As the semiconductor industries are moving beyond 22 nm node technology, the currently used stacked Ta/TaN diffusion barrier including a copper seed will be unable to fulfill the requirements for the future technologies. Due to its low resistivity and ability to perform galvanic copper fill without a seed layer, ruthenium (Ru) has emerged as a potential copper diffusion barrier. However, its crystallization and columnar nanostructure have been the main cause of barrier failures even at low processing temperatures (300 oC -350 oC). In this study, we have proposed and evaluated three different strategies to improve the performance of the ultrathin Ru film as a diffusion barrier for copper. The first study focused on shallow surface plasma irradiation/amorphization and nitridation of 5 nm Ru films. Systematic studies of amorphization and nitrogen incorporation versus sample bias were performed. XPS, XRD and RBS were used to determine the physico-chemical, crystallization and barrier efficiency of the plasma modified Ru barrier. The nitrogen plasma surface irradiation of Ru films at substrate bias voltage of -350 V showed an improved barrier performance up to 400 oC annealing temperatures. The barrier barely started failing at 450 oC due mainly to nitrogen instability. The second study involved only amorphization of the Ru thin film without any nitrogen incorporation. A low energy ion beam irradiation/amorphization on Ru thin film was carried out by using 60 KeV carbon ions with different irradiation doses. The irradiation energy was chosen high enough so that the irradiation ions pass through the whole Ru thin film and stop in the SiO2/Si support substrate. The C-ion fluence of 5×1016 atoms/cm2 at 60 KeV made the Ru film near amorphous without changing its composition. XRD and RBS were used to determine the relationship between crystallization and barrier efficiency of the carbon irradiated Ru barrier. The amorphized Ru film showed an improved barrier performance up to 400 oC annealing temperatures similar to the plasma nitrided Ru films. The barrier barely began to fail at 450 oC due mainly to crystallization. The third study focused on a study of Al doping of nitrided Ru thin films and their crystallinity with the aim of obtaining a completely amorphous Ru based barrier and stable nitridation. The addition of 4% Al and 14% of nitrogen in Ru produced a near amorphous film. Nitrogen in the film remained stable until the annealing temperature of 450 oC for 10 min in N2 atmosphere. Crystallization growth of the film was inhibited until 450 oC. At 500 oC, the crystallization of the Ru films barely started, but the degree of its crystallization is minimal. The Ru-Al-N film was demonstrated to be an effective diffusion barrier for copper until the annealing temperature of 450 oC and began to fail at 500 oC. The Al doping was shown to stabilize the nitrogen in the Ru thin film barrier inhibiting its crystallization and leading to improved diffusion barrier performance and a gain in processing temperatures of 150 oC -200 oC over the as prepared pure Ru thin film barriers.

Abstract The behaviors of the TaNx barrier layer that placed between the Cu metal and GaAs have been studied by using X-ray diffraction, X-ray photoelectron spectroscopy and scanning electron microscopy. The TaNx and Cu films were deposited on GaAs sequentially with RF magnetron sputter. With a 250 nm thick TaNx barrier layer, the Cu metal can be impeded from reacting with GaAs substrate at 575¢Jannealed for one hour. Within an As or Ga overpressure environment condition, the failure temperature still occurred below 600¢J. The failure of TaNx diffusion barrier layer for preventing the reaction of the Cu and GaAs was originated for the dissociation of the GaAs itself at 580¢J. The outgoing As atoms increased the deterioration speed of the TaNx film and reduced its blocking ability.

