- Email: [email protected]
Picosecond UV-laser ablation of Au and Ni films
applied
surface ELSEVIER
science
Applied Surface Science 96-98 (1996) 439-4442
Picosecond UV-laser ablation of Au and Ni films A. Rosenfeld *, E.E.B. Campbell Max-Born-lnstitutfiir Nichtlineare Optik und Kurueitspektroskopie, Postfach 1107, D-12474 Berlin, Germany
Received 22 May 1995
Abstract Single-shot ablation thresholds of gold and nickel films in the thickness range from 50 nm to 3.7 pm have been measured for 8 ps, 130 ps and, for comparison, 15 ns laser pulses at 248 nm. The metal films were deposited on fused silica substrates. The topology of the ablated films was determined by an AFM. For picosecond ablation one finds an interesting transition range where, for the 8 ps pulses, the threshold fluence initially drops as the film thickness is increased, in contrast to the 130 ps pulses where the threshold increases with increasing film thickness, as for ablation with the nanosecond pulses. In addition, comparison of the topological structure shows different behaviour for the Au and Ni films. The Ni films show typical material removal behaviour. The Au films show a melting character both for the 8 and 130 ps pulses. The nature of this difference is not yet clear and is the subject of further investigations.
1. Introduction Pulsed UV-laser ablation has become a useful tool for surface processing since it was first investigated in the early eighties [ 11. The dynamics of the ablation process is, however, strongly material dependent and remains under intense investigation. A topic of recent interest is the extent to which the ablation
process and the quality of the structures that can be obtained depend on the duration of the laser pulse. In particular, one would like to diminish and, if possible, remove the thermal effects which limit the resolution and quality of the laser processing that can be obtained for materials such as metal films when using standard nanosecond lasers.
* Corresponding author. Fax: +49 30 6392 1229; e-mail: [email protected].
0169-4332/96/$15.00 0 1996 Elsevier Science B.V. All rights reserved SSDI 0169-4332(95)00447-S
One of the first experiments to compare the ablation process of metals in the picosecond and nanosecond time regime was carried out by Corkum et al. [2]. They established that the threshold fluence for damage is independent of pulse duration below 1 ns for bulk copper and molybdenum. Kautek et al. [3] described the different behaviour of gold sheets for nanosecond and femtosecond excitation. They found that the heat affected zones generated by convenient nanosecond pulse lasers are approximately a hundred times larger than those generated by femtosecond laser treatment. The measurements of Matthias et al. [4] for Au and Ni films on fused silica substrates showed a linear dependence of the ablation threshold on film thickness for nanosecond pulses, as long as the thickness was larger than the optical absorption depth and smaller than the thermal diffusion length L, = (2~$,)‘/~, where K is the
440
A. Rosenfeld, E.E.B. Campbell/Applied Surf&e Science 96-98 (1996) 439-442
thermal diffusivity and tp the laser pulse duration. For thicknesses larger than L,, the thresholds are independent of thickness. For 0.5 ps pulses L, = 4 mn and is thus much smaller than the thinnest films investigated by Preuss et al. [5]. In this case the threshold fluence was found to be independent of the film thickness and approximately a factor 100 lower than for nanosecond ablation. In this paper we present results on the single shot damage thresholds of Ni and Au films of varying thickness on fused silica substrates for a laser wavelength of 248 nm and pulses in the picosecond range. We have found an interesting transition in the behaviour of the threshold fluence as a function of film thickness as the laser pulse duration is reduced. For ablation with 8 ps pulses the threshold fluence initially drops as the film thickness is increased, in contrast to the 130 ps and nanosecond results where the threshold increases with increasing film thickness.
2. Experimental
procedure
The homogeneous part of the output of an UV short pulse laser system was used to illuminate a pinhole which was then imaged with a quartz projection optics (working distance 25 mm) onto the sample. The laser system consists of a short pulse dye laser pumped by a N, laser. The output of this dye laser is spectrally narrowed by a grating in autocollimation or in gracing incidence to yield the two pulse durations of 8 or 130 ps respectively. The second harmonic of this radiation is amplified in a KrF laser (ASE < 12%). The pulse duration was measured with an autocorrelator (8 ps) or a streak camera (130 ps) and the laser energy was monitored on-line (the given laser fluence values have an experimental uncertainty of ca. 15%). The ablation measurements were carried out in a vacuum chamber with a background pressure of < 10e3 mbar. The sample was mounted on a q-stage which made it possible to get a fresh sample spot for each laser pulse. The film thickness varied from 50 nm to 3.7 pm for Ni and from 0.1 to 2 pm for Au. The threshold was defined as being the lowest laser fluence for which visible damage of the films could be observed.
