JPH0566012B2 - - Google Patents

Description

[Detailed description of the invention]

[0001]

TECHNICAL FIELD The present invention relates to a method for manufacturing a semiconductor device.

[0002] The present invention forms a transparent electrode in close contact with one surface of a semiconductor, and also includes elements constituting this transparent electrode or di- or trivalent P-type additives or pentavalent or hexavalent N-type additives in this electrode. A portion of the additive is added into the semiconductor inside the semiconductor by irradiating it with a laser or similar strong light energy, lowering the sheet resistance of the semiconductor in that region, and further increasing the contact between the electrode and the semiconductor in the doped region. The aim is to substantially integrate the

[0003] The present invention provides a semiconductor having a non-single crystal semiconductor mainly composed of a group 4 element and a transparent electrode mainly composed of an oxide or nitride containing an additive provided on the upper or lower surface of the semiconductor. The purpose of the device is to mutually dope elements or additives constituting a non-single-crystal semiconductor so that the semiconductor and the electrode are integrated or substantially integrated.

[0004]

2. Description of the Related Art Conventionally, thermal annealing has been known as a method for reducing the density of recombination centers or levels generated in a semiconductor device. This is done by annealing (heat slowing) in hydrogen or inert gas at a temperature of 300 to 700°C to produce semiconductors, especially single crystal semiconductors, or so-called MIS structures in which a gate insulator such as an insulated gate field-effect semiconductor device is provided on top of the semiconductor. In semiconductor devices (metal-insulator, especially silicon oxide-semiconductor, especially silicon), slow levels at the interface are canceled out or lattice strain in a single crystal semiconductor is removed.

[0005] Also, as high temperature annealing, 700 to 1200
℃For example, boron is added to a single crystal semiconductor at 1000℃.
It has been known that by implanting (B), phosphorus (P), arsenic (As), etc., and subsequent annealing, the amorphous state generated by this implantation can be transformed into a single crystal as if it were warm.

[0006] However, in all of these, the basic idea is to eliminate dangling bonds or vacancies in the crystal by making it more monocrystalline.

[0007] The present invention does not involve such a conventionally known thermal annealing method, but instead applies laser light or similar intense light energy (hereinafter collectively referred to as L-annealing) to a semiconductor, and as a result, the semiconductor surface or its surface is heated. This is intended to cure nearby semiconductors.

[0008] Furthermore, the present invention provides that such L-annealing is more effective for non-single crystals than for single crystals, and for non-single crystals, that is, polycrystalline or amorphous semiconductors or amorphous semiconductors formed on a substrate by a method such as a CVD method. It is particularly effective for hydrogen-containing amorphous amorphous formed by glow discharge method, plasma CVD method, etc. or polycrystal having a microscopic crystal grain size of 10 to 100 Å.

[0009]

[Problems to be Solved by the Invention] Since such non-single crystal semiconductors generally have an extremely large number of dangling bonds, they are essentially conductors heavily doped with impurities of 10 ¹⁹ to 10 ²¹ cm ^-3 . It is known to use the non-single crystal semiconductor as a semiconductor by bonding the unpaired junction with hydrogen and neutralizing it at the same time as forming a film in such a non-single crystal semiconductor. However, regarding the former, if the impurity is doped in a large amount (10 ²⁰ cm ^-3 to 50 atomic %), the impurity will precipitate, causing so-called segregation, and a lump of impurity will be generated in the semiconductor, resulting in no electrical activity. It becomes difficult to become. That is, the activity in the semiconductor (the amount of P or N type active in the semiconductor/the amount of impurities mixed in the semiconductor) has become extremely low, at 0.1 to 10%. On the other hand, in the case of a non-single crystal semiconductor doped with hydrogen, if an electrode is formed on the system or annealing is performed at a lower temperature of 300 to 700°C, the hydrogen in the semiconductor becomes a hydride such as Si- It was liberated from H bonds and released from the semiconductor as H ₂ , and the density of recombination centers increased due to thermal annealing.

