Micro-grooving of brittle materials using textured diamond grinding wheels shaped by an integrated nanosecond laser system

Figure 1a shows the laser conditioning configuration. A nanosecond pulse laser marking head (YLP-V2-1–100-30–30, IPG) was integrated on Z slide of an ultraprecision machine tool (Nanoform 250 Ultragrinder, Precitech). Two galvo motors in the marking head can adjust the laser machining area. The minimum spot size was about Ø50 µm in the focus plane, and the beam quality \({M}^{2}\) is 2. The beam quality is defined as a beam parameter to reflect how tightly a laser beam can be focused under certain conditions and it can be calculated by dividing the corresponding product for a diffraction limited Gaussian beam with same wavelength. More details about the laser beam are listed in Table 1. As shown in Fig. 1a, for laser dressing the grinding wheel (#1200 diamond grinding Ø 8 × 10 mm) was fixed on a tool holder which was then vacuum chucked on the machine spindle. The grinding wheel can move with the machine X slide for the fine adjustment to the laser marking head. A chromatic confocal probe (CHRocodile S, Precitec), which had 100 µm working range and 2 kHz sampling rate, was integrated into the ultra-precision grinding environment for on-machine surface measurement [27]. The integrated confocal probe was used for on-machine measuring the radial runout of wheels before and after laser truing. A USB microscopy (AM4815ZTL, Dino-Lite) was used for alignment and in-process monitoring laser conditioning and microstructure grinding process.

The grinding configuration is shown in Fig. 1b. A 25 × 25 × 5 mm glass ceramic substrate (Corning Macor) was vacuum-chucked on the spindle by a specially designed holder. The laser conditioned grinding wheel was clamped on an electro-spindle assembly (NR-3060S, Nakanishi). The rough and fine alignments were achieved by the USB microscopy and feeler gauges, respectively. The Clairsol 330™ coolant mist was used throughout the grinding process. The laser conditioned wheels and ground samples were ultrasonic-cleaned and then measured using a non-contact optical 3D microscopy (Alicona G4 InfiniteFocus). More parameters about the grinding wheel and the sample are listed in Table 2.

A preliminary was planned to explore the laser machining ability and find the laser processing parameters for truing, dressing, and profiling the grinding wheel. The four evaluated factors were pulse power \(e\), pulse overlap \({U}_{p}\), line overlap \({U}_{l}\), and loops, and the two responses were degree of graphitization and material removal volume. Three levels for each factor are estimated. An orthogonal test was designed according to [28] with four three-level factors L₉ (3⁴) as listed in Table 3. Totally nine groups of tests are performed. To exclude the environment disturbance, the experiment was performed at random order. The pulse overlap and line overlap were calculated by Eqs. (1) and (2). The material removal volume of each 2 × 2 mm square was measured by Alicona. The degree of graphitization was evaluated by observing the optical image of laser irradiated surfaces, where black surface represents serious graphitization of diamond grits.where the \(D\), \(S\), \(d\), \(f\), and \(pitch\) represent the wheel diameter, wheel rotation speed, spot diameter, pulse frequency, and spot interval, respectively.

Figure 2 shows images of laser ablated nine pockets according to machining parameters in Table 3. It is found that the laser irradiated surfaces in the 2nd, 4th, 5th, and 9th run were darker in color, indicating severe graphitization of the diamond abrasive grits. On the contrary, the laser machined surfaces in the 1st and 3rd run were bright, indicating less graphitization transition with small pulse energy (0.2 mJ) and low pulse overlap (20%). The range analysis was performed for both the material volume and graphitization. To quantitatively evaluate the influence of different factors on the graphitization, the index 1–4 is assigned to each pocket according to the burn degree. Range value \(R\), which can reflect the effect of estimated factors, was calculated bywhere \(i\) represents low, medium, and high level, \({y}_{i}\) is the output of each test from graphitization degree and material removal volume, and \({K}_{i}\) and \(\overline{{\text{K} }_{\text{i}}}\) are the sum and mean value of output at each level, respectively.

The impact of each factor is evaluated and ranked as shown in Table 4. It was found that the pulse energy dominates the material removal volume, and the pulse energy and pulse overlap are the two most important factors affecting the phase transition from diamond to graphite. To balance the material removal volume and degree of graphitization, the maximum pulse energy (i.e., \(e\) = 1 mJ) was chosen in laser truing for fast material removal, while small pulse energy (i.e., \(e\) = 0.2 mJ) was used in fine dressing to remove the graphitization layer and protrude the diamond grains.

