JPH03201585A - Manufacture of optical semiconductor element and growing method for low resistance semiconductor layer - Google Patents

Abstract

PURPOSE:To separate a pn homojunction position from a board-regrowth boundary and to reduce an increase in a rising voltage or a reverse leakage current by automatically diffusing impurities of a high concentration p-type layer of a p-type InP clad layer during growing. CONSTITUTION:When Zn concentration is so controlled as to be gradually reduced continuously from saturated concentration in a thermal equilibrium as separated from an active layer by using dimethyl zinc to reduce the saturated concentration to at least 40% or more, its growth is started, for example, at an arbitrary temperature from 550 to 600 deg.C, and ended at an arbitrary temperature from 620 to 680 deg.C to be controlled in the concentration. Then, gas flow in a reaction tube may not be disordered. Accordingly, irregularities in the composition, concentration in a growing wafer, fluctuations can be reduced. That is, doping concentration is varied by not altering dopant gas flow but fixing to a predetermined value.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention relates to a method for manufacturing optical semiconductor devices such as index-guided semiconductor lasers or double heterojunction light-emitting diodes. In particular, the present invention relates to a method of manufacturing a semiconductor light emitting device in which an active layer is surrounded by a semiconductor layer having a wider forbidden band width than the active layer.

The present invention also relates to a semiconductor thin film growth method, particularly m-
The present invention relates to a method for vapor phase growth of a group V compound semiconductor layer.

(Prior Art) In recent years, optical semiconductor devices have been actively developed. In particular, optical semiconductor devices using Peter junctions that can effectively inject current into their active layers have been developed and manufactured. Compared to LEDs, research on semiconductor lasers that utilize stimulated emission process, which has excellent high-speed response, is more active. Stable continuous oscillation at room temperature of a semiconductor laser was achieved using a double heterojunction structure. Nowadays, it is no exaggeration to say that semiconductor lasers almost always have this double heterojunction structure.

It is important for the semiconductor light-emitting cable of this building to satisfy the following three conditions. First, in order to increase luminous efficiency, current is efficiently constricted and concentrated only in a light-emitting region controlled by an extremely narrow cap. Second, in order to reduce the contact resistance of the device, electrodes should be formed over a wide area and adjacent to each other. Alternatively, the resistance value can be lowered by increasing the carrier concentration of the contact layer. Third, when high-speed modulation is required, such as in a light-emitting element for optical communication, the area of the part where the p-n junction is formed should be minimized to reduce the junction capacitance. That's true.

Among semiconductor light emitting devices for optical communications, mesa lasers that utilize a self-alignment process are examples that relatively satisfy the above three conditions, and are applied to GaInPAs/InP lasers (e.g. .

Y, Hirayama et al.: 11th
Int, Sem1conductorLaser
Conf., D:1. Ho5ton (1988),
or IEEE J.

Quantuit Electron, QB-25
, 1320 (1989), etc.).

This semiconductor laser is called a SA-CM (self-aligned current confinement mesa) laser. Similarly, as an example that relatively satisfies the above three conditions, DC-FBI ((Double
eChannel Planar Buried He
tero) laser (for example, Mito et al., 1982 Institute of Electronics and Communication Engineers National Conference, Optical and Quantum Electronics A
Proceedings 857), this laser has two grooves formed in a double heterostructure wafer, and buried crystal growth is performed using a pn reverse junction in areas other than the active region sandwiched between them.

In a similar manner, the manufacturing method and features will be explained with reference to engineering drawings for areas other than the buried active region.

FIG. 5 is a cross-sectional view showing the manufacturing process of a conventional mesa type pn reverse junction laser. First, as shown in FIG. 5(a), an n-type (100) InP substrate (hereinafter abbreviated as substrate) 10
n-1nP buffer and cladding with a thickness of about 1.5/a on top of 0)l! ! ! 101.1.55-band non-doped GaInPAs active layer 102 with a thickness of 0.1-I and a p-InP cap layer 10 with a thickness of 0.215I, which enables light emission in the 55-band.
3 are sequentially grown as crystals. Thereafter, as shown in FIG. 5(b), B was etched using the Sin film 200 as an etching mask.
GaInPAs active layer 1 using a methanol-based solution
Two stripe grooves 2 with a radius of 2 pm and a spacing of 1.54 on 02
Form 06. At this time, since the interval between the two stripe grooves ultimately becomes the width of the active layer, in order to obtain stable fundamental transverse mode oscillation and a low oscillation threshold,
It is controlled during s.

