DC plasma etching of silicon by sulfur hexafluoride. Mass spectrometric study of the discharge products

Polycrystalline silicon wafers were etched in dc discharges of SF₆. SF_x species were extracted from the discharges and measured with a mass spectrometer. A systematic procedure was used to measure the SF _x ⁺ signals such that they are indicators of events in the discharge close to the sample undergoing etching. The picture that emerges is remarkably simple and shows the relative stability of several SF_x species including SF₆, SF₄, SF₂, and SF which are shown to be extracted from the discharge both in the presence and absence of the silicon sample. When silicon is being etched on the cathode of the discharge cell, the only significant additional products are SiF₄ and S₂F₂. A comparison of blank and sample data for opposite substrate polarities shows that there is only a small cation-assisted etching effect and suggests that ions do not play an important role in the etching of silicon by SF₆ discharges.     

Plasma etching of refractory metals (W,Mo, Ta) and silicon in SF6 and SF6-O2. An analysis of the reaction products

The etching rates and reaction products of refractory metals (W, Mo, and Ta) and silicon have been studied in a SF₆-O₂ r.f. plasma at 0.2 torr. The relative concentrations of WF₆ and WOF₄ and the intensities of the WF _n ⁺ (n=3–5), WOF _m ⁺ (m=1–3), MoF _n ⁺ , and MoF _m ⁺ ions have been measured by mass spectroscopy. An analysis of the neutral composition of the plasma during etching of these metals and a comparison with the results obtained for silicon show that at least two species are involved for W and Mo etching: fluorine and oxygen atoms. A reaction scheme is proposed.     

Low-temperature dry etching of tungsten,dielectric, and trilevel resist layers on GaAs

Dry etching of common masking materials used in GaAs device technology, was examined down to temperatures of –30°C. The etch rates of SiN_x, SiO₂, and W in SF₆/Ar are reduced below 0°C, but the anisotropy of the etching is improved at low temperature. Microwave enhancement of the SF₆/Ar discharges produces increases in etch rates of several times at 25°C, but much lower increases at –30°C substrate temperature. The underlying GaAs surface shows increased S and F coverage after low-temperature etching, but these species are readily removerd either by anex-situ wet chemical cleaning step or an in-situ H₂ plasma exposure. Photoresist etching is less sensitive to temperature, and anisotropic profiles are produced between –30 and +60°C in pure O₂ discharges.     

Electron scattering and dissociative attachment by SF6 and its electrical-discharge by-products

Discrete electron-molecule processes relevant to SF₆ etching plasmas are examined. Absolute, total scattering cross sections for 0.2–12-eV electrons on SF₆, SO₂, SOF₂, SO₂F₂, SOF₄, and SF₄, as well as cross sections for negative-ion formation by attachment of electrons, have been measured. These are used to calculate dissociative-attachment rate coefficients as a function ofE/N for SF₆ by-products in SF₆.     

Formation of microstructures on silicon surface in a fluorinated plasma via the cyclic etching-passivation process

The conditions and causes of formation of microneedles and columnar structures on the surface of silicon during deep anisotropic etching in an SF₆/C₄F₈ plasma in the two-stage cyclic mode were determined. The appearance of microstructures is accelerated with an increase in the thickness of the fluorocarbon film formed on the silicon surface at the step of passivation in a C₄F₈ plasma and with an increase is the rate of etching the film in an SF₆ plasma. By means of X-ray photoelectron spectroscopy, it was shown that the formation of carbon residues of fluorocarbon film etching on the Si surface is enhanced under these conditions. Building-up on the surface in the cyclic process, the residues as a micromasking coating lead to the formation of microneedles and columnar structures.     

Photoassisted etching of silicon dioxide films

Theoretical approaches to resolving the problem of photoassisted etching of silicon dioxide as the most widely used protective layer in microelectronics are presented. A model of donor-acceptor interaction providing for the desolvation of the F^? ion involved in SiO₂ photoetching is proposed. The etchant compositions were optimized, and the effect of the most important factors on the etching process was examined. It has been shown that the maximal photoetching rate is 0.42 μm/min and the chemical contribution to etching is small, being about 0.02 μm/min.     

