from 30-sample scenario)
50 (per diem for operator)
400 (ODerator labor © $50/hrl
5,800 (from 30-sample scenario)
50 (per diem for operator)
400 (operator tabor <® S50/hr)
150/sample
61,400
767/sample
6,250
76.12/sample
300 (assume 6 days
for analysis, additional
labor and per diem
charge only vs. 80-
sample scenario)
61,400 (from 80-sample scenario)
400 (per diem for operator)
1.600 (oDerator labor © $50/hrl
6,250 (from 80-sample scenario)
400 (per diem for operator)
1.600 foDerator labor © $50/hrl
150/sample
63.400
211/sample
8,250
27.50/sample
This purchase scenario is based on a unit purchase for a single project. The cost per sample for this scenario
would significantly decrease if the purchased unit were used on multiple projects.
Training costs not included.
Travel costs not included.
Note:
a
b
c
Detection Limits
MDLs were determined using standard SW-846
protocols. MDLs were determined by collecting 10
replicate measurements on site-specific soil samples
having metals concentrations 2 to 5 times the expected
MDLs. These data were obtained from the same
samples used in the precision assessment. Based on
these 10 replicate measurements, a SD on the replicate
analysis was calculated. For the purpose of this
demonstration, these precision-based MDLs presented in
Table 5-3 are defined as 3 times the SD for each analyte.
The precision-based MDLs were obtained using a 100
second count time for the Cd109 source and a 60 second
count time for the Fe55 and Am241 sources using site-
specific soil samples. All the precision-based MDLs
were calculated for soil samples that had been dried and
ground in a sample cup.
Table 5-3 also lists MDLs reported by the
developer. The developer's MDLs were acquired using
a 200-second count time for each source with a Si02
blank free of any potential interferences but spiked with
the target analytes.
Because the developer's MDLs were based on 200-
second count times, whereas the precision-based MDLs
were calculated based on the shorter count times listed
above, the developer's MDLs were corrected for
comparison purposes. According to XRF counting
statistics, the precision-based MDLs will vary by the
square root of the count time. Therefore, the developer
MDLs for elements reported by the Cd109 source were
multiplied by a factor of 1.4(square root 2) and by a
factor of 1.82(square root 3.33) for the elements
reported by the Fe" and Am241 sources. The developer
MDLs listed in Table 5-3 have been corrected by the
factors listed above to account for count time
differences.
Another method of determining MDLs involves the
direct comparison of the analyzer data and the reference
method data. When these sets of data were plotted
against each other, the resultant plots were linear. This
60

-------
TABLE 5-3. METHOD DETECTION LIMITS—
TN 9000
Analyte
Developer
MDL8
(mg/kg)
Precision-
Based MDL
(mg/kg)
Field-Based
MDL
(mg/kg)
Antimony
58
55
68
Arsenic
35
60
100
Banum
16
60
975c
Cadmium
255
ND
247
Chromium
164®
372b
o o
(M in
443a
838b
Copper
62
85
195
Iron
157
ND
ND
Lead
20
45
46
Nickel
89
100
286
Zinc
50
80
165
Notes:
mg/kg Milligram per kilogram.
NO Not determined.
8 Chromium low based on Fe55 source.
b Chromium high based on Cd109 source.
c This MDL may be an artifact of the reference
data at concentrations below 200 mg/kg, or it
may be an artifact of barium concentration
determination by total analysis method.
is discussed in greater detail in the "lntennethod
Assessment" later in this section. As the plotted line
approached zero concentration there was a point at
which the analyzer data appeared to respond either
randomly or with the same reading for decreasing
concentrations of reference data. Figure 5-1 illustrates
this effect for copper. By determining the mean values
of this random or singular analyzer data and
subsequently two standard deviations around this mean,
it was possible to determine field-based MDLs for the
analyzer. These field-based MDLs are shown on Table
5-3.
Although the TN 9000 reported results for 24
analytes, only the target analytes are shown in Table 5-3.
Cadmium was reported only at very low concentrations
and a precision-based MDL could not be determined.
Iron was mostly found at concentrations in the tens of
thousands of mg/kg so that reasonable detection limits
could not be calculated. The precision-based MDLs
were generally higher than the developer's detection
limits, but usually within a factor of two. The field-
based MDLs were generally higher than the precision-
based MDLs. The differences between the developer's
MDLs and the precision- and field-based MDLs is
probably due to increased matrix interferences inherent
in environmental soil samples. Barium was the analyte
that showed the largest disparity between the developer's
MDL and the precision-based or field-based MDLs.
Throughput
The TN 9000 used a total source live-second count
time of 220 seconds or 3.7 minutes. With the additional
"dead" lime of the analyzer and the time required to
label each sample and store data between sample
measurements, the time required to analyze one soil
sample was between 5 to 6 minutes. At the beginning of
the demonstration, the PRC operator was able to analyze
8.5 in situ soil samples per hour. As he gained more
experience and became more efficient at operating the
TN 9000, he was able to analyze 9.5 in situ soil samples
per hour. The in situ sample throughput depended on
the weather conditions and the distance required to walk
from one sample point to another. In the intrusive mode
with the samples already prepared, the throughput was
increased to 9.5 to 10.5 samples per hour. The operator
found he was capable of analyzing an average of 100 soil
samples in a 10-hour day. The maximum number of soil
samples analyzed was 128 in a 12-hour day. This
throughput did not include the analysis of an average of
six QC samples, such as two SiOi blanks, two pure iron
sample calibration checks, and two NIST SRMs. These
QC analyses are recommended by the developer.
Sample analysis time did not include the time required
for sample handling and preparation or for data
downloading, printing, and documentation.
Considerable time was spent preparing the in situ
homogenized samples and the intrusive samples. The
sample homogenization process took approximately 5
minutes per sample, wet sieving took approximately 20
minutes per sample, and grinding and sieving took
approximately 10 minutes per sample. Approximately
0.5 hour was spent daily downloading the data to a PC
and obtaining a hard copy of the data.
Drift
For the TN 9000, drift was evaluated by reviewing
results from the daily analysis of NIST SRM 2710. This
SRM contained quantifiable levels of arsenic, barium,
copper, iron, lead, and zinc. NIST SRM 2710 data was
collected over 18 days, of which approximately 67
percent was collected at the ASARCO site and 33
percent at the RV Hopkins site. This data was reduced
to RSDs for the target analytes, and the percent drift
from the mean recovery of the true value. The percent
drift from the mean recovery for each of the 18 days is
shown on Figure 5-2. The RSD values for barium,
copper, iron, lead, and zinc were all less than 8 percent.
61

-------
3 100000
e
~ 10000
&
«3 1000
N
ra
< 100
10 100 1000 10000 100000
Reference Method Data (mg/Xg)
FIGURE 5-1. CRITICAL ZONE FOR THE DETER-
MINATION OF A FIELD-BASED METHOD DETECTION
LIMIT FOR COPPER: Between 100 and 200 mg/kg for
the reference data the linear relationship between the two
data sets changes. This point of change identified the
point at which field-based MDLs for the analyzer were
determined.
The RSD for arsenic was much higher at 18.2 percent.
This higher RSD for arsenic is probably an artifact of
interference from the much greater concentration of lead
in the sample. The developer has noted that in past
analyses of NIST SRM 2710, the precision of the arsenic
analysis in the presence of 5,500 ppm lead was 18
percent relative in a 100-second measurement. The low
RSD values indicate that for the concentrations of
analytes found in NIST SRM 2710, the TN 9000
exhibited little drift during the demonstration. This
minimal drift that did occur was less than the 10 percent
limit listed in the demonstration plan (PRC 1995) QAPP
for Level 3 data for ail analytes except arsenic.
Intramethod Assessment
Intramethod measures of the analyzer's performance
included its results on analyzer blanks, the completeness
of its results, its intramethod precision, its intramethod
accuracy, and its intramethod comparability. The
following sections discuss these four items.
Blanks
Analyzer blanks for the TN 9000 consisted of Si02
blanks. These blanks were routinely analyzed at the
beginning and end of the day. They were used to
monitor contamination of the probe due to such material
as residual soil left on the face of the probe. A total of
37 Si(>2 blanks were analyzed during the demonstration.
None of the primary analytes were detected in the 37
blanks. Iron was frequently detected at concentrations
ranging from 150 to 250 mg/kg. This small amount of
iron is actually present in the S1O2 matrix. These
concentrations of iron would not have significantly
affected the results of the soil samples because iron
concentrations in the soil samples were mostly greater
than 20.000 mg/kg.
Completeness
A total of 315 soil samples were analyzed four times
(four sample preparation steps) resulting in 1,260 sample
results. The TN 9000 produced results for 1,259 of the
1,260 samples for a completeness of 99.9 percent, above
the demonstration objective of 95 percent. The one
missing sample result was actually operator error and
was not due to analyzer malfunction. The PRC operator
failed to analyze one in situ sample at the ASARCO site.
Precision
Precision refers to the degree of repeatability or
agreement among individual measurements of the same
sample and provides an estimate of analyzer-induced or
random error. Precision for this demonstration was
expressed in terms of the percent RSD between replicate
measurements. The percent RSD is defined as the SD
divided by the mean concentration times 100. The
precision data for the target analytes is shown in Table
5-4. The TN 9000 performed 10 replicate measurements
on 12 soil samples that had analyte concentrations
ranging from less than 50 mg/kg to tens of thousands of
mg/kg. Each of the 12 soil samples underwent the four
different sample preparation methods providing 480
precision data points for each analyte. Since the
replicate analyses were taken without moving the probe
or sample, the resulting measurements reflect analyzer
precision and not method precision, which would include
sample preparation. The replicate measurements were
obtained using the source count times discussed in the
previous section of this report. For each detectable
analyte in each precision sample, a mean concentration,
SD, and RSD was calculated.
In this demonstration, the RSD for a given analyte
had to be less than or equal to 20 percent to be
considered Level 2 data and less than or equal 10 10 per-
cent to be considered Level 3 data. The precision of the
analyzer was defined by measurements in the 5 to 10
times the expected MDL range. The analyzer's
precision was below the 10 percent RSD required for
Level 3 data classification for all target analytes except
chromium (Table 5-4). Nickel, cadmium, and iron did
not have sufficient data to allow data quality conclusions
based on precision. Table 5-4 shows that chromium
precision in this concentration range was greater than 20
percent placing the chromium data in the Level 1
category. The decreased precision for chromium shown
in Table 5-4 was not unexpected as chromium is a
problematic analyte for FPXRF analysis. The average
62

