Continuous Wave Diode Laser Total Solid Sample Vaporization For Biological Sample Analysis Lead in Bovine Liver

Transcription

1 MICROCHEMICAL JOURNAL 54, (1996) ARTICLE NO Continuous Wave Diode Laser Total Solid Sample Vaporization For Biological Sample Analysis Lead in Bovine Liver JACK X. ZHOU, SUH-JEN JANE TSAI, 1 XIANDENG HOU, KARL X. YANG, AND ROBERT G. MICHEL 2 Department of Chemistry, University of Connecticut, 215 Glenbrook Road, Storrs, Connecticut Received February 16, 1996; accepted March 1, 1996 The potential of a 930 nm, 50 W continuous wave diode laser array as an efficient and compact solid sample vaporization tool was investigated. It was found that the power density of the focused laser output was not sufficient for ablation of metals, but vaporization of dry bovine liver powder was feasible. For convenience, a graphite furnace atomic absorption spectrometer was utilized to test the analytical performance of the setup via furnace impaction/deposition. The direct deposition of the vaporized materials onto the inner wall of a graphite tube without a transfer tube was shown to have better precision than deposition in a hot furnace through a transfer tube. Theoretical calculations and the experimental relationship between gas flow rate and particle collection efficiency are reported. Experimental data obtained with two types of ablation cells, straight through and tangential flow, indicated that the flow pattern inside the cell did not play a major role in terms of analytical performance. Aqueous standard calibration with a matrix modifier was found to be satisfactory. Determination of lead in bovine liver (SRM1577a) was performed accurately with the diode laser vaporization deposition setup. A lead mass detection limit of about 0.1 ng was achieved Academic Press, Inc. INTRODUCTION Solid sampling is known to have many advantages (1, 2) over conventional sample preparation in atomic spectroscopy, including reduced sample preparation time, avoidance of sample contamination during digestion and dilution, and minimization of loss of analyte during pretreatment. Laser ablation, as one of many solid sample introduction methods, has shown potential in solid sampling applications. Advantages of the use of a laser to ablate or vaporize small amounts of material include the ability to handle microgram masses of material, remote sampling potential, and surface analysis capability. A recent article has reviewed laser ablation and its applications in analytical spectroscopy (3). Most of the laser ablation work has been done with pulsed lasers (4) that can provide the high peak energy and high power densities that are necessary to reach the ablation threshold. In the plume generated by the laser, atoms and molecules have been spectroscopically observed by emission, absorption, or fluorescence. A variety of pulsed lasers have been used in laser ablation to do solid sampling. Some examples 1 On leave from Dept. of Applied Chemistry, Providence University, 200 Chung Chi Rd., Shalu, Taichung Hsien, Taiwan, R. O. C. 2 To whom correspondence should be addressed X/96 $18.00 Copyright 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

2 112 ZHOU ET AL. of the lasers that have been used include the carbon dioxide laser by Kurniawan et al. (5), the Q-switched ruby laser by Kwong et al. (6) and Richner et al. (7), the nitrogen laser by Kagawa et al. (8), the dye laser by Lewis et al. (9), the argon fluoride laser by Sneddon and co-workers (10 12), the xenon chloride laser by Pang et al. (13), and the Q-switched Nd:YAG laser by Sdorra et al. (14) and Kuzuya et al. (15). By use of these lasers, studies of the effects of laser intensities (10, 16), laser wavelengths (14, 17), and buffer gases (15 17) have been reported. Techniques such as averaging, reference normalization (7, 13), and internal standardization (18) have been investigated by some of the authors to improve analytical precision. In addition to direct observation of laser induced plumes above sample surfaces, ablated materials can also be swept by a carrier gas into a separate detection chamber. With a pulsed YAG laser as the ablation laser source, Arrowsmith et al. (19) compared different ablation cell designs for the efficient entrainment and transport of ablated materials. They found that entrainment efficiency was affected by the cell design, and transport efficiency was controlled by the transfer tubes. Kantor et al. (20, 21) designed a nebulization device with which laser generated aerosols were transported by nebulization into a flame and detected by atomic absorption. Experiments by Wennrich et al. (22, 23) employed a pulsed ruby laser with a graphite furnace atomic absorption spectrometer. Laser ablated material was deposited onto the inner wall of a graphite tube at an elevated temperature by argon as the carrier gas. A conventional graphite furnace heating regime was then followed. The authors investigated the influence of deposition temperature and carrier gas flow rate on the analytical signals. An optimum deposition temperature of about 300 K and optimum argon flow rate between 5 and 15 ml s 01 were found. Despite the advantages of pulsed laser ablation, several problems exist, including poor pulse-to-pulse reproducibility, sample homogeneity problems, and limited detection power due to the tiny amount of sample material ablated. Internal standardization (18) and reference normalization (7, 13) have been used to compensate for laser pulseto-pulse variation, but in some cases internal standardization failed to provide good results (14). Continuous wave (CW) lasers in this sense may offer an alternative as ablation sources. Long laser sample interaction time and stable output are the two possible advantages over pulsed laser ablation systems. A main concern with the application of CW lasers is their low brightness. For many solids, a power density in the range from 10 4 to 10 9 Wcm 02 is required for evaporation (24). A continuous wave laser vaporization system was demonstrated by Su et al. (25). In their work, a 50 W CW YAG laser was used at 1064 nm. Powdered samples were vaporized and material was carried by an argon gas stream into an ICP. A spiral flow type of cell was designed by the authors to compare with a straight through cell. It was shown that the spiral flow cell gave improved transport efficiency and analytical performance. In practical spectroscopic applications, CW diode lasers may be a perfect choice for field use instruments that take advantage of their compactness and light weight. Diode lasers were originally developed with output in the near-ir for use in telecommunications and data processing. When combined in the format of an array, such diode lasers are capable of high power output. CW diode laser arrays with power output as high as 100 W are available now. With appropriate focusing, a power density of over 10 5 Wcm 02 is possible. Diode arrays also have the advantages of low operating