Gas-phase reactions of SF5, SF2, and SOF with O(3 P): Their significance in plasma processing

Reactions of both SF₅ and SF₂ with O(³ P) and molecular oxygen have been studied at 295 K in a gas flow reactor sampled by a mass spectrometer. For reactions with O(³ P), rate coefficients of (2.0±0.5)×10^–11 cm³ s^–1 and (10.8±2.0)×10^–11 cm³ s^–1 were obtained for SF₅ and SF₂ respectively. The rate coefficients for reactions with O₂ are orders of magnitude lower, with an estimated upper limit of 5×10^–16 cm³ s^–1 for both SF₅ and SF₂. Reaction of SF₂ with O(³ P) leads to the production of SOF which then reacts with O(³ P) with a rate coefficient of (7.9±2.0)×10^–11 cm³ s^–1. Both SO and SO₂ are products in the reaction sequence initiated by reaction between SF₂ and O(³ P). Although considerable uncertainty exists for the heat of formation of SOF, it appears that SO arises only from reaction between SOF and O atoms which is also the source of SO₂. These results are discussed in terms of a reaction scheme proposed earlier to explain processes occurring during the plasma etching of Si in SF₆/O₂ plasmas. A comparison between the results obtained here and those reported earlier for reactions of both CF₃ and CF₂ with O and O₂ shows that there is a marked similarity in the free radical chemistry which occurs in SF₆/O₂ and CF₄/O₂ plasmas.     

A model for the etching of silicon in SF6/O2 plasmas

A model has been developed to describe the chemistry which occurs in SF₆/O₂ plasmas and the etching of silicon in these plasmas. Emphasis is placed nn the gas-phase free radical reactions, and the predictions n( the model are compared with experimental results. Forty-seven reactions are included, although a subset of 18 reactions describes the chemistry equally well. Agreement between the calculated and measured concentrations of stable products downstream of the plasma is better than a factor of 2. The need for additional kinetic data and fàr well-characterized diagnostic studies of SF₆/O₂ plasmas is discussed.     

The mechanism of formation of microneedles on the silicon surface in fluorinated plasma via the cyclic etching-deposition process

The mechanism of the formation of microneedles on the silicon surface in SF₆/C₄F₈ plasmas in the two-stage cyclic etching/deposition process is proposed. By means of scanning electron microscopy, it was shown that microneedle growth nuclei are nanosized entities of carbon nanofilaments. They are formed on the Si surface during the reactive ion etching of a fluorocarbon polymer film. As the number of etching/deposition cycles increases, the length of filaments increases and, beginning from a certain cycle, the filaments form a network of the fluorocarbon micromask needed for the formation of microneedles. A simulation of the microneedle formation by means of the hybrid string-cell representation of the profile and the Monte Carlo representation of the particle flux showed satisfactory agreement with the experimental data and the proposed mechanism.     

Nanostructured pyramidal black silicon with ultra-low reflectance and high passivation

In this study, nanostructured pyramidal black silicon is prepared by metal assisted chemical etching method, in which the silver nitrate (AgNO₃) is used as the metal catalyst. Effects of the concentration of AgNO₃ on passivation and optical properties of the black silicon are investigated. The experimental results show that at the AgNO₃ concentration of 0.03 M, the nanostructure length is about 300 nm, and the reflectance of the black silicon with a stack of silicon nitride (SiN_x) and aluminum oxide (Al₂O₃) is 0.8%, which is comparable to that of the conventional black silicon with micrometer-long nanowires. In addition, an acceptably low surface recombination rate of 42 cm/s can be obtained. Plasma chemical vapor deposited SiN_x is deposited well on the top of nanostructures of black silicon, but shows poor coverage at the bottom region. Spatial atomic layer deposited Al₂O₃ can conformally cover the nanostructures with high passivation quality. Simulation result indicates an improvement of 5.5% of conversion efficiency for the nanostructured pyramidal black silicon solar cell compared to industrial silicon solar cell. The short nanostructured pyramidal surface with low reflectance and high passivation is expected to be helpful for black silicon technology applied to photovoltaic applications.     

