1 Introduction

Laser-induced breakdown spectroscopy (LIBS) has been widely applied for element analysis in solid, liquid, and gas [1,2,3,4,5], because it offers a series of advantages such as fast response, no or minimal sample treatment and easy setup [6,7,8]. Using a high-power laser pulse to ablate the sample surface, a high-temperature plasma will be formed with plasma temperature higher than \(5\times 10^3\) K. Over time, the excited electrons in atoms and molecules decay to lower energy orbitals and emitting the photons. These photons, which have different wavelengths, reflect different components of the sample. By analyzing the spectral information of the sample, the composition of the sample can be obtained qualitatively or even quantitatively.

The sensitivity of LIBS depends on the elements, and the limit of detection (LOD) is in the order of ppm [8]. A number of techniques have been developed to improve the detection efficiency of LIBS, such as double-pulse [9,10,11], resonance [11, 12], nanoparticle [13,14,15], spatial confinement [16]. A. De Giacomo and co-workers applied nanoparticle enhanced LIBS (NELIBS) to liquid samples, enabling LIBS to quantify ppb concentration. By depositing silver nanoparticles on samples, they obtained an increase of 1–2 orders of magnitude in the LIBS signals [17, 18]. Fan Yang et al. [19] performed NELIBS and double-pulse LIBS (DP-LIBS) with femtosecond laser pulses on Si\(\mathrm O_2\) samples and obtained an enhancement factor above 10 at a low laser flux of 4.4 \(\mathrm mJ/cm^{2}\). Recently, Qiuyun Wang et al. [20] used Vortex beam to enhance the LIBS spectral intensity and achieved an enhancement factor about 2 compared with that obtained by Gaussian beam. The intensity in the middle of the Vortex beam is relatively low. A strong stagnation layer will be formed in the middle of the ring-shape plasma generated by the Vortex beam, relatively intense collision in the laser plasma will increase the plasma temperature manifesting as the enhancement of the spectral intensity. This is the first time using the Vortex beam to improve the LIBS signal. Vortex light is a hot topic in current research, it has a wide range of application prospects in ultra-large capacity optical communication [21], rotation detection [22], tweezers and manipulation [23], imaging [24], gravitational wave detection [25]. To our knowledge, correlational studies about the inter-pulse delay on femtosecond Vortex laser have not been reported yet. Utilizing the double-pulse configuration to measure the plasma evolution is a good way to understand the mechanisms of the signal enhancement [26,27,28]. The further study of the mechanisms would lead an improvement to the applications of Vortex beam LIBS.

In this paper, we investigate the enhancement of DP-LIBS of different samples in ambient environment using colinear NIR femtosecond Gaussian and Vortex pulses with the inter-pulse delay time changing in the range of several hundred picoseconds. The laser pulse energy is changed to exam the influence of pulse energy on the spectral signal enhancement for further improvements in sensitivity. The plasma densities generated by single Gaussian and Vortex pulse on the sample Al and Si are given to understand the measurement.

2 Experiment setup

Fig. 1
figure 1

Schematic diagram of the experimental setup, A1 reflection attenuation piece; A2 six stages attenuation piece; BS 5:5 beam-splitter; M1-M7 800 nm high-reflection mirrors; L1-L2 focusing lenses

