Chapter 4 Laser microfabrication of glass substrates by pocket scanning Precise laser microfabrication of glass is a high challenge task due to the stress-induced micro-cracks generated during laser ablation. In this chapter, the results of high quality glass microfabrication by low energy Nd:YAG laser ablation and pocket scanning technique are presented. The cracks formed around the edges by the pocket scanning are reduced significantly compared to laser direct scanning. The ablation depth is also enhanced greatly by the pocket scanning. It increases almost linearly with laser fluence and number of scanning loop. There is no saturation as observed in laser direct scanning. This technique of laser ablation with pocket scanning has high potential applications in glass precision engineering. 4.1 Background Glass micro-devices have many applications in analytical chemistry, biotechnology and microelectronics because of its excellent mechanical properties, chemical stability and transparent nature. Glass microstructure fabrication is the first step to fulfill its applications [83,84]. Current mechanical and chemical methods for glass microfabrication are affected either by poor 72
processing edge quality or toxic agents employed. Developments of glass microfabrication techniques are in strong demand in various industries. Pulsed laser ablation is a powerful micro-processing tool attributed to its small spot size with high localized energy [85]. However, due to the high bandgap and low thermal conductivity of glass, micro-cracks generated from thermal stresses affect the reliability and quality of the formed microstructures [86]. Crack-free glass microfabrication is a high challenge task for precise laser microfabrication. So far, different kinds of lasers have been proposed and studied for the glass microfabrication. The laser sources include CO 2 laser [87], excimer laser [88], Nd:YAG laser [89], F 2 laser [90] and ultrafast laser [91]. The drawbacks of CO 2 laser lie in that it provides relatively large beam spot and causes severe thermal effects due to its long wavelength, while the use of excimer laser suffers from the high running and maintenance cost. The studies of glass microfabrication by advanced F 2 and ultrafast lasers have obtained excellent results, but the extensive application of these lasers is still limited by their cost and stable problems. DPSS Nd:YAG laser is a low cost and mature light source for industrial applications. However, since most Nd:YAG laser light isn t absorbed by glass [91], the Nd:YAG laser ablation of glass substrate requires a high laser fluence and causes many micro-cracks due to the diffusion of excess energy and thermal effects. In this chapter, the pocket scanning technique is used to produce microstructures on glass. The pocket scanning is termed in computer assisted manufacturing (CAM) as parallel overlapped scanning paths. The low pulse 73
energy DPSS Nd:YAG laser (355 nm, 30 ns) with computer-generated toolpaths was employed. Micro-cracks formed around ablated grooves were reduced significantly compared to laser direct scanning. Ablation efficiency and edge quality were measured and compared for direct scanning and pocket scanning methods under various combinations of processing parameters. Precise glass microstructures were fabricated with the optimized setting of laser processing parameters. 4.2 Experimental setup The experimental setup for the glass precise microfabrication is shown in Fig. 2.6. The repetition rate of the DPSS Nd:YAG laser was set as 5 khz for glass microfabrication. The laser beam was focused by a Plano convex lens at a focal length of 50 mm. All the experiments were performed by focusing laser beam on the front surface of the glass. Laser fluence (pulse energy per irradiated area) was tuned in the range from 0.03 to 132.3 J/ cm 2. The CAM software was used to design the microstructures with the speed-controlled toolpaths. Laser ablation was carried out on lime sodium glass sheets with thicknesses of 100 µm and 700 µm. The samples were rinsed with acetone and deionized water before and after the laser ablation. Widths and depths of the processing results were measured from the cross section view of the profile under an optical microscopy. Figure 4.1 shows the absorption spectrum of the glass substrates for the irradiated light wavelength from 200 to 800 nm 74
