and Walter. Slavin for reviewing this ... Gardiner. PE, Stoeppler. M, Nurnberg. HW. Optimisation of the analytical conditions ... 4 Gardiner. PE, Stoeppler.
CLIN.
CHEM.
40/3,
431-434
(1994)
#{149} Automation
and Analytical
Techniques
Aluminum Determined in Plasma and Urine by Atomic Absorption a Transversely Heated Graphite Atomizer Furnace Cohn
Bradley
and
Fred
Ying
trace e!ements/metals/toxico!ogy
Since our initial application of the stabilized-temperature platform furnace (STPF) method for determining aluminum in serum and urine by atomic absorption spectroscopy (AAS) (1), several critical reviews (2-6) have been published, with suggested improvements in methodology.2 For the previous studies, a heated graphite atomizer (HGA) furnace of the Massmann type (7) was used, which produces nonisothermal temperature distribution along the tube. With the development of the side-heated constant-temperature graphite furnace, a uniform temperature can be obtained between the ends and the center of the tube (8). The formation of analyte atoms is thus more uniform, because atom condensation at the cooler tube ends, as occurs with an HGA furnace, is reduced.
Such
developments
should
prevent
“memo-
ry” effects,
which would be more noticeable as carryover after determinations of high concentrations of analytes. We compared the determination of aluminum in plasma and urine in two AAS systems, an original STPF method with the end-heated graphite furnace (HGA model) and a side-heated graphite tube with an integrated omizer
L’vov platform [transversely (THGA) model]. Matrix
with
Leung’
We compared a stabilized-temperature L’vov platform furnace containing an end-heated graphite atomizer (HGA) and transverse Zeeman background-correction system with a side-heated furnace system (transversely heated graphite atomizer; THGA) containing a longitudinal Zeeman background-correction system for the determination of aluminum in plasma and urine. The regression statistics for the correlation analysis of the two systems (slope coefficient = 0.995, intercept = -1.710, S,,, = 0.021 gIL) indicate that the systems generate comparable results. The newer technology of the THGA furnace with its more uniform and faster heating cycle allows a lower atomization temperature for aluminum, 2200#{176}C. Analyte carryover was significantly reduced in the THGA furnace system. The THGA system generates results equivalent to HGA in about one-third less time, thus making possible a greater throughput of samples in a busy laboratory. IndexIng Terms:
Spectroscopy
heated calibration
graphite atand modifl-
Trace Metals Laboratory, Department of Clinical Biochemistry, University Hospital, P.O. Box 5339, London, Ontario, Canada N6A 5A5. ‘Address for correspondence. Fax 519-663-3858. 2Nonstandard abbreviations: HGA, end-heated graphite atomizer; THGA, transversely heated graphite atomizer; STPF, stabilized-temperature platform furnace; AAS, atomic absorption spectrometer(-ry); and GFAAS, graphite furnace atomic absorption spectrometer(-iy). Received September 20, 1993; accepted October 25, 1993.
cation with magnesium nitrate used in both procedures. Zeeman either transverse or longitudinal, Materials
and
and Triton background was also
X-100 were correction, used.
Methods
Instrumentation. All instrumentation (Canada), Rexdale, Ontario. man graphite furnace atomic absorption kin-Elmer
was from PerWe used a Zeespectrometer
(GFAAS) (Model Z-5000) equipped with an autosampler (Model A840) and a data station (Model 10) for the original STPF method. For the other system, we used a THGA furnace AAS, Zeeman system (Model 4100ZL), equipped with an autosampler (Model AS7O), a 386 PC controller, a recirculating water cooler, and a fume extraction
system.
Supplies. Glassware was soaked in 100 milL nitric acid (ACS grade) for at least 1 h, rinsed several times in Milli-Q [Miffipore (Canada), Mississauga, Ontario] purified water, and dried. We used pyrolytic graphitecoated tubes with solid pyrolytic platforms (PerkinElmer) for the Z-5000 system, and transversely heated graphite tubes with integrated L’vov platforms for the 4100ZL system, also pyrolytically coated (Perkin-Elmer). Matrix modifier. Plasma, serum, and urine samples were diluted with an equal volume of 1.4 g/L magnesium nitrate (Suprapur grade; Merck, Darmstadt, Germany) solution prepared in purified water containing 2 mL/L Triton X-100 (Scintillation grade; BDH Chemicals, Toronto, Ontario). Standards, controls, and patients’ samples. A standard solution of aluminum, 1004 mgfL in 20 milL HNO3 (Inorganic Venture Standard; Anachemica Canada, Mississauga, Ontario), was diluted with purified water to make final concentrations of 25 to 300 pgfL. Pooled plasma from the Canadian Red Cross Blood Transfusion Center (London, Ontario; tested negative for hepatitis B surface antigen and H1V antibodies), with low aluminum concentrations (40 sitated
fewer
rather than the original (1) with a non-Zeeman wavelength maintained or higher. This is similar
for the which under
chose to use nm with the 309.3 nm in system, beour linear to published
(2,4,9), who also found higher linearMost of the patients’ samples we monpgL, so use of this wavelength necessample
repeats.