-------
c
01
o
O)
a
40
:= 20
-2 0
-4 0
.a
a
JX
Lh
-B-
Arsenic
*
*
B a riu m
Copper
q Analyte
Lead
Zinc
FIGURE 5-2. DRIFT SUMMARY—TN 9000: This graph illustrates drift over a period of 18 days. Each bar represents
a single measurement on a single day. The same sample was used throughout the demonstration.
RSD values for nickel and cadmium shown in Table 5-4
are biased high because of the low inherent nickel and
cadmium concentrations in the precision samples.
There was no significant sample preparation effect
on precision. This was expected because the method
used to assess precision during this demonstration was
primarily measuring analyzer precision, not total method
precision. There was a concentration effect on the
precision data. The precision samples were purposely
chosen to span a large concentration range to test the
effect of analyte concentration on precision. As the
concentration of the target analyte increased, the
precision increased (Figure 5-3). The largest increase in
precision occurred at concentrations two to three limes
the detection limit for that analyte. The precision
continued to increase until 1,000 to 2,000 mg/kg, then
stabilized above analyte concentrations of 2,000 mg/kg.
Table 5-4 shows that the RSD values were less than 10
percent for all analytes except chromium at
concentrations greater than 500 mg/kg.
Accuracy
Accuracy refers to the degree to which a measured
value for a sample agrees with a reference or true value
for the same sample. Intramethod accuracy was
assessed for the TN 9000 by using site-specific PE
samples and SRMs. Accuracy was evaluated through a
comparison of percent recoveries for each target analyte.
The TN 9000 analyzed six site-specific PE samples and
14 SRMs. The PRC operator knew the samples were
PE samples or SRMs but did not know the true
concentration or the acceptance range. These site-
specific PE samples and SRMs were analyzed in the
same way as all other samples.
The six site-specific PE samples consisted of three
from each of the two demonstration sites. These PE
samples were collected during the predemonstration
activities and sent to six commercial laboratories for
analysis by laboratory-grade XRF analyzers. The mean
measurement for each analyte was used as the true value
concentration. The 14 SRMs included seven soil, four
stream or river sediment, two ash, and one sludge SRM.
The SRMs were obtained from NIST, USGS,
Commission of European Communities-Community
Bureau of Reference, National Research Council-
Canada, and the South African Bureau of Standards.
The SRMs contained known certified concentrations of
certain target analytes.
These site-specific PEs and SRMs did not have
published acceptance ranges. As specified in the
demonstration plan (PRC 1995), an acceptance range of
80 to 120 percent recovery of the true value was used to
evaluate accuracy for the six site-specific PEs and 14
SRMs. Table 5-5 summarizes the site-specific PE and
SRM accuracy data for the target analytes for the TN
9000. Figures 5-4 and 5-5 show the true values, the
measured value, and percent recoveries for the
individual site-specific PEs and SRMs, respectively. No
figures were presented for analytes that had less than
three samples with detectable concentrations. True value
results from the site-specific PEs and SRMs with
concentrations less than the precision-based MDLs listed
in Table 5-3 were excluded from the accuracy
assessment.
The TN 9000 was 100 percent accurate for all
analytes in the site-specific PE samples with the
exception of chromium, nickel, and zinc. Overall, the
TN 9000 produced 37 out of 41 results within the 80 to
63
*	1 ' -^g

-------
TABLE 5-4. PRECISION SUMMARY—TN 9000

Mean RSO Values by
Concentration Range
Analyte
5-10
times
MDLa
50 - 500
Img/kg)
500 -
1,000
(mg/kg)
>1,000
(mg/kg)
Antimony
6.54 (8)
12.52 (16)
4.54 (4)
ND
Arsenic
5.33 (12)
9.68 (8)
4.39 (12)
2.87 (8)
Banum
4.02 (20)
NO
3.70 (40)
2.67 (8)
Cadmium
NO
29.84b (48)
ND
ND
Chromiumc
22.25 (12)
38.95 (12)
29.10 (4)
ND
Copper
7.03 (B)
19.02(24)
6.21 (4)
3.35 (12)
Iron
ND
ND
ND
1.78 (48)
Lead
6.45 (12)
9.69 (12)
5.34 (2)
3.68 (20)
Nickel
ND
30.85b (16)
ND
ND
Zinc
7.27 (16)
13.59 (24)
7.27 (16)
ND
Notes:
mg/kg Milligram per kilogram.
ND No data.
( ) Number of samples, including all (our preparation
steps, each sample represents 10 replicate
measurements. Numbers do not always add up
to 48 precision points because some samples
had analyte concentrations below the analyzer's
MDL
a Precision Based MDL.
s These values may be biased high because the
concentration of these analytes in the soil
samples was near the detection limit.
c Values calculated from chromium low results from
Fess source.
120 percent recovery acceptance range for all analytes in
the six site-specific PE samples. This translates into a
90.2 percent accuracy for all analytes. Two of the four
results that fell outside the acceptance Tange were only
slightly outside wiih a nickel recovery of 125 percent in
one sample and a zinc recovery of 79.1 percent in one
sample. The other two unacceptable results fell far
outside the acceptance ranges with a nondetect or 0
percent recovery for chromium in one sample and a zinc
recovery of 58.3 percent in one sample. The 58.3
percent recovery for zinc was for a PE sample that
contained 164 mg/kg zinc which is less than the field-
based MDL and less than 2 times the precision-based
MDL. With the exception of chromium, the TN 9000
produced mean percent recoveries near 100 percent for
all analytes (Table 5-5). These results were for analytes
that often showed concentrations spanning more than one
order of magnitude in the site-specific PE samples. A
detailed analysis of the SRM data is presented on Figure
5-5. The TN 9000 accuracy for the SRMs varied from
0 percent for chromium (only one SRM concentration
for chromium above the TN 9000's MDL) to 100
percent for antimony and iron in all SRMs. The iron
concentrations in the SRMj were in the tens of thousands
of mg/kg so ii was not surprising the TN 9000
performed well for iron. Some analytes such as barium,
copper, lead, and zinc had concentrations spanning one
or more orders of magnitude in the SRMs. Overall, the
TN 9000 produced 38 out of 58 results within the 80 to
120 percent recovery range for an accuracy of 65.5
percent. Of the 20 results that fell outside of the
acceptance range, four results were low, and 16 were
high. This ratio of high results to low, in addition to the
mean percent recoveries shown in Table 5-5, indicated
that, in general, the TN 9000 overestimated analyte
concentrations in the SRMs, especially for barium. The
lowest recovery produced by the TN 9000 was 67
percent for copper in the Canadian sediment SRM. The
highest recovery was 198 percent for barium in one of
the USGS soil SRMs. "Hie TN 9000 results for all
analytes were less than 2 times (200 percent recovery)
the reported SRM true value for all SRMs.
A more detailed analysis of the SRM data showed
that there was a matrix effect on the TN 9000 accuracy.
The TN 9000 produced 22 out of 24 or 91.7 percent of
the results within the acceptance range for ail target
analytes in the soil SRMs; 10 out of 19 or 52.6 percent
for the sediment SRMs; and 6 out 15 or 40 percent for
the ash and sludge SRMs. The greater accuracy for the
soil SRMs is expected since ii was using an FP
calibration based on the NIST soil SRMs. Only barium
recovery in the two USGS soil SRMs was outside
(above) the acceptance range. This demonstrates that the
TN 9000 is more accurate when analyzing SRMs that
closely match the matrix used to set the fundamental
parameters of the analyzer. The TN 9000 performed the
poorest on the one sludge SRM by overestimating all
analyte concentrations by a factor of 1.5 to 1.7. With
the sludge SRM removed from the data, the TN 9000
had percent recoveries less than 140 percent for all
analytes in all SRMs except for barium in one USGS
SRM.
The TN 9000 displayed almost identical accuracy
for the soil SRMs and the site-specific PEs (90.2
percent). This indicates that the matrix of the soil SRMs
matched the matrix of the site-specific samples well
enough such that the FP calibration based on the soil
SRMs produced results that were over 90 percent
accurate for site-specific samples. It also indicates that
SRMs of a sediment, ash, or sludge matrix are not as
suitable of accuracy checks when the FP calibration is
based on a soil matrix.

-------
40

b

«-
30 .
f-
O
w
*¦
K 20

T3
¦t-
	+
» -fr-ttytj-•«%. ^ +¦ , ++++. * . -f-
0	2	4	6
T housands
Lead Concentration (mg/kg)
50
40
30
20
1 0
4	6
Thousands
Copper Concentration (mg/kg)
FIGURE 5-3. PRECISION VS. CONCENTRATION FOR LEAD AND COPPER—TN 9000: These graphs illustrate the
TN 9000's precision as a function of analyte concentration.
Comparability
Intramethod comparability for the TN 9000 was
assessed through the analysis of four ERA PEs and four
CRM PEs. These eight samples were analyzed by the
TN 9000 in the same manner as all other samples. As
described in Section 3, these eight samples had certified
analyte concentrations determined by Methods
3050A/6010A. The ERA PEs had published PALs
based on a 95 percent confidence interval around each
certified concentration. The CRMs had a 95 percent
prediction interval (PI) associated with each certified
value. The ability of the TN 9000 to produce results
within the PALs or Pis and the percent recovery for
each of the analytes was used to evaluate the TN 9000
analyzer's intramethod comparability. True value
analyte concentrations in the ERA and CRM PEs that
were below the precision-based MDLs in Table 5-3 were
excluded from the intramethod comparability
assessment.
The TN 9000 performance data for all primary and
secondary target analytes for the eight samples are
summarized in Table 5-6. The measured values, true
values, and percent recoveries for all detectable analytes
are shown on Figure 5-6. No figure is shown for
chromium and nickel because there were only one and
two reported certified concentrations, respectively, for
these two analytes. For the ERA PEs, the TN 9000
produced 15 out of 29 results or 51.7 percent within the
acceptance range. For the CRMs, the TN 9000
produced 17 out of 23 results or 73.9 percent within the

-------
TABLE 5-5. ACCURACY SUMMARY FOR SITE-SPECIFIC PE AND SRM RESULTS—TN 9000
Analyte
n
Percent Within
Acceptance
Range
Mean
Percent
Recovery
Range of
Percent
Recovery
SD of
Percent
Recovery
Concentration
Range (mg/kg)
Site-Specific Performance Evaluation Samples
Antimony
4
100
96
85-105
8.82
51 - 2.253
Arsenic
3
100
94
87-101
7.07
424- 19.584
Barium
6
100
100
94-110
5.95
792 - 7.240
Cadmium
1
100
110
110
NA
353
Chromium
2
50
41
0-82
NA
939 - 3.800
Copper
5
100
96
85-120
14.5
300-7.132
Iron
6
100
97
87 -105
6.43
27,320 - 70.500
Lead
6
100
98
91 -103
4.34
292 - 14,663
Nickel
2
50
115
105 -125
NA
312-444
Zinc
6
67
85
58 -103
15.4
164 - 3,490
Soil Standard Reference Materials
Arsenic
3
100
101
89-115
13
105 - 626
Barium
5
60
130
98 -198
40
707 - 2,240
Copper
2
100
88
80-96
NA
131-2,950
Iron
3
100
98
95-102
3.7
28,900 - 35,000
Lead
5
100
100
80-115
13
101-5,532
Nickel
1
100
99
99
NA
299
Zinc
5
100
98
93-112
8.1
106-6.952
Sediment Standard Reference Materials
Antimony
1
100
100
100
NA
171
Arsenic
1
0
68
68
NA
. 211
Barium
3
33
125
107-139
16
335 - 414
Chromium
1
0
178
178
NA
509
Copper
4
50
100
67-139
31
99 - 452
Iron
1
100
99
99
NA
41,100
Lead
4
75
104
82-138
25
161 -5,200
Zinc
4
50
94
74-122
18
264 - 2,200
Ash and Sludge Standard Reference Materials
Arsenic
2
50
107
85-127
NA
136-145
Barium
2
50
123
117-130
NA
709-1,500
Copper
3
0
143
124-174
27
113-696
Iron
2
100
88
86-89
NA
77,800 - 94,000
Lead
2
50
122
91 -153
NA
72 - 286
Nickel
1
0
123
123
NA
247
Zinc
3
33
115
77-166
46
210-2,122
Notes:
n	Number of samples with detectable analyte concentrations.
SD	Standard deviation,
mg/kg	Milligrams per kilogram.
NA	Not applicable. Standard deviation not calculated for two or less results.
66


