The use of materials based on their mechanical behavior has moderated key aspects of the development of civilization. Starting perhaps with stone and wood, used more or less in forms provided directly by nature, progressing to the ferrous metallurgies that gave birth to the entire industrial revolution, and continuing with today's exotic high-strength, light-weight alloys and composites, materials have frequently been selected to fulfill the structural engineering needs of designers, builders, and inventors. Following these trends, studies of the mechanical behavior of solids, once the exclusive purview of blacksmiths and artisans, came to occupy the attention of scientists and engineers worldwide, and were foundational to the modern discipline of materials science and engineering.

In this issue of the MRS Bulletin, we have selected five articles that highlight the intellectual coupling of traditional mechanical behavior investigations with the use of materials in thin film forms. Such a coupling is not entirely new. In an ASM-sponsored seminar held in 1963, for instance, one of the papers presented was entitled simply “Mechanical Properties of Thin Films.” Within the last decade, however, and due largely to developments within the microelectronic and magnetic disk industries, awareness of the need to predict, control, and understand the mechanical behavior of thin film media has grown rapidly. Many of the most important failure mechanisms operative in integrated circuits, for example, such as stress and electromigration voiding, are mechanical in nature.

One of the simplest ways to measure the mechanical properties of a thin film is to deform it on a very small scale. Because indentation testing with a sharp indenter is one convenient means to accomplish this, nanoindentation, or indentation testing at the nanometer scale, has become one of the most widely used techniques for measuring the mechanical properties of thin films. Other reasons for the popularity of nanoindentation stem from the ease with which a wide variety of mechanical properties can be measured without removing the film from its substrate and the ability to probe a surface at numerous points and spatially map its mechanical properties. The utility of the mapping capability is illustrated in Figure 1, which shows several small indentations made at selected points in a microelectronic device. The hardness and modulus of the device were determined at each point. In addition to microelectronics, nanoindentation has also proved useful in the study of optical coatings, hard coatings, and materials with surfaces modified by ion implantation and laser treatment.

Evaluating the mechanical properties of thin films on thick substrates is tricky. Some useful techniques for mechanical property measurements on macroscopic film specimens exist, e.g., internal stress measurements by x-ray diffraction or by wafer buckling. Many conventional techniques, however, such as indentation for hardness measurement or scratch testing for evaluation of adhesion or wear resistance, yield results that are difficult or impossible to evaluate in terms of more fundamental film properties. In fact, these techniques are more practical engineering tools than tools for scientific study.

One solution is to miniaturize the test probe to dimensions approaching the film thickness. The nanoindentation technique for submicrohardness measurement described elsewhere in this issue by Pharr and Oliver is a recent example of such an approach. Still, for very thin films it is difficult to separate the influence of the substrate properties from the film properties, and the indentation process is inherently too complex to be well suited for determining fundamental materials parameters.

The rapidly evolving field of micromechanics offers some new possibilities in thin film characterization. Important film properties such as internal stress, elastic moduli, plastic yield limit, and fracture data can be extracted from experiments with micromachined test structures in the 10–1,000 μm size range.

Thin film stresses are important in many areas of technology. In the semiconductor industry, metal interconnects are prone to stress voiding and hillock formation. Stresses in passivation layers can lead to excessive substrate curvature which can cause alignment difficulty in subsequent lithographic processing. In other thin film applications, stresses can cause peeling from mechanical failure at the film-substrate interface. Beyond these issues of reliability, stress and the resulting strain can be used to tune the properties of thin film materials. For instance, strain, coupled with the magnetostrictive effect, can be utilized to induce the preferred magnetization direction. Also, epitaxial strains can be used to adjust the bandgap of semiconductors. Finally, the anomalous mechanical properties of multilayered materials are thought to be partially due to the extreme strain states in the constituents of these materials. To fully optimize thin film performance, a fundamental understanding of the causes and effects of thin film stress is needed. These studies in turn rely on detailed characterization of the stress and strain state of thin films.

X-ray diffraction and the elastic response of materials provide a powerful method for determining stresses. Stresses alter the spacing of crystallographic planes in crystals by amounts easily measured by x-ray diffraction. Each set of crystal planes can act as an in-situ strain gauge, which can be probed by x-ray diffraction in the appropriate geometry. Hence it is not surprising that x-ray diffraction is one of the most widely used techniques for determining stress and strain in materials. (For reviews of this topic, see References 5–7.) This article is a tutorial on the use of x-ray diffraction to extract the stress state and the unstrained lattice parameter from thin films. We present a handbook of useful results that can be widely applied and should be mastered by anyone seriously interested in stresses in crystalline thin films with a crystallographic growth texture.

