United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Las Vegas, NV 89193-3478
Research and Development
EPA/600/S4-88/011 August 1988
v>EPA Project Summary
Interlaboratory Evaluation of
SW-846 Methods 7470 and
7471 for the Determination of
Mercury in Environmental
Samples
Werner F. Beckert, J.E. Gebhart, J. D. Messman, and G. F. Wallace
The EPA protocols for SW-846
Methods 7470 and 7471 are cold-
vapor atomic absorption spectro-
metric (CV-AAS) methods for the
determination of mercury In aqueous
and solid environmental samples,
respectively. In continuation of a
previous single-laboratory study in
which a more sensitive mercury CV-
AAS method for environmental anal-
yses was evaluated, the revised CV-
AAS method has been subjected to
an interlaboratory study. The revised
CV-AAS system, operated in an
open configuration, incorporates a
dedicated gas sparging bottle for
reduction-aeration and an on-line
amalgamation/thermal desorptlon
step. With these modifications, the
CV-AAS method provides increased
sensitivity and also alleviates non-
specific background absorption in-
terferences so that instrumental
background correction is not re-
quired.
Silver-wool amalgamation cells,
mercury stock standard and spiking
solutions, a deionized water sample
with spiking instructions, a coal fly
ash reference material, and instruc-
tions for analysis by the amal-
gamation CV-AAS method were
sent to 18 participating laboratories
having prior experience with the
current EPA protocols for Methods
7470 and 7471. Ten of the 18 lab-
oratories were then invited to par-
ticipate In a more rigorous col-
laborative study. To evaluate the
revised Method 7470, three aqueous
sample types were analyzed: ground
water, waste water, and dilute nitric
acid. To evaluate the revised Method
7471, three solid sample types were
analyzed: marine sediment, incin-
erator fly ash, and municipal sewage
sludge. Some of the samples were
designated for spiking with inorganic
or organic mercury and also with
copper, a potential interferent
The analytical results reported by
the collaborating laboratories were
statistically examined in an attempt
to characterize the overall accuracy,
precision, and ruggedness of the
amalgamation CV-AAS method. In
general, the interlaboratory results
indicated that the amalgamation CV-
AAS method currently is not
sufficiently rugged for routine use
but, when properly implemented by
proficient laboratory personnel, may
serve as an alternative approach to
the recirculating CV-AAS method
described in the current EPA
protocols. To fully realize the ana-
lytical benefits of the amalgamation
CV-AAS method and to obtain
accurate and precise data, a
complete appreciation of the integral
factors for successful trace analyses
is essential, and a high level of
sophisticated operation and operator
skill is required.
-------
This Project Summary was
developed by EPA's Environmental
Monitoring Systems Laboratory, Las
Vegas, NV, to announce key findings
of the research project that is fully
documented in a separate report of
the same title (see Project Report
ordering information at back).
Introduction
Included in publication SW-846,
entitled "Test Methods for Evaluating
Solid Waste," by the Office of Solid
Waste and Emergency Response of the
U.S. Environmental Protection Agency
are two analytical protocols, Methods
7470 and 7471, for the determination of
mercury in aqueous and solid waste
samples by cold-vapor atomic absorp-
tion spectrometry (CV-AAS). In a
previous single-laboratory study, these
EPA protocols were evaluated and
revised to improve analytical perfor-
mance. The protocols were evaluated
using aqueous and solid environmental
samples of homogeneous and known
compositions in order to assess ac-
curacies and precisions of the methods
without introducing uncertainties due to
sample heterogeneities.
The results of the single-laboratory
study indicated, in general, that the
digestion procedures were satisfactory
for the samples analyzed; only minor
revisions that would improve, clarify, or
increase the flexibility of the digestion
procedures were recommended. The
recirculating CV-AAS method described
in the current EPA protocols was found
to be adequate only for mercury
determinations in samples of relatively
high mercury concentrations. Significant
modifications of the cold-vapor appa-
ratus were recommended and evaluated
to overcome its inadequacy for mercury
determinations in samples of low
mercury levels and to minimize inter-
ferences caused by non-specific
absorption of the primary mercury
radiation by organic vapors. Instrument
detectability was improved 10-fold by
the use of a gas sparging bottle as a
dedicated reduction-aeration vessel and
silver-wool amalgamation in the CV-
AAS system operated in an open
configuration. The on-line amal-
gamation/thermal desorption process of
the modified CV-AAS system also
effectively eliminated interferences from
water and organic matrix vapors so that
an instrument without dynamic bacl
ground correction capabilities could b
used. Good accuracy and precision wer
obtained with the amalgamation CV
AAS system for the analyses of foi
reference sediment materials.
