1. Introduction

The ability to perform real-time analysis of water contamination is highly desired in various industrial fields. Sensitive detection of sodium (Na) in water is of particular importance for monitoring water quality because the presence of Na causes corrosion in structural material. For example, ppb (µg/L) level detection is required by the safety regulations of power plants [1]. Inductively coupled plasma optical emission spectroscopy (ICP-OES) has conventionally been used for the elemental analysis of aqueous solutions. However, this method requires sample pretreatment; furthermore, it is a laboratory-based technique and therefore must be conducted off-site. As such, conventional analytical methods are inadequate for monitoring water contamination in real-time.

Laser-induced breakdown spectroscopy (LIBS) has attracted much attention as a simple technique for the elemental analysis of gas, solid, and liquid samples [2]. The advantages of LIBS include its capability to remotely detect multiple elements in real-time with minimal sample preparation [3]. Although this advantage fulfills the requirements for real-time water monitoring, the suitability of LIBS analysis for trace elements in liquid is limited by the low sensitivity and reproducibility of the spectral measurements. The drawbacks of liquid LIBS arise from the rapid quenching of plasma by the splashing and fluctuation of the liquid surface [4–7]. To increase the signal intensity of LIBS by extending the plasma lifetime, dual-pulse LIBS has been widely applied for liquid analysis [8–14]. Furthermore, various forms of liquid sampling have been explored to enhance measurement stability by eliminating the turbulence on the liquid surface, including jet flows [10,11,15–20], droplets [21–23], aerosols [24,25], and solid-phase conversion [26–29].

For flowing liquid, LIBS is implemented by relatively simple devices and is thus considered suitable for on-line and real-time water monitoring. Liquid jets have been applied to Na detection in aqueous solutions to attain a low limit of detection (LOD) [10,15,30–32]. Remarkably, some studies have yielded LODs for Na at the sub-ppb level [10,31]. For example, Lo et al. ablated a sheath gas-assisted jet (diameter = 0.5 mm) of HCl solutions of NaCl with an ArF laser (193 nm), producing an LOD of 0.4 µg/L [31]. Kuwako et al. combined dual-pulse LIBS with a 0.2 mm thick liquid jet and achieved an LOD of 0.1 µg/L [10]. Although these studies demonstrated ingenious ways to improve the detection limit of Na in liquid, the methods employed in these studies are not cost-effective and introduce more complexity to the experimental setup.

Recently, ultrathin liquid sheet jets were applied to LIBS measurements [33,34]. The LIBS signal intensity depended largely on the sheet thickness; the 20 µm sheet jet yielded the highest optical emission among a range of 5–80 µm jets [33]. The presence of an optimal thickness was attributed to a compromise between minimizing splash and maximizing ablation volume. This stands in contrast to the conventional cylindrical jet (> 100 µm), for which a thicker jet is favorable for the sensitive detection of Na, indicating that a large ablation volume is advantageous for achieving high emission intensities in this range of thickness [11]. Therefore, it is worth comparing the ultrathin sheet jet with the thick cylindrical jet in terms of the Na detection sensitivity of LIBS. In the present study, we examined the difference in Na detection sensitivity between LIBS with the sheet jet and LIBS with the cylindrical jet, combining both methods with conventional single-pulse ablation. To measure this difference, the LODs achieved for Na were evaluated for both liquid jets.

2. Experiment

The experimental setup used in the present study is shown schematically in Fig. 1. The apparatus consists of a liquid recirculation system, a Q-switched Nd:YAG laser (New Wave Research, Tempest-20, US), and a 0.3-meter Czerny-Turner spectrometer (Acton Research, Spectra Pro 2300i, US) equipped with a 1800 grooves/mm grating. The details of the liquid recirculation system were described in previous works [33,34]. We employed two different types of liquid jets: a liquid cylindrical jet and a liquid sheet jet. The cylindrical jet was prepared with a pipette-tip nozzle, which produced a stable laminar flow with a diameter of 500 µm. The liquid sheet jet was formed by a trapezoidal-shaped groove nozzle with a slit of 0.6 × 0.3 mm (Metaheuristic Japan, Type L, Japan). The thickness of the liquid sheet decreased along the flow direction, ranging from 80 to 5 µm [33]. The LIBS measurements were taken at a distance of 12 mm from the nozzle end, corresponding to a thickness of 20 µm.

 

Fig. 1. Experimental setup for the LIBS of Na in an aqueous solution.

Download Full Size | PDF

The fundamental output of the laser (1064 nm) was employed to generate luminous plasma on the surfaces of the liquid jets. The laser was operated at a repetition rate of 10 Hz and a pulse duration of 5 ns. The laser beam was focused onto the surfaces of the flowing liquid jets by a plano-convex lens with a focal length of 100 mm. The irradiation spot was adjusted so that the laser-induced air breakdown occurred just above the jet surfaces. The spot size of the laser beam was ≈ 200 µm at the liquid-air interface. Optical emission from the plasma was collected by a pair of plano-convex lenses and delivered to an optical fiber. The collection lenses were positioned at 90° with respect to the incident laser beam. The emission light was fed into the spectrometer and detected by an intensified charge-coupled device (ICCD) camera (Princeton Instruments, PI-MAX3: 1024iRB, US).

