Nanoporous Methylsilsesquioxane films are a leading candidate for low dielectric constant (low-κ) materials for microelectronic interconnect. Mechanical strength reduces rapidly with lower κ (increasing porosity) however, and there is a lack of techniques to characterize these properties in the κ∼2 range. This work reports application of surface acoustic wave spectroscopy and Brillouin light scattering to characterization of density/porosity and Young's modulus values of a range of Methylsilsesquioxane films from different manufacturers. We show that the results are validated by independent measurements and that nanoindentation measurements consistently overestimate stiffness. The behaviour of 3 sets of films show different stiffness-porosity relationships, the initially stiffer materials declining more rapidly with increasing porosity than the softer materials. This has important consequences for stiffness properties in the κ<2 region.

Presented theoretical paper concerns the investigation of SiGeC/Si heterojunction by the Grazing-angle Incidence X-ray Diffraction (GIXD) method. We consider a possibility in principal of the GIXD by the specific long-range harmonic variations of the germanium and carbon compositions in the thin SiGeC layer. Evaluation of the theoretically calculated coherent part of x-radiation scattered by the SiGeC layer points the way to the experimental direct investigations of the long-period structured intermediate transformation states of SiGeC layer that emerge owing to inhomogeneity of the strain field along the heterojunction surface.

Mechanical stress develops in silicide interconnection structures due to the volumetric shrinkage of newly grown silicide film during silicidation. Silicidation-induced stress of about 1 GPa was measured by inducing a reaction between a silicon wafer and a titanium film deposited on the wafer. The stress developed near the interface between the grown silicide film. The remaining silicon was analyzed using a finite-element method. The critical stress for delamination at the interface was determined by comparing the results of the estimated stress at the interface with the cross-sectional observation results of different interconnection-structures test samples. We also determined the critical thickness of TiSi2 and the diameter of silicide contacts for eliminating delamination at the interface.

We have shown previously that the dielectric constants of PECVD prepared SiCOH dielectrics can be extended to ultralow-k values of k=2.0. The reduction in the dielectric constants has been achieved by adding an organic precursor to the tetramethylcyclotetrasiloxane (TMCTS) used for the preparation of the SiCOH dielectric and annealing the films to remove the thermally less-stable organic CHx fractions from the films, thereby adding porosity and reducing the density of the films. To assess the effects of the reduction of the dielectric constant on other physical properties of the material, the density and composition of the films have been determined by RBS and FRES and the porosity has been measured by PAS/PALS, SAXS and ellipsometric porosimetry. In addition the films have been characterized structurally and topologically by TEM and AFM.

It has been found that addition of the organic precursor reduces the Si fraction in the films, however, there is no direct correlation between dielectric constant and film composition. The dielectric constant and density decrease with increasing porosity in the films, which reaches values of 30-39% for k values of 2.05. The pore size increases with decreasing k, however the diameter remains below 5 nm for k=2.05. This is significantly smaller than the pore size typically found in porous spin-on films and may provide an integration advantage compared to spin-on films having similar k values.

We present ab-initio modeling results including formation, migration, and activation energies for B diffusion through bulk and grain boundaries in polycrystalline HfO2 films. Modeling results clearly indicate that B can penetrate through a 40 Å HfO2 film via grain boundary diffusion, but not by bulk diffusion. SIMS analysis of B concentration profiles for polysilicon/HfO2/Si gate stacks after different anneals showed double B peaks at the interfaces and thus confirmed the modeling prediction.

In this work, non-volatile memory based on silicon nano trap was implemented by implanting silicon into oxide-nitride-oxide layers(ONO) on silicon. The charge storage effects in these memory structures were measured using capacitance-voltage techniques in polysilicon-ONO- silicon structures in terms of silicon implantation dose. Without silicon implantation, the devices did not the threshold shift (memory window). The memory window was found to be dependent on silicon implantation dose and DC bias (programming) voltage. The memory window did show any degradation by annealing in nitrogen or forming gas environment at 450 °C for 30 minutes. Nano trap memory elements were fabricated with 0.35 micron technology showed encouraging results up to 1e5 program/erase cylces.

