Spectrochimica Acta Part B 55 Ž2000. 1771᎐1785
A comparison of nanosecond and femtosecond laser-induced plasma spectroscopy of brass samples V. Margetic, A. Pakulev 1, A. Stockhaus, M. Bolshov, K. Niemax, R. Hergenroder ¨ U Institute of Spectrochemistry and Applied Spectroscopy (ISAS), Bunsen-Kirchhoff-Strasse 11, D-44139 Dortmund, Germany Received 19 May 2000; accepted 14 August 2000
Abstract The ablation of brass samples in argon shield gas by 170 fs and 6 ns laser pulses has been studied by optical emission spectroscopy of the evolving plasmas. Differences observed in the temporal behavior of the spectral line intensities are explained by the shielding effect of the Ar plasma for ns-pulses and the free expansion of the plasma of the ablated material in case of fs-pulses. Brass with different ZnrCu ratios were used as samples. Different types of crater formation mechanisms in the case of ns- and fs-pulses were observed. At 40 mbar argon pressure the thresholds of ablation were found to be ; 0.1 and ; 1.5 J cmy2 for fs- and ns-pulses, respectively. With an internal standardization of zinc to copper it is possible to correct for differences in the ablation rates and to obtain linear calibration curves. For optimum experimental conditions, narrower confidence intervals for the determination of unknown concentrations were found in case of fs-pulses. Within the range of the laser intensities used, no dependence of the ZnrCu line intensity ratio on the number of laser pulses applied to the same ablation spot was observed, neither for fs- nor for ns-pulses, which is interpreted as the absence of fractional vaporization. 䊚 2000 Elsevier Science B.V. All rights reserved. Keywords: Laser ablation; Femtosecond; Brass material; Optical emission spectroscopy
U
Corresponding author. Tel.: q49-231-1392-178; fax: q49-231-1392-120. E-mail address:
[email protected] ŽR. Hergenroder ¨ .. 1 On leave from the International Laser Center of the Moscow State University, Russia. 0584-8547r00r$ - see front matter 䊚 2000 Elsevier Science B.V. All rights reserved. PII: S 0 5 8 4 - 8 5 4 7 Ž 0 0 . 0 0 2 7 5 - 5
1772
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
1. Introduction Laser ablation ŽLA. is a popular and widespread sampling technique for direct solid sample analysis. In combination with LA, different analytical techniques for the measurement of ablated species have been developed ᎏ direct optical emission spectrometry of a laser plume Žthe technique is usually referred to as LIBS. w1x, laser-induced fluorescence of specific analytes of interest w2x, transport of the ablated material to inductively coupled plasmas ŽICP. followed by detection of the elements by emission ŽICP-OES. w3᎐5x or by mass spectroscopy ŽICP-MS. w4,5x. LA is attractive because it avoids time-consuming sample pretreatment procedures. However, there are still intensive discussions on the best laser system or the ideal laser wavelength for controlled ablation. Over the last few years different advanced laser systems have been tested for solid sampling, primarily for LA-ICP techniques: the quadrupled Nd:YAG Ž266 nm. w6x; the quintupled Nd:YAG Ž213 nm. w7x; and the ArF-excimer Ž193 nm. w8,9x. In principle, lasers at all wavelengths Žincluding the fundamental 1064-nm Nd:YAG. can be used for reproducible and accurate sampling. However, the experimental parameters, such as laser intensity and pulse length, buffer gas and buffer gas pressure, delay and duration of the gate width for data acquisition have to be optimized for each specific excitation wavelength w10,11x. Recently, there has been a trend toward UV lasers w12x and lasers with a shorter pulse duration w13,14x. It is hardly possible to formulate the general characteristics of an ‘optimal laser system’ without referring to a specific analytical problem or a special mode of detection. For example, the efficient transport of ablated matter to an ICP requires a fine aerosol Žwith solid particle diameters of less than a few m., whereas direct optical emission spectroscopy of the laser plume needs excited atoms and ions. The analytical problems which would be avoided or minimized by an ‘optimal laser’ are: non-linear calibration curves; fractional evaporation; limitation in spatial and depth resolution due to melting and re-solidification; and matrix dependence of the analytical signal. The under-
standing of the underlying physical processes that ultimately cause these problems gives guidance towards such an optimal laser system. Briefly, the sequence of events which leads to the removal of material from a metallic sample in a vacuum under the influence of intense short laser pulses is as follows w15᎐17x. The free electron gas of a metal is rapidly heated due to inverse Bremsstrahlung, and in the next step the energy of the hot electron gas is transferred to the lattice and thermalized in the bulk. The mechanisms responsible for material removal are thermal melting and evaporation, or some kind of explosive evaporation w18x. The details of the process depend on the laser fluence, pulse duration, laser wavelength and material properties, and are not yet fully understood w19x. If LA above a certain threshold is performed in a buffer gas, a plasma is ignited, heated and sustained due to the inverse Bremsstrahlung-absorption of the laser photons, which substantially complicates the whole picture w3x. The typical characteristic time scales of the different processes involved are: free electron heating and thermalization takes approximately 100 fs; hot electron gas cooling and considerable energy transfer to the lattice lasts a few