Daniel L. Flamm

Plasma etching of Si and SiO2—The effect of oxygen additions to CF4 plasmas

The plasma etching of silicon and silicon dioxide in CF4‐O2 mixtures has been studied as a function of feed‐gas composition in a 13.56‐MHz plasmagenerated in a radial‐flow reactor at 200 W and 0.35 Torr. Conversion of CF4 to stable products (CO, CO2, COF2, and SiF4) and the concentration of free F atoms ([F]) in the plasma were measured using a number of different diagnostics. The rate of etching, the concentration of F atoms, and the intensity of emission from electronically excited F atoms (3s 2 P–3p 2 P° transition at 703.7 nm) each exhibit a maximum value as a function of feed‐gas composition ([O2]); these respective maxima occur at distinct oxygen concentrations. For SiO2, the variation in etching rate with [O2] is accounted for by a proportional variation in [F], the active etchant. The etching of silicon also occurs by a reaction with F atoms, but oxygen competes with F for active surface sites. A quantitative model which takes oxygen adsorption into account is used to relate the etch rate to [F]. The initial increase of [F] with [O2] is accounted for by a sequence of reactions initiating with the production of CF3 radicals by electron impact and followed by a reaction of CF3 with oxygen. When [O2] exceeds ∼23% (under the present discharge conditions), [F] decreases due, probably, to a decrease in electron energy with an increase of oxygen in the feed gas.

The reaction of fluorine atoms with silicon

Fluorine atoms etch silicon with a rate, RF(Si) = 2.91±0.20×10−12T1/2nFe−0.108 eV/kT A/min, where nF (cm−3) is the atom concentration. This etching is accompanied by a chemiluminescent continuum in the gas phase which exhibits the same activation energy. These phenomena are described by the kinetics: (1) F(g)+Sisurf→SiF2(g), (2) SiF2(g) +F(g) →SiF*3(g), (3) SiF2(g) +F2(g) →SiF*3(g) +F(g), (4) SiF*3(g) →SiF3(g) +hνcontinuum where formation of SiF2 is the rate‐limiting step. A detailed model of silicon gasification is presented which accounts for the low atomic fluorine reaction probability (0.00168 at room temperature) and formation of SiF2 as a direct product. Previously reported etch rates of SiO2 by atomic fluorine are high by a constant factor. The etch rate of SiO2 is RF(SiO2) = (6.14±0.49)×10−13nF T1/2e−0.163/kT A/min and the ratio of Si to SiO2 etching by F atoms is (4.74±0.49)e−0.055/kT.

The design of plasma etchants

Theory and practice of plasma etching are critically reviewed. Some unifying principles are extended to explain a large body of experimental data, encompassing more than 20 substrate materials in dozens of etchant gas mixtures. These basic concepts can be used to select new etchants and plasma etching parameters.

On the role of oxygen and hydrogen in diamond‐forming discharges

Plasma emission actinometry has been used to study the mechanism by which small additions of oxygen (∼0.5%) enhance the rate of diamond deposition in a dilute (4%) CH4/H2 discharge at high temperature (900–1300 K). Increasing amounts of CH4 in the feed depress [H], while increasing the O2 concentration, up to ∼5%, produces a fivefold increase in atomic hydrogen in the discharge zone. Invoking a mechanism where diamond growth competes with the formation of an amorphous/graphitic inhibiting layer, these results and earlier studies suggest that oxygen (1) increases [H] which selectively etches amorphous/graphitic carbon, (2) accelerates reaction of this layer with molecular hydrogen, and (3) may itself act as a selective etchant of nondiamond carbon. As a result, the number of active diamond growth sites is increased and enhanced growth rates are observed. We also have grown diamond by alternating a CH4/He discharge with a H2/O2/He discharge and results are consistent with this mechanism. Instantaneous growt...

Comparison of XeF2 and F‐atom reactions with Si and SiO2

Silicon gasification by XeF2 is compared with F‐atom etching under conditions typical of those used in plasma etching. Temperatures ranged from −17 to 360 °C and XeF2 pressures were between 0.05 and 2 Torr. Silicon etching by XeF2 shows a sharply different etch rate/temperature dependence than the Si/F or Si/F2 reaction systems; there is no detectable reaction between XeF2 and SiO2 in contrast to the F‐atom/SiO2 system. These data indicate that physisorption can limit silicon etching by XeF2 and show that basic studies which use XeF2 as a model compound for the etching of silicon and SiO2 by F atoms should be interpreted with caution.

Plasma etching of Si and SiO2 in SF6–O2 mixtures

The products of reaction and etch rates of Si and SiO2 in SF6‐O2 plasmas have been studied as a function of feed composition in an alumina tube reactor at 27 mHz, 45 W, and 1 Torr. There is a broad chemical analogy with CF4‐02 plasmas. As in CF4‐02 mixtures, the rate of Si etching and 703.7‐nm emission from electronically excited F atoms each exhibit distinct maxima as a function of feed gas composition; these data support a model in which fluorine atoms, the etching species, compete with oxygen atoms for chemisorption on the Si surface. Without oxygen in the feed or Si in the reactor, no stable products could be detected. With an SF6‐O2 mixture in the absence of silicon, the final reaction products are F2, SOF4, and SO2F2. The product distribution was unaffected by small SiO2 substrates. When Si is etched, SiF4 is the only stable silicon‐containing etch product and SOF2 is formed in oxygen‐poor mixtures. Rapid etch rates (≳104 A/min for Si) can be obtained with a high selectivity in favor of silicon (Si:...

Diamond crystal growth by plasma chemical vapor deposition

We have grown diamond crystals and polycrystalline diamond films from CH4/H2/O2 gas feeds in a simple, high‐power density, 2450‐MHz discharge tube reactor. Single‐crystal growth rates over 20 μm/h have been achieved. The material has been analyzed using Raman spectroscopy, Auger spectroscopy, and x‐ray diffraction. Control of nucleation is a major problem for growing sound films, and the high temperatures currently required for growth will limit applications. Oxygen additions were necessary to deposit diamonds over the range of feed composition we studied.

Anisotropic etching of SiO2 in low‐frequency CF4/O2 and NF3/Ar plasmas

Anisotropic etching of SiO2 films is reported in low frequency (∼100 kHz), moderate‐pressure (0.35 Torr) CF4/O2 and NF3/Ar plasmas. Rates up to 2000 A/min were achieved with high selectivity over GaAs and InP substrates. The etching mechanism was studied with optical spectroscopy and downstream chemical titrations. Anisotropy is attributed to ion‐enhanced reactivity of fluorine atoms with SiO2 at rates up to two hundred times larger than purely chemical etching by fluorine atoms. Damage and product sputter desorption models of this process were evaluated. These two models are nearly mathematically equivalent at steady state, and show that the effectiveness of ions in etching by enhanced reaction is roughly 15 times that in physical sputtering under these conditions.

Hydrogen passivation of point defects in silicon

Laser melting of crystalline silicon introduces electrically active defects which are observed by capacitance transient spectroscopy. The electrical activity of these point defects is neutralized by reaction with atomic hydrogen at 200 °C.

Basic chemistry and mechanisms of plasma etching

A recent review of plasma etching is extended with discussions of similarity variables governing discharges, anisotropic oxide etching in fluorine and unsaturate‐rich plasmas, surface texture, the loading effect, and gas‐surface reactions.