CLIN.CHEM. 24/12, 2151-2154(1978)
Atomic AbsorptionSpectrophotometryof Chromiumin Serum and
Urinewith a ModifiedPerkin-Elmer603 Atomic Absorption
Spectrophotometer
Frederick J. Kayne,”2 G. Komar,2 H. Laboda,2 and Raymond E. Vanderlinde”2
Modification of a Perkin-Elmer 603 atomic absorption
spectrophotometer
by adding a high-intensity tungstenhalogen lamp forbackground correctionsignificantly
improved the detection limit
for elements that have analytical
The background correction procedure is intended to eliminate the nonspecifIc absorption caused by light scattering
from particulates, photon absorption not attributable
to atoms
wavelengths in the near-ultraviolet and visible regions.
Until the recent announcement
of the Perkin-Elmer
Model
5000, the background correction has been made in commer-
Chromium in human serum and urinecan be measured,
with a simplified sample-handling technique, in concentrations of less than 0.1 tg/liter. For comparison, the mean
value for chromium in the serum of eight men was 0.14
tg/liter.
AdditIonal Keyphrases:trace elements
.
normal values
(preliminary)
A recent
publication
showed
that
the
concentration
of
in sera from one small sampling of healthy humans
ranges from 0.1 to 0.2 ig/liter (1). These results, lower than
values previously reported,
were obtained by a careful neutron
activation analysis technique. A table presented by these
authors shows a clear trend in previously reported serum
chromium to follow the limits of analytical detectability
available to investigators involved with the various studies.
Our interest lies in measuring chromium in the serum and
urine of diabetic patients and controls, because of reports
during the past 20 years of a possible connection between
chromium and diabetes (2-4) and vascular diseases (5, 6) in
the human. This led us to attempt to improve atomic absorption methodology to levels of detectability that would be
satisfactory
for such analyses. We believed that flameless
atomic absorption spectrophotometry
would offer the best
approach because of its high sample-throughput
capability
and simple sample handling, which should decrease contamination with extraneous chromium.
Flameless atomic absorption spectrophotometry
is claimed
by various instrument manufacturers to have a sensitivity for
chromium of about 1-10 pg, corresponding to some 5-50 zl of
a solution with a chromium concentration
of 0.1-1.0 tg/liter.
Although the best models of commercial instruments apparently can measure in this range, this usually is true only for
aqueous samples, which have no complex matrixes. The major
difficulty with atomic absorption measurements has been the
necessity for accurate and reproducible background correction
during atomization, especially of serum or urine or the residual
matrix from their acid digestion.
chromium
Departments of Pathology and Laboratory Medicine’ and Biological Chemistry,2 Hahnemann Medical College and Hospital,
Philadelphia, Pa. 19102.
Received Aug. 20, 1978; accepted
Sept. 27, 1978.
of the element
being observed,
and optical-path
aberrations.
cially available instruments
by using the light from a deuterium arc lamp directed through the sample furnace. This light
is then alternately observed with the light from the hollowcathode source. In the past, the basic problem has been the
low photon flux rates available with the deuterium lamp at
the usual analytical wavelength for chromium, about 357-360
nm-a value in the near-ultraviolet.
Materials and Methods
Apparatus
We removed the deuterium lamp assembly from a PerkinModel 603 Atomic Absorption Spectrophotometer
and
substituted
a plane front surface mirror, to direct light
through the chopper in the usual background-correction
optical path. An air-cooled 100-W (12-V) tungsten-halogen
lamp
was mounted on the outside of the instrument, with the light
directed down through a shielded port cut in the mirror
housing and outside case. A Model 6324 lamp housing with
a blower (Oriel Corp., Stamford, Conn. 06902) is used to shield
and cool the lamp, as well as to facilitate alignment. An observation port on the side of the mirror box allows the lamp
image to be seen on the monochromator
entrance slit. The
resulting configuration yields very even illumination through
the furnace tube of sufficient intensity to balance fully the
photon signal produced by the hollow-cathode lamp, up to its
maximum operating current. An Oriel G-774-3300 filter in the
optical path removes stray light from this source. For stability
and versatility of control, the tungsten lamp is powered by a
Hewlett-Packard
6267 B power supply, operating in the
constant current mode. The operation of the instrument in
the background correction mode is exactly the same as the
unmodified version. To allow this, a shutter is placed in the
reference light path and a source of 60-120 V dc is supplied
to the deuterium lamp terminals to energize the background
correction circuit through the optically coupled switch that
is part of the instrument circuit. For convenience, the deuterium supply can simply be left connected and used to switch
on the circuit. With these simple modifications, the ability to
correct for nonspecific
background
is dramatically
improved.
