Laser interaction with materials which is determined by material properties and laser processing parameters, is a complex process involving multiple stages [1–3]. Femtosecond laser micromachining has shown its unique advantage to confine the amount of thermal diffusion due to short heating duration and rapid solid-to-vapor phase transition [4–6]. This fast energy deposition enables the femtosecond laser to be a versatile tool in processing materials for a smaller heat-affected zone, better quality ablation crater, and higher ablation rate. However, using a single femtosecond laser pulse can only exert limited impact on the ablation by changing laser processing parameters like pulse energy and pulse duration. To achieve effective ablation, it requires high power or multiple pulses in the femtosecond laser fabrication. Double-pulse femtosecond laser processing is used for surface ablation, and it has been shown to be a useful means to control material removal [7–10]. For different substrate materials, the double-pulse ablation mechanism is different. In the double-pulse femtosecond experiments for metals, when the time delay τd between the pulses is equal or longer than the electron-ion relaxation time τei, the efficiency of double-pulse-induced ablation decreases with respect to single-pulse irradiation [11–13]. For semiconductors, free-electron density is a crucial parameter for the evaluation of the ablation threshold of materials. In the case of double-pulse laser irradiation, the free electrons in semiconductors, such as Si and TiO2, can be excited by the first pulse irradiation. These free electrons cannot recombine with holes within 500 fs, which reduces the ablation threshold and allows the second pulse (SP) to generate enhanced ablation [14–16]. Different from metal and semiconductors, the ablation of dielectric materials undergoes nonlinear processes, including avalanche and multiphoton ionization [17–19]. The femtosecond laser irradiation can lead to a fast transient modification within pulse duration. For the double-pulse ablation of dielectric (fused silica), Nagata et al. believed that the first pulse (FP) functions as the preconditioning pulse and the SP effectively induces ablation [20]. Most of the reported studies are mainly focused on double-pulse-induced enhancement under the subpulses with laser fluence higher than the ablation threshold [21–23]. However, for the femtosecond laser ablation, the samples’ optical properties, such as absorption coefficient, can also be modified under the laser fluence lower than the ablation threshold within dozens of femtoseconds. This fast transient modification can be fully applied for the ablation enhancement by the following pulse duration at a short time delay close to pulse duration. Therefore, the related physical process and mechanism are completely different for us to explore.

In this paper, we systematically investigate the ablation enhancement by double femtosecond laser pulses at different laser fluences of the FP and different time delays. Under the condition of FP laser fluence higher than the ablation threshold, the optical absorption of the irradiated area increases due to the FP-induced ablation. Thus, the absorption of the SP can be enhanced. In the case of the FP laser fluence lower than the ablation threshold, the ablation enhancement is attributed to the light absorption difference between the liquid phase and solid phase at a time delay <330fs. By tuning time delay and total laser fluence, the change of light absorption can generate ablation enhancement of poly(ε-caprolactone) (PCL) film with ablation efficiency up to 338% compared to single-pulse irradiation. The laser fluence ratio between the two pulses can also be changed to further modulate the ablation enhancement of double-pulses irradiation. The understanding of the temporal overlapping double-pulse enhancement in the ultrashort laser micromachining process can significantly broaden its applications.

It has been claimed that the fundamental interaction processes between ultrashort lasers and dielectrics can be summarized as follows: (1) acceleration of the seed electrons in the laser field via inverse bremsstrahlung absorption (free carrier absorption); (2) generation of secondary electrons by collisional (avalanche) ionization; (3) electron–phonon coupling, i.e., transfer of energy from the electronic system to the lattice; (4) melting, vaporization, plasma formation, and/or ablation [17,18]. Based on this knowledge, the temporal behavior of the free-electron density in the conduction band n(t) can be described by a rate equation [18,24], ∂n(t)∂t=aI(t)n(t)+σkIk(t).(1)The first term on the right side of Eq. (1) describes avalanche ionization, and the second term means multiphoton ionization. I(t) is the time-dependent laser intensity and a is the avalanche coefficient. σk denotes a k-photon absorption cross section, and k is equal to the smallest number of photons needed to overcome the optical bandgap of the material. Equation (1) demonstrates that the bandgap of a dielectric has a major influence on its ablation efficiency, while the bandgap is affected by the material’s temperature. In case of temporal double-pulse irradiation, the free-electron density in the conduction band of the FP and SP can be defined as n1(t) and n2(t). Thus, the free-electron density of double-pulse ablation is described by the following equation: ∂n1(t)∂t=aI1(t)n1(t)+σk1I1k(t),(2)∂n2(t)∂t=aI2(t)[n1(t)+n2(t)]+σk2I2k(t).(3)

