Method for forming SIO2 by chemical vapor deposition at room temperature

BACKGROUND OF THE INVENTION

[0002] Chemical vapor deposition (CVD) is one of the most widely practiced thin film deposition techniques. Advances in CVD technology have fueled the deployment of many new technologies, including silicon microelectronic processing. One of the key goals for the fabrication of future silicon devices is lower deposition temperatures. These low growth temperatures will limit interlayer and dopant diffusion and facilitate the use of temperature sensitive substances such as polymers or biological materials.

[0003] The chemical vapor deposition of SiO2 is ubiquitous in silicon device fabrication. SiO2 CVD films compete effectively with thermal SiO2 that is formed by the reaction of oxygen with the silicon substrate at 900-1200 K. SiO2 CVD is performed at various temperatures that can be significantly lower than the required temperatures for thermal SiO2 growth. At high temperatures of ˜1200 K, excellent SiO2 films with properties close to thermal SiO2 can be grown using the reaction SiH2Cl2+2NO→SiO2+2N2+2HCl. At medium temperatures of ˜900-1000 K, very reasonable SiO2 dielectric films are deposited using tetraethyl orthosilicate (TEOS) decomposition. Several earlier investigations have reported the kinetics of SiO2 CVD using SiCl4 and H2O. These studies observed efficient SiO2 CVD only at temperatures greater than 900 K. At fairly low temperatures of ˜500-700 K, SiO2 films with a lower density than thermal SiO2 can be deposited using the reaction SiH4+O2→SiO2+2H2O.

[0004] SiO2 deposition at temperatures as low as room temperature has been the focus of recent research. Plasma processing is often used to lower film deposition temperatures. However, the drawbacks to plasma processing are particle contamination, surface damage from the energetic plasma species and high interface defect density. The use of novel molecular precursors has also been explored for low temperature SiO2 growth. However, these precursors are usually expensive, and the SiO2 films deposited with these precursors have not been device quality.

[0005] Amine catalysts have been used for the attachment of chlorosilanes and organosilanes to silica surfaces. See C. P. Tripp and M. L. Hair, J. Phys. Chem. 97, 5693 (1993) and J. P. Blitz et al., J. Amer. Chem. Soc. 109, 7141 (1987). The use of a catalyst has also recently been reported for SiO2 atomic layer deposition (ALD) using sequential surface reactions. These SiO2 ALD investigations used either pyridine (C5H5N) or ammonia (NH3) as the catalyst during sequential exposures to SiCl4 and H2O. See J. W. Klaus, O. Sneh, A. W. Ott and S. M. George, Surface Review & Letters 6, 435 (1999) and J. W. Klaus, O. Sneh, A. W. Ott and S. M. George, Science 278, 1934 (1997). These previous studies demonstrated that Lewis base molecules can catalyze SiO2 ALD at room temperature. Recent ab initio theoretical calculations have confirmed the catalytic effect of NH3 on the SiCl4 and H2O half-reactions occurring during SiO2 ALD. Y. Okamoto, J. Phys. Chem. B 103, 11074 (1999).

BRIEF DESCRIPTION OF THE DRAWINGS

[0006]
FIG. 1 is a graphical depiction illustrating the sequencing of reactants and catalyst addition for Example 1.

[0007]
FIG. 2 is a graph of SiO2 film thickness deposited by 30 consecutive NH3/SiCl4 pulses at 600 mTorr and 318 K versus NH3 pressure under the conditions of Example 2.

[0008]
FIG. 3 is a graph of NH3 thickness on the hydroxylated SiO2 surface at 318 K versus NH3 pressure under the conditions of Example 2.

[0009]
FIG. 4 is a graph of SiO2 film thickness deposited by 30 consecutive NH3/SiCl4 pulses at 100 mTorr and 318 K versus H2O pressure under the conditions of Example 3.

[0010]
FIG. 5 is a graph of H2O thickness on a hydroxylated SiO2 surface at 318 K versus H2O pressure under the conditions of Example 3.

[0011]
FIG. 6 is a graph of SiO2 film thickness deposited by 30 consecutive NH3/SiCl4 pulses at 600 mTorr H2O, 150 mTorr NH3 and 318 K versus SiCl4 pulse length under the conditions of Example 4.

[0012]
FIG. 7 is a graph of total SiO2 film thickness deposited versus number of SiCl4 pulses at 100 mTorr NH3, 300 mTorr H2O and 318 K under the conditions of Example 4.

[0013]
FIG. 8 is an atomic force microscope image of a SiO2 film deposited on Si(100) by 500 consecutive NH3/SiCl4 pulses at 600 mTorr H2O, 600 mTorr NH3 and 318 K under the conditions of Example 5.

