1. Intoduction

Laser-induced plasma formation in liquids has been used in various fields such as laser medicine [1], nanomaterial synthesis [2], micromachining [3], and photoacoustic applications [4]. In particular, the employment of optical emissions from the laser-induced plasma in water for spectroscopic analysis, the so-called underwater laser-induced breakdown spectroscopy (LIBS), is now emerging as an attractive sensing technique for geochemical analysis in marine applications [5]. This is driven by the rapid advance in vehicle platforms such as manned submersibles and remotely operated vehicles for the ocean observation, where the in-situ chemical sensors capable of noncontact, multi-component, and long-term analysis are highly demanded [6]. Over the past few years, several underwater LIBS devices have been successfully developed and deployed for submarine applications [5,7,8]. And more recently, many attentions have been paid to the deep-sea environments such as the hydrothermal vent regions where the concentrations of metal elements are extremely high [9].

From the technical point of view, when applying LIBS into the deep-sea, the pressure effect caused by different ocean depths (e.g., 2000 m depth corresponds to ∼20 MPa) is inescapable. It is known that the dynamics of laser-induced plasma strongly depends on the characteristics of the external environment in which the plasma is expanding [10]. In underwater LIBS, the plasma plume interacts with the surrounding cavitation bubble during its evolution. The high pressure of water can suppress the expansion of the bubble [11] and therefore have a great influence on the LIBS signals. The knowledge of the pressure effect on the plasma emission spectrum is essential for the application of LIBS in deep-sea. Up to now, several groups have reported the underwater LIBS studies at high pressure conditions. Lawrence-Snyder et al. firstly obtained the LIBS signals of Na, Ca, Li, K and Mn at the pressures up to 27.6 MPa [12], and demonstrated that there is little or no signal enhancement by using double-pulse LIBS for the pressures above 10 MPa [13]. This was attributed to the smaller and shorter-lived cavitation bubble induced by the first laser pulse at high pressures [14], and was further proved by the work of López-Claros et al. on the submerged solid target [15]. Thornton et al. reported the use of long-pulse laser which can improve the quality of LIBS spectra at high pressures up to 30 MPa [16,17]. However, their studies showed that the external pressure has a negligible effect on the spectral intensity and broadness [18]. Other studies concerned the underwater LIBS analysis at high CO₂ pressures for geologic carbon storage applications [19,20].

It can be summarized that most studies are devoted to the improvement of laser irradiation schemes such as using double-pulse or long-pulse lasers at high pressure conditions. We can still find a large dispersion about the pressure effects observed with different experimental parameters. In our previous works, we also investigated the pressure and laser energy dependence of LIBS signals in bulk water. The spectral results showed that the high pressure has a significant effect on the late stage of plasma evolution [21] and both the plasma temperature and electron density tend to be higher at higher pressures [22]. However, due to the limited plasma diagnostic method, the mechanisms of pressure effects on the LIBS signals from different plasma species are still not fully understood [23]. In this work, we investigated the spectral features of ionic, atomic and molecular emissions for the plasma in water at different pressures from 1 to 40 MPa. By comparing between the time-resolved spectra and shadowgraph images, we will especially show the important role of cavitation bubble at high pressures on the characterization of plasma emissions during the entire stages of plasma evolution.