measured by a scanning spectrophotometer (SHIMADZU UV-3103PC). The irradiated light with the wavelength less than 220 nm was totally absorbed by the sample, and the lime sodium glasses exhibit an absorption edge almost near 320 nm. Therefore, for the UV light at the wavelength of 355 nm, the material is practically transparent. A high laser fluence is required to cause the material removal. However, when concentrated energy is transferred from photons to the glass substrates, substantial heat diffusion induces thermal stress, resulting in the significant cracked edges. The cracks formed around the ablated edge were characterized with optical images. Crack sizes at different parameters were defined as the maximum lateral length of visible damages from the ablated edge. The surface quality after laser irradiation was also observed with SEM. 1.0 Normalized light absorption 0.8 0.6 0.4 0.2 Lime sodium glass thinkness: 700 µm 0.0 200 300 400 500 600 700 800 Wavelength (nm) Fig. 4.1 Absorption spectrum of the lime sodium glass for light wavelength from 200 to 800 nm. 75
4.3 Results and Discussion 4.3.1 Pocket scanning vs direct scanning The term of pocket scanning is used in CAM software as the use of an overlapped toolpath to scan inside a specific area. Schematic drawing of laser direct scanning and pocket scanning is shown in fig. 4.2. In microfabrication, the direct method scans the laser beam just along the designed structure edge, while the pocket method scans a series of parallel paths with desired overlapping, either the first path or the last path can be used to form the edge of the structure. A selected number of parallel paths are defined as one loop for the pocket scanning. Therefore, the direct scanning can be considered as a special case of the pocket scanning with only one single path in a loop. The sequent loops repeat along the same paths. 76
(a) (b) Fig. 4.2 Schematic drawings of (a) laser direct scanning and (b) laser pocket scanning methods. 77
Figure 4.3 shows the optical image of laser scanning results on glass surface by these two methods. Laser fluence was 35.7 J/cm 2 and scanning speed 0.8 mm/s. A trench with a width of 15 µm was fabricated by direct scanning with one loop. The edges of the trench exhibit extended splintering and cracking. The cracks are formed randomly at the rim of the trench and the maximum lateral size of the cracks is 43.4 µm. For comparison purpose, the same laser parameters were applied to the pocket scanning processing. One pocket loop, consisting of 10 parallel paths scanning from the left to the right, was used to scan on the sample surface. The distance between the centers of two parallel paths was set as 10 µm. Since the width of the directly formed trench is 15 µm, the overlapping of the adjacent trenches in the pocket scanning is 5 µm (overlapped 33%). It can be observed that there are significantly different morphologies for the starting edge and the end edge. At the starting edge formed by the first path, the sizes of formed cracks were almost the same as the direct scanning. It is because the same processing parameters were applied. In contrast, the sizes of cracks are reduced greatly at the end edge, and the crack distribution is more uniform than that at the opposite side. The improvement of the crack condition at the end edge has the potential to achieve high quality glass structures. In the following parts, the cracks generated in the pocket scanning will be specified as those formed at the end edge. 78
Fig. 4.3 Optical image of trenches fabricated on glass by laser direct scanning and pocket scanning with one scanning loop. Laser scanning speed is 0.8 mm/s at a laser fluence of 35.7 J/cm 2. The pocket scanning contains 10 parallel lines with an overlapping of 5 µm in one loop. 4.3.2 Crack size as functions of laser flunence and number of scanning loop for direct and pocket scanning Figure 4.4 shows the crack size as a function of laser fluence for the direct scanning and pocket scanning. The laser scanning speeds were 1.2 and 1.6 mm/s, respectively. In the pocket scanning, one loop of 10 parallel paths with the overlapping of 5 µm was scanned on the sample surface. It is found that in the direct scanning, the crack sizes are in the range from 30 µm to 50 µm, while in the pocket scanning, the crack sizes are greatly reduced, most of them are less than 30 µm. The minimum crack size appears at a laser fluence of 56.7 79
