Laser Ablation Sampling with Inductively Coupled Plasma Atomic Emission Spectrometry for the Analysis of Prototypical Glasses R. E. RUSSO AND X . L. M A 0 Lawrence Berkeley Laboratory Berkeley C A 94720 USA W. T. CHAN Department of Chemistry University of Hong Kong Hong Kong M. F. BRYANT Savannah River Technical Center Westinghouse Savannah River Co. Aiken SC 29808 USA W. F . KINARD Department of Chemistry College of Charleston Charleston SC 29424 USA Laser ablation sampling is presented as an alternative to dissolution procedures for elemental analyses of prototypical glasses using inductively coupled plasma atomic emission spectrometry. These glass samples were prototypes of vitrified waste products from the Savannah River Technology Center. The sanqles were not translated or rotated during laser sampling but were repetitively sampled at a single spot using a KrF excimer laser with a 10 Hz repetition rate.The time- dependent mass ablation rate was measured and is discussed. Silicon the major element in the matrix was used as an internal standard and excellent precision (s = 1-3%) was obtained. Quantitative analysis was demonstrated using known prototypical glass compositions. Preferential vaporization was investigated by comparing measured elemental ratios using a nanosecond excimer laser (1 = 248 nm) and a picosecond Nd YAG laser (fourth harmonic 1 = 266 nm). Keywords Laser ablation; inductively coupled plasma atomic emission spectrometry; preferential vaporization; glasses; direct solid sampling Laser ablation with subsequent sample introduction into an analytical source provides several attractive benefits for direct solid sample chemical analysis.'-6 Primarily laser ablation sampling can be applied to any solid material without sample preparation.Sample preparation involving acid and microwave digestion is time consuming and is often a source of contami- nation and analyte loss. Also laser sampling consumes only a small amount of the sample typically in the range of ng to pg per pulse. Therefore personnel exposure sample handling and instrument contamination are minimized which are especially important for toxic or radioactive samples. Further the sam- pling area is small in the range of pm for a focused laser beam. With multiple-spot sampling over the target surface the elemental spatial homogeneity can be evaluated.Depth profi- ling is also possible with repetitive pulsing at a single location on the surface. Laser ablation is a process that involves coupling of the photon energy of a laser beam into the surface of a solid resulting in evaporation ejection of atomic and ionic species ejection of fragments (particles) from the surface due to shock waves and a hybrid of these proce~ses.~-'~ The thermal evapor- ation component may be deleterious for analytical purposes Journal Joyrnal of Analytical Atomic Spectrometry because elements of high vapour pressure can be enriched in the vapour phase relative to the original solid sample (preferen- B Y using nanosecond and shorter UV laser pulses rapid heating and explosive ejection can minimize preferential vaporization and provide stoichiometric sampling of the However the laser-material interaction involves complex non-linear dynamic processes.The amount and composition of the sampled vapour will depend on the properties of the sample and the laser beam parameters."-'' This work involves UV laser ablation sampling with induc- tively coupled plasma atomic emission spectrometry (ICP- AES) for elemental analysis of prototypical glass samples from the Savannah River Site Vitrification Facility. These glass samples are prototypes for vitrified radioactive waste products. The influence of laser beam parameters (pulse width and power density) on the laser-glass interaction was studied by monitor- ing the temporal ICP emission intensity as the laser beam repetitively ablated the solid glass samples.Power density studies were conducted by varying the laser beam spot size on the glass surface with fixed laser energy and by using lasers with nanosecond and picosecond pulse durations. The amount of material removed