2680
Anal. Chem. 1988, 60,2680-2683
Low-Pressure Vaporization for Graphite Furnace Atomic Absorption Spectrometry D. C. Hassell, T. M. Rettberg,' F. A. Fort, and J. A. Holcombe* Department of Chemistry, The University of Texas at Austin, Austin, Texas 78712
A low-pressure graphite furnace atomic absorption spectrometry technique is presented which, for aqueous Pb samples at 0.15 Torr, results In a 2 orders of magnitude reduction In sensltlvlty and a worklng range that extends to about 1 pg. The limn of detection is 5 ng of Pb. Stkklng Is dlmlnlshed for refractory analytes (e.g., V), thus reducing memory effects and peak talllng. Solid sample analysis Is demonstrated for Pb In a phosphorlzed Cu alloy, for which low-pressure atomlzation minlmlzes Cu vaporizatlon while allowlng for quantltative determination of Pb when compared to a callbration curve prepared from aqueous standards. I n addition, the appearance of multlple peaks suggests the use of the technique for dlfferentiatlng between varlous forms or locatlons of Pb wHhin the sample. This is In contrast to atmospheric vaporlzatlon where only one broad peak Is observed.
Graphite furnace atomic absorption spectrometry (GFAAS) is typically conducted at 1 atm of pressure, usually with Ar or N2 Some researchers (1-5) have explored the use of elevated pressures within the furnace, which generally resulted in line broadening with a subsequent improvement in linearity of the working curve at the expense of sensitivity. In another pressure regime, Donega and Burgess (6) investigated lowpressure atomization using electrothermal vaporization from graphite, Ta, and W boats within a quartz tube. Although the low pressure was limited to 1 Torr, the trends were noteworthy: below 100 Torr the expected loss in sensitivity occurred, but the peaks were sharper relative to those obtained at 1 atm of pressure. This paper explores the use of low-pressure vaporization within a graphite tube furnace. The initial work was directed a t reducing memory effects of metals that may stick to graphite or form refractory species (e.g., carbides and oxides) and enhancing the release of metals from a solid matrix of similar volatility. Subsequent investigations with solid samples resulted in absorbance profiles containing multiple peaks, which seemed to correlate with analyte distribution in the sample. This preliminary work suggests that low-pressure vaporization offers improved direct solids analysis and potential elucidation of the analyte location within the sample (i.e., surface vs bulk). EXPERIMENTAL SECTION Apparatus. A Varian AA-875 spectrometer was interfaced to an IBM-compatible PC (PC's Limited; Austin, TX) via a Keithley System 570 ADC. Data collection and interpretation were performed with software written in ASYST (ASYST Software Technologies, Inc.; Rochester, NY). A Varian GTA-95 graphite tube atomizer was adapted for operation under vacuum conditions. Modified furnace chamber and end window assemblies were fabricated in the chemistry department machine shop, and Viton O-rings were fitted at junctions within the chamber assembly. Vaporization from the wall was used in all studies. Figure 1 shows the modified atomizer and general vacuum system configuration employed. The vacuum line entered the 'Present address: Varian Atomic Absorption Resource Center, 205 W. Touhy, Park Ridge, IL 60068.
furnace chamber normal to the furnace itself and was positioned as far away from the dosing hole as possible. The line consisted of a rotary vane vacuum pump (DUO-WB, Balzers; Hudson, NJ), a thermocouple vacuum gauge, and a bleed valve for the introduction of Ar. Pressure within the chamber was regulated with the bleed valve and differential pumping. With the bleed valve closed, a lower pressure limit of -0.05 Torr could be achieved. Reagents. Aqueous standards were prepared from 1000 ppm stock solutions. Solid sample analyses were performed by using NBS standard reference material (SRM) 1253a, consisting of small shards of phosphorized copper alloy, approximately 3 mm long, that were easily introduced through the furnace dosing hole. SRM 1253a is a milled version of the bulk SRM 1253; while the former is uncertified at this time, the bulk material has a certified Pb composition of 244 f 2 Ilglg. Procedure. Aqueous solutions were manually pipetted onto the furnace wall with a standard GC syringe fitted with a Teflon capillary. The furnace was then heated under 1atm of Ar to desolvate the sample. Solid samples, weighing 3-6 mg, were introduced directly through the dosing hole by using fine-pointed steel forceps. After the chamber was sealed, the system was pumped to the fiial pressure, which was controlled with fiie adjustments of the bleed valve, and the temperature program was then initiated. Since drying and ashing steps were unnecessary for solid metal samples, analysis times were much shorter than for solution analysis. Following atomization with the solid NBS samples, the residual Cu remained as a small sphere and was easily removed by suction with a disposable pipette inserted through the dosing hole.
