200
Anal. Chem. 1983,
(33) Yasuda, M.; Murayama, S. Specfrochim. Acta, Part B 1981, 368, 641-647. (34) Margrave, J . L., Ed. "The CharaCteriZatiOn of High Temperature Vapors"; Wiley: New York, 1967. (35) Hofmann, F. W.; Kohn, H.; Schneider, J. J . Opt. SOC.Am. 1961, 51, 508-51 1. (36) Young, C. "Tables for Calculating the Voigt Profile"; University of Michigan: Ann Arbor, MI, 1965 (ORA-05863). (37) Penkin, N. P.; Shabanova, L. N. Opt. Spectrosc. 1963, 14, 5-8. (38) Wagenaar, H. C.; Novotny, I.; de Galan, L. Soectrochim. Acta, Part B 1974, 298, 301-317. (39) van Trigt, C.; Hollander, Tj.; Alkemade, C. T. J. J . Quanf. Spectrosc. Radiat. Transfer 1965, 5 , 813-833. (40) L'vov, B. V. Specfrochim. Acta, Part B 1978, 338, 153-193.
55, 200-204 (41) Persson, J. A,; Frech, W.; Cedergren, A. Anal. Chim. Acta 1977, 92, 85-93. (42) Chakrabarti, C. L.; Hamed, H. A,; Wan, C. C.; Li, W. C.; Bertels, P. C.; Gregoire, D. C.; Lee, S.Anal. Chem. 1980, 52, 167-176. (43) L'vov. B. V.; Pelieva, L. A. Can. J . Spectrosc. 1978, 2 3 , 1-4. (44) Wall, C. D. Talanta 1977, 24, 755-757. (45) Sturgeon, R. E.; Berman, S.S.;Desaulnlers, J. A. H. Anal. Chim. Acta 1982, 134, 283-29 1.
RECEIVEDfor review May 25,1982. Accepted October 11,1982. This work was presented at the 8th FACSS Meeting, Sept 20-25, 1981, Philadelphia, PA. NRCC No. 20719.
Determination of Aflatoxins in Air Samples of Refuse-Derived Fuel by Thin-Layer Chromatography with Laser-Induced Fluorescence Spectrometric Detection Merlin K. L. Blcking,' Richard N. Knlseley," and Harry J. Svec Ames Laboratory and Department of Chemistry, I o w a State University, Ames, I o w a 5001 1
An analytlcal method Is descrlbed which allows determlnation of aflatoxins in a complex matrlx. An apparatus has been developed that quantitates fluorescent compounds on thlnlayer chromatography plates. A nltrogen laser excltatlon source produces a detectlon llmlt of 10 pg for four aflatoxlns. Aflatoxin B l has been found at levels up to 17 ppb In solld samples collected from the air at a plant whlch produces refuse-derlved fuel.
Mycotoxins have been defined as toxic compounds produced by fungal contaminants, with the toxicity syndromes resulting from ingestion of such contaminants termed mycotoxicoses. The fungal species (Aspergillus,Penicillium, and Fusarium (1))have been involved in reported mycotoxicoses. Mycotoxins were fiist identified as the cause of the mysterious "Turkey-X" disease in England, where a peanut meal contaminated with Aspergillus flaous contained a toxic component (2). The toxic compounds subsequently isolated were called, appropriately, aflatoxins, and many other mycotoxins have since been isolated and identified. The Ames Solid Waste Recovery System (ASWRS) accepts trash and garbage and processes it into refuse-derived fuel (RDF), which is used in the municipal power plant. This processing generates much airborne material of unknown danger to the public health. Studies of airborne bacteria levels in the processing plant have been found to be orders of magnitude above ambient locations outside the plant (3). Fungal levels are also higher than ambient levels. This work has been undertaken to evaluate the hazard to plant personnel posed by the high levels of fungi. Table I presents a summary of data obtained at the ASWRS during the period July-September 1978. These fungal levels are 2 to 3 orders of magnitude above those at locations outside the plant. Such data are indicative of a possible mycotoxin contamination problem. Initially, existing analytical techPresent address: Department of Chemistry, State University of New York-Buffalo, Buffalo, NY 14214.
