Laser induced breakdown spectroscopy inside liquids

Spectrochimica Acta Part B 101 (2014) 288–311

Contents lists available at ScienceDirect

Spectrochimica Acta Part B journal homepage: www.elsevier.com/locate/sab

Review

Laser induced breakdown spectroscopy inside liquids: Processes and analytical aspects V. Lazic a,⁎, S. Jovićević b a b

ENEA (UTAPRAD-DIM), Via. E. Fermi 45, 00044 Frascati (RM), Italy Institute of Physics, University of Belgrade, 11080 Belgrade, Serbia

a r t i c l e

i n f o

Article history: Received 27 May 2014 Accepted 5 September 2014 Available online 16 September 2014 Keywords: Laser induced plasma Laser induced bubble LIBS Liquid Ablation

a b s t r a c t This paper provides an overview of the laser induced breakdown spectroscopy (LIBS) inside liquids, applied for detection of the elements present in the media itself or in the submerged samples. The processes inherent to the laser induced plasma formation and evolution inside liquids are discussed, including shockwave generation, vapor cavitation, and ablation of solids. Types of the laser excitation considered here are single pulse, dual pulse and multi-pulse. The literature relative to the LIBS measurements and applications inside liquids is reviewed and the most relevant results are summarized. Finally, we discuss the analytical aspects and release some suggestions for improving the LIBS sensitivity and accuracy in liquid environment. © 2014 Elsevier B.V. All rights reserved.

Contents 1. 2.

Introduction . . . . . . . . . . . . . . . . . . . . . . . Fundamentals . . . . . . . . . . . . . . . . . . . . . . 2.1. Optical properties of water . . . . . . . . . . . . . 2.2. Laser induced breakdown in liquids . . . . . . . . . 2.3. Laser induced cavitation and shockwaves . . . . . . 2.4. Laser ablation in liquids . . . . . . . . . . . . . . 2.5. Single pulse versus dual or multi pulse laser excitation 3. LIBS measurements inside liquids . . . . . . . . . . . . . 3.1. Analysis of bulk liquids . . . . . . . . . . . . . . . 3.1.1. Plasma stability and effects of analytes . . . 3.1.2. Results of bulk-liquid analysis . . . . . . . 3.1.3. Effects of liquid pressure . . . . . . . . . . 3.1.4. Matrix effect . . . . . . . . . . . . . . . 3.2. Direct analysis of submerged solid targets . . . . . . 3.2.1. Single pulse LIBS on submerged solids . . . . 3.2.2. Dual pulse LIBS on submerged solids . . . . 3.3. Analysis of submerged targets under gas flow . . . . . 3.4. Analysis of submerged soft materials . . . . . . . . . 3.5. Analytical aspects . . . . . . . . . . . . . . . . . 4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . .

⁎ Corresponding author. E-mail address: [email protected] (V. Lazic).

http://dx.doi.org/10.1016/j.sab.2014.09.006 0584-8547/© 2014 Elsevier B.V. All rights reserved.

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . .

289 289 289 290 292 293 295 298 298 298 299 300 302 302 302 303 305 306 306 309 309 309

V. Lazic, S. Jovićević / Spectrochimica Acta Part B 101 (2014) 288–311

2. Fundamentals 2.1. Optical properties of water The most common liquid involved in the LIBS experiments is water, and in this section we discuss briefly its optical properties, relevant for the plasma formation and detection. A complex index of refraction ñ = n + ik contains the real part n that indicates the phase velocity, while the imaginary part k is related to the absorption losses when the electromagnetic wave propagates through the material. The real part n of the refractive index of water smoothly decreases with the wavelength in the range 200–1200 nm (Fig. 1 — top), which is of interest for LIBS [31]. Refractive index of pure water at room temperature and atmospheric pressure is between 1.394 and 1.326 in the range of 226–1013 nm [32]. The imaginary part k of the complex refraction index of water is also tabulated for

1.40

Refractive index

Laser induced breakdown spectroscopy (LIBS) is based on plasma generation by short and intense laser pulses focused on solid, liquid or gaseous samples. The spectrally resolved detection of the plasma radiation, which contains the excited atoms and ions originating from the sample, provides qualitative and quantitative information about the elements present in the material. Although LIBS has not yet reached very high sensitivity and accuracy compared to some other laboratory methods, it has been tagged as a superstar analytical technique [1] due to its inherent advantages such as: i) measurement capability in any environment: inside gases and liquids of any type and pressure; ii) characterization of any type of the sample: solid, liquid, gaseous, and aerosols; iii) sample preparation is not necessary; iv) the measurements are in real-time and might be implemented insitu; v) the probing is contactless and might be performed at distances also beyond 100 m; vi) the costs of the LIBS instruments are relatively low compared to other equipment with analogue performances; and vii) the instrument might be compacted to a portable format. Due to an increased importance and wide-spread use of the LIBS technique, different review papers have been recently published, mainly focused on: history of LIBS and the instrument developments [2], a comprehensive overview of the processes involved in LIBS [3], instrumentation and methodology for material analysis [4], portable and stand-off LIBS instruments [5,6], detection and modeling of the plasma and its parameters [7,8], biological [9,10] and biomedical applications [11,12], soil analysis [13], and some specific applications such as environmental monitoring, space exploration and cultural heritage [14]. The developments in LIBS over the years 2008–2012 are reviewed by F.J. Fortes et al. [15]. Presently, LIBS is the only available technique for direct elemental analysis of bulk liquids and submerged targets. The chemical characterization of bulk liquids might be employed for in-situ detection of leakages in industrial and power plants [16–18], other kinds of water contamination [19,20], geothermal winds in deep oceans [21, 22], and direct analysis of liquids inside transparent containers [23]. Characterization of the submerged materials could be exploited also for feedback control in laser surgery, usually performed with liquid coverage [24], then for recognition of underwater building materials and archeological objects [25,26], and determination of recent pollution or bio-activity in waters through sampling of seabed's surface [27]. The laser driven plasma formation and excitation on or inside liquids are not efficient processes because a great portion of the laser energy is expended for liquid vaporization. Both water and organic solutions contain hydrogen, which contributes to a rapid thermalization and cooling of the plasma. In water, due to abundance of oxygen atoms, different plasma species undergo rapid oxidation; this reduces availability of the excited analyte atoms or ions. Once the plasma is created inside liquid, the high density and nearly incompressible medium strongly confines the plume; the corresponding effects on the plasma evolution and LIBS signal are discussed in [28]. Plasma formation inside liquids is accompanied by emission of intense shockwaves, which affect the ablation threshold and rate of submerged targets. The plasma emission and expansion are followed by growth of a vapor cavity, which lifetime can exceed a few milliseconds. Although the LIBS applications inside liquids raised a large interest for in-situ measurements and a number of works reported the relative studies, the last review paper about underwater LIBS, by De Giacomo and co-workers, was published in 2007 [29]. Some discussion about LIBS in liquids is also given in one section of [28] — the review paper about the effects of the background environment on the plasma formation, evolution and its optical emission. The LIBS technique applied on or inside liquids is concisely discussed by Lazic in the book chapter 6 of [30], where following cases were considered:

ablation of a free surface (static, flowing and jet), breakdown on droplets and liquid aerosols, ablation of frozen solutions, breakdown in the media and sample ablation inside liquids. In the present paper we focus on LIBS measurements inside bulk liquids, relative both to the solutes and to the submerged solid samples. The relevant processes involved in plasma formation and evolution are described, and the results obtained by various research groups are discussed. Finally, we present different analytical aspects to take into account when performing the measurements and the data processing.

1.38

1.36

1.34

1.32 200

400

200

400

600

800

1000

1200

600

800

1000

1200

10

Absoprtion coefficient (cm-1)

1. Introduction

289

1 0.1 0.01 1E-3 1E-4

Wavelength (nm) Fig. 1. Refractive index (top) and absorption coefficient (bottom) of pure water at room temperature and pressure as a function of wavelength — data from [31]; the vertical lines indicate the wavelengths of Nd:YAG laser: 1064 nm (red), 532 nm (green), 355 nm (light blue) and 266 nm (violet). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

290


different wavelengths. The light attenuation along water column can be calculated from Beer–Lambert law considering that the absorption coefficient α is related to k through equation:

α¼

4πκ ðλÞ : λ

ð1Þ

For pure water at room conditions, dependency of α on the radiation wavelength is depicted in Fig. 1 — bottom. In the considered spectral range, water absorption changes over three orders of magnitude, and it has a transmission window in the blue–green region. This spectral window is favorable for the laser excitation in bulk water while the high absorption regions facilitate the plasma formation on water surface or droplets. Some of the most interesting LIBS applications regard the ocean exploration and the recognition of materials undersea. Austin and Halikas [33] exhaustively reviewed the literature on measurements of the real refractive index relative to distilled and seawater. Their report contains extensive tables and interpolation algorithms for the refractive index as a function of wavelength in the range 400–700 nm, salinity 0–35%, temperature 1–30 °C, and pressure 1–1080 atm. The refractive index n increases with salinity and pressure, and with lowering of the temperature. However, n changes by less than 3% at a single wavelength for the parameter variations over the entire range. Seawater contains various dissolved salts, which average is of about 35 parts per thousand by weight. The typical concentrations of Na, Mg, K and Ca are 10770, 1290, 380 and 412 parts per million (ppm), respectively. In normal conditions and for the wavelengths in visible region, the sea salts have very small effects on absorption and refraction. On the other hand, these salts in water increase molecular scattering [34], and this affects the transmission of the laser beam and of the plasma radiation. The scattering coefficient decreases with water pressure and grows with its temperature. Natural waters contain inorganic particulates and different types of organic materials with the equivalent diameter in the range from 0.1 μm (small colloids and viruses) to 100 μm (larger phytoplankton). Due to strong light scattering on particulates, the laser induced plasma formation in turbid waters might be difficult to achieve and would require a short optical path through the liquid media. 2.2. Laser induced breakdown in liquids LIB in liquids rely on electron cascade ionization (CI) or on direct ionization through multi-photon absorption [35]. Both mechanisms of plasma formation can occur also in solids and gases. The threshold irradiance required for the plasma formation depends on the medium properties such as ionization energy, type and concentration of the impurities. For a given medium, the LIB threshold depends on the laser beam characteristics: wavelength, pulse width, focusing angle and beam diameter. Experimentally, the LIB threshold in liquids is determined through detectable effects of laser radiation on media, like plasma emission, bubble formation and release of the shock waves. The first theoretical calculations of LIB threshold values were based on fitting of the experimental data with different simple formalisms, such as lucky electron model [36]. This model provided the evidence of CI in liquids as a primary mechanism for LIB when applying the pulses of duration in ps–ns range [37]. In combination with moving breakdown model, lucky electron approach provided an explanation for delayed plasma formation farther from the focal point i.e. towards the incoming laser beam. The basic assumption is that the electric field, oscillating perpendicularly to the beam propagation, creates the initial free electrons (lucky electrons) and that the breakdown occurs in layers where each layer establishes its own plasma (moving breakdown). Another theoretical approach for CI uses the rate equation formalism and

the Drude's model, which assumes a uniform scattering rate for all free electrons. The previously mentioned models do not take into account multiphoton ionization (MPI) nor include the impact of impurities inside liquids, so these models cannot be used for calculation of LIB threshold in general case. For MPI in liquids, the most frequently employed model is Keldysh's, which considers liquids as amorphous semiconductors [38]. Kennedy [39] provided first order model for estimation of LIB threshold that includes both MPI and CI regimes. By simulations and comparison with the experimental data, it was found that self-focusing in liquids plays an important role in the case of short laser pulses (ps and fs) [40]. To initiate the process of CI, some seed free electrons must be already present in the focal volume. These electrons absorb the laser photons during collisions with heavy particles; the process is named as inverse Bremsstrahlung absorption. The role of the heavy particles is to preserve momentum and energy. When a free electron gains the energy sufficient to ionize bound electron, their collision produces two free electrons of lower energies. Repetition of this process leads to a multiplication of free electrons i.e. to an electron cascade (breakdown). Nevertheless, in order to initiate the electron cascade, the rate of energy losses must be smaller than the rate of energy absorption in the focal volume. The energy losses are mainly due to inelastic collisions but there are also losses related to electron–ion recombination and to diffusion of electrons out of the focal volume. The diffusion is of particular importance for the breakdown induced by long pulses [35,41]. Plasma formation through MPI does not require an initial presence of free electrons, thus the process does not depend on impurities inside the liquid. The process is fast with respect to CI, and can occur also during short laser pulses (fs and ps) where the loss mechanisms can be ignored. On the other hand, the energy densities required for MPI are much higher than for CI, since MPI is a nonlinear process. Because of that, for ns pulses CI is a dominant mechanism for the plasma formation, except in the case of pure liquids where MPI is necessary to create the initial free electrons for the successive avalanche ionization. In non-distilled water, the seed electrons for CI mainly come from thermal excitation of the impurities. Consequently, the impurity concentration and type strongly affect the LIB threshold for ns pulses. Cascade ionization is a probabilistic process [35,40,42] because it depends on even slight density fluctuations in the focal volume; its threshold is usually determined experimentally for 50% or 100% probability of the breakdown events. The experimentally measured LIB threshold for excitation at 1064 nm with 7 ns long pulses is given in Table 1 [43]. For example, these data show that LIB threshold is 3–7 times higher in distilled than in tap water, where the differences are reduced with increasing of the calculated laser spot size. The dependence of the LIB threshold on the spot size was not always observed by other authors. Vogel et al. [44] mentioned that self-

Table 1 Experimental LIB threshold (IBD) for pulses of 7 ns duration at 1064 nm, measured for different water samples, corresponding to 100% and 50% breakdown probabilities [43]. Spot size [μm2] 70.7

132.9

495.0

Medium Distilled Saline Vitreous Tap water Distilled Saline Vitreous Tap water Distilled Saline Vitreous Tap water

IBD 100% [W/cm2] 10

2.1 · 10 1.85 · 1010 1.5 · 1010 3.0 · 109 1.55 · 1010 1.35 · 1010 1.17 · 1010 2.6 · 109 5.3 · 109 4.25 · 109 3.5 · 1010 1.5 · 109

IBD 50% [W/cm2] 2.5 · 1010 1.25 · 1010 1.02 · 1010 2.0 · 109 1.05 · 1010 7.2 · 109 7.8 · 109 1.5 · 109 2.5 · 109 2.25 · 109 2.0 · 109 0.9 · 109


focusing could be responsible for the measured decrease of the calculated threshold irradiance for larger spot sizes. With shortening of the pulse duration, MPI starts to compete with a relatively slow CI, and the influence of the impurities in liquids on the LIB threshold gradually decreases (Fig. 2), as confirmed experimentally [45]. A preferable path to water ionization depends on the laser wavelength. Dissimilarly from the excitation at 1064 nm with ns pulses, LIB threshold at 532 nm does not exhibit any impurity dependence (Fig. 2). These differences were explained by a number of photons necessary to start MPI in water, which ionization energy is 6.5 eV [38,46]: namely 3 photons at 532 nm or 6 photons at 1064 nm. For the wavelength of 532 nm MPI is a low order nonlinear process, which dominates the plasma initiation regardless of the impurity level. For the laser excitation at 1064 nm, the probability of MPI is very low and the dominant breakdown mechanism is CI, which threshold is dependent on impurities that provide the initial free electrons. Development of CI under infrared laser pulses is also favored by more efficient inverse Bremsstrahlung absorption, which coefficient is αIB ∝ λ3N2e . For small focusing angles, self-focusing i.e. filamentation of the laser beam might occur due to non-linear refraction index of media: n ¼ n0 þ n2 I

ð2Þ

where I is the field intensity, and both indexes, linear n0 and nonlinear n2, are wavelength dependent. Self-focusing arises due to Kerr's effect and it leads to local changes of refractive index of the medium along the beam path. Its appearance is connected with the critical input laser power Pcr, here defined for Gaussian laser beam [47]: P cr ¼

3:77λ2 8πn0 n2

ð3Þ

where λ is wavelength of the laser radiation. The experimentally measured values of Pcr and n2 differ from one work to another, as summarized by Wilkes et al. [48]. Recently, Helle and co-workers reported Pcr of only 1 MW when focusing the laser beam at 532 nm inside the distilled water; here measured filament's length was larger than 50 cm [49]. Due to increased local irradiance by self-focusing, the threshold (109 W/cm2) for Stokes-shifted Stimulated Raman Scattered (SRS) was exceeded and about 30% of the incident beam energy was converted into longer wavelength (shift 3550 cm−1). For high peak power ps and fs pulses at long wavelengths, the critical power Pcr is easily reached, particularly if focusing the beam with an

Fig. 2. Pulse-width dependence of LIB threshold at laser wavelengths of 1064 nm and 532 nm, for pure and impure waters or for all (pure and impure) at [35]. (Reprinted with permission from Ref. [35], copyright 1997, Elsevier).