The objective of this project was t
conduct an interlaboratory evaluation c
the revised EPA protocols for th
determination of mercury in environ
mental samples. The interlaborator
study was conducted in three phase;
Phase 1 - fabrication and testing of th
silver-wool amalgamation cells an
characterization of selected inter
laboratory study samples by the lea
laboratory; Phase II - preliminary eva
uation of the technical capabilities of th
participating laboratories; and Phase III
evaluation by the qualifying lab
oratories of the amalgamation CV-AA!
method for the determination of mercur
in a marine sediment reference materie
and in representative aqueous and solii
waste samples. The test samples con
tained concentrations of endogenou
mercury or were spiked with con
centrations of inorganic or organ!
mercury that were in the optimal range c
the amalgamation CV-AAS method /
Needle Valve
Charcoal
Trap
Glass Stopcock
\
Absorption
Cell
Tygon-to-Glass
Connections
Chrom-Alumel
Resistance Heating
Winding
Nitrogen
Cylinder
Flow Meter
Purging Cylinder
(Reduction-Aeration
Sample Cell)
Figure 1. Schematic diagram of the amalgamation CV-AAS system.
-------
statistical analysis of the mercury
concentration data submitted to Battelle
was conducted in an attempt to char-
acterize the interlaboratory accuracy,
precision and ruggedness of the amal-
gamation CV-AAS methods.
Experimental
Instrumentation
The schematic diagram of the
amalgamation CV-AAS system is
presented in Figure 1. Mercury vapor
evolved from the reduction-aeration cell
is trapped and concentrated by amal-
gamation on the silver-wool plug
positioned in the center of the
amalgamation cell. After the reduction-
aeration and the on-line amalgamation
steps are completed, mercury is de-
sorbed from the silver wool by resistance
heating to approximately 500°C. The
mercury vapor is transported via a
nitrogen-gas purge into the absorption
cell, and the maximum peak absorbance
is recorded. The appearance time of the
mercury peak on the strip-chart
recorder with a 1-second time constant
is approximately 17 seconds; the max-
imum peak absorbance occurs between
25 and 30 seconds after heating of the
silver-wool plug is initiated. The amal-
gamation/thermal desorption process
eliminates interfering water and organic
vapors prior to the mercury absorption
measurement without compromising the
instrument detection limit. Although
continuum-source background correc-
tion should also compensate for such
vapor interferences, the increased noise
level in most of the older atomic
absorption systems when operated in the
background-correction mode degrades
the detection limit. An additional benefit
of amalgamation CV-AAS is that the
method can be successfully implement-
ed even by using an atomic absorption
instrument that is not equipped with a
dynamic background-corrector acces-
sory.
Amalgamation Cells
To minimize experimental variability,
all mercury amalgamation cells used in
this study were manufactured, assem-
bled, and tested by Battelle staff. The
amalgamation cell, with an overall length
of 105 mm, consists of ST 12/5 male and
female ball joints annealed to opposite
ends of Pyrex glass tubing (5 mm i.d.). A
silver-wool plug (Fisher Scientific
Company, Fair Lawn, NJ) of 0.7 g, with a
mass uncertainty of 0.05 g, is inserted
into the Pyrex tube and then compacted
into a cylinder having approximate
dimensions of 5 mm diameter and 20
mm length.
A 92-cm length of 22-gauge
Chromel A wire (Fisher Scientific Com-
pany, Fair Lawn, NJ) is wrapped around
the Pyrex tubing, providing 30 windings
and a resistance of 22 ohms/ft. The 0.7-
g plug of silver wool quantitatively
amalgamates mercury vapor at a nitro-
gen carrier-gas flow rate of 0.55 L/min
with no apparent buildup of leak-
inducing backpressure.
Reagents and Standards
All reagents and standards specified
were reagent grade or better; the
deionized water was specified as ASTM
Type II water (ASTM D1193). All
glassware for sample digestions, sample
dilutions, and standard preparations must
be prewashed sequentially with an
aqueous detergent solution, mineral
acids, and Type II water.
