Chemical Vapor Generation: Atomic Absorption by Ag, Au, Cu, and Zn

Anal. Chem. 2000, 72, 3523-3531

Chemical Vapor Generation: Atomic Absorption by Ag, Au, Cu, and Zn Following Reduction of Aquo Ions with Sodium Tetrahydroborate(III) Aderval S. Luna,† Ralph E. Sturgeon,* and Reinaldo C. de Campos‡

Institute for National Measurement Standards, National Research Council of Canada, Ottawa, Ontario, Canada, K1A 0R9

Volatile species of Ag, Au, Cu, Zn, Cd, and As were generated at room temperature by the addition of sodium tetrahydroborate(III) to an acidified solution of the analytes. The vapor-phase species were rapidly transported to a heated quartz tube atomizer (QTA) for detection by atomic absorption spectrometry. A univariate approach was used to achieve optimized conditions and derive analytical figures of merit. The analytes are released from solution as molecular species (likely as metal hydrides) and are (partially) atomized in the QTA under nonoptimized conditions. Detection limits range from 1.8 (Zn) to 420 ng (Au). The efficiency of the generation process is estimated to be 92 ( 4% for Au. Loss of analyte during transport to the QTA was minimized through use of the minimum length of narrow-bore Teflon transfer line possible. Chemical vapor generation techniques are widely utilized for trace element detection. Of the numerous approaches available (i.e., cold vapor, halination, ethylation, propylation, oxidation, hydridization, etc.), those based on the generation of volatile hydrides enjoy greatest popularity.1,2 This stems from the ease of implementation, low reagent cost, high yield, and corresponding analytical benefits that accrue when hydride generation (HG) sample introduction is coupled with atomic spectrometric detection. Matrix separation, analyte concentration, enhanced detection power, and potential delineation of speciation lead to enhanced accuracy, precision, and reliability. The current scope of application of HG is rather limited, encompassing not only the “classical” suite of elements such as As, Sb, Se, Sn, Ge, Bi, In, Pb, Te, and Tl but also S3 and P.4 More recently, other elements have been appended to this group. Sanz-Medel et al.5 reported the generation of a volatile species of Cd (likely CdH2, which rapidly decomposes * Corresponding author: [email protected]. † Permanent address: Departamento de Quı´mica Analı´tica, Instituto de Quı´mica, Universidade do Estado do Rio de Janeiro, Brazil. ‡Permanent address: Departamento de Quı´mica, Pontifı´cia Universidade Cato´lica do Rio de Janeiro, Brazil. (1) Dedina, J.; Tsalev, D. Hydride Generation Atomic Absorption Spectrometry; John Wiley & Sons: New York, 1995; Chapter 2. (2) Dedina, J. Prog. Anal. Spectrosc. 1988, 11, 251-360. (3) Howard, A. G.; Russell, D. W. Anal. Chem. 1997, 69, 2882-2887. (4) Fujiwara, K.; Mignard, M. A.; Petrucci, G.; Smith, B. W.; Winefordner, J. D. Spectrosc. Lett. 1989, 22, 1125-1140. (5) Sanz-Medel, A.; Valde´s-Hevia y Temprano, M. C.; Bordel Garcı´a, N.; Ferna´ndez de la Campa, M. R. Anal. Chem. 1995, 67, 2216-2223. 10.1021/ac000221n CCC: $19.00 Published 2000 Am. Chem. Soc. Published on Web 07/01/2000

