LIBS Analysis of Geomaterials: Comparative Study of Basalt Plasma Induced by TEA CO 2 and Nd:YAG Laser in Air at Atmospheric Pressure

We present a study of the plasma generated by transversely excited atmospheric (TEA) CO 2 laser irradiation of a basalt sample. The plasma was induced in air at atmospheric pressure. The same sample was also analyzed using a commercial LIBS system based on Nd:YAG laser and time-gated detection. The main plasma parameters, temperature, and electron number density were determined and analytical capabilities of the two systems compared. Despite differences in laser wavelength, pulse duration, applied fluence, and signal detection scheme, the two systems are comparable in terms of element detectability and limits of detection. In both cases, all elements usually present in geological samples were identified. The estimated limits of detection for most elements were below 100ppm, while for Cu, Cr, and Sr they were around or below 10ppm. The obtained results led to the conclusion that simple, cost-effective TEA CO 2 LIBS system can find applications for geological explorations.


Introduction
e data on the chemical composition of geomaterials are of fundamental importance for exploitation and use of georesources, environmental studies, and space explorations. Because of that, development of rapid and accurate instrumental analytical methods for chemical analysis of geological materials is of prime importance. At the same time, the analysis of geomaterials is o en a challenging analytical task, as it requires determination of a large number of elements, over a wide range of concentrations, in a variety of complex matrices.
e well-established spectrochemical techniques that are mainly used in this eld are inductively coupled plasma optical emission spectroscopy (ICP-OES), inductively coupled plasma mass spectrometry (ICP-MS), and X-ray uorescence (XRF) spectroscopy. In recent years, a considerable progress has been made in the development of another spectrochemical technique, laser-induced breakdown spectroscopy (LIBS) [ ]. LIBS combines capability of providing fast multielemental analysis with no sample pretreatment, potential for in situ and remote analysis, the ability to perform surface and depth pro ling, and the ability to provide isotopic ratio information additionally to elemental composition [ ]. Regarding the analysis of geomaterials, LIBS has certain advantages compared to ICP and XRF. ICP-OES and ICP-MS have better analytical gures of merit such as accuracy and precision and limits of detection compared to LIBS. However, many geological samples are silicate rocks, and the main bene t of LIBS compared to ICP is that there is no need for solid sample digestion, that is, no need for complex and timeconsuming dissolution procedures. On the other hand, XRF spectrometry is a rapid method for in-eld elemental analysis of geological samples and requires simple and minimal sample preparation (e.g., fusion or preparation of pressed pellets) [ ]. ese characteristics are also common to LIBS. At the same time, compared to XRF, LIBS has higher sensitivity for elements like Mg, Si, Al, Ca, and K and especially for light elements like Li, Be, B, C, F, and Na. Also, unlike ICP and XRF, LIBS may be easily applied for depth pro ling by repeated ablation of the same location. In summary, each of the aforementioned analytical techniques is characterized by its own strengths and weaknesses, and the choice of the "best" technique is dictated by the analytical problem that has to be solved. In that sense, LIBS may be considered as spectrochemical technique complemental to ICP and XRF.
Due to its unique characteristics, LIBS can be used for elemental analysis of almost any type of material. However, the vast majority of LIBS applications deal with compact solid samples and, among them, perhaps the most analyzed sample types were geological materials. LIBS has been successfully applied, both in laboratory setting and in the eld, for chemical analysis of rocks, soils, sediments, and other natural materials [ , -]. One of the main problems in quantitative LIBS analysis of complex matrices (such as geological samples) comes from the matrix e ects which can reduce the accuracy and precision of the analysis [ , , ]. A number of calibration strategies were proposed for compensation of matrix e ects in LIBS, for example, di erent normalization procedures as well as calibration-free approach [ , ].
In this work, a recently developed LIBS system, based on transversely excited atmospheric (TEA) CO 2 laser and timeintegrated signal detection [ ], was applied to the qualitative analysis of basalt samples in air at atmospheric pressure. e basaltic rock was chosen as a representative of geological samples. Basalt is a very important rock as it is the most common rock in Earth's crust and is also an abundant rock on other satellites and planets, for example, Moon and Mars [ ]. TEA CO 2 laser-based LIBS has been previously applied for the analysis of basalt under simulated Martian atmospheric conditions (CO 2 gas; pressure: mbar) [ ]. It was shown that this LIBS system provides good analytical capabilities for geological studies under low-pressure CO 2 gas. Although LIBS under di erent atmospheric conditions o ers the possibility to improve the resolution, signal intensity, and signal-tonoise ratio or enable speci c applications, it also introduces a degree of complexity in the experimental setup. us, the present study was undertaken to examine the possibility of using TEA CO 2 laser-based LIBS for the analysis of geologic samples in air at atmospheric pressure. e same sample was also analyzed using a commercial LIBS system based on Nd:YAG laser and time-gated detection.
