Silica glass for fiber optics applications may perform better when manufactured under high pressure, according to research from Penn State and AGC Inc. in Japan. Researchers demonstrated that doing so reduced signal loss in the material. John C. Mauro, professor of materials science and engineering at Penn State, and his team used molecular simulations to evaluate the use of pressure when building optical fibers, showing that pressure quenching could reduce Rayleigh scattering loss by more than 50%. Rayleigh scattering occurs due to fluctuations in the glass’s atomic structure. The voids (yellow) in silica glass become much smaller when the glass is quenched at higher pressures. Courtesy of Yongjian Yang, Penn State. “Glass, on an atomic scale, is heterogeneous,” Mauro said. “It has an open porosity on an atomic scale that occurs randomly.” With pressure treatment of the glass, the homogeneity of the material would increase, and the number of microscopic holes in the structure would decrease, creating a higher mean density and less variability. “We were looking for the independent processes that can control mean and variance,” Mauro said. “We realized that the pressure dimension had not been explored previously.” In theory, Mauro and his team proved that pressure could improve the glass quality; Madoka Ono of AGC Inc.’s Materials Integration Laboratories and a professor in the Research Institute for Electronic Science at Hokkaido University, put the theory into practice. The results matched the simulation. “The optimum pressure we found was 4 gigapascals,” Mauro said. “But there is still a process challenge that needs to be addressed.” The glass must be formed and cooled under pressure while it is still in the glass transition phase, which is the point at which glass is not quite a solid and not quite a liquid. To achieve this, the researchers would need a pressure chamber capable of 40,000 atmospheres. Mauro told Photonics Media the technology isn’t quite ready. “It's possible, but we're not quite there yet. We can get about halfway there with a macroscopically sized sample,” Mauro said. The research was published in Computational Materials (www.doi.org/10.1038/s41524-020-00408-1).