The following samples were used in
this study:
• Coal fly ash (NBS-SRM 1633a) with
a certified mercury concentration of
0.16 ± 0.01 jig/g for a sample size
of at least 250 mg.
• Marine sediment (MESS-1 ,
National Research Council of
Canada) with a reference mercury
concentration of 0.171 ± 0.014 jig/g
for a sample size of at least 500 mg.
• Incinerator fly ash (crushed, sieved
and homogenized) with no
detectable mercury.
• Municipal sewage sludge (dried,
sieved and homogenized) with a
determined mercury concentration of
approximately I.4 \iglg.
• Ground water (filtered and acidified
with nitric acid) with no detectable
mercury.
• Waste water with no detectable
mercury.
• Deionized water (acidified with nitric
acid) with no detectable mercury.
Sample Preparation
The samples were prepared for
analysis by using the digestion
procedures specified in the revised
SW-846 Methods 7470 and 7471.
Liquid samples were digested with
sulfuric acid, nitric acid, permanganate
and persulfate at 95°C; solid samples
were digested either on a steam bath
with aqua regia and permanganate, or in
an autoclave with sulfuric acid, nitric acid
and permanganate. All sample digests
were diluted to calibrated volume in
100-mL volumetric flasks with deionized
water following the addition of the
sodium chloride-hydroxylamine hydro-
chloride reagent solution to reduce
manganese dioxide and excess
permanganate to the soluble divalent
manganese form. This procedure
permitted sampling of multiple aliquots
and further dilutions as needed. The
sample aliquot added to the reduction-
aeration vessel was diluted to 100 mL
with Type II water.
Results and Discussion
Phase / - Amalgamation Cell
Testing and Sample
Characterization
To minimize variability in the
instrumental responses, the silver-wool
amalgamation cells were fabricated and
individually tested at Battelle. The
parameter used to assess uniformity of
performance among the amalgamation
cells was the slope of their calibration
curves. The slopes of four-point
calibration curves for the first 8 cells
ranged from 0.0048 to 0.0065
absorbance units/ng of mercury, with an
average value of 0.0058 ± 0.0006
abs/ng. The average recovery of a 50-
ng mercury standard, using a calibration
curve with a slope of 0.0052 abs/ng, for
15 additional cells was 98 ± 6 percent,
with a recovery range of 89 to 108
percent. The overall testing results
indicate uniformity in performance among
the 23 amalgamation cells with respect to
amalgamation efficiency and thermal-
desorption characteristics. The
differences in instrument responses for
the different amalgamation cells are
expected to be small compared to the
overall measurement variabilities
between the participating laboratories.
Battelle staff analyzed NBS-SRM
1633a, with a certified mercury value of
0.16 ± 0.01 yg/g; the results of duplicate
0.2-g portions were 0.145 ng/g and
0.160 ng/g. The recoveries of inorganic
mercury predigestion spikes added to the
fly ash samples were 130 percent for
duplicate 10-ng mercury spikes, and
112 percent and 104 percent for
duplicate 50-ng mercury spikes. In
triplicate 0.2-g samples of incinerator
ash analyzed according to the revised
Method 7471, no mercury was detected.
The average mercury concentration
determined in 8 replicate 0.2-g portions
of the sewage sludge sample was 1.4
ug/g; the compiled results ranged from
approximately 1.2 ug/g to 1.4 ^g/g of
mercury. These results indicate that the
mercury content of the municipal sewage
sludge is homogeneous to approximately
± 0.1 mg/kg.
-------
The average recoveries of duplicate
90-ng and 180-ng Hg predigestion
spikes (as methyl mercuric chloride)
were 75 percent and 90 percent,
respectively, and the average recoveries
of duplicate 100-ng and 200-ng Hg
predigestion spikes (as inorganic
mercuric chloride) were 108 percent and
106 percent, respectively. The overall
recoveries of organic and inorganic
mercury added as predigestion spikes
are within ± 25 percent of full recovery.
No mercury was found in four
replicate 50-mL test aliquots of the
ground-water sample. The average
inorganic spike recoveries of four sample
replicates were 106 percent for 50-ng
mercury spikes and 105 percent for
100-ng mercury spikes.
Phase II - Preliminary
Laboratory Evaluation
Method evaluation materials were
sent to 19 laboratories participating in the
preliminary laboratory evaluation phase.
