Open Access
Add to BibSonomy
Abstract
The intermediate pulse duration regime between typical ultra-short and nanosecond pulses has been investigated using MHz-range bursts of 70 ps pulses emitted from a custom-made fiber laser source. The goal of this study was to observe and understand the processes involved during laser ablation on the timescales from picoseconds to nanoseconds, relevant due to pulses in bursts. We developed material processing approaches that enable similar behaviour as single 70 ps pulse ablation to ultra-short pulses in terms of quality and burst-mode behaviour like nanosecond pulses in terms of efficiency. The variability of the fiber laser operation modes was studied and compared to both ultra-short and nanosecond pulses from standard laser sources.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Laser micro-processing is already an established technology for many applications, with notable industrially relevant examples being prototyping of printed circuit boards [1], recently even fully flexible ones [2], or copper in general [3,4], and dicing and scribing of brittle materials including ceramics and semiconductors [5]. Most off-the-shelf laser processing systems employ either ultra-short pulses below 10 ps in duration or pulses several 100 ps long. Ultra-short pulsed lasers have become widely used in the last decade in various settings due to either low heat effects on the material surrounding the target zone or the wide range of possible uses of a single laser source, for applications ranging from surface modification [6] to two-photon imaging [7]. On the other hand, nanosecond lasers represent a mature technology, offering high efficiency of both light generation and utilization, as the ablation efficiency for material processing applications tends to be high, but lack the precision and final surface quality achievable with ultra-short pulses [8].
Technological developments have enabled high laser average power and pulse energies, sparking the second wave of scientific interest in generating and applying bursts of pulses. Bursts of pulses at widely ranging repetition rates from tens of MHz [9–12], through single GHz [4,13,14], up to near-THz [15–18] were generated by different groups by either splitting a single high-energy laser pulse or grouping subsequent pulses together. The latter approach is enabled by contemporary fiber laser concepts, reliably and repeatedly resulting in controlled changes of the ablation process. The use of bursts opened a new, previously unexplored, parameter space, leading to observation and quantification of the involved ablation related phenomena mainly governed by heat accumulation or incubation, improving the ablation efficiency and shielding and re-deposition effects lowering it [10,19,20].
Using an advanced, custom made fiber laser concept evolved from the design described in [21], we generated 70 ps laser pulses without the need for mode-locked oscillators, resulting in increased modularity and adaptability of the concept, as well as lower laser cost. Metal ablation efficiency decrease has been observed with increasing pulse duration from several ps (i.e. exceeding the mechanical relaxation time [22]) to tens of ps [23–25], with the highest efficiency corresponding to nanosecond laser ablation [26]. Some contradicting results were obtained showing an increase of ablation efficiency with increasing pulse duration from 0.2 ps towards 10 ps [3]. Ultra-short pulses also exhibit the lowest residual heat transferred to the surrounding material [10,23]. Nevertheless, contemporary fiber laser concepts offer high variability of the output light, enabling an optimization potential of using single pulses and bursts of pulses at variable intra-burst repetition rates. In this work, we sought to discover and develop mechanisms and approaches suitable for both high throughput and high precision laser processing through laser parameter optimization and understanding of the near-surface ablation mechanisms.
Controlling the laser operation mode enables tailoring of the ablation process when coupled with a fundamental understanding of near-surface light-matter interaction mechanisms. Our experiments were aimed at observation and understanding the near-surface phenomena involved. Comparisons of ablation efficiency, surface topography, near-surface heat accumulation and ablation by-products were employed to study the differences emerging from the use of various bursts of pulses and different pulse durations. We examined the effects of changing the number of pulses in a burst (pulses per burst – PPB) isolated from either total burst fluence or single pulse energy, as well as the effects of switching the pulse duration. The experiments were performed on two widely available materials, copper layers for printed circuit board production and bulk silicon, both extensively studied in thermal properties research [27,28] and laser processing applications [3,18,29–31]. The optical properties of said materials vary greatly, as linear absorption dominates in the metallic materials, while highly energetic pulsed NIR light in silicon is absorbed dominantly through nonlinear processes [32]. Differences in optical and thermal properties are reflected in different ablation process behaviour.
The aim of this study is to recognize the mechanisms of picosecond pulse laser ablation in the laser parameter space opened due to burst mode operation. Optimized operation modes were achieved through observation and quantification of the ablation phenomena caused by time-domain burst shaping. A crucial aspect of the research was probing of near-surface heat-accumulation and consequent changes of the ablation process observed in surface modification, cutting speed, ablated depth, ablated volume measurements and crater shape changes. New material processing applications that combine high efficiency of nanosecond laser pulses and high precision of ultra-short pulses can be implemented due to a high modality of the laser outputs in a single-laser multi-step process [33], combining different approaches within a single structure.
2. Materials and methods
The experimental setup was designed and used in a way that mimicked an actual industrial environment approach to laser-based material processing. At the same time, the experimental setup was maximally adaptable to capabilities of the fiber laser source and consequently to the following observation of the laser ablation mechanisms and effects of the time scales involved (schematic representation in Fig. 1(a)).
Fig. 1. a) Experimental setup schematic, mimicking a typical industrial-standard laser processing machine. b) Time-dependent pulse intensity, as obtained by the fast photodiode and captured by the oscilloscope. Observed FWHM pulse duration is (70 ± 10) ps. Inset shows the nearly uniform energy distribution among the pulses in a burst, where pulse-to-pulse stability remains within 5% even for the longest bursts (10 PPB).