3. Results and discussion 3.1. Ablation threshold The results of the determination of the ablation threshold of the Ni and Au films are shown in Figs. 1 and 2. In both cases a similar trend is observed. For Ni film thicknesses greater than 1 pm the ablation threshold remains constant with increasing film thickness but the absolute value decreases with decreasing pulse duration. For 15 ns and 130 ps pulses the threshold decreases as the thickness decreases below ca. 1 or 0.5 pm respectively but shows the opposite trend for 8 ps pulses i.e. the threshold increases as the film thickness is reduced. The nanosecond data are comparable with the results in Ref. [4]. A similar, but much less pronounced, trend can be seen in the Au results. As shown in Ref. [4] the nanosecond ablation can be described by considering a simple thermodynamic model in which the thermal diffusion length Lth governs the fluence thresholds for ablation. Assuming the ablation is still purely thermal and controlled by Lt,,, the model predicts an ablation threshold of 12 mJ/cm* for the Ni films using 0.5 ps pulses, in good agreement with the experimentally determined value of 20 mJ/cm* [5]. If we apply this simple model to our picosecond results, however, the predicted threshold values of 200 and 50 mJ/cm” for 130 and 8 ps pulses respectively are much lower than those observed experimentally. The 130 ps re-
.-. -0
0 .---
A -
--
10
--~--
13
A
2.0
d (elm) Fig. 1. Ablation threshold for 8 ps, 130ps and 15 ns laser pulses, wavelength 248 nm, as a function of Ni film thickness on fused silica.
A. Rosenfeld, E.E.B. Campbell/Applied Surface Science 96-98 (1996) 439-442
@ . _. ‘_A
0.0 D
.._............
1
A
A . . ..___.....__.....
2
3
4
d WI
441
the respective threshold fluence. The Au films shows a typical melting behaviour leading to an increased volume of the irradiated surface. Similar structures have been observed for many different metals, including the Au and Ni films investigated here, using nanosecond laser pulses. The behaviour of the Ni films is much more interesting. Fig. 4 shows that a clear decrease in the volume of the radiated surface is observed for the thicker film (3.7 pm) indicating that an initial melting of the metal film was not a prerequisite for material removal in this case. For the thinner film (0.48 pm) we can again see a small
Fig. 2. As in Fig. 1 for Au films on fused silica.
sults could be modelled by assuming a thermal diffusivity a factor of ten(!) larger than the bulk value of 0.19 cm* SC’ which also gives approximately the correct threshold value for ablation with 8 ps pulses with film thicknesses greater than 1 km. Exactly the same effect is observed for the gold films. It may be possible that the picosecond pulses alter the properties of the films in such a way that the thermal diffusivity increases or, alternatively, the reflectivity increases, both effects leading to an increased threshold fluence in this thermal model. The changes required are, however, so large that this does not appear to be a plausible explanation and the answer is more likely to be found in a non-thermal mechanism which serves to rapidly dissipate the absorbed energy. A number of other experiments have shown evidence that desorption and ablation from metals can occur by athermal mechanisms connected with surface properties, even with nanosecond pulses (see e.g. Ref. [6]). Also, the increase in the ablation threshold with decreasing film thickness for the 8 ps pulses cannot be explained with the simple thermal model. 3.2. Topology The topological structure of the ablated films may help to shed some light on the nature of the mechanisms leading to the results discussed above. Figs. 3 and 4 compare the structures observed with an AFM after 8 ps single pulse radiation of Au and Ni films for two different film thicknesses. The laser fluence used in each case was a factor of three higher than
Fig. 3. Al% images of two Au films irradiated with a single 8 ps, 248 nm laser pulse at a fluence 3 times over the threshold. Film thickness: (a) 0.4 pm, (b) 2.0 pm.
442
A. Rosenfeld, E.E.B. Campbell/Applied
Surface Science 96-98
(1996) 439-442
the behaviour approaches that of the Au films with a clear melting and volume increase. The difference between the Ni and Au films may somehow be related to the different thermodynamic properties such as the latent heat for melting and for vapourization which are very different for the two metals.
