Open Access
Add to BibSonomyAbstract
In this study we present a new measurement technique to investigate the timescales of back side ablation of conductive films, using Molybdenum as an application example from photovoltaics. With ultrashort laser pulses at fluences below 0.6 J/cm2, we ablate the Mo film in the shape of a fully intact Mo ’disc’ from a transparent substrate. By monitoring the time-dependent current flow across a specifically developed test structure, we determine the time required for the lift-off of the disc. This value decreases with increasing laser fluence down to a minimum of 21 ± 2 ns. Furthermore, we record trajectories of the discs using a shadowgraphic setup. Ablated discs escape with a maximum velocity of 150 ± 5 m/s whereas droplets of Mo forming at the center of the disc can reach velocities up to 710 ± 11 m/s.
© 2013 Optical Society of America
1. Introduction
The rapid development of industrial-grade ultrafast lasers, which operate at repetition frequencies in the range of hundreds of Kilohertz [1–4] has led to an increasing number of applications of such systems in today’s production processes. With pulse durations in the fs to ps range, ultrafast lasers allow material processing with little thermal damage compared to longer pulses [5–7]. A novel application of such lasers is found in the photovoltaic (PV) industry for the production of Copper-Indium-Selenide (CIS) based thin film solar modules. These modules are build up on a Molybdenum (Mo) coated glass substrate where the Mo forms the back contact of the module. In order to create single cells within a module, the Mo has to be electrically separated into 5–10 mm wide stripes by creating 20–50 μm wide trenches. The removal of the Mo can be done with laser pulses which either directly impinge the Mo side or pass through the glass, provided that the glass is transparent for the laser wavelength applied. First investigations of the so-called ‘laser-scribing’ of Mo were described in [8–12]. The approach to ablate the Mo film from the back side (i.e. the laser pulse passes the transparent substrate) appears to be promising mainly for two reasons: the required pulse energies for entire removal of the film are much lower compared to direct ablation and the achievable process speeds are considerably higher [13,14]. The first report of the usage of such a back side process in mass-production [15] seems to confirm these findings. Despite the recent interest in this topic, the ablation process itself is not fully understood and this may prevent a broader application.
First fundamental studies of “laser-induced back ablation” [16] of thin metal films by ultrashort laser pulses were done already in the 1980s by Mayer and Busch for plasma studies during laser-metal-vapor interaction [17]. Bullock et al. used both ns and ps pulses [16, 18] for back ablation of thin aluminum (Al) films. More recently, Ashitkov et al. presented an experimental study on back side ablated Al films with high temporal resolution as well as theoretical considerations [19, 20]. Similar processes were used for the spatially controlled transfer of materials (e.g. laser induced forward transfer) from one substrate to another both for short [21–24] and ultrashort pulses [25, 26]. The back side ablation of Mo from glass was studied by Domke et al. who analyzed time scales during the ablation process using a pump-probe microscopy setup [27–29]. By analyzing Newton’s rings with this direct observation technique, the authors observed first a bulging of the Mo films in less than 1 ns after the laser pulse, followed by an expansion phase lasting several nanoseconds. The breakaway time when the bulge is large enough and the tensile stress at the outer part of it exceeds a certain value was determined to approximately 20 ns for fluences slightly above the threshold for lift-off [30]. Regrettably the Newton’s rings are superimposed by diffraction rings as the ablation causes a hole, which forms an aperture for the light of the probe pulse. The superposition as well as changing optical properties of the ablated material, which influence the spacing of the Newton’s rings, limit the achievable resolution of the technique used.
Recent investigations by our group using a shadowgraphic setup [31] were not able to reveal the first nanoseconds of the ablation process and could only provide a rough estimate for the breakaway time. We now present a new measurement technique free from optical limitations. Using a specifically designed test structure, we analyze the temporal change of the current flow along a narrow Mo ‘bridge’ during and after the interaction with the laser pulse. With this setup a temporal resolution in the order of 1 ns can be achieved. It does not only allow to determine the breakaway time at which the Mo is lifted off, but also gives insight into the timescales of heating up the sample, vapor formation and cooling.
2. Experimental setup
The samples used in these investigations are made of a 3 mm thick soda lime glass (SLG) substrate with a sputter-deposited molybdenum layer of thickness d ≈ 360 nm. We use a commercially available ultrafast laser source (Trumpf “TruMicro 5000” Series, τ ≈ 10 ps (FWHM), λ = 515 nm) which is focused onto the back side of the Mo layer. Please note that all fluence values below refer to average fluences calculated by where Ep is the pulse energy and 2w0 the beam width (1/e2-diameter) at the focal plane.