ps; thermal diffusion in the bulk takes place on a time scale of 10y1 1 s; and the onset of thermal melting and ablation occurs after 10y1 0 s. From this an important conclusion can be drawn: a laser pulse of a duration longer than a few ps does not interact the whole time with the original thermodynamic state of the material. Instead, it interacts with different transient states, and with the plasma of evaporated material and buffer gas above the sample surface. The main part of the material is evaporated from molten metal, and preferential volatilization Žfractionation . of different elements in a sample with different melting temperatures occurs w20,21x. In such a case the stoichiometric composition of the gas phase does not adequately represent the composition of the bulk, and the accuracy of the analytical procedure deteriorates. The picture changes dramatically if a laser pulse with duration of approximately 100 fs or shorter is used. Now the laser interacts only with the electron sub-system of a material. Before the
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
material undergoes any changes in thermodynamic state, the laser pulse is over and most of the energy is deposited into the sample. Material removal occurs after the laser pulse. From this brief discussion it can be concluded that a fs-laser with pulse duration of approximately 100᎐200 fs and shorter should be closer to an ‘optimal’ laser system than other systems. On the other hand, it should be clear that not all analytical problems are addressed by this strategy. Changes in mass ablation rate from sample to sample due to differences in the structure of the material remain, as will be discussed later. To overcome this kind of problem, analytical approaches like internal standardization have to be used w22x. Recently, high-power fs-laser systems have been commercialized. Although the laser technology is relatively complex, the booming growth of micromachining w23x and surgery w24x applications provides an ongoing trend towards simplified, compact, all-solid-state fs-laser systems. The potential of fs-lasers for analytical applications, and in particular for laser ablation, have so far attracted less attention. The goal of our work is the investigation of the analytical characteristics of laser ablation with a fs-laser and the comparison of the fs- and ns-modes of ablation applying optical em ission spectroscopy. The experiments were performed with brass samples, since they are known to be notoriously difficult due to the significant differences in vapor pressure of the main components, Zn and Cu, and also due to structural differences.
2. Experimental section 2.1. Instrumentation The commercial fs-laser CPA-10 ŽClark-MXR Inc., MI, USA. was used for ablation. The laser system is an all-solid-state laser consisting of a master seed-laser and a Ti:sapphire amplifier, working under the principle of chirped pulse amplification ŽCPA. w25x. The seed-laser is based on the erbium-doped fiber ring laser w26x. The parameters of the seed-laser output radiation are:
1773
pulse energy, ; 0.15 nJ; mean power, ; 6 mW; and pulse duration, 150 fs. The parameters of the amplified output pulses are: central wavelength of the pulse spectrum, 775 nm; spectral bandwidth, approximately 5 nm; pulse energy, ; 0.5 mJ; pulse duration, 170᎐200 fs; and repetition rate from a single pulse, up to 10 Hz. At the repetition rate of 10 Hz the relative standard deviation of the pulse energy was approximately 5%. Important features in fs-laser ablation experiments are post- and pre-pulses, which are connected to the process of amplification. The ratios of post- and pre-pulse intensities to the intensity of the main pulse were found to be better than 100:1 and 500:1, respectively. The comparison of the fs- and ns-modes of ablation was provided with the same laser system. The CPA-10 can also deliver ns-pulses if the seed-laser output is blocked and the optimal delay for the Ti:sapphire Q-switch pulse is changed. The Ti:sapphire laser generated pulses of 6 ns duration without the seed radiation, with the same pulse energy as in the fs-mode. The beam profiles of ns- and fs-pulses were practically identical, and realignment of the beams was not necessary. The relative standard deviation of the ns-pulse energy was approximately 10%. The pulse duration of the fs-pulses was measured with an autocorrelator ŽPulseCheck, APE GmbH, Berlin, Germany., and the pulse energy was measured with a joule-meter ŽFieldMaster, Coherent.. The laser energy was varied using neutral density or color filters. It was checked that the optical components of the experimental arrangement Žbeam splitters, filters. did not affect noticeably the duration of the pulses. The beam profile was measured with a CCD camera in front of the sample chamber. It was found to be very close to a Gaussian shape. The schematic diagram of the experimental arrangement is shown in Fig. 1. The output laser beam was directed by a set of mirrors to the sample cell and focused onto the sample surface using a plane-convex quartz lens with 75.6-mm focal length. The best analytical signal was detected when the geometrical focus was located slightly above the sample surface. The sample chamber was a simple three-axis cell with two
1774
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
above mentioned spectrograph. A spectral window of approximately 10 nm in the 400᎐500 nm range could be recorded simultaneously. The spectral window could be changed by rotation of the grating with a stepping motor. 2.2. Samples
Fig. 1. Schematic of the experimental setup.
