TiN Stack Research Articles

Impact of bottom electrode and SrxTiyOz film formation on physical and electrical properties of metal-insulator-metal capacitors

Metal-insulator-metal capacitors with SrxTiyOz (STO) dielectric films on TiN, Ru, and RuOx bottom electrodes with TiN top electrodes were studied. Metastable perovskite STO films with compositions in the Sr/(Sr+Ti)∼54–64 at. % range were obtained by crystallization at 600 °C in N2 of dielectric stacks grown by atomic layer deposition consisting of Sr-rich STO films [Sr/(Sr+Ti)∼64 at. %] on thin interfacial TiOx layers. The significant decrease in equivalent oxide thickness (EOT) and STO lattice parameter observed with increasing TiOx thickness indicates full intermixing of the TiOx and STO layers during the crystallization anneal, which results in the formation of an STO layer with higher Ti content and higher dielectric constant. The Sr-rich STO on TiOx stacks crystallize with small grain size, favorable for low leakage (JG). A significant improvement in JG for e-injection from the bottom electrode is obtained when using RuOx, as compared to TiN or Ru. A milder JG improvement with RuOx bottom electrode is also seen for e-injection from the top TiN electrode, indicating that higher quality perovskite STO films are formed on RuOx, or equivalently, that their trap density is lower. We propose oxygen scavenging from the STO by TiN or Ru electrodes, eliminated or reversed when using RuOx, as an explanation for the improvement. Using an optimized RuOx/TiOx/STO/TiN stack we obtained leakage of 10−7 A/cm2 (at 0.8 V) and 0.4 nm EOT.

Epitaxial cubic HfN diffusion barriers deposited on Si (001) by using a TiN buffer layer

Cubic HfN (B1-NaCl) thin films were grown epitaxially on Si(001) substrates by using a TiN (B1-NaCl) buffer layer as thin as ∼10nm. The HfN∕TiN stacks were deposited by pulsed laser deposition with an overall thickness below 60nm. Detailed microstructural characterizations include x-ray diffraction, transmission electron microscopy (TEM), and high resolution TEM. The electrical resistivity measured by four-point probe is as low as 70μΩcm at room temperature. Preliminary Cu diffusion tests show a good diffusion barrier property with a diffusion depth (2Dτ) of 2–3nm after annealing at 500°C for 30min in vacuum.

Origin of tensile stress in the Si substrate induced by TiN∕HfO2 metal gate/high-k dielectric gate stack

The tensile stress induced by the metal TiN film in the atomic layer deposited HfO2∕TiN stacks has been found from the crystallite coalescence mechanism of the Volmer–Weber-type growth mode at the early stage of the TiN film formation. The higher tensile stress induced by 3nm TiN film than that by the 20nm TiN film resulted from the smaller grain size and the [200] orientation of the TiN layer. Electron energy loss spectrum profile shows that there is no significant elemental interdiffusion between HfO2 and TiN, which could contribute to stress relaxation.

Optimization of PVD Ti/CVD TiN Liner for 0.35 μm Tungsten Plug Technology

AbstractA novel plasma enhanced CVD TiN process was integrated with high density plasma sputter etch preclean (PCII) and 1:1.5 collimated PVD Ti (c-PVD Ti) process to deposit a Ti/TiN liner for tungsten contact and via plugs. The integrated liner process was optimized for a 0.35 μm nonsalicide CMOS device application. RF power and sputter depth used for contact preclean were the major process variants affecting the contact resistance, junction leakage and transistor threshold voltage. Low contact resistance was obtained for a c-PVD Ti thickness of ∼375 Å. Via resistance was significantly lower with c-PVD Ti/CVD TiN liner as compared to only a TiN liner. Contact resistance for c-PVD Ti/c-PVD TiN and c-PVD Ti/CVD TiN liners were comparable while contacts with conventional PVD Ti/TiN liner showed significantly higher values due to poor step coverage. Low junction leakage current was obtained for integrated c- PVD Ti/CVD TiN stack.

A thickness model for the [formula omitted] stack in the titanium salicide process module

A reliable TiSi 2 TiN stack thickness model is an essential component for modeling the titanium salicide process, and such a model is not well-developed in current process simulators and in the literature. To determine this model, a design of experiments was set up to examine five process variables, namely, as-deposited Ti thickness, reaction temperature, reaction time, As + implanted dose, and the reaction ambient pressure. Weight and sheet resistance measurements were used to evaluate the thickness and efficiency of reaction (%Ti converted to suicide). A good first-order linear model was obtained, with a residual standard deviation (variation of model from data) of ~30 Å. The model establishes quantitatively, that the TiSi 2 thickness is proportional to the as-deposited Ti thickness and reaction temperature, inversely proportional to the implanted As + dose, and weakly proportional to the reaction time. Since TiN and TiSi 2 are competing reactions, TiN exhibits inverse functional relationships with the variables, as compared with TiSi 2. The efficiency of the reaction also has been quantified by the model. The ambient pressure has been found to have no impact on either the TiSi 2 TiN stack thickness, or the reaction efficiency. The model has been validated by cross-sectional transmission electron microscopy, which agrees with the model prediction within experimental error.