182
Anal. Chem. 1980, 52, 182-185
would show different characteristics than the carbohydrate gels. The objective of this work has been to examine the interaction of various proteins with a chemically bonded diol phase. I t is already known that separation speeds and column efficiencies with such packings are much improved over classical polysaccharide and polyacrylamide gels (13-15). The lack of wide adaption of the silica gel based packings appears to be due to the belief that some proteins interact excessively with the stationary phase and that the bonded phase has poor long term stability. These problems have been addressed in this paper. It has been found that by control of the mobile phase (e.g. ionic strength) these problems can be minimized, permitting the diol phase to be used for the steric exclusion chromatographic separation of a variety of small proteins.
(18) White, A.; Handler, P.; Smith, E. L. "Principles of Biochemistry"; McGraw-Hill: New York, 1973; p 142. (19) Keil, B. I n "The Enzymes", Vol. 3, Boyer, P. D., Ed.; Academic Press: New York, 1971; pp 249-275. (20) Fruton, J. S. I n "The Enzymes", Vol. 3, Boyer, P. D., Ed.; Academic Press: New York, 1971; pp 119-164. (21) White, A.; Handler, P.; Smith, E. L. "Principles of Biochemistry"; McGraw-Hill: New York, 1973; p 133. (22) Peters, T. I n "The Plasma Proteins", Vol. I,Putnam, F. W., Ed.; Academic Press: New York, 1975; pp 133-181. (23) Putnam, F. W. I n "The Plasma Proteins", Vol. I,Putnam, F. W., Ed.; Academic Press: New York, 1975; pp 265-316. (24) Dehnge, R. J.; Smith, E. L. I n "The Enzymes", Vol. 3, Boyer, P.D., Ed.; Academic Press: New York, 1971; 81-118. (25) Habeeb, A. F. S.A.; Atassi, M. Z. Immunochem. 1971, 8, 1047-1059. (26) Habeeb, A. F. S.A. Arch. Biochem. Biophys. 1967, 719, 264-268. (27) Sigga, S."Quantitative Organic Analysis via Functional Groups", 3rd ed.; John Wiley and Sons: New York, 1965; pp 39-40. (28) Jay, R. R.. Anal. Chem. 1964, 36, 667-668. (29) Regnier, F. E., Fourth International Symposium on Column Liquid Chromatography, Boston, Mass., May 1979. (30) Straznesko, D. N.; Strelko, V. 6.; Belyakov, V. N.; Rubanik, S. C. J . Chromafogr. 1974, 102, 191-195. (31) Crone, H. D. J . Chromatogr. 1979, 92, 127-135. (32) Klapper, M: H.; Klotz, I.M. I n "Methods in Enzymology", Vol. 25, Hirs, C. H. W., Timasheff, S.N., Eds.; Academic Press: New York, 1972; pp 531-536. (33) Riordan, J. F.; Vallee, B. L. I n "Methods in Enzymology", Vol. 25, Hirs, C. H. W., Tirnasheff, S.N., Eds.; Academic Press: New York, 1972; pp 494-499. (34) Antonini, E.;Brunori, M. "Hemoglobin and Myoglobin in Their Reactions with Ligands": North-Holland Publishing Co.: Amsterdam, 197 1; pp 113-114. (35) Margoliash, E.; Schejter, A. Adv. Protein Chem. 1988, 27, 113-286. (36) Labouesse, J.; Gervais, M. Eur. J . Biochem. 1967, 2, 215-223. (37) Hofstee, 8. H. J. In "Protein Separations", Vol. 11, Catsimpoobs, N., Ed.; Plenum Press: New York. 1976; pp 245-278. (38) Hofstee, B. H. J.; Otillio, N. F. J . Chromatogr. 1078, 161, 153-163. (39) Strop, P.; Mikes, F.; Chytilova, Z. J . Chromatogr. 1978, 156, 239-254. (40) Meuarech, M.; Leicht, W.; Werber, M. M. Biochemistry 1976, 15, 2383-2387. (41) von der Haar, F. Biochem. Biophys. Res. Commun. 1978, 70, 1009- 1013.