SiC is used as the main diffusion barrier in the fuel spheres of the pebble bed modular reactor (PBMR). The PBMR is a modern high temperature nuclear reactor. However, the release of silver from the fuel spheres has raised some doubts about the effectiveness of this barrier, which has led to many studies on the possible migration paths of silver. The reported results of these studies have shown largely differing results concerning the magnitude and temperature dependence of silver being transported through the fuel particle coatings. Results from earlier investigations could be interpreted as a diffusion process governed by an Arrhenius type temperature dependence. In this study, the silver diffusion in 6H-SiC was investigated using two methods. In the first method a thin silver layer was deposited on 6H-SiC by vapour deposition while in the second method silver was implanted in 6H-SiC at room temperature, 350°C and 600°C to a fluence of 2×1016 silver ions cm-2. Finally the effect of neutron irradiation on the diffusion of silver was investigated for the samples implanted at 350°C and 600°C. Silver depth profiles before and after annealing were determined by Rutherford backscattering (RBS). Both isothermal and isochronal annealing were used in this study. Diffusion coefficients as well as detection limits were extracted by comparing the silver depth profiles before and after annealing. The radiation damage after implantation and their recovery after isothermal and isochronal annealing were analysed by Rutherford backscattering spectroscopy combined with channelling. The results of in-diffusion of silver into 6H-SiC at temperatures below the melting point (960°C) using un-encapsulated 6H-SiC samples with 100 nm deposited silver indicated no in-diffusion of silver; however, disappearance of silver occurred at these temperatures. For the encapsulated samples, no in-diffusion of silver was observed at 800°C, 900°C and 1000°C but silver disappeared from the samples’ surface and was found on the walls of the quartz glass ampoule. This disappearance of silver was established to be due to the wetting problem that existed between silver and SiC. The room temperature implantation resulted in a completely amorphous surface layer of approximately 270 nm thick. Epitaxial re-growth from the bulk was already taking place during annealing at 700°C and the crystalline structure seemed to be fully recovered at 1600°C, for samples that were sequentially isochronally annealed from 700°C in steps of 100°C up to 1600°C. However, no silver signal was detected at this temperature, which left certain doubts regarding the crystalline structure of the samples at this temperature. This was speculated to be due to thermal etching of the top original amorphous layer while the deeper amorphous layer was epitaxial re-growth from the bulk. The decomposition of SiC, giving rise to a carbon peak in the RBS spectra due to evaporation of Si, was clearly observed on the same samples at 1600°C. Isothermal annealing at 1300°C for 10 h cycles up to 80h caused epixatial re-growth from the bulk during the first annealing cycle (10h). No further epitaxial re-growth from the bulk was observed up to 80h. This was believed to be due to the amorphous layer re-crystallising into crystals that were randomly oriented to the 6H-SiC substrate. No diffusion of silver was observed at temperatures below 1300°C but silver seemed to form precipitates at these temperatures. Diffusion of silver towards the surface accompanied by silver loss from the surface began at 1300°C and was very high at 1400°C, with silver profiles becoming asymmetric and closer to the surface. The loss of silver was already taking place at 1100°C. This loss was found to be due to the following: diffusion of silver towards the surface; the mass flow of silver via holes that were observed to be becoming larger with higher annealing temperatures on SiC surfaces and thermal etching of SiC. Isothermal annealing at 1300°C for 10h up to 80h caused diffusion of silver during the first annealing cycle, while no further diffusion was observed for any further annealing at the same temperature up to 80 h. The diffusion coefficient was not calculated due to the lack of information on the structural evolution of SiC during the first annealing cycle. Isothermal annealing at 1300°C and 1350°C for 30 minute cycles up to 120 minutes caused high diffusion during the first cycle and reduced diffusion during the second cycle, while no diffusion was observed for any further annealing longer than the second cycle. The higher diffusion during the first 30 minutes was due to ion induced amorphization. The diffusion of silver in amorphised SiC was measured at different temperatures in the range 1300°C to 1385°C and yielded to Do ~ 1.4 × 10-12 m2s-1 and Ea ~ 3.3 × 10-19 J. These values were found to be approximately the same as the values of silver diffusion in polycrystalline CVD-grown SiC found by our group which were due to grain boundary diffusion: Do ~ 4×10-12 m2 s-1 and Ea ~ 4×10-19 J. Implantation of silver at 600°C retained crystallinity although distortions occurred in the implanted region while implantation at 350°C also retained crystallinity but more distortions occurred as compared to silver implanted at 600°C. This was caused by the fact that at 600°C, the displaced atoms were more mobile because of their higher thermal energy than at 350°C. The higher thermal energy increased the probability of the displaced atoms combining with their original lattice sites. Annealing of these samples at 1300°C, 1350°C and 1500°C caused the annihilation of some defects but certain others were retained. No diffusion of silver was observed during annealing of the samples (implanted at 350°C and at 600°C) at 1300°C, 1350°C and 1500°C but silver moved towards the surface at 1500°C. The upper limit of the diffusion coefficient of D < 10-21 m2s-1 was obtained at 1300°C. The movement of silver towards the surface was found to be due to thermal etching at 1500°C. Neutron irradiation of these samples caused no silver diffusion but silver -110mAg, due to -109Ag capturing a neutron during neutron irradiation, was detected in the samples. / Thesis (PhD)--University of Pretoria, 2010. / Physics / unrestricted

