Silicides: Fundamentals and Applications

The following sections are included:

• PREFACE

• CONTENTS

Part of a textbook on the structures, properties and applications of metal silicides, this paper gives a short description of some structural features of binary metal silicides, essentially the disilicides and compounds of the rare earth metals.

This paper gives a short summary of the structural chemistry of binary transition metal silicides, their structure and most essential features.

A brief review of the origin of the chemical bonding in transition metal disilicides, pointing out the existing differences between two prototypical families, i.e. near noble silicides and refractory silicides is presented. Some interesting and original issues concerning the stability trends among competing phases will be then addressed in either cases, showing how they take origin from the bond nature.

After a brief review about diffusion in pure metals and in intermetallic compounds, what is known about diffusion in silicides is analyzed. The topics covered are: 1) Diffusion during reactive phase formation with respect to dominant moving species and lattice versus grain boundary diffusion. 2) Self diffusion studies using metal isotopes and *Ge in lieu of unyieldy (short life times) Si isotopes, and 3) Diffusion of impurities (Si dopants).

The benefit of numerical simulations, purely thermodynamic or coupled thermodynamic-mass/heat transport, allowed and still allows to optimise different elaboration processes of silicides (microelectronics, composite materials, single crystal growth …). In order to allow an experimentator, usually not very familiar with thermodynamic data and their inclusion in numerical databases, to perform such a process simulation, the structure of the data, the classical models used and generally admitted conventions will be presented in this contribution. After this first step, a couple of recent case studies will illustrate what type of information can be expected to be obtained and what is the standard procedure of using this approach.

Binary silicides as contact materials in microelectronics, ternary silicides in composite materials and finally the case of single crystalline silicon carbide, a potential material for high frequency and high power applications will be presented.

The optical properties of silicides reflect the electronic structure of these compounds. The possibility of determining the dielectric function from first principle calculations and the comparison with the experimental results can improve the understanding of bonding. Furthermore, mesurements of the optical response allow a direct, non destructive characterization of some important parameters (e.g. the plasma frequency, or the gap energy) related to the transport properties of the material.

The electronic band structure of the semiconducting silicide Ir₃Si₅ has been studied with the LMTO-ASA method. The calculated LDA band gap is indirect and its width 0.78 eV. The minimum direct gap is calculated to 0.86 eV.

The structural and optical properties of β-FeSi₂ precipitates are discussed. Float zone single crystalline silicon substrates were implanted at 250 °C with iron ions to fluences of 1 to 5 × 10¹⁵ cm^-2 and annealed at 800 °C for several hours. Transmission electron microscopy revealed the formation of a band of small, highly strained precipitates just below the surface and a band of large, relaxed precipitates at a depth close to the anticipated ion range. These latter are shown to be the cause of the 1.54 μm luminescence. The extremely long lifetime of the luminescence decay (60 μs at 17 K) indicates that a weak oscillator strength characteristizes carrier recombination in iron silicide precipitates.

Few measurements of the optical properties of Ru₂Si₃ exist even though recent theoretical calculations suggest that this material may be the most attractive silicide for light emission applications. Here we report results of optical investigations of this semiconducting material by spectroscopic ellipsometry, UV-VIS-NIR transmission and reflectivity measurements, and Raman spectroscopy.

Theoretical and experimental data describing electronic band structure and its orbital composition in semiconducting silicides are reviewed and critically assesssed.

A review is given of the thermoelectric properties of semiconducting silicides which have shown a high potential for application in thermoelectric generators and thin film sensors. After an introduction of the main thermoelectric parameters the current understanding will be discussed of Mg₂Si and its alloys, MnSi_-1.73, Ru₂Si₃, ReSi_1.75 and β-FeSi₂. The compound β-FeSi₂ is discussed in more detail including recent results on the carrier mobility and thermopower of single crystals. It will be shown that there is still a potential in some silicides to improve their thermoelectric figure of merit.

Electrical transport properties of binary silicides in the form of thin film and single crystal will be reviewed and discussed. The presentation will be made by considering that most of the compounds behave as metals, with the resistivity, which increases with the temperature, and few of them are semiconductors. Several metallic silicides show, especially in the high-temperature limit, a resistivity-temperature dependence different from the classical linearity. In some silicide the resistivity increases more than linearly and in some other a saturation value at high temperatures is reached. The deviation from linearity, related to the intrinsic properties of the compound, can be affected by the presence of structural defects and impurities. The role played by such defects on the residual resistivity at low temperature will be considered. The tensor nature of the conductivity will also be presented and discussed in relation with the shape of the Fermi surfaces.