perpendicular horizontal paths for observation of optical emission through a 3-cm diameter fused silica window. The beam of the ablating laser was directed into the cell through a top window. The cell was mounted on a 3-axis holder with micrometer screws for precise positioning of the sample. The focusing lens was mounted on the cell holder. The focusing geometry could be changed and optimized by precise translation of the lens with a micrometer screw. Argon was used as a buffer gas. The reduced gas pressure was controlled by a pressure gauge ŽBaratron. and the gas flow was regulated by a needle valve. The gas pressure was varied between 1 and 1000 mbar. The optical emission of the plasma plume was collected by a lens with a 163-mm focal length and imaged 1:1 onto the entrance slit of a homemade Czerny᎐Turner grating spectrograph Ž2400 grovesrmm, spectral range 220᎐650 nm, reciprocal linear dispersion ; 0.4 nmrmm at 400 nm, aperture ratio ; 1:12.. Side-on observation was used throughout the experiment. The spectra were detected by an intensified diode array ŽPOSMA, Spectroscopy Instruments, Germany.. The detector head of POSMA comprises a micro-channel plate ŽMCP. image-intensified photodiode array ŽIPDA. with 1024 pixels. The pixel size was 25 m =2.5 mm, which corresponds to an approximate 10-pm spectral interval per pixel using the
The brass samples used in this study were delivered by the Wieland Werke ŽAG Metallwerke, Ulm.. The concentrations of both elements in the samples are listed in Table 1. In addition to the brass samples of pure technical grade, copper samples were used in the experiments. All samples were polished with 1-m grade diamondpaste to guarantee a smooth, flat surface.
3. Results and discussion 3.1. Morphological modification The craters due to the different mechanisms of material removal for ns- and fs-pulses are demonstrated in Fig. 2. The laser craters were generated on a fresh polished TD02 brass sample at an argon pressure of 40 mbar with the same focussing geometry, and with a pulse energy of 10 and 34 J for fs- and ns-pulses, respectively. This energy was slightly above the threshold energy for the plasma breakdown necessary for spectroscopic measurements. Only the very central part of the Gaussian beam of the ns-laser provided a sufficiently high fluence for surface damage. The threshold fluences for both modes Žfsrns. were estimated by plotting the crater depth vs. pulse Table 1 Composition of brass samples used in the study Sample
Cu Žwt.%.
Zn Žwt.%.
TD02 TD03 TD04 TD05 TD06 TD07
80.25 70.20 65.50 63.12 60.55 58.05
19.71 29.76 34.46 36.84 39.41 41.91
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
1775
The remarkable difference in the surface morphology modification for the two ablation modes is obvious. While the ns-crater shows the typical structure of a re-solidified melt with a number of re-deposited droplets, the fs-crater has a well-defined brim, and shows no melting and re-deposited droplets of melted matter. In the case of fs-pulses, the detection limit of our OES technique with a POSMA detector was approximately five-fold above the threshold energy for a surface modification, namely 0.5 J cmy2 Žpulse energy 10 J.. The main part of the spectroscopic and analytical investigations described in this paper was carried out with fluences 10᎐100 times higher than the fluences applied for the determination of the detection limit Ž3᎐50 J cmy2 .. The ablation rates were determined from the crater depths and diameter using an optical microscope. Within the uncertainty of such rough measurements, the experimentally obtained ablation rates were similar for both fs- and ns-pulses, varying from ; 0.7 to ; 1.3 mrshot within the fluence range 5᎐20 J cmy2 . 3.2. Characterization of the laser plume
Fig. 2. Laser craters on a fresh polished TD02 brass sample: Ža. after 45 pulses with the fs-laser Ž10 J, 0.7 J cmy2 .; and Žb. 45 pulses with the ns-laser Ž34 J, 2.7 J cmy2 .. The background gas is argon at 40 mbar.
fluences. At fluences close to the threshold, surface modification could be seen after a single shot. To improve the accuracy of the crater depth measurements, 20 laser shots were accumulated. The estimated fluence thresholds differed noticeably: approximately 0.1 J cmy2 for the fs- and approximately 1.5 J cmy2 for the ns-pulse. These data are in good agreement with published values for both fs- and ns-pulses w19x. The ablation rate for fs-pulses with fluences near the threshold was approximately 100 nm per shot. It should be noted that the fs-pulse threshold for surface damage was significantly lower than the threshold for a plasma ignition.
