Introduction

In most manufacturing applications with ultrashort pulse lasers, it is still an unsolved problem to productively use the now available high average laser powers of hundreds of watts while maintaining high precision. This is currently seen as one of the greatest challenges to further increase productivity in manufacturing with ultrashort pulse lasers [1]. In processing of metals, efficient high quality material removal requires the incident fluence to be within 5 to 15 times the ablation threshold [2]. To increase the material removal rate (MRR) within this regime by using more average laser power, either the pulse repetition rate or the irradiated area have to be increased accordingly. However, these measures quickly lead to a loss of precision of the processing results due to heat accumulation [3, 4].

Various approaches have been explored to boost MRR within these constraints such as multi-spot parallelization with diffractive optical elements [5, 6], beam shaping [7, 8], ultrafast scanning with rotating elements [9,10,11], and with hybrid scanners using acousto-optical elements [6, 12]. These approaches leverage technologies that enable a spread of the pulse energy onto an increased irradiated area, while avoiding heat accumulation by means of ultra-fast scanning.

In some cases, the use of burst pulses can increase productivity. The main effect is the division of single pulses with high energy into multiple burst pulses, leading to an increase in the energy-specific ablation volume per pulse. To realize an enhancement in ablation rate, shielding must be taken into account due to the short time intervals between burst pulses [13,14,15]. Long burst trains yield an exceptionally high energy-specific volume similar to single pulses of same duration. However, such processes operate outside the regime of evaporation alone [16,17,18,19].

A different strategy for high MRR and high quality results is splitting the process into roughing and finishing [20,21,22,23]. This approach employs subsequent steps in different ablation regimes. The first focuses on removing the majority of the material with high MRR at the expense of surface quality degradation. The second step then enhances surface quality, e.g. by remelting [24]. Building upon these strategies, recent research has been devoted to hybrid processes that involve iteratively performing limited quality ablation and high quality finishing [23].

All these processes were demonstrated starting on a smooth flat surface. This raises the question, whether it is possible to apply a high MRR regime for ablation starting on an arbitrary surface with high roughness or other deviations in shape and still achieve a polished surface finish with sub-µm roughness, which fulfills common industrial requirements towards application-ready parts [25,26,27].

This question is addressed in the present contribution, where we demonstrate the feasibility of this concept. We showcase this by a fully laser based process chain to remove support structures on a laser powder bed fusion (L-PBF) part and achieving a glossy polished surface after finishing. The approach is a sequential combination of closed loop controlled ultrashort pulse ablation, laser cleaning and polishing. The initial process precisely removes the rough saw cut support structures. An ultrashort pulse laser was chosen for the precision of a quasi-melt free process. This advantage has previously been demonstrated by enhancing geometrical accuracy of additively manufactured parts [28] and support structure removal on a small scale [22]. A depth sensor based on optical coherence tomography (OCT) was used for closed loop control of the ablation progress as described in [28]. However, to increase productivity the MRR was experimentally maximized in AlSi10Mg material by using the available (80 ± 1) W average laser power.

Following the laser ablation, a laser cleaning process was employed to eliminate debris deposits from the post-ablation surface. For this purpose, a nanosecond laser source operated in a specific regime that effectively removes debris without affecting the base material, as verified by light microscopy (LM) and scanning electron microscopy (SEM). Subsequent to laser cleaning, the base material underwent a laser polishing process utilizing a continuous wave disc laser, which was operated in a pulsed mode with rectangular pulse shapes. Three iterations between cleaning and polishing were performed for results fulfilling the requirements. The laser and scanning parameters for the polishing process were adjusted to create a melt pool depth larger than the surface structure depth, which enabled a remelting of the post-ablation surface with Ra = (24 ± 4) µm down to the final surface of Ra = (0.32 ± 0.04) µm.