2. Experimental setup

The schematic diagram of the experimental setup is shown Fig. 1. A Q-switched Nd:YAG laser (Beamtech Optronics, Dawa 200) was operated at the fundamental wavelength of 1064 nm with a repetition rate of 10 Hz and a pulse duration of 10 ns. The laser beam passed through a half-wave plate and a Glan prism which allowed a fine adjustment of the laser pulse energy. A portion of each laser pulse (∼ 8%) was reflected by the beam splitter and sent to a photodiode connected with an oscilloscope to monitor the laser energy. A long-pass dichroscope (Thorlabs, DMLP 900) was used to transmit the laser beam and reflect the plasma emission light. The laser beam was then focused into the high-pressure chamber filled with water solution to generate the plasma. The focusing lens L1 consisted of an achromatic doublet (f = 75 mm) and a meniscus lens (f = 100 mm) to minimize the spherical aberrations. A moderate laser energy of 10 mJ was used to have relatively stable LIBS signals with good signal-to-noise ratios [24]. The high-pressure chamber was a stainless-steel cube of 27 mL volume (3 × 3 × 3 cm) and equipped with three sapphire windows (diameter 10 mm, thickness 14 mm) for the laser beam entrance as well as plasma observation. The plasma was generated in the center of the chamber. The chamber can be pressurized using a mechanical pump (not shown in Fig. 1) to a maximum of 50 MPa with a precision of 0.1 MPa. A USB pressure sensor was mounted inside the chamber to monitor the pressure values during the experiments.

 

Fig. 1. Experimental setup for the spectroscopic and imaging measurement of laser-induced plasma in water at different pressure conditions.

Download Full Size | PDF

Spectroscopic measurements were performed by observing the plasma along the same path used for laser delivery. The plasma emission light reflected by the dichroscope was focused onto the fiber entrance by a fused silica plano-convex lens L2 (f = 50 mm). The spectrum was then recorded by a spectrometer (Princeton Instruments, IsoPlane SCT 320) equipped with an intensified CCD (ICCD) camera (Princeton Instruments, PI-MAX4-1024i). The used grating of the spectrometer was 300 lines/mm, and the slit width was 50 µm. Three wavelength windows were selected in the experiment (central wavelength of 410 nm for Ca I and Ca II lines, 554 nm for CaOH molecular band, and 670 nm for Li I line), and the wavelength range of each window is about 128 nm. Meanwhile, shadowgraph images of the cavitation bubble were also taken to show the bubble dynamics at different pressures. The imaging system consisted of a probe laser (Laser 2, Big Sky, ULTRA CFR), a second ICCD camera (Andor Technology, iStar DH 734i), and a pair of achromatic doublet lenses L3 (f = 50 mm) and L4 (f = 150 mm). The probe laser operated at 532 nm wavelength and 8 ns pulse duration, was expanded by a pair of plano-concave lens L5 (f = −30 mm) and plano-convex lens L6 (f = 60 mm) and was used for the back illumination. An interferential filter centered at 532 nm (1 nm bandwidth) was placed in front of the ICCD camera in order to block the spontaneous emissions from the plasma, and only 532 nm laser light was detected. The timings between the excitation laser, probe laser, and two ICCD cameras were controlled by a delay generator (Stanford Research Systems, DG 645). Water solutions in this experiment were made from CaCl₂ and LiCl dissolved in deionized water, with the concentrations of 1000 ppm Ca and 100 ppm Li to get good LIBS signals and alleviate the self-absorption effect.

3. Results and discussion

3.1 Time-resolved emission spectra

The LIBS spectra of Ca and Li were first recorded with an optimized delay of 100 ns and a gate width of 2 µs. Fig. 2 shows the Ca II 393 and 396nm line, Ca I 422 nm line, CaOH molecular band at 554 nm (B²Σ⁺−X²Σ⁺), and Li I 670 nm line at three pressures of 1, 20, and 40 MPa. We can see that the increase in pressure has an obvious impact on the LIBS signals. For both the Ca II lines, the spectral intensities increase gradually from 1 to 40 MPa. For Ca I line, CaOH band and Li I line, similar behaviors are found from 1 to 20 MPa, that the intensities at 20 MPa are clearly higher. However, from 20 to 40 MPa, the results are quite different. No obvious increase was observed for Ca I and CaOH from 20 to 40 MPa. After the background subtraction, the band intensity of CaOH is even lower at 40 MPa than that at 20 MPa. Whereas for Li I line, we can see that the peak intensity is clearly reduced from 20 to 40 MPa, regardless of the fact that both the spectral background and broadness are increasing with pressures. These results indicate the complex effects of pressure on the LIBS signals that rely on the analyzed lines.