The authors are aware that measurements of the temporal behavior of spectral line intensities can only give limited information on differences in the ablation process with fs- and ns-pulses. Strong continuous background radiation dominates in the first 100 ns and does not allow measurement of atomic and ionic lines with a good signal-to-noise ratio. Nonetheless, comparison of the line-intensity dynamics for the two cases shows some definite differences, which can be traced back to differences in the starting conditions of the plasma formation and evolution due to the laser pulse duration. The time evolution of the ionic and atomic lines of Cu and Mg in the laser plasma generated by ns- and fs-laser pulses of the same energy during the first s is shown in Fig. 3. The sample used was technical copper of unknown composition. The intensities were measured with a timegate of 100 ns for three different delays. The light was collected from the central part of the plasma without any spatial resolution in the direction of
1776
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
Fig. 3. Time evolution of atomic and ionic emission lines in a laser plasma generated by: Ža. fs-and Žb. ns-laser pulses. The background gas is argon at 40 mbar.
the slit. For fs-pulses the intensities of Mg and Cu ion lines wMgŽII., s 279.55 nm and 280.27 nm; CuŽII., s 283.73 nmx have their maxima near to zero delay and rapidly decrease with time. The intensities of neutral lines wCuŽI., s 282.43 nm; MgŽI., s 285.21 nmx slightly increase within the first 200 ns and slowly decrease later on. For ns-pulses the maximum intensities of the ionic lines are shifted towards later times. The intensities of the CuŽI. and MgŽI. lines gradually increase within the first 700 ns. It is interesting to compare the relative intensities of CuŽII. and CuŽI. lines. For fs-pulses the CuŽI. line is stronger than that of CuŽII. during the whole observation time, and the CuŽII. line decays rapidly within the first 200 ns. Contrary to this, the CuŽII. line is stronger at the very beginning of the plasma
evolution Ž0 delay. and the CuŽII.rCuŽI. ratio reverses after 200 ns for ns-pulses. The intensity of CuŽII. line decreases much more slowly than in the case of fs-pulses. The behavior of the ionic and atomic lines can be explained by differences in the laser heating of the buffer gas plasma. In case of fs-pulses, a buffer-gas plasma can be ignited only by the ejected material Želectrons, ions. and is not additionally heated by the laser radiation. The ablated material expands into a cold buffer gas. The number of ions created in the first stage of the ablation process can only decrease due to recombination. The number of excited neutrals increases slightly during the first 100 ns due to the ion recombination, and gradually decreases later on. In contrast, for ns-pulses, a buffer-gas plasma is ignited and sustained by absorption of laser radiation due to inverse Bremsstrahlung. The ablated species are efficiently atomized and ionized in hot plasma at the first stages of evolution. A rough estimation of the degree of ionization by the Saha᎐Eggert equation for an element with ionization potential of ; 7 eV, a plasma temperature of ; 10 4 K and a typical density of ablated atoms n 0 of ; 3 = 10 16 cmy3 , yields an ionization degree of more than 90% w22,27x. The much higher temperature of the buffer-gas plasma in the case of ns-pulses provides higher CuŽII.rCuŽI. line intensity ratios ŽFig. 3b. as compared to fs-pulses ŽFig. 3a.. The dynamics of ionic and atomic line intensities in a ns-pulse plasma can be qualitatively explained by efficient post-atomization and ionization of the ablated species in the hot gas plasma, followed by recombination and emission of excited neutral atoms at later stages of the plasma evolution w1x. Other significant differences can be found in the spatial distribution of the emission. Fig. 4 shows the emission of atomic and ionic lines of Cu and Mg as a function of the height over the sample. The intensities were measured with an integration time of 500 ns, a delay of 200 ns after the laser pulse, and a spatial resolution of 250 m. The pressure of the argon buffer gas was held at 140 mbar. The spatial resolution of 250 m was realized by moving a horizontal slit of 250-m width in front of the entrance slit of the
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
1777
Fig. 4. Spatial distribution of several Mg and Cu emission lines above a copper sample for ns and fs ablation at 140 mbar argon. Spatial resolution is 250 m. The irradiance is approximately 1 GW cmy2 and 10 TW cmy2 for the ns and the fs case, respectively. In the ns case the line represents a guidance for the eye, for the fs case a Gaussian curve is fitted to the data points.
spectrograph. The line emission intensities reflect the temperature distribution within the plume. The ns-plasma shows an asymmetry in the emiss io n d is tr ib u tio n . T h is c a n be explained by the laser-sustained detonation model. A strong compression of the front of the plume at an initial stage of the plasma evolution creates an absorbing zone for the laser light, which results in a temperature rise at the back side of the plume front w28,29x. Astonishingly, the emission of the fs-plasma looks as if it comes from a symmetrically expanding cloud. But even if a shock wave is generated by the expanding mate-
rial, there is no laser light, which causes an extra heating of this absorption zone. Further research needs to be conducted to explain this observation. Fig. 5 shows the typical time dependence of the Cu line intensity Ž s 521.82 nm. measured at two pressures with ns- and fs-pulses. In these experiments the emission was collected from the central part of the plasma, approximately 3 mm above the sample surface. Each data point was measured after 10 pre-shots and the data from 50 laser pulses were averaged. The best fit of the data was obtained with a two-exponential model. The two-exponential fitting with two different
Fig. 5. Time dependence of the Cu emission Ž s 521.81 nm. at two different pressures. The data points are fitted with two exponential decay functions.
