Radial tool wear analysis and compensation strategy for microhole arrays in sequential hybrid laser beam micromachining (LBMM) and micro-EDM

The integrated LBMM and µEDM system, developed in-house [37], features the LBMM system positioned horizontally offset from the µEDM setup. The workpiece remains within the µEDM workspace, accessible for both processes. The LBMM system features a 50-W fiber laser (1060-nm wavelength, 50-μm spot size, 200-mm focal length) controlled by an X/Y galvanometer scanner with ± 15° scan angle and 12-µrad resolution. Laser power was calibrated using a Gentec Pronto-250 power meter, which measures average output power. The calibration was performed across a range of 10% to 100% laser settings. Although the laser is rated at 50 W, the actual output at 100% was approximately 38.6 W. This represents a power loss of about 22.8%, which aligns with the data provided by the supplier of this fiber laser system. This loss occurs because the laser beam passes through several optical elements such as collimators, beam expanders, mirrors, and the galvanometric scanner before exiting the system. Each optical component contributes to reflection, scattering, or absorption losses, resulting in reduced power delivered to the workpiece surface. A calibration curve was established to map the percentage settings to actual output power. For this experiment, the laser was set to 90%, corresponding to a calibrated power of 35.4 W. The laser head is height-adjustable via a rotary handle and moves horizontally on a motorized linear stage, with process parameters managed through EZCAD Lite software. The µEDM setup, a three-axis CNC machine with 15-μm repeatability, operates in the same workspace and includes a µEDM spindle and microscopic camera mounted on the Z-axis plate. Custom software handles camera viewing, parameter adjustment, G-code execution, and axis monitoring. Figure 1 illustrates the machine configuration: during LBMM (Fig. 1a), the laser head positions above the workpiece; after retracting, the µEDM system moves in for machining (Fig. 1b), with the transition taking 36 s.

Bild vergrößern

A microscopic camera, mounted on the µEDM machine, is used to establish a common coordinate system for both the camera and the µEDM electrode [37], as depicted in Fig. 1a,b. During the calibration process, a blind reference hole is machined using µEDM at known coordinate positions. The microscopic camera is then moved using the CNC stages to align its crosshair overlay with the center coordinates of the reference hole. The x and y deviations, representing the offset distances between the electrode and the camera crosshair, are determined using linear scale feedback. These offset distances are subsequently utilized by the µEDM machine to transition seamlessly between the camera view and precise electrode alignment. Additionally, a second microscopic camera is positioned at the rear of the µEDM machine, as illustrated in Fig. 1c for optical microscopic measurement (OMM), characterization, and measurement purposes. It is specifically used to measure the axial and radial wear length of the tool. The measurement accuracy, along with the associated uncertainty, is determined by the resolution and repeatability of the CNC positioning stages, specified as 1 and ± 15 µm, respectively. Figure 1c also demonstrates the hybrid LBMM–µEDM integrated setup for the experimental work.

To study the tool wear characteristics for µEDM in the hybrid process, six sets of experiments, with different workpiece materials (copper and stainless steel) and thicknesses (200 and 600 µm), were conducted. The machining process involved the creation of an array of six LBMM pilot microholes, evenly spaced 1 mm apart in a 3 × 2 arrangement, followed by fine finishing of through-holes using µEDM with a 300-µm diameter tungsten electrode (as available in the lab). The size of the LBMM pilot hole is critical. If the pilot hole is too small, it increases the µEDM machining time due to limited access and inefficient material removal. Conversely, if the pilot hole is too large, it results in excessive HAZs, recast layer, and residual spatter, which may affect the final hole quality. Therefore, a suitable LBMM pilot hole size was determined through initial trials to optimize both machining efficiency and hole quality. The selection of LBMM parameters used in this study is described in the following section. Detailed parameters for LBMM and µEDM are tabulated in Table 1. The LBMM process used 75 loops count with a scanning speed of 10 mm/s, a 0° hatch angle, 0.01-mm line spacing, and a 50-µm focal spot size. A two-way hatch pattern was applied to enhance coverage uniformity and minimize unprocessed areas.

