1. Introduction

Laser-induced breakdown spectroscopy (LIBS), which is also known as laser-induced plasma spectroscopy (LIPS), has developed rapidly as an analytical technique over the past few years [1]. The technique uses a beam pulsed laser and a focusing lens to generate a plasma that ablates a small amount of a sample in the nanogram to picogram range. The emission spectrum is collected by a spectrometer and a photoelectric detector from the plasma. The spectral data contain information about the elements contained in the target sample. This information, which consists of specific wavelengths and relative line intensities, is important for analysis of the sample [2].

Researchers have been attempting to enhance the emission intensity in spectroscopy, which is a key parameter in LIBS [3–7 ]. For example, plasma confinement [6, 8], fast spark discharges [3], magnetic fields [4], nanoparticle-enhanced LIBS [9, 10], resonance LIBS [11], flame-enhanced LIBS [12, 13], and double pulse methods [14, 15] have all been studied. In double-pulse laser systems in particular, the sample is irradiated twice at the same spot with a time separation that is specified in the configuration [16–19 ]. The emission signal intensity is increased by using the double-pulse laser when compared with the use of a single pulse laser at the same laser energy [17, 18, 20]. During this period, a femtosecond laser pulse is also introduced into LIBS. When compared with long pulse lasers (i.e., nanosecond or microsecond-scale), the advantages of femtosecond LIBS include a low breakdown threshold, a small heat-affected zone, and faster broadband background decay [21]. In recent years, femtosecond double-pulse lasers have been used to perform LIBS [15, 22–24 ]. In most cases, the time separation of the femtosecond double-pulse is in the range from 0 ps to hundreds of ps, and the signal intensity can be enhanced several times [15, 24].

Femtosecond double-pulse techniques have good prospects for application to LIBS. In this context, we study the emission enhancement of LIBS from a silicon surface that has been irradiated by a femtosecond double-pulse laser. Unlike previous studies (two femtosecond pulses with the same wavelength) [15, 22–24 ], there are very few studies related to dual-wavelength double-pulse femtosecond laser irradiation for the enhancement of LIBS [25]. We intend to study the influence of the time separation on femtosecond double-pulses using the fundamental wavelength (800 nm) and the second harmonic wavelength (400 nm) from a Ti:sapphire laser.

2. Experimental details

A schematic drawing of the experimental apparatus used for the dual-wavelength femtosecond double-pulse laser-induced plasma spectroscopy measurements is shown in Fig. 1(a). The laser system is an one-box ultrafast Ti: sapphire amplifier (Coherent Libra). The full-width at half maximum (FWHM) of the pulse is 50 fs, the wavelength is 800 nm, and the repetition rate is 1 kHz. The maximum output energy of a single femtosecond pulse is approximately 4 mJ. The wavelength of 400 nm is generated using a β-phase barium borate (BBO) crystal as the second harmonic wavelength from the fundamental wavelength of 800 nm. The laser energy at 400 nm can be maximized by rotating the angle of incidence of the BBO crystal. We consider this 400 nm beam to be the main pulse and the zero reference time point in this experiment. The dual-color laser pulse is split into two sub-pulses using a dichroic mirror. A computer-controlled translation stage (Physik instrumente, M-505) with resolution of 1 μm is used to obtain the time separation between the 800 nm and 400 nm pulses by changing the optical path of the 800 nm pulse. The time separation range is from −350 ps to 350 ps. Using a combination of a Glan laser polarizer and a half-wave plate, the laser energy at 800 nm can be attenuated to the desired value. Then, the 400 nm and 800 nm beams are combined via another dichroic mirror. The combined beams are then directed by an off-axis parabolic mirror (focal length of 25 cm), while avoiding group velocity dispersion between the 800 nm and 400 nm pulses, onto the sample surface (the spot diameter is approximately 100 μm). The sample (p-type Si〈100〉, MTI KJ Group, 500 ± 10 μm thickness) is mounted on a computer-controlled X-Y-Z stage (Thorlabs, PT3/M-Z8), which guarantees that the sample location is renewed before each double-pulse shots. The emission spectra are focused on a fiber using two lenses (BK7), and are guided to the spectrometer (Spectra Pro 500, PI Acton, with 150 grooves/mm, 1200 grooves/mm, and 2400 grooves/mm gratings; the grating used has 1200 grooves/mm). The light is detected using an intensified charge-coupled device (ICCD, PI-MAX, Princeton Instruments) with 1024×256 pixels. A programmable timing generator (PTG, ST-133, Princeton Instruments) is used to provide the time delay between the laser pulse and the ICCD shutter, and to set the ICCD gate width. The ICCD is synchronized by an output signal from the femtosecond laser system (synchronization and delay generator, SDG). The timing diagram is shown in Fig. 1(b). Each data point is typically an average of 200 shots. The entire experiment is carried out in air at atmospheric pressure.

