F.M. d'Heurle

Nucleation of a new phase from the interaction of two adjacent phases: Some silicides

The reactions of metal layers with their silicon substrates resulting in the formation of various silicides are considered generally not only as phenomena common to all diffusion couples where new phases are formed, but also as typical of all transitions from two to three phases. The conditions under which such transitions will display the same characteristics as encountered in the usual one-to-two phase transitions (condensation, crystallization, boiling) are analyzed by comparison to the classical theory of nucleation. Because of the lack of knowledge about the exact values of the relevant parameters, the discussion is carried out mostly in descriptive thermodynamic terms. Although nucleation effects are analyzed in general terms, the main focus of attention is a class of reactions where nucleation dominates the formation of a new phase; a salient feature of these reactions is the absence of any equilibrium temperature, although the nucleation temperatures are relatively well defined within narrow limits. Nucleation effects are correlated to such material characteristics as the stability of the nucleated phases, and to such kinetic characteristics as the sequence of phase formation. The modification of the energy levels of the different phases brought about by stress, ion bombardment, or the replacement of usual phases by metastable ones, are considered with respect to their effect on nucleation processes. The nearly total absence of literature references to nucleation in metal-metal diffusion couples is discussed with respect to some specific aspects of the metal-silicon reactions.

Kinetics of formation of silicides: A review

The kinetics of silicide growth are classified into three different categories: (a) diffusion controlled, (b) nucleation controlled, (c) others (reaction rate controlled). These are analyzed with the aim of understanding both the phenomenology of growth and the specific atomic mechanisms of phase formation. Diffusion-controlled growth is discussed with respect to the Nernst-Einstein equation. Stress relaxation is considered as a possible cause of reaction-rate control. The relative merits of two different types of marker experiments are compared. A few silicides are discussed in terms of what can be inferred about diffusion mechanisms. The competition between reaction-rate and diffusion control phenomena is shown to have specific effects on the sequence of phase formation; it is also related to the formation of some amorphous compounds. Reactions between silicon and alloyed metal films are used to illustrate the respective influences of mobility and driving force factors on the kinetics of silicide growth; they can also be used to underline the dominance of nucleation over diffusion in some silicide formation processes.

Note on the origin of intrinsic stresses in films deposited via evaporation and sputtering

An attempt is made to provide a broad and brief review (with a sufficient number of references) of the question of intrinsic stresses in films deposited via evaporation and sputtering. Films deposited via evaporation are usually found initially in a disordered state, at the limit in an amorphous condition. Very broad thermodynamic principles imply that disorder is usually accompanied by an increase in volume (with few exceptions, e.g. water). Any relaxation from a disordered state to a more ordered state will, therefore, be accompanied by a decrease in volume and the formation of tensile stresses. The situation is modified in films deposited in the presence of impurities. Then one often finds the formation of compressive stresses. Several mechanisms may account for this result: (a) direct diffusion of interstitial impurities in the bulk of the underlying film: (b) surface and grain boundary diffusion of the impurities leading to compound formation (and swelling) in the grain boundaries, and at low temperatures in intergranular voids; (c) some attention is paid to a third model where compressive stresses could result from impurity adsorption not at the top surface of the growing films, but one monolayer below. The presence of compressive stresses in films deposited via sputtering is briefly reviewed in terms of the atomic peening mechanism. An experiment exploring the interrelationship between the purification effect of ion bombardment, thus causing tensile stresses, and the more normal formation of compressive stresses is discussed.

Formation of thin films of CoSi2: nucleation and diffusion mechanisms

Abstract The reaction of thin cobalt films with their silicon substrates leads to the formation (in sequence) of Co 2 Si, CoSi and CoSi 2 . The formation of the first two compounds has already been studied in detail, but the formation of the disilicide remains to be elucidated. It is important to understand better the formation of CoSi 2 because of its similarities with NiSi 2 which is well known to form via nucleation-controlled processes. With very thin films, only 20–30 nm of cobalt, clear evidence for nucleation phenomena has been obtained. It is believed that nucleation explains, at least partly, the complex kinetics which are observed during the formation of thick (several hundred nanometers) layers of CoSi 2 . The coincidence of the temperatures where nucleation and diffusion take place makes it impossible to identify clearly the diffusion kinetics as, for example, with Co 2 Si, CoSi or NiSi, or the nucleation process as, for example, with NiSi 2 . As with other compounds the importance of the nucleation effects results from the small driving force for the transition from CoSi to CoSi 2 . An attempt is made to compare quantitatively the role of nucleation in the formation of CoSi 2 over crystalline silicon and over amorphous silicon. An experiment with an implanted marker confirms the expectation that the formation of CoSi 2 , like that of the isomorphous NiSi 2 , occurs mostly through the motion of the metal atoms.

The formation of Cu3Si : marker experiments

Abstract Implanted rare gas markers, as well as a thin tungsten marker, show that the silicide Cu 3 Si, which is the first phase to form on the reaction of a copper film with a silicon substrate, grows by the motion of copper atoms. The same process seems to be at work in the formation of the silicide by ion mixing, but the one result obtained in that case is not unambiguous since the observed disappearance at the surface of the implanted marker might be due as well to interface drag effects as to the specificity of the mobile atoms. The low temperature of formation of the silicide, about 200°C, is discussed in terms of what is known about point defects and diffusion in the very similar compound Cu 3 Sn.

