1. Introduction

Laser finishing is a process for strengthening glass surfaces by healing exposed flaws to create a smooth surface. In this process, which is similar to fire polishing, a laser heats the glass to the point where the glass softens and flows under surface tension causing the surface to become smooth [1,2]. Typically, a carbon dioxide laser operating at 10.6 um is used for this process since this wavelength is readily absorbed in oxide glasses and these lasers are commercially available with high output powers. This process has been widely studied, particularly in fused silica, for a variety of applications including increasing the damage threshold in laser optics [2], laser polishing of optical components [3], healing laser damage spots [4,5], forming micro-lenses [6], and smoothing etched features [7].

In this work we apply the laser finishing process to the edges of thin glass sheets widely used in the consumer electronics industry. Considering the fabrication of LCD displays as an example, during the manufacturing process the edges of the glass sheets are exposed and may be required to contact alignment pins for precise orientation of display components. A weak or flawed edge may cause failure of the glass sheet during this process. Also, the accurate alignments required during display manufacture require that a precision edge be imparted to the glass. This is usually achieved with multiple steps of mechanical grinding with decreasing grit size which can be done at very high speeds (meters per minute). One significant disadvantage of the mechanical grinding process is the generation of particles that can adhere to the glass surface and require multiple washing steps to remove. Unfortunately, this process is not perfect, and some particles can be left adhered to the glass which can interfere with the manufacturing process for LCDs [8]. The requirements for surface cleanliness are becoming ever more stringent as the resolution of displays increases and pixel sizes decrease.

A significant advantage of the laser finishing process is the elimination of the multiple grinding steps and the creation of a smooth, strong edge without the generation of particles. These advantages apply quite generally to finished glass edges and not just in display manufacture used as an example above. However, the laser finishing process has several challenges which need to be understood before it can be utilized in a manufacturing environment. First, the glass part to be finished is usually preheated to minimize thermal gradients in the sample during laser processing [2]. Without this preheating step, the thermal stresses in the glass can become so high that a few millimeters of the edge will sheer off or the whole sample will shatter. Second, the thermal cycle imposed on the glass during laser finishing usually creates a tensile stress at the surface [1,2]. The presence of a tensile stress is likely to cause any flaws created on the edge to grow over time and can result in the failure of the finished part. The glass properties can also have a significant impact on the finishing process. For example, in [2], it was shown that fused silica with a low coefficient of thermal expansion (CTE) could be laser finished at room temperature but higher CTE glasses required preheating although the preheat temperature in this work was limited to 350°C. Laser polishing of optical glass with preheat temperatures of 600°C was reported in [3] but data for only a single preheat temperature was presented and no stress results were reported.

The purpose of the work presented here is to determine how to minimize the residual tensile stress in laser finished edges of thin glass sheets and we explore a range of laser processing parameters, such as laser power, laser exposure time, preheat temperature and cooling rate. We have studied this for different glasses typically used in the display industry covering a wide range of CTE values and for preheat temperatures as high as 700°C. We demonstrate, for the first time, that the correct choice of processing conditions can achieve a near zero residual stress or even a compressive stress on a laser finished edge.

2. Laser processing system

A laser processing system was developed that enabled the preheating of sheet glass samples and accurate positioning of them to enable laser heating of the sample edges. At the start of the finishing process a sample is loaded on to a standard laboratory hotplate at a temperature of 400°C. The hotplate has been modified to include locating pins to precisely position the glass samples in a repeatable manner. The glass is then picked up by a heated vacuum chuck on the end of a four-axis robot arm from Epson. The temperature of the vacuum chuck can be set from room temperature up to 700°C.

As illustrated in Fig. 1a, the glass overhangs the chuck by five millimeters on each side and because of the low thermal conductivity of glass the edges can be up to 100°C cooler than the vacuum chuck when the chuck is heated to ∼600°C. Therefore, the robot positions the glass above an infra-red heater, powered through a variable transformer, which provides sufficient energy to raise the edge temperature of the glass to be the same as the set point of the chuck. The temperature of the glass edge is calibrated as a function of vacuum chuck temperature and infra-red heater power using a 0.5mm diameter K-type thermocouple from Omega Engineering.