Each package contained: (1) a tested
silver-wool amalgamation cell, (2) in-
structions for the installation and use of
the amalgamation cell, (3) two samples
and appropriate evaluation standard so-
lutions, (4) instructions for the prep-
aration and analysis of the samples, and
(5) instructions for reporting the ana-
lytical results to Battelle. One lab-
oratory withdrew at the beginning of
Phase II after receiving the package,
leaving a total of 18 participating
laboratories. The two samples shipped to
the participating laboratories consisted of
deionized water for the evaluation of
revised Method 7470, and NBS-SRM
1633a (Coal Fly Ash) for the evaluation of
revised Method 7471. A spiking solution,
containing mercury at a concentration of
1.5 mg/L, was included for spiking the
deionized water sample. The participants
were instructed to add 50 nL of the 1.5-
mg/L spiking solution (equivalent to 75
ng Hg) to a 20-mL aliquot of the water
sample as a predigestion spike. The
laboratories were also provided a stock
solution containing mercury at a
concentration of 1000 mg/L to be used
for preparing intermediate and final
calibration standards.
For sample analyses, the partic-
ipants were instructed to: (1) construct a
4-point calibration curve comprised of
the absorbances for a reagent blank,
10-ng, 50-ng, and 100-ng Hg stan-
dards, (2) analyze the spiked water
sample according to the revised Method
7470, (3) analyze the coal fly ash sample
according to the revised Method 7471,
and (4) prepare and analyze one reagent
blank each for the revised Methods 7470
and 7471.
The slopes of the calibration curves
generated by the 18 participating
laboratories for the two revised methods
ranged from 0.0001 to 0.0056 ab-
sorbance units/nanogram. Ten labora-
tories reported slopes between 0.003 and
0.006 absorbance units/nanogram for at
least one of the revised methods. Battelle
researchers have found that this is a
typical range for the slope of the
calibration curve obtained by using the
amalgamation CV-AAS system ac-
cording to the specific instructions
described in revised Methods 7470 and
7471. Ten laboratories reported reagent
blank values less than 0.03 absorbance
unit. Battelle researchers have consis-
tently measured absorbance values
between 0.02 and 0.03 absorbance units
for reagent blanks analyzed according to
the revised Methods 7470 and 7471.
Based on a typical instrument calibration
slope of 0.005 absorbance units/
nanogram, reagent blank absorbances in
this range correspond to approximately 5
ng Hg. Reagent blanks for aqueous
mercury calibration standards, consisting
only of the stannous chloride reductant
and deionized water, generally produce
0.001 absorbance unit; this is negligible
relative to the absorbance values for the
reagent blanks carried through th<
digestions. Some of the apparent lov
reagent blank values reported by th<
laboratories may be suppressed in (host
cases in which the reported calibratioi
slopes were lower than the typical range
of values.
Two laboratories did not submi
results for the spiked deionized wate
sample, and five laboratories did no
report results for the coal fly ash sample
The results reported by one laboratory
were excluded as obvious outliers. Tht
other laboratories reported mercury
concentrations for the spiked wate
sample ranging from 0.00076 to 0.007^
mg/L, and for the coal fly ash sample
from 0.04 to 1.1 mg/kg. Thus, the
calculated mercury recoveries for the
water samples ranged from 20 percent (<
5-fold negative bias) to 200 percent (e
2-fold positive bias), and those for the
coal fly ash samples ranged from 2E
percent (a 4-fold negative bias) to 69C
percent (a 7-fold positive bias).
A summary of the statistics across al!
laboratories (except the obvious outlier]
for the mercury measurements on the
spiked deionized water and coal fly ash
samples is presented in Table 1. The
means and standard deviations, reporteC
as percent recoveries, were calculated for
all laboratories and also for only those
eventually selected for Phase III. After the
outlier results had been removed, the
remaining data were smoothly dispersed,
and no additional outliers were obvious.
The data reported by some of the
participating laboratories for these
preliminary sample analyses are in
appreciable error. Possible contributing
factors to the wide range of results,
especially for the coal fly ash sample,
include high absorbance values for the
reagent blanks resulting from sample
digestions, a propagation of errors in the
calculated data submitted to Battelle, and
the general lack of operator experience
with the technique of silver-wool amal-
Table 1. Statistical Summary of Mercury Measurements on Spiked Deionized Water and Coal Fly Ash
Percent Recoveries
Number of Laboratories
Spiked Deionized Water
Coal Fly Ash
All Labs3
Mean
Standard Deviation
Phase III Labs
Mean
Standard Deviation
99
53
110
57
230
200
230
210
a Recoveries for the obvious outlier laboratory were not used in the calculations of the means and the standard
deviations.