LITERATURE CITED Morris, C. J. 0. R.; Morris, P. "Separation Methods in Biochemistry"; John Wiley and Sons: New York, 1976; pp 413-470. Belew, M.; Porath, J.; Fohiman, J.; Janson, J . 4 . J . Chromatogr. 1978, 147, 205-212. Lin, A. W . 4 . M.; Castell, D. 0. Anal. Biochem. 1978, 86. 345-356. Andrews, P. In "Methods of Biochemical Analysis", Vol. 18, Glick, D., Ed.; John Wiley and Sons: New York, 1970; pp 1-53. Messing, R. A. J . Am. Chem. Soc. 1989, 91, 2370-2371. Anderson, D. M. W.; Dea, I.C. M.; Hendrie, A. Talanta 1971, 18, 365-394. Mizutani, T., Mizutani, A. J . Chromatogr. 1979, 168, 143-150. Mizutani, T., Mizutani, A. J . fharm. Sci. 1978, 67, 1102-1105. Shechter, I. Anal. Biochem. 1974, 58, 30-38. Hawk, G. L.; Cameron, J. A.; Dufault, L. B. Prep. Biochem. 1972, 2 , 193-203. Fukano, K.; Komiya, K.; Saski, H.; Hashimoto, T. J . Chromatogr. 1978, 166, 47-54. Engeihardt, H.; Mathes, D. J . Chromatogr. 1977, 142, 311-320. Mathes, D.; Engelhardt, H. Naturwissenschaften 1979, 66, S5 1-52. Regnier, F. E.; Noel, R. J . Chromatogr. Sci. 1978, 14, 316-320. Unger, K. K.; Becker, N. P. (a) 28th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, 1977, Cleveland, Ohio, Abs. No. 171; (b) Fourth International Symposium on Column Liquid Chromatography, Boston, Mass., May 1979. Righetti, P. G.; Caravaggio, T. J . Chromatogr. 1978, 127, 1-28. Richards, F. M.; Wyckoff, H. W. In "The Enzymes", Vol 4, Boyer, P. D., Ed.; Academic Press: New York, 1971; pp 647-806.
RECEIVED for review June 8, 1979. Accepted October 5, 1979. The authors thank the National Institute of Health Grant GM 15847 for financial assistance. Contribution No. 57 from the Institute of Chemical Analysis.
Analytical Parameters for Determination of Chromium in Urine by Electrothermal Atomic Absorption Spectrometry M. W. Routh Varian Instrument Group, 6 7 7 Hansen Way, Palo Alto, California 94303
The analytical parameters for the detemlnation of chromium in urine by electrothennal atomic absorption spectrophotometry are evaluated. The reduction of and compensation for nonspecific absorption interferences is the primary factor limiting analytical accuracy. Reduction of nonspecific absorption to an acceptable level was achieved using a hydrogen diffusion flame as a supplement to the inert gas sheathlng the graphite absorption cell. Simultaneous continuum source background correction using a deuterium arc lamp proved adequate for nonspecific absorption compensation. The typical limit of detection for chromium In the urine matrix, using 20-I.~Lsample Injection volumes, Is 0.2 hg/L.