<p>Tantalum diffusion barrier coatings were investigated as a way to improve the conversion efficiency of CIGS (copper indium gallium diselenide) solar cells.  Tantalum coatings were deposited upon silicon and stainless steel foil substrates using direct current magnetron sputtering (DcMS) and high power impulse magnetron sputtering (HiPIMS).  The coatings were characterized using scanning electron microscopy (SEM).  Cross-sectional scanning electron micrographs revealed that the HiPIMS coatings appeared denser than the DcMS coatings.</p>

Tantalum diffusion barrier coatings were investigated as a way to improve the conversion efficiency of CIGS (copper indium gallium diselenide) solar cells.  Tantalum coatings were deposited upon silicon and stainless steel foil substrates using direct current magnetron sputtering (DcMS) and high power impulse magnetron sputtering (HiPIMS).  The coatings were characterized using scanning electron microscopy (SEM).  Cross-sectional scanning electron micrographs revealed that the HiPIMS coatings appeared denser than the DcMS coatings.

The failure mechanisms of the tantalum-based nitride diffusion barrier using between copper metal and the SiGe/Si layers grown with UHV/CVD have been studied. The TaN and Cu films were deposited with RF sputtering technique. The structure of these films was analyzed by X-ray diffraction. The stoichiometry of TaN was characterized by XPS (X-ray photoelectron spectroscopy). The morphology of the films was examined with SEM and the microstructure of the interface between several layers was observed with TEM. With comparing the XRD patterns of the samples which were annealed in the different temperatures, the failure temperature of the TaN barrier layer can be identified and the failure mechanism of this barrier layer cab be elucidated with TEM observation. The results revealed that the deposited TaN film with low sputtering power had better performance for preventing the Cu atoms diffusing into the SiGe layer. The high composition of Ge in the SiGe alloy degraded the blocking ability of the TaN barrier layer due to the released the existed strain between the SiGe and Si. When the failure temperature was reached, The Cu3Si phase was formed first in the interface of the TaN/SiGe and inside the TaN film. If the annealed temperature went higher, the TaSi2 phase also was formed. Compared with SiGe/Si and Si substrate, the TaN diffusion barrier layer has a higher failure temperature in Si than those in SiGe layer.

abstract: Miedema's plot is used to select the Cu/metal barrier for Cu metallization.The Cu/metal barrier system selected should have positive heat of formation (Hf) so that there is no intermixing between the two layers. In this case, Ru is chosen as a potential candidate, and then the barrier properties of sputtered Cu/Ru thin films on thermally grown SiO2 substrates are investigated by Rutherford backscattering spectrometry (RBS), X-ray diffractometry (XRD), and electrical resistivity measurement. The Cu/Ru/SiO2 samples are analyzed prior to and after vacuum annealing at various temperatures of 400, 500, and 600 oC and at different interval of times of 0.5, 1 and 2 hrs for each temperature. Backscattering analysis indicate that both the copper and ruthenium thin films are thermally stable at high temperature of 600 oC, without any interdiffusion and chemical reaction between Cu and Ru thin films. No new phase formation is observed in any of the Cu/Ru/SiO2 samples. The XRD data indicate no new phase formation in any of the annealed Cu/Ru/SiO2 samples and confirmed excellent thermal stability of Cu on Ru layer. The electrical resistivity measurement indicated that the electrical resistivity value of the copper thin films annealed at 400, 500, and 600 oC is essentially constant and the copper films are thermally stable on Ru, no reaction occurs between copper films and Ru the layer. Cu/Ru/SiO2 multilayered thin film samples have been shown to possess good mechanical strength and adhesion between the Cu and Ru layers compared to the Cu/SiO2 thin film samples. The strength evaluation is carried out under static loading conditions such as nanoindentation testing. In this study, evaluation and comparison is donebased on the dynamic deformation behavior of Cu/Ru/SiO2 and Cu/SiO2 samples under scratch loading condition as a measure of tribological properties. Finally, the deformation behavior under static and dynamic loading conditions is understood using the scanning electron microscope (SEM) and the focused ionbeam imaging microscope (FIB) for topographical and cross-sectional imaging respectively. / Dissertation/Thesis / M.S. Materials Science and Engineering 2010