Table I. Occurrence of Fungi at the Ames Solid Waste Recovery System, July-September 1978 CFU/m3 a
Aspergillus flavus A. fumigatus A. niger A. ochraceus A. terreus A . niveus Penicillium sppb Fusarium spp
22000
53000
13000 360 600
a Colony forming units per cubic meter. species.
240 4800 1200
Unidentified -~
niques ( 4 ) were found to be inadequate to identify and quantify the mycotoxins which were present. It was necessary to initiate new procedures for the determination of mycotoxins in a complex, highly fluorescent matrix. Previous work (5) has demonstrated that some interfering components in extracts from an RDF processing area may be effectively removed. This separation used simultaneous exclusion and partition mechanisms on a column packed with Styragel. This has been combined with other modified techniques into a new analytical method for aflatoxins in air samples collected a t an RDF processing plant. Detection limits in the low picogram range have been reported for aflatoxins on thin-layer chromatography (TLC) plates (6). However, the techniques were somewhat specialized and the method has been applied only to standards, not real samples. In this study, a pulsed nitrogen laser was chosen as the excitation source for the determination of the highly fluorescent aflatoxins. Previous work (7) has measured the laser-induced fluorescence of aflatoxins directly on a TLC plate, with a detection limit of 0.2 ng. This was the only reported application of laser light to TLC analysis of aflatoxins. However, the supporting equipment was crude and unsuitable for quantitative work. In the present study, techniques have been developed to allow removal of interfering components in the matrix without
0003-2700/83/0355-0200$01.50/00 1983 American Chemlcal Society
ANALYTICAL CHEMISTRY, VOL. 55, NO. 2, FEBRUARY 1983
loss of aflatoxino present at low levels. With t h e use of an internalstandard,quantitation on TLC plates by laser-induced fluorescence was possible at levels near 1 ppb.
wGm
EXPERIMENTAL SECTION General Considerations. All glassware in contact in contact with any mycotoxins was cleaned in a KOH/alcohol bath. Clean glassware was rinsed with dilute acid, distilled water, and acetone and then air-dried. No carry-over problems were observed. Purification of Solvents. All solvents were filtered through a fine glass frit (4--5 pm) or 0.5-rm Teflon filter. The following were reagent grade and used as received formic acid, ethyl ether, trifluoroacetic acid. Pesticide-grade hexane was used as received. Chloroform was purified as previously reported (5) and then distilled to remove a blue fluorescing impurity. Acetonitrile (2 L) was refluxed over KMnO, for 24 h and distilled under nitrogen through a Vigreux column. The first 10%) of the distillate was discarded. The remaining distillate was passed through a column containing 100 g of acidic alumina (activity Grade I). The purified solvent showed a UV cutoff below 190 nm. Acetone (2 L) contained a blue fluorescing impurity and was refluxed several hours over KMnO,. This was followed by dis. tillation through a Vigreux column under nitrogen. Several small pieces of Na metal were dissolved in 2 L of methanol, followed by several grams of iodine. After being refluxed for several hours the solvent was distilled (8). Several small pieces of Na metal were dissolved in isoamyl alcohol and the solvent was distilled immediately. This removed an unidentified yellow impurity. Tetrahydrofuran (THF) was purified by refluxing for 1h over CaH,, followed by distillation under nitr0ge.n (8). Pyridine was dried over KOH overnight, followed by distillation over barium oxide (8). Acetyl chloride was distilled under nitrogen. Mycotoxins. The four aflatoxins, B1,B2, G1, and G2, were obtained from Aldrich Chemical Co. Stock solutions (10 ppm) of aflatoxins were prepared in methanol. All dilutions were prepared with a CHC1, stock solution to which the internal standard had been added a t a level of approximately 200 pg/5 pL. Standards were applied to the TLC plate as 5-pL aliquots of standard solutions. Calibration curves were obtained by plotting the area ratio (peak/internal standard) vs. amount applied to the plate. I n t e r n a l Standard. The internal standard chosen was the acetate derivative of aflatoKin G2a. One milliliter of a 10 ppm by weight solution of aflatoxin G2a (Sigma Chemicals) in CHC1, was cooled t o 0 “C. One drop of dry pyridine was added. Two drops of acetyl chloride were added during a 20-min period. The solvent was removed as quickly as possible and the residue was dissolved in CHCl:, and purified by preparatory TLC in 10% CH3CN/CHC13 The derivative, identified by its RF value (slightly less than that of G2), was scraped from the plate and eluted with 