Full size image

Figure 1 shows the schematic diagram of the collinear DP-LIBS system used for our measurement. The femtosecond laser pulses with pulse duration of 33 fs, pulse energy up to 5.0 mJ, and centeral wavelength of 800 nm are supplied by a Ti:Sapphire chirped-pulse amplification laser system operating at a repetition rate of 1 Hz. The laser beam is split into two beams by the first 5:5 beam splitter (BS1), then merged into a collinear beam at the second 5:5 beam splitter (BS2). A six stages attenuation piece (A1) and a reflection attenuator (A2) are placed into the beampath to change the total laser pulse energy. During the measurement, the total pulse energy is set to be 2.60 mJ and 2.03 mJ, respectively. A Vortex wave plate (VR2-808, LBTEK) is placed in the pump beampath to produce the Vortex beam. Then the laser beams are focused by a 70 mm focal length lens (L1) onto the sample target to produce the laser-induced plasma. The diameter of the laser spot on the target is about 100 \(\mathrm \mu\)m. The inter-pulse delay time (\(\Delta t\)) is controlled by changing the optical path difference of these two pulses through a translation stage. To ensure that every laser pulse hits a fresh area, the sample is fixed on a three-dimensional translation table which kept moving during the measurement. A lens (L2) is applied to collect the emission signal with a crossing angle of \(\mathrm 90^{\circ }\) to the focused beam. The spectral signal is then coupled into an optical fiber-pigtailed spectrometer (Mechelle 5000) and recorded by an intensified charge-coupled device (ICCD). Air plasma will be unavoidably produced by focusing the femtosecond pulse in air. The length of air plasma correlates with the Rayleigh length, which is about 0.21mm, i.e., air plasma is generated in a very short region. To minimize the influence of air plasma, the position of laser focus is adjusted by moving the lens L1 installed on a displacement platform. A semiconductor spectrometer (Ocean USB4000) is applied to measure the spectral signal in real time, and the optimal focusing position is determined when the maximum spectral signal is obtained. Gating of the spectral data collection is achieved by the internal electronics of the spectrometer, which is triggered by the laser output signal. The measured spectral range is 200–900 nm with a resolution of 0.16 nm. The gate detection width and the gate detection delay are defined as \(t_g\) and \(t_d\), respectively. In terms of the double-pulse configuration, \(t_d\) is controlled by adjusting the ICCD, which triggered by the laser pulse. To ensure the reliability of the data, each measurement is accumulated 60 times.

3 Results and discussion

Fig. 2
figure 2

The typical LIBS spectra of Al (a) and Si (b) obtained by single Gaussian pulse (black line), double Gaussian pulses (red line), and Vortex–Gaussian pulse (blue line) with a total pulse energy of 2.60 mJ, respectively

Full size image

LIBS spectra of Al and Si targets are measured in double- and single-pulse configurations. Results in Fig. 2 are the LIBS spectra of Al I (394 nm, 396 nm) and that of Si I (288 nm), the DP-LIBS spectra are measured at the best inter-pulse delay time. The intensities of all these Al lines are greatly enhanced in double-pulse configuration. Compared with the result obtained in single-pulse configuration, an enhancement factor of 10 is achieved in the case of Vortex–Gaussian configuration for Al sample, which is about 1.5 times higher than that of the double Gaussian scheme. For Si sample, the enhancement achieved in the Vortex–Gaussian configuration is at the same level as that in the double Gaussian one, and both spectral intensities are about 20 times higher than that obtained with the single pulse one.

Fig. 3
figure 3

The spectral intensity of DP-LIBS (double Gaussian and Vortex–Gaussian configurations) varying with the inter-pulse delay time. Panels (a) and (c) are the results measured on the sample Al (396 nm) and Si (288 nm) with an incident total pulse energy of 2.60 mJ in Vortex–Gaussian configuration, respectively, while panels (b) and (d) are these results obtained with the same incident pulse energy in double Gaussian scheme

Full size image

With an incident total laser pulse energy of 2.60 mJ, Fig. 3a and b displays the dependences of the spectral intensity measured on the Al sample varying with the inter-pulse delay time in the DP-LIBS configuration. For Al sample under Vortex–Gaussian beam configuration, as shown in Fig. 3a, the spectral intensity arrives the maximum point at 5 ps, then decreases rapidly with further increase of the inter-pulse delay time. For comparsion, it can be found in Fig. 3b that the weakest spectral intensity with the double Gaussian beam case is detected when the inter-pulse delay time is 0 ps. With the increase of the inter-time delay time, the intensity of atomic emission spectra increases in tens of picoseconds, then reaches its maximum at about 100 ps. In both schemes, the variation tendency of signal intensity with the inter-pulse delay time is basically similar, but the turning point of the evolution curve is different. The mechanism of femtosecond laser interaction with metal can be used to explain the enhancement of double-pulse LIBS signal. When the femtosecond laser is irradiated on the surface of Al, a large number of free electrons are excited by absorbing the photons in the laser pulse. The free electrons reach a quasi-thermal equilibrium state through the spin–spin relaxation process, which is known as free electron gas. The free electron heating process of Al is very short, on the same order of magnitude as the pulse width of femtosecond laser and will be completed in around 1 ps. If a second femtosecond pulse hits the surface of the material during free electron heating process, its influences on the ablation process of the material can be ignored, i.e., the ablation result of double pulse is similar as that a single pulse of the same energy. It just takes a little bit longer time to heat up the electrons. That is why a relatively low LIBS signal enhancement is achieved when the inter-pulse time delay is less than 1 ps.

