High-rate laser microprocessing using a polygon scanner system

High-pulse repetition frequency (PRF) ultrashort laser sources supplying high average laser power of more than 100 W have recently become available, allowing excellent machining quality and high processing throughput at the same time. Initial studies in this field are focused on microvias generation,¹ microhole processing in steel and copper foils,² refractive index changes in transparent materials,³ and laser ablation of microcavities.^4,5 In addition, the potential of the technology has been already demonstrated by means of specific machining examples, such as micropyramids and embossing tools.^6–9 

The influence of laser fluence and pulse repetition rate on material removal rate (MRR) and process efficiency was studied in Refs. 6 and 9 by using a high-PRF picosecond laser source. By applying an average laser power of 3 W, a maximum removal rate of 0.16 mm³/min was achieved with pulse repetition rates between 50 and 100 kHz. Due to constant average laser power, the pulse energy was gradually reduced while the pulse repetition rate increased up to 300 kHz, causing the decrease of the ablation rates. Furthermore, higher ablation rates can be achieved performing multipass irradiation in the “low-fluence” regime while the laser fluence is above the threshold limit. Therefore, it has been shown that the material removal rate can be increased up to 7 mm³/min at a pulse repetition rate of 20 MHz.¹⁰ 

However, by using high-PRF laser sources fast beam deflection systems are required prospectively permitting a processing speed of several 100 m/s. A first approach of a fast laser beam deflection system was realized by means of an acousto-optic deflector.⁷ Therewith, a laser engraving process was performed on the outer mantle of a fast rotating cylinder with a processing speed of 40 m/s while the material removal rate was 3 mm³/min.

Laser micromachining with considerably higher scan speeds was conducted by the use of a polygon scanner in combination with a single mode cw fiber laser.¹¹ In this study, processing speeds of 300 m/s have been demonstrated by irradiating a continuous wave laser beam of 2 kW laser power and a spot size of 21 μm on metal sheets.

First, results obtained in large-area microprocessing demonstrated once more the potential of the innovative high-PRF ultrashort pulse laser technology. A ripple pattern was generated on stainless steel with a dimension of 80 × 80 mm in width and length while the processing rate was 25 cm²/min.¹⁰ 

Research activities with high-PRF ultrashort pulse laser radiation yielded novel phenomena in laser beam material interaction, those are in contrast to low repetition rate ultrashort pulse laser processing.^2,5,12,13 On the one hand, a decrease of the ablation rate was observed with pulse repetition rates of several 100 kHz. For this, plasma shielding induced by interaction of the incident laser pulse with the plasma plume, which originated from the previous laser pulses, was suggested. On the other hand, the ablation rate rises when laser pulses with a frequency in the megahertz range impinge onto the surface of a poor heat-conducting material like stainless steel, potentially due heat accumulation effects. Moreover, a further phenomenon called microcone formation on the scale of some 10 μm was observed in high-PRF ultrashort pulse processing with femtosecond laser pulses. Shape, size, and density of the structures could be influenced by the laser parameters. Even a selective elimination of the structures was realized that is in contrast to picosecond laser pulses.¹⁴ Microcone formation can modify the properties of technical grade surfaces with respect to wettability, roughness, and reflection characteristics.

In conclusion, high average power high-PRF laser sources combined with ultrafast beam deflection systems seems to be a promising technology in order to scale the productivity of high-PRF ultrashort pulse micromachining with respect to processing throughput and large-area laser processing. In this work, a novel machining setup consisting of a high-PRF high average power picosecond laser source and an ultrafast polygon scanner was investigated. The study was concentrated on high-rate ablation of bulk material as well as selective large-area thin film removal from bulk substrates. Finally, the potential of this technology was demonstrated on large-area processed samples.

In this study, a high-PRF picosecond laser source of the PX series from Edge wave GmbH emitting a linearly polarized Gaussian beam at the fundamental wavelength of 1064 nm was used. Including all losses in the optical beam path, the laser supplied an average power of 75.9 W onto the work piece surface. The maximum repetition rate was 20 MHz while the pulse width was 10 ps. Further, the maximum available single pulse energy decreased with rising repetition rate as a consequence of limited average laser power. For ultrafast beam deflection across the sample surface, an in-house developed polygon scanner was utilized. As a special feature, the polygon scanner system is equipped with an additional deflection unit permitting two-dimensional raster scanning. The laser beam was focused onto the material surface with a spot size of 44 μm by using an f-theta objective which allows a maximum scan field of 325 × 325 mm. With this setup scan speeds v_sc up to 800 m/s were achieved. A schematic of the setup is plotted in Fig. 1.