to Cd0(g)) by reduction of Cd(II) with sodium tetrahydroborate(III) in aqueous solutions of vesicles of didodecyldimethylammonium bromide. Sturgeon et al.6 detected a volatile species of copper by merging streams of acidified copper(II) with sodium tetrahydroborate(III), and Wang et al.7 studied the generation of Cd and Cu species using a movable reduction bed hydride generator. The transient existence of free atoms of several transition and noble metals in solution8 following their reduction with sodium tetrahydroborate(III) suggests that many more elements may be amenable to hydride generation for analytical application than earlier speculated. Moor et al.9 recently reported analytical figures of merit for detection of volatile species of Rh, Pd, Ag, and Au following their solution-phase reduction by a stream of tetrahydroborate(III) and rapid introduction into an inductively coupled plasma mass spectrometer (ICPMS). Although such compounds appear to be quite unusual, they are well-known in inorganic chemistry. The spectroscopic properties of the diatomic gaseous hydrides of many elements have been characterized since the 1960s, but their generation and study has typically been confined to high-temperature, high-pressure (hydrogen) batch reactors.10 This study was undertaken to establish a parametric optimization of the variables influencing vapor generation of Ag, Au, Cu, and Zn following batch reduction of analyte solutions with tetrahydroborate(III). Detection was effected by transporting the vapors to a heated quartz tube atomizer (QTA) mounted in an atomic absorption spectrometer (AAS). Although the reported figures of merit may not compare favorably with other techniques used for detection of these analytes, such comparison was not the aim of this study. To the best of our knowledge, this is the first report on the generation and characterization of volatile species of Ag, Au, and Zn using this methodology. EXPERIMENTAL SECTION Instrumentation. A Perkin-Elmer model AAnalyst 100 atomic absorption spectrometer fitted with an electrically heated quartz tube atomizer and deuterium background correction was used (6) Sturgeon, R. E.; Liu, J.; Boyko, V. J.; Luong, V. T. Anal. Chem. 1996, 68, 1883-1887. (7) Tian, X. D.; Zhuang, Z. X.; Chen, B.; Wang, X. R. Analyst 1998, 123, 627632. (8) Panichev, N.; Sturgeon, R. E., Anal. Chem. 1998, 70, 1670-1676. (9) Moor, C.; Lam, J. W. H.; Sturgeon, R. E. J. Anal. At. Spectrom. 2000, 15, 143-149. (10) Seto, J. Y.; Morbi, Z.; Charron, F.;, Lee, S. K.; Bernath, P. F.; Le Roy, R. J. J. Chem. Phys. 1999, 110, 1-12.

Analytical Chemistry, Vol. 72, No. 15, August 1, 2000 3523

Figure 1. FIAS manifold, generation cell and QTA arrangement. GLS, gas/liquid separator; P1, P2, peristaltic pump.

throughout this work. Perkin-Elmer hollow cathode lamps were used as line sources for all elements except As, for which a PerkinElmer electrodeless discharge lamp and system II power supply were used. The instrument was operated via a PC using PerkinElmer AA WinLab software. As this software does not include silver, gold, cadmium, copper, and zinc as elements for which hydride generation is an option, the default conditions for arsenic were used for construction of a working file and the appropriate wavelength and lamp current were set at the optimum values for each of these elements. Samples were processed using a Perkin-Elmer flow injection system model FIAS 400. A chemifold was constructed, as illustrated in Figure 1. Of the two peristaltic pumps on the FI unit, one was used to fill a 30-µL loop with the sodium tetrahydroborate(III) reductant solution during the “fill” step whereas the second 3524 Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

Table 1. FI Program Used for Chemical Vapor Generation step prefill 1 2 3 a

time (s) pump 1 (rpm)a pump 2 ( rpm)a valve read step 5 10 25 5

60 60 0 0

0 0 80 0

fill fill inject fill

read

Yellow-yellow Tygon pump tubing.

