American Journal of Anal yt ical Chemistry, 2011, 2, 752-756
doi:10.4236/ajac.2011.27086 Published Online November 2011 (
Copyright © 2011 SciRes. AJAC
A Method for the Measurement of Mercury in Human
Whole Blood
Alicia E. Stube, Helene H. Freiser, Charles R. Santerre*
Department of Nut r it i on Sci e n c e , Purdue University, West Lafayette, Indiana, USA
E-mail: *
Received July 26, 2011; revised September 1, 2011; accepted September 12, 2011
A method for measuring total mercury in human whole blood using Thermal Decomposition-Amalgamation/
Atomic Absorption Spectrophotometry (TDA/AAS) was developed and applied to a study of women that
were fish consumers. This method has a limit of detection of 0.33 μg/L. The blood mercury concentrations
measured ranged from 0.74 μg/L to 14.80 μg/L, with a mean of 3.36 μg/L. Accuracy was within 15% of the
expected value at the lower concentrations and within 10% at higher concentrations. Some 560 analysis were
completed in about three weeks and the mean error in precision was 1.8% when measured in duplicate. It
was concluded that this method is viable for use in clinical settings, with the benefit of small sample volumes
and minimal sample preparation.
Keywords: Mercury, TDA/AAS, Human Blood
1. Introduction
Mercury (Hg) status can be assessed by hair or blood
biomarkers [1]. Humans who consume certain types of
fish may have higher mercury concentrations in their
system [2]. Methyl mercury is readily absorbed from the
gut into the bloodstream, where it binds to red blood
cells [3]. Previous analytical methods for measuring
blood mercury tested only red blood cells; however
whole blood methods are now gaining ground [4]. Use of
whole blood reduces sample preparation time and elimi-
nates uncertainty relating to cell number. Centrifugation
and separation of red blood cells also present an oppor-
tunity for analyte loss and increases the overall uncer-
tainty in measurement. Reducing the time and steps in-
volved in sample preparation facilitates measurement of
mercury in clinical settings.
The traditional methods for measuring mercury are
cold-vapor atomic absorption spectroscopy (CVAAS)
and inductively-coupled plasma mass spectrometry
(ICP-MS) [5,6]. These methods are accepted as reliable
and sensitive; however, throughput of samples is slow,
with high equipment and reagent costs. The samples are
exposed to concentrated acids and extended heating,
which increase analyte loss [7]. Thermal Decomposition-
Amalgamation/Atomic Absorption Spectrophotometry
(TDA/AAS) is an emerging technique for rapidly testing
large numbers of samples [8]. Samples are introduced
directly to the instrument and analyzed in minutes with-
out the need for extensive preparation [9]. This high
throughput method allows large number of samples to be
analyzed accurately and rapidly with less analyte loss
than the conventional methods listed above. TDA/AAS
also contributes to the movement towards green chemis-
try, requiring no preparatory reagents or hazardous che-
micals. This technology has already been utilized for
analysis of avian blood samples with high mercury con-
centrations [4]; however, because of the low concentra-
tion of mercury in human blood there is currently no re-
liable TDA/AAS method. This paper proposes a method
for TDA/AAS analysis of whole blood that would be
usable for clinical applications with subjects within typi-
cal human blood mercury concentrations.
2. Materials and Methods
2.1. Instrumentation and Che micals
The Tricell DMA 80.3 Direct Mercury Analyzer (Mile-
stone, Inc., Shelton, CT) was utilized for mercury meas-
urements. Quartz sample boats of 1.5 ml volume were
obtained from the same vendor. The auto-sampler of the
DMA 80.3 instrument had forty slots that could be filled
with sample boats. A stock mercury solution (Accustan-
dard, New Haven, CT) of 0.1 μg/ml mercury in 5% nitric
acid was used to calibrate the instrument and act as a
running standard. SRM 966 (NIST, Gaithersburg, MD),
Toxic Elements in Bovine Blood, level 1, was used as
Standard Reference Material (SRM). A single human
blood sample from an individual low in mercury was
used to establish a baseline and for spiking. For verifica-
tion, four samples of blood from 70 subjects were util-
ized, for a total of 280 samples. Blood was drawn into
EDTA K3 tubes (Greiner BioOne, Monroe, NC). These
samples were stored frozen at –80˚C and then thawed
and subjected to vortex mixing for at least 30 seconds
immediately prior to analysis.