3 CONTINUOUS WAVE DIODE LASER SAMPLE VAPORIZATION 113 voltage, small size, high efficiency in conversion from electricity to light, and long lifetime of operation. Also, high power YAG lasers with diode laser pumping have been demonstrated to have advantages similar to those offered by a diode laser. One example is a YAG laser at Stanford University pumped by 25 fiber-coupled diode lasers to give a coherent 40 W CW output in a TEM 00 mode (26). Recently, a very promising oscillator amplifier diode laser has been reported (27) that can produce a CW, coherent, diffraction limited output. Estimated power density with the laser beam can be about Wcm 02. Fiber coupled diode lasers are also very attractive for laser ablation/vaporization. Such lasers are likely to be extremely flexible for use in portable ablation instruments. With a fiber of a typical diameter of several hundred micrometers, a very small spot size can be produced with an appropriate focusing system to deliver high laser power density. High power diode laser arrays have been demonstrated by researchers at Spectra Diode Labs (28) and McDonnell Douglas (29) and are commercially available. Applications with high power diode laser arrays in material processing (30), and medical applications (31) have been reported. However, to the knowledge of the present authors, no publication has shown the potential of a laser diode array as a tool for analysis by laser vaporization/ablation. Solid sampling for direct determination of lead and other trace elements in biological samples is important in meat inspection (32), residue control (33), and standard reference material production (34 36). Klein et al. (33) attempted to establish a practical monitoring system for chemical contamination in agriculture. In their project, dairy cows were considered as indicators of environmental contamination. The key to this project was to quickly determine the distribution of chemical contaminants in several bovine tissues. Solid sampling atomic absorption was used by the authors to determine lead and cadmium in bovine liver and kidney. In this article, the use is reported of a compact CW diode laser array for total biological sample vaporization. The driving force behind the project was to develop a field use analytical instrument. For this purpose an efficient and compact solid sample introduction technique becomes critical. For the sake of convenience, a laboratory, table top, graphite furnace atomic absorption spectrometer (GF-AAS) was utilized initially to evaluate the analytical performance of the ablation setup. Lead determination in bovine liver was performed with the laser vaporization-aas system. The power density of the focused laser output was not sufficient for ablation of metals, but vaporization of dry bovine liver powder was feasible. EXPERIMENTAL The experiments were conducted with a diode laser array and a graphite furnace atomic absorption spectrometer. Diode Laser Array A prototype diode laser array, model ISL-50 from McDonnell Douglas (McDonnell Douglas Aerospace, St. Louis, MO), was used for solid sample vaporization. The laser was a 2-D array made up of over 320 independent lasing elements and linearly polarized at 930 nm with a linewidth of 5 nm full-width-at-half-maximum (FWHM). Microlenses were contained in front of the diodes to collimate the output beam. Typical beam transverse divergence and lateral divergence were 0.23 and 0.5, respecah02$$ :54:39 mica AP: MICROCHEM

4 114 ZHOU ET AL. FIG. 1. Schematic diagrams of ablation cell geometries. (a) A commonly used straight through cell where carrier gas flowed horizontally across the cell, and (b) a tangential flow cell in which tangential flow carrier gas was introduced. Both cells were made of Pyrex glass. Teflon flanges were used to hold a quartz focusing lens and provided seal for the assembly. tively. The output aperture was 2 cm by 2 cm. The ISL-50 could be operated in either CW or modulated mode. When operating CW, a maximum power of 50 W was provided. The output power could be continuously varied via control of the current. In order to cool the diode laser array system, a refrigerated water recirculator was used (Model CFT-33, Neslab Instruments, Porismouth, NH). The cooling water temperature was regulated at 20 C and the laser was operated in CW mode. Atomic Absorption The atomic absorption measurements were performed with a Perkin Elmer model Zeeman/5000 GF-AAS instrument and an HGA-500 atomizer (Perkin Elmer, Norwalk, CT). The integrated absorbance of the atomic absorption signal was measured and denoted as A i. The instrument was operated with Zeeman background correction. Ablation Cells Two types of ablation cells were constructed for a comparison of their analytical performance. These were a commonly used straight through cell and a tangential flow cell as shown in Fig. 1. Both cells had cylindrical geometry and were made of Pyrex glass. Teflon flanges with o-ring grooves were used at both ends of the cells as depicted in Fig. 2a. A Quartz bi-convex lens was glued on the top flange to provide a seal and focusing. Seated onto the bottom flange was a graphite sample holder made from POCO Graphite AXF-5Q1 (POCO Graphite, Decatur, TX)) rod stock, of a height determined by the focal length of the focusing system. Samples A powdered standard reference material of bovine liver (SRM1577a) from the National Institute of Standards and Technology (Gaithersburg, MD) was used. The certified value for lead content in this powder was { mg g 01.