from the solid and introduced into the ICP (mass ablation rate) preferential vaporization of volatile components and elemental quantification using Si as an internal standard were addressed. These studies emphasize the understanding of laser sampling of prototypical glasses and are applicable to analysis using either ICP-AES or ICP mass spectrometry (MS). The research demonstrates that precise quantitative elemental analysis of prototypical glass samples is viable by laser ablation sampling. tial vaporization) rendering the analysis EXPERIMENTAL A diagram of the experimental system is shown in Fig.1. The primary components are an excimer or a Nd YAG laser an ICP and spectrometer with a photodiode-array detector and a microcomputer with associated data acquisition electronics. The KrF excimer laser has a pulse width of approximately 30 ns and A = 248 nm. An iris is placed between the laser and the focusing lens to limit the laser beam diameter to 6 mm so that the spatially more homogeneous portion of the excimer beam is used for these experiments. The energy of the laser at the sample surface is 30 mJ. The relative standard deviation of Analytical Atomic Spectrometry April 1995 Vol. 10 295Microcomputer (.I I I Laser sampling chamber t Iris Carrier gas IN Fig.1 Diagram of the LA-ICP-AES experimental configuration. The chamber is mounted on an xyz translation stage for focus and lateral adjustments. A nanosecond KrF excimer laser (1 = 248 nm pulse energy = 30 mJ) and picosecond Nd YAG laser (1 = 266 nm pulse energy = 10 mJ) were used (s,) of the pulse energy was measured to be 4%. The Nd YAG laser has a pulse width of 35 ps and the energy of the fourth harmonic (A=266 nm) is approximately 10 mJ. This laser was also pulsed at 10 Hz with an s in pulse energy of approximately 8%. The laser beam is focused on to the glass samples with a plano-convex lens with a 20 cm focal length. The effective focal length is 18.3 cm owing to a higher refractive index of the lens (fused silica) at the UV wavelengths. The samples are always placed before the effective focus of the lens.Depending on the position of the lens and the energy of the laser pulse the typical power density at the target surface is in the range 108-109 W cmP2 for the excimer laser and approximately two orders of magnitude higher for the Nd:YAG laser. Power density is estimated from the energy of the laser beam pulse width and measured spot area. For the excimer laser (30 ns 30 mJ per pulse) the diameter of the spot on the glass surface varies from approximately 1100 pm ( lo8 W cmV2) to 200 pm (3 x lo9 W cme2). The spot diameter is approximately 400 pm (240 x lo9 W The ICP is a Plasma Them 2500D operating at a forward power of 1.25 kW. The flow rates of the outer intermediate and central Ar gas are 15 1 and 11 min-I respectively.The central gas flow rate is regulated with a needle valve and monitored with a mass flow meter to ensure reproducible flow. A photodiode-array (PDA) spectrometer is used to monitor multiple wavelengths from the ICP simultaneously. T!e spec- trometer focal length is 0.32 m and dispersion is 25 A mm-' with a 1200 groovesmm-' grating. The PDA detector has 1024 pixels each of which is 25 pm x 2.5 mm. The PDA length is about 2.6 cm. Therefore the wavelength coverage is 60 nm with the 1200 groovesmm-' grating. The PDA detector is electrically cooled to reduce dark-current background. The ICP discharge is imaged with a 20cm focal length bi-convex singlet lens on to the entrance slit of the spectrometer with a demagnification ratio of 0.75. Typical slit width and slit height for the 35 ps Nd:YAG laser.Table 1 Composition (YO) of major elements for the SRTC glass samples* are 50 and 5 mm respectively. The observation height in the ICP is 10-15 mm above the load coil. A diagram of the laser sampling chamber is shown in Fig. 2. The main body has 0.d. 3.18 cm and i.d. 2.54 cm. The internal length of the chamber is 3cm. The chamber window is detachable to facilitate easy replacement. The window can be damaged by a high laser fluence and/or deposits from ablated samples. The sample holder is