Currently, the most common LIBS system used for the analysis of geological samples comprises Nd:YAG laser and time-gated detection [ -]. Compared to it, the main advantage of TEA CO 2 laser-based LIBS is simpli cation of the instrumentation. e characteristics of TEA CO 2 laser (wavelength and pulse duration) make this laser especially suitable for time-integrated measurements, which precludes the use of delay generators and expensive gated detectors required for time-resolved measurements. Di erences in laser wavelength and pulse duration for the two laser sources ( . m, ns for TEA CO 2 compared to . nm, ns for Nd:YAG) inevitably in uence the ablation process, the plasma parameters, and plasma expansion dynamics. Excitation wavelength has a strong in uence on the lasertarget and laser-plasma interactions. Mass ablation rate per pulse ( ) is higher for shorter wavelengths ( ∼ −4/9 ) [ ]. However, when operating with infrared wavelengths, dominant mechanism for laser-plasma interaction is Inverse Bremsstrahlung (IB) absorption. e rate of IB absorption ( IB) increases as the laser wavelength increases ( IB ∼ 3 ) [ ]. us, for Nd:YAG plasma, a higher ablation e ciency may be expected, while laser-plasma interaction is expected to be much stronger in the case of CO 2 laser. E cient IB absorption reheats the plasma, causing an additional plasma excitation and expansion. In LIBS, only a fraction of the analyte mass is excited and is capable of producing detectable optical emission. Higher excitation e ciency in CO 2 plasma may compensate the lower ablation e ciency. In addition, the dimensions and lifetime of plasma increase with pulse duration. e plasma takes longer to decay and, hence, the emission lasts longer. With the long-pulse irradiation, the excitation of the plasma is rather gradual, and, hence, it is suitable to obtain highly excited plasma. e analytical capabilities of the two LIBS systems were compared in terms of detectability and limits of detection (LODs). A simpli ed single-sample method was applied for estimation of LOD [ ]. Elemental composition of the basalt sample was determined by ICP analysis. Also, plasma parameters, temperature, and electron number density for plasma created by CO 2 and Nd:YAG laser pulses were evaluated and compared.
e rock sample used in the present work was the same as the one used in our previous study [ ], that is, a Tertiary basaltic rock from the Balkan Peninsula, classi ed according to the TAS scheme as basaltic trachyandesite [ ]. A sample was prepared by cutting the original natural basaltic rock to a piece about mm long and ∼ mm thick. e sample was used as is, without any preparation. Elemental composition of the sample was determined by ICP analysis. Samples for ICP analysis were prepared by alkaline fusion procedure [ ]. For the major elements, mg of powdered material was fused with LiBO ux and then dissolved in dilute HNO 3 . For the trace elements, HF/HClO 4 /HNO 3 dissolution was undertaken using mg of sample. Five analyses were completed to determine measurement precision. Elemental composition of the sample is given in Table . . . TEA CO 2 LIBS System. e experimental setup is described in prior publication [ , ] and is brie y summarized here. A commercial TEA CO 2 laser system, developed at Vinča Institute, was used as the energy source for plasma generation on the basalt samples [ ]. Pulsed TEA CO 2 laser emits radiation at . m, with an output peak power typically in the order of megawatts. e laser-optical pulse has a gain-switched spike, followed by a slowly decaying tail. e full width at half maximum (FWHM) of the spike Journal of Chemistry T : Elemental composition of basaltic rock obtained using ICP-OES analysis [ ].