[0014]
FIG. 9 is a graph of SiO2 film thickness deposited by 30 consecutive NH3/SiCl4 pulses at 600 mTorr H2O and 150 mTorr NH3 versus substrate temperature under the conditions of example 6.

SUMMARY OF THE INVENTION

[0015] This invention is a chemical vapor deposition process for forming a SiO2 layer on a substrate comprising reacting water with a silicon precursor compound having the structure SiX4, Si(NR2)4, Si(OH)a(OR)4-a or SiHb(OR)4-b wherein each R is an alkyl group, each X is independently a halogen atom, and a and b are numbers from 0-4, in the presence of the substrate at a temperature of between about 290 K and 350 K and in the presence of ammonia or a Lewis base that is a gas under the conditions of the chemical vapor deposition process.

[0016] This process permits the formation of high quality SiO2 coatings on various substrates, most notably on a silicon substrate, using a low temperature process. The process is easily controllable to form conformal SiO2 layers of a desired thickness. Lower deposition temperatures are particularly important to limit interlayer and dopant diffusion as thin film device sizes approach nanoscale dimensions.

DETAILED DESCRIPTION OF THE INVENTION

[0017] In this invention, an SiO2 layer is formed on a substrate by the reaction of water with a silicon precursor compound. The silicon precursor compound has the structure SiX4, Si(NR2)4, Si(OH)a(OR)4-a or SiHb(OR)4-b wherein each R is an alkyl group, each X is independently a halogen atom, and a and b are numbers from 0-4. X is preferably chloride ion. R is preferably a C1-C4 alkyl such as methyl, ethyl, n-propyl, isopropyl, n-butyl, t-butyl or sec-butyl. a and b are each preferably from 0-3, more preferably from 0-2 and most especially 0. SiCl4 and tetraethyl orthosilicate (TEOS, where each R is ethyl) are especially preferred silicon precursors. The overall reaction is represented by one or more of the (unbalanced) equations

SiX4+2H2O→SiO2+4HX (I)

Si(NR2)4+2H2O→SiO2+4HNR2 (II)

Si(OH)a(OR)4-a+2H2O→SiO2+4ROH (III) and

SiHb(OR)4-b+2H2O→SiO2+4ROH (IV)

[0018] The substrate can be any substrate on which an SiO2 layer can be deposited. A highly preferred substrate is silicon, especially an Si(100) wafer. In addition to applications in manufacturing semiconductor devices, low temperature SiO2 CVD may be used to deposit insulating and protective SiO2 coatings on a variety of materials, including thermally sensitive materials such as polymers and biological materials. Among such other substrates are nitrides such as aluminum nitride, boron nitride or silicon nitride; oxides such as alumina, titania, or zeolites, metals, carbides such as titanium carbide, boron carbide, silicon carbide, tungsten carbide, and the like. In most cases, the SiO2 layer will become chemically bound to the substrate through oxygen atoms on the substrate surface.

[0019] The catalyst is ammonia or a Lewis base that is a gas under the conditions of the reaction. The Lewis base preferably has a nitrogen atom. Those having H-bond strengths towards Lewis acid sites similar to ammonia and pyridine are especially suitable. Among such catalysts are pyridine, primary, secondary or tertiary amines such as monoalkyl amines (ethyl amine, n-propyl amine, n-butyl amine, t-butyl amine and the like); dialkyl amines such as dimethyl amine, diethyl amine, di-n-propyl amine, methyl ethyl amine and the like) and trialkyl amines (trimethyl amine, triethyl amine and the like). Ammonia is most preferred on the basis of catalytic activity and ease of use.

[0020] In accordance with the invention, the foregoing reaction is conducted in a chemical vapor deposition process. In general, this process comprises subjecting the substrate to the reactant under conditions of temperature and pressure so that the reactants and catalyst are in the gaseous state. The reactants react on the surface of the substrate to form an SiO2 layer on its surface. At the temperature range of interest, 290-350 K, especially 313-333 K, pressures are selected so that the reactants assume a gaseous state. In general, the substrate is contacted with reactants at a pressure that is preferably less than 50 Torr, more preferably less than 10 Torr and even more preferably less than 5 Torr. Thus, the process is generally conducted within an evacuated chamber into which the reactants and catalysts can be fed under controlled subatmospheric pressures. The reaction vessel is preferably adapted to be evacuated to low pressures such as 1×10−6 Torr, preferably 1×10−7 Torr, especially 1×10−8 Torr. Suitable such equipment is described in A. W. Ott, J. W. Klaus, J. M. Johnson and S. M. George, Thin Solid Films 292, 135 (1996), incorporated herein by reference.

