JPH0332209B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a plasma vapor phase method for forming non-single crystal silicon carbide on a formation surface using a reactive gas made of a hydride or halide having a carbon-silicon bond.

By diluting such a reactive gas with helium instead of hydrogen, nitrogen, or argon, the present invention achieves an optical energy band that causes less damage to the surface to be formed, has excellent film thickness uniformity, and has a faster film growth rate. The purpose is to form a silicon carbide film having a width (hereinafter referred to as Eg) of 2.3 eV or more at a low temperature of 100 to 500°C.

The present invention further provides an impurity gas containing valent impurities such as B, Al, Ca, and In, such as diborane (B ₂ H ₆ ), and an impurity gas containing V-valent impurities, such as phosphin (PH ₃ ) or arsine. (AsH ₃ ) is gradually added to form a P-type layer, then an I-type layer and an N-layer closely on the substrate having the surface to be formed.
The purpose is to form layers in the order of PIN.

Traditionally, non-single crystal semiconductors were called amorphous (hereinafter simply referred to as
It is known as a typical example of silicon (AS) produced using the plasma vapor phase method. This is expected to be applied to photoelectric conversion devices such as solar cells. However, when attempting to make such a device or to obtain a visible light emitting element using a semi-single crystal semiconductor, there was a need to develop a window material having a wide Eg of 2.3 to 3.5 eV.

This material is silicon carbide (SixC _1-x (0<x<
1)) is a typical example. However, there have been attempts to make silicon carbide by decomposing and reacting a carbide gas such as methane (CH ₄ ) with a silicide gas silane (SiH ₄ ) in a plasma atmosphere. However, silicon carbide obtained by such a method has a macroscopic value of SixC _1-x (0<x<1).
However, there are many silicon clusters and many carbon clusters, making it impossible to create homogeneous silicon carbide. For this reason, the optical Eg
It is impossible to achieve a voltage higher than 2.0 eV, and generally only 1.6 to 1.8 eV, which is exactly the same as silicon, can be obtained.
Furthermore, if you try to obtain more than 2.0eV instead of the narrow Eg, the discharge power will be 200 to 500W.
As a result of these reactions, the reaction spuds caused spatter (damage) on the formed surface, making it completely impossible to electrically form a PIN junction and obtain desirable diode characteristics.

Therefore, in the present invention, in order to eliminate this drawback, a material having a carbon-silicon bond is used as the starting material, ie, a reactive gas. That is, hydrides or halides having a carbon-silicon bond, such as tetramethylsilane (Si(CH ₃ ) ₄ ) (simply referred to as TMS), tetraethylsilane (Si(C ₂ H ₅ ) ₄ )Si
(The first feature is that reactive gases such as (CH ₃ ) ₃ Ci, Si(CH ₃ ) ₂ C1 ₂ , Si(CH ₃ ), and C1 ₃ are used.

Furthermore, in the present invention, such a reactive gas is introduced into an atmosphere in which electromagnetic energy is applied and a plasma state is generated, and C-H bonds, Si-C1 bonds,
As the Si-C bond is broken, hydrogen recombines to the dangling bonds of C and Si, creating a C-H bond and Si-
To prevent the formation of H bonds, helium is used as the carrier gas instead of hydrogen. In that case, other things being the same, TMS/H
When e = 1/1 to 30 and TMS/H ₂ = 1/1 to 30, the growth rate of the film can be increased by 3 to 9 times, and the uniformity of the formed film is better than that of hydrogen. In the case of , the variation in film thickness was ±6%, but it was possible to reduce it to ±3%, resulting in an extremely uniform film. Unlike active gases such as Ar, He has a small molecular ratio and has the highest ionization energy of 2.5 eV, so even if it is turned into plasma, it does not sputter on the surface on which it is formed, and it can be used in films with PIN junctions. The effect was also large.