2.1 Experimental setup
We have used an in-house made NIR pulsed fiber laser emitting 70 ps long pulses at a 1030 nm wavelength, operating at variable inter-burst repetition rates in the hundreds of kHz range, coupled with a fast, external electro-optical modulator serving as a shutter. The laser design is a minor upgrade from our previous work [21], featuring a high-quality output beam (M2 = 1.12 ± 0.06). The time-dependent intensity distribution of the laser pulse was captured by an ultrafast fiber-coupled photodetector (AlphaLas UPD-15-IR2-FC, <15 ps rise time) and observed on a high-performance oscilloscope (Agilent DSO81204B Infiniium – 12 GHz bandwidth, 40 GSa/s) – a sample trace is presented in Fig. 1(b). The maximum pulse energies achieved by the laser were around 20 µJ.
The custom fiber laser source was based on a gain-switched seed diode, coupled to an acousto-optic modulator (AOM) and four fiber amplifier stages. By design of the laser source, the emitted pulses were grouped in bursts of 1-10 pulses, pre-compensated by the AOM for the uniform pulse energy to be achieved throughout the burst. Consequently, each sub-pulse had the same pulse duration as a single pulse and the pulse energy was uniformly distributed among sub-pulses (traces shown in insets of Fig. 1(b)). Two different intra-pulse repetition rates were defined by the AOM, 20 MHz and 40 MHz, as both are meaningful from a laser design standpoint and enable better research of nanosecond-scale phenomena involved in near-surface heat accumulation. Fiber amplifiers were set for either high average power or high pulse energy, and the all-fiber design guaranteed a high beam quality and no ellipticity.
Bursts of 1-10 pulses were guided to fast x-y galvo-scanners and were focused through a 100 mm f-theta lens onto the sample surface, resulting in a calculated 1/e2 laser spot diameter of 19 ± 1 µm, a maximal pulse fluence of about 8 J/cm2 and a maximal total burst fluence of about 90 J/cm2, both calculated from the pulse energy measured on material and the calculated laser spot diameter. The laser beam was always focused on the top material’s surface, which was ensured by a camera coupled into the laser beam path before the f-theta lens (as shown in Fig. 1(a)).
For comparison, we repeated the experiments using two established industrial systems for rapid prototyping, ProtoLaser ST and ProtoLaser R (both made by LPKF Laser & Electronics AG), integrating a nanosecond and single picosecond laser source, respectively. The ProtoLaser ST is a tabletop system aimed primarily for PCB prototyping applications (ablation of Cu layers), while the ProtoLaser R is a multipurpose high-precision laser structuring system. The diameters of the focused laser beams are specified as 22 µm and 15 µm for the ProtoLaser ST and ProtoLaser R devices, respectively. Both systems feature comparable optical setups to our experiment and use NIR laser sources that were set to power levels and repetition rates matching the rest of the experiments. The nanosecond laser in the ProtoLaser ST was set up in a way that emulated a maximum energy 5 PPB burst, as the total durations were comparable – 100 ns duration of the 5 PPB burst and the 120 ns single pulse. We used a combination of 100 kHz pulse repetition rate and 6.0 W average power output power measured at the sample plane. The resulting nanosecond pulse energy was 60 µJ, equal to the burst energy of a 5 PPB burst at a 12 µJ pulse energy. The picosecond laser in the ProtoLaser R was set to 100 kHz pulse repetition rate and 1.2 W average power, resulting in a 12 µJ pulse energy at the sample plane, enabling a direct comparison of ablation efficiency and effects to a single 70 ps pulse. Scanning speeds were set as described in the approach I below to achieve spot-to-spot overlaps from 0.2 to 0.9 for all laser systems.
2.2 Laser processing approaches
Experiments were set up for the observation of four different effects on the laser processing:
- 1) Isolation of time-scale effects within bursts by repeating experiments with the only variable being the intra-burst repetition rate.
- 2) Burst duration effects at a constant pulse energy, by observing burst-mediated changes to the ablated depth per pulse measured at various energy levels – at each PPB number, for this, a scan from below the threshold energy to the maximal available energy in the laser system was performed.
- 3) Burst duration effects at constant burst energy, by observing the ablation process using the same burst energy split into a different number of PPB.
- 4) Direct comparison with industrial standard laser sources using both significantly shorter (1 ps) and longer (120 ns) pulse durations, in terms of postprocessing surface quality evaluation and ablation efficiency measurements.
The first effect mainly isolates heat accumulation and incubation within a burst from shielding and re-deposition processes, due to short heat diffusion times. The second and third effects are expected to show the changes due to the interplay of heat accumulation and incubation increasing the ablation efficiency and shielding decreasing it at different experimental conditions. The direct comparison with other laser sources offers an insight into how pulse duration alone contributes to both efficiency and process quality (in terms of heat-induced effects and re-deposited material around the processed zone).
Two industry standard and widely researched, yet optically and thermodynamically fundamentally different materials were chosen for laser processing experiments: copper layers and bulk silicon. First, the copper layer for PCBs was used (18 µm thick metal layer on a dielectric substrate; Cu FR4, Žibret Laminati), and second a bulk semiconductor substrate (0.65 mm thick polished p-doped silicon, 1-10 Ωm resistivity; Siltronix). Due to the material’s form factor and different target measurements, the experiments aimed to observe the above-mentioned effects were performed using three different laser processing approaches:
- I. Channels in copper layers: An effective way of determining the ablation depth and efficiency in layered materials is cutting of single-line channels, simultaneously enabling the observation of ablation by-products on the surrounding virgin surface and diminishing the effects of the uneven surface on single ablation craters. Channels were cut by scanning the chosen line multiple times until the resulting channel formed a cut through the complete material thickness, with the number of scans being increased in steps to achieve an exact cut. Figure 2(a) shows an example of n-1 repetitions not resulting in a cut-through, n repetitions resulting in an exact cut, and n+1 repetitions already damaging the substrate.