After the free electron gas reaches thermal equilibrium, these free electrons then transfer energy to the surroundings through the outward-radiating phonons, which is called electroacoustic coupling process. The time required for this process is within 1–10 ps and increases with the increase of the laser flux [30]. Next comes the phonon dynamics of the metal lattice, which requires several picoseconds relaxation times to bring its energy distribution close to thermal equilibrium. At high laser flux, some of the lattice begins to melt before the electroacoustic coupling reaches equilibrium. The electrical and thermal conductivity of melted metal will be strongly reduced compared with that at room temperature [31]. Hence, the energy of the second pulse is absorbed by the melted Al plasma produced by the first pulse rather than released into the surrounding solids, suggesting that more energy is confined to the plasma. As the liquid Al gets hotter, it expands over time and begins to separate from the sample surface when it reaches its vaporization point. It expands vertically at the same order of magnitude as sound travels through the material. If a second pulse hits the surface of the sample during this process, it will interact with the plasma plume. In the case of relatively high laser flux, the energy of the second laser pulse can be directly coupled to these particles in the plasma generated by the first pulse, which is equivalent to the reheating effect of the plasma generated by the first pulse. There is a plasma shielding effect in the early time of solid plasma formation, because of the small volume and high density of the plasma plume. The energy of the second pulse is partly absorbed by the plasma produced by the first pulse through inverse bremsstrahlung radiation, which further heats up the plasma and ionizes, resulting in enhanced radiation of the isosomes. As the inter-pulse delay time increases, the plasma density decreases to below the critical value of plasma shielding effect and the laser can irradiate through the plasma plume to the bottom of the material for ablation. The reheating effect caused by the second pulse on the plasma generated by the first pulse is reduced, inducing the decrease of all LIBS signal with the inter-pulse delay time. The investigation of femtosecond laser pulses ablation on metal by A. Semerok et al found that the double pulse effect is similar to that of single pulse with the inter-pulse delay time less than 1ps, and the inter-pulse delay time from 1ps to 10ps corresponds to the transition regime with the partial plasma shielding manifesting as the rapid increase of emission signal [29].

For the Vortex–Gaussian and double Gaussian cases, the main differences are the initial plasma density and the mechanism of interaction between the plasma and second laser beam. In the case of Vortex–Gaussian scheme, due to the ring-shape of plasma formed by Vortex beam, there will be a weak plamsa density in the center at beginning. According to the formula of electron density \(\Delta \lambda _{FWHM} =2\omega \left(\frac{N_{e} }{10^{16} }\right)\), the electron density \(N_e\) is proportional to full width at half maximum (FWHM) of the emission spectra. As shown in Table 1, the FWHM of single Vortex beam is smaller than the single Gaussian beam, i.e., the plasma produced by the single Vortex pulse is relatively thiner. The energy of the second pulse can be absorbed by the plasma generated by the first pulse through inverse Bremsstrahl radiation in dense plasma, which further heats up and ionizes the plasma and finally results in enhanced radiation of the plasma. As the inter-pulse delay time increases, the electron density of the plasma decreases below the critical electron density, the laser can irradiate through the plasma plume to the bottom of the material for ablation, the reheating effect on the plasma generated by the first pulse is weakened and the overall LIBS signal begins to decrease. When the inverse bremsstrahlung effect is weakened, the second Gaussian beam can interact with atoms in plasma, leading to emission signal intensity increases rapidly at early 10 ps in Vortex–Gaussian scheme. Then the ring-shape plasma expands on each other forming a plasma block layer in the center. The plasma in this region has a higher density and temperature [15]. Plasma shielding effect becomes the main factor for the absorption of the second laser energy. Since the plasma density is weaker compared to the plasma produced by the Gaussian beam, the electron density of the plasma drops below the critical electron density and is faster than the Gaussian case. Vortex beam leads to a change in the interaction process between the Al plasma and the second laser beama achieving a better signal enhancement with short inter-pulse delay times..