was used to supply air as the moving medium for delivery of the loop volume of reductant to the generation cell during the “inject” mode. Table 1 summarizes the FI program used for this operation. Following the inject step, the AA instrument was triggered to record the resulting transient signal, which was saved to disk for

evaluation. Tygon pump tubing was used throughout, and all other conduits consisted of poly(tetrafluoroethylene) (PTFE) tubing. The generation cell has been described earlier11 and consisted of a 1 cm diameter × 3 cm long glass tube containing a medium porosity glass frit onto which the sample aliquot was pipetted. A premixed flow of Ar and air entered the tube from below the frit and was used to effect gas/liquid separation as well as transfer of the volatile reaction products to the heated QTA. This was achieved by fitting the gas/liquid separator with a ground glass cap having an inlet for the reductant solution and an outlet for the vapor stream. Calibrated rotameters were used to establish the necessary air/Ar ratio and flow rates. Reagents. Stock solutions (1000 µg/mL, 99.99%) of As(III), Cu(II), Ag(I), Au(I), Zn(II), Cd(II), Ni(II), and Pd(II) were obtained from SCP Science (Montreal, PQ, Canada). Working standards of lower concentration were prepared by dilution of the stocks using 18 MΩ‚cm deionized, reverse osmosis water (DIW) obtained from a mixed-bed ion-exchange system (NanoPure, model D4744, Barnstead/Thermolyne, Dubuque, IA). High-purity sub-boiling distilled HCl and HNO3 were prepared in-house and used for acidification of the working standards. A 1% (m/v) solution of sodium tetrahydroborate(III) in 0.05 M NaOH was prepared daily or more frequently if required. Both reagents were purchased from Alfa Aesar (Ward Hill, MA). A 1000 mg/L stock solution of high-purity Mg(NO3)2 was prepared by dissolution of high-purity Mg in sub-boiling distilled HNO3 and stored in a precleaned screw-capped polypropylene bottle. All solutions were prepared in a clean room providing a class 10 working environment. All glassware was soaked for at least 24 h in 1 M nitric acid and rinsed with DIW before use. Compressed air and argon gas were obtained from Praxair (Mississauga, ON, Canada). Procedures. For the generation of volatile analyte species, a 10-20 µL volume of acidified sample solution was typically injected onto the frit of the gas/liquid separator using an adjustable microliter pipet, the ground glass cap was attached, an air/Ar gas mixture was supplied to the cell at a total flow rate of 40-100 mL/min, and the FIAS program (Table 1) was initiated. The instrument was optimized for the element of interest in accord with the manufacturers’ recommendations; a read delay of 1015 s was used in conjunction with a baseline offset correction (BOC) of 2 s. Both peak height and integrated absorbance of the transient signal were recorded, and the signal was stored to disk for further evaluation. A univariate optimization approach was undertaken to establish, for each element, the best conditions for volatile species generation, transport, and atomization. This necessitated examination of the effects of acid type (HCl or HNO3) and concentration, sodium tetrahydroborate(III) concentration, sample volume, transfer gas flow rate, air/Ar mixture, transfer line dimensions and materials, and QTA temperature. The need for background correction was assessed for each element as was the nature of the species transferred to the QTA. (11) Sturgeon, R. E.; Willie, S. N.; Berman, S. S. Anal. Chem. 1985, 57, 23112314.

Figure 2. Typical background-corrected transient signal obtained for the generation of Ag from a solution of 1.2 M HCl spiked with 150 ng of Ag(I).

Using optimized conditions, analytical figures of merit were established, including the following: detection limit, precision of replicate measurement, sensitivity, linear range, efficiency of generation, and major source of the blank. The QTA was routinely and periodically cleaned after every 2 weeks of operation using a dilute solution of HF. Although no deposits of any description were visible, the sensitivity for all elements would gradually deteriorate until such time as the tube was cleaned. Additionally, the blank for Cu became noticeably elevated with continued use as a consequence of deposition of metal vapors onto the tube surface which could subsequently be re-released into the gas phase by the action of hydrogen radicals introduced into the QTA during blank generation. Deposition of metal vapors onto the tube surface likely results in the creation of active sites for subsequent loss of analyte. RESULTS AND DISCUSSION Optimum conditions for the generation and detection of each of the analytes studied was undertaken using single-element solutions. Silver and gold are primarily used as the exemplary elements in the ensuing discussion to highlight the common impact of experimental variables on all analytes. The discrete nature of the sampling process resulted in the generation of transient signals which were characterized by measurement of both their peak height and integrated absorbance (peak area). Figure 2 illustrates the typical response for the generation of Ag from a solution of 1.2 M HCl spiked with 150 ng of Ag(I). The characteristics of the pulse are determined by the kinetics of generation/evolution of the gaseous species convoluted with its gas transport to, and residence time/atomization parameters within, the QTA. Typically, signals from all elements possessed half-widths on the order of 2-10 s under the conditions used and were often unsymmetrical in shape, likely due to the nonuniform delivery of the tetrahydroborate(III) reductant to the generation cell. Optimization of Generation/Detection. As is well-known with the usual suite of hydride-forming elements, generation is most efficient from acidic solution and the metals studied here present no exceptions. Figure 3 illustrates the effect of acid concentration on response from Ag. An optimum concentration of 1.2 M HCl is evident for integrated response, beyond which Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