Mathematical simulation of gas-phase deposition of epitaxial silicon films in the Si-C-H and Si-H systems

The effect of experimental conditions on the magnitude and uniformity of the deposition rate of epitaxial silicon obtained by chemical deposition from the gas phase in the SiCl₄-H₂, SiHCl₃-H₂, and SiH₄-H₂ systems (in the temperature ranges from 1300 to 1520 K for the chloride and 1270 to 1370 K for the silane systems) has been examined. Chloride and silane processes are compared.Translated fromIzvestiya Akademii Nauk. Seriya Khimicheskaya, No. 7, pp. 1217–1222, July, 1995.     

Gas-phase reactions in plasmas of SF6 with O2 in He

Processes which occur in microwave discharges of dilute mixtures of SF₆ and O₂ in He have been examined using a flow reactor sampled by a mass spectrometer. Two classes of experiments were performed. In the first set of experiments, mixtures containing 6×10¹¹ cm^–3 SF₆, 6×10¹⁶ cm^–3 He, and O₂ in the range (0–3.6)×10¹³ cm^–3 were passed through a 20-W 2450-MHz microwave discharge. The gas mixtures arriving at a sample point downstream from the discharge were examined for SF₆, SF₄, SOF₂, SOF₄, SO₂F₂, SO₂, F, and O. In the second class of experiments, rate coefficients were measured for the reactions of SF₄ with O and O₂ and for the reaction of SF with O. The rate coefficient for the reaction of SF with O was found to be (4.2±1.5)×10^–11 cm^–3 s^–1. SF₄ was found to react so slowly with both oxygen atoms and oxygen molecules that only upper limits could be placed on the rate coefficients for these reactions. These values were 2×10^–14 cm³ s^–1 and 5×10^–15 cm³ s^–1 for reactions with O and O₂ respectively. The observed distribution of products from the discharged mixtures is discussed in terms of the measured rate coefficients.     

Etching of Silicon Nitride in CCl2F2, CHF3, SiF4, and SF6 Reactive Plasma: A Comparative Study

Silicon nitride is an important material layer in various types of microelectronic devices. Because of continuous integration of devices, patterning of this layer requires a highly selective and anisotropic etching process. Reactive ion etching is one of the most simple and popular plasma processes. The present work is an experimental analysis of primary etch characteristics in reactive ion etching of silicon nitride using chlorine- and/or fluorine-based organic and inorganic chemistries (CCl ₂ F ₂+O ₂ , CHF ₃+O ₂ , SiF ₄ +O₂, SF₆+O ₂ , and SF ₆+He) in order to obtain a simultaneous etch selectivity against polysilicon and silicon dioxide. A recipe, in CCl ₂ F ₂ /O ₂ plasma chemistry, which provides acceptable etch characteristics, along with a reasonable simultaneous selectivity against polysilicon and silicon dioxide, has been formulated.     

Gas-phase combination reactions of SF4 and SF5 with F in plasmas of SF6

Reactions of both SF₄ and SF₅ with F have been studied at 295 K in a gas-flow reactor sampled by a mass spectrometer. The rate coefficient for the combination reaction of F with SF₄ to produce SF₅ was found to increase from (0.9 to 3.0)×10^–12 cm³ s^–1 when the helium bath gas number density was increased from (2 to 26)×10¹⁶ cm^–3. The values obtained here are three orders of magnitude higher than a recent estimate of the high-pressure value based on the modelling of photochemical studies. The experimental results have been compared with RRKM and master equation calculations in which a simplified Gorin model has been used to determine the structure of the transition state. These calculations show that reasonable agreement can be obtained between the experimental data and the calculation if a small (2 KJ/mol) activation energy is assumed. The rate coefficient for the reaction between SF₅ and F to produce SF₆ was found to be independent of helium bath gas number density within the range given above. The value obtained for the rate coefficient was 9×10^–12 cm³ s^–1 with an uncertainty of a factor of 2. This value is close to that of 1×10^–11 cm³ s^–1 computed from the simplified Gorin model and to the value of 1.7×10^–11 cm³ s^–1 deduced from modelling of photochemical experiments.     

Mass spectrometer-wall probe diagnostic of Ar discharges containing SF6 and/or O2: Reactive ions in etching plasmas

Positive and negative ions of Ar/SF₆ and Ar/SF₆/O₂ plasmas (etching plasmas) and of Ar/O₂ plasmas (cleaning plasmas) in Pyrex tubes have been investigated using a mass spectrometer-wall probe diagnostic technique. The measurement of negative ions proved to be a very sensitive method for the detection of wall material. In etching plasmas with small admixtures of SF₆, oxygen was found as the only representative of wall material. At larger amounts of SF₆, silicon could be detected. In cleaning plasmas with small admixtures of O₂ applied to a previously etched Pyrex surface, fluorine was found, indicating the reversal of fluoridation by oxygenation.     

