JP2005086135A - Epitaxial wafer for heterobipolar transistor and manufacturing method thereof - Google Patents

Abstract

An epitaxial wafer for a heterobipolar transistor having a high activation rate of carbon added as a p-type impurity to a base layer and a method for manufacturing the same are provided.
In an epitaxial wafer for a heterobipolar transistor including a collector layer 3, a base layer 4, and an emitter layer 5 on a semi-insulating InP substrate 1, trimethyl phosphorus or the like is used as a raw material for epitaxial growth of all or part of the emitter layer 5. By using this organophosphorus compound, hydrogen atoms taken into the base layer 4 are consumed, and the activation rate of carbon impurities is improved.
[Selection] Figure 1

Description

  The present invention relates to an epitaxial wafer for heterobipolar transistors used for optical communication, wireless communication, signal measurement, and the like, and a method for manufacturing the same.

A heterobipolar transistor (hereinafter referred to as HBT) includes a collector layer, a base layer, and an emitter layer, and is characterized in that the band gap of the emitter layer is made larger than that of the base layer. In general, the collector layer is undoped or n-type, the base layer is high-concentration p-type, and the emitter layer is andb or n-type. In an HBT having a p-type base layer, leakage from the base of holes, which are minority carriers, to the emitter is suppressed by the difference in the band gap, so that a high gain can be obtained.
In HBT, an epitaxial wafer with a collector layer, a base layer and an emitter layer grown on a semiconductor substrate using an epitaxial growth apparatus is etched using a fine processing technique to expose each layer, and electrodes are formed thereon. Produced.
In an HBT using a GaAs substrate, GaAs is generally used for the collector layer and the base layer, and AlGaAs or InGaP is used for the emitter layer.
It has been reported that an HBT can be manufactured using an InP substrate and operates at a higher speed than an HBT manufactured using a GaAs substrate (Non-patent Document 1).
An outline of an HBT epitaxial wafer using an InP substrate is shown in FIG. 1, and an outline of the HBT is shown in FIG. The HBT epitaxial wafer is required to have high accuracy with respect to the film thickness and impurity concentration of each layer. In particular, the base layer needs to be a thin film and is required to have a high p-type carrier concentration by doping p-type impurities at a high concentration. Abrupt changes in composition and impurity concentration are required at both the interface between the collector layer and the base layer, and between the base layer and the emitter layer.

When using an InP substrate, it is common to use InGaAs for the collector layer and the base layer, and InP for the emitter layer. It is also possible to use InGaP or InAlP having a larger band gap than InP as the emitter layer. The InGaAs cab layer on the emitter layer is grown for the purpose of reducing the electrical resistance between the emitter layer and the emitter electrode.
In recent years, HBTs using GaAsSb as a base layer have been reported (Non-patent Document 2). In this report, an InP substrate is used and InP is used for both the collector layer and the emitter layer. Due to the band gap relationship between InP and GaAsSb, the leakage of minority carriers from the base layer to the emitter layer is strictly suppressed, so the operation speed is extremely high.

  As a main method of manufacturing an epitaxial wafer made of a compound semiconductor, a metal organic chemical vapor deposition method (MOCVD) in which an organic metal diluted with a carrier gas and a group V hydride are used as raw materials and a semiconductor substrate heated in the gas is held. And a molecular beam epitaxy method (MBE method) in which a metal element or a group V element is supplied as a molecular beam to a substrate in a container kept in a high vacuum. Compared with the MBE method, the MOCVD method is superior in mass productivity in that a plurality of large-area substrates can be grown simultaneously. In particular, in the growth of a compound semiconductor containing phosphorus such as InP, since the vapor pressure of phosphorus is high, it is difficult to apply the MBE method, and the MOCVD method is advantageous.