3525

Figure 3. Effect of acid concentration on normalized peak height and area response from 200 ng of Ag: (b) HCl, integrated absorbance; (1) HNO3, integrated absorbance; (O) HCl, peak height; (3) HNO3, peak height.

sensitivity decreases. This likely arises as a consequence of the rapid consumption of the reductant at higher acidities. Relative sensitivity is 20% lower when HNO3 is used (at an optimum concentration of 1.0 M), due to the resulting oxidizing medium compromising the reduction of the metal and consuming the reductant.12 Optimum concentrations of HCl were found to be 0.6 M for Au, Cd, and Cu and 1.2 M for As and Zn. With the exception of Cd (0.75%), a 1% (m/v) solution of the reductant was selected for study in that a compromise was established between additional slight increases in signal intensity and monotonic increases in the magnitude of the blank as this reagent concentration increased. Additionally, stabilization of the reductant solution was more difficult at higher concentrations, this being particularly important in light of the relatively small volumes used for reaction and the need to minimize bubble formation in the reductant loop. As a consequence, variable reductant volumes delivered to the reaction cell impacted on the overall repeatability of the measurements in most cases, due to the contribution of the blank to the response. It is significant to emphasize that no signals were detected for Ag, Cu, Au, and Zn when relatively large volume batch reactions (>1 mL) were attempted. Rapid gas/liquid separation of the volatile product appears to be essential, suggesting that the generated analyte species are highly unstable in the aqueous phase. Earlier reports on the generation of Cu6,7 and several other metals9 support this conclusion. As such, reactions were conducted using a minimal sample volume, typically in the range of 10-20 µL. Small reaction volumes promote rapid and complete mixing of the sample and reductant and offer a high surface/ volume ratio for efficient escape of the volatile product. Approximately 0.7 mL of H2 is rapidly evolved (initial 2 s) following addition of 30 µL of reductant to the acidified analyte solution. An auxiliary source of carrier gas supplied to the bottom of the gas/liquid separator (Figure 1) enhanced the efficiency of this process. In the most severe case, for Zn, relative signal intensity decreased 75% as sample volume increased from 10 to 50 µL. Effect of Carrier Gas. Peak height and integrated response are determined by the overall efficiencies of analyte generation, (12) Risnes, A.; Lund, W. J. Anal. At. Spectrom. 1996, 11, 943-948.

3526 Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

Figure 4. Effect of Ar carrier gas flow rate on normalized integrated absorbance response from analytes: (b) 100 ng of Ag; (O) 10 ng of As; (1) 500 ng of Au; (3) 10 ng of Cd; (9) 100 ng of Cu; (0) 10 ng of Zn.

