Experimental verification of femtosecond laser ablation schemes by time-resolved soft x-ray reflective imaging

1. Introduction

Femtosecond laser irradiation above the ablation threshold induces the extreme condition far from thermodynamical equilibrium at high temperature, high pressure with high density charge carriers [1]. These extreme conditions cannot easily be achieved by other experimental techniques, and expected to lead to a novel fabrication technique for designing new materials. In addition, femtosecond laser processing can be applied to the micro fabrication for various materials such as metals [2, 3], semiconductors [4], and insulators [5]. To make the femtosecond laser processing technique more accurate and controllable, the understanding of the dynamics of the laser ablation process is important. However, in particular, the dynamics of the femtosecond laser ablation is still not clear at the early stage. Several studies on the time-resolved imaging of femtosecond laser ablation process have been performed on various materials [6–9].

The numerical simulation study of femtosecond laser ablation process has been carried out by some groups. Lewis et al. simulated the femtosecond laser ablation process by using Lennard-Jones potential based on the two-dimensional molecular dynamics model [10]. They also carried out the three-dimensional simulation of the femtosecond laser ablation process in insulator (silicon) [11]. They categorized the ablation process into four types according to the magnitude of fluence, that is, the spallation, homogeneous nucleation, fragmentation, and vaporization. Garrison et al. also reported the modeling of short pulse laser ablation by using Lennard-Jones potential and discussed the threshold behavior of ablation [12]. More recently, Zhigilei et al. simulated short pulse laser ablation of metals [13]. They indicated that melting, spallation, and phase explosion are taking place simultaneously and are closely intertwined with each other.

The dynamics of the surface morphology has been extensively studied by optical reflectivity and Newton ring-like interference with a time resolution of picosecond by Sokolowski-Tinten et al.[8]. Dilation of the ablation front at a speed of sound velocity was observed. Compared with visible light, soft x-ray has extremely short wavelength and the penetration depth of soft x-ray is very shallow. Therefore, soft x-ray is the most suitable to observe the nanometer scale morphology of the solid surface. Lindenberg et al. observed the formation of nano-bubbles in femtosecond laser ablation by using the time-resolved hard x-ray diffuse scattering [14]. Barty et al. observed the ablation dynamics of free-standing SiN films by using the soft x-ray beam from the free electron laser [15]. The ablation process and the produced nanoparticles was also studied by inner-shell transitions [16–18]. As for the surface morphological change during the femtosecond laser ablation process, we have reported the observation of the surface expansion in the early stage of femtosecond laser ablation by using soft x-ray interferometer [19].

The change in optical reflectivity of laser excited material is caused not only by the change of bulk properties due to excited electrons or phase transition (melting), but also by surface deformation and morphology in nanometer scale. The reflectivity of soft x-ray does not depend on the details of the lattice structure or chemical bonds but depends simply on the spatial distribution of the atomic density and the roughness of the interface. The reflectivity depends critically on the material density within ten to twenty nanometers from the surface. Another remarkable advantage of soft x-ray is its high sensitivity on the roughness, owing to the short wavelength [20]. According to X-Ray Database [21], when the surface roughness increases from 1 nm to 5 nm, the reflectivity reduction exceeds 80 % for our experimental condition. In addition, 20 nm-thick density transition region also causes the reduction of reflectivity of about 80% for the same condition. This high sensitivity of the reflectivity of soft x-ray to surface morphology brings us rich information about the temporal behavior of the ablation front, which is useful for understanding femtosecond laser ablation dynamics. In this paper, we demonstrate a pump and probe reflective imaging of the platinum (Pt) surface during the femtosecond laser ablation by using the laser-driven plasma induced soft x-ray laser (SXRL) as a probe beam.