-------
gamation. Some laboratories measured
reagent blank values as high as 0.6
absorbance unit, which is equivalent to
approximately 120 ng of mercury based
on a calibration slope of 0.005
absorbance units/ng; such a reagent
blank value exceeds the linear range of
the amalgamation CV-AA8 method.
Battelle staff discussed the results with
many of the laboratories; recommen-
dations for improvements in controlling
mercury contamination of the reagent
blank and test sample, as well as
suggestions for implementation of the
amalgamation CV-AAS system, were
provided. In general, the high dispersion
of data between laboratories for both
samples indicates a lack of ruggedness
for the amalgamation CV-AAS method
in this testing phase of the study.
Phase ///- Interlaboratory
Evaluation
The results of the preliminary
sample analyses in Phase II were used
as a qualitative guideline but not as
absolute criteria for the selection of 10
laboratories for participation in Phase III.
Five laboratories obtained results
which were in close agreement with the
target mercury values for both samples.
Five other laboratories obtained less
satisfactory results but exhibited suf-
ficient interest and enthusiasm in the
silver-wool amalgamation technique;
these 5 laboratories were judged to be
capable participants in Phase III.
Two different sample sets, denoted
as Option A and B and which comprised
three solid and three aqueous samples,
were shipped to the participants in Phase
III, together with detailed instructions.
Some of the samples required addition of
a predigestion spike. Option A samples
were supplied in duplicate and Option B
samples in triplicate. The samples were
shipped unspiked to avoid potential
losses of trace mercury spikes added to
samples with a reducing matrix. Pre-
digestion spiking solutions containing
organic and inorganic mercury were sent
to the participants with spiking in-
structions. One solid sample in Option B
was also to be spiked with copper, a
potential interferent in CV-AAS analyses
for mercury.
The slopes of the calibration curves
reported by the 10 participating lab-
oratories ranged from 0.0010 to 0.0065
absorbance units/nanogram of mercury.
Six of the 10 laboratories reported slopes
within the range experienced by Battelle
staff (0.003 and 0.006 absorbance units/
nanogram). The absorbance values for
the reagent blanks ranged from -0.004
to 0.075 absorbance units. Eight of the
10 laboratories reported reagent blank
values less than 0.03 absorbance unit.
Based on a typical calibration slope of
0.005 absorbance unit/nanogram, a re-
agent blank value of 0.03 absorbance
unit corresponds to a maximum con-
centration of 6 ng Hg in the reagent
blank. In general, the 10 laboratories in
Phase III reported a lower dispersion and
a narrower range of results for calibration
and reagent blank data than the 18
laboratories in Phase II. The 10 Phase-
Ill laboratories represent a more con-
sistent range of personnel skills and
demonstrated competence in trace ana-
lysis. The lower dispersion in the data
may also be a direct result of the
acquired experience and familiarity with
the silver-wool amalgamation technique
by the 10 laboratories from their
participation in the Phase II study.
The statistical summaries of the
interlaboratory study results are
presented in Tables 2 and 3. The means
and standard deviations are reported in
concentration units for those environ-
mental samples originally having
nondetectable endogenous mercury
concentrations, and in percent recoveries
for those samples having detectable
target mercury concentrations. The per-
cent recoveries are calculated as 100
times the measured concentration
divided by the target concentration.
Outlier tests were performed on the
laboratory means for each option within
each sample type. One laboratory was
identified as an outlier for most samples,
and its results were excluded from
statistical treatment for all sample types.
The average mercury concentration
or percent recovery over all laboratories
in a given option is reported in the third
column of Tables 2 and 3. The fourth
column contains the pooled within-
laboratory standard deviations of the
mercury concentrations or percent
recoveries. The within-laboratory var-
iances are simply calculated as the
averages of the variances for the
individual laboratories. The between-
laboratory standard deviations of the
mercury concentrations or percent
recoveries are reported in the fifth
column. This statistic characterizes the
variabilities in the true average mercury
concentration or percent recovery values
across the laboratories. The between-
laboratory variances are calculated
according to the following formula:
212 2
S =-(S_-S )
B n x w
where
is the variance of the average percent
recoveries for an individual laboratory, n
is the number of sample results used to
calculate the average percent recovery
for each laboratory, and
w
is the within-laboratory variance. If
2
S_
x
is less than
then
w
B
is set to zero. The total standard
deviations are reported in the last
column. The total variance is simply the
sum of the within-laboratory and
between-laboratory variances.