gas/liquid separation, transport, and atomization. The carrier gas flow rate and composition were found to have a significant impact on response as they influence the latter three processes. Initial gas/liquid separation is likely largely effected by the pulsed release of H2 coevolved with the analyte and may reflect the need for a small sample volume for this process to occur efficiently and rapidly. Once in the gas phase, rapid, quantitative transport of the volatile species to the atomization cell must be achieved, necessitating a compromise between minimal transport time to, and maximal residence time within, the QTA. Figure 4 shows the effect of Ar carrier gas flow rate on normalized response from all analytes. An optimum is evident for each analyte, below which little or no transport to the QTA can be achieved and above which the residence time, and possibly atomization efficiency, are compromised. Loss of analyte to the walls of the transport tubing by adsorption processes is problematic and appears to be dependent on the exposed surface area (more likely, the number of active sites at which adsorption occurs). Narrow-bore (1.64-mm-i.d.) PTFE tubing was used for all experiments. Doubling the inner diameter to 3.91 mm increased the conduit surface area 2.4-fold and resulted in a corresponding decrease in integrated response from Ag by 2.6-fold. Doubling the length of this (3.91-mm-i.d.) transport line from 23.4 to 46.8 cm caused a 7.8-fold decrease in the peak area response. This decreased response is larger than anticipated from a purely geometric increase in the surface area of the transport line. Clearly, additional factors may come into play that influence the loss process, such as a change in the laminarity or turbulence in the gas flow pattern, which may alter the efficiency of exposure of the analyte to the surface during its residence time in the transport line. Substitution of Tygon tubing of the same dimensions for PTFE attenuated the signal by 46fold, illustrating the significant impact of increased surface reactivity on the loss processes rather than simple loss of analyte due to gas-phase decomposition of unstable species during its increased residence time in the lengthened transfer line. Although these latter conclusions can be deduced from a simple consideration of the physical aspects of the analyte transport process, the composition of the carrier gas was also found to exert a substantial effect on the signal intensity due to its impact on the atomization efficiency. Welz and Guo13 com-

Figure 5. Effect of addition of air to the carrier gas stream on normalized peak area response from analytes. Optimum total gas flow (mL/min) given in parentheses: (b) 100 ng of Ag (40); (O) 10 ng of As (100); (1) 500 ng of Au (40); (3) 10 ng of Cd (60); (9) 100 ng of Cu (100); (0) 10 ng of Zn (80).

mented earlier on the need to add a small amount of oxygen (typically 1%) to the carrier gas line used for gas/liquid separation during flow injection hydride generation with the QTA. The small volume of solutions normally used with the FIA technique often contains insufficient amounts of dissolved oxygen to promote the hydrogen radical abstraction reactions necessary for efficient atomization of the hydride-forming elements.13-15 This same problem was evident in this study, and as both sample and reductant volumes are very small and the sample volume is additionally purged by Ar prior to the addition of the reductant, the problem is likely more severe. Analyte signals were significantly enhanced when air was added to the carrier gas line, as illustrated by the data in Figure 5. Gold is most noticeably affected, the signal increasing by 20-fold in the presence of an optimum amount of oxygen. As the concentration of oxygen is increased beyond the optimum, response decreases, possibly as a consequence of enhanced analyte oxide formation. At the other extreme, an air flow/total gas flow rate greater than 0.6 had to be avoided in an effort to prevent (audible) explosive combustion in the QTA when H2 was admitted during sample introduction. Effect of QTA Temperature. The optimum concentration of oxygen is also influenced by the temperature of the QTA, the latter decreasing as the former increases.14 Although this interplay was not specifically examined here, the impact of the QTA temperature on response was found to be significant. Figure 6 shows the dramatic effect of QTA temperature on signals from elements such as Cu, Au, and Ag. Irrespective of the fact that a nonthermal process of hydride atomization likely occurs in the QTA (i.e., a hydrogen radical abstraction process1,14), the residence time of the atomic vapor so produced (hence, signal response) will be influenced by the temperature of the QTA and the volatility of the metal once the hydrogen radical cloud has dissipated. Significantly higher temperatures may be required to achieve good sensitivity for Cu, Au, and Ag, likely because of their rapid adsorption or condensation onto the relatively cool walls of the QTA. Attempts to detect Ni and Pd under similar experimental (13) Welz, B.; Guo, T. Spectrochim. Acta, Part B 1992, 47, 645-658. (14) Welz, B.; Sperling, M. Atomic Absorption Spectrometry; Wiley-VCH: Weinheim, 1999. (15) Dedina, J.; Matousˇek, T. J. Anal. At. Spectrom. 2000, 15, 301-304.

