Self-Calibrated Laser-Induced Breakdown Spectroscopy for the Quantitative Elemental Analysis of Suspended Volcanic Ash

Abstract

Real-time analysis of fine ash in volcanic plumes, which represent magma fragments expelled from the crater during explosive eruptions, is a valuable tool for volcano monitoring and hazard assessment. To obtain the chemical characterization of the juvenile pyroclastic material emitted in volcanic plumes, many analytical techniques can be used. Among them, laser-induced breakdown spectroscopy (LIBS) is the one that can most easily be adapted to advanced applications in extreme environments. In this paper, LIBS experiments based on self-calibrated approaches are used to determine the elemental composition of suspended volcanic ash. To simulate the conditions of dispersed volcanic ash in the atmosphere, different sizes of volcanic ash samples are suspended in the air by laser-induced shockwaves in a dedicated chamber, and a parametric study is carried out to establish the optimal experimental conditions for recording usable plasma emission spectra for each ash size. The quantitative analysis is performed using a self-calibrated analytical method, including calibration-free LIBS, which is based on the calculation of the spectral radiance of a uniform plasma in local thermodynamic equilibrium. The method accounts intrinsically for self-absorption since it modifies the intensity of spectral lines and thus leads to an underestimation of the elemental fraction. An intensity calibration of the spectra based on the measurements of Fe lines intensities was also used in this work to deduce the apparatus response from the spectrum itself and avoid the use of standard calibration lamps. Results demonstrate the potential of real-time measurements of elemental fractions in volcanic ash with good agreement with the literature composition.

Keywords

Laser-induced breakdown spectroscopy LIBS calibration self-absorption atomic emission

Introduction

Laser-induced breakdown spectroscopy (LIBS) is the optical emission spectroscopy of the plasma produced by laser–matter interaction. The induction of the plasma is obtained when the laser irradiance is beyond the characteristic breakdown threshold of the irradiated sample. Atoms, ions, and molecules in the plasma phase are excited by electron collisions and emit photons, thus generating spectra characteristic of the atomic and molecular structure of the emitters. The plasma radiation is collected through an optical fiber and analyzed by a spectrometer allowing the elemental analysis.¹

Among the growing number of applications where LIBS is employed, real-time measurement in hostile environments is still the most promising field of application of this technique. This is mainly due to the peculiar characteristics of LIBS measurements as it can provide some advantages that cannot be obtained with any other analytical tool. These peculiarities are related to the fast response, stand-off measurements, absence of sample preparation, multi-elemental detection, robust and flexible set-up, and the possibility of self-calibrated approaches.²

The robustness of self-calibrated LIBS has been particularly illustrated in space exploration. It has been shown that for asteroid investigations, the instrumentation could be calibrated on the emission spectrum itself, allowing a decrease in the potential payload of the LIBS instrumentation and increasing the accuracy of the elemental analysis.^3,4 Besides, it has been demonstrated that this technique can function remotely in the most hostile environments as is the case of the LIBS instrument implemented in rovers operating on Mars surface, thus opening the way to compact on-flight instruments for other applications.⁵ In this paper, the use of a LIBS methodology for the analysis of fine volcanic ash suspended in air is discussed. This case study can represent a further strong boost for LIBS elemental analysis in hostile environments.

In the past two decades, geophysical volcano studies have made significant progress in understanding volcanic processes, eruption dynamics, and forecasting, but most research efforts focused on predicting eruptions, managing emergencies, evaluating the impact on local economies, particularly agriculture-dependent livelihoods near volcanoes, and characterizing gas emissions, e.g., carbon dioxide (CO₂), hydrochloric acid (HCl), hydrogen fluoride (HF), hydrogen sulfide (H₂S), and sulfur dioxide (SO₂).^6–9 Thus, a relatively small amount of research has been carried out for the chemical characterization of fine solid particles (ash) emitted in volcanic plumes, which is crucial since they represent magma fragments expelled from the crater during explosive volcanic eruptions.

As a matter of fact, the elemental composition of fine volcanic ash offers insight into magma properties, thus enabling potential correlation between ash and source volcanoes. Moreover, in the case of persistent and long-lasting eruptions, the elemental composition of the volcanic ash and its fluctuations that can occur in time is able to give information about the evolutionary stage of the magmatic reservoir and anticipate rapid changes in eruptive conditions, thus helping to plan early warning procedures.¹⁰ Since volcanic plumes are rich in fine ash, a fast way to know their composition is the real-time analysis using a compact instrument based on LIBS.

From an experimental point of view, measuring on-flight ash in the scenario of a real volcanic plume poses several strict requirements to any analytical tool:

The sample cannot be pretreated.

Measurement must be fast to have a real-time response.

The number of ash particles is unknown and continuously changing because of the turbulence in the plume, thus the elemental analysis should be independent of the amount of sampled ash.

The instrument itself may be subjected to changes in the instrumental response due to possible deterioration of the apparatus while approaching the volcano, and the optical response can also change because of the hot environment or to the deposition of particles and condensation of gases on the detection optics.

In this paper, it will be demonstrated that self-calibrated LIBS can overcome all the above-mentioned issues, retrieving all the required information, from instrumental calibration to elemental analysis, just from the direct measurement of the sample spectra. This provides a significant advantage in terms of flexibility of the set-up, instrumental payload, and robustness of the LIBS sensor. As an example, if the ChemCam or the SuperCam instruments used by NASA for exploring Mars, which represent the most advanced LIBS tools for elemental analysis in a hostile environment, is compared to the LIBS approach of the present paper, it is possible to outline the importance and the modernity of the proposed LIBS methodology. As an example, the NASA LIBS sensor is based on a fixed instrumental calibration and on standard samples included in the instrument. The measurement needs continuous calibration and the accuracy of the analysis, when feasible, mainly depends on the stability of the set-up and on the similarity of the onboard standards to the real samples being analyzed on the planet surface. With our proposed approach, no standards are required, and each measurement is individually calibrated, which can speed up the analysis, at the same time decreasing the potential payload. This enables LIBS to avoid any matrix effects and adapt the response to the actual instrumental conditions.