  Carbon is attracting attention as a p-type impurity that can be doped at a high concentration. Carbon, which is a group IV element, is incorporated into the position of a group V element in many III-V compound semiconductors and becomes an acceptor. Since carbon has a smaller diffusion coefficient in a compound semiconductor than zinc which is taken into the position of a group III element and behaves as a p-type impurity, the deterioration of the impurity concentration profile due to diffusion in the solid phase is small. Also, carbon halides such as carbon tetrabromide, which are often used as carbon source gases in MOCVD, do not adsorb on the piping of the epitaxial equipment or the walls of the reaction vessel, and have a steep impurity profile by switching the gas. An epitaxial wafer can be obtained.

  When carbon is used as a p-type impurity, there is a problem that the p-type carrier concentration does not coincide with the carbon concentration and part of the carbon may become inactive. This phenomenon appears remarkably when carbon is doped into InGaAs, and it is reported that 90% or more of carbon becomes inactive (Non-patent Document 3). Even when GaAsSb is doped with carbon, inactivation of part of the carbon has been confirmed (Non-patent Document 4). The reason why a part of carbon becomes inactive is that hydrogen atoms generated by thermal decomposition of hydride as a V group material and hydrogen gas as a carrier gas are taken into the compound semiconductor at the same time as carbon incorporation during epitaxial growth. is there. Since these hydrogen atoms supply electrons to carbon, carbon does not function as an acceptor, and the p-type carrier concentration becomes smaller than the carbon concentration.

By annealing a carbon-doped p-type compound semiconductor in an atmosphere in which no hydrogen exists, such as in a nitrogen atmosphere, hydrogen atoms in the semiconductor can be removed. However, in the epitaxial wafer for HBT, the base layer is covered with the emitter layer, and the diffusion of hydrogen atoms in the ANDOR or n-type compound semiconductor is extremely slow, so that the epitaxial wafer is annealed in a nitrogen atmosphere. Cannot remove hydrogen atoms incorporated into the base layer.
In the epitaxial growth apparatus, it is possible to create an atmosphere in which no hydrogen gas or group V hydride exists after the base layer is grown. However, in the subsequent growth of the emitter layer, hydrogen atoms are supplied by thermal decomposition of the hydride that is a group V raw material and the hydrogen gas that is the carrier gas, so that the hydrogen atoms diffuse again in the base layer and deactivate the carbon. To do.

As described above, it is difficult to reduce the incorporation of hydrogen atoms into the base layer. In the conventional method, the supply amount of the hydrogen atom as a supply source of hydrogen atoms is reduced as much as possible, the epitaxial growth temperature and the pressure in the reaction vessel are optimized, etc., to reduce the intake amount of hydrogen atoms. However, the fact is that a lot of hydrogen remains in the grown epitaxial wafer.
In an HBT epitaxial wafer using InP as a substrate manufactured by a conventional method, most of carbon impurities are inactivated by hydrogen atoms. In order to obtain the required p-type carrier concentration, excessive carbon must be doped, and the deterioration of crystallinity due to excessive carbon is jeopardized.
Further, since the base layer is exposed in the process of processing the epitaxial wafer for HBT fabrication, part of the incorporated hydrogen atoms is desorbed, and as a result, the p-type carrier concentration is changed. Such side reactions are difficult to control and adversely affect the uniformity of the HBT product.

In the above-mentioned Non-Patent Document 1 and Non-Patent Document 3, in a state where a base layer is partially exposed in the course of microfabrication, high-temperature annealing is performed in a nitrogen gas atmosphere to desorb hydrogen atoms, thereby deactivating carbon. The influence of the conversion is reduced. In this method, hydrogen atoms in the base layer immediately below the emitter layer are desorbed from the exposed surface of the base layer after diffusing horizontally with the substrate along the base layer. In this way, the diffusion distance becomes long, and the distance varies depending on the location. Therefore, it is difficult to make the distribution of hydrogen atoms after annealing uniform. In addition, this method increases the number of processes and may cause thermal damage to the finely etched surface.
There is also a problem in reliability after completing the HBT as a device operating at high speed through all the steps. This is because some of the hydrogen atoms are gradually desorbed during use. Changes in device characteristics due to a gradual change in the p-type carrier concentration accompanying desorption of hydrogen atoms, and changes in characteristics due to hydrogenation of electrodes and the like are in danger.

