Spectroscopic observation of interface states of ultrathin silicon oxide
01 May 1996-Journal of Applied Physics (American Institute of Physics)-Vol. 79, Iss: 9, pp 7051-7057
TL;DR: In this article, the amount of the energy shift of the substrate Si 2p3/2 peak measured as a function of the bias voltage was analyzed for 3.6-nm-thick silicon oxide/n-Si(100) metal-oxide-semiconductor devices.
Abstract: Interface states in the Si band gap present at oxide/Si(100) interfaces for ∼3‐nm‐thick Pt/2.1∼3.6‐nm‐thick silicon oxide/n‐Si(100) metal–oxide–semiconductor devices are investigated by measurements of x‐ray photoelectron spectra under biases between the Pt layer and the Si substrate, and their energy distribution is obtained by analyzing the amount of the energy shift of the substrate Si 2p3/2 peak measured as a function of the bias voltage. All the interface states observed using this new technique have discrete energy levels, showing that they are due to defect states. For the oxide layer formed in H2SO4+H2O2, the interface states have three density maxima at ∼0.3, ∼0.5, and ∼0.7 eV above the valence‐band maximum (VBM). For the oxide layer produced in HNO3, two density maxima appear at ∼0.3 and ∼0.7 eV above the VBM. The energy distribution for the oxide layer grown in HCl+H2O2 has one peak at ∼0.5 eV. The 0.5 eV interface state is attributed to the isolated Si dangling bond defect. The 0.3 and 0.7 eV ...
Citations
More filters
TL;DR: In this paper, the leakage current of the SiO2 layer formed with 61 wt'% HNO3 at its boiling temperature of 113'°C has a 1.3 nm thickness with a considerably high density leakage current.
Abstract: Ultrathin silicon dioxide (SiO2) layers with excellent electrical characteristics can be formed using the nitric acid oxidation of Si (NAOS) method, i.e., by immersion of Si in nitric acid (HNO3) solutions. The SiO2 layer formed with 61 wt % HNO3 at its boiling temperature of 113 °C has a 1.3 nm thickness with a considerably high density leakage current. When the SiO2 layer is formed in 68 wt % HNO3 (i.e., azeotropic mixture with water), on the other hand, the leakage current density (e.g., 1.5 A/cm2 at the forward gate bias, VG, of 1 V) becomes as low as that of thermally grown SiO2 layers, in spite of the nearly identical SiO2 thickness of 1.4 nm. Due to the relatively low leakage current density of the NAOS oxide layer, capacitance–voltage (C–V) curves can be measured in spite of the ultrathin oxide thickness. However, a hump is present in the C–V curve, indicating the presence of high-density interface states. Fourier transformed infrared absorption measurements show that the atomic density of the SiO...
210 citations
TL;DR: In this article, the optical properties of Si surfaces can be significantly and reversibly altered by standard microelectronic treatments, and that stable, high optical quality surface passivation layers will be critical in future Si micro- and nanophotonic systems.
Abstract: Utilizing a high quality factor (Q~1.5×10^6) optical microresonator to provide sensitivity down to a fractional surface optical loss of alpha s [prime] ~10^–7, we show that the optical loss within Si microphotonic components can be dramatically altered by Si surface preparation, with alpha s [prime] ~1×10^–5 measured for chemical oxide surfaces as compared to alpha s [prime] <=1×10^–6 for hydrogen-terminated Si surfaces. These results indicate that the optical properties of Si surfaces can be significantly and reversibly altered by standard microelectronic treatments, and that stable, high optical quality surface passivation layers will be critical in future Si micro- and nanophotonic systems.
141 citations
TL;DR: In this paper, the chemistry and physics of thin films and interfaces govern the behavior of microelectronic devices, and different spectroscopies, in particular X-ray photoemission spectroscopy (XPS), have been used to determine the physics and chemistry of a variety of thin materials and interfaces.
Abstract: The chemistry and physics of thin films and interfaces govern the behavior of microelectronic devices. In this review, we examine how different spectroscopies, in particular X-ray photoemission spectroscopy (XPS), have been used to determine the physics and chemistry of a variety of thin films and interfaces commonly found in microelectronic devices. Electron spectroscopies are particularly attractive because the electron escape depths are comparable to the thickness of the films, allowing one to probe the bulk of the film and the interfaces simultaneously. We also give examples of how one can use the maximum-entropy method, a relatively straightforward analysis technique, to derive composition depth profiles of thin films from angle-resolved XPS data. Two technologically important classes of thin films are discussed. First, we examine thin inorganic films that are important high dielectric constant materials for CMOS devices or DRAM applications. Second, we discuss metal/polymer interfaces that are important in low dielectric constant device packaging. We conclude by discussing future developments and applications.
76 citations
TL;DR: In this paper, the energy distribution of interface states at ultrathin oxide/Si(100) interfaces is obtained using a new method, i.e., x-ray photoelectron spectroscopy measurements under biases between the metal overlayer and the Si substrate of the MOS devices.
Abstract: The energy distribution of interface states at ultrathin oxide/Si(100) interfaces is obtained using a new method, i.e., x-ray photoelectron spectroscopy measurements under biases between the metal overlayer and the Si substrate of the metal-oxide-semiconductor (MOS) devices. Ultrathin thermal oxide layers formed at 450 °C in oxygen have an interface state peak near the midgap and it is attributed to isolated Si dangling bonds with which no atoms in the oxide layer interact. On the other hand, thermal oxide layers formed at 650 °C have a two-peaked structure, one peak above and the other below the midgap, and they are attributed to Si dangling bonds with which an oxygen or Si atom in the oxide layer interacts weakly. The density of the interface states, especially that near the midgap, decreases drastically by cyanide treatment, i.e., the immersion of Si in a KCN solution for a few seconds followed by a rinse in boiling water, performed before the oxide formation. It is suggested that cyanide ions penetrat...