Scanner speeds were set accordingly to achieve spot-to-spot overlap between 0.2 and 0.9, defined as one minus the ratio between the position displacement and the focused beam diameter. For example, at a 100 kHz (burst) repetition rate and a 1000 mm/s scanning speed, the consequent bursts were separated by 10 µm, resulting in an overlap of approximately 0.5. Finally, using the number of PPB, spot-to-spot overlap (o), and the copper layer thickness (D), the ablated depth per pulse was calculated as $d = D\ast ({1 - o} )/({n\ast \textrm{PPB}} )$.
- II. Separate ablation craters in silicon were chosen due to a highly homogeneous silicon surface for ablation efficiency measurements and ablation by-product observation. Scanners were programmed to generate long straight lines with a fixed scanning speed of 4000 mm/s, to create separate ablation craters by placing a single burst per spot (Fig. 2(b)). At a 100 kHz (burst) repetition rate, craters were isolated by 40 µm offsets.
- III. Pockets in bulk silicon were milled to improve the precision of ablation depth measurements, found to be difficult during the single crater topography analysis. 2 mm x 2 mm sized pockets were milled in the surface of bulk silicon (Fig. 2(c)) by scanning a crossed grid of straight lines with a defined grid density of 100 lines/mm. Scanner speeds were set accordingly to achieve spot-to-spot overlaps between 0.2 and 0.9 in the fashion described above.
Fig. 2. Different material processing approaches: a) cutting of channels, b) single-burst ablation craters, and c) milling of pockets.
The range of spot-to-spot overlaps allowed the monitoring of the effects of re-deposition, as it spanned the range of distances from highly overlapping spots (0.9) to barely overlapping (at 0.2). In short, the approach I enabled us to study the response of copper layer material to surface ablation, comparing the effects to ablation data using either ultra-short pulses with sub-10 ps duration or nanosecond pulses. Approaches II and III were used on bulk silicon and also enabled studies of ablation efficiency, surface properties, and other ablation-related phenomena.
Different spot-to-spot overlaps affected the laser processing dynamics on the microsecond time scales due to laser repetition rates in the multi-kHz regime, while bursts of pulse happened on a nanosecond scale on account of the underlying fiber laser design, and laser pulses themselves occurred on a time scale of picoseconds. The multitude of variable and controlled time scales involved in the experiment enabled us to study the ablation processes in detail.
We have measured the resulting structures with high magnification optical microscopy in both bright field and dark field imaging modes (Olympus BX53M microscope with Olympus MPlanFL 20x and 50x objectives). The optical measurements with focus plane recognition resulted in high vertical position sensitivity that was used for channel and single-crater dimension measurements. Craters in bulk silicon were imaged by darkfield optical microscopy and measured by a custom Matlab edge recognition script, typically averaging 10-20 craters at each laser setting.
3. Results and discussion
3.1 Copper layer
Cutting of channels, an approach shown in Fig. 2(a), was chosen for the complete evaluation of burst laser ablation in a copper layer. We measured the ablated depth per pulse and channel width. The resulting copper channels were examined in terms of surface morphology and quality, the latter mainly affected by the heat-induced effects.
3.1.1. Burst duration effects
The results regarding burst duration effects at constant pulse energy have been presented on the graph in Fig. 3, showing ablated depth per pulse in nm for two different intra-burst repetition rates. To achieve constant pulse energy at various burst settings, we used different burst repetition rates, due to laser design limitations (between 100 kHz and 167 kHz) and correspondingly, average power. The ablated depth per pulse was calculated as described in the Laser processing approaches subsection. The ablated depth per pulse would be independent of the burst length if there were no incubation/heat accumulation effects, as well as no shielding/redeposition effects.
Fig. 3. a) Ablated depth in copper per pulse, using 40 MHz and 20 MHz bursts with constant pulse energy (12 µJ) as a function of the PPB for various spot-to-spot overlaps. Increased uncertainty obtained at high PPBs was due to an increased average power resulting in fewer passes needed to cut through the copper layer.
This set of experiments was done at a fixed pulse energy of 12 µJ on the sample. By fixing the pulse energy, we isolated the effects of variable intra-burst burst lengths, and by repeating the whole set of experiments at two intra-burst repetition rates, we isolated the time-scale effects within bursts. At equal pulse energies and the pulse duration of 70 ps, the ablated depth per pulse was uniformly increased with an increasing PPB. The increase was greater when a 40 MHz intra-burst repetition rate was used compared to a 20 MHz repetition rate and was largely invariable to the spot-to-spot overlap caused by changes in the scanning speed. Both results are shown on the graph in Fig. 3.
The ablated depth per pulse is increased when processing with bursts up to a factor of 10 compared to single pulse processing, meaning that the amount of ablated material increases up to 10-fold per pulse or per µJ of energy input (as pulses all have equal energy). The volume of ablated material per unit energy is a commonly used measure of ablation efficiency. The increase in ablated depth per pulse becomes gradually larger with an increasing PPB number, signifying long-lived heat accumulation or a sustained molten phase. The difference between the incubation effects observed at different repetition rates can be explained by typical time constants imposed by the heat diffusion speed, i.e., thermal diffusivity, however detailed thermal simulations would be needed for confirmation.
Increasing the number of PPB at constant pulse energy leads to a linear increase in the total energy deposited per spot. Heat-induced effects were observed by other groups even with highly energetic ultra-short pulses [34]. We repeated the previous experiments at constant burst energy and varied burst lengths, again using different intra-burst repetition rates, to observe the heat accumulation and incubation separately from shieling and material re-deposition. We effectively split the energy carried within a single pulse into multiple pulses at constant burst energy of 16 µJ on the sample. This was the upper experimental limit for single pulse energy, so we could not scale the experiment to higher energies that would enable a more direct comparison with the previous results. Nonetheless, we observed an increase in ablated depth per pulse (and ablation efficiency) when splitting the energy to multiple pulses, with an optimum reached at 2-3 PPB (graphs in Fig. 4). Again, no clear dependency of the results was observed on the spot-to-spot overlap, and contrary to the constant pulse energy experiments, no clear change due to the intra-burst repetition rate change was detected. This could be a consequence of lower pulse energy at all settings above 1 PPB, leading to a much lower heat accumulation overall, regardless of the timescales involved.