 

Fig. 2. Comparison of emission spectra of (a) Ca II 393 and 396nm, (b) Ca I 422 nm, (c) CaOH molecular band at 554 nm, and (d) Li I 670 nm at pressures of 1, 20, and 40 MPa. The spectra are recorded at a delay of 100 ns with a gate width of 2 µs and with an accumulation of 1000 laser shots.

Download Full Size | PDF

Time-resolved spectra were then taken to show the pressure effects on the temporal evolution of the plasma emission. Fig. 3 shows the spectral intensity of Ca II 393 nm, Ca I 422 nm, CaOH molecular band at 554 nm, and Li I 670 nm as a function of delay at the pressures of 1, 10, 20, 30, and 40 MPa. The used gate width is 50 ns for the delays from 50 to 450 ns and 300 ns for the delays from 500 to 2000 ns, to obtain a good signal-to-noise ratio of the spectral lines. The recorded spectra were then normalized to a gate width of 100 ns (gain of the ICCD kept constant) for the evaluation of spectral intensity presented in Fig. 3. The error bars of the spectral intensities are deduced from 8 replicated measurements. First in Fig. 3(a), we can see that the spectral intensity of Ca II line decreases rapidly under 1 MPa with a short lifetime of 300 ns. As the pressure increases, the decay becomes much slower (reaching a lifetime of 500 ns at 40 MPa) and shown with a gradually enhanced intensity especially at later times. While at early times before 100 ns, the enhancement is quite limited. For atomic and molecular emissions, the persistence is much longer than ionic emission during the fast cooling of plasma in water [25]. As shown in Fig. 3(b), for Ca I, the behavior before 1000 ns is similar to Ca II, where the intensity is higher at higher pressures and with a slower decay. However, we can observe a sharp reduction at the late stage for 30 and 40 MPa. And such reduction occurs earlier at 40 MPa (∼ 1000 ns) than that at 30 MPa (∼ 1500 ns). This reduction behavior was also observed for CaOH molecular band (in Fig. 3(c)) and Li I line (in Fig. 3(d)), and was shown to be more pronounced due to the longer lifetimes of CaOH and Li I compared with Ca I. These time-resolved results explain well the observed spectra in Fig. 2 that recorded integrally from 100 to 2000 ns.

 

Fig. 3. Time-resolved emission intensity of (a) Ca II 393 nm, (b) Ca I 422 nm, (c) CaOH molecular band at 554 nm, and (d) Li I 670 nm at pressures of 1, 10, 20, 30, and 40 MPa.

Download Full Size | PDF

3.2 Shadowgraph images of the cavitation bubble

To better understand the pressure effects on the dynamics of laser-induced plasma in water, shadowgraph images of the laser-induced cavitation bubble were taken from 100 to 2000 ns at the pressures of 1, 10, 20, 30, and 40 MPa. The results are shown in Fig. 4. First, we can see elongated and multiple bubbles formed at early times along the laser focal axis due to the multiple independent breakdown in water [26]. In particular, at 100 ns, no obvious difference was observed for the pressures between 1 and 40 MPa. This can be explained by the fact that, the initial plasma heated by a nanosecond laser pulse in water can reach temperatures on the order of 10⁴ K, and plasma pressures as high as 10³−10⁴ MPa [27,28]. Therefore, the external pressure of several tens MPa will have a negligible influence on the early stage of bubble evolution. And this corresponds well with the spectral results in Fig. 3 that there is a weak dependence of the plasma emission on the external pressure at early times before 100 ns.

 

Fig. 4. Time-resolved shadowgraph images of laser-induced bubble at pressures of 1, 10, 20, 30, and 40 MPa. The corresponding delays are indicated on the left side of the figure.