2.1. Shadowgraphic setup
The setup for shadowgraphic observation of the ablated Mo discs consists of two lasers, the ultrafast processing laser and a second one for illumination. A fast camera (PCO AG “hsfc pro”) is used for sequential image recording. This camera consists of 4 ICCD (intensified charge-coupled device) cameras with a minimum gate time of 3 ns which are coupled via beam splitters. The processing laser is focused onto the back side of the Mo layer using a f = 180 mm plano-convex lens. The beam width obtained at the focal plane was determined to 2w0 ≈ 33 μm using the method described in [32] via film side ablation of thin Mo layers. Note that in the case of back side ablation, the effective beam width at the SLG/Mo interface is smaller. This effect was already observed for λ = 1064 nm with a comparable pulse duration [33] and seems to be even more pronounced for the shorter wavelength used in this study. We think this effect is most likely caused by self-focusing inside the 3 mm thick SLG but this will not be discussed here.
The second laser, a pulsed, incoherent diode laser (Cavitar “Cavilux”, λ = 685 nm, τ = 2 μs) is spatially and temporally synchronized to the processing laser and illuminates the ablated Mo fragments perpendicular to the direction of the processing laser. The camera is positioned in the beam path of the illuminating laser and in this way the shadows of the ablated fragments are imaged by means of a 20x microscope objective. A simplified layout of our setup is shown in Fig. 1, details can be found in [34]. The fast ICCD camera system allows us to acquire 4 images of the same molybdenum disc with definable delays after the laser pulse and a temporal accuracy of ±1 ns between individual images. The point in time when the laser pulse of the processing laser hits the Mo sample (t = 0) can be determined with an uncertainty of about 3 ns by imaging the stray light of the pulse inside the glass. The average velocity of Mo objects can be determined by measuring their distance to the SLG surface on the recorded images. The uncertainty of the position measurements was estimated to about 2 μm.
2.2. Setup for electrical measurements
The setup for these measurements consists of the processing laser described above and a galvo scanner (Scanlab AG “IntelliSCAN”) with a f = 330 mm f-theta lens, leading to a beam width of 2w0 ≈ 39 μm at the focal plane. A constant voltage source, including a battery (UB = 2.6 V), a cascade of capacitors and a 50 Ω resistor, is connected to the test structure and ground (GND), respectively. The second connector of the test structure is fed into the 50 Ω input of a digital storage oscilloscope (DSO). In this way the voltage change over time, which is measured at the DSO to GND can be directly related to the change in resistivity of the test structure.
Figure 2 shows the circuit diagram of the setup. A similar electrical setup was used by Brunco et al.[35] for time-resolved temperature measurements during laser-matter interaction with ns lasers.
To achieve the desired temporal resolution in the order of 1 ns we use a fast DSO (“WaveRunner 606Zi”, LeCroy Corp., rise time < 600 ps, sampling rate up to 20 GS/s). Furthermore, it is important to take capacitances and inductances of the components used in the setup into account. For this reason, an Al-Ni-ceramic resistor (suited for frequencies up to 2 GHz) is chosen as 50 Ω resistor inside the constant voltage source and fast ceramic disc capacitors are used for voltage stabilization.
Figure 3 shows the test structure used for electrical measurements, which is scribed into a Mo coated SLG sample using the galvo scanner. The important part is the small Mo ’bridge’ at the center forming the only electrical connection from the left to the right part of the structure as shown in Fig. 3(a). After electrically connecting the structure according to Fig. 2 and Fig. 3(b), respectively, a small current flows over the Mo bridge and via the DSO to GND. Then a single laser pulse is released, which leads to the ablation of one Mo disc at the center of the bridge, see Fig. 3(c). During that ablation and the subsequent cooling phase, the voltage U(t) is recorded using the DSO. This data allows for calculating the change in resistivity of the Mo bridge by
Here RDSO denotes the 50 Ω resistor of the DSO, UB the voltage of the battery, U0 ≡ U(t0) the initial voltage and U(t) the voltage at a specific time t measured at the DSO respectively.
Fig. 3 a) Sketch of the test structure scribed into the Mo. The region at the center (red dotted) shows the interrupted scribe. b) Perspective view of the central region highlighted in a). The yellow arrow is pointing to the position where the laser pulse hits the Mo bridge. c) Microscope image of the central region (dash-dotted in b)) after the ablation.