2. Experimental

The laser-driven plasma soft x-ray laser in Japan Atomic Energy Agency (JAEA) was used as the probe beam [22]. The system provides 7 ps soft x-ray pulses, and it delivers soft x-ray photons at 89.2 eV (13.9 nm) with a photon number exceeding 10⁹ (equivalent to 50 nJ/pulse as the output of the laser). Owing to this bright laser pulse, it is possible to capture the ultrafast phenomena in single-shot measurement. Figure 1 shows the schematic setup of the femtosecond laser pump and the soft x-ray probe microscopy. The reflective measurement was carried out in vacuum with a pressure lower than 10⁻³ Pa. The soft x-ray probe beam with a wavelength of 13.9 nm from the plasma was focused in front of a sample by the concave mirror. The sample image was transferred onto the detector by the imaging mirror with a magnification factor of 19.1. The injection mirror was placed before the sample in order to adjust the incident grazing angle θ (24.5 degrees) for the sample. The reflectivity is estimated about 30 % for the soft x-ray laser. The detector was a back illuminated charge coupled device (CCD) for soft x-ray (Princeton Instruments, PIXIS-XO:2048B).

 

Fig. 1 Schematic of the experimental setup for near IR pump and soft x-ray probe microscopy. The soft x-ray probe beam was focused in front of a sample by the concave mirror (imaging mirror A) and the sample image was transferred onto the detector by the concave mirror (imaging mirror B). Femtosecond laser beam was focused by a lens onto the sample surface, and the pumping energy was adjusted by using a combination of a half-wavelength plate and a polarizing beam splitter

Download Full Size | PPT Slide | PDF

The pumping laser used for ablation was a Ti:Sapphire laser system based on chirped pulse amplification (Thales laser, alpha10). The laser emitted 80 fs pulses of linearly polarized light at a central wavelength of 795 nm. The emitted pulses were focused by a lens (f = 600 mm) onto the sample surface at nearly normal incidence. The pumping energy was adjusted by using a combination of a half-wavelength plate and polarizing beam splitter. To evaluate the pulse energy on the sample for each shot, a fraction of the incident pulse was sampled by a photodiode joule meter (OPHIR, PD10-SH-V2). Figure 2 shows the spatial beam profile of the pump beam at the sample surface measured by a beam analyzer. The spatial profiles of the cross sections along x–x′ and y–y′ lines are also shown. The open circles and the solid curves show the measured and the fitting results with Gaussian functions, respectively. The Gaussian fitting was in good agreement with the experimental data with the accuracy better than 1.5 %. The focal spot size (1/e) on the sample surface was measured to be about 73 μm. The typical pump energy, peak fluence, and excitation intensity on the sample surface were 170 μJ, 4.1 J/cm², and 1 × 10¹⁴ W/cm², respectively.

 

Fig. 2 Beam profile of pump (Ti:Sapphire laser) pulse measured by SPIRICON beam analyzer. The spatial profiles of the cross sections cut along x–x′ and y–y′ lines are also shown. The open circles show measurement results and the solid curves show the result of Gaussian fitting.

Download Full Size | PPT Slide | PDF

The oscillators of Ti:Sapphire laser system and soft x-ray lasers were electronically synchronized in order to control timing between the pump and probe pulses. The time origin was determined by simultaneous observation of the signals of soft x-ray and multiphoton photoemission signal of the pump pulse, using an X-ray streak camera (Hamamatsu Photonics, C4575-01). The sample was Pt thin film with 300 nm thickness evaporated on fused silica substrates. We chose platinum for the sample because of its high reflectivity to the soft x-ray. The initial surface roughness was measured by atomic force microscope (AFM) and the root mean square (RMS) deviation was found less than 1.0 nm.

