OSA's Digital Library

Optics Express

Optics Express

  • Editor: Michael Duncan
  • Vol. 13, Iss. 7 — Apr. 4, 2005
  • pp: 2731–2741
« Show journal navigation

Deterministic control of thin film thickness in physical vapor deposition systems using a multi-aperture mask

John Arkwright, Ian Underhill, Nathan Pereira, and Mark Gross  »View Author Affiliations


Optics Express, Vol. 13, Issue 7, pp. 2731-2741 (2005)
http://dx.doi.org/10.1364/OPEX.13.002731


View Full Text Article

Acrobat PDF (231 KB)





Browse Journals / Lookup Meetings

Browse by Journal and Year


   


Lookup Conference Papers

Close Browse Journals / Lookup Meetings

Article Tools

Share
Citations

Abstract

A technique for controlling the thickness profile of a thin film in physical vapor deposition systems is reported. The technique uses a novel mask design with apertures of varying dimension to selectively deposit the required film thickness at predetermined locations across the aperture of the substrate. The technique has been used to correct the thickness uniformity of a 55 mm diameter, 280 µm thick, lithium niobate wafer to less than 0.5 nm rms, and also to improve the uniformity of deposited films in an Ion Beam Sputtering system to better than 0.5% over a 50 mm aperture.

© 2005 Optical Society of America

1. Introduction

Physical vapor deposition of thin films is used in the fabrication of a wide range of components in fields such as optics, microelectronics, and magnetic storage media. For example, in optics, components such as dielectric mirrors, anti-reflection coatings, and spectral filters for telecommunications and photonic applications are all fabricated from multi-layer thin film coatings. These components require precision control of thickness and uniformity of dielectric thin films. In addition, selective deposition of dielectrics is also used to correct the surface of optical components after polishing. For these and other applications the spatial variation of thickness of the deposited layer must be controlled to a high degree.

There are a number of existing techniques routinely employed for controlling the spatial thickness variation of deposition in physical vapor systems. Techniques for improving the uniformity include rotary, planetary, or randomised motion of the substrate through the beam during deposition; sometimes including passage across a stationary uniformity shutter; see for example Section 9.5 of Reference 1. Spatial variation in thickness has also been demonstrated by using slots of varying width to selectively aperture the beam during deposition2 and also by raster scanning the substrate behind a small aperture to build up a spatially varying layer3. Both techniques have been used to improve the quality of precision optics.

This paper describes a technique for achieving deterministic control of the spatial thickness variation of deposited thin films using a multi-aperture mask. The technique has been used to improve the thickness uniformity of a 76 mm lithium niobate wafer, reducing the peak-to-valley thickness variation from 30 nm to approximately 2.4 nm (0.45 nm rms) over a working aperture of 55 mm; and also to improve the inherent thickness uniformity of deposition in an Ion Beam Sputtering (IBS) system from around 3% to less than 0.5% of the total thickness over a 50 mm diameter aperture.

This aperture mask technique offers similar or improved uniformity control and a greatly enhanced spatial correction capability compared to previously reported techniques. It also has the advantage of requiring little or no motion of the beam, mask or substrate and does not greatly extend the time taken to deposit an uncorrected film of a similar thickness.

2. Deposition control

2.1 Aperture widths and thickness uniformity

To control the spatial thickness variation of the deposited film, a mask containing a number of apertures of different widths was positioned between the deposition source and the substrate. The width of each aperture was determined by both the local flux of the deposition beam and also the thickness of film required at that point. The number and density of apertures determined the spatial resolution of the thickness correction being sought — typically a spatial resolution of the order of 1×2 mm was used which requires around 2000 apertures to cover a mask diameter of 65 mm. The mask was designed using a ‘checker board’ pattern in which alternate unit cells of the pattern contained an aperture of specific width. This allowed the highest density of apertures to be used while maintaining the mechanical integrity of the mask. To achieve the 1×2 spatial resolution, each unit cell was 1×1 mm, and the width of the aperture was determined from the average film thickness required over two adjacent unit cells. Figure 1(a) shows a schematic of a typical mask, in this instance designed to deposit a film with a ramped variation in thickness from left to right, as discussed below. During deposition the substrate was dithered in a linear fashion in the direction indicated by the arrow in Fig. 1(b) and by an amplitude equal to the pitch of the apertures along a given row.

Fig. 1. (a) Schematic of a 60×60 mm aperture mask used to apply a ramped thickness profile to a deposited thin film. (b) Detail of the aperture distribution showing the direction and amplitude of applied dither.