[0021] The chemical vapor deposition process includes the simultaneous addition of the reactants, or at least one of the reactants and the catalyst, so that some mixing occurs in the vapor phase. Typically, the silicon precursor is added simultaneously with water or the catalyst, preferably both. A preferred mode of addition is to introduce water vapor continuously or over relatively long periods. During the water addition, catalyst and silicon precursor are added intermittently in “pulses” of short duration. In this manner, there are periods of time in which only water is being introduced into the reaction, and other periods of time in which water and catalyst are being added simultaneously and still other periods of time in which water and the silicon precursor (and still more preferably, the catalyst) are added simultaneously. This permits the surface of the substrate to first become coated with a thin layer of water molecules, which are then coated with a thin layer of catalyst molecules, and then with a thin layer of the silicon precursor, at which point the reaction between the water and silicon precursor takes place, forming a very thin SiO2 layer. A certain amount of mixing of the reactants and/or catalyst will occur in the vapor phase, and a certain amount of simultaneous deposition of these materials onto the substrate will occur. Repeated sequential applications of the reactants and catalyst form or increase the thickness of the layer. It is believed that a contiguous SiO2 layer may be created from “islands” of SiO2 that are formed during the first or first several applications of the reactants and catalysts, and which grow and are joined together during subsequent applications of the reactants and catalysts.

[0022] In an especially preferred method, water vapor is used not only as a reactant, but also as purge stream that helps to flush away excess quantities of the catalyst and silicon precursor as well as by-products of the reaction. In this especially preferred method, water vapor can be introduced continuously into the reactor, with the catalyst and silicon precursor being “pulsed” (i.e., added to the reactor under pressure for an interval of time) into the reactor periodically. The catalyst and silicon precursor “pulses” are preferably performed with the catalyst being introduced first, followed by the silicon precursor. In an especially preferred process, the catalyst is pulsed into the reactor, and the silicon precursor is pulsed into the reactor near the end of the catalyst pulse, so that the silicon precursor is introduced simultaneously with the last portion of the catalyst pulse. Between pulses of catalyst/silicon precursor, it is preferred to purge the reactor with water or an inert gas, in order to remove excess raw materials and by-products of the reaction.

[0023] The amount of reactants and catalyst that are introduced in a given addition will depend on the pressure and the length of time of the addition. For catalyst and silicon precursor additions, it has been found that most efficient SiO2 layer growth is achieved when each addition is sufficient to deposit approximately a monolayer of molecules on the surface of the substrate. Adding greater amounts contributes little to the rate of SiO2 growth and may tend to cause the SiO2 layer to become less uniform, whereas adding substantially less tends to slow the rate of SiO2 layer growth.

[0024] For any particular apparatus and substrate, suitable pulse lengths and pressures can be determined empirically. A preferred set of conditions includes adding H2O in pulses or continuously until an H2O partial pressure of about 50 to about 5,000 mTorr, especially 100 to 500 mTorr, is achieved. Catalyst pulsing is preferably done to achieve a catalyst partial pressure of about 50 to about 2500 mTorr, preferably about 50 to about 1000 mTorr, more about 100 to about 500 mTorr, especially about 100 to 300 mTorr. Silicon precursor pulsing is preferably done to achieve silicon precursor partial pressure of from about 2 to about 2500 mTorr, preferably from about 5 to about 500 mTorr, more preferably from about 5 to about 200 mTorr.

[0025] The process is continued until a film of SiO2 of the desired thickness is deposited onto the substrate. The film thickness may range from about 10 nm to any desired thickness, provided that the deposition process is repeated enough times. The process of the invention is particularly suitable for forming layers of from about 50 to about 1000 nm.

[0026] HCl, dialkylamines and/or alcohols are produced when the water reacts with the various silicon precursors. The catalyst may then form reaction by-products, particularly ammonium chlorides when HCl reacts with NH3. These by-products may become affixed to the surface of the growing SiO2 film. Purging the reactor between reaction sequences tends to remove these by-products as well as other impurities that may form. Thus, it is preferred to separate each sequence of reactions with a purge of an inert gas, or preferably, water vapor. The purge may continue for up to 1 minute or more, depending on the particular conditions and design of the reactor, but often a purge of from about 1 to about 50 seconds, especially from about 10 to about 50 seconds, is adequate to remove most of the by-products and form a good quality film.