⁵⁷Fe conversion electron Mössbauer spectroscopy has been applied to the characterization of bulk stable and epitaxially stabilized iron silicides, and iron-silicon interfaces. The different Mössbauer parameters and their related physical information are discussed in terms of the local electronic surrounding of the iron nucleus. This technique provides also important and unique information on other systems such as ternary transition metal silicides as well as other silicides where the iron nucleus can be used as an efficient probe.

The kinetics of phase formation according to parabolic and linear parabolic laws are analyzed first for the case of single phase growth. The ideas are then extended to the case where two phases grow competitively. A section is devoted to difficult nucleation when the driving force is very small. Special attention is devoted to what is known about the very initial stage of a reaction, and what this means for later stages of growth. Problems such as the nature of the predominant diffusing species, lattice versus grain boundary diffusion, the identity of the so-called first phase to form, and distinctions between thin film and bulk reactions are discussed. In the final chapter, mention is made of "bulk" synthesis from powders, thin film and bulk reactions are compared.

Thermodynamic and diffusion models are given to describe the morphological evolution of the diffusion zone during solid state interactions in binary and ternary silicide systems. In the case of diffusion-controlled process in a temary system, reactions can be interpreted using chemical potential diagrams. However, in some cases a periodic layered morphology is found in the transition zone, which is not fully understood and it is difficult to predict a priori. Silicide formation in systems based on dense silicon nitride and non-nitride forming metals can be explained by assuming a nitrogen pressure build-up at the contact surface. This pressure determines the chemical potential of Si at the interface and, hence, the product phases in the diffusion zone.

The synthesis and the properties of cubic silicides are reviewed, such as NiSi₂ and CoSi₂ with the fluorite structure, which can be grown epitaxially on a large scale. Especially, we focus on the interface and surface structures of CoSi₂/Si(111) and CoSi₂/Si(100) and the corresponding electronic properties. Recent results obtained by scanning tunneling microscopy and ballistic-electron-emission microscopy (BEEM) are shown to yield information about buried interfaces down to an atomic scale. The favourable interface structure is also responsible for the epitaxial stabilization of cubic silicide structures which do not occur in bulk form, but which can also be grown as single crystalline films. They are very common among Fe silicides, and occur even for CoSi₂ as a transition state.

An introduction into ion beam synthesis, a technique capable to grow single crystalline silicides, is given. Ion beam synthesis is a two step process which involves high dose ion implantation into a heated silicon substrate and subsequent high temperature annealing. An alternative technique to grow single crystalline silicides is molecular beam allotaxy, a modified molecular beam epitaxy technique. Both techniques use an additional high temperature anneal to form uniform layers and to anneal the crystal defects. The physics involved in this particular growth process will be explained and experimental examples will be given. Furthermore, a new self-assembled patterning technique for thin epitaxial CoSi₂ layers is presented. Presently, the method allows the fabrication of 50 nm lines in these layers. Some innovative applications of epitaxial silicides are also presented.

CoSi₂ has become a standard material in advanced MOS processes. Several silicidation processes have been investigated and discussed to improve the silicide formation in view of its self-alignment, its electrical performance and its manufacturing robustness. In this chapter the various CoSi₂ technologies are presented and compared.

Three methods have recently been developed to enhance the formation of the low-resistivity C54 phase of TiSi₂, the most widely used silicide contact in microelectronic devices. These methods are:

(1) ion implantation of a transition metal into the Si prior to Ti deposition;

(2) deposition of a thin transition metal interlayer between the Si and Ti;

(3) codeposition of Ti alloyed with a transition metal.

Each of these methods decreases the C49-C54 transformation temperature by more than 100 ºC and improves the probability of phase formation in narrow lines by increasing the nucleation site density. In this paper, we identify the aspects of phase formation which are shared by these three methods, review the methodology by which they were developed, and summarize the applications to silicon devices. Mechanisms responsible for the enhanced formation of C54 TiSi₂ are reviewed, on the basis of a combination of temperature-controlled in situ measurements of resistance, x-ray diffraction and optical scattering, coupled with ex situ studies of phase formation and morphology. The main mechanisms are identified as enhanced nucleation of the C54 phase by a reduction of grain size in the C49 phase and the creation of crystallographic templates of the C40 disilicide phase and the metal-rich Ti₅Si₃ phase.