291

Fig. 3. Plasma in distilled water excited with ns pulses: left — for different energies (μJ) and focusing angle of 22°; right — for different focusing angles and laser energy of 10 mJ [44].

optical system having a low numerical aperture (NA). Filamentation was observed in distilled water also with low energy IR ns pulses for focusing angles below 2° [44]. By increasing the laser energy (Fig. 3 — left) or reducing the NA of the optical system (Fig. 3 — right) the plasma becomes elongated towards the laser and it shields more efficiently the focal volume. Toker et al. [50] investigated the LIB produced by ns pulses at 1064 nm inside water. In these studies, the spatially and the temporally resolved Schlieren and shadowgraphy techniques were employed. The experiment demonstrated that, except for very low laser energies, the breakdown and the successive cavitation occur in multiple sites along the beam propagation (Fig. 4). The length of the laser spark column was anomalously longer than the theoretically predicted values, with the length proportional to the laser energy. Among the possible explanations for this effect, the authors mention self-focusing of the laser beam. A discrete structure of the breakdown was attributed to the inclusion particles present in tap water, which trigger the plasma formation. Successively, the same experimentation was comparatively performed on tap water and on four different alcohols; here, the interferometry measurements were also introduced [51]. In all cases, the initial plasma formation and cavitation in multiple sites were observed, considered to be caused by impurity inclusions. In later detection times, after about 100 ns from the laser pulse, the growing micro-plasmas unify into

Fig. 4. Examples of dark field Schlieren pictures of LIB in tap water. Laser pulse energy is 74 mJ, time delay of the diagnostic light: (a) 1 ns and (b) 14 ns [50]. (Reproduced with permission from Ref. [50], copyright 2004, Elsevier).

292


continuous structure. At longer delays, warmed channels in the focal volume were detected in all liquids; the corresponding temperature rise was about 47 K and 82 K in water and ethanol, respectively. The main observed differences between LIB in water and in alcohols are the following: (a) warmed channels in alcohols generate cylindrical shockwaves, while in water they are spherical; and (b) velocities of the expanding spherical shockwaves and of the micro-bubbles in water are remarkably higher than in alcohols. Recently, a discrete structure of the initial plasma was observed also in distilled water [52], and this fact raises doubts about the impurity occlusions as the only source for plasma formation in multiple sites. The striking feature in different pictures shown in [50–52] and in Fig. 4 is a rather regular distance between the plasma micro-sites. Evans and Camacho-López [52] hypothesized that the discrete, regular plasma structure is caused by ponderomotive self-focusing of the polarized laser beam: every time the beam collapses in the space, the electron cavitation is created and its position coincides with that of the microplasma. The same locations are the centers for the successive bubble cavitation 2.3. Laser induced cavitation and shockwaves Inside a weakly compressible liquid environment, the expansion of the laser induced plasma is inhibited. The confining effect of the liquid leads to considerably higher local temperatures and pressures than in a gaseous environment because the plasma expansion and its adiabatic cooling proceed more slowly. In liquid environment, a large portion of the input laser energy might be transformed into mechanical energy (Section 2.5). Initially, the hot plasma expands at supersonic velocity; it compresses the surrounding liquid and creates a shockwave, which pressure might exceed 10 kbar [53]. In addition to the shockwave, a thin layer of vapor and diffused gas is formed around the high temperature plasma. The local liquid evaporation continues and the vapor expands almost adiabatically, causing a growth of a cavity containing both the vapor and the diffused gas [35]. During the growth of laser formed bubble (LFB) the pressure inside decreases due to both volume enlargement and vapor condensation across the interface. Immediately after formation, the bubble has a very high inner pressure and it expands too rapidly for establishing the equilibrium with the hydrostatic pressure of liquid. At a certain point, the bubble pressure becomes lower than the pressure of the surrounding media. The increasing difference between the internal and the external pressure slows down the bubble expansion and brings it to a halt. At this point, the kinetic energy of the liquid during bubble expansion has been transformed into potential energy of the expanded bubble. Knowing the maximum radius Rmax of fully expanded bubble, it is possible to estimate bubble energy EB [54]: EB ¼

4π 3 ðp −pv ÞRmax 3 0

ð4Þ

where p0 is the hydrostatic pressure, and pv is the vapor pressure inside the bubble (for water vapor pv = 2330 Pa at 20 °C and p 0 = 0.1 MPa [55]). Inside the fully expanded cavity, the vapor temperature approaches that of the surrounding liquid while the pressure is reduced at the saturated vapor pressure. At this point, pv b bp0 [56] and the bubble begin to shrink until the rate of condensation cannot offset the volumetric reduction. The final phase of the bubble collapse produces a rapid increase of the inside gas temperature and pressure, giving rise to a second shockwave and to re-expansion of the cavity. Oscillations of LFB may continue for many cycles of expansion and collapse [56,57], with total duration of a few milliseconds. The time tc of the first bubble collapse is approximately 2 te, where the latter period refers to the expansion time. A common way to

measure the bubble's lifetime tc is time resolved laser scattering [58]. In the absence of a rigid boundary, the maximum bubble radius Rmax can be calculated from Rayleigh's equation [59]: Rmax ¼

tc qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1:83 ρ =p −p 0

0

ð5Þ

v

where ρ0 = 998 kg/m3. The maximum radius of LFB increases with the pulse energy [60] and the pulse width up to nanosecond range [41]. Due to the large specific volume of water vapor, bubbles of considerable size are formed. For ns pulse excitation in water, Rmax might reach 4 mm, with the corresponding expansion time tc of about 400 μs [61]. In the case of LFB generated by ablation of a solid target, the growing cavity is approximately hemispherical. During compression, the bubble deforms at the rigid boundary. In the successive cycle, an additional bubble deformation occurs due to a jet development during the final collapse [62]. The laser induced plasma formation and cavitation in multiple sites often occur in the LIBS experiments. Peel et al. [63] observed a number of small initial bubbles, which expansion brought to a coalescence and formation of an almost spherical cavity. In the experiment described in [58], cavitation in multiple sites was evident, but the bubbles remained distinct over their entire evolution. Here, the maximum lateral gas bubble expansion was achieved for relatively low laser pulse energy; at higher irradiances the plasma was elongated and produced two or more spherical cavities of smaller diameters (see Section 3.5). Thornton et al. studied comparatively the cavitation in water, induced by pulses with the energy of 25 mJ and duration of 20 ns and 150 ns, respectively [64]. Longer pulses produce lower irradiance, so the plasma initiation was observed in well localized volume; the corresponding cavitation develops from only two centers that remain well distinct during the bubble's lifetime. For the shorter laser pulse, the LIB threshold was achieved well before the focal plane; the cavitation and the shockwaves developed from several centers along the beam propagation; the initial cavities partially coalesced during time. The longer pulse efficiently heats the initially produced plasma and the final, overall volume of the vapor cavity and the corresponding collapse time are larger than in the case of 20 ns pulses. Symmetrically collapsing bubble, generated faraway from rigid boundaries, might produce sonoluminescence light emission [65,66]. This short living radiation is characterized by a broad continuum spectra [66] attributed to ion-electron recombination inside the compressed cavity. Sonoluminescence was observed only during the first collapse, followed by generation of an intense shockwave, which causes important energy losses. During the successive rebounds and collapses, the bubble energy is not sufficient to produce sonoluminescence again [67]. If the LIBS measurements employ non-gated detectors, which typical integration time is in the order of milliseconds, the acquired spectra might be affected by sonoluminescence [27], occurring much later than the LIBS signal. Bubbles inside liquids produce size-dependent light scattering [68,69]. The incident rays reflect and refract at the wall of a cold bubble because of its low refraction index nb with respect to that of the surrounding media nw. The bubble's inner pressure and temperature vary by few orders of magnitude during its evolution. The corresponding changes of the refractive index were estimated in [70]: inside a high pressure LFB, the refractive index is larger than 1.23 while at the maximum expansion it drops down to 1.00. During the collapsing phase the vapor pressure and nb grow again. Refractive index nw of the surrounding water at room temperature and pressure is between 1.39 and 1.33 (Fig. 1). Due to difference among nw and nb, the spherical cavity acts as a negative lens both for incident and out-coming light. The maximum lens power occurs at the bubble's full expansion; at this point the LFB measured in [70] corresponds to a lens with focal length of only f = − 11 mm at 1064 nm.


293

Fig. 5. (a) Inhomogeneities inside the bubble at different delays from the first laser pulse (Schlieren photography); (b) craters produced by the second pulse at the corresponding interpulse delays [70]. (Reproduced with permission from Ref. [70], copyright 2012, American Institute of Physics).

In dual pulse (DP) LIBS experiments the second laser pulse is sent through the bubble already formed. If the bubble is already well expanded and cooled, a strong beam defocusing must be taken into account. Defocusing by LFB enlarges the ablation crater generated by the successive laser pulse and reduces the ablation efficiency. Inside the expanded, cooled LFB a not uniform vapor condensation occurs (Fig. 5a) and this further perturbs the beam intensity distribution [70]. As a consequence, the ablation craters generated through LFB are not only enlarged by defocusing, but also irregular (Fig. 5b). The plasma radiation generated inside LFB by the second laser pulse is deflected at the cavity interface, as depicted in Fig. 6a. The rays initially traveling upward with respect to the substrate plane are even more deviated from the axis of the collection system than the parallel rays. Differently, some of the rays initially propagating downward, if not hitting the substrate, might be captured by the detector but this involves only a narrow span of the initial propagation angles. Consequently, the external plasma layers almost cannot be captured by the collecting optics and a dark ring appears at the detector plane, as shown in Fig. 6b. In the specific case, the radiation from more than 50% of the plasma volume escapes the detector, causing the large signal losses. Simultaneously, Snell's reflections at the cavity wall redistribute the observed plasma intensity and create a bright spot in proximity of the bubble center (Fig. 6c); this effect was evident after attenuating the plasma

emission on the detector [57]. Formation of this bright spot inside the expanded bubble partially recovers the LIBS signal (Section 2.5). Defocusing and reflection of the light by the vapor cavity increase at shorter radiation wavelengths due to larger refractive index of media (Fig. 1). Both processes alter the externally detected spectral distribution of the plasma produced by DP excitation. 2.4. Laser ablation in liquids The presence of liquid layer above the sample increases the efficiency of the laser ablation (LA) compared to solid–gas interface. This enhancement is due to a high acoustic pressure with long duration [71, 72], where the intense shockwaves remove also molten and not vaporized sample layers; as a consequence, also the ablation threshold is significantly reduced [72]. Comparison between the ablation rates of aluminium in air and inside water is shown in Fig. 7. In the case of dry ablation, the material removal rate rapidly saturates with the applied radiant exposure. The corresponding pressure wave, measured by a piezoelectric transducer placed below the sample, slowly grows with the laser irradiance. Underwater, both the ablation rate and the peak pressure steadily increase with the laser irradiance in the examined range. Convective bubble motion in liquids prevents the debris re-deposition and makes LA process cleaner than in gas surrounding.

Fig. 6. Position dependent collection of the secondary plasma generated inside the expanded bubble: (a) illustration, nb and nw are bubble and water refractive indices, respectively; (b) photogram of the secondary plasma without and (c) with the attenuating filter [57]. (Reproduced with permission from Ref. [57], copyright 2013, Elsevier).

294


Fig. 7. Ablation of aluminium in air (black) and in water (red) as a function of radiant exposure: left — ablation rate and right — peak pressure [72]. (Reproduced with permission from Ref. [72], copyright 2008, American Institute of Physics).

Material removal during LA in liquids has the following phases: I target evaporation by the laser pulse and by the initially hot plasma; the process typically occurs in time scale up to 100 ns [73] because of the fast plasma quenching inside liquids. IIa explosive evaporation [74,75] and expulsion of molten material by the compression wave moving towards the sample surface [28,76,77]; this phase has a duration of several hundreds of nanoseconds. IIb slow target evaporation through backward reheated target, into the growing bubble.

III during the collapse of the vapor cavity, which might occur hundreds of microseconds after the laser pulse, very high temperatures and pressures are created again; this leads to an additional expulsion of the target material from small and deep pits [78]. The phase IIb from above depends on the experimental conditions and it was not always observed. Recently, this phase of plasma formation has been reported for underwater LA of aluminium target (Fig. 8). The late plasma formation, detected between 1 and 30 μs from the laser pulse, was explained by the backward propagation of the initial

Fig. 8. Plasma after SP ablation of aluminium underwater photographed at different delays from the laser pulse [57]. (Reproduced with permission from Ref. [57], copyright 2013, Elsevier).


hot vapor flow, de-accelerated and rebounded by the surrounding media [79,80]. Such process occurs only in certain experimental conditions, where the initially formed plasma detaches from the target. The backward heat flow raises locally the sample temperature, even above 8000 K [80] and this might induce a new phase of the plasma growth. Variation of the ablation rate of a submerged brass sample was observed when changing the hydrostatic pressure [81]. Here, the ablation was performed by 8 ns long pulses having the energy of 10 mJ, with the corresponding irradiance on the target of about 4 GW/cm2. Increasing the water pressure from 0.1 MPa to 10 MPa the ablation craters enlarge about 16% while their depth reduces about 62%. Further increase of the pressure up to 30 MPa did not produce appreciable variations in the crater's dimensions. The reasons for so important changes in the ablation when raising the pressure from the atmospheric to 10 MPa have not yet been explained. Material ablation inside liquids leads to formation of nano-particles (NPs), as reviewed in [82,83]. During the phases I and IIa, which might partially overlap in the time scale, vapors and fragments of molten material are injected into the surrounding liquid where they solidify in few hundreds of ns. Besides the initial fragment expulsion and late material ejection during the bubble collapse (phase III), the exact processes of NP generation are still under investigation. Vapor plume condenses inside the bubble growing from the ablation site, with typical condensation time of ~100 μs [84,85]. The condensed NPs have smaller diameter compared to those formed by ejection of molten material [86]. Production of NPs depends on the laser parameters and the target material, and on the liquid's properties: chemical structure, polarity, viscosity, and refractive index. The medium properties and its pressure [87] play key roles in the post-ablation processes such as agglomeration and aggregation. A relative amount of NPs after the ablative laser pulse and during the successive cavitation was measured by Soliman et al. [76] by applying shadowgraphy and laser scattering. This work demonstrated that most NPs that formed after the end of the laser pulse remain confined inside the vapor cavity during its first evolution cycle, terminating with collapse. 2.5. Single pulse versus dual or multi pulse laser excitation Emission intensity of the plasma produced inside liquids is generally lower than in gaseous environment due to different factors relative to the plasma formation and de-excitation. Laser plasma initiation in liquids has a reduced efficiency with respect to air surrounding due to light absorption by the medium, radiation scattering on suspended particles and micro-bubbles [51,82], and input energy losses on liquid vaporization. Once the plasma is formed it has a high density, which causes strong shielding of the remnant portion of the laser pulse [41, 88]. The energy shielded is mainly lost by absorption A while reflection R and scatter S on the plasma together amount to only a few percent of the incident laser energy Ein [89]. Fraction of Ein absorbed by the plasma depends on the pulse duration, laser wavelength, focusing conditions, and irradiance with respect to the LIB threshold. For ns pulses at 1064 nm the energy absorbed by plasma might exceed 90% of Ein [89], and this is favorable for the plasma excitation and LIBS analysis of liquid sample. However, such strong light attenuation by plasma reduces the effective irradiance on a submerged target and the corresponding ablation rate, so the LIBS signal from solid samples might be poor also when using high pulse energies. The energy balance during LIB in water was studied by Vogel et al. [89] and according to their measurements, the absorbed laser energy is dissipated into: AEin ¼ Es þ Eb þ Ev þ Er þ Ex

ð6Þ

where Es is the shock wave energy, Eb is the bubble energy, Ev is the evaporation energy, Er is the energy of the plasma radiation and Ex is an unknown, missing part for the full energy balance.