For this study, the LIBS experiments were carried out in a dedicated chamber where different sizes of volcanic ash samples were suspended in air by laser-induced shockwaves, in order to mimic the same conditions that would occur in a volcanic plume. The laser-induced plasma (LIP) spectrum was recorded and analyzed, and the results were compared with the average composition values found in the literature. The elemental analysis was performed using the calibration-free (CF) approach and, in order to improve the analytical results, within the Boltzmann plot-based technique for temperature determination, self-absorption was taken into account in the calculation of the spectral lines intensities.¹¹ In addition, the radiometric calibration of the spectra was carried out using the measurements of Fe lines intensities in the ash spectrum to deduce the apparatus response from the spectrum itself and avoid the use of radiometric standards.

The experiments were conducted on real ash samples belonging to pyroclastic deposits of two different volcano eruptions: (i) The 2004 eruption of Grimsvötn Volcano, Iceland (GSV 3-1)¹² and (ii) the Pomici di Mercato eruption of Somma Vesuvius volcano (MC 1-0).¹³

The results show that volcanic ash can be effectively suspended in air with appropriate experimental conditions, and their plasma emission spectra are good examples for demonstrating that LIBS is an adequate elemental technique for real-time analysis of magma-feeding components and that the present approach can be extended to other LIBS applications in a hostile environment.

Materials and Methods

Ash Samples

The solid fine particles emitted during explosive eruptions are multicomponent. They can contain juvenile, crystal, and lithic fragments, and each component is characterized by its specific chemical composition. The relative proportions of these components can change with grain size, inducing a variation of the geochemical composition as a function of grain-size fraction.^10,14 Therefore, coarser ash could have as main components “lithic fragments", while for finest ash, the major components could be “juvenile fragments". For this reason, the compositions of the analyzed ash are not supposed to be the same. From the literature, it is deduced that ash smaller than 250 μm is mainly composed of juvenile fragments that refer to freshly emitted volcanic material of magma origin that has not undergone changes during transportation in air or sedimentation.¹⁰

During this research, we selected two ash samples finer than 125 μm since they are the most representative of the magma responsible for feeding the eruption. For comparison purposes, we also selected ash with a fraction coarser than 250 μm from the same deposits.

From the 2004 eruption of Grimsvötn volcano (GSV), three ash samples of different sizes have been analyzed: GSV (S1) < 125 μm, GSV (S2) > 125 μm and GSV (S3) > 250 μm. From the Pomici di Mercato eruption of Somma Vesuvius volcano (MC), two ash samples of two sizes have been analyzed: MC (S1) < 125 μm and MC (S2) > 250 μm. Note here that ash > 125 μm has a grain-size fraction between 125 and 250 μm, while ash > 250 μm has a grain-size fraction between 250 and 500 μm.

In Figure 1, pictures of two samples of GSV ash taken using an optical microscope show the grain-size difference between the ash analyzed during the experiments. Also, the shape and texture are different, mainly because they contain different components (juvenile, crystal, and lithic fragments) that are distributed in varying quantities. The elemental analysis will reveal in a later section the different chemical composition of these samples.

Table I shows the elemental composition of GSV and MC ash samples taken from the literature. Unlike in the present LIBS experiment, the ash analyses were performed using an energy dispersive spectrometer (EDS) and X-ray fluorescence (XRF) on the GVS 3-1 sample and MC 1-0, respectively.^12,13

Table I.

Chemical composition of the selected samples (wt%). Chemical data are taken from previous works which are obtained from EDS analysis on juvenile particles for GVS 3-1 samples,¹² and from powdered juvenile fragments XRF analysis for MC 1-0 samples.¹³

Wt%	SiO₂	TiO₂	Al₂O₃	FeO–Fe₂O₃	MnO	MgO	CaO	Na₂O	K₂O	P₂O₅
MC 1-0	60.96	0.13	19.51	2.14	0.21	0	1.71	8.33	7.02	0
GSV 3-1	48.25	2.60	14.00	12.04	0.12	7.8	8.94	5.94	0.32	< 0.01

Calibration-Free Laser-Induced Breakdown Spectroscopy (CF-LIBS) Analytical Methodology

Calibration-free laser-induced breakdown spectroscopy (CF-LIBS) was employed to eliminate the need for matrix-matched standards as well as the dependence of the analysis from the number of sampled particles, to get a simultaneous multi-elemental analysis of volcanic ash starting from the plasma emission spectrum.

Calibration-free laser-induced breakdown spectroscopy (CF-LIBS) analysis was first introduced by Ciucci et al.¹⁵ in 1999 for metallic alloys. It is based on the local thermodynamic equilibrium (LTE) of the LIP and applied to multi-elemental detection. The LTE assumption means that electron collisions prevail on radiative processes and that particles follow the Maxwell–Boltzmann distribution at each spatial and temporal coordinate. In this condition, the following relations are applied to obtain the elemental analysis:

Boltzmann's law predicts the statistical distribution of energy level populations of species in the plasma giving the population density of an excited electronic state N_u by the relation

N_{u} = N_{0} \frac{g_{u}}{Z (T)} e^{- E_{u} / k T}

(1)

where N₀ is the total population density of the emitter, g_u is the upper-level degeneration, Z(T) is the partition function, E_u is the energy of the upper level, k is the Boltzmann constant, and T is the excitation temperature.

Saha's law describes the ionization equilibrium by providing the ratio between the densities of two successive ionization states. It is given by:

\frac{N_{0}^{A +}}{N_{0}^{A}} = \frac{1}{N_{e}} {(\frac{2 π m k T}{h^{2}})}^{3 / 2} \frac{Z_{e} Z_{A +} (T)}{Z_{A} (T)} e^{- E_{i} / k T}

(2)

where

N_{0}^{A}

and

N_{0}^{A +}

are the atomic and ionic number density, respectively, N_e is the electron density, E_i is the ionization potential of the species with charge i, Z_e is a constant equal to 2, Z_A+ is the partition function of the ionic species A⁺ at temperature T, Z_A is the partition function of A at temperature T, and h is the Plank constant.