5 CONTINUOUS WAVE DIODE LASER SAMPLE VAPORIZATION 115 FIG. 2. Block diagrams of the laser diode ablation deposition system. (a) Off-line deposition in which a graphite tube at room temperature was fitted tightly onto a plastic pipette tip and the vaporized materials were carried by argon gas to impact onto the inner wall of the tube; (b) on-line deposition in which vaporized materials were transported by argon gas through a transfer tube and a glass Pasteur pipette to a graphite furnace atomic absorption spectrometer held at an elevated temperature. In calibration procedures, when a matrix modifier was used, it consisted of 0.2 mg of diammonium hydrogen phosphate (Grade 1, Johnson Matthey Chemicals Ltd.) and 0.01 mg magnesium nitrate (Speciality Products, %).

6 116 ZHOU ET AL. TABLE 1 AAS Furnace Programs Used in the Experiment On-line deposition Off-line deposition Step Temperature ( C) Ramp (s) Hold (s) Internal flow (ml min 01 ) Record (s) Read (s) Experimental Procedures Dry bovine liver sample (SRM1577a) was weighed with an analytical balance to a precision of {0.01 mg, and was placed into the graphite sample holder inside the ablation cell followed by assembly of the cell with its flanges. Figure 2 shows block diagrams of the vaporization deposition setup for the ablation/vaporization system. Laser vaporized materials were deposited onto the inner wall of a graphite tube either on-line in a GF-AAS instrument at an elevated temperature, or off-line at room temperature for off-line deposition experiments. In both experiments, argon gas was used as a carrier gas. As shown in Fig. 2a, a graphite tube was placed tightly onto a clean plastic pipette tip (Fisherbrand reference tip, Fisher Scientific, Pittsburgh, PA) that was connected to the ablation cell for room temperature off-line deposition. The ISL-50 diode laser array was fired at maximum power for a fixed period of time to vaporize the sample for deposition in the graphite tube. Vaporization typically lasted about 3 min which was determined by the amount of sample. The graphite tube was then transferred to the GF-AAS and analysis followed. In on-line deposition as depicted in Fig. 2b, a borosilicate glass nozzle (Pasteur pipette, Fisher Scientific, Pittsburgh, PA) was used, which was linked to the ablation cell through a polypropylene tube of a length of about 40 cm. The nozzle was held with its tip in the sampling hole of the graphite tube. The nozzle did not touch the wall of the hole. During the deposition period, the graphite tube was maintained at a preprogrammed, elevated temperature, and the internal argon flow on the GF-AAS was turned off. All GF-AAS measurements were done with the furnace programs as listed in Table 1 and at an analytical absorption line of nm. RESULTS AND DISCUSSION Focusing Considerations Diode lasers are known to have large divergence in their beams. For a given diode laser, the choice of a suitable focusing element becomes critical for ablation/ vaporization purposes. A measure of the diode laser source intensity and beam divergence, by use of the small angle approximation, can be expressed as brightness L (W cm 02 sr 01 ) L Å P Power density Å, (1) AV V