water cooled to maintain a constant temperature at the sample during long repetitive sampling experiments. The chamber is sealed from the atmos- phere with O-rings at the window and the sample holder. Carrier gas flows from the window toward the sample to minimize the deposition of material on the window.Laser sampled material is delivered to the ICP via a Teflon tube (20 cm x 3 mm i.d.). The chamber is mounted on an xyz- translator to position the sample relative to the laser beam. A chard of glass is mounted in the laser-sampling chamber with double-stick tape. The sample is not translated or rotated during laser sampling. Instead it is repetitively ablated at several surface locations using the 10Hz pulse rate from the laser. The samples are as-received chards of Savannah River Technology Center (SRTC) prototypical glass. The glasses are formed by melting a mixture of oxide reagents; five components (Mn Fe Si A1 and Mg) are listed in Table 1 for several samples. Si02 is the dominant matrix species at approximately 50% and the other four oxide concentrations range from 1.4% to 12.8%.These species were chosen for this work based on their relatively high concentrations and strong elemental spec- tral emission in order to measure good signal-to-background (ICP) ratios using the ICP-PDA system. Each glass sample contains fifteen additional oxides that only vary slightly among the samples; the physical thermal and optical properties are similar. The centre wavelength of the spectrometer is set at Ar gas OUT I Sample Window chamber \ I I c:p \ - Base I Sample O-ring 1 1 t Ar gas IN Fig. 2 Detailed diagram of the LA sampling chamber. The chamber is machined out brass rod. The window is a 1 mm thick quartz plate Oxide Batch 1 Batch 2 Batch 3 Batch 4 Blend 1 Px HM MnO 2.1 1 1.73 1.87 3.11 2.05 2.07 2.15 12.8 11.1 11.7 11.7 10.9 13.3 7.78 SiO 50.2 52.1 52.6 50.1 51.9 46.5 55.8 A1203 4.88 4.63 3.44 3.43 4.16 2.99 7.15 MgO 1.42 1.42 1.42 1.43 1.41 1.41 1.49 Fe203 * Fifteen additional oxides comprise each batch to total 100%.Variation in oxide concentrations from batch to batch are small. Optical and thermal properties are similar. Sample names (Batch Blend Px HM) represent correspondence to expected conditions at SRTC. 296 Journal of Analytical Atomic Spectrometry April 1995 Vol. 10288 nm to monitor the entire range of emission lines from the five selected elements simultaneously. A typical background- corrected spectrum is shown in Fig. 3. RESULTS AND DISCUSSION Mass Ablation Rate Time Response The amount and composition of material ablated from the prototypic glass will be dependent on the laser beam param- eters.By using the excimer laser with a power density of 1.2 GW cm-2 the elemental emission intensity shows that the mass ablation rate changes with time during repetitive pulsing. Fig. 4 shows the temporal emission responses measured for Mn and Si respectively in the ICP. Similar behaviour is observed for all the elements listed in Table 1. The five graphs in Fig. 4(a) and (b) represent the mass ablation time response during repetitive sampling at five separate locations on a chard from batch 1 using fixed laser and ICP conditions. The spots (craters) are spaced 5mm from each other. The changing intensity level during repetitive pulsing at each spot demon- strates that the mass ablation rate does vary during the experiment.The exact time response of mass ablation during repetitive sampling depends on a number of factors including the sample surface and bulk properties laser beam energy and lens focusing arrangement. The time behaviour is not due to a change in power density as the crater is developed; the confocal parameter of the focused beam is several orders of magnitude longer than the final crater depth (approximately 1000 pm after 900 pulses). However the changing aspect ratio of the crater can influence the laser beam energy coupling into the material. The time profile will not always look like the data shown in Fig. 4 depending primarily on the sample and laser beam. For some metal alloy samples we measured a constant-level steady-state intensity over