Table 1 Electron density of Al and Si plasma generated by single Gaussian pulse and single Vortex pulse at 2.6 mJ pulse energy
Full size table

For sample Si, as shown in Fig. 3c, the signal intensity decreases gradually until the inter-pulse delay time reaches 100 ps, then it keeps in the same level in Vortex–Gaussian configuration. In case of double Gaussian scheme (Fig. 3d), the best signal enhancement is achieved at 0 ps, then there is a rapidly descent witnin 0–10 ps. At range of 10–50 ps, the intensity of the emission spectrum drops slightly, in other words, remains unchanged almost. The study of underdense plasma generated with double femtosecond pulses found the spectral signal reaches its maximum at zero inter-pulse delay time [32], which well conforms to the experimental phenomenon of Si. If a laser pulse is applied to interact with semiconductor materials, valence band electrons transit to conduction band through single photon or multi-photon absorption. For a given pulse energy, the shorter the duration of laser pulse is, the nonlinear absorption increases significantly and the more easily photoexcitation occurs. At about 10 fs, decoherence appears. With the electrons’ transitions to the excited state due to the energy absorption from the laser field, the redistribution of electrons in different energy levels reaches the quasi-equilibrium. This time scale is about 0.1 ps. When the distribution of electrons achieves the Fermi-Dirac equilibrium, the electron temperature is much higher than the lattice temperature. 0.1–1 ps, the electronic system transmits energy to the lattice through electro-acoustic coupling and the electrons gradually cool down. Subsequently, thermal melting and ablative destruction of materials occur with thermal diffusion. In several picoseconds, the surface of the sample will be continuously melted. With the arrival of a second laser pulse at the surface of the sample, instead of the semiconducting silicon at room temperature, it interacts with the hot metal-like liquid plasma accompanied by strong absorption. As is exhibited in Table 1, the electron density in the plasma formed in semiconductor is lower than that in metal. The plasma generated by the Vortex beam has a ring-shape distribution with a relatively lower density, compared with the point plasma formed by the Gaussian pulse for sample Al. For the sample Si, the plasma excitation process depends on valence band electrons absorbing single- or multi-photon excitation to the conduction band [33], which is mainly related to the laser flux. For the measurement in double-pulse case, the ring-shape plasma formed by the first Vortex beam cannot form an effective blocking layer and most energy of the second pulse is deposited on the molten silicon. Thus, the reheating effect on the plasma is weakened. When the inter-pulse delay time is 0 ps, the interference effect between the two laser pulses leads the intensity of electric field to its maximum, which produces stronger ablative effect on the Si sample. According to the results in Table 1, the electron density of the plasma produced by the Vortex beam is similar with that produced by the Gaussian pulse. The plasma formed by Vortex light has larger spatial distribution and faster diffusion. Hence, the intensity of the emission signal decreases faster than that of the double Gaussian pulse and the signal enhancement factor is basically the same as that of the double Gaussian pulses. With a large inter-pulse delay time, the second laser pulse can traverse the plasma simply to interact with the metal-like liquid plasma. But this liquid’s reflectivity is about three times that of solid silicon and the laser energy absorption depth is three orders of magnitude smaller than that of solid Si at room temperature. The energy absorption of the second laser pulse is limited to a very thin layer. All these factors lead to the decrease of the signal enhancement in Si sample in the double-pulse configuration. The absorption ability of the plasma to the laser energy is positively related to the electron density [34] because of the inverse bremsstrahlung radiation, the electron density of the plasma generated by the first pulse is higher than that of Si at room temperature, and the energy of the second laser is absorbed in large quantities. Therefore, the plasma signal intensity generated by double pulse is much higher than that generated by single pulse. The enhanced effect of double-pulse depends on the density of the plasma produced by the first laser ablation, but the optimal delay time is different in the case of Gaussian and Vortex beam.

The interaction mechanism between laser and semiconductor is different from that of metal sample. When a laser pulse is applied to interact with Si sample, valence band electrons transit to conduction band through photon absorption. The effect of interference increases the intensity of the laser field and further the number of electrons transiting to the conduction band. For the sample Al, which contains a large number of free electrons, denser plasma can be produced at a lower laser pulse energy. The plasma shielding effect is more significant. During the plasma diffusion process, energy is transferred to the inner plasma by inverse bremsstrahlung radiation, resulting in the enhancement of the spectral signal in a relative larger range of inter-pulse delay time.