4. Conclusions The investigation of the thickness dependence of the fluence threshold has been carried out for Ni and Au films on a fused silica substrate with picosecond and nanosecond laser pulses at 248 nm. The results can not be explained with a simple thermal diffusion model. As the pulse duration is decreased from 130 to 8 ps there is increasing evidence of a non-thermal desorption or ablation mechanism as inferred from the behaviour of the threshold fluence and the topology of the irradiated surfaces. More work is required to identify the exact nature of the processes leading to these results.
Acknowledgements Financial support from the BMBF through grant “Grundlagenuntersuchungen number 13N6591/1 zur Materialstrukturierung und -modifizierung mit ultrakurzen Laserimpulsen” is gratefully acknowledged. Fig. 4. AFM images of two Ni films irradiated with a single 8 ps, 248 mn laser pulse at a fluence 3 times over the threshold. Film thickness: (a) 0.48 pm, (b) 3.7 pm.
References
decrease in the volume but the effect is very small considering that the fluence is three times the threshold fluence. This is also the film thickness regime where the threshold fluence increases with decreasing thickness and it seems to be much more difficult with short laser pulses to remove material from these films compared to films over 1 pm thick. The structures obtained with 130 ps pulses lie between the ns and 8 ps results: for thin films we see a volume decrease but, as the film thickness increases,
[l] R. Srinivasan and V. Mayne Banton, Appl. Phys. Lert. 41 (1982) 576. [2] P. Corcum, F. Brunel, N. Sherman and T. Srinivasan-Rao, Phys. Rev. Lett. 44 (1990) 1847. [3] W. Kautek and J. Krllger, Proc. SPIE 2207 (1994) 600. [4] E. Matthias, M. Reichling, J. Siegel, O.W. KZding, S. Petzoldt, H. Skurk, P. Bitzenberger and E. Neske, Appl. Phys. A 58 (1994) 129. [5] S. Preuss, E. Matthias and M. Stuke, Appl. Pbys. A 59 (1994) 79. [6] See e.g. R.F. Haglund and R. Kelly, in: Fundamental Processes in Sputtering of Atoms and Molecules (SPUT92), Mat.-Fys. Medd. 43 (1993) 527.
surface ELSEVIER
science
Applied Surface Science 96-98 (1996) 439-4442
Picosecond UV-laser ablation of Au and Ni films A. Rosenfeld *, E.E.B. Campbell Max-Born-lnstitutfiir Nichtlineare Optik und Kurueitspektroskopie, Postfach 1107, D-12474 Berlin, Germany
Received 22 May 1995
Abstract Single-shot ablation thresholds of gold and nickel films in the thickness range from 50 nm to 3.7 pm have been measured for 8 ps, 130 ps and, for comparison, 15 ns laser pulses at 248 nm. The metal films were deposited on fused silica substrates. The topology of the ablated films was determined by an AFM. For picosecond ablation one finds an interesting transition range where, for the 8 ps pulses, the threshold fluence initially drops as the film thickness is increased, in contrast to the 130 ps pulses where the threshold increases with increasing film thickness, as for ablation with the nanosecond pulses. In addition, comparison of the topological structure shows different behaviour for the Au and Ni films. The Ni films show typical material removal behaviour. The Au films show a melting character both for the 8 and 130 ps pulses. The nature of this difference is not yet clear and is the subject of further investigations.
1. Introduction Pulsed UV-laser ablation has become a useful tool for surface processing since it was first investigated in the early eighties [ 11. The dynamics of the ablation process is, however, strongly material dependent and remains under intense investigation. A topic of recent interest is the extent to which the ablation
process and the quality of the structures that can be obtained depend on the duration of the laser pulse. In particular, one would like to diminish and, if possible, remove the thermal effects which limit the resolution and quality of the laser processing that can be obtained for materials such as metal films when using standard nanosecond lasers.
* Corresponding author. Fax: +49 30 6392 1229; e-mail: [email protected].
