Silicon Carbide Single Crystal and Method of Etching the Same

Information

  • Patent Application
  • 20080050301
  • Publication Number
    20080050301
  • Date Filed
    July 06, 2005
    19 years ago
  • Date Published
    February 28, 2008
    16 years ago
Abstract
Single-crystal silicon carbide is obtained, which finds a wide variety of applications including semiconductors, and an etching method for single-crystal silicon carbide is provided, by which nitrogen trifluoride plasma is used in single-crystal silicon carbide to obtain a smooth surface. In order to obtain single-crystal silicon carbide having the smoothness (surface roughness) within 150 nm and the material, nitrogen trifluoride-containing gas is subjected to plasma excitation to obtain single-crystal silicon carbide with a smooth surface. It is preferable that the pressure of nitrogen trifluoride gas is in the range of 0.5 to 10 Pa. It is also preferable that the flow rate of the nitrogen trifluoride gas is in the range of 5 to 15 sccm.
Description
TECHNICAL FIELD

The present invention relates to single-crystal silicon carbide and the etching method thereof and, more particularly, relates to an etching method for single-crystal silicon carbide by which nitrogen trifluoride is brought into contact with single-crystal silicon carbide.


BACKGROUND ART

Our daily lives are increasingly dependent on electric energy and demand for electric power is expected to grow over the long term. Under these circumstances, it is particularly important to save conversion loss of electric power in electric energy systems. It has already been found that silicon semiconductors most commonly used in currently existing semiconductor electronics are theoretically limited in terms of physical properties. There are problems that no further reduction in the conversion loss of electric power or the operation of device at a higher speed, at a higher temperature or with high pressure resistivity can be expected. As such, silicon carbide has become the focus of attention as a semiconductor material for the next generation.


Silicon carbide is characterized by a wide band gap, high-thermal conductivity, low-thermal expansion ratio, etc., and expected to find applications such as high-power/high-frequency devices and power devices as a semiconductor material, because of the physical properties which are not achieved by silicon. It is, however, extremely difficult to provide high-precision processing to silicon carbide due to its great thermal and chemical stability. Chemical etching treatment of silicon carbide requires high temperatures from 600 to 800° C. and also tends to form an isotropic cross section. In contrast, dry etching treatment such as reactive ion etching is able to simultaneously provide physical etching by cations present inside plasma and chemical etching by radicals, thereby an anisotropic etching high in the selectivity is provided and productivity is also improved. At the present time when semiconductor devices are available at a higher density and a greater integration, more elaborate high-precision pattern processing is demanded. Under these circumstances, silicon carbide is required to have a surface smoothness after etching treatment. It is noted that the above-described reactive ion etching is mainly used as an etching method in semiconductor industries. The following Non-patent Documents are referred to as documents covering research on the above-described dry etching of silicon carbide by plasma in which fluorine-based gas is used.

  • [Non-patent Document 1] P. H. Yih A. J. Steckel, J. Electrochem. Soc., 140, 1813 (1993)
  • [Non-patent Document 2] J. Sugiura, W. J. Lu, K. C. Cadien, and A. J. Steckel, J. Vac. Sci. Technol., B4, 349 (1986)
  • [Non-patent Document 3] J. W. Palmour, R. F. Davis, T. M. Wallet, and K. B. Bhasm, J. Vac. Sci. Technol., A4, 590 (1986)
  • [Non-patent Document 4] G. Kelner, S. C. Binari, and P. H. Klein, J. Electrochem. Soc., 137, 213 (1990)
  • [Non-patent Document 5] J. W. Palmour, R. F. Davis, P. Astell-Burt, and P. Blackborow, Mater. Res. Soc. Symp. Proc., 76, 185 (1987)
  • [Non-patent Document 6] W.-S. Pan and A. J. Steckel, J. Electrochem. Soc., 137 213 (1990)
  • [Non-patent Document 7] R. Padiyath, R. L. Wright, M. I. Caudhry, and S. V. Babu, Appl. Phys. Lett., 58, 1053 (1991)
  • [Non-patent Document 8] Wang J. J., Lambers E. S., Pearton S. J., Ostling M., Zetterling C. M., Grow J. M., Ren F., and Shul R. J., Vac. Sci. Technol., A16, 2204 (1198)
  • [Non-patent Document 9] G. Kelner, S. C. Binari, and P. H. Klen, J. Electrochem. Soc., 134, 253 (1987)


DISCLOSURE OF THE INVENTION
Problems to be Solved by the Invention

In general, where plasma of fluorine-containing gas such as nitrogen trifluoride gas is brought into contact with silicon carbide, it is known that, depending on how to allow the fluorine-containing gas to contact with plasma, irregularities called spikes are formed on the surface of single-crystal silicon carbide. Each of the above Non-patent Documents covers a description that etching can be performed on the surface of silicon carbide by changing a type of various fluorine-containing gases but does not disclose any means for providing a high-precision etching on the surface of the single-crystal silicon carbide concerned.


An object of the present invention is to provide a single-crystal silicon carbide excellent in smoothness and an etching method for single-crystal silicon carbide by which nitrogen trifluoride plasma are used to etch the surface of single-crystal silicon carbide at a high precision.