-------
10000
i 1000
120
100 «
>
o
o
0)
so a.
c
0)
u
60
40
s.
Antimony
|M eaeured Value ~ TnieValue	O Pereenl Recovery
10000
= 1000
100
Barium
|M easured Value ~ True Value
120
»>
100 w
>
o
u
O
80 K
— 60
£
40
~ Percent Recovery
100000
at
c
.2
c
<3
10000
100
m
120
100 £
>
o
o
80 g
60
40
£
Arsenic
I Measured Value OTrueValue
~ Percent Recovery
10000
w 1000
«
u
c
100
125
Copper
|Meaiured Value ~True Value	~Pereenl Recovery
o>
E_
c
.9
re
™ a
90
70
2 50
30
10
I-
r
pi
-i—H
•
1
11
f

120
>»
100 ®
>
o
o
«
80 a
60 w

o
u
o
80 ^
c
60 ®
40
~ P ercent Recovery
10000
— 1000
100
Zinc
|M enured Value ~True Value
120
100 «>
>
o
e>
(C
80
60
40
e
0
1
~Percent Recovery
FIGURE 5-4. StTE-SPECIFIC PE SAMPLE RESULTS—TN 9000: These graphs illustrate the relationship between the
TN 9000's data (measured values) and the true values for site-specific PE samples. The gray bars represent the percent
recovery for the TN 9000. Each set of three bars (black, white, and gray) represent a single site-specific PE sample.
67
. —-

-------
FIGURE 5-5. SRM RESULTS—TN 9000: These graphs illustrate the relationship between the TN 9000's data
(measured values) and the true values for the SRMs. The gray bars represent the percent recovery for the TN 9000.
Each set of three bars (black, white, and gray) represent a single SRM sample.
B00
550
c
.2
15
300
50
1

160
120 u
g
&
"c
0)
o
£
40
Arsenic
I Measured Value ~ True Value
a Percent Recovery
10000
.8 1000
15
8
100
zoo
150 5
o
. 100 g
I£
50
Barium
I Measured Value ~ True Value
~ Percent Recovery
100

80
0)
i I 60
5 g
I ^
o
c
40
20

D
I
120
100
0)
>
o
o
80 £
c
V
u
60 £
40
tron
|Measired Value a True Value
~Percent Recovery
.Q
To
c
.
120 ai
g
|
80 §
&
40
O Percent Recovery
c
.9
10000
1000
100
10
Zinc
(Measured Value oTrue Value
125 o
100 c
q Percent Recovery
acceptance range. With the ERA and CRMs combined,
the TN 9000 produced 32 out of 52 results or 61.5
percent within the acceptance range. Based on the data
presented in Table 5-6, the TN 9000 analyzers's results
were more comparable to the CRMs than the ERA Pes.
Also, the mean percent recovery was nearer 100 percent
for all analytes in the CRMs versus the ERA PEs except
for arsenic. The better comparability to the CRMs
versus the ERA PEs may have been an artifact of the
low analyte concentrations in the ERA PEs. With the
exception of iron, the analyte concentrations in the ERA
PEs were all less than 350 mg/kg which is less than 5
times the precision-based MDL for most of the analytes.
The TN 9000 greatly overestimated antimony
concentrations in the ERA PEs and barium
concentrations in the ERA PEs and CRMs. These
results were expected because FPXRF techniques (or
68

-------
Notes:
n	Number of samples with detectable analyte concentrations.
SD	Standard deviation,
mg/kg	milligrams per kilogram.
NA	Not applicable. Standard deviation not calculated for two or less results.
TABLE 5-6. PE AND CRM RESULTS—TN 9000
Analyte
n
Percent Within
Acceptance
Range
Mean
Percent
Recovery
Range of
Percent
Recovery
SD of
Percent
Recovery
Concentration
Range (mg/kg)
ERA Performance Evaluation Samples
Antimony
3
100
311
270 - 344
38
56-99
Arsenic
4
100
101
72-120
21
65 - 349
Banum
4
0
762
446-1.064
272
111 - 319
Cadmium
2
0
172
156-188
NA
90-131
Copper
4
75
131
113-174
28
88-196
Iron
4
0
195
168-240
34
7.130- 10.400
Lead
4
75
112
72-146
32
52 - 208
Nickel
1
0
169
169
NA
135
Zinc
3
67
114
107-121
9.7
101-259
Certified Reference Materials
Antimony
1
100
149
149
NA
4,955
Arsenic
1
100
108
108
NA
397
Barium
2
0
270
193 - 347
NA
342 - 566
Cadmium
2
100
115
101 -129
NA
362-432
Chromium
1
100
99
99
NA
161,500
Copper
4
100
92
61 -142
35
279 - 4.792
Iron
3
67
110
78-154
40
6,481 -191,650
Lead
4
75
103
66-139
30
120- 144,740
Nickel
1
100
108
108
NA
13,279
Zinc
4
50
92
41 -130
38
546 - 22.217
total metals analytical methods) often produce antimony
and barium results much higher than those obtained from
Methods 3050A/6010A (Kane 1993). The TN 9000 also
produced results for iron and nickel in the ERA PEs that
were much higher than the certified values. Again,
these are two analytes that the acid leaching technique of
Method 3050A will not achieve 100 percent recovery.
Therefore, it was not surprising that the TN 9000's
results were higher for iron and nickel. For all
analytes in the ERA PEs, only two out of 29 percent
recoveries were less than 100 percent. This indicated
that the TN 9000 generally gave higher results for PEs
that had values certified by Methods 3050A/6010A,
especially when the analyte concentrations were less than
S limes the precision-based MDL.
Intermethod Assessment
The comparison of the TN 9000 results to the results
of the reference methods was performed using the
statistical methods detailed in Section 2. The purpose of
this statistical evaluation was to determine the
comparability between data produced by the analyzer and
that produced by the reference laboratory. If the
FPXRF data were statistically equivalent to the reference
data and had acceptable precision (10 percent RSD), they
met the Level 3 data quality criteria. If they did not
meet the Level 3 criteria, but it could be mathematically
corrected to be equivalent to the reference data, they met
the Level 2 criteria. If the analyzer did not meet the
Level 3 criteria, and the statistical evaluation could not
69
,

-------
10000
o>
•£
E. 1000
c
.2
c
o
a
100
10
ilt
Q.
400
0)
300 o
U
ft;
cr
c
v
V
£
100
Antimony
|M eatured Value ~ True Value	~Percent Recovery
10000
1100
— 1000
9)
>
o
U
©
c
c
6)
u
a>
&
Barium
I Measured Value qT rue Value
QPercent Recovery
10000
200
w 1000
Copper
|M eatured Value qTrue Value
qP ercent Recovery
1000000
100000
£
I 10000
rj
c	1000
(J
I 100
10
M
ISO
125 o
>
o
u
®
100 <*¦
- 75
50
Lead
(Measured Value pTrue Value
~Percent Recovery

450
350
•£ 250
CD
C
O)
£ 150
3
50
120
100
B0
60
40
O
o
0>
ct
c
a
u
Arsenic
IMeaiured Value ~ True Value
~ Percent Recovery
650
200
150
UJ
tr
l—i	100
Cadmium
|M eaiured Value ~ True Value
QPercenl Recovery
1000000
— 100000
250
o
o
10000
1000
|M eaiured Value

fron
~ Percent Recovery
100000
oi
— 10000
C
.2
a
1000
100
£
l
i
ISO

125
>.

©

>

o
100
o

EC
75
c
o

o

o
50
a.
25

Zinc
|M eaaurad Value QTrueVelua
qPercent Recovery
FIGURE 5-6. PE and CRM SAMPLE RESULTS—TN 9000: These graphs illustrate the relationship between the TN
9000's data (measured values) and the tiue values for the PE and CRM samples. The gray bars represent the percent
recovery for the TN 9000. Each set of three bare (black, white, and gray) represent a single PE or CRM sample.
70

-------
identify a predictable bias in the data, but the analyzer
identified the presence or absence of contamination with
at least a 95 percent accuracy rate, the data was
classified as Level 1.
The TN 9000 analyzer was configured to report
concentrations for all of the target analytes. The
developer recommends that reported concentrations less
than three times their associated SDs should not be
considered valid data. This analyzer reported two values
for chromium. The chromium high values were based
on sample analysis by the Cd'09 source and the
chromium low values were based on sample analysis by
the Fe55 source.
The regression parameters for the six primary
analytes are shown in Table 5-7. The regression
analysis of the entire data set showed that arsenic,
copper, lead, and zinc had ? s at or above 0.92. In the
cases of arsenic, lead, copper, and zinc, the slopes and
y-intercepts were not significantly different from 1.0 and
0.0, respectively. Barium and chromium had i^s
ranging from 0.79 to 0.67. Based on the slope of the
regressions, the analyzer tended to overestimate barium
and chromium concentrations by a factor of two relative
to the reference method. The slope values in Table 5-7
were determined by plotting the FPXRF data on the x-
axis and the reference data on the y-axis.
The next step in the data evaluation involved the
assessment of the potential impact of the variables: site,
soil type, and sample preparation step on the regression
analysis (Table 5-7). Based on this evaluation, there was
no apparent impact of either the site or soil variables on
the regression. The sample preparation variable
exhibited the greatest impact on the regression analysis,
producing up to 1.3 fold (up to 0.19 r2 units) increases
in the r2 (Table 5-7). Generally, the largest shift in the
r2 was exhibited between the in situ-unprepared and in
situ-prepared analyses (Figures 5-7 and 5-8). Sample
homogenization accounted for between 80 to 100 percent
of the total increase in the r2 experienced across all
sample preparation steps. This makes sense due to the
fact that the homogenization step assured that the
analyzer and the reference method were analyzing
essentially the same sample. Arsenic and copper
analyzer data met Level 3 data quality criteria prior to
inital sample homogenization. For lead the initial
sample homogenization (in situ-prepared) improved the
comparability between the two data sets to the point that
the analyzer met the Level 3 criteria. The initial sample
preparation step (homogenization) improved the
regression-based data for the zinc analysis; however, the
t-test indicated the two data sets were different so the
analyzer produced Level 2 data quality for zinc through
all sample preparation steps. The remaining primary
target analytes, barium and chromium, never exceeded
data quality Levels 2 or 1, respectively. The chromium
data was considered Level 1 because the precision was
greater than 20 percent.
The impact of the site and soil type variables was
then assessed for each of the four sample preparation
steps (Tables 5-8 and 5-9). This evaluation was
conducted for lead and zinc only. These were the only
primary analytes exhibiting relatively even concentration
distribution among the site and soil variables. Copper
and barium appear to exhibit site and soil effects.
However, a closer examination of the data shows that the
reported concentrations were either approaching
instrument MDLs or they reported a very narrow range
of concentrations. This held for the site and soil
variables. No clear relationship was observed for these
variables and the comparability of the technology's data
with the reference method data. A minor trend was
noticed for zinc. The loam soil always produced the
poorest correlations; however, these correlations still
met the Level 2 criteria.
Within the four sample preparation steps, the effect
of contaminant concentration was also examined. The
data sets for the primary target analytes were sorted into
the following concentrations ranges: 0 to 100 mg/kg,
100 to 1,000 mg/kg, and greater than 1,000 m/kg. The
regression analysis for each target analyte and for each
sample preparation step was rerun on these
concentration-sorted data sets. A review of these results
showed no consistent improvement in either the r2 or the
standard error for any of the concentration-based data
sets. This indicates that there is no concentration effect
and that the regression analyses associated with the
entire data set are most representative of the relationship
between the analyzer data and the reference data. After
examining the analyzer and reference data plots, a slight
shift in the slope of the plot was noticed at approximately
2,000 mg/kg (Pigure 5-7). When the data was assessed
in the 0 to 2,000 mg/kg and greater than 2,000 mg/kg
concentration ranges, a definite concentration effect was
noticed. The regression parameters were generally
better for the data in the 0 to 2,000 mg/kg concentration
ranges. Lead exhibited the greatest effect, this
comparison consistently produced lower i^s in the
greater than 2,000 mg/kg range. Identification of the
exact cause of the concentration effect is beyond the
scope of this project. This effect does not appear to
strongly effect data quality, and it is less pronounced for
the TN 9000 relative to the TN Lead Analyzer. Possible
causes include changes in reference method accuracy at
higher concentrations due to analyte interferences, and
shifts in FPXRF performance at higher concentrations
due to detector characteristics, or inherent characteristics
of the FP calibration.

