United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S4-89/015 Sept. 1989
&EPA Project Summary
USEPA Method Study 35:
SW-846 Method 3005:
Acid Digestion of Waters for
Total Recoverable or Dissolved
Metals for Analyses by Flame
Atomic Absorption
Spectroscopy
Kenneth W. Edgell
An interlaboratory collaborative
study was conducted to determine
the precision and bias (recovery) of
Solid Waste (SW-846) Method 3005 for
total recoverable metals by flame
atomic absorption on twenty-one ele-
ments in ground water. Method 3005
is entitled "Acid Digestion of Total
Recoverable or Dissolved Metals For
Analyses By Flame Atomic Ab-
sorption Spectroscopy or Inductively
Coupled Plasma Spectroscopy" and
includes instructions for quality con-
trol, sample preparation and analysis
of samples by AA-Flame.
The study design was based upon
Youden's non-replicate plan for col-
laborative tests of analytical meth-
ods. Each water type was spiked with
six concentrations (as three Youden
pairs) of the twenty-one test ele-
ments and was digested using a
nitric/hydrochloric acid procedure
and analyzed by flame atomic ab-
sorption Spectroscopy. Test data
from three spiked ground water
sources were compared against re-
agent water as a control. The result-
ing data were analyzed using
USEPA's computer programs entitled
"interlaboratory Method Validation
Study" (IMVS). This study produced,
for each element, measures of
precision and mean recovery for the
acid digestion/flame atomic absorp-
tion Spectroscopy and compared the
performance of the method between
each water type and reagent water.
This study was conducted by The
Bionetics Corporation at the direction
of the Environmental Protection
Agency, Quality Assurance Research
Division, Environmental Monitoring
Systems Laboratory, Cincinnati, Ohio
under EPA Contract No. 68-03-3254.
This report covers a period from Sep-
tember 10, 1986 to December 21,
1987. Analytical work was completed
in September 1987.
This Project Summary was devel-
oped by EPA's Environmental Monitor-
Ing Systems Laboratory, Cincinnati,
OH, to announce key findings of the
research project that is fully docu-
mented in a separate report of the
same title (see project ordering in-
formation following this document).
Introduction
The Hazardous Waste Management
facility permit regulations were promul-
gated in July 1982 (40 CFR 265} and pro-
vide performance standards for the mon-
itoring of ground waters at hazardous
waste sites. To facilitate these standards,
certain analytical methodology will *be
-------
required to assess the degree of ground
water contamination at and around the
area of the site. Test Methods for Evalu-
ating Solid Waste, Physical/Chemical
Methods, (SW-846), November 1986,
Third Edition, is intended to provide a
unified, up-to-date source of information
on sampling and analyses related to
compliance with Resource Conservation
and Recovery Act (RCRA) regulations.
The success of these pollution control ac-
tivities, particularly when legal action is
involved, depends upon the reliability of
the data provided by the -laboratories.
Therefore, it is necessary to determine
the bias (recovery) and precision of the
methodology in interlaboratory method
validation studies,
EMSL-Cincinnati of the USEPA devel-
ops/selects analytical methods and pro-
vides quality assurance (QA) support to
the Office of Solid Waste (SW) as re-
quired by regulations. The QA program is
designed to maximize the reliability and
legal defensibility of water quality infor-
mation collected by the Agency, the pri-
mary regulating authorities in the states,
and by the private sector and commercial
laboratories performing compliance
analyses. The responsibility for providing
QA support is assigned to the QA Branch
of EMSL-Cincinnati. One QA activity is to
conduct interlaboratory method validation
studies to obtain precision and recovery
statements for the Agency's operating
program such as for the Office of Solid
Waste.
This report describes an interlabora-
tory method validation study on Method
3005 entitled "Acid Digestion of Waters
for Total Recoverable or Dissolved Met-
. als for Analyses by Flame Atomic Ab-
sorption Spectroscopy or Inductively
Coupled Plasma Spectroscopy". In this
study 21 trace metals were investigated
using the total recoverable digestion pro-
cedure with flame atomic absorption
Spectroscopy. Nine commercial labora-
tories were selected by the Quality As-
surance Branch of EMSL-Cincinnati to
participate in this study based on tech-
nical, criteria. The Bionetics Corporation,
as primary contractor to the Quality As-
surance Branch of EMSL-Cincinnati, was
responsible for the collection and char-
acterization of three ground waters for
use as test waters in the study and the
subsequent spiking with the analytes. Ad-
ditional Bionetics activities included anal-
yses of the samples to confirm the true
concentrations, preparation of user in-
structions and report forms, preparation
and distribution of the sample, screening
of returned data for gross errors, and
drafting of the final report. The raw data
were evaluated statistically by the Quality
Assurance Research Division of EMSL-
Cincinnati using a series of computer
programs entitled "Interlaboratory Meth-
od Validation Studies" (IMVS).