Means for Solving the Problem

In order to solve the above problem, the present inventors have evaluated in detail etching conditions which allow contact with the surface of single-crystal silicon carbide, and have found an etching method for single-crystal silicon carbide having a smooth surface and have completed the present invention.


According to the present invention, single-crystal silicon carbide is that having smoothness (surface roughness) within 150 nm on the basis of observation by an atomic force microscope.


Further, the present invention mainly deals with an etching method for single-crystal silicon carbide in which nitrogen trifluoride plasma is used to provide etching on the surface of single-crystal silicon carbide. Still further, the present invention mainly deals with an etching method for single-crystal silicon carbide in which a preferable pressure of nitrogen trifluoride gas is in the range of 0.5 to 10 Pa. When the pressure of nitrogen trifluoride gas is lower than 0.5 Pa, smoothness is not sufficiently provided on the surface of single-crystal silicon carbide. When the pressure of nitrogen trifluoride exceeds 10 Pa, spikes will be formed on the surface of single-crystal silicon carbide, which is not preferable.


In addition, the present invention mainly deals with an etching method for single-crystal silicon carbide in which a preferable flow rate of nitrogen trifluoride gas is in the range of 5 to 15 sccm (standard cc per minute). When the flow rate of nitrogen trifluoride is lower than 5 sccm, as described above, the single-crystal silicon carbide will be etched at an extremely slow rate. In contrast, when the flow rate of nitrogen trifluoride exceeds 15 sccm, spikes are formed on the surface of single-crystal silicon carbide, as described above, which is not preferable.


Further, the present invention mainly deals with an etching method for single-crystal silicon carbide in which RF power (reflected power:applied power) is preferably in the range of 50 W to 100 W. When the RF power is lower than 50 W, the etching rate is reduced, which is not preferable. When the RF power exceeds 100 W, the above etching is provided in a spike form, which is not preferable either. It is preferable that a reaction time is in the range of 4 to 15 minutes.


Still further, the present invention mainly deals with an etching method for single-crystal silicon carbide in which ion etching by rare gas radical and down-flow etching by nitrogen trifluoride gas is repeatedly performed to make the surface smooth.


In particular, it is preferable that an etched surface of single-crystal silicon carbide which is etched by plasma under the pressure of nitrogen trifluoride gas of 10 Pa or more and given a spike form is subjected to the treatment which repeats ion etching by rare gas radical and down-flow etching (hereinafter, referred to as DFE) by nitrogen trifluoride gas. In this instance, the DFE is defined as a process of performing etching, with the damage to samples resulting from plasma being reduced.


It is preferable that the pressure of argon gas in performing ion etching is in the range of 0.1 to 10 Pa, the RF power is from 50 to 100 W, and the reaction time is from 5 to 15 minutes.


Further, it is preferable that down-flow etching is performed under the pressure of nitrogen trifluoride gas in the range of 0.5 to 10 Pa and the RF power is in the range of 50 to 100 W. A preferable reaction time is in the range of 5 to 15 minutes.


Still further, it is preferable to repeat ion etching and down-flow etching more than 40 times.


So-called nitrogen trifluoride mixed gas in which oxygen gas is mixed with the above-described nitrogen trifluoride gas is also allowed to act on single-crystal silicon carbide to etch single-crystal silicon carbide, thereby it makes it possible to further reduce spikes formed on the surface of single-crystal silicon carbide. The oxygen gas is preferably supplied at concentrations from 5 to 20% and more preferably from 10 to 20%.


After the above-described nitrogen trifluoride mixed gas is allowed to act on single-crystal silicon carbide to etch single-crystal silicon carbide, ion etching by rare gas radical and down-flow etching by nitrogen trifluoride gas may be repeated to obtain a smooth surface. It is preferable that the conditions of repeating the ion etching and the down-flow etching are the same as those described above.


The present invention also mainly deals with an etching method for single-crystal silicon carbide in which single-crystal silicon carbide is selected from a 3C type, 4H type and 6H type. Single-crystal silicon carbide itself is available in various types other than the above-described three types. It is, however, possible to perform an extremely smooth etching by using the single-crystal silicon carbide selected from these three types so as to attain the surface roughness of 150 nm or less after the surface treatment by nitrogen trifluoride. Single-crystal silicon carbide of the 4H type is wider in bandgap, namely, 3.2 eV, and has less defects. This silicon carbide is highly expected to find an application as a power device to be used in an electric automobile, etc., because it is now available in a larger diameter and at a high efficiency, in addition to the above-described excellent properties. Single-crystal silicon carbide of the 6H type is slow in electron mobility but similar to silicon carbide having the 4H type structure. Further, single-crystal silicon carbide of the 3C type is narrow in band gap, namely 2.9V, and has high defects. However, since this silicon carbide can be manufactured at low temperatures and also made larger in diameter, it is highly expected to find applications such as a general inverter for which a cost benefit is demanded.