Fig. 4. a) Ablated depth in copper per pulse, using 40 MHz and 20 MHz bursts with constant burst energy (16 µJ) as a function of the PPB for various spot-to-spot overlaps.
In this case, excessive addition of pulses into bursts caused a decline in the ablated depth per pulse. We believe that the main reason for this was due to the pulse energy scaling inversely with increasing PPB, causing less energy of a Gaussian distributed pulse to be utilized for effective ablation and more energy from the distribution’s tails to be lost to ineffective heating of the surroundings, staying below the ablation threshold (as per [25]). The previously observed heat accumulation responsible for the increased efficiency of long bursts was not high enough to counter this, mainly due to much lower pulse energy. Consequently, at higher burst energies the highest efficiency (depth) would presumably shift to a higher PPB number, however further experiments are needed to confirm this.
The invariability of the results of both fixed pulse energy and fixed burst energy experiments on spot-to-spot overlaps suggests that no significant energy transfer is occurring on the time scale of microseconds. The observed ablated depth per pulse (ablation efficiency) changes with burst duration and complementary effects on morphology point towards a highly efficient energy accumulation on 10-20 ns timescales, causing an increased temperature of the surrounding copper. In short time frames, defined by the intra-burst repetition rate, 25 ns for 40 MHz and 50 ns for 20 MHz, the heat cannot entirely diffuse away from the laser beam position fast enough, which contributes to the increased ejected molten material and the creation of burrs on the edge of the cut (results presented in Fig. 5). On the timescale of 50 ns, the energy accumulation mechanisms are significantly less pronounced compared to the faster repetition rate.
Fig. 5. Structural morphology of copper channels using intra-burst repetition rate of 40 MHz, 60% spot-to-spot overlap, and a) fixed pulse energy (12 µJ on the sample), b) fixed burst energy (16 µJ on the sample). Dashed lines on left channel sides indicate top channel edge (blue lines) and the outside affected zone border (yellow lines). Scale bar equals 50 µm.
Complementary to the ablated depth measurements, we observed changes to the structural morphology of the copper channels for both described experiments. For bursts with constant pulse energy, an example is shown in Fig. 5(a), at 40 MHz intra-burst repetition rate and 60% spot-to-spot overlap, with the number of pulses incrementing from 1 PPB to 10 PPB. For bursts with constant burst energy, an example at the same repetition rate and overlap is shown in Fig. 5(b).
Burst duration increase at constant pulse energy leads towards a significant increase of heat-induced effects. The overall channel quality drops with the increasing PPB, we observed onset of visible melt accumulation at 3 PPB, and with longer bursts also melt ejection. Copper channels also widen with the increasing PPB, therefore the ablation efficiency increases even faster than the ablated depth since the ablated volume is linearly correlated with the channel width.
On the other hand, the burst duration increase at constant burst energy leads towards improved overall channel quality, as individual pulse energies get smaller with the increasing PPB number. Copper channels shrink in width with the increasing PPB (Fig. 4(b)), therefore the ablation efficiency, in this case, decreases faster than the ablated depth.
At high pulse energies, the excess energy is mainly in the central part of the Gaussian intensity distribution, while at low pulse energies the excess energy not causing ablation, is distributed in the distribution’s tails. The heat-related effects are more pronounced in the first case due to a larger amount of energy available for conversion into heat as well as it being concentrated in the middle of the beam path. The quality changes observed in the experiment carried out at constant pulse energies are in part due to the described effect and in part due to using a single pulse energy well above the ablation threshold (estimated at 0.4 ± 0.1 J/cm2 more than 10-times below the fluences used). The energy build-up from multiple pulses with energies well above the threshold causes the reduced quality observed.
3.1.2. Comparison with standard laser sources
The ablation efficiency and overall surface quality achieved by our burst laser were compared with two industrial standard laser processing systems using NIR lasers with either 1 ps (ProtoLaser R) or 120 ns duration (ProtoLaser ST), the latter marketed specifically for printed circuit board prototyping. The same copper cutting technique was used as a benchmark, with resulting measurements presented for various spot-to-spot overlaps in Fig. 6(a), and optical images of resulting morphologies presented in Fig. 6(b).
Fig. 6. a) Comparison of ablation efficiency at different laser cutting regimes (1 ps pulses, 70 ps pulses at 1 PPB, 5 PPB, 10 PPB, and 120 ns pulses), and different spot-to-spot overlaps. b) Structural morphology of copper channels - comparison between laser sources and modes of operation at a 60% spot-to-spot overlap. Dashed lines on left channel sides indicate the top channel edge (blue lines) and the outside affected zone border (yellow lines). Scale bar equals 100 µm.
The nanosecond ablation regime proved to achieve the highest ablation efficiency overall, with a decline at high spot-to-spot overlaps (0.8 and above). At high overlaps, the burst regime at 10 PPB and 40 MHz inter-burst repetition rate turned out to be the most efficient. The ablation efficiency comparison was used to compare different laser power levels and presented as a factor of speed increase over the 70 ps 1 PPB regime, which was defined to the value of 1. The burst repetition rate was constant at 100 kHz for all data sets, while pulse and burst energy were not. The average powers of various laser sources used, as measured on the sample, are presented in Table 1 for comparison of the laser processing systems and operation regimes.
Table 1. Average power of the laser sources, measured at the sample’s position.