 

Fig. 1 (a) Schematic drawing of the apparatus. Components include mirror (M), dichroic mirror (DM), off-axis parabolic mirror (PM), Glan laser polarizer (G), half-wave plate (HWP), and sample (S). (b) Timing diagram.

Download Full Size | PDF

3. Results and discussion

First, the laser energy at 400 nm is adjusted to a fixed value of 180 μJ, which is the maximum energy in this experiment, by rotating the angle of the BBO, and the laser energy at 800 nm can be attenuated to 180 μJ and 360 μJ using the Glan laser polarizer and a half-wave plate. To obtain experimental data with a good signal-to-noise ratio, the delay time of the ICCD is set at 2 μs using the SDG of the laser system, and the gate width of the ICCD is set to 400 ns. This time delay can avoid the scattered light at 400 nm, and produces the best signal-to-noise ratio. Using the grating with 150 grooves/mm at the center wavelength of 500 nm in the spectrometer, the only observed emission line that was collected with the BK7 lenses is that of Si (I) at 390.55 nm from the 3s ²3p ²(¹ S ₀) ← 3s ²2p4s(¹ P ₁) transition. The 1200 grooves/mm grating is thus selected to improve the resolution of the 390.55 nm spectral line.

The spectral intensity distribution produced by the dual-wavelength femtosecond double-pulse laser-induced Si plasmas in the 389.8 nm – 391.2 nm spectral range with double-pulse time separation is shown in Fig. 2. The laser fluences of the double-pulse are 4.6 J/cm ² (i.e., 2.3 J/cm ² of the 800 nm pulse + 2.3 J/cm ² of the 400 nm pulse), and 6.9 J/cm ² (4.6 J/cm ² of the 800 nm pulse + 2.3 J/cm ² of the 400 nm pulse). As shown in the figure, at a relatively high double-pulse laser fluence (6.9 J/cm ²), the emission intensity is higher than the low double-pulse laser fluence (4.6 J/cm ²). The high laser energy causes the plasma spectroscopy to be stronger and its emission will last for a longer decay time [26]. Also, the spectral intensity of the Si plasma is found to depend on the specific time separation of the dual-wavelength femtosecond double-pulse. It is possible that the emission intensity is enhanced because more material is ablated from the sample surface under femtosecond double-pulse laser irradiation as a result of a surface transformation excited by the first laser pulse [27]. The enhancement may also be caused by the plasma reheating effect of the second laser pulse [28]. The hydrodynamic effect that is generated by the first laser pulse may modify the propagation of the femtosecond laser pulse and the expansion processes of the plasma plume produced by the second laser pulse [23]. Therefore, the emission intensity of dual-wavelength femtosecond double-pulse LIBS can be optimized by selecting a specific time separation for the two sub-pulses.

 

Fig. 2 Contour map of spectral intensity with the time separation of the double-pulse laser. The fluence of the femtosecond double-pulse laser is: (a) 4.6 J/cm ² (2.3 J/cm ² at 800 nm + 2.3 J/cm ² at 400 nm), (b) 6.9 J/cm ² (4.6 J/cm ² at 800 nm + 2.3 J/cm ² at 400 nm).