Procedure
Method Summary
The method consists of acidifying a
100 mL sample with 2 mL of concen-
trated nitric acid and 5 rnL of concen-
trated hydrochloric acid, which is then
heated to about 90°C in a hood until the
volume has been reduced to 15-20 mL.
The sample is then filtered to remove in-
soluble materials. Sample volume is ad-
justed to 100 mL and analyzed for metals
by AA-direct aspiration. The method is
applicable to the following elements:
Aluminum
Antimony
Arsenic
Barium
Beryllium
Cadmium
Calcium
Chromium
Cobalt
Copper
Iron
Lead
Magnesium
Manganese
Molybdenum
Nickel
Potassium
S&lenium
Silver
Sodium
Thallium
Vanadium
Zinc
Arsenic and selenium were not tested
in this study because these two elements
are not recommended for analyses by
flame atomic absorption Spectroscopy.
Either furnace technique atomic absorp-
tion or ICP analysis are recommended for
arsenic and selenium analyses.
Selection of Participants
Twenty-four commercial laboratories
responded to an abstract in the Com-
merce Business Daily for participants in
the method validation study. Technical
proposals were evaluated based upon
laboratory experience and quality control
within the laboratory and those labora-
tories whose proposals were acceptable
were further evaluated based on their
performance on preaward trace metal
samples. The final participants selected
for the formal study were the top nine
laboratories of the preaward performance
evaluation study.
Selection and Collection of
Water Samples
The types of wafer selected for the
IMVS included reagent water as the con-
trol and three unfiltered ground waters
from monitoring wells at different haz-
ardous waste sites: a) Chem-Dyne,
Hamilton, Ohio, b) Wayne Disposal,
Detroit, Michigan, and c) Coshocton Sank
tary Landfill, Coshocton, Ohio. It shoi
be noted that the unfiltered Wayrrs
ground water contained significant
amounts of calcium, chromium, nickel,
and cobalt in the suspended solid frac-
tion. Suspended solids concentrations in
the unfiltered Chem-Dyne, Coshocton
and Wayne waters were 142 mg/L, 69
mg/L, and 168 mg/L, respectively. Only
the unfiltered waters were used in this
study.
Description of Interlaboratory
Method Validation Study
The design of the study is based upon
Youden's original non-replicate design for
collaborative evaluation of analytical
methods. According to this design,, two
samples are prepared In pairs such that
the analyte concentrations of the pairs
vary between 5-20% of the mean of the
pairs. Youden pairs are prepared at three
different concentration levels, for each
water, to cover the optimum range of the
method specified in the SW-846 methods
manual, A copy of the method and the
instructions for sample preparation, with
data report forms, were supplied to par-
ticipants by USEPA. Each participant was
required to analyze a quality control sam-
ple after every sixth sample and deta
mine acceptable results within contt^
limits. If out of control, the problem was
to be corrected and the last six samples
rerun. The primary objective of the study
was to establish mathematical relation-
ships which express the precision and
mean recovery of the returned data as a
function of mean recovery and true con-
centration, respectively. (See Table 1.)
Results and Discussion
The primary objective of this study
was to characterize the performance of
Method 3005 in terms of mean recovery,
overall precision, and single-analyst pre-
cision on each of 21 trace metals in four
different water matrices.
Rejection of Outliers
The analytical data, processed through
the IMVS. programs, underwent three out-
lier tests. First, the Youden's Laboratory
Ranking Procedure was used to detect
and reject data having a large systematic
error associated with a particular labora-
tory. If the majority of a laboratory's data
for a particular element-water combina-
tion was either biased high or biased low,
compared to the other laboratories, tha
laboratory failed the lab ranking proc"
dure, and all of their data were rejected
-------
H
Table 1. Study 35, SW-846 Method 3005,
(in mg/L)
Water Type Aluminum,
Applicable Cone. Range
Reagent Water 1
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 1
Single-Analyst Precision
Overall Precis/on
Mean Recovery
Applicable Cone. Range
Reagent Water 2
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 2
Single-Analyst Precision
Overall Precision
Mean Recovery
Applicable Cone. Range
Reagent Water 3
Single-Analyst Precision
Overall Precision
]Mean Recovery
Ground Water 3
Single-Analyst Precision
Overall Precision
Mean Recovery
X = Mesn Recovery
C = True Value for the
, Trace Metals, AA Flame Weighted Linear Regression Equations for Mean Recovery and Precision
Al Barium, Ba Beryllium, Be Calcium, Ca Cadmium, Cd
(5.00 - 50.00)
Sfl
S
- 0.056X
= O.OS2X
-0.18
-0.13
X = 0.599C + 0.77
Sfl
S
= 0.030X
= 0.068X
X = 0.978C •
Sfl
S
X
SR
S '
X
(5.00 - 50.