50% CH3CN/CHC13. After evaporation, the derivative was dissolved in CHC1,. Thin-Layer Chromatography. TLC plates were obtained as 20 X 20 cm plates, prescored to 5 X 20 cm (Anasil 0,TLP099, Analabs, Inc., North Haven, CT). Plates were purified by allowing them to develop in methanol overnight. After removal from the methanol, the plates)were dried in a warm vacuum desiccator for 1 h. Storage was in an atmosphere of 48% relative humidity (saturated KSCN solution). Exclusion Chromatography. The Styragel column was prepared as previously reported ( 5 ) ,with the peristaltic pump replaced by a single piston pump (Analabs, Inc.; Model A-60-S). Florisil. The design and operation of the Florisil column and switching have been discussed (9). Apparatus. The TLC fluorescence scanning system consisted of a pulsed N2 laser which emitted light at 337.1 nm. The laser output was focused to a sharp line with a cylindrical quartz lens, striking the plate a t an angle of approximately 45”. Fluorescence normal to the plate was focused onto the entrance slit of a I/,-rn monochromator, set a t 440 nm. The current produced by the photomultiplier tube was amplified and directed, through an RC time constant circuit, to a digital voltmeter, an X-Y recorder, and an integrator. A block diagram of the system is given in Figure 1.
201
4
,/>
ELECTROMETER/ INTEGRATOR
RECORDER
N2 U S E R
T L C PIATE
Figure 1. Block diagram of the TLC fluorescence scanning system.
VIF Converter
Figure 2. Block diagram of the integrator.
The TLC plate was held in a screw-driven slide assembly on the bottom of an enclosed box containing the focusing and light collecting optics. The laser was focused on the TLC plate, which was scanned manually across the beam. The screw drive also turned a 10-turn helical potentiometer. The voltage drop acros!s the potentiometer, provided by a 9-V battery, drove the X axis of the X-Y recorder, allowing scanning in either direction and permitting the operator to stop or start the scan when necessary. Figure 2 describes the integrator developed for this system. Arl optical interrupter mounted on the screw of the plate drive s i g naled the integrator every quarter turn. This turned the timer integrated circuit (IC) “on”, allowing the V/F converter and counter to collect data from the voltage output of the amplifier The length of time the IC was ”on” was determined by the RCconstant of the IC, which could be varied with a 10-turn potentiometer. The result was a histogram of data points, the sum of which was proportional to the area of the peak. Approximately 25-30 points were collected for an average spot on a TLC plate. The X axis for this system was position instead of time. Since the integrator only “counted” when “triggered”, the scan could be stopped a t any position (time) without loss of data. The pulse rate of the laser (20 Hz) and the noise made it difficult to maintain a proper base line with the offset on the amplifier. A 500-ms time constant was placed in the circuit after the amplifier, influenciing the pen on the X-Y recorder, digital voltmeter, and the integrator. Stable base lines were obtained, but the scan rate was reduced to 0.5-1 min/peak. Typical reproducibility values (percent relative standard deviation) for a set of developed spots were 1.7,1.5,2.7, and 1.6% for aflatoxins B1,B2, G1, and G2, respectively. These reproducibility values included the fact that photodecomposition decreased the peak area after each scan. Actual reproducibility values (Le., not including decomposition) should be lower. The Analytical Method for Aflatoxins in Refuse-Derived Fuel. An air sample waEi obtained by collection on a quartz fiber filter (2500 QAS filter from Pallflex Products, Putnam, CT) with a 99.9% collection efficiency for 0.3-pm particles. The sample was scraped from the filter and mixed thoroughly and a weighed aliquot removed. The sample (2.0 g) was placed in a glass-stoppered flask with 50 mL of methanol and $shakenon a wrist-action shaker for 1h. The sample was removed by filtration. The methanol was removed with a rotary evaporator and the residue suspended in 30% CH3CN/CHC13. After being thoroughly mixed, the solution was filtered to remove particulates larger than 5 pm and diluted to 10 mL. The sample was injected onto the Styragel column, which used a 30% CH,CN/CHC13 mobile phase at 1.00mL/min. The fraction containing the aflatoxins was diverted to the Florisil column. The elution time of the aflatoxins had been determined previously
202
ANALYTICAL CHEMISTRY, VOL. 55, NO. 2, FEBRUARY 1983
Figure 3. Typical fluorescence scans of aflatoxins on developed TLC plates. Direction of development is to the left. The peaks represent (left to right), B1, 82, G1, G2: (a) 1 ng each, (b) 100 pg each, (c) 10 pg each.