J/cm 2. As laser fluence increases from 19.8 to 56.7 J/cm 2, for scanning speed of 1.2 mm/s, the crack size decreases from 44.5 to 31.5 µm for the direct scanning; while for the pocket scanning, it decreases from 24.9 to 10.8 µm. When the laser fluence is higher than 56.7 J/cm 2, the crack size increases accordingly. For different scanning speeds at 1.2 and 1.6 mm/s, it is also found that the lower scanning speed results in smaller cracks for both methods. Maximum crack size (µm) 60 50 40 30 20 Glass 355 nm, 30 ns Repetition Rate: 5 khz Direct scanning Pocket scanning Scanning speed: 1.2 mm/s 1.6 mm/s 10 20 40 60 80 100 120 140 Laser fluence (J/cm 2 ) Fig. 4.4 Dependences of crack size on laser fluence by the direct scanning and pocket scanning at the scanning speeds of 1.2 mm/s and 1.6 mm/s and a scanning loop of 1. Detailed examination of the crack size in terms of the laser scanning speeds was carried out in fig. 4.5. The scanning speeds were from 0.8 to 5 mm/s at laser fluences of 19.6 J/cm 2, 56.7 J/cm 2 and 132.3 J/cm 2, respectively. The parameters of the pocket scanning (10 parallel paths, 5 µm overlapping) 80
were the same to the previous part. It is obvious that associated with increasing scanning speed, the cracks formed by the pocket scanning increase, while they don t change much in direct scanning. It is also found that the increase of crack size for the pocket scanning is even more rapid than that for direct scanning. At a large scanning speed of 5 mm/s, the crack sizes are nearly the same for both of these two methods. It indicates that a lower scanning speed should be desired for reducing the cracks for the pocket scanning. Maximum crack size (µm) 70 60 50 40 30 20 Glass 355 nm, 30 ns Repetition Rate: 5 khz Laser fluence: 19.6 J/cm 2 Direct scanning Pocket scanning 10 0 1 2 3 4 5 Laser scanning speed (mm/s) (a) 81
60 Maximum crack size (µm) 50 40 30 20 Glass 355 nm, 30 ns Repetition Rate: 5 khz Laser fluence: 56.7 J/cm 2 Direct scanning Pocket scanning 10 0 1 2 3 4 5 Laser scanning speed (mm/s) (b) Maximum crack size (µm) 70 60 50 40 30 20 Glass 355 nm, 30 ns Repetition Rate: 5 khz Laser fluence: 132.3 J/cm 2 Direct scanning Pocket scanning 10 0 1 2 3 4 5 Laser scanning speed (mm/s) (c) Fig. 4.5 Dependences of the maximum crack size on the laser scanning speed by laser direct scanning and pocket scanning at laser fluences of (a) 19.6 J/cm 2, (b) 56.7 J/cm 2 and (c) 132.3 J/cm 2. The scanning loop is set as 1. 82
It is generally understood that the formation of cracks is due to thermal effect in nature. The irradiated laser pulses provide the photonic energy to the glass, the excited electrons dissipate excess energy into the lattice by generating phonons. For the laser pulse duration in the nanosecond range, heat transferring from hot electrons to the lattice plays a significant role to cause the rise of lattice temperature, subsequently generate thermally induced stresses [92]. The smaller crack size in the pocket scanning can be explained as following: the bandgap energy of glass (~ 8 ev) is much higher than the photon energy of 355 nm laser (~ 3.4 ev) [93], electrons cannot absorb laser energy linearly to cause the evaporation of materials at a low laser fluence. As the laser fluence increases, high photon density leads to the non-linear (e.g. multiphoton) absorption. Material ablation occurs when electrons are excited sufficiently to ionize atoms or molecules. However, due to the poor thermal conductivity, a portion of the heat energy is concentrated at the nearby region. Ablation and heat induced stresses take place simultaneously and result in crack formation around the ablated trench in the direct scanning. In the case of the pocket scanning, laser after the first path scanning irradiates on the surface where the cracks are already formed by the previous scanning. The presence of the cracks can increase the local field intensity by a factor of η 4 for transparent materials, where η is the refractive index [94]. As a result, efficient laser ablation takes place due to the much high absorption at the defect region, and the thermal stresses are reduced corresponding to the less heat diffusion energy. Therefore, the cracks generated from the subsequent scanning are less than the initial one to form a sharp ablation edge. 83
The presence of minimum crack size at a critical laser fluence can be attributed to the balance between laser ablation and thermal accumulation. For low laser fluence just above the damage threshold, heat dissipation into the lattice occurs to cause cracks at the surface. The energy required to vaporize transparent materials is much higher than that to cause them to fracture [95]. When laser fluence increases, more energy contributes to the ablation through non-linear absorption, the crack sizes decrease responding to less heat diffused. Smaller crack size can be obtained. Further increase of laser fluence causes the energy for heat accumulation to increase, resulting in the increased manifestation of residual stresses around the ablation region [96]. As a result, the size of stress induced crack increases with laser fluence. Another possible mechanism can be due to the generation of shock wave and breakdown plasma during nanosecond laser ablation. Such processes take place at a high laser fluence to cause fractures in the vicinity of the interaction zone [97]. The increase of crack size with scanning speed can be related to the decrease of pulse overlapping. As the scanning speed increases, subsequent pulse irradiates a position farther away the cracked area formed by previous pulses. This reduces the interaction between the laser and the cracked area to cause more energy diffused as heat. The crack size increases accordingly. 