High-Power Laser Material Processing via Numerical Simulation of Deformable Mirror-Controlled Beam Shaping

High-power laser fabricating (HLMP) has become a mainstay of high-quality production, allowing incision, welding, surface hardening, and additive production to be performed with a previously unseen level of control. Such processes are important for industries such as aerospace, automotive, and medical tool manufacturing, where precision and repeatability at the micron level [1] are often a requirement. As a foundation of these processes, it must be understood that ability to control the spatial and temporal distribution of the laser power is critical. This is due to the fact that the laser intensity profile ultimately dictates the heat input, molten pool dynamics, and final microstructure of the substrate. For instance, in laser cutting of chrome steel, a Gaussian beam profile may additionally result in choppy strength deposition, causing thermal distortion or microcracks, while a tailored "top-hat" profile can enhance part quality and reduce the heat-affected zone (HAZ) [2]. Despite its essential position, traditional beam-shaping strategies, including static diffractive optical elements (DOEs), constant aspheric lenses, or segment plates, are afflicted by intrinsic limitations. These static systems cannot adapt to dynamic processing conditions, along with fluctuating thermal hundreds or varying cloth reflectivity, which are ubiquitous in high-electricity applications (>1 kW) [3]. For example, thermal lensing, a phenomenon wherein lens heating alters the beam's focal length, remains a continual undertaking in non-stop-wave laser welding, frequently requiring costly hardware recalibration mid-process [4].

To overcome those obstacles, adaptive optics, specifically deformable mirrors (DMs), have garnered extensive attention. Deformable mirrors (DMs) use piezoelectric or electrostatic actuators to change their surface shape. This capability allows for adjustments to correct wavefront errors and manage beam features like intensity, focal spot size, and divergence [5]. This is helpful for high-power laser materials processing (HLMP), where quick adjustments to beam form are needed for changing between tasks like cutting and surface texturing, or for fixing substrate warping during production. Work in lower power settings (<500 W) show what DMs can do [6]. For example, reshaping the beam from Gaussian to Bessel reduced the heat-affected zone (HAZ) by 30% when laser scribing silicon. Also, study [7] used a DM to improve power delivery in laser hardening of titanium alloys, which resulted in a 15% increase in surface hardness consistency.

However, using these methods for strong structures is difficult. High densities (>1 MW/cm²) create large thermal stresses on DM substrates, leading to actuator drift, surface change, and reduced reflection [8]. The short-term thermal-mechanical link between the mirror and the laser beam, made worse by lasting contact with multi-kilowatt lasers, has not been well studied, which leaves a major gap in designing stable DM systems for industrial use. This problem is amplified by a lack of standard ways to measure DM ability under intense conditions. While computer models, like finite element analysis (FEA) of actuator response times or heat load distribution, have been suggested [9], experiments in real HLMP situations are rare. For example, current models don't consider actuator hysteresis and thermal creep under multi-kilowatt loads. These factors are important for predicting how well DMs will perform over time in industrial uses. For example, no study has measured how actuator hysteresis or thermal creep affect beam stability during fast laser cutting of metals that can withstand high temperatures. Current control methods for DMs are taken from astronomy or biomedical imaging and are not well suited for the fast feedback loops (under a millisecond) required in HLMP [10]. The low-power studies show what DMs can potentially accomplish, but their results cannot be directly applied to high-power systems because of thermal and mechanical problems. This suggests that we need a careful study of DMs in high-power settings. Solving these problems is necessary to get the most out of DMs in industrial laser systems.

This paper explores objectives to address these gaps through 3 interrelated goals. First, we develop a high-power-compatible DM system, which combines a 127-actuator piezoelectric mirror with a closed-loop control system, relying on real-time beam diagnostics from a high-speed Shack-Hartmann wave front sensor. Second, we systematically characterize the dynamic beam properties, including spatial intensity uniformity, focal shift tolerance, and thermal load behavior at operational power levels of up to three kW, using a combination of beam profiling cameras and pyroelectric sensors. Third, we investigate the responsiveness of our system by using it to laser cut 316L stainless steel, which is used routinely in surgical implants for its corrosion resistant properties. Metrics, including cut area roughness (Ra), HAZ width, and microhardness gradients, are compared with baseline motor effects from static beam shaping. By linking DM actuator dynamics with material effects, this research improves the fundamental understanding of adaptive optics in HLMP as well as outlines details for the contractor to use in developing tailored solutions for industry.

3.1 Deformable mirror system design

The deformable replicate (DM) system contains a 128-actuator non-stop face-sheet mirror (Boston Micromachines Corp.), imparting a stroke range of ±5 µm and a response time of <10 ms, permitting dynamic beam modulation. A closed-loop control gadget, including a Shack-Hartmann wavefront sensor, employs proportional-integrating-derivative (PID) feedback algorithms to adjust the mirror's surface deformation in real time. The management regulation is defined as:

$\mathrm{u}(\mathrm{t})=\mathrm{K}_{\mathrm{p}} \mathrm{e}(\mathrm{t})+\mathrm{K}_{\mathrm{i}} \int_0^{\mathrm{t}} \mathrm{e}(\tau) \mathrm{d} \tau+\mathrm{K}_{\mathrm{d}} \frac{\mathrm{de}(\mathrm{t})}{\mathrm{dt}}$

where, u(t) is the actuator displacement, e(t) is the wavefront error, and $\mathrm{K}_{\mathrm{p}}$, $\mathrm{K}_{\mathrm{i}}$, and $\mathrm{K}_{\mathrm{d}}$ are PID gains optimized via iterative tuning [22]. This configuration guarantees particular beam shaping, crucial for minimizing thermal distortions in high-power packages.