For sample atomization we used a Perkin-Elmer HGA 2200
Graphite Furnace Unit equipped with the temperature
ramping accessory and the optical temperature
sensor.
Elmer
CLINICALCHEMISTRY,
Vol. 24,
No. 12, 1978
2151
Table 1. Chromium Analysis Conditions
Dry: lOOs,RAMP8Os; 110#{176}C
Char: 60 S,NO RAMP; 1100 #{176}C
C
A
_______________
.064
.064
.059
-----
2.5 ug IL
,,rr
.055
.052
053
“-:
______
::-H
.047
.046
Purified nitric acid was obtained from the National Bureau
of Standards (prepared by sub-boiling distillation)
and the
J. T. Baker Chemical Co. (Ultrex grade) and 30% H2O2 from
Fisher Scientific Co. K2Cr2O7 was a product of Fisher, and
both the solid AR-grade reagent and 1000 mg/liter Certified
Atomic Absorption Standard were used. The water used for
all procedures and washing was first de-ionized, then distilled
in glass, and finally passed through a “Nanopure 3” ion exchange and ultrafiltration
system (Barnstead
Co., Boston,
Mass. 02132). Any chromium in the water supply was below
our level of detection. Solutions were kept in polyethylene or
polystyrene containers [washed in HNO3 according to the
procedure of Moody and Lindstrom (7) when necessary] and
checked for contamination
by chromium.
Factors
Using first the unmodified and then the modified atomic
absorption instrument, we have been able to derive conditions
for analysis of our chromium-containing
samples that apparently are optimized for the general case. (We have assumed
nothing about the type or number of different chromiumcontaining compounds in our biological samples.)
We then empirically
selected conditions
(Table 1) for
sample introduction,
drying, charring, gas flow, and atomization that are as near as possible to optimal for all the sorts
of samples tested, including the standard K2Cr2O7 used separately and for standard additions.
Analytical results were computed from standard curves or
individual standard measurements,
after correction for the
appropriate
dilutions and blank absorptions.
Atomic absorption signals were recorded with a Perkin-Elmer
Model 56
recorder, the peak heights detected by the recorder were
measured and compared with the digital absorbance readout
on the instrument
itself, and the concentrations
were calculated.
Apparent chromium content varies widely with different
sample-preparation
techniques
(8, 9). Most of these techniques involve preliminary ashing or digestion, followed by
programmed atomization in a graphite furnace similar to ours.
This apparent variability led us to adopt procedures utilizing
oxidation in strong nitric acid solution, aided by peroxide. The
lowest blank values have been obtained when digestions were
done in fused-silica 16 X 100 mm test tubes fitted with
threaded Teflon caps (Scientific Glass and Instrument
Co.,
Houston, Tex. 77001).
Samples of biological materials for chromium analysis were
prepared by two different procedures. For solid-tissue samples, each sample was weighed directly into the fused-silica
tube, 0.5 ml of nitric acid and 0.5 ml of peroxide were then
added, the tube tightly capped and incubated overnight at 80
CLINICAL CHEMISTRY,
Fig. 1. Sample recorder tracings from Perk In-Elmer Model 603
atomic absorption spectrophotometer before and after modif ication
Conditionsas described in text. Numbers represent meter readout values, with
chromium concentrations given for each set. A, before modification; B, after
modification; C, reproducibility at low chromium concentrations
#{176}C
(we used a Fisher aluminum-block
incubator).
For samples
of serum or urine, 0.5 ml was placed into the tubes described
previously.
After overnight incubation,
the tubes were removed and allowed to cool to room temperature. The contents
of the tubes were then diluted to a final volume of 2.5 ml with
de-ionized distilled water, and 20-id portions were removed
from the solution with Eppendorf automatic pipettes and
injected into the graphite furnace. Replicate samples were run
and the peak height measurements
obtained
were .compared
with those for standards.
Standard
chromium
solutions were
added to either water, the acid-peroxide
matrix, or sampleacid-peroxide
mixtures
before they were incubated
at 80 #{176}C
overnight.
Alternatively, the method of standard additions was used
and appropriate
samples of standard inorganic chromium
were added to samples of biological materials that had already
been oxidized and diluted for the atomic absorption measurement. Although most analytical procedures for chromium
reported previously have involved separate ashing as a main
step, we find it unnecessary.
We attempted
to keep our
method as simple as possible, not only to increase sample
throughput, but also to minimize the possibility of introducing
extraneous sources of contamination
into the sample. We
believe the following results of analysis for chromium in several reference materials show the method to be satisfactory.
Results
Figure 1, showing recorder traces, illustrates
instrument
Sample Preparation
Vol. 24, No. 12, 1978
0.32uglL
.047
2.5 iaglL
Reagents
2152
O.39ug/L
.053
.045
current: 20 mA
Analytical
-.
