Atsushi Ogura

Characterization of Strain for High-Performance Metal-Oxide-Semiconductor Field-Effect-Transistor

Strain evaluation in a small area is required because the extremely short channel length in state-of-the-art metal–oxide–semiconductor field-effect transistors (MOSFETs) leads to a narrow and shallow channel region. The strain in this limited area strongly affects the device performance owing to carrier mobility modification. We used UV–Raman spectroscopy with a quasi-line-shape excitation source and a two-dimensional charge-coupled-device detector in order to evaluate the strain distribution in Si or Si-on-insulator (SOI) substrates with a patterned SiNx film. As results, the strain was concentrated at the SiNx/Si interface and SiNx film pattern edge. A large tensile (compressive) strain was induced by the SiNx film with inner tensile (compressive) stress in the space region that corresponds to a channel region of the n- or p-MOSFETs. We assume that these large strains in the space region are the origin of the mobility enhancement in n- or p-MOSFETs. Furthermore, in addition to the size effect of channel length, we confirmed that the strain could be controlled by changing SiNx film thickness, film stress, and the substrate (SOI or bulk-Si). The quantitative evaluation of strain by means of simulation is also discussed.

A Comparative Study of Nitrogen Gas Flow Ratio Dependence on the Electrical Characteristics of Sputtered Titanium Nitride Gate Bulk Planar Metal–Oxide–Semiconductor Field-Effect Transistors and Fin-Type Metal–Oxide–Semiconductor Field-Effect Transistors

Sputtered titanium nitride (TiN) metal gates fabricated at different nitrogen gas flow ratios [RN=N2/(N2+Ar)] from 17 to 100% have systematically been investigated, and the RN dependence on the electrical characteristics of fabricated TiN metal gate bulk planar metal–oxide–semiconductor field-effect transistors (MOSFETs) and fin-type MOSFETs (FinFETs) are clarified. It is experimentally found that the electrical characteristics of FinFETs such as subthreshold slope (S-slope) and mobility are almost independent of RN, while those of bulk planar MOSFETs markedly deteriorate with increasing RN. These experimental results are discussed from the viewpoint of the device structure differences and the direction of energetic nitrogen atoms in sputtering. The threshold voltage (Vth) of FinFETs can be adjusted to about 100 mV by controlling RN without device performance degradation. These results are very useful in setting an appropriate Vth for TiN gate-last FinFETs.

Study of Strain Induction for Metal–Oxide–Semiconductor Field-Effect Transistors using Transparent Dummy Gates and Stress Liners

Strain induction was studied on a sample that had a dummy gate tetraethyl orthosilicate–silicon dioxide (TEOS–SiO2) and SiN film by UV-Raman spectroscopy with high spatial and high wave-number resolution. The UV laser penetrated through the dummy gate that was transparent to UV light, which enabled us to evaluate strain in the channel of the metal–oxide–semiconductor field-effect transistor (MOSFET) model. Furthermore, we compared stress profiles obtained by finite element (FE) calculations with those obtained by UV-Raman measurements. There was a difference between the stress profiles in the line-and-space pattern sample and in the dummy-gate sample; large compressive (tensile) strains were concentrated at the channel edges in the dummy-gate sample with the compressive (tensile) stress liner, although both tensile and compressive strains existed at the channel edge in the line-and-space pattern sample. The results from UV-Raman spectroscopy were consistent with those obtained by the FE calculation.

Improvement of CVD SiO2 by Post Deposition Microwave Plasma Treatment

We evaluated density and chemical bonding states of the CVD SiO2 film with and without plasma treatment to clarify an effect of the plasma treatment. It was found that the chemical bonding states were homogenized by the plasma treatment from the results of X-ray photoelectron spectroscopy. In addition, an increase of the film density was also observed. These results indicate the densification of SiO2 film, suppression of bond-angle fluctuation, and decrease of impurities (e.g. Hydrogen, Nitrogen and so on) in the SiO2 film. These results can well explain the improvement of the electrical properties by the plasma treatment. Furthermore, UV-Raman measurement was performed to evaluate the modification of Si stress and crystal quality at the SiO2/Si interface.