Studies on the significance of various trace elements in human nutrition, as well as toxicology, continue to flourish 0003-2700/80/0352-0182$01,00/0
as a result of the constant improvement of analytical methods and instrumentation ( I ) . Chromium, which has been identified as essential for maintenance of normal glucose metabolism (2),is one of those elements. It has been suggested that the determination of chromium in urine is a viable means for evaluating chromium utilization (2). Flame atomic absorption has frequently been the method of choice for trace metal measurement in biological systems (3). The utilization of electrothermal atomic absorption spectrophotometry (EAAS) allows access to lower analyte concentrations and in many cases simplified sample preparation procedures. As a result, several groups of workers have reported the determination of chromium in urine using graphite furnace atomic absorption methods (4-8). However, each group also reported difficulties, using conventional instrumentation, related to nonspecific absorption (background) which interfered with the measurement of chromium. Several 0 1979 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 52, NO. 1, JANUARY 1980
Table I. Analytical Conditions for the Determination of Chromium in Urine A V E L E I\ G - I-
0500
I
7
357 9 -111 0 5 qr-3
'i""
SPECTRA. B A \ D W I D T H
@ A C K 2 F. C U k D COR R E CT I O', I 1 E ASU G E hl E k T n ? C DE sa'upi. E V ~ L U M E
YES >EA< HEICHT
20 100 C P O S E C 1000 C 5 0 S E C
3PY Ash ATOI.l I LE
II
2300 C 1 5 S E C SO0 C SEC '
RA'.lP H A T E S h E A T i GAS
CAL'BRATION
183
MT90GEh S T A ;U D A P 3 A D D I - C \ S
300
solutions to the interference were employed with varying degrees of success. These include cycling through two different dry, char, and atomize temperature cycles ( 4 , 5 ) ,replacement of the deuterium arc background corrector with a tungstenhalogen lamp and an external power supply (B), and utilization of a specially constructed continuum emission wavelength modulated atomic absorption spectrophotometer ( 7 ) . The purpose of this investigation was to develop and evaluate parameters suitable for the direct determination of chromium in urine using EAAS and conventional atomic absorption instrumentation. The parameters investigated include ashing and atomization temperatures and heating rate; peak height vs. peak area measurement; effect of acid on chromium absorbance; effect of sheath gas composition on accuracy, precision, sensitivity, and nonspecific absorbance; quality and consistency of graphite atomization components; and limit of detection in a urine matrix using background correction. The suitability of carbon rod atomization and deuterium arc background correction in view of these parameters is discussed.
EXPERIMENTAL Instrumentation. A Varian Techtron AA-775 Atomic Absorption Spectrophotometer (AAS),CRA-90 carbon r o d atomizer, ASD-53 automatic sample dispenser, and Varian 9176 chart recorder with a 0.1-s time constant were used for all measurements. Quantitative measurements were accomplished using the electronic peak height/peak area measurement and recall facility of the AA-775. The chart recorder was used primarily as a diagnostic aid during the parameter investigation phase of these experiments and as a visual aid during sample measurement. Double beam background correction was accomplished using a built-in highintensity deuterium arc. The initial analytical conditions employed are listed in Table I. All graphite tubes used were pyrolitically coated by the manufacturer (Ultra-Carbon) and available commercially (Varian). Reagents. Chromium standards were prepared by diluting a 1000 pg/mL stock solution (Arro Laboratories), in the form of ammonium chromate, with deionized distilled water. The standards were prepared daily and were 1 N with respect to HN03 or HC1. Sample Preparation. Urine samples were collected over a 24-h period in acid-washed polyethylene bottles, acidified with Ultrex HN03 (J.T. Baker Chemical Co.) and frozen until analyzed. Six samples were collected from healthy male adults from 26 to 41 years of age. Thawed, acidified samples were individually prepared for analysis utilizing the standard additions technique by diluting aliquots 1 to 1 with blank or chromium standard solutions. No other sample preparation was used. NBS standard reference material (SRM) 1571 was prepared by drying an accurately weighed portion of sample ( - 1 g) and digesting in 10 mL of Ultrex HN03 for about 15 min. Two mL of perchloric acid (J. T. Baker) were added and the solution was evaporated to moist salts. The moist residue was taken up in 2 mL of HN03 and diluted to 10 mL with deionized distilled water.
