JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 Comparison Between Infrared and Ultraviolet Laser Ablation at Atmospheric Pressure-Implications for Solid Sampling Inductively Coupled Plasma Spectrometry Christian Geertsen* Alain Briand Frederic Chartier Jean-Luc Lacour Patrick Mauchient and Sten SjostromS Laboratoire de Spectroscopie Laser Commissariat a L’Energie Atomique Service de Physique d’Experimentation et d’Analyse Centre d’Etudes de Saclay BSt. 397 9 7 79 7 Gif-sur-Yvette Cedex France. Jean-Michel Mermet Laboratoire des Sciences Analytiques Bit. 308 Universite C. Bernard Lyon 7 69622 Villeurbanne Cedex France. 17 The efficiency of laser solid sampling was investigated as a function of several experimental parameters under experimental conditions similar to those used for laser ablation (LA) inductively coupled plasma spectrometry. This was done by studying the amount of material removed as a function of melting temperature of the (metallic) matrix material laser wavelength and laser energy and by studying the plasma ignition in air and argon buffer gases as a function of laser wavelength.It was found that direct LA is the major process responsible for the removal of material in the case of a UV laser as opposed to with an IR laser where shielding of the laser radiation by the absorbing plasma limits direct LA and increases the temperature of the plasma. The consequences of this difference between IR and UV laser radiation are considerable and lead to a superior performance of UV laser sampling in every analytical aspect reproducibility matrix effects quantification spatial resolution and sensitivity.Keywords Laser-produced plasma; solid sampling; laser ablation; inductively coupled plasma spec- trometry; trace analysis Laser ablation (LA) for direct solid sampling and subsequent analysis was introduced in the early 1960s. This technique presents several important features such as direct analysis of both non-conducting and conducting materials localized lat- eral analysis no sample preparation prior to analysis and operation at atmospheric pressure. However in spite of the important potential of this sampling technique it is only recently that results have been obtained that compare favour- ably with other solid sampling techniques. This is mainly owing to two major developments during the past few years.Firstly the lasers used for laser solid sampling today are more reliable stable and easy to handle than in the past. Secondly the use of inductively coupled plasma mass spectrometry (ICP-MS) for the detection of the ablated material offers a performance previously unmatched in terms of sensitivity time of analysis and ease of use. All commercially available instruments for LA-ICP-MS are equipped with a pulsed Nd:YAG laser operating in the IR at its first harmonic (1064 nm). This choice of laser is based on two essential reasons firstly it is today the cheapest and simplest choice of laser for these kinds of applications and secondly the results of for example Gray’ and Arrowsmith’ have shown that high sensitivities can be achieved typically in the sub-pg g-’ range for solid samples.However in spite of the analytical efforts of numerous laboratories the technique still exhibits an unsatisfactory reproducibility (above loo/,) for many applications. This is for example the case for quality control of many industrial products where a precision of the order of 0.1% is required for the major elements and a few percent for the minor and trace elements. The levels of precision required for trace and minor elements are obtainable by ICP atomic emission spectrometry (AES) if * On leave from the Pechiney Centre de Recherches de Voreppe t To whom correspondence should be addressed. 1 On leave from the Department of Physics Chalmers University BP 27 38340 Voreppe France. of Technology 5-412 96 Sweden.liquid nebulization sample introduction is used. The shot-to- shot stability of the laser itself is of the order of a few percent and averaging over many laser shots should lead to reproduci- bilities of better than 1%. Hence the basis for the moderate accuracy and precision of LA-ICP-MS is to be found in the physics of the laser-surface interaction. The purpose of this paper is to investigate the ablation efficiency as a function of several experimental parameters. This was done by studying the amount of material removed as a function of matrix material laser wavelength and laser energy and by studying the plasma ignition in different buffer gases as a function of laser wavelength. Experiment a1 Three sets of experiments were performed. The experimental arrangement (see Fig.1) for the production of the plasmas consisted