Next, as shown in FIG. 5(e), a p-InP layer 104 and an n-InP layer 10 are formed in the two grooves 206 as current blocking layers.
5.

Crystals are selectively grown using the 5in2 film 200. After that, as shown in FIG. 5(d), only the SiO□ film 200 is etched, and then the p-InP layer 106 with a thickness of 2 is formed, and the p''-GaInPA with a composition of, for example, 1.3/a in terms of the emission wavelength is etched.
Contact q 107, film thickness 0.2. InP cap layer 108 is sequentially crystal-grown.

Next, as shown in FIG. 5(e), the cap layer 1 is
After peeling off 08, Au-ZnTIt electrode 2 with a width of 30I1m
02 is formed by a lift-off method, and then alloyed. Subsequently, etching is performed until the active layer 102 is exposed to form a mesa. At this time, p-InP layer 1.0
If HCI is used for etching of 6, the etching will stop precisely at active) eJ102 due to its selectivity.

Thereafter, only the active layer outside the two grooves is selectively removed using a sulfuric acid-based etchant (for example, sulfuric acid + hydrogen peroxide + water = 10:1:1). This solution has almost no effect on InP. Therefore, the lateral progress of etching is W
It automatically stops at the InP layers 104 and 105, which are current blocking layers, and a desired mesa shape can be obtained with extremely high reproducibility. In addition, the void 205 is formed by the above etching.
is formed.

Next, as shown in FIG. 5(f), 5 lo
After depositing the t polish 201, a window is opened at the top of the mesa, and an Au-Cr electrode 203 is deposited on the entire surface as a p-side electrode. In addition, after polishing the substrate 100 side to a thickness of about 100 mm, the Au-Gettt electrode 20 mm was used as the n(ll electrode).
4 total 204 completed. As a result, a buried mesa type pn reverse junction semiconductor laser is completed.

If a laser that oscillates in a single mode with a wavelength of 1.557a is required, the active layer 102 is
A two-layer structure is formed in which a GaInPAs optical guide layer having a wavelength, for example, a composition of 13-1 and lattice-matched to InP is grown on a GaInPAs active layer 102 having a composition in the 5-band range, as shown in FIG. 5(b). Before forming the stripe grooves, a diffraction grating with a period matching the desired wavelength is created on the top of the optical guide layer (not shown), and then by continuing the above-described creation method, a distributed feedback type (DFB
: Distributed Feed Back) A buried mesa type pn reverse junction semiconductor laser is completed.

However, in the manufacturing process of this type of buried mesa type pn reverse junction semiconductor laser, the width of the active layer is limited.

Although it has the advantage of being able to accurately control the width of the buried layer, it is inevitable that the growth will be interrupted twice due to the process. Therefore, the surface of the GaInPAs layer 102 and the buffer I
n-InP layer 101 which becomes an nP layer and ρ which becomes a current blocking elbow
-InP104 interface and p-InP cladding layer 1
06 and the ρ-InP layer 105, each of the outer layers 05 is once exposed to the atmosphere and then regrown. This interface is each G
It is also the heterojunction interface of aInPAs and InP, and the homopn junction interface of DynP, and the crystal surface is roughened by the introduction of unnecessary impurities due to exposure to -degree atmosphere and during waiting during the temperature rise during regrowth. This has been an obstacle to improving the electrical characteristics of the bond.

The junction of the embedded InP needs to be optimized from the viewpoint of reducing the junction capacitance and increasing the rising voltage of the junction to reduce current leakage to areas other than the active layer and increase output. Generally, as mentioned above, the substrate,
Or, at the interface between the surface exposed to the atmosphere after interruption of growth and the epitaxial growth layer thereon, contamination of crystals or disordered crystals may exist, and if a hetero interface is created, The electrical properties are inferior to those produced continuously without interrupting growth. For example, there have been problems such as the rise voltage not rising to a desired level and the amount of current leakage in the forward and reverse directions being large.

On the other hand, introducing a moderate p-type impurity into the non-doped GaInPAs active layer 102 has the advantage of speeding up the process. It has been difficult to introduce impurities into the active layer 102 because the p-type impurity diffuses into the active layer 102, penetrating the p-n junction or heterojunction interface, and preventing oscillation. As a measure to overcome these problems, p-I
nPt? When regrowing the X-flow blocking layer 104, p -1,n P cladding N106, 11 layers are provided at the bottom of it, and a part of the n-type cladding layer 10) is formed by diffusion during growth.
There is a technique in which Zn, which is a p-type impurity, is diffused to a part of the active layer 102.