In
our
comparison
of
paired
patients’
samples
40 pg/L gave similar results at either wavelength. With Zeeman AAS systems, these findings confirm that the 396.2-nm aluminum line provides a signal-to-noise ratio equivalent to that at the 309.3-nm wavelength, but with enhanced linearity (10). When we compared the 4100ZL system with the Z-5000 system for plasma and serum quality-control specimens at 396.2 nm (Table 2), the results were equivalent at the midrange (-61--88 pgfL). With lower values (-6-19 pg/L), the Z-5000 system gave slightly higher results, but the ranges overlapped. With higher values (-200 LgfL), the 4100ZL gave slightly higher results. These findings are not clinically significant, and the differences may not be distinguishable when larger sample populations are compared. Correlation. We determined the aluminum values on 34 patients’ samples, using both instruments at 396.2 nm. A slope of 0.995, y-intercept of -1.710, and S of 0.021 gfL were found for the regression analysis of samples with aluminum values of 2-112 1gfL. The correlation is very acceptable between these two analytical systems. The linear regression is shown graphically in Fig. 1 (three values >55 pgfL are not shown).
Table 2. Aluminum in patients’ and control samples determined with 41 OOZL and Z-5000 AAS systems. System Patients’
n plasma
Al conc, samples
4100ZL
18
4100ZL Quality-control
10 samples
(mean 11.2,
1iaJL
and range) 1-38.5
(12.4,
119.2, 41-225 (mean ± 2 SD)
2_40)a
(119.5,
41_226)a
Plasma 4100ZL Z-5000
17 17
15.9 18.6
± ±
1.4 1.5
61.5 61.6
±
66.1 66.0
±
±
1.6 2.0
88.3
±
1.1
87.9
±
2.1
±
8.7 5.2
Serum 4100ZL Z-5000 a
309.3
15 6.7 ± 0.3 15 7.0 ± 0.5 nm; all other results measured at
396.2
1.0 2.4 nm.
210 204
±
±
Calibration. Two methods for the calibration of AAS for aluminum determinations are currently used. One is based on the use of aqueous standards, and the other uses standards added to a similar matrix such as serum or plasma. Because of the speciation of this metal, the pH of the aqueous solution would have to be near 3 or >8 to ensure that all the aluminum remains in solution (2). In plasma or serum at the physiological pH of 7.4, most of the aluminum that is protein bound would be precipitated with an acid matrix diluent (>1 mmolfL acid). The sample should be mixed well to ensure that all
the
after
aluminum centrifugation.
should
be
is extracted
The
similarly
treated
avoid possible atomization chemical modifier-treated precipitation technique, calibration,
can
produce
into
acidified
the
acidic
aqueous
supernate
standards
with chemical modifiers to variations with the acidserum samples (2). The acid which uses aqueous standard results similar to those of the with a serum-based calibrator
direct-calibration method (11), but is more time consuming and requires larger sample volumes, and the acid reduces the useful life of the graphite furnace. In our use of nonacidifled aqueous standards diluted with the same matrix modifier as for plasma on the 4100ZL, only -70% of the matrix-based aluminum is detected, similar to that reported by others using a fourfold dilution of serum with 1 milL Triton X-100 and analyzing with a Hitachi Zeeman GFAAS (12). Because we did not veriiy all the different effects such as matrix modifiers and type of tubes on the atomization response (2), we chose to use plasmaor serumand urine-based calibration standards throughout this study, as we reported earlier (1). Accuracy and precision. The accuracy of the Z-5000 system for determining plasma aluminum was assessed from previous recovery studies (1), as well as from external quality assurance programs. The Centre de Tonicologie du Quebec interlaboratory comparison program has ranked our procedure for aluminum as 100% accurate for 1992, with a ranking of 1st of the 100 participating laboratories. We have also scored high in the Guildford Trace Element external quality assessment scheme (Guildford, Surrey, UK), and the Worldwide Interlaboratory (Poitiers Cedex, France) aluminum quality-control
trace elements group. Although we have external controls on the 4100ZL system, the close correlation between the patients’ and internal quality-control results found for both systems suggests that similar accurate results would be obtainable with the 4100ZL. The within-run and between-run precision was deternot
tested
mined
on
trol
serum
6.5
to 210
these
the gfL
(n = 20 each) for aluminum Alumbun pg/I. Z-5000
MS
Fig. 1. Linear correlation of patients plasma aluminum determined with the Z-5000 and the 41 OOZL MS
tions n
=
34, slope
=
0.995, y.intercept= -1.710,
and S,
=
0.021
concentrasystems.