Advances in the methods of deposition and characterization of crystalline films of submicron thickness have been dramatic over the past two decades. A principal motivation for development of this technology has been the potential for use of thin film semiconductor materials in electronic and optoelectronic devices. The main function of these devices is to control transport of electrons in a way which permits high spatial density, and in which the carriers are highly mobile, that is, they show fast response with little power consumption. Spatial control of mobile electrons can be facilitated by combining materials, forming a material heterostructure, with one or more of these materials being a thin film. Carrier confinement is enforced by a difference in energy band structure across the interface acting as a barrier. The exploitation of this physical effect in device design is called bandgap engineering. A great deal of attention has been focused on film/substrate systems involving the III-V compounds (InGaAs/GaAs, for example), as well as on II-VI compounds for optical applications (ZnSe/GaAs, for example). Current efforts are also directed toward SiGe/Si and GaAs/Si systems to exploit well-developed silicon device technology.

Narrow, passivated metal lines are generally used as interconnects in VLSI microcircuits at the chip level. In most metals, high electric current densities lead to a mass flow of constituent atoms accompanying the current of electrons. Electromigration (EM) has long been considered an important reliability concern in the semiconductor industry because the current-induced atomic fluxes can give rise to void formation and open circuits, or hillock formation and short circuits between nearby interconnects. The problem is exacerbated because of the continued trend of increasing the density of the devices on the chip. This means that the line widths of the interconnects have been reduced and are now in the submicron range; correspondingly, the current densities have increased and may be as high as 106 A/cm2. Recently, thermal-stress-induced damage in metallizations has also been recognized as an important reliability concern, perhaps of the same gravity as EM. Thermal stresses in the metallizations are caused by the different thermal expansion coefficients of the metal and the substrate. Stress-induced void and hillock formation are the main causes of in terconnect failures before service. More recently, concern has been growing that thermal stresses or thermal-stress-induced voids may enhance the subsequent electromigration damage during the service life of the microchips.

For simplicity, this article addresses the case of pure aluminum metallizations on oxidized silicon substrates. However, much of what is said applies to other metal-rigid substrate systems as well, most notably to various aluminum and copper-based metallizations on ceramic substrates. The present treatment emphasizes void formation and growth in the metallizations during nd after cooldown from elevated temperatures, or those due to electromigration in service or testing conditions. Many of the mechanisms we explain are also applicable to hillock formation under compressive stresses, whether due to EM or thermal cycles during manufacturing.

I would like to raise a question of interest to many of us here today: “Why have we not been able to find a room-temperature superconductor?” I have a scenario for why not that can be illustrated by projecting ahead 98 years. It is 2089, the tercentenary of the French Revolution. The revolt against science that started building up in the last decade of the 20th century has reached a crescendo and a revolutionary tribunal has erected a huge guillotine. Following in the tradition of Lavoisier's trial and execution, the presidents of the Materials Research Society, the American Chemical Society, and the American Physical Society have been found to be enemies of the people and have been sentenced to be guillotined. The president of the Materials Research Society steps up. In his final words, he says: “I have no regrets. We've supplied you with the pole vaults which have made possible a new world's record above 30 feet, we've given you automobile bumpers that prevent damage in crashes at 60 miles per hour, and we've given you tennis rackets that are big enough to allow you to cover from center court to the alley without having to take a step.” With that, he bravely puts his head on the block. Down crashes the knife but inexplicably it stops just before it reaches his neck. According to the rules he steps down a free man. Then the American Chemical Society president steps up and says: “I don't have any regrets either. We have given you body centered cubic diamond that makes it possible to build indestructible houses, we have given you disposable diapers that are biodegradable in 30 seconds, and we've reseeded the stratosphere with ozone. We have done our job.” She courageously puts her head on the block, and the guillotine comes down. Again it stops short and she walks away, a free woman. Then the president of the American Physical Society says: “I don't have any regrets either. We've circled the equator with our newest SSC and have given you the Higgs boson, we've given you light that you can squeeze until it hurts, and we've discovered over 100 theoretical models which should lead to room-temperature superconductors. We haven't given you any real ones because there aren't any—we've tested all the possibilities. But, by the way, while we've been standing here, I've done a back-of-the-envelope calculation and I've found out wha's wrong with this guillotine. If you give me a screwdriver I can fix it.”