Both single-pulse regimes of picosecond duration were matched in terms of pulse energy for the most relevant comparison. The nanosecond source was set to match the burst energy of the 5 PPB (40 MHz intra-burst repetition rate) regime in a long pulse as the total durations were closely comparable. The combined results show that 70 ps pulses in 1 PPB regime exhibit the lowest ablation efficiency, but through burst-mode regime optimization (high intra-burst repetition rate, 5 or more PPB, high pulse energy), it is possible for a 70 ps laser to perform similarly to a nanosecond laser in terms of ablated depth per pulse (and consequently cutting speed/ablation efficiency) and also heat-induced effects.
The surface topography comparison of copper is shown for all the above-mentioned regimes in Fig. 6(b), the quality of which is finest using either 1 ps or 70 ps pulses. A much wider channel created by 1 ps pulses compared to 70 ps pulses in 1 PPB regime points to a much lower ablation threshold, thus also causing ablation well into the Gaussian intensity distribution tails. Comparing the different burst-mode regimes of the 70 ps laser, the quality worsens with the addition of pulses in a burst, gradually becoming indistinguishable from orders in magnitude longer nanosecond pulses in terms of channel width, heat-induced effects such as melting and melt ejection and burr on the edge of the channel.
3.2 Bulk silicon
The second set of experiments was performed on silicon, which exhibits different thermodynamic and optical properties compared to copper. Two significant differences are: deeper linear optical absorption depth and lower thermal diffusivity in Si compared to Cu at elevated temperatures [35,36]. A higher value of thermal diffusivity likely results in a bigger heat-affected zone in a homogeneous material under the same heating conditions. This approximation cannot be used on a single picosecond time scale, but it can be on longer time scales of bursts, where the heat-affected zone is roughly the maximum of the optical penetration depth and thermal diffusivity length [37]. We found that linear absorption alone is not enough to warrant ablation at the parameters used, as only around 0.3% of incoming energy is absorbed in the top 1 µm of the material at the 1.03 µm wavelength [38,39]. Therefore, the absorption of a laser pulse happens through both linear and nonlinear mechanisms.
First, we observed single-burst ablation craters in silicon surface and compared the results with a model for ultra-short pulsed laser ablation. The model predicts the ablation crater area ${\textrm{A}_\textrm{c}}$ as a function of the pulse energy:
where ${\textrm{w}_0}$ is the laser beam waist radius, Ep is the pulse energy, and Eth the threshold energy for ablation. The expression was established for laser processing by Liu et al. [40] and widely adopted since Furmanski et al. [41], by taking into account the Gaussian spatial intensity distribution of the pulse and basic ablation properties. We have found a good agreement between the model and the experimental data for single-pulse bursts as shown on the graphs in Fig. 7, where we used ${\textrm{w}_0}$ and ${E_{th}}$ as the free parameters for the fit. The agreement was found in both the general function shape and fit parameter values - converging to 8.4 ± 0.1 µm and 3.1 ± 0.1 J/cm2, respectively – compared to the calculated focused beam radius in the Experimental setup subsection.Fig. 7. a) Ablation of single craters into Si surface using 40 MHz bursts with constant pulse energy. Images show 1 PPB craters at various energies and the corresponding points on the graph are labelled accordingly (pulse energies chosen: 2.9 µJ, 4.4 µJ, 8 µJ, and 16 µJ on the sample). Scale bar equals 25 µm. The graph shows how the crater area changes with the increase of burst energies for bursts ranging from 1 PPB to 10 PPB and the model fitted to 1 PPB data. b) Ablation of single craters into Si surface using 40 MHz bursts with constant burst energy of 16 µJ on the sample. Images show the corresponding craters, except for 2 PPB – upper 13 µJ, lower 22 µJ, 3 PPB – upper 14 µJ, lower 19 µJ, and 4 PPB – upper 13 µJ, lower - 19 µJ. Scale bar equals 25 µm. The graph shows data gathered on dimensional scaling of the observed ablation craters.
The agreement with the model confirms that 70 ps pulses of low to medium fluences work in a regime similar to ultra-short pulse laser ablation, indicating a negligible energy transfer to the surrounding material in the range of used parameters. The model enables a continuous prediction of ablation effects between the measured data points, and limited predictions outside of the current system’s laser parameter space (e.g., at higher pulse energy), allowing for better optimization of material processing. A visual comparison of single-pulse (1 PPB) craters at various fluences from just above the threshold to medium fluence at the upper limit of our system is shown in Fig. 7(a) (selected examples). The area is changing according to the logarithmic law described in the equation above. The absence of melt ejection again points towards ablation behaviour mimicking ultra-short pulses.
Evaluation of burst-mode effects was planned in the same fashion as on the copper layer, with a goal to separately evaluate intra-burst repetition rate effects on the ablation process, burst duration effect at constant pulse energy, and burst duration effects at constant burst energy. Ablation of single craters was chosen as the evaluation method (method described in section 2). To avoid loss of precision in crater depth measurements due to surface irregularities and high aspect ratio, the ablated volume per burst was separately measured by milling of pockets. While the measurement does not accurately provide a single-crater depth due to changes in the ablation process [42], the relative scaling with energy is expected to be closely comparable.
The complete data on crater area energy dependency, for different PPB numbers at 40 MHz intra-burst repetition rate, is presented in the graph in Fig. 7(a). We observed an increasing deviation from the logarithmic law with the PPB number, as seemingly longer bursts behave ever more unlike a single ultra-short pulse of such energy would behave. Due to this, we did not apply the described logarithmic law to higher PPB data, as the general functional shape does not match the data. Apart from this behaviour, ablation with longer bursts does not increase the crater width. We observed a slight width decrease, where the values for crater area stayed below the line of the single-pulse ablation model throughout the burst energy range (graph in Fig. 7(a)). On the other hand, the depth of ablation increased significantly, signifying both a very small percentage of energy carried in the Gaussian intensity distribution tails and the presence of a strong incubation effect. The observed changes in the process led to significant or even prevalent melt ejection at high burst energies. This may be caused by a sustained long-living molten phase near the binodal border, at least for burst energies above 10 µJ, where also the crater area data showed a change in behaviour (steepness). The strength of depth increase phenomena is best illustrated through the measurement of the ablation depth increase in the milled pockets with the PPB number at equal burst energies (Fig. 7(b)).