several minutes.The fact that mass ablation rate changes with time may complicate analytical utility. As shown in Fig. 4 the initial laser-material interaction is always measured to be more imprecise than the resultant long-term mass ablation behav- iour. For the five spots the signal intensity for the first 100 pulses (the initial peak in the profiles) differs significantly (s up to 50%). The precision for the five spots after 60s of pulsing improves to s = 5-8%. Although the mass ablation rate changes these data are very useful because the emission temporal profile for Mn exhibits the same mass ablation behaviour as that from,Si. The intensity ratio of Mn to Si is constant both in time and between spots except for the first 10 s which is omitted in Fig.4. Similar graphs are obtained .z 2000 c 1000 0 1 250 260 270 280 290 300 310 320 Wavelengthtnrn Fig. 3 Typical background-corrected spectrum for laser sampling of SRTC prototypical glass. The central wavelength of the PDA spec- trometer is set at 288 nm to cover the emission lines from Mn Mg Fe A1 and Si. Integration time 5 s for Mg Fe and A1 when compared with Si. Si was used as the internal standard because it is the dominant matrix element. The integrated intensity ratios for Mn Fe Mg and A1 to that of Si are listed in Table 2. The ratios represent the spectroscopic values not the mass abundance ratio in the solid sample. The spectroscopic intensity depends on the emission line strength and detector efficiency at the particular wavelength and also the amount of the material.The intensity ratios have excellent precision; for the five repetitions s varies from 0.3 to 1.7%. This excellent precision for successive spots demonstrates the improvement in the analytical figures of merit by repetitive pulsing and using an internal standard. The constant ratios demonstrate that there is no time- dependent fractionation of these elements over this time at this power density. Also these data demonstrate that the glasses are homogeneous in elemental composition throughout the crater depth. The crater diameters are about 200 pm and the depth is of the order of 1000 pm after 900 laser pulses. The sampling rate is therefore about 1 pm per laser shot. Depth profiling with this resolution or better is possible.Therefore the elemental ratio to Si provides an indication of homogeneity in the concentration distribution. Mass Ablation Versus Power Density The mass ablation rate at a particular time during repetitive pulsing is also dependent on the laser beam power density. By using the excimer laser with a fixed beam energy and translating the focusing lens to obtain spot sizes from approximately 1000 to 200 pm the power density is varied from 0.13 to 2.12 GW cmP2. The amount of glass sample ablated increases with increasing power density and reaches a maximum at about 0.3 GW cm-2. Fig. 5 shows the data for Fe and Si as a function of power density. Each data point in these graphs represents the averaged intensity from five spots measured after 60 s of pre-ablation.The reason for the plateau in intensity or mass ablation rate is believed to be plasma shielding. When a laser-induced plasma initiates at the target surface it can absorb or reflect a portion of the laser pulse energy thereby reducing the energy available for removing material mass.I7 Analysis of crater volumes for the same number of pulses verified that the mass ablation rate was reduced at the higher power densities; the plateau is not due to particle size or transport changes. Part of the plateau is due to the change in area of the spot which will be discussed in detail in a subsequent paper. Relative to Si the intensity ratio for Fe remains essentially constant over this power density range (Fig. 5). Similar behav- iour is measured for Mg Al and Mn.Tables 3 and 4 list normalized measured elemental intensities and their ratios to Si respectively. Although the amount of material ablated changes with power density the absolute intensity change is not significant if the power density changes only slightly especially in the plateau region (> 0.3 GW cm-2). Therefore