7 CONTINUOUS WAVE DIODE LASER SAMPLE VAPORIZATION 117 where P is the beam power (W), A is the beam cross section (cm 2 ), and V (sr) is the solid angle of the beam. The ISL-50 unit contains microlenses in front of each individual diode to collimate the output beam. Ideally, good collimation of the output beam with a lateral divergence of about 8 mrad and transverse divergence of 3 mrad can be achieved. This beam quality corresponds to a brightness of ú0.25 MW cm 02 sr 01. Theoretically, this implies that focused laser spots with greater than 10 5 Wcm 02 power density are feasible with a simple focusing lens. A spot size S (mm) of a beam of a divergence of u (mrad) focused with a focal length of f (mm) can be estimated by the equation S Å uf. (2) If a 10 cm focal length lens is used, with a laser lateral divergence of 8 mrad and transverse divergence of 3 mrad, a spot size of 800 mm by 300 mm can be expected in the diffraction limited case with Eq. (2). This corresponds to a power density of 10 4 Wcm 02. A similar calculation will yield a power density of about 10 5 Wcm 02 for a lens with 4 cm focal length. These power density levels are sufficient for biological solid sample vaporization. Figure 3 shows experimental beam profiles of the ISL-50 laser, measured with a linear CCD array detector. Figures 3a and 3b are the far field beam cross section distributions obtained at the exit aperture of the laser. The lateral beam FWHM was estimated to be about 20 mm and transverse beam FWHM of 7 mm. A 10 cm focal length bi-convex lens was initially used for focusing. However, a very slow evaporation rate of the bovine liver sample was observed with the lens. This was due to a poor spot size formed with the 10 cm lens as shown in Figs. 3c and 3d. A two spot pattern also was seen on the focal plane. With the beam dimensions shown in Figs. 3c and 3d, the estimated power density, with the main spot, was only about Wcm 02. At this power density level, vaporization would not have been expected. A 4 cm focal length bi-convex lens was used instead. Figures 3e and 3f are the corresponding beam cross section profiles measured with the CCD array. These results showed a lateral FWHM of about 1.8 mm with a transverse FWHM of about 0.8 mm. The calculated power density approached Wcm 02. From our experiments, this proved to be adequate to vaporize the bovine liver. The difference between calculated and experimental spot sizes resulted from many factors: First, the laser may have been used with excessive current before it was acquired by the present authors. Second, severe aberrations are associated with the use of a spherical lens, when used with an input beam of large aperture. Third, the prototype unit had a relatively poor beam quality associated with it. Probably the diodes had degraded with time, and damage of individual diodes was probably responsible for the degraded beam quality. Partially burned emitters would result in an output beam with multiple spots. Consequently, a change in the emitter beam divergence would occur that would influence the output beam collimation. A high quality laser beam, with proper care in operation, and a carefully designed focusing mechanism are the keys to the success of a diode laser ablation/vaporization system. In a situation where a smaller and cleaner spot is desired, an aspherical lens should be considered for focusing. This

8 118 ZHOU ET AL. FIG. 3. The ISL-50 diode laser beam lateral and transverse cross section distributions measured with a linear CCD array at: (a and b) the laser exit aperture; (c and d) the focal plane of a 10 cm bi-convex lens; (e and f) the focal plane of a4cmbi-convex lens. The full width at half maximum for the lateral and transverse cross sections were about 20 and 8.0 mm for (a) and (b); 4.2 and 1.9 mm for (c) and (d); and 1.8 and 0.8 mm for (e) and (f), respectively. would significantly reduce the spherical aberrations that dominate with a low f number lens, and provide a higher laser power density. On-Line and Off-Line Deposition Lead recovery was used as one of the criteria to compare on-line deposition with off-line deposition. Standard aqueous lead solutions were used without matrix modifier as the reference for calculation of lead recovery, although matrix modifier was used in the section on analytical performance discussed later. The recovery was calculated from the ratio of the signal from laser-vaporized liver with the signal from the same mass of aqueous standard vaporized normally in the furnace. For on-line deposition, it was found that the lead recovery peaked at about 300 C. This result agrees with the data published by Chamsaz et al. (37). The typical lead

9 CONTINUOUS WAVE DIODE LASER SAMPLE VAPORIZATION 119 TABLE 2 Lead Recovery a Data Obtained with On-Line Furnace Deposition Amount of lead Sample # (ng) vaporized Recovery (%) a { RSD Å 49% a The recovery was calculated with the signal from laser vaporized bovine liver sample divided by the signal from aqueous standard in graphite furnace. No matrix modifier was used with the aqueous standard solution. recovery calculated at the temperature was about 74%. Taking into account that matrix modifier would increase the aqueous standard signal size by about 10%, a lead recovery of about 66% would be expected with matrix modifier added to the aqueous standards. The experiments performed at this temperature failed to give good precision. Table 2 shows some of the lead recovery data collected with the on-line deposition method. The reasons for the poor precision may have been the following: The transfer tube and glass nozzle may have introduced significant contamination. The temperature gradient along the tube may have caused adsorption/dilution of the vaporized particle flow. Figure 4 shows experimental results from a blank ablation. After total vaporization of a bovine liver sample, a clean sample holder was used to perform blank ablation under the same conditions as a sample ablation. Strong residual lead absorbance was found, that changed as a function of the number of blanks (Fig. 4). This evidence supports the adsorption/ contamination hypothesis, as the clean sample holder eliminated the possibility of a residual source from the incomplete vaporization of the previous sample run. The ablated particles, adsorbed to the wall of the transfer tube from the previous sample ablation, were believed to be flushed out by hot argon flow during subsequent blank ablations. After three blank ablations, the signal damped to below 5% of the original level. A thermally controlled transfer tube might be useful to minimize the adsorption due to the temperature gradient. In contrast to on-line deposition, off-line deposition at room temperature showed low recovery data. Typical recovery was around 15%, but the elimination of the transfer tubing significantly reduced the risk of contamination. Furthermore, at room temperature, it was feasible to use plastic pipette tips as deposition nozzles to avoid possible contamination introduced by the glass nozzle. As a result, much better precision (14 16%) was obtained with the off-line laser ablation and deposition scheme, as indicated by data discussed later in association with Table 3. The off-line deposition method was therefore adopted for the rest of the experiments.