Pump-probe shadowgraph imaging was performed to analyze the laser-induced hydrodynamics of the liquid jets. The second harmonic (532 nm) of a Q-switched Nd:YAG laser (Continuum, Surelite II, US) was used as a probe pulse triggered by a varying delay time from the 1064 nm pump pulse. The laser-liquid interaction region was illuminated by the probe beam and imaged with an ICCD camera (Hamamatsu, C2925-0, C6558, Japan) operating at a gate width of 40 ns to obtain time-resolved images of the liquid jets and the laser-induced shockwave.

The sample solutions used in this study were prepared by diluting an ICP single-element standard solution of Na (Kanto Chemical, Japan) in distilled water. The concentrations of Na in the sample solutions were confirmed using a quadrupole based ICP mass spectrometer (Seiko Instruments Inc.,SPQ9200, Japan).

3. Results and discussion

3.1 Optimization of experimental parameters

To optimize the experimental conditions for the LIBS measurements, we examined signal-to-background ratios (SBRs) for the Na 588.99 nm line from an aqueous solution containing 1 mg/L Na as a function of laser energy and the ICCD gate delay after the laser pulse. The LIBS spectra were acquired by accumulating 100 laser shots with a fixed ICCD gate width of 5 µs. Generally, the use of the sheet jet led to more intense plasma emission compared to the cylindrical jet. Figure 2(a) shows the emission spectra of Na obtained from the two liquid jets under the optimum conditions described below. It is evident that the sheet jet enhanced the line intensity. Figure 2(b) illustrates the laser energy dependence of the SBRs. All data were obtained at a gate delay optimized for a given laser energy. The SBRs for the sheet jet increased monotonically with increasing laser energy up to 125 mJ/pulse before dropping sharply at 135 mJ/pulse; this drop was probably due to the plasma shielding effect, in which dense plasma generated by the leading edge of the laser pulse absorbs the remaining part of the laser energy [35]. The SBR for the cylindrical jet varied slightly with increasing laser energy, reaching a slight maximum at 75 mJ/pulse. Thus, the laser energies were set to 125 mJ for the liquid jet and 75 mJ for the cylindrical jet.

 

Fig. 2. (a) Emission spectra of Na in the aqueous solution (1 mg/L) obtained from the 500 µm cylindrical jet and the 20 µm sheet jet. Each spectrum represents an accumulation of 100 laser shots. (b) The SBR variation of the Na 588.99 nm line with laser pulse energy.

Download Full Size | PDF

Figure 3(a) shows the dependence of the Na line intensity on the ICCD gate delay. In addition to the significant differences in the line intensity between the liquid jets, there were also differences in their temporal behavior; the optical emission from the sheet jet reached maximum intensity at a delay of 25 µs, while that from the cylindrical jet reached maximum intensity at a delay of 15 µs. A delay time of tens of microseconds is longer than those normally required for LIBS measurements. It is conceivable that the long delay could result from the low norm temperature for Na detection (∼4000 K), at which the population of the upper level for the 589 nm emission reaches a maximum [36]. The difference in the optimal delay time between the two jets leads us to infer that the plasma generated on the sheet jet cooled down more slowly than that generated on the cylindrical jet. Consequently, it took a longer time for the plasma on the sheet jet to reach the norm temperature. This means that the use of a sheet jet enhances plasma lifetime. The lifetime of the background continuum emission was also longer for the sheet jet. Consequently, both liquid jets reached their maximum SBRs at a delay of 35 µs, as shown in Fig. 3(b).

 

Fig. 3. (a) Intensity variation of the Na 588.995 nm line with gate delay. These data were obtained at a laser energy of 125 mJ/pulse for the sheet jet and 75 mJ/pulse for the cylindrical jet. (b) The SBR variation of the Na 588.995 nm line with gate delay.