The whole laser conditioning chain is schematically shown in Fig. 3. Firstly, the radial runout of the grinding wheel was controlled within ± 1.25 µm by the integrated gage. Then laser truing with \(e\) = 1 mJ, \(f\) = 30 kHz, \(S\) = 1000 RPM, and \(t\) = 5 min was performed to reduce the wheel radial offset and create the straight base line by repetitively scanning along the axial direction, as shown in Fig. 3a. The total removal depth was 100 µm with twice 50 µm radial infeed. After tangential truing, the laser dressing with low fluence (\(e\) = 0.2 mJ, \(f\) = 30 kHz, \(S\) = 1000 RPM, and \(t\) = 3 min) was carried out to remove only the metal bonding and protrude diamond grains in Fig. 3b. The desired pattern was then produced through the layer-by-layer irradiating mode as shown in Fig. 3c.

In order to improve the accuracy of profiling, a radial compensation strategy was adopted and schematically described in Fig. 3d. Along the laser beam propagation direction, most energy was focused within a narrow depth of focus (\(DoF\)). The energy transferred to the grinding wheel will induce heat-affected zone (HAZ) where the metal bonding material or diamond grains will be ablated if their fluence threshold is reached. However, when the truing line fed along radial direction, the laser should retreat a distance \(AB\) to place the \(DoF\) on the new exposed area of the grinding wheel. In this case, the \(DoF\) was evaluated to be ~ 850 µm by Eq. (6). The radial compensation distance \(AB\) for each laser scanning was calculated according to Eq. (7). The radius of the grinding wheel was 4 mm. For example, if the profiling infeed \(a\) is set as 30 µm for the 1^st laser beam scan, the grinding wheel should move away 489 µm along the X slide.where \(R\), \(\lambda\), \({M}^{2}\), \(r\), \(a\), and \(n\) are the wheel radius, laser wavelength, beam quality, spot radius, profiling infeed value, and radial infeed times, respectively.

Figures 4a−c shows the surface characteristics of the metal bond diamond grinding wheel at different laser conditioning phases, and the radial runout measured by the confocal probe is plotted in Figs. 4d−f. The surface of an as-received new grinding wheel (Fig. 4a) is very rough with some contaminants/bonding material left on it. Large co-axiality and radial runout were observed after mounting the wheel in the spindle which are resulted from the mounting error and the surface unevenness of the grinding wheel (Fig. 4d). After 5-min laser truing, redundant bonding and abrasive materials were ablated by the high fluence laser and a precise concentricity of the grinding and spindle axis was achieved. The measured radial runout was reduced to ~ 20 µm as shown in Fig. 4e. Followed by another 10-min laser fine truing (as illustrated in Fig. 3d), the outer layer material of the grinding wheel was fully ablated. Since only a small part of the laser was projected to the wheel surface, the absorbed and accumulated energy was small, and the material removal volume was slow. The grinding wheel surface after fine laser truing is shown in Fig. 4c and less than ~10 µm radial runout was achieved (Fig. 4f).

Figure 5 shows the surface topography and the statistical texture parametric analysis results of grinding wheels processed under different laser conditioning stages. The root mean square height \({S}_{q}\) and arithmetical mean height \({S}_{a}\) were selected as indicators to surface smoothness, and the peak height \({S}_{pk}\) value was used to describe the grain protrusion. As shown in Fig. 5a, the as-received grinding wheel had irregular surface (\({S}_{a}\) 16.3 µm and \({S}_{q}\) 22.8 µm) and abrasive grains were fully covered by bronze bonding. After 10-min laser fine truing, most diamond grains were pulled off which led to massive pits and converted graphite in situ as shown in Fig. 5b. The surface of grinding wheel was smooth with reduced surface roughness (\({S}_{a}\) = 1.4 µm and \({S}_{q}\) = 1.9 µm). To ensure good grinding performance, the protrusion height of grit after dressing should be within 1/3 ~ 1/2 of the grit size [18]. Therefore, another 5-min laser tangential dressing with low fluence was conducted and the bronze bonding materials were melted and then blowed away by the compressed air. A large number of diamond grains were disclosed again as shown in Fig. 5c. The surface roughness of grinding wheels is \({S}_{a}\) 9.9 µm and \({S}_{q}\) 14.2 µm, and the grain protrusion height was approximately 5.3 µm as indicated by \({S}_{pk}\).

Figure 6 shows the textures and cross-sectional profiles of diamond grinding wheels shaped by laser and mechanical dressing tip, respectively. As shown in Fig. 6a, b, riblet structures with different dimensions were designed and well shaped by the integrated laser beam. The laser structured grinding wheel has better form uniformity, and the shape and dimensions of structures can be flexibly controlled. In contrast, the mechanically profiled grinding wheel using diamond nib had large form deviation due to the diamond tip wear and elastic back-off of the grinding wheel (Fig. 6c, d). An important notice is that the laser profiling accuracy depends on the stability and energy distribution of laser pulse. The actual dimension of the laser processed structures was slightly larger than the designed dimensions. This is mainly caused by the overlapping of laser spot at the corners. The positioning error of the galvo motors, the loosening and pull-out of grains near the irradiated surface, and the re-deposition of melted bronze bonding materials on the top of wheel can also reduce form accuracy of shaped surface structures. The re-deposition layer has less strength than diamond grains and will be removed shortly in grinding process.