Materials and Methods

Process Chain

The main idea to achieve high MRR and sub-µm roughness is to approach each goal with a separate laser process. The proposed process chain is depicted in Fig. 1. First, the support structures are removed using closed loop laser ablation in a high MRR process regime. Second, laser cleaning and polishing is performed iteratively to enhance surface quality. The goal was to achieve a sub-µm Ra roughness in conjunction with a surface free from debris and superficial porosity. The roughness was measured using laser scanning microscopy. The existence of debris and porosity was visually determined using light microscopy (see Sect. “Surface characterization”.).

Fig. 1
figure 1

Flow chart of the process chain. After the closed loop ablation with high MRR, a subsequent repetition of laser cleaning and laser polishing is performed. N presents the number of iterations required for sub-µm roughness and a surface free from debris and superficial porosity

Full size image

The required number of iterations N until the surface meets the set characteristics was determined during process development. In an application, the developed process chain can be repeated N times according to the findings without intermediate checks. Therefore, the time required for the manual quality assessment during development is not incorporated into the final discussion on total processing time.

Ablation with Closed Control Loop

The first step in the process chain is the removal of the support structures using closed loop laser ablation, see Fig. 1. The closed control loop is required for precise results due to the variances in the geometry of additively manufactured parts, the rough saw cutting process to separate the part from the base plate, and the unknown inner structure of the material.

For the closed loop control, measuring was performed with an OCT sensor CHRocodile 2 from Precitec Optronik, which was coaxially superimposed to the processing beam by means of a dichroitic mirror before the scanner aperture. The OCT is an interferometric distance sensor, which measures the length of the optical beam path through the scanner to the surface of the sample. Deflection of the scanner mirrors therefore yields an increase in measured distance even for a flat sample, which leads to a distorted topography measurement. This effect is extensively described in [29]. To calibrate and compensate the effect, a flat sample was measured as a reference geometry according to the proposed procedure [29].

The OCT sensor was set to a measuring frequency of 70 kHz. The waist diameter of the OCT beam was measured to (69 ± 3) µm on the surface of the sample. Independently of the beam diameter, the sensor provides an axial measuring accuracy of up to 1 µm. A self-developed software application was used for the closed loop process control. It is designed to trigger inline measurements of the process using the OCT sensor, process and analyze the data and automatically adjust the beam path during processing according to the difference between set topography and the actual topography. This control system was previously published in [22, 28, 30, 31].

High MRR Process Regime

The process chain is based on the idea, that the MRR of the ablation process can be maximized without any quality constraints. The subsequent cleaning and polishing are used to realize the desired surface characteristics. Therefore, the maximum material removal rate achievable with the given setup was experimentally determined on the L-PBF material. Using a conventional open loop process, 3 × 4 mm2 sized pockets were produced on the sample. This L-PBF specimen was ground flat beforehand to enable a precise measurement of the ablation depth. The ablation volume was measured using a Laser Scanning Microscope (LSM), see Sect. “Surface characterization”. The material removal rate is defined by MRR = Vmat / temission, where Vmat is the removed material volume. The time of emission temission is the accumulated time intervals of laser emission during the scans. Any unproductive times of jumps between scan lines are not considered. Therefore, this MRR value is independent of the specific ablation geometry or the properties of the scanner system.

A Carbide laser from Light Conversion was used with the beam characteristics shown in Table 1. At a maximum average beam power PL,usp of (80 ± 1) W, this system is able to output pulse energies between 1.6 mJ (at 50 kHz) and 40 µJ (at 2 MHz).

Table 1 Beam characteristics of the ultrashort pulse laser for laser volume ablation
Full size table

The optical setup consisted of a 2D galvanometer scanner with an f-theta lens of a focal distance of 340 mm. To determine the maximum MRR, the peak fluence was varied between 1.2 J/cm2 and 48 J/cm2. The fluence was set by changing the pulse energy and pulse repetition rate while using the constant PL,usp of (80 ± 1) W. The scan speed was constant at the maximum of 3 m/s. Scanning was performed at a hatching distance of 20 µm and an incrementally rotating scan orientation over 140 scans. The angle increment between each layer was set to 42° according to [32].

