Laser polishing and laser shape correction of optical glass

Research has already been conducted on laser polishing of fused silica since 1982 by Temple.¹ Using defocused cw CO₂ laser radiation and a meandering scanning strategy, an initial roughness of rms = 0.33 nm could be reduced by Temple to rms 0.27 nm (measuring field 500 × 400 μm²). Nowak et al. showed in Ref. 2 the first laser polishing method with pulsed CO₂ laser radiation for industrial usage by fabricating micro optics. Richmann et al. and Hildebrand et al.^3,4 investigated laser polishing with a quasiline with cw CO₂ laser radiation. Due to a fast movement of the defocused laser beam (d_S > 6 mm), a homogenous temperature distribution in the quasiline on the surface was achieved. Using this quasiline for polishing an initial roughness Sa in the range of 300–700 nm on flat fused silica samples was reduced to a Sa value below 15 nm.⁵ 

In addition to the lens polishing,⁶ further applications of laser polishing are presented by some recent studies. Cormont et al. use in Ref. 7 the laser polishing process for scratch removal on fused silica optics. According to Matthews et al., the laser damage threshold increases with CO₂ laser treatment of large optics.⁸ Choi et al. produced microlens arrays by polishing gratings.⁹ Further fundamental studies of different research groups show the increasing relevance of laser polishing process of glass.^10–15 However, the so far achieved roughness with laser polishing is not sufficient for imaging optics due to a roughness Sa > 1 nm for spatial wavelengths of λ > 100 μm. In addition, most researches were focused on fused silica. In this paper, recent results are presented for three types of optical glass (BK7, S-TIH6, and fused silica).

To achieve results with a laser-based process for fabricating high quality optics, a further process step for shape correction is necessary. One of the methods for shape correction is ion beam figuring (IBF).^16,17 As well as laser polishing, the IBF is a contactless processing. The removal rate is up to 0.03 mm³/s with a tool size (ion beam diameter) of 0.5 mm.¹⁶ However, the economic efficiency is significantly reduced by using another technology after laser polishing to finish optics with high precision and low roughness. In this paper, another approach to correct the lens shape is applied by using the same laser source that is used for laser polishing. By applying this method, laser beam locally irradiates the glass material. The absorption of laser energy leads to heating of a thin surface layer above the evaporation temperature, so that the material is ablated. According to some work of using laser radiation to ablate glass material, ablation depths in the range of several nanometers are reached.^18,19 Wlodarczyk et al. investigated in Ref. 20 a laser-based generation of holographic structures of glass and achieved ablation depths with single pulses (t_P = 10 μs) less than 10 nm. Heidrich et al. presented high precision laser ablation for shape correction by using either cw CO₂ laser or ultrashort pulse laser radiation.^21,22 However, both methods possess poor reproducibility and low spatial lateral resolution (>250 μm (Ref. 22)), which are not yet sufficient for shape correction.

CO₂ laser radiation with a wavelength of λ = 10.6 μm is absorbed by silicate glass up to 80%. The optical penetration depth δ_opt depends on the temperature of the absorbing material. At room temperature, the optical penetration depth is about 20 μm, while the depth decreases to approximately 4 μm at evaporation temperature.^23,24 Hence, silicate glass can be heated up in a thin surface layer.

The absorbed laser energy is transferred into heat in a thin surface layer of glass and results in reduction of viscosity. With the joint effect of surface melting and surface tension of a glass material, roughness of the surface is reduced and smoothing of the surface is achieved. Glass ablation occurs, once the evaporation temperature of the glass is reached with sufficient laser intensity.

Therefore, it is necessary for the laser polishing process to control the temperature of the glass surface precisely below evaporation temperature. In the current experiment, the temperature control is realized by using a pyrometer which measures the temperature of the radiated glass surface in situ. The laser power is controlled by comparing the measured temperature with a preset process temperature T_Process (known as set-point temperature) through an external proportional–integral–derivative (PID)-controller. Temperature fluctuations are controlled smaller than 20 K. The temperature is measured by a pyrometer made by DIAS Infrared (measuring frequency: 600 Hz) at the center of the quasiline for the PID-control (see Fig. 1).