EFFECTS OF THE INVENTION

In the present invention, single-crystal silicon carbide is used to perform etching treatment by using nitrogen trifluoride mixed gas containing nitrogen trifluoride or oxygen gas under optimal conditions, thereby it makes it possible to obtain single-crystal silicon carbide which is quite smooth on the surface. Further, where the pressure of nitrogen trifluoride is 10 Pa or more under the above etching conditions, spikes are formed on the etched surface. In this instance as well, etching treatment by Ar+ ion and down-flow etching by nitrogen trifluoride are alternately repeated several times, thereby the single-crystal silicon carbide with smoothness is obtained. It is further secured to find applications such as an electricity device and a power device due to availability of the above-described smoothness.


EMBODIMENTS

Hereinafter, an explanation will be made for the present invention with reference to the embodiments. However, the present invention shall not be construed to be restricted to these embodiments.


(Preparation of Samples of Single-Crystal Silicon Carbide and Introduction Into Chamber)


4H-type single-crystal silicon carbide, 6H-type single-crystal silicon carbide and 3C-type single-crystal silicon carbide, each of which measures 7.0×5.0×1.0 (mm), are prepared and these samples are introduced into a plasma chamber.



FIG. 1 is a drawing briefly showing the plasma chamber used in experiments. Reference numeral 1 denotes a gas feeding port; 2 denotes a grounding electrode; 3 denotes a sample, 4 denotes an RF electrode; 5 denotes a CCD camera; 6 denotes a flow controller; 7 denotes a chamber body; 8 denotes a valve; 9 denotes a flow meter; and 10 to 13 respectively denote storage tanks of argon (Ar) gas, oxygen (O2) gas, nitrogen (N2) gas and nitrogen trifluoride (NF3) gas. The chamber body 7 is made with stainless steel (SUS 304) and provided with a cylindrical window portion 7a projecting from the side surface of the chamber body 7. The window portion 7a is formed by fitting a transparent member such as glass into the outermost periphery so that the internal part can be observed by a CCD camera 5. Further, a tubal portion 7b into which an electric wiring for connecting the RF electrode 4 to a power source is inserted is provided on a side opposite the window portion 7a of the chamber body 7. In FIG. 1, the grounding electrode is given on the upper side and the RF electrode is given on the lower side. However, in reality, these two electrodes can be freely interchanged.


When individual samples are placed on the RF electrode on a lower side to induce plasma, a space charge region which is called an ion sheath is formed around the RF electrode due to a difference in mobility between electrons and ions, thereby a region where electrolysis undergoes a substantially similar change is provided. When cations generated inside a plasma region are charged into an ion sheath region, ions are accelerated by a vertical electrical field generated in the region and allowed to enter in the vertical direction to the samples, thereby a physical collision is caused.


On the other hand, neutral molecules such as radicals in the plasma are allowed to enter, without having orientation, thereby a chemical reaction is caused. More specifically, it is possible to perform reactive ion etching having both characteristics of physical etching and chemical etching. The chamber is structured so as to effect discharge under vacuum from the lower part of the chamber by using a rotary pump and an oil diffusion pump, with the ultimate vacuum being 1.3×103 Pa. A liquid nitrogen trap is provided between the etching chamber and the oil diffusion pump, thereby it makes it possible to reduce a quantity of reactive gas dissociated during the etching which will flow into the oil diffusion pump or the rotary pump. It is, therefore, possible to effect discharge under vacuum until the pressure reaches a level of 1.33×10−3 Pa and further discharge for 60 minutes continuously.


(Method for Etching Treatment)


After the discharge under vacuum, a needle valve (KOFLOC model RK-1200, made by Kojima Seisakusho Co., Ltd.) is used to adjust the pressure of nitrogen trifluoride (made by Mitsui Chemicals Inc.) to a predetermined level, which is then allowed to stand for 5 minutes. After the flow rate becomes stable, RF plasma having 13.56 MHz is induced.


Further, a matching unit MU-2 (made by SAMCO International Inc.) is also used to make adjustment so that the RF power can be minimized.


During the plasma induction, cooling water kept at a constant temperature is circulated through a sample base, by which the samples are kept at 50° C. After completion of the etching, discharge under vacuum is effected again for approximately 10 minutes to remove activated species remaining inside the chamber, thereby reactions with the samples are prevented. Thereafter, in order to prevent the effect of adsorption of H2O and O2 present in the air to the greatest possible extent, N2 is introduced to a level of 1 atm and the samples are then removed from the chamber. Also in order to prevent the effect of the air, the thus removed samples are immediately placed into a vacuum desiccator containing drying agents and oxygen absorbing agents (ultimate vacuum of 1.33×102 Pa) and retained therein.


Further, the change in the etching rate resulting from the change in the types of samples (single-crystal silicon carbide substrate), the change in the flow rate of nitrogen trifluoride, pressure and applied power as well as the smoothness (surface roughness) of etched samples are measured by an atomic force microscope to provide embodiments 1 to 25 and comparative examples 1 to 7, the results of which are summarized in Table 1. In this instance, where the surface roughness of etched single-crystal silicon carbide is within ±150 nm, the silicon carbide concerned is designated as ∘, and where the surface roughness is particularly within ±50 nm, it is designated as ⊚. In contrast, where the surface roughness is larger than ±150 nm, it is designated as ×. Further, oxygen gas is mixed with nitrogen trifluoride gas at a predetermined pressure to perform etching in such a way that the concentration of oxygen gas in the mixed gas can be 10% and 20% respectively in the embodiments 10 and 11, the results of which are shown in Table 1.