Figure 2a shows the measured MRR over the peak fluence. The maximum MRRmax = (21 ± 2) mm3/min is achieved at a peak fluence of 3.2 J/cm2 and a pulse repetition rate of 750 kHz. These parameters were used for the support structure removal. The SEM image in Fig. 2b shows the resulting surface of the machined pocket created at MRRmax. The presence of bumps is a typical result of heat accumulation [3], leading to a rugged surface. The structures with a lateral dimension of 70–100 µm depicted in Fig. 2b exhibit a roughness Ra = (30 ± 3) µm and Rz = (147 ± 11) µm.

Fig. 2
figure 2

a Measured MRR of the ablation process of 3 × 4 mm2 pockets on AlSi10Mg material using the maximum laser power of (80 ± 1) W, scan speed of 3 m/s, and variable peak fluences between 1.2 J/cm2 and 48 J/cm2. A total of 140 scans per pocket with a hatching distance of 20 µm and incrementally rotating angles per scan were used. The error bars are derived from an estimation of the measurement errors. The MRRmax = (21 ± 2) mm3/min is achieved at a peak fluence of 3.2 J/cm2 and a pulse repetition rate of 750 kHz. b The SEM image of the ablation basis at MRRmax depicts bump formation which is unwanted in most applications. The roughness values are measured using a Laser Scanning Microscope

Full size image

Laser Cleaning

In order to remove debris from the ablated surfaces, laser cleaning was applied as a subsequent step, see Fig. 1. The laser cleaning was done with a short pulse fiber laser TruMark 5020 from TRUMPF, which was mounted in a laser cell TRUMPF TruMarkStation 5000. The laser parameters used are summarized in Table 2.

Table 2 Beam characteristics of the laser for cleaning
Full size table

The laser process was performed with a 2D scanner, using an f-theta lens with a focal length of 254 mm. To reduce the fluence, the beam was defocused on the sample. The cleaning was performed at a scanning speed of 3 m/s and track offset of 70 μm over three scans resulting in a total area rate of 42 cm2/min. These settings have proven to be effective for debris removal on additive manufactured AlSi10Mg [33].

Laser Polishing

For laser polishing, a TruDisk 4002 Yb-disk laser was used from TRUMPF with the process parameters given in Table 3.

Table 3 Beam characteristics of the laser for polishing
Full size table

The used optics I-PFO by TRUMPF has a focal distance of 600 mm. The treated area was scanned in linear hatchings with a line spacing of 50 µm at a scanning speed of 200 mm/s, resulting in a pulse overlap of 81.0%, a track overlap of 95.2% and an area rate of 6 cm2/min. These parameters were found to be optimal for vertical printed L-PBF AlSi10Mg surfaces [34].

In order to protect the samples from negative influences of atmospheric oxygen during the polishing, the process was performed in a sealed chamber under a purified Argon gas atmosphere. The residual oxygen content was controlled with the oxygen measurement device of type Pro 2 plus from Orbitalservice.

Surface Characterization

Images of the intermediate states of the sample were captured with a LM Axio Zoom V16 by Zeiss and a SEM JSM-6490LV by JEOL. For ablation volume measurements, topographical analysis, and roughness measurements the LSM VK-9710 K and VK-X3000 by Keyence were used.

Material Sample

The used samples were created by L-PBF on a TruPrint 1000 Multilaser from TRUMPF, which is equipped with two infrared fibre lasers. The experimental investigations were executed on the aluminum alloy AlSi10Mg. The used aluminum powder from Heraeus has a particle size distribution of D10 of 21.3 µm to D90 of 56.9 µm with an average powder grain diameter of 35.7 µm. The samples were built with a slicing thickness of 20 µm. The layers were generated with a laser power of 175 W and scanning speed of 1400 mm/s. The support structure was produced with laser power of 150 W and a scanning speed of 2000 mm/s. During the fabrication process the chamber was flooded with Argon until a residual oxygen content COxy < 0.3 vol.-% was reached.