-------
TABLE 5-7. REGRESSION PARAMETERS8 BY VARIABLE—TN 9000
Variable
Arsenic
Barium
Copper
n
r2
Std. Err.
Y-lnt.
Slope6
n
r2
Std. Err.
Y-lnt.
Slope"
n
r2
Std. Err.
Y-lnt.
Slope"
All Data
816
0.949
0.15
0.16
0.96
1223
0.787
0.11
1 87
0.50
959
0.951
0 16
0.32
0.94
Asarco Site
604
0.964
0.13
0.11
0.98
627
0.412
0.09
2.26
0.31
824
0.966
0.13
0.10
1.00
RV Hopkins Site
3
ND
ND
ND
ND
393
0.869
0.12
1.67
0.58
135
0 488
0 16
1.16
0.59
Sand Soil
359
0.970
0.13
O.OB
0.97
385
0.078
0.06
2.62
0.12
378
0.951
0.13
0.05
1.00
Loam Soil
445
0.962
0.12
0.14
0.97
444
0.611
0.11
1.78
0.53
444
0.963
0.12
0.28
0.96
Clay Soil
3
ND
ND
ND
ND
393
0.869
0.12
1.67
0.58
135
0.488
0.16
1.16
0.59
In Situ-Unprepared
204
0.909
0.20
0.34
0.B8
305
0.675
0.14
1.86
0.50
250
0.888
0.24
0.54
0.87
In Situ-Prepared
205
0.9S1
0.09
o.os
0.96
306
0.770
0.11
1.91
0.48
241
0.966
0.13
0.28
0.93
intrusive-Unprepared
202
0.9S3
0.09
0.02
1.01
306
0.865
0.08
1.91
0.49
228
0.981
0.10
0.13
0.99
Intrusive-Prepared
201
0.965
0.13
0.14
0.99
306
0.853
0.09
1.86
0.52
239
0.976
0.11
0.25
0.98
TABLE 5-7 (Continued). REGRESSfON PARAMETERS3 BY VARIABLE—TN 9000

Lead
Zinc

Chromium
[Low)

Variable
n
r2
Std. Err.
Y-lnt.
Slope"
n
r2
Std. Err.
Y-lnt.
Slope"
n
i2
Std. Err.
Y-lnt.
Slope"
All Data
1177
0.956
014
0.19
0.96
1062
0.926
0.13
0.24
0.96
277
0.782
0.15
1.85
0.41
Asarco Site
792
0.951
0.14
0.16
0.97
734
0.918
0.13
0.27
0.94
S3
0.003
0.08
2.43
0.03
RV Hopkins Site
387
0.953
0.13
0.29
0.93
341
0.934
0.13
0.18
1.00
184
0.673
0.17
1.60
0.49
Sand Soil
351
0.9S7
0.13
0.13
0.96
323
0.945
0.13
0.21
0.94
40
0.047
0.07
2.67
-0.13
Loam Soil
440
0.953
0.13
0.20
0.97
413
0.888
0.12
0.34
0.93
53
Q.032
0.09
2.30
0.12
Clay Soil
387
0.953
0.13
0.29
0.93
341
0.934
0.13
0.18
1.00
184
0.673
0.17
1.60
0.49
In Situ-Unprepared
296
0.871
0.23
0.38
0.90
281
0.824
0.20
0.49
0.87
100
0.B65
0.12
8 43
1.31
In Situ-Prepared
294
0.979
0.09
0.14
0.96
266
0.959
0.09
0.24
0.94
77
0.513
0.27
1.91
0.36
Intrusive-Unprepared
297
0.983
0.09
0.11
0.99
265
0.957
0.10
0.07
1.02
47
0.770
0.14
1.79
0.43
Intrusive-Prepared
294
0.979
0.10
0.14
1.00
265
0.955
0.11
0.11
1.03
49
0.748
0.17
1.50
0.56
TABLE 5-7 (Continued). REGRESSION
PARAMETERS" BY VARIABLE—TN 9000
Variable
Chromium (High)
n
r2
Std. Err.
Y-lnt.
Slope"
Alt Data
160
0.674
0.17
1.93
0 43
Asarco Site
16
ND
ND
ND
ND
RV Hopkins Site
143
0.692
0.17
1.53
0.57
Sand Soil
10
ND
ND
ND
ND
Loam Soil
8
ND
ND
ND
ND
Clay Soil
143
0.692
0.17
1.53
0.57
In Situ-Unprepared
39
0.617
0.19
1.96
0.43
In Situ-Prepared
35
0.735
0.15
1.93
0.42
Intrusive-Unprepared
42
0.631
0.18
1.98
0.41
Intrusive-Prepared
45
0.768
0.17
1.77
€.51
Notes:
a	Regression parameters based on log10 transformed data.
"	Slope values determined by plotting FPXRF data on the x-axis
and the reference data on the y-axis.
Int. Y-lntereept.
Std. Err. Standard error,
n	Number of data points.
NA Not applicable, analytes not present in significant quantities
to provide meaningful regression.
ir^
72

-------
100000
h srtu-unprepared-Arsenic
10000
100 1000 10000 100000
Reference Data (mg/kg)
tntrusn/e-unprepared-Arsenic
100000

re
o
o
o
o
O)
z
10000 -
1000
10	100 1000 10000 100000
Reference Data (mg/kg)
100000
h situ-unprepared—Lead
o>
t*
I
«
<5
O
o
o
o
O)
10000
1000
100 1000 10000 100000
Reference Data (mg/kg)

Intrusive-unprepared—Lead
100000

o>


10000


re


§ 1000


o


o


o> 100


z
K


10
I I t

10 100 1000 10000 100000

Reference Data (mg/kg)
situ-prepared—Arsenic
100000
o
o
o
o
cn
Z
10000
100 1000 10000 100000
Reference Data (mg/kg)

Intrusive-prepared-Arsenic
100000

O)


10000
•"••I"
re
+

Q 1000


o


o
o


o 100


Z


H


10
i i i

10 100 1000 10000 100000

Reference Data (mg/kg)

In situ-prepared--Lead
100000

S*


gi 10000


re


g 1000


o


o


§ 100


z


t-


10
i i i

10 100 1000 10000 100000

Reference Data (mg/kg)


fritrusive-prepared--Lead
100000

"S


£> 10000


a


3 1000


o
j&f

§ ioo
-

z


t-
¦r ¦'

10
r i l

10 100 1000 10000 100000

Reference Data (mg/kg)
FIGURE 5-7. SAMPLE PREPARATION EFFECT ON ARSENIC AND LEAD RESULTS: These graphs illustrate the
effect of sample preparation on the comparability between the TN 9000 data and the reference data.

-------
a
Arsenic	CrLo	Copper	Zinc	Iron	Antimony
Barium	CrHI	Lead	Nickel	Cadmium
A naiyle
C3 Data Quality Level	g Sam pie preparation Slep
FIGURE 5-8. HIGHEST DEGREE OF DATA QUALITY AT THE LOWEST DEGREE OF SAMPLE PREPARATION—
TN 9000: This graph illustrates the highest data quality level that the TN 9000 met for each analyte. The small black
squares show the lowest degree of sample preparation that produced the data quality level shown.
To examine the effect of count times on the
analyzer's comparability, a subset of 26 samples from
the RV Hopkins site were reanalyzed using twice the
original count times. This increased the rs for both
chromium and copper measurements from 0.09 to 0.23
units, respectively. Antimony, arsenic, barium,
cadmium, lead, nickel, iron, and zinc did not show as
great an effect.
This analyzer reported concentration data for all
four secondary analyies. When the entire data set was
evaluated, the antimony results met the Level 2 criteria
(1^= 0.87, Slope=0.80, and y-intercept = 1.9 ppm).
The remaining secondary analytes met the Level 1 data
quality criteria. Similar to the primary analyies, sample
preparation appeared to be the dominant variable. Both
iron and nickel met Level 2 data quality criteria after the
first sample preparation step. Cadmium reached the
Level 2 data quality criteria after the third sample
preparation step. Antimony met Level 2 data quality
criteria for all four sample preparation steps.
74
ViiAff

-------
TABLE 5-8. REGRESSION PARAMETERS3 BY THE SAMPLE PREPARATION VARIABLE
AND SOIL TYPE—TN 9000

Arsenic
Barium
Copper



n
r2
Std. Err.
Y-lnt.
Slope"
n
r2
Std. Err.
Y-lnt
Slope"
n
r2
Std Err.
Y-lnt
Slope"
Soil Type

In Situ-Unprepared


Sand Soil
90
0.924
0.19
0.34
0.86
95
0.015
0.07
2.92
-0.06
94
0.918
0.15
0.34
0.88
Loam Soil
112
0.911
0.18
0.26
0.93
110
0.570
0.12
1.77
0.51
113
0.852
0.25
0.45
0 92
Clay Soil
2
ND
ND
ND
ND
98
0.742
0.17
1.69
0.57
44
0 494
0.18
1.21
0.61



In Sltu-Prepa
red




Sand Soil
92
0.986
0.09
0.04
0.97
97
0.078
0.08
2.55
0.15
95
0.973
0.09
-0.05
1.03
Loam Soil
112
0.981
0.08
0.15
0.95
112
0.509
0.15
1.73
0.55
113
0.979
0.09
0.27
0.94
Clay Soil
4
ND
ND
ND
ND
98
0.855
0.11
1.81
0.52
33
0.592
0.11
1.36
0.45

Intrusive-Unprepared



Sand Soil
87
0.989
0.08
•0.05
1.02
99
0.304
0.04
2.51
0.18
93
0.977
0.09
-0.11
1.07
Loam Soil
111
0.987
0.07
0.06
1.01
111
0.698
0.09
1.82
0.52
114
0.988
0.07
0.26
0.96
Clay Soil
3
ND
ND
ND
ND
98
0.929
0.08
1.72
0.57
24
0.394
0.15
0.82
0.75

Intrusive-Prepared
Sand Soil
90
0.980
0.11
0.09
0.99
95
0.293
0.03
2.63
0.13
95
0.970
0.10
0.03
1.04
Loam Soil
111
0.983
0.08
0.13
1.00
111
0.742
0.09
2.27
0.79
112
0.987
0.07
0.24
0.98
Clay Soil
3
ND
ND
ND
ND
99
0.938
0.09
1.56
0.63
34
0.541
0.16
0.98
0.68
TABLE 5-8 (Continued). REGRESSION PARAMETERS8 BY THE SAMPLE PREPARATION
VARIABLE AND SOIL TYPE—TN 9000