The electrode was flattened at the bottom before each set of array hole machining to ensure a consistent electrode shape across all sets of microholes. The pregrinding was performed using the block electrical discharge grinding (EDG) method. The influence of flattening the bottom of the electrode on radial wear is not within the scope of this study. However, this step was carried out to eliminate variability in electrode shape and to minimize potential bias in wear measurements. The flatness of the electrode was visually verified using the microscopic camera, and care was taken to ensure that no significant axial or radial wear was present before machining was performed.

Surface detection was performed for each hole, setting the Z-position to zero and retracting the electrode 100 µm above the surface. To ensure complete cuts through the 200 and 600-µm thickness of workpieces, a programmed depth of 600 and 1000 µm was used, respectively. Using a microscopic camera, images of the microelectrode were captured after each machining hole, allowing for precise measurements of axial and radial wear length. Furthermore, the machined samples were cleaned with isopropyl alcohol (IPA), followed by cleaning in an ultrasonic bath [38]. A scanning electron microscope (JSM-IT100 from JEOL Ltd.) was used to characterize the machined hole. Third-party software named ImageJ [39] was used for characterization and subsequent measurement purposes.

The selection of workpiece and electrode materials in LBMM–µEDM is important for achieving high precision and efficiency. In µEDM, the electrode material affects the accuracy and shape of the machined microfeatures. Tungsten was chosen as the electrode material due to its high thermal wear resistance and low tool wear rate [40]. In this study, a cylindrical tungsten electrode with a 300-µm diameter was used to ensure consistent machining performance.

For the workpiece, copper and stainless steel (SS) were selected due to their suitability for both EDM and laser machining. Copper was chosen for its high electrical and thermal conductivity, which ensures efficient material removal during µEDM and good laser energy absorption in LBMM. Stainless steel was selected for its high strength, corrosion resistance, and dimensional stability, making it an excellent choice for precision microfabrication. Both materials are commonly used in biomedical devices, aerospace components, industrial filtration, and inkjet nozzles, where microhole arrays require high accuracy, durability, and performance.

This section focuses on the characterization of key performance metrics, including axial and radial wear length measurement methods and hole quality assessment.

The performance of the hybrid LBMM–µEDM process is strongly influenced by key machining parameters from LBMM, including laser power (P), scanning speed (SS), loop count (L), and programmed diameter (PD). For this study, predrilled LBMM holes were used as the initial step to facilitate the subsequent µEDM process. The selection of parameters was guided by their effects on LBMM entry and exit diameters, recast layer area, LBMM processing time, and the final hybrid machining time. Laser power dictates the energy delivered per pulse, scanning speed controls the exposure time on the workpiece, and loop count determines the total number of laser passes, affecting the cumulative material removal and taper formation. The experiments were carried out on stainless steel workpieces with a thickness of 600 µm. A full factorial experiment was performed using laser powers of 10%, 50%, and 90%, scanning speeds of 10, 100, and 1000 mm/s, and loop counts of 75, 150, and 225. In a separate study, the programmed diameter was varied with values of 80, 180, and 280 µm.