RESULTS AND DISCUSSION Figure 2 illustrates absorbance profiles for various concentrations of the aqueous P b standards at (a) 760 and (b) 0.15 Torr; correspondingcalibration curves are shown in Figure 3. Peak shapes are similar for both pressures, indicating that both vaporization processes are governed by comparable mechanisms. As can be seen in this figure, a loss in sensitivity of approximately 2 orders of magnitude is experienced under reduced pressure conditions. The expected increase in curvature in the calibration curve due to the narrower absorption line profile is also seen in Figure 3b. This curvature permits an extended working range for vacuum vaporization and may prove worthwhile for samples in which maximal GFAAS sensitivity is not needed, or if a wide range of concentrations is expected. An estimated detection limit of 5 ng of Pb for the low-pressure method was calculated from the curve by using the criterion S I N = 3. Percent relative standard deviation (%RSD) ranged from 2% to l l % for the range 1000-40 ng of Pb. Figure 4 illustrates application of the technique to the determination of V, an element not readily amenable to GFAAS analysis due to the probable formation of refractory species on the graphite surface (7). Atmospheric pressure studies using He and Ar as sheath gases indicated that a higher diffusion coefficient alone did not ameliorate the severity of tailing. However, a reduced pressure of 60 Torr resulted in
0003-2700/88/0360-2680$01.50/00 1988 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 60, NO. 24, DECEMBER 15, 1988 /
2881
Sample Inucduction Pon Seal
VltO
-+--
Lightpath
--
End Window Assembly
124 ITn
Modified GTA-95 Atomizer
p-TO Flgure 1.
4
6
ng Pb
a
Modified atomizer and vacuum system configuration.
'"T
2
0
AI Bled Valvs and Mechanical Pvmp
Detail of Furnace Chamber
a
b Flgure 3. Calibration of Ar; (b) 0.15 Torr.
I 210
i!o
6'0
elo
ib,
rb .
ng Pb curves from profiles in Figure 2: (a) 760 Torr
11
T I M E in1
Flgve 2. Absorbance profiles for aqueous Pb standards: (a) 760 Torr of Ar, sample mass (ng), 2, 4, 6, 20; (b) 0.15 Torr, sample mass (fig), 40, 100, 200, 400, 800, 1000.
virtual elimination of tailing and an attenuation in sensitivity similar to that obtained for Pb. These data for V suggest the potential utility of low-pressure vaporization for quantitative analysis of refractory metals or metals that tend to form stable surface species with the atomizer material. The data also suggest that reduced pressures may be useful for the "clean cycle" by reducing the maximum temperature and time required to adequately remove residual materials from the furnace. In many instances, this would result in a significant extension of tube lifetimes with only a alight increase in sample analysis times. Furthermore, the reduced sensitivity may promote usage of
I
21.0
4I.o
d.0
d o
ib.
ib
.
A.
T I I E lo)
Flgure 4. Absorbance profiles for aqueous V standards, with atomization at 2800 O C , 318.5-nm line: (a) 25 ng (760 Torr of He): (b) 125 ng (60 Torr of He).
vacuum vaporization for GFAAS in preference to flame systems in which flame temperatures are insufficient to provide reproducible atomization efficiencies. Solids Analysis. The analytical conditions used for the analysis of phosphorized Cu (SRM 1253a) for the determination of P b are shown in Table I. An increased residence time and the probable contribution of analyte readsorption and desorption from the graphite and Cu surfaces yielded one
2882
ANALYTICAL CHEMISTRY, VOL. 60, NO. 24, DECEMBER 15, 1988 a
Table I. Instrumental Parameters
IV
temperature program thermal pretreatmenta atomization ramp time, s 2 hold time, s 1400 temp, O C pressure wavelength spectral bandwidth lamp current
1
8 1800
b
0.15 Torr 283.3 nm 0.5 nm 5 mA
a This step was used to initiate data collection and establish base line (zero absorbance).
210
4!0
610
610
ib
15,
la
TIME Is1
k
Figure 8. Absorbance profiles for SRM 1253a (3.35 mg), 0.15 Torr, atomization at 1800 OC: (a) 283.3-nm Pb line; (b) 244.2-nm Cu line.