We use a TTL signal provided by the laser at a defined time prior to the optical pulse to trigger the DSO. In this way the point in time when the laser pulse hits the sample (t0) can be determined with an uncertainty of < ±1.5 ns. This value is mainly limited by the jitter between the optical pulse and the TTL signal as well as the rise time of the DSO.
3. Results
Figure 4 shows a sequence of an ablated Mo disc recorded with the setup described in Sect. 2.1 using a laser fluence of F = 0.23 J/cm2. In the first picture, Fig. 4(a), recorded 22 ns after the pulse one can already see the bulged Mo coating. 400 ns later, in Fig. 4(b), the disc has moved about 44 μm away from the surface. Furthermore, small Mo droplets can be recognized in the right part of this picture. In Fig. 4(c) the Mo disc has turned and a small hole inside the disc can be recognized. This phenomenon was already reported in [31] and will not be discussed in this context.
Fig. 4 Sequence of an ablated Mo disc at different times after the laser pulse (t = 0) for F = 0.23 J/cm2. The position of the Mo surface is highlighted with the white line.
In order to determine the velocities of the escaping Mo disc and droplets, we measure their distance to the substrate at different delay times. In all cases we find a linear movement of both the disc and droplets within the time interval analyzed. Table 1 shows the results of disc and droplet velocities for different fluences (average values for 5 to 10 measurements in each case), including the standard errors of the linear regression. As can be seen in the table, the droplets form only at fluences above F = 0.21 J/cm2 and for F = 0.26 J/cm2 their velocity already exceeds the speed of sound in air. In contrast, the Mo discs do not reach higher velocities than 150 m/s even for the highest fluence applied.
Table 1. Measured velocities of the Mo disc (vd) and fastest droplet (vp) for different fluence values including standard errors of the linear regression.
By varying the delay time between the laser pulse and the first picture we tried to determine the point in time when the disc has left the Mo surface. Unfortunately the spatial resolution is not high enough to distinguish between disc and possible Mo vapor or droplets. Therefore we can only estimate the required time for lift-off by extrapolating the positions measured at later times towards x(t) = 0 in Fig. 5. Due to high fluctuations of the velocities determined for higher fluences we do this extrapolation only for F = 0.18 and 0.23 J/cm2. Here we find t ≈ 26 ns in both cases, but the errors of the linear regression lie already in the range of the values obtained, which makes an accurate determination impossible.
For this reason we use the setup described in Sect. 2.2 to determine the time required for the lift-off of the Mo disc. A typical signal observed at the DSO on a μs timescale is shown in Fig. 6(a). When the laser pulse hits the test structure at t = 0, a sudden voltage drop is observed, followed by a slow increase up to a level, which is significantly lower than the initial voltage. When we release a second laser pulse onto the same structure, see Fig. 6(b), we observe a drop from the level previously reached, followed by a similar increase back to the same level.
Fig. 6 Electrical measurement data for a fluence of F = 0.39 J/cm2. The time t = 0 refers to the point in time when the pulse hits the sample. a) First shot on test structure. b) Second shot on the same test structure (remaining Mo bridge).
By analyzing the voltage signal on a ns timescale (Fig. 7), we see not only a sharp voltage drop but an intermediate level lasting several ns, followed by a second decrease. This intermediate level is not present when we release a second pulse onto the same test structure. By varying the incident fluence, we observe that the width of this intermediate level decreases and the steepness of the second voltage drop increases for higher fluences. Note that the exponential-like increase to the final level, as shown in Fig. 6, cannot be observed on the ns timescale but is of course present in the signal. Figure 7 shows a selection of acquired measurement data for fluences between F = 0.17 and 0.49 J/cm2. The threshold for lift-off of complete discs was determined to F = 0.15 ± 0.1 J/cm2.
Fig. 7 Selected examples of measurement data for F = 0.17, 0.22, 0.30, 0.39 and 0.49 J/cm2 (from top to bottom). The time t = 0 refers to the point in time when the pulse hits the sample. The points in time when the observed voltage drop from the intermediate level starts and when it ends are labeled as t1 and t2, respectively. Note that an individual offset was added to the measurement data in y-direction for the sake of readability.
