In this paper we present a low-energy ion scattering (LEIS) study of the site exchange of Fe adatoms with Ag atoms from the Ag(100) surface. The time-of-flight (TOF) spectra obtained at low temperatures have been interpreted with a newly developed LEIS simulation program MATCH. After low temperature deposition (˜50 K) of Fe atoms oil the Ag(100) surface, the Fe atoms occupy adatom positions. These Fe adatoms exchange sites with Ag atoms from the first layer, starting at a temperature of 130(10) K.

In situ transmission electron microscope (TEM) observations of TiSi2 C49-to-C54 phase transformations were recorded on video tape and photographic negatives, and transformation front velocities (vTF) were measured from the data. The samples studied in this work include: C49-TiSi2 blanket films and 0.2 μm wide, 7 mm long lines on undoped poly-Si and 0.13 μm wide, 10 μm long lines on either B- or As-doped poly-Si. The in situ TEM analysis showed that blanket TiSi2 films on undoped poly-Si fully transformed at 830°C and had an average vTF of 0.5 ± 0.2 μm/s. The transformation occurred from a sparse nucleation density of ˜0.1 site/μm2. The 0.2 μm wide lines transformed at temperatures greater than 885°C, and the average vTF was 1.1 ± 0.2 μm/sec. Agglomeration started for both the blanket film and the 0.2 μm wide lines at temperatures above 900°C. In situ x-ray diffraction (XRD) analyses of C49-TiSi2 on B- and As-doped poly-Si showed that blanket films completely transformed to C54-TiSi2 at T = 835–843°C, but 0.13 μm wide, varying length lines did not fully transform, even when rapid thermal annealed to 1025°C. From in situ TEM analysis of 0.13 μm wide, 10 μm long C49-TiSi2 lines on B-doped poly-Si, a distinct transformation was not observed. Instead, the lines slowly agglomerated, and electron diffraction of the agglomerated regions showed that the film had transformed to C54-TiSi2. An individual line either 1) completely transformed and agglomerated or 2) remained as C49-TiSi2 and did not agglomerate. Approximately 85% of all lines transformed. In situ TEM analysis of 0.13 μm wide, 10 μm long lines of C49-TiSi2 on As-doped poly-Si also showed an indistinct transformation to the C54 phase along with agglomeration, and 100% of the lines transformed.

Silicide reaction between co-deposited TiSix layers and Si was studied. It was shown that an amorphous TiSix layer, with x − 0.5–1, deposited between amorphous Si and the Ti film led to a significant reduction in the observed C49→C54 TiSis transformation temperature. Amorphous TiSix layers generated by the pre-deposition of Ti and pre-annealing also enhanced the nucleation of the C54-TiSi2 phase. The efficacy of the interfacial TiSix layer was demonstrated on undoped and heavily doped n- and p-type α-Si, but was found to be absent on single crystal Si. Possible mechanisms of these phenomena are discussed.

The thermal stability of thin CoSi2 films formed from four layered structures: Co/Si, TiN/Co/Si, Co/Ti/Si and TiN/Co/Ti/Si has been studied using four-point-probe (FPP), TEM and RBS/channeling. It is found that thermal instability can be influenced by the polycrystalline structure in CoSi2. Growing epitaxial CoSi2 film on Si (100) using TiN capping and Ti interfacial layers can improve the thermal stability of thin CoSi2 film.

The TiB2/TiSi2 bilayer is considered as a diffusion barrier in metallization system with Cu. The TiSi2 sublayer serves as a contact and also as an additional diffusion barrier against boron, which outdiffuses from TiB2 at high temperature annealing. The attempts to form TiSi2 by vacuum annealing of TiB2/Ti film, which was obtained by co-sputtering from elemental targets are described. The composition and the structure of the films were analyzed by Auger electron spectroscopy (AES), X-ray diffraction (XRD) and high-resolution cross-sectional TEM (HRXTEM). The Cu/TiB2/(Ti-Si)/n-Si contacts were investigated using current-voltage (I–V) on Schottky diode structures, which were prepared on n-type Si (100). The thermal stability of the TiB2/(Ti-Si) barrier was studied by structural and electrical analysis.

It was observed that the lowest sheet resistance of 5.1 Ω/‪ was obtained after 850 °C annealing for 30 min, however the resulting Ti-Si layer is practically still amorphous and contains only a very small fraction of C54 TiSis in the form of microcrystallites. This layer also contained Ti5Si3 as indicated by XRD. The barrier height of Cu/TiB2/(Ti-Si)/n-Si Schottky diodes is ˜0.6 V and it does not show significant changes in the range 400–700 °C. Electrical monitoring is a very effective tool to detect deterioration of the barrier. No penetration is observed by AES at 700 °C, while the I–V curve shows changes in properties.