[0010]

[Means for Solving the Problems] The present invention eliminates such drawbacks, and provides divalent, trivalent or pentavalent,
When a hexavalent additive is added, its activity is increased to nearly 100% by increasing crystallization, which in turn increases the electrical conductivity in the semiconductor;
Then, due to this process or low-temperature annealing at 300 to 700°C, neutralized recombination centers such as hydrogen or halogen elements that are released are added back into the semiconductor in a chemically active state, and dangling bonds are removed. The density of recombination centers in the semiconductor is lowered by combining with the semiconductor.

[0011] In addition, the present invention moves some of the additives or constituent elements of the electrode formed on the upper surface of the semiconductor, particularly the transparent electrode, into the semiconductor during L-annealing, and the boundary is changed from the previous surface. It is characterized by the fact that it has expanded from the concept of ``to'' to the concept of ``area''. As a result, the impurity activity of the semiconductor under the electrode is increased, and its electrical conductivity is extremely high and has a conductivity close to that of metal. That is, it has been found that the Fermi level can be brought to a substantially degenerate state.

[0012] Examples of the present invention used in the present invention will be described below with reference to the drawings.

【0013】

Embodiment FIG. 1 shows an embodiment of a semiconductor device used in the present invention.

[0014] A semiconductor substrate 1 is shown in FIG. 1A.
A typical example of this semiconductor substrate is a single crystal semiconductor such as silicon. This single crystal semiconductor is on the top surface.
Even if an MIS structure is provided, a part of the semiconductor substrate may be doped with impurities by ion implantation or the like, so that it becomes partially non-single crystal. In the present invention, such a semiconductor was subjected to L-annealing.
The laser used for L-annealing was a CW laser. The output was 10-70W. The position was continuously scanned using a mirror. In this way, the semiconductor layer near the surface of the semiconductor substrate at a depth of 0.1 to 3 μm was annealed. However, this L-annealing was not very effective in eliminating the interface states at or near the semiconductor-insulating film interface. In addition, it is not very effective in preventing leakage of minute currents due to minority carriers flowing in semiconductors.

[0015] In order to eliminate such drawbacks, the present invention
This semiconductor was immersed in an atmosphere maintained at one atmospheric pressure or less containing neutralized recombination centers such as hydrogen in a chemically active state excited by radio frequency induction. The temperature of this atmosphere can also be room temperature (-70 to +200°C). 0.1 to 100M from the outside of the furnace under reduced pressure
Hydrogen or an inert gas such as helium or a portion of halogen elements such as chlorine and fluorine are added at 0.01 to 3 atomic % by high frequency induction at Hz, for example 13.5 MHz.
The atmosphere was excited to a concentration of . Thus, for example, hydrogen can be the hydrogen of the generating group chemically active with _H2 , H ^* or H ⁺ . This hydrogen penetrates into semiconductors or insulators without any hindrance, and can be found in semiconductors, insulators, or dangling bonds in semiconductors such as silicon existing at their interfaces, or dangling bonds in insulators such as silicon or oxygen in silicon oxide. It combined with the pair bond and electrically neutralized it.

[0016] As a result, the semiconductor layer destroyed by ion implantation etc. has a defect density of 10 ²² cm ^-3
We were able to lower it to 10 ¹⁹ - 10 ¹⁷ cm ^-3 and further reduce it to 1/10 - 1/50. In particular, laser annealing can reduce the defect density of impurity layers that constitute the sources and drains of MIS and FETs without widening the junction, whereas induction annealing can reduce the defect density of impurity layers that constitute the sources and drains of MIS and FETs without widening the junction, whereas induction annealing can reduce the defect density at or near this junction or between the semiconductor and insulating film. This was effective in reducing the number of dangling bonds and levels at the interface. In addition, while laser annealing anneals areas closer to the interface, L-annealing neutralizes and anneals defects at a depth of 3 to 10 μm from the semiconductor surface, which cannot be processed. Induction annealing was extremely effective.