In general, the maximum utilization rate η_f,max of a single polygon facet was limited by the technical factors of the geometrical arrangement of the optical components of the polygon scanner. For the polygon scanner used in this study, the maximum utilization rate η_f,max was determined to be 49%. With respect to the used f-theta objective, this corresponds to a maximum processable length l_max of 325 mm. In conclusion, for laser processing with the maximum possible length l_max of 325 mm performed with more than one facet a duty cycle with a ratio of 49:51 has to be considered. As a result, the maximum achievable effective processing speed v_eff,max corresponds to the duty cycle and is approximately half of the scan speed v_sc, see the following equation:

Furthermore, a shorter processing length results in a less effective speed v_eff given by the ratio η_f of the current processed length l_c and the maximum length l_max. Hence, the effective processing speed v_eff can be calculated from the below equation

In this study, the ablation characteristics of two different types of materials were investigated: 1.5 mm thick technical grade stainless steel (AISI 304) for high-rate volume ablation and a 75 nm thick silicon nitride thin film deposited on a multicrystalline silicon wafer in order to demonstrate high-rate large-area ablation.

To identify the maximum achievable removal rate cavities were processed into stainless steel by using the “line-by-line” and “layer-by-layer” raster scan regime characterized by the lateral pulse distance d_p between two consecutive incident pulses and the hatch distance d_h between lines. D_c is the cavity depth and n_s is the number of scans representing the quantity of processed layers. The averaged ablated volume per single pulse V_sp can be calculated according to the below equation, taken from Ref. 15 

The cavity depth d_c was determined at five individual positions at the cavity bottom by means of a confocal point sensor MicroScan from NanoFocus AG. From Eq. (3), the ablation rate d_z can be estimated considering the average number of incident laser pulses per area associated with the focal spot radius w₀

The MRR as an indicator for the process efficiency results from the averaged ablated volume per single pulse V_sp and the repetition rate f_r and is given by

For large-area laser processing, the efficiency is characterized by the processing rate A_p which can be calculated in the following equation:

It is noteworthy, that the processing rates A_p are realistic rates provided by the setup because the effective processing speed v_eff was used.

The surface roughness S_a was measured in accordance with ISO 25178 utilizing a measurement arrangement consisting of the Confocam C101 (confovis) and the LV100D-U microscope (Nikon). The measurement data were analyzed with the mountains map^® software.

In this work, high-rate laser ablation of stainless steel by applying an average laser power of 75.9 W of a high-PRF picosecond laser source and an ultrafast polygon scanner was studied. Even at the highest pulse repetition rate of 20 MHz, the supplied laser fluence at the material surface of 0.5 J/cm² exceeds the ablation threshold for stainless steel H_th ∼ 0.1–0.15 J/cm² by a factor of 4.

The ablation experiments were carried out by applying the maximum average laser power of 75.9 W onto the stainless steel surface, while the lateral pulse distance and the hatch distance between lines were kept constant at 10 μm. By variation of the repetition rate between 1 and 20 MHz, the laser fluence dropped continually beginning at 10 J/cm² at 1 MHz down to 0.5 J/cm² at 20 MHz. Figure 2 depicts the ablation rate as a function of the repetition rate for 100 scans. The ablation rate was taken from the cavity depth obtained.

In addition, the laser fluence versus the repetition rate is plotted. The maximum ablation rate of 23 nm/pulse was achieved at the lowest investigated repetition rate of 1 MHz corresponding to the maximum applied laser fluence of 10 J/cm². With rising repetition rate, an exponential decrease of the ablation rate down to 3 nm/pulse was observed mainly caused by lower laser fluences.

In order to gain information on the process efficiency, the volume ablation rate and the material removal rate reveal more detailed facts. Figure 3 plots the volume ablation rate per laser pulse as well as the material removal rate versus the repetition rate for a number of 100 scans. As expected, the largest volume ablation rate was achieved by using the lowest investigated repetition rate corresponding to the highest laser fluence of 10 J/cm². By applying higher repetition rates, the volume ablation rate decreases, as shown in Fig. 2. However, the material removal rate, which results from scaling of the volume ablation rate with the repetition rate increased with rising repetition rate although the amount of the volume ablation rate dropped. The maximum material removal rate was found to be 5.4 mm³/min at a laser fluence of 0.5 J/cm² and a repetition rate of 20 MHz. The highest ablation efficiency for ultrashort laser pulses was indicated for laser fluences about 7.4 times above the ablation threshold.¹⁶ Within the presented investigations performed with a picosecond laser system, the supplied fluence at the highest material removal rate exceeds the threshold fluence by a factor of 4. So it can be assumed that the material removal rate can be maximized by irradiating laser pulses with a laser fluence in the range of 0.8 J/cm². For this, a higher average laser power is needed. It is noteworthy to point out that the maximum material removal rate of 5.4 mm³/min is related to the scan speed of the polygon scanner, which was 200 m/s. Considering the maximum utilization rate of the polygon scanner of 49% used in the investigations, the maximum effective speed was 98 m/s yielding to a maximum material removal rate of 2.65 mm³/min.