ULSI packing density calls for sub-micron line widths, but on n-type polysilicon this can lead to incomplete titanium silicide C49 to C54 phase transformation after a conventional two step rapid thermal anneal (RTA). This study compares three different ion beam amorphization techniques: preamorphization, ion beam mixing and silicide amorphization, aiming at a complete phase transformation for submicron silicide lines. For preamorphization, an arsenic implantation at moderate energies (35–75 keV) was used to amorphize the top layer of the polysilicon prior to the titanium deposition. Ion beam mixing used a high-energy (200 keV) arsenic implantation after the titanium deposition to create an amorphous mix of silicon and titanium. These two methods did, each by themselves, lead to an increased fraction of C54 silicide grains already during the low temperature RTA, and a complete phase transformation during the subsequent high temperature RTA. Both methods lowered the thickness difference between titanium silicide on p- and n-type silicon. Silicide amorphization with 75 keV arsenic or 100 keV antimony, applied before the second RTA, did not significantly improve the silicide phase transformation.

The ion-beam-sputtered polycrystalline SiGe film and its doping properties have been studied. Boron and phosphorus have been doped into the sputtered poly-SiGe film by ion implantation and diffusion. To activate the implanted impurities, both rapid thermal annealing and fiirnace annealing have been used. The electrical measurements show that boron and plhosphorus can be doped into sputtered SiGe films and effectively activated by both ion implantation with post-annealing and diffiision. Hall mobilities as high as 31 cm2/V-s and 20 cm2/V.s have been obtained in B-difflhsed and P-diffused SiGe films, respectively. The x-ray diffraction spectra of the sputtered Sifie filhn show its typical polycrystalline structure with (111), (220) and (311) as the preferential orientations.

We demonstrate that the addition of a molybdenum interlayer between titanium and silicon enhances the formation of C54 TiSi2, without bypassing the formation of the C49 TiSi2 phase. In situ x-ray diffraction analysis during rapid thermal annealing, at a rate of 3 °C/s, was used to study the phase formation sequence of TiSi2 starting from a blanket bilayer of Ti on Mo on a polycrystalline Si substrate. It was shown, as in the case without the Mo layer, that the C49 TiSi2 phase forms first, followed by the C54 TiSi2 phase. The results were similar for undoped or arsenic, boron, and phosphorous doped polycrystalline silicon substrates. The temperature range over which the C49 phase is stable is reduced, on average, by 80 °C. The lower end of the range (appearance of C49) is increased by approximately 60 °C and the upper end of the range (disappearance of C49) is decreased by about 20 0C. The orientation of the C49 phase differs in that both the C49(131) and C49(060) orientations are observed, compared to the case without the Mo layer where only the C49(131) orientation is observed.

We have demonstrated that the formation of C54 TiSi2 on Boron-doped single crystal silicon substrates, under RTA annealing conditions in a Nitrogen ambient, leads to a thicker TiN capping surface layer, thinner silicide layer, higher C49 to C54 transformation temperature and greater interface roughness compared to C54 TiSi 2 formation on undoped single crystal silicon substrates. Titanium films 32 nm thick were deposited on undoped and boron-doped single crystal silicon substrates. The films were annealed at 3 /C/isn nitrogen to final quenching temperatures between 500 °C and 900 TC. Ex-situ four point probe sheet resistance, cross sectional transmission electron microscopy (XTEM), high resolution transmission electron microscopy (HRTEM) and x-ray diffraction (XRD) were used to analyze the resulting TiN on TiSi2 bilayer. The C49 to C54 transformation occurs circa 760 TC and 810 TC for the undoped and boron-doped cases respectively. HRTEM observations reveal a thick 20 nm TIN layer on the C54 TiSi2 film in the boron-doped case but only fine dispersed TiN particles embedded on the top of the silicide in the undoped case. It was observed that the resultant silicide in the boron-doped case was thinner and the TiSi2 /Si(100) interface is rougher. XRD and TEM analysis show that in the boron doped case, there is a preferred C54 (040) orientation compared to a random orientation for the undoped case.