Figure 4a illustrates the effect of scanning speed and loop count on the LBMM processing time for laser powers of 10%, 50%, and 90%, with a programmed diameter of 180 µm. It can be observed that processing time decreases significantly with increasing scanning speed, as higher speeds reduce the duration of laser exposure per pass. Higher loop counts, on the other hand, lead to longer laser machining times. For example, at a low scanning speed of 10 mm/s, increasing the loop count from 75 to 225 results in processing times rising from 22 to 65 s. Therefore, a loop count of 75 was selected for this study. Moreover, although very high scanning speeds, such as 1000 mm/s, reduce processing time, they were not chosen because, as shown in Fig. 4b, they result in smaller entry diameters, which would increase the overall machining time in the subsequent µEDM finishing process and will be discussed later. Figure 4b depicts the effect of the power and scanning speed at a fixed loop count of 75 and LBMM programmed diameter of 180 µm. At high power (90%), the largest entry diameters are achieved at low scanning speeds (225 µm at 10 mm/s), with sufficiently large exit diameters to ensure proper debris flow. Ideally, the large entry and exit diameters are preferred, as these through-holes facilitate efficient flushing (during the µEDM finishing process), which is critical for successful hybrid machining. Furthermore, the LBMM programmed diameter has a significant impact on the final hybrid machining performance due to the formation of the recast layer. Importantly, the LBMM entry diameter, including the recast layer, must remain below 300 µm, which corresponds to the µEDM electrode size, to ensure that the recast layer can be fully removed during the subsequent µEDM finishing process. Figure 4c shows the variation of LBMM entry diameter (including and excluding recast layer) and hybrid machining time for different programmed diameters, at 90% laser power, 10 mm/s scanning speed, and a loop count of 75. At a programmed diameter of 80 µm, the hybrid machining time remains relatively long (30 min), similar to the machining time of pure µEDM. In contrast, a programmed diameter of 280 µm results in the fastest hybrid machining time (4 min) and the highest LBMM entry diameter, as the larger diameter improves debris flushing efficiency. However, the LBMM entry diameter, along with the recast layer, is 423 µm, which is larger than the 300-µm electrode used in this study. This leaves part of the recast layer unremoved from the edge of the final hole, resulting in poor hole quality. A typical morphological comparison of the final hole after the hybrid process is illustrated in Fig. 4c, showing the differences between programmed diameters of 180 and 280 µm. The 180-µm hole exhibits complete removal of the recast layer from the edge of the finished hole, while the 280-µm hole still retains some recast material. Therefore, a programmed diameter of 180 µm was selected, as it provides a reasonable machining time of 7 min while maintaining acceptable hole quality. To summarize, the selected LBMM parameters for efficient hybrid machining are 90% laser power, 10 mm/s scanning speed, 75 loops, and a 180-µm programmed diameter. These settings provide sufficient entry and exit diameters for effective flushing, complete removal of the recast layer formation, and balance machining efficiency with hole quality.

Fluence, also known as energy density, refers to the laser energy applied per unit area (J/cm²) during the LBMM process, and cumulative fluence represents the total energy delivered to the material over the entire number of laser pulses. In this study, the cumulative fluence was varied by adjusting the laser power, scanning speed, and loop count. In addition, the pulse number was determined from the laser processing time and laser frequency. The equation for the cumulative fluence is shown in Eq. (2). The effect of the cumulative fluence and laser pulse number on the LBMMed hole’s taper angle is shown in Fig. 4d,e, respectively. It should be noted that only LBMMed holes that pierced through were plotted in the figures. LBMMed holes that did not pierce through were excluded, as their taper angle could not be measured. Specifically, at 10% laser power, none of the holes pierced through. At 50% power, the hole at a scanning speed of 225 mm/s did not pierce through. At 90% power, all holes successfully pierced through. It can be inferred from Fig. 4d,e that the taper angle decreases with increasing cumulative fluence and pulse number. At both 50% and 90% laser power, a steep reduction in taper angle is observed at lower fluence and pulse numbers, followed by a gradual decrease that approaches a near-constant value of around 4–5° at higher inputs. Across the entire range, the 90% laser power condition tends to produce lower taper angles compared to 50% power, indicating improved hole quality at higher power settings. This suggests that higher cumulative fluence and larger pulse numbers reduce the diameter difference between the entry and exit holes, thereby minimizing tapering in LBMMed holes. Therefore, 90% laser power was selected in this study, as it resulted in the minimal taper angle along with the largest entry and exit diameters. In the subsequent section, we present the compensation strategy developed to mitigate the effect of the initial taper induced by the LBMM process through subsequent µEDM finishing, which constitutes the primary focus of this study.

where P_App is the applied laser power (W), t is the processing time (s), and A is the spot area (cm²).

Figure 5a,b illustrates the evolution of electrode geometry during continuous machining on stainless steel material with a thickness of 600 µm for the pure µEDM and hybrid LBMM–µEDM processes, respectively. It can be inferred from Fig. 5a that the electrode subjected to successive pure µEDM experiences more pronounced axial wear while largely maintaining its diameter. In contrast, the electrode used in the hybrid process, as shown in Fig. 5b, develops a tapered, conical shape due to increased radial tool wear length.

