Direct Wafer Bonding and Layer Exfoliation of Sapphire

Abstract

The process of sapphire direct wafer bonding has been previously explored, and it has been shown that at room temperature, sapphire may be directly bonded to a variety of materials, including silicon and germanium. However, its applications have been mostly directed toward creating substrates for silicon-on-insulator (SOI) structures. Recent developments in consumer electronics have created new interest in sapphire’s application for device displays, and a detailed process by which sapphire may be successfully integrated has become necessary. In this paper, we lay out the initial steps for direct wafer bonding and layer exfoliation of bonded (0001) sapphire wafers for this purpose. Bonding was performed at room temperature in air, and three cleaning steps – Standard Clean 1 (SC1), oxygen plasma exposure, and a combination of Standard Clean 1 and oxygen plasma cleaning – were investigated with regards to their effectiveness in cleaning the sapphire surface as well as rendering the surfaces hydrophilic. The SC1 cleaning step involved a ten minute immersion in a 1:1:5 NH4OH:H2O2:H2O solution followed by a 1 minute rinse in deionized (DI) water. The oxygen plasma cleaning involved a 1 minute exposure to a 200 W O2 plasma in vacuum, followed by a 1 minute rinse in DI water. Following cleaning, all samples were annealed at 450 °C for 15 hours with an approximately 3 °C/min ramp rate up and down to maximize the bond strength. Transmission infrared images were taken before and after annealing, and used to quantify the percent area bonded of the total interfacial area, which was used as a rough indicator of bond quality. It was found that usage of a SC1 or oxygen plasma cleaning step resulted in the highest percent area bonded. To further quantify the bond strength, samples were subjected to the Maszara crack-opening test. It was found that the Standard Clean 1 cleaning step resulted in the highest bond strengths among three conducted tests. Further investigations using x-ray photoelectron spectroscopy (XPS) determined that the effectiveness of the Standard Clean 1 step above the other two investigated methods was due to the increased presence of hydroxyl groups as a result of surface hydration of the sapphire surface during the cleaning. These hydroxyl groups served to increase the surface charge, which contributed to greater initial attraction between the two wafer surfaces. Further investigations into the determination of parameters required for layer exfoliation are also discussed. Simulations performed on SRIM reveal that using an implantation energy of 360 keV results in a layer thickness of approximately 2 µm. Previous work conducted by our group has also shown that, in contrast to previous research done on sapphire layer exfoliation, implantation at elevated temperatures is not necessary, and can be replaced by implantation at lower temperatures (i.e. room temperature) in order to control the defect state, followed by a two-step heat treatment at low (< 500 °C) and then elevated temperatures to maximize bond strength, sufficiently initiate defect nucleation and coarsening, and finally, achieve smooth layer transfer. For this purpose, the effectiveness of a 15 hour anneal at 150 °C, 200 °C, and 250 °C respectively was investigated for the purpose of determining the temperature for the initial anneal step. It was found that annealing at 250 °C resulted in comparable percent area bonded to the previously-used 450 °C anneal. As a result of this investigation, it is concluded that for sapphire direct wafer bonding and layer exfoliation, a SC1 cleaning step is preferred to prepare wafer surfaces for bonding. Additionally, exfoliation parameters such as implantation voltage and heat treatment temperature were determined for future application.