Download Full Size | PDF

After that, the high temperature and pressure generated within the bubble cause a fast expansion of the bubble volume in all directions. During the expansion, the pressure inside the bubble decreases with time because of the increased volume of the bubble, and when it reaches the saturated vapor pressure which is lower than the pressure of the surrounding water, the bubble collapses [29]. It is therefore clear that the bubble expansion and collapse phases depend strongly on the external pressure of water. As shown in Fig. 4, both the size and the lifetime of the cavitation bubble decrease dramatically with the external pressure as a consequence of faster dynamics. For example, at 1 MPa, the cavitation bubble still undergoes a rapid expansion at 2000 ns and the multiple cavities tend to be merged into a single one during the expansion. Whereas for 20 MPa, the bubble begins to shrink at 2000 ns. At higher pressures, the bubble expansion is greatly suppressed and it collapses at 2000 ns for 30 MPa and at 1500 ns for 40 MPa. Although the collapse produces a rapid increase in the temperature and pressure inside the bubble [29], no re-expansion of the bubble was observed due to the great energy loss during the bubble collapse phase.

The results in Fig. 4 agree well with the previous studies on the laser-induced bubble dynamics at high pressures, both in the framework of underwater LIBS [15,16] as well as laser ablation in liquids for nanomaterial synthesis [30,31]. Here, what we focus is the role of cavitation bubble on the characterization of plasma spectral emission under different pressures from a temporal evolution point of view. Hence, by comparing between the results in Fig. 4 and Fig. 3, we can yield the following picture: (1) the spectral emission depends weakly on the external pressure at the early stage of plasma; (2) as time evolves, the bubble is more compressed by the higher external pressure, leading to a positive confinement effect to maintain the plasma emission (corresponding to the slower decay before 1000 ns as shown in Fig. 3); (3) however, at very high pressures, the bubble collapses extremely fast and occurs even earlier than the persistence of plasma, leading to a negative quenching effect to terminate the plasma emission (corresponding to the sharp reduction for 30 and 40 MPa as shown in Fig. 3). We can assume that once after the passage of the laser pulse, the bubble evolution is primarily determined by the external pressure. The pressure dependence of the cavitation bubble will in turn determine the pressure dependence of the spectral emission from the plasma inside the bubble. This is reasonable in a view of energy balance during the breakdown process in water. According to the work of Vogel et al. [32], for a 10-mJ, 6-ns laser pulse at 1064 nm, a large percentage of the laser pulse energy (29.4%) is transformed into the cavitation bubble energy, while the energy of plasma radiation is negligible (6 × 10⁻⁴%). At high pressures where the bubble collapses before the cooling of plasma, the plasma will gain energy from the confinement effect caused by bubble collapse, but quench immediately after the collapse phase.

Note that the prevailing emitting species in the plasma are also different during the plasma evolution [10]. For Ca II, it appears earlier with a very short lifetime because it has higher ionization energy, so that it suffers no quenching effect when the pressure is high. While for atomic and molecular emissions, both the confinement and the quenching effects may exist during the entire lifetime of plasma. The persistence of CaOH is longer than Ca I, because CaOH molecules are formed by the recombination between Ca atoms and OH radicals [25]. This gives rise to a more pronounced quenching effect for CaOH at the late stage of plasma compared with Ca I. That’s also the case of Li I, whose lifetime is longer than Ca I because of the lower upper level energy of Li I (1.85 eV) compared with Ca I (2.93 eV). From these discussions, we can therefore understand the spectral features between the different plasma species when increasing the pressure as observed in Fig. 2 and Fig. 3.