295

Noack and co-workers [90] showed that the laser pulses of 6 ns at 1064 nm and with energy of 10 mJ, focused to achieve irradiance well above the LIB threshold in water, deposit 96.1% of Ein in the breakdown region; the transmitted portion is only 2.6%. The deposited energy is then consumed in the following way: the main loss is due to shock wave emission, with Es = 58.9% of the absorbed energy; bubble cavitation further consumes Eb = 29.4%, while evaporation energy was Ev = 6.5%. Only a very small part of the absorbed energy, about 0.06% is transformed into optical radiation, which detection is the principle of the LIBS technique. The energy partition during LIB depends on the experimental conditions and the examples for ps and ns pulses are shown in Fig. 9 [89]. A general trend is that a decrease of the pulse duration and its energy leads to: i) less efficient plasma absorption and weaker mechanical effects, although some discrepancy might arise for short fs pulses where non-linear effects become important [90]; ii) increased losses in liquid evaporation; and iii) reduced conversion of the laser pulse energy into optical radiation. The initially formed laser plasma in liquids has high electron density, which might exceed 1020 cm−3 [41]. Because of the high electron density the Bremsstrahlung emission, analogue to a black-body radiation, is very intense. The plasma rapidly cools due to large losses into mechanical energy (Es + Eb) and liquid evaporation (Ev); this produces a high rate of electron–ion recombination, which emission is also continuum. The consequences on the LIBS signal are the following: the plasma emission is of short duration — it extinguishes in a few microseconds [20] or less [16,20]; the spectra are dominated by an intense continuum component and after its decay, only the lines from low excitation levels emerge; the spectral lines are strongly broadened due to Stark effect and self-absorption in dense plasma. Poor spectral properties of single pulse (SP) LIBS signal require careful optimization of the acquisition gate and delay, both usually in the order of 100 ns. The analytical sensitivity of SP LIBS measurements in liquids is low, having the typical limits of detection (LODs) in the order of 10–100 ppm, while higher sensitivity was achieved only on alkali elements (Li and Na) [20]. It has been shown that the enlargement of the laser pulse duration from 20 ns to 150 ns at the equivalent energy enhances the LIBS signal intensity and reduces the line broadening. However, this advantage of long ns pulses has not been yet widely exploited due to a lack of the commercial, cost-effective laser sources with a sufficient energy per pulse. DP laser excitation notably enhances the LIBS signal in various experimental set-ups, as summarized in [29]. Inside liquids, the first laser pulse produces a cavitation bubble while the successive, probing pulse excites the plasma inside the already formed gas–vapor environment. The spectra detected after the second laser pulse have a relatively narrow line emission (Fig. 10) due to a reduced plasma collisional broadening and self-absorption in less dense plasma with respect to that formed after the first laser pulse [28,29,91]. Lower density plasma also produces a less intense Bremsstrahlung emission, so the atomic and the ionic spectral lines emerge from the continuum already after few tens of nanoseconds from the laser pulse. The plasma duration is longer than in SP LIBS, which together with early appearance of the spectral lines, allows acquiring the LIBS spectra over larger time intervals. Longer plasma lifetime in DP LIBS with respect to SP excitation indicates that the plasma cooling is slower i.e. the energy losses are reduced. The second laser pulse, comparing to the first one, is more efficiently transformed into optical radiation and the overall plasma emission is much more intense. Here, we report an example of the bubble dynamic measured by laser scattering technique in our experiment described in [58]. Fig. 11 shows the scattered signal after SP and DP induced breakdown in water using 8 ns long pulses at 1064 nm, with energies of 72 mJ and 150 mJ, respectively. The second pulse was delayed for 100 μs from the first one and it coincides with the maximum bubble expansion. Although an interference filter was placed before the detector in order

296


Fig. 9. Energy balance in LIB of water: laser at 1064 nm and focusing angles at 22° and 14° for ns and ps pulses, respectively [89]; for estimating shock wave energy Es two different approaches were applied and the final value considered as the average between the two results. (Reproduced with permission from Ref. [89], copyright 1999, Springer-Verlag).

to transmit only HeNe laser light used for illumination, the arrival of both excitation pulses is well visible in Fig. 11 due to plasma generation, which continuum emission is transmitted by the filter. The second pulse rapidly enlarges the cavity radius, which extrapolated collapse time tc increases from 228.5 μs (SP) to 256.2 μs (DP). Knowing tc from the experimental data, we calculate the maximum bubble radius Rmax by the Rayleigh relation (5). The estimated Rmax is 1.23 mm and 1.38 mm for SP and DP, respectively. Although the second laser pulse has energy twice higher than the first pulse, it enlarges the radius of the already existing bubble for only 12%. In order to evaluate the losses due to cavitation, we calculate the energy Eb of a spherical bubble by formula (4). The estimated bubble energy after the first pulse is 0.77 mJ, and 1.08 mJ after the second one laser pulse. Considering also that the water column had length of 32 mm and so absorbs about 50% of the laser pulse energy, the fraction of the input laser energy transformed to the cavitation is very small (less than 2.2%) with respect to the results reported in [89]. In our experimental conditions the focusing angle was about twice smaller than in [89] and the plasma was formed in multiple sites, which share the energy deposited in the focal region. Anyway, the above estimated values of Eb indicate that the bubble energy after DP excitation is mainly supplied by the first laser pulse (71%), while the difference (29%) is due to the plasma re-excitation by the second, here twice more energetic pulse. Assuming

that the secondary plasma is produced at the center of the macroscopic bubble generated by the first laser pulse, liquid evaporation is delayed with respect to the laser pulse, i.e. it starts when the plasma expands up to the bubble wall. This fact prevents energy losses due to evaporation during the initial stage of the secondary plasma. In the absence of an immediate liquid evaporation by the second pulse, and due to the plasma re-excitation inside vapor environment, less dense than a liquid, the expected shock wave intensity is lower than after the first laser pulse. All these considerations imply that the plasma energy losses into mechanical effects and evaporation after the second pulse are significantly lower than in the case of the SP; this also explains a large LIBS signal increase under DP excitation. DP LIBS signal intensity is a complex function of interpulse delay Δt (Fig. 12) [92], which optimization should be performed considering the period tc of the bubble produced by the first laser pulse. During the bubble expansion after the first laser pulse (area 1 in Fig. 12), DP LIBS signal decreases with Δt due to a growing defocusing power of the bubble itself, which affects both the incident laser and the exiting plasma radiation (Section 2.3). However, the fully expanded bubble (area 2 in Fig. 12) has the maximum reflectivity at the vapor–liquid interface, which is responsible for the formation of a bright central spot through multiple reflections of the internal plasma emission [57]. The radiation from this central spot is not significantly deflected at the bubble wall


297

Bubble radius, SP (mm)

4

3

2

1

2

1

3

0 0

100

200

300

400

500

600

400

500

600

Plasma peak intensity, DP (a.u.)

15

12

9

6

3 78.8 μs

0 0

Fig. 10. LIBS spectra from submerged steel sample, acquired with 15 μs long gate width opened 1 μs before the arrival of the analytical pulse: (a) SP excitation, E = 277 mJ and (b) DP excitation with E1 = 92 mJ, E2 = 171 mJ, and Δt = 30 μs [61]. (Reproduced with permission from Ref. [61], copyright 1997, Elsevier).

and can be captured by the collecting optical system. The redistribution of the plasma radiation by Snell's reflections allows detecting also external plasma layers, otherwise deflected out of the collection system by the LFB. In this way, DP LIBS signal partially recovers for interpulse delays matching the maximum expansion of the previously formed bubble. If the second laser pulse is sent through the collapsing bubble (area 3 in Fig. 12), the plasma emission deteriorates rapidly with the

SP E1=72 mJ DP, E1=72 mJ, E2=150 mJ

Voltage on PMT (mV)

4

3

2

1

0

tc1 0

100

200

tc2 300

400

Time (μs) Fig. 11. Determination of the bubble collapse times tC1 and tC2 after SP and DP excitation respectively, by laser scattering technique described in [58]. (Reprinted with permission from Ref. [58], copyright 2007, Elsevier).

100

200

300

Time Δt from the first pulse (μs) Fig. 12. Dependence of DP LIBS signal on the bubble evolution, for the plasma produced on submerged aluminium target: top — bubble radius after SP ablation, data from [70]; bottom — measured plasma intensity (circles) after the second pulse sent at different delays Δt from the first one [92]; the red line connects the points away from peaks/valleys; timing for the pressure wave transition through focal volume is indicated (crosses). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) (Reproduced with permission from Ref. [70], copyright 2012, American Institute of Physics). (Reprinted with permission from Ref. [92], copyright 2013, Elsevier).

bubble diameter. The reasons for such signal behavior were not yet explained. Previously we saw that inside liquids the largest part of the laser energy deposited inside focal volume transforms into energy of shockwaves. In laboratory conditions the liquid is physically confined and the initial, strong pressure wave generated by the first laser pulse is reflected by the cell's walls and by the sample backplane if dealing with a solid target. The reflected pressure waves transit the focal volume and alter locally the refractive index [57], which plays a very important role both in plasma formation i.e. beam propagation, and in detection of the plasma emission. In Fig. 12, the peaks of DP LIBS signal correspond to the interpulse delays coincident with the transition of the acoustic echoes through the focal volume. Such coincidence itself might increase the DP LIBS signal several times. The secondary plasma expands into the vapor cavity formed by the first laser pulse, and its radiation might be detected for more than 10 μs, which is comparable to the plasma lifetime in atmospheric surrounding. In [92], the exponential decay coefficient of the secondary plasma generated on a solid target was measured for different interpulse delays during the first bubble's cycle. The obtained decay coefficients were 13.8 and 11.3 μs for Δt corresponding to the early growing (20 μs) and well expanded bubble (150–450 μs), respectively. From these results it is clear that the secondary plasma has longer duration if generated inside the compressed, high pressure bubble i.e. soon after its formation or during its late collapsing

298


phase. Inside the high-pressure cavity the plasma thermalizes rapidly, thus producing a more uniform volumetric emission, but the plasma lifetime is longer because: a) small cavity volume hinders the expansion and cooling of the secondary plasma; and b) heat exchange with liquid through a relatively small bubble–liquid interface is much more limited than in the case of the large bubble. For the same reasons the plasma temperature is higher inside a not yet expanded small bubble. De Giacomo et al. [29] reported the measurements of temperature, electron density and line widths for the plasma generated on titanium plate inside water by using DP excitation with different interpulse delays. In the chosen experimental conditions, the bubble generated by the first laser pulse reaches the maximum expansion after about 130 μs. The temperature of the secondary plasma, measured through atomic lines over relatively long acquisition gate (9 μs) shows higher values (T1 = 12,200 K) when delivering the second laser pulse to the dense rather than to the fully expanded bubble (T2 = 10,300 K) (Fig. 13a). Simultaneously measured plasma temperature through ionic lines has similar dependency but with higher values, attributed to recombination well before the end of the acquisition gate. The corresponding measured electron densities are Ne1 ≈ 9 · 10 17 and Ne2 ≈ 1.5 · 10 16 cm − 3 (Fig. 13b). The line narrowing with the bubble expansion is shown in Fig. 13c. The same authors point out that chemical reactions play a very important role in DP LIBS plasmas. Dissociation energy of H2O molecules is much lower than of O2, present in air surrounding. For this reason, there is an abundance of oxygen atoms in underwater plasma and chemical reactions between the analyte (Ti) and the surrounding (vapor) are more important than in air. Underwater, TiO emission was detected already from the plasma beginning while in air its emission appears only for the acquisition delays longer than 10 μs. The emission from CaOH was also detected in the early plasma produced inside water solution containing calcium chloride, by applying both SP and DP excitation [16]. The plasma intensity and the overall quality of the LIBS signal are significantly improved by DP excitation [26,61,77,91,93,94] and the detection limits are for 1–2 orders of magnitude lower than in SP LIBS. The exception regards DP LIBS in high pressure liquids, as it will be discussed next. A further, tenfold enhancement of the LIBS signal intensity was observed when sending a weak pre-pulse shortly before the second, analytical pulse [58]. In such case, the pre-pulse reheats the vapor and increases its pressure, temperature and refractive index. Then the successive analytical pulse, sent before the bubble regains the equilibrium through expansion, has low losses into mechanical energy and it remains well focused inside the cavity and the detection volume. Here, also defocusing of the out-coming plasma emission is significantly reduced and the signal collection efficiency rises dramatically. 3. LIBS measurements inside liquids 3.1. Analysis of bulk liquids LIBS applied on bulk liquids is aimed to detect its impurities, and here we discus different aspects of such analyses. 3.1.1. Plasma stability and effects of analytes LIBS measurements of bulk liquids deal with an extremely unstable signal from one laser shot to another because the breakdown is a probabilistic process, which threshold strongly depends on the impurities present in dissolved state or as colloids or particles [20,95]. The presence of the dissolved gases in liquids is an additional source for the LIB instabilities. The bubbles provoke the beam scattering and might also act as preferable centers for the breakdown formation [96]. Small gas bubbles are always present in natural waters and in flowing circuits. Bubbles are also the final products of the breakdown in liquids. Measurements of the LIB probability in liquids as a function of the laser energy were exploited for determination of concentration and

Fig. 13. Dependence of the plasma temperature (a), electron density (b) and line width (c) in DP LIBS on titanium target in water [29]. (Reproduced from Ref. [29], with permission, copyright 2007, Elsevier).

size of colloids or particles in water [17,20,95,97]. It is important to note that in the fixed experimental conditions the breakdown probability, consequently also the LIBS signal, might vary for one order of magnitude for the same matrix (water) when slightly changing the impurity content. An example of the LIB probability for different kinds of waters, measured by focusing the laser beam (532 nm, 12 ns long pulses, energy 160 mJ) with f = 40 mm lens, is shown in Fig. 14 [97].