the irregular surface of glass chards and thus the slight variation in lens-to-sample distance should not influence the mass ablation behaviour. Also by using Si as the internal standard the influence of power density on the mass ablation behaviour is minimized. The fact that the ratio remains constant for these elements also indicates that preferential vaporization is not significantly influenced by power density over this range. Previously we demonstrated that the Zn Cu ratio from brass samples varied over this same power density range.2 In that work we obtained the correct elemental ratio by using > 1 GW cm-2 as verified by dissolution and solution analysis of the brass sample.The effect of the glass-sample surface condition on laser ablation sampling was studied further. Several chards from Batch 1 were melted in a Pt crucible at 1075°C for about Journal of Analytical Atomic Spectrometry April 1995 Vol. 10 297800 600 400 200 0 600- 400 200 I Mn 2500 2000 1500 1000 500 - - - - - - - I I I 1 1 1 1 I 3000 2500 2000 1500 1000 500 0 800 Mn 600 400 200 2500 2000 1500 1000 500 0 I Mn 3000 2500 2000 1500 1000 500 0 L 800 3000 0.3 OB4 * 0-4 7 0.2 1 I 0.2 C g 0.1 0 d ::I 0.1 ____I 1 0 40 80 120 160 Fig.4 Temporal mass ablation rate profiles for Mn and Si from ICP-ALES during repetitive sampling at five locations of SRTC prototypical glass using the nanosecond excimer laser repetition rate of the laser is 10 Hz and ratio of the Mn to Si data Table 2 Emission intensity ratios with respect to Si and the precision for the major elements from sample Batch 1 Wavelength/nm 257.61 273.96 279.55 285.21 309.27 Intensity ratio 0.3279 0.2676 2.4360 0.2223 0.4176 Standard deviation 0.0010 0.001 1 0.0181 0.0039 0.0065 ~ Relative standard deviation (Yo) 0.3 1 0.42 0.74 1.73 1.56 10min. The re-solidified sample adhered to the crucible and had to be broken out. Laser ablation of these samples provided the same intensity ratios and precision as measured for the original irregular shaped glass chard.Therefore sample prep- aration (smoothing the sample surface) is not necessary. The irregular shaped glass chard can be used as long as repetitive pulsing is used to ‘condition’ the sample surface. Laser ablation sampling can be significantly influenced by different gas atmospheres; only argon was used in these studies. Previously we measured higher ma% ablation rates in helium 298 Journal of Analytical Atomic Spectrometry April 1995 VuZ. 10Table 3 Normalized emission intensity versus lens-to-sample distance (laser power density) 0.9 .- ci! 0.7 2 d 0.5 2 .g 0.3 .- .I- c .I- t 0.1 Lens-to-sample distancelcm 14.3 14.8 15.8 16.8 17.3 17.6 ps Nd YAG laser ( C) - - ns excimer laser - ; Spot size/ cm 0.12 0.10 0.08 0.06 0.04 0.03 Power density/ GW cm-' 0.13 0.21 0.34 0.53 1.19 2.12 Normalized signal intensity Mn Fe MgO) Mg(11) Si A1 0.57 0.58 0.58 0.58 0.55 0.60 0.68 0.7 1 0.68 0.70 0.66 0.72 1.01 1.04 1.04 1.06 0.96 1.10 1.06 1.07 1.09 1.12 1.03 1.20 0.98 1 .oo 0.98 1 .oo 0.97 1.04 1 .oo 1 .oo 1 .oo 1 .oo 1 .oo 1 .oo Table 4 Normalized emission intensity ratios with respect to Si versus laser power density Power density/ ~ GWcm-2 Mn/Si 0.13 1.04 0.21 1.02 0.34 1.05 0.53 1.03 1.19 1.01 2.12 1.00 Normalized signal intensity ratio Fe/Si Mg(r)/Si Mg(rI)/Si 1.05 1.05 1.06 1.06 1.03 1.05 1.08 1.08 1.11 1.04 1.06 1.09 1.02 1.01 1.03 1 .oo 1 .oo 1 .oo Al/Si 1.09 1.07 1.14 1.16 1.06 1 .oo L 5000 ? 4000 > ._ +- C - 3000 2000 1000 0 Si .@ . 0 0 0 0 0.3 ' .r" 1 0.1 0 0.5 1.0 1.5 2.0 2.5 Power density/GW cm-2 Fig. 5 ICP-AES signal intensity versus laser power density for LA sampling of SRTC prototypical glass.Each data point represents the average of intensity for five spots obtained 60 s after pre-ablation at 10 Hz. Data are from nanosecond KrF excimer laser sampling for metal sampleslg and stoichiometric sampling of brass in oxygen. Also laser sampling generates a large amount of particles. Incomplete vaporization and atomization in the ICP and deposition on the sampling cone of an ICP-MS are potential problems.20 The amount of sample ablated and the size distribution of the particles should be