K. Kurishima, et al., Jpn. J. App1. Phys. Vol. 37 (1998) p. 1353 M.W.Dvorak, et al., IEEE Elec. Dev. Lett. Vol. 22 (2001) p.361 N. Watanabe, et al., J. Cryst. Growth Vol. 195 (1998) p. 48 Oda, et al., 63rd Joint Lecture on Applied Physics 26a-YD-10 (2002)

Electronic devices that operate at high speed have been demanded due to the increase in optical communication speed due to the increase in communication demand and the increase in frequency used in wireless communication, and HBT is promising. Further, the HBT using the InP substrate operates at a higher speed than that using the GaAs substrate.
In the HBT using the InP substrate, InGaAs or GaAsSb doped with carbon is used as the base layer. In the case of producing an epitaxial wafer for HBT by the MOCVD method, there is a problem that a part of carbon is inactivated by hydrogen atoms during epitaxial growth. Inactivation by hydrogen atoms not only reduces the p-type carrier concentration, but also makes it difficult to ensure device uniformity and stability.

SUMMARY OF THE INVENTION An object of the present invention is to provide an HBT epitaxial wafer with a small amount of hydrogen atom incorporation by MOCVD method with excellent mass productivity under the present situation where reduction of carbon inactivation by hydrogen atoms is approaching the limit. It is.
By using the epitaxial wafer of the present invention, it is possible to produce a higher quality HBT. By using the epitaxial wafer manufacturing method of the present invention, an epitaxial wafer with a small amount of hydrogen uptake can be obtained.

In order to solve the above problems, the present invention is configured as described in the claims. That is,
The hetero bipolar transistor epitaxial wafer using an InP substrate and including a collector layer, a base layer, and an emitter layer according to claim 1, wherein the base layer is a p-type compound semiconductor containing carbon as an impurity, The emitter layer is made of InP, InGaP, or InAlP grown using an organic phosphorus compound as a raw material to form an epitaxial wafer for a hetero bipolar transistor.

  Further, according to claim 2, a heterobipolar using an InP substrate and including a collector layer, a base layer, a first emitter layer in contact with the base layer, and a second emitter layer in contact with the first emitter layer In the epitaxial wafer for a transistor, the base layer is a p-type compound semiconductor containing carbon as an impurity, and the first emitter layer is grown from an organic phosphorus compound as a raw material. This is an epitaxial wafer for a hetero-bipolar transistor that is InAlP.

  In addition, as described in claim 3, in claim 1 or claim 2, as the organic phosphorus compound, any one or more of trimethyl phosphorus, dimethyl phosphorus, monomethyl phosphorus, triethyl phosphorus, diethyl phosphorus, and monoethyl phosphorus are used. The grown epitaxial bipolar transistor wafer is a hetero bipolar transistor.

  In addition, as described in claim 4, when growing the epitaxial bipolar transistor epitaxial wafer according to any one of claims 1 to 3, the organic phosphorus compound is supplied after the base layer is grown. A method for producing an epitaxial wafer for a heterobipolar transistor includes a step of producing an atomic layer of phosphorus at the interface between the base layer and the emitter layer.

  In addition, as described in claim 5, when growing the epitaxial wafer for a hetero bipolar transistor according to any one of claims 1 to 3, a part of the carrier gas during the growth of the emitter layer Alternatively, a method of manufacturing an epitaxial wafer for a hetero-bipolar transistor using nitrogen gas for all of them is used.

  In addition, as described in claim 6, in the epitaxial wafer manufacturing method according to claim 4 or 5, when the phosphorus atomic layer at the interface between the base layer and the first emitter layer is formed. Thus, a method for producing an epitaxial wafer for a hetero-bipolar transistor using nitrogen gas for a part or all of the carrier gas is used.

  According to the present invention, it is possible to provide an epitaxial wafer having a high activation rate of carbon added as a p-type impurity to the base layer. By using this epitaxial wafer, a heterobipolar transistor excellent in operating characteristics could be realized.

