A deterministic surface correction technique has been used to improve the surface figure of two fused silica optical flats over a diameter of 60 mm with no measurable degradation in their surface quality at spatial frequencies of ≤750 mm-1. The surface corrections were achieved by selective ion beam sputtering (IBS) deposition of an index-matched dielectric layer through a multi-aperture mask. Two flats were corrected, one finished on a pitch lap, the other on a Teflon lap to give two distinctly different surface roughness characteristics. The microroughnesses measured on a TOPO-WYKO profilometer were 3.0 Å and 7.2 Å respectively. Both optics were improved to better than λ/100 peak-to-valley and in each case the surface correction process preserved or potentially improved the microroughness of the optic.
© 2008 Optical Society of America
The requirements on ultra-precision optical surfaces, and the ability to produce them, are continually evolving as more exacting operational specifications are demanded by users. For example it is now common at CSIRO’s Australian Centre for Precision Optics (ACPO)  to receive requests for optics with surface roughness of less than 1 Å rms and overall figure accuracy of better than 10 nm. While achieving either Ångström level surface roughness or surface figures of <10 nm peak-to-valley is possible independently, it can be difficult to achieve both specifications simultaneously due to the differing requirements of each surface finishing process.
Recent developments in sub-aperture finishing such as magneto-rheological finishing (MRF) [2,3], and precessions polishing  can achieve quite remarkable overall surface figures, however, the results available to date for the resultant surface roughness indicate that this parameter is not well maintained by these processes and the rms surface roughness is typically around 5 Å.
In this work we investigate the effect on surface roughness of a recently reported selective surface correction technique [5,6]. The technique uses a multi-aperture mask to deposit material selectively in an ion beam sputtering (IBS) machine. The technique has been used to correct two fused silica optical flats of 75 mm diameter, prepared in ACPO’s optical workshop, over their operating aperture of 60 mm. The first flat was prepared using a Teflon lap, which is known to produce flats of very accurate surface figure; and the second was produced using a pitch lap to give a higher quality, and distinctly different, surface roughness . Both flats were then corrected a total of three times using the multi-aperture technique and the surface figure and roughness were measured at each stage to detect and track any changes in the surface roughness, and to determine the quality of the correction.
2. Measurement of the optical flats
A key aspect of any surface correction technique is the metrology used to determine the initial and subsequent surface quality of the optic. This section will provide details of the measurement techniques used.
2.1 Figure measurements
Prior to correction we measured all substrates in ACPO’s ultrahigh-precision, calibrated Large Aperture Digital Interferometer (LADI) . The diameter of the flats is 75 mm, whereas the calibrated aperture is 320 mm wide. Therefore, to enhance the accuracy of the measurement, five measurements were taken in different locations under the aperture, roughly in a cross formation as in Fig. 1, and the best three of them (in terms of noise and consistency) were averaged and used to generate the surface correction masks. This averages out some residual instrument artifacts due to small-scale spatial patterns, and the faint streaks generated by speckle averaging through the rotating diffuser disk within the LADI.
To enhance the measurement accuracy further, we used an error-correcting seven-sample phase-shifting formula  with start-phase averaging , which suppresses all higher-order nonlinearities in the intensity modulation during phase shifting except for multiples of seven. Non-linearities of such high order are usually of very low amplitude and therefore negligible. If we keep the fringes fluffed out for measurement, our measurements improve as the substrates become flatter, since we then measure a field of very similar phase values within a fraction of a fringe, with the concomitant decrease in periodic phase-calculation errors.
Since we are aiming for λ/100 accuracy, it is necessary to measure the substrates with at least that accuracy, and the LADI is capable of exceeding λ/100 accuracy over the full aperture, and commensurately better over smaller sections. The limit to accuracy is not given by large-scale figure uncertainties in this case, but by small-scale noise and structure. Fig. 2 provides an example for the attainable uncertainty between sub-aperture measurements from different locations in the aperture. The maximum rms deviation between any pair of these maps over the 65 mm aperture used to generate the masks was 0.38nm.
We repeated measurements in this manner after each correction step.
2.2 Roughness measurements
The surface roughness characteristics of the optics were measured using a TOPO-WYKO profilometer with a measurement window of 117×154 µm2. The measurements were repeated 5 times at random locations on the substrates and averaged. This removes the random influence from the substrate surface and creates a good approximation of the residual instrument errors. By subtracting this average from an individual measurement, we get a reading for the average rms roughness of a given surface area. Another way to measure roughness is to calculate a difference map from two successive measurements and to divide the resulting rms roughness by 20.5 . Both methods give the same result within less than 0.5 Å, even within the small data sets that we have used here. Besides the rms figures, the visual appearance of the surfaces gives a very good indication of the different surface finishing techniques used and the spatial scale of the roughness, as shown in Fig. 3.
The difference in smoothness can be seen very clearly, and emphasizes why pitch polishing is preferred for surface finish. Our task in this study is to maintain the smoothness of the pitch-polished substrate while improving its figure error to values matching (or exceeding) those typically associated with Teflon polishing.
The first attempts at correcting the flats were not successful due to an accidental misalignment of the correction mask to the substrate. The masks were flipped about the vertical axis, which can be discerned from the two symmetric blemishes seen in Fig’s. 4d & 4e shown below. The correction process copied an original indented blemish at approximately 2 o’clock on the Teflon polished optic and imprinted it as an inverted hump at approximately 10 o’clock. This processing error significantly degraded the overall figure of the Teflon polished optic due to a left-right asymmetry of its initial figure but luckily did not greatly affect the pitch polished optic as this optic had a better initial left-right symmetry.