299

Fig. 14. LIB probability on different water samples [97]. (Reproduced with permission from Ref. [97], copyright 2001, Elsevier).

The lowest LIB probability occurs in pure water while the highest one was obtained on river water where the largest concentration of impurities is expected (dissolved and organic materials, particles and bubbles). Even if the plasma is generated inside bulk liquids with pulses having the energy well above the threshold for 100% LIB probability, the impurities affect the plasma ignition time [43] and position [17], with consequences on the plasma heating by the remnant portion of the laser pulse and on the final radiation intensity. If applying DP laser excitation, the LIBS signal is even more unstable than for SP measurements because of variability in the position and size of cavitation bubble or bubbles. LIB probability for DP laser excitation (1064 nm, pulses 8 ns long, interpulse delay Δt = 75 μs) as a function of the impurity concentrations and for two different sets of pulse energies is shown in Fig. 15 [98]. By increasing the impurity concentration up to 100 ppm (Mg) in distilled water, the LIB probability inside the focal region, measured over the section of 3 mm diameter, rapidly deteriorates. Simultaneously, the plasma position becomes very unstable, as evident from the increase in the measured standard deviation. Knowing that LIB probability increases with the presence of solutes, the observed reduction of breakdown events inside the detection region is clearly caused by the spatial movement of the plasma towards the focusing lens. 3.1.2. Results of bulk-liquid analysis The first LIBS analysis of bulk liquids, by both SP and DP excitation, had been reported in by Cremers et al. [16]. The emission lines were detected by a photomultiplier (PMT) placed at the exit plane of a high resolution scanning spectrometer. In a water solution containing chloride salts, under SP laser excitation at 1064 nm, the ionic (Ca II) and the atomic (Ca I) emission lines were observed during first 2 μs and 3 μs from the laser pulse, respectively. By using 1 μs large acquisition gate delayed for 0.5 μs with respect to the laser pulse, the obtained LODs for alkali metals were in range of 0.006 mg/L for Li up to 1200 mg/L for Be (see Table 2). Hydrogen emission was not detected while oxygen lines were very weak; OH band head at 306.4 nm was present also in the early plasma stage. Under DP excitation the corresponding atomic lines (O I and H I) became strong and their maximum enhancement was achieved for interpulse delay Δt = 18 μs. In such conditions, the LOD for beryllium was reduced 15 times with respect to the SP measurements. Successively, different papers reported the analytical results obtained by SP or DP LIBS and with quite different outcomes. Here, we compare the SP sampling of water containing CaCl2 [16,21,99,100] or CaCO3 [101]. The achieved LODs for Ca in water vary from 0.13 mg/L

Fig. 15. Number of breakdown events (%) in the detection region as a function of Mg concentration, for two sets of laser pulse energies: (a) E1 = 37 and E2 = 156 mJ and (b) E1 = 91 and E2 = 214 mJ [98]. (Reproduced with permission from Ref. [98], copyright 2005, Elsevier).

to 25 mg/L (Table 2) although in all experiments the laser excitation was at 1064 nm. For the water containing CaCl2, the emission lines from Ca II doublet around 395 nm were always detected together with a number of Ca I lines. Under excitation at 532 nm with low energy pulses (3.4 mJ) the ionic emission was more intense than the atomic one [100], while the opposite effect occurred for more energetic pulses (30 mJ) at 1064 nm. In the case of CaCO3 added to water [101], Ca II emission was not detected at all, perhaps due to a low Ca concentration (10 ppm) with respect to the previously mentioned references. In the same work, only the resonant atomic lines of Al I were detected for the solution containing even 1000 ppm of AlCl3, and this was explained by fast decay of highly excited states with respect to the plasma continuum radiation. Similarly, in [102] H I and Zn I lines originating from highly excited states were not detected by SP LIBS; they were observed only under DP excitation and over a limited range of interpulse delays corresponding to a large, already developed LFB after the first laser pulse. Knopp and co-workers [20] noted that the detection sensitivity was very low for d- and f-transition elements and no emission lines were observed for Hg and Er even at the concentrations of 10 g/L and 2.7 g/L, respectively. Due to fast plasma quenching inside liquids, in SP LIBS the timing for the signal acquisition becomes critical. Lawrence-Snyder et al. [102] measured the impurities in water at pressures up to 200 bar; and reported the optimal acquisition gate delay and width, of only 50 ns and 200 ns, respectively. Similar results regarding the optimization of the acquisition gate are reported in other works, although some emission lines were detectable during more than 2 μs [16,21]. Differently, in [103] the optimized gate delay for SP bulk liquid analyses at pressures up to 30 MPa was of 400 ns, well longer than in the previous studies. Here, the focusing was very tight, obtained by an objective with 10× magnification. The LODs were not determined, but from the reported

300


Table 2 Examples of the detection limits achieved by LIBS inside water with impurities, at atmospheric pressure. Impurities

Laser excitation

Element

LOD

Ref.

Metal salts or boric acid

SP, 1064 nm, 76 mJ, 15 ns, 10 Hz DP, 2° laser: 1064 nm, 125 mJ, 15 ns, 10 Hz

SP, 500 nm, 22 mJ, 28 ns

CaCl2

SP, 1064 nm, 30 mJ, 10 ns, 10 Hz SP, 532 nm, 3.4 mJ, 10 ns, 10 Hz DP, 1064 nm, 95 mJ, 9 ns, 5 Hz 1064 nm, 175 mJ, 7 ns, 5 Hz, Δt = 150 μs

0.006 mg/L 0.014 mg/L 1.2 mg/L 0.2 mg/L 1.0 mg/L 0.8 mg/L 1200/80a mg/L 20 mg/L 0.013 mg/L 0.0075 mg/L 0.13 mg/L 6.8 mg/L 12.5 mg/L 500 mg/L 25 ppm 50 ppm 47 ppb 1 ppm 17 ppm 40 ppm 0.2 mg/Lb

[16]

Metal salts

Li Na K Rb Cs Ca Be Al Li Na Ca Ba Pb Cd Ca Ca Ca Cr Zn Cr Mg Mg Mn Cr

0.034 ppmb 0.39 ppmb 0.92 ppmb

[105]

Standard solutions

Chrome-alum MgSO4 MgSO4, MnCl2•7H2O, or Cr(CH3COOH)3

a b

SP, 1064 nm, 18 mJ, 5 ns, 1 Hz DP, 1064 nm, E1 = 90 mJ, 8 ns, 10 Hz E2 = 214 mJ, Δt = 75 μs Multi-pulse, 1064 nm, 10 Hz 5 pre-pulses Etot = 72 mJ, analytical pulse 144 mJ

[20]

[100] [94]

[19] [96]

SP/DP Data filtering applied [96]

data at pressure of 30 MPa, they are slightly below 10 ppm, 1 ppm and 0.1 ppm for Ca, Mg, K and Li, respectively. The proposed set-up was exploited by the same group, from the University of Tokyo, Japan, for the LIBS device I_SEA (In-situ Seafloor Element Analyzer). This device performs multi-element chemical analysis of both liquids and submerged solids. The instrument is enclosed in a cylinder 1.5 m long with a diameter of 0.3 m, and weighs 110 kg in air. The device has two optical setups, a direct optic for generating the plasma in bulk liquid and a fiber optic setup for probing of solid samples underwater (see Section 3.2.1). The direct optic contains the objective 10 ×, mounted on the hull. The I_SEA instrument was fixed on-board of the ROV Hyper-Dolphin and was successfully tested at depth of 200 m [103]. The elements detected in water were Ca, Mg, K and Li; a C2 Swan band was also observed in the spectra and this was attributed to the suspended organic particulates. Systematic measurements of DP LIBS signal in liquids as a function of Δt for two collinear beams reveal a complex dependence [104] similar to that shown in Fig. 12 and relative to a submerged solid target. With orthogonal DP laser excitation, the optimum inter-pulse delay for analyte detections was much longer than in previously cited work, and corresponds to 150 μs where probably the maximum bubble expansion occurs [94]. The differences in the DP LIBS signal dependence on timing between the two pulses, measured by various research groups, might also be attributed to different detection systems and optical alignments [92]. By using two orthogonal beams [94], the highest DP LIBS signal was achieved with a slight spatial mismatching between the two laser spots; this case probably corresponds to the initiation of the secondary plasma in proximity of the vapor–liquid interface (see Section 3.5). The same authors applied a relatively long acquisition gate delay from the second pulse (1.7 μs) in order to suppress an intense and spectrally wide emission from H lines, otherwise, they were partially masking the analytical lines from Ca, Cr and Zn. So the obtained LOD for Ca in water was 47 ppb (Table 2), much lower than in any other reported case. St-Onge et al. [23] applied a collinear DP LIBS to characterize isotonic solutions directly through transparent, closed bottles. The optimum signal was obtained for a relatively short interpulse delay, namely 30 μs. Here, it was necessary to limit the laser repetition rates to

only 1.25 Hz in order to avoid disturbances by the gas bubbles formed after the previous breakdown events. Sodium concentration in solutions was measured using doublets at 569 nm and 589 nm. The first, weaker and non-resonant transition was much less subjected to selfabsorption and allowed to obtain a linear calibration range up to 0.9% w/v. The resonant transitions around 589 nm, although easily selfabsorbed, were more suitable for the analyte detection at very low concentrations. These lines show also self-reversal for shorter values of Δt, corresponding to a higher gas pressure inside the LFB. Substituting the first laser pulse in DP excitation by a sequence of less intense pulses leads to a manifold enhancement of the LIBS signal [58]. In this way, LOD for Mg in distilled water was reduced from 0.2 ppm [98] to 0.034 ppm [105]. 3.1.3. Effects of liquid pressure SP LIBS signal as a function of liquid pressure was studied in [21,99, 106] with purpose to perform in-situ chemical sensing in deep ocean. It was found that the pressure effect on LIBS signal strongly depends on the laser pulse energy and the selected analytical lines. For low laser pulse energies (up to 30 mJ) the intensities of the atomic lines from Na, Ca and Li were almost independent on the water pressure in the range 1–270 bar while the intensity of Mn I triplet around 403 nm even increases twice with the pressure [106]. The emission lines from Mn I had the maximum intensities at very low laser energy (11 mJ) independently on the pressure (Fig. 16 bottom). The maximum intensity of Na I doublet around 589 nm was always achieved for the pulse energies close to 22 mJ (Fig. 16 top). The LIBS signal reduction with the laser energy, if it exceeds a certain, relatively low level, was attributed to moving of the breakdown out of the focal volume. However, we must consider that the optimizations of the laser pulse energies were not performed on a single solution containing all the abovementioned analytes. Instead, the measurements regard one solution containing, for example, only 100 ppm of Na, or a solution with 5000 ppm of Mn added to the water already containing 2540 ppm of NaCl. In the latter case the impurity content is much higher than for the first solution, thus we might expect remarkably lower laser energy threshold for both the plasma formation and its elongation.


301

Fig. 16. SP LIBS signal from Na (top) and Mn (bottom): left — line peak as a function of laser energy, at two water pressures; right — spectral lines at pressure of 276 bar and for different laser energies [106]. (Reproduced with permission from Ref. [106], copyright 2007, The Optical Society).

The plasma lifetime does not seem significantly affected by the pressure in the range 1–270 bar [21]. Regarding the line broadening, for the Li I emission acquired at delay of 600 ns the line width increases with pressure about 20%; the corresponding increase of the line width, measured with the acquisition delay of 2000 ns, is about 5 times [21]. From here, it seems that the density of early plasma does not depend importantly on hydrostatic pressure. The plasma expansion is clearly suppressed at higher pressures, leading to larger widths of the emission lines due to Stark broadening. Efficient plasma confinement by high pressure media also increases self-absorption i.e. causes an additional line broadening. Changes of the water temperature between 27 °C and 99 °C did not produce noticeable effects on the SP LIBS signal [106]. Recently, SP LIBS applied for element detection in seawater was tested for the pressures up to 40 MPa, equivalent to that in the ocean's depth of 4000 m [22]. These results add some more details with respect to [21] and lead to the identical conclusion: the early plasma stage (first 100 ns) is independent on the pressure and its effects on intensity and width of the spectral lines occur only at longer delays from the laser pulse. Here, the optimized experimental parameters correspond to the laser pulse energy of 40 mJ. Atomic K line at 766 nm was detectable up to 1800 ns from the laser pulse at pressure of 20 MPa; below or above this pressure the same line had shorter lifetime. For example, at 40 MPa the measured lifetime of the same line was of 700 ns, compared to 1100 ns at atmospheric pressure. Fig. 17 shows the intensities of different analytical lines as a function of the pressure, measured by using the acquisition delay and gate of 100 ns and 5 μs, respectively. For all the examined transitions, the maximum LIBS signal was obtained around 20 MPa. The chosen Ca I and Mg I lines did not show important intensity variations with pressure while K I peak rapidly decays with further pressure growth. The anomalous behavior of K I line was attributed to the low upper energy level (1.62 eV) and the low ionization energy of the element (4.34 eV). The examined ionic line Ca II grows in intensity with increasing the pressure, where probably hotter and denser plasma exists in the late stage. The authors explain the pressure effects in the following way: i) the early plasma has always high pressure, in the order of 1000 MPa [107]; consequently, the much lower

pressure of the surrounding liquid has no important influence on the plasma itself; ii) when the plasma pressure decreases down to that of the surrounding, the external pressure compresses back the plasma, causing an increase of its temperature and electron density; and iii) the plasma gains energy by compression, more efficient in high pressure environment, so the liquid pressure might keep high or even increase the plasma emission intensity. Thornton et al. performed comparative SP LIBS measurements with laser pulses of 25 mJ and duration of 20 ns and 150 ns, respectively; the pressure of the water solutions was varied between 0.1 MPa and 30 MPa [64]. The optimized acquisition gate delay from the laser pulse was 400 ns, and the integration time was 500 ns. The excitation by the longer pulse enhanced the line emission intensities for factor 2.2× (Ca I line at 423 nm) or 1.6× (K I line at 766 nm) with respect to the short

Fig. 17. Pressure dependence of spectral line intensities from different elements in seawater; SP laser excitation with energy of 40 mJ and pulse duration of 10 ns [22]. (Reproduced with permission from Ref. [22], copyright 2014, Royal Society of Chemistry).