[0017] FIG. 1B shows a semiconductor layer 1 formed on a substrate 3. Semiconductors or semiconductor layers are made using a thermal decomposition method using whole silicides such as silane.
It was formed at a temperature of ~900°C. In order to fabricate this semiconductor layer, CVD (Chemical Vapor
Deposition) was invented by the present inventor in 1973.
1389. Furthermore, the invention was carried out based on the glow discharge method, plasma CVD method, etc., patent application No. 53-67507 (filed on June 5, 1978) filed by the inventor. The semiconductor layer 1 formed by this method is made of a non-single crystal semiconductor, and multiple junctions such as PN junctions, PIN junctions, PNPN...PN junctions are formed selectively or approximately parallel to the substrate surface in the semiconductor. Furthermore, such a non-single crystal semiconductor is provided with an insulated gate field effect transistor or a semiconductor device integrated therewith. For example, application No. 53-124022 (1978) which is an invention of the present inventor.
(October 7, 2016).

[0018] When such a non-single-crystal semiconductor is selectively or entirely subjected to L-annealing similar to that shown in FIG. By combining these, we were able to reduce the density to 1/10 ³ to 1/10 ⁵ . However, at the same time, the elements constituting the semiconductor combine with hydrogen, etc. and are neutralized, and some of the dangling bonds change from Si-H bonds to Si-, creating dangling bonds instead. I let it happen.
At this time, it was found that hydrogen bonds with itself through Si--H, and only remains in a stable state as _H2 in the semiconductor. That is, Process 1 Si-H+H-Si→Si-Si+H ₂ Process 2 Si-H+H-Si→2Si-+H

[0019] It has been found that in the case of a large number of process 2, crystallization is promoted more, and even though the density of recombination centers is closer to single crystallization than process 1, it increases. In other words, even though the electrical conductivity has increased by about 100 times as the silicon covalently bonds with each other in process 1 and approaches a single crystal, the density of recombination centers is lower than that created by glow discharge, etc. For coatings, before L-annealing it is 10 ¹⁷ to 10 ¹⁸
cm ^-3 to 10 ¹⁸ to 10 ¹⁹ cm ^-3 , and it was found that although the hydrogen content in this semiconductor remained unchanged at about 20 to 30 mol%, it increased by one order of magnitude.
In other words, this fact means that free hydrogen bonds with each other,
It was found that the hydrogen could not fully recombine with the dangling bonds of silicon in an extremely short period of time.

[0020] In addition, since the non-single crystal semiconductor film formed by low pressure CVD method etc. does not contain recombination center neutralized substances in advance, the crystal grain boundaries are reduced by 0.1μ to 50μ from 10 to 1000Å by L-annealing. It was possible to increase the size of the crystal and make it even more monocrystalline. The same effect can be obtained by using a ruby laser or glass laser (output 10 to 1000 MW) with a pulse width of 10 to 100 ns instead of the CW oscillation described above. As a result, the defect density in an intrinsic semiconductor that is not doped with P- or N-type impurities (in this case also includes semiconductors doped with background-level impurities) can be reduced to ¹⁰ cm. did it. However, for use as a semiconductor, this density must be lowered to 10 ¹⁴ to 10 ¹⁶ cm ^-3 or lower. Furthermore, in order to similarly reduce the density in a portion deeper than the surface of the semiconductor layer, the present invention is characterized in that induction annealing is added at the same time as or after this L-annealing. This induction annealing may be performed by chemically exciting the above-mentioned neutralized product in advance at a position away from the substrate using microwaves and guiding it onto the substrate.
Microwave has an output of 30-200W, for example 2.46GHz
was used. The reaction system is below 1 atm, e.g. 0.01~
The pressure was 10 Torr, and the atmosphere was hydrogen or a neutralized hydrogen solution with 30 to 50% helium added. By placing the present semiconductor device in such an atmosphere for 10 minutes to 1 hour, the defect density described above can be reduced to 10 ¹⁵ to ^{10 16 .}
It was possible to reduce it to cm ^-3 . This defect density is determined by the glow discharge method, plasma
It has no relation to the CVD method, cluster vapor deposition method, low pressure CVD method, vacuum vapor deposition method, ion plating method, etc., and by combining the L-annealing and induction annealing of the present invention, semiconductors can be manufactured without much dependence on the manufacturing method. We were able to bring it closer to its original state.

[0021] FIG. 2 shows another embodiment of the present invention, in which a transparent electrode is formed on a semiconductor.