The relative expanded standard uncertainty ranged from -%, in ppm concentration range, to -% for the highest measured concentrations. is about ns, and the tail is about s. About % of the total irradiated laser energy is consisted in the initial spike. Typical output pulse energy was mJ. e sample surface was irradiated by laser light focused using a ZnSe lens (focal length: mm). e angle of incidence of the laser beam with respect to the surface was ∘ . e sample was mounted on a rotating sample holder. Single-shot craters well separated in space were achieved by rotation of the sample, while repeated ablation of the same circular path was avoided by lateral movement of the sample. e optical emission from the plasma was viewed in the direction parallel to the target surface (Figure (a)). A timeintegrated space-resolved laser-induced plasma spectroscopy (TISR-LIPS) was applied [ , -]. Spatial resolution was achieved by changing the position of the plasma along the direction of the incoming laser beam. e focusing lens and the target holder were placed on positioning stages (threeaxis translation stage; travel: mm; resolution: m) to maintain focus during experiments. For spatially resolved measurements, it was essential to nd plasma region, where spectral continuum background intensity is low, while the discrete line signal intensity is high. e optimization was performed by monitoring the signal-to-background ratio (SBR) of several elements (Mg: . nm, Ca: . nm, Mn: . nm, and Fe: . nm lines) as a function of the viewing position. Di erent spatial regions of the plasma were sequentially examined by moving the target holder in steps of m. Highest SBR values were obtained at distances mm away from the sample surface. us, in all reported experiments, the observing position was set to mm from the target surface. e horizontal part of the plume was projected by an achromatic lens on the entrance slit of the monochromator (entrance slit width: m; height: mm; magni cation: : ). e monochromator was equipped with di raction grating with lines/mm (dispersion: . nm/mm; blaze at nm in the rst order). For the time-integrated measurements, Apogee Alta F CCD camera was used. e CCD consists of × pixels, each of × microns. e total active area is . × . mm. e laser was operated at . Hz repetition rate, while the shutter of the nongated CCD detector was opened for s. Measurement time was much longer than plasma lifetime; however, the main contribution to the time-integrated spectrum comes from a very limited temporal window. Averaging of many acquired spectra was a procedure used to attenuate the variation of spectral emission, that is, to compensate pulse-to-pulse variations. Each recorded LIBS spectrum is an average spectrum, obtained by accumulation of consecutive spectra from di erent locations on the sample. is procedure was as a rule repeated in triplicate, and the resulting spectra were averaged. us, all reported spectra correspond to an average of laser pulses. A relative standard deviation of measured LIBS signal intensity was -%.
. . Nd:YAG LIBS System. Basalt sample was also analyzed by using commercial LIBS system based on pulsed Nd:YAG laser and time-gated detection (Figure (b)).
is was a LIBSCAN-system which was purchased from Applied Photonics Ltd. (UK). As laser source, it is equipped with a Qswitched Nd:YAG laser emitting at nm (model Quantel Big Sky CFR Ultra GRM), pulse duration of ns, frequency of Hz, and energy up to mJ. e system possesses Journal of Chemistry T : Spectroscopic constants of the Fe II lines (Figure ) used in Boltzmann plot temperature determination.
3.50 + 08 spectrometer channels covering -nm spectral range. e acquisition is made on an integration time of . ms (the smallest) with . s delay time from the laser pulse. e signals were averaged on laser pulses. e statistical variability in the LIBS signal was approximately %.
A diameter of a focused laser spot was ∼ mm for TEA CO 2 laser and ∼ . mm for Nd:YAG laser, that is, in both cases well above the grain sizes of the rock sample (< mm) [ ].
us, the in uence of grain size on accuracy and precision of the LIBS analyses was not expected to be large [ ]. e main contribution to uncertainty comes from uctuations due to laser energy and high degree of surface roughness.