When growing the P+ layer and the P layer continuously, it is necessary to change the flow rate of the doping gas significantly (for example, from point 304 to point 303 in Figure 1), and when the flow rate changes, the total gas flow rate to the reaction tube changes. This caused problems such as fluctuations in the composition of the semiconductor layer during growth. Moreover, the reproducibility was poor.

Hereinafter, a conventional method for growing a low-resistance semiconductor layer will be explained. In recent years, attention has been paid to semiconductors, especially vapor phase growth of ■-v group compound semiconductor layers. Expected as an optical communication device], , 55. Long wavelength semiconductor lasers and optoelectronic integrated circuits (optoelectronj, csIntc
GaInPAs/T occupies an important position in the manufacturing process of
Vapor phase epitaxy is preferred for growing nP-based thin films, and among them, metal organic chemical vapor deposition (Mstal Organic C
Chemical VaporDeposition:
The MOCV[l) method is noted in Note 14.

High performance GaInP with wide modulation frequency band and high output
As/InP double hetero (Double, e t(e
tcro: D) I) Structure In order to produce a laser diode, a highly doped p-type 1 with low resistance is used.
It is believed that an nP layer is required. In addition, a highly doped p-type semiconductor layer was essential as a means to reduce contact resistance.

However, the concentration of zinc, which is generally widely used as a p-type dopant, is saturated at a certain value when doping is performed by epitaxial vapor phase growth, and even if the flow rate of the zinc raw material is increased beyond that point, the hole concentration will decrease. It did not increase and was limited to the so-called saturation concentration.

Several studies have been conducted regarding zinc doping into MOCVD-grown InP.

Razeghi et al. (J, Crystal, Growth
64.76 (1983)), A, w, Ne1s
on et al. (J, Cryst, Gro%Ith 6g,
102 (1984)), R.5axena et al. (J.
Crystal, Growth, 77°591 (19
As seen in reports such as 86)), as the flow rate of dimethylzinc or diethylzinc increases, the zinc concentration increases proportionally until the zinc concentration incorporated into the InP crystal is below the level of I is the saturation concentration at so-called thermal equilibrium (approximately 3×to”(2)−3〉
It didn't get any better than that. - Figures 2 and 3 show the doping characteristics of zinc into InP using the intrashell low pressure MOCVD method.
Shown in the figure. It was investigated that in the unsaturated region, for a constant dimethylzinc or diethylzinc flow rate, the zinc concentration decreases with increasing temperature and decreases with increasing growth rate. The above-mentioned MOCVDF& was based on normal pressure and reduced pressure MOCvD methods, and the growth temperature range was between 550°C and 670°C, which is practical.

As described above, zinc, which is expected to be a p-type impurity in InP, cannot be doped into the crystal to a level higher than 3XIO"cm-', and the resulting hole concentration is also limited to below this value. Therefore, a P-type layer with lower resistance could not be produced using zinc as a dopant.

(Problems to be Solved by the Invention) As mentioned above, according to the conventional manufacturing of optical semiconductor devices, techniques for automatic diffusion doping into the active layer and separation of the homo pn junction interface and the regrowth interface are required. It was 1
It has been difficult to change the impurity concentration in a continuously growing crystal without changing the total gas flow rate to the reaction tube. Furthermore, there has been a desire for countermeasures against physical deterioration of the mesa structure itself, such as changes in shape during regrowth.

The first invention was made in consideration of the above circumstances, and its purpose is to minimize the surface deterioration of semiconductor layers exposed on the surface during growth, such as active layers and mesa structures. , and gradually change (especially decrease) the p-type impurity concentration without greatly disturbing the steady state of the gas flow rate, air flow, pressure, etc. flowing into one reaction tube including the dopant gas flow rate.
The objective is to form the cladding layer while At this time, near the active layer (at the start of regrowth), p-type impurities are automatically added to a part of the active layer and n-type rad m during subsequent growth.
By introducing impurities up to the thermal equilibrium saturation concentration of the p-type dopant in the p-type semiconductor layer, which is such that it diffuses into a portion of the p-type semiconductor layer.

An object of the present invention is to provide a method for manufacturing a semiconductor light emitting device that can perform high-speed modulation and has high output.

As in the example mentioned above, conventionally, there are impurities such as zinc in InP whose concentration saturates at a constant value with respect to the amount of impurity feed and the carrier concentration cannot be increased, and these impurities are used to develop devices. When it is desired to fabricate a device, there is a problem in that the carrier concentration of the device cannot be raised above that value. This was one of the factors that hindered the reduction of parasitic resistance and contact resistance.