= 0.030X
= 0.073X
= 0.955C
= 0.039X
+ 0.03
- 0.05
* 0.78
00)
+ 0.00
-0.05
- 0.12
- 0.14
= 0.085X -*• 0.02
= 0.94 7 C
- 0.16
(5.00 - 50.00)
SR
S •
X
SR
S
X
= 0.046X
= 0.090X
= 7.007C
= 0.027X
= 0.092X
= 0.998C
+ 0.03
+ 0.04
- 0.32
+ 0.11
- 0.17
- 0.25
(1.05 - 20.90)
SR = 0.050X
S = 0.1 21X
X = 0.967C
SR = 0.085X
S = 0.777X
X = 0.645C
(7.05 -27,
* 0.02
-0.07
-0.02
+ 0.07
-0.00
+ 0.42
00)
SR = 0.077X - 0.02
S = 0.074X
X = 0.926C
+ 0.06
-0.06
SR = 0.044X - 0.07
S = 0.062X
X =0.9370
(1.05 -21
SR - 0.034X
S = 0.119X
X = 7.078C
+ 0.02
- 0,04
.00)
•* 0.09
-0.03
- 0.72
SR = 0.746X -0,06
S = 0.274X
X =0.498C
+ 0.06
+ 0.50
(0.04 - 2.00)
SR = 0.038X + 0.00
S = 0.066X + 0.00
X = 0.956C - 0.00
SR = 0.079X + 0.00
S = 0.045X + 0.00
X = 0.967C + 0.00
(0.04 - 2.00)
SR = 0.073X + 0.00
S = 0.026X * 0.07
X = 0.966C - 0.07
Sfl = 0.009X * 0.00
S = 0.031 X + 0.00
X =Q.961C •* 0-00
(0.04 - 2.00)
SR = 0.072X * 0.00
S = 0.032X * 0.0 7
X = 0.986C - 0.00
Sfl = 0.074X * 0.00
S = 0.037X * 0.07
X =0.984C - 0.00
(90.00 - 300.00)
SR
S
)( ss
SR
S
X
= 0.022X
= O.T70X
0.905C •*
= 0.002X
= 0.772X
= 0.998C
(24.00 - 75
Sfl
S
x -
SR
S
X
SR
S
X
SR
S
X
= O.J86X
= 0. 755X
= 0.95 7 C
= 0.029X
= 0.045X
= 0.997C
(6.00 - 27
* 7.96
-2.02
1 72.72
+ 6.26
-5.47
-0.43
.00)
- 3.89
- 7.83
+ 0.28
+ 0.01
+ 1.86
+ 0.93
.00)
= 0.743X - 0.88
= 0.282X
= 0.878C
= 0.049X
= 0.368X
= 0.866C
- J.47
+ 0.43
+ 0.23
-0.04
-0.80
SR
S
X
Sfl
s
X
Sfl
s
X
Sfl
s
X
Sfl
s
X
Sfl
s
X
(0.04 - 7,86)
= -0.003X + 0.02
= 0.027X * 0.07
= 7.007C * 0.00
= 0.079X * 0.00
= 0.027X * 0.07
= 0.986C + 0.00
(0.04 -1.86)
= 0.013X + 0.01
= 0.015X + 0.01
= 0.995C - 0.00
= 0.016X * 0.00
= 0.027X •* 0.07
= 7.007C -0,00
(0.04 - 7.86)
= 0.074X + 0.00
= 0.038X * 0.00
= 0.997C - 0.00
= 0.072X * 0,00
= 0.057X -* 0.00
= 0.978C - 0.00
Concentration
for that element-water combination. Sec-
ond, zero, negative, and non-detected
data were rejected as unusable data.
Finally, the Thompson Outlier Test was
used to reject individual outliers.
For the entire project, the IMVS pro-
gram rejected 1403 data points (20.6%)
of the 6804 data points submitted. The
percentage of rejected data did not vary
significantly among water types. The low-
est rejection was for reagent water 1, and
the highest rejection was for ground
water 3. Molybdenum had the lowest
number of rejected data points (24) and
copper had the highest (102). Of the nine
laboratories participating in the study
three accounted for 786 or 56% of all re-
jected data. The other six laboratories
ranged from 74 to 130 rejected data
points.