For the experimental investigations two rectangular plates with 3 mm thickness were built up for the parameter study and the demonstration of the process chain. The parts were cut-off from the built platform by a water cooled saw. Figure 3 shows the residual support structure on the sample. The support material is 2 mm high and the ridges 0.2 mm wide. The spacing of the ridges is 0.5 mm as shown in the cross section. In some places top left and right the sawing left a closed lid between the ridges. To determine the ablated volume of the support structures, the sample’s mass was measured before and after the ablation process with a precision scale I 2000 D by Sartorius and the ablated volume was calculated using the density of 2.67 g/cm3 provided by manufacturer Heraeus.

Fig. 3
figure 3

a Picture of the backside of the L-PBF specimen, made from AlSi10Mg, with saw cut support structures. Field for ablation marked with dashed contour. b The profile measured across support structure remnants. This is the initial state of the sample surface on which the ablation, cleaning and polishing process is demonstrated

Full size image

The demonstration of the process chain was conducted on a rectangular test area (Fig. 2 dashed line, 40 × 30 mm2) within the residual support structure of the specimen plate (50 × 40 mm2, Fig. 3). The ablation geometry is designed as a pocket with a depth of 2.2 mm in order to ensure removal of all support structures and the porous initial layer of the print. The side walls of the pocket are designed with draft at an inside corner angle α of 114° to prevent edge deepening.

Results and Discussion

Closed Loop Ablation

The state of the sample after ablation is shown in Fig. 4. The ablation process with closed loop control took in total 184 min to remove 1592 mm3 of material. Thereby, it successfully removed all the support structures within the specified field (a). As to be expected from the chosen processing parameters, the surface appears dark indicating residual debris and severe roughness (b). However, all the support structures are removed leaving a flat surface as evidenced by the cross sectional height profile (c). The resulting surface exhibits a roughness of Ra = (24 ± 4) µm and Rz = (148 ± 21) µm.

Fig. 4
figure 4

Resulting surface after closed loop ablation. a Light microscope image of the 40 × 30 mm ablation field. b Laser scanning microscope image of a subsection topography. c Cross section of the surface after ablation including the slopes with a corner angle α of 114° and measured roughness values

Full size image

Examining the topography section measured with LSM (b), it reveals that the microstructures still indicate the former presence of the ridges. Coarser bumps have formed in those areas where more material was removed. The average roughness Ra = (24 ± 4) µm is in the range of the roughness Ra = (30 ± 4) µm measured in the parameter study for the used processing parameters (Fig. 2). The total processing time of 184 min includes the accumulated time of emission corresponding to the MRRmax of (21 ± 2) mm3/min and all unproductive down times, which divide into:

  • intermediate measuring with the OCT

  • unproductive moves of the scanner to jump between scan vectors

  • additional intermediate jump vectors due to the spacing of the support structure.

The OCT enables for concurrent surface measurement during the ablation process, thus avoiding the need for intermediate measurements with the OCT. However, this approach was not implemented in this case, as it would not have detected the potential redeposition of material in areas that had already reached the target depth. Therefore, iterative intermediate OCT measurements of the entire sample surface were used in this study instead.

Cleaning and Polishing

Figure 5 shows a LM and SEM image of a subsection of the sample before and after cleaning. The cleaned surface appears brighter in the LM image (a) which indicates that less particles like oxide or smudge are present on the surface. The surface structure in the cleaned area is not changed compared to the uncleaned surface as shown by the SEM image (b), hence the process did not remove any bulk material. The positive impact on appearance becomes clear after polishing.