Bild vergrößern

This manuscript attempts to explain the different tool wear characteristics observed experimentally for pure µEDM and the LBMM–µEDM process. Figure 6a–d illustrates the spark discharge regions on electrodes during pure µEDM and hybrid laser-based µEDM processes. In Fig. 6a, in pure µEDM, most of the spark discharge occurs at the bottom surface of the electrode as it is positioned above the workpiece. This uniform discharge ensures consistent wear at the bottom. In Fig. 6b, as the electrode machines within the workpiece, the spark discharge is distributed across both the bottom surface and the sidewalls of the electrode. This balanced discharge helps maintain the electrode’s shape, minimizes taper, and ensures precise machining.

Bild vergrößern

For the hybrid laser-based µEDM process in Fig. 6c, when the electrode is positioned above the workpiece before penetration, the spark discharge is localized to a small region at the bottom corner of the electrode, with minimal activity at the center of the bottom surface. In Fig. 6 d, during machining within the workpiece, the spark discharge is concentrated along the sidewalls of the electrode, promoting radial wear length. This limited discharge activity in the center reduces axial wear, resulting in distinct machining characteristics compared to the pure µEDM process.

This study investigates the variation in tool wear characteristics when machining stainless steel and copper workpieces with thicknesses of 600 and 200 µm, using a 300-µm electrode in the hybrid LBMM–µEDM process. The evolution of electrode geometry during continuous machining is depicted in Figs. 5 and 7. Based on visual inspection, it can be inferred from the figures that radial wear length is more significant on stainless steel compared to copper. Furthermore, insignificant or no axial wear was observed on the 200-µm thick workpiece (for both copper and stainless steel), in contrast to the 600-µm thick workpiece.

Bild vergrößern

For in-depth analysis, the axial and radial wear lengths during the hybrid LBMM–µEDM, as well as the microhole’s taper angle measurement, are summarized in Fig. 8a–c. Apart from the absence of axial wear on the 200-µm thick workpiece, stainless steel recorded a higher average axial tool wear compared to the copper workpiece in the 600-µm thick workpiece, as shown in Fig. 8a. The reason for this is that the µEDM processing time for copper is much shorter compared to stainless steel, as shown in Fig. 8d. The presence of axial wear in the 600-µm thick workpieces (both stainless steel and copper) but its absence in the 200-µm thick workpieces in Fig. 8a can be attributed to the differences in machining conditions and spark distribution, as explained in Fig. 6. Thicker workpieces, such as the 600-µm ones, require deeper machining, leading to prolonged spark exposure probability on the bottom surface of the electrode, which promotes axial wear. In contrast, the shallower penetration required for the 200-µm thick workpieces reduces the chance of spark exposure on the electrode’s bottom surface, effectively almost eliminating axial wear. However, the radial wear length increases continuously with the increase in the hole sequence number, as shown in Fig. 8b for both materials and thicknesses. Stainless steel, for both thicknesses, shows a higher radial wear length compared to copper. In µEDM machining, copper is widely recognized as an easier material to machine compared to stainless steel [41]. The higher radial wear length observed in stainless steel compared to copper is due to the higher machining time, as depicted in Fig. 8d. The longer interaction time of stainless steel with the electrode contributes to increased tool wear. From a material properties perspective, stainless steel has lower thermal conductivity and a higher melting point than copper, which causes localized heat accumulation and increases the likelihood of wider spark dispersion around the electrode edges, resulting in greater radial wear. Additionally, copper’s lower electrical resistivity enables more efficient and frequent spark discharges, resulting in a higher discharge rate and faster material removal. In contrast, the higher electrical resistivity of stainless steel leads to a lower discharge rate, requiring prolonged spark activity to achieve the same depth, thereby increasing tool wear, especially along the radial direction [42].

Bild vergrößern Bild vergrößern

The graph of machining time and discharge pulse number, as shown in Fig. 8d,e, highlights that the 600-µm thick workpieces have significantly longer machining times and higher discharge pulse numbers compared to the 200-µm thick workpieces. This indicates that the deeper machining required for the 600-µm thickness involves more prolonged spark activity, which contributes to the observed tool wear, axially and radially. However, the increase in machining time is not proportional to the increase in thickness. In hybrid LBMM–µEDM processes, this nonlinear scaling is influenced by several factors. During µEDM, deeper cavities reduce flushing efficiency, causing debris accumulation and unstable discharges, which slow down the MRR [43]. As the machining depth increases, the risk of short circuits and arcing also rises, resulting in more frequent pauses in the process. Moreover, the workpiece thickness has a significant effect on the resulting cut quality during laser machining [44]. This is due to increased laser energy loss as the beam penetrates deeper into the material, where scattering, surface reflection, and thermal diffusion become more pronounced. These combined effects make the machining time grow at a faster rate than the workpiece thickness.