[0027] The following examples are provided to illustrate the invention, but are not intended to limit its scope. All parts and percentages are by weight unless otherwise indicated. These examples are conducted in an apparatus as described in A. W. Ott, J. W. Klaus, J. M. Johnson and S. M. George, Thin Solid Films 292, 135 (1996). The apparatus includes a sample load lock chamber, a central deposition chamber, and an ultra high vacuum chamber for surface analysis. The central deposition chamber is designed for automated dosing of gas phase precursors under a wide variety of conditions. The deposition chamber is pumped with either a 175 l/s diffusion pump equipped with a liquid N2 trap or two separate liquid N2 traps backed by mechanical pumps. This chamber has a base pressure of 1×10−7 Torr. The central deposition chamber contains an in situ spectroscopic ellipsometer (J. A. Woolam Co. M-44) that collects ellipsometric data simultaneously at 44 visible wavelengths. The ellipsometer is mounted on ports positioned at 80° with respect to the surface normal. Gate valves on the ports protect the birefringent-free ellipsometer windows from deposition during film growth. The surface analysis chamber houses a UTI-100C quadrupole mass spectrometer and is pumped by a 210 l/s turbo pump to a base pressure of 1×10−8 Torr. Mass spectometric analysis of the gases in the central deposition chamber can be performed using a controlled leak to the surface analysis chamber.

[0028] The sample substrates were 0.75×0.75 inch Si(100) wafers. The Si(100) samples were p-type, boron-doped with a resistivity of ρ=0.1−0.4 Ωcm. A 3000 Å Mo film deposited on the backs of the samples was used for resistive heating to >1100 K. The Si(100) samples were cleaned in HF to remove the native oxide. The Si(100) surface was further cleaned in vacuum by annealing at 875 K for 1 minute. This anneal was followed by a high frequency H2O plasma discharge at 300 K to hydroxylate the surface and remove surface carbon contamination. The cleaning procedure leaves a very thin SiO2 film on the Si(100) substrate.

EXAMPLE 1

SiO2 Deposition Onto Si(100) Wafers

[0029] SiO2 deposition experiments were performed using the reaction sequence illustrated in FIG. 1. A steady-state pressure of H20 was maintained in the central deposition chamber by continuously flowing H2O through the chamber. The H2O pressure was controlled by varying the conductance between the chamber and the pumps using an in-line butterfly valve. The NH3 catalyst and SiCl4 precursor were then introduced into the deposition chamber using automated valves. The NH3 pressure was controlled using the regulator on the NH3 gas cylinder. The SiCl4 precursor was pulsed into the deposition chamber after establishing the NH3 pressure. The SiCl4 exposure was defined by opening the SiCl4 automated valve for 50 ms. This SiCl4 pulse yielded a peak SiCl4 pressure of ˜5 mTorr as measured by a capacitance manometer on the deposition chamber. Following the NH3/SiCl4 pulse sequence, the NH3 and SiCl4 valves were closed, and the continuous H2O flow purged the chamber. A purge time of 20-60 secs was required to reestablish the steady-state H2O pressure.

[0030] The NH3 catalyst can form complexes with HCl reaction product to yield NH4Cl salt. NH4Cl has a vapor pressure of 4×10−5 Torr at 300 K. The H2O purge also removed the NH4Cl from the SiO2 surface and the chamber walls.

[0031] The SiO2 film thickness was evaluated after 30 identical NH3/SiCl4 pulses using the in situ spectroscopic ellipsometer. Bulk SiO2 optical constants were initially assumed because the refractive index of very thin films is difficult to determine. The accuracy of this assumption was tested by collecting ex situ data on a 1000 Å thick SiO2 film at multiple angles of incidence. This examination permits an accurate determination of the SiO2 optical constants. The measured index of refraction of 1.43±0.03 for the catalyzed SiO2 CVD films agrees with the refractive index of 1.46 for dense thermal SiO2 films.

EXAMPLE 2

Comparison of Film Growth With and Without NH3 Catalyst

[0032] When example 1 is repeated without using any NH3 catalyst, no SiO2 CVD film growth was observed at any temperature between 300-700 K for SiCl4 and at H2O pressures as high as to 2 Torr. In contrast, the addition of a small amount of NH3 initiated immediate SiO2 growth at 313-333 K. The effect of the NH3 catalyst was quantified by measuring the SiO2 film thickness deposited by 30 consecutive NH3/SiCl4 pulses at 318 K versus NH3 pressure. These in situ spectroscopic ellipsometry measurements are shown in FIG. 2. The solid line is intended only to guide the eye. The steady-state H2O pressure was 600 mTorr and the SiCl4 pulse length was fixed at 50 ms for the results shown in FIG. 2. H2O was allowed to purge the deposition chamber for 20 seconds between each NH3/SiCl4 pulse. The H2O/NH3/SiCl4 reaction cycle occurred with the following sequence: Flow H2O (600 mTorr)/Flow NH3(0-3 Torr)/Pulse SiCl4(50 ms)/Close SiCl4 and NH3 valves/Flow H2O (20 sec). FIG. 2 shows that the deposited SiO2 film thickness increases rapidly with increasing NH3 pressures up to 3.0 Torr. NH3 pressures >3.0 Torr were not attainable during the reaction cycles when the deposition chamber was connected to the pumps. Given the total time required for the 30 NH3/SiCl4 pulses, the catalyzed SiO2 CVD rates varied from 0.4 Å/min at 50 mTorr NH3 to 8.1 Å/min at 3.0 Torr NH3.