of a laser a focusing lens and the target for the first two sets of experiments with the addition of a charge-coupled device (CCD) camera with imaging optics for the third set of experiments. Laser I Lens f= 100 rnm I7 Diaphragm Lens f=25 rnm Plasma camera Fig. 1 Experimental arrangement for LA of solid targets. The intensi- fied charge-coupled device (CCD) camera with imaging optics and the delay unit were used for the registration of the plasma emission in Experiment 318 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 Experiment 1 Several laser sources emitting at different wavelengths were used to determine the ablation efficiency on a copper target in air buffer gas as a function of laser energy and laser wavelength see Fig.1. The laser sources used were the first three harmonics of a Q-switched Nd:YAG (Quantel YG-585-10) and two excimer lasers with XeCl emitting at 308 nm (Lambda-Physik EMG 102 MSC) and with ArF emitting at 193 nm. The laser energy was controlled with a power meter (Scientech 372). The pulse durations were approximately 10 ns for the Nd:YAG and 30 ns for the excimer lasers. The laser beams were incident on the target at normal angle and focused on the target using a plano-convex lens with a focal distance of 100 mm (Suprasil for the UV wavelengths and Herasil for the IR). The focal spot area was approximately 2 x lop4 cm2 for the Nd:YAG and 1 x The ablated mass defined by the displaced volume was measured by integrating the volume of the crater with a high- performance scanning profile-meter (laboratory made).The position of the sample relative to the lens was selected to be that which gave the smallest crater size. cm2 for the excimer lasers. Experiment 2 Several laser sources emitting at different wavelengths were used to determine the ablated mass uersus fusion temperature of several targets in air buffer gas at different wavelengths at a laser energy of 85 mJ per laser pulse corresponding to a fluence of 90 J cmp2 for the excimer lasers and 425 J cmP2 for the Nd:YAG laser. The laser sources used were the first ( 1064 nm) and second (532 nm) harmonics of the Nd:YAG and the XeCl excimer laser emitting at 308 nm. The samples covered a range of melting temperatures from 700 (zinc) to 3700 K (tungsten).The laser beams were focused on the surface in the same way as in Experiment 1. Experiment 3 The role of the buffer gas and the laser wavelength in ignition of the plasma was investigated by means of time-resolved images of the emission from the laser produced plasma. The ablation was performed with the first harmonic (1064 nm) as well as the third harmonic (355 nm) of an Nd:YAG laser (Quantel Compact YG-585-30). The pulse energy was 10mJ per pulse a value commonly encountered in LA,3 and the duration of the laser pulse was 6ns [full width at half maximum (FWHM)]. The energy was adjusted by slightly varying the angle of a dielectric mirror reflecting 100% at 45" and not by adjusting the supply voltage since this changes the energy distribution in the laser beam.The laser beam was focused at right angles onto the surface of the sample using a plano-convex lens with a focal distance of 100 mm. The positioning of the samples was carried out by moving the sample until the highest acoustic levels were found. It is experimentally very easy to determine this position and this method was used rather than the technique described above as the plasma images exhibited the same characteristic features within a range of about 10mm around the optimal sample position. At the first harmonic (1064 nm) it was found that the highest acoustic level was obtained when the laser beam was focused slightly in front of the sample (2mm) whereas the laser was focused on the surface in the case of the third harmonic.The positioning of the sample at the focal point is often done by determining the position where maximum mass is ablated but the proposed method is much simpler and the results obtained by the two methods compare well with one another for UV lasers it is well-known that ablated mass increases with the acoustic ~ i g n a l ~ and for IR lasers we have observed that crater diameters do not change whether the focus is slightly in front of or on the surface of the target. Irradiation areas on the target were determined using the 'knife-edge' m e t h ~ d . ~ For the IR laser radiation the diameter was approximately 150-170 pm (60-80 pm at the focal point) while in the UV the diameter was approximately 100-120 pm. Measurements were made for aluminium and copper targets and were carried out at atmospheric pressure both in air and in argon buffer gas.Images of the plasma were taken with an intensified gated CCD camera (Hamamatsu C4346-Ol) with a gate time of 3 ns at right angles to the laser beam. A biconvex lens with a focal distance of 25mm was used for the imaging of the plasma giving a magnification of 10 x on the CCD. A diaphragm with a diameter of 3 rnm (yielding aroundfll0) ensured that spheri- cal aberrations were small while two colour filters (Schott BG3 and GG400) limited the spectral range to 400-500nm and thus virtually eliminated chromatic aberrations and stopped scattered laser radiation from reaching the CCD detector. Each image was taken after about 1000 laser shots in the same place on the target. This choice was based on earlier findings that the optical emission signal evolves during the first laser shots and is stable thereafte~-.~.~ The CCD camera was triggered from the laser trigger output.A jitter of about 2 ns was measured for this trigger signal. The time delay for image registration was measured relative to the beginning of the laser pulse. The intensity of the light emission is represented by false colours in the images. Results and Discussion In the first set of experiments the amount of material removed from a copper target as a function of wavelength and laser energy was investigated. In Fig. 2 it is shown that the ablation efficiency (ablated mass per unit energy per unit surface) as a function of laser fluence is more than one order of magnitude higher (20 times) for a UV laser than for an IR laser or even a visible laser at 200 J cm-2.It should be noted that for the same output energy of approximately 200 mJ (corresponding to 210Jcm-2 for the excimer and 1000Jcm-2 for the Nd:YAG) the excimers emitting in the UV ablates a factor of 7 more mass than the Nd:YAG emitting in the IR. Similar results were also obtained for targets made of molybdenum nickel and tungsten while the dependence on laser wavelength of the amount of material removed from a zinc target was less pronounced. In the second set of experiments the ablated mass versus melting temperature of different targets in air buffer gas at three different laser wavelengths was determined see Fig. 3. In the case of UV radiation a linear correlation exists between the ablated mass and the melting temperature of the sample material.In the case of IR radiation the differences in ablated mass are considerably larger and there exists no such simple 2 70 1 0 200 400 600 800 1000 1200 1400 Laser fluence/J cm-' Fig. 2 Ablation efficiency (mass ablated per laser fluence) on a copper target in air buffer gas as a function of laser energy for four different lasers. A ArF 193nm; B XeCl 308nm; C Nd YAG 355nm; D Nd:YAG 532 nm; and E Nd:YAG 1064 nmJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 19 Melting-point/K Fig. 3 Ablated mass versus fusion temperature of several targets (the element indicated was the major element in the target >99%) in air buffer gas at different laser wavelengths at a laser energy of 85 mJ per laser pulse (corresponding to a fluence of 90 J cm-' for the excimer laser (A XeCl 308nm) and 425 Jcm-' for the Nd:YAG lasers (B 532 nm; and C 1064 nm) relation between the ablated mass and the melting temperature of the sample material. The small number of measuring points in the case of visible laser radiation indicates a similar behav- iour as for IR radiation.This indicates a fundamental difference in the physics involved in the ablation process and might have important analytical consequences since it drastically reduces matrix effects in the case of UV laser sampling. The results obtained here for the IR laser agree well with results previously published by Kawaguchi et al.,' Ishida and Kubota' and Iidal' and indicate an almost exponential decrease in the ablated mass as a function of melting temperature in the case of an IR laser.In the third experiment three series of images of the emission from laser produced plasmas were obtained with a time resolution of 3 ns. The first series of four images shows plasmas produced by an IR laser on a copper target in argon buffer gas at a delay of approximately 30ns after the beginning of the laser pulse (6ns FWHM) (see Fig.4). The experimental conditions for the second series of images were identical to the first series except that air was used instead of argon as buffer gas; a representative plasma is shown in Fig. 5(a). The last series of images shows a plasma produced a UV laser on a copper target in argon buffer gas at a delay of approximately 30 ns after the beginning of the laser pulse [see Fig.5(b)]. The first observation is that the plasmas produced by an IR laser in argon are considerably more complex than those produced by a UV laser see Figs. 4 and 5(b) exhibiting several distinct