Closure equation. From Eqs. 1 and 2, considering that the emission line intensity is proportional to N_u it is possible to retrieve the relative total number density of the species A, $N_{0, A}^{T O T} = N_{0}^{A} + N_{0}^{A +}$ . If all the n species of the plasma can be detected in the spectra, then with the following closure equation the plasma composition is determined:

\sum_{n} C N_{0, n}^{T O T} A M_{n} = 100

(3)

where AM_n is the atomic mass of the n species and C is a normalization constant related to experimental conditions and setup.

Using these hypotheses, Ciucci et al.¹⁵ implemented an algorithm that allows plotting a multi-elemental Boltzmann diagram from the intensities of the lines and deduces the elemental composition. A further improvement has been accomplished by including self-absorption and estimating the Boltzmann distribution through a direct simulation of the spectra. This approach is based on minimizing the difference between the experimental spectra and the ones obtained assuming the Boltzmann distribution of the emitting species at the temperature obtained by the fitting of the experimental spectrum. The detailed description of the method can be found in the literature,^16,17 while the validation of the technique and physical model is described in many previous publications.^4,18–21 Details about all calculations employed in this work are explained in Hermann's patent¹¹ and briefly reported in the Supplemental Material.

One of the most crucial disadvantages of CF-LIBS is the need for as short as possible a detection time, i.e., gate width, which is required to ensure the validity of the LTE condition. The use of a small gate width can reduce the signal-to-noise ratio (S/N) which is the main factor containing the analytical information. As already demonstrated in several applications,^22,23 this issue can be correctly compensated for if we work with the concept of apparent temperature instead of excitation temperature (the latter is generally assumed to be the electronic temperature).²⁴ Briefly, if during the gate width, the plasma is in LTE condition at different temperatures and the distribution of the logarithm of the emission coefficient, corrected by the spectroscopic parameters (see Supplemental Material), is still linear when plotted as a function of the energy of the upper level, then we can approximate the following:

\begin{aligned} N_{0, 1}^{A} & (- \frac{E_{u}}{k T_{1}}) \frac{1}{Z (T_{1})} + N_{0, 2}^{A} \begin{matrix} (- \frac{E_{u}}{k T_{2}}) \end{matrix} \frac{1}{Z (T_{2})} \\ + N_{0, 3}^{A} (- \frac{E_{u}}{k T_{3}}) \frac{1}{Z (T_{3})} \dots = N_{0, T O T}^{A} R (- \frac{E_{u}}{k T_{a p p}}) \frac{1}{Z (T_{a p p})} \end{aligned}

(4)

where the subscript in the first member indicates the microstate in LTE at a given temperature and R is a residual factor due to the approximation. According to this approximation, the temperature obtained during the whole gate width, namely T_app, is not the average temperature in the considered time and it does not have any physical meaning, but it may be interpreted as the temperature of a virtual Boltzmann distribution followed by the spectral intensities, and therefore it allows applying CF-LIBS at the same temperature. It is important to underline that the deviation from the average Boltzmann distribution due to this approximation proportionally affects the concentration of all the elements and consequently, it will be mostly mitigated in the closure equation (Eq. 3). This strategy allows using longer gate widths and improving the S/N. As discussed in the work by Bousquet et al.,²⁴ a low S/N can affect the experimental Boltzmann distribution to a larger extent than the use of the above-mentioned approximation of the apparent temperature.

Experimental

Laser-induced breakdown spectroscopy (LIBS) experiments were carried out in a stainless-steel chamber which was built to simulate a plume of flying ash (Figure 2).¹⁰ The chamber was equipped with fused silica optical windows to admit the laser beam and to capture the light emitted by the plasma. All trials were performed in air as a surrounding gas under ambient pressure conditions. The laser source was a neodymium-doped yttrium aluminum garnet (Nd:YAG) laser, specifically the Quantel Q-smart 850 model, operating at a wavelength of 1064 nm with a pulse duration of 6 ns. The laser can operate at a repetition rate of up to 10 Hz and deliver energies of up to 615 mJ per pulse. During the experiments, we used laser energies ranging from 100 to 300 mJ. The laser beam was focused using a plano-convex lens with a 5 cm focal length, positioned along one side of the chamber, as illustrated in Figure 2. Plasma emission light, passing through a fused silica window positioned on a second arm of the chamber, perpendicular to the main one, was collected using a biconvex lens with a focal length of 15 cm and focused on optical fiber connected with an echelle spectrometer (Aryelle 200, LTB). The spectrometer was coupled with an intensified charge-coupled device (ICCD) detection system (New iStar, Andor). The ICCD camera was synchronized with the laser source by a pulse generator (Stanford DG 535). The spectra were wavelength- and intensity-calibrated, respectively, with a low-pressure mercury lamp and a standard radiometric light source.

The ash was positioned on a holder that was placed on the bottom of the chamber and the laser pulses were focused on different distances above the samples. Thanks to the shockwaves generated by the laser pulses inside the chamber, the ash was dispersed in the chamber atmosphere and kept suspended in the air. The LIBS experiments were then performed by generating the plasma in air with and without suspended ash.

A parametric study allowed us to deduce the optimal conditions for quantitative analysis of volcanic ash. For the small ash samples, largely composed of juvenile ash, GSV 3-1 (S1) and MC 1-0 (S1) whose sizes are <125 μm, the optimal laser energy and distance between the laser beam and the sample were 150 mJ and 1.5 cm, respectively, while for the bigger ash samples GSV 3-1 (S2), GSV 3-1 (S3), and MC 1-0 (S2), they were 300 mJ and 0.5 cm, respectively.

In this work, two measurement modes were employed, i.e., the acquisition of the temporal evolution of the plasma produced in air and the one produced on ash suspended in air, with the goal of understanding the evolution of its parameters in both cases and the acquisition of plasma emission spectra for the CF-LIBS analysis. For the temporal evolution of plasma emission in the air, a gate width of 100 ns was used, while for the temporal evolution of the plasma spectra in the air with suspended ash, a gate of 250 ns was used, with a progressive delay of 250 ns from the laser pulse for all the plasma persistence.