addressed especially for ICP-MS. Picosecond Versus Nanosecond Laser Sampling A primary concern in this and other laser ablation sampling studies is demonstrated by the data in Fig.6. The data in Fig. 6(a) show the time-dependent mass ablation rates for Mg and Si using the nanosecond excimer laser at 1 GW cm-2 (similar to the data in Fig. 4). The time-dependent mass ablation rate using the picosecond Nd:YAG laser with a power density of 240 GW cm-2 for the same two elements is shown in Fig. 6(b). Similarly to the nanosecond sampling the initial interaction for picosecond laser sampling provides dras- tic variations in the intensity level versus the longer time response. A primary difference between the nanosecond and picosecond ablation can be seen by plotting the ratio of these data [Fig. 6(c)J. For the same prototypical glass sample the intensity ratio of Mg to Si is different from the two lasers. 10000 I 1 9000 8000 7000 6000 $ 3 0 0 0 p I L .e 2000 I I I I I J -Ef Si 8000 2000 ' I I I 1 I I 1 1 I 0 20 40 60 ao 100 Time/s Fig.6 Temporal mass ablation rate profiles of Mg and Si during repetitive laser sampling of SRTC prototypical glass using (a) nano- second pulsed excimer laser (A = 248 nm) and (b) picosecond pulsed Nd YAG laser (A = 266 nm).(c) Mg Si intensity ratio for the two lasers Journal of Analytical Atomic Spectrometry April 1995 Vol. 10 299Table 5 Emission intensity of Mg Al Mn and Fe with respect to Si using the picosecond Nd YAG (3 = 266 nm) and nanosecond KrF excimer ( A = 248 nm) lasers after 60 s of pre-ablation Laser Nd YAG KrF excimer Power density/ GW cm-' MgO)/Si Mg(II)/Si Al/Si Mn/Si Fe/Si 240 1.2 2.56 2.18 0.66 0.40 0.41 0.24 0.52 0.3 1 0.38 0.24 Again the ratios are of spectroscopic intensities not the mass abundance of the solid sample.The glass samples were not dissolved and analysed using solution nebulization. Therefore we do not know which intensity ratio is correct. A complete mechanism to describe the different elemental ratios measured using nanosecond and picosecond pulses would require an understanding of energy coupling mechan- isms energy dissipation hydrodynamic expansion gas dynamic expansion particle size distribution gas entrainment plasma interaction with the laser beam and target surface and solid sample excitation characteristics in the ICP. Such extens- ive studies have been conducted for the past 45 years yet still cannot accurately describe the ablation mechanism. A compo- nent of the laser material interaction is believed to be thermal vaporization which may lead to preferential vaporization at some point during laser ablation sampling of glass.Assuming partial thermal behaviour we can qualitatively explain the picosecond uersus nanosecond data based on melting- and boiling-points of the oxides. The emission intensity ratios for Mg Al Mn and Fe obtained with nanosecond excimer and picosecond Nd YAG laser sampling (after 60 s) are listed in Table 5. Both Mg and A1 have a higher intensity ratio using the higher power-density picosecond Nd YAG laser whereas Mn and Fe have the same ratios for both lasers. The melting-points of the Mg and A1 oxides are significantly higher than that of Si while Mn and Fe oxides have lower melting points (Table 6 ) .A preferential vaporization mechanism may explain the differences in inten- sity ratio. For the nanosecond excimer laser the sample is heated to a temperature that is not high enough for complete removal of the more refractory oxides (Al Mg). The oxides with lower melting points (Mn Fe) than this induced surface temperature can vaporize preferentially and be enhanced in the vapour phase. As the picosecond Nd:YAG laser has a higher power density the sample temperature can rise faster and to a greater extent than that of the nanosecond excimer laser. Based on this argument the intensity ratios (to Si) for elements with higher melting-point oxides (Mg and Al) can be larger with the picosecond Nd YAG laser. The argument can be extended to the boiling-points of the individual oxides (Table 6 ) .An interesting observation is that SiO MgO and A1203 have well defined boiling points