This Small Business Innovation Research (SBIR) Phase I project will develop the first automated sweep pattern generator for production scale electron beam deposition systems that will enable better control over layer optical thickness, improving optical coating performance for many critical applications. Electron beam deposited optical coatings are an enabling technology for telecommunications, defense, satellites, and high energy lasers. However, the electron beam process is limited by coating layer thickness variation resulting from uneven evaporation of the coating material and requires uniform electron beam sweep pattern. The project will use thermal imaging capabilities and a newly proposed design theory to develop an automatic sweep pattern generator that will enable more uniform evaporation. In addition to reducing layer thickness variation for applications requiring significant source material utilization by allowing more uniform depletion, this new pattern generator will enable higher precision coatings and deposition of more layers from a source crucible. The strategy is to leverage the company's extensive process knowledge and industrial collaborators to deliver significant value in a product that is simple to use and easily integrated into production systems.
The broader impact/commercial potential of this project will be realized in optical coating performance, an enabling technology for many critical precision optics applications including high energy lasers, lithography, telecommunications, defense, and satellites. Electron-beam vacuum deposition is dominantly used to produce these coatings because of its superior economics and process versatility. However, the industry is constantly demanding more stringent tolerances on layer thickness and composition while increasing the number of layers. Emerging foreign competition necessitates optimizing production objectives such as throughput, quality, and cost. U.S. manufacturing must both tighten its tolerances and lower its costs if it is to continue to play a leadership role in the 1.8 billion dollar high-end precision optical coating market. Domestic e-beam system vendors are struggling to compete with foreign vendors who sell turn-key coating systems with high precision processes pre-configured. These foreign vendors are able to charge a premium due to the difficulty of developing advanced processes. In order to facilitate advanced process development in production, vendors seek to provide high precision systems with advanced instrumentation including programmable sweep controllers and optical monitors. The automated sweep pattern generator will work synergistically with these components to enable a high precision system that can be optimized automatically.
The objective of this Phase I SBIR project was to demonstrate the proof-of-concept of an automatic sweep pattern generator for electron beam deposition for manufacture of precision optical coatings. Optical coatings manufactured by electron beam deposition are an enabling technology for many important applications in telecommunications, defense, satellites, and high energy lasers, and are typically comprised of multiple layers of different optical materials that are designed to modulate light passing through them. Electron-beam vacuum deposition is widely used to produce these coatings because of its superior economics and process versatility. Emerging foreign competition necessitates improving production objectives such as throughput, quality, and cost if the US is to continue to play a leadership role in the 1.8 billion dollar high-end precision optical coating market. Optical coating performance is often determined by each layer’s thickness variation and the industry is demanding more stringent tolerances on layer thickness and composition while increasing the number of layers. A major source of variation in the e-beam process is non-uniform evaporation of the material due to uneven heating of the source material by the electron gun. In an attempt to compensate for non-uniform heating, digitally programmable sweep controllers were developed to modify how the beam is swept across the material. However, these problems persist because system operators lack a systematic method to design sweep patterns that compensate for e-gun non-uniformities. Our approach to this problem was to develop vision-based experimental models of the e-gun system that capture the impact of each digital pixel of the programmable sweep controller. Using this model, we developed a system that iteratively obtains an improved sweep pattern design, evaluates the performance of the design and identifies problems, and adjusts the sweep pattern performance to achieve good evaporation performance. We found that three distinct capabilities were needed, and during Phase I developed and tested the following three algorithms: - an iterative algorithm to develop an initial pattern to ensure that the starting pattern was broad and roughly at uniform magnitude. - a system identification method that provides an accurate linear model of the impact of each sweep controller pixel on the current sweep pattern - an analysis algorithm to determine which minimal set of pixels should be utilized to span the desired evaporation area. Overall, the proposed method was shown to improve sweep pattern performance in terms of spatial coverage and uniformity, as measured by performance metrics developed as part of this effort. In one experimental evaluation, sweep pattern coverage increased by 400% from the coverage of the standard pattern included with the sweep controller. In addition, we also developed the underlying hardware capabilities needed to automate sweep pattern development, including reduction of vision analysis time by a factor of 8 to enable effective real-time video analysis and development of a custom interface to one of the widely used sweep controllers in order to increase capabilities and eliminate operator intervention. During Phase I, several e-beam process issues were discovered and evaluated in terms of their effect on pattern performance. The most significant issue observed was e-gun non-uniformities related to how the system was setup that limit sweep controller positioning performance, which can therefore limit pattern improvement. This suggests that proper setup procedures need to be established before pattern development can be initiated. Adjustment of the e-gun magnetics is a significant undertaking for which there is currently no systematic optimization technique. Although significant pattern improvement can be achieved with the iterative system ID procedure, there are still a number of issues that must be overcome that could limit performance and market potential. It remains to be evaluated whether the inherent variability of the e-beam process may overshadow the performance increases achieved by an automated sweep pattern generator. Cyber Materials will explore partnership opportunities with leading e-beam hardware vendors who can help inform a solution to overcome these performance limiting issues.