.,etI
IV
11
IV
n
Y
9
.o
0
2'0
410
8:o
8'0
ib
lb
1I
T I M [SI
Flgure 5. Absorbance profile for Pb in SRM 1253a, 760 Torr of Ar, atomization at 1800 OC, 281.4-nm Pb line.
large broad peak with no distinguishing features for the atmospheric pressure vaporization of P b from the Cu shard. This is illustrated in Figure 5, for which a less sensitive P b line was required in order to obtain usable absorbance values. In fact, it was very difficult quantitatively to remove the Pb from the Cu sample without succumbing to total vaporization of the Cu. At a reduced pressure of 0.15 Torr, a final atomization temperature of approximately 1800 "C provided a sharp rising edge for the Pb signal while minimizing Cu vaporization. Background interference due to Cu vaporization was discounted by observation of both the deuterium background signal and a less sensitive Cu resonance line at 244.2 nm. The Cu absorbance signal indicated that the matrix vaporized to some extent at this temperature, but always well after the Pb signal abated (Figure 6). Although Cu vaporization did not interfere with the Pb absorbance profile, it was sufficient to coat a thin layer of Cu on the cooler ends of the furnace, adjacent to the electrode contacts. After about 10 firings this Cu layer pulled the pyrolytic coating away from the furnace ends, possibly due to thermal expansion differences. This resulted in partial blockage of the optical path and less integral contact between the furnace and the electrodes; arcing often occurred a t this point. The Pb concentration in SRM 1253a Cu alloy, determined by comparison of the peak area for the unknown with a calibration curve derived by using aqueous standards, was found to be 268.3 f 10.6 rg/g, which is within 10% mass error compared to the Pb concentration in the certified bulk SRM 1253 (244 f 2 rg/g). A very striking feature of the absorbance profiles was the consistent appearance of four peaks for each alloy sample (Figure 6). These peaks were thought to be due to either different chemical forms of P b within the sample or changes in the relative location of the analyte (e.g., surface vs bulk). To test this hypothesis, samples were modified by
t 00
a
TIME
Flgure 7. Absorbance profiles for SRM 1253a, 0.15 Torr, atomlzatbn at 1800 OC, 283.3-nm Pb line: (a) unmodified sample (Figure ea), (b) sample that was electroplated with Cu, (c) sample that was Immersed in 1000 ppm Pb solution.
(i) decreasing the surface Pb coverage by electrodepositing Cu onto the surface, or (ii) increasing the surface Pb coverage by momentarily immersing the sample into a 1000 ppm Pb solution. Figure 7b shows the results for the sample onto which Cu was electrodeposited. Peak number I1 is significantly decreased compared to the concomitant peak in Figure 7a. The spontaneous deposition of Pb onto the surface of the Cu shard produced the absorbance profile shown in Figure 7c, normalized to the Pb concentrations in 7a and 7b. (The peak area is approximately twice the value that would be expected if the sample were unaltered.) In this instance only three peaks appear, the second being very large compared to the second peak of 7a, and the third appearing between peaks I11 and IV of 7a. These results were not anticipated nor can they be readily explained in this introductory study. Nevertheless, the enhancement of selected peaks in the profile suggests the possibility of using GFAAS with vacuum vaporization for differentiating various forms or locations of the P b within a given sample. This particular aspect has been explored by Jackson and co-workers (8, 9) using atmospheric pressure GFAAS with synthetic alumina samples for which Pb was intrcduced onto the surface and into the bulk during synthesis. While Jackson et al. were able to distinguish separate peaks even when atomization occurred under 1 atm of pressure, their work dealt with a relatively volatile analyte in a refractory matrix. In contrast, the results contained herein deal with an analyte/matrix pair with not so dissimilar volatilities.
Anal. Chem. 1988, 60, 2683-2686
The use of low-pressure vaporization for the Cu samples greatly reduced the residence time of P b within the system and promoted the definition of distinct peaks as well as the removal of the Pb from the molten Cu droplet. Present work is aimed at enhancing the surface analysis capabilities of the technique by determining the origin of these peaks. Furthermore, elucidating mechanisms for low-pressure vaporization should facilitate the application of this technique to refractory analytes and solid matrices. ACKNOWLEDGMENT We thank Vahid Majidi for his valuable assistance. LITERATURE CITED (1) L'vov, B. V. Spectrochemical Anal~eisby Atomic Absorpthm Spectrometry; Hllger: London, 1970.
2683
(2) Sturgeon, R. E.; Chakrabarti, C. L.; Bertels, P. C. Spectrochim. Acta, Part B 1977, 328, 257-77. (3) Sturgeon, R. E.; Chakrabarti, C. L. Rag. Anal. At. Spectrosc. 1978, 1 , 5.