by injection of standards and was usually 11-14 min after injection. After the fraction had been trapped on Florisil, interfering components were removed by injection of 1.2 mL of 20% methanol/CHC13through the Florisil. Aflatoxins were eluted into a 10-mL pear-shaped flask with 5.0 mL of 4% HzO/acetone, and the solvent removed with a rotary evaporator. The sample was dissolved in 100 HLof 1.5% acetone/CHCl, and then 50 FL of hexane was added. A 5-hL aliquot of the internal standard was added and the entire solution applied to a TLC plate with the TLC sample applicator (IO). The spot was predeveloped in a 30% ether/acetonitrile solution and dried with the applicator. The plate was developed 1 h in formic acid/isoamyl alcohol/hexane/CHC13 (1.6 + 4.0 + 5.0 + 89.4). After preliminary drying in the applicator, the plate was dried in a vacuum desiccator for 30 min. The plate was placed in position and scanned across the focused laser. The amplifier offset and range switches were used to put the background “on scale“, and the output was monitored with an X-Y recorder. Background correction was performed manually and background area subtracted from the recorded integral. From the area ratio of peak to internal standard, the amount of aflatoxin present was determined by using the calibration curves. To prevent confusion due to variations in TLC development, it was necessary to develop one set of standards only on each plate. By scanning the standards frst, one could determine the ”location” of aflatoxins in real samples. Identical RF values for sample and standard provided additional confirmation of aflatoxins.
RESULTS AND DISCUSSION Figure 3 represents typical scans of developed TLC plates at three different concentrations of aflatoxins. The direction of development was to the left in the figure. The individual peaks, left to right, were B1, B2, G1, and G2. Note that 10 pg appeared to be the detection limit for this system. At the amplifier sensitivity used to detect 10 pg, plate irregularities became the limiting factor. Also, photodecomposition occurred to such an extent that, below 100 pg, only one scan across the spots was allowed. Subsequent scans showed significant (i-e., >25%) loss. In Figure 3a, the base line was easily adjusted to zero with the amplifier offset. The resolution was sufficient so that the integration was stopped or started at the valley between peaks. However, as shown in Figure 3b,c, below 500 pg the change in base line in the region of interest was too great, relative to peak heights, to allow proper use of the offset. Figure 3b,c illustrates the method of background correction used. The offset was adjusted so that the lower background level, usually a t a lower RF than that of G2, had a positive value. The stop/start points for the integration of each peak are indicated by the vertical lines. The horizontal line representing “zero” is also shown. The area of the resulting trapezoid was then measured. In a separate experiment, it was determined that the conversion factor for this system was
Flgure 4. Quenching effects for RDF samples on TLC plates: (A) fluorescence scan of an RDF sample developed on a TLC plate, spiked with aflatoxlns and internal standard: (B) identical levels of aflatoxin standards and internal standard developed on the same plate. 0
;
1.0
4
a 4
w
I
+/
/
a 4 0
0
-1.0 O
L
1
log
2
CONC.
3
4
(pg)
Flgure 5. log (peak arealinternel standard) ratio vs. log concentration (pg) calibration curve for aflatoxin B1.