The extracted times (t1, t2) are denoted in Table 2 where t1 refers to the point in time when the second decrease starts and t2 to the time when it ends (i.e. when the lowest voltage level is measured, see also Fig. 8). The values for t1 and t2 in Table 2 are averaged values over 4 to 7 measurements for every fluence value. As can be seen in the table, the shortest time t2 of about 21 ns is already measured at 0.3 J/cm2 and there seems to be no further decrease for the two higher fluence values. Due to the occurrence of first damage or ablation of the substrate at the SLG/Mo interface for fluences exceeding about 0.5 J/cm2 no higher fluence was applied.
Fig. 8 Model assumption of the different stages during ablation in cross-sectional view (top). An example of a measured signal (F = 0.3 J/cm2) and the corresponding voltage levels are shown in f). a) Initial state, b) first voltage drop (U1 → U2) due to heating, c) bulging due to vapor formation, d) voltage drop (U2 → U3) due to lift-off, e) final state after cool-down.
Table 2. Measured times between laser pulse (t = 0) and voltage drops. The times ti refer to the point in time when the observed voltage drop from U2 to U3 starts (t1) and when it ends (t2), respectively. See also Fig. 8.
Note that we have to adjust the width w of the Mo bridge according to the ablation spot diameter d for every fluence value in order to obtain a good signal-to-noise ratio. Typically, w is set to w = d + 4 μm. Therefore, the absolute height of the voltage drops shown in Fig. 7 cannot be compared directly, as different widths result in different resistances of the structure and thus in different voltage levels.
4. Discussion and modeling
Both the measured velocities of the disc and the dependence of the velocity on the fluence meet the expected behavior as observed in similar experiments [24, 25]. It is interesting to note that the Mo disc does not reach velocities higher than about half the speed of sound whereas the droplets easily exceed this velocity. One possible reason might be the aerodynamic resistance of the disc. However, the measured velocity within the time window regarded is uniform, i.e. no deceleration is observed. For low fluences the extrapolated values for the lift-off time of the disc are in good agreement with that obtained using the electrical measurements. Admittedly the errors of the extrapolation are high, thus a final conclusion on the lift-off time would only be possible in case of a significant improvement of the spatial resolution of our shadowgraphic setup.
To explain the temporal behavior of the voltage signal, a closer look at the different phases of the ablation is necessary. Figure 8 illustrates a simple model of these phases and the corresponding voltage levels Ui measured. The different phases and the underlying assumptions can be explained as follows:
In the initial phase, prior to the laser pulse, the structure is at room temperature (RT) and the Mo bridge is fully intact (constant voltage U1, Fig. 8(a)). After the deposition of energy by the laser pulse and its thermalization, the resistivity of the Mo bridge increases and within <1 ns the first drop in voltage at the DSO is observed (U1 → U2). Within the next couple of nanoseconds vapor forms at the SLG/Mo interface and causes a bulging of the Mo film, see Fig. 8(b) and 8(c). During this period a certain amount of energy is transformed into latent heat of vaporization. The temperature of the bridge remains approximately constant and hence the resistivity, too (voltage level U2). The vapor formation causes an expansion of the Mo bulge. When it is large enough, the stress limit of the Mo film at the edge of the bulge is exceeded and the disc starts to lift-off and finally escapes perpendicularly to the surface of the substrate as shown in Fig. 8(d). At this stage the second voltage drop U2 → U3 is observed because of a fast decrease of the cross-sectional area which results in an increased resistivity. In the final phase the remaining Mo cools down to RT. Therefore, the resistivity decreases and accordingly an increase in voltage U3 → U4 is observed, see Fig. 8(e).
According to this model assumptions, the point in time t2 refers to that situation, at which the Mo disc has completed the lift-off from the surrounding Mo. Indeed, the measured times for t2 (Table 2) using our test structure are in good agreement with values observed by Domke et al.[30] using a pump-probe setup. Interestingly, our measured lift-off times decrease with increasing fluence only down to a value of about 21 ns which stays the same for 0.3 J/cm2 < F < 0.49 J/cm2. This is probably due to the fact that a certain minimum of time is required to vaporize the material and to build up the pressure necessary for lift-off.
In order to verify that the observed voltage steps can originate from heat induced change in resistivity and a change in geometry, respectively, we use a simple 2D finite element model (FEM) of the test structure. This model is created with the software “Abaqus FEA” (Dassault Systèmes, v6.12) which solves the 2-dimensional heat flow equation and computes e.g. values of the local current density by taking into account the temperature dependence of the electrical resistivity and the heat capacity, respectively. Our transient simulation lasts for up to 100 μs after the laser pulse and includes several changes of the simulation conditions. The resulting four important stages and the underlying conditions can be described as follows:
- initial phase, structure at room temperature.