Channel-stress study on gate-size effects for damascene-Gate pMOSFETs with top-cut compressive stress liner and eSiGe

Damascene gate process enhances the drivability in shorter gate length region, as compared to conventional gate 1st process for pFETs with compressive stress SiN liner and embedded SiGe. The origin of the gate length effect is investigated for the first time by using the UV-Raman spectroscopy. Moreover, the relationship between channel strain and gate width for damascene gate pFETs is analyzed and the effect is also demonstrated. It is found that channel strain is considerably enhanced in shorter gate length and narrower gate width by the combination of damascene gate process and stress enhancement techniques.

Study of stress effect on replacement gate technology with compressive stress liner and eSiGe for pFETs

The stress effect at the channel region of pFETs with compressive stress liner (c-SL) and eSiGe using replacement gate technology is firstly investigated in detail based on the combination of UV-Raman spectroscopy and 3D stress simulation. The gate length effect for the channel stress is confirmed by measurement and simulation. Moreover, the Ion dependence on the channel width is also investigated. It is found that the lateral stress along the channel is enhanced at the edge beside STI, resulting in high Ion at narrow gate width region.

UV-Raman Spectroscopy Study on SiO2/Si Interface

We evaluated stresses and crystal qualities at SiO2/Si interfaces formed in dry O2 an O radical using UV-Raman spectroscopy. The stresses induced by oxidation at 1050C were higher than those at 900C and 1000C. It was also shown that crystal quality of Si and stress at SiO2/Si interface formed in dry O2 depend on the degree of dilution of oxygen and the oxidation time. The SiO2/Si interface formed in O radical exhibits larger compressive stress and better crystal quality as compared with that formed in dry O2. Introduction The oxidation is one of the most important processes using for the fabrication of integrated circuit. It is essential to control and characterize SiO2/Si interface because the increasing impacts of variation in electrical characteristics have become one of the most crucial issues for extremely scaled down LSIs [1]. In this study, we evaluated stress and crystal quality at SiO2/Si interface formed in dry O2 and O radical using UV-Raman spectroscopy. Experiment Si(100) substrates were annealed at 1200C in pure Ar atmosphere for 30min to obtain atomically flat surface [1]. After etching the oxide films in a mixed solution of HCl/HF, the substrates were oxidized in O2/Ar atmosphere at 900C, 1000C and 1050C. The percentage of dilution of O2 with Ar was 10% and 100%. These two oxidation atmosphere are abbreviated hereafter as 10% O2 and 100% O2. Oxide film formed in O radical was fabricated using microwave-excited high-density and lowion-energy Kr/O2 mixture plasma at 400 C. Thicknesses of oxide films formed by dry O2 and O radical were 7 nm. The stresses and crystal quality in Si at SiO2/Si interface were evaluated using UV-Raman spectroscopy excited by Ar ion laser ( =364 nm), whose depth of penetration into the Si substrate was approximately 5 nm. The full-width at half maximum (FWHM) of Raman spectra reflects crystal quality. Wider (narrower) FWHM indicates poorer (better) crystal quality. Raman peak shift shows stress. Lower (higher) peak shift means tensile (compressive) stress. The value of Raman peak shift indicates stress magnitude. Raman spectroscopy system used in this study was described in detail elsewhere [2]. Results and Discussion Fig. 1 shows stress distributions measured at 100 positions over 0.25 x 0.25 cm for the oxide film, which was formed in 10% O2 at 1000 C (red) and 1050C (blue), using UV-Raman spectroscopy. Here, the horizontal axis indicates the Raman peak shift. According to this figure, the compressive stresses were observed at SiO2/Si interface formed in the dry O2 at both 1000 C and 1050C. The Raman peak shifted toward higher wavenumber with increasing oxidation temperature from 1000C to 1050C. However, the approximately similar Raman peak shifts were observed for dry oxidation at 900C and 1000C. It is considered that the viscous flow occurred above the temperature between 1000C and 1050C [3]. In Fig. 2 the dependence of a) FWHM and b) Raman peak shift on Tox (oxidation-temperature) measured for the oxidation in 10% O2 (red) and 100% O2 (blue) are compared with that measured for the oxidation using O radical. In this figure the average of values measured at 100 positions over 0.25 x 0.25 cm is only shown, because the deviations were almost the same in all samples. The observation that FWHM measured for 10% O2 is narrower than that measured for 100% O2 implies better crystal quality in the former case. Raman peak shift was also affected by dilution of oxygen, indicating that longer oxidation time results in a larger compressive stress at SiO2/Si interface. Namely, it was shown that crystal quality and stress at Si substrate surface oxidized in dry O2 depend on dilution of oxygen and oxidation time. For the oxidation using O radical the averaged values in FWHM and Raman peak shift were found to be 2.513 cm and 0.055 cm, respectively. The SiO2/Si interface formed in O radical exhibits larger compressive stress as compared with that formed in dry O2. However, crystal quality at SiO2/Si interface formed in O radical was similar to that formed in dry O2 at 900 C, implying that large compressive stress does not necessarily result in poor crystal quality. References [1] R.Kuroda et al., SSDM, pp. 706-707 (2008). [2]A. Ogura et al., Jpn. J. Appl. Phys. 45. 3007 (2006). [3] E.Kobeda et al., J. Vac. Sci. Technol. B6 (2), Mar/Apr (1988).