Approval for human subject research was obtained
from the Purdue University Institutional Review Board,
Research Project Number 0709005855. All subjects sign-
ed a consent form detailing the research procedures and
any possible risks that might be incurred as a result of
2.2. Calibration
Cells 0 and 1 of the DMA 80.3 were calibrated using di-
fferent volumes of the standard mercury solution in quar-
tz boats (i.e. at 0.05, 0.1, 0.2, 0.3, 0.5, 1, 2, 3, 5, 7 ng of
mercury) for Cell 0. That range was extended to 10 ng of
mercury for Cell 1. Cell 2 was not calibrated as its meas-
urement range was well above the expected mercury
levels in human blood. The linearity of the calibration
curve was evaluated using this dataset and a best fit
model was established using the DMA 80.3’s internal
software. An S-curve was used to fit the data, resulting in
a correlation equation of
0.0003 0.0015*
0.0734 0.0013
Hg Hg
 
and a coefficient of correlation, R2, of 1.00. An injection
of 20 μL of 100 µg/L standard was used to confirm the
calibration and was measured within 2% of the expected
value. The limit of detection with this calibration was
0.33 µg/L for a 150 μL blood sample with 0.0495 ng
mercury. Below 0.33 µg/L, measurements became im-
precise due to background noise and residual mercury on
the boats.
The calibration was confirmed daily at the start of
each run using a 20 μL injection of standard solution
containing 2 ng mercury. If the verification sample was
off by more than 5%, a second 2 ng of mercury sample
was run to establish a calibration factor to be applied to
all of that day’s measurements. If the verification was off
by more than 10%, a new calibration curve was estab-
lished. The standard solution was used in conjunction
with a 150 μL injection of SRM 966 level 1 to ensure
interday repeatability.
2.3. Instrument Parameters
Various parameters were tested using human blood, until
a method which provided acceptable recovery and preci-
sion was found. These parameters included changes to
sample volume and drying/decomposition conditions.
EPA method 7473 was used as a starting point, as it had
previously been validated as an effective and reliable
method to measure mercury in fish tissues [10]. A me-
thod provided by the manufacturer was also considered,
but found unsuitable for whole blood analysis. With lar-
ger number of samples, this method resulted in residue
accumulation in the catalyst tube and cells, wearing
down the mercury vapor lamp and ultimately resulting in
greater uncertainty between duplicate samples.
The optimal process conditions were found to be a
sample size of 150 μL, drying at 120˚C for 2 min 20 sec,
1 min ramp to 650˚C, and finally decomposition at
650˚C for 3 min 30 sec. This set of conditions was cho-
sen based on precision and reproducibility over a number
of sampling days, as well as completeness of combustion.
Sample volumes greater than 150 μL tended to boil over
in the oven and overwhelm the catalyst tube with an ac-
cumulation of ash. However, at smaller volumes, there
was not enough absolute mercury present in the sample
for reliable analysis.
Development of the drying/decomposition ramp was
also important to avoid boiling the sample out of the
boats too quickly. Too high an initial drying temperature
would result in boats bubbling over and loss of volume.
In these cases, a blackened crust was observed on the top
and sides of the boat. The ramp was used to slowly heat
the dried samples, avoiding volume loss. The decompo-
sition temperature recommended by the EPA method for
mercury analysis in wastewater [8] was too low to prop-
erly combust the complex matrix of whole blood and
would leave blackened residue in the boats after heating.
Raising the temperature to 650˚C on the DMA 80.3 re-
sulted in complete combustion.
The remaining instrument parameters for sample ana-
lysis included a purge time of 30 sec, the amalgam time
was set to 12 sec and the recording time measured 30 sec.
Before a quartz boat could be used again after a blood
analysis, it needed to be cleaned by going through one
cycle at a drying temperature of 300˚C for 1min and de-
composition at 650˚C for 3 min.
2.4. Precision and Accuracy Study
Two samples of human blood, one from a non-fisheater
and one from a regular fisheater, were used to test the
Copyright © 2011 SciRes. AJAC
feasibility of the method. Once a method with less than
10% spread between triplicate samples was identified,
spiked samples of the low mercury blood were used as to
calculate precision and accuracy. Quartz boats were util-
ized for blood samples and standards; a blank in the form
of an empty nickel boat was placed at the beginning and
end of every sample run, as well as between each indi-
vidual sample to ensure any residual mercury from the
previous sample was burned off.
The baseline low mercury human blood sample, 150 μL
for each concentration, was spiked with fixed amounts of
standard mercury solution at 2.2, 5, 10, 20 and 30 μL.