After this alignment issue was rectified, the optical correction process for each flat was carried out in two further stages. The first stage was designed to correct approximately 30% of the error function and the second stage corrected the residual. This allowed the deposition rate of the IBS system through the multi-aperture mask to be confirmed. The depositions were carried out in a Veeco Spector Ion Beam Sputtering machine using a silica target and a deposition rate of approximately 0.4 nm/s at the centre of the plume before occlusion by the mask. Details of the deposition process, including details of mask and substrate position and substrate motion can be found in References 5 and 6.
The evolution of the surface figures of the optics due to the selective depositions is shown in Fig. 4. In each case the first image is of the optic after the first abortive run. The same z-scale has been used for the images of each respective optic to emphasize the improvement in overall flatness seen after each subsequent deposition. Images from the TOPO profilometer showing the surface roughness after the first abortive deposition and the final stage of correction are given in Fig. 5.
The surface flatness and surface roughness values for the pitch and Teflon lap finished optics at each stage of the correction over their working apertures of 60 mm are given in Tables 1 and 2 respectively. Note that the initial values of surface roughness are for the substrates before the abortive first deposition was carried out.
The difference in starting surface roughness of each optic is immediately obvious with the pitch lap surface having the better characteristic values.
As mentioned above, the initial degradation in the surface figure of the Teflon finished optic was due to an accidental misalignment of the mask to the substrate. Conversely, after the first deposition, both surfaces roughness values appear markedly improved. The reason for this is as yet unclear; however since the change is evident in both substrates, and both measurements constitute averages, it is likely that the detected change is real.
After the second deposition, the roughness deteriorates slightly, but again it does so in both our data sets. Our body of data is too small to conclude with certainty that the roughness of the substrates has improved; however, it seems safe to say that the difference in surface qualities is at least maintained between the pitch-polished and the Teflon-polished substrate, and that the smooth surface of the pitch-polished flat has maintained its characteristics through the deposition process.
In both instances, the overall surface figure of the optics were improved to significantly better than λ/100 (measured at 633 nm) over the working diameter of 60 mm.
A surface correction technique previously developed by the authors, and put into realization here for the first time on reflective optics, allows the fabrication of precision-polished small to medium sized optics with surface figures of λ/100 peak-to-valley and better. Since the surface correction does not adversely affect the surface roughness of the optical surface this technique now allows us to fabricate ultra-precision optics in more deterministic fashion. An initial high quality precision-polished optical surface is fabricated with very low surface roughness, but moderate figure error. The surface can then be corrected in a deterministic fashion using the selective deposition process to give the required overall surface figure. This results in a significant reduction in the iterative process of manual polishing and precision metrology that was hitherto associated with “impossible” combinations of surface figure and smoothness.
The authors would like to thank Katie Green for many useful discussions during the preparation of this manuscript. The precision-polished substrates used for this work were fabricated by Jeff Seckold and Edita Puhanic, from the Australian Centre for Precision Optics, at CSIRO Materials Science and Engineering.
References and Links
2. H. B. Cheng, Z. J. Feng, and Y. W. Wang, “Surface roughness and material-removal rate with magnetorheological finishing without subsurface damage of the surface, “J. Opt. Technol. 72, 865–871 (2005), http://www.opticsinfobase.org/abstract.cfm?URI=JOT-72-11-865. [CrossRef]
3. F. H. Zhang, G. W Kang, and Z. J. Qiu, “Surface Roughness of Optical Glass Under Magnetorheological Finishing,” Key Eng. Mat. 259–260, 662–666 (2004). [CrossRef]
4. D. Walker, D. Brooks, A. King, R. Freeman, R. Morton, G. McCavana, and S. -W. Kim, “The ‘Precessions’ tooling for polishing and figuring flat, spherical and aspheric surfaces,”Opt. Express 11, 958–964 (2003), http://www.opticsinfobase.org/abstract.cfm?URI=oe-11-8-958. [CrossRef] [PubMed]
5. J. Arkwright, I. Underhill, N. Pereira, and M. Gross, “Deterministic control of thin film thickness in physical vapor deposition systems using a multi-aperture mask,” Opt. Express 13, 2731–2741 (2005), http://www.opticsinfobase.org/abstract.cfm?URI=oe-13-7-2731. [CrossRef] [PubMed]
7. A. J. Leistner, E. G. Thwaite, F. Lesha, and J. M. Bennett, “Polishing study using Teflon and pitch laps to produce flat and supersmooth surfaces,” Appl. Opt. 31, 1472–1482 (1992). [CrossRef] [PubMed]
8. P. S. Fairman, B. K. Ward, B. F. Oreb, D. I. Farrant, Y. Gilliand, C. H. Freund, A. J. Leistner, J. A. Seckold, and C. J. Walsh, “300-mm-aperture phase-shifting Fizeau interferometer,” Opt. Eng. 38.8, 1371–1380 (1999). [CrossRef]
9. K. G. Larkin and B. F. Oreb, “New seven-sample symmetrical phase-shifting algorithm,” Proc. SPIE 1755, 2–11 (1993). [CrossRef]
10. P. Hariharan, “Phase-shifting interferometry: minimization of systematic errors,” Opt. Eng. 39, 967–969 (2000). [CrossRef]