1778
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
characteristic times can be explained by different mechanisms dominating the plasma evolution at different stages. At the beginning of the plasma evolution, fast electron three-body recombination dominates, while cooling of the plasma is the dominating process at later times. It can be seen that there are two distinctive decay regimes: for longer times Ž) 4 s. the decay times 2,ns and 2,fs in the ns- and fs-cases are quite similar, and in both cases they decrease approximately twofold when the ambient argon pressure is increased from 40 to 140 mbar. This can be explained by faster cooling of the expanding plume at higher pressures of the buffer gas. Any memory of the starting conditions has been lost at these times. Therefore, the plasma evolution is similar for ns- and fs-pulses. The differences in plasma dynamics for ns- and fs-pulses are more pronounced at the beginning of the plasma evolution. Before ; 4 s, there is practically no influence of the buffer gas pressure on the emission decay time 1,ns in the ns-case, and there is an approximately two-fold decrease in the decay time 1,fs at 140 compared to 40 mbar in the fs-case. At this stage the ns-plasma is
still very hot because of the efficient inverse Bremsstrahlung heating during the laser pulse. This heating mechanism becomes more efficient with increasing pressure and the plasma is sustained for longer periods w19x. In the case of a fs-pulse, the plume of ablated material propagates without any extra heating and the dynamics of plume evolution is significantly affected by the ambient buffer-gas. This mechanism is responsible for the differences in the pressure dependence of the emission line intensities. Fig. 6 shows this pressure dependence of the intensity of the same copper atomic line Ž s 521.82 nm. integrated over the first 5 s for the fs- and ns-ablation modes. The same optical arrangement as for the data in Fig. 5 was used. Evidently, the dependence of the emission intensity on the pressure at the first stages of the plasma evolution is much less pronounced in the ns- than in the fs-case. 3.3. Analytical results A series of measurements were made to study the influence of the laser pulse duration on the
Fig. 6. Pressure dependence of the Cu emission Ž s 521.81 nm. integrated over the first 5 s after the laser pulse. Pulse energy, 400 J.
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
line intensities of the analytes. Most experiments were performed with brass samples. The emission o . lines of CuŽI. 521.820 nm Ž4d 2 D5r2 ᎐4p 2 P3r2 3 3 o and ZnŽI. 481.053 nm Ž5s S᎐4p P2 . were selected for the following reasons. Both lines have comparable transition probabilities; the energy difference between the upper levels of both lines is relatively small Ž; 3700 cmy1 .; and neither of the lines interfered with emission lines of other elements in the sample or of argon. Unfortunately, these lines could not be detected simultaneously within the same spectral window of the spectrograph. Other pairs of lines, e.g. 330.25 nm ŽZn.r328.51 nm ŽCu. or 518.2 nm ŽZn.r520.09 nm ŽCu., were not used because the line strengths were too different and line intensities of the weaker lines dropped down below the level of detection within the first few s. To avoid the problems with line self-reversal in hot laser plasma, the strong Cu line at 324.75 nm, frequently used in LA-OES experiments, was not selected. To find the best integration time for line intensity measurements, the temporal behavior of Cu and Zn line intensities and the intensity ratio for ns- and fs-modes of ablation were measured. These temporal profiles were measured with a gate time of 100 ns by changing stepwise the delay between the laser pulse and a gate pulse. After approximately 2᎐5 s the intensity ratio of the lines IZn rICu is constant within the experimental error bars. During these first s atomization of the ablated droplets or micro-particles is completed w22x. After approximately 15᎐20 s the line intensities of both elements were too low, and extended integration of the emission did not provide any improvement in the sensitivity. Therefore, the line intensities were integrated between 2 and 20 s. The measurements were performed at 40 and 900 mbar argon pressure. The 40-mbar level was chosen because of the maximum emission intensities in the fs-case. With ns-pulses the difference between the maximum signal at 80 and the signal at 40 mbar was less than 10%. A pressure of 900 mbar was selected because it is close to the atmospheric pressure used in LA-ICP-OESrMS. The coupling of laser
1779
ablation with ICP-MS is planned for the next stage of our investigations. Non-linear calibration curves for Zn and Cu in the case of LA of brass samples are well documented for LA-ICP techniques with different laser wavelengths and for ns- and ps-pulses w30,31x. It is interesting to investigate the same peculiarities of the calibration plots of these elements in the case of LA of the brass samples with fs-pulses. As the first step for such investigations the question of fractional evaporation in the case of fspulses was studied. An enhancement in the emission signal of one component due to fractional evaporation of the sample during the ablation process should be accompanied by a depletion of this component in the laser-affected zone. To check for preferential evaporation, the ratio Izn rICu was measured for successive