Finishing quartz glass surfaces with laser radiation: analyses of the parameters for process optimisation/automation.

Hecht, Kerstin ; Bliedtner, Jens ; Mueller, Hartmut 等

1. INTRODUCTION

From state of the art methods of polishing surfaces with laser radiation are already known. So it's possible e.g. to reduce processing time of metallic injection moulding moulds from 30min/[cm.sup.2] to a few seconds per [cm.sup.2] (ILT, 2008). Also in polishing cast materials good surface quality is reached (ILT, 2003). The laser material processing of glass has made good progress whereas the high-precision finish (for optical parts) causes still problems. It's barely possible to reach good surface quality without creating thermal tensions (LZH, 2005). To finish the quartz glass parts in this present report a 1.5kW C[O.sub.2]-laser is used. The polishing is carried out by a regional short melting of a thin surface layer with a defocused beam. Using adjusted laser power stock removal is prevented and the smoothing happens just because the surface tension of the melting layer. This tension is responsible that the profile peaks were levelled and the valleys were filled. The analysis should also show, how the starting roughness and the changeable laser parameters influence the final surface quality. It's interesting too how the parameters have to be changed to reach requested roughness values.

The Optimisation of the process parameters should be carried out by measure temperature and surface quality. This quality is detected by the help of a stylus instrument.

2. EXPERIMENTS

Fig. 1 shows the experimental setup. It's a portal system where the laser beam is coupled in and leaded over different mirrors to a scanner system. In this scanner mirrors move the beam in lines with a maximum speed of [v.sub.s] = 3m/s. Because of this high velocity a polishing line is generated on the glass part (part size: 20 x 20 x ~3[mm.sup.3]). The feed rate [v.sub.f] is realized with the axis of the portal system and is diversified in different (polishing) sections ([s.sub.1], [s.sub.2], [s.sub.3]). The pyrometer measuring spot for the temperature recording is carried on behind the laser line. The measuring system detects the surface temperature (T) and the developing T-curve is used to optimise the process parameters. In Fig. 4 two T-curves (continuous lines) are visible.

[FIGURE 1 OMITTED]

The task is it first to adjust the parameters feed rate and length of the polishing sections in a way, that T raises steep ([s.sub.1]), stays constant during the finishing process ([s.sub.2]) and has just a small peak at the end of the part ([s.sub.3]). The steep rise guarantees that enough energy is inserted to have a good finishing result from the beginning. To reach uniform polishing results, constant T along the whole surface is deciding. To prevent unnecessary additive tensions in the material and rounding the rear edge of the part a heat accumulation has to be avoided. As you can see regarding the curve V034_C28 (Fig. 4) it's possible to improve the unfavourable temperature distribution from trial V002_C03 by optimise [v.sub.f] in the 3 sections.

Beside temperature the surface quality is a criterion for process optimisation. To check influence of starting roughness ([Ra.sub.1]) (before finishing) the parts were divided in three experimental groups (after defined roughness groups).

experimental group group A group B group C
maximum Ra [[micro]m] 1,6 0,8 0,4
roughness group N7 N6 N5

The optimisation of the process parameters is carried out in every group and during the experiments the following parameters were changed:

* feed rate [v.sub.f] in 3 sections (already discussed)

* laser output power P, 400 ... 700W

* beam velocity [v.sub.s], 400 ... 1000 mm/s

3. RESULTS

After more than 120 tests the influences of the machining parameters and [Ra.sub.1] could be studied in detail. Altogether it turned out that the reachable surface quality depends on [Ra.sub.1]. This connexion gets very obvious regarding group A. It's actually observed that it isn't possible to polish the parts of this group without stock removal. Therefore these parts of group A were no longer considered in this paper. The influence of [Ra.sub.1] changes also with variation of the other parameters. Regarding the dependency of laser output power (P) on Ra (Fig. 2) there is no connexion between [Ra.sub.1] and final roughness ([Ra.sub.2]) visible at all. Observable is that with increasing P the surface is smoothed more. In the areas of hight P there is already stock removal.

[FIGURE 2 OMITTED]

The laser beam velocity (vs) has also an effect on reachable surface roughness. The influence of [Ra.sub.1] stays partly important while diversifying vs. Considering Fig. 2 it's obvious that Radistribution changes only frictional. With higher [v.sub.s] less energy is inserted into the glass part and so the surface smoothing abates. However it seems that (in the chosen limits) [v.sub.s] has no such significant influence on the [Ra.sub.2] as P for example.