TABLE 1NitrogenCrystaltrifluorideEtched surfacestructure ofFlowRFEtchingReactionsiliconPressureratepowerSmoothnessratetimecarbide(Pa)(sccm)(W)note 1)(A/min.)(min)RemarksEmbodiment 14H0.510100?70010Embodiment 24H110100?65010Embodiment 34H21010055010Embodiment 46H21010063010Embodiment 53C21010062010Embodiment 64H1010100?50010Embodiment 74H215100?60010Embodiment 84H2107550010Embodiment 94H2510055010Embodiment 104H0.510100?80010OxygenconcentraEmbodiment 114H0.510100?75010OxygenconcentraEmbodiment 126H21010075010OxygenconcentraEmbodiment 136H21010070010OxygenconcentraEmbodiment 143C31010070010OxygenconcentraEmbodiment 153C31010073010OxygenconcentraEmbodiment 166H0.510100?65010Embodiment 176H1015100?60010Embodiment 186H25100?56010Embodiment 196H0.51060?55010Embodiment 206H0.51080?60010Embodiment 213C0.510100?65010Embodiment 223C1010100?60010Embodiment 233C25100?60010Embodiment 243C2108065010Embodiment 253C215100?60010Comparative4H21040X30010example 1Comparative4H210130X60010example 2Comparative4H1010120X50010example 3Comparative4H2010100X60010example 4Comparative4H220100X60010example 5Comparative4H23100?45010example 6Comparative4H0.110100?40010example 7
note 1): smoothness: ⊚ within ±50 nm, ◯ larger than 50 nm but 150 nm or less, X larger than ±150 nm


Further, in order to conduct comparative experiments, 4H-type single-crystal silicon carbide is used to perform etching treatment by allowing the conditions to change within a scope under the etching conditions of the present invention. Still further, evaluation is made for further smoothness of the surface of single-crystal silicon carbide given in a spike form (measured by an atomic force microscope). More specifically, prior to introduction of nitrogen trifluoride, sputtering treatment is performed in advance on the surface of single-crystal silicon carbide only by using Ar+ ion (one type of rare gases) at 3 Pa, and DFE treatment is then performed by using nitrogen trifluoride. A two-stage treatment made up of sputtering by Ar+ ions and a subsequent treatment by fluorine radicals (fluorine radicals generated by DFE) is repeated. The number of such repetitions and the smoothness of single-crystal silicon carbide samples are evaluated in the embodiments 26 to 28, the comparative example 8 and the comparative example 9, the results of which are summarized in Table 2.

TABLE 2Crystalstructureof siliconcarbideobtainedDown-flowthroughetching (DFE)reaction(a)withNitrogenEtching by Ar+nitrogentrifluorideions (b)RepetitiontrifluoridePressureRF powerPressureRF powernumber ofSmoothnessalone(Pa)(W)(Pa)(W)(a) and (b)note 2)Embodiment4H5310031504026Embodiment6H5310031504027Embodiment3C5310031504028Comparative4H53100315035Xexample 8Comparative4H53100315025Xexample 9
note 2): smoothness: ⊚ within ±50 nm, ◯ larger than 50 nm but 150 nm or less, X larger than ±150 nm


Further, after ion etching is performed for single-crystal silicon carbide similarly as above, oxygen-containing nitrogen trifluoride mixed gas prepared by mixing oxygen gas with nitrogen trifluoride gas is used to perform DFE of single-crystal silicon carbide. This two-stage treatment of single-crystal is performed several times under different conditions in the embodiments 29 to 31, the comparative example 10, and the comparative example 11, the results of which are shown in Table 3.

TABLE 3Crystalstructureof siliconcarbideobtainedthroughreactionwithnitrogenDown-flowtrifluorideetching (DFE)gas which(a)Etching by Ar+containsNitrogenions (b)oxygen intrifluorideRFRFRepetitionthe range ofPressurepowerPressurepowernumber ofSmoothness10% to 20%(Pa)(W)(Pa)(W)(a) and (b)note 3)Embodiment4H5310031504029Embodiment6H5310031504030Embodiment3C5310031504031Comparative4H53100315035Xexample 10Comparative4H53100315025Xexample 11
note 3): smoothness: ⊚ within ±50 nm, ◯ larger than 50 nm but 150 nm or less, X larger than ±150 nm


Next, an explanation will be made for the identification of chemical species in nitrogen trifluoride plasma, measurement of concentrations of the chemical species and measurement of etching rate. First, an explanation will be made for a method used for these measurements.


(Method for Identifying Chemical Species in Nitrogen Trifluoride Plasma and Measuring the Concentration of the Chemical Species)


The following emission spectroscopic analysis is conducted to identify chemical species in nitrogen trifluoride plasma and measure the concentration of the chemical species. Light emitted inside a chamber is collected from an observation window made with Pyrex (registered trade mark) by using a CCD camera and received by bundle fibers. The light is made incident into a diffraction grating of the computer-controlled spectrometer, intensified and detected by using a CCD element, which is converted to digital signals. Then, the signals are captured in the computer for monitoring. The spectrometer and related instruments used in this experiment include SPG-120PM, AT-120PM, AT-120PL, AT-100AP, AT-100PCC (made by Shimadzu Corporation). In this experiment, attention is focused on the spectra assigned to N+ ion at 353.2 nm and spectra assigned to fluorine radical at 703.7 nm.


