1. Introduction

Ultrashort laser pulses offer the possibility of high precision manufacturing of microstructures in various kinds of materials due to their extreme peak intensities and short interaction timescales. Therefore, the laser material interaction process has been subject to detailed experimental as well as theoretical investigations since the mid 1990s. These investigations cover the influence of principle laser parameters, like pulse duration and pulse energy [1–8] or wavelength [9] on the threshold fluence and average ablation depth per pulse. A temporal analysis of the surface ablation process is possible by shadowgraphy imaging or transmission sensing of the ablation products and plasma separating from the surface. In addition, the study of velocity, size and distribution of the ablated particles provides an insight into the material decomposition mechanisms [10–14]. Analytical and numerical models have been developed to describe the experimental observations [15–20]. While these investigations reveal the physical process of the laser material interaction for ultrashort pulses, they are all related to the ablation process at the materials surface only.

For many industrial applications, like laser drilling, however, the ablation process in the depth of the material is important. In contrast to surface ablation the pulse propagation as well as the removal of the ablation products are modified within the capillary and therefore the process parameters are subject to change with increasing depth [21,22]. In addition the erosive effect of the generated plasma might act as a secondary ablation mechanism. This effect is generally present, also in single pulse ablation at the surface [23], but its importance can significantly increase within a capillary [24,25]. Further examples for this problem are the influence of the polarization on the hole cross-section [26] or even the complete stop of the ablation process at a certain depth. However, a direct observation of the ablation process inside the hole is not feasible when dealing with non-transparent materials like metals. The simplest approach to gain insight is the measurement of the breakthrough-time for a certain sample thickness as a measure for the average ablation rate [27]. In addition, several other methods have been developed to study the ablation process in the depth of the sample. Most of them involve polishing the sample until the longitudinal section of the drilled hole becomes visible [28,29]. This laborious technique requires a high level of experience, but it enables the access to the complete hole shape and the materials composition [30]. Despite that only a stepwise evaluation but no in situ observation of the drilling progress is possible with this approach, which means different holes need to be drilled to varying depths, requiring a stable and reproducible processes for each step. Another possibility is drilling close to the side surface of the sample (distance < 100 µm) [31]. The observation of the bulging of the side surface allows a time-dependent investigation of the drilling depth and speed, but gives no information about the cross-section of the hole and offers only limited resolution. A precise in situ depth-profiling with scan rates in the kHz regime is possible by optical coherence tomography (OCT) [32]. This technique is suitable for online process control [33], though other feature information beyond depth is only limited. The preparation of a sandwich structure of glass – metal plate – glass enables the in situ imaging of the evolution of the ablated material inside the capillary (a thin slit in the plate acts as the hole) [34]. However, here the ablation takes place in an artificial environment which does not correspond to the real drilling process. Furthermore, it is limited to few pulses due to the redeposition of material on the glass surfaces which consequently become non-transparent.

In this paper we demonstrate the use of silicon to investigate the ablation process in the depth of the material. Silicon has the advantage to be still transparent when illuminated with light at a frequency below or at the band edge (i.e. at 1.06 µm) which can at the same time be detected with standard devices based on silicon technology. When laser radiation above the band edge is used for the ablation process, the single photon absorption process is dominant and the material appears opaque to the ablation laser. This way, detrimental side effects, like nonlinear propagation or absorption effects which can lead to a different interaction/ablation regime, when e.g. glass is drilled, can be avoided. Thus, although silicon is a semiconductor, it can be considered as a model system for the ablation of non-transparent solids (including metals). The ablation behavior and the process of matter removal of silicon has already been studied, showing the similarity to those of metals in terms of ablation rate and threshold [35], characteristics of the ablation products [12] and simulation of the involved mechanisms [18].

2. Experimental setup

In our experiments we used a silicon sample with a width of 500 µm (typical wafer thickness). With optically polished surfaces we measured a transmission at perpendicular incidence of about 26% for the illumination wavelength of 1060 nm. This is mainly due to the strong Fresnel reflection of approx. 31% at each surface, because of the high refractive index of 3.5. However, this transmission still allows a trans-illumination concept for the observation of the drilling process. Figure 1 shows a schematic diagram of the setup.

 

Fig. 1 (Color online) Principle experimental setup for trans-illumination imaging of the laser drilling process in the silicon sample.

Download Full Size | PDF

For the illumination of the sample a fs laser oscillator (HighQ femtoTRAIN) with a wavelength of 1060 nm, a repetition rate of 82 MHz and an average power of 140 mW is used as the light source. The use of an ultra short pulsed laser opens the additional possibility of measurements and imaging of the ablation process with high time resolution (e.g. to trace ablation particles etc.). In this case a certain delay between the ablating laser pulse and the illuminating pulse as well as a pulse selection mechanism is required in addition. However, such schemes are outside the scope of this paper and will be discussed elsewhere. A telescope optic ensures an almost homogeneous illumination of the sample. The image of the sample is projected onto the CCD camera by a microscope objective (NA 0.1). Hence the magnification and the corresponding resolution can be adapted based on the focal length of the objective and the distance to the detector. We used a standard CCD camera (Basler scA1000-20fm) with a silicon based sensor. Since the 1060 nm radiation is at the band edge of silicon the sensitivity of the sensor is still high enough to yield a high quality image. An additional filter (Thorlabs FEL1050) in front of the camera is to be used in order to block scattered light from the drilling process, which would otherwise glare the whole image.