10 120 ZHOU ET AL. FIG. 4. Relationship between residual absorbance and blank ablation times for on-line deposition. After total sample evaporation, blank ablation was performed with the same transfer tubing and a clean sample holder for the same length of time as in a loaded ablation. See text for details. Effect of Carrier Gas Flow Rate Ideally, an ablation and effluent transfer system should transport particles in a wide size distribution to the detection system. In practice, aerosol transportation and deposition on a graphite tube surface is a complex process. Many factors are involved including size and structure of the laser ablated particles, flow rate of the carrier gas, geometry and properties of the graphite surface, and nozzle diameter. These factors may all have contributions to the particle collection efficiency during deposition of laser ablated materials, but some of their exact behaviors are unknown. If the deposition process can be considered as a single stage impaction, size discrimination in the collection of particles will occur. Theoretical and experimental studies have shown (38, 39) that a cut-off particle size is associated with single-stage impaction. Under TABLE 3 Determination of Lead in Bovine Liver by Aqueous Calibration and Standard Addition Methods Aqueous calibration Standard addition With matrix Without matrix With matrix Without matrix Sample Certified modifier modifier modifier modifier Bovine liver { { { { { (SRM 1577a) (mg g 01 ) (mgg 01 ) (mgg 01 ) (mgg 01 ) (mgg 01 ) (14, n Å 7) a (14, n Å 7) a (16, n Å 7) a (15, n Å 7) a a RSD%, n Å number of determinations.

11 CONTINUOUS WAVE DIODE LASER SAMPLE VAPORIZATION 121 FIG. 5. Theoretical relationship between lead particle diameter (mm) collected and the average argon gas flow rate (liter min 01 ) for a single stage impactor with a nozzle diameter of 1.5 mm. In the calculation, the Stoke s number of was assumed. the assumption that the particle motion is governed by Stoke s law (40), the lower limit of the particle diameter (mm) in the collection can be calculated with the equation d p Å 9 StkhD 1/2 j r p UC c 10 4, (4) where Stk is Stoke s number, h is the viscosity of the carrier gas (g cm 01 s 01 ), D j is the nozzle orifice diameter (cm), r p is the particle density (g cm 03 ), U is the average nozzle exit flow velocity (cm s 01 ), and C c is the Cunningham correction factor. At a nozzle diameter of about 1.5 mm, the theoretical relationship between the minimum particle size collected and the average gas flow rate was calculated as shown in Fig. 5. As can be seen from the plot, an increase in the flow rate would result in a decrease in the minimum particle size collected. The highest collection efficiency, or largest number of particles collected, would be at high gas flow rate. With a nozzle diameter of 1.5 mm and a nozzle-to-surface distance of about 5.0 mm, the effect of flow rate in the range from 0.5 to 6.0 liter min 01 was investigated with the diode laser ablation system. The experimental relationship between normalized absorbance of lead and argon flow rates is shown in Fig. 6. At low flow rate, the signal size increased with the flow indicating improved collection efficiency. Beyond the flow rate of 2.0 liter min 01, the signal became insensitive to increased flow rate. The minimum particle diameter collected at 2.0 liter min 01 was estimated, by theoretical calculation, to be 0.8 mm. This can be compared with an experimental particle size measurement that has been attempted by Arrowsmith et al. (19). In their work, with a pulsed YAG laser

12 122 ZHOU ET AL. FIG. 6. Experimental relationship between the normalized absorbance signal and argon gas flow rate. At low flow, the signal size increased with an increase in flow rate. At high flow, the signal size became insensitive to the change in flow rate. for molybdenum ablation, the particle size distribution at the end of a transfer tube was measured by light scattering. Their results showed a maximum in the distribution at 0.2 mm. Only a relatively small number of particles (õ0.05% of the total) were detected at 5 12 mm. This indicates that the small range of particle size detected by the transport deposition system in the present work was partially responsible for the low lead recovery (15%). On the other hand, a small population of large particles probably accounts for a large percentage of the ablated mass. Theoretical calculation (19) has shown that transfer losses become serious for particles smaller than mm and larger than 2 mm for a parabolic flow, but the exact behavior of the cell transfer function still remains unknown. It is believed that matching the transfer function between the cell and the impactor will ensure a maximum transfer and collection efficiency. Further investigations on the ablated particle size distribution in a CW laser generated plume will be necessary for a better understanding of the relative contributions to the transport and collection efficiency of the system. Effect of Ablation Cell Geometry Straight through cells are the most widely used ablation chambers in the literature (Fig. 1a). The simplicity of the geometry is the main advantage of these types of cells. During ablation, laser radiation incident from the top window generates a plume near the sample surface, and vaporized materials are entrained onto the carrier gas flow stream and transported out of the chamber. Inevitably, the carrier gas stream in this configuration will disturb the upward expanding plume and result in an irregular flow pattern. The gas flow through the horizontal tubes can also suffer from gravitational loss of the particles. In an attempt to study the effect of cell geometry on the analytical