This radial tool wear length directly impacts the tapering of the microhole array, as clearly shown in Fig. 8c. For both thicknesses of stainless steel, the taper angle increased linearly after successive machining, while copper maintained a consistently low taper angle. Axial wear can be easily compensated for by performing surface detection before machining each hole. However, radial wear length must be compensated for by adjusting the programmed machining depth. In these experiments, the programmed machining depth was set to a constant value for each successive operation: 1000 µm for the 600-µm thick material and 600 µm for the 200-µm thick material, leaving an additional 400 µm to compensate for radial wear. This compensation was sufficient for the copper workpiece, where the maximum radial wear recorded was 335 µm, as shown in Fig. 8b, resulting in low taperness in the microholes, as seen in Fig. 8c. However, this was insufficient for the stainless steel material, as the radial wear length exceeded 400 µm. To address this issue, a dynamic compensation strategy has been proposed, which will be discussed in the next section.

This section proposes an offline compensation method to mitigate the tapering effect in the machining of successive microhole arrays in hybrid LBMM–µEDM. Axial wear is compensated by performing surface detection before machining each hole. However, radial wear length must be compensated by adjusting the programmed machining depth. It is important to clarify that the aim of this study was not to reduce tool wear itself but rather to study its characteristics and propose a compensation strategy to achieve high-quality holes with reduced taper angle and efficient machining time. By increasing the programmed machining depth using the proposed dynamic compensation formula, the negative effects of both RTWL and ATWL on the final hole geometry can be significantly minimized. This approach enables more uniform hole profiles and enhances dimensional consistency, despite the inevitable occurrence of tool wear during the process. Moreover, more detailed studies on the effect of laser parameters in sequential LBMM and µEDM processes have been conducted in previous work [45].

Using the experimental data, the radial wear length can be predicted and incorporated into the programmed machining depth calculation. Figure 9a–c illustrates the electrode position during µEDM in the hybrid process with a tapered electrode at different stages: initially, after surface detection, before exiting the workpiece, and after exiting the workpiece.

Bild vergrößern

Initially, the tapered electrode performs surface detection to eliminate the axial wear from the previous hole machining, as shown in Fig. 9a. If the programmed machining depth d is set equal to the workpiece thickness t, the hole will become tapered due to the taper of the electrode, as illustrated in Fig. 9b. To prevent this, the programmed machining depth must be adjusted to account for both the workpiece thickness and radial wear length, with a safety factor f. This factor was introduced to ensure that machining is consistently carried out within the nontapered portion of the electrode, considering several sources of uncertainty in the process. The selected value compensates for uncertainties in radial tool wear length measurement, nonuniform workpiece thickness, surface detection accuracy, and CNC stage positioning, which has a specified tolerance of ± 15 µm.

Although the value of f = 1.2 was not derived through formal optimization, it was conservatively selected to provide a sufficient safety margin for consistent and reliable through-hole machining. Using a lower value could result in insufficient depth, incomplete breakthrough, or an increased failure rate due to the accumulation of positional and measurement errors. On the other hand, selecting a higher safety factor would unnecessarily increase machining time and may lead to excessive overcutting. Therefore, f = 1.2 was adopted as a practical compromise between machining reliability and process efficiency. To ensure consistent through-hole machining and compensate for tool wear and process uncertainties, a depth adjustment strategy is introduced. The programmed machining depth is formulated as in Eq. (3):where \({d}_{n}\) is the programmed machining depth in micrometer with respect to the hole sequence number n, f is the safety factor, t is the workpiece thickness, and \({RTWL}_{n}\) is the radial tool wear length. The radial tool wear length \({RTWL}_{n}\) was linearly approximated based on the experimental data, providing a predictive model for adjusting the programmed machining depth, as illustrated in Fig. 9d. Table 2 summarizes the equation for programmed machining depth for stainless steel and copper, considering workpiece thicknesses of 600 and 200 µm, using a 300-µm electrode.