In this work, the possibility of Cu electrodeposition on Ru-Ta alloy thin films is explored. Ru and Ta were sputter deposited on Si substrate with different composition verified by RBS. Four point probe, XRD, TEM and AFM were used to study the properties of Ru-Ta thin films such as sheet resistance, crystallinity, grain size, etc. Cyclic voltammetry is used to study the Cu electrodeposition characteristics on Ru-Ta after various surface pretreatments. The results provide insights on the removal of Ta oxide such that it enables better Cu nucleation and adhesion. Bimetallic corrosion of Cu on modified Ru-Ta surface was studied in CMP related chemicals. In Cu interconnect fabrication process, the making of trenches and vias on low-k dielectric films involves the application of fluorocarbon plasma etch gases. Cu microdots deposited on Ru and Ta substrate were treated by fluorocarbon plasma etch gases such as CF4, CF4+O2, CH2F2, C4F8 and SF6 and investigated by using x-ray photoelectron spectroscopy, contact angle measurement and electrochemical techniques. Micropattern corrosion screening technique was used to measure the corrosion rate of plasma treated Cu. XPS results revealed different surface chemistry on Cu after treating with plasma etching. The fluorine/carbon ratio of the etching gases results in different extent of fluorocarbon polymer residues and affects the cleaning efficiency and Cu corrosion trends.

Carbon nanotube templated microfabrication (CNT-M) is a method that allows high-aspect ratio structures to be made for microelectromechanical systems (MEMS) devices. One concern when making monolithic electrical devices using CNT-M is that the aluminum oxide diffusion barrier will create too large of a resistance in the device. However, in developing CNT based MEMS devices, it has been observed that an electrical DC current is capable of transport from a conductive substrate, across the aluminum oxide, and through to the CNT structure grown on top of it. This thesis attempts to determine the mechanisms responsible for current being able to cross the aluminum oxide diffusion barrier easily through sample characterizations. Principally, current-voltage measurements, electron microscopy, XEDS, and SIMS analysis are used to characterize the various samples and determine the process responsible for the observed phenomenon. Through these techniques, it is determined exposure to ethylene gas during the CNT growth recipe used in our lab, regardless of whether CNTs grow on the sample or not, is necessary to cause a drop in resistance across the aluminum oxide, but the that the overall content of iron and carbon in the aluminum oxide do not correlate with this drop in resistance.

Lower k and low-leakage silicon carbonitride (SiCxNy ) films were fabricated using single precursor by using radio-frequency (RF) plasma-enhanced chemical vapor deposition (PECVD). We explored precursors with (1) cyclic-carbon-containing structures, (2) higher C/Si ratio, (3) multiple vinyl groups, as well as (4) the incorporation of porogen for developing low-k SiCxNy films as etch-stop/diffusion barrier (ES/DB) layer for copper interconnects in this study. SiCxNy films with k values between 3.0 and 3.5 were fabricated at T≦ 200 o C, and k~4.0-4.5 at 300-400 °C. Precursors with vinyl groups yielded SiCxNy films with low leakage, excellent optical transmittance and high mechanical strength due to the formation of cross-linked Si-(CH2)n-Si linkages.