Characterization of Multilayer Optical Coatings on Microstructured Surfaces
An Analysis of Optical Interference Coatings on Diffraction Gratings Using SEM and Related Techniques
James Oliver - MSC 507 - April 29, 2004
I. The Omega-EP laser system currently under construction at the Laboratory for Laser Energetics (LLE) will require gratings for pulse compression of 1053nm light from the nanosecond-regime to the picosecond-regime. In the past, grating structures have been created in photoresist and overcoated with a thin film of gold to achieve high reflectivity. This creates a grating that may be quite efficient, but the gold has residual absorption that significantly limits the maximum fluence that may be incident on the grating surface. Since multilayer dielectric coatings are capable of producing the reflectivity and laser damage resistance necessary for this application, the creation of a grating/multilayer combination is being investigated for use on the Omega-EP system. Other groups currently provide multilayer dielectric coatings with pulse-compression gratings etched in the top layer (Jobin Yvon, Lawrence Livermore National Laboratory)[1-2]; LLE is investigating the use of continuous-enfolded gratings for this application . It is proposed that a series of gratings be fabricated, etched, and coated with a multilayer dielectric. The results of this effort will be evaluated using scanning electron microscopy.
II. Gratings were created by the Optical Imaging Sciences Group (OISG) at LLE, and thin film coatings were applied by the Optical Manufacturing Group (OMAN) at LLE. Imaging and analysis were performed at the University of Rochester River Campus Electron Microscopy Facility, in conjunction with the SEM Practicuum class taught by Brian McIntyre.
III. Samples will be created and provided at all relevant stages of progress, including patterned photoresist, etched fused silica surface, and finished component with the multilayer deposited on the microstructured surface. It will be necessary to provide multiple samples, each at a different stage of production rather than measuring a single sample as it is processed, due to the destructive nature of the sample preparation.Samples will be cross-sectioned and then coated with a thin layer of sputtered gold. The grating profile and relative coating thickness over this profile will be mapped using SEI and BSE imaging in the SEM. The relative mapping of layers close to the etched surface compared with layers further removed in the coating stack will also be evaluated, by quantifying the modulation at each layer interface. X-ray spectrometry will be used to evaluate the presence (or lack thereof) of residual contaminants from the presence of the photoresist on the silica surface.
 B.W. Shore, M.D. Perry, J.A. Britten, R.D. Boyd, M.D. Feit, H.T. Nguyen, R. Chow, G.E. Loomis, and L. Li, “Design of High-Efficiency Dielectric Reflection Gratings,” J. Opt. Soc. Am. A 14 (5), 1124 (1997).
 H. Wei, L. Li, “All-dielectric reflection gratings: a study of the physical mechanism for achieving high efficiency,” Appl. Opt. 42, 6255-6260 (2003).
 L. Li, J. Hirsh, “All-dielectric high-efficiency reflection gratings made with multilayer thin-film coatings,” Opt. Lett. 20, 1349 (1995).
Two sets of grating samples were created in photoresist, with groove densities of 1740 lines/mm. The first set possessed somewhat of a sinusoidal shape, with a modulation of approximately 0.35mm. The second set of gratings had a flatter structure on the top, with a reduced modulation, on the order of 0.18 mm. The first set of gratings was overcoated with a multilayer dielectric coating consisting of alternating layers of tantalum pentoxide and silicon dioxide, deposited at ambient temperature to preserve the photoresist structure. Etching difficulties encountered in the OMAN ion etch system led to delays in producing a CEG on a glass microstructure (i.e. the photoresist pattern transferred into the substrate), but a sample was eventually etched and overcoated with a hafnium dioxide/silicon dioxide multilayer deposited at 200°C. Degradation of the grating structure due to the addition of multiple interference layers was evaluated based on the images obtained, providing an understanding of the growth direction (orientation/pattern) of the coating.
Fused silica substrates, 100mm in diameter by 1mm thick, were spin-coated with a thin layer of photoresist. High- and low-modulation grating structures were exposed and then developed in this photoresist layer. The grating component was cleaved to provide a relatively clean break of the structure perpendicular to the grating grooves, by scoring the back of the substrate and breaking using a custom vacuum chuck. After sputter coating the sample with a thin layer of gold, the cross-section was then imaged using in-chamber secondary-electron imaging at 30,000X. These images are shown in Fig. 1-2, respectively.
One of the samples produced in the batch with that shown in Fig. 1 was then ion milled, in an attempt to transfer this pattern into the fused silica substrate. It is widely accepted that features of this size require reactive ion etching (RIE) in order to successfully transfer the pattern into the substrate. However, it was prudent to confirm that ion milling would not work, since OMAN possessed an ion milling system and the controls for the gases are much simpler than for RIE. The severe degradation of the pattern due to the ion milling process is shown in Fig. 3; it was necessary to reduce the magnification to 10,000X to allow the grating structure to be readily apparent. At the higher magnification, the cratered surface tended to dominate rendering the grating unrecognizable. Nevertheless, it is obvious that the structure has been destroyed by the ion milling process, practically eliminating the diffractive properties of the component.