For the integrated LIBS measurements of air with suspended ash, a delay with respect to the laser pulse of 0.5 μs and 1 μs with a gate width of 5 μs was employed for improving the signal-to-noise ratio (S/N) for the elemental analysis. Each acquired spectrum was an average of 20 spectra measured in the same experiment. Each integrated measurement was repeated three times and subsequently averaged for the CF-LIBS analysis. To analyze the emission intensities of specific peaks at specific wavelengths, all the spectra were adjusted with the corresponding radiometric calibration curve and the transmission curve of the fused silica window. After each experiment, the chamber underwent a complete cleaning process to prevent cross-contamination.

Results and Discussion

Intensity Calibration Using LIP Spectrum

As usual, an intensity calibration procedure is needed to account for the different instrumental efficiency in the different spectral regions, which requires standard radiometric lamps. Nonetheless, calibration lamps’ spectra have to be recorded using the same optical path and in the same experimental conditions as the ones used for plasma spectra acquisitions. This is difficult or unattainable in the case of in situ recordings of LIP spectra of volcanic ash released during volcanic emissions.

The appropriate solution would be to find a way to correct the apparatus response using a methodology that completely avoids the use of radiometric sources.

A calibration procedure that exploits the emission spectrum of the ash plasma itself instead of calibration lamps, and based on the measurements of Fe lines intensities was used in this work to deduce the apparatus response. Iron was chosen because of its high number of emission lines in geological sample spectra. From a number of spectral windows, in which the emission lines used for the elemental analysis were chosen, multiple Fe(I) emission lines were selected (Table II). Then, the correction procedure started with measuring the electron density from an appropriate line and the excitation temperature using a Boltzmann plot of Fe(I) lines selected from one narrow spectral window of the recorded spectrum, in order to minimize the influence of the apparatus response (here T is obtained from spectral range 7 of Table II).^3,25 After that, the intensities of each other Fe emission line were calculated. At the same time, the intensities of these same Fe lines were measured from the recorded spectra. By calculating the ratio between measured intensities I_meas and calculated intensities I_cal of Fe(I) emission lines, we deduced a calibration factor for each spectral window. For accuracy, we used N spectral lines for each spectral window to obtain an average value for the calibration factor calculated by the relation³

G = \frac{\sum_{N} (I_{N, m e a s} / I_{N, c a l})}{N}

(5)

Table II.

Chosen spectral ranges with selected Fe(I) emission lines used for the calibration intensity procedure.

Spectral ranges (nm)	Wavelength of Fe emission lines (nm) for intensity calibration
(1) 270–285	271.902; 272.357; 274.24; 275.014; 276.202; 276.752; 276.929; 278.793; 278.81; 280.698; 281.328; 382.555; 283.243; 284.397
(2) 290–305	293.69; 294.787; 295.394; 296.689; 297.313; 297.323; 298.144; 299.442; 299.951; 300.094; 302.584; 303.738; 304.76
(3) 305–320	305.744; 305.908; 306.724; 309.989; 317.544; 319.692
(4) 322–337	322.206; 322.578; 322.779; 328.675; 329.202; 330.597; 330.634; 336.954
(5) 340–355	340.745; 341.313; 341.85; 342.711; 344.06; 344.098; 346.586; 347.545; 347.67; 349.057; 352.604; 352.616; 353.655; 354.207; 355.851
(6) 356–371	357.009; 357.388; 358.119; 358.698; 360.32; 360.545; 362.146; 364.784; 368.599; 368.745
(7) 370–385	370.108; 370.924; 371.993; 372.256; 372.761; 373.486; 374.556; 374.826; 374.948; 375.823; 376.378; 376.553; 376.719; 378.788; 379.5; 379.954; 381.296; 381.584; 382.042; 382.444; 382.588; 382.782; 384.043; 384.104; 385.637; 385.991
(8) 387–402	387.857; 388.628; 388.851; 389.565; 390.294; 392.792; 393.029; 398.177; 398.395; 400.524
(9) 398–413	402.186; 404.581; 406.359; 407.173; 411.854
(10) 420–435	420.202; 421.936; 423.593; 425.011; 425.078; 426.047; 427.176; 429.412; 430.79; 431.508; 432.576
(11) 438–453	438.354; 440.475; 441.512; 445.911; 447.601; 449.456; 452.861
(12) 485–500	489.149; 492.05; 493.4; 495.759
(13) 518–533	519.145; 519.234; 522.686; 522.718; 523.294; 532.417; 532.803
(14) 538–553	538.336; 539.712; 541.519; 543.452; 544.691; 545.56; 550.677
(15) 555–570	557.284; 557.608; 558.675; 561.564; 563.826; 566.251
(16) 570–595	570.599; 575.312; 576.299; 585.221; 585.608; 586.718; 587.948; 590.567; 591.624; 593.018; 593.465
(17) 620–635	619.155; 623.072; 623.839; 625.425; 625.636; 630.15; 631.53; 631.801; 633.682
(18) 640–655	640; 640.801; 641.164; 643.084; 656.921
(19) 710–725	716.444; 718.731; 720.738
(20) 730–745	729.283; 729.304; 730.793; 731.107; 738.93; 741.866; 749.164; 749.506; 751.102

When the factor G is obtained for each spectral window used for the elemental analysis, all intensities of all chosen spectral lines for CF-LIBS were calibrated with the appropriate factor.

Figure 3a shows all spectral ranges chosen for the selection of iron lines. It also reports the factor G_app obtained from the convolution of the instrumental response of the Echelle spectrometer (measured with the standard lamps) and the transmittance of the chamber window and the factor G obtained from the ratio between measured and calculated intensities of Fe lines. These show that the two calibration factors have a similar trend with wavelength, thus confirming that the intensity ratio procedure can efficiently replace the use of radiometric sources to account for the response of all experimental components. Figure 3b shows a zoom of the spectral range 7 that contains Fe(I) lines used to measure the excitation temperature with a Boltzmann plot.