Oxidation of silicon-germanium alloys. I. An experimental study

The oxidation of polycrystalline SixGe1−x films with different compositions (i.e., different values of x) is carried out in pyrogenic steam at 800 °C for various lengths of time. It is found that the oxidation is enhanced by the presence of germanium and that the enhancement effect is more pronounced for the films richer in germanium. A mixed oxide in the form of either (Si,Ge)O2 or SiO2–GeO2 is found at the sample surface if the initial SixGe1−x contains more than 50% of germanium. However, a surface silicon cap layer of about 14 nm is found to have a significant impact on the oxidation of the Si0.5Ge0.5 films; it leads to the growth of about 115-nm-thick SiO2 which is about four times that of the SiO2 resulting from the oxidation of the cap layer itself. On the SixGe1−x films with only 30% of germanium, the SiO2 continues to grow after oxidation for 180 min resulting in 233-nm-thick SiO2 which is about 2.4 times greater than the SiO2 grown on 〈100〉 silicon substrates. Rejection of germanium results in p...

Reaction of silicon with films of CoNi alloys: Phase separation of the monosilicides and nucleation of the disilicides

Abstract CoNi alloy films with compositions varying from 5 to 95 at.% Co were deposited onto silicon substrates by evaporation from electron beam sources. After annealing, the reaction products were analyzed by Rutherford backscattering spectrometry, Auger electron spectroscopy, secondary ion mass spectrometry and X-ray diffraction as a function of the annealing temperature (from 450 to 750 °C). The silicide formation is influenced by the isomorphism of the compounds Co2Si and Ni2Si, and CoSi2 and NiSi2, and the lack of isomorphism of CoSi and NiSi. It is observed that the formation of the monosilicides is accompanied by a non-uniform distribution of the metallic elements as a function of depth. The formation of the alloyed disilicides always occurs at temperatures much lower than those required to form NiSi2 from pure nickel. For most cases the alloyed disilicides form at temperatures even lower than that required for the formation of CoSi2 (pure). The position where the formation of the disilicides begins varies with the composition of the alloys. In some cases the initial formation of the disilicide occurs at the free surface and is followed by inward growth; in other cases this occurs inside the complex monosilicide layer. Such effects illustrate quite dramatically the importance of nucleation phenomena in the formation of these disilicides. Nucleation of the disilicide phase at the interface between CoSi and NiSi is attributed to the contribution of the entropy of mixing to the driving force for the transformation, clearly illustrating the importance of entropy in some classes of solid state reactions.

Formation of silicide thin films by solid state reaction

Abstract Because of the practical interest of silicides, the M-Si thin film reaction has been thoroughly investigated over the last two decades. This allows a systematic analysis of M-Si reactions and suggests that observations on thin film reactions may in some respects differ from those on bulk reactions: sequential appearance of phases, absence of certain phases, very rapid kinetics of formation. The scope of this paper is to review the existing knowledge, to describe the thin films “peculiarities” and to tentatively explain them taking into account the specificities of reactive diffusion and the role of the experimental conditions encountered in M-Si thin film reactions (low temperature, high density of interfaces, small thicknesses).

The formation of disilicides from bilayers of Ni/Co and Co/Ni on silicon: Phase separation and solid solution

Abstract Bilayers of Ni/Co and Co/Ni were deposited onto silicon by means of electron beam evaporation. The annealing behavior was investigated as a function of time in the temperature range 475–550 °C by Rutherford backscattering spectrometry, Auger electron spectroscopy, scanning electron microscopy and X-ray analysis. In this temperature range it is observed that the disilicide (solid solution) nucleates and grows from a film containing both the monosilicide of cobalt and that of nickel. These nearly insoluble monosilicides are separated into layers; the disilicide grows from the interface between the NiSi and the CoSi phase. If there are equal amounts of the two metals this interface is located approximately midway between the upper surface and the silicon substrate. The disilicide grows simultaneously both toward the free surface and toward the silicon substrate. The growth kinetics of the disilicide have been analyzed and are found to be complex. The general behavior of the disilicide formation is found to be similar to that found for Co-Ni alloy films and as for the latter case indicates the primacy of nucleation over diffusion as the controlling mechanism in the disilicide formation. The site of nucleation of the disilicide at the NiSi-CoSi interface is explained by a model based on (a) the phase separation of the two monosilicides and (b) the high entropy of mixing of the mixed disilicide. In agreement with this model the nucleation of the mixed disilicides is shown to occur at temperatures much lower than those for either CoSi 2 or NiSi 2 . Given the common crystal structure and the almost identical lattice parameters, the mutual solubilities of CoSi 2 and NiSi 2 follow from elementary concepts of alloy phase theory. However, the behavior of the monosilicides is more unexpected. The limited solubilities of the monosilicides, CoSi and NiSi, in each other are shown to result from specific features of the electronic band structure of these two compounds, features which are not found for example in the isomorphous disilicides CoSi 2 and NiSi 2 .

Thermal stability of silicide on polycrystalline Si

Abstract The thermal stability of silicide on polycrystalline Si (poly-Si) has been investigated. At elevated temperatures, the silicide/poly-Si layered structure becomes morphologically unstable, because of the grain growth of poly-Si. The driving force for the high temperature instability is the reduction of the grain boundary energy and surface energy of the poly-Si. In situ stress measurement shows that the grain growth is accompanied by a decrease in the compressive stress of as-deposited poly-Si. A change in crystallographic texture from (110) to (111) is also observed during the grain growth, indicating a process similar to that of secondary grain growth. A study of the grain growth kinetics shows that the grain growth of poly-Si is enhanced by fast diffusion in the silicide, but is not rate limited by the diffusion of the dominant diffusion species in the silicide. A strong correlation is found between the onset temperature for plastic deformation in the silicide and that of the grain growth in poly-Si.