Furthermore, when such a reactive gas is used, the reactor can be heated to 1 atm or less, particularly 0.01 to 10 torr, typically 0.3 to 10 torr.
Even with electromagnetic energy of 50W or less under a pressure of 0.6torr, for example, 0.1 to 100MHz, especially 13.56MHz,
Alternatively, it is possible to form a coating at 1 to 4 GHz, especially 2.45 GHz. i.e. low energy plasma
It could be used as a CVD device.

Furthermore, if a high-energy plasma atmosphere of 50 to 500 W is applied, the formed silicon carbide becomes microcrystalline, and as a result, in the P type or N type, boron or phosphorus is added by 0.5 to 10% (here, (B ₂ H ₆ or
When the ratio of PH ₃ )/(carbide silicide gas + silicide gas) is added, the electrical conductivity at low energy is 10 ^-9 to 10 ^-6 (Ωcm) ^-1 to 10 ^-6 to 10 ^-3 (Ωcm) ^-1 , which was approximately 1000 times higher.

And its optical Eg is 1.6 ~ like silicon
2.3-3.5eV instead of 1.8eV typically 2.5-3.2eV
It was possible to have In addition, when 0.01 to 5 mol% of diborane or phosphine is added to this, the silicon carbide (SixC _1-x 0<x<0.5) has an AS structure and an activation energy of 0.3 in the low energy method.
~0.6eV. In addition, in the high energy method, 0.01~
It was possible to make it a P- or N-type semiconductor with a voltage of 0.1 eV. Furthermore, silicon carbide obtained using this high-energy method has a microcrystalline structure with a size of 5 to 200 A.
We were able to create a structure called SAS.
In such SAS, the ionization rate of the acceptor or donor of the P or N type impurity is determined.
It had 97-100% and was able to activate all of the added impurities.

The plasma vapor phase method of the present invention will be explained below with reference to the drawings.

FIG. 1 shows an outline of a plasma CVD apparatus using the present invention.

In FIG. 1, a substrate 1 having a surface to be formed is held by a quartz jig, and as shown in the drawing, there are seven stages and two rows, 14 in total. The substrates are spaced apart by 10 to 40 mm, typically 20 to 25 mm, and the jig is arranged so that the reactive gas is evenly injected into the gaps between the substrates. The surface to be formed is the lower surface of the substrate, and the upper surface is covered so that it does not become the surface to be formed. This is due to the decomposition and reaction of reactive gases, which cause the reaction products to adhere uniformly and form a film, and during the formation of this film, flakes, etc., released from the reaction tube wall fly off and are deposited on the top surface in large numbers due to gravity. This is because the flakes fall and cause pinholes to occur. For this reason, it is extremely important to set the surface to be formed on the bottom surface in consideration of mass production yield. Further, in the reactor 25 into which the substrate 1 is inserted, electrodes 9 and 10 are provided above and below so that an electric field of electromagnetic energy is applied perpendicularly to the substrate. An electric furnace 5 is provided outside this electrode, and the substrate 1 is heated to 100 to 500°C, typically 300°C.

The reactive gas is carrier gas helium.
From 14, diborane, which is a valent impurity,
Phosphine, which is a valence impurity, was introduced from point 15, and silane, which is a silicide gas, which is a valence additive, was introduced from point 16.

Also, reactive gas TMS with carbon-silicon bonds
20, it is stored in a stainless steel container 21 because it is liquid in the initial state. This container is controlled at a predetermined temperature by an electronic constant temperature layer 22.

This TMS has a boiling point of 25° C., and the rotary pump 12 was evacuated through the valve 11 to maintain the inside of the reactor at 0.01 to 10 torr. In this way, TMS can be vaporized at a pressure lower than 1 atmosphere without any particular heating.
Introducing this vaporized TMS at 100% concentration into the reactor via a flow meter is carried out in the container 21 as in the conventional method.
Compared to the method of emitting reactive gases by bubbling them, the flow rate can be controlled with high precision, which is technically important.

