The first part of this paper constitutes a summary account of reactive phase formation as seen from a purely kinetic point of view. In the longer second part a more detailed analysis of phase formation based on microscopic, and calorimetric observations is presented. It is shown that these phenomena also tend to be dominated by kinetic factors. This explains why purely kinetic considerations (which should always prevail in any case at some point) usually provide a fairly accurate account of the growth of new chemical phases forming on a solid substrate, regardless of the physical state (solid, liquid or gas) of the second reactant.

Titanium silicides were prepared by coevaporation of Ti and Si on Si substrates at intermediate substrate temperatures followed by high temperature annealing. Depending on the deposition conditions, transmission electron diffraction analyses revealed two different halo patterns from the as-deposited samples. Variations in the deposition conditions included substrate temperature, deposition rate, and film thickness. Radial distribution functions were calculated to estimate the short range ordering of the amorphous phases. The interatomic distances of all the titanium silicide compounds were also calculated in order to compare them with the atomic ordering of amorphous phases. Phase transition from these amorphous phases to the first crystalline silicide is discussed in terms of kinetic variations as well as the atomic ordering.

The stability of C54 Ti(Si1−yGey)2 films in contact with Si1−xGex substrates was investigated. The titanium germanosilicide films were formed from the Ti − Si1−xGex solid phase metallization reaction. It was observed that Ti(Si1−yGey) 2 initially forms with the same germanium content as the Si1−x Gex substrate (i.e., y = x). Following the initial formation of TiM2 (M = Sil−yGey), silicon and germanium from the substrate diffuse into the TiM2 layer, the composition of the TiM2 changes, and Si1−z Gez precipitates form along the TiM2 grain boundaries. The germanium content of the Ti(Sil−y Gey)2 decreases, and the Sil−z Gez precipitates are germanium rich such that y < x < z. This instability of the TiM2 film and the dynamics of the germanium segregation were examined using the Ti-Si-Ge ternary equilibrium diagram. The relevant region of the ternary diagram is the two phase domain limited by a Si-Ge solid solution and a TiSi2 − TiGe2 solid solution. In this study first approximation Ti(Sil−y Gey)2 -to- Sil−xGex tie lines were calculated on the basis of classical thermodynamics. The tie line calculations indicate that for C54 Ti(Sil−yGey)2 to be stable in contact with Sil−xGex, the compositions of the two phases in equilibrium must be such that y < x. The specific compositions of the two phases in equilibrium depend on the temperature and the relative quantities of the two phases. The dynamic processes by which the Ti(Si1−yGey)2/Si1−x. Gex, system progresses from the as-formed state (y = x) to the equilibrium state (y < x) can be predicted using the tie line calculations.

Interactions between Ti and SiO2 thin films have been studied in self-aligned transistor process in the MOS Integrated Circuits Technology. A thermodynamic study of this interaction was conducted on the titanium and silicon dioxide chemistry. The Gibbs free energy was analyzed in the 25 to 1000°C range and it was concluded that the 11/3 Ti + SiO2 --> 1/3 Ti5 Si3 + 2 Ti reaction has the lowest free energy. Ti thin films were deposited by sputtering over dry silicon oxide and then sintered in atmospheric pressure furnace - RTP at argon ambient. The samples were analyzed by X-RAY Diffraction (XRD), Rutherford Backscatering Spectroscopy (RBS) and fourpoint- probe resistivity measurements. It was showed that experimental results can be modeled by theoretical analysis.

In this paper we report the cohesion energy curves for different CoSi2 structures calculated by a semiempirical tight binding scheme. We set up a stability hierarchy among them and provide a kinetic model which could explain very recent and apparently contradictory Molecular Beam Epitaxy findings concerning the stability of a new pseudomorphic phase, i.e. the defected CsCl.

Diffusion processes in silicide thin films play a key role both during their formation by reactive diffusion or during their use. However our unique source of information was provided by indirect analysis: growth of thin films and dopant redistribution which are quite difficult to analyse in terms of diffusion mechanisms. Recent tracer experiments conducted in bulk silicides are presented. They allow a determination of both volume (v) and grain boundary (gb) diffusion coefficients. Contrary to what is observed in certain intermetallic compounds no fast volume diffusion mechanism was found. The main difference with the behaviour of pure metals is a slightly higher value of the ratio Qgb/Qv which makes gb diffusion an efficient process in a wider temperature range.