whereas Mn02 and Fe,O undergo several phase transitions and release 0 as temperature increases.18 Again only considering the thermal component of the interaction the heated volume is expected to be raised to a higher temperature during the picosecond than the nanosecond pulse (the energy is the same only the time is different). For the higher temperature the vapour pressure of the Al,O MgO and SiO will be increased whereas Table6 Thermal properties of the major oxides in the SRTC glass samples'8 Oxide Melting-point/"C Boiling point/"C * * - 1564 Fe203 1594 SiOZ 1723 2230 A1203 2072 2980 MgO 2852 3600 - Mn304 * Undergoes several phase transitions before reaching a stable boiling compound.the additional temperature could be dissipated in ancillary phase transitions for the MnO and Fe203. Of course preferential vaporization based on melting boiling and phase transitions is not the complete mechanism to explain these data; the n-ature of the laser material interaction is a complicated convolution of many mechani~ms.l~-'~ The actual vapour pressure-temperature behavior is complicated by the high pressure from the induced shock wave. Also preferential vaporization may be due in part to the laser-induced plasma initiated over the glass surface. Convective and/or radiative heat transfer from this plasma may influence the amount of mass ablated and its composition. The influence of plasma heating on preferential vaporization has not been investigated.Another parameter that may have influenced these data is laser beam ~oherence.'~*'~ The excimer laser with unstable resonator optics has entirely different spatial coherence to the solid-state picosecond Nd YAG laser. It is interesting that neither time response curve in Fig. 6(c) approaches the other even after creating a deep (several hundred micrometres) crater in the glass. It might be expected that forced congruency would act to regulate the ablated mass ratios if preferential vaporization was the dominant factor influencing these data. Although the mechanism to explain these data cannot be confirmed by these studies heuristically these data show that the composition of sample in the ICP is effected by the laser beam properties.Quantitative Analysis Although the above discussion points out complexities in laser ablation sampling most systems will not offer such variations in laser beam properties (picosecond versus nanosecond pulses). For a fixed set of conditions a suite of similar samples may exhibit similar ablation behaviour. Analytical accuracy could be verified by dissolving one of the samples and 'calibrating' the instrument response. For the seven SRTC prototypical glasses the oxide concentrations vary only slightly providing similar physical thermal and optical properties among these samples. A quantitative study does show that linear calibration graphs can be measured when Si is used as the internal standard (Fig. 7). The ordinate is the measured emission intensity ratios (to Si) and the abscissa is the prepared oxide concentration ratios. Again each glass chard was sampled at five different spots over its surface and the signal intensity was recorded after 60 s of pre-ablation; s is typically 1-3%.The correlation coefficients (r) for these data are 0.98-0.99 except for that of Mg which is 0.70. The lower correlation for the Mg data may be due in part to the small difference in concentration ratio among the different batches. Also as demonstrated earlier Mg ablation was influenced more than the other three elements by the pulse width of the laser. However these linear calibration graphs demonstrate that laser ablation sampling is a viable approach for direct chemical analysis of solid prototypical glasses. These studies only measured the major elements in the glass because of the low sensitivity of our photodiode-array detector.The applicability of these results to the other trace elements in the glass could be investigated using a more sensitive ICP detector system. A charged-coupled device (CCD) detector which is three orders of magnitude more sensitive than the 300 Journal of Analytical Atomic Spectrometry April 1995 Vol. 100.4 0.3 0.2 >. 