 

Fig. 1. System for laser finishing glass edges. (a) Schematic of glass sample held on heated vacuum chuck. (b) Schematic of rotating polygon mirror for laser heating the glass edge.

Download Full Size | PDF

The glass edge is finished by heating it with energy absorbed from a CO₂ laser beam. Activation of the laser is controlled by the same computer that operates the robot. The laser is focused on to the glass edge by an f-theta lens with an effective focal length of 100 mm from Special Optics. The laser spot size is approximately 300 um in diameter. The laser is scanned along the edge of the glass at a rate of 1 KHz using a 13-sided rotating polygon mirror from Lincoln Laser as depicted in Fig. 1b. The combination of scanner and f-theta lens produce a focused laser spot moving at approximately 100 m/s and essentially creates a 100 mm long line focus that is directed on to the glass edge. The diffusion distance for thermal energy absorbed by the glass is given by L=√(τκ), where τ is time and κ is the thermal diffusivity [9]. The thermal diffusivity for typical glasses used in display devices is ∼4 × 10⁻⁷ m²/s (calculated from glass properties given in [10]). In the 1 ms between laser scans the thermal energy deposited in the glass diffuses over a distance of approximately 20 um. This is similar to the absorption depth for silicate glasses at 10.6 um [11]. Therefore, at high scan frequencies the low thermal diffusivity of glass means that the sample effectively sees a constant heat source. During the laser finishing process the glass is monitored by a thermal camera and the temperature of the edge as a function of time can be recorded. When one edge of the glass sample has been finished in this way the laser is shut off and the sample rotates 90 degrees to finish the next edge. This continues until all four edges have been finished.

When the laser finishing is completed the sample is placed on to a second hotplate heated to 400°C. When this hotplate is full the power to it is switched off and it is allowed to cool naturally to room temperature over approximately one hour. The samples are then placed in a cassette for storage or transport to various test locations.

3. Laser finishing experiments

3.1 Process optimization

To optimize the laser finishing process the laser power and exposure time must be chosen carefully. The absorption of silicate glasses is very high at the 10.6 um wavelength used here [11] and trying to heat the glass too rapidly can lead to very high surface temperatures causing reboil of dissolved gas or volatilization of some glass components [12]. Additionally, prolonged heating of the glass causes the edge to flow to the point where the glass sheet starts to bulge. This deviation from flatness is undesirable from a glass handling perspective.

Figure 2 and Fig. 3 show microscope images of several samples processed in an optimization experiment performed with 0.4 mm thick Corning EAGLE XG glass. In this experiment the samples were preheated to 600°C, the laser exposure time (defined as the time for which the laser was turned on) was fixed at two seconds, corresponding to two thousand scans of the laser along the glass edge, and the laser power varied. Before laser processing, the edges were prepared by mechanical grinding with a 400-grit wheel to create a consistent initial edge quality for each laser processing experiment. Furthermore, applying the laser finishing process to this edge eliminates the need for any further mechanical grinding steps that would otherwise be required to reduce the size of any edge flaws and increase strength.

 

Fig. 2. Microscope images of laser finished edges for various laser power densities on Corning EAGLE XG glass. The laser exposure time was 2 seconds and the samples were preheated to 600°C in all cases.

Download Full Size | PDF

 

Fig. 3. Microscope images showing the cross section of laser finished edges for various laser power densities. The position of each image corresponds with that of Fig. 2.

Download Full Size | PDF

Figure 2 shows the laser finished edges for several different laser power densities ranging from 155 to 288 Watts/cm². The power density here is defined as the average laser power incident on the glass sample divided by the area of the edge of the sample. These views are all taken normal to the finished edge in the plane of the sample. At the lowest power density of 155 W/cm² there is no apparent change in the edge from the as received sample. Increasing the power density to 183 W/cm² clearly modifies the edge but there is still some roughness. At 210 W/cm² and higher the edge appears to be completely smooth but for power densities greater than 250 W/cm² the edge is seen to be bulging.