302


pulse. By using the long pulses, the line widths increase slightly with the hydrostatic pressure. In the case of the short pulses, the line broadening at 30 MPa was much more severe because the plasma and the bubble expansion are less efficient than for the long laser pulses. Increase of the static liquid pressures reduces the radius and the lifetime of LFB [60,105], which minimum inner pressure becomes higher. As a result, the optimum Δt in DP laser excitation shifts towards shorter intervals (Fig. 18). Such conditions approach to those of the SP laser excitation also in terms of the plasma density, i.e. the line broadening and the continuum spectral intensity. For these reasons, DP LIBS loses efficiency at high liquid pressures. The signal enhancement by DP excitation was almost absent above the static pressure of 100 bar, where the bubble was generated by a laser pulse with energy of 30 mJ [102]. The author suggested that the pressure range for the LIBS signal enhancement by DP excitation might be enlarged by increasing the energy of the first pulse or by applying a multi-pulse laser excitation, both aimed to produce a larger vapor bubble.

water salinity and the LODs based on the analytical ionic lines are lower in the absence of the salt [105]. Addition of any analyte (impurity) lowers the LIB threshold, particularly for long IR pulses. This also alters the plasma dimensions and shape, and the successive cavitation. Spatial extension of the plasma towards the focusing lens is one of the reasons for a fast LIBS signal saturation in liquids with the analyte concentration. For example, in [19] the LIBS signal under SP excitation rapidly saturates when adding Cr in concentrations already above 100 ppm. Jiang-lai et al. [100] observed the saturation of Ca line for the element concentrations above 200 ppm. Suspended particles decrease the breakdown threshold [17] and also alter the plasma shape and its parameters. From the other side, the same particles scatter the laser light along the beam path, and due to a simultaneous lowering of the breakdown threshold, there is some intermediate particle concentration in liquids for which the LIBS signal is maximal [20]. For example, here measured LOD was 1000 times lower for particle born Ba (ErBaCu3Ox) than for the dissolved salt (BaCl2). 3.2. Direct analysis of submerged solid targets

3.1.4. Matrix effect The matrix effect in LIBS analysis of bulk liquids is much stronger than for a sampling in gas surrounding. Cremers et al. [16] found that the Li line intensity at equivalent concentrations was 2–4 times higher in organic solutions (methanol, acetone, and ethanol) than in water. This difference was attributed to the lower values of boiling point, heat capacity and heat of vaporization of the organic liquids, leading to a more efficient atomic excitation. There are also differences in shockwave and bubble evolution between the liquids [51], which consume the input laser energy and influence the plasma excitation and successive cooling. Under SP laser excitation at 1064 nm, a steady decrease of the line intensity ratio Ca II/Ca I was observed when adding NaCl to the aqueous solution in concentrations above 10 mg/L [16]. Another research group detected more intense atomic (Ca and Li) and continuum emissions in artificial seawater than in pure water, both of them pressurized to 30 MPa [103]. They performed the calibration for Li and Ca only in seawater, and concluded that the matrix effects are significant. More systematic studies of effects of water salinity on DP LIBS signal were performed in [105] where pure water, seawater and their volumetric mixture 70%–30% were considered. The most intense continuum and ionic emissions were obtained on the solutions prepared from pure water, while the atomic line intensities were enhanced by the presence of NaCl salt. Similar results under DP excitation were obtained in [106], where addition of 254 ppm of NaCl to distilled water increased SNR of Ca I line from 22 to 30. The presence of Na in liquid, which ionization energy is low, reduces the LIB threshold and increases the plasma length, as well as the plasma electron density. Higher electron density leads to more efficient recombination, thus to a higher atomic-to-ionic ratios in the plasma. For these reasons, the calibration curves change with

Fig. 18. Intensity of Zn I line at 481.05 nm in DP excitation as a function of interpulse delay at different water pressures [102]. (Reproduced with permission from Ref. [102], copyright 2007, Society for Applies Spectroscopy).

Laser ablation of materials covered by liquids is much more efficient than the ablation in gas surrounding (Section 2.4). This increase of the ablation rate does not correspond to a more intense LIBS signal inside liquids because of more efficient plasma confinement, cooling and chemical reactions compared to the gas environment. The particles expulsed from the melted sample layer shortly after the laser pulse [57,91] are analogue to hot black-body radiators and they also contribute to an intense continuum emission [108]. The particles ejected by ablation or those formed later by clustering [109,110] tend to accumulate close to the sample surface. The presence of the particles generated by laser ablation is responsible for the radiation loss of the successive pulses through scattering, and might induce also the plasma formation away from the target. 3.2.1. Single pulse LIBS on submerged solids Different published results related to SP LIBS measurements on submerged solid samples [91,61,29,111,112] state a very short plasma duration, in the order of 100 ns. Similarly to the case of the breakdown in liquids, the plasma emission is dominated by the continuum component. The detected element transitions originate only from low excited states and the lines are very broad due to high plasma electron density and self-absorption. The spectra obtained by SP LIBS might allow to recognize the main sample constituent, for example, to identify a type of the metal. Also for such identification, a gated detector is necessary to separate the signal from the continuum emission. For example, Suzuki et al. [112] acquired the SP LIBS spectra from Ti, Al and Pt target inside water by using a not gated CCD. In all three cases the spectral lines were not distinguished. The observed differences among the spectra regard only the intensity and the spectral distribution of continuum radiation, which is associated with temperature of the plasma as a blackbody radiator. Usually, the LIBS analyses involve the laser pulses of duration below 10 ns. Some recent experiments [108,113,114] demonstrated that enlargement of the laser pulse duration significantly improves the quality of SP LIBS spectra also from the solid targets (Fig. 19). The ablation rate is lower for longer pulses [113] because, before reaching the sample, their later part is efficiently absorbed by the plume. The beam absorption by the plasma causes an additional plasma excitation and expansion, which lead to an enhancement of the emission lines and to lowering of the continuum level, respectively. In this way, well resolved analytical lines (Fig. 19) were obtained by ablating copper with long pulses having the energy of only 1.7 mJ [114]. The same research group performed the spatially resolved LIBS signal detection after applying 100 ns long pulses [115]. When shifting the collection point away from the hot plasma center, the emission lines from the solid target were narrower. By collecting the signal close to the plume edge, they


303

Fig. 19. SP LIBS spectra from copper target in water, obtained by pulses with energy 1.7 mJ having different durations; the acquisition gate delay and width are 1 μs and 5 μs, respectively [114]. (Reprinted with permission from Ref. [114], copyright 2006, American Institute of Physics).

obtained the well resolved analytical lines also when applying long integration times, characteristic for not gated detectors. The same research group tested 150 ns long pulses with energy of 5 mJ, for the LIBS measurements on solid targets inside water pressurized up to 30 MPa [116]. The emission lines from copper target, detected by using the gate delay of 600 ns and the gate width of 500 ns, were well resolved. The line widths of the triplet around 515 nm increase about 20% when raising the hydrostatic pressure from 0.1 MPa to 30 MPa. Dynamic of the plasma produced by SP ablation inside liquids still needs the systematic studies. Contrary to some of the previously mentioned results, in different works [57,78,79,108,117,118] the plasma emission was detectable well beyond the first few hundreds of nanoseconds from the laser pulse. In [108], the plasma produced by applying 16 mJ pulses (1064 nm) on copper plate in water was clearly observed at delays of 1000 ns and 2000 ns for pulse durations of 20 ns and 150 ns, respectively. During underwater ablation of graphite [117] the plasma emission was still detectable after 1 μs from the laser pulse. Here, the optical density of C2 Swan band head (0,0) was also estimated from the LIBS spectra: after 1200 ns from the laser pulse, its emission was only 20% lower than the maximum value occurring at delay of 200 ns. Recently, it has been demonstrated that underwater LA of aluminium target by SP might lead to a late plasma formation, which emission was detectable after even 30 μs from the laser pulse (Section 2.4). Another paper reports well resolved emission lines obtained on a zinc plate in water after SP ablation with pulses of 10 ns length [118] and at liquid pressures between 0.1 MPa and 30 MPa. Inside this pressure

range, the maximum line emission intensities were detected after 400 ns from the laser pulse. The optimized acquisition gate starts at this delay, and so obtained LIBS spectra were of good quality (see Fig. 20 — left). The experimental data show that at pressure of 30 MPa the emission lines are more intense and of longer lifetime than at 0.1 MPa. In both cases, the spectral lines were detected also at delays longer than 1000 ns with respect to the laser pulse (Fig. 20 — right). Here measured line widths do not change significantly with the pressure. 3.2.2. Dual pulse LIBS on submerged solids Application of DP excitation on submerged solid targets might lead to the signal enhancement even for two orders of magnitude with respect to SP [29,61,111]. So significant LIBS signal enhancement is linked not only to a larger ablated mass, but mainly also to the presence of the vapor cavity, created by the first laser pulse (Section 2.5). For example, the measured increase of the ablated mass underwater by DP compared to SP was about 4 and 2 times for aluminium and brass sample, respectively [61]. This enhancement in the ablation is much lower than the observed LIBS signal increase under DP excitation. Here, the emission lines from Fe after SP ablation of steel sample were not even observed, whereas different, relatively intense Fe lines were detected under DP excitation. The influence of the nanoparticles trapped inside the bubble on the emission intensity of the secondary plasma has not yet been sufficiently investigated. From one side, NPs inside the vapor cavity scatter the incoming second beam and thus reduce the ablation rate. On the other hand, NPs have much lower melting temperature than the

Fig. 20. Plasma emission after SP ablation of zinc in water: left — the spectrum inside pressurized liquid, acquired with delay of 400 ns from the laser pulse (duration b 10 ns); right — temporal behavior of the emission underwater at two liquid pressures [118]. (Reproduced with permission from Ref. [102], copyright 2007, Society for Applies Spectroscopy).

304


corresponding bulk material [119]. Small particles can be fully evaporated inside the secondary plasma [120], and contribute to an increase of DP LIBS signal. The main role of the first pulse in LIBS is to create a vapor bubble above the sample surface, which gaseous environment is favorable for the plasma properties after the second, delayed laser pulse, as demonstrated in [121]. Here, the sample submerged in water was an aluminium plate and the first laser pulse had energy of only 0.2–0.4 mJ. This energy was not sufficient to produce a detectable plasma emission but it generated a vapor bubble with the maximum radius around 700 μm, reached after about 30 μs from the laser pulse. The second laser pulse with energy of only 1 mJ and delayed for 15–50 μs from the first pulse, produced well resolved Al I doublet around 395 nm even in the spectra acquired without gating. For the interpulse delays corresponding to the still growing or lately shrinking bubble, the emission lines were strongly enlarged or even self-absorbed. Increasing the second pulse energy up to 10 mJ, the spectral lines collected by a not gated system exhibited self-reversal for all interpulse delays, probably due to very dense and intense secondary plasma in its early stage. The important outcome of this work is that DP LIBS performed with very low laser energies is suitable for recognition of the main target constituents also by using a non-gated detector. The same paper [121] explores another interesting issue, namely the importance of the superposition between two collinear beams in DP LIBS applied on submerged targets. Reduction of DP LIBS signal intensity was observed when shifting the second beam from the first one about 60 μm. This shift is comparable to the laser spot diameter on the target. Further displacement of the second beam, for 90 μm from the first one, causes the disappearance of the LIBS signal after the second laser pulse. Here, the systematic measurements were not performed and the paper does not provide the exact experimental parameters, such as the pulse energies applied, the interpulse delay, the exact bubble shape and dimensions, the distance between the optical elements and the laser beam properties. Based on these preliminary results, the authors conclude that the second laser beam must overlap with the first one, and that the other methods for generating the bubble before the analytical laser pulse (ultrasonic radiation or boiling by the target heating) cannot produce the LIBS signal with quality analogue to DP excitation with two well aligned beams. However, we believe that before reaching such conclusions, a systematic experimental work must be performed first. For example, we attempted to simulate by ZEMAX (ZEMAX LLC, Redmond, WA, USA) the optical system based on the data given in [121]. Assuming that a perfectly hemispherical vapor bubble was generated by the first laser shot, and that the bubble is sufficiently expanded to have refractive index 1.0, we obtained the results shown in Fig. 21. Here, we considered that the beam is displaced from the bubble center only along one axis, and that the beam diameter incident on the bubble is about 11% of the bubble radius R. The beam shift d from the bubble center for d = 0.5R (Fig. 21b) causes a slight laser spot deformation on the sample, with the equivalent diameter D = 1.02D0, where D0 is the beam diameter on the target when d = 0. For the shift d = 0.65R the beam partially hits the target out of the bubble (Fig. 21c). This external part of the beam, if it creates the plasma on the target, behaves as a defocused SP. Simultaneously, the overall spot on the sample is visibly deformed, with the equivalent diameter D = 1.27D0. The corresponding average fluence is reduced about 38% with respect to the centered beam (d = 0), and this could be critical when the second laser pulse operates close to the ablation threshold. When the second beam is shifted about 0.75R, total reflections occur for the external rays while those closer to the bubble center are strongly deviated and they reach the sample completely out of the bubble. By careful observation of Fig. 21c–d it is evident that the maximum laser flux is achieved already at the bubble–water interface, after which the beam diverges. This means that in DP LIBS of bulk liquids by two shifted laser beams, the second pulse will eventually produce the breakdown at the bubble interface where its irradiance is maximal. If the secondary plasma initiates

Fig. 21. Ray tracing through a hemi-spherical vapor bubble with radius R and refractive index 1.0 inside water, and for different beam displacements d from the bubble center: (a) d = 0; (b) d = 0.5R; (c) d = 0.65R; and (d) d = 0.76R.

closer to the bubble interface, the losses in liquid evaporation are higher. All the mentioned effects might explain the missing plasma formation by the second laser beam if decentered from the first one and when operating with very low pulse energies. From the present discussion, we might also conclude that larger bubbles with respect to the incident beam diameter are less sensitive to misalignment between the two beams. One interesting application of LIBS in liquids regards the recognition of materials in seawater, like in the case of underwater archeology [25, 26]. It was observed that the spectra generated by DP on the targets placed inside seawater had strongly absorbed emission in proximity of the resonant Na lines around 589 nm [111]. This effect was explained by a distributed presence of sodium atoms inside the vapor bubble, which excitation by the second laser pulse is delayed because the plasma evolves starting from the target. Fortunately, the resonant Na atomic transitions are spectrally distant from the most important analytical lines from other elements and they do not compromise the material identification. By applying DP LIBS on targets inside seawater, the characteristic spectral features were observed from different metallic samples of archeological interest, such as iron, bronze, gold or silver alloy [26]. It was also possible to distinguish marble from not precious calcareous rock due to much higher content of Si, Fe and Mn in the last sample type (Fig. 22). In the same experiment, the plasma emission from a submerged wood was not observed at all; this was attributed to the swelling of cellulose fibers after absorbing water, with the consequent


305

Fig. 22. Comparison of underwater DP-LIBS spectra from marble (a) and calcareous rock (b), used for identification of C, Si, Fe and Mn recorded under the same experimental conditions [26]. (Reproduced with permission from Ref. [26], copyright 2005, Elsevier).

increase of the ablation threshold. On the same dry wood in air, the LIBS spectra were intense and revealed also the trace elements. Presently, there are very few published results about LODs from submerged solid targets and they regard only DP excitation, as summarized in Table 3. The sensitivity of LIBS on solid targets inside water is rather low, in the order of 100–1000 ppm. The data presented in [111,122] indicate that a better plasma stoichiometry from copper alloys, known for a strong fractionation, was achieved in DP underwater analysis than in air; this was evident only when using a short acquisition gate i.e. during the first 400 ns from the second laser pulse. The authors hypothesized that this improvement in accuracy is due to more homogeneous plasma inside the bubble than in air and to the conditions closer to LTE. For longer signal acquisition delays, changes in the ratios of the line intensities between Cu, Pb and Sn occur; this was attributed to different reactivities of these species with the confined water vapor [122]. In [26] the quantitative analysis of binary copper alloy Cu–Sn produced an excellent agreement with the results obtained by SEM–EDX (scanning electron microscopy with energy dispersive X-ray spectroscopy). However, on the alloy Cu–Pb–Zn, the LIBS measured values were not satisfactory although the plasma emission was acquired shortly after the second laser pulse, i.e. where a better stoichiometry was hypothesized. 3.3. Analysis of submerged targets under gas flow Introduction of a pressurized gas flow above a submerged solid target removes locally the liquid and provides the gaseous environment without need to create first the vapor bubble, like in DP experiments

which require more complex, bulky and expensive systems. The application of the gas flow has further advantages: the effective laser energy available for ablation and plasma excitation is higher because the input energy is not lost in liquid evaporation and it is not absorbed and scattered by the liquid and the floating particles. In this way, it is possible to operate also in turbid waters, for example, close to the sediment sea bottom. The analytical performances are equivalent to those in gas surrounding. The strict requirement for this technique is that the target to analyze is solid, not releasing the particles. In the case of target coverage by soft materials, like sediments or algae, an intense gas flux might provide the target cleaning prior to the laser sampling. The first arrangement for underwater LIBS measurements by applying a gas flow was reported by Beddows et al. [123]. The laser radiation was delivered to the solid sample through an optical fiber of 550 μm diameter and 20 m length, surrounded by a tube for pressurized gas. The same fiber was exploited for transportation of the LIBS signal backward to the spectrometer. The main limitations of this set-up are related to the damage threshold of the fiber where the maximum launch energy of nanosecond pulses was below 50 mJ, and to the optical attenuation of the fiber, relative to both the laser and the plasma radiation. The laboratory tests inside water at atmospheric pressure were performed with flows of air, nitrogen and argon. In the last case, the LIBS signal was slightly higher than for the previous two gases. The optimal gas pressure was of about 2 bar, corresponding to the pressure difference between the gas and the water ΔP = 1 bar. On steel samples the obtained LODs for Cr, Mn and Si were of 310, 325 and 455 ppm respectively.