Dry etching characteristics of III–V semiconductors in microwave BCl3 discharges

Electron cyclotron resonance (ECR) BCl₃ discharges with additional rf biasing of the sample position have been used to etch a variety of III–V semiconductors. GaAs and Al_xGa_1–xAs (x = 0–1) etch at equal rates in BCl₃ or BCl₃/Ar discharges, whereas SF₆ addition produces high selectivities for etching GaAs over AlGaAs. These selectivities are in excess of 600 for dc biases of –150 V, and fall to 6 for biases of –300 V. If the dc biases are kept to – 100 V, there is no measurable degradation of the optical properties of the GaAs and AlGaAs. The AlF₃ formed on the AlGaAs surface during exposure to BCl₃/SF₆ plasmas can be removed by sequential rinsing in dilute NH₄OH and water. In-based materials (InP, InAs, InSb, InGaAs) etch at slow rates with relatively rough morphologies in BCl₃ plasmas.     

Sorption and Diffusion of SF6 in Silicalite Crystals

Kinetic and equilibrium data for sorption of SF₆ in silicalite have been determined by the ZLC method. The equilibrium constants and adsorption energy are comparable with the values reported previously for SF₆-NaX. Intracrystalline diffusion is relatively rapid with diffusivities of order 10_–7– 10^–8 cm²·s^–1 at temperatures in the range 30–90°C. The implications for the use of SF₆ as a probe molecule for assessing the integrity of silicalite membranes are considered.     

Infrared Spectroscopic Analysis of Plasma-Treated Si(100)-Surfaces

 Infrared reflection spectroscopy (specular reflection, attenuated total reflection) has been applied in combination with spectroscopic ellipsometry and electron microscopy to analyze the surface structure of plasma-treated Si(100) surfaces. It is shown that plasma treatments in oxygen and fluorine or chlorine-containing gases cause the formation of a thin surface layer having thicknesses of a few nanometers. The layer was identified to consist of SiO₂ for treatments in an oxygen plasma. Analyses of layers formed by treatments in a fluorine-containing plasma do not confirm the generally assumed model. Different Si-F vibration modes were identified in the surface layer caused by a SF₆ plasma. They correlate, however, with SiF and SiF₂ molecules. There are no indications of the existence of the generally assumed SiF₄. Neither has SiOF₂ been proven in layers produced by etching in a SF₆/O₂ plasma.     

Mass spectrometric study of SF6-N2 plasma during etching of silicon and tungsten

The reaction products in the SF₆-N₂ mixture rf plasma during reactive ion etching of Si and W have been measured by a mass spectrometric method. Two kinds of cathode materials were used in this work; they were stainless steel for the Si etching, and SiO₂ for the W etching. The main products detected in the etching experiments of Si and W included SF₄, SF₂, SO₂, SOF₂, SOF₄, SO₂F₂, NSF, NF₃, N₂F₄, N_xS_y, NO₂, and SiF₄. In the W etching with the SiO₂ cathode, additional S₂F₂, N₂O, and WF₆ molecules were also obtained. The formation reactions about the novel NSF compound and the sulfur oxyfuorides were discussed.     

Mechanism of etching and of surface modification of polyimide in RF and LF SF6-O2 discharges

Etch rates of Kapton H polyimide film in SF₆-O₂ plasmas (0.25 torr) were studied as a function of the input gas mixture, the excitation frequency (25–450 kHz; 13.56 MHz), and the biasing mode. The treated surface was examined by X ray photoelectron spectroscopy (ESCA), scanning electron microscopy (SEM), and contact angle measurement. The ion and neutral species of the plasma were sampled and analyzed by mass spectrometry. Etch rates are found to depend on the positive ion flux and the degree of dissociation of neutral molecules. Plasma-treated surfaces are always covered with a deposited material (C_nH_mO_xF_y) which partially obstructs the etching reaction by a masking effect and causes surface roughness. A proposed kinetic analysis of the etching mechanism is in good agreement with the experimental data.