Download Full Size | PDF

Figure 3 shows the spectra of several double-pulse time separations selected from Fig. 2 for the two laser energy combinations. In this figure, in the left column (a, c), the laser fluence is 4.6 J/cm2 (2.3 J/cm ² at 800 nm + 2.3 J/cm ² at 400 nm); in the right column (b, d), the laser fluence is 6.9 J/cm ² (4.6 J/cm ² at 800 nm + 2.3 J/cm ² at 400 nm). The selected time separations are −300 ps, −25 ps, 0 ps, 5 ps, and 300 ps. When compared with the time separation of 0 ps, the emission intensity of the Si dual-wavelength femtosecond double-pulse LIBS is significantly enhanced when the time separations are −25 ps and 5 ps. However, at the longer time separations (i. e., −300 ps and 300 ps), the emission intensity is lower than that for 0 ps. For 400 nm + 800 nm (bottom row, Fig. 3), as the time separation increases, the ion and electron densities increase rapidly. The maximum values of the electron and ion density have been observed at delay times of about 5 – 10 ps [29], after which the density decreases with increasing double-pulse time separation, and weakens the absorption of the second pulse (800 nm). Because the longer wavelength pulse can heat the plasma more efficiently [30, 31], the emission intensity will decrease because of the weaker absorption (800 nm). For 800 nm + 400 nm (top row, Fig. 3), the shorter wavelength pulse can reduce the plasma shielding effects, and can thus penetrate the plasma more easily [19, 32], so that the influence of plasma density is relatively weak, and the emission intensity thus depends on the interaction between the laser (400 nm) and the sample. The 800 nm laser pulse excites the silicon at the surface and melts it electronically [27]. Some of the absorbed energy is retained by the liquid and creates a phase boundary that propagates inward at roughly the speed of sound [33, 34]. The liquid depth typically reaches a maximum within 20 – 30 ps [33, 34], and then drops slowly. The second pulse (at 400 nm) is coupled more strongly to the liquid phase, producing stronger spectral emission. When the time separation is ±300 ps, various physical processes have become very weak, and the double-pulse becomes equivalent to two single pulses. Each pulse needs to overcome the valence band of silicon, which leads to the consumption of a great deal of energy. The spectral intensity will thun decrease. However, for the time separation of 0 ps, both pulses just overcome the valence band of silicon once, the energy loss is low. These results are different from those of Si plasma spectroscopy induced by a same-wavelength femtosecond double-pulse [27, 35]. This shows that the choice of double-pulse time separation is very important when attempting to obtain the stronger spectral emission for dual-wavelength femtosecond double-pulse LIBS. At the local thermodynamic equilibrium, the emission intensity for spectroscopy, corresponding to a transition from level k to level i, is given by I_λ = F_expN(A_kig_k/λU(T_p))(exp(−E_k/(k_bT_p))) [15, 36]. Here, A_ki is the transition probability, g_k and E_k are the degeneracy and the energy of the upper level k, U(T_p) is the partition function, λ is the emission wavelength, k_b is the Boltzmann constant, N is the total number density of a species at a given ionization stage, T_p is the plasma temperature, and F_exp is the experimental coefficient with respect to the efficiency of the optical detection system. According to this equation, for the same spectral line, most of the parameters (A_ki, g_k, E_k, U(T_p), λ, k_b, F_exp) are the same. The only possible differences are thus in N and T_p. Therefore, the enhanced mechanism of femtosecond double-pulse LIBS [23,35] is as follows: the increase in the spectral line intensity is due to more of the sample mass (N) being ablated from the target surface by the femtosecond double-pulse laser; the plasma reheating by the second pulse in the femtosecond double-pulse increases the plasma temperature (T_p), and thus enhances the plasma emission.