3.3 Plasma temperature and electron density

The pressure effects caused by bubble dynamics are also proved on the physical properties of plasma. Fig. 5 shows the temporal evolution of the plasma temperature and electron density in water at the pressures of 1, 10, 20, 30, and 40 MPa. The error bars of the temperature and electron density are deduced from 8 replicated measurements. The temperature is estimated from the intensities of Ca I 422 nm and Ca II 393/396 nm lines by the Saha-Boltzmann plot method [33], and three typical Saha-Boltzmann plots are given as the inset in Fig. 5(a). The electron density is calculated from the Stark-broadened profile of the Li I 670 nm line with a Lorentzian fit [34]. This method for electron density assumes that the Stark effect is the dominant broadening mechanism, in comparison with other possible broadening mechanisms (e.g., pressure broadening). The Stark broadening is corrected by subtracting the instrumental broadening measured by a standard low-pressure Hg lamp. The broadening coefficient of Li I 670 nm line (1.38×10⁻³ nm at 10000 K) is from the Griem's book [35]. It should be noted that for the plasma in water, the high pressure will increase collision broadening due to atomic and ionic interactions, which may lead to an overestimation of the calculated electron density in this case [34]. As shown in Fig. 5(a), the initial temperatures measured at 50 ns are above 12 000 K which are similar between different pressures. It decreases rapidly to ∼ 8000 K at 250 ns for 1 MPa, and at 300, 350, 400, 450 ns for 10, 20, 30, and 40 MPa, respectively. This means that the stronger confinement effect caused by the compressed bubble at higher pressures can lead to an increase in plasma temperature that corresponds to the enhanced LIBS signals. At later times, the temperature calculation is not possible due to the absence of Ca II lines. However, the pressure effects on the late stage of plasma are clearly confirmed by the temporal evolutions of the electron densities as shown in Fig. 5(b).

 

Fig. 5. Temporal evolution of (a) temperature and (b) electron density of the plasma in water at pressures of 1, 10, 20, 30, and 40 MPa. Typical Saha-Boltzmann plots used for estimating the temperature are given as the inset in Fig. 5(a).

Download Full Size | PDF

From Fig. 5(b), we can see that the electron densities are always higher at higher pressures, that explains the increased continuum background and line broadening from 1 to 40 MPa as observed in Fig. 2. At lower pressures less than 10 MPa, the expansion time for cavitation bubble is long enough to complete the plasma recombination, so that the electron density decreases exponentially as a function of delay. At higher pressures, due to the fast collapse of the cavitation bubble, the plasma volume decreases quickly and the density of particles inside the plasma increases. That’s why we can observe a significant increase in electron density at the late stage of plasma for the pressures from 20 to 40 MPa. In these cases, the temporal evolutions of electron density are actually dominated by the dynamics of the cavitation bubble. The time for the minimum value of electron density in Fig. 5(b) (for example 800 ns at 30 MPa) corresponds to the time that the cavitation bubble reaches its maximum expansion (shown in Fig. 4). Therefore, these results may offer us a new spectroscopic approach for the investigation of laser-induced bubble dynamics at high pressures in water. From the application point of view, it seems that a properly enhanced pressure is beneficial for improving the underwater LIBS signals that both higher intensity and longer persistence of the plasma emission can be achieved. This would further highlight the use of underwater LIBS technique in the deep-sea, since a large proportion of the interesting ocean areas including the hydrothermal vents and mineral deposits are located within the depth between 1000-3000 m [36], corresponding to a pressure window of 10-30 MPa as shown in this work.

4. Conclusions

In conclusion, we have investigated the spectral features of laser-induced plasma in water at different pressures from 1 to 40 MPa. By comparing between the time-resolved spectra and shadowgraph images, we demonstrate that the dynamics of the cavitation bubble at high pressures plays a key role on the characterization of plasma emission. The initial plasma emission depends weakly on the external pressure. As time evolves, the bubble is more compressed by the higher external pressure, leading to a positive confinement effect to maintain the plasma emission. However, at very high pressures, the bubble collapses extremely fast and occurs even earlier than the cooling of plasma. The plasma will gain energy from the bubble collapse phase, but quench immediately after the collapse, leading to a sharp reduction in the plasma persistence. These effects caused by bubble dynamics explain well the observed spectral features of ionic, atomic and molecular emissions, and are further proved by the temporal evolutions of the plasma temperature and electron density. The present results in this work give not only some insights into the laser-induced plasma and bubble dynamics in high pressure liquids but also better understanding for the application of underwater LIBS in the deep-sea.