Figure 2 shows the location of the mask in the IBS machine during deposition. In this instance, the source to substrate distance was approximately 30 mm and the mask to substrate distance was 9 mm, although the mask location was found to be not very critical in this particular machine due to the large sputtering target used for the deposition (The Ta2O5 was sputtered from a 356 mm diameter tantalum target using a 1200eV Argon ion beam from a 160 mm ion source resulting in an effective spot size of approximately 254 mm wide by 180 mm high). A further advantage of the large size of the source and the offset between the mask and substrate is that the hard edges of the apertures were blurred out in a manner analogous to a pin-hole camera. This effect is shown schematically in Fig. 3. The blurring of the aperture edges and the dither motion of the substrate ensured that the pattern of the mask apertures was smoothed out and no ‘print-through’ of the mask design onto the substrate surface occurred during deposition (this point is further discussed in Section 2.2). Apart from adjusting the amplitude of the dither to match the pitch of the apertures, it was not found to be necessary to optimise the deposition parameters in any specific manner in order to achieve the reported results.

Fig. 2. Location of the aperture mask with respect to the deposition source and substrate in the IBS machine.
Fig. 3. Effect of the finite sized target and offset between aperture mask and substrate.

To calibrate the effect of aperture width, a series of trial depositions were carried out using Ta2O5 on a 50 mm glass substrate using:

1. no mask

2. a uniform mask with apertures of equal width

3. a ramped mask whose aperture widths varied linearly from one side of the mask to the other (as shown in Fig. 1).

These masks were laser cut from 0.002 inch (~50 µm) thick stainless steel shim and had unit cells dimensions of approximately 1 mm - this dimension had no specific importance; it was simply used to accommodate the throw of the first translation stage used to test the technique. The laser cutting process was externally sourced and had a quoted accuracy of +/- 20 µm. The ramped mask, and the masks described in later Sections, had a minimum aperture width of 20% of the unit cell (~200 µm) in order to prevent effects from excessive depth/width aspect ratios.

The deposited films were measured using a Cary Spectrophotometer with a spatial resolution of ~5 mm and an estimated error of +/-0.5% of the total thickness. The deposition time for each trial was varied to give a film thickness of approximately 600 nm in each case; and the calculated deposition rate was used to compare the results. The calculated rates are shown in Fig. 4. The important features of these results are:

1. The inherent deposition rate varied by approximately 3% from centre to edge of the substrate and had an approximately cosinusoidal variation

2. The uniform mask gave a uniform reduction in deposition rate, ie the reduction in thickness was independent of position in the beam

3. Once the inherent uniformity was factored out, the ramped mask gave a linear variation in rate, ie the deposition rate varied directly with aperture width with no degradation of deposition rate due to reduction in aperture width.

Fig. 4. Results of the deposition rate trials carried out using a) no mask, b) a uniform mask, and c) a ramped mask.

Following these trials, a mask was fabricated to correct for the inherent thickness variation of the deposited film. In this instance the aperture widths varied inversely with the thickness variation. Again, a trial deposition was carried out using this mask, and the results with and without the correction mask are plotted in Fig. 5 as a percentage variation in the deposited thickness. Using the correction mask improved the thickness variation from 3% to 0.5% of the total deposited thickness. It is worth noting that this was a one-off result using a very basic algorithm that set the aperture width simply based on the measured rate at a given point on the deposition plane. No iterative algorithm or repetitive measurements were used to optimise the design of the mask.

No effects were observed due to filling of the apertures during these initial experiments or for the corrective processes discussed in Section 2.2; however, if this technique is to be used for long term improvements in deposition uniformity then it will be necessary to consider the effects of filling of the apertures during deposition. A possible solution will be to fabricate more than one mask and to acid etch the deposited species on alternate masks after each deposition run. This will be the subject of future investigations.

Fig. 5. Measured thickness profiles of the deposited films; a) along the horizontal axis; and b) along the vertical axis.