A deeper understanding was sought through observations of both crater diameter and ablation depth for bursts with constant burst energy, again effectively splitting the energy carried within a single pulse to multiple within a burst. The different intra-burst repetition rates were used at a burst energy of 16 µJ on the sample. Visual morphology comparison of silicon craters shown in Fig. 7(b) was done at a 40 MHz intra-burst repetition rate. We observed a transition of the ablation properties through multiple regimes, from a clear ablation crater typically associated with ultra-short pulses at 1 PPB to dominant melt ejection at 10 PPB. Minimal visible melting occurred at 2 PPB with slight changes of the crater edge visible, which quickly reached a visible presence of melting at 3 PPB and ejection of the first micro-droplets. With the further splitting of energy into longer bursts, the melting became more pronounced with clear melt ejection visible at 5 PPB and transitioning into a dominant melt ejection regime at 10 PPB. The crater area was the biggest when processing with a lower number of PPB, which could be attributed to Gaussian spatial intensity distribution and ablation threshold levels. On the other hand, the depth was increased with the addition of pulses in bursts up to 9-fold compared to the 1 PPB regime. Our proposed explanation is that the energy efficiency of melt ejection is significantly better compared to material evaporation, as melt ejection can be expected upon reaching the boiling temperature [43], not requiring the energy for a full phase transition into vapour. The visual morphology comparison shows a complete transition from prevalent full ablated volume vaporization to prevalent melt ejection.
As the thermal diffusivity of silicon is rather low compared to Cu, it leads to a more efficient energy accumulation during the burst. We observed a reduced ablation threshold in terms of energy per pulse with multi-pulse bursts compared to single pulses (graph in Fig. 7), leading to a conclusion that the first few pulses either caused no ablation or a small amount of removed material while mainly preheating the material. This regime led to the highly increased melting observed. The overall ablation efficiency increases more than 5-fold with the addition of pulses in bursts, but the quality is affected and becomes comparable to typical nanosecond laser ablation results [44].
The ablation depth was measured separately as precise single crater analysis is very complex for spatial dimension, surface irregularities, and aspect ratios of our craters. Therefore, we performed milling of pockets into the bulk silicon as described in Fig. 2(e). The ratio of depths obtained at equal burst energies and fixed scanning parameters were used to compare the depth of ablation in Fig. 7(b).
4. Conclusion
This study was designed to experimentally observe and quantify the processes involved during laser ablation of surface layers on the timescales of tens of picoseconds to tens of nanoseconds relevant up to bursts of 70 ps. We have observed the effects of near-surface residual heat accumulation and redistribution as well as shielding and material re-deposition, both causing various changes to the ablation process due to bursts of pulses, postprocessing structure quality, and surface modifications, as well as to the overall ablation efficiency, presented through ablated volume, crater area and depth measurements.
We have compared the single-pulse ablation using 70 ps pulses in NIR to different burst mode settings, possible with the custom fiber laser source. To put the results into perspective, we have evaluated them by both comparing the ablation efficiency and surface morphology to industrial standard laser sources on a case study of copper layer ablation as well as by comparing the measured data to existing ultra-short pulse ablation models (ablation of silicon).
The following findings were observed:
- 1. Heat accumulation effects on intra-bursts timescales: silicon ablation behaves similarly on 25 ns and 50 ns timescales (below 40% difference in ablation efficiency), while for copper ablation the shorter time leads up to a 200% increase in ablation efficiency (e.g. at 5 PPB).
- 2. Heat accumulation on inter-burst time scales: varied spot-to-spot overlaps reveal a significant increase in efficiency on 10 µs timescales, only at the highest fluence settings - using high-energy bursts above 7 PPB.
- 3. Energy-scaling behaviour on silicon of single 70 ps pulse ablation at low to medium fluences is in line with the model for ultra-short pulse ablation.
- 4. Ablation efficiency increases due to operation in bursts of pulses up to 10-fold on copper and up to 5-fold on silicon, but processing quality degraded on both materials (e.g. increased heat affected zone, melt ejection to the surrounding virgin material surface).
- 5. Operation in bursts of pulses enabled higher average power output, higher fluences on material, and a choice of compromises between maximal processing quality and speed (i.e., ablation efficiency).
- 6. The comparison with industrial-standard laser sources shows a similar precision to single-picosecond lasers reached with optimized parameters using single 70 ps pulses and a similar efficiency to nanosecond lasers reached using energetic bursts of 10 PPB.
The findings are important for the utilization of gain-switched fiber laser designs, being easier to design, make, and maintain compared to ultra-short laser sources.