izgJL
4100ZL
and
system
by
using
six
quality-con-
plasma concentrations with a range of aluminum. The CVs for within-run tests with aluminum at 7 pgfL was 4.2%, and at 88 ,u.gfL was 1.2%. The respective be-
tween-run CVs were 8.6% and 2.4%. These are comparable with results for serum and urine obtained with the Z-5000 system (1). Some drawbacks of the newer system include the slightly lower sensitivity for aluminum determination, CLINICAL
CHEMISTRY,
Vol. 40, No. 3, 1994
433
which can be compensated for with the use of a larger sample volume, and the slightly more difficult alignment of the autosanipler probe into the graphite tube opening, as well as the positioning over the platform, because of the fixed position of the 4100ZL graphite furnace and its smaller size. The probe adjustments are infrequent, because the sampler is very reproducible. In conclusion,
we
found
that
the
4100ZL
system
pro-
vides a reliable aluminum determination that is comparable in detection limits and precision with the Z-5000 system. The newer technology of the THGA furnace, however, incorporates several advantages. The integrated L’vov platform in the THGA achieves a faster heating cycle, with the result that aluminum atomization takes place at 2200#{176}Cinstead of the 2300#{176}Cin the HGA system. This allows a more rapid and uniform generation of the atomization signal. Coupled with faster drying and pyrolysis steps, the net result is a reduction in analysis time. The lower atomization temperature and absence of an acidic matrix also help to extend the useful life of the graphite furnace. We also noted less carryover of sample-to-sample contamination after analysis of high aluminum concentrations. With the Z-5000, it was often necessary to run a water blank after a high sample to recover a low background. The even heat distribution of the THGA tube avoids the memory
effect
problem.
We thank John Smith, Perkin-Elmer (Canada) for the use of the AAS (Model 4100ZL); Cynthia Priest Bosnak, Perkin-Elmer, Senior Product Specialist, for her technical advice on atomic spectroscopy on the Model 4100ZL; and Walter Slavin for reviewing this manuscript.
434
CLINICAL
CHEMISTRY,
Vol. 40, No. 3, 1994
References 1. Leung
FY, Henderson AR. Improved determination nuni in serum and urine with use of a stabilized platform furnace. Clin Chem 1982;28:2139-43.
of aluini-
temperature
2. Gardiner PE, Stoeppler M, Nurnberg HW. Optimisation of the analytical conditions for the determination of aluminum in human blood plasma or serum by graphite furnace atomic-absorption spectrometry: part 1. Examination of the various analytical conditions. Analyst 1985;110:611-7. 3. Slavin W. An overview of recent developments in the determination of aluminum in serum by furnace atomic absorption spectrometry. J Anal At Spectrom 1986;1:281-5. 4 Gardiner PE, Stoeppler M. Optimisation of the analytical conditions for the determination of aluminum in human blood plasma and serum by graphite furnace atomic absorption spectrometry: part 2. Assessment of the analytical method. J Anal At Spectrom 1987;2:401-4. 5. Savory J, Brown S, BertholfRL, Mendoza N, Wills MR. Aluminum [Review]. Methods Enzymol 1988;158:289-301. 6. Hewitt CD, Winborne K, Margrey D, Nicholson JRP, Savory MG, Savory J, Wills MR. Critical appraisal of two methods for determining aluminum in blood samples. Clin Chem 1990;36:
1466-9. 7. Massmann H. Vergleich von Atomabsorption und Atomfluoreszenz in der Graphitekuvette. Spectrochim Acta 1968;23B:21526. S. de Loos-Vollebregt MTC, de Galan L, van Uffelen JWM. Longitudinal a.c. Zeeman AAS with a transverse heated graphite furnace. Spectrochim Acta 1988;43B:1147-56. 9. Taylor A, Walker AW. Measurement of aluminum in clinical samples. Ann Clin Biochem 1992;29:377-89. 10. Manning DC, Slavin W. The choice of an analytical Zeeman AAS wavelength for aluminum. At Spectrosc 1986;7:123-6. 11. Bradley C, Leung FY, Slavin W, Henderson AR. Directcalibration method for determination of aluminum in serum is comparable with the protein-precipitation technique. Clin Chem 1985;31:1882-4. 12 Wang ST, Pizzolato S, Demshar HP. Aluminum levels in normal human serum and urine as determined by Zeeman atomic absorption spectrometry. J Anal Toxicol 1991;15:66-70.