Metal-silicon compounds have been studied since the first days of I.C. technology from both a fundamental and practical point of view. In particular, silicides of transition and refractory metals are characterized by high electrical conductivity and low contact resistance to silicon. For these reasons they are nowadays widely used for contacts, for source and drain and gate conductors, and for buried interconnections. Most silicides are compatible with the I.C. technology, but they integration in an already defined production flow is not trivial, and frequently requires the introduction of diffusion barriers. In this paper some of the results obtained in the daily characterization work performed for developing and debugging silicide processes and understanding the fundamental physics behind, are presented. The reasons for the noticeable time gap between the publication of a paper on a new silicide, and its incorporation in a commercial device is also explained, taking into account the unavoidable limitation imposed to the experimental conditions by a very complex process flow.

Micro-Raman spectroscopy is a useful technique to identify the two phases of TiSi₂, namely, C54 and C49. It allows one to identify the local crystallographic phase even in submicron patterns, and provide information about the formation of the C54 phase: density of nucleation sites and growth rates. More recently, a new method based on polarized Raman measurements on the C54 single-crystal has been developped; it provides images of the C54-TiSi₂ polycrystalline film microstructure. In this paper the main features of the micro-Raman imaging technique, and its applications to TiSi₂ are presented.

At room temperature, the stress in thin films of silicides formed by solid state reaction is mainly tensile due to a considerable mismatch in thermal expansion coefficients between the silicide and the silicon substrate. During solid state reaction, the few published results report a compressive stress, which can be explained by the volume change between the silicide and either the metal, or the silicon. A tensile stress has nevertheless been observed in the epitaxy of CoSi₂ and NiSi₂ on silicon. The intrinsic stress during solid state reaction has been obtained from in-situ measurements of substrate bending. The measured overall force in a stack of metal and silicide layers provides evidence of stress relaxation during isothermal experiments. A qualitative model has been proposed to describe the experimental observations with a reasonable agreement. More recent preliminary in-situ XRD obtained from the Pd/Si system are also presented. They seem to confirm previous curvature measurements.

A popular and long-standing view of the Schottky barrier height (SBH) has been based on the concept of Fermi level pinning. A charge due to interface/surface states was thought to give rise to the necessary interface dipole which renders the SBH insensitive to structure. Such a view is slowly yet steadily falling out of favor as more and more results from epitaxial interfaces, particularly epitaxial silicide interfaces, become available. The now firmly established relationship between the interface structure and the SBH not only challenged the basis of Fermi level pinning but also brought forth the issue of barrier height inhomogeneity. It was through the study of the latter subject that a coherent picture of the formation of the SBH began to emerge. In these lectures, basic concepts, old and new, of the SBH will be reviewed. Experimental results which changed our views of the SBH will be highlighted. The SBH inhomogeneity issue and its impart will be explained. And, finally, the present status and future prospects of SBH research will be discussed.

The traditional Fowler plot of detector's photoyield is found to be nonlinear for values close to the barrier height of PtSi-Si junctions. A model that takes all scattering mechanisms into account provides a successful description for the experimental results. The photoemission spectrum is independent of temperature in agreement with this model. The detector's atmosphere influences the photoemission spectrum for wavelength of 3 μm or less. A strong loss in the detector's response at 3 μm results from ice formation on the detector's surface during cooling; it can be avoided by using a more effective pumping system.

An overview of the structure and oxidation behavior of metal-rich silicides of the A₅Si₃-type is presented. In particular, bonding, crystal structure, and oxidation resistance at elevated temperature of these silicides modified by the addition of small amounts of light elements are addressed. A₅Si₃-type silicides and light element doped derivatives occur in three distinct structure types: Mn₅Si₃, W₅Si₃ and Cr₅B₃. The relationship between the structures and the role of light elements in the types of structures stabilized are discussed. Oxidation behavior of select compositions at elevated temperatures is presented. Boron doping of Mo₅Si₃ and carbon doping of Ti₅Si₃ results in dramatic improvements in oxidation resistance of these materials, whereas the doping effect for V₅Si₃ and Y₅Si₃ appears to be ineffective. The oxidation behavior of these materials is dependent on many factors and potential improvement in oxidation resistance by light element modification is not easily predicted.

The literature on the epitaxial growth of CoSi₂ via solid-state interaction between Co films and Si substrates is reviewed. Different mechanisms are discussed. Both thermodynamic and kinetic factors are important for the epitaxy of CoSi₂ to occur. Reduction of atomic mobility coupled with minimization of total free energy favor the growth of CoSi₂ in accordance with the substrate orientation.

CoSi₂ layers, 30 nm thick, were prepared from Co films deposited on Si and reacted either at 650ºC or 800ºC . The morphological stability of these layers during annealing at 850ºC or 1150ºC was monitored by sheet resistance measurements. In agreement with models proposed by others, the layers with the smaller grains display better stability

The following sections are included:

• AUTHOR INDEX

• SUBJECT INDEX