[0033] The data in FIG. 2 suggests that enhancement of SiO2 CVD with NH3 pressure results from a progressive increase in NH3 coverage with NH3 pressure. In situ ellipsometry was used to measure the NH3 coverage on a fully hydroxylated SiO2 surface. A refractive index of n=1.325 was utilized based on the previous measurements for liquid ammonia at 289 K as reported by W. L. Jolly and C. J. Hallada, “Liquid Ammonia” in Non-Aqueous Solvent Systems, ed. by T. C. Waddington; Academic Press: London, 1965, Chapter 1. FIG. 3 displays the NH3 coverage versus NH3 pressure at 318 K. For NH3 pressures <1 Torr, the NH3 thickness increases rapidly with pressure. In this monolayer regime, the NH3 molecules are adsorbing strongly to surface SiOH* groups. (As used herein, an asterisk * denotes an active surface species.) The NH3 thickness increases much more slowly at NH3 pressures >1 Torr. At these pressures, the NH3 molecules are adsorbing less strongly onto themselves and form a NH3 multilayer.

[0034] The Brunauer-Emmett-Teller (BET) adsorption model was used to fit the ellipsometric data. See J. B. Hudson, Surface Science: An Introduction; Butterworth-Heinemann: Troy, N.Y., 1992. The BET model relates NH3 coverage to: τ1, the NH3 lifetime on the silanol surface; τ2, the NH3 lifetime on the ammonia multilayer; n0, the NH3 density in one complete monolayer; and I, the incident NH3 flux. A molecular density in the first monolayer of n0=7.6×1014 molecules/cm2 was obtained from the density of liquid ammonia. The density of liquid ammonia also yields a thickness of 3.6 Å for the NH3 monolayer. The BET fit is shown as the solid line in FIG. 3. The NH3 lifetimes of τ1=3×10−6 s and τ2=7×10−9 s at 318 K were obtained from this fit.

[0035] Additional ellipsometric measurements of the NH3 coverage versus NH3 pressure at different temperatures were consistent with an adsorption energy of E=9.4±0.8 kcal/mol and a desorption preexponential of v0=1.0×1012±0.8 s−1 for NH3 on the silanol surface. Assuming a desorption preexponential of v0=1.0×1012 s−1, the NH3 lifetime of τ2=7×10−9 s yielded an adsorption energy of 5.6 kcal/mol for NH3 on the ammonia multilayer. This adsorption energy is in excellent agreement with the heat of vaporation of 5.58 kcal/mol for ammonia at the boiling point.

[0036]
FIG. 2 displays two SiO2 CVD growth regimes versus NH3 pressure. For NH3 pressures <1 Torr, the SiO2 film thickness deposited by 30 NH3/SiCl4 pulses increases nearly linearly with NH3 pressure. In contrast, the SiO2 film thickness deposited by 30 NH3/SiCl4 pulses increases more slowly for NH3 pressures>1 Torr. This SiO2 growth behavior is explained by the NH3 adsorption isotherm at 318 K displayed in FIG. 3.

[0037] The NH3 adsorption isotherm shown in FIG. 3 exhibits a turnover at 1 Torr. At this pressure, the NH3 molecules have nearly completed the first monolayer formation by binding to most of the SiOH* surface species. At pressures >1 Torr, the NH3 molecules adsorb less strongly as a result of NH3 multilayer formation. This decrease in the NH3 adsorption efficiency at >1 Torr is directly reflected by the SiO2 deposition versus NH3 pressure.

EXAMPLE 3

Dependence of SiO2 CVD on H2O Exposure

[0038] The dependence of catalyzed SiO2 CVD on the H2O exposure was examined by measuring the SiO2 film thickness deposited by 30 consecutive NH3/SiCl4 pulses at 318 K for various H2O pressures. The NH3 pressure was 100 mTorr. Each SiCl4 pulse was defined by opening the automated valve for 50 ms to produce a peak SiCl4 pressure of ˜5 mTorr in the chamber. The H2O was allowed to purge the chamber for 20 seconds between each NH3/SiCl4 pulse.