In ultrashort pulse laser processing, there is a controversial discussion regarding efficiency analyses between picosecond laser ablation and femtosecond laser ablation. The maximum achieved material removal rate of 5.4 mm³/min in stainless steel, based on the scan speed, is lower in comparison to the removal rate performed with high-PRF femtosecond laser pulses reported in Ref. 10. There, a scan speed based material removal rate of 6.8 mm³/min was determined with an applied laser fluence of 0.85 J/cm² and a repetition rate of 19.3 MHz by using a galvanometer scanner system. While the material removal rate achieved with femtosecond laser pulses was obtained at the optimum laser fluence of eight times above the threshold level, the rate for picosecond laser pulses was lower whereas the applied fluence was below the optimum fluence.

Against this background, similar material removal rates can be assumed at identical laser fluences. Therefore, at first glance there is no preference in efficiency between the picosecond and the femtosecond regime. But, if the average laser power input is taken into account, in the picosecond regime an average laser power of 75.9 W was applied. The investigations with femtosecond laser pulses were performed with an average laser power of only 31.7 W. To put the rates on a comparable basis, the rates were normalized with the laser power leading to a material removal rate of 0.09 and 0.21 mm³/W min for picosecond laser pulses and femtosecond laser pulses, respectively. Even though identical material removal rates of 6.8 mm³/min, the laser-power-normalized rates differ by a factor of more than 2 indicating a more efficient ablation process of stainless steel for processing with femtosecond laser pulses.

For a constant total energy input within the ablation process, two different regimes were compared: the “high-fluence” regime with 10 J/cm² and the low-fluence regime with 0.5 J/cm². Because of the maximum average laser power of 75.9 W, the repetition rate was 1 MHz for 10 J/cm². Choosing the maximum repetition rate of 20 MHz, the laser fluence of the pulses irradiated to the material surface amounts to be 0.5 J/cm². In order to provide a constant total energy input in the high-fluence regime 100 scans and in the low-fluence regime 2000 scans were performed. Furthermore, the processing time was identically. For the high-fluence regime, a volume ablation rate of 2.05 mm³/min was achieved. However, in the low-fluence regime an increase of the ablation rate of 2.58 mm³/min was obtained corresponding to a growth of 25%.

Beside the ablation rates, the surface quality and the roughness of the ablated areas were evaluated. In Fig. 4 scanning electron microscope (SEM) photographs are presented, demonstrating the surface morphology obtained with various laser fluences. It can be seen that the surface quality was strongly influenced by the applied processing parameters. The smoothest surface was obtained by applying the lowest fluence of 0.5 J/cm² and the highest repetition rate of 20 MHz. Highly regular ripple formation with a spatial period of around 1 μm, which correlates with the laser wavelength of 1064 nm of the laser beam,^17–21 as well as starting microcrater development in the grooves between ripples can be seen. Neither debris nor molten bulges can be observed on the structure surface indicating a minor thermal load of the material, in spite of high average power impinging on the sample. This is the result of the low applied laser fluence of 0.5 J/cm² and very high scan speed of 200 m/s. Further, with increasing laser fluence the surface appears more roughly. On the bottom of the cavity microcraters became more pronounced, growing into deeper regions followed by formation of microcones when laser fluence was increased up to the high-fluence regime with gradually lower scan speed down to 10 m/s at a laser fluence of 10 J/cm². The surface structures were covered with ripples.

The results of surface roughness measurement, listed in Fig. 4(f), correlate with the surface morphologies presented in Figs. 4(a)–4(e). The smoothest surface, achieved with the highest repetition rate of 20 MHz and the lowest laser fluence of 0.5 J/cm² is characterized by a surface roughness of 0.6 μm. Surface roughness increases with higher laser fluences and sinking repetition rates due to pronounced formation of microcraters and microcones. The achieved roughness value of 1.9 μm at a laser fluence of 2 J/cm² is close to the value reported in Ref. 22. In similar experiments in stainless steel, performed with picosecond laser radiation at a fluence of 2 J/cm², a repetition rate of 150 kHz, and a comparable ratio between lateral pulse distance and focal spot diameter, a profile roughness value R_a of 1.6 μm was determined. In conclusion, by applying low fluences and high repetition rates, maximum material removal rates and best surface quality were achieved at the same time. These results show a similar tendency observed and discussed in high-PRF laser ablation with femtosecond laser pulses.¹⁰ 

Up to now, laser microprocessing was limited on comparatively small areas of a few square centimetres. However, with the investigated laser system, a maximum scan field of 325 × 325 mm corresponding to an area of 0.1 m² can be processed in one pass. The surface of a stainless steel sheet with a dimension of 265 mm in length and 250 mm in width was laser processed, presented in Fig. 5.