13 CONTINUOUS WAVE DIODE LASER SAMPLE VAPORIZATION 123 performance of the ablation deposition system, a straight through cell and a tangential flow cell were constructed. In the cell design of Fig. 1b, the inlet and outlet for the gas were arranged tangentially with respect to the ablation chamber. As a result, the incoming gas from the bottom of the cell had an upward tangential flow and then went through a tangential exit toward the detection system. A more stable gas flow pattern was expected as a consequence of the tangential cell design. Analytical Performance The use of aqueous standard solutions and standard addition were tested as calibration methods for evaluation of analytical performance. The aqueous standard was initially applied to the same graphite holder that was used for the solid sampling, followed by ablation/vaporization with the diode laser. It was found that adsorption of the analyte onto the surface of the sample holder caused very poor lead recovery and precision. A totally pyrolytic graphite platform (Perkin Elmer, Norwalk, CT) was placed on top of the graphite holder to improve the situation. After application of aqueous standard to the platform, the diode laser power was increased to about 15 W at a rate of 30 W min 01 to dry the solution and then set at full laser power for 3 min for vaporization. Although no temperature measurement was made, it was observed that the platform quickly and uniformly turned to red under full laser power, which indicated a quick heat transfer and small temperature gradient along the platform. The platform was able to reduce the analyte loss due to adsorption, and provide a quick and uniform vaporization. Figure 7 shows the calibration curves by the aqueous calibration and by standard addition obtained with the platform. For Fig. 7a, the aqueous standard was vaporized from the platform and the atomic absorption signal from the deposition tube was then measured. Data were obtained both with and without matrix modifier added to the deposition tube after deposition, and before atomic absorption measurements. The aqueous calibration curve achieved with the matrix modifier displayed a significantly higher sensitivity by 45%, with a lead recovery of around 15%. No explanation was available for the large effect of the matrix modifier in Fig. 7a for laser vaporization compared to the approximately 10% effect on lead aqueous standard solution vaporized normally in the graphite furnace. Calibration with standard addition was also explored. Bovine liver powder (2 mg) was added onto the platform after application of aqueous standard. The sample and standard together were dried for 30 s with about 15 W of diode laser radiation. Full laser power was applied after the dry period to perform the ablation/vaporization. The calibration data obtained with and without matrix modifier added to the deposition tube are illustrated in Fig. 7b. Statistical data showed no significant difference between these two curves. A possible reason for this may be that the solid sample, when vaporized with the aqueous solution, acted as its own matrix modifier. Determination of lead in bovine liver powder (SRM 1577a) was performed based on each of the calibration curves shown in Fig. 7. By use of each calibration curve, the amount of lead in known weights of sample was calculated by use of the absorbance signals they produced. The experimental results of the determination of lead in bovine liver obtained with the different calibration methods are listed in Table 3. By use of Student s t test performed at the 95% confidence level, no statistical difference was found between the results obtained by the aqueous calibration with matrix modifier and the certified reference value. The

14 124 ZHOU ET AL. FIG. 7. Calibration curves for the diode-ablation deposition-aas detection setup with (a) aqueous standard solution and (b) standard addition method. The solid triangles ( ) and solid circles (l) represent aqueous standard calibration data obtained with and without matrix modifier added to the deposition tube, respectively. The x axis is the amount of lead in different weights of bovine liver sample. The regression functions were A i Å / m Pb and A i Å / m Pb, respectively. The circles ( ) and solid squares ( ) represent standard addition calibration data achieved with and without matrix modifier added to the deposition furnace. The regression functions were A i Å / m Pb and A i Å / m Pb, respectively. A L vov graphite platform was used for the sample vaporization. t test showed, at the 95% confidence level, that the certified value was significantly different from the results obtained with the other calibration methods, which illustrated that matrix modification was necessary for calibration, and that standard addition did not work. In general, standard addition would only be expected to work if the sample and standard were to vaporize under identical conditions. As standard addition did not give accurate results, it appears that addition of the aqueous standard and the solid bovine liver together on the platform did not create identical vaporization conditions for sample and standard. That aqueous calibration did work, if matrix modifier was added to the deposition tube, follows atomic absorption practice. This is encouraging for an ablation/vaporization approach which has often been shown to be difficult to calibrate, but the precision of the measurements was not excellent, which means that these results cannot be considered to be definitive. In order to study the effect of gas flow patterns inside the ablation cell on the system analytical performance, bovine liver powder vaporization was carried out with both straight through and tangential flow ablation cells. It was found that, in the case