 

Fig. 3 Spectral intensity at the time separations of −300 ps, −25 ps, 0 ps, 5 ps, and 300 ps. The laser fluences of the femtosecond double-pulse laser are: (a, c) 4.6 J/cm² (2.3 J/cm ² at 800 nm + 2.3 J/cm ² at 400 nm), (b, d) 6.9 J/cm ² (4.6 J/cm ² at 800 nm + 2.3 J/cm ² at 400 nm).

Download Full Size | PDF

To aid to understanding of the detailed information of the dual-wavelength femtosecond double-pulse enhancement effect, Fig. 4 (a) shows the evolution of the spectral intensity with the time separation of the double-pulses. The line is Si (I) at a wavelength of 390.55 nm. The energies of the femtosecond double-pulse laser are 2.3 J/cm ² (800 nm) + 2.3 J/cm ² (400 nm), and 4.6 J/cm ² (800 nm) + 2.3 J/cm ² (400 nm). We see that the variation in the emission intensity is highly sensitive to the different time separations of the dual-wavelength double-pulse. Based on this plot, and beginning from the time zero, two main regions can be distinguished. In the range from 0 ps to ±300 ps, the LIBS emission intensity increases rapidly with increasing time separation between the 800 nm and the 400 nm pulses, and then begins to decreases. When we compare the evolution of both sides from the time zero, the time separations required to reach the maximum emission intensity are different. For longer time separations, the emission intensity gradually decreases, and the emission intensity is lower when compared with the time separation of 0 ps. However, on the right axis (where the pulse at 400 nm is earlier than the pulse at 800 nm), the evolution of the emission intensity is different to that on the left axis. The time separation of the obtained maximum emission intensity is shorter than that on the left axis, and the emission intensity decreases rapidly. The decay rate is high. At a time separation of approximately 100 ps, the intensity remains almost constant, and the constant value is lower than the intensity for 0 ps time separation. For the maximum emission intensity on the left or right axis, when the pulse at 400 nm is earlier than the pulse at 800 nm (right axis), the maximum emission intensity obtained by the dual-wavelength femtosecond double-pulse laser at a specific time separation is higher when compared with the corresponding value on the left axis. This observed result is obviously different to those reported in earlier studies on same-wavelength femtosecond double-pulse LIBS [27, 35, 37]. This result indicates that a shorter wavelength laser may be preferable for plasma generation, while a longer wavelength delayed laser pulse is more appropriate for plasma reheating [30, 38]. The wavelength dependence demonstrates that the electron temperature is consistently higher for longer wavelength pulses rather than short wavelength pulses, whereas the opposite behavior is observed for the electron number density [31].

 

Fig. 4 (a) Spectral intensity vs. the time separation for femtosecond double-pulse laser; (b) enhancement ratio vs. the time separation for femtosecond double-pulse laser. The laser fluences of the femtosecond double-pulse laser are 4.6 J/cm ² (2.3 J/cm ² at 800 nm + 2.3 J/cm ² at 400 nm), and 6.9 J/cm ² (4.6 J/cm ² at 800 nm + 2.3 J/cm ² at 400 nm).

Download Full Size | PDF

The interaction of the laser and the matter is dependent on the wavelength dependence of the laser absorption by both the sample and the plasma. The shorter wavelength laser pulse is more efficient in coupling to the target because of reduced shielding effects, which results in increased mass ablation when compared with the longer wavelength pulses [19, 32]. Also, the longer wavelength pulses can heat the plasma more efficiently by inverse Bremsstrahlung (IB) [30, 31]. In femtosecond laser ablation of semiconductors (i.e., silicon), the laser pulse energy is first absorbed by the valence band electrons through interband absorption because of a lack of free electrons. Therefore, single-photon and multi-photon (MP) absorption are both important for the femtosecond irradiation of silicon. The electrons that are excited into the valence band will absorb more of the laser pulse energy by the IB absorption process, in a similar manner to the cases in metals. The absorbed laser energy of the sample can be divided into two parts: the thermal energy of the electrons, and the energy that was consumed to overcome the band gap. Obviously, it is easier for the 400 nm laser with high photon energy to overcome the band gap and obtain the high energy than the 800 nm laser [39], and it is also easier for the 400 nm laser with its higher frequency to penetrate the high density plasma. However, it is easier for the 800 nm laser to be absorbed by the electrons based on the IB, which followed the wavelength with an exponent of three [30].