In practice, if the flow meter is clogged, 1 in the drawing
Helium was introduced from 7.

These reactive gases were mixed with helium as a carrier gas at a predetermined ratio and introduced into the reactor 25. Electromagnetic energy is applied between the electrodes 9 and 10 by, for example, high frequency (13.56 MHz), thereby applying an electric field in a direction that compacts the film accumulated on the surface to be formed. By doing this, the presence of voids etc. in the formed silicon carbide or silicon was reduced by utilizing the flight of the reactive gas moved by the electric field. Further, in this plasma discharge, a spatial reaction is mainly used in which reactive gases are mixed through the mixing chamber 8 and then decomposed or reacted in the excitation chamber 26 to form reaction products on the substrate. As electromagnetic energy, DC high frequency was mainly used from the power source 4. Of course, microwaves (1 to 4 GHz) may also be used. In this way, a silicon carbide film was formed on the surface to be formed. For example, the substrate temperature is 300℃, the high frequency energy output is 25W,
TMS 5 c.c./min, He 100 as carrier gas
cc/min. At TMS/He=20, a film growth rate of 160 A/min could be obtained. This is the first
If carrier gas 13 in the figure is changed to hydrogen, TMS/
Even if H ₂ = 20 and all other conditions are the same, 25A/
The growth rate is only about 1/6th of that of the previous film. When the carrier gas is helium, silicon carbide (C/Si4/
A growth rate 5 to 8 times faster than hydrogen dilution can be obtained not only when producing carbon-excess silicon carbide (1), but also in all areas of so-called SixC _1-x (0<x<1). . This has a huge impact on the efficiency of collecting the materials used,
This is essential for reducing the cost of the formed semiconductor device.

Electromagnetic energy 10~200W instead of 25W and 10,
The results were the same even when the power was changed to 25, 50, 100, and 200W, and a significantly higher film growth rate was obtained using helium.

In addition, the uniformity of the formed film also varies with hydrogen dilution.
5%, you can get less than ±2%,
This made a large contribution when used as a semiconductor device.

Furthermore, instead of using only helium as the carrier gas, when hydrogen was simultaneously mixed with He until the ratio of H ₂ /He reached 1/1, the film growth rate also decreased accordingly.

The present invention further provides a PIN junction on the substrate for such silicon carbide.

That is, as shown in FIG. 2A, which is a vertical cross-sectional view, P-type silicon carbide (SixC _1-x 0<x<1) 2 is deposited on a substrate such as a stainless steel substrate on which a metal electrode is formed.
8. I-type silicon carbide or silicon 29 and N-type silicon carbide (SixC _1-x 0<x<1) 30 are provided, and a transparent conductive film 32 of metal oxide or heavy compound such as ITO is further formed on the upper surface thereof. It is something that In the semiconductor 31 having the PIN structure, TMS and diborane were added as B ₂ H ₆ /TMS=1 to 5% as shown in FIG. 1 from above the surface to be formed. Then, the energy band width is 2.7
~3.0 eV, and the band width did not become smaller as when diborane was added to silane in an amount of 1% or more. After forming the P-type layer 28 in this manner, TMS is added to intrinsic or substantially intrinsic silicon or to this silicon in the thickness direction to gradually reduce the energy width,
Created silicon carbide or silicon as an intrinsic or substantially intrinsic semiconductor. This is shown in Figure 1.
Eg can be changed from 3.5 eV to 1.8 eV by introducing TMS and introducing silane from 16 to change SiH ₄ /TMS from 0 to ∞.

For example, in a photoelectric conversion device such as a solar cell, the intrinsic semiconductor layer 29 is formed to a thickness of 0.4 to 1μ, and the Eg is a P layer (more than 2.0e, especially 2.3 to 3.3eV), 28-I layer (1.5
~2.0eV) 29-N layer (more than 2.0eV, especially 2.3~
3.3eV) 30, add 0.5 to 5 mol% of PH ₃ with TMS as the main component again on the top surface of 29.
was added to form N-type SixC _1- x30 to a thickness of 100 to 500A.