4.15 integrator areas/mm2. This number was then subtracted from the area obtained from the scan to obtain the peak area. All previous steps in the analytical scheme have shown essentially quantitative recoveries. However, when 500 pg each of the aflatoxins was added to a real sample just before application to the plate, a 50-70% recovery was obtained. The recovery did not depend on the concentration of aflatoxins added to the plate and dilution of the background did not increase recovery. No specific chemical interferences could be identified and the emission spectra did not change in the presence of the matrix. This indicated an absorption type of quenching by reducing the incident intensity. Other surface phenomena may also be responsible. Such fluorescence losses were corrected through the use of an internal standard. For this study, it was necessary to synthesize an internal standard which exhibited the appropriate absorption, fluorescence,and chromatographic retention characteristics. The compound chosen was the acetate derivative of the alcohol, aflatoxin G2a. It was formed by esterification of G2a with acetyl chloride. Figure 4 illustrates results with the internal standard. The lower curve is a TLC scan of developed standards. The upper curve represents the same level of aflatoxins and internal standard added to a real sample. Although a difference in peak heights was expected, the ratio between the area (or height) of the internal standard and the aflatoxins remained constant. The logarithmic calibration curve for aflatoxin B1 is presented in Figure 5 . The concentration range was 10 pg to 5000 pg of aflatoxin standard applied to the TLC plate.
ANALYTICAL CHEMISTRY, VOL. 55, NO. 2, FEBRUARY 1983
203
w Figure 6. Fluorescence scan of a developed plate showing 17 ppb
aflatoxin E1 in RDF-2. The percent relative standard deviation ranged from less than 5% at 200 pg to sllightly less than 30% at 10 pg. As expected the percent relative standard deviation was less at intermediate concentrations, where the aflatoxins and internal standard were present in similar amounts. In Figure 5, the data points showed good correlation with a second-order polynomial Nonlinearity at high concentration could indicate selc-quenching or electronic saturation. Theoretical discussions (1I ) have indicated that TLC fluorescence calibration curves should show nonlinearity at higher concentrations. However, the nonlinearity was due to photomultiplier saturation since the curves became linear when the photomultiplier voltage was reduced. The data were linear a t concentrations below 250 pg. Analysis of Real Samples. An air sample (RDF-1) was collected a t the Ames Solid Waste Recovery System during the period (11/18/81 to 11/19/81). A second sample (RDF-2) was collected (3/5/82). The aflatoxin concentration in each sample was calculated by using eq 1,where X was the amount of aflatoxin found on concn (ppb) =
xv
Vinj W
(1)
the TLC plate (picograms), V the volume to which the extract was diluted (milliliters), 'Vinj the volume injected onto the Styragel column (190 MLin this case), and W the weight of the sample (grams). A standard addition experiment was performed with RDF-1. The calculated intercept, after addition of from 20 to 200 pg of aflatoxins, was 260 pg of aflatoxin B1. By use of the calibration curve and the ratio from the unspiked sample, a value of 250 pg of B1 was obtained. These concentrations represented 6.9 and 6.5 ppb B1, respectively. The standard addition curves for the other aflatoxins showed zero or near zero intercepts. Figure 6 shows a higher level of B1 present in RDF-2. This sample aliquot contained 650 pg of B1 (17 ppb). The percent relative standard deviation for the analysis of this sample was 20%. The peak appearing next to the internal standard was not G2, but an artifact in the sample matrix that could not be eliminated by the analytical method. The peak interfering with G2 was shown to be distinct from G2 in the additions experiment. A background sample (4.0 g), dust from a vacuum cleaner, was treated similarly and Figure 7 shows 120 pg of B1 (1.5 ppb) for this sample. The peak interfering with 6 2 is also present in tliiis sample. The only question remaining was whether or not the peak labeled B1 in Figure 6 actually was aflatoxin 131. At this point, the sample had undergone three chromatographic separations and a precipitation step. An interfering component must have
Flgure 7. Fluorescence scan of a developed plate showing what could be 1.5 ppb aflatoxin 81 in the blank. The arrows indicate the expected positions for the four aflatoxlns.