Mean Recovery
The mean recovery, X, of the retained
sata, was compared to the true values
supplied to the IMVS programs for each
element at each spike concentration in
each test water. These individual values
are presented in Appendix C, in the main
report, as 732 separate values. Sub-
jecting these values for each water type
to linear regression analysis yielded re-
gression equations where mean recovery,
X, was related to the true value, C, over
the entire concentration range. The slope,
m, of these regression equations can be
used to estimate the percent recovery of
each element as long as the intercept, b,
is small (i.e., less than 5% of the value
slope times true concentration, mC).
Examination of the regression equations
presented in Table 1 indicates that this
criterion can be met for elements in all
water types except calcium, sodium,
magnesium, potassium, iron and
chromium. For these elements, percent
recovery estimates were calculated by in-
serting a midpoint concentration, from the
range studied, for the value C. The X
value obtained divided by the midpoint
concentration, C, times 100, will yield
percent recovery. Only two elements had
recoveries outside of the 90%-110% in-
terval across water types: barium (85%)
and silver (65%).
A closer look at the barium recovery
estimates revealed large recovery differ-
ences for ground water 1 and ground
water 3. These matrices appeared to
have adversely affected barium recovery,
possibly due to elevated sulfate levels,
Reagent water recovery for barium was
98%.
Using all Youden pairs, the regression
equation for mean recovery of silver in-
dicated an average recovery of 65%
across all water types. The poor recovery
data are attributed to the high level
spikes (approximately 4 mg/L) which ex-
ceed the solubility of the acid mixture
used in the method. The mean recovery
of silver at the high concentration level
was 53% across all water types. The
-------
V
Table 1. (Continued)
Water Type
Applicable Cone. Range
Reagent Water 1
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 7
Single-Analyst Precision
Overall Precision
Mean Recovery
Applicable Cone. Range
Reagent Water 2
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 2
Single-Analyst Precision
Overall Precision
Mean Recovery
Applicable Cone. Range
Reagent Water 3
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 3
Single-Analyst Precision
Overall Precision
Mean Recovery
X = Mean Recovery
C = True Value for the
Cobalt, Co
(0.96 - 6.00)
SR = 0.003X + 0.04
S = 0.027X + 0,02
X = 7.077C -0.00
Sfl = 0.076X + 0.05
S = 0.067X - 0.03
X = 7.084C - 0.06
(0.36 - 6.00}
SR = 0.023X + 0.01
S = 0.053X - 0.07
X = 7.077C - 0.00
SR = 0.01 IX + 0.07
S = 0.045X - 0.07
X = 0.988C * 0.07
(72,00 - 60.00}
SR = 0,02 5X + 0.79
S = 0.056X + 0.77
X = 7.040C -0.73
SR = 0.032X + 0.79
S = 0.056X + 0.07
X =7,062C - 7.06
Concentration
Chromium,Cr
(0.44 -10.50)
SR = 0.029X + 0.02
S = 0.087X + 0.07
X = 7.064C + 0.07
SR = 0.045X + 0.07
S = 0.095X + 0.00
X = 7.067C - 0.02
(0.44 -70.50}
SR = 0.070X + 0.00
S = 0.040X + 0.02
X = 7.065C - 0.07
SR = 0.027X + 0.07
S = 0. 736X + 0.03
X = 7.057C + 0.07
(73.70 -43.80}
SR = 0.078X + 0.30
S = 0.004X + 7.27
X = 7.038C + 7.37
SR = 0.020X - 0.78
S = 0.073X + 2.66
X =7.f07C - 0.07
Copper, Cu
(0.15 -5.00)
SR = 0.004X + 0.07
S = 0.028X + 0.00
X = 0.974C + 0.07
SR = 0.079X + 0.00
S = 0.026X + 0.01
X = 0.970C + 0.07
(0.75 -5.00}
SR = 0.070X + 0.00
S = 0.076X + 0.00
X = 0.982C - 0.00
SR = 0.072X + 0.00
S = 0.023X + 0.07
X =0.969C - 0.00
(0.75 -5.00}
SR = 0.004X + 0,07
S = 0.023X + 0.00
X = 0.983C + 0.07
SR = 0.072X + 0.07
S = 0.037X + 0.07
X =0.992C + 0.00
Iron, Fe
(2.70 - 77.30}
Sfl = 0.028X + 0.04
S = 0.057X - 0.05
X = 0.958C - 0.03
SR = 0.033X - 0.07
S = 0.083X-0.75
X = 0.992C * 0.27
(6.75 - 22.50}
Sfl = 0.77X + 0.77
S = 0.044X + 0.73
X = 0.949C + 0.65
SR = 0.035X + 0.07
S = 0.770X -0.28
X = 0.964C - 0.03
(2.70 - 77.30}
SR = 0.073X + 0.08
S = 0.065X - 0.04
X = 0.935C + 0.03
Sfl = 0.048X - 0.06
S = 0.043X + 0.79
X =0.969C - 0.37
Potassium, K -
(7.00 - 34.50)
SR = 0.049X -0.16
S = 0.1 03X -0.24
X = 7.060C - 0.56
SR = 0.778X-0.73
S = 0.732X-0.09
X = 7.097C - 0.79
(0.33 - 74.20}
SR = 0.060X + 0.07
S = 0.069X + 0.06
X = 7.042C - 0.07
SR = 0.057 X + 0.04
S = 0.095X - 0.00
X =0.987C - 0.07
(0.33 - 74.20;
SR = 0.040X + 0.03
S = 0.073X + 0.08
X = 7.738C -0.08
i
SR = 0.049X + 0.06
S = 0.071X + 0.05
X = 1.189C -0.05
mean silver recovery for the low and mid-
dle Youden pairs across all water types
was 85%. It has been shown that silver is
soluble at a concentration of 0.05 mg/L in
a hydrochloric acid/nitric acid mixture but
forms a significant precipitate at 0.5
mg/L. The optimum concentration range
specified in SW-846 Method 3005,
however, is 0.1 to 4.0 mg/L. Further
investigation is needed concerning the
acid mixture and optimum range of the
method.