A quantitative analysis of diffusion processes during silicide formation is then possible. As an example we propose a comparison between the kinetics of growth of thin films and bulk diffusion couples in the Co/Si and Ti/Si systems. Providing that attention is paid to: i) the laws of growth which are slightly different for a phase growing simultaneously with others (bulk) and one phase growing alone (thin films), and ii) the grain size of the growing phase which is strongly dependant on temperature and thicknesses excellent agreement is obtained between the two sets of measurements. Moreover the growth rates may be calculated quite accurately from the values of the volume and gb tracer diffusion coefficients. This stresses and quantifies the role of interfacial diffusion in thin films behaviour.

The lattice diffusion of arsenic in CoSi2 has been studied in the temperature range from 750°C to 950°C. Two types of bulk samples were used: single crystals prepared by a modified Czochralski pulling technique from a radio frequency levitated melt and polycrystals synthesised by quenching from the melt. The latter samples were subsequently annealed in vacuum at 900°C and displayed grain sizes in the millimetre range. Starting from an ion implanted arsenic profile with two different doses (5·1014 and 5·1015 cm−2) the concentration versus depth profiles after annealing at different temperatures and different times were measured using secondary ion mass spectrometry (SIMS). Contrary to previous studies by other authors substantial diffusion has been observed with an activation energy of 3.3 eV and a pre-exponential factor of 7.37 cm2/s for the diffusion coefficient. These values are very close to the self diffusion coefficient of Si in CoSi2 suggesting that the As atoms migrate via thermal vacancies on nearest neighbour lattice sites by a similar type of mechanism as the Si (and Co) atoms. In the high dose implanted polycrystalline samples arsenic precipitation occurred which gives an estimate for the solid solubility in the 1019 atoms/cm3 range at 800 °C.

It has been observed that the sheet resistance of a Ti-salicided polysilicon-gate electrode or source/drain region increases significantly as the dimension reaches the lower sub-micron range. The resistance of platinum and nickel silicide (PtSi and NiSi), however, does not increase with reduced linewidth. We have studied PtSi and NiSi films with deep sub-micron linewidths on single crystal or poly-Si substrates. In this study, the material properties such as sheet resistance, grain structure and surface morphology of these silicide films in confined geometries are reviewed and compared with TiSi2. Process windows for forming and maintaining these silicides are explored.

The formation and the relaxation of NiSi2 films with and without Au are examined by scanning electron microscopy, X-ray diffraction and Rutherford backscattering spectrometry. We studied the solid state reactions between a Ni(7 at.% Au) thin film and a Si substrate which occurs during the solid phase epitaxy before the formation of NiSi2. We show that the addition of Au to the Ni film drastically affects the silicides formation: Ni2Si and NiSi appear simultaneously and the nucleation temperature of NiSi 2 is lowered. The solubility of Au in the three silicides is limited which induces a precipitation of Au. Depending on temperature this precipitation takes various forms: Au enriched surface layer or Au clusters at inner interfaces. The films lattice parameters both parallel and perpendicular to the interface are also measured and compared to the lattice parameters of bulk samples which have been made by solidification from the melt. The relaxation modes are deduced from these measurements.

Decreasing contact dimensions coupled with the need for planarization to accommodate multiple layers of metal have created many challenges for the contact etch module. For example, contact etch processes are often required to stop on thin titanium silicide while at the same time forming high aspect ratio, straightwalled contacts. In this paper, the impact of various dielectric compositions and contact etch process parameters on etch profile, selectivity, and contact resistance is presented for the formation of high aspect ratio, submicron contacts to thin TiSi2 layers. The etch profile is formed by RIE using a mixture of CHF3 and various amounts of CF4. Surprisingly, the sidewall angle and selectivity to silicide showed little dependence on the percent CF4. Contact resistance measurements, however, varied greatly with percent CF4 and contact aspect ratio. The variation of contact resistance with etch chemistry was attributed to a variation in the extent of fluorocarbon polymer film formation, which in turn depends on the ratio of carbon to fluorine in the plasma. Finally, post contact etch treatments were examined for efficiency in removing the polymer films from the high aspect ratio contacts.