B
Wavelength; 357.9 nm
Slit: 0.7nm
Lamp
Q46u9 IL
.058
Atomize: 8 5, NO RAMP; 2500 #{176}C
Argon gas flow: 40 mI/mm
Flow time: 3 s, STOP
Temperature sensor: ON
Maximum power: ON
performance.
Especially
apparent
the improved
is the reduction
in signal intensity seen after the atomization peak while the
graphite tube is still at atomization
temperature.
Unfortunately, there is no reliable means of quantitating
this improvement. Although instrument manufacturers
recommend
use of background absorbance values at which the sample and
background
beams remain balanced, this is a static measurement, probably not representative of the dynamic process
of atomization such as takes place when the instrument
is
actually in use. By a static criterion (neutral-density
filters)
the background correction of our modified instrument is at
least 1.4 A at 357.9 nm, approximately
the value given by
Perkin-Elmer
Corp. for background-correction
performance
in the ultraviolet region.
Although
this modification
does not involve any of the
signal-handling
processes of the instrument,
the improved
atomization
conditions
and operation
of the hollow-cathode
CORR. . 9g72
SLOPE: .1106
INTCP.
Table 2. Chromium Concentration In Reference
Materials
-0003
7
z
ua Cr/a (±SD
0
Labeled
Material
Found
value
Kodak Gelatin Reference Material 41 (±1)
47 (±2)
0.
0
‘/)
no.TEG-50-C
w
Kodak Gelatin Reference
I-
no.lEG-SO-B
NBS SRM 1577 BovineLiver
-J
w
CHROMIUM,
Fig. 2. Determination
14g/L
Least-squares regression line is paralleled by a ± I SD line. Bars represent the
range of three or four replicate analyses
CORR.
SLOPE
.996
1025
INTCP
.0139
z
0
I-.
0.
44
(± 1)
47 (±1)
0.060
(±0.006)
0.088
NBS SRM 1659 Brewer’s Yeast
1.85
2.12
NBS Cooperative Round Robin
Brewer’sYeastno. 19
(±0.29)
1.90
(±0.28)
(±0.05)
not given
(±0.012)
of chromium detection limits (aqueous
solutions) under conditions described in text
7
Material
S
0
U,
w
>
I-
lamp near its optimum
current range (20 mA instead of 10
mA) makes it possible for us to detect chromium
in a concentration
of about 55 ng/liter,
which is 1.1 pg/20-tl sample.
This sensitivity
is obtainable
both for samples in aqueous
solution and those in the highly acid matrix used for oxidation
of samples, as shown in Figures 2 and 3. (Also, the relatively
low chromium
contamination
of water and HNO3 is best illustrated
by these figures.) A maximum
value of 0.14 ag/liter
can be estimated
from extrapolation
of the determinations
made in the 8 mol/liter acid matrix. The modification
reported
here has thus increased analytical
detectability
of chromium
with our system between five- and 10-fold and has fully removed the uncorrected
black-body
radiation response caused
by graphite-tube
glow (Figure 1).
Table 2 presents our results for chromium in specific reference materials. These materials were oxidized by the procedure described above and the data on chromium are based
-J
w
CHROMIUM.
ug/L
Fig. 3. Determination of chromium detection limits in HNO3 (8
mol/liter) as described in Fig. 2
The negative x-intercept represents 0.14g of the chromitmi per liter introduced
with the reagents. “Relative Absorption”
is 100 X A
12
0
w
>9
0
U
I
z
3
B
C
0
F
E
6
r
2
4
5
8
1.0
12
1.4
CHROMIUM.
-1rirl
1.6
1.8 2Q
2.2
‘22
ug/L
Fig. 4. Histogram describing the distribution of 66
randomly
on results obtained
with inorganic standards.
Figure 4 shows the distribution
of data on chromium
in
some unselected
urine samples. These data were obtained by
directly injecting the sample into the graphite furnace. The
range of chromium concentrations
is considerably
below those
previously reported. Most of our values are near 0.9 pg/liter;
a few exceed 2 zg/liter, but most of these are less then 10
pig/liter. Some of these samples also were measured after being
oxidized with nitric acid and peroxide
as described
above.
These values, listed in the figure legend, are expressed
as the
ratio of the chromium
concentration
in the urine sample
measured
after oxidation
to the concentration
in the same
sample analyzed
directly. All values for chromium
concentrations
reported
are referenced
to Fisher Inorganic
Chromium AAS Standard analyzed in the acid matrix by direct
addition to the graphite furnace.
Our data indicate that our method allows routine determination of as little as 0.1 .ig of chromium
per liter in samples
of biological fluids. The principal
reason for this sensitivity
is our improved
analytical
instrumentation
as well as the
sample
preparation
procedure,
which routinely
results
in
blank values of less than 0.1 gfliter
chromium.