A strategy was developed for controlling hillock formation in thin metal films by controlling the fiber texture to be of a relatively “weak” orientation. Two-dimensional molecular dynamics (MD) simulations were performed to determine the parameter dependencies of texturing under ion beam assisted deposition. Simulations showed that even for film orientations that have a lower number of nearest neighbor surface bonds, the reduction in sputtering rate by ion channeling will favor the growth of the grains aligned with their channeling direction in the direction of the ion beam. Higher energies should result in greater sputtering and a higher surface roughness. Confirmatory experiments were performed by growing Al films using ion beam assisted deposition in which the Ne ion beam was normal to the substrate surface. For all energies above 0 eV/atom, the fiber texture contained a (220) component and, at high normalized energies, the fiber texture was heavily (220) dominated. Subsequent annealing at 450°C for 30 min. resulted in hillock formation in the PVD (physical vapor deposition) condition, a reduction in the hillock density by two orders of magnitude in the 120 eV/atom condition and complete elimination of hillocking above 800 eV/atom. Although the surface roughness increased with ion beam energy as modeled by MD, the surface became smoother during annealing. These results show that the fiber texture can be controlled in a thin metal film in such a way as to eliminate hillock formation, that molecular dynamics simulation is a valuable predictive tool for guiding experiments in the development of thin film microstructures and that ion beam assisted deposition is an effective, practical tool for controlling microstructures of thin metal films.

In this study, the relation between interdiffusion in Al/ TiN thin film couples and diffusion barrier stability has been investigated. The TiN diffusion barrier was deposited by reactive sputtering in an Ar-N2 gas mixture. The stoichiometry of TiN was achieved by varying the nitrogen-to-argon ratio in the sputtering gas. Interdiffiusion occurred after annealing for 30 minutes at 475°C and 575°C. Diffusion processes and interfacial reactions with respect to TiN stoichiometry were investigated via Auger Electron Spectroscopy (AES) depth profiling and X-ray diffraction (XRD). The barrier reliability with respect to the stoichiometry changes was established; nitrogen-deficient TiN films result in a high degree of interdiffiusion and decomposition at annealing temperatures of 475°C and 575°C. AI3Ti and AlN intermetallic compounds were formed at the interfaces. The sheet resistance of Al films was measured by four-point probe method. Resistance increases for all the annealed films were due to interdiffusion between Al and TiN. The degree of interdiffusion was analyzed by using AES and XRD.

The surface morphology of a UV excimer lamp photodefimed palladium activation layer and its effect on subsequent electroless copper deposition have been studied on 96% alumina substrates using an atomic force microscope (AFM). The morphology of the activation layer and the density of the palladium catalyst are important factors affecting the initial stages of the electroless deposition and the microstructure of the copper film. Under activation conditions on which a subsequent electroless copper deposit exhibited good adhesion characteristics, the palladium activation was found to have a nodular structure with a large distribution of palladium particle sizes of the order of 10 nm at grain boundaries and the edges of grain steps, with smaller activation particles observed on smoother areas of the alumina grains. The influence of substrate morphology on the nature of the catalytic sites and thence, on the nature of the subsequently deposited metal was investigated, both at the nucleation stage, and as the autocatalytic electroless deposition reaction proceeded. The recrystallisation of the initial electroless deposit, the morphology of the resulting copper films and its relationship to catalyst particle size are reported. The principal determining factors in the substrate and activation layer morphology which control the microstructure of the electroless copper deposit are discussed.

Platinum(Pt) films were sputter-deposited on Si02/Si substrates under the mixed gas atmosphere of Ar and O2. Under certain deposition conditions, the films were oriented such that the (100) direction is normal to the substrate surface. The formation of the (100) texture was affected by the gas pressure and film thickness. After annealing at 650 °C for 1 hour, (100) oriented Pt films with the resistivity of pure Pt were obtained. The annealed Pt films all passed a tape adhesion test and had no defects such as hillocks or pinholes. The experimental results from this work are presented.

In this study, defect-free Pt films with good adhesion were deposited on SiO2/Si substrates by a two-step magnetron sputtering. This method consists of the first sputtering step using Ar/O2 gas mixture and the second step using Ar. After two-step deposition, an annealing process was followed at 600-1,000 °C in ambient atmosphere. In the first step, oxygen containing Pt films were deposited. Oxygen incorporated in the Pt films completely diffused out during the high temperature annealing. After the annealing process, the film became dense without catastrophic failures such as hillock, pinhole or buckling. Adhesion strength of films produced by this process was good enough to pass a tape test. It is believed that the good adhesion and the observed microstructural evolution are related to the oxygen in Pt films introduced during the first sputtering step. Adhesion, microstructural evolution and the role of oxygen in Pt films are briefly discussed.

Formation and phase transformation of non-BCC δ-A15 Cr thin films prepared by R. F. magnetron sputtering have been characterized. Processing parameters such as working pressure, deposition time and temperature were found to affect the formation of δ-A15 Cr films. Using differential scanning calorimetry, we have demonstrated for the first time that the phase transformation of the δ-A15 Cr phase to the equilibrium c-BCC Cr phase is an irreversible, exothermic, first-order transition. At a heating rate of 10°C/min, the onset and peak temperatures of transformation were determined to be 428°C and 437°C, respectively. Our post-deposition annealing study by X-ray diffraction further verified the occurrence of transformation.