[0022] In FIG. 2A, the substrate 3 is made of glass, ceramic or a composite material such as glass epoxy, Kapton,
It may be an insulating substrate made of an organic material such as polyimide, a conductive substrate made of stainless steel, titanium or titanium nitride, or a composite substrate in which a conductor is selectively provided on the above-mentioned insulating substrate.

[0023] A semiconductor layer 1 was formed to have a non-single crystal structure on these substrates. This semiconductor was manufactured using a plasma CVD method using silicide as the main component. This semiconductor contains PN junction, PIN junction or PNPN
...PN multiple junction, PINI...IPIN multiple junction were formed. The thickness of the semiconductor layer is 0.5-5μ thick. Furthermore, a conductive film 2 made of tin oxide, indium oxide, antimony oxide, or a mixture thereof or a nitride of tin, indium oxide, antimony oxide, or a mixture thereof is applied to this upper surface as a single-layer or multi-layer electrode using the same plasma CVD method.
It was manufactured to a thickness of 0.05 to 3μ. This conductive film is optically transparent and is characterized by low absorption of laser light and visible light.

Furthermore, L-annealing is added to this FIG. 2A, and a transition region 5 is provided at the boundary between the transparent electrode 2 and the semiconductor layer 1 as shown in FIG. or nitrogen, as well as indium (In), gallium (Ga), aluminum (Al), which exhibits P-type conductivity in semiconductors,
Boron (B) or zinc (Zn), cadmium (Cd)
was added as an additive. In particular, metals have properties when used as a single substance, and in semiconductors, In has a P-type conductivity type.
Alternatively, a mixed additive of In and B was effective in extremely increasing the P-type conductivity in this transition region.

[0025] This L-annealing increases the melting amount of In and B by 10 to 10 ³ from the concentration of 10 ²⁰ cm ^-3 which is the melting limit.
It has the characteristics of being twice as high, ^{supersaturated and} not causing segregation.
Addition of 3 atomic % was extremely effective in increasing conductivity without causing scattering of impurities with respect to holes. In the present invention, the diameter of the crystal grain boundary of the non-single crystal semiconductor is further increased by L-annealing from 10 to 10.
The size is 1 to 50 μ compared to 1000 Å, and by approaching a single crystal, the conductivity can be increased by 10 to 10 ³ times.

[0026] However, in order to prevent the generation of dangling bonds due to this L-annealing,
Induction annealing was performed on the dangling bonds, and active hydrogen was added to the dangling bonds to electrically neutralize them.
By doing so, it is possible to improve the photoelectric conversion efficiency in the short wavelength region on the light transmitting side of a photoelectric conversion device, especially a solar cell, etc., and furthermore, it is possible to improve the photoelectric conversion efficiency in the short wavelength region on the side where light passes through the photovoltaic device, especially in a solar cell, etc.
We were able to achieve a correction effect of 95-100% in the wavelength range of .

[0027] Also, if you want to make the semiconductor under the transparent electrode N-type, the additives to the transparent electrode should be pentavalent additives such as antimony (Sb), arsenic (As), and phosphorus (P), or tellurium ( 6 such as Te), selenium (Se)
Additives of 10 ^{to 20} cm ^-3 to transparent electrodes of tin oxide or nitrides such as tin nitride or antimony nitride.
It may be added at a concentration of 30 atomic %. Among these additives, particularly Sb or a mixture of Sb and P, the semiconductor layer directly under the electrode is similarly made N-type by L-annealing, and the concentration exceeds the solid solubility limit of the amount added, causing segregation. We were able to make it an N-type with an activity close to 100%.

[0028] By such L-annealing, the non-single crystal semiconductor progresses to single crystallization, and some components or additives of the transparent electrode are deposited to a depth of 50 to 5 × 10 ³ Å.
It was possible to dope to an extremely shallow depth of 500 Å.
This doped surface is a transition region that can be in close contact with both the electrode and the semiconductor, and its resistivity is between 10 ^-1 and ^{10 -4} Ωcm.
^-1 , which is close to a metal, and is estimated to be in a degenerate state at the Fermi level in terms of quantum theory. In addition, because this transition region is thin, in photoelectric conversion devices, electrons are excited by short-wavelength light.
Because hole pairs were generated and the recombination centers of both were neutralized with a neutralizer such as hydrogen, it was possible to guide them to the electrode without recombining.