HAX-PES Study of SiN Film for Charge Storage Layer in High Performance SONOS Type Flash Memory Cell

1 School of Science and Technology, Meiji University 1-1-1 Higashimita, Tama-ku, Kawasaki, 214-8571, Japan Phone: +81-44-934-7324 E-mail: d_kose@isc.meiji.ac.jp 2 TOKYO ELECTRON AT LTD 1-8 Fuso-cho, Amagasaki, Hyogo, 660-0891, Japan 3 Japan Synchrotron Radiation Research Institute (JASRI) 1-1-1 Koto, Saya-cho, Sayo-gun, Hyogo, 679-5198, Japan 4 Research Fellow of the Japan Society for the Promotion of Science 8 Ichiban-cho, Chiyoda-ku, Tokyo, 102-8472, Japan

Channel Strain Analysis in Damascene-gate pMOSFETs on Si (100) and (110) Substrate by Conventional and Cross-sectional Raman Spectroscopy

Channel strain in damascene gate pMOSFETs with compressive stress liner (c-SL) and embedded SiGe (eSiGe) were studied by micro-Raman spectroscopy with a quasi-line-shape UV excitation (λ=363.8nm). The channel strain profiles were obtained by the conventional mea-surement from the surface after dummy gate removal. The compressive strains at the channel edges were larger than that at the channel center for the relatively long gate length (Lgate). As the Lgate became smaller, although it became hard to recognize the strain profile, the compres-sive strain at the channel center increased by the superposition of the strain at the channel edges. However, channel strain disappeared in the measurement data for the channel length less than 160 nm. Thus, we extended the laser exposure time from 10 to 40 minutes to extract the channel strain component from obtained Raman spectra. The Raman peaks consisted of two or three peaks for the Lgate less than 160 nm. By multi peak fitting, we have succeeded in measuring the extremely large stress of - 2.4 GPa in the channel of Lgate = 30 nm pMOSFET. We also per-formed the cross-sectional measurements for the samples before and after metal-gate/high-k gate stack formation. Channel strain profiles were obtained similar to those by the conventional mea-surement. Extremely high device performance can be clearly explained by the compressive stress derived from the Raman measurements both in the Lgate dependence and eSiGe effect. We also demonstrated that Raman spectroscopy using cross-sectional measurement can evaluate the channel strain even in the MOSFETs after gate stack formation.

New analysis of heavily doped boron and arsenic in shallow junctions by X-ray photoelectron spectroscopy

Chemical bonding states of boron (B) in shallow P+/N junctions were studied by soft X-ray photoelectron spectroscopy (SXPES). The concentration profiles of B having different binding energies were successfully determined the SXPES combined with the step-by-step etching of Si substrates. The concentration profiles of B having the lowest binding energy can be assigned as activated B, which agreed quite well with those of holes determined by the Hall measurements, while those having the middle and highest binding energies must be attributed to deactivated B. Effects of the spike-RTA and flush lamp annealing (FLA) were compared regarding the concentration profiles of B and UV Raman spectroscopy. Arsenic (As) doped layers were also studied by the X-ray photoelectron spectroscopy and the two different bonding states were revealed for As atoms embedded in Si substrates.