plasmas. It is also clear from Fig. 4 that the reproducibility is poor for the plasmas produced by an IR laser in argon. The plasmas produced by an IR laser in air buffer gas also had a complex structure with two hot spots but the reproducibility was better than for the argon buffer gas. The plasmas produced by a UV laser in argon buffer gas Fig. 5(b) were confined close to the surface and consisted of one single plasma. The reproducibility of the plasma images with UV was considerably better than with IR produced plasmas.It should be noted that the experimental conditions used to obtain the images in Fig. 4 correspond well with the operating conditions used in commercial LA-ICP-MS apparatus. The lack of reproducibility of the plasma ignition observed in Fig. 4 probably explains the poor reproducibility observed in LA-ICP-MS measurements. The images suggest the following physical interpretation. When a high-power laser is focused on a metal target the intense radiation leads to rapid heating and evaporation of the solid. At the same time free electrons are created either by thermoelectric emission by multiphoton photoemission" or by multiphoton ionization of the metallic vapour or the ambient gas. These electrons absorb the incoming photons by inverse bremsstrahlung. This is necessarily a three-body process for reasons of conservation of both energy and momentum.The third body can be either an atom or an ion but the cross- sections are several orders of magnitudes greater for ions than for atoms. Therefore the heating of the electrons begins slowly before they acquire enough energy to ionize the gas collision- ally. This starts the cascade that leads to breakdown. In addition since the inverse bremsstrahlung cross-section of the ions relate to the wavelength as [1-exp(-hc/AkTjA3 the plasma will be much more absorbing in the IR than in the UV.12 Figs. 4 and 5(a) show that with IR radiation the absorption of the plasma is so strong that the energy is deposited at the leading edge of the plasma where it can lead to secondary or multiple breakdowns.This is not the case with UV laser produced plasmas where no breakdown of the buffer gas was observed [Fig. 5(b)] confirming results previously reported by Autin et al.13 The experimental observation that the highest acoustic levels were obtained when the laser beam was focused slightly in front of the sample (2 mm) with an IR laser whereas the beam was focused on the surface for a UV laser is another confirmation of a breakdown of the buffer gas in the case of IR laser produced plasmas. As a consequence of its higher absorption cross-section the IR plasma is hotter than the UV plasma14 and less energy in the laser beam interacts directly with the sample surface. In other words the plasma produced by the IR laser acts as a shield that only transfers a small part of the incoming laser energy towards the solid surface. These findings support the view that the UV and IR plasmas are essentially different as they interact in fundamentally different ways with the incoming laser radiation.The absence of multiple breakdowns in air buffer gas for IR produced plasmas could be explained by an easier ionization of argon gas than air and thus a higher inverse bremsstrahlung absorption coefficient due to the resulting ions. Although the ionization potential of argon is higher than that of nitrogen and oxygen metastable states in argon can serve as relay states and thereby facilitate the onset of a breakdown cascade which would explain the seemingly lower ionization potential of argon.15 Photographs taken with a scanning electron microscope of the crater forms obtained with UV and IR lasers respectively are presented in Fig.6. In the case of UV [see Fig. 6(c) and ( d ) ] there is a crater with well defined borders surrounded by a circular surface (with a diameter of approximately 600 pm) that is only slightly affected. With an IR laser [see Fig. 6(a) and (b) the crater formation is totally different without any clear limit between the actual crater and the surrounding affected zone. In addition the presence of a 'lip' of metal (indicated by an arrow) that had flowed out of the crater before it was solidified is visible in Fig. 6(b). The presence of the large affected region visible in Fig. 6(a) proves that it is the IR plasma that erodes the sample since the distances of heat diffusion in the targets used here (copper and aluminium16) are of the order of only 1 pm in 10 ns and 10 pm in 1 ms and thus too small to allow the heat deposited by the laser in the focal spot to diffuse over such a large surface.As a consequence of the shielding effect present in the case of an IR laser the relative importance of direct LA and plasma erosion are different for IR and UV lasers with UV radiation the direct laser interaction regime lasts the duration of the laser pulse (6 ns) whereas the importance of the plasma erosion is negligible [see Fig. 6(c)]. With IR radiation the direct laser interaction regime lasts only a fraction of the laser pulse (the time to reach breakdown probably less than 1 ns according to a model developed by Rosen et a1.17) and together with the fact that this plasma is considerably hotter the plasma erosion with an IR laser was found to be very important for the crater formation [see Fig.6(a) and (b)] and possibly also for the removal of material.