Lead
Zinc
Chromium (Low)

n
r2
Std. Err.
Y-lnt.
Slope"
n
r2
Std. Err.
Y-lnt.
Slope"
n
r2
Std. Err.
Y-lnt.
Slope"
Soil Type
In Situ-Unprepared
Sand Soil
87
0.920
0.17
0.36
0.86
80
0.950
0.11
0.40
0.85
19
ND
ND
ND
ND
Loam Soil
112
0.879
0.20
0.47
0.87
106
0.753
0.17
0.70
0.78
28
ND
ND
ND
ND
Clay Soil
96
0.776
0.28
0.74
0.81
92
0.861
0.20
0.38
0.97
56
0.688
0.17
1.81
0.43

In Situ-Prepared
Sand Soil
89
0.978
0.09
0.04
0.98
81
0.968
0.10
0.23
0.94
16
ND
ND
ND
ND
Loam Soil
109
0.971
0.10
0.15
0.97
101
0.923
0.10
0.37
0.90
18
ND
ND
ND
ND
Clay Soil
98
0.982
0.08
0.24
0.93
84
0.976
0.07
0.18
0.97
42
0.715
0.15
1.46
0.51

Intrusive-Unprepared
Sand Soil
88
0.984
0.08
0.01
1.01
79
0.963
0.11
0.00
1.03
3
ND
ND
ND
ND
Loam Soil
113
0.983
0.08
0.12
1.00
101
0.935
0.10
0.21
0.98
4
ND
ND
ND
ND
Clay Soil
98
0.989
0.07
0.21
0.95
86
0.979
0.08
0.06
1.03
40
0.793
0.12
1.32
0.58

Intrusive-Prepared

Sand Soil
87
0.972
0.11
0.11
1.00
77
0.955
0.12
0.12
1.00
3
ND
ND
ND
ND
Loam Soil
109
0.974
0.10
0.12
1.02
106
0.937
0.10
0.20
1.00
3
ND
ND
ND
ND
Clay Soil
99
0.988
0.07
0.17
0.98
81
0.978
0.08
0.05
1.06
43
0.842
0.13
0.76
0.80
75

-------
TABLE 5-8 (Continued). REGRESSION
PARAMETERS3 BY THE SAMPLE PREPARATION
VARIABLE AND SOIL TYPE—TN 9000
Soil Type
Chromium (High)
n
r2
Std. Err.
Y-lnt.
Slope"
In Situ-Unprepared
Sand Soil
3
ND
ND
ND
ND
Loam Soil
6
ND
ND
ND
ND
Clay Soil
36
0.544
0.19
1.82
0.48

In Situ-Prepared
Sand Soil
3
ND
ND
ND
ND
Loam Soil
4
NO
ND
ND
ND
Clay Soil
30
0.816
0.12
1.28
0.63

Intrusive-Unprepared
Sand Soil
3
ND
ND
ND
ND
Loam Soil
3
ND
ND
ND
ND
Clay Soil
37
0.779
0.14
1.14
0.69

Intrusive-Prepared
Sand Soil
3
ND
ND
ND
ND
Loam Soil
3
ND
ND
ND
ND
Clay Soil
39
0.794
0.16
1.44
0.62
Notes:
a	Regression parameters based on log10 transformed data.
b	Slope values determined by plotting FPXRF data on the
x-axis and the reference data on the y-axis.
Int. Y-lntercept.
Std. Err. Standard error,
n	Number of data points.
NA Not applicable, anatytes not present in significant quantities
to provide meaningful regression.
76

-------
TABLE 5-9. REGRESSION PARAMETERS3 BY THE SAMPLE PREPARATION VARIABLE AND
SITE TYPE—TN 9000
Site Name
Arsenic
Barium
Copper
n
r*
Std. Err.
Y-lnt.
Slope"
n
r2
Std. En.
Y-lnt
Slope6
n
r2
Std. Err.
Y-lnt
Slope6
In Situ-Unprepared
ASARCO Site
202
0.916
0.19
0.31
0.89
206
0.364
0.11
2.19
0.32
203
0.912
0.21
0.19
0.97
RV Hopkins Site
2
NO
ND
ND
ND
98
0.742
0.17
1.69
0.57
44
0.494
0.18
1.21
0.61

In Situ-Prepared
ASARCO Site
204
0.982
0.09
0.08
0.97
207
0.477
0.11
,2.10
0.38
207
0.980
0.10
0.03
V01
RV Hopkins Site
4
ND
ND
ND
ND
98
0.855
0.11
1.81
0.52
33
0.592
0.11
1.36
0 45

Intrusive-Unprepared
ASARCO Site
19B
0.986
0.08
0.00
1.02
207
0.600
0.06
2.29
0.30
208
0.984
0.09
0.04
1.02
RV Hopkins Site
3
ND
ND
ND
ND
98
0.929
0.08
1.72
0.57
24
0.394
0.15
0.82
0.75

Intrasvie-Preparad
ASARCO Site
201
0.980
0.10
0.11
1.00
206
0.583
0.05
2.35
0.27
208
0.982
0.09
0.11
1.02
RV Hopkins Site
3
ND
ND
ND
ND
99
0.938
0.09
1.66
0.63
34
0.541
0.16
0.98
0.68
TABLE 5-9 (Continued). REGRESSION PARAMETERS® BY THE SAMPLE PREPARATION
VARIABLE AND SITE TYPE—TN 9000

Lead
Zinc
Chromium (Low)

n
r*
Std. Err.
Y-lnt.
Slope6
n
r2
Std. Err.
Y-lflt
Slope"
n
r2
Std. Err.
Y-lnt.
Slope6
Site Name
In Situ-Unprepared
ASARCO Site
200
0.885
0.20
0.38
0.88
186
0.860
0.16
0.51
0.83
47
ND
ND
ND
ND
RV Hopkins Site
96
0.776
0.28
0.74
0.81
93
0.867
0.20
0.43
0.95
56
0.688
0.17
1.81
0.43

In Situ-Prepared
ASARCO Site
196
0.975
0.10
0.09
0.98
183
0.947
0.10
0.28
0.92
34
ND
ND
ND
ND
RV Hopkins Site
98
0.982
0.08
0.24
0.93
84
0.976
0.07
0.18
0.97
42
0.715
0.15
1.46
0.51

Intrusive-Unprepared
ASARCO Site
200
0.980
0.09
0.05
1.02
181
0.939
0.12
0.13
1.00
6
ND
ND
ND
ND
RV Hopkins Site
98
0.989
0.07
0.21
0.95
86
0.979
0.08
0.06
1.03
40
0.793
0.12
1.32
0.58

Intrusive-Prepared


ASARCO Site
197
0.970
0.11
0.12
1.01
184
0.939
0.12
0.17
1.00
5
ND
ND
ND
ND
RV Hopkins Site
99
0.988
0.07
0.17
0.98
81
0.978
0.08
0.05
1.06
43
0.842
0.13
0.76
0.80
77

-------
TABLE 5-9 (Continued). REGRESSION
PARAMETERS3 BY THE SAMPLE PREPARATION
VARIABLE AND SITE TYPE—TN 9000
Site Name
Chromium (High)
n
r2 | Std. Err.
Y-Int
Slope"
In Situ-Unprepared
ASARCO Site
9
ND
ND
ND
ND
RV Hopkins Site
36
0.544
0.19
1.82
0.48

In Situ-Prepared
ASARCO Site
7
ND
ND
ND
ND
RV Hopkins Site
30
0.816
0.12
1.28
0.63

Intrusive-Unprepared
ASARCO Site
5
NO
ND
ND
ND
RV Hopkins Site
37
0.779
0.14
1.14
0.69

Intrusive-Prepared
ASARCO Site
6
ND
ND
ND
ND
RV Hopkins Site
39
0.794
0.16
1.44
0.62
Notes:
8	Regression parameters based on log10 transformed data.
b	Slope values determined by plotting FPXRF data on the
x-axis and the reference data on the y-axis.
Int. Y-lntercept
Std. Err. Standard error,
n	Number of data points.
NA Not applicable, analytes not present in significant quantities to
show meaningful regression.
78


-------
Section €
Applications Assessment and Considerations
The two TN Spectrace analyzers evaluated during
this demonstration are designed 10 produce quantitative
data on the concentration of metais in soils, sludges, and
other solids. The TN Spectrace "Soils Application"
software was used for calibration and quantitation to
maximize instrument performance and account for
common soil-related matrix interferences. The FP
calibrations were fine tuned with NIST soil SRMs to
further improve data comparability. These analyzers arc
designed for field use and exhibited roggedness through
a variety of environmental operating conditions. These
analyzers never experienced failures resulting in down
time throughout the I-month field demonstration.
During this time, over 1,260 samples were analyzed by
each FPXRF. The short training video provided by the
developer was sufficient to allow basic field operation of
either analyzer. The developer also offers a training
class in the use of the analyzers and this training coupled
with on-line technical support was sufficient to allow
uninterrupted operation and no data loss throughout the
demonstration.
Comparison of TN Lead Analyzer and reference
data indicated thai the analyzer generally provided, at a
minimum. Level 2 quality data as defined in this study.
This data quality level is applicable to most field
screening applications. The data produced by this
analyzer was linearly correlated to the reference data.
In addition, this linear correlation appears to hold over
5 orders of magnitude. The linear relationship between
the analyzer and the reference method would indicate
that if 10 to 20 percent of the samples analyzed were
submitted for reference method analysis, any TN Lead
Analyzer bias for Level 2 data could be evaluated and
the measured concentrations could be corrected to
simulate the reference data. In the case of copper, lead,
and arsenic, after the initial sample hotnogenizations, the
TN Lead Analyzer's data was statistically equivalent to
the reference data. This analyzer also exhibited analyzer
precision similar to the reference method, indicating a
high degree of reproducibility.
- The TN Lead Analyzer is generally operated with
relatively short count times and has only one radioactive
source. The single radioactive source limits the number
of analytes which can be detected. The TN Lead
Analyzer's "Soils Application" software can report
concentrations for arsenic, chromium, iron, copper,
zinc, and manganese in soil samples. The shorter count
times and the single radioactive source combine to
generally increase the possible sample throughput
and detection limits, but decrease the analyzer accuracy.
Two hundred to 300 samples were analyzed in a 10-hour
day during the demonstration. The advantages and
disadvantages of the TN Lead Analyzer are summarized
in Table 6-1.
Comparison of TN 9000 and reference data
indicated that the TN 9000 generally produced Level
3 quality data for arsenic, copper, lead, and nickel after
the initial sample homogenixation. This indicates that
the TN 9000's data was statistically equivalent to the
reference data for these analytes. For the other target
analytes the TN 9000 produced Level 2 data except for
chromium. As with the TN Lead Analyzer, if 10 to 20
percent of the samples analyzed by the TN 9000 were
submitted for reference method analysis, bias in the TN
9000 for Level 2 data could be determined and the data
could be corrected to simulate the reference data. In
addition, this analyzer exhibited instrument precision
similar to the reference method, indicating high
instrument reproducibility.
The TN 9000 can use up to three radioactive sources
allowing analysis of a large number of metals in soils.
The TN 9000's "Soils Application" software can report
concentrations for potassium, calcium, titanium,
chromium, manganese, iron, cobalt, nickel, copper,
zinc, arsenic, selenium, rubidium, strontium, zirconium,
molybdenum, mercury, lead, uranium, thorium, silver,
cadmium, antimony, tin, and barium. The TN 9000
generally uses longer count times, which are
proportional to the number of sources used in
79
****¦> •r" -