3. Results and discussion

Figures 3(a)–3(g) show the soft x-ray reflective images of a Pt surface at different delay times after the irradiation of the pump pulse with the peak fluence of 4.1 J/cm². The scale-bar (50 μm) shown in Fig. 3(a) applies to all images, while the contrast was optimized for each image. The intensity non-uniformity observed outside the disk-shaped dark area stems mainly from the spatial non-uniformity of the soft x-ray. The image at t = +∞ observed by an optical microscope is also shown in Fig. 3(h). In this figure, the height of the central part of the circular disk is lower than the initial surface, which is surrounded by the ring-like protrusion (so-called ablation rim, shown by an arrow). We define the ablation crater as the area inside this protrusion. At 10 ps (Fig. 3(b)), a disk-shaped dark area was found on the center of irradiated area. This means that the ablation phenomena already started at t = 10 ps. The reflectance outside of the disk-shaped dark area gradually decreases and reaches the value in the central area at 160 ps. The diameter of disk-shaped area at t = 160 ps coincides with that at t = +∞. Since the height of the surface dilation associated with the ablation process is less than 100 nm [19], which is three orders of magnitudes smaller than the spot size, the observed behavior at each radial position is regarded as the phenomenon simply dependent on the local fluence at that point. The sharp contrast between the central dark area and the surrounding bright ring seen at t = 40 ps indicates that the ablation dynamics changes drastically at the fluence at the boundary. The size of disk-shaped dark area at t = 10 ps was about 75 % of the size of the dark area at t = +∞, which corresponds to the crater, and grows slowly to the final size till 160 ps.

 

Fig. 3 (a)–(g) are snapshots of the sample surface probed by soft x-ray laser at different times after exposure to the pump pulse (Peak fluence : 4.1 J/cm²). The arrow under (h) shows the direction of the incident soft x-ray. To overcome the spatial fluctuation of the X-ray probe, the contrast has been optimized to better visualize their most characteristic features. The scale-bar shown in (a) applies to all images (a)–(h). The intensity profiles along the dotted lines are shown in Fig. 4, where points A and D correspond to the left and right ends of the dotted line. The yellow arrows shown in (b) and (c) indicate the ring position. The yellow arrows shown in (h) shows the edge of the ablation crater. For comparison, the image at t = +∞ observed by an optical microscope is also shown in (h).

Download Full Size | PPT Slide | PDF

We have plotted the horizontal cross-sections of the snapshot in Fig. 4 along the dotted lines in Figs. 3(b)–3(f). These plots were normalized to the averaged values between C and D, where the significant reflectivity change was not observed. The difference in the temporal behavior of the reflectivity indicates that the observed phenomena are strongly fluence dependent. In order to discuss the fluence dependence in detail, we label each radial point, which reflects the local fluence, from A to D as shown in Fig. 4(a). In strongly excited area (from A to B), drastic and fast reflectivity drop was observed. In moderately excited area (from B to C), the decay of the reflectivity is slower than that in the strongly excited area, and the reflectivity reaches its minimum at t = 160 ps. In weakly excited area (beyond the point C toward D), laser-induced reflectivity change was not observed. In addition, the point C, where the reflectivity dip was observed at t = 10 ps and t = 40 ps, coincides with the position of the edge of reflectivity drop at t = 160 ps. This drop corresponds to the ring structure distinctly seen at t = 10 ps and t = 40 ps in Fig. 3.

 

Fig. 4 (a) The fluence distribution of the pump beam. The classification of the laser fluence is also shown in this figure; (i) strongly excited area (between 1.2 J/cm² (B) and 4.1 J/cm² (A)), (ii) moderately excited area (between 0.5 J/cm² (C) and 1.2 J/cm² (B)), and (iii) weakly excited area (below 0.5 J/cm² (C)). (b) The reflectivity at a grazing angle of 24.5 degrees is plotted as a function of the radial position (fluence dependence).

Download Full Size | PPT Slide | PDF

To evaluate the reflectivity change in the area between B and C, the time evolution of the relative reflectivity at the position of 47.1 μm was plotted as a function of the delay time in Fig. 5. The dotted line shows the exponential fit to the data with a decay time of 36 ps. The fitting result reproduces well the experimental data.

 

Fig. 5 Temporal evolution of the reflectivity is shown in moderately excited area. These values are obtained at the position of 47.1μm, and normalized by the reflectivity of the weakly excited area between C and D. The dotted line shows the exponential fit to the data assuming the decay time of 36 ps.

