Library

Notes: Top nitride (TN) and bottom nitride (BN) configurations of the a-SiN1.6:H/a-Si:H interface produced by plasma enhanced chemical vapor deposition (PECVD) have been characterized by using metal/a-SiN1.6:H/a-Si:H (MNS) devices and thin-film transistors (TFT). We observed that: (i) for MNS devices, the resonant frequency activation energy Eac from admittance, saturates at a higher value for the TN ((approximately-equal-to)0.3 eV) than for the BN ((approximately-equal-to)0.1 eV) MNS devices and the resonant frequency preexponential factors are found to be, respectively, in the 1012 s−1 and 108 s−1 range, (ii) in strong accumulation, the source-drain current activation energy is nearly similar for TN and BN TFTs (Edc(approximately-equal-to)0.1 eV), a slightly higher value being found in the latter configuration, (iii) the analysis of the transfer characteristics yields a very similar density of deep interface states (DOS) (approximately-equal-to)4–5×1012 cm−2 eV−1 in both configurations, and (iv) the effective field-effect mobility is higher for the BN (μFE(approximately-greater-than)0.5 cm2/V s) than for the TN (μFE(approximately-less-than)0.3 cm2/V s) TFTs.The observed difference in mobilities for both TFTs structures is most likely associated with very high source/drain contact resistances in the TN TFT rather than with the quality of both interfaces. To explain some of these experimental results, the surface- and buried-channel models are proposed for TN and BN TFT configurations, respectively. The buried-channel concept is based on evidence of recycling/intermixing of nitrogen atoms into a-Si:H deposited on a-SiN1.6:H by PECVD. The nitrogen tail would produce a Si-rich a-SiNx:H alloy transition layer, followed by a nitrogen-doped n-type layer. The doped layer corresponding to the buried-channel formation is predicted to be located beyond 30 A(ring) from the BN interface. In contrast, the surface-channel proposed for the TN interface is based upon this interface being considered as atomically sharp. The MNS results (i) are consistent with this model. In the case of TN MNS they are explained by thermionic emission of electrons from the traps to the conduction band edge in a-Si:H and in the case of BN MNS by hopping in the defective a-SiNx:H interlayer. The result (ii) on TFTs may be attributed to a broadening of the linear part of the conduction band tail of a-Si:H in the channel region, due to recycled nitrogen atoms.

Notes: We have investigated the effects of different post-deposition temperature anneals and N-H concentrations, on the generation of ultraviolet (UV)-induced two-coordinated nitrogen dangling bonds in plasma-enhanced chemical vapor deposited (PECVD) silicon nitride films using electron spin resonance (ESR). It is shown that the nitrogen dangling bond is activated by post-deposition anneals with temperatures as low as 500 °C in N-rich PECVD nitride films followed by subsequent UV broadband illumination. It also appears that there is a possible correlation between the initial N-H concentration in the films and the concentration of generated nitrogen dangling bonds following the anneal/UV sequence. We also report the charge state associated with the nitrogen dangling bond using a combination of ESR and capacitance versus voltage measurements; these measurements suggest that this two-coordinated defect is electrically neutral when paramagnetic.

Notes: In this paper we have analyzed the influence of the mask channel length (LM) on the performance of the 55-nm-hydrogenated amorphous silicon (a-Si:H) thin-film transistors (TFTs), incorporating nitrogen-rich hydrogenated amorphous silicon nitride gate dielectric and phosphorus-doped microcrystalline silicon (n+μc-Si:H) source/drain (S/D) contacts. In our TFTs the n+μc-Si:H S/D contacts have a specific contact resistance around or below 0.5 Ω cm2. We have shown that in our TFTs a field-effect mobility and threshold voltage are dependent on LM, and this dependence is most likely due to the influence of the S/D contact series resistance on TFTs characteristics. Finally, we have demonstrated that if the mask channel length is extended by a ΔL (which is a distance from the S/D via edge at which the electron injection/collection is taking place) the field-effect mobility and threshold voltage are independent of the channel length. In such a case μFE, VT, and ON/OFF current ratio around 0.76 cm2/V s, 2.5 V, and 107, respectively, has been obtained.

Notes: Ultraviolet light induced changes in polyimide liquid-crystal alignment films were investigated. Infrared, UV-visible, x-ray photoelectron spectroscopy, and electron-spin-resonance measure- ments indicated that bond breaking and subsequent oxidation reactions occur in polyimide films (SE7210, OCG284, DuPont 2555 and 2540) during the broadband UV illumination in air. Mechanical rubbing has no effect on the optical and magnetic properties but it causes the removal of the UV-exposed film. Capacitance–voltage measurements indicate that there is a slight decrease in dielectric constant and creation of net negative charges in the film after UV exposure. Sur- face tension of polyimide films before and after UV illumination and changes in the pretilt angle of the polyimide surface following UV exposure have also been studied. The decrease in pretilt angle following UV illumination is attributed to an increase in surface tension. Our results indicate that a simple UV technique can be used to achieve domain divided liquid-crystal pixel electrode design with improved viewing characteristics. © 1996 American Institute of Physics.