Here I1(t) and σk1 are time-dependent laser intensity and the k-photon absorption cross section of the FP, while I2(t) and σk2 are related to the SP. In this case, the absorption of the SP is highly dependent on the FP laser fluence. Stuart et al. reported that for a 100-fs laser interaction with dielectric material, only the last part of the laser pulse experiences strong absorption or reflection [24]. Thus, if the SP can be introduced in the last part of the FP, the absorption of the SP should be increased considerably. These free-electron density changes can be described by the femtosecond laser-generated temporal distribution of temperature (for a detailed calculation result of free-electron density at single-pulse and double-pulse irradiation, see Appendix A). As shown in Fig. 1, the temperature distribution of the temporal double-pulse irradiation can be considered in three cases, including the FP laser fluence being higher than ablation threshold, lower than ablation threshold, and lower than the melting threshold. Here, T_m and T_v are the melting temperature and vaporization temperature, respectively. While the FP laser fluence is higher than the ablation threshold, the SP irradiates on the material surface at a time delay (within the pulse duration) after the FP ablation. The ablation results in the change of surface roughness, which leads to the increase of absorption of the SP laser energy, as shown in Fig. 1(a). On the other hand, while the laser fluence of a subpulse is slightly lower than surface ablation, the FP can only melt the target. In this case, the material is required to absorb enough latent heat for phase change from liquid to vapor. While the material’s absorption coefficient of the liquid phase is typically much higher than that of the solid phase, the energy absorption of the SP enhances after the FP irradiation, as shown in Fig. 1(b). If the phase difference-induced light absorption increase can overcome the latent heat, the double-pulse irradiation can generate considerable enhancement of surface ablation. This phase-change-induced absorption enhancement allows two subpulses, with laser fluence lower than ablation threshold to achieve effective ablation. However, if the laser fluence of the subpulse is too low, there is a weak melting effect induced by the FP irradiation, which is not beneficial for the surface ablation enhancement, as shown in Fig. 1(c). Thus, the temporal femtosecond laser double-pulse-induced ablation should exhibit different behaviors under different laser fluences of the FP.

Based on the above theoretical prediction, the surface ablation experiments are designed and carried out by an 800 nm Ti:sapphire femtosecond laser (Mira 900, Coherent) of 150-fs pulse duration, repetition rate of 80 MHz, and 3 W maximum average power. The double pulses are obtained by a modified Michelson interferometer. As shown in Fig. 2, the laser beam is divided into two parts by a polarizing beam splitter (PBS) and then guided into two arms. One of the two pulses is temporally delayed with respect to the other one by controlling the optical path (arm) length using a precise linear translation stage. In this way, temporal delays Δτ between 0 and 33 ps at a resolution Δτ of ∼66fs are realized between the two parallel polarized pulses. The time delay of temporal double pulses in this study ranges from 0 to 594 fs. A shutter is located in front of a half-wave plate to achieve the desired number of double pulses or single-pulse irradiation. In our experiment, the exposure time is 0.5 s, which means the pulse number is 4×107. As the reference for the same total laser power of double-pulse laser ablation, the single-pulse ablation is processed by covering one arm of the interferometer under the case of twice the laser power compared to the double pulse. The half-wave plate allows for tuning laser fluence and different translation stages to adjust the delay individually in each arm. The isolator prevents the laser damage due to strong reflective light from the sample. The two beams are recombined by the other PBS and focused normally onto the target in ambient air by an objective lens of f=3.7mm with a Gaussian spot of a radius (1/e2) of ∼100μm. The laser ablation is systematically performed on the 20-μm thickness PCL film. Owing to the biocompatibility and biodegradability with good transparency (∼90%) of PCL, it has been extensively studied for biological engineering applications such as nanofibers, implants, and drug delivery. The flexible mechanical properties (Young’s modulus, elasticity, tensile strength, elongation at break value) of PCL make it suitable for biomedical applications as flexible surface enhanced Raman scattering (SERS) substrates [25,26]. In addition, the surface hydrophilicity of PCL can be modified under excimer laser-induced tissue engineering [27]. The polymer surface is mounted on a computer-controlled x–y–z 3D translation stage and positioned perpendicularly to the beam propagation direction in the focal plane. A laser scanning confocal microscope (LEXT OLS 4100, Olympus) and a surface profiler (Alpha-Step 500, Tencor) are employed to characterize the surface morphology of the ablated samples. In this study, the surface ablation volume is estimated by considering both the depth and diameter of the ablated area depending on the volume formula of the elliptic cylinder: V=π·a·b·h, or elliptic cone: V=π·a·b·h/3, where h is the ablation depth, and a and b are semimajor and semiminor axes of ellipse, respectively.