[0039]
FIG. 4 shows that the SiO2 deposited by 30 NH3/SiCl4 pulses increases rapidly with increasing H2O pressure from 150 mTorr to 1.5 Torr. The solid line is shown only to guide the eye. Based on the time needed for the 30 NH3/SiCl4 pulses, the catalyzed SiO2 CVD rate varied from 0.12 Å/min at 150 mTorr H2O to 3.0 Å/min at 1.5 Torr H2O.

[0040] The enhancement of the SiO2 CVD versus H2O pressure results from a progressive increase in the H2O coverage with H2O pressure. In situ ellipsometry was used to measure the H2O coverage on a fully hydroxylated SiO2 surface. A refractive index of n=1.33±0.05 was determined for the H2O multilayer. FIG. 5 displays the H2O coverage on the silanol surface versus static H2O pressure at 318 K. For H2O pressures <500 mTorr, the H2O thickness increases rapidly with pressure. In this monolayer regime, the H2O molecules are adsorbing strongly to surface SiOH* groups. The H2O thickness increases more slowly at H2O pressures >500 mTorr. At these H2O pressures, the H2O molecules are adsorbing less strongly onto themselves and form a H2O multilayer.

[0041] The Brunauer-Emmett-Teller (BET) adsorption model was again utilized to fit these ellipsometry results. This fit is displayed as the solid line in FIG. 5. From this data and measurements at additional temperatures, the BET model yielded a H2O surface adsorption energy of E=12.1±0.9 kcal/mol and a desorption preexponential of v0=3.3×1014±0.9 s−1 on the silanol surface at 318 K. A comparison of . 4 and 5 indicates that the SiO2 CVD rate and the H2O surface coverage both increase versus H2O pressure. This correspondence continues in the H2O multilayer regime at H2O pressures >500 mTorr. In contrast to the NH3 dependence, the dependence of the catalyzed SiO2 CVD on H2O pressure displayed in FIG. 4 first increases slowly versus H2O pressure and then increases more rapidly at H2O pressures >500 mTorr.

EXAMPLE 4

Dependence of SiO2 CVD on SiCl4 Exposure

[0042] Ellipsometric measurements also examined the dependence of the catalyzed SiO2 CVD on SiCl4 exposure. FIG. 6 displays the SiO2 film thickness deposited by 30 consecutive NH3/SiCl4 pulses at 318 K versus SiCl4 pulse length. The SiCl4 peak pressure during each SiCl4 pulse varied from ˜5 mTorr for 50 msec pulse lengths to ˜150 mTorr for 1000 msec pulse lengths. The H2O pressure was 600 mTorr and the NH3 pressure was 150 mTorr. The H2O was allowed to purge the system for 20 seconds between each NH3/SiCl4 pulse.

[0043]
FIG. 6 indicates that there are two distinct SiO2 CVD growth regimes as the SiCl4 pulse length is varied from 10 msec to 1000 msec. The solid line connects the points and is displayed to direct the eye. SiCl4 pulse lengths <300 msec result in SiO2 deposition that is directly proportional to the SiCl4 pulse length. In contrast, SiCl4 pulse lengths ≧400 msec result in SiO2 deposition that is nearly independent of SiCl4 pulse length.

[0044] The saturation of the SiO2 deposition at longer SiCl4 pulse lengths is believed to result from NH4Cl formation. The NH4Cl salt forms from the complexation of the HCl reaction product with the NH3 catalyst. More HCl reaction product and NH4Cl salt is produced during the longer SiCl4 pulse lengths. Eventually the NH4Cl poisons the surface because the NH4Cl formation rate exceeds the NH4Cl sublimation rate. The vapor pressure of the NH4Cl salt varies from 4×10−5 Torr at 300 K to 3×10−4 Torr at 318 K. The problem of NH4Cl salt formation does not occur with the Si(NR2)4, Si(OH)a(OR)4-a or SiHb(OR)4-b silicon precursors that do not contain halogens.

[0045]
FIG. 7 displays the ellipsometric measurements of the SiO2 film thickness deposited after various numbers of NH3/SiCl4 pulses. The H2O pressure was 300 mTorr and the NH3 pressure was 100 mTorr. The SiCl4 pulse length was fixed at 50 msec. The H2O purged the system for 20 seconds between each NH3/SiCl4 pulse. FIG. 7 shows that the catalyzed SiO2 CVD is directly proportional to the number of NH3/SiCl4 pulses. The measured SiO2 CVD rate was 0.24 Å/pulse or 0.72 Å/min at 318 K. The refractive index of these thicker SiO2 films was determined to be n=1.43±0.03.