The reaction between sputtered Ti thin films and heavily arsenic doped Si(100) is studied. The use of an arsenic implantation to pre-amorphize the Si substrate and the choice of the substrate temperature during Ti sputtering are both found to have a significant effect on subsequent TiSi2 reactions. Cross-sectional transmission electron microscopy reveals that an amorphous TiSix layer is formed at the interface between Si and as-sputtered Ti. The thickness of this interfacial layer increases with the sputtering temperature. After rapid thermal anneals in nitrogen, the sheet resistances of TiSi2 thin films grown with the pre-amorphization step and a high sputtering temperature (450°C) are generally lower than films processed under other conditions. This apparent reduction in the temperature for the polymorphic C49 to 54 phase transformation in TiSi2 is shown to originate from a higher nucleation density of the C54-TiSi2 phase. These dependencies of the silicide reaction are ascribed to the interfacial amorphous TiSix layer. In increasing the nucleation density of the C54-TiSi2 phase, the amorphous TiSix layer is speculated to either act as a direct nucleation source for the C54-TiSi2 phase, or lead to more defective C49-TiSi2 structures which facilitate the C54-TiSi2 nucleation.

To improve Ti SALICIDE process, Si preamorphization by arsenic before Ti sputtering has been studied in two parts: process characterization and fundamental studies. Sheet resistance (Rs) reduction by the preamorphization is more pronounced on thinner and narrower-line silicide formation. At 60keV implantation energy, there is an optimum arsenic dose for the improvement. Through the treatment, more uniform silicide layer can be formed, indicated by the improved Rs uniformity. In the fundamental study, preamorphization appears to have little effect on promoting C49-to-C54 phase transformation. It is suggested that the treatment is able to enhance the reaction rate between Ti and amorphous Si, and results in C54-TiSi2 with larger grains and consequently slightly lower resistivity.

The effects of small concentrations of metallic impurities have been studied in conjunction with the formation of titanium disilicide. We report that, by introducing small quantities of a refractory metal such as molybdenum or tungsten at or near the titanium/silicon interface, the temperature required to form the C54 phase TiSi2 can be reduced by as much as 100°C. Furthermore, the resulting C54-TiSi2 film exhibits small (∼ 0.2μm) grain size and improved thermal stability. This discovery has the potential to reduce the complexity and cost associated with forming low resistivity TiSi2 on submicron structures and to significantly improve the titanium silicide process window for future sub-half-micron VLSI applications.

The direct nucleation and growth of Ti silicide on the surfaces of Si(100) and amorphous Si were studied. Silicide phase formation depended on the temperature and the stoichiometry of deposition and the crystallinity of the substrate. A very low temperature, − 500°C, for the nucleation of the low-resistivity C54-TiSi2 phase was observed on amorphous Si. Stoichiometric and uniform TiSi2 layers were grown with the depositions of pure Ti. On crystalline Si, uniform TiSi2 layers were also grown at ∼ 500°C with a co-deposited template layer. The much reduced C54 formation temperature is discussed in terms of a possible circumvention of precursor amorphous silicide phases during surface nucleation.

In this work the conditions of forming a bi – layer structure of TiN/TiSi2 thin film on Si (100) substrate is investigated. Two methods of producing this structure were used: a) Deposition of Ti on Si (100), followed by reactive sputtering to obtain TiN on top of this layer and b) codeposition of Ti and Si on Si (100) and then deposition of TiN by reactive sputtering. The reactive sputtering was carried out in a mixture of N2/Ar with 20% N2. This amount is believed to be optimal for obtaining good quality and stoichiometric TiN films. Annealing is essential to form TiSi2 and it was performed either in the sputtering chamber immediately after the deposition or by rapid thermal annealing (RTA). The structure of the specimens was analyzed by X-ray diffraction using step scanning, Auger electron spectroscopy (AES) and transmission electron microscopy (TEM). TEM analysis was done on cross sectional specimens and also electron diffraction results were recorded. Resistivity measurements were performed by four – point probe method. The influence of TiN on the silicide formation was established. The results indicate that in the presence of TiN the phase TiSi2 was obtained, but in its absence Ti5 Si3 is formed under the same conditions of deposition and annealing. The stress distribution was investigated by Hall – Williamson curves and it was found that TiN stabilizes the silicide film and no peeling was observed. The effectiveness of TiN as diffusion barrier against Al and Si penetration was tested at 500°C/lh. It was found, that under these conditions, the TIN/TiSi2 structure is about the same, as before the heat treatment. No Al penetration is observed.