The second invention was made in consideration of the above circumstances, and its purpose is to increase the impurity introduction efficiency in the semiconductor*S growth process and introduce a higher concentration of impurities into the crystal. The object of the present invention is to provide a technology for manufacturing a low-resistance semiconductor layer.

[Structure of the invention]

(Means for solving mliI) The first aspect of the present invention relates to a method for manufacturing an optical semiconductor device, in which a compound semiconductor layer is grown while being doped with an impurity having a higher vapor pressure than at least one constituent element of the compound semiconductor layer. When manufacturing an optical semiconductor element forming a semiconductor layer, the impurity concentration in the compound semiconductor layer is determined from the doping saturation concentration of the impurity in thermal equilibrium in the semiconductor layer to 40% of this saturation concentration.
%, the flow rate of the doping gas is set as a predetermined value, and the growth temperature is changed and controlled. The method is characterized in that the compound semiconductor layer is formed by increasing the impurity flow rate above a flow rate at which the doping rate increases as the doping rate increases. The second aspect of the present invention relates to a method for growing a resistive semiconductor layer, which includes growing a compound semiconductor layer while doping it with an impurity having a higher vapor pressure than at least one constituent element of the compound semiconductor layer. , the impurity concentration in the compound semiconductor layer is determined from the doping saturation concentration of the impurity in thermal equilibrium in the semiconductor layer to 4 of this saturation concentration.
Within the concentration range of 0%, the flow rate of the doping gas is set at a predetermined value, and the growth temperature is changed and controlled. Furthermore, the third aspect of the present invention is that when growing a low resistance compound semiconductor layer doped with an impurity having a higher vapor pressure than at least one constituent element of the compound semiconductor layer, the concentration of the impurity is lower than the concentration of the impurity in the semiconductor layer. For concentrations exceeding the saturation concentration at thermal equilibrium in the semiconductor layer, the impurity flow rate is increased beyond the flow rate at which the doping concentration increases as the growth rate of the semiconductor layer increases, and growth is performed at a growth rate of 7IIIa/hour or more. A method for growing a characteristically low-resistance semiconductor layer is provided.

(Function) The gist of the method for manufacturing an optical semiconductor device according to the present invention is that when forming a p-type rad layer or a current blocking layer, only the growth temperature is changed to change the impurity concentration without changing the gas flow rate. By increasing the distance from the active layer, the p-type impurity decreases from the doping saturation concentration during vapor phase growth of the cladding semiconductor layer to at least 4% of the saturation concentration.
By providing a structure that continuously gradually changes (particularly decreases) to a concentration higher than 0%, the buried homo pn junction interface is separated from the substrate-regrowth interface by automatic diffusion of the impurity, and also through the heterojunction interface. The purpose is to automatically diffuse impurities to a part of the active layer so as to have an arbitrary concentration distribution.

In addition, we fabricated the above structure by gradually increasing the growth temperature from the start of growth of the current blocking layer and cladding, and also minimized the deterioration of the shape of the Peter junction interface, pn junction interface, and mesa. be. in particular.

The amount is appropriately selected, and the growth start temperature and growth end temperature are selected without changing the pressure inside the reaction tube.

According to the present invention as described above, the p-n homojunction position can be separated from the substrate-regrowth interface by self-diffusion of impurities in the highly concentrated p-type layer of the p-InP cladding layer during growth, and the rise voltage can be reduced. This makes it possible to increase the current and decrease the reverse leakage current, improving the current confinement ratio. Furthermore, by appropriately selecting the growth temperature, p-type impurities can be introduced into the active region while suppressing deterioration of the shape of the mesa provided at the end of the active layer. Therefore, a semiconductor light emitting device capable of high-speed response can be obtained.

Using the low-pressure organic gold diversion vapor phase growth method has good reproducibility in practice, and if dimethylzinc is used as the raw material for Zn, which is a p-type impurity, the flow rate can be easily controlled to an arbitrary constant value. Yes, and the reproducibility is good. Furthermore, dimethylzinc was used to increase the concentration of Zn as it moves away from the active layer.
When controlling the concentration to be at least 40% of the saturation concentration by continuously gradually decreasing from the saturation concentration at thermal equilibrium, for example, the growth is controlled from 550 °C to 60 °C.
Starting from any temperature between 0°C and 620°C to 6
If the concentration is controlled by ending the reaction at an arbitrary temperature between 80°C, there is no need to disturb the gas flow in the reaction tube (for example, points 305 to 303 in Fig. 1).
It is possible to reduce variations in composition, concentration, etc., fluctuations, etc. within a wafer during growth.