Fig. 5
figure 5

a Light microscope image of a cleaned square within the ablated surface. The cleaned area appears brighter, indicating the removal of the oxide layer and smudges on the surface. b Scanning electron microscope image of clean and unclean surface

Full size image

Figure 6 shows LM images of the polished area after each of the three conducted cleaning and polishing iterations according to the flow chart in Fig. 1. The image of the first iteration shows dark discoloration due to trapped impurities in the surface structure from the ablation process, that were deposited on the surface during remelting. It can be assumed that the initial cleaning step insufficiently eliminated the debris due to the profound surface structures remaining after laser ablation, see Fig. 2. Additionally, some porosity remained on one side of the area. Hence, a second and third iteration of cleaning and polishing was performed. With each step, the magnitude of discoloration and superficial porosity progressively diminished. After the third iteration, a shiny surface containing few isolated pores is created which exhibits a roughness of Ra = (0.32 ± 0.04) µm and Rz = (2.2 ± 0.3) µm.

Fig. 6
figure 6

Evolution of surface after cleaning and polishing. Three iterations are sufficient to remove most of the superficial pores and all of the debris. The roughness measurement reveals significantly lower Ra and Rz values compared to the condition after ablation

Full size image

The achieved final surface is shown in Fig. 7. The LM image (a) gives an overview over the specimen including the support structure brim. The main treated area shows a pure aluminum surface free from any visual contamination and a glossy finish. In (b), a SEM image of the overlapping area of all three processes is shown, which highlights the major improvement in surface quality realized with this laser based strategy.

Fig. 7
figure 7

L-PBF sample after removal of support structures. a Light microscope image of the ablation and polishing field. b Detail view of the transition from support structures to ablated surface and polished final result captured with the scanning electron microscope. The initial state of the support web, the ablated area and the cleaned and polished area are visually separated by dashed lines

Full size image

Regarding the processing time, ablation was completed in 184 min, cleaning in a total of 0.7 min and polishing in a total of 4.5 min. Hence, cleaning and polishing only contribute an additional 2.8% of processing time. This represents a minor increment when considering the advantage of utilizing the full (80 ± 1) W of beam power for high MRR ablation in the first place.

In general, the achieved time savings during ablation scales with the cube of the characteristic dimension L, neglecting the influence of the geometry on the processing time. In contrast, the additional time for finishing only scales with L2, rendering this approach particularly advantageous for larger parts. Consequently, a critical threshold exists for very small ablation volumes, below which the time saved by high MRR ablation does not compensate the additional finishing time. In this case, opting for the direct removal of support structures in a high-quality regime becomes more favorable.

Conclusions

This study addresses the question, whether it is possible to achieve a surface quality with sub-µm roughness using a high productivity laser-based processing chain on an arbitrary geometry of high roughness and shape deviations. An application is support structure removal on parts made from laser powder bed fusion. The investigated procedure consists of closed loop laser volume ablation and subsequent laser cleaning and polishing steps. We improve productivity during ablation at the expense of surface quality and restore the finish in subsequent cleaning and polishing steps.

The results show, that an ultrashort pulse laser commonly used for high precision post-processing can also be operated in a high productivity regime for the task at hand. We utilized the full available beam power of (80 ± 1) W to maximize the material removal rate up to (21 ± 2) mm3/min. The cleaning process then iteratively removed debris from the post-ablation surface. Suitable laser and scanning parameters for polishing create a molten layer of sufficient thickness to reduce the roughness from the ablation stage by two orders of magnitude from Ra = (24 ± 4) µm down to Ra = (0.32 ± 0.04) µm. The result is a smooth, glossy surface.

The approach of separately optimizing productivity and finish is especially useful for upscaling, as the time savings from optimized ablation are proportional to the volume, whereas the additional time for finishing only scales with surface area.

In this work, the three steps were carried out on separate laser machines, because the processes require different pulse durations ranging from ultrashort to continuous wave in pulsed mode. Advances in the development of lasers with highly flexible beam characteristics may result in laser systems, that offer the required capabilities in a single device in the near future [35]. Consequently, such development would enable the execution of this process chain on a unified laser machining system.