  FIG. 1 shows an outline of a cross section of an epitaxial wafer for HBT (heterobipolar transistor), and FIG. 2 shows an outline structure of a cross section of a heterobipolar transistor using an InP substrate. 1 and 2, 1 is a semi-insulating InP substrate, 2 is an InP buffer layer, 3 is an InGaAs collector layer, 4 is an InGaAs base layer, 5 is an InP emitter layer, 6 is an InGaAs cap layer, 7 is a collector electrode, 8 Represents a base electrode, and 9 represents an emitter electrode.

  If hydrogen atoms can be desorbed from the carbon-doped base layer during the growth of an HBT epitaxial wafer using an InP substrate, an epitaxial wafer with less inactivation of carbon impurities by hydrogen atoms can be realized. In the present invention, the epitaxial growth of the entire emitter layer, or the portion of the emitter layer that is close to the base layer is the first emitter layer, and the epitaxial growth is performed with a raw material that has a low supply of hydrogen atoms, such as an organic phosphorus compound. A significant reduction in the hydrogen atom concentration was achieved.

  The effect of using an organophosphorus compound has a stronger hydrogen atom desorption effect than a treatment in an atmosphere in which no hydrogen atom exists, such as annealing in a nitrogen gas atmosphere. This phenomenon is manifested by the consumption of hydrogen atoms in the semiconductor due to the decomposition of the organophosphorus compound on the semiconductor surface by the reaction illustrated in the formula (1), and is a phenomenon discovered by the present invention. is there.

(Decomposition of trimethyl phosphorus on the semiconductor surface) (Formula 1) Formula (CH ₃ ) ₃ P + 3H (hydrogen in the semiconductor) → P (phosphorus in the semiconductor) + 3CH ₄
The effect of consuming hydrogen in the semiconductor due to the decomposition of the organophosphorus compound is stronger as the number of alkyl groups is larger. Further, the effect of the ethyl group is weaker than that of the methyl group because a plurality of reactions shown in the formula (2) occurs. Tertiary butylphosphine, which has been attracting attention as a MOCVD raw material with low toxicity, decomposes as shown in the formula (3) at a temperature lower than the temperature of the semiconductor substrate during epitaxial growth and does not supply an alkyl group to the substrate surface. There was no effect of consuming hydrogen.

(Decomposition of ethyl phosphorus compound on semiconductor surface) (Formula 2) Formula (C ₂ H ₅ ) −P + H (hydrogen in semiconductor) → P (phosphorus in semiconductor) + C ₂ H ₆
(C ₂ H ₅ ) -P → P (phosphorus in the semiconductor) + C ₂ H ₄ + H (hydrogen supplied to the semiconductor surface)
(Thermal decomposition of tert-butylphosphine in the vicinity of the semiconductor) (Formula 3) Formula (CH ₃ ) ₃ C—PH ₂ → (CH ₃ ) ₂ = CH ₂ + PH ₃
Since diffusion of hydrogen atoms is slow in an undoped or n-type semiconductor, even if the thickness of the first emitter layer grown using an organophosphorus compound is small, hydrogen atoms from hydrides supplied in the subsequent epitaxial growth Is less diffused into the base layer.

Hydrogen atoms necessary for the decomposition of the organophosphorus compound can be supplied not only from within the semiconductor but also from a hydrogen carrier gas. In order to strengthen the effect of desorption of hydrogen atoms in a semiconductor by the above-described organophosphorus compound, it has been found effective to replace part or all of the carrier gas with nitrogen gas.
The base layer thickness, carrier concentration, and crystallinity are the main factors that determine the characteristics of the HBT. In the present invention, it is not necessary to change the epitaxial growth conditions of the base layer from those according to the conventional method. Therefore, the base layer having the optimum crystallinity obtained by the conventional method can be easily reproduced. There was a concern that carbon originating from an organophosphorus compound would be mixed into the emitter layer, but the carbon concentration in the emitter layer was below the measurement limit of secondary ion mass spectrometry.

