Abstract
During electrostatic bonding, anodic oxidation of the anode material, for instance silicon, is thought to be the essential step in the bonding mechanism, leading to the formation of a permanent, strong and vacuum-tight bond. Despite the perceived importance of this step in the bonding mechanism of this well-established bonding technique, there is little experimental evidence for anodic oxidation during electrostatic bonding. One reason is that a thin (approximately 10–20 nm) amorphous anodic oxide layer is difficult to detect adjacent to an amorphous cation-depleted glass. Here, silicon–Pyrex and aluminium–Pyrex electrostatic bonds are made and the anodic oxidation process is studied directly using transmission electron microscopy. The consumption of silicon is demonstrated by the movement of the crystalline–amorphous interface compared with a marker under the original silicon–Pyrex interface. The formation of an anodic silica layer can also be demonstrated using electron-energy-loss spectrometry. An amorphous reaction layer 5–20 nm thick is formed during the bonding cycle. For aluminium anode materials bonded at 450°C a nanocrystalline γ-Al2O3 reaction layer is formed, which can be readily detected by transmission electron microscopy. At a bonding temperature of 350°C, no such crystalline reaction layer can be detected between Pyrex and aluminium.
Acknowledgements
The authors would like to thank the UK Department of Trade and Industry for funding the project through the Postgraduate Training Partnership scheme between the University of Cambridge and TWI as well as Norges forskningsråd (NFR) for funding project 140553/420. We would also like to thank Dr D.G. Hasko of the Microelectronics Group, Cavendish Laboratories, University of Cambridge, Cambridge, UK for supplying the silicon with a buried oxide layer, and S. Schlautmann from the Transducers Science and Technology Group, Twente University, Enschede, The Netherlands, for supplying the bonds with microchannels.