(Method for Measuring Etching Rate)


A series of photolithography technologies are used to pattern an AI mask (99.99% made by Nilaco Corporation) on the surface of single-crystal silicon carbide. In order to measure the etching rate, the samples are submerged into a phosphoric acid solution (H3PO4:HNO3:CH3COOH:H2O=8:3:1:1) approximately for 5 minutes after etching to dissolve and remove the AI mask from the surface of the samples. A step between the etched surface by the above treatment and the surface not etched by the AI masking treatment is measured by using a laser displacement meter (made by Keyence Corporation).


Since a 100 nm or greater step which allows the measurement by a laser displacement meter is required in establishing an etching time, the etching time is established to be 10 minutes or longer. The etching rate referred to in the text or drawings is a value obtained by dividing the step measured after etching treatment with time etched. Further, the measurement is made at 1000 sites and expressed by their average value.


Next, an explanation will be made for the results and discussion on the above measurements.


(Results and Discussion on the Identification of Chemical Species and Measurement of Concentration of the Chemical Species)


Results of emission spectroscopic analysis of nitrogen trifluoride plasma conducted under the conditions that the pressure of nitrogen trifluoride is 10 Pa, the applied power is 100 W and the flow rate of nitrogen trifluoride is 10 sccm are shown in FIG. 2, where (a) indicates a result obtained by conducting an emission spectroscopic analysis in a state that samples are not placed in a chamber, and (b) indicates a result obtained by conducting the analysis in a state that the samples are placed. As apparent from FIG. 2, no peak is newly observed even if the samples are placed.


However, the radial strength of the fluorine radical (F ) is decreased when the samples are placed. Peaks assigned to a nitrogen system such as N+ ion, N2+ ion and N2 are observed in the vicinity from 350 nm to 400 nm, and peaks assigned to fluorine radicals are observed in the vicinity from 600 nm to 750 nm. Reactions of generating fluorine radicals and nitrogen ions in NF3 plasma are shown as follows.


Further, N2+ ions and fluorine radicals are consumed in a great quantity on etching of silicon carbide, and their chemical species are mainly involved in etching of silicon carbide.

NF3→NF2.+F.   (1)
NF2.→NF.+F.   (2)
NF.→N.+F.   (3)
NF.→N++F.+e   (4)
N.→N++e   (5)
2N.→N2   (6)
N2→N2++e   (7)


Then, in measuring concentrations in this experiment, evaluation is made for the dependence of concentrations of chemical species on the etching rate in reactive ion etching of silicon carbide and on the spike formation by measuring the concentrations of N2+ ions and fluorine radicals, with attention focused on the peaks assigned to N+ ions at 353.2 nm and peaks assigned to fluorine radicals at 703.7 nm.


An emission spectroscopic analysis is to be initiated one minute and 30 seconds later after the induction of plasma. Further, since the peak strength is dependent on the number of particles, it is expressed as the concentration in this experiment. First, samples are placed inside a chamber, and evaluations are made for a change in peak strength of N2+ ions and fluorine radicals when the pressure of nitrogen trifluoride is allowed to change under the conditions that the applied power is 100 W and the flow rate of nitrogen trifluoride is 10 sccm, the results of which are shown in FIG. 3.


Fluorine radicals are greatly increased in peak strength with an increase in pressure of nitrogen trifluoride from 2 Pa to 5 Pa and gradually increased in peak strength with an increase in pressure of nitrogen trifluoride over the range of 5 Pa to 20 Pa. Further, N2+ ions are greatly increased in peak strength in pressure up to 10 Pa but gradually increased in peak strength with a further increase in pressure of nitrogen trifluoride. On the basis of these findings, it is clear that chemical species in plasma are also increased in concentration with an increase in pressure of nitrogen trifluoride.


(Results and Discussion on Measurement of Etching Rate)


The peak strength assigned to fluorine radicals at 703.7 nm is evaluated for a change with the lapse of time with reference to nitrogen trifluoride plasma generated in a state that no silicon carbide is placed on a sample base under the conditions that the applied power is 100 W, the pressure of nitrogen trifluoride is 10 Pa and the flow rate of nitrogen trifluoride is 10 sccm, the result of which is shown in FIG. 4. Fluorine radicals are increased in concentration approximately for 4 minutes after the induction of plasma but kept constant thereafter. Thus, the etching time is established to be 10 minutes, during which the dissociation of nitrogen trifluoride sufficiently proceeds and the depth of etching can be fully measured.


The pressure of nitrogen trifluoride is allowed to change from 2 Pa to 30 Pa to perform etching and the etching rate of silicon carbide is measured under the constant conditions that the applied power is 100 W and the flow rate of nitrogen trifluoride is 10 sccm, the result of which is shown in FIG. 5. As apparent from FIG. 5, the etching rate is increased monotonously with a decrease in pressure of nitrogen trifluoride when the pressure is lower than 2 Pa and with an increase in pressure of nitrogen trifluoride when the pressure is higher than 3 Pa.