laser shots on the same spot of the sample. Figs. 7 and 8 show representative results for two different pressures and two different energy levels. After a definite number of pre-shots Žwhich are indicated as the argument values on the abscissa in Figs. 7 and 8. the line intensities of the subsequent 10 shots were accumulated and further processed. The selected lines of Cu and Zn were sequentially measured in two different spectral windows. The procedure was repeated twice for each element. Error bars were calculated from the data sets of the two measurements. To improve the signal-to-noise ratio in the highpressure range, higher laser energies had to be used. Fractionation should appear as a variation of the IZn rICu ratio dependent on the number of shots. Within the experimental error bars, no change in the intensity ratio based on the number of shots was detected, either for fs- or ns-laser pulses. Differences in the intensity ratios from the certified ratios stem from changes in the plasma conditions. This result corresponds well with the published data w22,30,31x for ns-lasers with intensities larger than 1 GW cmy2 . For the fs-pulse energy of 4 J Ž0.4 J cmy2 . there is some indication of the effect Žsee Fig. 7.. However, within the given precision of our measurements a final statement cannot be given. Further detailed
1780
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
Fig. 7. ZnrCu line intensity ratio as a function of the number of shots for fs laser ablation at two different pressures. Each point represents the accumulation of 10 shots.
experiments with fs-pulses of low energy and an improved optical arrangement need to be performed to come to a more definite conclusion. It is noteworthy that at higher pressure the line intensity decreases faster in the fs-case than in the ns-case Žsee Fig. 6.. However, the precision of the intensity ratio is still higher with fs- than with ns-pulses, a trend that has been already found in the comparison of ns- and ps- ablation w4x. Fig. 9 shows the measured intensities of two Cu and one Zn lines obtained with fs- and ns-pulses at low pressure. The dependence of the line intensities on the ZnrCu ratios is similar. The curves are highly non-linear and show decreasing Cu line intensities with increasing Cu content, which is similar to the trend reported by Gagean and Mermet w30x who studied the influence of
different laser wavelengths on the ablation process in detail. Their detector was an inductively coupled plasma atomic-emission spectrometer. A complementary study of the efficiency of laser ablation was thoroughly investigated by Borisov et al. w31x, who used three different laser systems with 35 ps, and 6 and 30 ns at 532, 266 and 252 nm, respectively. In w31x the ablated material was transported to and detected in an inductively coupled plasma mass-spectrometer. These measurements showed that non-linearity in calibration graphs during laser ablation of brass samples are independent of the laser pulse length or wavelength, and are most probably connected with changes in the mass ablation rate. This behavior was explained by structural changes that influence the melting temperature w30x and, as specu-
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
1781
Fig. 8. ZnrCu line intensity ratio as function of the number of shots for ns laser ablation at two different pressures. Each point represents the accumulation of 10 shots.
Fig. 9. Calibration curves for Cu and Zn for fs and ns ablation at low pressure. The non-linearity does not depend on the wavelength of the emitting line or the duration of the laser pulse.
1782
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
lated in w31x, the reflectivity of the sample surface. Therefore, the mass ablation rate is a function of the ZnrCu ratio. This picture is complicated by the observation w30x that there is a relation between the non-linearity of the Cu calibration and the thermal pre-treatments of samples during their production. Good calibration for Zn was obtained when internal standardization against the Cu signal was applied. The same behavior was found in our experiments with fs-pulses and direct detection of the emission spectra of a laser plasma. Therefore, the observed non-linearity of the calibration plots for the two elements is not attributed to changes in the plasma conditions. Most likely the non-linearity is due to changes in the ablated mass. Under the same pressure conditions and with a similar laser energy, calibration graphs of the
ZnrCu line ratio were obtained. The results are shown in Figs. 10 and 11. High laser energy and low pressure is favorable, as can be expected from the previous measurements. Linear calibration curves were obtained in all cases. The linear regression shows that the correlations for ns-ablation at higher pressure or at lower energy are poor. In general, the calibration curves for different pulse energy and pressure in the fs-case have better correlation coefficients. To numerically compare the precision of an analysis for fs- and ns-pulses, we removed one standard ŽTD03, ZnrCu ratio 0.424. from the set of standards used for construction of calibration graphs for both cases. This standard was used as an ‘unknown’ sample and the ratio was determined using the calibration graphs for fs- and ns-mode with 40 mbar and the highest energy
Fig. 10. Calibration for the ZnrCu ratio at different pressures and laser energies for fs laser pulses.