The kinetics of the C49 to C54 TiSi2 phase transformation in nitrogen ambient have been investigated in a temperature range from 700 °C to 800 °C for a range of titanium film thicknesses (135 Å to 350 Å) using sheet resistance measurement, Auger electron spectroscopy(AES), Rutherford backscattering spectroscopy(RBS) and transmission electron microscopy(TEM). About 80% of the titanium converts to titanium silicide with the rest converting to titanium nitride. The activation energies obtained for the C49 to C54 transformation in nitrogen ambient are lower, at least by 2–2.25 eV, than that obtained for the transformations occuring either in argon ambient or vacuum environment. This has been explained with a model involving stress state of titanium silicide film with titanium nitride overlayer.

In this work we investigated the influence of the polycrystalline silicon substrate on the titanium silicide formation process. The results are compared to those obtained when single crystal silicon wafers are used. We observed that the polycrystalline substrate affects both the kinetics of formation of the phase TiSi2-C49 and the temperature of transition from TiSi2-C49 to TiSi2-C54. The temperature of this phase transition is also influenced by the presence of phosphorous in the polycrystalline silicon substrate. In order to prevent degradation of the silicide film formed on polycrystalline silicon, the maximum temperature of RTP has to be lower than 900°C. Due to the high chemical reactivity between titanium and oxygen present in the ambient, we also investigated the use of a protective cap. For this purpose, a thin amorphous silicon layer was sputter-deposited sequentially on Ti films without breaking the vacuum.

The reaction between an amorphous silicon film, with a titanium film, yields the formation of a titanium silicide film useful in the fabrication of local interconnection lines (straps) for integrated circuits. In this work we look at the influence of the films thickness ratios on the silicide formation on silicon dioxide. For structures with 70 or 80 nm of amorphous silicon deposited on 40 nm of titanium, we found that after a low temperature treatment of 650 °C, nominally utilized in the first processing step for the formation of self-aligned silicides, the formation of the metastable TiSi2 (C 49) and Ti5Si3 phases occurs. For silicon rich structures i.e.. 90 nm of amorphous silicon on 40 nm of titanium, besides the stable TiSi2 (C54) the metastable phases mentioned above are also detected. In this case strong morphological changes, and silicon precipitates are observed.

Bilayer of TiB2/TiSi2 was deposited by magnetron co-sputtering on silicon and alumina substrates, and this structure was investigated for structural and electrical properties. Substrate bias and annealing in vacuum have been applied to vary the film properties. X-ray diffraction (XRD) and cross-sectional transmission electron microscopy (XTEM) were used to characterize the structure, and chemical composition was characterized by Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS). Resistivity was measured by four probe method. Diffusion barrier properties were studied by AES. As deposited films are amorphous with resistivities of about 40 μΩcm. Post deposition annealing in vacuum shows that the amorphous titanium boride film is very stable. Crystallization starts above 1000°C as seen by XRD, and the crystallization temperature depends on the thickness of TiB2. TiSi2 C54 forms in the temperature range 586°C - 922°C, when TiB2 still remains in amorphous form. The TiSi2 sublayer serves as an additional effective diffusion barrier, preventing outdiffusion of boron from TiB2 into the Si substrate.

Detailed analysis of Ti and/or TiSi2 islands growth have been made by UHV-STM observations after Ti deposition and subsequent annealing. It is shown that islands growth mode changes drastically at about 500 U for both cases on Si(111)-7×7 and on H-terminated Si(lll)-l×l. In the temperature regime higher than 500 °C, activation energies of islands growth are 1.12eV and 0.56eV for Si(M11)-7×7 and H-terminated Si(111) respectively. It is speculated that residual H-atoms combined with Si dangling bonds lowered surface diffusion activation energy.