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 ( b) Fig. 4 Images of the optical emission from a plasm,a produced by an Nd:YAG laser emitting at 1064 nm (IR) on a copper target in argon buffer gas at a delay of approximately 30 ns (after the beginning of the laser pulse). The laser beam is incident from the left and the target is situated vertically on the images at the right-hand side of the line shown above the plasmas (reflections on the metallic surface gives the impression that the plasma was situated inside the target).All four images (a)-@) are taken under identical experimental conditions Fig. 5 Images of the optical emission from a plasma produced by an Nd:YAG laser emitting at (a) 1064 nm (IR) ion a copper target in air buffer gas at a delay of approximately 30 ns (after the beginning of the laser pulse) and (b) 335 nm (UV) on a copper target in argon buffer gas at a delay of approximately 30 ns. The laser beam is incident from the left and the target is situated vertically on the image at the right-hand side of the line shown above the plasmas (reflections on the metallic surface gives the impression that the plasma was situated inside the target) In a publication by Hager," the relative elemental response for LA-ICP-MS was investigated using an Nd:YAG laser operating in the IR (Q-switched and free running).The results obtained confirmed that a thermal process is responsible for the evaporation in an IR laser. In a model developed by the author it was assumed that all the energy in the laser beam was deposited as heat in the target and that no screening effect was present. A formula was derived for the mass evaporated (for one specified matrix) that only included the heat of vaporization and the temperature of the solid which demon- strated good agreement between experiment and theory. Our interpretation based on the results presented by Hager," isJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL. 9 (a) (6) 21 - 100 pm 100 pm - 7 100 pm 100 pm Fig. 6 Scanning electron microscope images of the craters produced by 1000 laser shots from (a) and (b) an IR laser and (c) and ( d ) a UV laser in a copper target.The craters shown in Fig. 7(a) and (c) are given in greater detail in Fig. 7(b) and ( d ) respectively. The lip of molten metal is indicated by an arrow that the thermal model worked so well because it was the plasma (duration of some ps) rather than the laser (duration of 6 ns) that was responsible for the selective thermal evapor- ation of the elements. As has already been pointed out the curves in Fig. 2 show that another much more efficient ablation process is at work when using a UV laser. The crater images show that this process takes place with only minimal disturbance of nearby material. The process is thus non-selective and even though plasma erosion is also taking place the UV LA is globally less selective than IR LA because of the relative efficiencies of the two processes. It has previously been shown by Leis et d7 that excellent analytical signals can be obtained with LA optical emission spectrometry using an Nd:YAG laser at its fundamental IR frequency at a reduced pressure of the argon buffer gas of 140 hPa.The results obtained in the present study are not however do not contradict these results. According to the investigations of Rosen and Weyl,” the irradiation threshold for the break- down cascade is approximately the inverse of the pressure of the buffer gas times the square of the wavelength (pA2)-’ Hence when Leis et ~ 1 . ~ used 1064 nm instead of 355 nm they off-set the variation in breakdown thresholds due to the wavelength change by decreasing the pressure and could thus perform the ablation without screening by the plasma or formation of multiple plasmas.Conclusion The results presented in this paper demonstrate large differ- ences between UV and IR lasers for laser solid sampling. The mass ablated by a UV laser is considerably larger than for an IR laser under operating conditions similar to those employed for