4.3.3 Glass microfabrication by optimized parameters To meet the requirement of high speed fabrication in industries, the higher scanning speed is required. In fig. 4.5, it can be observed when the scanning 84
speed was less than 1.2 mm/s, the crack sizes do not decrease much at a low speed. The speed of 1.2 mm/s can be used as one of the optimized processing parameters. Figure 4.6 shows the SEM images of the glass microstructure fabricated by the optimized parameters at a laser fluence of 56.7 J/cm 2, a scanning speed of 1.2 mm/s and 10 parallel paths with an overlapping of 5 µm. 15 repeated loops were used to cut through a 700 µm glass substrate with high quality edges. The crack around the edge is less than 10 µm. The ablated sidewall is very smooth and vertical to the surface, no obvious slope can be observed. There are some cracked edges at the rear side of glass. It can be due to that when laser cuts through the glass substrate, suddenly ejected materials fracture the neighboring edge to cause cracks there. Further improvement of the edge quality may be achieved by fabricating the microstructures with a protective layer, such as coating a photoresist layer at the rear side. 85
(a) (b) Fig. 4.6 SEM images with magnifications of (a) 25 and (b) 100 of a glass microstructure fabricated by the pocket scanning at a laser fluence of 56.7 J/cm 2, a scanning speed of 1.2 mm/s and a scanning loop of 15. 86
4.3.4 Ablation depth versus laser fluence and number of scanning loop for direct scanning and pocket scanning The processing depth is another important parameter needed to be considered. Figure 4.7 shows the laser ablation depth as a function of laser fluence at a scanning speed of 5 mm/s for the direct scanning and pocket scanning. For the pocket scanning, 10 paths were used at different laser fluences. The ablation depth of glass increases from 12 to 61 µm for the direct scanning, while from 82 to 208 µm for the pocket scanning, as laser fluence increases from 20 to 132.3 J/cm 2. It is clear that the ablation depth is much higher for the pocket scanning than the direct scanning. It can also be observed from fig. 4.7 that the ablation depth for the direct scanning saturates at about 50 µm for laser fluence above 100 J/cm 2. While for the pocket scanning, the ablation depth increases almost linearly with laser fluence. More substrate material removal is attributed to the overlapped scanning and increased light absorption of laser energy at the cracked area. Furthermore, the enlarged width of the ablated trench also leads to the increase of ablation depth in the pocket scanning. More debris generated during laser ablation is easier to be ejected out of the ablated trench. Less debris accumulated around the trench provides much stronger laser interaction with glass substrate than the direct scanning. 87
Ablation depth (µm) 200 150 100 Glass 355 nm, 30 ns Scanning speed: 5 mm/s Repetition rate: 5 khz Scanning loop: 1 Pocket scanning 50 Direct scanning 0 25 50 75 100 125 Laser fluence (J/cm 2 ) Fig. 4.7 Ablation depth versus laser fluence for the direct scanning and pocket scanning. Laser scanning speed is 5 mm/s and a scanning loop of 1. Figure 4.8 presents the ablation depth versus number of scanning loop at a laser fluence of 132.3 J/cm 2 and a scanning speed of 5 mm/s. It is found that the ablation depth of the pocket scanning increases almost linearly at 220 µm/loop. Three loops of the pocket scanning can cut through the 700 µm glass substrate. For the direct scanning, the ablation depth starts to saturate at 450 µm after 25 loops. It can be also found in the figure that up to 50 loops of the direct scanning cannot cut through this glass sheet. The saturation of ablation depth is mainly due to the accumulation of the ablated debris, which limits the further laser interaction with the glass substrate. The wide trench fabricated by the pocket scanning allows the ablated materials to be ejected out more easily since the nearby area has already been ablated away. The ablation depth for the pocket scanning increases greatly. 88
