M. J. Kim

5nm ruthenium thin film as a directly plateable copper diffusion barrier

Interfacial stability of electroplated copper on a 5nm ruthenium film supported by silicon, Cu∕(5nmRu)∕Si, was investigated using Rutherford backscattering and high-resolution analytical electron microscopy. Transmission electron microscopy (TEM) imaging shows that a 5nm Ru film is amorphous in contrast to the columnar microstructures of thicker films (20nm). Direct Cu plating on a 5nm Ru film yielded a homogeneous Cu film with over 90% plating efficiency. It is demonstrated that 5nm Ru can function as a directly plateable Cu diffusion barrier up to at least 300°C vacuum anneal. TEM reveals an interlayer between Ru∕Si, which expands at the expense of Ru upon annealing. Electron energy loss spectroscopy analyses show no oxygen (O) across the Cu∕(5nmRu)∕Si interfaces, thereby indicating that the interlayer is ruthenium silicide (RuxSiy). This silicidation is mainly attributed to the failure of the ultrathin Ru barrier at the higher annealing temperature.

Diffusion studies of copper on ruthenium thin film a plateable copper diffusion barrier

Diffusion studies were carried out on physical vapor deposited Cu/Ru(∼20 nm)/Si samples using secondary ion mass spectroscopy (SIMS) and transmission electron microscopy (TEM). Back side SIMS depth profiling revealed well-defined interfaces and showed that Cu interdiffusion was impeded by Ru thin film up to 450°C vacuum annealing. TEM showed a 20-22 nm Ru barrier layer with a columnar microstructure oriented vertically with respect to Si substrate. TEM results corroborate with SIMS data to indicate stability of the Ru film barrier for annealing temperatures up to 450°C. Direct Cu electroplating on ultrathin Ru barrier layers (<20 nm) was investigated in sulfuric acid. The electroplated Cu film is shiny, smooth, and without agglomeration under scanning electron microscopy. Excellent adhesion between interfacial layers was confirmed by the scribe-peel test. The interfacial characterization results indicate that Ru thin film is a promising candidate as a directly plateable Cu diffusion barrier.

Outdiffusion of La and Al from amorphous LaAlO3 in direct contact with Si (001)

We have evaluated the thermal stability of Al2O3/LaAlO3/Si (001) stacks with atomic force microscopy, x-ray diffraction, transmission electron microscopy, and secondary ion mass spectrometry using a back side polishing approach. Crystallization of the amorphous LaAlO3 film was found to occur for rapid thermal anneals (RTA) above 935u2009°C for 20 s, in flowing N2. Penetration of Al and La into the underlying Si (001) is clearly observed for RTA at or above 950u2009°C for 20 s in flowing N2.

Effect of N incorporation on boron penetration from p+ polycrystalline-Si through HfSixOy films

We demonstrate that incorporating N in Hf-silicate films reduces B penetration through the dielectric film. By modeling the B depth profiles, we calculated the B diffusivities through Hf-silicate (HfSixOy) to be ∼2× higher than the corresponding diffusivities for Hf-silicon oxynitride (HfSixOyNz). B diffusion through grain boundaries after HfSixOy crystallization is believed to be responsible for the enhanced B diffusivity observed. Suppression of crystallization observed in HfSixOyNz films is attributed to the lower Hf content in the films and the incorporation of N. The decreased B penetration observed in HfSixOyNz is a combination of absence of grain boundaries and the fact that N blocks B diffusion by occluding diffusion pathways. The minimum temperatures for B penetration are estimated to be approximately 950 and 1050u200a°C for HfSixOy and HfSixOyNz, respectively.

Boron penetration studies from p+ polycrystalline Si through HfSixOy

We present detailed B penetration studies from B-doped polysilicon through alternate gate dielectric candidate HfSixOy films. No detectible B penetration is observed for annealing times as long as 20 s after 950u200a°C. Considerable B incorporation into the Si substrate is observed for annealing temperatures higher than 950u200a°C. By modeling the B depth profiles, we calculated the B diffusivities through HfSixOy to be higher than the corresponding diffusivities for SiO2. B diffusion through grain boundaries after HfSixOy crystallization is proposed to be responsible for the enhanced B diffusivity observed.