15 CONTINUOUS WAVE DIODE LASER SAMPLE VAPORIZATION 125 FIG. 8. Solid sampling data for the lead in bovine liver with (a) a straight through and (b) a tangential flow cell with regression functions of A i Å / 0.037m Pb and A i Å / 0.033m Pb, respectively. The statistical data showed no significant difference in the sensitivity and precision between the two cells. In both experiments, no graphite platform was used. The solid sample was vaporized from a POCO graphite sample holder. of solid sampling, the use of a platform did not affect analytical performance. Therefore, in both experiments, bovine liver powder was vaporized directly from the POCO graphite sample holder, without the platform. Figures 8a and 8b illustrate the experimental data for the straight through and tangential flow cells, respectively. For both cells, the recovery of the lead in bovine liver was about 15% and at the 95% confidence level, the Student s t test showed statistically no significant difference in sensitivity between the cells. To determine the standard deviation with each ablation chamber, a normalized signal was used. This was calculated as the integrated absorbance divided by the amount of lead in each weighed bovine liver sample aliquot. For 11 replicates, the experimental data are listed in Table 4. The F test at a confidence level of 95% showed no statistical difference in precision obtained with the straight through cell

16 126 ZHOU ET AL. TABLE 4 Precision Data of Lead Determination with Straight Through and Tangential Flow Cells Normalized signal and standard deviation Cell type Short term a Long term a Tangential flow { { (Absorbance ng 01 ) (Absorbance ng 01 ) (12, n Å 11) b (24, n Å 17) b Straight through { (Absorbance ng 01 ) (18, n Å 11) b a In the short term measurement, all data were obtained within 24 h; in the long term measurement, data were collected over a period of several days. b RSD%, n Å number of determinations. (12%) and the tangential flow cell (18%). This precision result is about the same as that reported by Wennrich et al. (23). In their experiments, an RSD value of 13% was obtained with pulsed ruby laser and an on-line deposition method by integrated absorbance measurement of lead in a solid sample. Su et al. (25) were able to achieve statistically the same RSD (9%) in their CW laser ablation work for lead determination. A comparison of the experimental data for the two cells indicated that the effect of the gas flow pattern did not play a major role in the analytical performance of the system. The rather small absorbance signal (õ0.10), determined by the sample loading capacity of the ablation cell and the deposition efficiency, could have a more prominent effect on the precision than that of the gas flow patterns. According to definition, the mass detection limit is the amount of analyte that produces a signal equivalent to three times the standard deviation of the noise associated with the blank. The mass detection limit for the lead determination based on the calibration curve of Fig. 8b was calculated to be about 0.1 ng. Seemingly, the overall collection efficiency of only 15% leaves room for improvement in the detection sensitivity. Optimization of the deposition parameters will probably enhance the collection efficiency and hence, the detection sensitivity. Further improvement in the detection sensitivity can be obtained with a more sensitive detection technique. CONCLUSIONS High power diode lasers can be used for direct solid sample analysis and have potential for portable analytical instrumentation that could take advantage of the light weight, compactness, and high efficiency of diode lasers. The experimental results have shown the potential for the success of the use of an incoherent diode laser array in solid sample vaporization. With a power density of about Wcm 02, bovine liver powder could be totally evaporated. The CW diode laser was able to produce high stability laser power output which is potentially superior to that operated in the pulsed mode. However, in this initial experiment the precision of measurement was

17 CONTINUOUS WAVE DIODE LASER SAMPLE VAPORIZATION 127 not superior to that achieved by others with pulsed lasers. The small absorbance signal size used on the detection system could be responsible for the system precision performance. A detection limit for lead determination in a bovine liver powder standard was estimated to be about 0.1 ng with a precision of around 12% in RSD. There still exists room for further improvement. First, a better laser beam quality with higher brightness is essential for tight focusing. The higher power density would ensure more complete solid sample evaporation, and probably would lead to improved precision and limit of detection. Second, a more efficient vaporization chamber design and transfer mechanism deserves continued investigation. With the help of a totally pyrolytic graphite platform, calibration with aqueous standard plus matrix modifier was successful although the standard addition method failed to give statistically correct values. From the experimental data, the tangential flow cell did not play a major role in improvement of the system precision, even though it was designed to create a more favorable flow pattern compared to the straight through flow cell. It is believed that the laser ablated/vaporized particle size distribution, the transfer function of the cell, the collection function of the deposition system, and the match between them are the limiting factors in the system performance. The particle size distribution in the primary plume generated by laser radiation is dependent on the laser power density and laser solid interactions. In the power density range from 10 4 to 10 7 Wcm 02, the resulting vapor consists of polyatomic particles and molecular species (24). The relatively low power density used in this work probably gave rise to droplets and clusters (41). To improve the analytical performance, at least an order of magnitude increase in the laser power density would be needed to ensure a complete vaporization and well controlled particle size. An improvement in the low lead recovery (15%) of the system is necessary for a more sensitive lead determination. The low collection efficiency associated with the graphite furnace impactor was the main cause of the low recovery. ACKNOWLEDGMENTS The authors gratefully acknowledge both the Perkin Elmer Corporation for use of the 5000 atomic absorption spectrometer and the support of the Photonics Research Center of the University of Connecticut for the loan of laser equipment. In particular, we acknowledge Dr. Chandra S. Roychoudhuri and Weiqun Chen for their technical assistance in this work. Helpful discussions with Dr. Mark S. Zediker at McDonnell Douglas and assistance in AAS operation by Robert F. Lonardo are also gratefully acknowledged. Jack X. Zhou was financially supported in part by a research fellowship from the Photonics Research Center. REFERENCES 1. Miller-Ihli, N. J. Anal. Chem., 1992, 64, 964A. 2. Brown, A. A.; Lee, M.; Kullemer, G.; Rosopulo, A. Fresenius Z. Anal. Chem., 1987, 328, Russo, R. E. Appl. Spectrosc., 1995, 49, 14A. 4. Darke, S. A.; Tyson, J. F. J. Anal. At. Spectrom., 1993, 8, Kurniawan, H.; Kobayashi, T.; Kagawa, K. Appl. Spectrosc., 1992, 46, Kwong, H. S.; Measures, R. M. Anal. Chem., 1979, 51, Richner, P.; Borer, M. W.; Brushwyler, K. R.; Hieftje, G. M. Appl. Spectrosc., 1990, 44, Kagawa, K.; Ohtani, M.; Yokoi, S.; Nakajima, S. Spectrochim. Acta B 1984, 39, Lewis II, A. L.; Piepmeier, E. H. Appl. Spectrosc., 1983, 37, Hwang, Z. W.; Teng, Y. Y.; Li, K. P.; Sneddon, J. Appl. Spectrosc., 1991, 45, Lee, Y. I.; Sawan, S. P.; Thiem, T. L.; Teng, Y. Y.; Sneddon, J. Appl. Spectrosc., 1992, 46, Lee, Y. I.; Thiem, T. L.; Kim, G. H.; Teng, Y. Y.; Sneddon, J. Appl. Spectrosc., 1992, 46, Pang, H. M.; Wiederin, D. R.; Houk, R. S.; Yeung, E. S. Anal. Chem., 1991, 63, 390.