2.2 Thickness profiling

The technique was then used to add a specific profile to the deposited film. The principle motivation for this aspect of the project was to improve the thickness uniformity of a large aperture lithium niobate Fabry-Perot etalon. These devices are becoming increasingly popular for solar observation and have an extremely high requirement for thickness uniformity across the etalon aperture4. The etalons are typically fabricated from a Z- or Y-cut lithium niobate wafer and are prepared using pitch and Teflon laps to a typical physical thickness tolerance of a few 10’s of nm5, 6. The optical thickness variation of the wafers was measured by silvering each side of the wafer to form a high finesse etalon. The transmission of a high finesse etalon as a function of angle of incidence becomes very sharply peaked for reflectivities above ~95%, and provides a simple technique for determining the thickness variation across the aperture. To measure the thickness variation the silvered wafer was rotated in front of a collimated beam from a stabilised Helium Neon laser and the transmitted beam was imaged onto on a ccd array. By noting the angle at which maximum transmission was achieved for each pixel, the thickness variation was calculated for all points across the aperture. The magnification used to image the transmitted beam and the pixel count of the ccd array gave a spatial resolution of 0.6 mm and the sharply peaked transmission response allowed sub-nanometre resolution in the measured thickness variation. Further details of this measurement technique will be the subject of a future publication7. After the measurements have been completed the silvered surfaces were removed by soaking in nitric acid for a few seconds so that the corrective layers to be applied. Figure 6 shows the measured thickness variation of a, 280 µm thick lithium niobate wafer being processed for use as a Fabry-Perot etalon. The peak-to-valley variation in thickness was approximately 30 nm over the working aperture of 55 mm diameter. Although this is close to the best uniformity attainable from traditional lap polishing of this material6 it still falls short of the current requirements for solar observation, hence the need for post-polishing correction. Using the measured thickness variation map, a suitable correction mask was fabricated by matching the aperture widths to the amount of additional material needed at each location to flatten the wafer. A schematic of the correction mask used to correct this wafer is shown in Fig. 7. Note that the aperture widths vary inversely with the measured thickness of the wafer at each equivalent location, and also that the overall diameter of the mask extends beyond the intended working aperture of the etalon to avoid any edge effects on the deposited surface.

Fig. 6. Measured thickness variation of a 55 mm diameter 280 µm thick lithium niobate wafer before correction.
Fig. 7. Schematic of the aperture mask used to correct the wafer shown in Fig. 6.

As for the uniformity improvement, the mask was positioned between the source and substrate, and the substrate was dithered during deposition to smooth out the deposited layer. The checker-board nature of the mask reduced the maximum deposition rate of the IBS machine by a factor of approximately 2; however, the total deposition time for the correction was still less than 400 seconds.

Figure 8 shows the measured thickness variation of the same wafer after correction. The Z-scale used in this figure is the same as that shown in Fig. 6 to allow direct comparison of the thickness variations in each case. Clearly, the initial thickness variation has been largely removed to give a peak-to-valley variation of 2.4 nm (0.45 nm rms). Figure 9(a) and 9(b) show horizontal and vertical thickness profiles after correction, measured across the centre of the wafer.

Fig. 8. Measured thickness variation of the wafer shown in Fig. 6 after correction.
Fig. 9. Horizontal and Vertical thickness variation profiles of the wafer shown in Fig. 8.

The dappled effect seen on the surface in Fig. 8 is not fully understood at this time, but has been shown to be present before the corrective layer was applied and so cannot be due to print though of the mask design during deposition. To illustrate this, the large scale background variation of the surface before correction was removed from the measured data by subtracting a 2D polynomial fit to the surface. Figure 10 shows the resulting thickness map and again, the dappling is clear. Careful inspection of Figs. 8 and 10 will show that the features remain unchanged before and after correction.

Fig. 10. Thickness variation of the wafer in Fig. 6 after the large scale variations have been removed using a 2D polynomial fit to the surface

Since the possibility of ‘print through’ is an obvious concern with this technique, as mentioned in Section 2.1, particular effort was put in to measuring the thickness profile before and after correction. Figure 11 shows the measured thickness across the centre of the wafer before and after correction (the residual large scale variation has again been removed by subtracting a polynomial fit to the data). The spatial resolution of these data is approximately 0.6 mm and the thickness resolution is less than 1 nm. No evidence of print through was seen which indicates that any effect is at least within the thickness resolution of the measurement, or less than 1.3% of the total film thickness (Note that the total deposited thickness over the full mask diameter was approximately 80 nm).

Fig. 11. Measured horizontal thickness variation across the centre of the wafer shown in Fig. 8: a) before and b) after correction (with large scale variations removed).