Precision
To compare the overall and single-
analyst precision and the percent relative
standard deviations (%RSD), the mean
recovery was calculated by using as the
true concentration (C) a mid-point value
in the concentration range studied. This
calculated mean recovery was inserted
into the overall and single-analyst stan-
dard deviation regression equations given
in Table 1. The overall %RSD for a single
water type ranged from 2.0% for copper
to 40.3% for silver. The average overall
%RSD for all water types ranged from
2.7% for copper to 33.0% for silver. The
average %RSD for all elements/water
type was 8.5%.
The single-analyst (SA) %RSD for a
single water type ranged from 0.1% to
36.0% for barium and silver, respectively.
For all water types and elements, the
average SA %RSD ranged from 2.9% to
5.1%. The average SA %RSD or all
elements/water types was 4.3%, which is
almost one-half of the overall %RSD.
Effects of Water Type
The recovery and precision estimates
across water types were subjected to an
analysis of variance test to determine the
effect water types had on the results. Es-
tablishment of a statistically significant
effect due to matrix type does not neces-
sarily mean that the effect was of prac-
tical importance. Practical importance
was determined by reviewing the retained
data and judging whether the statistically
significant matrix effects were influenced
by the retention of several errant data
points or the non-uniform recovery of one
Youden concentration pair. If no
anomalies were observed, the statistically
significant matrix effect was considered
to be of practical importance.
Fifteen elements had 18 occurrences
that implied a statistically significant ef-
fect by one of the matrix waters. Thirteen
of the eighteen occurrences show small
differences in the percent recovery and
%RSO values compared to their reagent
water counterparts. Only four elements
(barium, cobalt, iron, and lead) showed
practical differences in the percent r
covery and %RSD values compared to
-------
r*ft/e 1. (Continued)
,*amr Type
Applicable Cone. Range
Reagent Water 1
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 1
Single-Analyst Precision
Overall Precision
Mean Recovery
Applicable Cone. Range
Reagent Water 2
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 2
Single-Analyst Precision
Overall Precision
Mean Recovery
Applicable Cone. Range
Reagent Water 3
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 3
Single-Analyst Precision
Overall Precision
Mean Recovery
X = Mean Recovery
C = True Value for the
Magnesium, Mg
Manganese, Mn
(30.00 - 125.00)
SR = O.OOOX
S = 0.052X
X = 0.960C
SR = O.OT8X
S = O.T47X
X = 0.902C
+ 1.51
+ 1.03
+ 1.61
+ 0.62
- 1.45
+ 7.22
SR
S -
X ;
SR
S '
X -
(4.50 - 30.00)
Sfl = O.OT9X
S = 0.030X
X = 0.855C
Sfl = O.OT7X
S = 0.050X
X = 0.859C
* 0.04
-0.72
* 2.29
* 0.11
-0.04
+ 1.87
SR
S
X
Sfl
S
X
(5.00 - 30.00)
Sfl = 0.022X
S = 0.070X
X = 0,9570
Sfl = 0.053X
S = O.T03X
X =0.9230
Concentration
+ 0.01
-0.22
•*• 0.29
* 0.24
-0.04
* 0.23
Sfl
S
X
Sfl
S
X
(0.05 -3.35)
= 0.008X + O.OT
= 0.052X * 0.02
= 0.983C * 0.00
= 0.030X * 0.00
= 0.054X * 0.00
= T.002C * 0.00
(7.T2-3.90)
= 0.087X - 0.08
= 0.083X - 0.05
= T.OT8C - 0.07
= 0.077X - 0.03
= 0.064X - 0.00
= 0.992C - O.OT
(0.08 - 3.35)
= 0.006X * 0.02
= 0.060X •*• 0.02
= 1.010C * 0.01
= 0.015X * O.OT
= 0.05TX * 0.07
= T.OTOC -O.OT
Molybdenum, Md
(0.75 - 37.50)
Sfl
S
X
Sfl
S
X
- 0.055X
= 0.223X
= 0.95 1C
= 0.053X
= 0.792X
= 0.9670