To demonstrate the validity of this self-calibrated method, a comparison between the CF-LIBS procedure performed with the intensities of each element corrected with the G factor of Eq. 5 and the one performed with the intensities of each element corrected with the apparatus response as obtained with standard lamps is shown in the next paragraph.

CF-LIBS Of Suspended Ash Samples

An important step of these experiments was to suspend the volcanic ash in air inside the chamber, so as to mimic their behavior in volcanic plumes. This step was described in detail in De Giacomo et al.¹⁰ Briefly, the first laser pulse inside the stainless-steel chamber generated a shockwave in the medium above the ash samples that dispersed the ash. Since each laser pulse is followed by a shockwave, the multiple interactions between shockwaves generated by the subsequent laser shots allowed the suspending of an ash plume with unknown and random particle concentration. Thus, from the second laser pulse, the ash inside the chamber were able to enter the air plasma undergoing particle vaporization. In this case, from N laser shots, N−1 of them could produce plasma emission spectra with emission lines of air and elements composing the volcanic ash.

To check the shockwave ability to suspend the ash and the quality of the acquired plasma emission spectrum, different volcanic ash were studied during the experiments. Their plasma spectra were recorded, and the presence of emission lines of elements contained in the ash was verified. The samples selected for the experiments were basically composed of juvenile constituents, so their chemical composition could be easily attributed to the same composition of fresh erupting magma, but they had different sizes. Suspension and vaporization of solid particles were easier for the finest particles. Nonetheless, for what concerns the studied ash, the parametric study showed that optimal experimental conditions could be found for each ash sample size.

The comparison between plasma emission spectra induced in the air and the ones induced on the suspended ash was investigated to understand the processes occurring during particle ablation. During experiments, when the laser irradiance exceeds the breakdown threshold of air, a plasma is generated, resulting from the ionization and atomization of air components. In our case, i.e., in the presence of suspended ash in the air, it has been demonstrated in previous publications that the primary mechanism for vaporizing particles is the interaction between plasma and particles in the air, rather than directly inducing the plasma on particles, although particles can decrease the breakdown threshold of the atmospheric gas.^26,27 Observation suggests that, inside the chamber, ash particles were atomized and ionized in the air because of the high temperature and high electron number density of the air LIP. Thermalization of ash species and air plasma occurs fast as a consequence of the high number of fast electrons in the air plasma. Then, it was possible to obtain plasma emission spectra with lines of elements composing ash samples. As mentioned above, the presence of particulate locally lowers the breakdown threshold of the air by inducing the spark in different points around the focal point. Due to this effect, the spectrum obtained by averaging 20 shots is not merely the result of an average of 20 identical spectra, but rather it ensues from averaging different portions of the produced plasma and can sometimes also include spectra of air without ash particles if the latter miss the probing plasma volume. For this reason, a classical analysis based on standard calibration curves would be unfeasible. On the contrary, in the proposed approach, the elemental analysis is directly based on the detected emission signal, thus making it possible to determine the average ash composition in the probed plasma volume during the 20 shots. Moreover, the proposed methodology is not dependent on the number of sampled particles, because the composition is directly recovered from the plasma emission without any restriction to calibration experiments that would require the same number of sampled particles or the use of some normalizing factor to consider the dependence of the net intensity on the amount of sample used for the calibration.

Figure 4 shows the comparison of the LIBS spectrum in pure air and with ash suspended in air, in the case of sample MC S1 ash. The emission spectrum acquired from plasma induced in air shows the emission lines of N(I), N(II), H(I) and O(I), whereas spectra acquired from plasma induced on flying ash, besides the air elements, clearly show the emission lines of the elements contained in the sample, such as Ca(II), Fe(I), Si(I), Al(I) and Na(I).

It is interesting to observe that the emission lines of the elements from the air, i.e., H(I), N(I), O(I), present a very similar S/N ratio, denoting that the ash particles have a negligible effect on the plasma induced in air. Note that the spectrum of O(I) in pure air is saturated because of an instrumental artifact and not because of self-absorption. However, comparing the base of the O(I) peaks in pure air and with ash it is observable they are generated by a similar population of emitters. As a matter of fact, as already reported,¹⁰ electron number density of the plasma induced in air and with ash in air present identical values during the temporal evolution within the experimental errors (Supplemental Material).

The elements that compose the ash can all be identified while suspended inside the chamber and then quantified by CF-LIBS. Table III presents elements identified in ash spectra and all selected lines for the measurements.

Table III.

Selected emission lines for the CF-LIBS elemental analysis. The wavelength and spectroscopic data of each line can be found in the NIST and Kurucz databases.

Element	Wavelength of emission lines (nm) for CF-LIBS
Si	Si(I) 288.15; Si(I) 390.55
Al	Al(I) 394.40; Al(I) 396.15
Mg	Mg(I) 518.36; Mg(I) 383.82; Mg(I) 516.73; Mg(I) 517.27
Mn	Mn(I) 380.67; Mn(II) 294.92; Mn(II) 293.93
Na	Na(I) 588.99; Na(I) 589.59
Ca	Ca(I) 428.93; Ca(I) 430.25; Ca(I) 422.67
Ti	Ti(I) 363.54; Ti(I) 364.26; Ti(II) 334.18; Ti(II) 323.45; Ti(II) 337.28
Fe	Fe(I) 356.53; Fe(I) 382.58; Fe(I) 363.14; Fe(I) 302.06; Fe(I) 358.11
K	K(I) 766.48; K(I) 769.89

Element	Wavelength of emission lines (nm) for Ne measurements
H, Si, Ca	H(I) 656.28; Si(I) 288.15; Ca(I) 422.67

Element	Wavelength of emission line (nm) for T measurements with two lines
Ca, Si	Ca(I) 430.25; Ca(II) 317.93; Si(I) 390.55; Si(II) 385.60