Over all experiments T-curves of every polished part were recorded. They have to be optimised in each group by changing [v.sub.f] in the three sections. Afterwards their appearance doesn't change much. Just the relation of P and vs and the maximum temperature ([T.sub.max]) seems still interesting. Due to the pyrometric measurement the measured T-values don't reflect real T in the interaction zone on the glass surface. It's not possible to measure these temperatures contact-free so far. It's assumed that T in the interaction zone is about 2230[degrees]C--the vaporisation point of quartz. That's most likely because there is always a light steam of sublimate while polishing. The T-measure-point is selected in a way that it is located always short behind the polishing line. The thickness of the parts is not constant for each group that's why the pyrometer has to be adjusted time and time again. From this it fallows that the distance between measuring spot and interaction zone changes slightly. But even the smallest change of this distance leads to variations in the measured T-values. That's why the position of the curves is not significant but their appearance however is. Fig. 3 shows the curves of [T.sub.max] in relation to P (left) and vs (right). It's obvious that with higher P the temperature rises. Against that higher [v.sub.s] causes lower [T.sub.max]. The degree of surface smoothing ([Ra.sub.1]/[Ra.sub.2]) follows the T-profile so far. It turned out that the measured T-value can be used for process automation. For this automation the aim is to observe T continuously (online) and adjust parameters accordingly to the T-profile.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

4. PROCESS AUTOMATION AND CONCLUSION

Regarding results from experiments laser power is interesting for automation. It has the most significant influence on surface-T and quality. So a laser control system was installed to control P. In the scheme in Fig. 1 you can see aditionally the connection of the devices with the control unit (CU). It's not necessary to describe all technical details of this CU because it isn't part of the investigation. It was constructed for different control tasks and this is just one example of using it.

Return to the present study it's important to know, that the CU works on basis of temperature measurement data were T symbolise corresponding P-values. The pyrometer is connected with the CU and this has a connection to the laser system. You can determine a temperature (/power) control routine. In this case you teach times and temperatures to the CU and P would be adapted accordingly. So there is a permanent comparison between measured T and selected. (Kamann, 2008)

To apply the CU there are some problems to solve. Remember the measuring point of the pyrometer adjusted short behind the polishing line. This "line" is created because the laser needs 50ms to pass the surface. The pyrometer detects the surface temperature with a sampling rate of 250ms. So the laser spot is sometimes nearer and sometimes farther the measuring spot which leads to different measured values. The CU would adapt P permanently and you'll get no constant surface temperature. It's necessary to average the measured values according to the FIFO-method with the CU too. The system works with an average value out of 10 measured values. Regarding Fig. 4 (discontinuous line) it's visible that it's possible to realise the whished T-distribution without different parameters or polishing sections by using the CU. This picture shows another problem still to solve. In the beginning it last too long until the CU and/or the pyrometer recognise the developing surface-T. It even rises out of range and there is a lot stock removal. On the other hand the CU is able to keep T within a limit of 25K.

To solve the problem of overheat maybe different measuring spot positions could be tested. Lay the spot into the polishing line to reduce the influence of heat conduction is one possibility. But altogether further investigations have to fallow. In addition it would be also necessary to develop and apply a construction to hold and adjust the pyrometer high precisely.

5. REFERENCES

Fraunhofer Institut fur Lasertechnik-ILT; (2003). Fraunhofer ILT Annual Report 2003, Available from: http://publica. fraunhofer.de/eprints/N-35932.pdf, Accessed: 2007-08-06

Fraunhofer-ILT (2008). Laser-Beam Polishing of Injection Molded Tools, Available from: http://www.ilt.fraunhofer.de/ eng/100607.html, Accessed: 2008-05-06

Kamann, J. Entwicklung und Aufbau eines Laserleistungsregelsystems fur temperaturgefuhrte Materialbearbeitungsprozesse, 2008, diploma thesis

LZH e.V. (2005) Innovative manufacturing processes for polishing glass surfaces, Final Report, Available from: http://www.laser-zentrum-hannover.de/en/projects/index.php Accessed: 2007-08-06