Notes: We report results of electron paramagnetic resonance, photothermal deflection spectroscopy, and capacitance-voltage measurements on amorphous hydrogenated silicon nitride (a-SiNx:H) thin films exposed to ultraviolet (UV) illumination. It has been previously shown that exposure to UV light activates silicon dangling-bond defects, i.e., K0 centers, in a-SiNx:H thin films. Here, we demonstrate that the initially UV-activated K0 center can be irreversibly annihilated at long illumination times. Because this effect seems to scale with H content of the measured films, we propose that hydrogen may be passivating the K0 defects during the extended UV exposure. We also show that films subjected to long UV exposures trap charge as efficiently as those having much larger K0 concentrations. A few possibilities to explain this effect are discussed. © 1995 American Institute of Physics.

Notes: The correlations between structural, chemical, electrical, optical properties of in situ phosphorus-doped hydrogenated microcrystalline silicon prepared by plasma-enhanced chemical vapor deposition have been studied by high-resolution transmission electron microscopy, secondary-ion mass spectrometry, electrical conductivity, and optical measurements. Microcrystallinity has been observed at a substrate temperature as low as 100 °C with a 1% dilution of (1% PH3/SiH4) in H2. In situ phosphorous-doped hydrogenated microcrystalline silicon is best grown at 200–300 °C in terms of microstructure, H and P content, and dopant activation. The effects of thermal processing and the use of silicon nitride cap deposited prior to anneal on the structure and properties of phosphorous-doped hydrogenated microcrystalline silicon are also reported. The use of a silicon nitride capping layer is shown to inhibit recrystallization of hydrogenated microcrystalline silicon during rapid thermal anneal.

Notes: We have reproduced and extended some recently reported electron spin resonance (ESR) measurements related to the nature of the dominant charge traps in silicon nitride. We detected defect centers at low temperatures using ESR, in both as-deposited and ultraviolet-irradiated silicon nitride powders and films prepared by low-pressure chemical vapor deposition (LPCVD). Only two silicon dangling bond defects were observed in the silicon nitride, one at g=2.003 (⋅Si≡N3), and the other at g=2.005 (⋅Si≡Si3). The signal intensity at g=2.003 is by far the dominant signal in the LPCVD films and powders subjected to UV illumination; the signal at g=2.005 is only about 3% of its intensity. These results support the idea that there is just one dominant silicon paramagnetic center (⋅Si≡N3), which is responsible for charge trapping in silicon nitride.

Notes: The effects of positive and negative bias stress on hydrogenated amorphous silicon nitride/hydrogenated amorphous silicon (a-SiNx:H/a-Si:H) structures are investigated as a function of stress time, and stress temperature. It is shown that bias stress induces a parallel shift of the capacitance voltage (C-V) characteristics. The direction of the C-V shift depends on the sign of the applied stress voltage, while the magnitude of the C-V shift depends on stress time and temperature in a manner which is identical to that observed in a-Si:H thin-film transistors. In addition, it is shown that positive bias stress increases the number of localized states in the a-Si:H mobility gap, but negative bias stress does not. However, the observed increase cannot account for the corresponding C-V shift. These results lead us to conclude that the C-V shift is not induced by dangling bond defects in a-Si:H but rather by carrier trapping in the insulator.

Notes: UV-induced enhancement of the subgap optical absorption has been observed in thin films of chemically vapor deposited silicon nitride. The threshold for this process is ∼4 eV. Visible light bleaches out much of the enhanced absorption and 200–400 °C thermal anneals remove the rest, leaving the optical properties of the nitride in the "as-grown'' state. The characteristics of this creation and bleaching process suggest that UV light "optically dethermalizes'' the distribution of carriers in localized band tail states. Correlation of our data with recent electron spin resonance results suggests that many of the trapped carriers may be singly occupied silicon dangling bonds.

Notes: We show that silicon dangling bonds with different nearest-neighbor configurations and energy levels can be resolved by electron spin resonance (ESR) in silicon nitride. Using an in situ bias technique on large-area metal-nitride-silicon structures, we demonstrate that the ESR line in Si-rich nitride consists of an inhomogeneous distribution of discrete components at different g values. We show that trapping of holes occurs at a site with a g value of 2.0052 corresponding to a pure Si environment, while electron trapping occurs at a site with a g value of 2.0028, corresponding to a pure N environment.