For the drilling we used a laser system (Trumpf TruMicro5050) with a wavelength of 1030 nm, a pulse duration of 8 ps and pulse energies up to 125 µJ at 400 kHz repetition rate. The actual pulse energy and repetition rate is controlled by an internal electro-optic modulator. The laser beam is circularly polarized due to a quarter wave plate and focused on the sample surface perpendicular to the illuminating beam. We used a meniscus lens with a focal length of 100 mm, resulting in a spot size of ca. 25 µm.

3. Experimental results

Figure 2 shows as an example a drilling process recorded with our technique. Here we used 25 µJ pulse energy, which corresponds to a fluence of 8 J/cm², at a repetition rate of 100 Hz in the percussion drilling geometry. The borehole locally reduces the transmission of the sample, because it induces additional surfaces where the light is scattered or reflected due to Fresnel reflection. Hence, the longitudinal section of the hole is represented by the dark area in the images. The images are miscolored to highlight the outline of the borehole. The number of pulses applied to the sample N is used as a reference to the progress of the drilling process. In Fig. 2 only steps with a significant change in the shape of the borehole are represented to illustrate its evolution.

 

Fig. 2 (Color online) Trans-illumination images of the drilling process. The number of pulses applied is given below each picture. Arrows indicate the occurrences of bulges, change in drilling direction and the formation of multiple hole ends. Video online, with linear timescale (Media1).

Download Full Size | PDF

For the first N=500 pulses, see Fig. 2(a-e), the percussion drilling forms the typical tapered shape with a funnel-shaped entrance due to the Gaussian beam profile of the drilling laser. At the bottom there is a hemispherical dome. The average ablation rate is about 500 nm/pulse for the first 200 pulses and thereafter decreases to an average of 120 nm/pulse for the period from N=200 to N=500. The entrance diameter is approximately 25 µm and the depth 125 µm at this time, corresponding to an aspect ratio of about 5.

In the further progress the forward drilling continues, however at a significantly decreased drilling rate (ablation depth per pulse is 50 nm/pulse and 30 nm/pulse for Figs. 2(f) and (g), respectively). In addition there is also ablation which now occurs at the side walls of the hole, leading to an increased hole diameter at a random depth, see the arrow in Fig. 2(f). This effect might be caused by internal reflections within the capillary or the erosive influence of the generated plasma [23,24]. The bulges may increase by subsequent pulses and also new ones may occur (compare Figs. 2(f) and (g)). Such effects are not traceable using cut-out polishing techniques of different holes described above.

At N=3200, see arrow Fig. 2(h), the internal reflections at the bottom of the hole as well as the deviation of the beam due to the interaction with particles and plasma lead to a change in the direction of the drilling process which is now no longer determined by the incident laser beam itself. At random the drilling can proceed in different directions, leading to the formation of multiple borehole tails, see Fig. 2(i) and arrow in Fig. 2(j). The subsequent observation of the drilling process up to N=15000 pulses, see Fig. 2(k), showed no further increase in the depth of the hole. The diameter of the hole increases slightly and the internal walls are eroded due to the ablation caused by the side wings of the Gaussian beam profile. Therefore the central part of the hole in Fig. 2(k) looks more cylindrical compared to the tapered shape in Fig. 2(j) and before. The fluence in the wings of the beam is much lower than in the middle and the ablation rate is low [4,33], therefore this is a slow effect which still continues even at a high number of applied pulses. It will stop when the diameter of the hole has reached a size where the fluence at the hole walls drops below the ablation threshold.

4. Conclusion

The concept of transmission imaging of silicon samples can be used as a feasible technique to directly observe the borehole shape evolution in (ultra) short pulse drilling where features of micrometer size need to be visualized. We exemplarily recorded the drilling process for a hole with a final entrance diameter of approx. 35 µm and a maximum depth of 250 µm as a proof of principle. This technique is capable of visualizing special features and defects occurring in a certain depth or when a certain aspect ratio is reached. This includes bulges on the side walls, changes in the direction of the drilling progress and the formation of multiple hole tails at the bottom. All these observations are in agreement with the previous investigations of the hole shape in metals, where a cut-and-polish post processing technique was used [21,28,29,35]. These effects have been attributed to internal reflections, the influence of the capillary on the pulse propagation within the borehole and the interaction of the laser beam with the plasma and/or ablated particles. However, a direct observation and therefore a direct proof of these explanations was so far prevented by the complex post-processing techniques required.

In the future this imaging concept can be used to reveal the processes which appear in the depth of the material due to the high aspect ratio of the structures under investigation. This involves the ablation mechanism as well as the material removal and possible interaction with following laser pulses. It will help to develop a more detailed understanding of the ablation process and lead to a better control of processing parameters in order to improve the quality of laser micromachining. With regard to applications, the obtained information may appear extremely valuable to increase the processing speed and to improve the hole quality.