Table 3 Examples of the detection limits achieved by DP LIBS on solid targets underwater. Sample

Laser excitation

Element

LOD

Ref.

Steel

DP, 1064 nm, 8–10 ns, 10 Hz, Δt = 30 μs E1 = 199 mJ, E2 = 146 mJ* E1 = 82 mJ, E2 = 159 mJ** DP, 532 nm, 7 ns, 10 Hz, Δt = 120 μs, E1 = 10 mJ, E2 = 75 mJ

Marble

DP, 1064 nm, 8 ns, 10 Hz, Δt = 55 μs, E1 = 82 mJ, E2 = 180 mJ

520 ppm 367 ppm 1200 ppm 1190 ppm 0.03%w 0.08%w b486 ppm b0.72%

[61]

Copper alloy

Cu* Cr* Mn** Si** Pb Sn Fe Si

[122] [26]

306


The first in-situ undersea LIBS trials on solid samples are described in [25]. The instrument is composed of the main unit for the signal generation and detection, and of an underwater optical probe connected to the main unit by a 40 m long umbilical. The latter contains a quartz optical fiber for transmission of laser and plasma radiation, and a channel for the pressurized air flow. The spectra were recorded at the wavelengths above 350 nm due to attenuation in UV by the long optical fiber. The initial, laboratory measurements were successfully performed on a set of different archeological materials such as pottery (Fig. 23 — right), bronze, precious metals, iron, and bones. During the campaign, the instrument was connected to an autonomous power generator and operated from a vessel while the probe handling was provided by a professional diver (Fig. 23 — left). The system was tested in-situ up to the depth of 30 m by applying the air flow at 5 bar. The measured LIBS signal decreases with depth and this was explained by more dense initial plasma, which reduces transmission of the laser pulse tail to the sample. However, the spectra obtained at the maximum depth still had a very good quality and allowed to recognize the sample composition and also to distinguish different types of copper alloys. The signal was not importantly affected by changing the angle between samples and the probe in the range 0–40°. This fact facilitates the positioning also in the conditions of strong underwater currents. The same probe could be also operated by a remote operated vehicle (ROV) equipped with a robotic arm. 3.4. Analysis of submerged soft materials In-situ elemental characterization of the sediment surface by LIBS might provide important environmental and geological data, as well as information about recent biological activity in waters [124]. The feasibility of performing semi-quantitative LIBS analyses on sediments directly underwater by using non-gated detectors was assessed in [27]. Sediments are soft materials, and underwater they naturally generate suspended particles above their surface, particularly in the presence of strong currents close to the sea bottom. The concentration of these particles increases after the laser induced plasma formation due to the intrinsic shock waves, which blast away the material from top layers. The sediment particles act both as centers for micro-plasma formation and as the radiation scatterers. Removal of the suspended material by flushing water above the sample surface is not a viable solution because the flux would provoke an additional water turbidity by removing particles from the sediment top. For similar reasons, it would not be useful to flush a gas above the examined surface with the aim to create sample/ air interface. An additional difficulty in LIBS measurements on the underwater sediments is caused by the sample's roughness, which is naturally present and also induced by the laser pulses. The sample's surface

roughness compromises the precise laser beam focusing, particularly when using short focusing lengths, necessary to minimize the water absorption along the optical path. All the mentioned effect lead to severe fluctuations in the breakdown position, the plasma emission intensity (Fig. 24) and the plume temperature from one laser shot to another. The optimized conditions for underwater sediment analysis in [27] correspond to DP laser excitation at 1064 nm, with pulse energies of E1 = 60 mJ and E2 = 240 mJ, respectively; the pulses were separated by Δt = 75 μs. The beam focusing system has the equivalent focal length of 35 mm and it was placed angularly with respect to the sample in order to prevent deposition of the gas bubbles, formed by the breakdown, on the lens–water interface. The laser repetition rate was of only 0.5 Hz because otherwise, the LIBS signal rapidly decays due to excessive particle accumulation above the sediment surface. The plasma formation on the sediment surface produces a relatively intense signal. With the high resolution (HR) spectrometer it was possible to detect, also by applying only one DP sequence (Fig. 24a), some elements in concentrations of 60 ppm (Table 4). If the plasma formation was induced slightly before the sample surface, the plasma continuum component dominated the spectra (Fig. 24b). In the case of the breakdown formation on the sediment particles, dispersed above the sample surface, the spectra had a very intense continuum emission (Fig. 24c), occasionally caused also by sonoluminescence produced by the collapsing vapor bubble. The corresponding emission lines were very weak or even absent. Due to strong shot-to-shot fluctuations in the plasma temperature, the quantitative analyses were obtained only after applying an appropriate data processing procedure. The latter selects automatically only the spectra characterized by similar plasma parameters, here related to the continuum spectral distribution. For example, the plasma temperature estimated from Planck's law for blackbody radiator indicates the fluctuations between 9000 K and 11,000 K, approximately. Such large temperature variations lead to changes of the here considered line intensity ratios up to a 3.5 times, and do not allow performing the calibration. However, selecting only the spectra with similar plasma temperatures, made it possible to obtain well correlated calibration graphs. These results indicate that the direct quantitative LIBS analyses of submerged sediments or other soft materials underwater are feasible. The procedure for the spectral selection requires separate data registering after each laser shot, which could improve significantly the measurement sensitivity and accuracy [98]. 3.5. Analytical aspects Depending on the liquid type and impurities, there is optimal pulse energy for the LIBS signal generated inside a bulk liquid. Increasing

Fig. 23. In-situ trials on the Mediterranean Sea: left — the diver working with the probe; right — LIBS spectra from an archeological pottery obtained in laboratory conditions [25]. (Reproduced with permission from Ref. [25], copyright 2012, Elsevier).


307

Fig. 24. Single shot LIBS spectra from natural sediment (left) and illustrated plasma position (right): (a) optimal for the analysis; in the presence of weak (b) and intense (c) breakdown above the surface; the emission lines belong to Ti+ [27].

the laser energy over this value leads to the signal deterioration. In Fig. 25 we show an example of the spectrally integrated plasma intensity along the axis of a tightly focused (NA = 0.6) laser beam in water, for different laser energies [52]. The maximum local plasma intensity point corresponds to the laser energy of only 0.65 mJ. Doubling of the laser pulse energy leads to the plasma elongation and to multiple spatial peaks, each with much weaker intensity than at the optimal energy. Not localized, weaker plasma means also that its excitation, favorable for the SP LIBS signal detection, is less efficient. Simultaneously, the maximum diameter of the vapor bubble/bubbles growing from the breakdown site/sites and the intensity of the secondary plasma in DP excitation, are reduced (Fig. 26) [58]. In summary, the highest LIBS signal from bulk liquids, both by applying SP and DP excitation, corresponds to the laser pulse energies just below the threshold for the plasma formation in multiple sites. This threshold irradiance decreases with the impurity content in liquid and causes an additional, strong matrix effect with the following implications:

— in order to achieve the minimum LOD for one analyte, the excitation should be optimized for the solution containing the lowest analyte concentration — for the fixed laser excitation, increase of the analyte concentration or of any other impurity, leads to a rapid saturation of the calibration curves — the calibration curves are strongly dependent on other impurities and cannot be generated for a universal case, even in a single liquid type.

— the optimal experimental parameters and the LIBS signal intensity depend on the impurity level

Table 4 The minimum certified concentrations of minor/trace elements, detected on reference samples, by non-gated HR system, in a single-shot acquisition; the effective LOD is lower than the values reported here. Element

Min. concentration (ppm)

Sample

Ba Cu Mn Ti

59 114 375 56

ES6, loess (France) NIST2710, contaminated soil NBS1646, Estuarine sediment ES1 — marine deposit (Italy)

Fig. 25. Plasma intensity distribution along the beam path in water for different laser pulse energies; the distance was measured from the beam waist [52]. (Reprinted with permission from Ref. [52], copyright 2010, American Institute of Physics).

308


Fig. 26. Images of the vapor bubble and the plasma after the second laser pulse (DP excitation), taken by the ICCD with integration time 10 μs; the energy of the first laser pulse is progressively increased from (a) to (d) [58]. (Reproduced with permission from Ref. [58], copyright 2005, Elsevier).

Formation of the absorbing plasma along the beam path reduces not only the intensity of the secondary plume, but also the number of breakdown events inside the focal volume. Differently from the LIBS analysis in gas surrounding [125], the correlation between detected plasma continuum level and intensities of the analyte emission lines is poor in liquids [98]. Consequently, the signal normalization on the continuum emission, in both SP and DP measurements, is not adequate. Much better reproducibility of the LIBS signal might be achieved by separate acquisition of the LIBS spectra after each laser shot instead of their accumulation, followed by elimination of anomalous, low intensity spectra from the data set and successive averaging of the filtered data. The optimal limit for the filtering of the spectral intensities depends on the experimental conditions and on

the sample solution, and might lead to a lowering of the LODs even for a factor 7 (Fig. 27). The analogue data filtering improves the detection limits in LIBS analysis of submerged solid samples. Here, the accumulation of the ablated or newly formed particles along the beam path causes strong fluctuation of the plasma intensity from one shot to another. For these reasons, in laboratory conditions the liquid should be frequently exchanged or the fresh liquid should be flushed through the cell. When sampling at a fixed point it is necessary to avoid the signal degradation due to development of a deep crater. This problem is much more severe inside liquids than in gas environment, as in the first case the ablation rate is significantly higher. Furthermore, the sampling inside liquids requires use of relatively short focal lengths in order to reduce the light attenuation by the liquid itself. Such a condition makes the sampling more sensitive to the target roughness. The particles generated by previous laser shots disturb the formation of the plasma and the successive bubble growth, which properties are important for DP LIBS inside liquids. During ablation of aluminium in water with laser pulses at 1064 nm and duration of 20 ns, the relative standard deviation (RSD) of the bubble radius was as high as 30% and 10% for sampling at the fixed point and on moving target, respectively [110]. In the corresponding conditions, the bubble radius reduces for 10% already after 17 laser shots at the fixed point, or after 80 pulses when moving the target. In the latter case the bubble size reduction is exclusively caused by accumulation of the particles in the surrounding liquid and not by the crater development. The same work also reports an occasional formation of a secondary bubble on top of the cavity generated by the first laser pulse. Such effect was explained by initiation of the secondary plasma on nanoparticles, which tend to accumulate in the proximity of the bubble wall. We add here also another consideration: if using long focal lengths to concentrate the laser beam, like 200 mm in [110], the beam diverges after hitting the spherical bubble wall. In the specific case, the estimated spot diameter is about 12% larger on the target than on the bubble wall. This means that the secondary plasma has less probability to be initiated on the target and more chances to be triggered on nanoparticles close to the bubble wall.

Fig. 27. Comparison of the spectra obtained by DP LIBS in distilled water containing Mg concentration of 5 mg/L by: (a) summing the spectra over 1000 laser shots and (b) summing only the spectra with the Mg peak above 50% of maximum peak among the data assembly [98]. (Reproduced with permission from Ref. [98], copyright 2007, Elsevier).


In following, we resume some important considerations for LIBS measurements inside liquids: a) plasma generation — The beam path through liquid should be as short as possible, thus to minimize the light absorption and scattering — The use of large numerical apertures of the focusing optics is preferable, in order to obtain well localized plasma. In the case of DP LIBS, this prevents that the beam diverges inside the vapor cavity formed by the first laser pulse — The pulse energy should be below the threshold for the plasma elongation and successive cavitation in multiple sites; this threshold lowers with the impurity content in liquids, both in form of solutes and particles — When sampling solid targets, the reproducibility of the LIBS signal and the cavitation is much better if moving the target or the probe — Liquid flushing above a submerged target or frequent liquid exchange in laboratory conditions, reduces the accumulation of the particles — The bubbles formed by the previous breakdown events tend to move vertically and attach to solid surfaces; their accumulation on optical elements should be avoided — Longer nanosecond pulses produce more intense and longer plasma emission on solids. — In SP LIBS on solids, better signal is obtained when detecting the late formed plasma, if present — In DP LIBS, the second beam should be centered on the previously formed bubble in order to minimize defocusing by the bubble itself and total reflections at the vapor–liquid interface — The most intense DP LIBS signal correspond to interpulse delays well shorter than the bubble expansion time; the transition of the reflected pressure waves, generated after the breakdown induced by the first laser pulse, might further enhance the signal intensity — If high spectral resolution is required for detection of some elements, the optimum interpulse delay corresponds to the well expanded bubble b) data acquisition and processing — The optical collecting system, besides a short path inside liquids, should have a large NA, particularly when detecting the secondary plasma inside the vapor bubble, which behaves as a powerful negative lens — The plasma intensity, position and its parameters have strong shot-to-shot fluctuations inside liquids. The sensitivity of the measurements increases if registering separate spectra after each laser shot, and selecting and summing only the spectra with the wanted characteristics — In the presence of strong signal fluctuations, selection of the spectra corresponding to similar plasma parameters is necessary for obtaining quantitative results from not very large data sets — Before calculating the plasma parameters, it is necessary to correct the spectra for the absorption inside the liquid — For DP LIBS, the complete procedure for the spectral correction would require the simulations of the losses due to the wavelength dependent, light deflection by the bubble out of the collecting system. The contribution of the Snell's reflection on the collection efficiency is more difficult to evaluate. 4. Conclusions In this paper we discussed various processes involved in LIBS spectroscopy inside liquids, such as: light transmission in water — the most common media, plasma formation in single or multiple sites, shockwave emission, successive vapor cavitation and collapse, enhanced ablation in liquids and nanoparticle formation, as well as the beam coupling through the vapor bubble, like in the case of dual pulse excitation. Regarding DP