We believe
that our approach
represents
a great improvement
in rapid
measurement
of the trace amounts
of chromium
present in
biological
matrices.
This is especially
significant
when one
realizes that the results are comparable
with those obtained
by neutron
activation
analysis,
as seen with the reference
materials reported in Table 2 and as described in the following
selected subject urine samples analyzed
section.
Chromium concentrations determined by direct analysIs as descilbed In the text
For comparison, values for chromium concentration determined on oxidized
samples divided by the value as measureddirectly are, for each group: A, 2.9
± 0.9; B, 3.6,4.5; C, 3.7 ± 0.9; 0,2.5 ± 0.5; E, 2.6 ± 0.7; F, 2.1 ± 0.4; 0,2.0
± 0.6; J, 2.4; K, 0.9 ± 0.2
Serum Chromium
We measured
chromium
chosen healthy volunteers
in serum from eight casually
by the preparation
and analysis
CLINICAL CHEMISTRY,
Vol. 24, No. 12, 1978
2153
procedures
just described.
The mean value determined
was
0.14 gig/liter, in excellent
agreement
with the value of 0.16
igfliter
recently reported
from a neutron activation
analysis
study (1). Although
our mean value is near the lower level of
detectability,
this agreement
seems to validate it as actually
representing
the chromium
content
of normal serum. The
approximately
10-fold difference
in values for chromium
concentration
between serum and urine analyzed
the same
way (HNO3-H202
oxidation)
also validates
the lack of extraneous
chromium
contamination
down to the lower levels
of chromium
concentration
reported
here.
We have not yet evaluated
the new Perkin-Elmer
Model
5000 instrument,
but it may be capable of efficiently
making
such background-corrected
measurements
as reported
here.
In any event, the modifications
reported
here are simple
enough for a competent
laboratory
to make and the costs are
far less than an investment
in new instrumentation.
However,
some of the enhanced
performance
in the system reported
here over that previously
reported
in the literature
may be
traced additionally
to the use of the HGA 2200 furnace with
the optical temperature
accessory.
This combination
allows
fast attainment
of atomization
temperature
and an increased
sensitivity
in general for chromium
analysis.
Our ability to
detect chromium
concentrations
in biological fluids down to
50 ng/liter should be of interest to some analytical laboratories
and we hope it will lead to answers to the many questions
concerning
the role of chromium
in human nutrition
and the
development
of certain pathophysiological
states.
This work was supported
2154
in part by grant AM 20830 from the NIA
CLINICAL CF#{128}MISTRY,
Vol. 24, No. 12, 1978
MDD, NIH,
USPHS,
and training
grant
5-T01-GM02198
from
NIGMS, NIH, USPHS.
References
1. Versieck, J., Hoste, J., Barbier,
mium and cobalt in human
serum
F., et a!., Determination
of chroby neutron activation analysis. Clin.
Chem. 24, 303 (1978).
2. Mertz, W., Chromium occurrence and function in biological systems, Physiol. Rev. 49, 136 (1969).
3. Doisy, R. J., Minerals and trace elements-chromium.
In Diabetes,
Obesity and Vascular Disease. Metabolic
and Molecular
Interrelationships,
H. M. Katzen and R. J. Mahler, Eds., John Wiley and Sons,
New York, N.Y., 1977, pp 596-604.
4. Underwood,
E. J., Chromium. In Trace Elements
in Human and
Animal Nutrition.
Academic Press, New York, N.Y., 1977, pp
258-270.
5. Newman, H. A. I., Leighton, R. F., Lanese, R. R., and Freedland,
N. A., Serum chromium and angiographically
determined
coronary
artery disease. Clin. Chem. 24,541(1978).
6. Punsar, S., Wolf, W., Mertz, W., and Karvonen, M. J., Urinary
chromium
excretion and
atherosclerotic
manifestations
in
two Finnish
male populations. Ann. Clin. Res. 9, 79 (1977).
7. Moody, J. R., and Lindstrom,
R. M., Selection and cleaning of
plastic containers for storage of trace element samples. Anal. Chem.
49, 2264 (1977).
8. Grafflage, B., Buttgereit, G., Kuebler, W., and Mertens, H. M., Die
Messung Spurenelemente
Chrom und Mangan im Serum mittels
flammenloser Atomabsorption. J. Clin. Chem. Clin. Biochern. 12,287
(1974).
9. Wolf, W. R., and Greene, F. E., Preparation of biological materials
for chromium analysis. In Accuracy in Trace Analysis: Sampling,
Sample Handling, Analysis,
I, P. D. La Fleur, Ed., U.S. Govt.
Printing Office, Washington, D.C., 1976, pp 605-610.