Fig. 4
figure 4

The spectral intensity of DP-LIBS (Vortex–Gaussian) varying with the inter-pulse delay time. Panel (a) is the results measured on the sample Al (396 nm), while panel (b) is the results obtained with the same incident pulse energy on sample Si (288 nm)

Full size image

The total pulse energy is then reduced to be 2.03 mJ to further examine the enhancement of the emission signal obtained in Vortex–Gaussian configuration varying with the inter-pulse delay time. Results in Fig. 4a and b are the dependences of the intensity of emission signal on the inter-pulse delay time measured in the sample Al and Si, respectively. For Al, the spectral intensity increases faster than that measured with the incident pulse energy of 2.60 mJ and it rapidly reaches its maximum in the early 10 ps. As the inter-pulse delay time is tuned to be lager than 50 ps, the variation of the spectral intensity is similar to that measured with higher pulse energy. We attribute this to the decrease of plasma density. The maximum DP-LIBS signal is obtained when the plasma shield is weakened. The smaller the laser flux is, the smaller the inter-pulse delay time is required. The spectral intensity of emission from Si decreases rapidly in the early 50 ps, then turns to increase at low-incident laser energy. The decrease of first incident laser energy has a great influence on the reflectance of metal-like liquid plasma, resulting in enhanced deposition effect of the second laser energy. With this relatively low pulse energy, the laser pulse ablates the Al and Si forming a thin plasma, which expands very quickly causing the earlier appearance of the turning point. Due to the expansion of the plasma generated by the first laser pulse, the reheating efficiency is reduced. Compared with the case of high pulse energy in sample Si, the plasma density decreases and the diffusion increases. When the inter-pulse delay time increases to the 100ps, the plasma shielding effect weakens and the energy of the second laser beam is gradually deposited onto the sample. The final effect is equivalent to the first laser pulse and the second acting on the sample together [35], resulting in the slightly enhancement of the spectral signal during 100–200 picoseconds.

Two main factors may cause the above differences. The Vortex beam that ablated the different sample targets formed the ring-shaped plasma, but different sample have different ablation mechanisms. Vortex beam interacting with the Al sample can produce more plasma with a lower pulse energy, so the dependences of the spectral enhancement on the inter-pulse time delay achieved with different total pulse energy are similar for the measurements performed in the Vortex–Gaussian configuration. For the Si sample, there is a great dependence on the laser energy and the plasma density increases obviously at high flux Vortex beam. The second reason is the different energy absorption mechanisms to the second laser pulse. For Al plasma, the energy of the second laser pulse is absorbed mainly by reverse bremsstrahlung radiation and direct excitation, while the energy is absorbed by metal-like liquid plasma in Si sample, inducing the different trend of the intensity of the emission signal with the inter-pulse delay time

4 Conclusion

In this work, femtosecond Vortex and Gaussian pulses are applied to perform LIBS measurement on the sample Al and Si with a home-built collinear double-pulse experimental setup. Comparing with the result obtained in single pulse case, the LIBS signals of Al have an enhancement factor about 10 and 7 in Vortex–Gaussian and double Gaussian configurations, respectively, while that of Si is about 20 in both configurations. The appearance of spectral enhancement means the improvement of the detection limit of elements. The inter-pulse delay time is changed in both Vortex–Gaussian and double Gaussian configurations to examine the variations of LIBS signals, exhibiting different tendencies on the different samples. Due to the different spatial distribution of Vortex and Gaussian pulses, the plasma densities generated by the Vortex pulse is lower than that by the Gaussian pulse on the sample Al. When a second pulse is then introduced to irradiate on the sample with a inter-pulse delay time, the volume and manner for the energy absorption of the second laser pulse by the plasma generated by the first pulse are different, inducing the different enhancement factors and variations of the measured LIBS signals. The material properties of semiconductor are different with metal, ring-shape plasma is generated by the Vortex pulse on the sample Si; however, the plasma densities generated by the Vortex pulse are similar with that by the Gaussian pulse on the sample. This explains the similar enhancement factor and different variation tendencies of the LIBS signals measured with the sample Si in Vortex–Gaussian and double Gaussian configurations. It can be found that the introduction of Vortex light is an effective approach to improve the detection limit of LIBS, meanwhile the material properties is also an important factor .