Element	Wavelength of emission line (nm) for T measurements using the Saha–Boltzmann plot
Fe	Fe(I) 374.55; Fe(I) 302.56; Fe(I) 344.06; Fe(I) 344.09; Fe(I) 346.58; Fe(I) 347.54; Fe(I) 349.05; Fe(I) 379.05; Fe(I) 358.11; Fe(I) 358.49; Fe(I) 354.20; Fe(I) 371.99; Fe(I) 360.88; Fe(I) 361.87; Fe(I) 381.58; Fe(I) 382.44; Fe(I) 382.58; Fe(I) 407.17; Fe(I) 404.58; Fe(I) 427.17 Fe(I) 432.57 Fe(II) 272.75; Fe(II) 273.95; Fe(II) 275.57; Fe(II) 516.90

In Figure 5, two plasma emission spectra of different sizes of the same ash are shown (two samples from the same volcano). The spectra have different intensities but the same element lines intensity ratios, which shows that both ash are close in their elemental composition and illustrates the advantages of using CF-LIBS to determine the ash composition.

In order to verify whether the plasma fulfilled the requirements for apparent LTE conditions during our experiment, since it is necessary to apply CF-LIBS, the electron density, and temperature of each LIP with ash were measured from their integrated LIBS spectra. As mentioned earlier, electron density was measured from Stark broadening of Si and Ca lines (Table III), and the temperature was measured by using the emission coefficient ratio of two lines from the same element (Ca) and verified by Fe lines Boltzmann and Saha–Boltzmann plots (Supplemental Material). These are the starting data for the spectrum simulation and CF-LIBS elemental analysis.¹¹ It was observed that the McWhirter criterion, which provides a necessary but not sufficient condition for the electron density required to achieve a balance between electron impact mechanisms in the plasma and to ensure LTE condition,²⁸ was fully satisfied, thus the measured temperature could be assumed for all elements in the ash plasma. Table IV presents the electron number densities and the temperatures of different ash plasmas used in the CF-LIBS analyses.

Table IV.

Measured temperature and electron density of LIP with different ash suspended in the air.

Ash sample	Temperature (K)	Electron density (cm^–3)
GSV-S1	10 000 ± 400	1.0 × 10¹⁷
GSV-S2	10 100 ± 900	8.5 × 10¹⁶
GSV-S3	9500 ± 600	5.1 × 10¹⁶
MC-S1	9700 ± 500	9.4 × 10¹⁷
MC-S2	9000 ± 400	5.6 × 10¹⁷

Once our parametric study allowed us to deduce the optimal conditions for analysis, we applied CF-LIBS to quantify all elements present in all five ash samples. The identified elements and the corresponding emission lines are reported in Table III. The results obtained with standard radiometric calibration are shown in Tables V and VI, which contain oxides composition of ash samples suspended in the air. Note here that since our analyses are carried out in air, the oxygen concentration has to be considered in the 100% normalization procedure because it cannot be directly measured. The oxygen elemental fraction is then deduced using stoichiometric calculations of the oxides components present in the ash samples. The results are in good agreement with the concentrations in the literature. It can be noticed that major elements such as Si and Al are estimated with the highest precision, while minor elements are estimated with lower precision compared to the composition reported in the literature. The resulting accuracy is from 5% to 10%, for major elements, and from 15% to 20% for minor elements. Note also here that estimating the mass fraction of sodium in some samples was very difficult, because the line used for the measurement has a large optical thickness. Thus, the concentrations of the other elements are measured, and the mass fraction of Na is deduced by adjusting all the elemental concentrations in the calculation loop.

Table V.

Oxides composition of GSV 3-1 (S1) sample (grain size < 125 µm), GSV 3-1 (S2) sample (grain size > 125 µm) and GSV 3-1 (S3) sample (grain size > 250 µm).

	Literature	GSV 3-1 (S1)	GSV 3-1 (S2)	GSV 3-1 (S3)
Oxide	Wt%
SiO₂	48.25	47.98	49.37	49.96
Al₂O₃	14.00	14.32	13.25	15.47
FeO–Fe₂O₃	12.04	11.69	11.67	10.32
K₂O	0.32	0.28	0.37	0.28
CaO	8.94	9.85	10.22	8.65
Na₂O	5.94	6.04	5.93	5.01
TiO₂	2.60	2.31	2.73	5.50
MnO	0.12	0.11	0.30	0.09
MgO	7.80	7.42	6.16	4.72

Table VI.

Oxides composition of MC 1-0 (S1) sample (grain size < 125 µm) and MC 1-0 (S2) sample (grain size > 250 µm).

	Literature	MC 1-0 (S1)	MC 1-0 (S2)
Oxide	Wt%
SiO₂	60.96	58.72	57.59
Al₂O₃	19.51	18.72	21.40
FeO–Fe₂O₃	2.14	2.55	2.24
K₂O	7.02	7.65	6.38
CaO	1.71	3.20	3.70
Na₂O	8.33	8.73	8.10
TiO₂	0.13	0.20	0.22
MnO	0.21	0.23	0.37

For samples taken from the same volcano, we can observe different elemental compositions due to the ash size differences. The elemental fractions of the finest ash from both volcanic samples (S1) are closer to the reference values and with better calculation precision compared to the ones obtained for coarse ash. Therefore, it is essential to consider the heterogeneous composition of volcanic ash as a contributing factor to these differences and some of the other differences between our data and the ones from the literature obtained with EDS and XRF analyses.^12,13 As explained earlier, juvenile volcanic ash is composed of glass fragments and a certain amount of minerals and lithics, which have quite different compositions. So, it is crucial to conduct a study of different ash sizes because, in a real-life scenario, it is reasonable to hypothesize that the chemical composition of ash may vary across different sampling locations.