In contrast to the conventional method, which required large changes in the dopant gas flow rate to change the doping concentration, this method
This method changes the doping concentration without fixing the dopant gas flow rate to a certain constant value.

That is, above a certain dimethylzinc flow rate, the concentration of Zn incorporated into the crystal is saturated almost regardless of the growth temperature, so any dimethylzinc flow rate range close to the saturation region (for example, region 301 in FIG. 1) One point is to take advantage of the effect that the doping concentration can be controlled to a concentration suitable for application to optical semiconductor devices only by changing the growth temperature.

Next, the gist of the method for growing a low-resistance semiconductor layer according to the present invention is to add impurities to the semiconductor thin film that have a higher vapor pressure than the constituent elements at the positions to be replaced, especially in the saturated region (depending on the supply amount of the impurity raw material). On the other hand, when growing while introducing a material that has a region in which the amount of impurities incorporated into the crystal hardly changes, it is possible to increase the crystal growth rate while flowing the impurity flow rate higher than the flow rate that saturates the impurity. The purpose of this is to increase the amount of impurities incorporated into the crystal and the carrier concentration. With this, a low resistance semiconductor layer can be formed. Furthermore, by changing only the growth conditions that determine the growth rate, the carrier concentration in the semiconductor layer can be made high.

As described above, according to the present invention, it is possible to form a semiconductor layer with a high concentration, which could not be achieved under conventional growth conditions, so that contact resistance can be reduced by using it as a contact layer when an element is manufactured. Further, by using this high concentration layer for the cladding, the output of the semiconductor laser can be significantly improved. Also, for example, MOCV is used as a crystal growth method.
Since the film thickness can be easily controlled using the D method, it has the advantage that it can also be used as a thin film for a diffusion source with a constant total amount of impurities.

(Example) Hereinafter, details of the present invention will be explained with reference to illustrated embodiments.

FIG. 4 is a sectional view showing the manufacturing process of a mesa type pn reverse junction laser showing the first embodiment of the present invention. First, Figure 4 (
As shown in a), n-type (n = IXIO"Re-3)
(001) InP substrate 100 with a thickness of about 1.5. n of
-InP buffer/cladding layer 11 (n: I
O"al-3), non-doped Ga1nPAs active layer 12 with a composition thickness of 0.1- that enables emission in the 1.55. band
, thickness 0.2. The p(nP cap layer 13 is successively crystal-grown. Then, as shown in FIG. 4(b), 5un2
Using the film 20 as an etching mask, a Br methanol system (for example, I(Or:Br:H,0=17:1
: 100) using a solution of GaInPAs active layer 12
width 24, spacing 1.5. Two stripe grooves 26 are formed. At this time, since the distance between the two stripe grooves ultimately becomes the width of the active layer, it is controlled between 1 and 1.57 m (73 m) in order to obtain stable fundamental transverse mode oscillation and a low oscillation threshold.

Next, as shown in FIG. 4(c), p-InPJil 14 (for example, (p=1
x10"an-'), n-InP layer 15 (e.g. n =
Crystals of Ix10"cxa-") are selectively grown on a 20-layer SiO□ film. After that, Fig. 4(d)
As shown in the figure, after etching only the film 20, the film thickness is 2.
1 m of p-InPn soil layer, the emission wavelength is e.g. 1.3
p"-GaInPAs contact layer 1') with a composition of μ
p>10L! Iex-3>, an InP cap layer 18 having a thickness of 0.2 μm is successively grown.

Next, as shown in FIG. 4(e), after peeling off the cap layer 18, an Au-Zn electrode 22 having a width of 304 mm is formed by a lift-off method, and then alloying is performed. Perform etching to form a mesa. At this time, if HCI is used for etching the p-InPn soil layer, the etching will stop precisely at the active layer 12 due to its selectivity. After that, with a sulfuric acid-based etchant (for example, sulfuric acid + hydrogen peroxide + water = 10: 1: 1),
Only the active layer outside the two grooves is selectively removed. This solution has little effect on InP, so the lateral progress of etching is limited to InPM114.1, which is the current blocking layer.
It automatically stops at point 5, and the desired mesa shape can be obtained with extremely good reproducibility.