The high dispersion of the
interlaboratory data precludes any
statistical treatment of the results for the
revised Methods 7470 and 7471 that is
more rigorous than the analyses
conducted and summarized in Tables 2
and 3, respectively. As shown in Table 2
for the revised Method 7470, the total
standard deviations for Options A and B
are similar for each of the 3 test samples.
However, as shown in Table 3 for the
revised Method 7471, neither option
demonstrates a consistently lower total
standard deviation for all samples. For
example, the total standard deviations of
the results in Option A are 4-fold lower
than those in Option B for the incinerator
fly ash and the marine reference
sediment samples. However, the total
standard deviation in Option B is
approximately 2-fold lower than that in
Option A for the spiked incinerator fly ash
sample. The total standard deviations of
the results in both options are similar for
the unspiked and the spiked municipal
sewage sludge sample. The relative
differences between the total standard
deviations in Options A and B by the
revised Method 7471 for the 5 samples
appear to be random.
With the exception of the spiked
waste-water sample in Table 2 and of
the spiked incinerator fly-ash sample in
Table 3, the grand means using the
-------
Table 2.
Statistical Summary of Mercury Measurements Using Revised Method 7470s
Sample
Samples Having Nondetectable Target Concentrations &
Ground Water
Ground Water
Samples Having Detectable Target Concentrations c
Waste Water, Spiked
Waste Water, Spiked
Dilute Nitric Acid, Spiked
Dilute Nitric Acid, Spiked
Option
A
B
A
B
A
B
Mean
4
6
140
150
82
82
Pooled
Standard
Deviation
1.8
1.5
17
15
27
31
Between
Standard
Deviation
1.4
2.1
45
39
27
17
Total
Standard
Deviation
2.3
2.6
48
42
38
36
a Results for an outlier laboratory in Option A were not used in the statistical analyses for any sample.
b Results are in absolute concentration units (ng).
c Results are in percent recovery.
Table 3.
Statistical Summary of Mercury Measurements Using Revised Method
Sample
Samples Having Nondetectable Target Concentrations &
Incinerator Fly Ash
Incinerator Fly Ash
Samples Having Detectable Target Concentrations c
Marine Sediment MESS- 7
Marine Sediment MESS- 7
Incinerator Fly Ash, Spiked
Incinerator Fly Ash, Spiked
Municipal Sewage Sludge
Municipal Sewage Sludge
Municipal Sewage Sludge, Spiked
Municipal Sewage Sludge, Spiked
Option
A
B
A
B
A
B
A
B
A
B
Mean
0.021
0.074
110
110
74
39
90
88
97
90
Pooled
Standard
Deviation
0.0090
0.021
6.5
48
8.6
15
14
20
26
12
Between
Standard
Deviation
0.0076
0.041
10
0
32
11
18
21
25
26
Total
Standard
Deviation
0.012
0.046
12
48
34
19
22
30
36
28
a Results for an outlier laboratory in Option A were not used in the statistical analyses for any sample.
b Results are in absolute concentration units (mg/kg).
c Results are in percent recovery.
revised Methods 7470 and 7471,
respectively, are generally within 25
percent of the target values. However,
the total standard deviations for all
sample types represent relative standard
deviations as high as 50 percent.
Because of such highly dispersed
results, in addition to the contradictory
results of the copper interference test,
the revised Methods 7470 and 7471
presently lack the necessary ruggedness
to be successfully used for routine
testing by analytical laboratories without
additional analyst training and demon-
stration of proficiency.
Conclusions
The results of this interlaboratory
study indicate that only a few of the
collaborating laboratories demonstrated
proficiency in incorporating the silver-
wool amalgamation apparatus into their
CV-AAS systems within a relatively
short learning period. These laboratories
reported calibration slopes in an
acceptable range based on Battelle's
experience with the amalgamation CV-
AAS method. Only a few of the
laboratories were able to perform
accurate and precise measurements on
both aqueous and solid environmental
samples.
