Contact bonding, including direct-bonding in a historical and recent context of materials science and technology, physics and chemistry

Abstract

Abstract Bonding is a subject matter, which on the one hand is at least as old as written history, and on the other hand is as modern as ultrahigh-vacuum (UHV) technology. In this paper, we present the main threads of its historical evolution and modern evaluation. Bonding has always been a high-tech technology, which used to be governed by an ‘object in view,’ and nowadays is governed by the ‘state-of-the-art.’ Direct-bonding, i.e. the glueless joining of two solid bodies, is more or less embodied in what we have called ‘contact bonding,’ i.e. a large variety of bonding and annealing techniques. Reasonably weak van der Waals attractions are transferred into strong chemical bonds by annealing. Sir Isaac Newton was the first to see direct-bonding, as testified by his famous central black spot surrounded by ‘Newton rings,’ established between an optical contact of a flat and a convex optical surface. Before World War II, direct-bonding was mainly applied in classical optical instruments (such as interferometers); after World War II it was primarily applied in semiconductor technology, optoelectronics, micromechanics and microelectromechanics. This leads to the need for the thinning of one of the wafers for appropriate applications, such as silicon-on-insulator (SOI). More recently, direct-bonding has been investigated for a large variety of materials, thus leading to significant upgrades in terms of flatness, smoothness and cleanliness. A polishing strategy is one consequence of this, which we will deal with in some detail. During the last decade of the 20th century, great progress was made in UHV-bonding, a technology comparable to lateral solid-phase epitaxial growth (SPEG). Bonding and crystal growing have, therefore, become united disciplines. Wafer thinning now has a new impact, for example, by dedicated ion implantation and low-temperature annealing, called ‘smart-cut.’ A great deal of effort has been exerted to master lattice mismatch in the form of dislocations, i.e. compliant layers. The outlook of these technologies is promising, to say the least, and might one day surpass the physical limits of those of bulk monocrystalline materials such as silicon. All these subject matters are treated step-by-step in this paper. We take a phenomenological approach, sometimes alone or in combination with other disciplines, but not specifically application-directed. The paper covers pragmatic issues and also treats know-how.