The laser ablation is systematically performed on the 20-μm thickness PCL films. Prior to the double-pulse experiment, the single-pulse ablation is studied. Figures 3(a)–3(f) show the surface ablation results by the single-pulse femtosecond laser irradiation at a laser fluence of 80-160μJ/cm2. The Gaussian distribution of laser irradiation results in ablation and drills holes in the center under 96-160μJ/cm2 laser irradiation. Due to the low melting temperature of PCL (about 60°C) and large pulse number, there is an obvious heat-affected zone surrounding the ablation crater. Most of the central craters are the region where the material is completely removed from the PCL film. No surface ablation is observed when the laser fluence is lower than 80μJ/cm2. The corresponding horizontal and vertical cross-section profiles of these irradiated samples are shown in Figs. 4(a)–4(d). It can be observed that most of the center hole structures penetrate through the PCL film. In this case, the material removal can also be estimated based on the variation of ablation width. In addition, the anisotropic ablation effect in the horizontal and vertical direction is attributed to the laser spot’s energy distribution. For the single-pulse ablation, as shown in Figs. 4(a) and 4(c), there is an obvious hump on the edge due to heat effect-induced resolidification. However, Figs. 4(b) and 4(d) show that the edges in the ablation area are flat by using double-pulse ablation, which indicates the double-pulse-induced surface ablation is stronger compared to the single-pulse ablation.

The variation of ablation volume versus laser fluence for a single-pulse regime is shown in Fig. 4(e). In this case, the depth of the hump should be neglected. Thus, the depth is estimated as the distance from the baseline y=0 to the bottom of the hole. It can be observed that the ablation volume increases from 0.32×10−5 to 4.5×10−5mm3, since laser fluence changes from 80 to 160μJ/cm2. For the double-pulses ablation, at a time delay of 264 fs, it can be found that most of the ablated craters have larger diameters with respect to the single-pulse crater, as shown in Figs. 3(g)–3(l). The corresponding heat-affected zones also become bigger. The volumes of the double-pulse laser ablation with total laser fluence from 80 to 160μJ/cm2 are 0.89×10−5, 4.46×10−5, 5.32×10−5, 6.43×10−5, 7.17×10−5, and 8.46×10−5mm3, respectively, which are much larger than those for the single-pulse ablation, as shown in Fig. 4(e). For the irradiation of 160μJ/cm2 total laser fluence, the laser fluence of each subpulse is higher than the ablation threshold (80μJ/cm2). In this case, the SP arrives at the polymer after the FP ablation. The absorption of the SP laser energy by the polymer is enhanced due to the FP laser-induced surface modification, which results in a maximum enhancement of ablation efficiency of about (9.06−4.5)/4.5×100%=101% (for detailed ablation results, see Appendix A). However, for the FP laser fluence lower than the ablation threshold at 80μJ/cm2, for instance, the ablation volume induced by 112μJ/cm2 total laser fluence (56μJ/cm2 for each subpulse) irradiation of the polymer can reach about 5.91×10−5mm3 at a time delay of 132 fs (for detailed ablation results, see Appendix A), which is 4.38 times the 112μJ/cm2 single-pulse-induced ablation (1.35×10−5mm3). Thus, the corresponding maximum increase of ablation efficiency can reach about (5.91−1.35)/1.35×100%=338%.

Figures 5(a)–5(f) show the ablation enhancement factors by double-pulse femtosecond laser ablation at different laser fluences. All these figures are plotted with the same y-axis scale for easy comparison. The ablation enhancement factor is defined as σ=ViVo, where Vi and Vo are the ablated volumes by double-pulse and single-pulse ablation, respectively. It can be observed that most of the ablation by double pulses can achieve obvious enhancement with respect to the single-pulse ablation. The ablation enhancement factor is mostly larger than 1 within a 300-fs time delay, which means the double-pulse-induced ablation is more effective compared to single-pulse ablation at these time delays. However, the ablation enhancement factor reduces and is lower than 1 at time delay >330fs. In this case, the double-pulse femtosecond laser-induced surface ablation becomes weaker than single-pulse ablation. Furthermore, the ablation enhancement factor decreases obviously when the total laser fluence is too high at a laser fluence of 160μJ/cm2 [Fig. 5(f)] or too low of a laser fluence of 80μJ/cm2 [Fig. 5(a)]. For the 160μJ/cm2 double-pulse laser ablation, the FP of 80μJ/cm2 can generate the surface ablation. Then the absorption of the SP (80μJ/cm2) is enhanced by the FP-induced surface modification. However, for the subpulse with insufficient laser fluence for the surface ablation (lower than 80μJ/cm2), the FP can only melt the polymer surface. In this case, the material is required to absorb enough latent heat for the phase change from liquid to vapor. However, the absorption coefficient (α) of nonlinear optics materials is described by the equation: α≈2k·(1−n), where k is the wavenumber related to the wavelength, and n is the material’s refractive index [28]. In our experiment, the wavelength is 800 nm for both the FP and the SP, which means the absorption coefficient is inversely proportional to the material’s refractive index. For the PCL, the refractive index decreases with the temperature (n=1.463 for 293.15 K, and n=1.455 for 313.15 K) [29]. Thus, the absorption capability of melted PCL is higher than that of the solid PCL, which means the energy absorption of the SP on PCL film enhances after the FP irradiation. This enhancement of absorption can overcome the latent heat and generate considerable surface ablation. Therefore, the double-pulse laser irradiation of a total laser fluence ranging from 96 to 128μJ/cm2, as shown in Figs. 3(b)–3(d), can generate strong ablation enhancement compared to the case of 144 and 160μJ/cm2 [Figs. 5(e) and 5(f)]. However, under each subpulse of 40μJ/cm2 laser fluence ablation, as shown in Fig. 5(f), the enhancement of surface ablation becomes weaker. The lower laser fluence of the FP results in a weak melting effect, which leads to low absorption of the SP laser energy. In this case, the 80μJ/cm2 double-pulse ablation is similar to the single-pulse ablation at a time delay <330fs. By increasing time delay from 330 to 594 fs, the surface ablation decreases abruptly. Therefore, the surface ablation of PCL film can be well tuned by the laser fluence of the FP.