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
1783
Fig. 11. Calibration for the ZnrCu ratio at different pressures and laser energies for ns laser pulses.
Žsee Figs. 10 and 11.. The values obtained for ZnrCu ratios were Ž0.410" 0.032. and Ž90.420" 0.065. for the fs- and ns-cases, respectively. The confidence intervals which are given within the brackets are calculated for a level of confidence, Ps 0.95. While the mean values agree reasonably, the confidence interval for ns-pulses is twofold larger than for fs-pulses. Thus, ablation with fs-pulses provides better precision for the direct analysis of brass samples compared to ns-pulses.
4. Conclusion It has been shown that there are significant differences in the ablation process under atmospheric and reduced buffer-gas pressures when using ns- and fs-laser pulses. Experimentally observed differences in the temporal evolution of
atomic and ionic line intensities for fs- and nspulses were explained by the absence of a laserheated buffer-gas plasma above the sample surface in case of fs-pulses, and post-atomization and -ionization of the ablated species in a hot gas plasma in the case of ns-pulses. An expanding, recombining plasma of ablated species is created in the case of laser ablation with fs-pulses. The ablation process and subsequent excitation of atoms is more reproducible for fs- than for nspulses. This results in a lower ablation threshold, a higher sensitivity and an improved precision. Non-linear calibration graphs for Cu and Zn were found for both lasers. The non-linear behavior is very similar to that observed for LA-ICPOESrMS techniques and was attributed to structural changes in the alloys, which result in different ablation rates for the samples with different Zn᎐Cu compositions, both in the ns- and fs-cases.
1784
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785
By internal standardization of the Zn to the Cu line intensity, a linear calibration was obtained. Using one of the standards as an unknown sample, its ZnrCu ratio was determined from the fsand ns-laser calibration graphs. While the mean values calculated for both cases coincide reasonably, the confidence interval for ns-pulses was two-fold larger than for fs-pulses. For the applied pulse energy range 20᎐500 J and laser intensity range 10 8 ᎐10 13 W cmy2 , no changes in the IZn rICu line intensity ratio for successive ablation pulses on the same spot were observed, either in the fs- or ns-case.
w10x
w11x
w12x
w13x
Acknowledgements Financial support from the Deutsche Forschungsgemeinschaft ŽProject HE 1941r2-1. is gratefully acknowledged.
w14x
w15x
References w1x F. Leis, W. Sdorra, J.B. Ko, K. Niemax, Basic investigations for laser microanalysis I: optical emission spectrometry of laser-produced sample plumes, Mikrochim. Acta II Ž1989. 185᎐190. w2x W. Sdorra, A. Quentmeier, K. Niemax, Basic investigations for laser microanaylsis II: laser-induced fluorescence in laser-produced plasma plumes, Mikrochim. Acta II Ž1989. 201᎐209. w3x R.E. Russo, Laser ablation, Appl. Spectrosc. 49 Ž9. Ž1995. 14A᎐28. w4x L. Moenke-Blankenburg, Laser-ICP spectrometry, Spectrochim. Acta Rev. 1 15 Ž1993. 37. w5x S.A. Darke, J.F. Tyson, Interaction of laser radiation with solid materials and its significance to analytical spectrometry, J. Anal. At. Spectrom. 8 Ž1993. 145᎐209. w6x S.E. Jackson, H.P. Longerich, G.R. Dunning, B.J. Fryer, The application of laser-ablation microprobe-inductively coupled plasma-mass spectrometry ŽLAM-ICP-MS. to in situ trace element determinations in minerals, Can. Mineral. 30 Ž1992. 1049᎐1064. w7x T.E. Jeffries, S.E. Jackson, H.P. Longrich, Application of a frequency quintupled Nd:YAG source Ž216 nm. for laser ablation inductively coupled plasma mass spectrometric analysis of minerals, J. Anal. At. Spectrom. 13 Ž1998. 935᎐940. w8x R.R. Loucks, Annual Report, ANU, 1995. w9x D. Gunther, R. Frischknecht, C.A. Heinrich, H.-J. ¨ Kahlert, Capabilities of an argon fluoride 193-nm ex-