The series of standards was analyzed five times. For each
concentration, a predicted mercury concentration was
also calculated, taking into account the approximate mer-
cury content of the baseline blood. Measurements were
compared to calculated concentrations, and the percent
error was used as an estimate of accuracy. For precision
calculations, the mean and relative standard deviation for
the five samples was determined.
2.5. Verification Study
Human blood samples were used to check the reference
range and confirm that the method was useful for analysis
of a large number of human blood samples. Before and
after each day’s run of human blood samples, a 150 μL
sample of SRM 966 was analyzed, and the results were
used to formulate a control chart. Quality control limits
were set at two standard deviations from the mean. Runs
where the SRM value fell outside the three standard de-
viation control limits were removed and the samples done
that day re-analyzed after a recalibration. The samples for
the verification study were done in duplicate. If more than
10% error was observed between duplicate samples, the
samples were reanalyzed. Data from the control chart
were tracked to observe between day precision.
3. Results and Discussion
3.1. Assessment of Method
Accuracy and precision were measured using spiked b-
lood samples. Baseline blood was taken from a non-fish
eater and estimated to contain 0.122 μg/L mercury after
five sample runs. This blood was used as a baseline to
represent the low anchor of the range of expected human
blood mercury concentrations. Aliquots of the baseline
blood were spiked at five concentrations using a standard
mercury solution; the mean, standard deviation, and cal-
culated errors in precision and accuracy are shown in
Table 1.
At each concentration, the measured DMA 80.3 values
were slightly lower than the expected blood mercury
levels. Accuracy was within 15% of the expected value
at the lower concentrations and within 10% at higher
concentrations. When 2.2 μL of standard solution were
added (expected concentration 1.6 μg/L), 87.5% of the
mercury was recovered and measured by the instrument.
When the spiked amount was raised to 30 μL, 95.6% of
the mercury was recovered. Loss of mercury may also be
attributable to the volatility of the standard mercury solu-
tion; some of the spiked mercury may have been lost
before the sample could be introduced into the instru-
ment [7].
Error in precision was very low across all of the
spiked blood samples, consistently below 5%. This did
not appear to relate to concentration, as values were si-
milar at both high and low concentrations. Overall the
blood method showed excellent recovery in spiked sam-
ples as well as intra-day precision.
3.2. Method Validation via Analysis of Human
Blood Samples
A total of 280 different blood samples from 70 different
female subjects were used to evaluate the method’s fea-
sibility in clinical applications. These measurements are
shown in Figure 1.
Table 1. Precision and accuracy of spiked blood samples.
Known Hg
Measured Hg
Std Dev
Error in
Error in
1.60 1.40 0.036 12.0% 2.6%
3.45 3.08 0.047 10.7% 1.5%
6.79 6.18 0.220 8.9% 3.5%
13.45 13.00 0.454 3.7% 3.5%
20.12 19.25 1.234 4.3% 6.4%
Figure 1. Blood mercury concentrations measured by TDA
/AAS. Mean is represented as a solid line.
Copyright © 2011 SciRes. AJAC
These samples were run in duplicate and intra-sample
precision was measured. Blood mercury concentrations
ranged from 0.74 μg/L to 14.80 μg/L, with a mean of
3.36 μg/L and the 50th percentile at 2.76 μg/L. All sam-
ples were above the detection limit and all had errors in
precision of less than 10%; the mean error was 1.8%
when measured in duplicate.
The current EPA reference dose (0.1 μg/kg body wei-
ght-day) has been shown to provide a concentration in
human blood of 5.8 μg/L [11]. Recurring levels above
this value is believed to cause harm, in particular to the
developing fetus. Most of the blood samples collected in
this study registered lower than that threshold, as evident
from Figure 1, but some fish-eating subjects had unusu-
ally high values. For those individuals quick corrective
actions might be needed, highlighting the advantage of
the described method.
Each sample required about 10 minutes for analysis;
when analyzed in duplicate, about twenty-four minutes
was spent on the analysis of each sample. In total, the
time required to measure mercury concentration in these
280 samples of blood (560 analyses) was about three
weeks if samples were prepared and loaded during the 40
hour work week. If samples are prepared and loaded
during nights and weekends as well, sample throughput
can be increased further. Preparing the samples and fill-
ing the auto-sampler tray required less than an hour so
most of the time spend for each lot consisted of unat-
tended data collection. By comparison, it can take up to a
day of preparation work to analyze samples by CVAAS
or ICP-MS.