Figure 6. Influence of QTA temperature on normalized integrated absorbance response from analytes. All other parameters are as summarized in Table 2: (b) 100 ng of Ag; (O) 10 ng of As; (1) 500 ng of Au; (3) 10 ng of Cd; (9) 100 ng of Cu; (0) 10 ng of Zn.

Figure 7. Transient response (no background correction) from the generation of volatile species of Au from 0.6 M HCl spiked with 1000 ng of Au(III).

conditions were unsuccessful and may be attributed to their significantly lower vapor pressures rather than poor generation efficiency.9 This is supported by the substantial depletion of these elements (78 and 66%, respectively, as measured by ICPMS) from reacted analyte solutions following addition of reductant to 200ng masses of these analytes. As the boiling points of Cd and Zn (765 and 907 °C, respectively) are lower than the maximum operating temperature of the QTA, these elements remain completely vaporized and, as such, exhibit an optimum atomization temperature which may reflect a balance achieved between the rate of atomization and the residence time of analyte within the observation volume. Sanz-Medel et al.5 concluded that reaction of Cd with tetrahydroborate(III) results in the formation of an unstable hydride which subsequently decomposes to the free metal (cold vapor) in an unheated QTA. In accordance with this, the data in Figure 6 show that Cd atomic absorption appears well below the temperature required for detection of Zn and As, reflecting the instability of this species. Lower QTA temperatures for atomization of Cd were thus not examined. Figures 7-9 show transient response from the generation of Au, Zn, and Cu, respectively. All three exhibit similar half-widths and no evidence for significant tailing. Despite the relative volatilities of these three metals (i.e., Zn . Cu > Au) there is no apparent correlation between peak shape and volatility. The multiple peaks and shoulders are likely connected with the Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

3527

Figure 8. Transient response (no background correction) from the generation of volatile species of Zn from 1.2 M HCl spiked with 10 ng of Zn(II).

Figure 9. Transient response (no background correction) from the generation of volatile species of Cu from 0.6 M HCl spiked with 200 ng of Cu(II).

generation/release process. The pulse of hydrogen released with the analyte species forms a hydrogen radical cloud (by interaction with oxygen) in the QTA, which gives rise to the measurable free atom AAS signal. Free atoms of the analyte are then rapidly removed from the observation volume by condensation on the relatively cool surfaces of the QTA once the radical cloud dissipates. Despite the fact that the QTA is clearly not the optimum atomization cell for the detection of these elements, it is adequate for preliminary study and characterization of the phenomenon of vapor generation of several transition and noble metals. Table 2 summarizes the optimum geneartion/detection conditions established for the elements examined. Species Identification. As noted earlier, the spectroscopic properties of the diatomic gaseous hydrides of many transition and noble metals have been studied for several decades.16,17 These molecular compounds have bond energies on the order of a few electronvolts,18 but investigations have been limited to use of the high-temperature King furnace in atmospheres of H2.10 Several (16) Transition Metal Hydrides; Muetterites, E. L., Ed.; Marcel Dekker: New York, 1971; pp 11-31. (17) Transition Metal Hydrides; Dedieu, A., Ed.; VCH: New York, 1992; pp 1-59. (18) Huber, K. P.; Herzberg, G. Constants of Diatomic Molecules; van Nostrand: New York, 1979.