A statistical treatment of the
interlaboratory results indicates that the
amalgamation CV-AAS method cur-
rently lacks sufficient ruggedness to be
successfully used by analytical lab-
oratories for routine testing but, when
properly implemented by proficient lab-
oratory personnel, it may serve as an
alternative to the recirculating CV-AAS
method described in the current EPA
protocols of Methods 7470 and 7471.
However, to fully realize the analytical
benefits of the amalgamation CV-A/
method and to obtain accurate ai
precise data, a complete appreciate
and awareness of the critical facto
affecting successful trace analyses a
essential, and a high level
sophisticated operation and operator si
is required. Moreover, because of tl
increased sensitivity of the amalgamate
CV-AAS method, improved analytic
techniques and the use of high-puri
reagents are required to minimize m<
cury contamination that would not be
problem when using the recirculati
CV-AAS method with samples of high
mercury concentrations.
To achieve better results, many
the laboratories reported that they had
modify the operating parameters of t
silver-wool amalgamation apparat
from those specifically described in t
revised protocols. This may have be
-------
necessary when the silver-wool
amalgamation apparatus was not as-
sembled and configured properly ac-
cording to the directions in the protocols,
or when the amalgamation apparatus had
to be reconfigured to fit a specific
commercial or customized CV-AAS
system. The different operating con-
ditions used by some laboratories may
have contributed to the high dispersion in
the amalgamation CV-AAS data
submitted to Battelle.
Most of the comments from the
collaborating laboratories addressed the
increased sensitivity and the longer
analysis time of the amalgamation CV-
AAS method. Many laboratories
recognized the need for higher operator
skill and the importance of the need for
high-purity reagents to minimize mer-
cury contamination. The benefit of an
amalgamation CV-AAS method with
increased sensitivity but at the expense
of longer analysis time was challenged
by some laboratories from a business
economics viewpoint; the longer analysis
time would increase sample analysis
costs to the laboratory that would have to
incur the additional costs or charge
higher analysis fees to the customer.
Although the faster recirculating CV-
AAS method was considered adequate
for the majority of their samples, other
laboratories expressed special needs for
the increased sensitivity of the amal-
gamation CV-AAS method and
indicated that it would be valuable in
research and for difficult analysis
problems. Some laboratories expressed
enthusiasm about the amalgamation
CV-AAS method and indicated that they
plan to incorporate this method into their
laboratory operations for additional
flexibility. The extent of time and cost
constraints for adapting the amal-
gamation CV-AAS method to a specific
sample workload may be influenced by
whether the emphasis of the analytical
laboratory is on commercial testing or on
research, development and specialty
problems in a support function of a larger
organization.
In conclusion, it is recommended
that the CV-AAS methods be used to
characterize waste samples for mercury
as described below:
(1) The current EPA protocols for the
recirculating CV-AAS method
should be used for analyses when
the threshold concentration value for
mercury is relatively high and when
the samples do not contain volatile
organic compounds that could cause
nonspecific absorption interferences.
(2) The amalgamation CV-AAS method
should be used for the analyses of
samples containing low mercury
concentrations which can not be
easily determined by the re-
circulating CV-AAS method.
(3) The amalgamation CV-AAS method
should also be used for the analyses
of samples containing volatile
organic compounds which will cause
nonspecific absorption interferences
and inaccurate results with the
recirculating CV-AAS method.
Because of the greater sensitivity
and the more sophisticated apparatus of
the amalgamation CV-AAS system, it is
emphasized that the laboratory analysts
be properly trained and that they
demonstrate proficiency with the
amalgamation method before initiating
regulatory analyses for mercury.
-------
The EPA author, Werner F. Beckert, (also the EPA Project Officer, see below)
is with the Environmental Monitoring Systems Laboratory, Las Vegas, NV.
Judy E. Gebhart, Jerry D. Messman, and Gordon F. Wallace are with the
Battelle Columbus Division, Columbus, OH 43201-2693.
The complete report, entitled "Interlaboratory Evaluation of SW-846 Methods
7470 and 7471 for the Determination of Mercury in Environmental
Samples," (Order No. PB 88-196 001/AS; Cost: $14.95, subject to
change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Las Vegas, NV 89193-3478
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S4-88/011
0000329 PS
y S EHVIR PROTECTION AGiNCY
REGION 5 LIBRARY
230 S OfAR8QRH STREET
CHICAGO IL 60604
------- |