The procedural principle of laser polishing is shown in Fig. 1. The defocused laser beam, controlled by a galvanometric scanner, moves with scan speed v_scan up to 10 m/s vertical across the surface. In addition, the laser scanner moves in the horizontal direction with feed speed v_feed in the range of 1–5 mm/s. Due to the fast movement of the laser beam compared to the feed speed, the scan tracks can be considered as quasiline. The interaction time is defined as

To prevent dust particles from deposition on the glass sample, the experimental setup is placed in a flow box. The remaining dust particles lead to surface defects while the surface layer is melted. With the flow box an ISO 5 purity level is achieved. The experimental setup is shown in Fig. 2. For laser polishing, a CO₂ laser is used with a maximum power of 1.5 kW. A ZnSe lens with f = 450 mm focuses the laser beam to a minimum diameter of d_s = 650 μm (1/e²) with a Gaussian intensity profile. To avoid cracks and reduce the thermal distortion, the glass samples can be preheated with heating plates or in a cleanroom oven up to 600 °C.

To quantify the roughness Sa of a laser polished surface, four measurements with different magnifications are made with a white light interferometer. By using a band pass filter, the roughness Sa is determined for discrete spatial wavelength intervals and plotted in a roughness diagram. The spatial wavelength intervals used for the analyzation in this paper are shown in Table I.

Figure 3 displays the procedural principle of laser beam figuring. Compared to laser polishing by remelting, for laser beam figuring modulated laser radiation is used. The laser beam is moved by a galvanometric scanner in an unidirectional scanning strategy across the surface. The pulse peak power P_P, the scan speed v_scan, the track pitch dy, and the frequency f_rep are kept constant during the processing. The pulse distance dx is determined by the frequency and the scan speed

For this process, a CO₂ laser named Mikrostorm™ by FEHA Lasertec²⁵ with a maximum output power of 200 W is used. The laser beam is focused with a ZnSn lens (f = 200 nm) resulting in a focus diameter of d_s = 500 μm (1/e²). A measurement of the laser intensity distribution is shown in Fig. 4. According to Ref. 26, the laser beam has a Gaussian shape with a steepness of the slope of n = 2.078. Laser pulses are generated by an acousto-optic modulator (AOM). Both the beam position and the pulse duration are set by a scanner software. A typical shape of a laser pulse and the AOM trigger pulse used for the investigations is shown in Fig. 4. The experimental setup for laser beam figuring is displayed in Fig. 5. An extraction system is used in order to remove the ablated glass material from the processing area. For the positioning of the sample a x,y,z-translation axis is used.

Experiments of laser beam figuring are conducted on flat fused silica samples. No preheating is used for processing. To determine the ablation depth of a specific parameter set, test fields were ablated with a dimension of 5 × 5 mm². The fields were measured by white light interferometry, and the ablation depth was determined by the mean height difference of the initial surface height and the ablated surface. The error bars in the diagrams show the standard deviation of the ablation depth in the area of one test field.

In previous investigations, the process temperature for laser polishing of fused silica was mentioned by approximately 50 °C below evaporation temperature²⁷ (T_process ≈ 2200 °C). However, the process temperature is strongly influenced by the varying viscosity characteristics of different types of glass. For example, the process temperature for fused silica is nearly below the evaporation temperature because of its high viscosity even at high temperatures. At the process temperature, the viscosity of fused silica is approximately 10⁴ Pas (working point of glass). Hence, a lower viscosity in the surface layer cannot be achieved without evaporation.

In Fig. 6, results of Richmann²⁷ are presented for fused silica in three states: initial state, laser polished, and conventionally polished with a form quality λ/20. In the diagram, the roughness Sa is plotted against the spatial wavelength λ. The Sa value of the initial state is 200–300 nm. The roughness of the laser polished sample shows a typical curve for laser polished glass surfaces in dependence of the spatial wavelength. After laser polishing or conventional polishing, the microroughness is both reduced to under 0.1 nm for spatial wavelength λ < 100 μm. The microroughness with λ < 100 μm implies small path length for the material relocation. As shown in Fig. 7, less melt flow is necessary to smooth the microroughness than for smoothing the waviness. For laser polishing, interaction time t_int < 7 s is sufficient for the relocating of the material in the peaks to the valleys. In order to reduce the roughness for spatial wavelength λ > 100 μm, it is necessary to increase the interaction time of the laser radiation and the material. However, it leads to an increase of shape deviation of the glass surface due to thermal bending as well. As shown in Fig. 6, the roughness for laser polished glass is higher than the conventionally polished for λ > 100 μm. In summary, microroughness is efficiently reduced by laser polishing while a waviness for λ > 100 μm remains on the fused glass surface (see Fig. 7). Figure 12 shows visually the white light interferometer image of the laser polished glass surface with residual waviness.