3528

Analytical Chemistry, Vol. 72, No. 15, August 1, 2000

experiments were thus undertaken to elucidate the identity of the volatile species observed in this study. Attempts to measure the UV-visible spectra of the generated analytes utilizing microgram quantities of metals and a diode array spectrometer fitted with a 10-cm-path length quartz cell at 150 °C were unsuccessful. Although the atomic absorption lines of Hg at 253.7 nm (control element) and Cd at 228.8 nm were readily discerned, confirming the presence of Cd cold vapor, neither atomic nor molecular features in the range 200-800 nm were evident for any of the other species investigated. This is a consequence of molecular absorption coefficients generally being 3-4 orders of magnitude inferior to atomic absorption coefficients, i.e., 107 vs 104 L mol-1 cm-1 for the most intense Π-Π* transitions in molecules.19,20 Several pieces of experimental data provide indirect evidence for the formation of volatile molecular hydrides of the analytes studied. The possibility exists that the signals generated occur as a result of the convection of an aerosol cloud arising from the relatively violent reaction between the reductant and the acidified sample. The presence of an aerosol containing the analyte should produce scattering of the source radiation and detection of a concomitant background absorption signal in every case. This was not observed. Background correction was, however, required for Cd and As due to attenuation of the source beam by water vapor entering the cell. Thus, most measurements could be undertaken without the need for background correction. It is possible that no background correction is required because the density/size of the aerosol cloud is too small to give rise to a detectable signal and/or it does not exhibit absorption bands within the band-pass of the AAS instrument. In a deliberate attempt to induce aerosol formation without the attendant possibility of generation of volatile analyte species, the “generation” reaction was simulated by replacing the tetrahydroborate(III) solution with a solution of Na2CO3. Rapid release of CO2 following addition of this reagent to the acidic sample solution would be expected to transfer similar quantities of the analyte containing aerosol to the QTA. No signals were obtained for any of the analytes studied under these conditions. Addition of 150 µg of Mg(II) to the sample solution (3000 µg/mL as MgCl2) had a minimal effect on the response from Ag, attenuating the signal by only 6%. As any aerosol phase formed must reflect the composition of the generation solution, it is expected that the analyte fraction transported to the QTA under such generation conditions would be embedded in droplets containing high concentrations of Mg(II) which, during drying in the QTA, would rapidly undergo hydrolysis and convert to an aerosol of refractory MgO. This would be expected to induce a substantial attenuation in the atomic absorption response from any analyte; none was observed. The substantial effect that the presence of air has on the response from all elements suggests that the atomization process is similar to that occurring for the usual suite of hydride-forming elements, i.e., a hydrogen radical abstraction process.1,14 This implies that the volatile species of Cu, Ag, Au, and Zn are molecular hydrides. The influence of the QTA temperature on response also supports atomization of a gaseous analyte as opposed to an aerosol-containing species. At the temperature (19) Agterdenbos, J.; van Noort, J. P. NM.; Reters, F. F.; Bax, D.; Ter Heege, J. P. Spectrochim. Acta, Part B 1985, 40, 501-515. (20) Ingle, J. D.; Crouch, S. R. Spectrochemical Analysis; Prentice Hall: Englewood Cliffs, NJ, 1988; p 389.

Table 2. Optimum Conditions for Generation/Detection element, nm

HCl (M)

NaBH4 (% m/v)

Ar flow (mL/min)

air flow rate/ total gas flow rate

QTA temp (°C)

sample vol (µL)

Ag, 328.1 As, 196.7 Au, 242.8 Cd, 228.8 Cu, 324.7 Zn, 213.9

1.2 1.2 0.6 0.6 0.6 1.2

1.00 1.00 1.00 0.75 1.00 1.00

20 80 30 40 80 60

0.50 0.20 0.25 0.33 0.20 0.25

1000 900 1000 800 1000 900

10 20 10 20 10 10

Table 3. Figures of Merit Ag

As

Au

Cd

Cu

Zn

integrated response blank, ng mo, ng LOD, ng (n ) 10) precision,a % RSD upper massb g gl, %

41 ( 7 0.49 ( 0.02 7 10 (50) 300 80-95

0.5 ( 0.4 0.040 ( 0.002 0.5 20 (4) 20 ndc