- t = 0: laser pulse at the center of the Mo bridge. The effect of the laser pulse is modeled by the application of a heat source with a two-dimensional Gaussian distribution, i.e. the absorbed energy of the pulse is used to heat up the material. The 1/e2-width was set to 28 μm, which is about 30 % below the value determined in Sect. 2.2. This value is an estimate of the effective beam width at the SLG/Mo interface and was obtained by using the method of [32]. The absorbed energy was calculated by taking into account the reflectance of the air/SLG and the SLG/Mo interfaces. Both values are calculated using Fresnel’s equations with the following values of the (complex) refractive index: nAir = 1, nSLG = 1.52 [36], nMo = 3.59 + 3.79i[37]. Note that our simple 2D model does not include phase changes. Moreover, as it is a 2D model, it only considers lateral heat flow, i.e. the heat flow perpendicular to the surface of the 360 nm thin Mo film is neglected.
- t = 25 ns: structure after lift-off of the Mo disc. At this point in time we simply ’remove’ the central part of the bridge in our model, i.e. the energy stored in the disc also disappears. This assumption is justified by the fact that the Mo disc including a certain amount of energy stored both in the disc and in a gaseous Mo phase is removed from the system after lift-off.
- t = 100 μs: final stage, structure with removed Mo disc cooled down to approximately room temperature.
For calculating the electrical resistivity of the structure, we use data for electrical conductivity ρ(T) from [38–40]. As these values differ in the order of up to 10% for higher temperatures, the FEM calculations are conducted in two scenarios: a ’best case’ and a ’worst case’, where the lowest and highest reported values for resistivity are used, respectively. This approach is also applied for the thermal conductivity λ(T) as reported in [41, 42]. Regarding the data for heat capacity Cp(T), we use the values reported in [38]. In order to simplify the model, the density of Mo is assumed to be constant within the entire temperature range. Please note that the material data used was obtained mostly for bulk Mo and not for thin films, which might lead to an additional error of the values calculated.
In order to visualize the simulation results, we present the current density at the central part of our test structure in Fig. 9. As the current density both depends on the geometry and the temperature of the material, the figure shows a combination of these two influencing factors.
Fig. 9 Simulation results for current density (arbitrary units) at the central part of the test structure (refer to Fig. 3). From left to right: a) initial state, b) after heat deposition by the laser pulse, c) after lift-off of the disc, d) final state after cool-down of the bridge.
In the first image, Fig. 9(a), the current is equally distributed. In the second image, Fig. 9(b), the heat source introduced leads to an increased resistivity particularly at the center and forces the current to the colder outer parts of the bridge with comparatively low resistivity. This causes an overall increase in resistivity of the test structure and consequently a voltage drop at the DSO. After the lift-off of the central part, two bars remain at the side of the Mo bridge, see Fig. 9(c). Due to the energy deposited by the ’wings’ of the Gaussian intensity distribution, their temperature is higher at the side, which is now facing the hole. It can be observed that the current mainly flows at the colder outer parts of the remaining Mo bars. The lift-off and the resulting change in geometry leads to a further increase in overall resistivity and consequently a further voltage drop is observed at the DSO. When the remaining Mo has cooled down to approximately RT, the current is uniformly distributed over the remaining Mo bars, see Fig. 9(d). The resistivity decreases and an increase of the voltage at the DSO is observed, however, not to the original value due to the removed Mo disc.
Table 3 shows both the measured data and the simulation results for F = 0.22 J/cm2. The simulated overall voltage step U1 − U4 is in good agreement with the measurement results which means that the observed increase in resistivity of our test structure at RT can be fully explained by the change in geometry due to the removed Mo disc. However, both voltage steps U1 − U2 and U2 − U3 are over-estimated by about a factor of 1.5 – 2. Thus, it seems that our simple model over-estimates the resistivity of the test structure after the laser pulse, which originates most likely from an over-estimation of the temperatures reached. One explanation for this could be the neglected heat flow from the Mo to the SLG substrate. Another more likely explanation could be the impact of the neglected phase changes. Our model uses the entire energy absorbed by the Mo to heat up the structure. But in reality, the energy transferred to latent heat of fusion and vaporization is not available for heating up the structure. Therefore, a more complex model including thermodynamic phase changes could lead to better agreement with the measured values.