0169-4332/96/$15.00 0 1996 Elsevier Science B.V. All rights reserved SSDI 0169-4332(95)00447-S
One of the first experiments to compare the ablation process of metals in the picosecond and nanosecond time regime was carried out by Corkum et al. [2]. They established that the threshold fluence for damage is independent of pulse duration below 1 ns for bulk copper and molybdenum. Kautek et al. [3] described the different behaviour of gold sheets for nanosecond and femtosecond excitation. They found that the heat affected zones generated by convenient nanosecond pulse lasers are approximately a hundred times larger than those generated by femtosecond laser treatment. The measurements of Matthias et al. [4] for Au and Ni films on fused silica substrates showed a linear dependence of the ablation threshold on film thickness for nanosecond pulses, as long as the thickness was larger than the optical absorption depth and smaller than the thermal diffusion length L, = (2~$,)‘/~, where K is the
440
A. Rosenfeld, E.E.B. Campbell/Applied Surf&e Science 96-98 (1996) 439-442
thermal diffusivity and tp the laser pulse duration. For thicknesses larger than L,, the thresholds are independent of thickness. For 0.5 ps pulses L, = 4 mn and is thus much smaller than the thinnest films investigated by Preuss et al. [5]. In this case the threshold fluence was found to be independent of the film thickness and approximately a factor 100 lower than for nanosecond ablation. In this paper we present results on the single shot damage thresholds of Ni and Au films of varying thickness on fused silica substrates for a laser wavelength of 248 nm and pulses in the picosecond range. We have found an interesting transition in the behaviour of the threshold fluence as a function of film thickness as the laser pulse duration is reduced. For ablation with 8 ps pulses the threshold fluence initially drops as the film thickness is increased, in contrast to the 130 ps and nanosecond results where the threshold increases with increasing film thickness.
2. Experimental
procedure
The homogeneous part of the output of an UV short pulse laser system was used to illuminate a pinhole which was then imaged with a quartz projection optics (working distance 25 mm) onto the sample. The laser system consists of a short pulse dye laser pumped by a N, laser. The output of this dye laser is spectrally narrowed by a grating in autocollimation or in gracing incidence to yield the two pulse durations of 8 or 130 ps respectively. The second harmonic of this radiation is amplified in a KrF laser (ASE < 12%). The pulse duration was measured with an autocorrelator (8 ps) or a streak camera (130 ps) and the laser energy was monitored on-line (the given laser fluence values have an experimental uncertainty of ca. 15%). The ablation measurements were carried out in a vacuum chamber with a background pressure of < 10e3 mbar. The sample was mounted on a q-stage which made it possible to get a fresh sample spot for each laser pulse. The film thickness varied from 50 nm to 3.7 pm for Ni and from 0.1 to 2 pm for Au. The threshold was defined as being the lowest laser fluence for which visible damage of the films could be observed.
3. Results and discussion 3.1. Ablation threshold The results of the determination of the ablation threshold of the Ni and Au films are shown in Figs. 1 and 2. In both cases a similar trend is observed. For Ni film thicknesses greater than 1 pm the ablation threshold remains constant with increasing film thickness but the absolute value decreases with decreasing pulse duration. For 15 ns and 130 ps pulses the threshold decreases as the thickness decreases below ca. 1 or 0.5 pm respectively but shows the opposite trend for 8 ps pulses i.e. the threshold increases as the film thickness is reduced. The nanosecond data are comparable with the results in Ref. [4]. A similar, but much less pronounced, trend can be seen in the Au results. As shown in Ref. [4] the nanosecond ablation can be described by considering a simple thermodynamic model in which the thermal diffusion length Lth governs the fluence thresholds for ablation. Assuming the ablation is still purely thermal and controlled by Lt,,, the model predicts an ablation threshold of 12 mJ/cm* for the Ni films using 0.5 ps pulses, in good agreement with the experimentally determined value of 20 mJ/cm* [5]. If we apply this simple model to our picosecond results, however, the predicted threshold values of 200 and 50 mJ/cm” for 130 and 8 ps pulses respectively are much lower than those observed experimentally. The 130 ps re-
.-. -0
0 .---
A -
--
10
--~--
13
A
2.0
d (elm) Fig. 1. Ablation threshold for 8 ps, 130ps and 15 ns laser pulses, wavelength 248 nm, as a function of Ni film thickness on fused silica.
A. Rosenfeld, E.E.B. Campbell/Applied Surface Science 96-98 (1996) 439-442
@ . _. ‘_A
0.0 D
.._............
1
A
A . . ..___.....__.....