For laser polishing of fused silica typical parameters are feed speeds of 1–2.5 mm/s and a defocused laser beam diameter of 7 mm (1/e²). The area rate, defined as the maximum area per time which can be polished with the given laser source (1.5 kW CO₂ laser source), is 1 cm²/s.

In Fig. 8, the viscosity in dependence on the temperature is shown for fused silica, BK7, and S-TIH6. The curves are calculated with the Vogel–Fulcher–Tammann equation, and the viscosity data are from the data sheets of the materials. In comparison to fused silica, the viscosity of BK7 and S-TIH6 decreases to lower values before evaporation occurs during laser processing. Hence, laser polishing is possible for these types of glass with lower viscosity. But previous results for laser polishing of BK7 and S-TIH6 with a process temperature T_process = T_eva − 50 °C show an increase of the roughness Sa to above 1 μm in the spatial wavelength range of 100–4000 μm. Due to the low viscosity in the surface layer, the waviness is increased by an induced melt flow. The induced melt flow can be avoided at higher viscosity near the working point of glass (see marked area in Fig. 8).

For both BK7 and S-TIH6, the best results are achieved with the process temperature near the temperature of the working point. The roughness diagrams for BK7 and S-TIH6 are shown in Figs. 9 and 10. Both diagrams present the initial roughness (mechanically grinded) and the roughness of the laser polished (T_process = T(η ≈ 4 dPas) = 900 °C). The waviness after laser polishing (λ > 100 μm) is one order of magnitude higher than for fused silica. Figure 11 displays directly the white light interferometer images of the initial grinded S-TIH6 and the laser polished surface with different scales. A significant reduction due to the laser polishing of the roughness is observed. However, compared to the laser polished surface of fused silica (Fig. 12), an induced waviness can be observed and no residual waviness from the initial grinded surface.

Another difference in the thermal properties of the three types of glass is the thermal shock resistance. BK7 and S-TIH6 both have lower thermal shock resistance. Therefore, it is necessary to preheat the glass samples during the laser polishing process to avoid cracks. The preheating temperature T_preheat is 600 °C. The higher thermal shock resistance of fused silica allows laser polishing without preheating.

Typical process parameters for laser polishing of BK7 and S-TIH6 are feed speeds of v_feed = 5–8 mm/s and a defocused laser beam diameter of 10 mm (1/e²). The area rate is up to 5 cm²/s, which is 5 times higher compared to fused silica.

With laser polishing as a remelting process, the form accuracy of the surface geometry is not changed apart from thermal distortion. Hence, form errors cannot be corrected by laser polishing. As seen in the results of laser polishing of fused silica, roughness with spatial wavelength λ > 100 μm is not completely reduced by laser polishing. For manufacturing of imaging optics, a further process step is necessary for the final shape correction.

Besides the melting of a thin surface layer with laser radiation, material can be ablated as well. With a precise ablation, the shape of optics is able to be corrected and the residual waviness can be eliminated. In Fig. 12, a typical fused silica surface is shown after laser polishing. The required minimum resolution of laser beam figuring is determined by the peak height of the surface, which has to be ablated. As shown in Fig. 12, a lateral resolution <100 μm and a vertical resolution <5 nm should be achieved by laser beam figuring.

To ablate glass material selectively on the surface, the pulse duration is adjusted to the height of the surface. All other laser parameters are kept constant. Hence, at a surface position where no ablation is needed, the laser pulse is adjusted to the minimum pulse duration. With the minimum pulse duration no ablation occurs. An increase of the pulse duration leads to an increase of the ablation depth. As shown schematically in Fig. 13, with a pulse distance of dx = 20 μm, the pulse duration is adjusted to the height of the glass surface and hence a selective laser beam figuring is possible.

To study the influence of the pulse duration on ablation depth, test fields are ablated with a dimension of 5 × 5 mm² on flat fused silica samples. All test fields are processed with constant pulse peak power P_P = 50 W, and pulse distance is kept equal to track pitch dx = dy = 20 μm. The scan speed v_scan is 20 mm/s, which is determined by frequency (f_rep = 1000 Hz) and pulse distance. In Fig. 14, white light images of three ablated test fields are shown. For t_P = 36 μs, the ablation depth reaches 4 ± 1 nm. The ablation is in the magnitude of the initial roughness. To visualize the ablation depth without the influence of the initial roughness, in Fig. 15 white light images of the initial surface, of the ablated test field and the subtraction of the height of the two measurements are shown. With laser beam figuring, homogenous ablation depths are achieved.