w16x
w17x w18x
w19x
w20x w21x
w22x
w23x
w24x
cimer laser for laser ablation inductively coupled plasma mass spectrometry microanalysis of geological materials’, J. Anal. At. Spectrom. 12 Ž1997. 939᎐944. W. Sdorra, K. Niemax, Basic investigations for laser microanalysis III: application of different buffer gases for laser-produced sample plumes, Microchim. Acta 107 Ž1992. 319᎐327. W. Sdorra, J. Brust, K. Niemax, Basic investigations for laser microanalysis IV: the dependence on the laser wavelength in laser ablation, Microchim. Acta 108 Ž1992. 1᎐10. D. Gunther, S.E. Jackson, H.P. Longerich, Laser abla¨ tion and arcrspark solid sample introduction into inductively coupled plasma mass spectrometers, Spectrochim. Acta Part B 54 Ž1999. 381᎐409. A. Semerok, C. Chaleard, V. Detalle, J.-L. Lacour, P. Mauchien, P. Meynadier, C. Nouvellon, B. Salle, P. Palianov, M. Perdix, G. Petite, Experimental investigations of laser ablation efficiency of pure metals with femto-, pico- and nanosecond pulses, Appl. Surf. Sci. 311 Ž1999. 138᎐139. A.P.K. Leung, W.T. Chan, X.L. Mao, R.E. Russo, Influence of gas environment on picosecond laser ablation sampling efficiency and ICP conditions, Anal. Chem. 70 Ž1998. 4709᎐4716. B.N. Chichkov, S. Momma, F. von Alvensleben, A. Tunnermann, Femtosecond, picosecond and nanosecond ¨ laser ablation of solids, Appl. Phys. A 63 Ž1996. 109᎐115. D. von der Linde, H. Schuler, ¨ Breakdown threshold and plasma formation in femtosecond laser᎐solid interaction, J. Opt. Soc. Am. B 13 Ž1996. 216᎐222. S. Preuss, A. Demchuk, M. Stuke, Sub-picosecond UV laser ablation of metals, Appl. Phys. A 61 Ž1995. 33᎐37. A. Miotello, R. Kelly, Laser-induced phase explosion: new physical problems when a condensed phase approaches the thermodynamic critical temperature, Appl. Phys. A 69 Ž1999. 67᎐73 ŽSuppl.. S. Amoruso, R. Bruzzese, N. Spinelli, R. Velotta, Characterization of laser-ablation plasmas, J. Phys. B 32 Ž1999. R131᎐R172. J. Baldwin, Q-Switched laser sampling of copper᎐zinc alloys, Appl. Spectrosc. 24 Ž1970. 429᎐435. E. Cromwell, P. Arrowsmith, Fractionation effects in laser ablation inductively coupled plasma mass spectrometry, Appl. Spectrosc. 49 Ž1995. 1652᎐1660. J.B. Ko, W. Sdorra, K. Niemax, On the internal standardization in optical emission spectrometry of microplasmas produced by laser ablation of solid samples, Fresenius Z. Anal.Chem 335 Ž1989. 648᎐651. H.K. Tonshoff, F. von Alvensleben, A. Ostendorf, G. ¨ Kamlage, S. Nolte, Micromachining of metals using ultrashort laser pulses, Int. J. Electr. Mach. 4 Ž1999. 1᎐6. F.H. Loesel, J.P. Fischer, M.H. Gotz, ¨ C. Horvarth, T. Juhasz, F. Noack, N. Suhm, J.F. Bille, Non-thermal ablation of neural tissue with femtosecond laser pulses, Appl. Phys. B 66 Ž1998. 121᎐281.
V. Margetic et al. r Spectrochimica Acta Part B: Atomic Spectroscopy 55 (2000) 1771᎐1785 w25x D. Strickland, G. Mourou, Compression of amplified chirped optical pulses, Opt. Commun. 56 Ž1985. 219᎐221. w26x K. Tamura, E.P. Ippen, H.A. Haus, L.E. Nelson, 77-fs pulse generation from a stretched-pulse mode-locked all-fiber ring laser, Opt. Lett. 18 Ž1993. 1080᎐1082. w27x W. Sdorra, Ph.D. Thesis, ISAS, University of Dortmund, 1991. w28x F. Garrelie, C. Champeaux, A. Catherinot, Expansion dynamics of the plasma plume created by laser ablation in a background gas, Appl. Phys. A 69 Ž1999. 55᎐58 ŽSuppl.. w29x W. Pietsch, B. Dubreuil, A. Briand, A study of laser-pro-
1785
duced copper plasma at reduced pressure for spectroscopic applications, Appl. Phys. B 61 Ž1995. 267᎐275. w30x M. Gagean, J.M. Mermet, Study of laser ablation of brass materials using inductively coupled plasma atomic emission spectrometric detection, Spectrochim. Acta Part B 53 Ž1998. 581᎐591. w31x O.V. Borisov, X.L. Mao, A. Fernandez, M. Caetano, R.E. Russo, Inductively coupled plasma mass spectrometric study of non-linear calibration behavior during laser ablation of binary Cu᎐Zn alloys, Spectrochim. Acta Part B 54 Ž1999. 1351᎐1365.