The epitaxial wafer for HBT of the present invention was produced using an epitaxial growth apparatus by the MOCVD method. Trimethylaluminum, trimethylgallium, and trimethylindium were used as raw materials for the group III elements Al, Ga, and In, respectively. Arsine was used as the As raw material, and trimethylantimony was used as the Sb raw material.
Various organic phosphorus compounds and phosphine were used as raw materials for phosphorus, which is a feature of the present invention.
Hydrogen gas and nitrogen gas were used as the carrier gas.
Carbon was used as an impurity for making the base layer p-type, and carbon tetrabromide was used as a raw material. In <Examples 1 to 12> shown below, the collector layer and the emitter layer were undoped unless otherwise specified. When growing the n-type collector layer and the n-type emitter layer, Si was used as an n-type impurity, and disilane was used as a raw material.
As the substrate, (001) plane semi-insulating InP was used.

  The carbon concentration and hydrogen concentration in the base layer are measured by removing the emitter layer made of InP, InGaP or InA1P by etching at room temperature after growing the epitaxial wafer, exposing the base layer, and then measuring the p-type carrier concentration electrically. Was determined by After etching, annealing is performed at 500 ° C. in a nitrogen gas atmosphere to desorb hydrogen, and the p-type carrier concentration measured after cooling to room temperature in the nitrogen atmosphere is defined as the carbon concentration. The hydrogen concentration was obtained by subtracting the carrier concentration. Secondary ion mass spectrometry was performed on several samples, and it was confirmed that the carbon and hydrogen concentrations determined by electrical measurement coincided with the carbon and hydrogen concentrations determined by secondary ion mass spectrometry.

In the following examples, the ratio between the p-type carrier concentration immediately after etching and the carbon concentration was expressed as the activation rate. As the activation rate is larger, it means that the carbon of the base layer after epitaxial growth is electrically active, that is, less hydrogen atoms are taken up.
The outline of the epitaxial wafer for HBT according to the present invention is the same as that shown in FIG. A feature of the present invention is that the emitter layer in contact with the base layer is epitaxially grown using an organic phosphorus compound.
In order to facilitate the etching of the emitter layer made of InP, InGaP or InA1P and to facilitate the measurement of the carbon and hydrogen concentrations of the base layer, an epitaxial wafer not including an InGaAs cap layer was also used in the following examples. It was confirmed by secondary ion mass spectrometry that the presence or absence of the cap layer did not affect the carbon and hydrogen concentrations of the base layer.

<Example 1>
A 200 nm InGaAs collector layer, a 50 nm InGaAs base layer, a 50 nm InP emitter layer, and a 70 nm InGaAs cap layer were epitaxially grown on an InP substrate at a growth temperature of 450 ° C.
Trimethyl phosphorus was used as a phosphorus raw material for the InP emitter layer. For comparison, an InP emitter layer was grown using phosphine, which is a conventional method, as a phosphorus material. Combined with the electrical characteristics, secondary ion mass spectrometry was also performed.
FIG. 4 shows the concentration distribution in the depth direction of carbon and hydrogen of an epitaxial wafer produced by the conventional method, and FIG. 3 shows the concentration distribution in the depth direction of carbon and hydrogen of the epitaxial wafer of the present invention. As is clear from the comparison between FIG. 4 and FIG. 3, the hydrogen concentration in the base layer is greatly reduced by the present invention. Further, no carbon is observed in the InP emitter layer.
The values obtained by the electrical measurement are as follows. The carbon concentration is 7.0 × 10 ¹⁹ cm ⁻³ and the activation rate is 7% in the conventional method. In the present invention, the carbon concentration is 7.0 × 10 ¹⁹ cm ⁻³ and the activation rate is 95%. Met. This corresponds to a hydrogen concentration of 6.5 × 10 ¹⁹ cm ⁻³ in the conventional method and a hydrogen concentration of 2.5 × 10 ¹⁸ cm ⁻³ in the present invention, and the hydrogen concentration is greatly reduced to 1/25. It is shown that.
According to the present invention, it has become possible to produce an epitaxial wafer for HBT without using phosphine, which is a specific high-pressure gas and dangerous to handle, in addition to realizing a significant reduction in hydrogen concentration.