The periodic heating is studied using hyperbolic heat wave propagative equation complex variables. The flux reversal is observed. Fin assemblies may reduce the heat accumulated.

In this work, Ru and RuO2 films have been investigated for potential applications in emerging CMOS gate electrode and memory capacitor bottom electrode applications. Films were deposited on SiO2 using chemical vapor deposition (CVD) and low power plasma assisted CVD (PACVD) in a 200-mm wafer deposition cluster tool using a metal beta-diketonate precursor [Bis (2,2,6,6-tetramethyl-3,5-heptanedionato) (1,5-cyclooctadiene) ruthenium (II)]. Hydrogen and oxygen were employed as the reactive gases to deposit, respectively, Ru and RuO2, over a wafer temperature range from 320°C to 480°C. The resulting film properties were analyzed using cross-sectional scanning electron microscopy (CS-SEM), four point resistance probe, x-ray photoelectron spectroscopy (XPS), Rutherford backscattering spectrometry (RBS) and x-ray diffraction (XRD). Both Ru and RuO2 films could be deposited with minimal carbon concentration (∼5 at. %). The purity of the films was also reflected in the as-deposited resistivity of the films, which was as low as 47 μΩ-cm, and was strongly dependent on processing conditions. In order to assess thermal stability, the films were subsequently annealed in forming gas and oxygen ambients for 60 min at 650°C. It was observed that, generally speaking, CVD RuO2 films were stable, with respect to resistivity, in oxidizing ambients, while annealing in a reducing ambient resulted in significant film densification and reduction of the film resistivities to as low as 43 μΩ-cm. Ru films demonstrated good adhesion after anneals in oxidizing, but not in reducing ambients.

To extend the scaling limit of thermal SiO2, in the ultra thin regime when the direct tunneling current becomes significant, members of our group embarked on a program to explore the potential of silicon nitride as an alternative gate dielectric. Silicon nitride can be deposited using several CVD methods and its properties significantly depend on the method of deposition. Although these CVD methods can give good physical properties, the electrical properties of devices made with CVD silicon nitride show very poor performance related to very poor interface, poor stability, presence of large quantity of bulk traps and high gate leakage current. We have employed the rather newly developed Hot Wire Chemical Vapor Deposition (HWCVD) technique to develop the a:SiN:H material. From the results of large number of optimization experiments we propose the atomic hydrogen of the substrate surface prior to deposition to improve the quality of gate dielectric. Our preliminary results of these efforts show a five times improvement in the fixed charges and interface state density.

Thermally grown Si3N4 films in NH3 are known to have a higher dielectric constant and a higher N concentration than silicon oxynitrides, although they incorporate hydrogen atoms that induce hot electron carriers during subsequent high temperature processing. Further, a silicon nitride is difficult to grow over about 6 nm thick, due to self-limiting growth. One alternative is SiOxNy post-nitrided with NH3.

In this work, we study the scope of improvement of Ar annealed nitrided oxynitrides as a function of annealing temperature and duration. Secondary ion mass spectroscopy (SIMS) studies of the nitrogen and hydrogen profiles suggest increasing N and H removal with increasing annealing time and temperature. Electrical characterizations have been performed to determine the total charge (Qox) and interface trap (Dit) densities at different processing conditions, before and after the annealing step. Post-annealing steps are not found to yield improvements of the electrical properties of these dielectric films. Instead, sometimes Qox is even seen to increase (e.g., after a 30 min Ar anneal at 1000 °C). Therefore, an optimization of such annealing steps is essential in designing nanodielectrics with desired nitrogen amounts and N concentration profiles as well as in understanding related process-structure-function relationships.

Atomic layer deposition is a promising technique for the deposition of ZrO2 thin films for high-k gate dielectric applications. However, there are a number of problems associated with existing Zr precursors such as ZrCl4 and [Zr(OBut)4]. In this paper, we examine the ALD of ZrO2 using the new alkoxide complexes, [Zr(OBut)2(dmae)2], [Zr(OPr1)2(dmae)2] and [Zr(dmae)4] (dmae = [OCH2CH2NMe2]), and compare the results with data obtained using [Zr(OBut)4].