(4) Hoenig, M.; Vanderstappen, R.; van Hoeyweghen, P. Analusls 1978, 6 , 433. (5) Fazakas, J. Spectrochim. Acta, Part B 1982. 378, 921-7. (6) Donega, H. M.;Burgess, T. E. Anal. Chem. 1970, 42, 1521-4. (7) Wendl, W.; Mirller-Vogt, 0. Spectrochim. Acta, Part B 1984, 398, 237-42. (8) Karwowska, R.; Jackson, K. W. Spectrochim. Acta, Part B 1986, 418. 947-57. (9) Hinds, M. W.; Jackson, K. W. J . Anal. At. Spectrom. 1987, 2 , 441-5.
RECEIVED for review February 25,1988. Accepted September 19,1988. Financial support for this project was provided by National Science Foundation Grant CHE-8704024 and Welch Foundation Grant F-1108.
Micromolar Protein Concentrations and Metalloprotein Stoichiometries Obtained by Inductively Coupled Plasma Atomic Emission Spectrometric Determination of Sulfur Jacob Bongers, Cynthia D. Walton, a n d David E. Richardson* Department of Chemistry, University of Florida, Gainesville, Florida 3261 1 J o h n U. Bell Department of Physiological Sciences, University of Florida, Gainesville, Florida 3261 1
The concentrations of several mlcromolar sdutlons of protelns with known sulfur contents were determlned by Inductively coupled plasma atomic emlsslon spectrometry (ICP-AES) of sulfur In the vacuum ultraviolet. These values are compared to concentrations obtalned by using spectrophotometric measurements and published speclflc and mdar absorptlvltles based on varlous conventlonal methods of protein determlnation. The two sets of values are In close agreement, Indlcatlng that ICP-AES of sulfur Is an accurate means of determlnlng mlcrogram quantltles of protelns. Standard devlatlons are wlthln 2% of the mean values obtalned for data sets of SIXrepllcate measurements. Dilute buffered proteln solutions may be directly pumped Into the nebullzer; sample dlgestlon and other speclal sample preparatlons are not necessary. I t Is also demonstrated for several metalloprotelns that multlelement ICP-AES Is an excellent means of determlnlng stokhlometrles of bound metal Ions as both proteln and metal assays may be rapidly performed on a slngle sample.
Many investigators in biochemistry and related areas are unaware that proteins may be quantified by inductively coupled plasma atomic emission spectrometry (ICP-AES)even if the protein in question contains no metallic elements. Temperatures in an ICP torch are sufficiently high to excite many nonmetallic elements, and several analytically useful emission lines in the vacuum ultraviolet have been observed (1). Of particular interest in this study is the prominent sulfur line a t 180.73 nm, which has an estimated detection limit of 15 pg/L (2). Proteins with known contents of the sulfur-
bearing residues cysteine and methionine can thus be quantified at the microgram level by ICP-AES provided the instrument is equipped with evacuated or otherwise oxygen-free optics. ICP-AES determinations of total sulfur in tissue samples (3, 4) and the use of an ICP-AES instrument as a sulfur detector in the high-performance liquid chromatography (HPLC) of tissue extracts (5)have been reported. The purpose of this report is to demonstrate the viability of pumping micromolar purified protein solutions directly into the nebulizer of the ICP-AES instrument and determining protein concentrations based on the emission of sulfur at 180.73 nm. It is also shown for several metalloproteins that this method permits the precise stoichiometries of bound metal ions to be determined by multielement ICP-AES of both sulfur and metals. EXPERIMENTAL SECTION Preparation of Samples and Standards. Horse kidney and rat liver metallothioneins were isolated and purified by a literature method (6). Cd7metallothioneinwas prepared by adding excess CdC12to a solution of native equine metallothionein. An excess of the reducing agent dithioerythritolwas added to ensure complete reduction of cysteine sidechains. After 24 h the reaction mixture was fractionated on a Sephadex G-50 column. Horse heart cytochrome c was obtained from Sigma Chemical Co. and purified by carboxymethylcellulose ion-exchange chromatography. The other proteins used in this investigation were obtained as lyophilized powders from Sigma and used without further purification. The ribonuclease sample was dissolved in 10 mM Tris/HCl buffer at pH 7.8. All other protein samples were dissolved in 5 mM phosphate buffer at pH 7.2. All samples were concentrated
0003-2700/88/0360-2683$01.50/00 1988 American Chemical Society