309

LIBS in liquids, reviewed in 2007 [29], the successive research works introduced new, important findings, some of them contrary to the previous statements. The most intense DP LIBS signal does not occur if sending the second pulse when the vapor bubble created by the first pulse is fully expanded, but well before this expansion occurs. The excitation and the signal collection through denser bubble lead to lower light losses because the vapor refractive index is closer to that of surrounding liquid; consequently, the light defocusing by the bubble itself is low. The corresponding plasma lifetime is longer than inside rarefied bubble. Transitions through the focal volume of the initially formed pressure waves, reflected from the nearby interfaces, lead to a manifold DP LIBS signal increase. Here, we explained that strong signal enhancement by DP excitation is also caused by the reduced energy losses into shockwaves, cavitation and liquid evaporation. The importance of the spatial overlap between the two laser beams is also discussed. From the practical point of view, LIBS characterization of bulk liquids or submerged targets is presently the only viable way for measuring the elemental composition in the absence of a sample-gas interface. The examples regard direct liquid characterization inside transparent containers, local measurements inside natural waters, feasible also in deep oceans, and undersea recognition of solid materials. Until now, only two undersea trials have been reported. One regards characterization of archeological materials up to depth of 30 m, the depth being limited by the length of fiber probe connected to the instrument placed on the boat. The other instrument, mounted on a ROV, performed preliminary tests on waters and solids at depth of about 200 m. Realization and testing of other types of underwater LIBS instruments would require consistent funding to build-up a compact submergible system and test it in campaigns. The experimental results obtained on solid submerged targets by using the compact, non-gated spectrometers, are very promising for building up the LIBS equipment for underwater exploration. In order to increase the chances for employing the LIBS in-situ, it is of outmost importance to perform further studies in view of optimizing the laser excitation and scaling the power of one or more lasers. Some works, here cited, report the LIBS spectra adequate for the material recognition underwater by using the pulse energy of only 1 mJ. By applying higher pulse energies from one or two laser sources, the detection limits in the order of 10 ppb and 100 ppm have been achieved for elements present in water and in submerged solids, respectively. In the present review we also pointed out some analytical aspects regarding optimal laser excitation, matrix dependent calibration, optical system for beam delivery and signal collection, sampling conditions and signal processing. LIBS spectroscopy of bulk liquids or submerged solid samples is a very promising research area from the standpoints both of basic science and applications. Fundamental studies of the plasma formation, cavitation and successive beam coupling through the bubble, described in this work, are of interest not only for increasing performances of LIBS but also in other fields like biomedical, wet material processing and synthesis of nanoparticles. Acknowledgments The authors thank Marijana R. Gavrilović, PhD student, for literature survey. References [1] D. Winefordner, I.B. Gornushkin, T. Correll, E. Gibb, B.W. Smith, N. Omenetto, Comparing several atomic spectrometric methods to the super stars: special emphasis on laser induced breakdown spectrometry, LIBS, a future super star, J. Anal. At. Spectrom. 19 (2004) 1061–1083. [2] L. Radziemski, D. Cremers, A brief history of laser-induced breakdown spectroscopy: from the concept of atoms to LIBS 2012, Spectrochim. Acta B 87 (2013) 3–10. [3] D.W. Hahn, N. Omenetto, Laser-Induced Breakdown Spectroscopy (LIBS), part I: review of basic diagnostics and plasma–particle interactions: still-challenging issues within the analytical plasma community, Appl. Spectrosc. 64 (2010) 335A–366A.

310


[4] D.W. Hahn, N. Omenetto, Laser-Induced Breakdown Spectroscopy (LIBS), part II: review of instrumental and methodological approaches to material analysis and applications to different fields, Appl. Spectrosc. 66 (2012) 347–419. [5] J.L. Gottfried, F.C. De Lucia Jr., C.A. Munson, A.W. Miziolek, Laser-induced breakdown spectroscopy for detection of explosives residues: a review of recent advances, challenges, and future prospects, Anal. Bioanal. Chem. 395 (2009) 283–300. [6] F.J. Fortes, J.J. Laserna, The development of fieldable laser-induced breakdown spectrometer: no limits on the horizon, Spectrochim. Acta B 65 (2010) 975–990. [7] C. Aragón, J.A. Aguilera, Characterization of laser induced plasmas by optical emission spectroscopy: a review of experiments and methods, Spectrochim. Acta B 63 (2008) 893–916. [8] I. Gornushkin, U. Panne, Radiative models of laser-induced plasma and pumpprobe diagnostics relevant to laser-induced breakdown spectroscopy, Spectrochim. Acta B 65 (2010) 345–359. [9] A.K. Pathak, R. Kumar, V.K. Singh, R. Agrawal, S. Rai, A.K. Rai, Assessment of LIBS for spectrochemical analysis: a review, Appl. Spectrosc. Rev. 47 (2012) 14–40. [10] D. Santos Jr., L.C. Nunes, G.G. Arantes de Carvalho, M. da Silva Gomes, P.F. De Souza, F. De Oliveira Leme, L.G. Cofani dos Santos, F.J. Krugg, Laser-induced breakdown spectroscopy for analysis of plant materials: a review, Spectrochim. Acta B 71–72 (2012) 3–13. [11] S.J. Rehse, H. Salimnia, A.W. Miziolek, Laser-induced breakdown spectroscopy (LIBS): an overview of recent progress and future potential for biomedical applications, Med. Eng. Technol. 36 (2012) 77–89. [12] V.K. Singh, A.K. Rai, Prospects for laser-induced breakdown spectroscopy for biomedical applications: a review, Laser Med. Sci. 26 (2011) 673–687. [13] V.S. Burakov, S.N. Raikov, N.V. Tarasenko, M.V. Belkov, V.V. Kiris, Development of laser induced breakdown spectroscopy method for soil and ecological analysis, J. Appl. Spectrosc. 77 (2010) 595–608. [14] R. Gaudiuso, M. Dell'Aglio, O. De Pascale, G.S. Senesi, A. De Giacomo, Laser induced breakdown spectroscopy for elemental analysis in environmental, cultural heritage and space applications: a review of methods and results, Sensors 10 (2010) 7434–7468. [15] F.J. Fortes, J. Moros, P. Lucena, L.M. Cabalín, J.J. Laserna, Laser-induced breakdown spectroscopy, Anal. Chem. 85 (2013) 640–669. [16] D.A. Cremers, L.J. Radziemski, T.R. Loree, Spectrochemical analyses of liquids using the laser spark, Appl. Spectrosc. 38 (1984) 721–729. [17] C. Walther, C. Bitea, W. Hauser, J.I. Kim, F.J. Scherbaum, Laser induced breakdown detection for the assessment of colloid mediated radionuclide migration, Nucl. Inst. Methods Phys. Res. B 195 (2002) 374–388. [18] A. Sarkar, V.M. Telmore, D. Alamelu, S.K. Aggarwal, Laser induced breakdown spectroscopic quantification of platinum group metals in simulated high level nuclear waste, J. Anal. At. Spectrom. 24 (2009) 1545–1550. [19] S. Koch, W. Garen, M. Muller, W. Neu, Detection of chromium in liquids by laser induced breakdown spectroscopy (LIBS), Appl. Phys. A 79 (2004) 1071–1073. [20] R. Knopp, F.J. Scherbaum, J.I. Kim, Laser induced breakdown spectroscopy (LIBS) as an analytical tool for the detection of metal ions in aqueous solutions, Fresenius J. Anal. Chem. 355 (1996) 16–20. [21] M. Lawrence-Snyder, J. Scaffidi, S.M. Angel, A.P.M. Michel, A.D. Chave, Laserinduced breakdown spectroscopy of high-pressure bulk aqueous solutions, Appl. Spectrosc. 60 (2006) 786–790. [22] H. Hou, Y. Tian, Y. Li, R. Zheng, Study of pressure effects on laser induced plasma in bulk seawater, J. Anal. At. Spectrom. 29 (2014) 169–175. [23] L. St-Onge, E. Kwong, M. Sabsabi, E.B. Vadas, Rapid analysis of liquid formulations containing sodium chloride using laser-induced breakdown spectroscopy, J. Pharm. Biomed. Anal. 36 (2004) 277–284. [24] C. Jeong, P.S. Tsai, D. Kleinfeld, Prospect for feedback guided surgery with ultrashort pulsed laser light, Curr. Opin. Neurobiol. 22 (2012) 24–33. [25] S. Guirado, F.J. Fortes, V. Lazic, J.J. Laserna, Chemical analysis of archeological materials in submarine environments using laser-induced breakdown spectroscopy. On-site trials in the Mediterranean Sea, Spectrochim. Acta B 74–75 (2012) 137–143. [26] V. Lazic, F. Colao, R. Fantoni, V. Spizzichino, Recognition of archeological materials underwater by laser induced breakdown spectroscopy, Spectrochim. Acta B 60 (2005) 1014–1024. [27] V. Lazic, F. Colao, R. Fantoni, V. Spizzichino, S. Jovićević, Underwater sediment analyses by laser induced breakdown spectroscopy and calibration procedure for fluctuating plasma parameters, Spectrochim. Acta B 62 (2007) 30–39. [28] A. De Giacomo, M. Dell'Aglio, R. Gaudiuso, S. Amoruso, O. De Pascale, Effects of the background environment on formation, evolution and emission spectra of laserinduced plasmas, Spectrochim. Acta B 78 (2012) 1–19. [29] A. De Giacomo, M. Dell'Aglio, O. De Pascale, M. Capitelli, From single pulse to double pulse ns-laser induced breakdown spectroscopy under water: elemental analysis of aqueous solutions and submerged solid samples, Spectrochim. Acta B 62 (2007) 721–768. [30] V. Lazic, LIBS analysis of liquids and of materials inside liquids, in: S. Mussazzi, U. Perini (Eds.), Laser Induced Breakdown Spectroscopy: Theory and Applications, Springer & Verlag, 2014. [31] G.M. Hale, M.R. Querry, Optical constants of water in the 200-nm to 200-μm wavelength region, Appl. Opt. 12 (1973) 555–563. [32] P. Schiebener, J. Straub, Refractive index of water and steam as function of wavelength, temperature and density, J. Phys. Chem. Ref. Data 19 (1990) 677–715. [33] R.W. Austin, G. Halikas, The Index of Refraction of Sea Water. SIO Report 76-1, January 1976, Scripps Institution of Oceanography, San Diego, California, USA, 1976. 121. [34] C.D. Mobley, Optical Properties of Water in Handbook of Optics, in: M. Bass (Ed.), 2nd EdMcGraw-Hill, Inc., 1994

[35] P. Kennedy, D. Hammer, B. Rockwell, Laser-induced breakdown in aqueous media, Prog. Quant. Electron. 21 (1997) 155–248. [36] W. Shockley, Problems related to p–n junctions in silicon, Solid State Electron. 2 (1961) 35–60. [37] F. Docchio, P. Regondi, M.R.C. Capon, J. Mellerio, Study of temporal and spatial dynamics of plasmas induced in liquids by nanosecond laser pulses. 2: Plasma luminescence and shielding, Appl. Opt. 27 (1988) 3669–3674. [38] F. Williams, S.P. Varma, S. Hillenius, Liquid water as a lone-pair amorphous semiconductor, J. Chem. Phys. 64 (1976) 1544–1549. [39] P.K. Kennedy, A first-order model for the computation of laser-induced breakdown thresholds in ocular and aqueous media. I. Theory, IEEE J. Quantum Electron. QE-31 (1995) 2241–2249. [40] Q. Feng, J.V. Moloney, A.C. Newell, E.M. Wright, K. Cook, P.K. Kennedy, D.X. Hammer, B.A. Rockwell, C.R. Thompson, Theory and simulation on the threshold water breakdown induced by focused ultrashort laser pulses, IEEE J. Quantum Electron. QE-33 (1997) 127–137. [41] J. Noack, A. Vogel, Laser-induced plasma formation in water at nanosecond to femtosecond time scales: calculation of thresholds, absorption coefficients, and energy density, IEEE J. Quantum Electron. 35 (1999) 1156–1167. [42] H. Schmidt-Kloiber, G. Paltauf, E. Reichel, Investigation of the probabilistic behavior if laser induced breakdown in pure water and in aqueous solutions of different concentrations, J. Appl. Phys. 66 (1989) 4149–4153. [43] F. Docchio, P. Regondi, M.R.C. Capon, J. Mellerio, Study of temporal and spatial dynamics of plasmas induced in liquids by nanosecond laser pulses. 1: Analyses of the plasma starting times, Appl. Opt. 27 (1988) 3661–3668. [44] A. Vogel, K. Nahen, D. Theisen, J. Noack, Plasma formation in water by picoseconds and nanosecond Nd:YAG laser pulses — part I: optical breakdown at threshold and superthreshold irradiance, IEEE J. Sel. Top. Quant. Electron. 2 (1996) 847–860. [45] F. Docchio, C.A. Sacchi, J. Marshall, Experimental investigation of optical breakdown threshold in ocular media under single pulse irradiation with different pulse durations, Lasers Ophthalmol. 1 (1986) 83–93. [46] D. Grand, A. Bernas, E. Amouyal, Photoionization of aqueous indole: conduction band edge and energy gap in liquid water, Chem. Phys. 44 (1979) 73–79. [47] J.H. Marburger, Self-focusing: theory, Prog. Quant. Electron. 4 (1975) 35–110. [48] Z.W. Wilkes, S. Varma, Y.-H. Chen, H.M. Milchberg, T.G. Jones, A. Ting, Direct measurements of the nonlinear index of refraction of water at 815 and 407 nm using single-shot supercontinuum spectral interferometry, Appl. Phys. Lett. 94 (2009) 211102. [49] M.H. Helle, T.G. Jones, J.R. Penano, D. Kaganovich, A. Ting, Formation and propagation of meter-scale laser filaments in water, Appl. Phys. Lett. 103 (2013) 121101. [50] G. Toker, V. Bulatov, T. Kovalchuk, I. Schechter, Micro-dynamics of optical breakdown in water induced by nanosecond laser pulses of 1064 nm wavelength, Chem. Phys. Lett. 471 (2009) 244–248. [51] T. Kovalchuk, G. Toker, V. Bulatov, I. Schechter, Laser breakdown in alcohols and water induced by λ = 1064 nm nanosecond pulses, Chem. Phys. Lett. 500 (2010) 242–250. [52] R. Evans, S. Camacho-López, Pump-probe imaging of nanosecond laser-induced bubbles in distilled water solutions: observations of laser-produced-plasma, J. Appl. Phys. 108 (2010) 103106. [53] J. Noack, A. Vogel, Single-shot spatially resolved characterization of laser-induced shock waves in water, Appl. Opt. 37 (1998) 4092–4099. [54] C.E. Brennen, Cavitation and Bubble Dynamics, Oxford University Press, Oxford, 1995. [55] CRC Handbook of Chemistry and Physics, in: R.C. Weast (Ed.), CRC Press, Boca Raton, 1988. [56] I. Akhatov, O. Lindau, A. Topolnikov, R. Mettin, N. Vakhitova, W. Lauterborn, Collapse and rebound of a laser-induced cavitation bubble, Phys. Fluids 13 (2001) 2805–2819. [57] V. Lazic, J.J. Laserna, S. Jovicevic, Insights in the laser-induced breakdown spectroscopy signal generation underwater using dual pulse excitation — part I: vapor bubble, shockwaves and plasma, Spectrochim. Acta B 82 (2013) 42–49. [58] V. Lazic, S. Jovicevic, R. Fantoni, F. Colao, Efficient plasma and bubble generation underwater by an optimized laser excitation and its application for liquid analyses by laser-induced breakdown spectroscopy, Spectrochim. Acta B 62 (2007) 1433–1442. [59] R.T. Knapp, J.W. Daily, F.G. Hammitt, Cavitation, McGraw-Hill, New York, 1971. [60] B. Wolfrum, T. Kurz, O. Lindau, W. Lauterborn, Luminescence of transient bubbles at elevated ambient pressures, Phys. Rev. E. 64 (2001) 046306. [61] E. Pichahchy, D.A. Cremers, M.J. Ferris, Elemental analysis of metals under water using laser-induced breakdown spectroscopy, Spectrochim. Acta B 52 (1997) 25–39. [62] E.A. Brujan, K. Nahen, P. Schmidt, A. Vogel, Dynamics of laser-induced cavitation bubbles near elastic boundaries: influence of the elastic modulus, J. Fluid Mech. 433 (2001) 283–314. [63] C.S. Peel, X. Fang, S. Ahmad, Dynamics of laser-induced cavitation in liquid, Appl. Phys. A 103 (2011) 1131–1138. [64] B. Thornton, T. Sakka, T. Masamura, A. Tamura, T. Takahashi, A. Matsumoto, Longduration nano-second single pulse lasers for observation of spectra from bulk liquids at high hydrostatic pressures, Spectrochim. Acta B 97 (2014) 7–12. [65] M.P. Brenner, Single-bubble sonoluminescence, Rev. Mod. Phys. 74 (2002) 425–484. [66] R. Hiller, S.J. Putterman, B.P. Barber, Spectrum of synchronous picoseconds sonoluminescence, Phys. Rev. Lett. 69 (1992) 1182–1184. [67] C.D. Ohl, O. Lindau, W. Lauterborn, Luminescence from spherically and aspherically collapsing laser induced bubbles, Phys. Rev. Lett. 80 (1998) 393–396.