Microstructural development of Fe and Cu in Cu/Fe multilayers of layer thickness 1.5–10 nm prepared on Si, Ge, and MgO substrates by ion beam sputtering has been studied using x-ray diffraction and cross-sectional transmission electron microscopy (TEM). High-angle x-ray results show an fcc Cu structure and a distorted bcc structure in the Fe layers at 5 nm-layer-thickness and smaller, and bcc Fe (110) and fcc Cu (111) peaks in the 10 nm-layer-thickness samples. Lowangle x-ray diffraction indicates that the layers in the samples grown on MgO substrates have a more uniform and smooth layered structure than the multilayers grown on Si and Ge substrates, which results from larger grains in the MgO substrate samples for the same layer thickness. Relationships among growth, microstructure, and interfaces with layer thickness are discussed.

Thin film compliant substrates can be used to extend the critical thickness in mismatched overlayers. A metastability model has been coupled with recent experimental strain relief data to determine the critical thickness of InGaAs epilayers grown on GaAs compliant substrates of variable thickness. The results of this model are also compared to other compliant substrate critical thickness models.

Previous studies of the structure of epitaxial Fe/Pt(001) multilayers have shown that the structure of the Fe layers depends on the Fe layer thickness. For small thicknesses (< 12 Å) the Fe is fcc, while for greater thicknesses (> 22 Å) the Fe has a bcc structure. We speculated that this transition of the Fe structure is a result of intermixing with subsequently deposited Pt. We explore this transition further by examining superlattice satellite peaks around the (111) fundamental reflection, and by using grazing incidence x-ray scattering (GIXS) to examine the structure of epitaxial Fe/Pt(001) multilayers during sputter deposition. The GIXS observations show that the initial growth of Fe is in the bcc structure; thin bcc Fe layers are partially transformed to fcc by subsequent deposition of Pt. The transformation is thermodynamically favorable because the intermixed fcc phase has lower epitaxial mismatch and bulk chemical free energy than the bcc phase.

We have measured strain relaxation and islanding in Ge films grown by molecular beam epitaxy on Si(111) at substrate temperatures between 450°C and 700°C in real time with reflection high energy electron diffraction (RHEED). At 450°C, we observe an oscillation of the surface lattice parameter for the first three bilayers (BL), followed by a sharp 2D–3D growth mode transition, when transmission diffraction features appear in RHEED. The surface lattice parameter then begins to relax at an initial rate of about 0.5%/BL. The mechanisms of island growth and strain relaxation change with growth temperature. At 500°C the surface lattice parameter begins to relax after only 1BL; at 550°C relaxation begins immediately. However, 3D spots do not appear until after 3.5BL at either temperature. The initial rate of strain relaxation decreases with increasing temperature until, at 700°C (when 3D spots never appear), it is only 0.04%/BL. This behavior may be explained by a temperature-dependent roughness length scale, as well as by differences in dislocation nucleation at low and high temperatures. At low temperature, atomic force microscope images show the development of small (1000Å), faceted islands with aspect ratios (height/width) on the order of 0.07. The formation of well-defined facets is inhibited at higher temperatures. At 700°C, islands grow very large (lμm) from the outset, with aspect ratios less than 0.015. These islands cannot thicken much, because dislocations can glide in easily at their edges. The islands grow laterally quickly, and the strain in the “new” islands is not substantially less than that in the “old.” At 700°C, 28% of the Ge/Si misfit strain may be relieved by diffusion.

The high-resolution x-ray diffraction technique was used to determine the entire strain tensor as a function of x-ray penetration depth in a 400 nm W/Ta/W trilayer, which consisted of 150 nm W, 100 nm Ta, and 150 nm W. The strain tensor was calculated in a laboratory reference frame to determine the variation of in-plane strains, εxx, and εyy, and the normal strain εzz as a function of x-ray penetration depth. Two different methods were then used to determine the strains εxx, εyy, and εzz in each layer of trilayer, and both indicated that the strain states in the W layers were fe same, and that the strains εxx and εyy. in the W and Ta layer were the same magnitude. The difference between the Ta and W arose witK the normal strain, εzz, and was due to a Poisson contraction effect. The average residual stresses in the trilayer were determined with the HRXRD data, and also with sin2ψ and a substrate curvature technique, double-crystal diffraction topography. All three techniques indicated that the average biaxial stress in the trilayer was ∼1.0 GPa.