Ablation depth (µm) 800 600 400 200 Cutting through Pocket scanning Glass (thickness: 700 µm) 355 nm, 30 ns Repetition rate: 5 khz Laser fluence: 132.3 J/cm 2 Scanning speed: 5 mm/s Direct scanning 0 0 10 20 30 40 50 Number of scanning loop Fig. 4.8 Ablation depth versus number of scanning loop for the direct scanning and pocket scanning. (Glass thickness: 700 µm, laser scanning speed: 5 mm/s and laser fluence: 132.3 J/cm 2 ). The optical cross-section images of the deep trenches fabricated by the direct scanning and the pocket scanning are shown in figs. 4.9 (a) and (b). It indicates that there are some redeposited materials accumulating at the bottom of the narrow trench during the direct scanning. They block laser light for the further ablation of glass substrate. While in the case of the pocket scanning, there are no such redeposited materials observed at the bottom. It demonstrates that the redeposition of ablated glass materials hampers the increase of ablation depth in the direct scanning. 89
(a) (b) Fig. 4.9 Optical cross section images of deep trenches fabricated by (a) direct scanning of 10 loops and (b) pocket scanning of 1 loop at a laser fluence of 132.3 J/cm 2 and a scanning speed of 5 mm/s. 90
4.3.5 Energy-dispersive X-ray spectrum analyses The interaction between the laser and glass substrate can be studied by comparing the chemical compositions of glass surface before and after the laser irradiation. Energy-dispersive X-ray spectrum (EDX) analyses for the original and laser irradiated glass surfaces are shown in figs. 4.10 (a) and (b). The compositions determined by the two spectra are nearly identical due to the almost same peak positions and intensities. The results suggest that the chemical compositions of the surfaces are not changed upon the interaction with the laser light. Therefore, the main mechanism of glass removal can be attributed to the photothermal laser ablation. Since energy of a photon from the 355 nm laser is not sufficient to break a typical Si-O bond (3.6 ~ 4.2 ev) [98], laser ablation can be attributed to the multiphoton absorption at the cracked areas to dissociate such chemical bonds. Other physical changes such as heating and melting can only respond to a small portion of the overall surface modification. 91
Fig. 4.10 EDX spectra of chemical compositions of (a) original and (b) laser irradiated glass surfaces. 4.3.6 High quality glass microstructure formation by laser pocket scanning The high ablation efficiency and the good control of edge quality enable the pocket scanning technique to be a promising tool in the precise engineering 92
of glass. Figures 4.11 (a) and (b) demonstrate the capabilities of the laser pocket scanning to create arbitrary contours on the glass sheets with a thickness of 100 µm. 200 µm width cantilevers with crack sizes less than 10 µm were fabricated at a laser fluence of 56.7 J/cm 2, a laser scanning speed of 1.2 mm/s and a scanning loop of 3. Further deposition of a thin layer of piezoelectric materials or a metallic thin film with a counter electrode can make the cantilever functioned as an inertial sensor [99]. (a) 93
(b) Fig. 4.11 Photographs of glass microstructures (thickness: 100 µm) fabricated by the optimized parameters of a laser fluence of 56.7 J/cm 2, a scanning speed of 1.2 mm/s and a scanning loop of 3. 4.4 Summary The feasibility of glass microfabrication by UV DPSS Nd:YAG laser (355 nm, 30 ns) was investigated. The results show that severe cracks can be developed on glass during laser ablation by the direct scanning. Pocket scanning was employed to scan the laser beam along parallel overlapped paths with the last path to form the microstructure edge. It was found that by using the pocket scanning, the cracks around the structure edges can be reduced significantly to a few micrometers. The crack size has a minimum value at a critical laser fluence and increases with the laser scanning speed. The ablation 94
depth also increases without the saturation in the direct scanning. The presence of smaller crack size and higher ablation depth for the pocket scanning is attributed to the high optical energy absorption at the defect region. The wide trench fabricated by the pocket scanning also causes the ablated materials to be ejected out more easily to increase the ablation depth. Optimized processing parameters of a laser fluence of 56.7 J/cm 2, a laser scanning speed of 1.2 mm/s and 10 parallel paths with an overlapping of 5 µm were applied to fabricate high quality glass microstructures with crack sizes less than 10 µm. 95