-------
TABLE 6-1. ADVANTAGES AND LIMITATIONS—TN LEAD ANALYZER
Advantages
Disadvantages
Field portable — weighs less than 20 pounds,
battery lifetime of 8 hours
Limited analytes — did not report antimony,
barium, cadmium, or nickel
High sample throughput — 25 samples per hour
Single excitation source
Can conduct in-situ measurements or measure
prepared samples in cups
MDLs 5 to 10 times higher than SW-846 Methods
3050A/6010A
Rugged and reliable — no downtime; achieved
data completeness of 100 percent
Higher cost per sample relative to SW-846
Methods 3050A/6010A for sampling and analysis
events of fewer than 30 samples
Easy operation — one day training, no prior
experience necessary
Poor comparability (r2 less than 0.80) and precision
(RSD greater than 20 percent) for chromium
Real-time data
Field-based MDLs are generally 2 to 3 times higher
than developer-supplied or precision-based MDLs
All measured radiation levels below occupational
limits
Poorer accuracy for ash and sludge SRMs
Lower cost per sample relative to Methods
3050A/6010A, for sampling and analysis events
of greater than 80 samples

Drift less than ±10 percent for all analytes
monitored
Produces EPA Level 2 quality or better data
•	Arsenic, copper, and lead — Level 3
•	Chromium, iron, and zinc — Level 2
EPA-approved method for FPXRF
(SW-846 Update 4)
Data is strongly linearly related to Methods
3050A/6010A data — r2 values greater than 0.92
for all analytes except chromium
FP calibrations that can be fine tuned with site-
specific samples
Good precision (reproduceable results) —
Percent RSD values less than 10 percent at 5 to
10 times the MDL for all analytes except
chromium
Accuracy of 100 percent for all analytes in soil
SRMs
Generally not susceptible to soil matnx effects
Matrix-specific FP calibrations in "Soils
Applications" software
Can be used on soils exhibiting over 30 percent
saturation by weight
80


-------
analysis. The longer count times and multiple sources
generally increase accuracy and lower the detection
limits, but decrease sample throughput. Eighty to 100
samples were analyzed in a 10-hour day during the
demonstration. The advantages and disadvantages of the
TN 9000 are summarized in Table 6-2.
For both analyzers, there was no apparent effect of
site or soil type on performance. This demonstration
identified sample preparation as the most important
variable with regard to analyzer performance. Both of
these analyzers can be applied in an in situ mode. The
data from this demonstration indicated that when
operated in the in situ mode, the user most probably
would not be able to show a strong correlation between
FPXRF and reference data. This may not be due to
instrument error, but rather to inherent spacial variability
of contamination, even within an area as smalt as the 4-
inch by 4-inch grid sampled during this demonstration.
The greatest increase in correlation between the FPXRF
data and reference data for both analyzers was achieved
after the initial sample homogenization. Further sample
preparation such as sieving or drying and grinding, is
most cases did not greatly improve the comparability.
This more involved sample preparation was generally
needed to improve the quality of chromium data. This
was indicative of the general problematic nature and
influence of panicle size of chromium determination by
FPXRF. With the exception of antimony and cadmium,
both analyzers produced field-based MDLs 2 to 3 times
greater than the dcve)aper-supplied or precision-based
MDLs.
The costs associated with the use of these analyzers
suggests an economy of scale. To be cost effective
relative to the reference methods (when leasing the
analyzers), a sampling and analysis event using either of
these analyzers would need to analyze around 00
samples. At this point, the FPXRF analysis becomes
Less costly than the reference methods. Although the
analyzers may provide data of lesser quality or data not
currently accepted by regulatory agencies, their
applications should be considered due to the quantity of
information produced relative to conventional analysis.
In some cases, more data of a lower quality would be
more useful than only a few data points at the highest
data quality. This is a dilemma commonly associated
with field analysis.
Based on this demonstration, both of these analyzers
are well suited for the rapid real-time assessment of
metals contamination in soil samples. Although in
several cases the analyzers produced data statistically
equivalent to the reference data, generally confirmation
analysis will be required or requested for iTXRf
analysis as is indicated in the draft Method 6200. If 10
to 20 percent of the samples analyzed by either analyzer
are submitted for reference method analysis, instrument
bias, relative to standard methods such as Methods
3050A/6010A. can be corrected. This will only hold
true if the analyzers and the laboratory analyze similar
samples. This was accomplished in this demonstration
by thorough sample homogenization Bias correction
allows analyzer data to be corrected so that it
approximates the Methods 3050A/601QA data. The
demonstration showed thai these analyzers exhibit a
strong linear relationship with the reference method data
over a 5 orders of magnitude concentration range. For
optimum correlation, samples with high, medium, and
low concentration ranges from a project must be
submitted for reference method analysis.
These analyzers can provide rapid assessment of the
distribution of metals contamination at a hazardous waste
site. This data can be used to characterize general site
contamination, guide critical conventional sampling and
analysis, and monitor removal actions. This demon-
stration suggested that in some applications and for some
elements, the data may be statistically similar to the
reference data. The drafting of SW-846 Method 6200
"Field Portable X-Ray Fluorescence Spectrometry for
the Determination of Elemental Concentrations in Soil
and Sediment'' may help in the acceptance of this data
for some Level 2 applications and possibly Level 3
applications. The analyzer data can be produced and
interpreted in the field on a daily or per sample basis.
This real-time analyst allows the use of-contingency-
based sampling for any application; and greatly increases
the potential for meeting project objectives on a single
mobilization. These analyzers are powerful tools for site
characterization and remediation. They provide faster
and less expensive means of analyzing metals
contamination in soil.
The following general guidance on FPXRF
application applies to both TN Spectrace analyzers, and
was based on draft SW-846 Method 6200.
General operation of FPXRF instruments will vary
according to the developer protocols. Before operating
any FPXRF instrument, the developer's manual should
be consulted. Most developers recommend that their
instruments be allowed to warm up for IS to 30 minutes
before analysis of samples. This will help alleviate drift
or energy calibration problems later in analysis.
An FPXRF instrument should be operated according
to the developer's recommendations. There are two
modes in which FPXRF instruments can be operated: in
situ and imnisive. The in situ mode involves analysis of
an undisturbed soil or sediment sample. Intrusive
analysts involves collection and preparation of a soil or

-------
TABLE €-2. ADVANTAGES AND LIMITATIONS—TN 9000
Advantages
Disadvantages
Field portable — weighs less than 20 pounds,
battery lifetime of 4 to 5 hours
Method detection limits 5 to 10 times higher than
Methods 3050A/6010A
High sample throughput — 10 samples per hour
Higher cost per sample relative to Methods
3050A/6010A for sampling and analysis events
of fewer than 30 samples
Can conduct in-situ measurements or measure -
prepared samples in cups
Poor comparability (r2 less than 0.85) and
precision (RSD greater than 20 percent) for
chromium; chromium only met Level 1 data
quality
Rugged and reliable — no downtime; data
completeness of 99.9 percent
Field-based MDLs are generally 2 to 3 times
higher than developer-supplied or precision-
based MDLs
Easy operation — one day training, no prior
experience necessary
Poorer accuracy for sediment, ash, and sludge
SRMs
Real-time data

All measured radiation levels below occupational
limits
Lower cost per sample relative to Methods
3050A/6010A, for sampling and analysis events of
greater than 80 samples
Produces EPA Level 2 quality or better data for
most analytes
•	Arsenic, copper, lead, nickel — Level 3
•	Barium, iron, cadmium, antimony, zinc — Level 2
EPA-approved method for FPXRF
(SW-846 Update 4)
Data is linearly related to Methods 3050A/6010A
data — r2 values greater than 0.92 except for .
barium and chromium
FP calibrations that can be fine tuned with site-
specific samples
Good precision {reproduceabte results) — percent
RSD values less than 10 percent at 5 to 10 times
the MDL for all analytes except chromium
Accuracy of 90.2 percent and 91.7 percent for all
analytes in the site-specific PEs and soil SRMs,
respectively
Generally not susceptible to soil matrix effects
Three excitation sources allowing for analysts of
over 30 elements
Matrix-specific FP calibrations in 'Soils
Applications" software
Can be used on soils exhibiting over 30 percent
saturation by weight

-------
sediment sample before analysis. Some FPXRF
instruments can operate in both modes of analysis, while
others are designed to operate in only one mode. The
two modes of analysis are discussed below.
For in situ analysis, one requirement is that any
large or nonrepresentative debris be removed from the
soil surface before analysis. This debris includes rocks,
pebbles, leaves, vegetation, roots, and concrete.
Another requirement is that the soil surface be as smooth
as possible so that the probe window will have good
contact with the surface. This may require some
leveling of the surface with a stainless-steel trowel.
During the demonstration, this modest amount of sample
preparation was found to take less than 5 minutes per
sample location. The last requirement is that the soil or
sediment not be saturated with water. Developers state
that their FPXRF instruments will perform adequately
for soils with moisture contents of 5 to 20 percent, but
will not perform well for saturated soils, especially if
ponded water exists on the surface. Data from this
demonstration did not see an effect on data quality from
soil moisture content. Source count times for in situ
analysis usually range from 30 to 120 seconds, but
source count times will vary among instruments and
depending on required detection limits.
For intrusive analysis of surface soil or sediment, it
is recommended that a sample be collected from a 4- by
4-inch square that is 1 inch deep. This will produce a
soil sample of approximately 375 grams or 250 cm3,
which is enough soil to fill an 8-ounce jar. The sample
should at a minimum be homogenized, and at a
maximum, dried, and ground before analysis. The data
from this demonstration indicated that advanced sample
preparation, beyond homogenization, does not greatly
improve data quality. Sample homogenization should
thoroughly mix the sample, or homogenization can be
conducted by kneading a soil sample in a plastic bag.
One way to monitor homogenization when the sample is
kneaded in a plastic bag is to add sodium fluorescein dye
to the sample. After the moist sample has been
homogenized, it is examined under an ultraviolet light to
assess the distribution of sodium fluorescein throughout
the sample. If the fluorescent dye is evenly distributed
in the sample, homogenization is considered complete;
if the dye is not evenly distributed, mixing should
continue until the sample has been thoroughly
homogenized. During the demonstration, the
homogenization procedure using the fluorescein dye
required 3 to 5 minutes per sample.
Once the soil or sediment sample has been
homogenized, it can be dried. This can be accomplished
with a toaster oven or convection oven. A small aliquot
of the sample (20 to 50 grams) is placed in a suitable
container for drying. The sample should be dried for 2
to 4 hours in the convection or toaster oven at a
temperature not greater than 150 °C. Microwave drying
is not recommended. Field studies have shown that
microwave drying can increase variability between the
FPXRF data and reference method analysis. High levels
of metals in a sample can cause arcing in the microwave
oven, and sometimes slag will form in the sample.
Microwave oven drying can also melt plastic containers
used to hold the sample.
The homogenized dried sample material can also be
ground with a mortar and pestle and passed through a
60-mesh sieve to achieve a uniform panicle size.
Sample grinding should continue until at least 90 percent
of the original sample passes through the sieve. The
grinding step normally takes an average of 10 minutes
per sample.
After a sample is prepared, an aliquot of the sample
should then be placed in a 31-mm polyethylene sample
cup (or equivalent) for analysis. The sample cup should
be one-half to three-quarters full at a minimum. The
sample cup should be covered with a 2.5-micrometer
Mylar (or equivalent) film for analysis. The rest of the
soil sample should be placed in a jar, labeled, and
archived for possible confirmation analysis. All
equipment including the mortar, pestle, and sieves must
be thoroughly cleaned so that any cross contamination is
below the MDLs of the procedure or data quality
objectives of the analysis.