Table 3. Measured and calculated voltage drops for F = 0.22 J/cm2. Refer to Fig. 8 for labeling of voltages Ui.
5. Conclusion
In this work we present a measurement technique free from optical limitations which allows to study the timescales of back side ablation of thin conductive films. On a Mo coated glass substrate (as commonly used in production of CIS thin film solar cells) we create a specifically developed test structure. This structure has a small constriction, which forms a sensitive temperature-dependent resistor. Using a fast reacting voltage source and an oscilloscope, we monitor the current flow over this resistor while a single ultrashort laser pulse (τ ≈ 10 ps) is fired onto it. In this way we study the first nanoseconds of ablation as well as the cool-down of the non-ablated material which occurs on a microsecond time scale. Furthermore, we compare our findings with observations of the ablation products – fully intact Mo ’discs’ – using a shadowgraphic setup.
With increasing fluence F we find increasing velocities of ablated Mo discs up to a maximum of 150 ± 5 m/s at F ≈ 0.6 J/cm2. However, liquid droplets of Mo forming at the center of the disc can reach velocities up to 710 ± 11 m/s. By using our test structure we determine the time required for vaporization and for the subsequent bulging of the thin Mo layer until lift-off. This time decreases with increasing fluence but only down to a minimum of about 21 ± 2 ns which is observed already for a fluence of 0.3 J/cm2. We use a simple 2D FEM to verify the voltage drops measured with our test structure. Our hypothesis that the observed voltage drops can originate from the ascribed events of heating and lift-off can be validated with this model. However, it over-estimates the resistivities reached within the test structure by about a factor of two. We assume this is mainly due to neglected phase changes which occur at the central part of the test structure at the Mo/SLG interface and thus lead to an overestimated temperature of the sample.
In summary, our study contributes to a better understanding of the back-side ablation process. Thus it may help to spur the implementation of such processes in different application areas, e.g. from photovoltaics to laser induced forward transfer.
Acknowledgments
We acknowledge financial support by the German Federal Ministry of Education and Research (BMBF) under contract Nos. 13N11783 and 13N11788 (“Tailored for next PV, T4nPV”).
References and links
1. F. Röser, T. Eidam, J. Rothhardt, O. Schmidt, D. Schimpf, J. Limpert, and A. Tünnermann, “Millijoule pulse energy high repetition rate femtosecond fiber chirped-pulse amplification system,” Opt. Lett. 32, 3495–3497 (2007) [CrossRef] [PubMed] .
2. J. Kleinbauer, D. Eckert, S. Weiler, and D. H. Sutter, “80W ultrafast CPA-free disk laser,” Proc. SPIE 6871, 68711B (2008) [CrossRef] .
3. P. Russbueldt, T. Mans, G. Rotarius, J. Weitenberg, H. D. Hoffmann, and R. Poprawe, “400W Yb:YAG innoslab fs-amplifier,” Opt. Express 17, 12230–12245 (2009) [CrossRef] [PubMed] .
4. R. Knappe, T. Herrmann, B. Heinrich, and A. Nebel, “Novel picosecond lasers for fast micro-machining,” in “Proceedings of the Third International WLT-Conference on Lasers in Manufacturing, LIM 2005 ”, 723–728 (2005).
5. B. Chichkov, C. Momma, S. Nolte, F. Von Alvensleben, and A. Tünnermann, “Femtosecond, picosecond and nanosecond laser ablation of solids,” Appl. Phys. A: Mater. Sci. Process. 63, 109–115 (1996) [CrossRef] .
6. S. Nolte, C. Momma, H. Jacobs, A. Tünnermann, B. Chichkov, B. Wellegehausen, and H. Welling, “Ablation of metals by ultrashort laser pulses,” J. Opt. Soc. Am. B 14, 2716–2722 (1997) [CrossRef] .
7. D. Breitling, A. Ruf, and F. Dausinger, “Fundamental aspects in machining of metals with short and ultrashort laser pulses,” Proc. SPIE 5339, 49–63 (2004) [CrossRef] .
8. A. D. Compaan, I. Matulionis, and S. Nakade, “Laser scribing of polycrystalline thin films,” Opt. Laser Eng. 34, 15–45 (2000) [CrossRef] .