Next, the etching rate is measured when the applied power is allowed to change from 50 W to 130 W to perform etching under the constant conditions that the flow rate of nitrogen trifluoride is 10 sccm and the pressure of nitrogen trifluoride is 10 Pa, the result of which is shown in FIG. 6. In this instance, the pressure of nitrogen trifluoride is established to be 10 Pa because observations are also made for a difference in spike formation depending on the applied power under the pressure of nitrogen trifluoride which allows spikes (needle-like projections) to form on the surface of silicon carbide. As apparent from FIG. 6, the etching rate is also increased with an increase in applied power.


Next, the etching rate is measured when the flow rate of nitrogen trifluoride is allowed to change from 2 sccm to 20 sccm to perform etching under the constant conditions that the pressure of nitrogen trifluoride is 10 Pa and the flow rate of nitrogen trifluoride is 10 sccm, the result of which is shown in FIG. 7. The etching rate is also increased with an increase in the flow rate of nitrogen trifluoride until the flow rate is 5 sccm but substantially kept constant in the range of 5 sccm to 15 sccm. Then, the etching rate is increased with an increase in flow rate of nitrogen trifluoride to 20 sccm. The above-described tendency of the etching rate is well in conformity with an increasing tendency of fluorine radicals.


The relationship between the concentration of fluorine radicals and the etching rate is evaluated with respect to the change in pressure, change in applied power and change in flow rate, the results of which are summarized in FIG. 8. It is apparent from FIG. 8 that a high etching rate (536 Å/min) is obtained with respect to a change in pressure under the pressure of nitrogen trifluoride of 2 Pa, although the concentration of fluorine radicals is low. Except in a case where the pressure of nitrogen trifluoride is 2 Pa, the etching rate is increased with an increase in concentration of fluorine radicals with respect to any change in pressure, applied power or flow rate of nitrogen trifluoride.



FIG. 9 shows the relationship between the concentration of N2+ ions and the etching rate which is not necessarily limited to the pressure of nitrogen trifluoride of 2 Pa in a case of N2+ ions. The etching rate is also increased with an increase in the concentration of N2+ ions in the case of N2+ ions, which is not necessarily limited to the pressure of nitrogen trifluoride of 2 Pa. Incidentally, not only in the case where the pressure of nitrogen trifluoride is 2 Pa but also in the case where the applied power is 50 W and the flow rate is 2 sccm, the etching rate is great only in the case where the pressure of nitrogen trifluoride is 2 Pa, although the concentrations of fluorine radicals and N2+ ions are low. Where the concentration of chemical species in plasma is low, a mean free pass of the chemical species is made long, the kinetic energy on collision of cations against silicon carbide is increased, which may result in an easy advance of physical energy. Therefore, a similar phenomenon should occur also in the case where the applied power is 50 W and the flow rate is 2 sccm. Since cations are accelerated depending on the strength of the electric field in an ion sheath, the applied power of 100 W is required in obtaining a sufficient kinetic energy.


Where the pressure of nitrogen trifluoride is 2 Pa, the mean free path of cations such as N2+ ions is increased due to a low concentration of chemical species in plasma. More specifically, a negative bias potential of the RF electrode may prolong a distance of cations to be accelerated and result in an increase in kinetic energy. Therefore, carbons hardly etched under the pressure of nitrogen trifluoride of 2 Pa can be effectively removed by physical etching with cations due to a difference in binding energy with fluorine (binding energy: Si—F: 130 kcal/mol, C—F: 107 kcal/mol), thereby a smooth surface and a high etching rate are obtained.


On the other hand, where the pressure of nitrogen trifluoride is 10 Pa, fluorine radicals are changed according to a change in applied power and flow rate, and the etching rate is increased with an increase in fluorine radicals. Since the carbon-rich silicon carbide surface is generated to form an innumerable number of spikes, an increased concentration of the chemical species in plasma reduces the mean free path of cations and fails in sufficiently obtaining kinetic energy. Then, carbon is ablated by cations at a reduced speed and, as a result, silicon may be preferentially etched by fluorine radicals.


A scanning electron microscope (SEM) is used to photograph the surfaces of single-crystal silicon carbide samples etched for 10 minutes under the conditions that the applied power is 100 W, the flow rate of nitrogen trifluoride is 10 (sccm) and the respective pressures of NF3 gas of 2, 10 and 20 Pa. The photos of the samples (SEM images) are respectively shown in FIG. 10. It is apparent from FIG. 10 that those etched under the conditions that the applied power is 100 W, the flow rate of nitrogen trifluoride is 10 (sccm) and the pressure of nitrogen trifluoride is 2 Pa are free of spikes on the surface of single-crystal silicon carbide samples. However, it is also apparent that spikes are formed on the etched surface where the pressure of nitrogen trifluoride is increased to 10 Pa and 20 Pa.