While the diffractive properties of the component were destroyed, this sample could still provide useful information in the form of energy-dispersive x-ray analysis (EDAX). It is important to characterize the etching process, to evaluate the presence of contaminants in the component surface due to degradation of the ion source grids (molybdenum), sputtering of fixturing materials in the process chamber (stainless steel, aluminum), or remnants of the photoresist processing (carbon). This sample was evaulated using EDAX, with the results shown in Fig. 4.
The EDAX analysis shows that the etching (milling) process is being performed cleanly, without contaminants from the system being left on or imbedded in the surface. This is critical for high-laser-damage components for systems such as Omega-EP, due to the extremely high fluences incident on the components.
While difficulties in ion milling/ion etching were being addressed, it was decided to characterize the coating of a microstructured surface using a photoresist grating. This limits the flexibility in the coating process, since most evaporated thin films (especially oxides) require elevated temperatures during deposition. By adding heat during the deposition process, the film tends to more fully oxidize (lower absorption) and has greater surface mobility (more dense films). A 7-layer tantalum pentoxide/silicon dioxide high reflector was deposited on the high modulation photoresist grating shown in Fig. 1. This process was carried out in a Davis & Wilder 28" coating chamber equipped with electron-beam evaporation and a planetary rotation system. The results of this deposition, which was carried out at < 50°C, are shown in Fig. 5-6.
There is a great deal to be learned from the 7-layer tantala/silica coating on the diffraction grating. First, the diffractive properties are practically eliminated within 4 layers. The interfaces after layers 5-7 are rapidly approaching planar. This will result in a standard "specular" reflection from the coating within these layers, greatly reducing the diffraction efficiency of the componenet. Severe voids, or "cracking," are also apparent between the grating features in layer 2. These are due primarily to self-shadowing of the surface by the accumulating coating on the adjoining grating features. The extremely low surface mobility of the coating molecules contributes to this phenomenon. It is expected that the use of a higher energy deposition process, such as ion-assisted deposition or sputtering, would allow improved mapping of the grating surface.
A second set of grating samples were created using the low-modulation grating structure etched into a fused silica wafer. Since the structure was etched, removing the photoresist, the resulting grating was capable of withstanding the same temperatures as fused silica. This allowed the deposition of a hafnia/silica multilayer coating at 200°C, which was expected to possess improved packing density. Furthermore, the reduction in surface modulation should minimize self-shadowing, producing less void in the film. The 14-layer hafnia/silica multilayer result is shown in Fig. 7. The modulation of the surface features are preserved a greater distance through the coating structure, and the void fraction is greatly reduced. Careful inspection shows that the remains of the periodic structure are still evident at the top surface of the coating, although essentially all diffractive properties have been eliminated by this point in the coating.
The original concept of this project was to characterize the grating shape at each interface, showing how the shape changed as a function of deposited film thickness or layer count. Mathematical characterization of the shape of the interfaces is rather difficult, since the shapes are quite irregular and arbitrary curve fits to each layer will provide little insight. Instead, it was decided to characterize the degradation of the modulation at each interface. Accordingly, a simple program was written to digitize points on the micrograph for transfer to Microsoft Excel for analysis.
Since the shape transfer through the coating layers did not lend itself to mathematical curve fitting, the amplitude of the modulation was instead determined at each interface. The image was digitized within the program as shown in Fig. 10:
The points from the image in Fig. 10 were then exported, the modulations calculated, and the degradation of the modulation was plotted as a function of increasing layer count. For the purposes of this plot the initial thin hafnia layer on the surface of the fused silica grating was neglected, since its thickness is not comparable to that of the other layers (and it should not, therefore, provide equivalent degradation to the surface contours). The change in modulation as a function of layer count is shown in Fig. 11.
The reason for this degradation is apparent by considering the growth of the film as occurring normal to the localized groove structure. Each of the grooves will tend to grow together as the thickness of the coating progresses through the deposition. This is similar to a growing circle or sphere. As you move far enough away from the center of the sphere, a small area appears more and more planar. This problem is not as simple as centering the sphere, however, due to the influence of other grooves which are also growing. This behavior is shown in Fig. 12.
It would be interesting to create individual features that are widely separated on the substrate surface, in order to determine a transfer function for each feature prior to analyzing the composite of many features. This should allow the modification of process parameters, or even the selection of a suitable process, capable of accurately mapping the desired grating structure throughout the multilayer.
Gratings were created in photoresist at two different modulation heights with a groove density of 1740 lines/mm. Photoresist gratings were cleaved, coated with gold, and imaged using secondary electron imaging. A sample was etched, imaged and evaluated using EDAX, to evaluate potential contamination due to sputtering of ion source grids or sample fixturing. It was determined that the etching process did not introduce significant contamination on the sample surface. Both low- and high-modulation grating structures were overcoated with multilayer dielectric coatings, and the results imaged using backscatter and secondary electron imaging. The rapid degradation of the groove structure is apparent in both samples, although the use of the low-modulation surface with a heated deposition process improved the transfer of the grating structure and minimized void formation in the coating. The degradation of the groove structure appears to be proportional to 1/x, due to film growth that is relatively normal to the localized surface features, rather than the average substrate surface. Significant challenges remain to deposit a multilayer with a high layer count (> 20 layers) over a grating structure using electron-beam evaporation. Analysis shows that the current process is limited to 4-6 layers before the grating shape is severely degraded.