Next, as shown in FIG. 4(f), a Sin
After depositing the layer 21, a window is opened at the top of the mesa, and an Au--Cr electrode 23 is deposited on the entire surface as a p-side electrode.

Further, after polishing the substrate 100 side to a thickness of approximately 100 μm, an Au-Ge electrode 24 is deposited as an n-side electrode. As a result, a buried mesa type pn reverse junction semiconductor laser is completed.

If a laser with single mode oscillation at a wavelength of 1.554 is required, the active layer 12 may be oscillated at a wavelength of 1.55μ.
On top of the GaInPAs active layer 12 having a composition of
For example 1.3. G that has a composition of and is lattice matched to InP
A two-layer structure is formed by growing an aInPAs light guide layer, and a diffraction grating with a period matching the desired wavelength is created on the top of the light guide layer before forming the stripe grooves shown in FIG. 4(b). (Illustration is omitted). After that, by continuing the above-described production method, the distributed feedback buried mesa type p
An n-reverse junction semiconductor laser is completed.

Between (b) and (c) in FIG. 4, and between (c) and (d), the P-InP current blocking layer 14 and the p-InP cladding layer 16 are formed as the first An example of the doping characteristics of Zn when dimethylzinc is used as a Zn impurity raw material in the 1M0CVD method is shown in FIGS. 1 to 3. In FIG. 3, trimethylindium (TMIn) and phosphine (PH3) are used for MOCVD growth of InP, and dimethylzinc is used for the dopant concentration.

The flow rates of TMIn and PH3 are 1. I x 10
-'mol/win and 4,2 X 10"" m
This is a typical value of 1/1 inch, and each raw material is diluted with an inert gas such as hydrogen or nitrogen, and introduced into a reaction tube controlled at a reduced pressure of 200 Torr at a total flow rate of 1012/1 inch.

From FIG. 1, in order to change the dopant concentration from a constant dopant concentration to another lower concentration in region 302, it is necessary to greatly change the dopant gas flow rate when the growth temperature is kept constant; If the flow rate of the growth temperature was not changed, it was necessary to change the growth temperature significantly. However, in region 301, there is an advantage that in order to change the dopant concentration from a constant dopant concentration to another lower concentration, the growth temperature only needs to be changed slightly without changing the dopant gas flow rate.

(For example, from point 305 to point 303 in FIG. 1) Utilizing this region 301, when growing the current blocking layer 14 and the p-InP cladding layer 16, the concentration of Zn can be continuously changed by changing only the growth temperature. By growing while changing the impurity concentration, it is possible to grow a current blocking layer and a cladding layer with a continuous impurity concentration distribution. Growth temperature 5
Since the growth rate does not change between 50° C. and 700° C. unless the flow rate of TMIn is changed, it is easy to gradually change the Zn concentration. It is also known that, for example, when Zn in InP exceeds 1.8×10"Ql-", the diffusion into the n-type layer becomes faster.

Semiconductor lasers generally require a p-type cladding with an impurity concentration of 10"Ql""3 or more, so for example, in the vicinity of the active layer, the concentration of Zn in InP at a thermal equilibrium of 2.3 If you want to use the saturation concentration to appropriately control automatic diffusion into the active layer during subsequent growth, and want the flat part of the Zn concentration to be lXl0"an-', dimethylzinc should be used at, for example, 9X10-7 mol/
By starting growth at 550°C while supplying n+in to the reaction tube, and gradually increasing the growth temperature to 670°C, and then fixing the growth temperature there, it is possible to obtain the desired impurity concentration distribution. For example, a structure such as ρ+P can be created without greatly changing the dopant gas flow rate.

In the above example, it was considered that the supply amount of dimethylzinc could be changed analogously, but the flow rate of dimethylzinc could be changed from, for example, 5,6XIQ-'mol/min to 1.45XIQ-'mol/min.
Even if it is only possible to select digitally between X 10-' mol/win, it can be adapted by selecting the temperature at the start of growth, its heating rate, and final temperature to desired values.

Particularly in the case of InP-based materials, as shown in FIG.

The concentration of Zn in InP is almost independent of the growth temperature, and is saturated at about 3XIO"al-" above a certain dimethylzinc flow rate. Furthermore, the lower the growth temperature, the higher the concentration of introduced impurities for the same dimethyl zinc flow rate. Therefore, by temperature manipulation as in the above example, the saturation concentration of about 3
It has the advantage of being able to control the desired concentration from 10" am-3 to I x 10"as-".