<Example 2>
The same structure as in Example 1 was grown at 400 ° C. In the conventional method, the carbon concentration was 7.0 × 10 ¹⁹ cm ⁻³ and the activation rate was 7%, and in the present invention, the carbon concentration was 7.0 × 10 ¹⁹ cm ⁻³ and the activation rate was 90%. The reason why the activation rate is smaller than that of Example 1 is considered to be that the decomposition rate of methyl groups is slowed by the lower growth temperature, and the elimination of hydrogen is reduced. Although it was feared that carbon was mixed into the emitter layer due to the slow decomposition of the methyl group, in the secondary ion mass spectrometry, no carbon was observed in the InP emitter layer.

<Example 3>
A 200 nm InGaAs collector layer, a 50 nm InGaAs base layer, and a 100 nm InP emitter layer were epitaxially grown on an InP substrate at a growth temperature of 450 ° C. The 100 nm emitter layer was divided into a first emitter layer in contact with the base layer and a second emitter layer in contact with the first emitter layer, and growth was performed so that the total was 100 nm. Trimethyl phosphorus was used as the phosphorus material for the first emitter layer, and phosphine was used as the phosphorus material for the second emitter layer.
In all the wafers, the carbon concentration of the base layer was 7.0 × 10 ¹⁹ cm ⁻³ . The activation rate is 45%, 60%, 80%, 85%, 90%, and 95% for wafers having a thickness of the first emitter layer of 0.5 nm, 1 nm, 2 nm, 5 nm, 10 nm, 20 nm, and 50 nm. 95%. Even if the InP layer grown using trimethyl phosphorus as a raw material is extremely thin, the effect of desorbing hydrogen atoms is high. If the thickness is 10 to 20 nm or more, re-diffusion of hydrogen atoms supplied from phosphine during the second emitter growth can be completely prevented.

<Example 4>
In producing the structure shown in Example 3, trimethyl phosphorus was supplied after the base layer was grown, the surface was covered with phosphorus, and a structure in which the raw material of the phosphorus atomic layer at the interface between the base layer and the emitter layer was trimethyl phosphorus was produced. . Trimethylindium and phosphine were supplied after the formation of the phosphorus layer at the interface. This structure corresponds to that the thickness of the first emitter layer is one molecular layer in Example 3.
Even if the InP layer using trimethyl phosphorus as a raw material is a single molecule, the hydrogen desorption effect due to the methyl group during interface formation worked, and an activation rate of 22% was obtained. This value is large compared to 7% according to the conventional method.

<Example 5>
A 200 nm InP collector layer, a 50 nm GaAsSb base layer, and a 100 nm InP emitter layer were epitaxially grown on an InP substrate at a growth temperature of 450 ° C. The 100 nm emitter layer was divided into a first emitter layer in contact with the base layer and a second emitter layer in contact with the first emitter layer, and growth was performed so that the total was 100 nm. Trimethyl phosphorus was used as the phosphorus material for the first emitter layer, and phosphine was used as the phosphorus material for the second emitter layer. Phosphine was used as a raw material for phosphorus in the InP collector layer.
In all the wafers, the carbon concentration of the base layer was 8.0 × 10 ¹⁹ cm ⁻³ . In the conventional method without the first emitter layer, the activation rate was 60%. The influence of hydrogen atoms is smaller in the case of using GaAsSb than in the case of using InGaAs as the base layer even by the conventional method.
The activation rate is 70%, 75%, 82%, 87%, 92%, 96% for wafers having a thickness of the first emitter layer of 0.5 nm, 1 nm, 2 nm, 5 nm, 10 nm, 20 nm, and 50 nm. 96%. Even if the InP layer grown using trimethyl phosphorus as a raw material is extremely thin, the effect of desorbing hydrogen atoms is high. If the thickness is 10 to 20 nm or more, re-diffusion of hydrogen atoms supplied from phosphine during the second emitter growth can be completely prevented.
An activation rate of 96% was also obtained when all the InP emitter layers were grown using trimethyl phosphorus as a raw material.
It was also possible to grow the InP collector layer using trimethyl phosphorus as a raw material. By using trimethyl phosphorus as the raw material for phosphorus, it is possible to avoid the use of phosphine, which is a specific high-pressure gas.
As shown in this embodiment, the present invention is also effective for an HBT epitaxial wafer using GaAsSb as a base layer.