V. Lazic, S. Jovićević / Spectrochimica Acta Part B 101 (2014) 288–311 [68] D.S. Langley, P.L. Marston, Critical-angle scattering of laser light from bubbles in water: measurements, models, and application to sizing of bubbles, Appl. Opt. 23 (1984) 1044–1054. [69] R.G. Holt, L.A. Crum, Mie scattering used to determine spherical bubble oscillations, Appl. Opt. 29 (1990) 4182–4191. [70] V. Lazic, S. Jovicevic, M. Carpanese, Laser induced bubbles inside liquids: transient optical properties and effects on a beam propagation, Appl. Phys. Lett. 101 (2012) 054101. [71] A. Kruusing, Handbook of Liquids-Assisted Laser Processing, Elsevier, 2007. [72] H.W. Kang, H. Lee, A.J. Welch, Laser ablation in a liquid-confined environment using a nanosecond laser pulse, J. Appl. Phys. 103 (2008) 083101. [73] T. Sakka, K. Takatani, Y.H. Ogate, M. Mabuchi, Laser ablation at the solid–liquid interface: transient absorption of continuous spectral emission by ablated aluminium atoms, J. Phys. D. Appl. Phys. 35 (2002) 65. [74] J.H. Yoo, S.H. Jeong, R. Greif, R.E. Russo, Explosive change in crater properties during high power nanosecond laser ablation of silicon, J. Appl. Phys. 88 (2000) 1638–1649. [75] S.I. Kudryashov, S. Paul, K. Lyon, S.D. Allen, Dynamics of laser-induced surface phase explosion in silicon, Appl. Phys. Lett. 98 (2011) 254102. [76] W. Soliman, N. Takada, K. Sasaki, Growth processes of nanoparticles in liquidphase laser ablation studied by laser-light scattering, Appl. Phys. Express 3 (2010) 035201. [77] A. Vogel, V. Venugopalan, Mechanisms of pulsed laser ablation of biological tissues, Chem. Rev. 103 (2003) 577–644. [78] N. Takada, T. Nakano, K. Sasaki, Formation of cavitation-induced pits on target surface in liquid-phase laser ablation, Appl. Phys. A 101 (2010) 255–258. [79] S. Tao, Y. Zhou, B. Wu, Y. Gao, Infrared long nanosecond laser pulse ablation of silicon: integrated two-dimensional modeling and time-resolved experimental study, Appl. Surf. Sci. 258 (2012) 7766–7773. [80] Y. Zhou, S. Tao, B. Wu, Backward growth of plasma induced by long nanosecond laser pulse ablation, Appl. Phys. Lett. 99 (2011) 051106. [81] B. Thornton, T. Takahashi, T. Ura, T. Sakka, Cavity formation and material ablation for single-pulse laser-ablated solids immersed in water at high pressure, Appl. Phys. Express 5 (2012) 102402. [82] A.V. Simakin, V.V. Voronov, G.A. Shafeev, Nanoparticle formation during laser ablation of solids in liquids, Phys. Wave Phenom. 15 (2007) 218–240. [83] S. Venugopal Rao, G. Krishna Podagatlapalli, S. Hamad, Ultrafast laser ablation in liquids for nanomaterials and applications, J. Nanosci. Nanotechnol. 14 (2014) 1–25. [84] S.I. Dolgaev, A.V. Simakin, V.V. Voronov, G.A. Shafeev, F. Bozon-Verduraj, Nanoparticles produced by laser ablation of solids in liquid environment, Appl. Surf. Sci. 186 (2002) 546–551. [85] S.B. Wen, X.L. Mao, R. Grief, R.E. Russo, Expansion of the laser ablation vapor plume into a background gas. I. Analysis, J. Appl. Phys. 101 (2007) 023114. [86] C.Y. Liu, X.L. Mao, S.S. Mao, R. Greif, R.E. Russo, Particle size dependent chemistry from laser ablation of brass, Anal. Chem. 77 (2005) 6687–6691. [87] A. De Giacomo, A. De Bonis, M. Dell'Aglio, O. De Pascale, R. Gaudiuso, S. Orlando, A. Santagata, G.S. Senesi, F. Taccogna, R. Teghil, J. Phys. Chem. C 115 (2011) 5123–5130. [88] D.X. Hammer, E.D. Jansen, M. Frenz, G.D. Noojin, R.J. Thomas, J. Noack, A. Vogel, B.A. Rockwell, A.J. Welch, Shielding properties of laser-induced breakdown in water for pulse durations from 5 ns to 125 fs, Appl. Opt. 36 (1997) 5630–5640. [89] A. Vogel, J. Noack, K. Nahen, D. Theisen, S. Busch, U. Parlitz, D.X. Hammer, G.D. Noojin, B.A. Rockwell, R. Birngruber, Energy balance of optical breakdown in water at nanosecond to femtosecond time scales, Appl. Phys. B 68 (1999) 271–280. [90] J. Noack, D.X. Hammer, G.D. Noojin, B.A. Rockwell, A. Vogel, Influence of pulse duration on mechanical effects after laser-induced breakdown in water, J. Appl. Phys. 83 (1998) 7488–7495. [91] A. De Giacomo, M. Dell'Aglio, F. Colao, R. Fantoni, V. Lazic, Double-pulse LIBS in bulk water and on submerged bronze samples, Appl. Surf. Sci. 247 (2005) 157–162. [92] V. Lazic, J.J. Laserna, S. Jovicevic, Insights in the laser induced breakdown spectroscopy signal generation underwater using dual pulse excitation — part II: plasma emission intensity as a function of interpulse delay, Spectrochim. Acta B 82 (2013) 50–59. [93] K. Rifai, S. Laville, F. Vidal, M. Sabsabi, M. Chaker, Quantitative analysis of metallic traces in water-based liquids by UV-IR double-pulse laser-induced breakdown spectroscopy, J. Anal. At. Spectrom. 27 (2012) 276–283. [94] W. Pearman, J. Scaffidi, S.M. Angel, Dual-pulse laser-induced breakdown spectroscopy in bulk aqueous solution with an orthogonal beam geometry, Appl. Opt. 42 (2003) 6085–6093. [95] N.M. Thang, R. Knopp, H. Geckeis, J.I. Kim, H.P. Beck, Detection of nanocolloids with flow-field flow fractionation and laser-induced breakdown detection, Anal. Chem. 72 (2000) 1–5. [96] N.F. Bunkin, A.V. Lobeyev, Influence of dissolved gas on optical breakdown and small-angle scattering of light in liquids, Phys. Lett. A 229 (1997) 327–333. [97] T. Bundschuh, R. Knopp, J.I. Kim, Laser-induced breakdown detection (LIBD) of aquatic colloids with different laser systems, Colloids Surf. A Physicochem. Eng. Asp. 177 (2001) 47–55. [98] V. Lazic, F. Colao, R. Fantoni, V. Spizzichino, Laser-induced breakdown spectroscopy in water: improvement of the detection threshold by signal processing, Spectrochim. Acta B 60 (2005) 1002–1024.

311

[99] P.M. Michel, A.D. Chave, Single pulse laser-induced breakdown spectroscopy of bulk aqueous solutions at oceanic pressures: interrelationship of gate delay and pulse energy, Appl. Opt. 47 (2008) G122–G130. [100] W. Jiang-lai, L. Yuan, L. Ying, C. Kai, G. Jin-jia, Z. Rong-er, Time resolved laserinduced breakdown spectroscopy for calcium concentration detection in water, Optoelectron. Lett. 7 (2011) 0065–0067. [101] A. De Giacomo, M. Dell'Aglio, O. De Pascale, Deposition of high-resolution carbon/ carbon multilayers on large areas for X-ray optical applications, Appl. Phys. A 79 (2004) 1035–1042. [102] M. Lawrence-Snyder, J. Scaffidi, S.M. Angel, A.P.M. Michel, A.D. Chave, Sequentialpulse laser-induced breakdown spectroscopy of high-pressure bulk aqueous solutions, Appl. Spectrosc. 61 (2007) 171–176. [103] B. Thornton, T. Masamura, T. Takahashi, T. Ura, K. Ohki, T. Sakka, Development and field testing of laser-induced breakdown spectroscopy for in situ multi-element analysis at sea, Proc. IEEE Oceans 12 (2012) (120514-002). [104] P.M. Michel, A.D. Chave, Double pulse laser-induced breakdown spectroscopy of bulk aqueous solutions at oceanic pressures: interrelationship of gate delay, pulse energies, interpulse delay, and pressure, Appl. Opt. 47 (2008) G131–G143. [105] V. Lazic, Laser-induced plasma underwater: spectroscopic applications, enhancement of the emission and bubble cavitation. PhD thesis, p. 102–110. [106] A.P.M. Michel, M. Lawrence-Snyder, S.M. Angel, A.D. Chave, Laser-induced breakdown spectroscopy of bulk aqueous solutions at oceanic pressures: evaluation of key measurement parameters, Appl. Opt. 46 (2007) 2507–2515. [107] A. Vogel, S. Busch, U. Parlitz, Shock wave emission and cavitation bubble generation by picosecond and nanosecond optical breakdown in water, J. Acoust. Soc. Am. 100 (1996) 148–165. [108] H. Oguchi, T. Sakka, Y.H. Ogata, Effects of pulse duration upon the plume formation by the laser ablation of Cu in water, J. Appl. Phys. 102 (2007) 023306. [109] S.I. Dolgaev, A.V. Simakin, V.V. Voronov, G.A. Shafeev, F. Bozon-Verduraz, Nanoparticles produced by laser ablation of solids in liquid environment, Appl. Surf. Sci. 186 (2002) 546. [110] G. Cristoforetti, M. Tiberi, A. Simonelli, P. Marsili, F. Giammanco, Toward the optimization of double-pulse LIBS underwater: effects of experimental parameters on the reproducibility and dynamics of laser-induced cavitation bubble, Appl. Opt. 51 (2012) B30–B41. [111 A. De Giacomo, M. Dell'Aglio, F. Colao, R. Fantoni, Double pulse laser produced plasma on metallic target in seawater: basic aspects and analytical approach, Spectrochim. Acta B 59 (2004) 1431–1438. [112] H. Suzuki, H. Nishikawa, I.- Yin, S. Lee, Laser-induced breakdown spectroscopy at metal–water interfaces, Phys. Chem. Commun. 5 (2002) 88–90. [113] T. Sakka, S. Masai, K. Fukami, Y.H. Ogata, Spectral profile of atomic emission lines and effects of pulse duration on laser ablation in liquid, Spectrochim. Acta B 64 (2009) 981–985. [114] T. Sakka, H. Oguchi, S. Masai, K. Hirata, Y.H. Ogata, Use of a long-duration ns pulse for efficient emission of spectral lines from the laser ablation plume in water, Appl. Phys. Lett. 88 (2006) 061120. [115] A. Matsumoto, A. Tamura, K. Fukami, Y.H. Ogata, T. Sakka, Single-pulse underwater laser-induced breakdown spectroscopy with nongated detection scheme, Anal. Chem. 85 (2013) 3807–3811. [116] B. Thornton, T. Sakka, T. Takahashi, A. Tamura, T. Masamura, A. Matsumoto, Spectroscopic measurements of solids immersed in water at high pressure using a longduration nanosecond laser pulse, Appl. Phys. Express 6 (2013) 08240. [117] T. Sakka, K. Saito, Y.H. Ogata, Confinement effect of laser ablation plume in liquids probed by self-absorption of C2 Swan band emission, J. Appl. Phys. 97 (2005) 014902. [118] B. Thornton, T. Ura, Effects of pressure on the optical emissions observed from solids immersed in water using a single pulse laser, Appl. Phys. Express 4 (2011) 022702. [119] T. Castro, R. Reifenberger, E. Choi, R.P. Andres, Size-dependent melting temperature of individual nanometer-sized metallic clusters, Phys. Rev. B 42 (1990) 8548–8557. [120] V. Hohreiter, D.W. Hahn, Plasma–particle interactions in a laser-induced plasma: implications for laser-induced breakdown spectroscopy, Anal. Chem. 78 (2006) 1509–1514. [121] T. Sakka, A. Tamura, T. Nakajima, K.o Fukami, Y.H. Ogata, Synergetic effects of double laser pulses for the formation of mild plasma in water: toward non-gated underwater laser-induced breakdown spectroscopy, J. Chem. Phys. 136 (2012) 174201. [122] A. De Giacomo, M. Dell'Aglio, A. Casavola, G. Colonna, O. De Pascale, M. Capitelli, Elemental chemical analysis of submerged targets by double-pulse laser-induced breakdown spectroscopy, Anal. Bioanal. Chem. 385 (2006) 303–311. [123] D.C.S. Beddows, O. Samek, M. Liska, H.H. Telle, Single-pulse laser-induced breakdown spectroscopy of samples submerged in water using a single-fibre light delivery system, Spectrochim. Acta B 57 (2002) 1461–1471. [124] R. Barbini, F. Colao, V. Lazic, R. Fantoni, A. Palucci, M. Angelone, On board LIBS analysis of marine sediments collected during the XVI Italian campaign in Antarctica, Spectrochim. Acta B 57 (2002) 1203–1218. [125] V. Lazic, R. Fantoni, F. Colao, A. Santagata, A. Morone, V. Spizzichino, Quantitative laser induced breakdown spectroscopy analysis of ancient marbles and corrections for the variability of plasma parameters and of ablation rate, J. Anal. At. Spectrom. 19 (2004) 429–436.