RESULTS AND DISCUSSION The shapes of atomization peaks in EAAS are dependent ) on the complex relationship of atomization time ( T ~ and residence time of the atomic vapor ( T ~ )(9). For a given
$00
500
Sl;DflAlE,
600
'00
800
Csrc'
Figure 1. Chromium signal vs. atomization ramp rate. 0 = peak height,
= peak area
number of available analyte atoms, decreasing the atomization time by increasing the atomization ramp rat e should produce larger peak absorbances of shorter duration. However, the peak area should remain somewhat constant. A comparison of peak height and peak area measurement of chromium absorbance in aqueous solution as a function of atomization ramp rate is shown in Figure 1. A constant analyte quantity, 20 pL of 25 ng/mL chromium standard, was used for the comparison. As predicted, the peak height increases nearly linearly with ramp rate while the peak area remains approximately constant. Although the best sensitivity was achieved a t a ramp rate of 800 "C s-l, it was found that the best reproducibility was obtained in the peak height mode a t an atomization ramp rate of 600 "C s&. The latter ramp rate was chosen since it provided the optimum combination of sensitivity and reproducibility in the carbon rod atomizer. It has been suggested that the substitution of argon in place of nitrogen as the inert sheathing gas can offer several advantages, including enhanced sensitivity ( I O , 11) and reduced base-line disturbances ( 7 ) . In this work, however, no significant differences were observed with the carbon rod using nitrogen or argon, under the conditions employed. As a result, nitrogen was used as the inert sheath gas for the remainder of the experiments in this study. The effect of ash temperature on the nonspecific absorbance signal derived from the urine matrix was determined. The background absorbance signals from urine samples made 1 N in either hydrochloric or nitric acids were measured using ash temperatures from 400 to 1400 "C. A t 1300 "C, the nonspecific absorbance is reduced, in either acid/urine matrix, about 90%. A similar study of the effect of ash temperature on the chromium atomic signal was made in the two different acid matrixes. At the same 1300 "C ash temperature which efficiently volatilizes the urine matrix, a significant loss of chromium also occurs in either the chloride or nitrate matrix. As a result of these volatility comparisons, a compromise ash temperature is indicated owing to the similarity in the volatilities of the urine matrix and the onset of chromium atom formation. Although 1200 "C is the ash temperature of choice based solely on comparative volatilities, loss of chromium during the ash cycle at 1200 "Cleads to sufficient uncertainty in the measurements of urinary chromium, resulting in poorer precision in the results. Therefore, an ash temperature of lo00 "C was chosen to ensure maximum retention of chromium during the ash cycle and optimum precision in the chromium determinations. Simultaneous background correction using a continuum source (deuterium arc) was subsequently used to compensate for the residual nonspecific absorbance. A chart recorder tracing of the signals obtained from chromium in urine due to the background and under background correction conditions is shown in Figure 2. The signals correspond to 20-pL injections of acidified urine spiked with 2.5 ng/mL of chromium standard. The peak obtained from the spiked
184
ANALYTICAL CHEMISTRY, VOL. 52, NO. 1, JANUARY 1980
1
li
i
Ii
0 02 ABS
BACKGROUND ONLY
'
i
NO H2' BACKGRC J Z D CORRECTED
Figure 2. Comparison of background only and background corrected peaks for chromium in urine
urine, attributable to chromium, is an order of magnitude smaller than the background superimposed upon it. Consequently, a small error in background compensation could mean a significantly larger error in the chromium measurement. Clearly, any means of modifying the atom production environment which would effectively increase the analyte signal-to-background ratio should improve background correction accuracy and therefore the overall analytical accuracy and precision. One straightforward means of altering the atom production environment is to change the sheath gas or combine two or more gases enveloping the atomizer tube. The addition of hydrogen to the inert sheath gas during ashing and atomization has been shown to enhance the signals obtained for some elements in a variety of matrixes (12-14). This improvement in sensitivities for some 14 elements has been attributed to the creation of a more reducing environment, facilitating atom formation processes. For some elements, this may be related to a reduced tendency toward carbide formation on the graphite surface in the presence of hydrogen. For other elements which atomize through the oxide form, the presence of hydrogen probably promotes completion of the oxide dissociation reaction or makes metal oxide formation unfavorable by reducing the partial pressure of oxygen in the diffusion flame environment. In the case of chromium, Shrader and co-workers (14) showed that the degree of enhancement diminished with hydrogen flow rate, and ranged between +30% (at 0.5 L/min) t o -15% (at 3 L/min). This loss at higher hydrogen flow rates is