18 128 ZHOU ET AL. 14. Sdorra, W.; Brust, J.; Niemax, K. Mikrochim. Acta, 1992, 108, Kuzuya, M.; Matsumoto, H.; Takechi, H.; Mikami, O. Appl. Spectrosc., 1993, 47, Sdorra, W.; Niemax, K. Mikrochim. Acta, 1992, 107, Geertsen, C.; Briand, A.; Chartier, F.; Lacour, J-L.; Mauchien, P.; Sjöström, S.; Mermet, J-M. J. Anal. At. Spectrom., 1994, 9, Quentmeier, A.; Sdorra, W.; Niemax, K. Spectrochim. Acta B, 1990, 45, Arrowsmith, P.; Hughes, S. K. Appl. Spectrosc., 1988, 42, Kantor, T.; Polos, L.; Fodor, R.; Pungor, E. Talanta, 1976, 23, Kantor, T.; Bezur, L.; Pungor, E.; Fodor, R. Spectrochim. Acta B, 1979, 34, Wennrich, R.; Dittrich, K. Spectrochim. Acta B, 1982, 37, Wennrich, R.; Dittrich, K. Spectrochim. Acta B, 1987, 42, Lincoln, K. A. Int. J. Mass Spectrom. Ion Phys., 1974, 13, Su, G. Z.; Lin, S. L. J. Anal. At. Spectrom., 1988, 3, Shine, R. J.; Alfrey, A. J. Jr.; Byer, R. L. Opt. Lett., 1995, 20, Lang, R. J. High Power Flared Semiconductor Lasers: Theory and Experiment, CLEO, Vol. 8, 1994 OSA Technical Digest Series, p. 24. Optical Society of America, Washington, DC, Cross, P. S.; Harnagel, G. L.; Streifer, W.; Scifres, D. R.; Welch, D. F. Science, 1987, 237, Krebs, C. A.; Vivian, B. D. Proc. SPIE, 1988, 893, Chen, W.; Roychoudhuri, C. S.; Banas, C. M. Opt. Eng., 1994, 33, Vogel, A.; Dlugos, C.; Nuffer, R.; Birngruber, R. Laser Surg. Med., 1992, 11, Lucker, E.; Rosopulo, A.; Kreuzer, W. Fresenius Z. Anal. Chem., 1987, 328, Klein, J.; Schmidt, H.; Dirscherl, C.; Muntau, H. Fresenius Z. Anal. Chem., 1987, 328, Mohl, C.; Grobecker, K. H.; Stoeppler, M. Fresenius Z. Anal. Chem., 1987, 328, Lucker, E.; Rosopulo, A.; Kreuzer, W. Fresenius Z. Anal. Chem., 1991, 340, Lucker, E.; Konig, H.; Gabriel, W.; Rosopulo, A. Fresenius Z. Anal. Chem., 1992, 342, Chamsaz, M.; Sharp, B. L.; West, T. S. Talanta, 1980, 27, Sneddon, J. Anal. Chem., 1984, 56, Sneddon, J. Appl. Spectrosc., 1989, 43, Hinds, W. C. Aerosol Technology. Wiley, New York, Vam Deijck, W.; Balke, J.; Maessen, F. J. M. J. Spectrochim. Acta B, 1979, 34, 359.