An Mo gate work function control technique which uses annealing or N+ ion implantation has been reported by Ranade et al. We have fabricated Mo-gate MOS diodes, based on their report, with 5-20 nm SiO2 and found that the gate leakage current was increased as the N+ implantation dose and implantation energy were increased. Although a work function shift was observed in the C-V characteristics, a hump caused by high-density interface states was found for high-dose specimens. Nevertheless, a work function shift larger than -1V was achieved. However, nitrogen concentration at the Si surface was about 1x1020 cm-3 for the specimen with a large work function shift.

We present experimental results regarding the thermodynamic stability of the high-k dielectrics ZrO2 and HfO2 in contact with Si and SiO2. The HfO2/Si interface is found to be stable with respect to formation of silicides whereas the ZrO2/Si interface is not. The metal oxide/SiO2 interface is marginally unstable with respect to formation of silicates. Cross-sectional transmission electron micrographs expose formation of nodules, identified as silicides, across the polysilicon/ZrO2/Si interfaces but not for the interfaces with HfO2. For both ZrO2 and HfO2, the X-ray photoemission spectra illustrate formation of silicate-like compounds in the MO2/SiO2 interface.

Inserting a titanium layer between cobalt and silicon can control the phase formation sequence for cobalt silicide. In this study we show that sputtered Ti combines with surface oxide to form a stable permeable membrane <20Å thick. Below 500°C, the restricted diffusion of Co and Si through this interlayer allows Co2Si and CoSi2 to form above and below the interlayer, respectively. Because the interlayer is very thin, the type of substrate doping dominates the reaction sequence at higher temperatures. Given sufficient cobalt supply, at 500-600°C the initial phases transform to CoSi for an n-type Si substrate. For a p-type substrate the Co2Si phase converts to CoSi while the bottom CoSi2 phase remains stable. Above 650°C, the deposited cobalt fully converts to the final CoSi2 phase.

Tantalum Pentaoxide, an alternative to SiO2, as a high-k dielectric for DRAM and MOS applications, faces the problem of interface mismatch at silicon. SiO2 or Si3N4 interfacial layer could help in overcoming this problem. The higher band offsets of these materials also help in the reduction of leakage currents at low electric fields. Here we study the physical and electrical characteristics of Ta, oxidized in O2:NH3 ambient, and without any other interface layer. This is done to check if N/H moves to the interface, and thus improves the electrical properties. XRD studies of the film, showed the presence of Ta2O5. Peaks corresponding to TaSi2, un-oxidized tantalum and TaN were also found in the film. But the intensity of these peaks decreased with the reduction of NH3 content. Thus a higher oxygen content could reduce the content of TaN and unoxidized tantalum. FTIR analysis however showed strong Ta=O and Si-O peaks. For the MOS capacitors, due to the presence of resistive components, the maximum capacitance was reduced, compared to that of pure Ta2O5 films. Oxide charges in the films were observed to be around 1.9E10 cm-2. But the traps in these films were found to be almost negligible as observed from the negligible hysteresis in the C-V characteristics. Films with N/H showed lesser oxide charges by an order of magnitude, as compared to pure Ta2O5 films.

Deformation of interconnect structures at the back-end of microelectronic devices during processing or service can have a pronounced effect on component reliability. Here, we use atomic force microscopy (AFM) to study plastic deformation and interfacial sliding of Cu interconnects lines on Si. The behavior of both stand-alone Cu lines and lines embedded in a low K dielectric was studied. Following thermal cycling, changes were observed in the in-plane Cu line dimensions, as well as the out-of plane step height between Cu and dielectric in single layer structures. These were attributed to differential deformation of the Cu/Si and Cu/dielectric material pairs due to thermal expansion mismatch, accommodated by interfacial creep. These results are discussed in light of previous work on the mechanism of interfacial creep. Some preliminary results on the distortion of Cu lines due to package-level stresses are also presented.