However, this method can also be used with materials other than InP and dopants other than Zn. For example, the present invention can be applied to a case where the active layer is made of GaInP or GaInAIP and the cladding layer is made of GaInAIP or GaAs. The same applies to a laser having a cladding layer made of GaAlAs and an active layer made of GaAs or GaAlAs.

FIG. 4 shows an example of a laser in which current confinement is achieved by a pn reverse junction of current blocking layers 14 and 15.
The cladding is of the pn forward junction type, and can be applied to optical semiconductor devices such as lasers that perform current confinement based on the difference in rising voltage from the Peter junction.For example, the current blocking layer 14 in the example of FIG. 15 is.

Even in the case of p-InP being formed only, by starting the growth temperature at a low temperature (for example, 550°C) at the start of regrowth, 1, for example, about 3×10"01-
At the saturation concentration of Zn in InP at thermal equilibrium of
"an-3. In this case.

Zn moderately diffuses into the active layer from the regrowth interface of the heterojunction, increasing speed, and separating the pn homojunction interface from the regrowth interface, reducing leakage current and strengthening the current confinement effect, increasing optical output. There are advantages such as an increase in

Furthermore, the present invention can be applied to the electronic device manufacturing process, and H
It can also be applied to devices using heterojunctions such as EMT (high electron mobility transistor) and HBT (hetero bipolar transistor). Also.

It is of course applicable to homozygous devices.

In the above example, Zn was particularly used as a dopant for the P-type semiconductor layer, but it goes without saying that the dopant for the N-type semiconductor layer may be an impurity that causes concentration saturation in thermal equilibrium, such as atoms that are not related to the conductivity type, such as Er. It can also be applied to rare earth elements such as , Eu, and S.

The details of the method for growing a low-resistance semiconductor layer according to the present invention will be explained below with reference to the illustrated embodiments.

FIG. 3 shows the growth rate dependence of the hole concentration in zinc-doped InP grown by MOCVD, showing the first embodiment of the present invention. The growth temperature is 620°C.

Figure 3(a) shows the hole concentration when the growth rate is changed in the saturation region 401 in Figure 2, and Figure 3(b)
shows the hole concentration when the growth rate is changed in the non-saturated region 402 in FIG.

In the non-saturated region 402, the incorporation of zinc, which belongs to the group II, into the crystal is due to competition with In, which belongs to the group II, and the growth rate is determined in proportion to the flow rate of In under this growth condition. Therefore, under a constant dimethylzinc flow rate, the hole concentration decreases as the growth rate increases (see FIG. 3(b)). However, the saturated region 401
As a result of extensive research, it has become clear that, on the contrary, the hole concentration increases as the growth rate increases (see Figure 3 (a)).
0. 4.5X10"C!I-'. This value is the highest hole concentration ever achieved by zinc in InP. By further increasing the growth rate, it is possible to further increase the hole concentration. If you consider using it as a contact layer, the difficulty in controlling the film thickness that accompanies an increase in the growth rate can be avoided, so the highly doped p-type layer obtained in this way can be used as an effective low-concentration layer. Can be used as a resistive semiconductor layer.Contact resistance is determined by an impurity concentration of 3 x 10"
It was reduced to one tenth or less compared to the case of Ql-3.

Here, the flow rate of dimethylzinc in the saturated region is
In the figure, any flow rate may be used as long as the flow rate is higher than the flow rate at which the concentration of Zn is saturated. Also,
The growth temperature may be any temperature as long as it is within the temperature range where InP can grow.

In the above example, a reduced pressure organometallic chemical vapor deposition method with a reaction tube pressure of 2.7 x LO'Pa is used, and the PH3 flow rate is 4.
Although 2×10 3 mol/min is adopted, this value may be any other value. Further, although zinc was used in the above example, it is natural that other P-type dopants with high vapor pressure such as Cd and Be may be used. Further, any dopant having a doping saturation region like this dopant can be applied. In the above example, especially p
Although Zn is used as a dopant for the n-type semiconductor layer, the dopant for the n-type semiconductor layer can naturally be an impurity that causes concentration saturation, such as atoms unrelated to the conductivity type, such as Er,
It can also be applied to rare earth elements such as Eu and 5I11.

Furthermore, the present invention can also be applied to the process of creating electronic devices.
HEMT (high electron mobility transistor) and IIBT (
It can also be applied to the growth of devices using heterojunctions, such as heterojunctions (hetero-bipolar transistors).

Moreover, it is of course applicable to homojunction devices.

Further, the present invention can be applied not only to m-■ group compound semiconductors but also to U-Vt group compound semiconductors having high vapor pressure by increasing the growth rate.