3.2 Experimental setup

The processing system (Figure 2) includes a 3 kW IPG YLS-3000 fiber laser (wavelength: 1070 nm) coupled with collimating lenses (f=150 mm) and a beam splitter to direct 5% of the beam to a thermal camera (FLIR X6900sc) for in-situ beam profiling. A CNC stage (Aerotech ALS130) presents µm-level positional accuracy for pattern translation. The DM is located downstream of the collimator to modulate the beam before focusing (f=200 mm) onto the workpiece. Table 2 summarizes critical parameters for reproducibility. The beam profiler captures intensity distributions, whilst the CNC level ensures constant processing paths.

Table 2. Key parameters of the experimental setup

2.png

Figure 2. Effect of beam shaping on surface roughness

3.3 Experimental protocol

Samples (stainless steel 316L) were processed using a Gaussian beam (M² = 1.2) at fixed parameters: 1.5 kW power, 100 mm/s scan speed, and 0.5 N load. Post-processing metrics (e.g., hardness, penetration depth) were measured using Vickers indentation and optical profilometry (Table 3). Table 3 highlights increased hardness (33% avg.) and HAZ expansion due to Gaussian beam-induced thermal gradients [23-26].

Table 3. Baseline results for Gaussian beam processing

3.4 Dynamic beam shaping

Custom beam profiles (Top-Hat, Bessel) were generated by means of iteratively adjusting the DM's floor using Zernike polynomial-based total section maps. Real-time beam modulation was synchronized with the CNC stage to observe transient results. For example, a Top-Hat profile decreased surface roughness ($\mathrm{R}_{\mathrm{a}}$) with the aid of 25% in comparison to Gaussian beams, attributed to uniform electricity distribution (Figure 3).

3.png

Figure 3. HAZ width vs. laser power & scan speed

3.5 Parameter optimization

A design of experiments (DoE) approach examined interactions between laser power (0.5–3 kW), test velocity (50–300 mm/s), and beam diameter (50-200 µm). Response surface methodology (RSM) identified optimum combinations (2 kW, 200 mm/s, 100 µm beam) for minimizing the HAZ width (≤40 μm) while maximizing hardness (180 HV). The relationship is modeled as:

HAZ Width $=0.8 \mathrm{P}^{0.5} \cdot \mathrm{v}^{-0.3} \cdot \mathrm{~d}^{0.4}$

where P, v, and d denote power, speed, and beam diameter, respectively [23].

3.6 Data acquisition and analysis

Post-manner metrics were analyzed using ANOVA to evaluate statistical significance (p<0.05). Thermal stability was evaluated through cyclic heating (100–500℃), revealing <5% hardness degradation for dynamically formed beams, in comparison to 12% for Gaussian beams.

The integration of dynamic beam shaping via a deformable mirror (DM) has demonstrated considerable enhancements in laser fabric processing, as evidenced by the experimental consequences. The superior uniformity of strength distribution in shaped beams (e.g., Top-Hat) minimizes localized thermal gradients, a vital aspect in lowering micro-crack density (Table 4) and enhancing surface smoothness (Ra = 0.65 µm). This is consistent with the conclusions of the studies [27-29] which differed crack suppression to a reduced thermal pressure in uniformly irradiated zones. However, the 18% increase in hardening depth which lies in Table 5 exceeds the 12% increase reported by the research [30] for equivalent beams, presumably because of the enhanced PID regulation algorithm [11] which provides accurate beam regulation over excess feed rate of the scanning manipulation.

An important distinction from previous work is the scalability of the DM Australia DM gadget. While previous research [31] demonstrated thermal stability at ≤2 kW, the contemporary setup maintained ≤0.2 µm surface glide at 3 kW, highlighting advancements in actuator materials and cooling designs. Nevertheless, industrial adoption faces demanding situations. The high price of DM systems (about 2 to three× traditional optics) and the complexity of real-time closed-loop manipulation necessitate specialized information, restricting accessibility for small-scale producers. These boundaries echo issues raised by the study [11] (384-09, 1999), which emphasized the need for cost-discount techniques, such as modular actuator arrays or machine learning-based control simplifications [26].

Figure 7 illustrates the trade-off between premature investment and long-term gains in precision, suggesting that high-extent manufacturing justifies the initial expense.

7a.png

(a)

7b.png

(b)

Figure 7. Cost-effectiveness comparison (a) Relative cost of Gaussian vs. DM-based systems (b) Processing quality (HAZ width, surface roughness) vs. system cost

Furthermore, the advanced overall performance of Top-Hat profiles over Bessel beams in decreasing HAZ width contrasts with findings presented in reference [27], which suggested comparable results for each profile in aluminum alloys. This discrepancy can also stem from differences in fabric thermal conductivity (180–190 W/mK for steel vs. 205 W/mK for aluminum), which affect warmth dissipation costs. Future studies need to discover beam-shaping efficacy across a broader variety of substances and geometries. Future research will explore combining this laser shaping technique with a subsequent annealing stage, a process proven to enhance the crystallinity and physical characteristics of functional films like In₂O₃ [32], to achieve superior material properties. In the end, even as dynamic beam shaping offers transformative capacity for high-power laser processing, its business viability hinges on addressing fee and complexity challenges. Collaborative efforts between academia and enterprise to standardize, manipulate interfaces, and optimize actuator designs should accelerate adoption, as tested by means of current pilot implementations in aerospace element production [33-35]. Future work could investigate whether the precision of deformable mirror-controlled beam shaping could be applied to polymer synthesis or surface functionalization, similar to the molecularly imprinted polymers used for selective adsorption in environmental applications [36].