[0046] The linear increase in the SiO2 film thickness versus number of NH3/SiCl4 pulses indicates that the SiO2 deposition kinetics remain constant versus time. This behavior argues against permanent NH4Cl salt formation under these reaction conditions. Salt incorporation could poison the SiO2 surface and degrade the SiO2 CVD reaction efficiency. Performing the SiO2 CVD reaction at 318 K enhances the NH4Cl sublimation rate and prevents the NH4Cl salt from impairing the SiO2 CVD reaction.

EXAMPLE 5

Surface Properties of SiO2 Films

[0047] The surface topography of the SiO2 films was imaged using a NanoScope III atomic force microscope (AFM) from Digital Instruments operating in tapping mode. The AFM images were collected within 1-2 hours of removing the samples from vacuum to prevent surface contamination by ambient dust particles. Scan lengths of 250 nm-1.2 μm were performed using a 1.2 μm scanning head. The AFM images were conditioned to remove AFM artifacts using the software provided by Digital Instruments.

[0048]
FIG. 8 displays a side view and top view of a SiO2 film deposited at 318 K after 500 NH3/SiCl4 pulses. The H2O pressure was 600 mTorr and the NH3 pressure was 600 mTorr. The SiCl4 pulse length was fixed at 100 ms. This SiO2 film had a thickness of ˜1000 Å. The surface topography in the AFM image reveals island structures that have grown together and resemble “shingles on a roof”. The root mean square (rms) surface roughness obtained from these AFM images was ±15 Å. In comparison, the roughness of the initial cleaned Si(100) substrate was ±2 Å (rms).

[0049] The observed surface structure in the AFM image suggests that the catalyzed SiO2 CVD nucleates at various sites on the cleaned Si(100) surface. As the SiO2 growth progresses, the individual SiO2 islands eventually merge with each other. The spatial frequency distribution or power spectral density of the SiO2 surface topography was compared directly with the initial cleaned Si(100) substrate. The spatial frequencies that comprise the SiO2 surface topography are weighted more heavily towards lower frequencies or longer wavelengths. This observation is consistent with the larger island features that dominate the surface topography of the catalyzed SiO2 CVD) films.

EXAMPLE 6

Temperature Dependence of SiO2 Films

[0050] The temperature dependence of the catalyzed SiO2 CDV is displayed in FIG. 9. The H2O and NH3 pressures were 600 mTorr and 150 mTorr, respectively. The SiCl4 pulse length was fixed at 50 msec. The SiO2 film thickness deposited by 30 consecutive NH3/SiCl4 pulses decreases as the temperature is increased from 313 K to 333 K. The solid line is shown only to connect the data points and direct the eye. Given the total time required for the 30 NH3/SiCl4 pulses, the SiO2 deposition rate drops from 7.5 Å/min at 313 K to 0.14 Å/min at 333 K. The reduction in the SiO2 growth results primarily from the reduction in the SiCl4, H2O and NH3 coverages at higher temperatures.

[0051] The catalyzed SiO2 CVD exhibits a pronounced decrease with temperature as displayed by FIG. 9. The reduction in SiO2 growth at higher temperatures is the result of two competing processes. The net reaction rate is controlled by the SiCl4, H2O and NH3 coverages and the rate at which the SiCl4 and H2O reactants are converted to products. The coverage of the reactants and catalyst should exhibit an exponential decrease with temperature. In contrast, the reaction rate should display a exponential increase with temperature. The reduction in the catalyzed SiO2 CVD rate with increasing temperature indicates that the SiCl4, H2O and NH3 coverages must dominate the observed SiO2 growth.

EXAMPLE 7

Desorption of NH4Cl

[0052] The reaction of the NH3 catalyst with HCl to form the NH4Cl salt has important implications for catalyzed SiO2 CVD when using the silicon precursors containing chlorine. NH4Cl salt formation can convert the NH3 catalyst to a chemically inactive form.

[0053] The NH4Cl salt can build up on the SiO2 surface during long SiCl4 pulse lengths and degrade the SiO2 deposition. This NH4Cl salt formation explains the saturation of the SiO2 deposition at SiCl4 pulse lengths >400 ms in FIG. 6. However, the effect of the NH4Cl salt can be negligible if NH4Cl deposition is minimized by pulsing the SiCl4 reactant and providing sufficient time for NH4Cl sublimation between SiCl4 pulses. To facilitate the removal of the NH4Cl salt, the SiO2 surface temperature was maintained slightly above room temperature at 313-333 K to promote the sublimation of the NH4Cl salt back to NH3 and HCl. The desorption rate of NH4Cl was rapid enough for complete removal of the NH4Cl salt during the H2O purge of 20 s between NH3/SiCl4 pulses.