To demonstrate the grinding performance of laser conditioned diamond grinding wheels, three typical microstructures (riblet grooves, rectangle grooves, and pillar array) were machined. The riblets and rectangle grooves were processed with plunge grinding mode, while the pillar array was fabricated by twice plunge grinding with the workpiece 90° apart. The structure height was designed as 150 µm for riblets (textured grinding wheel shown in Fig. 6a) and 100 µm for rectangle grooves and pillar array (textured grinding wheel shown in Fig. 6b). Other parameters were set as \(S\) = 40 000 rpm, \(DoC\) = 10 µm/pass, \(feedrate\) = 5 mm/min, and \(v\) = 16.7 m/s, respectively.

Figure 7 shows the overview and close-up of the ground riblets on ceramic glass. Two groups of riblets with different periods and depths were machined on the ceramic glass. The ground structure has uniform profile accuracy along the 14 mm length, and the surface roughness \({S}_{a}\) was measured to be 1.6 µm. It is found that the period of riblet structure on the ground sample was 450 µm and 300 µm, respectively, which match the period of structure on the laser profiled grinding wheel (Fig. 6a). The measured depths of riblet structures were 120 µm for Group 1 and 140 µm for Group 2, which are slightly smaller than the designed grinding depth (150 µm) due to the macro-wear of the grinding wheel. Some cracks and fractures were observed on the ground surface by optical microscopy, indicating that the ceramic glass was removed under brittle removal method. During the grinding, the normal and tangential load between abrasive grains and workpiece accelerates the generation of lateral, median, and radial cracks. These cracks grow along different directions and intersect with each other to achieve the material removal. The brittle fracture leads to uncontrollable and larger material removal volume in machining hard and brittle materials; thus, the machined structure had larger dimension than the laser profiled structure on grinding wheel. Besides, the macro-wear along radial direction of the wheel can also deteriorate the form accuracy.

Figure 8 shows the overview and close-up of the ground rectangular grooves (14 mm × 4 mm) on the ceramic glass. The period of ground structures was 400 µm and 600 µm, which also match very well with the pitch of structure on wheel. The ground surface was rougher (\({S}_{a}\) 2.1 µm), and structure height was 20 µm shorter than design because of the serious wheel wear. Compared to the riblet structures in Fig. 7, the machined rectangular grooves have better form accuracy, indicating that the shaper transferability of rectangular texture on wheels is better. The reason might be that the grinding wheel wear is more stable for wheel with rectangular texture.

Figure 9a shows the measurement results of ground pillar array in a 5 mm × 6 mm square area. A variety of pillars with different dimensions were generated by the proposed crossing plunge grinding. Three groups of pillars, as indicated by red, black, and yellow squares, were extracted for detailed analysis (Fig. 9b−d). An important notice is that the depth of ground pillar array decreased gradually from the left bottom corner (~80 µm) to the top right corner (~50 µm) because of mounting unevenness of the sample. The height consistency can be improved by flat grinding the sample before fabricating microstructures.

The surface topography of the grinding wheel after machining is shown in Fig. 10. Macro-wear and micro-wear were observed, and the average height of structure left on the grinding wheel was reduced to around 70 µm after the grinding, indicating ~ 40 µm wear in height as compared with profile height being 110 µm (Fig. 6b). During the grinding process, mechanical and thermal loads are applied to the textured grinding wheel. Grains and bonding material are broken and pulled out under the intermittent normal and tangential force. Therefore, macro-wear happens along the radial direction of the grinding wheel. To quantitatively depict the wear condition of the textured grinding wheel, the grinding ratio, defined as the material removal volume divided by the volume of wheel wear, was estimated to be 6.3–6.5 in machining riblets and rectangle grooves. Moreover, the micro-worn of the texture of the grinding wheel was assessed by surface roughness parameters. It is found that the areal roughness reduced to \({S}_{a}\) 1.5 µm and \({S}_{q}\) 1.9 µm and the abrasive protrusion decreased from \({S}_{pk}\) 5.3 µm to \({S}_{pk}\) 1.4 µm. A possible reason for the micro-wear of the grits is that during the contact between grains and workpiece, the grinding force was gradually accumulated and led to grain fracture and bond breakage once the local load was unendurable. New grains were exposed but usually with blunt and negative cutting edges, thus inducing larger grinding force and smaller protrusion height.