[0029] In addition, in this invention, since it is forcibly annealed by L-annealing, some elements, such as oxygen or nitrogen, may locally react with the elements constituting the semiconductor, resulting in local lower silicon oxide or Failure modes such as making silicon nitride and using it as an insulating film
It was extremely reliable as it did not cause any generation when left at high temperatures of 100 to 150°C.

[0030] In FIG. 2C, the transparent electrode 2 is formed on the lower side,
This is the case where the semiconductor layer 1 is formed on the upper side.
In such a case, if the substrate 3 is made of glass or the like, annealing using laser light incident from the lower glass side is preferable. However, if the semiconductor layer is as thin as 0.05-2μ, L-annealing may be performed through the semiconductor layer from above.

[0031] As a result, the transition region 5 was formed as shown in FIG. 2D, similar to FIG. 2B. Depending on the direction of laser light irradiation, the crystal grain size of the semiconductor layer increases; if the laser beam is irradiated from below, the lower part of the semiconductor layer will be larger and the upper part will be smaller; if the laser beam is irradiated from the upper side, as in FIG. 2B, the semiconductor layer will become larger. The upper part of 1 becomes larger and the lower part becomes smaller as a crystal. From this, it was found that the crystal grain size in the depth direction can be controlled by the irradiation direction, intensity, and irradiation time of the laser beam.

[0032] FIG. 2E shows a case where transparent electrodes are formed on the upper side 2 and further on the lower side 4 with the semiconductor layer 1 sandwiched therebetween. As a result, by L-annealing, the transition region 3 is made to be P type and the transition region 6 is made to be N type, so that a so-called P-N junction can be suitably made. Of course, in the embodiment shown in the drawings, the lower electrode 4 is formed by forming a conductor electrode made of a compound of Sn and Sb on the base metal on the substrate, and utilizing the reflection of the laser beam from the upper side of the lower electrode. A method may be used in which a part of this electrode is added to the semiconductor layer. Conversely, it is also possible to create a NIP junction by having the upper electrode contain a pentavalent or hexavalent additive, and the lower electrode containing a trivalent or divalent additive.

[0033] After these L-anneals, the unpaired bonds generated by the L-anneals in the entire semiconductor layer are neutralized by induction annealing with H, He, etc., which are recombination center neutralizers, and are made electrically inactive. This is extremely important for operating the semiconductor device.

[0034] In FIGS. 2A and 2C, it is not explicitly stated that a semiconductor layer of N or P type conductivity is formed on the substrate or a semiconductor layer, and that a PN junction or other junction is formed within this semiconductor layer. Ta. but
In the CVD method, plasma CVD method, glow discharge method, etc., semiconductors of these conductivity types are manufactured by adding B as an impurity for P-type and P for N-type at the same time as forming the semiconductor layer. do it. In addition, this concentration exceeds the solid solution limit, and in non-single crystal semiconductors, the activity is only 3 to 30%, so
These can be increased to 90 to 100 by performing L-annealing.
%, and it has become possible to have extremely structural sensitivity as a semiconductor.

[0035] FIG. 2G is an example in which transparent electrodes are selectively provided on the conductor layer.

[0036] As a result, a shear joint (5 to 200 Å)
can be made as 5, 5' as shown in FIG. 2(H).

[0037] FIG. 3 is an example of a manufacturing apparatus for implementing the present invention. The outline of the device will be explained based on the drawings and the description as before.

[0038] Substrate 11 on which a semiconductor is formed
from the input chamber 20 to the output chamber 21 by the loader 28. Chamber 23 is 0.01~
Since the reaction is carried out under a reduced pressure of 100 Torr, especially 0.1 to 10 Torr, the neutralized gas is hydrogen from 15, an inert gas such as helium from 16, and a halogen element such as HCl from 17. Further, the exhaust gas is exhausted by a vacuum pump 19 via a needle valve 18.

[0039] The laser beam is transmitted from the laser 12 to the mirror 13.
The wafer is then scanned onto the substrate to perform L-annealing. In this device, a high frequency induction furnace is installed outside the chamber at the same position where the laser is irradiated. This high frequency induction furnace 22 adopted a voltage heating method, and used 13.56MHz and 100W to 1KW.
Thereafter, a furnace 25 is provided to perform low-temperature annealing of the entire structure at a temperature of 300 to 700°C, and a special high-frequency induction furnace 24 is provided independently behind the furnace 25. This induction furnace may also be of a parallel plate type so as to face the substrate 11.