<Example 6>
A 200 nm InGaAs collector layer, a 50 nm InGaAs base layer, and a 100 nm InP emitter layer were epitaxially grown on an InP substrate at a growth temperature of 450 ° C. Various organic phosphorus compounds were used as the raw material for phosphorus in the InP emitter layer. In all the wafers, the carbon concentration of the base layer was 7.0 × 10 ¹⁹ cm ⁻³ . When using phosphine, which is a conventional method, the activation rate was 7%.
When trimethyl phosphorus was used, an activation rate of 95% was obtained as shown in Example 1. The activation rate was 85% for dimethyl phosphorus and 65% for monomethyl phosphorus.
An organophosphorus compound having an ethyl group was also tried. The activation rate was 40% for triethyl phosphorus, 18% for diethyl phosphorus, and 12% for monoethyl phosphorus.
In addition, when tertiary butyl phosphine was used, there was no difference from the case where the conventional phosphine was used, and the activation rate was 7%.

<Example 7>
The carrier gas used in the growth of the InP emitter layer having the structure shown in Example 6 was replaced with hydrogen gas to form nitrogen gas.
Various organic phosphorus compounds were used as the raw material for phosphorus in the InP emitter layer. In all the wafers, the carbon concentration of the base layer was 7.0 × 10 ¹⁹ cm ⁻³ . When phosphine, which is a conventional method, was used, the activation rate was 9%, which was slightly improved from the case where hydrogen gas was used as a carrier.
When trimethyl phosphorus was used, an activation rate of 98% was obtained, and the hydrogen concentration was further reduced to half or less compared with the case where hydrogen carrier was used. The activation rate was 92% for dimethyl phosphorus and 80% for monomethyl phosphorus.
An organophosphorus compound having an ethyl group was also tried. The activation rate was 55% with triethyl phosphorus, 25% with diethyl phosphorus, and 18% with monoethyl phosphorus.
In addition, when tertiary butyl phosphine was used, there was no difference from the case where the conventional phosphine was used, and the activation rate was 9%.
Even when nitrogen gas was used as a carrier, the carbon concentration in the emitter layer was below the detection limit by secondary ion mass spectrometry in all the samples shown in this example.

<Example 8>
The carrier gas shown in Example 7 was a mixed gas of hydrogen and nitrogen, and the effect of the carrier gas was confirmed.
When the ratio of nitrogen is 0% (only hydrogen), 25%, 50%, 75%, and 100%, and the InP emitter layer having the structure described in Example 6 is grown using trimethyl phosphorus, the activation rate is They were 95%, 95%, 97%, 97%, and 98%, respectively.
When grown using monomethyl phosphorus, the activation rates were 65%, 68%, 75%, 77%, and 80%.
The activation rate can also be increased by using part of the carrier gas as nitrogen. Further, the higher the ratio of nitrogen in the carrier gas, the higher the activation rate can be obtained.

<Example 9>
In the method of manufacturing the interface between the base layer and the emitter layer shown in Example 4 with trimethyl phosphorus, nitrogen gas was used as a carrier gas only when trimethyl phosphorus was supplied. This increased the activation rate from 22% to 30%.

<Example 10>
In Example 1 to Example 9, the InP emitter layer was an InGaP emitter layer (the proportion of Ga in group III elements was 5%, 10%, and 20%). When the emitter layer is made of InGaP having a larger band gap than InP, it is expected that the band gap difference from the base layer is increased and the characteristics of the HBT are improved.
Since the base layer was the same as in Examples 1 to 9, there was no change in the carbon impurity concentration in the base layer. Also, the activation rate of carbon impurities was almost the same as in Examples 1 to 9.