With laser pulses supplied at a repetition rate of 20 MHz and a lateral pulse distance of 40 μm, a scan speed of 800 m/s was realized. If the current processed length of 265 mm is taken into account, an effective processing speed of 320 m/s resulted. Further, it yielded to a processing rate of 7680 cm²/min for one scan which is evident with a processing time of 5.2 s while the hatch distance was also 40 μm. On closer inspection, the processed field is characterized by barrel and pincushion distortion arising from the so far uncorrected optical system of the polygon scanner. However, the scanner control provides a feature for scan field correction which will be tackled in a next step in order to improve the accuracy of the scanner system.

High-rate large-area laser ablation of a thin film was conducted on 6 in. (156 × 156 mm) multicrystalline silicon wafers covered with an approximately 75 nm thick silicon nitride layer for photovoltaic application. The challenge of the investigation was to remove the layer within one pass without any damage of the silicon substrate.

First, a grid of lines with a hatch distance of 100 μm was laser processed by applying a fluence of 1 J/cm² and a repetition rate of 10 MHz while the lateral pulse distance was 20 μm corresponding to a scan speed of 200 m/s. Figure 6 shows an optical microscope image of the processed wafer.

The silicon nitride layer was completely removed within the lines showing a width of 30 μm without damage of the underlying silicon. Considering the effective processing speed, a processing rate of 2800 cm²/min was obtained yielding to a processing time of the entire wafer surface of 5.2 s. A more detailed consideration revealed slight variations of the hatch distance between adjacent lines. This is due to insufficient manufacturing precision of the polygon wheel, which is also a task of prospective scan field correction. By applying a laser fluence of 2 J/cm², a repetition rate of 5 MHz, a lateral pulse distance, and a hatch distance of 20 μm, and a resulting scan speed of 100 m/s the silicon nitride layer was completely removed from the silicon wafer. For a processed field of 136 × 136 mm the processing rate was 245 cm²/s, according to a processing time of 45.5 s. As a special feature of the polygon scanner, multiple line segments can be produced while one facet passes the laser beam realized with fast laser switching as a preliminary stage of a so called pixel mode. The challenge is an exact timing of the scanner and the control at high scan speed, in order to process the adjacent line segments with a high lateral precision. The performance capability was demonstrated by selective ablation of the silicon nitride layer, shown in Fig. 7. The pattern consisted of 16 ablated fields with a line segment width of 4 mm processed at a scan speed of 200 m/s. It is remarkable that there is only a marginal spatial jitter regarding the starting points of the line segments, attesting a very high degree of temporal synchronization of the scanner system.

High-rate laser microprocessing by using a high average power high-PRF picosecond laser source in combination with an in-house developed polygon scanner system, equipped with a two-dimensional scanning capability, has been investigated in this study. By applying a raster scanning regime, high-rate ablation of bulk material as well as high-rate large-area processing was analyzed in detail.

Stainless steel was irradiated with a maximum average laser power of 75.9 W. The highest material removal rate of 5.4 mm³/min was achieved in the low-fluence regime at the highest available repetition rate of 20 MHz. In addition, with these parameters the best surface quality was obtained, characterized by a minimum surface roughness of 0.6 μm.

With the presented laser system, the capability of high-rate large-area laser processing was demonstrated on a stainless steel sheet. A field size of 265 mm in length and 250 mm in width was processed with a processing rate of 7680 cm²/min for one scan. Further, thin film ablation of a silicon nitride layer from bulk silicon substrate was investigated. The layer was completely removed in one pass in a field of 136 × 136 mm without any damage of the silicon while the processing rate was 245 cm²/min. Furthermore, selective thin film removal was performed. For this, a grid of lines with a hatch distance of 100 μm was laser processed with a processing rate of 2800 cm²/min.

The presented results have been conducted in the course of the project “Innoprofile Transfer-Rapid Micro/Hochrate-Laserbearbeitung” (03IPT506X), funded by the Federal Ministry of Education and Research.

Udo Loeschner is graduated in Physical Engineering from the University of Applied Sciences Mittweida (Germany) in 1998. Thereafter, he has been a R&D engineer at the laser institute of the same university. Since 2006 he joined the Rapid Micro Tooling research group to investigate laser microprocessing with short and ultrashort pulse laser technologies. He completed his Ph.D. thesis at the Technische Universitaet Ilmenau in 2007. In 2011, he became an endowed professorship and represents the appointment area “high rate laser processing.”