Funding
Javna Agencija za Raziskovalno Dejavnost RS (L2-9240, P2-0270).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. R. Rozman, B. Kmetec, B. Podobnik, D. Kovačič, and E. Govekar, “Optimisation of direct laser structuring of printed circuit boards,” Appl. Surf. Sci. 254(17), 5524–5529 (2008). [CrossRef]
2. F. Dupont, S. Stoukatch, P. Laurent, and J. M. Redouté, “Fine pitch features laser direct patterning on flexible printed circuit board,” Opt. Lasers Eng. 126, 105869 (2020). [CrossRef]
3. A. Žemaitis, P. Gečys, M. Barkauskas, G. Račiukaitis, and M. Gedvilas, “Highly-efficient laser ablation of copper by bursts of ultrashort tuneable (fs-ps) pulses,” Sci. Rep. 9(1), 12280 (2019). [CrossRef]
4. G. Bonamis, E. Audouard, C. Hönninger, J. Lopez, K. Mishchik, E. Mottay, and I. Manek-Hönninger, “Systematic study of laser ablation with GHz bursts of femtosecond pulses,” Opt. Express 28(19), 27702–27714 (2020). [CrossRef]
5. J. Bonse, K.-W. Brzezinka, and A. J. Meixner, “Modifying single-crystalline silicon by femtosecond laser pulses: an analysis by micro Raman spectroscopy, scanning laser microscopy and atomic force microscopy,” Appl. Surf. Sci. 221(1-4), 215–230 (2004). [CrossRef]
6. D. W. Müller, T. Fox, P. G. Grützmacher, S. Suarez, and F. Mücklich, “Applying Ultrashort Pulsed Direct Laser Interference Patterning for Functional Surfaces,” Sci. Rep. 10(1), 3647 (2020). [CrossRef]
7. M. Sibai, H. Mehidine, F. Poulon, A. Ibrahim, P. Varlet, M. Juchaux, J. Pallud, B. Devaux, A. Kudlinski, and D. Abi Haidar, “The Impact of Compressed Femtosecond Laser Pulse Durations on Neuronal Tissue Used for Two-Photon Excitation Through an Endoscope,” Sci. Rep. 8(1), 11124 (2018). [CrossRef]
8. B. N. Chichkov, C. Momma, S. Nolte, F. von Alvensleben, and A. Tünnermann, “Femtosecond, picosecond and nanosecond laser ablation of solids,” Appl. Phys. A 63(2), 109–115 (1996). [CrossRef]
9. J. Mur, J. Petelin, N. Osterman, and R. Petkovšek, “High precision laser direct microstructuring system based on bursts of picosecond pulses,” J. Phys. D: Appl. Phys. 50(32), 325104 (2017). [CrossRef]
10. D. J. Förster, S. Faas, S. Gröninger, F. Bauer, A. Michalowski, R. Weber, and T. Graf, “Shielding effects and re-deposition of material during processing of metals with bursts of ultra-short laser pulses,” Appl. Surf. Sci. 440, 926–931 (2018). [CrossRef]
11. B. Neuenschwander, B. Jaeggi, E. V. Zavedeev, N. R. Arutyunyan, and S. M. Pimenov, “Heat accumulation effects in laser processing of diamond-like nanocomposite films with bursts of femtosecond pulses,” J. Appl. Phys. 126(11), 115301 (2019). [CrossRef]
12. A. Žemaitis, M. Gaidys, P. Gečys, M. Barkauskas, and M. Gedvilas, “Femtosecond laser ablation by bibursts in the MHz and GHz pulse repetition rates,” Opt. Express 29(5), 7641–7653 (2021). [CrossRef]
13. C. Kerse, H. Kalaycıoğlu, P. Elahi, B. Çetin, D. K. Kesim, Ö. Akçaalan, S. Yavaş, M. D. Aşık, B. Öktem, H. Hoogland, R. Holzwarth, and F. Ö. Ilday, “Ablation-cooled material removal with ultrafast bursts of pulses,” Nature 537(7618), 84–88 (2016). [CrossRef]
14. B. Neuenschwander, B. Jaeggi, D. J. Foerster, T. Kramer, and S. Remund, “Influence of the burst mode onto the specific removal rate for metals and semiconductors,” J. Laser Appl. 31(2), 022203 (2019). [CrossRef]
15. R. Stoian, M. Boyle, A. Thoss, A. Rosenfeld, G. Korn, I. V. Hertel, and E. E. B. Campbell, “Laser ablation of dielectrics with temporally shaped femtosecond pulses,” Appl. Phys. Lett. 80(3), 353–355 (2002). [CrossRef]
16. A. Klini, P. A. Loukakos, D. Gray, A. Manousaki, and C. Fotakis, “Laser Induced Forward Transfer of metals by temporally shaped femtosecond laser pulses,” Opt. Express 16(15), 11300–11309 (2008). [CrossRef]
17. C. Gaudiuso, G. Giannuzzi, A. Volpe, P. M. Lugarà, I. Choquet, and A. Ancona, “Incubation during laser ablation with bursts of femtosecond pulses with picosecond delays,” Opt. Express 26(4), 3801–3813 (2018). [CrossRef]
18. J. Mur and R. Petkovšek, “Near-THz bursts of pulses – Governing surface ablation mechanisms for laser material processing,” Appl. Surf. Sci. 478, 355–360 (2019). [CrossRef]
19. A. Semerok and C. Dutouquet, “Ultrashort double pulse laser ablation of metals,” Thin Solid Films 453-454, 501–505 (2004). [CrossRef]
20. M. Spellauge, J. Winter, S. Rapp, C. McDonnell, F. Sotier, M. Schmidt, and H. P. Huber, “Influence of stress confinement, particle shielding and re-deposition on the ultrashort pulse laser ablation of metals revealed by ultrafast time-resolved experiments,” Appl. Surf. Sci. 545, 148930 (2021). [CrossRef]
21. J. Petelin, B. Podobnik, and R. Petkovšek, “Burst shaping in a fiber-amplifier chain seeded by a gain-switched laser diode,” Appl. Opt. 54(15), 4629–4634 (2015). [CrossRef]
22. J. Winter, M. Spellauge, J. Hermann, C. Eulenkamp, H. P. Huber, H. P. Huber, M. Schmidt, and M. Schmidt, “Ultrashort single-pulse laser ablation of stainless steel, aluminium, copper and its dependence on the pulse duration,” Opt. Express 29(10), 14561–14581 (2021). [CrossRef]
23. C. Y. Chien and M. C. Gupta, “Pulse width effect in ultrafast laser processing of materials,” Appl. Phys. A 81(6), 1257–1263 (2005). [CrossRef]
24. A. Ancona, S. Döring, C. Jauregui, F. Röser, J. Limpert, S. Nolte, and A. Tünnermann, “Femtosecond and picosecond laser drilling of metals at high repetition rates and average powers,” Opt. Lett. 34(21), 3304–3306 (2009). [CrossRef]
25. B. Neuenschwander, B. Jaeggi, M. Schmid, V. Rouffiange, and P.-E. Martin, “Optimization of the volume ablation rate for metals at different laser pulse-durations from ps to fs,” in Laser Applications in Microelectronic and Optoelectronic Manufacturing (LAMOM) XVII (International Society for Optics and Photonics, 2012), 8243, p. 824307.