+ 0.04
•*• 0.05
-0.02
* 0.06
+ 0.03
•* 0.04
(0.75 - 37.50)
Sfl
S
X
Sfl
S
X
= 0.032X
= O.T38X
= 7.009C
= 0.036X
= 0.760X
= t.004C
* O.OT
-0.02
- 0.03
•*• 0.02
-0.05
-»• 0.06
(0.75 - 37.50)
Sfl
S
X
Sfl
S
X
= 0.030X
= O.T04X
= T.T46C
= 0.026X
= 0.7TSX
= 7.7300
* 0.04
•*• 0.03
- 0.08
* 0.03
-0,03
* 0.00
Sodium, Na
(27.50- 110.00)
SR = 0.070X - 2.03
S = 0.042X - 0. TO
X = 0.9600 - 0.85
Sfl = 0.075X •* 0.90
S = 0.025X - T.45
X = 0.955C - 0.63
(77.00 - 47.30)
Sfl = O.OOTX •*• 0.34
S = 0.045X -0.76
X = T.007C - 0.82
SR = 0.079X - 0.01
S = 0.045X + 0.06
X = 0.947C + 0.47
(550.00 - 2750.00)
Sfl = 0.055X - 8.96
S = 0.042X + 0.61
X = 0.982C + 77.94
Sfl = 0.020X + 09.76
S = O.T25X -0.47
X =7.0570 - T3.96
Nickel, Ni
(0.25 - 5.50)
SR
' S
X
Sfl
S
X
= 0.022X
- 0.025X
= 1.002C
= 0.026X
= 0.024X
= 0.9930
* 0.07
-0.07
-0.00
+ 0.00
* 0.05
-O.OT
(0.25 -5.50)
SR
S
X
Sfl
S
X
SR
S
X
= 0.013X
= 0.050X
= 7. OTTO
= 0.072X
= 0.036X
= 7.0080
(4.5 - 27.
= 0.077X
= 0.043X
= 0.992C
* 0.03
+ 0.03
-0.02
•*• 0.04
* 0.06
-0.04
50)
+ 0.11
+ 0.00
+ 0.01
Sfl = 0.034X - 0.03
S
X
= 0.707X
= 0.9750
+ 0.34
+ 0.66
their reagent water and were considered
to be of practical importance.
Minimum Detection Limits
(MDLs)
The average MDL for each water
type/ element, in almost all cases, agreed
closely, indicating virtually no effect of
water type on the magnitude of the
MDLs. A comparison of these MDLs with
the MDLs listed in SW-846 for ground
waters showed close agreement for bari-
um, beryllium, magnesium, vanadium,
cadmium, chromium, manganese, nickel,
molybdenum, and zinc. Elements with
higher MDLs than listed SW-846 MDLs
were aluminum, calcium, iron, sodium,
antimony, and thallium. Elements with
lower MDLs than the SW-846 MDLs were
potassium, cobalt, copper and lead.
These lower detection limits determined
f this IMVS study include zero re-
reported by the participating lab-
oratories. Therefore, the average stan-
dard deviation was lower, which subse-
quently reduced the MDLs as calculated
in the IMDL equation.
Evaluation of Laboratory
Performance with Low
Concentration Samples
The final objective of the study was to
evaluate the performance of the par-
ticipating laboratories' performance with
samples containing trace metal at con-
centrations slightly above the MDLs re-
ported in SW-846 methods. Therefore,
the three ground waters and reagent
water were spiked with Youden pair con-
centration levels approximating 5x MDL.
These were compared with the lowest
Youden pair concentrations near the low-
est part of the optimum range reported in
the SW-846 methods; this was approxi-
mately 10x MDL.
The overall %RSD and single-analyst
%RSD were used to evaluate per-
formance. Analytes that could not be
evaluated due to high background con-
centration levels in the natural matrices
included calcium, iron, potassium, mag-
nesium, manganese, and sodium. Ground
water 3 data was also excluded for nickel,
copper, cobalt, chromium, and zinc for
the same reason.
The overall and single-analyst %RSDs
were computed by the IMVS program for
each water type. The overall %RSD was
averaged between each of the Youden
pairs and reported for each water type,
with the single-analyst %RSDs for each
analyte. To facilitate a general conclusion
for all water types at a 5x MDL concen-
tration level versus the lower concentra-
tion level of the optimum range, the
overall %RSDs were combined, and an
average was calculated. The single-
analyst %RSDs were similarly treated.