[0054] The formation of the NH4Cl salt may also have a beneficial effect on the energetics of catalyzed SiO2 CVD. The formation of NH4Cl from HCl and NH3 is exothermic by 42 kcal/mole. Consequently, NH4Cl formation provides an enhanced thermodynamic driving force for SiO2 formation from H2O and SiCl4 and may help to reduce the overall reaction activation barrier.

[0055] However, NH4Cl salt formation alone does not explain the entire catalytic effect of the amine bases. NH3 also has a strong catalytic effect for the silicon precursors such as Si(OR)4 that do not contain chlorine. This catalytic effect is independent of NH4Cl salt formation because organosilanes do not produce an HCl reaction product that is necessary to form the NH4Cl salt.

[0056] The method of the present invention can be operated at reaction conditions where NH4Cl salt formation does not degrade the SiO2 CVD growth rate. FIG. 7 illustrates that SiO2 films can easily be deposited at rates of 0.24 Å per NH3/SiCl4 pulse or 0.72 Å/min at 318 K. Higher SiO2 deposition rates of >16 Å/min can be achieved by varying the SiCl4, H2O and NH3 pressures.

[0057] Also, the deposited SiO2 films are very smooth as revealed by FIG. 8. The SiO2 films display a surface roughness of only ±15 A (rms) for a deposited SiO2 film thickness of 1000 A. This minimal surface roughness corroborates that NH4Cl salt formation does not interfere with the catalyzed SiO2 growth.

EXAMPLE 8

SiO2 Film Quality and Stoichiometry.

[0058] Rutherford Backscattering (RBS) experiments were performed on several catalyzed SiO2 CVD films to evaluate the film quality and stoichiometry. The RBS investigations employed a 2.45 MeV He2+ primary ion beam. The backscattered ions were collected from the sample at ˜20° off-specular. The H content of the films was also evaluated by collecting the H+ions that were forward-scattered from the SiO2 film.

[0059] The RBS measurements obtained an elemental ratio of 2.1/1 for O/Si. These results reveal that the deposited film is nearly stoichiometric SiO2. The N/Si atomic ratio was below the detection threshold of the RBS measurements, i.e. <0.02%. This small upper limit for N indicates that the NH3 catalyst is not incorporating into the SiO2 film when using the SiCl4 and H2O reactants.

[0060] The RBS analysis also revealed that the Cl/Si atomic ratio was <0.02%. This low ratio is consistent with a complete reaction between the H2O and SiCl4 reactants. Modest levels of carbon (˜1%) and some heavy metal contaminants were also detected in the SiO2 films. The carbon impurities probably are derived from small hydrocarbon partial pressures in the deposition chamber that is pumped by a liquid N2 trapped diffusion pump and operates at a base pressure of 1×10−7 Torr.

[0061] Rutherford backscattering (RBS) investigations did not observe any N or Cl contaminants in the SiO2 films. Thus, although the NH4Cl may interfere with SiO2 deposition under some reaction conditions, there is no evidence that NH4Cl is incorporated in the SiO2 CVD films. These RBS results argue that the growth temperatures of 313-333 K during the SiCl4 and H2O reactions were sufficient for complete removal of the NH4Cl salt from the SiO2 surface during the H2O purge.

[0062] NH3 can react with SiCl4 to form Si3N4 under certain conditions. SiCl4 and NH3 have also been employed earlier to obtain Si3N4 (ALD) using sequential surface reactions. However, the SiCl4 and NH3 half-reactions during Si3N4 ALD have a negligible rate at T <700 K. Further, the reaction between SiCl4 and NH3 in the gas phase is also inconsequential at room temperature.

[0063] In accordance with the present invention, it has been demonstrated that NH3 can be used as a catalyst for the SiCl4+2H2O→SiO2+4HCl and Si(OR)4+2H2O→SiO2+4ROH reactions at room temperature. The precursors are inexpensive and widely available. Because the catalyzed process can be used at such low temperatures, it facilitates the use of temperature sensitive substrates such as polymers or biological materials in electronic devices. The catalyzed CVD process in accordance with the present invention may be applied to the growth of other materials.

Method for forming SIO2 by chemical vapor deposition at room temperature

Information

Publication Number

Date Filed

Date Published

Inventors

CPC

US Classifications

International Classifications

Abstract

Description

Claims

CROSS-REFERENCE TO RELATED APPLICATION

Provisional Applications (1)