Finally, a set of experiments was carried out on the same samples, and this time the experimental response deduced from the self-calibration procedure based on the measurements of Fe lines intensities was used to calibrate the ash spectra (see the Intensity Calibration Using LIP Spectrum section above). All intensities of the spectral lines were normalized using the calibration factor G (Eq. 5) and CF-LIBS analysis was applied. In Figure 6, the spectra of the ash sample GSV-S1 are reported, respectively, with no radiative calibration, with standard lamp calibration, and with the self-calibration method proposed in this work. As can be observed the spectra corrected with a standard lamp and with self-calibration are similar except for those optically thick lines where self-absorption should be optimized further.

The obtained results of two ash samples are shown in Table VII which contain oxides composition of GSV (S1), and MC (S1) ash suspended in the air. The concentrations are obtained with lower accuracy due to the lower precision of the calculated apparatus response compared to G_app deduced from the calibration lamps (standards). Nevertheless, the obtained precision of elemental fractions is sufficient for volcanic plumes study and for the purpose of our analyses, which consists of performing CF-LIBS analyses without any calibration.

Table VII.

Oxides composition of GSV 3-1 (S1) and MC 1-0 (S1) samples (grain size < 125 µm).

	GSV 3-1 (Lit)	GSV 3-1 (S1)	MC 1-0 (Lit)	MC 1-0 (S1)
Oxides	Wt%
SiO₂	48.25	49.93	60.96	55.86
Al₂O₃	14.00	14.96	19.51	20.85
FeO–Fe₂O₃	12.04	11.22	2.14	2.70
K₂O	0.32	0.47	7.02	5.08
CaO	8.94	6.66	1.71	5.67
Na₂O	5.94	6.92	8.33	9.46
TiO₂	2.60	3.16	0.13	0.28
MnO	0.12	0.06	0.21	0.10
MgO	7.80	6.62	–	–

To sum up, CF-LIBS is a technique that can perform the quantitative analysis of ash suspended in the air without any preliminary measurement or calibration. Only the recording of plasma emission spectra is enough to identify and quantify elements composing volcanic ash. Thus, this technique can be employed for real-time characterization of material emitted from volcanic plumes, and a compact LIBS instrumentation can be developed for in situ measurements.

Conclusion

In this paper, a self-calibrated LIBS approach that considers self-absorption during calculations was applied to analyze volcanic ash of different sizes under experimental conditions close to those in the field. Ash were suspended in the air in a chamber by laser-induced shockwaves. A LIP was generated in the air with and without suspended ash. Recorded plasma spectra enabled the detection and quantification of the ash components without any previous sample preparation, calibration, or any strict requirement of conventional laboratory tools, such as matrix-matched standards and experimental reproducibility. The intensity calibration of the ash spectra was first carried out with standard calibration lamps, then with a calibration factor deduced from the ratio of measured and calculated Fe lines intensities. Elemental fractions of the ash elements were close to the values found in the literature, despite the heterogeneous nature of volcanic ash components, which confirmed the robustness of CF-LIBS analysis. Furthermore, concentrations obtained with the self-calibration factor resulted in having enough precision for the desired applications which consist of estimating the elemental concentration of the major oxides with a higher precision than minor elements and traces.

Therefore, CF-LIBS can be employed for real-time characterization of material emitted in volcanic plumes since concentrations and system calibration parameters are extracted directly from the ash emission spectrum itself. Such an implementation has a clear usefulness, especially for field applications in harsh environments where full control of the experimental parameters is not possible.

It is important to underline that applying the present LIBS approach in a real-life volcanic scenario, where sampled ash can have different sizes and consequently different compositions, would measure a composition resulting from the averaging of all the sampled particles. This, in turn, implies that further complementary techniques may be required in order to retrieve all the necessary information about the activity of the investigated volcano.

Supplemental Material

sj-docx-1-asp-10.1177_00037028241241076 - Supplemental material for Self-Calibrated Laser-Induced Breakdown Spectroscopy for the Quantitative Elemental Analysis of Suspended Volcanic Ash

Supplemental material, sj-docx-1-asp-10.1177_00037028241241076 for Self-Calibrated Laser-Induced Breakdown Spectroscopy for the Quantitative Elemental Analysis of Suspended Volcanic Ash by Aya Taleb, Marcella Dell’Aglio, Rosalba Gaudiuso, Daniela Mele, Pierfrancesco Dellino and Alessandro De Giacomo in Applied Spectroscopy

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was carried out within the RETURN Extended Partnership and received funding from the European Union Next-Generation EU (National Recovery and Resilience Plan–NRRP, Mission 4, Component 2, Investment 1.3–D.D. 1243 2-8-2022, PE0000005).

ORCID iD

Alessandro De Giacomo

Marcella Dell’Aglio

Supplemental Material

All supplemental material mentioned in this paper is available in the online version of the journal.

References

Miziolek Andrzej

Palleschi

Schechter

. Laser-Induced Breakdown Spectroscopy (LIBS): Fundamentals and Applications. New York: Cambridge University Press, 2006.

Hahn

D.W.

Omenetto

. “Laser-Induced Breakdown Spectroscopy (LIBS), Part II: Review of Instrumental and Methodological Approaches to Material Analysis and Applications to Different Fields”. Appl. Spectrosc. 2012. 66(4): 347–419.

Dell'Aglio

De Giacomo

Gaudiuso

De Pascale

, et al. “Laser Induced Breakdown Spectroscopy of Meteorites as a Probe of the Early Solar System”. Spectrochim. Acta B. 2014. 101: 68–75.

Dell'Aglio

López-Claros

Laserna

J.J.

Longo

, et al. “Stand-Off Laser Induced Breakdown Spectroscopy on Meteorites: Calibration-Free Approach”. Spectrochim. Acta B. 2018. 147: 87–92.

Maurice

Wiens

R.C.

Saccoccio

Barraclough

, et al. “The ChemCam Instrument Suite on the Mars Science Laboratory (MSL) Rover: Science Objectives and Mast Unit Description”. Space Sci. Rev. 2012. 170: 95–166.

Scarpa

Gasparini

. “A Review of Volcano Geophysics and Volcano-Monitoring Methods”. In: Monitoring and Mitigation of Volcano Hazards”. Berlin; Heidelberg: Springer, 1996. Pp. 3–22.