2
3
4
d WI
441
the respective threshold fluence. The Au films shows a typical melting behaviour leading to an increased volume of the irradiated surface. Similar structures have been observed for many different metals, including the Au and Ni films investigated here, using nanosecond laser pulses. The behaviour of the Ni films is much more interesting. Fig. 4 shows that a clear decrease in the volume of the radiated surface is observed for the thicker film (3.7 pm) indicating that an initial melting of the metal film was not a prerequisite for material removal in this case. For the thinner film (0.48 pm) we can again see a small
Fig. 2. As in Fig. 1 for Au films on fused silica.
sults could be modelled by assuming a thermal diffusivity a factor of ten(!) larger than the bulk value of 0.19 cm* SC’ which also gives approximately the correct threshold value for ablation with 8 ps pulses with film thicknesses greater than 1 km. Exactly the same effect is observed for the gold films. It may be possible that the picosecond pulses alter the properties of the films in such a way that the thermal diffusivity increases or, alternatively, the reflectivity increases, both effects leading to an increased threshold fluence in this thermal model. The changes required are, however, so large that this does not appear to be a plausible explanation and the answer is more likely to be found in a non-thermal mechanism which serves to rapidly dissipate the absorbed energy. A number of other experiments have shown evidence that desorption and ablation from metals can occur by athermal mechanisms connected with surface properties, even with nanosecond pulses (see e.g. Ref. [6]). Also, the increase in the ablation threshold with decreasing film thickness for the 8 ps pulses cannot be explained with the simple thermal model. 3.2. Topology The topological structure of the ablated films may help to shed some light on the nature of the mechanisms leading to the results discussed above. Figs. 3 and 4 compare the structures observed with an AFM after 8 ps single pulse radiation of Au and Ni films for two different film thicknesses. The laser fluence used in each case was a factor of three higher than
Fig. 3. Al% images of two Au films irradiated with a single 8 ps, 248 nm laser pulse at a fluence 3 times over the threshold. Film thickness: (a) 0.4 pm, (b) 2.0 pm.
442
A. Rosenfeld, E.E.B. Campbell/Applied
Surface Science 96-98
(1996) 439-442
the behaviour approaches that of the Au films with a clear melting and volume increase. The difference between the Ni and Au films may somehow be related to the different thermodynamic properties such as the latent heat for melting and for vapourization which are very different for the two metals.
4. Conclusions The investigation of the thickness dependence of the fluence threshold has been carried out for Ni and Au films on a fused silica substrate with picosecond and nanosecond laser pulses at 248 nm. The results can not be explained with a simple thermal diffusion model. As the pulse duration is decreased from 130 to 8 ps there is increasing evidence of a non-thermal desorption or ablation mechanism as inferred from the behaviour of the threshold fluence and the topology of the irradiated surfaces. More work is required to identify the exact nature of the processes leading to these results.
Acknowledgements Financial support from the BMBF through grant “Grundlagenuntersuchungen number 13N6591/1 zur Materialstrukturierung und -modifizierung mit ultrakurzen Laserimpulsen” is gratefully acknowledged. Fig. 4. AFM images of two Ni films irradiated with a single 8 ps, 248 mn laser pulse at a fluence 3 times over the threshold. Film thickness: (a) 0.48 pm, (b) 3.7 pm.
References
decrease in the volume but the effect is very small considering that the fluence is three times the threshold fluence. This is also the film thickness regime where the threshold fluence increases with decreasing thickness and it seems to be much more difficult with short laser pulses to remove material from these films compared to films over 1 pm thick. The structures obtained with 130 ps pulses lie between the ns and 8 ps results: for thin films we see a volume decrease but, as the film thickness increases,
[l] R. Srinivasan and V. Mayne Banton, Appl. Phys. Lert. 41 (1982) 576. [2] P. Corcum, F. Brunel, N. Sherman and T. Srinivasan-Rao, Phys. Rev. Lett. 44 (1990) 1847. [3] W. Kautek and J. Krllger, Proc. SPIE 2207 (1994) 600. [4] E. Matthias, M. Reichling, J. Siegel, O.W. KZding, S. Petzoldt, H. Skurk, P. Bitzenberger and E. Neske, Appl. Phys. A 58 (1994) 129. [5] S. Preuss, E. Matthias and M. Stuke, Appl. Pbys. A 59 (1994) 79. [6] See e.g. R.F. Haglund and R. Kelly, in: Fundamental Processes in Sputtering of Atoms and Molecules (SPUT92), Mat.-Fys. Medd. 43 (1993) 527.