Download Full Size | PPT Slide | PDF

In the following, we will discuss the origin of the reflectivity drop at t = +∞. As the depth of the crater is 150 nm (one half of the total thickness), the low reflectivity of Si substrate has no contribution to the reflectivity drop. The origin of the reflectivity drop is ascribed to the random nano-structure at the bottom of the crater as discussed below. It is reported that the femtosecond laser induces nanometer-scaled roughness on metal surface [23, 24]. We examined the surface morphologies inside the crater at t = +∞ by AFM. The typical values of the surface roughness were several tens of nanometers, and the correlation lengths along the surface plane of the roughness were smaller than 1 μm. For evaluating the X-ray reflectivity of a rough surface, the Nevot-Croce factor is known to be a good approximation when the roughness spectrum has a high spatial frequency (>1μm⁻¹) [25]. According to this formulation, we estimated the reflectivity to be negligibly small (less than 10⁻⁴⁰). Thus, we ascribed the observed reflectivity drop at t = +∞ to the surface roughness.

Next, we will discuss the origin of the transient soft x-ray reflectivity change. As the origin of the change of the intensity of the reflected soft x-ray, there are several possibilities, that is, ejected plasma, nanoparticles and change of the surface morphology of the sample. As the density of the plasma is very low, its effect was evaluated as negligible at similar experimental condition [19]. The generation of nanoparticles of the target material is also the possible origin of the observed reflectivity drop. However, the density of nanoparticles is not so high as to disturb the transmission of soft x-ray [26–28]. Thus, we suppose that the observed reflectivity drop is caused by the surface morphologies of the target materials.

The reflectivity of soft x-ray strongly depends on the reduction of material density, the density gradient near the surface, and the surface roughness. Our observation showed that the temporal behavior of the soft x-ray reflectivity is distinctly different in three areas. In the area between A and B, the reflectivity drop appeared faster than the time resolution of our system (< 7 ps). In the area between B and C, the reflectivity drop is far slower (36 ps) than that in the strongly excited area. In addition, ring-like reflectivity drop with a relatively fast emergence is observed at t = 10 ps and t = 40 ps at the position C.

One possible explanation of our experimental results can be given by the model proposed by Zhigilei et al.[12, 13], where metal ablation is treated assuming electron and lattice temperatures independently (two-temperature model). In their simulation, separation of a liquid layer (spallation) from the substrate is observed above ablation threshold. The slow reflectance change observed in B–C area may correspond to this ablation scheme, because the surface of the liquid layer will keep relatively flat and sharp interface for a while after separation owing to a the large surface tension. The relatively slow dilation of the surface agrees with the interferometric measurements in our previous report [19]. This assignment explains also the sharp edge at B, because the threshold-like behavior is found in the simulation [12]. The phase explosion then corresponds to the strongly excited area (A–B).

The position of the ring-like structure (the point C) corresponds to the sharp threshold for spallation [12]. However, it is not possible to understand why the reflectance drop occurs in a very short time (within 10 ps) and why it occurs only in a very narrow fluence range, based on the existing simulations. More detailed numerical simulation for specified material and the calculation of the expected optical properties of the interface in the soft x-ray region is needed for the precise comparison between the theoretical model and the experimental results.

4. Conclusion

Ti:Sapphire laser pump and soft x-ray laser probe microscopy was carried out for the observation of femtosecond laser ablation dynamics on Pt film. By using Gaussian profiled beam, the fluence dependence of laser-induced phase transition process was deduced. The dynamical behavior of soft x-ray reflectivity showed discontinuous dependence on irradiation fluence. In addition, at t = 10 ps and t = 40 ps, the fine ring structure was observed at the radial position of the ablation crater edge. Establishment of the theoretical model for ablation process supported by experimental information will contribute to the prediction of the laser fabrication scheme by numerical simulation, and we are extending this experiment to other materials for the better understanding of the femtosecond laser ablation phenomena.