Dopant penetration studies through Hf silicate

We present a study of the penetration of B, P, and As through Hf silicate (HfSixOy) and the effect of N incorporation in Hf silicate (HfSixOyNz) on dopant penetration from doped polycrystalline silicon capping layers. The extent of penetration through Hf silicate was found to be dependent upon the thermal annealing budget for each dopant investigated as follows: B(T⩾950°C∕60s), P(T⩾1000°C∕20s), and As (T⩾1050°C∕60s). We propose that the enhanced diffusion observed for these dopants in HfSixOy, compared with that of SiO2 films, is related to grain boundary formation resulting from HfSixOy film crystallization. We also find that, as in the case of SiO2, N incorporation inhibits dopant (B, P, and As) diffusion through the Hf silicate and thus penetration into the underlying Si substrate. Only B penetration is clearly observed through HfSiON films for anneals at 1050u2009°C for durations of 10 s or longer. The calculated B diffusivity through the HfSixOyNz layer is D0=5.2×10−15cm2∕s.

Phosphorus and arsenic penetration studies through HfSixOy and HfSixOyNz films

Phosphorus and arsenic penetration studies from P- and As-doped polycrystalline silicon through HfSixOy and HfSixOyNz (18% N) alternate gate dielectric candidates films into Si(100) are presented using a combination of chemical etching and secondary ion mass spectrometry (SIMS). Penetration is observed through HfSixOy after 1050 and 1000u200a°C rapid thermal annealing for As and P, respectively. In contrast, As or P dopant penetration is at the SIMS limit of detection for HfSixOyNz films. By modeling the P and As depth profiles in the Si substrate, their respective diffusivities in HfSixOy are higher than the corresponding diffusivities in SiO2. The enhanced dopant diffusivity observed in HfSixOy is proposed to be due to grain boundary formation in the dielectric after crystallization from annealing.

High performance gate first HfSiON dielectric satisfying 45nm node requirements

We show an ALD based HfSiON gate dielectric scaled to 1 nm EOT with excellent performance and reliability. Furthermore, the HfSiON dielectric films are integrated in a gate first approach that includes a 1000degC-5s anneal. It is also demonstrated that this 1 nm EOT HfSiON can achieve electron and hole mobilities comparable to that of SiON. This progress is enabled due to better understanding of the relationship between charge trapping, HfSiON thickness and crystallinity. Performance and reliability improvement is attributed to reduced charge trapping due to suppressed crystallization of the optimized HfSiON films

Hafnium silicate formation by ultra-violet/ozone oxidation of hafnium silicide

Abstract We report the room temperature growth of hafnium silicate by ultra-violet/ozone oxidation of hafnium silicide. Hafnium silicide was deposited by magnetron sputtering on hydrogen terminated (1xa00xa00) Si. The film was then exposed to UV radiation while in an O2 ambient. Hafnium silicate films are obtained with no detectable SiOx interfacial layer as characterized by X-ray photoelectron spectroscopy and high-resolution transmission electron microscopy.

Interfacial Diffusion Studies of Cu ∕ ( 5 nm Ru ) ∕ Si Structures Physical Vapor Deposited vs Electrochemically Deposited Cu

In contrast to physical vapor deposition (PVD), the electrochemical deposition (ECD) process is dependent upon substrate resistivity. ECD of Cu on ultrathin Ru diffusion barriers remains a technological challenge due to large resistivity increase over a wide plating area. Results are presented from the comparative investigation of interfacial stability and Cu diffusion processes in PVD and ECD Cu/(5 nm Ru)/Si structures. Cu can be conformally electroplated onto (5 nm Ru)/Si surfaces (ca. 1 cm 2 ) with over 94% efficiency. However, lesser uniformity and conformality of ECD Cu are observed on (5 nm Ru)/Si samples with larger surface areas. The transmission electron microscopy (TEM) reveals that ECD Cu film is less densely packed (ca. 70 nm) than PVD Cu. HRTEM studies in conjunction with surface analyses using optical microscopy and four-point probe resistivity measurements show that 5 nm Ru can successfully impede Cu diffusion up to 300°C for 10 min, but fails at 450°C. Interfacial profiling data obtained from back side secondary-ion mass spectrometry (SIMS) analysis agree with TEM results. X-ray photoelectron spectroscopy (XPS) investigation on nitric acid-etched PVD and ECD Cu/(5 and 20 nm Ru)/Si samples shows the presence of residual ECD Cu after annealing, suggesting that ECD Cu diffuses further into Ru than PVD Cu.