Figure 3 shows cross-sectional images of each sample and the position of each image corresponds with those in Fig. 2. These images clearly demonstrate that as the edge becomes more finished (i.e. more of the initial roughness is removed) then there is a greater rounding of the edge. The optimum value of laser power density for this combination of glass and preheat temperature seems to lie in the range 210-235 W/cm² as this will give the greatest level of flaw removal without causing the glass to bulge. This optimization process was repeated for each glass type, each glass thickness and each preheat temperature investigated in this work.

3.2 Laser finishing of high CTE glasses

Three different aluminosilicate glass compositions were chosen for laser finishing experiments and their properties are summarized in Table 1. Rectangular samples of each glass were prepared from fusion drawn sheets and the edges ground with a 400-grit wheel to give a final dimension of 60 × 44 mm. The Corning EAGLE XG samples were 0.5 mm thick and the glass A and glass B samples were 1.0 mm thick.

Table 1. Properties of glasses used in laser finishing experiments

View Table

Laser finishing experiments were performed on each glass for a range of preheat temperatures up to a maximum of 700°C using a fixed laser heating time of two seconds. As described above, the laser power was optimized for each glass and each preheat temperature to maximize flaw reduction but without causing bulging of the sample. It was found that EAGLE XG glass could be finished for preheat temperatures in the range 450 to 700°C while glass A could be laser finished at preheat temperatures in the range 500°C to 700°C. At lower preheat temperatures the glass was observed to shatter immediately after the laser heating was completed. The highest CTE composition (glass B) could not be laser finished at any of the preheat temperatures used in these experiments. The glass would usually shatter after being loaded on to the vacuum chuck. No further experiments were performed with this glass.

The optimized power density used for each glass as a function of preheat temperature is plotted in Fig. 4. For both glasses the required power density decreases as the preheat temperature increases. This is simple to understand since the energy that has to be put into the glass to reach the required temperature for laser finishing decreases as the preheat temperature is raised. Also, the required power density for EAGLE XG is higher than for glass A since the softening point is higher (971°C vs 900°C) and the temperature of EAGLE XG has to be higher for the edge finishing process to work. The power density for finishing EAGLE XG at a preheat temperature of 600°C in this case was found to be 273 W/cm² which is slightly above that observed in section 3.1. The reason for this was not clear but may have been due to some misalignment in the system.

 

Fig. 4. Optimized laser power densities for edge finishing of EAGLE XG and glass A as a function of preheat temperature.

Download Full Size | PDF

The residual stress in the laser finished samples was measured for each preheat temperature using a StrainScope imaging polarimeter from ilis GmbH [13]. The instrument determines the stress induced birefringence by measuring the rotation of polarized light passing through the sample. The stress in the sample is then calculated from the birefringence using the stress-optic coefficient.

Figure 5 plots the residual stress after laser finishing as a function of distance from the laser finished edge for both EAGLE XG and glass A for several different preheat temperatures. The plots show that for most samples there is significant tensile stress close to the laser finished edge which turns into a compressive stress at a distance of 2-3 mm inside the sample. As the preheat temperature applied to the glass before laser finishing is increased, the magnitude of the tensile stress decreases and the penetration depth for the tensile stress in to the glass tends to increase. Most significantly, at a preheat temperature of 700°C we observe compressive stress in glass A which extends further in to the sample than the tensile stress observed at lower preheat temperatures. Also, for a preheat temperature of 650°C the residual tensile stress at the laser finished edge in glass A is very small at <4 MPa.

 

Fig. 5. Residual stress in laser finished glass sheets. (a) EAGLE XG, (b) Glass A. Positive values represent tension.

Download Full Size | PDF

The peak value of the residual stress parallel to the finished edge is plotted in Fig. 6 as a function of preheat temperature. The graph clearly shows that the residual stress decreases with increasing preheat temperature in both glasses. Linear regression curves show good fits to the data for both glasses and the trendline for glass A is somewhat steeper than for EAGLE XG. The residual stress in EAGLE XG drops from 86 MPa at a 450°C preheat to only 22 MPa with a 700°C preheat. The trendline predicts that zero residual stress could be achieved for a preheat temperature of 800°C but this was beyond the range of our equipment.