Jenkins

S.F.

Spence

R.J.S.

Fonseca

J.F.B.D.

Solidum

R.U.

, et al. “Volcanic Risk Assessment: Quantifying Physical Vulnerability in the Built Environment”. J. Volcanol. Geotherm. Res. 2014. 276: 105–120.

Galle

Arellano

Bobrowski

Conde

, et al. “A Multi-Purpose, Multi-Rotor Drone System for Long-Range and High-Altitude Volcanic Gas Plume Measurements”. Atmos. Meas. Tech. 2021. 14(6): 4255–4277.

Arellano

S.R.

Hall

Samaniego

Le Pennec

J.-L.

, et al. “Degassing Patterns of Tungurahua Volcano (Ecuador) During the 1999-2006 Eruptive Period, Inferred from Remote Spectroscopy Measurements of SO₂ Emissions”. J. Volcanol. Geotherm. Res. 2008. 176(1): 151–162.

10.

De Giacomo

Dell’Aglio

Salajkvá

Vaníčková

, et al. “Real-Time Analysis of the Fine Particles in Volcanic Plumes: A Pilot Study of Laser Induced Breakdown Spectroscopy with Calibration-Free Approach (CF-LIBS)”. J. Volcanol. Geotherm. Res. 2022. 432: 107675.

11.

Hermann

. System and Method for the Quantitative Analysis of the Elementary Composition of Matter by Laser-Induced Plasma Spectroscopy (LIBS). CNRS Patent WO2010052380A1. Filed 2008. Issued 2010.

12.

Dürig

Mele

Dellino

Zimanowski

. “Comparative Analyses of Glass Fragments from Brittle Fracture Experiments and Volcanic Ash Particles”. Bull. Volcanol. 2012. 74: 691–704.

13.

Mele

Sulpizio

Dellino

La Volpe

. “Stratigraphy and Eruptive Dynamics of a Pulsating Plinian Eruption of Somma-Vesuvius: The Pomici di Mercato (8900 Years BP)”. Bull. Volcanol. 2011. 73(3): 257–278.

14.

Mele

Costa

Dellino

Sulpizio

, et al. “Total Grain Size Distribution of Components of Fallout Deposits and Implications for Magma Fragmentation Mechanisms: Examples from Campi Flegrei Caldera (Italy)”. Bull. Volcanol. 2020. 82(31): 31. https://doi.org/10.1007/s00445-020-1368-8

15.

Ciucci

Corsi

Palleschi

Rastelli

, et al. “New Procedure for Quantitative Elemental Analysis by Laser-Induced Plasma Spectroscopy”. Appl. Spectrosc. 1999. 53(8): 960–964.

16.

Gerhard

Hermann

Mercadier

Loewenthal

, et al. “Quantitative Analyses of Glass via Laser-Induced Breakdown Spectroscopy in Argon”. Spectrochim. Acta B. 2014. 101: 32–45.

17.

Hermann

Lorusso

Perrone

Strafella

, et al. “Simulation of Emission Spectra from Nonuniform Reactive Laser-Induced Plasmas”. Phys. Rev. E. 2015. 92(053103): 1–15. https://doi.org/10.1103/PhysRevE.92.053103

18.

Gerhard

Taleb

Pelascini

Hermann

. “Quantification of Surface Contamination on Optical Glass via Sensitivity-Improved Calibration-Free Laser-Induced Breakdown Spectroscopy”. Appl. Surf. Sci. 2021. 537(147984): 1–7. https://doi.org/10.1016/j.apsusc.2020.147984

19.

Hermann

Grojo

Axente

Gerhard

, et al. “Ideal Radiation Source for Plasma Spectroscopy Generated by Laser Ablation”. Phys. Rev. E. 2017. 96(053210): 1–6. https://doi.org/10.1103/PhysRevE.96.053210

20.

Hermann

Gerhard

Burger

Craciun

, et al. “Progress in Calibration-Free Laser-Induced Breakdown Spectroscopy”. Spectrochim. Acta B. 2022. 200: 106595. https://doi.org/10.1016/j.sab.2022.106595

21.

Poggialini

Campanella

Cocciaro

Lorenzetti

, et al. “Catching Up on Calibration-Free LIBS”. J. Anal. At. Spectrom. 2023. 38(9): 1751–1771.

22.

Aragón

Aguilera

J.A.

. “Spatial and Temporal Scaling and Common Apparent Excitation Temperature of Laser-Induced Plasmas Generated at Constant Irradiance with Different Pulse Energies”. J. Appl. Phys. 2008. 103(1): 013310.

23.

Aguilera

J.A.

Aragón

. “Apparent Excitation Temperature in Laser-Induced Plasmas”. J. Phys. Conf. Ser. 2007. 59(1): 210–217.

24.

Bousquet

Gardette

Motto-Ros

Gaudiuso

, et al. “Plasma Excitation Temperature Obtained with Boltzmann Plot Method: Significance, Precision, Trueness, and Accuracy”. Spectrochim. Acta B. 2023. 204: 106686.

25.

Taleb

Shen

Mory

Cieslik

, et al. “Echelle Spectrometer Calibration by Means of Laser Plasma”. Spectrochim. Acta B. 2021. 178: 106144.

26.

Carranza

J.E.

Hahn

D.W.

. “Assessment of the Upper Particle Size Limit for Quantitative Analysis of Aerosols Using Laser-Induced Breakdown Spectroscopy”. Anal. Chem. 2002. 74(21): 5450–5454.

27.

Hahn

D.W.

Lunden

M.M.

. “Detection and Analysis of Aerosol Particles by Laser-Induced Breakdown Spectroscopy”. Aerosol Sci. Technol. 2000. 33(1–2): 30–48.

28.

Cristoforetti

De Giacomo

Dell'Aglio

Legnaioli

, et al. “Local Thermodynamic Equilibrium in Laser-Induced Breakdown Spectroscopy: Beyond the McWhirter Criterion”. Spectrochim. Acta B. 2010. 65(1): 86–95.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.21 MB