caused by a dilution of the inert sheath gas with a lower molecular weight gas, effectively increasing the rate of diffusion of analyte from the graphite tube. In this study, marginal signal enhancement was encountered when using the nitrogen-hydrogen diffusion flame. However, a significant reduction in nonspecific absorbance was observed, independent of hydrogen flow rate, as shown in Figure 3. The peaks shown were obtained using the background-only mode, i.e., using only the deuterium source. The indicated peaks obtained during the atomization cycle using an ash temperature of 1000 "C for 50 s, an atomize temperature of 2300 "C, and an atomize ramp rate of 600 "C s-', correspond to the least volatile nonatomic components of the urine matrix. The use of the nitrogen-hydrogen diffusion flame permits a significant reduction in nonspecific absorbance, in this case by a factor of approximately 3. Several complex factors may be contributing to this phenomenon. When the hydrogen
WIT? H,
Figure 3. Effect of the nitrogen-hydrogen diffusion flame on urine background absorbance 0 080 -7
0
I 0
10 20 3 0 40 50 CO v C E U T R A T I O N C I R O ' l l t l h l ~ n gm
I
I
I
I
20
40
6C
80
I
I 100
ABSOLUTE i ' i E I G H T C H R O ' ~ I I U h ' ~ 1 0 J'
Flgure 4. Chromium calibration curves in 3 different matrixes. 0 = 1 N nitric acid; A = urine diluted 1:l with nitric acid, final solution 1 N with respect to nitric acid; I3 = 1 N hydrochloric acid. Analytical conditions are those of Table I
flame ignites, it forms an extended high temperature sheath around the graphite tube enhancing volatilization of matrix salts in the tube. This in turn leads to a decrease in the nonspecific contribution to the absorption peak. I t has also been suggested that a portion of the nonspecific absorption signal may be due to scattering by volatilized carbon in the particulate form and that hydrogen effectively reduces this signal by hydrocarbon formation (15). I t is likely that both enhanced volatilization and reduced scattering by particulates contribute to the overall reduction of nonspecific absorbance. While nonspecific absorbance cannot be entirely eliminated, the use of the nitrogen-hydrogen diffusion flame, in conjunction with the appropriate ash temperature, effectively reduces the nonspecific contribution to an absorbance level approaching the same order of magnitude as the chromium level in urine. This reduction permits background correction accuracy levels acceptable for routine work. While there is no standard urine sample currently available that is certified for chromium content, NBS standard reference material number 1571, bovine liver, was analyzed for chromium content using the analytical conditions listed in Table I. The results were 0.084 f 0.003 Hg/g which compares favorably with the certified value of 0.088 f 0.012 pg/g. Based on these results, the analytical conditions employed appear suitable for the determination of naturally occurring chromium in a t least one form of biological material. In order to minimize matrix interferences during calibration, the method of standard additions was chosen. The fact that standard additions is necessary is demonstrated in Figure 4 which shows varying calibration slopes for nitric acid, hydrochloric acid and urine/nitric acid matrixes. The responses
ANALYTICAL CHEMISTRY, VOL. 52, NO. 1, JANUARY 1980
185
Table 11. Results of Chromium Analysis in Urine by Carbon Rod Atomic Absorption M E A ? RESULT
109,r,
RANGE O F RESULTS
0 4 5 to 2 35 ?5 13to20
PRECIS'OIL
50
0
25 CHR0:'lUlvl
30
i
-
20
13
I
to imprecision from the use of deuterium arc background correction or the urine matrix. The consistency and lifetimes of the graphite components were also investigated. While the relative sensitivities for chromium among all of the graphite tubes used varied by as much as 50%, the analytical results were in good agreement with each other, independent of which tube was used. Additionally, we experienced neither the multiple peaks nor the inconsistent peak shapes reported by other workers (7). Typical graphite lifetimes were on the order of 80 to 125 firings. A summary of the analytical results is shown in Table 11. The mean value found for chromium in urine in the test population was 1.09 pg per 24-h sample with a range of 0.45 to 2.35 pg per 24-h sample. On a concentration basis, the mean value was 0.79 pg/L. The measurement precision was 10 to 20% RSD a t the 1 pg per liter level. Based on 20-pL sample injections, the limit of detection in urine diluted 1:l with water, using the experimental parameters reported here, is 0.2 Fg/L. We have found the determination of chromium in urine, using standard additions, to be achievable a t the rate of 30 determinations per hour using single measurements, or 15 per hour using duplicate readings. The analytical results of this study indicate a typical chromium content in urine of approximately 1 pg per 24-h sample (corresponding to less than 1pg/L), in accordance with other recent studies (7, 8), but in disagreement with other earlier published values using graphite furnace atomization (4-6). This disparity in results appears to be related to the stringent requirements for background correction in the chromium/urine system.