It is, therefore, apparent that the etching is performed by subjecting nitrogen trifluoride gas to plasma excitation, thereby single-crystal silicon carbide having an extremely smooth surface is obtained. Further, where the pressure of nitrogen trifluoride is 10 Pa or more under the above-described etching conditions, spikes are formed on the etched surface. In this instance, however, it is also apparent that ion etching treatment by Ar+ ion and down-flow etching treatment by nitrogen trifluoride are alternately repeated several times, thereby single-crystal silicon carbide having an extremely smooth surface is obtained.


Next, the etching rate of oxygen-containing nitrogen trifluoride mixed gas in plasma is measured, chemical species are identified and the concentration of the chemical species is measured. An explanation for their methods will be omitted here, because they are similar to those described above. In identifying the chemical species and measuring the concentration of the chemical species, N2+ ion, fluorine radical and oxygen radial are evaluated, with attention focused on the respective spectra of 353.2 nm, 703.7 nm and 777 nm.


(Results and Discussion on Measurement of Etching Rate)


Samples are etched for 10 minutes and 60 minutes under the constant conditions that a total pressure is 10 Pa, the applied power is 100 W and a total flow rate is 10 sccm, to evaluate the change in the etching rate in association with the change in the oxygen concentration, the results of which are shown in FIG. 11. In this instance, the total pressure is established to be 10 Pa, because chemical etching is to mainly participate in reactions. As apparent from FIG. 11, as compared with samples etched for 10 minutes, those etched for 60 minutes exhibit a decrease in the etching rate of SiC at all oxygen concentrations. However, the etching rate is increased with an increase in oxygen concentration both for the samples etched for 10 minutes and those etched for 60 minutes, a maximum value is exhibited at the oxygen concentration of 10%. Further, the etching rate is monotonously decreased when the oxygen concentration is further increased.


(Results and Discussion on Identification of Chemical Species and Measurement of Concentration of the Chemical Species)


An emission spectroscopic analysis is conducted under the above-described conditions to evaluate the concentration of chemical species NF3/O2 plasma, the results of which are shown in FIG. 12. First, the concentration of fluorine radicals is increased with an increase in oxygen concentration, a maximum value is exhibited when oxygen is added at the concentration of 10%. The concentration of fluorine radicals is monotonously decreased with a further increase in oxygen concentration. In contrast, in measuring the concentration of N2+ ions, the peak strength exceeds an upper detection limit of the equipment where the oxygen concentration is in the range of 0% to 10%, and any concentration higher than the limit cannot be measured. As shown in FIG. 13, where the concentration of N2+ ions is measured under the pressure of NF3 of 5 Pa, the concentration is increased to a maximum level at the oxygen concentration of 2% and monotonously decreased with a further increase in oxygen concentration. Further, the concentration of fluorine radicals is increased to a maximum level at the oxygen concentration of 10% similar to the where the pressure of NF3 is 10 Pa. On the basis of the above findings, the concentration of N2+ ions measured under the pressure of NF3 of 10 Pa is considered to be similar in tendency to that measured under the pressure of NF3 of 5 Pa. Finally, the concentration of oxygen radicals is gradually increased at the oxygen concentration of up to 60% and gradually decreased with a further increase in oxygen concentration. As a result, the etching rate of SiC is considered to increase with an increase in fluorine radicals resulting from the addition of oxygen. However, where the oxygen concentration is in the range of 20% to 30%, the etching rate is greatly increased more than a case of NF3 plasma alone, although the concentration of fluorine radicals is lower.


Since fluorine radicals are high in surface charge density and electrostatic repulsion is also great between fluorine atoms, attack by fluorine radicals is less likely to occur at a part where C—F couplings are formed to some extent on the surface of SiC. On the other hand, in view of the fact that oxygen radicals are also high in surface charge density but an oxygen/carbon binding energy is greater than a fluorine/carbon binding energy, carbons at a site abundant in C—F couplings are also exposed to attack by oxygen radicals to cause reactions between oxygen radicals and carbons, and easily removed as COx. As a result, it is estimated that the etching rate is increased and formation of spikes is inhibited in NF3/O2 plasma. More specifically, it is understood that due to addition of oxygen, not only are fluorine radicals increased but also oxygen radicals are allowed to react with a part of carbon in SiC and removed as COx, the etching rate of SiC is greatly influenced. These findings clearly exhibit that oxygen-containing nitrogen trifluoride mixed gas is subjected to plasma excitation to perform etching, thereby it makes it possible to obtain single-crystal silicon carbide having an extremely smooth surface. It is also apparent that even if an etched surface is in a spike form, etching treatment by Ar+ ions and down-flow etching treatment by nitrogen trifluoride are repeatedly provided to the surface alternately several times, thereby it makes it possible to obtain single-crystal silicon carbide having an extremely smooth surface.