A method for removing SiO2 and producing an ordered Si(100) surface using Sr or SrO has been developed. In this technique, a few monolayers of Sr or SrO are deposited onto the as received Si(100) wafer in a ultrahigh vacuum molecular beam epitaxy system. The substrate is then heated to ∼800°C for about 5 minutes, the SiO2 is removed to leave behind a Sr terminated Si(100) surface. This Sr terminated Si(100) surface is well suited for the growth of crystalline high k dielectric SrTiO3 films. Temperature programmed desorption measurements were carried out to understand the mechanism of removing SiO2 from Si(100) using Sr or SrO. The species we observed coming off the surface during the temperature cycle was mainly SiO and O, no significant amount of Sr containing species was observed. We conclude that the SiO2 removal is due to the catalyst reaction SiO2 + Sr (or SrO) → SiO (g) + O + Sr (or SrO). The reaction happened through several intermediate steps. The reaction SiO2 + Si → 2SiO (g) at the SiO2/Si interface is limited and the pit formation is suppressed. The main roles that Sr or SrO play during the oxide removal process are catalysts promoting SiO formation and preventing further etching and the formation of pits in the substrate.

In this work, we report the effects of nitrogen on electrical and structural properties in TaSixNy /SiO2/p-Si MOS capacitors. TaSixNy films with various compositions were deposited by reactive sputtering of TaSi2 or by co-sputtering of Ta and Si targets in argon and nitrogen ambient. TaSixNy films were characterized by Rutherford backscattering spectroscopy and Auger electron spectroscopy. It was found that the workfunction of TaSixNy (Si>Ta) with varying N contents ranges from 4.2 to 4.3 eV. Cross-sectional transmission electron microscopy shows no indication of interfacial reaction or crystallization in TaSixNy on SiO2, resulting in no significant increase of leakage current in the capacitor during annealing. It is believed that nitrogen retards reaction rates and improves the chemical-thermal stability of the gate-dielectric interface and oxygen diffusion barrier properties.

We have successfully grown epitaxial cubic (B1-NaCl structure) tantalum nitride films on Si (100) and (111) substrate using a pulsed laser deposition technique. A thin layer of titanium nitride was used as a buffer medium. We characterized these films using X-ray diffraction, high resolution transmission electron microscopy and scanning transmission electron microscopy (Zcontrast). X-ray diffraction and high-resolution transmission electron microscopy confirmed the single crystalline nature of these films with cubic-on-cubic epitaxy. The epitaxial relations follow TaN(100)//TiN(100)//Si(100) on Si(100) and TaN(111)//TiN(111)//Si(111) on Si(111). We observed sharp interfaces of TaN/TiN and TiN/Si without any indication of interfacial reaction. Rutherford backscattering experiments showed these films to be slightly nitrogen deficient (TaN0.95). High precision electrical resistivity measurements showed excellent metallic nature of these films. We also tried to deposit TaN directly on silicon, the films were found to be polycrystalline. In our method, TiN plays a key role in facilitating the epitaxial growth of TaN. This method exploits the concept of lattice matching epitaxy between TaN and TiN and domain matching epitaxy between TiN and Si. We studied the diffusion barrier properties of these films by growing a thin layer of copper on the top and subsequently annealing the films at 500°C and 600°C in vacuum. Cu diffusion layer was about 2nm after 600°C annealing for 30min. This work explores a promising way to grow high quality TaN diffusion barrier on silicon for copper interconnection.

This paper discusses the issues that arise when ab excess of Si or SiO flux is used to adjust the Deal -Grove model to match the observed growth in the early stages of oxidation. The author uses the insight gained from the modeling of the oxidation of Si-Ge alloys as a background to the discussion. A recent simulation of the thin oxide problem by M. Uematsu et.al. invokes the role of “excess Si” near the interface to adjust the D&G reaction coefficient and also add additional flux to the growing oxide.This paper assesses some of the points in this model as well as the relation with sub bonded Si and/or the reaction barrier layer to oxidationmechanism. A recent proposal by Pasquarello et.al opens new insight into the way the interface may adjust to the growing oxide which challenges our previous mechanistic notions.