58 citations
TL;DR: In this paper, the performance of 〈indium-tinoxide (ITO)/silicon oxide/n-Si(100) junction solar cells is improved by immersing Si wafers in a potassium cyanide solution before the ITO deposition.
Abstract: The performance of 〈indium-tin-oxide (ITO)/silicon oxide/n-Si(100)〉 junction solar cells is improved by immersing Si wafers in a potassium cyanide solution before the ITO deposition. It is found from x-ray photoelectron spectroscopy measurements that about 3% monolayer cyanide (CN−) ions are present on the Si surface after the cyanide treatment. The temperature dependence of the current–voltage curves shows that the band bending in n-Si is increased by the cyanide treatment. The increase in the band bending is attributed to an upward Si band edge shift caused by the presence of CN− ions at the oxide/Si interface and/or in the oxide layer. Conductance–voltage measurements show that the density of trap states considerably decreases after the cyanide treatment. The conductance decrease is attributed to the passivation of interface states by the adsorption of CN− ions on Si dangling bonds.
50 citations
References
More filters
TL;DR: In this paper, the bonding of Si atoms at the SiO2/Si interface was determined via high-resolution core level spectroscopy with synchrotron radiation, and a model of the interface structure was obtained from the density and distribution of intermediate oxidation states.
Abstract: The bonding of Si atoms at the SiO2/Si interface is determined via high-resolution core level spectroscopy with synchrotron radiation. For oxides grown in pure O2, the SiO2/Si interface is found to contain Si atoms in intermediate oxidation states with a density of 1.5 ± 0.5 × 1015 cm−2. From the density and distribution of intermediate oxidation states, models of the interface structure are obtained. The interface is not abrupt, as evidenced by the non-ideal distribution of intermediate oxidation states and their high density (about 2 monolayers of Si). The finite width of the interface is explained by the bond density mismatch between SiO2 and Si. Annealing in H2 is found to influence the electrical parameters by removing the Pb centers that pin the Fermi level. The distribution of intermediate oxidation states is not affected.
1,543 citations
TL;DR: In this article, a method of determining the energy distribution of surface states at silicon-silicon dioxide interfaces by using low-frequency differential capacitance measurements of MOS structures is described.
Abstract: A method of determining the energy distribution of surface states at silicon-silicon dioxide interfaces by using low-frequency differential capacitance measurements of MOS structures is described. Low-frequency measurements make it possible to determine the silicon surface potential as a function of MOS voltage directly from the experimental data without requiring knowledge of the Si doping profile. No graphical differentiations are required to determine the surface state density from the experimental curves, and errors introduced by uncertainties in the silicon doping density are reduced. Also, it is shown that the measurements can be used to determine the relative lateral uniformity in the characteristics of the oxide and interface under the MOS field plate. Nonuniformities can result in large errors in the surface-state density derived from MOS capacitance measurements. Measurements are presented and interpreted for both n- and p-type silicon samples prepared by bias-growing the oxide in steam.
629 citations
TL;DR: In this paper, a direct tunneling theory is formulated and applied to high-speed thin-oxide complementary metal-nitride-oxide-silicon (MNOS) memory transistors.
Abstract: A direct tunneling theory is formulated and applied to high-speed thin-oxide complementary metal-nitride-oxide-silicon (MNOS) memory transistors. Charge transport in the erase/write mode of operation is interpreted in terms of the device threshold voltage shift. The threshold voltage shift in the erase/write mode is related to the amplitude and time duration of the applied gate voltage over the full range of switching times. MNOS memory devices ( X_{o}=25 \Aring, X_{N} = 335 \Aring ) exhibit a \Delta V_{th} = \plusmn3 V for an erase/write t_{p} = 100 ns, which corresponds to an initial oxide field strength E_{ox}= 1.2 \times 10^{7} V/cm. The direct tunneling theory is applied to the charge retention or memory mode in which charge is transported to and from the Si-SiO 2 interface states. The rate of charge loss to interface states is influenced by electrical stress which alters the interface state characteristics. We discuss the fabrication of complementary high-speed MNOS memory transistors and the experimental test procedures to measure charge transport and storage in these devices.
223 citations
94 citations
TL;DR: In this article, a series of semi-empirical calculations on threefold-coordinated silicon at the Si/${\mathrm{SiO}}_{2}$ interface were performed on finite clusters of atoms with use of hydrogen terminators in an unrestricted Hartree-Fock formalism.
Abstract: We present a series of semiempirical calculations on threefold-coordinated silicon at the 〈111〉 Si/${\mathrm{SiO}}_{2}$ interface. These were performed on finite clusters of atoms with use of hydrogen terminators in an unrestricted Hartree-Fock formalism wherein we include lattice relaxations. We have calculated defect electrical levels as well as ESR hyperfine parameters. The agreement with Brower's principal hyperfine data is excellent. On the strength of this agreement, we assign the superhyperfine shoulders to spin density on three second-nearest-neighbor silicon atoms in the crystalline silicon. Our agreement with electrical data is good; we obtain a positive U of between 0.3 and 0.6 eV, depending upon the method of calculation. Finally, we predict the existence of a spin-dependent deep-level transient-spectroscopy signal at high pressure.
75 citations