<Example 11>
In Example 1 to Example 9, the InP emitter layer was an InAlP emitter layer (the proportion of Al in group III elements was 5%, 10%, and 20%). When the emitter layer is made of InA1P having a larger band gap than InP, it is expected that the band gap difference from the base layer is increased and the characteristics of the HBT are improved.
Since the base layer was the same as in Example 1 to Example 9, there was no change in the carbon impurity concentration in the base layer. Also, the activation rate of carbon impurities was almost the same as in Examples 1 to 9.

<Example 12>
In Examples 1 to 11, disilane was supplied during the growth of the collector layer and the emitter layer to obtain an n-type having a concentration of 5.0 × 10 ¹⁷ cm ⁻³ . There was no change in the carbon concentration in the base layer, and the activation rate of the carbon impurities was equivalent to that shown in Examples 1 to 11.
According to Embodiments 1 to 12 of the present invention, it is possible to provide an epitaxial wafer having a high activation rate of carbon added as a p-type impurity to the base layer. By using this epitaxial wafer, it is possible to produce a heterobipolar transistor having excellent operating characteristics.

The schematic of the cross section of the epitaxial wafer for hetero bipolar transistors illustrated by embodiment of this invention. 1 is a schematic cross-sectional view of a heterobipolar transistor using an InP substrate exemplified in an embodiment of the present invention. The figure which shows carbon and hydrogen concentration distribution of the epitaxial wafer for hetero bipolar transistors by this invention measured by the secondary ion mass spectrometry. The figure which shows carbon and hydrogen concentration distribution of the epitaxial wafer for hetero bipolar transistors by the conventional method measured by the secondary ion mass spectrometry.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Semi-insulating InP substrate 2 InP buffer layer 3 InGaAs collector layer 4 InGaAs base layer 5 InP emitter layer 6 InGaAs cap layer 7 Collector electrode 8 Base electrode 9 Emitter electrode

Claims (6)

  In an epitaxial wafer for a heterobipolar transistor using an InP substrate and including a collector layer, a base layer, and an emitter layer, the base layer is a p-type compound semiconductor containing carbon as an impurity, and the emitter layer is made of an organic phosphorus compound as a raw material An epitaxial wafer for a hetero-bipolar transistor, characterized in that it is InP, InGaP or InAlP grown as:   In an epitaxial wafer for heterobipolar transistors using an InP substrate and including a collector layer, a base layer, a first emitter layer in contact with the base layer, and a second emitter layer in contact with the first emitter layer, the base layer is made of carbon. A hetero-bipolar, characterized in that the first emitter layer is InP, InGaP or InAlP having a thickness of one or more molecular layers grown from an organic phosphorus compound as a raw material. Epitaxial wafer for transistors.   3. The heterobipolar transistor according to claim 1, wherein the organic phosphorus compound is grown using at least one of trimethyl phosphorus, dimethyl phosphorus, monomethyl phosphorus, triethyl phosphorus, diethyl phosphorus, and monoethyl phosphorus as the organic phosphorus compound. Epitaxial wafer.   4. When growing the epitaxial bipolar transistor epitaxial wafer according to claim 1, the organic phosphorus compound is supplied after the base layer is grown, and is supplied to an interface between the base layer and the emitter layer. A method for producing an epitaxial wafer for a heterobipolar transistor, comprising a step of producing an atomic layer of phosphorus.   4. When growing the epitaxial bipolar transistor wafer according to claim 1, nitrogen gas is used for a part or all of the carrier gas during the growth of the emitter layer. A method for producing an epitaxial wafer for a heterobipolar transistor, characterized in that:   6. The method of manufacturing an epitaxial wafer according to claim 4, wherein nitrogen is contained in a part or all of a carrier gas when forming an atomic layer of phosphorus at an interface between the base layer and the first emitter layer. An epitaxial wafer manufacturing method characterized by using a gas.

Images