LA-ICP-MS. Considering that the detection limits are already very good sub-ppm and that the amount of sample material that can be accepted by an ICP mass spectrometer is limited this advantage of UV laser sampling might not be capable of being exploited to its full extent in LA-ICP-MS. An excimer laser could because of the large masses ablated (Fig. 2) lead to blocking of the sampler; for the purposes of ICP-MS a frequency-tripled Nd:YAG laser would be quite sufficient.However if the detection of the ablated material is performed by ICP-AES the larger mass ablated by a UV excimer laser can be exploited to its full extent possibly leading to a similar analytical performance as LA-ICP-MS.20 Time resolved images of the optical emission from the early evolution of laser produced plasmas revealed a complex struc- ture and poor reproducibility in the case of an IR laser while a reproducible plasma with a simple geometrical structure was obtained with a UV laser. It is probable although no evidence is given in this paper that the reproducibility and complexity of the plasma emission is related to the reproducibility in22 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY JANUARY 1994 VOL.9 LA-ICP-MS measurements in which case considerably better reproducibility is to be expected if a UV laser is used instead of an IR laser for the laser sampling. The lateral spatial resolution of the laser sampling for metallic targets is given approximately by the laser focal spot size in the case of a UV laser while it is determined by the size of the plasma for an IR laser. Hence considerably better spatial resolution is obtained with UV laser sampling. Direct LA taking place during a few ns is the major process responsible for the removal of material in the case of a UV laser. This is in contrast to the case with an IR laser where shielding of the laser radiation limits the direct LA and increases the temperature of the plasma leading to an increased selective removal of material by the plasma.As a consequence the risk of selective vaporization is minimized if UV laser sampling is used for LA-ICP-MS. Measurements of the ablated mass versus fusion temperature strongly suggest a considerably simpler quantification of the signal with UV laser sampling. Hence for the operating conditions pertaining to LA-ICP-MS the results presented in this paper strongly indicate that UV laser sampling is far superior to IR laser sampling in every analytical aspect reproducibility matrix effects spatial resolution quantification and sensitivity. To conclude it is the authors' firm belief that the use of frequency- tripled ND:YAG laser sampling in LA-ICP-MS would lead to a new breakthrough of this technique with considerably better analytical characteristics.The authors express their appreciation to B. Dubreuil and V. Martin-Daguet for helpful discussions and L. Battery for his assistance in taking the SEM micrographs. S.S. acknowl- edges support from the Swedish-French research exchange foundation and the Swedish Institute. This work was in part funded by the French Research Group on Analytical Laser Ablation (GRAAL) with the following participants C.E.A. CNRS DILOR Jobin Yvon Pechiney Renault Sollac Saint- Gobain and the French Ministkre de la Recherche. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 References Gray A. L. Analyst 1985 110 551. Arrowsmith J. W. Anal. Chem. 1987 59 1437. Moenke-Blankenburg L. Spectrochim. Acta Rev. 1993 1 1. ' Chen G. and Yeung E. S. Anal. Chew. 1988 60 2258. Jones R. D. and Scott T. R. Laser Focus World 1993 123 185. AndrC N. Briand A. Lacour J.-L. Mauchien P. Semerok A. and Sjostrom S. in preparation. Leis F. Sdorra W. KO J. B. and Niemax K. Mikrochim. Acta 1989 11 185. Kawaguchi H. Xu J. Tanaka T. and Mizuike A. Bunzeki Kagaku 1982 31 El85 Ishida R. and Kubota M. J. Spectrosc. SOC. Jap. 1972 21 16. Iida Y. Spectrochim. Acta. Part B 1990 45 1353. Petite G. Agostini P. Guizard S. Martin P. and Trainham R. Laser Ablation of Electronic Materials ed. Fogarassy E. and Lazare S. Elsevier 1992 p. 12. Weyland G. M. and Rosen D. Phys. Rev. A 1985,31 2300. Autin M. Briand A. Mauchien P. and Mermet J. M. Spectrochim. Acta Part B 48 851. Girault C. Ph.D. Thesis UniversitC de Limoges France 1990. Zel'dovich Ya. and Raizer Yu. Sou. Phys. JETP (Engl. Transl.) 1965 20 772. von Allmen M. Laser Beam Interaction With Materials Springer- Verlag Berlin 1987. Rosen D. I. Hastings D. E. and Weyl G. J. Appl. Phys. 1982 53 5882. Hager J. W. Anal. Chem. 1989 61 1243. Rosen D. I. and Weyl G. J. Phys. D 1987 20 1264. Chartier F. Ph.D. Thesis 1'Universite Claude Bernard Lyon 1991. Paper 3/03535B Received June 21 1993 Accepted September 17 1993