For the 800 nm + 400 nm pulse combination, when compared with the 400 nm + 800 nm combination, the variation of the emission intensity can also be caused by higher mass ablation due to better laser-target coupling. The mechanism that emerges from this analysis is as follows. The first pulse (800 nm) superheats the silicon surface and melts it electronically [27]. Some of the atoms then absorb sufficient photons to produce electronically excited Si emission, and an ablation crater is also formed. Some of the absorbed energy is retained by the liquid and creates a phase boundary that propagates inward at roughly the speed of sound. The liquid depth typically reaches a maximum within tens of picoseconds [40], and then drops away slowly. The second pulse (400 nm) couples more strongly to the liquid phase, producing a larger fraction of the electronically excited Si, and thus causing the enhanced spectral emission. Because the shorter wavelength pulse can interact for longer periods with the high-density regions of the plasma, the pulse should penetrate and reach the liquid phase of the sample more easily. This may be explained by the deeper ablation crater produced by this combination [19]. The change in the first pulse melting the surface determines the change in the spectral intensity of the plasma.

When the long wavelength pulse (800 nm) is fired at a specific time after the short wavelength pulse (400 nm), the first laser pulse (400 nm) is used to pre-excite the silicon, and the electron density in this region increases to a very high level within a few picoseconds [41]. Subsequently, the electron density will then decrease rapidly due to electron recombination and diffusion effects [41]. At a certain time separation, the energy of the long wavelength pulse (800 nm) is mainly absorbed by the electrons. The absorption efficiency thus depends on the electron density. The change in the electron density determines the change in the spectral intensity of the plasma.

Figure 4(b) shows the enhancement ratio with the time separation of the dual-wavelength femtosecond double-pulse laser. The enhancement ratio of the emission intensity of the dual-wavelength femtosecond double-pulse laser is calculated from the results in Fig. 4(a) by dividing the emission intensities at the different double-pulse time separations by the emission intensity at 0 ps time separation. The enhancement ratio plot is similar to that of the emission intensity. The enhancement ratio at the lower laser energy (2.3 J/cm ² (800 nm) + 2.3 J/cm ² (400 nm)) is higher than that at the higher laser energy (4.6 J/cm ² (800 nm) + 2.3 J/cm ² (400 nm)). At the lower laser energy, the maximum value of the enhancement ratio is approximately 2.4. However, the maximum value of enhancement ratio value is only about 1.4 at the higher laser energy. For longer time separations (about ±300 ps), the enhancement ratio is 0.5 for both laser energy combinations. Therefore, we can optimize the emission intensity of dual-wavelength femtosecond double-pulse LIBS by varying the time separation of the double-pulse.

4. Conclusions

In conclusion, we have demonstrated the enhancement of the optical emission generated by dual-wavelength femtosecond double-pulse laser-induced Si plasma in the atmosphere. The spectral intensity of the Si plasma is dependent on the specific time separation of the dual-wavelength femtosecond double-pulse. Unlike the same-wavelength femtosecond double-pulse, the different wavelength combinations of the double-pulse will affect the time taken to reach the maximum emission intensity, and this behavior is analyzed by comparing the evolution of the emission intensity on both sides of the axis. When the 400 nm pulse arrives earlier than the 800 nm pulse, the increase in intensity is rapid and the decay rate of the intensity is high. At longer time separations (300 ps), the enhancement ratio is 0.5, and the enhanced efficiency of the emission intensity generated by the dual-wavelength double-pulse is low. This indicates that shorter wavelength lasers may be preferable for plasma generation, while a delayed laser pulse at a longer wavelength is more appropriate for reheating of the plasma. These results may also be used to provide a better way to optimize dual-wavelength femtosecond double-pulse LIBS.