In FIG. 2A, 32 is a transparent conductive film for light incidence. With such a structure, since light is not absorbed by impurities in the N layer 30, all of the light can be introduced into the I layer.
The P layer 28 and the N layer 30 sandwiching the Eg have a wide Eg, and the depletion layer generated between them makes it possible to separate pairs of electrons and holes toward the electrodes. As a result, at AM1 (100 mW/cm ² ), a conversion efficiency of 10 to 12% could be obtained with a 1 cm ² cell.

However, this stacking order is changed from layer 28 to N layers to layer 30.
If it is made into a P layer, only about 1 to 2% can be obtained, and it is extremely important that the first coating formed is P-type silicon carbide.

FIG. 2B shows a transparent substrate 37, on which a transparent conductive film 32 is provided. In contact with this surface, P-type silicon carbide 28, I-type silicon carbide or silicon 29, and N-type silicon carbide 30 are initially formed as in 15. The thickness of the I layer is 20~1000A, especially 50~
The thickness of this I layer, which is extremely thin at 200 A, is favorable for the diode characteristics obtained in addition to the electrodes 32 and 33. In other words, it is important that the leakage does not occur in the opposite direction. In this PIN junction, the P layer may be a so-called single heterojunction with a small Eg window like the I layer. In this way, a barrier could be created especially for holes with a short lifetime, so that sufficient light emission was still possible. Also, this I layer is P28-I2
In 9-N30, Eg is generally in a W-N-W relationship and has a double heterojunction. Therefore, when current was passed in the forward direction, electron holes gathered in this I layer, recombined with each other, and emitted light. In order to increase the efficiency of this light emission, it is important to set the stacking order on the surface to be formed in the present invention as P-I-N; conversely, if it is set as N-I-P, then
Part of the type impurity mixes into the I layer, making it substantially N-type. For this reason, good diode characteristics could not be obtained.

This is due to the fact that silicon carbide usually has an N type even though it is considered to be intrinsic due to the method of the present invention, and furthermore, the mobility of holes in the I layer is 1/100 to 1/100 of that of electrons.
This is presumed to be due to the fact that it is 0.

Furthermore, even if a carbide gas and a silicide gas are reacted instead of using a carbide silicide gas such as TMS in the plasma CVD method as in the present invention, the structure shown in FIG. 2B does not exhibit diode characteristics. , no luminescence was observed.

From this, it is not a sufficient condition that the material is a stoichiometric mixture of carbon and silicon, and it is extremely important that carbon and silicon are sufficiently bonded. When examining the infrared absorption spectrum, the method of the present invention shows that the method of the present invention has a broad absorption range of 600 to 1000 cm ^-1 with a peak at about 800 cm ^-1 , and that Si-C bonds are sufficiently generated because other bonds are extremely rare. I have been able to prove that I am doing so.

[Brief explanation of drawings]

FIG. 1 shows an apparatus for producing silicon carbide using the plasma vapor phase method of the present invention. FIG. 2 is a vertical sectional view of a semiconductor device obtained by the method of the present invention.

Claims (1)

[Claims] 1. P is placed on the formation surface on which the electrode is provided on the substrate.
After forming a silicon carbide film having a conductivity type of
A film having a PIN structure is obtained by forming an intrinsic or substantially intrinsic semiconductor layer on the film and, after the step, forming a semiconductor layer having an N-type conductivity on the intrinsic semiconductor layer. When forming a silicon carbide film having P-type conductivity, carbon-
A silicide carbide gas of a hydride or halide having a silicon bond and silane (SiH ₄ ) are used as reactive gases, and are introduced into a plasma atmosphere maintained at 1 atmosphere or less to cause a decomposition reaction,
1. A method for producing a P-type silicon carbide film, characterized in that the film is formed by simultaneously introducing an impurity gas containing a valent impurity into the atmosphere.