The ratio of the %RSDs (i.e., %RSD
for 5x MDL divided by the %RSD for the
lower concentration level) provided a
-------
Table 1. (Continued)
Water Type
Lead, Pb
Antimony, $t>
Vanadium, V
Zinc, Zn
Applicable Cone. Range
Reagent Water 1
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 1
Single-Analyst Precision
Overall Precision
Mean Recovery
Applicable Cone, Range
Reagent Water 2
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 2
Single-Analyst Precision
Overall Precision
Mean Recovery
Applicable Cone, Range
Reagent Water 3
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 3
Single-Analyst Precision
Overall Precision
Mean Recovery
(0.75 - 18.80)
SR = 0.040X + 0,07
S = 0.046X + 0.05
X = 1.037C + 0.00
SR = 0.030X + 0,07
S = Q.054X - 0.01
X - 0.949C + 0.12
(0.75 - 18.80)
SR = 0.025X + 0.04
S = 0.043X + 0.02
X = 1.024C + 0.02
SR = 0.027X - 0.01
S = 0.038X + 0.05
X = 7.0T0C * 0.04
(0.75 - 18.80)
SR = 0.013X + 0.07
S = 0.022X + 0.73
X = 1.034C + 0.01
SR = 0.04SX + 0.03
S = 0.072X - 0.01
X = 0.811C + 0.77
(0.75 -37.50)
SR = 0.012X + 0.04
S = 0.042X + 0.19
X = 1.039C + 0.76
SR = Q.032X + 0.03
S = 0.045X * 0.74
X = T.058C + 0.09
(0.75 - 37.50)
SR = 0.037X + 0.03
S = 0.046X * 0.06
X = 1.049C + 0.05
SR = 0.005X + 0.77
S = 0.049X * 0.75
X =1.022C - 0.03
(0.75-37.50)
SR = 0.022X + 0.08
S = 0.030X + 0.06
X = t.OSOC * 0.02
SR = 0.008X + 0.14
S = 0.031 X + 0.13
X =1.031C -0.02
(1.88 - 93.80)
SR = 0.047X + 0.04
S = 0.121X - 0.09
X - 1.051C -0.16
SR = 0.024X •*• 0.08
S = O.J23X * 0.77
X = 7.092C - 0.00
(1.88 - 93.80)
SR = 0.035X + 0.06
S = Q.062X + 0.11
X = 1.045C - 0.04
SR = 0.040X •*• 0.02
S = 0.107X -0.03
X =7.050C -0.12
(1.88 - 93.80)
SR = 0.063X - 0.07
S = 0.104X -0.17
X = 1.091C -0.08
SR = 0.020X + 0.12
S = 0.115X -0.04
X =1.091C + 0.10
(0.05 - 1.13)
SR = 0.023X + 0.00
S « 0.018X + 0.01
X = 0.979C + 0.00
Sfl - 0.020X + 0.00
S =• 0.042X + 0.01
X = 0.975C + 0.00
(0.05- 1.13)
SR = 0.002X + 0.01
S = 0.013X + 0.01
X = 0.985C - 0.00
SR = 0.039X + 0,00
S = 0.053X + 0.07
X = 0.950C + 0.07
(0.22 -1.13)
SR = 0.004X + 0.01
S = 0.013X + 0,07
X = 0.962C + O.OT
SR ~ 0.005X + 0.01
S = 0.007X + 0.01
X =0.966C + 0.01
X
c
- Mean Recovery
— True Value for the Concentration
measure of the magnitude of the increase
in the %RSD when operating on a routine
basis at 5x MDL. For most trace metals,
the overall %RSD increased, anywhere
from 3.8 times to 1.1; antimony showed
no increase, and slight decreases in the
overall %RSD were observed for lead
and cadmium. The largest overall %RSD
at the 5x MDL level occurred for
aluminum; this can be explained in part
by the fact that the concentration was
only one-tenth of the lower limit of the
optimum range, and significant quantities
of indigenous aluminum were present in
each of the three natural matrices.
The smallest overall %RSD at the 5x
MDL level occurred for copper. The
average overall %RSD for all analytes in
the lower concentration level was 15.3%
whereas the 5x MDL was 22.1%.
The ratio of the single-analyst %RSD
varied from 0,7 for lead to 3.8 for
aluminum. Generally, the single-analyst
%RSD increased with decreasing con-
centration, but lead, silver and cadmium
showed slight decreases with decreasing
concentration. On the average, for all
analytes, the lower concentration had
8.8% RSD, and the 5x MDL samples had
10,8% RSD.