26. A. Semerok, C. Chaléard, V. Detalle, J.-L. Lacour, P. Mauchien, P. Meynadier, C. Nouvellon, B. Sallé, P. Palianov, M. Perdrix, and G. Petite, “Experimental investigations of laser ablation efficiency of pure metals with femto, pico and nanosecond pulses,” Appl. Surf. Sci. 138-139, 311–314 (1999). [CrossRef]
27. E. Matthias, M. Reichling, J. Siegel, O. W. Käding, S. Petzoldt, H. Skurk, P. Bizenberger, and E. Neske, “The influence of thermal diffusion on laser ablation of metal films,” Appl. Phys. A 58(2), 129–136 (1994). [CrossRef]
28. H. Wada and T. Kamijoh, “Thermal Conductivity of Amorphous Silicon,” Jpn. J. Appl. Phys. 35(Part 2, No. 5B), L648–L650 (1996). [CrossRef]
29. M. Hashida, A. F. Semerok, O. Gobert, G. Petite, Y. Izawa, and J. F.- Wagner, “Ablation threshold dependence on pulse duration for copper,” Appl. Surf. Sci. 197-198, 862–867 (2002). [CrossRef]
30. J. Byskov-Nielsen, J.-M. Savolainen, M. S. Christensen, and P. Balling, “Ultra-short pulse laser ablation of copper, silver and tungsten: experimental data and two-temperature model simulations,” Appl. Phys. A 103(2), 447–453 (2011). [CrossRef]
31. D. Bachman, Z. Chen, R. Fedosejevs, Y. Y. Tsui, and V. Van, “Permanent fine tuning of silicon microring devices by femtosecond laser surface amorphization and ablation,” Opt. Express 21(9), 11048–11056 (2013). [CrossRef]
32. T. Boggess, K. Bohnert, K. Mansour, S. Moss, I. Boyd, and A. Smirl, “Simultaneous measurement of the two-photon coefficient and free-carrier cross section above the bandgap of crystalline silicon,” IEEE J. Quantum Electron. 22(2), 360–368 (1986). [CrossRef]
33. T. Kurita, K. Kawai, T. Morita, T. Iguchi, and Y. Kato, “Simultaneous material processing by picosecond-pulse bursts and nanosecond pulses arbitrarily generated by a directly modulated laser diode system,” OSA Continuum 3(6), 1711–1720 (2020). [CrossRef]
34. J. Schille, L. Schneider, and U. Loeschner, “Process optimization in high-average-power ultrashort pulse laser microfabrication: how laser process parameters influence efficiency, throughput and quality,” Appl. Phys. A 120(3), 847–855 (2015). [CrossRef]
35. K. Yamamoto, T. Abe, and S. Takasu, “Thermal Diffusivity of Crystalline and Liquid Silicon and an Anomaly at Melting,” Jpn. J. Appl. Phys. 30(Part 1, No. 10), 2423–2426 (1991). [CrossRef]
36. J. P. Holman, Heat Transfer (McGraw-Hill, 2008).
37. T. V. Kononenko, S. V. Garnov, S. M. Pimenov, V. I. Konov, V. Romano, B. Borsos, and H. P. Weber, “Laser ablation and micropatterning of thin TiN coatings,” Appl. Phys. A 71(6), 627–631 (2000). [CrossRef]
38. M. A. Green, “Self-consistent optical parameters of intrinsic silicon at 300 K including temperature coefficients,” Sol. Energy Mater. Sol. Cells 92(11), 1305–1310 (2008). [CrossRef]
39. C. Schinke, P. Christian Peest, J. Schmidt, R. Brendel, K. Bothe, M. R. Vogt, I. Kröger, S. Winter, A. Schirmacher, S. Lim, H. T. Nguyen, and D. MacDonald, “Uncertainty analysis for the coefficient of band-to-band absorption of crystalline silicon,” AIP Adv. 5(6), 067168 (2015). [CrossRef]
40. J. M. Liu, “Simple technique for measurements of pulsed Gaussian-beam spot sizes,” Opt. Lett. 7(5), 196–198 (1982). [CrossRef]
41. J. Furmanski, A. M. Rubenchik, M. D. Shirk, and B. C. Stuart, “Deterministic processing of alumina with ultrashort laser pulses,” J. Appl. Phys. 102(7), 073112 (2007). [CrossRef]
42. B. Jaeggi, B. Neuenschwander, S. Remund, and T. Kramer, “Influence of the pulse duration and the experimental approach onto the specific removal rate for ultra-short pulses,” in Laser Applications in Microelectronic and Optoelectronic Manufacturing (LAMOM) XXII (International Society for Optics and Photonics, 2017), 10091, p. 100910J.
43. K. Hirano, R. Fabbro, and M. Muller, “Experimental determination of temperature threshold for melt surface deformation during laser interaction on iron at atmospheric pressure,” J. Phys. D: Appl. Phys. 44(43), 435402 (2011). [CrossRef]
44. D. Paeng, J.-H. Yoo, J. Yeo, D. Lee, E. Kim, S. H. Ko, and C. P. Grigoropoulos, “Low-Cost Facile Fabrication of Flexible Transparent Copper Electrodes by Nanosecond Laser Ablation,” Adv. Mater. 27(17), 2762–2767 (2015). [CrossRef]