3. Conclusions

A technique has been demonstrated for controlling the spatial variation in thickness of a deposited thin film. This technique has proved to be highly effective for deterministic correction of precision polished substrates and is also a promising technique for improving the uniformity of deposited thin films. In addition, this technique offers the advantage that the improved uniformity and surface corrections only require a linear dither motion of the order of 1 to 2 mm to be applied to either the substrate or mask during deposition, and the total deposition time is only changed by approximately a factor of 2 compared to the maximum deposition rate of the chamber.

The technique has been used successfully to improve the uniformity of deposition in an IBS system to better than 0.5%; and also to correct the measured thickness profile of a 55 mm aperture lithium niobate wafer to less than 0.5 nm rms.

Acknowledgments

The authors would like to thank Roger Netterfield, David Farrant, Jeff Seckold and Svetlana Dligatch for useful discussions and assistance during the project, and also Wayne Stuart and Edita Puhanic for preparing the lithium niobate substrates used.

References and links

1.

P. W. Baumeister, Optical Coating Technology (SPIE Press, Washington, 2004), Chapter 9 and references therein. [CrossRef]

2.

J. R. Kurdock and R. R. Austin, “Correction of Optical Elements by the Addition of Evaporated Films,” Physics of Thin Films, Academic Press, (1978).

3.

J. M. Mackowski, L. Pinard, L. Dognin, P. Ganau, B. Lagrange, C. Michel, and M. Morgue, “Different approaches to improve the wavefront of low-loss mirrors used in the Virgo gravitational wave antenna,” Applied Surface Science86–90 (1999). [CrossRef]

4.

C. D. Prasad, S. K. Mathew, A. Bhatnagar, and A Ambastha, “Solar photospheric and chromospheric observations using a lithium niobate Fabry-Perot etalon,” Experimental Astronomy125–133, (1998). [CrossRef]

5.

A. J. Leistner, “Teflon polishers: their manufacture and use,” Applied Optics293, (1976).

6.

J. A. Seckold, “Precision flat polishing of lithium niobate,” Optical Fabrication and Testing OSA Technical Digest SeriesOptical Society of America, Washington, D. C. (1996).

7.

P. S. Fairman, D. I. Farrant, and J. W. Arkwright, CSIRO Industrial Physics are preparing a manuscript to be called “High resolution metrology of high-finesse Fabry Perot structures”

OCIS Codes
(220.4610) Optical design and fabrication : Optical fabrication
(310.1620) Thin films : Interference coatings
(310.1860) Thin films : Deposition and fabrication

ToC Category:
Research Papers

History
Original Manuscript: February 23, 2005
Revised Manuscript: March 23, 2005
Published: April 4, 2005

Citation
John Arkwright, Ian Underhill, Nathan Pereira, and Mark 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/oe/abstract.cfm?URI=oe-13-7-2731


Sort:  Journal  |  Reset  

References

  1. P. W. Baumeister, Optical Coating Technology (SPIE Press, Washington, 2004), Chapter 9 and references therein. [CrossRef]
  2. J. R. Kurdock and R. R. Austin, �??Correction of Optical Elements by the Addition of Evaporated Films,�?? Physics of Thin Films, Academic Press, (1978).
  3. J. M. Mackowski, L. Pinard, L. Dognin, P. Ganau, B. Lagrange, C. Michel, M. Morgue, �??Different approaches to improve the wavefront of low-loss mirrors used in the Virgo gravitational wave antenna,�?? Applied Surface Science 86-90 (1999). [CrossRef]
  4. C. D. Prasad, S. K. Mathew, A. Bhatnagar, A Ambastha, �??Solar photospheric and chromospheric observations using a lithium niobate Fabry-Perot etalon,�?? Experimental Astronomy 125-133, (1998). [CrossRef]
  5. A. J. Leistner, �??Teflon polishers: their manufacture and use,�?? Applied Optics 293, (1976).
  6. J. A. Seckold, �??Precision flat polishing of lithium niobate,�?? Optical Fabrication and Testing OSA Technical Digest Series Optical Society of America, Washington, D. C. (1996).
  7. P. S. Fairman, D. I. Farrant, and J. W. Arkwright, CSIRO Industrial Physics are preparing a manuscript to be called �??High resolution metrology of high-finesse Fabry Perot structures�??.

Cited By

Alert me when this paper is cited

OSA is able to provide readers links to articles that cite this paper by participating in CrossRef's Cited-By Linking service. CrossRef includes content from more than 3000 publishers and societies. In addition to listing OSA journal articles that cite this paper, citing articles from other participating publishers will also be listed.


« Previous Article  |  Next Article »

OSA is a member of CrossRef.

CrossCheck Deposited