BRIEF DESCRIPTION OF THE DRAWINGS


FIG. 1 is a drawing briefly showing a plasma chamber used in the present invention;



FIG. 2 shows the results of emission spectroscopic analysis for nitrogen trifluoride plasma conducted under the conditions that the applied power is 100 W and the flow rate of nitrogen trifluoride is 10 sccm, where (a) is a graph where no sample is placed and (b) is a graph where samples are placed;



FIG. 3 is a graph showing the change in peak strength of N2+ ions and fluorine radicals when the pressure of nitrogen trifluoride is allowed to change under the conditions that the applied power is 100 W, the flow rate of nitrogen trifluoride is 10 sccm and the reaction time is 10 minutes;



FIG. 4 is a graph showing the result obtained by evaluating the change with the lapse of time in peak strength assigned to fluorine radicals at 703.7 nm in nitrogen trifluoride plasma generated in a state that no silicon carbide is placed on a sample base under the conditions that the applied power is 100 W, the pressure of nitrogen trifluoride is 10 Pa and the flow rate of nitrogen trifluoride is 10 sccm;



FIG. 5 is a graph showing the result obtained by measuring the etching rate of silicon carbide when the pressure of nitrogen trifluoride is changed from 2 Pa to 30 Pa to perform etching under constant conditions that the applied power is 100 W, the flow rate of nitrogen trifluoride is 10 sccm and the reaction time is 10 minutes;



FIG. 6 is a graph showing the etching rate obtained when the applied power is changed from 50 W to 130 W to perform etching under constant conditions that the flow rate of nitrogen trifluoride is 10 sccm, the pressure of nitrogen trifluoride is 10 Pa and the reaction time is 10 minutes;



FIG. 7 is a graph showing the result obtained by measuring the etching rate of silicon carbide when the pressure of nitrogen trifluoride is changed from 2 Pa to 30 Pa to perform etching under constant conditions that the applied power is 100 W, the flow rate of nitrogen trifluoride is 10 sccm and the reaction time is 10 minutes;



FIG. 8 is a graph showing the relationship between the concentration of fluorine radicals and the etching rate with reference to a change in pressure, change in applied power and change in flow rate;



FIG. 9 is a graph showing the relationship between the concentration of N2+ ion and the etching rate;



FIG. 10 shows SEM images of sample surfaces photographed after etching is performed for 10 minutes under the pressure of nitrogen trifluoride of 2 Pa, 10 Pa or 20 Pa;



FIG. 11 is a graph showing the change in the etching rate according to the change in oxygen concentration observed when etching is performed for 10 minutes and 60 minutes under constant conditions that a total pressure is 10 Pa, the applied power is 100 W and a total flow rate is 10 sccm;



FIG. 12 is a graph showing the results obtained by evaluating the concentration of chemical species in NF3/O2 plasma on the basis of emission spectroscopic analysis under constant conditions that a total pressure is 10 Pa, the applied power is 100 W and a total flow rate is 10 sccm; and



FIG. 13 is a graph showing the results obtained by evaluating the concentration of chemical species in NF3/O2 plasma on the basis of emission spectroscopic analysis under constant conditions that a total pressure is 5 Pa, the applied power is 100 W and a total flow rate is 10 sccm.




DESCRIPTION OF SYMBOLS




  • 1: Gas feeding port


  • 2: Grounding electrode


  • 3: Sample


  • 4: RF electrode


  • 5: CCD camera


  • 6: Flow controller


  • 7: Chamber body


  • 8: Valve


  • 9: Flow meter


  • 10: Storage tank of argon gas


  • 10: Storage tank of oxygen gas


  • 12: Storage tank of nitrogen gas


  • 13: Storage tank of nitrogen trifluoride gas


Claims
  • 1. Single-crystal silicon carbide in which smoothness observed by an atomic force microscope is within ±150 nm.
  • 2. An etching method for single-crystal silicon carbide, wherein nitrogen trifluoride gas is subjected to plasma excitation to treat the surface of single-crystal silicon carbide.
  • 3. The etching method for single-crystal silicon carbide according to claim 2, wherein the pressure of nitrogen trifluoride gas is in the range of 0.5 to 10 Pa.
  • 4. The etching method for single-crystal silicon carbide according to claim 2, wherein the flow rate of nitrogen trifluoride gas is in the range of 5 to 15 sccm.
  • 5. The etching method for single-crystal silicon carbide according to claim 2, wherein the applied power is in the range of 50 to 100 W when nitrogen trifluoride gas is subjected to plasma excitation.
  • 6. The etching method for single-crystal silicon carbide according to claim 2, wherein ion etching by rare gas radicals and down-flow etching by nitrogen trifluoride gas are repeated as additional treatment to obtain a smooth surface.
  • 7. An etching method for single-crystal silicon carbide, wherein oxygen gas-containing nitrogen trifluoride gas is subjected to plasma excitation to treat the surface of single-crystal silicon carbide.
  • 8. An etching method for single-crystal e silicon carbide comprising a step in which oxygen gas-containing nitrogen trifluoride gas is subjected to plasma excitation to treat the surface of single-crystal silicon carbide, and a step in which ion etching by rare gas radicals and down-flow etching by nitrogen trifluoride gas are repeated as additional treatment to obtain a smooth surface.
  • 9. The etching method for single-crystal silicon carbide according to claim 2, claim 7 or claim 8, wherein the above-described single-crystal silicon carbide is selected from 3C type, 4H type or 6H type.
Priority Claims (1)
Number Date Country Kind
2004-201617 Jul 2004 JP national
PCT Information
Filing Document Filing Date Country Kind 371c Date
PCT/JP05/12473 7/6/2005 WO 1/8/2007