an identical RF value, absorb light at 337.1 nm, and fluoresce at 440 nm. For these reasons it was likely that the peak was B1. In addition, the peak shape was similar to the standards and the shape did not change in the standard additions experiment. The peak interfering with G2 was shown to be distinct from G2 in the additions experiment. Finally, an RDF sample was exposed to HzO and a small (5-10 pL) amount of trifluoroacetic acid and heated for 10 min at 40-50 "C. Thie reaction was an accepted confirmatorJy test, as it converted B1 and B2a (4). After removal of the HzO, followed by TLC development, the peak corresponding to B1 had disappeared. An identically treated standard produced the same results. Thus, the presence of B1 in the RDF sam. ples was confirmed. In conclusion, the analytical method has allowed determination of a specific compound present in low levels in EL complex matrix. The general method described here should be applicable to a variety of other matrices (9). Health Risk Assessment. It was of interest to evaluate the exposure level of workers in the plant to aflatoxin B1. An air sampler passing 1400 L min-l for 8 h would sample 6.7 X lo5 L. A total of 4.7 g was collected, giving an average level of 7.0 X lo4 g L-I of particulates in the air. Assuming 15 ppb ng L-' B1 in the (15 ng/g) B1 in the sample gave 1.9 X air. An average person inhales 6 L min-l (12) or 2880 L day-' (8 h). The total exposure to B1 was then 0.3 ng day-'. However, only approximately 25% (w/w) of the particulates were respirable (Le., reach the lungs), reducing the exposure level to 70-80 pg day-l. Presently, the long-term mutagenic effects of this type of exposure are unknown.
ACKNOWLEDGMENT The authors wish to express their appreciation to Mal Iles for design and construction of the laser and much of the supporting electronics,
Registry No. Aflatoxin B1,1162-65-8;aflatoxin BP,7220-81-7; aflatoxin G1, 1165-39-5;aflatoxin G2,7241-98-7. LITERATURE CITED Goidbiatt, L. A. J . A m . OilChem. SOC.1977, 54,302A. Sargeant, K.; Sheridan, A.; O'Keiiy, J.; Carnaghan, R. B. A. Nature (London) 1961, 792, 1096. Lembke, L. L.; Kniseiey, R . N. Appl. Env. Microbiol. 1980, 4 0 , 888. Horwitz, W., Ed. "Official Methods of Analysis of the A.O.A.C."; Association of Official Analytical Chemists: Washington, DC, 1980; Chapter 26. Bicking, M. K.; Kniseley, R. N. Anal. Chem. 1980, 52,2164. Lee, K. Y.; Poole, C. F.; Ziatkis, A. Anal. Chem. 1980, 52,837. Berman, M. R.; Zare, R. N. Anal. Chem. 1975, 47, 1200. Gordon, A. J.; Ford, R . A. "The Chemist's Companion"; Wiiey-Interscience: New York, 1972; pp 431-436. Bicking, M. K. L.; Kniseley, R. N.; Svec, H. J. J . Assoc. Off. Anal. Chem ., submitted for publication.
204
Anal. Chem. l983, 5 5 , 204-208
(IO) Bicking, M. K. L.; Kniseley, R. N.; Svec, H. J., unpublished results. (11) Hurtubise, R. J. Anal. Chem. 1977, 49, 2180. (12) Ruppei, G. “Manual of Pulmonary Function Testing”; Mosby: St. Louis,
1975; pp 13-15.
RECEIVED for review June 9, 1982. Resubmitted October 13,
1982. Accepted October 25, 1982. This research was supported by Department Of ‘Ontract w-7405Eng-82, Office of Health and Environmental Research, Pollutant Characterization and Safety Research Division (GD01-02-04-3).
Reduction of Matrix Interferences in Furnace Atomic Absorption Spectrometry J. J. Sotera,” L. C. Cristiano, M. K. Conley, and H. L. Kahn Instrumentation Laboratory Inc., Analytical Instrument Division, 1 Burtt Road, Andover, Massachusetts 0 18 10
The trace contamlnants Te, Bi, TI, and Pb were determined in complex refractory samples (high-temperature nickel alloys) by furnace atomic absorptlon. Four different techniques for sample introductlon were lnvestlgated. Conventional plpettlng onto the walls of a graphlte cylinder produced severe Interferences. These Interferences were reduced by the use of a mlcroboat and also by the use of an aerosol deposltlon system. The combinatlon of this aerosol system with the use of the microboat essentially ellminated the Interferences discussed in this study.