9. D. Ruthe, K. Zimmer, and T. Höche, “Etching of CuInSe2thin films–comparison of femtosecond and picosecond laser ablation,” Appl. Surf. Sci. 247, 447–452 (2005) [CrossRef] .
10. J. Hermann, M. Benfarah, S. Bruneau, E. Axente, G. Coustillier, T. Itina, J.-F. Guillemoles, and P. Alloncle, “Comparative investigation of solar cell thin film processing using nanosecond and femtosecond lasers,” J. Phys. D: Appl. Phys. 39, 453 (2006) [CrossRef] .
11. S. Zoppel, H. Huber, and G. Reider, “Selective ablation of thin Mo and TCO films with femtosecond laser pulses for structuring thin film solar cells,” Appl. Phys. A: Mater. Sci. Process. 89, 161–163 (2007) [CrossRef] .
12. H. P. Huber, F. Herrnberger, S. Kery, and S. Zoppel, “Selective structuring of thin-film solar cells by ultrafast laser ablation,” Proc. SPIE 6881, 688117 (2008) [CrossRef] .
13. H. P. Huber, M. Englmaier, C. Hellwig, A. Heiss, T. Kuznicki, M. Kemnitzer, H. Vogt, R. Brenning, and J. Palm, “High speed structuring of CIS thin-film solar cells with picosecond laser ablation,” Proc. SPIE 7203, 72030R (2009) [CrossRef] .
14. G. Heise, M. Englmaier, C. Hellwig, T. Kuznicki, S. Sarrach, and H. P. Huber, “Laser ablation of thin molybdenum films on transparent substrates at low fluences,” Appl. Phys. A: Mater. Sci. Process. 102, 173–178 (2011) [CrossRef] .
15. T. Dalibor, S. Jost, H. Vogt, A. Heiß, S. Visbeck, T. Happ, J. Palm, A. Avellán, T. Niesen, and F. Karg, “Towards module efficiencies of 16% with an improved CIGSSe device design,” in “Proceedings of the 26th EUPVSEC ”, 2407–2411 (2011).
16. A. B. Bullock and P. R. Bolton, “Laser-induced back ablation of aluminum thin films using picosecond laser pulses,” J. Appl. Phys. 85, 460–465 (1999) [CrossRef] .
17. F. J. Mayer and G. E. Busch, “Plasma production by laser-driven explosively heated thin metal films,” J. Appl. Phys. 57, 827–829 (1985) [CrossRef] .
18. A. B. Bullock, P. R. Bolton, and F. J. Mayer, “Time-integrated reflectivity of laser-induced back-ablated aluminum thin film targets,” J. Appl. Phys. 82, 1828–1831 (1997) [CrossRef] .
19. S. I. Ashitkov, N. A. Inogamov, P. S. Komarov, V. V. Zhakhovsky, I. I. Oleynik, M. B. Agranat, G. I. Kanel, and V. E. Fortov, “Strength of metals in liquid and solid states at extremely high tension produced by femtosecond laser heating,” AIP Conf. Proc. 1464, 120–125 (2012) [CrossRef] .
20. S. Ashitkov, M. Agranat, G. Kanel’, P. Komarov, and V. Fortov, “Behavior of aluminum near an ultimate theoretical strength in experiments with femtosecond laser pulses,” JETP Lett. 92, 516–520 (2010) [CrossRef] .
21. V. Schultze and M. Wagner, “Blow-off of aluminium films,” Appl. Phys. A: Mater. Sci. Process. 53, 241–248 (1991) [CrossRef] .
22. Y. Nakata and T. Okada, “Time-resolved microscopic imaging of the laser-induced forward transfer process,” Appl. Phys. A: Mater. Sci. Process. 69, S275–S278 (1999) [CrossRef] .
23. R. Fardel, M. Nagel, F. Nüesch, T. Lippert, and A. Wokaun, “Shadowgraphy investigation of laser-induced forward transfer: Front side and back side ablation of the triazene polymer sacrificial layer,” Appl. Surf. Sci. 255, 5430–5434 (2009) [CrossRef] .
24. K. S. Kaur, R. Fardel, T. C. May-Smith, M. Nagel, D. P. Banks, C. Grivas, T. Lippert, and R. W. Eason, “Shadowgraphic studies of triazene assisted laser-induced forward transfer of ceramic thin films,” J. Appl. Phys. 105, 113119 (2009) [CrossRef] .