Results and Discussion
LIBS spectra of basalt were recorded in air at atmospheric pressure. All reported spectra are spectra averaged over many laser shots, and the measurements were taken at a number of locations on the sample surface (a fresh surface was exposed to each incident laser pulse) to avoid the problem with sample heterogeneity. Spectra obtained on the basaltic rock sample with a single TEA CO 2 laser pulse using a nongated detection, covering the spectral region from to nm, are shown in Figure . e spectra were taken mm from the target surface. For comparison, spectra obtained using LIBS system equipped with pulsed Nd:YAG laser and time-gated detection are also shown. Most of the elements present in the sample were identi ed (Table ), including those present in concentrations around or below ppm, like Co, Cr, Ni, and V. As it may be seen from Figure , there is no di erence with regard to element detectability for the two LIBS systems used, despite the fact that the applied laser intensities di er by one order of magnitude ( MW/cm 2 and / MW/cm 2 for TEA CO 2 laser and Nd:YAG laser, resp.). us, it may be concluded that excitation conditions in CO 2 plasma are suitable for detection of all elements usually present in geological samples. e dissociation, ionization, and excitation processes taking place in plasma are related to characteristic parameters of plasma such as temperatures ( ) and electron number densities ( ). us, it was of interest to determine these parameters for plasma created by CO 2 and Nd:YAG laser pulses. e excitation temperature was evaluated by Boltzmann's plot method [ ] using the equation where is the wavelength of the emitted light, is the integrated line intensity of the transition involving an upper level ( ) and a lower level ( ), is statistical weight of upper level, is the transition probability, ℎ is Planck's constant, is the velocity of light in vacuum, is the total number density, ( ) is the partition function, is excitation energy, and is the Boltzmann constant. For the determination of excitation temperature of the plasma, eleven iron ionic (Fe II) lines were selected as shown in Figure . Considering a relatively narrow spectral interval used, the chromatic sensitivity was not taken into account. e line identi cations and di erent spectroscopic parameters such as wavelength, transition strength ( ), energy , and accuracy of transition probability (B+ ≤ 7%; B ≤ 10%; C+ ≤ 18%) are listed in Table based on reference data [ ]. In case of TEA CO 2 LIBS, the whole set of spectral lines listed in Table was used to obtain Boltzmann's plot, whereas in case of Nd:YAG LIBS, lines numbers and were excluded as they were not resolved. e estimated temperature of CO 2 laser-induced plasma was K, while that of Nd:YAG plasma was K and K for mJ and mJ pulses, respectively (Figure ). For both LIBS systems, under applied experimental conditions, high temperature plasma was created. Higher temperatures were obtained in Nd:YAG laser-induced plasma. However, it should be noted that estimated plasma temperatures are not directly comparable, because time-integrated measurements for CO 2 laser plasma provide average values for the spatially selected plasma zone, while for the Nd:YAG LIBS the values correspond to a speci c time window. e electron number density was determined from the measured Stark width of Al I . nm line (Figure ). : LIBS spectra obtained on the basalt sample with a single TEA CO 2 laser pulse using a nongated detection and with Nd:YAG laser and time-gated detection (delay: . s; integration time gate: . ms). In both cases, the plasma was induced in air at atmospheric pressure.
F : Fe II spectral lines (listed in Table )  these values are not directly comparable, as already discussed for the temperature evaluation. Nevertheless, they provide indication of the excitation conditions in plasma. For both laser systems, spectral data were processed by calculating the integrated peak area of the emission line, the baseline continuum emission intensity, and the root mean square (RMS) noise of the continuum radiation intensity in regions adjacent to the emission line.
e signal-to-noise ratio (SNR) was calculated as the integrated peak area A, divided by the width of the peak area , times the absolute value of the RMS noise: SNR = /( × rms) [ ]. Signal-tonoise (SNR) ratio for the elements under study, obtained for TEA CO 2 and Nd:YAG laser-induced basalt plasma, is shown in Table . Si A simpli ed single-sample method was applied for estimation of the limit of detection (LOD) [ ]. LOD was calculated using the formula LOD = (3 × )/SNR, where is a known analyte concentration obtained by ICP measurements (Table ). Comparison of estimated limits of detection obtained for the two laser systems is shown in Figure . As it can be seen, comparable or even better LODs were obtained for TEA CO 2 LIBS system. Also, there is no obvious correlation between the energy of the Nd:YAG laser pulse and the estimated LODs.

Conclusions
Laser-induced breakdown spectroscopy is based on TEA CO 2 laser and has been applied to elemental analysis of basalt sample. e analysis was done in air at atmospheric pressure without any sample treatment. Results of qualitative analysis were compared to those obtained using a commercial LIBS system based on pulsed Nd:YAG laser. It may be concluded that, with regard to detectability and limits of detection, less complex and cost-e ective TEA CO 2 LIBS system has comparable characteristics to the system based on Nd:YAG laser and time-gated detection.

Conflicts of Interest
e authors declare that there are no con icts of interest regarding the publication of this paper.  [ ] A. Ruzicka, G. A. Snyder, and L. A. Taylor, "Comparative geochemistry of basalts from the moon, earth, HED asteroid, and mars: Implications for the origin of the moon, " Geochimica et Cosmochimica Acta, vol. , no. , pp.
[ ] D. Campos, S. S. Harilal, and A. Hassanein, " e e ect of laser wavelength on emission and particle dynamics of Sn plasma, " Journal of Applied Physics, vol. , no. , p. , .