〔Effect of the invention〕

As described above, according to the method for manufacturing an optical semiconductor device of the present invention, the p
It is possible to increase the rise voltage of the n homojunction and reduce the leakage current. In addition, a diffraction grating provided above or below the active layer by appropriately selecting the Fil length,
Pu impurities can be introduced into the active region while suppressing deterioration of the shape of the mesa structure and the like. Therefore, a semiconductor light emitting device capable of high-speed response and oscillation can be obtained. Furthermore, according to the present invention, once growth has started, even if a continuous concentration gradient rather than a stepwise one is desired, only the temperature needs to be controlled. The above structure can be produced without changing the gas flow rate sent to the tube.

Next, according to the method of growing a low-resistance semiconductor layer according to the present invention, it is possible to form a semiconductor layer with a high concentration, which could not be achieved under normal growth conditions, so that it can be used as a contact layer when manufacturing an element, thereby increasing the contact resistance. can be reduced. For example, contact resistance was reduced by more than two orders of magnitude. Also,
By using this high concentration layer as a cladding layer, the output of the semiconductor laser can be significantly improved. Furthermore, if the MOCVD method is used as the crystal growth method, for example, the film thickness can be easily controlled, so there is an advantage that it can be used as a thin film for a diffusion source with a constant total amount of impurities.

[Brief explanation of drawings]

Figures 1 to 3 show M used in the present invention.
A diagram showing an example of the doping characteristics of Zn into InP by the OCVD method, FIG. 4 is a cross-sectional view showing the manufacturing process of an optical semiconductor device according to an embodiment of the present invention in the order of steps, and FIG. 5 is a diagram showing the conventional method. All of these are cross-sectional views showing the manufacturing process of the optical semiconductor device in the order of steps. 100-n-InP substrate. 11... n-InP buffer (cladding) layer, 12-
...Non-doped GaInPAs active layer. 13...p-InP cap layer, 14-p-InP current blocking layer, 15-n-InPtt flow blocking layer, 16-p-InP cladding layer, 17-p"GaInPAs contact layer, 18-・・p
-InP cap layer, 26...stripe groove. 22-Au-Zn electrode, 23-Au-Cr electrode, 24
-Au-Ge electrode, 20.2l-=Si02 film, 301
...The concentration of Zn taken into the crystal for a constant flow rate of dimethylzinc does not change much even if the growth temperature is changed, and furthermore, the concentration of Zn taken into the crystal becomes saturated as the flow rate of dimethylzinc increases. The previous area. 302...The concentration of Zn taken into the crystal changes significantly when the growth temperature is changed for a constant dimethylzinc flow rate, and the Znlll concentration in the crystal is not yet saturated as the dimethylzinc flow rate increases. In the region, 303-=620°C, the Zn concentration is l x 10''
Points indicating the dimethylzinc flow rate at which the Zn concentration becomes 2,3X10"Ql-", which is the saturation concentration in this example, at 304...620°C, 305... 303 point and a point indicating the growth temperature at which a saturation concentration can be obtained without changing the dimethyl zinc flow rate, 401...A region where the hole concentration is saturated with an increase in the dimethyl zinc flow rate.402...Dimethyl zinc A region where the hole concentration increases with respect to the flow rate.

Claims (2)

[Claims] (1) When manufacturing an optical semiconductor device in which a compound semiconductor layer is grown by doping with an impurity having a higher vapor pressure than at least one constituent element of the compound semiconductor layer, the impurity concentration in the compound semiconductor layer is , within a concentration range from the doping saturation concentration of the impurity at thermal equilibrium of the semiconductor layer to 40% of this saturation concentration, the flow rate of the doping gas is set to a predetermined value and the growth temperature is varied and controlled. For a concentration exceeding the doping saturation concentration in thermal equilibrium of the layer, a compound semiconductor layer is formed by increasing the impurity flow rate beyond the flow rate at which the doping rate increases as the growth rate of the semiconductor layer increases. A method for manufacturing semiconductor devices. (2) When growing a low-resistance compound semiconductor layer doped with an impurity having a higher vapor pressure than at least one constituent element of the compound semiconductor layer, the concentration of the impurity is the saturated concentration at thermal equilibrium of the semiconductor layer. For concentrations exceeding , increasing the impurity flow rate above the flow rate at which the doping concentration increases as the growth rate of the semiconductor layer increases,
A method for growing a low-resistance semiconductor layer, characterized in that growth is performed at a growth rate of 7 μm or more per hour.