[0040] As a result, a discharge occurs in the chamber, and hydrogen and other chemically active groups of generating groups (radicals) are doped into the semiconductor.
It was possible to neutralize it by combining with the unpaired bond.
In addition, conventionally, L-annealing could only be performed in air, but by doing so, it can be performed in hydrogen, an inert gas, especially helium, and as a result, a characteristic of ring-shaped L-annealing on the irradiated surface is achieved. It was possible to reduce the occurrence of striped patterns.

[0041] In the present invention, the Q-switch pulse oscillation laser or
Although a CW laser is used, a lamp of xenon or the like may be used to generate a flash or the like to produce a similar effect. The temperature of the substrate is raised and cooled extremely rapidly, and even though microscopic movement of the semiconductor or additives in the semiconductor can be carried out at high temperatures and in a substantially molten state, large movements such as segregation of impurities cannot occur, and heat A feature of this method is that impurities or additives with a concentration higher than the solid solubility limit in the annealing method are added to the semiconductor without precipitation.

[0042] In the previous embodiments of the invention,
The transparent electrode is left as is. However, it goes without saying that this electrode may be removed once with an etching solution and a new transparent electrode may be formed again. In addition, the first transparent electrode is made of, for example, nitride.
After forming to a thickness of 100 to 1000 Å, photoannealing
Furthermore, the second transparent electrode may be formed of an oxide to a thickness of 0.1 to 2 μm.

[0043] Furthermore, the previous embodiments of the present invention have been described using silicon as the main semiconductor. But Si _x Ge _1-x
(0<x<1), Si _x Sn _1-x (0<x<1), Si _x C _1-x
(0.5<x<1) or Group 4 semiconductors such as Sn or Group 3 and 5 compound semiconductors such as GaAs and GaAlAs,
Furthermore, or in a part of the semiconductor, Si _x O _2-x (0<x<
2) Form a part of such a semiconductor with a lower oxide or lower nitride such as Si _x N _4-x (0<x<4), and continuously change the energy band width to a W-N structure. It goes without saying that other semiconductors may also be used.

[0044] In the embodiments of the present invention, the transparent electrode is mainly an oxide conductive transparent electrode such as tin oxide, indium oxide, or antimony oxide.
However, chemically more stable nitride conductive transparent electrodes such as tin nitride, indium nitride, antimony nitride, titanium nitride, and germanium nitride may be used, and silicon nitride and mixtures thereof may be used as conductive transparent electrodes. Good too.

[0045] In addition, it goes without saying that the present invention can also be applied to a semiconductor device in which a very thin nitride film is provided that allows a tunnel current of 10 to 50 Å to flow at the boundary between a semiconductor layer and a transparent oxide electrode.

【0046】

[Effects of the present invention] Furthermore, the semiconductor device of the present invention is applicable not only to photoelectric conversion devices, especially solar cells, but also to all semiconductor devices such as integrated circuits using MIS.FET, light emitting elements, semiconductor lasers, transistors, diodes, etc. It goes without saying that it can be done.

[Brief explanation of the drawing]

FIG. 1 shows an example of a semiconductor device for implementing the present invention.

FIG. 2 shows an example of a semiconductor device for illustrating another embodiment of the present invention.

FIG. 3 is an example of a manufacturing apparatus for implementing the present invention.

[Explanation of symbols]

1 Semiconductor layer 2 Conductive film 3 Board 4 Lower electrode 5 Transition area.

Claims (1)

[Claims] 1. After performing optical annealing by irradiating one main surface of a non-single-crystal semiconductor mainly composed of a group 4 element with a laser or similar intense light energy, the semiconductor is subjected to high-frequency or microwave treatment. 1. A method for manufacturing a semiconductor device, comprising placing the semiconductor device in a hydrogen, halogen element, or inert gas atmosphere in a plasma state, and performing thermal annealing in a heated atmosphere at a temperature of 300 to 700°C.