In the two diagrams of Fig. 16, the ablation depths of the test fields are plotted against the pulse duration. Up to an ablation depth of approximately 15 nm, a nearly linear increase of the ablation depth is observed in dependence of the pulse duration. For z_ab > 15 nm, the ablation depth increases with a nearly exponential rise. The depression of glass surface by pulsed CO₂ laser radiation can be explained with two different physical mechanisms.

Due to the pulsed laser processing and the resulting rapid cooling, the glass structure in the surface is changed. The resulting change of the density leads to a densification of glass material and thus to a quasiablation.²⁹ This effect occurs already before reaching the evaporation temperature on the surface. The second mechanism is the evaporation of material. Hence, the nearly linear increase of the ablation depth is a densification. By exceeding the evaporation temperature, the depression is a combination of densification and ablation.

A requirement of the laser beam figuring is that the roughness should not be increased after the ablation process. Otherwise, the glass surface must be laser polished again. The iteration step could increase the processing time significantly. Figure 17 presents the roughness of the ablated test fields in relation to the ablation depth. The roughness in the test fields up to ablation depth of 25 nm (t_P = 50 μs) is equivalent to the initial roughness (mechanically polished) and is not induced by the ablation. An increase of the roughness is observed for ablation depths > 25 nm.

In order to reduce the waviness with laser beam figuring, a precise selective ablation of the surface is necessary. Hereby, the surface profile is first measured with a white-light interferometer. With the data of the measurement, a scanner script is generated. Afterward the sample is processed by the CO₂ laser through laser beam figuring. Notably, the sample should be placed precisely under the laser scanner, so that the peaks are precisely ablated. In this study, the positioning of the laser beam is realized by using a software, which is developed by ILT. With this software, scripts are generated for controlling the beam position (galvanometric scanner) and the pulse duration (AOM). Input data for the software are a surface measurement for the spatial coordinates and the characteristic ablation curve to set the pulse duration on each position depending on the surface height. A flowchart of the software is displayed in Fig. 18.

With an area rate of 0.4 mm²/s, the surface shape is corrected by laser beam figuring. An exemplary result for laser beam figuring on a mechanical polished surface is presented in Fig. 19. The initial roughness of the test surface is Sa = 4.4 nm and rms = 5.2 nm. After laser beam figuring, the roughness is reduced to Sa = 1.5 nm and rms = 1.9 nm. It is observed in the white-light interferometer images that the microroughness is not increased. These results show the feasibility of the selective ablation. Ablation depth below 25 nm is achievable by modulating the laser pulses.

With laser polishing, the time for polishing of optics can be significantly reduced. The study presents results of laser polishing for three types of glass. For fused silica, an area rate up to 1 cm²/s, for BK7 and S-TIH6 up to 5 cm²/s can be achieved. It is concluded that the process temperature and the preheating temperature depend on the thermal properties of different types of glass. A process temperature near the evaporation temperature for laser polishing of BK7 and S-TIH6 leads due to a viscosity below 4 dPas to an induced waviness. An induced waviness can be avoided by polishing near the working point of glass (η ≈ 4 dPas). For both types of glass, a preheating temperature of 600 °C is necessary to avoid cracks due to the low thermal shock resistance.

For fused silica, the reduction of the waviness is limited through laser polishing because of its high viscosity even at evaporation temperature. As a result, a residual waviness with lateral dimension >100 μm and vertical dimension <50 nm remains on the laser polished surface. In order to achieve not only small roughness but also high shape precision laser beam figuring is presented for a precise selective ablation. By using laser beam figuring depressions less than 5 nm are reached. A precise shape correction is achieved through selective laser ablation, in which only pulse duration is modulated.

So far, laser beam figuring is conducted on mechanically polished glass. In further investigations, tests with laser beam figuring will be applied on laser polished surfaces.

This work has been partially funded by the Federal Ministry of Education and Research (BMBF). The authors would like to thank the Federal Ministry of Education and Research for their generous sponsoring of the research project: “RapidOptics” (Funding Reference: 13N13294).