 

Fig. 6. Residual stress in laser finished EAGLE XG and Glass A as a function of preheat temperature. Symbols represent measured data and solid lines are linear fits. Positive values of stress represent tension.

Download Full Size | PDF

For preheat temperatures above 650°C the residual stress in glass A is compressive and, for a 700°C preheat temperature, the magnitude of the compressive stress is as high as 30 MPa. It is also clear from Fig. 6 that, since the residual stress changes from tension to compression at preheat temperatures of approximately 650°C, if we finish glass A at this temperature we can generate an edge with very little residual stress. Alternatively, we can finish the glass using preheat temperatures above 650°C to generate edges with compressive stress that would be expected to resist the growth of any flaws remaining after laser treatment.

3.3 Laser finishing with a controlled cool down

Considering the results of the previous section, modifications to the laser finishing process were sought that could either reduce the residual stress or allow the finishing process to occur at a reduced preheat temperature. The residual stress can be completely eliminated by annealing but this is generally quite a slow process requiring the glass to be heated above the anneal point and cooled over a period of hours. As an alternative we chose to try cooling the glass more slowly by gradually reducing the laser power to zero once the laser finishing step was completed. This was accomplished by reducing the laser power in a series of discrete steps where the duration and power of each step could be controlled by the same computer controlling the robot system.

Examples of the glass edge temperature, measured with a thermal camera, generated using this technique are shown in Fig. 7. The samples are preheated to a nominal temperature of 600°C (This temperature represents the set-point of the vacuum chuck as measured by a thermocouple. The thermal camera reading was often slightly higher). The black line in Fig. 7 represents the thermal cycle of the glass edge when only laser finishing is performed as described in section 3.2. In this case the laser is turned on for two seconds and then shut off. The glass temperature reaches a maximum of approximately 1300°C and then cools naturally. The red line is an example of laser finishing followed by a slow cool down process that is implemented by decreasing the laser power in six steps over a period of thirty seconds. The laser finishing step is identical to the previous case with the laser being active for two seconds. The first reduction in power after the laser finishing step is completed causes a rapid drop in temperature of the glass edge by approximately 200C and no further shaping of the edge occurs.

 

Fig. 7. Temperature of glass edge during laser finishing process recorded by a thermal camera.

Download Full Size | PDF

Samples of EAGLE XG and glass A were laser finished using the controlled cool down process for preheat temperatures ranging from 500 to 700°C. When this process is applied to all four sides of a glass sample, the sample is spending upwards of two minutes on the heated vacuum chuck. To determine if this amount of heating was having any effect on the glass several control samples were also created at each preheat temperature. For each control sample, the sample went through all of the same process steps as for the laser finished samples but without the laser being turned on. Therefore, the control sample spends the same amount of time on the heated vacuum chuck and undergoes the same preheating and cool down cycles as the laser finished samples but receives no laser heating. The residual stress at the edge was measured using the same technique described previously and is plotted in Fig. 8. Peak stress data generated by just the laser finishing step described in section 3.2 is included for comparison.

 

Fig. 8. Residual stress in laser finished glass edges with a controlled cool down depicted in Fig. 7. (a) Peak residual stress in EAGLE XG; (b) Stress vs distance in EAGLE XG; (c) Peak residual stress in Glass A; (d) Stress vs distance in Glass A. In (a) and (b) points represent measured data and solid lines represent linear fits to the data. No trendline was fitted to the data for the control samples in (a).