lLlil -
0 1
0 2 , i g LITER
ADDED q g ml
Figure 5. Typical background-corrected atomization peaks for chromium in urine by standard additions
'@
L "IT OF DETECTION I \ URINE [BASED O Y 20 u i INJECTIONS1
24 H R SAMPLE 24 HR SAMPLE
1
1 1 ,
10
AeSOLJTE
, , \ I
I
l
l
10
g
100
v E GHTCHROP
U'
1
1 1000
Figure 6. Precision vs. absolute weight of chromium 0 = peak height, fI= peak area
from the urine/nitric acid matrix are offset to pass through the origin by an amount corrsponding to the urinary chromium content. This facilitates comparison of slopes for the purpose of determining the need for standard additions. The calibration slope for the urine/nitric acid matrix indicates a characteristic concentration of approximately 6 pg on an absolute weight basis and 0.3 ng/mL, on a concentration basis, using 20-pL sample injections. Since the AA-775 has the capability of computing and storing standard additions calibration lines using a linear least-square regression algorithm, this facility was used to obtain direct concentration readout of all results by the method of standard additions. Chart recorder tracings for a typical set of standard additions peaks are shown in Figure 5. The duplicate sets of peaks correspond to 20-pL injections of sample containing 5 ng/mL chromium added, 2.5 ng/mL chromium added, and no chromium added. Although the chromium concentrations typically encountered in urine are very low, usually less than 1 ng/mL, the precision a t these levels is about 10 to 15% relative standard deviation (RSD), permitting acceptable quantification. For the solutions containing added chromium, the precision improves to better than 10% RSD owing to the higher signal-to-noise ratios encountered a t the higher chromium levels. Plots of the precision in percent RSD as a function of the absolute weight of chromium are shown in Figure 6. The weight scale is plotted as a logarithmic function. The data were acquired in the background corrected mode using 9 replicates a t each point plotted, 20-pL injections, and aqueous solutions which were 1 N in nitric acid concentration. The precision and detection limit values were calculated using the mean-to-sigma definition defined by Rowe and Routh (16). The chromium detection limit is approximately 3 pg with better precision in the peak height mode a t all chromium concentrations except near the detection limit. This anomalous behavior of the precision data near the detection limit was judged insignificant, with better precisions for most samples being obtained in the peak height mode. More importantly, precision measurements in the urine matrix and in aqueous solutions were nearly the same at equivalent chromium concentrations, indicating minimal contributions
LITERATURE CITED H. J. Sanders, Chem. Eng. News, March 26, 1979, 27-46. W. Mertz, Physiol. Rev., 4S, 163 (1969). G. D. Christian and F. J. Feldman, "Atomic Absorption Spectroscopy: Applications in Agriculture, Biology, and Medicine", Wiley-Interscience, New York, 1970. I. W. F. Davidson and W. L. Secrest, Anal. Chem., 44, 1808 (1972). K.H.Schakr, H.G. Essing, H.Valentin, and G. Schicke, 2.Win. Chem. Klin. Biochem., I O , 434 (1972). R. T. Ross, J. G. Gonzalez, and D. A. Segar, Anal. Chim. Acta, 63, 205 (1973). 8. E. Guthrie. W. R. Wolf, and C. Veiilon, Anal. Chem., 50, 1900 (1978). F. J. Kayne, G. Komar, H. Laboda, and R. E. Vanderiinde, Clln. Chem.. 24, 2151 (1978). R . E. Sturgeon, Anal. Chem., 40, 1255A (1977). K. C. Thompson, R. G. W e n , and D. R. Thomerson, Anal. Chim. Acta, 74. 289 (1975\. R. ErSturgeonand C. L. Chakrabarti. Prog. Anal. At. Spectrosc., I, 5 (1978). M. D. Amos, P. A. Bennett, K. G. Brodie, P. W Y. Lung, and J. P. Matousek, Anal. Chem., 43, 211 (1971). Y. E. Araktingi, C. L. Chakrabarti, and I. S.Maines, Specfrosc. Len.. 7, 97 (1974). D. Shrader, B. R. Culver, and R. Ippolito, 26th Pittsburgh Conference on Analvtical Chemistw and Aoolied Soectroscoov. , . . . Cleveland, Ohio, 1975. paper no. 453. . D. J. Johnson, T. S.West, and R . M. Dagnall, Anal. Chim. Acta. 68, 171 11973). C. J . ' R o w ~and M. W. Routh, ResJDev., 28, 74 (1977)
RECEIVEDfor review July 6, 1979. Accepted October 29, 1979.