0.03 0.04 0.05 0.06 0.07 4- .- 2.2 - 8 - 2.1 2.0 1.9 1.8 0.026 0.028 0.030 0.032 0.25 0.20 0.15 0.10 0.15 0.20 0.25 0.30 0.5 0.4 0.3 0.2 (d) AI:Si 0.050 0.075 0.100 0.125 0.150 Concentration ratio Fig. 7 Calibration graphs for measured elemental Si ratio versus the prepared (nominal) concentration. Nanosecond excimer LA sampling of SRTC prototypical glass with ICP-AES PDA detector will be used in future studies.Ideally an ICP-MS would provide the best sensitivity for performing these studies. CONCLUSION Laser ablation sampling for ICP-AES is a viable approach for directly measuring the elemental composition of SRTC proto- typical glasses without dissolution procedures. For these glasses pre-ablation for 60 s before data acquisition minimizes the effect of sample surface condition and geometry. By using silicon as the internal standard excellent precision and quanti- tative analysis with matrix-matched standards were demon- strated; the precision was better than 1% for several of the elements. Stoichiometric sampling is a vital issue for direct chemical analysis of solids and this work showed that the picosecond Nd YAG laser provided a different analysis from the prototypic glass samples compared with that of the nano- second excimer laser.However most systems would not offer such drastic differences in laser pulse width and dissolution of several samples could be performed to calibrate the particular instrument. The suite of SRTC prototypical glass samples seemed to ablate similarly as evidenced by the linear Cali- bration graphs obtained. The ratio of elements to the Si internal standard remained constant over a wide power density range using the nanosecond UV excimer laser pulses. Therefore changes in the power density during sampling will not effect the accuracy especially as a crater develops at the sampling location. Time response measurements showed that the ratio remained constant throughout a 1000 pm depth indicating the homogeneity of elements in these glass samples.The authors acknowledge Mark Shannon and Marvin Kilgo for technical assistance. This research was supported by the US Department of Energy Office of Basic Energy Sciences Division of Chemical Sciences under Contract No. DE-AC03-76SF00098 and by Savannah River Technology Center. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 McLeod C. W. Routh M. W. and Tikkanen M. W. in Inductively Coupled Plasma in Analytical Atomic Spectrometry ed. Montaser A. and Golightly D. W. VCH New York 2nd edn. 1992 ch. 16. Chan W. T. and Russo R. E. Spectrochim. Acta Part B 1991 46 1471. Walder A. J. Abell I. D. and Platzner I. Spectrochim. Acta Part B 1993,48 397. Drake S. A. and Tyson J. F. J. Anal. At. Spectrom. 1993 8 145. Moenke-Blankenburg L. Schumann T. Ganther D. Kuss H. M. and Paul M. J. Anal. At. Spectrom. 1992 7 251. Durrant S. F. and Ward N. I. Fresenius’ J. Anal. Chem. 1993 345 512. Phipps C. R. and Dreyfus R. W. in Laser Ionization Mass Analysis ed. Vertes A. Gijbels R. and Adams F. Wiley New York 1993 ch. 4. Klocke H. Spectrochim. Acta Part B 1969 24 263. Dabby F. W. and Paek U.-C. IEEE J. Quantum Electron. 1972 8 106. Olander D. R. Yagnik S. K. and Tsai C. H. J. Appl. Phys. 1988,64 2680. Thompson M. Chenery S. Brett L. J. Anal. At. Spectrom. 1990 5 49. Chan W. T. Mao X. L. and Russo R. E. Appl. Spectrosc. 1992 46 1025. Chenery S. Hunt A. and Thompson M. J. Anal. At Spectrom. 1992 7 647. Baldwin J. M. Appl. Spectrosc. 1970 24 429. Ready J. F. Eflect of High-Power Laser Radiation Academic Press New York 1971. Hughes T. P. Plasmas and Laser Light Wiley New York 1975. Von Allmen M. Laser Beam Interactions with Materials - PhysicaE Principles and Applications Springer New York 1987. Handbook of Chemistry and Physics ed. Lide D. R. CRC Press Boca Raton FL 75th edn. 1994. Mao X. L. Chan W. T. Shannon M. A. and Russo R. E. J. Appl. Phys. 1993 74 4915. Crain J. S. Houk R. S. and Smith F. G. Spectrochim. Acta Part B 1988 43 1355. Paper 4/05 15 7B Received August 23 1994 Accepted November I 1994 Journal of Analytical Atomic Spectrometry April 1995 Vol. 10 301