Download Full Size | PDF

3.2 Sensitivity of Na detection

Figure 4 shows calibration curves for Na obtained from the peak intensity of the Na 588.99 nm line. The data were obtained with an ICCD gate width of 20 µs and a gate delay of 35 µs. The gate widths longer than 20 µs brought about complex spectral profiles due to the appearance of optical emissions from radical molecules. The data were acquired by averaging five replicate runs, each of which accumulated 500 laser shots. The maxima of the relative standard deviations (RSDs) were 6% for the sheet jet and 12% for the cylindrical jet. The calibration curve obtained from the sheet jet experiment showed nonlinear behavior, indicating that the self-absorption of the Na line was not negligible above 0.5 mg/L of Na (Fig. 4). Therefore, a linear fit was performed for the concentrations below 0.5 mg/L. The LODs for Na were calculated with the formula 3σ/S, where σ is the SD of the blank signal, and S is the slope of the calibration curve. Based on the fitted curves, the LODs were determined to be 0.57 µg/L for the sheet jet and 10.5 µg/L for the cylindrical jet. While the sensitivity yielded by the cylindrical jet was almost equivalent to or slightly greater than those obtained in previous studies of liquid LIBS [15,30,37,38], the use of the sheet jet increased the sensitivity by a magnitude of more than one order, thus achieving sub-ppb Na detection. The slopes of the calibration curves showed that the sensitivity of Na detection achieved by the sheet jet was approximately five times higher than that achieved by the cylindrical jet. In addition, the SD of the background signal from the sheet jet was more than three times smaller than that of the background signal from the cylindrical jet. An LOD value close to that of the commercial ICP-OES apparatus (∼ 0.5 µg/L) [39] was achieved by combining a single-pulse LIBS with a liquid sheet.

 

Fig. 4. Calibration curves of Na obtained from a liquid sheet jet and a liquid column jet.

Download Full Size | PDF

3.3 Time resolved shadowgraphs of laser-ablated liquid jets

The difference between the laser ablation processes of the two jets is particularly noticeable in the behavior of the splash from the jet surfaces. Figure 5 shows the temporal evolutions of the shadowgraph images recorded after the laser pulse hit the front surfaces of the liquid jets of 1 mg/L Na aqueous solution. Note that the shadows of the sheet jet appear to be thicker than 20 µm due to the thick rim formed on the side of the sheet. As observed in our previous report [33], the splash from the sheet jet emerged predominantly on the rear surface of the jet [Fig. 5(a)]; the splash on the front side is barely discernible from the 5 µs delay. Thus, the front surface of the sheet jet eluded splashing. In contrast, the cylindrical jet produced big splashes on both the front and rear surfaces of the jet [Fig. 5(b)].

 

Fig. 5. Temporal evolutions of the shadowgraph images for (a) the liquid sheet jet with a thickness of 20 µm and (b) the liquid cylindrical jet with a diameter of 500 µm.

Download Full Size | PDF

This stark difference in splashing behavior indicates that the incident laser energy was transformed into the mechanical energy of water to a greater extent for the cylindrical jet than for the sheet jet. This transformation is also reflected in the difference in shock wave propagation on the rear surface between the two jets; the shock front on the rear surface of the cylindrical jet grew faster than that of the sheet jet, as seen in the shadowgraphs at the 0.3 and 1 µs delays (Fig. 5). These results probably stemmed from the larger volume of the laser-liquid and plasma-liquid interaction in the cylindrical jet. Moreover, the 20 µm sheet jet is so thin that the liquid around the plasma core is easily blown away to the rear side of the jet by the shock wave formation, creating a hole on the jet surface. Due to the absence of liquid, the splash is no longer provoked on the front surface of the jet in the subsequent stage.

As has been widely recognized in previous studies of liquid LIBS [4–7], the liquid surface splash leads to absorption, scattering, and cooling of the plasma. Therefore, the plasma formed on the cylindrical jet is prone to be quenched more rapidly than that formed on the sheet jet. The sheet jet reduces the splash effectively, producing plasma with a longer lifetime and more intense emission. These advantages more than compensate for the small volume of ablated liquid intrinsic to the sheet jet LIBS. Consequently, optical emission is more intense for the liquid sheet, as shown in Fig. 2. Thus, the liquid sheet jet is much more suitable for liquid LIBS. Furthermore, splash suppression on the front side effectively prevents contamination of the optics.

4. Conclusions

The Na detection sensitivity of liquid LIBS was investigated using two types of liquid jets of aqueous NaCl solutions: a liquid sheet jet and a liquid cylindrical jet. The sheet jet yielded much more intense plasma emission than that the cylindrical jet, although it produced a smaller amount of ablated liquid. This occurred because the laser-induced plasma on the sheet jet avoids splashing, resulting in more intense and more persistent emission. The LODs for Na were determined to be 0.57 µg/L for the sheet jet and 10.5 µg/L for the cylindrical jet. Thus, LIBS with the sheet jet significantly outperforms LIBS with the typical cylindrical jet. The LOD value attained by LIBS with the sheet jet was almost equivalent to that attained by the commercial ICP-OES apparatus. In particular, we highlight that this sensitivity was achieved by the simple setup of a single pulse LIBS for liquid samples. This raises the obvious question of whether such sensitive detection is achievable for other elements in liquid. We are currently exploring the detection of other alkali metals in liquid. Preliminary experiments on the rubidium (Rb) detection in RbCl aqueous solutions have suggested that LIBS with the sheet jet provides an LOD for Rb comparable to that obtained for ICP-OES (5 µg/L [39]). This indicates that the LIBS using an ultrathin liquid sheet jet offers a promising approach to in situ analysis of trace alkali metals in aqueous solutions. Further application of the proposed technique is the detection of toxic heavy metals in liquid, which will be of particular importance in a wide variety of industrial fields.