-------
Section 7
Developer Comments and Technology Update
TN Spectrace submitted both editorial and technical
comments on the TN Lead Analyzer and TN 9000 draft
ITER. The developer's comments are presented
verbatim in italics. PRC's responses follow each
developer comment.
1.	The theory section 4 should be in a separate
section from the specific instrument discussion.
The XRF theory discussion has been moved to
Section 2.0 (Introduction).
2.	Details should be provided (in an appendix) of
the analytical data of the various SRM, CRMs. PEs,
etc. mentioned in the report. According to my
recollection, if we made any fine tuning of our
standard FP calibration (as you mention in a few
places) it would have been based only on the NIST
27XX soils and not any PE samples. 1 am curious
what these and the CRMs comprised especially in
view of your comparative results.
It would also be of value to all participants if a
more detailed profile could be provided of the actual
lab soil results. You have indicated some break
point anomaly in the results at 2000 mg/kg. 1 note
that both Pb and As span this range and it would
help to know if both elements ranged together in any
systematic way.
All the raw analytical data for the FPXRF
analysis of the SRMs. CRMs, and PEs and the true
analyte concentrations in these samples, as well as
the actual reference laboratory results, are contained
in the technical evaluation report (TER), which can
be obtained from Mr. Stephen Billets, EPA-Las
Vegas. Both lead and arsenic showed the anomaly
at 2,000 mg/kg as displayed on Figure 5-4 of the
ITER. In general, for the soil samples from the
ASARCO site, the arsenic and lead concentrations
decreased and increased systematically together.
3.	Pg xiii: I'd appreciate the inclusion of my name
in the Acknowledgments.
Your name (Peter Berry) has been added in the
Acknowledgments.
4.	Pg 2: First para. At top right of page: it is
erroneous to state that the Lead Analyzer can be
"operated at shorter count times" and "can provide
a four- to five-fold increase in sample throughput"
just because qfthe "type and activity of its source."
While its source is nominally 6 times the miUicurie
content of the source in the 9000, its useful 22 kev
emission is in fact less intense because of the
attenuating shims within the capsule. These are
necessary to balance the 22 kev output with the
increase output of the 88 kev gamma ray which only
beneficial for the PbK analysis in wall paint. The
main reasons for the increase in sample throughput
relative to the 9000 are the reduced Cd exposure
time (i.e. 60 versus 100 sec) and the fact that oniy
source is used in the measurement protocol. The
9000 uses at least an additional 120 seconds for the
exposure of the other two excitation sources.
The text has been modified to indicate that the
use of only one source at a shorter count time
increased the sample throughput for the TN Lead
Analyzer.
5.	Pg 2: Next para. Concerning the listed
elements for data quality 2 on the 90001 have some
questions concerning Cr and Zn later. The Co
seems to be out of place. 1 can find no data in the
report.
The word cobalt has been changed to cadmium.
6.	Pg 2: Next para. I think the statement that the
analyzers exhibit precision similar to the reference
method should be clarified by the addition of "at the
84
* ' ¦ '¦
.1

-------
5-10 MDL concentration level" or something equivalent.
Otherwise one might wonder why the XRF MDLs which
depend on the precision are not also comparable with
those of the reference methods. In regard to the MDL
statement in the next sentence it should be emphasized
that the "developer-provided" MDL values are based on
the use of a longer measurement (by a factor of 2-3) over
that used in this study. Since precision based MDLs
depend on the square root of the measurement time, the
values of the "developer-provided" MDLs used for
comparison in this report should have been increased by
a factor of 1.4 (for the Cd excited elements) and 1.8 (for
the Fe and Am excited elements). That would then
obviously reduce the factor of difference with the field
based data.
The MDLs have been modified as you
requested in the subsection discussing MDLs. (See
comments 24 and 35).
7.	Pg 9: 2nd para. Was the 10 mesh sample
homogenized prior to dispensing into cups?
Yes. The word "homogenized" has been added
to the first sentence of this paragraph for
clarification. Figure 2-2 also indicates the sample
was homogenized prior to passing it through the 10-
mesh sieve.
8.	Pg 9/10: Joining para. Why is the word
compound used when intercomparisons are based on
elements? The same para also talks about
"duplicates" and "nondetects, " does this refer to
XRF measurements? It is confusing since the last
sentence says that the reference method "reported
measurable values "for all samples.
The sentence you commented on has been
removed from the text as it is not relevant because
the reference methods reported measurable values
for all samples.
9.	Pg JO: Para at bottom of left column. The third
sentence indicates the MDL is 3 times the relative
standard deviation. This is incorrect, the MDL is
expressed in terms of absolute deviations.
The method of obtaining the MDL from the /in'n
spread around the point where a plot of the XRF
results versus their reference values shows little or
no change appears somewhat unusual but should be
equivalent to the method of analyzing repetitive
measurement on samples of low concentration. The
suggested method however could yield very different
re<sfohhe collection of data points at the selected
^/threshold level was limited in number or inclusive of
samples with unusually wide-varying concentrations
of matrix elements, having a strong effect on the
analyte precision. It would be informative to see a
table of these data values to see if there are any
"non-anatyte" concentration anomalies. It is also
somewhat unclear how the "threshold" and "2-
sigma range" values are arrived at. Is this a
subjective judgment? In a later table (Pg 61) of the
9000 MDLs. a comment is made in the case of Ba
that its "Method MDL" may be high because of
possible incomplete digestion in the lab procedure.
Might the same be true for other elements?
The word "relative" has been deleted from the
text. There were no nonanalyte concentration
anomalies noted in the data. It is agreed that the
threshold and 2-sigma range values are somewhat
subjective as indicated in the text. No changes have
been made to the text based on this comment.
10. Pg 11: In the para at the top of the left column,
there is a statement that the microwave drying was
a "minor variable " yet later on pg 17 it is shown to
be a major concern. This is confusing.
This sentence has been removed from the text.
In the next para, it is not true that the count
time is a minor variable on the precision, especially
when precision is derived from replicate
measurement. XRF precision con be expected to
change as the square root of the time. What were
the "longer" and "shorter" times tested in the study
and will the results of the measurements be
published? It is also important to note that any
measurement of precision will in itself be subject to
some uncertainty based on the number of
measurements employed. If only 10 replicate
measurements are used the uncertainty in the
calculated precision value will be about 20%
relative.
The word "minor" has been removed from the
text. The longer count times were twice as long.
This language has been added to the text. No
precision or replicate measurements were collected
using the twice as long count times. The word
"replicate" has been removed from the text.
Samples were reanalyzed using the longer count
times to see if there was an effect on comparability
to the reference laboratory. There is a short
discussion of the longer count time data at the end
of Sections 4 and 5.
In the next para, it was good that the radiation
measurements were made and as concluded by the
85


-------
measurements (and reported in the specific instrument
section) no significant radiation levels were detected. I
think these results should be briefly summarized in the
Executive section. In the draft report of the 6200
method, tsection 5) it is being proposed thai radiation
monitoring will be performed routinely during operation.
Are you aware of this? This seems to be an unnecessary
requirement.
A sentence has been added to the Executive
Summary discussing the low radiation levels
detected while operating these two analyzers.
11.	Pg 16. The formulae and footnotes contain
several errors which I think will be obvious on close
examination.
The formulae have been modified to indicate
the correct subscript terms and the square root has
been removed from the denominator.
12.	Pg 17. I commented on the microwave drying
issue earlier.
Changes were addressed based on developer
comment number 10.
13.	Pg 21. There is a typo in the lower left cell of
the table.
The word "postdigestion* has been spelled
correctly.
14.	Pg 22. Re Table 3-2, the mgfkg unit should be
inserted directly into the table alongside LRL.
The units "mg/kg" have been added to Table 3-
2.
15.	Pg 32. There is a typo "TNLead Analyzer" in
the figure caption.
The words "TN Lead Analyzer" have been
removed from the figure caption.
16.	Pg 33, A reference to the source of the
published {recovery) result should be included in the
table.
A footnote has been added to Table 3-5 to
indicate that the recovery results were obtained from
the addendums to the NIST SRM certificates.
17.	Pg 34. I suggest you break out the "principles
etc. "to a new section. 1 offer the following text as
a possible INTRO to give a better understanding of
the energy dispersive method and appreciation of the
non-destructive (sample prep) considerations.
"Principles of the FPXRF Analysis Method"
FPXRF analyzers operate on the principle of
energy dispersive XRF Spectrometry whereby the
characteristic energy components of the excited x-
ray spectrum are analyzed directly via their energy
proportional response in the x-ray detector.
Compared to the traditional methods of wavelength
diffraction, energy dispersion affords highly efficient
full-spectrum measurement which enables the use of
low intensity excitation sources (such as
radioisotopes) and compact design battery-powered
field-ponable instruments. Many FPXRF instrument
designs based on various energy dispersive detector
technologies are now widely used for composition
analysis in the industrial and environmental arena.
Such applications can also make the best use of the
essentially nondestructive nature of the XRF
measurement technique. Mainly it is sufficient that
the sample be homogeneous or at least prepared to
the extent that it is effectively homogeneous on the
scale of the x-ray penetration, typical x-ray
penetration depths might range from about O.J to 1
mm for the x-rays of most targeted metal
contaminants in the environmental samples.
'Fluorescent x-rays are produced by exposing
a sample to x-rays having an energy etc etc "
The discussion of FPXRF theory has been
moved to Section 2.0 (Introduction) and your
INTRO paragraph has been added to the discussion.
18. Pg 35: The remark that Kabs is apprax=sum of
emission energies is not true if all the energies are
summed. It applies only if the "alpha " energies are
summed.
The word "alpha" has been added to this
sentence.
/ assume there will be a new subsection in the
Lead Analyzer section entitled "Operation and
Background Information. " In that section and in the
corresponding section of the TN 9000 (pg 56)
description, the paragraphs which describe the Hgl
detector should read The TN Lead Analyzer (or TN
9000) uses Hgl2 semiconductor detector that
achieves an Mn-Ka. x-ray resolution of better than
300 eV. The detector is operated at a moderately
sub-ambient temperature controlled by a low power
86