Figures 6(a)–6(c) show the single-pulse-induced surface ablation at the laser fluence from 32 to 48μJ/cm2. It can be observed that no very clear melting effect can be observed while the laser fluence is lower than 40μJ/cm2. Thus, the melting threshold is about 40μJ/cm2. The surface ablation can be further tuned by changing the ratio of laser fluence between the FP and the SP at the same total laser fluence. With a total laser fluence of 112μJ/cm2, the laser fluence of the FP is set to 32, 48, 56, 64, and 80μJ/cm2; meanwhile, the corresponding laser fluence of the SP is 80, 64, 56, 48, and 32μJ/cm2, respectively. As shown in Fig. 6(d), for an FP of 32μJ/cm2 and SP of 80μJ/cm2, the ablation enhancement can only be achieved at time delays from 0 to 66 fs, while the double-pulse ablation at time delays from 132 to 594 fs shows lower ablation volumes compared to the single pulse. In this case, the melting due to the low FP laser fluence is weaker; thus, the following pulse cannot generate considerable ablation. However, under the FP of 80μJ/cm2 and the SP of 32μJ/cm2, the ablation enhancement can be achieved at the time delay of 0–330 fs. In this condition, the 80μJ/cm2 FP results in surface ablation, which is beneficial to increase the absorption of the SP. Thus, 80μJ/cm2 FP and 32μJ/cm2 SP can achieve ablation enhancement within a wide range of time delay (66 to 330 fs). As shown in Fig. 6(e), the laser fluence of the FP is further changed to 48μJ/cm2 and 64μJ/cm2 at a total laser fluence of 112μJ/cm2. 48 and 64μJ/cm2 are close to the ablation threshold, and a considerable ablation enhancement can be achieved within 330 fs and then reduced with a time delay >330fs, which is in accordance with the results shown in Fig. 5. Therefore, temporal overlapping double-pulse irradiation can reduce the ablation threshold for the femtosecond laser ablation with a short time delay. The ablation enhancement can be well tuned by time delay and laser fluence ratio between the two subpulses.

In this paper, the surface ablation enhancement of temporal overlapping double-pulse femtosecond laser irradiation is studied at different time delays and laser fluences. The ablation enhancement mechanism is investigated by the temperature distribution model and experiment results. Compared to the single-pulse femtosecond laser ablation, the double-pulse femtosecond laser ablation can be enhanced mostly within twice the pulse duration, and then decreases with time delay. When the laser fluence of the FP is higher than the ablation threshold, the enhancement of the laser-induced surface ablation is attributed to the absorption increase induced by the FP. In the case of the subpulse laser fluence lower than the ablation threshold, the considerable ablation enhancement can also be achieved by the FP-induced absorption improvement because of the increase of avalanche ionization. The surface ablation enhancement can be further tuned by changing the ratio of two subpulse laser fluences. Under the same total laser fluence, a high laser fluence for the FP can generate more effective surface ablation. This approach provides a novel way and understanding of the efficiency increase for ultrashort laser processing.

Acknowledgment. We wish to thank Dr. Yang Li and Dr. Huagang Liu for the discussions on double-pulse laser experiments. The support provided by China Scholarship Council (CSC) during a visit of Zhenyuan Lin to the National University of Singapore is acknowledged.