Conclusions and
Recommendations
Method 3005 is recommended for the
analysis of all specified elements except
silver in ground water matrices. The
linear regression equations obtained from
this study can be used to predict the pre-
cision and mean recovery of the covered
elements at any concentration in the
ranges investigated in this study.
Except for silver and barium, the mean
recovery across the three water types
studied ranged from 91% to 108% with
two exceptions. Barium showed ex-
tremely low mean recoveries from ground
water 1 (67%) and ground water 3 (50%).
Barium recovery from reagent water was
98%.
For silver, the mean recovery of each
of the three Youden pairs decreased sig-
nificantly with increased concentration in
the range studied (0.1-4.0 mg/L). The
digestion procedure in SW-846 Method
3005 employs hydrochloric acid, and pre-
cipitation occurred at the high con-
centration levels. Further studies should
be conducted to determine the optimum
silver concentration range for this method
or the appropriate acid mixture to be
used in the digestion procedure.
The. overall standard deviation ex-
pressed as %RSD, averaged across the
waters studied, ranged from 2% to 19%.
The only exception was silver with a
%RSD of 34%.
The single-analyst standard deviation
expressed as the previously defined SA
%RSD, averaged across all water types,
ranged from 1% to 7%. Silver was aoain
the only exception at 22% withj
%RSD ranging from 14% to 36%.
-------
Table 1. (Continued)
Water Type
Applicable Cone. Range
Reagent Water 4
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 1
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 2
Single-Analyst Precision
Overall Precision
Mean Recovery
Ground Water 3
Single-Analyst Precision
Overall Precision
Mean Recovery
X = Mean Recovery
C = True Value for the
Silver, Ag
(0.09 -4.13)
SR
S
= 0.21 3X •
- 0.418X -
X = 0.578C *
SR
S
= 0.360X
= 0.415X -
X = 0.534C +
SR
S
= O.J67X
= 0.247X -
X = 0.696C -f
SR
S
X
= 0.1 43X
= 0.279X -
= 0.7290 +
• 0.00
0.02
0.04
-0.00
0.00
0.03
-0.01
0.01
• 0.02
-0.00
0.01
0.02
SR
S
X
SR
S
X
SR
$
X
SR
S
X
Thallium,
Ti
(0.94 - 20.70)
- 0.038X
= 0.054X
= 0.945C
= 0.0 20X
= 0.063X
= 0.987C
= 0.031 X
= 0.087X
= 0.960C
= 0.009X
= 0.041 X
= /.oooc
+ 0.01
•* 0.08
* 0.04
•*• 0.07
* 0,0?
* 0.07
+ 0.07
- 0.01
+ 0.09
+ 0.12
+ 0.07
+ 0.07
Concentration
tatistical comparisons were per-
on all analytes to determine ef-
fects of water type. Statistically signifi-
cant effects were found for barium,
beryllium, cobalt, chromium, iron, potas-
sium, magnesium, manganese, molyb-
denum, sodium, nickel, lead, antimony,
thallium, and zinc. The effect was not,
however, considered of practical im-
portance for any analytes except barium,
cobalt, iron, and lead (See Treatment of
Data section in the main report for
details.)
Interlaboratory MDLs were determined
for 21 trace elements. Comparison with
MDLs specified in SW-846 showed close
agreement for barium, beryllium, mag-
nesium, vanadium, cadmium, chromium,
manganese, nickel, molybdenum, silver,
and zinc. Elements exhibiting greater
MDLs than those in SW-846 methods
were aluminum, calcium, iron, sodium,
antimony and thallium. Elements ex-
hibiting MDLs less than those in SW-846
methods were potassium, cobalt, copper,
and lead.
The SW-846 methods specify opti-
mum concentration ranges for trace met-
al analyses which usually begin at 10x
MDLs. A pair of" spiked samples were in-
rl*ded in this study to obtain overall and
e-analyst precision data at approxi-
mately 5x MDL. The percent overall
%RSD of the 10x MDL samples averaged
16.0% for ail elements; this agreed well
with 22.7% for the 5x MDL level. The
single-analyst %RSD for the tOx MDL
samples averaged 9.1% for all elements;
this agreed well with 11.2% for the 5x
MDL level.
-------
Kenneth W. Edgetl is with The Bionetics Corporation, Cincinnati, OH 45246
Edward L Berg is the EPA Project Officer (see below).
The complete report, entitled "USEPA Method Study 35: SW-846 Method 3005:
Acid Digestion of Waters for Total Recoverable or Dissolved Metals for
Analyses by Flame Atomic Absorption Spectroscopy," (Order No. PB 89-
190 573/AS; Cost: $28.95, subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
United States Center for Environmental Research
Environmental Protection Information
Agency Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S4-89/015
------- |