-------
thermoelectric (pettier) cooler in the measurement
probe."
The description of the Hgl2 detector in Sections
4 and 5 has been changed per your comment.
In the next para and in several other places of
this and the TN 9000 section, 1 notice the use of the
words 'source-detector beryllium window" in
connection with the sample positioning. This should
be corrected in all places to read "plastic-film probe
measurement window." At the end of the para
substitute the words "	and the sample.
contained in a thin windowed plastic cup is placed
over the probe measurement window and beneath a
swing-down safety shielded enclosure." These same
words should be used in the 9000 description.
The description of the probe window and the
analysis of soil samples in cups has been changed in
Sections 4 and 5 per your comment.
In the next para, delete the word "backscaaer"
after FP and substitute "calibrated." For some
reason our FP method has got confused with the
Backscaxter method used in the Metorex instrument.
It is not the same. The same correction is needed
on pg 56.
The word "backscatter" has been deleted and
the word "calibrated" inserted in Sections 4 and S.
19.	Pg 36: 2nd para Replace "Be-window" with
"mylar-window."
This change was made per developer comment
number 19.
20.	Para 4: Change the wording "or directly from
an alternating current	"to read "orfrom an
alternating current AC line via a plug-in adaptor
unit. " The minimum hours to recharge batteries is
14. not 16.
The changes were made in Sections 4 and 5 per
developer comment number 20.
There is a typo "silicone " in last para on this
page.
The word "silicon" is how spelled correctly.
21.	Pg 37 Table. Change the resolution spec to
"<300 eV. " Repeat on Pg 57. Add the words
"with shim inserts'to the source name. Change the
Electronic Unit Oper. Temp Range to that of the
probe unit. Repeat on Pg 57. For the Power
Source add 230 to the AC spec. Repeat on Pg 57.
All changes were made to Tables 4-1 and 5-1
per developer comment number 21.
22.	Pg 38: Mid left column. Needs a rewrite as
follows:
"A reliability check of the TN Lead Analyzer
was carried out by measuring a check sample daily.
This check comprised of a 50 second measurement
of a pure Fe sample. By this one measurement a
verification was obtained of: (1) the fluorescent
element sensitivity: (2) the spectrometer energy
resolution; and (3) the spectrometer energy
calibration. To be acceptable the measured relative
x-ray intensity of iron had to be greater than 0.95
and the equivalent intensity of Mn and Co had to be
less than 0.006. Relative intensity refers to the new
value relative to that obtained at the time of the
initial instrument calibration. If the intensity
conditions were not met	etc	"
A similar rewrite will be needed for the 9000
(Pg 58).
The rewrite has been included in the
"Reliability" subsection of Sections 4 and 5.
23.	In the next para it might be less alarming to
some readers if the words "The potential for
exposure to radiation...." were used. Likewise for
the 9000 on pg 59.
The change has been made to Sections 4 and 5
per developer comment number 23.
24.	Pg 40 re MDLs: An indication of the 60 sec
data acquisition time should be given on the table.
An additional column was added to Table 4-3
that lists developer MDLs that were corrected for
the difference in count times (200 seconds versus 60
seconds). A footnote has been added to the table to
indicate the developer MDLs were corrected for the
variance in count times.
25.	Re Throughput: 2nd line, substitute the word
"system "for "deteaor.'
The change has been made per developer
comment number 25.
26.	Pg 43 last para: To the best of our recollection
no calibration tune-up was performed on the PE
87

-------
samples, but some tune-up was performed using some of
the NIST soil SRMs. Thus some of the wording in this
para will need rework.
This paragraph has been modified to indicate
that the FP calibration was fine-tuned to the NIST
soil SRMs and not to the site-specific PE samples.
There is much memion of CAMS in this work.
We have no knowledge of what these materials
contain to comment on the data, but it underscores
the need for additional detail by way of "certs " and
material descriptions" to be included as an appendix
to the report.
This information can be found in the TER
which can be provided to you by Mr. Billets of
EPA-Las Vegas.
27.	Pg 51: The identity "TNLead Analyzer" needs
to be incorporated into the figure caption as is the
case of the 9000 figure on pg 77.
The figure caption has been changed to identify
the TN Lead Analyzer.
28.	Pg 52: 1st para. The comment about the
analyzer sensitivity to the particle size for Cr
analyzer, and for other elements to a different
degree is well placed. However it should not be
construed that this is a specific instrument
limitation. It is a limitation of XRF in general.
Thus a general remark about XRF of lower Z
element x-rays and coarse panicle effects might be
in order in this section.
Language has been added to this paragraph to
indicate that particle size will effect chromium
analysis for all FPXRF analyzers.
29.	Pg S3. Figure. The meaning of the numbers on
the Sample prep axis become clear after a while but
it would help if there was a "key" at the foot of the
figure.
This is a valid comment; however, there was
insufficient space at the bottom of the figure to
include a description of all the sample preparation
steps.
30.	Pg 56: In addition to the suggestions made in
common with the Lead Analyzer section (especially
about the Be window) I would like to see in the pan
at the foot of the left column the following: In place
of 'can be used" use "is exposed." In place of
"three separate source exposures are required" use
"three source exposures are sequenced
automatically."
The changes have been made per developer
comment number 30.
31.	Pg 57: Make the same changes to the Table as
were indicated for the Lead Analyzer and, in the
text, regarding the AC operation and charge times.
There is a typo in the probe operating range given
in the first para. In the last para the words "pointed
at" should be replaced with "placed in contact
with."
The changes have been made per developer
comment number 31.
32.	Pg 58: In the last para of the left column,
concerning the developer calibration adjustment, we
would like to stress that no site-specific sample
adjustments were made but some adjustments were
made to accommodate some the NIST soil SRMs.
The text has been changed to indicate NIST soil
SRMs were used and not site-specific soil samples.
In the preceding sentence, the "Be " should read
"plastic"
The change has been made per developer
comment number 32.
In the first para of the right column, there is an
appeal for a PgUP and PgDN to scroll data. This
has always been a feature of the instrument and is
described in the manual (pg 4-3). I guess we didn't
do a good enough job in the training.
This sentence has been deleted. It was realized
after the conclusion of the field work that this option
was available on the analyzer.
33.	Pg 58: In the last para, the rewrite from the
Lead Analyzer (pg 38) should be used.
Changes were made per developer comment
number 33.
34.	Pg 59: Note the earlier suggestion about the
"radiation exposure " item.
Changes were made per developer comment
number 34.
35.	Pg 60:, Para in left column: Instead of
"Cadmium was detected only	7 think

-------
"Cadmium was present only	" is more
appropriate.
The word "detected" was deleted and the word
"reported" was inserted.
/ would like to suggest a change in the MDL
Table 5-3 and corresponding modification to the text
in this paragraph. In the table, could we modify, or
append in brackets, the Developer MDL values to
reflect heir expected value for the shorter
measurement time employed? As I have said in
several places, our predicted values are for 200
seconds on each source. They need to be adjusted
for the 100/60/60 second regimen. If the change is
made, the adjusted values will bexl.8 larger for Sb.
Ba, Cd, CrLo. and xl.4 larger for the Cd 109
elements CrHi, Cu, Fe. Pb, Ni, and Zn.
The single value presently in the Cr box should
have been two values 90(a> and 263With these
changes there is seen to be less difference between
the Developer and Precision values, especially when
one considers that the precision based data {from
only 10 replicates) can be in error by as much as
20%. The difference with ihe field data is still hard
to explain without seeing the raw data. It could well
be due to the presence of strongly absorbing
elements in the field samples.
The developer MDLs have been corrected to
account for the count time difference of 200 versus
100 seconds or 200 versus 60 seconds.
Also on Pg 60. under throughput, change "of
the detector" to "of the system."
The changes have been made per developer
comment number 35.
36. Pg 61. Regarding Drift: The apparent drift
attributed to the As results is more than likely a
statistical fluctuation since if one looks at the
precision of the As analysis in the presence of5500
ppm Pb (as in N2710) it will be seen to be approx
18% relative in a 100 second measurement. In
order to satisfy the <10% drift criteria the
statistical error would have to be reduced. A
measurement time of300 seconds would have to be
used. On the other hand it has to be recognized that
As is less precise in the presence ofPb at >x 10 the
As concentration.
Language has been added to this paragraph to
indicate thai the precision for arsenic may be up to
18 percent relative in the presence of 5.500 ppm
lead.
Regarding the blanks: It seems a pity that the
results of repetitive Si02 measurement were not also
used to develop a precision based MDL. That value
should have been more akin to the developers
precision number. The small amount ofFe reported
on the blank is actually present in the Si02-
The objective of the demonstration was to
calculate MDLs based on real field samples not on
a clean Si02 matrix to give a user a realistic value
for field use. That is why replicate measurements
were not conducted on the silicon dioxide blank.
The fact about the iron being present in the silicon
dioxide has been added to the text.
37.	Pg 63: Next to the last para, last sentence. For
which elements were the 9000 results 2 times the
SRM value, and for which SRMs? This statement
does not fit with most of the preceding remarks.
The TN 9000 results for analyses in all SRMs
were less than 2 times the true value. The text has
been changed to clarify this sentence.
38.	Pg 64: Last para in left column. As stated
previously, there was no calibration tune-up as we
recall on the PE samples, only the SRMs. Judging
by the data in Table 5-5 the calibration was near
perfect for the NIST soils, but not so for the ash and
sludge. This is not surprising considering their
different nature.
This paragraph has been rewritten to indicate
that the FP calibration tuneup was based on the
NIST soil SRMs and not on the site-specific soil
samples.
Concerning the Intermethod Assessment (Pgs 70-78)
I have only a few comments on this section
since I don't fully understand some of the
methodology in the classification. It obviously
contains a lot of data which could be interpreted in
many ways.
39.	One area I had a problem with was the Cr
results in Table 5-7 which I note (compared to those
of Table 5-8) seem to have included the analyses of
many "ND" sample points. When the results of
Table 5-8 are examined they give the impression that
they could have satisfied the level 2 condition. Why
didn't Cr warrant a higher rating?
89

-------
The "ND" symbol indicates that the regression
parameters were "not determined" or calculated for
chromium because there were too few samples that
provided quantifiable chromium by FPXRF analysis
at the ASARCO site. This lack of regression
parameters did not impact the data quality level for
chromium. The chromium data was considered
Level 1 because the precision for chromium was
above 20 percent. This clarification has been made
in the text and in Tables 5-7, 5-8, and 5-9.
40.	Another case is Zn, which looks like a definite
level 3 based on the regression statistics. Which
statistic reduced it to level 2?
The inferential statistics, or paired t-test,
indicated that the zinc data from the TN 9000 and
from the reference laboratory were statistically
significantly different. Therefore, according to the
demonstration plan (PRC 1995), it was considered
Level 2. This clarification has been made in the
text.
41.	None of the tables include detailed results for
Sb and Cd yet their bottom line analysis appears in
Figure 5-7.
The tables only show the regression parameters
for the six primary analytes. whereas Figure 5-7
also includes results for the four analytes. A
discussion of the data quality level for the four
secondary analytes is presented at the end of Section
5. The text has been clarified to indicate that only
the six primary analytes are presented in the tables.
42. Comment during telephone conversation with
Mr. Peter Berry: Explain inconsistency in slope
values in tables.
The regression parameters found in the tables in
Sections 4 and 5 were determined using a statistical
program with the FPXRF data plotted as the
dependent variable (x-axis) and the reference data as
the independent variable (y-axis), thus giving the
reader an idea of a correction factor that may have
to be used when comparing FPXRF data generated
in the field to formal laboratory data. Since the data
was regressed in this fashion, a slope value less than
1.0 generally indicates that FPXRF was
overestimating concentrations as compared to the
reference method while a slope value greater than
1.0 generally indicates the FPXRF was
underestimating concentrations as compared to the
reference method. A footnote has been added to ail
the tables to explain this inconsistency.
90

-------
Section 8
References
Havlick, Larry L., and Ronald D. Crain. 1988. Practical Statistics for the Physical Sciences. American
Chemical Society. Washington, D.C.
Kane, J. S., S.A. Wilson, J. Lipinski, and L. Butler. 1993. "Leaching Procedures: A Brief Review of their
Varied Uses and their Application to Selected Standard Reference Materials." American Environmental
Laboratory. June. Pages 14-15.
Kleinbaum, D. G., and L. L. Kupper. 1978. "Applied Regression Analysis and Other Multivariable Methods."
Wadsworth Publishing Company, Inc., Belmont, California.
Morgan, Lewis, and Bockius. 1993. ROD Scan Database of EPA Superfund Records of Decision.
PRC Environmental Management, Inc. 1995. "Final Demonstration Plan for Field Portable X-ray Fluorescence
Analyzers."
U.S. Environmental Protection Agency. 1993. "Data Quality Objectives Process for Superfund-Interim Final
Guidance." Office of Solid Waste and Emergency Response. Washington, D.C. EPA/540/R-93/071.
91

-------