Download Full Size | PDF

Figure 8(a) demonstrates that for a given preheat temperature the slow cool down process lowers the residual tensile stress in EAGLE XG compared to just using a simple finishing step. Also, for a preheat temperature of 700°C the residual stress becomes compressive. The transition from tension to compression in EAGLE XG can be seen in the stress versus distance plot of Fig. 8(b). At a preheat temperature of 600°C the peak tensile stress of 38 MPa occurs close to the laser finished edge. When the preheat temperature is raised to 700°C the stress becomes compressive but now the peak of the stress, approximately 9 MPa, occurs 3.5 mm away from the laser finished edge. This contrasts with the compressive stress seen in Glass A in Fig. 5 and Fig. 8(d) where the peak of the compressive stress occurs at the edge.

Figure 8(c) and (d) show a similar impact of the slow cool down process on Glass A. When the slow cool down is used the preheat temperature where the residual stress passes through zero is reduced from 650°C to approximately 610°C. For a 700°C preheat temperature the compressive stress has been increased to 48 MPa and shows a similar shape to that seen in Fig. 5b where the peak stress is close to the finished edge.

Figure 8(a) and (c) also show that compressive stress was generated in the control samples that were only subjected to preheating and not laser finished. At a 700°C preheat temperature the compressive stress in EAGLE XG was 29 MPa while in Glass A it was 48 MPa which is identical to the compressive stress remaining after laser processing. While this is an interesting result, the unfinished edge will still have larger flaws than the laser finished edge and is still expected to be weaker. Laser finishing with a slow cool down is expected to provide the most improvement in strength because of the removal of flaws.

4. Discussion

Ultimately, we wish to understand the residual stress in the glass sheets after laser edge finishing in terms of both the glass properties and the laser processing parameters. Residual stress in glass arises when different parts of a glass sample are subjected to different thermal histories. One well known example of this is the tempering of plate glass to produce a compressive stress on the surface [14]. While we do not yet have a full physical description for our process we can make a few general comments. There is a clear trend in Fig. 6 and Fig. 8 that residual tensile stress decreases as the glass preheat temperature is increased. The likely reason for this is that the temperature gradient between the glass edge during laser finishing and the bulk of the sample is reduced as the preheat temperature is raised. The residual tensile stress is also substantially decreased by using the slow cooling process introduced in section 3.3. The slow cooling allows further reductions in thermal gradients within the glass compared to the rapid cooling that occurs when the laser is abruptly turned off.

Based on the difficulty observed in finishing high CTE glasses (e.g. glass B in Table 1 of this work and BK-7 in [2]) we expected to see higher residual stress in the high CTE glass. However, as shown by the data of Fig. 6 we actually saw lower stress in the higher CTE glass. If we look at Fig. 6 carefully we note that the curves for the two glasses cross at a preheat temperature of approximately 500°C. If glass A survived the laser finishing process at temperatures below this point, we would indeed expect to see higher residual stress than in EAGLE XG. However, at the elevated preheat temperatures used in our experiments we are approaching or exceeding the strain point of the glasses undergoing laser finishing and it is possible that viscoelastic behavior of the glass is starting to influence the observed residual stress.

5. Summary

In this work we have described a laser finishing process for the edges of glass sheets intended to produce strong edges with minimal flaws and a reduced need for grinding and the associated particle generation. Laser processing typically leaves glass with tensile stress that can lead to the growth of flaws remaining on the edge, but we demonstrated that, by appropriately choosing the process conditions, the residual stress can be almost eliminated or be made compressive. We believe that this is the first time that such residual stress conditions have been observed in laser finished glass and that this is of significant value to the laser materials processing field. The elimination of the residual tensile stress means that laser processing can be performed without requiring an additional annealing step. The generation of compressive stress will hinder the growth of flaws at the glass surface and thus increase the strength of the glass.

Compressive stress as high as 9 MPa has been demonstrated in EAGLE XG glass and as high as 48 MPa in an alternative aluminosilicate glass composition with a higher coefficient of thermal expansion. The process conditions required to achieve a particular stress condition depend on material properties, such as CTE, and further study is required to determine the exact nature of the link between glass properties, residual stress, and the crossover from tensile to compressive stress. Also, glasses with very high values of CTE remain difficult to finish with the laser processing technique described here and additional work is required to understand how laser processing can be applied to these materials.