25. D. G. Papazoglou, A. Karaiskou, I. Zergioti, and C. Fotakis, “Shadowgraphic imaging of the sub-ps laser-induced forward transfer process,” Appl. Phys. Lett. 81, 1594–1596 (2002) [CrossRef] .
26. A. Kuznetsov, C. Unger, J. Koch, and B. Chichkov, “Laser-induced jet formation and droplet ejection from thin metal films,” Appl. Phys. A: Mater. Sci. Process. 106, 479–487 (2012) [CrossRef] .
27. M. Domke, G. Heise, I. Richter, S. Sarrach, and H. P. Huber, “Pump-probe investigations on the laser ablation of CIS thin film solar cells,” Phys. Procedia 12, 399–406 (2011) [CrossRef] .
28. M. Domke, S. Rapp, M. Schmidt, and H. P. Huber, “Ultrafast pump-probe microscopy with high temporal dynamic range,” Opt. Express 20, 10330–10338 (2012) [CrossRef] [PubMed] .
29. M. Domke, S. Rapp, and H. Huber, “Ultra-fast movies resolve ultra-short pulse laser ablation and bump formation on thin molybdenum films,” Phys. Procedia 39, 717–725 (2012) [CrossRef] .
30. M. Domke, S. Rapp, M. Schmidt, and H. Huber, “Ultra-fast movies of thin-film laser ablation,” Appl. Phys. A: Mater. Sci. Process. 109, 409–420 (2012) [CrossRef] .
31. D. Bartl, A. Michalowski, M. Hafner, A. Letsch, S. Nolte, and A. Tünnermann, “Time-resolved study of back side ablated molybdenum thin films by ultrashort laser pulses,” Appl. Phys. A: Mater. Sci. Process. 110, 227–233 (2013) [CrossRef] .
32. J. M. Liu, “Simple technique for measurements of pulsed gaussian-beam spot sizes,” Opt. Lett. 7, 196–198 (1982) [CrossRef] [PubMed] .
33. G. Heise, M. Domke, J. Konrad, S. Sarrach, J. Sotrop, and H. P. Huber, “Laser lift-off initiated by direct induced ablation of different metal thin films with ultra-short laser pulses,” J. Phys. D: Appl. Phys. 45, 315303 (2012) [CrossRef] .
34. A. Michalowski, D. Walter, P. Berger, and F. Dausinger, “Diagnostics of drilling process using ultrashort laser pulses,” in “Proceedings of the Fourth International WLT-Conference on Lasers in Manufacturing, LIM 2007 ”, 563–567 (2007).
35. D. P. Brunco, J. A. Kittl, C. E. Otis, P. M. Goodwin, M. O. Thompson, and M. J. Aziz, “Time-resolved temperature measurements during pulsed laser irradiation using thin film metal thermometers,” Rev. Sci. Instrum. 64, 2615–2623 (1993) [CrossRef] .
36. R. A. Synowicki, B. D. Johs, and A. C. Martin, “Optical properties of soda-lime float glass from spectroscopic ellipsometry,” Thin Solid Films 519, 2907–2913 (2011) [CrossRef] .
37. D. W. Lynch and W. R. Hunter, Handbook of Optical Constants of Solids, E. D. Palik, ed. (Academic Press Inc., 1985), chap. VI, pp. 303–313.
38. K. Maglic, N. Perovic, and G. Vukovic, “Specific heat and electric resistivity of molybdenum between 400 and 2500 K,” High Temp. - High Pressures 29, 97–102 (1997) [CrossRef] .
39. R. S. Hixson and M. A. Winkler, “Thermophysical properties of molybdenum and rhenium,” Int. J. Thermophys. 13, 477–487 (1992) [CrossRef] .
40. T. Hüpf, C. Cagran, G. Lohöfer, and G. Pottlacher, “Electrical resistivity of high melting metals up into the liquid phase (V, Nb, Ta, Mo, W),” J Phys.: Conf. Ser. 98, 062002 (2008) [CrossRef] .
41. S. Taluts, V. Zinov’yev, V. Polev, and S. Il’inykh, “Thermal diffusivity and conductivity of molybdenum in solid and liquid states.” Phys. Met. Metall. 58, 191–194 (1984).
42. A. Kurichenko, S. Taluts, I. Korshunov, A. Ivlieu, and V. Zinov’ev, “Thermal diffusivity and thermal conductivity of molybdenum and molybdenum-titanium bimetal at high temperatures,” High Temp. - High Pressures 21, 437–439 (1989).