Analytical interferences with furnace atomizer atomic absorption determinations can be severe even for elements and matrices that are normally interference-free with flame atomic absorption. To cope with these interferences, we have employed the method of additions, matrix modification, and/or solvent extraction. These techniques, however, are timeconsuming, cumbersome, likely to add contamination, and by no means always successful. Numerous researchers have reported on analytical systems which reduce matrix interferences commonly experienced in furnace atomizer atomic absorption spectrometry. Woodriff et al. (1) reported that their constant temperature furnace was free from some matrix interferences commonly experienced with pulsed-type furnace atomizers. In 1977, L’vov (2) s u g gested that the placement of a graphite platform within a cylindrical pulsed-type furnace should approximate a constant temperature design. The platform reduces vapor-phase interferences probably because the platform temperature lags behind that of the cuvette, resulting in a delay in atomization until the gases in the furnace are a t a higher and more nearly constant temperature. Numerous researchers have given examples of the reduction of furnace interferences by the addition of the L’vov platform to the Massman-style furnace cuvette (3-7). Using a similar approach, Holcombe (8) has inserted a loose fitting “plug” into a slot milled in the top of cylindrical furnace. During the heating cycle, the plus temperature lags behind the wall temperature by as much as 900 “C. This effect permits the use of an extremely high temperature “ashing” stage to vaporize the analyte from the wall and to recondense the material on the “plug”. During the atomization cycle, the analyte is atomized from this second surface into a much hotter environment, which tends to reduce matrix interferences. As an alternate approach, Kahn e t al. (9) reported interference reduction in furnace atomic absorption with an aerosol
deposition technique. This technique deposits the sample into a hot graphite tube in the form of an aerosol mist. I t was theorized that, because the aerosol dries on contact with the graphite tube, solid-phase interferences associated with the formation of large crystals during solvent evaporation would be eliminated. In a later work Conley e t al. (10) combined the interference-reducing aerosol deposition technique with the use of a graphite insert in a graphite cuvette. The insert, a graphite microboat, was originally intended for use in solid sampling rather than for interference reduction (11-13). The work herein described is an investigation of the matrix interferences experienced in the determination of Pb, Bi, Te, and TI in high-temperature nickel alloys with the following sample introduction systems: (1) manual pipetting into a cylindrical cuvette; (2) manual pipetting into a microboat seated in a rectangular cuvette; (3) aerosol deposition into a cylindrical cuvette; and (4) aerosol deposition into a microboat seated in a rectangular cuvette. To investigate the various sample introduction techniques, we selected a particularly difficult matrix in the hope that conclusions could be extrapolated to many different types of samples. High-temperature nickel alloys are challenging because they require a corrosive acid mixture for dissolution and the nickel base cannot be selectively volatilized prior to atomization of the analyte elements.
EXPERIMENTAL SECTION Apparatus. All of the analytical work presented here was performed with an Instrumentation Laboratory IL551 Video I atomic absorption spectrophotometer,IL Visimax hollow cathode lamps, and the IL655 CTF (controlled-temperature furnace) atomizer. An IL254 FASTAC system (flame/furnace autosampling technique with automatic calibration) was used to introduce samples in the form of an aerosol mist into the furnace atomizer. A schematic of the FASTAC aerosol deposition device is shown in Figure 1. The sample is aspirated through a pneumatic nebulizer into a spray chamber (similar to a flame AA premix system) where it is converted to an aerosol. The coarser droplets run down the waste tube, and the finer ones are transferred onto the graphite cylinder which is maintained at a temperature of approximately 150 OC. The aerosol dries immediately on contact with the graphite. When microboats are used in the IL furnace atomizer, the cylindrical graphite is replaced with a rectangular cuvette (Figure 2). The microboat is placed in a slot machined in the top of the cuvette. To combine aerosol deposition with the use of a microboat, we drilled a hole 2.25 mm in diameter into the forward wall of the tube for sample introduction. As with the cylindrical cuvette, the rectangular cuvette and microboat are maintained at ap-
0003-2700/83/0355-0204$01.50/0 0 1983 American Chemical Society