4. Conclusions
The method described here is a viable way to measure
total mercury concentration in whole blood. Mercury in
human blood can be measured reliably and accurately in
less than ten minutes per sample. This method can be
useful for large clinical studies with high volumes of
samples to analyze. Since only a 150 μL sample volume
is required, it may also not be necessary to draw venous
blood to measure an individual’s blood mercury levels. A
fingerstick with a deep puncture blade can draw up to 1
mL of blood, sufficient for analysis via this method. Fin-
gersticks are faster and less invasive than venous blood
draws. Collection of fingerstick samples makes meas-
urement of blood mercury more accessible and practical
for situations outside of the clinical setting. The ability to
utilize small sample volumes combined with a rapid turn-
around time makes measuring blood levels just as viable
as hair concentrations for efficient assessment of mercury
exposure. Use of this method for analysis of total mercury
in human whole blood provides a green alternative that is
simple, reliable and time efficient.
5. Acknowledgements
Financial support for this study was obtained from the
US Department of Agriculture, National Institute of
Food and Agriculture (No. 07-51110-03804). Collection
of blood samples was done by Doug Maish, EMT-P,
Purdue University.
6. References
[1] T. W. Clarkson and L. Magos, “The Toxicology of Mer-
cury and Its Chemical Compounds,” Critical Reviews in
Toxicology, Vol. 36, No. 8, 2006, pp. 609-662.
[2] E. Oken, J. S. Radesky, R. O. Wright, D. C. Bellinger, C.
J. Amarasiriwardena, K. P. Kleinman, H. Hu and M. W.
Gillman, “Maternal Fish Intake during Pregnancy, Blood
Mercury Levels, and Child Cognition at Age 3 Years in a
US Cohort,” American Journal of Epidemiology, Vol.
167, No. 10, 2008, pp. 1171-1191.
[3] L. W. Chang, “Toxicology of Metals,” Lewis Publishers,
Boca Raton, 1996, p. 1052.
[4] J. T. Ackerman, C. A. Eagles-Smith, J. Y. Takekawa, J.
D. Bluso and T. L. Adelsbach,” Mercury Concentrations
in Blood and Feathers of Prebreeding Foster’s Terns in
Relation to Space Use of San Francisco Bay, California,
Habitats,” Environmental Toxicology and Chemistry, Vol.
27, No. 4, 2008, pp. 897-908. doi:10.1897/07-230.1
[5] APHA. 3112, “Metals by Cold-Vapor Atomic Absorption
Spectrometry,” In: M. A. H. Franson, Ed., Standard
Methods for the Examination of Water and Wastewater,
Ame- rican Public Health Association, Washington D. C.,
Vol. A, 1998, pp. (3-22) - (3-24).
[6] APHA. 3120, “Inductively Coupled Plasma (ICP) Me-
thod,” In: M. A. H. Franson, Ed., American Public Health
Association, Washington D.C., Vol. B, 1998, pp. (3-38) -
[7] D. D. Afonso, Z. Arsian and A. J. Bednar, “Assessment
of Matrix-Dependent Analyte Stability and Volatility
during Open-Vessel Sample Dissolution for Arsenic, Ca-
dium, Mercury and Selenium,” Microchimica Acta, Vol.
167, No. 1-2, 2009, pp. 53-59.
[8] EPA, “Method 7473: Mercury in Solids and Solutions by
Thermal Decomposition, Amalgamation, and Atomic
Adsorption Spectroscopy,” Agency UEP, 1998.
[9] S. J. M. Butala, L. P. Scanlan and S. N. Chaudhuri, “A
Detailed Study of Thermal Decomposition, Amalgama-
tion/Atomic Absorption Spectrophotometry Methodology
for the Quantitative Analysis of Mercury in Fish and
Hair,” Journal of Food Protection, Vol. 69, 2006, pp.
Copyright © 2011 SciRes. AJAC
Copyright © 2011 SciRes. AJAC
[10] J. A. Lasrado, C. R. Santerre, S. M. Shim and J. R. Stahl,
“Analysis of Mercury in Sportfish Tissue by Thermal
Decomposition, Amalgamation/Atomic Absorption Spec-
trophotometry,” Journal of Food Protection, Vol. 68, 20
05, pp. 879-881.
[11] L. Trasande, P. J. Landrigan and C. Schechter, “Public
Health and Economic Consequences of Methyl Mercury
Toxicity to the developing Brain,” Environmental Health
Perspectives, Vol. 113, 2005, pp. 590-596.