The linearized radial tool wear length (RTWL) equation varies based on both the workpiece thickness and the material type, as summarized in Table 2. However, for a given material, the coefficients and constants of the RTWL equations can be estimated through linear interpolation between two known thickness values. For instance, Table 2 presents RTWL equations for stainless steel (SS) at thicknesses of 200 and 600 μm. To estimate the RTWL equation for SS at any intermediate thickness within this range, linear interpolation of the coefficient and constant can be employed. The accuracy of the interpolated RTWL equation can be improved by reducing the gap between the experimentally measured thickness values. Another constraint is the available length of the electrode mounted on the collet, which limits the maximum achievable machining depth. Additionally, the model assumes a limited number of holes per electrode, since progressive wear will eventually require repositioning or replacement of the electrode. It is important to highlight that the current RTWL model is specific to the tested material and electrode diameter and therefore may not be directly applicable to other material–electrode combinations. To improve generalizability, a more comprehensive, data-driven approach is recommended. This would involve building a large data set encompassing various workpiece materials, electrode diameters, and machining parameters. Such a dataset could support the development of a universal model using predictive algorithms or machine learning techniques, enabling accurate RTWL estimation across a wider range of conditions.

To verify the compensation strategy, the same experiment was conducted using stainless steel with a thickness of 600 µm and a 300-µm electrode diameter but with the compensation method applied. The safety factor f was set to 1.2. Table 3 compares the programmed machining depth with and without compensation, emphasizing the dynamic adjustment of the programmed depth. From the table, it can be observed that compensation was only possible for the first hole, as the machining depth exceeded both the total radial wear length and the workpiece thickness. However, by employing the proposed compensation method, all radial wear lengths are effectively compensated, as indicated by the green text.

Figure 10 presents a morphological comparison of microhole arrays produced using three different approaches: pure µEDM, hybrid machining without compensation, and hybrid machining with compensation. It is evident that pure µEDM produces microholes with a minimal taper angle. However, this comes at the expense of longer machining time. To support this observation, Fig. 11 provides a quantitative comparison of machining performance in terms of taper angle and machining time, based on three repeated trials (n = 3), with error bars representing the minimum and maximum values. As shown in Fig. 11b, the pure µEDM approach results in the longest machining time. This trade-off highlights the advantage of adopting LBMM–µEDM, which optimally balances machining speed and precision by leveraging the strengths of both processes. Moreover, as shown in Fig. 11, it is worth noting that the pure µEDM process exhibits greater variability in machining time, while the hybrid process without compensation shows higher variability in taper angle, as indicated by the wider range between the minimum and maximum values represented by the error bars.

Bild vergrößern

Bild vergrößern

Furthermore, Fig. 10 also demonstrates that hybrid machining without compensation exhibits a noticeable tapering issue, particularly evident in the exit holes, where the hole size progressively decreases throughout the machining sequence. This phenomenon leads to an increase in taper angle with successive machining, as illustrated in Fig. 11a. In contrast, hybrid machining with compensation effectively addresses this tapering problem, resulting in a lower taper angle that is comparable to that of pure µEDM. On average, the compensation method results in a 7 × reduction in taper angle compared to the hybrid process without compensation. Additionally, the taper angle in hybrid machining with compensation is 2 × smaller compared to pure µEDM.

In addition, hybrid machining with compensation exhibits only a slight increase in machining time compared to hybrid machining without compensation. This is primarily due to the need for deeper machining during the compensation process, which adds to the overall machining time. However, both hybrid machining methods are significantly faster than pure µEDM, as shown in Fig. 11b. On average, the hybrid methods achieve a 4 × reduction in machining time compared to pure µEDM.

In the hybrid process, residual spatters resulting from LBMM are still visible, as shown in Fig. 10. This is primarily because the µEDM electrode diameter is insufficient to completely remove the spatter during machining. Potential solutions include using a larger µEDM electrode diameter or reducing the size of the LBMM hole to ensure better overlap during the finishing process. Additionally, chemical etching has been identified as an effective postprocessing method to further eliminate residual spatter from LBMM and enhance surface quality, as demonstrated in previous studies [46].

These findings confirm that the compensation method effectively mitigates the tapering issue in the successive machining of microhole arrays by accurately compensating for radial tool wear length.