JP2003007998A - Low resistance nitride semiconductor and method for manufacturing the same - Google Patents

Abstract

(57) Abstract: Provided is a p-type nitride semiconductor structure having a high hole concentration and a low resistance, and a method for manufacturing the same. SOLUTION: A semiconductor 1 composed of a nitride semiconductor doped with a p-type impurity and having a lattice constant a1 and a semiconductor 2 having a lattice constant a2 different from a1 are joined in a state containing lattice strain, thereby forming a nitride semiconductor. One dopant is activated by 5% or more. 600 ° C in the joined state
The activation heat treatment is performed at the above temperature.

Description

Detailed Description of the Invention

[0001]

[Industrial application] Green based on nitride semiconductors
When manufacturing an ultraviolet light emitting device, a high-power electronic device, or the like, it is necessary to increase the carrier concentration of the p-type nitride semiconductor and reduce the resistance in order to improve device characteristics. In particular, increasing the concentration of a semiconductor having a wide band gap is effective for confining light and electrons. The present invention relates to a structure and a method for manufacturing such a high-concentration p-type layer of a nitride semiconductor.

[0002]

2. Description of the Related Art Researches on fabrication of high-concentration p-type nitride semiconductors have been actively conducted. This is because the p-type nitride semiconductor layer has a high resistance and thus generates heat, and because the carrier concentration is low, it is difficult to form a good ohmic junction and the threshold value of the device increases, which adversely affects the device characteristics. is there. Heretofore, Mg atoms have been used as acceptors in the production of p-type nitride semiconductors. The following methods have been reported for activating the acceptor.

It is excited by electron beam excitation.

A high frequency is applied.

Heat treatment is performed in a nitrogen atmosphere.

[0006] Any of the above methods is a method of terminating the bond of Mg-H and activating the Mg acceptor. At present, it is the most convenient method, but it is the mainstream.

Further, a method of using a metal catalyst layer on the semiconductor surface for activation of the Mg acceptor has been reported (Waki, Fujioka, Ojima, Proceedings of the 48th Joint Spring Conference on Applied Physics, Spring 2001). p.416-31p-K-
3). In this method, Ni is thinly vacuum-deposited on the GaN surface doped with Mg, and then heat treatment is performed.
In the conventional heat treatment, the Mg acceptor was not activated unless the temperature was about 600 ° C. or higher, but in the method using the thin Ni catalyst layer, the Mg acceptor was activated even at a low temperature of 200 ° C. It is considered that this is because the thin Ni acts as a metal catalyst layer and promotes desorption of hydrogen on the GaN surface, so that the Mg acceptor is activated even at a low temperature. However, even when the Ni catalyst layer is used, the obtained hole concentration is the same as when the catalyst layer is not used.

On the other hand, as a method for producing a high-concentration p-type layer, there are the following reports.

A nitride semiconductor having a narrow band gap (having a large dielectric constant) is used.

A method using a co-doping method. That is, it is a method of simultaneously doping an n-type dopant and a p-type dopant (n-type dopant <P-type dopant) to form and stabilize a donor-acceptor pair, and to stabilize excess acceptors to a high concentration. .

In the method (1), InGaN having a narrower band gap than GaN is generally used. InGaN is Ga
Since the dielectric constant is larger than N, the binding energy of holes calculated from the hydrogen atom model becomes small. Therefore, the activation energy of the acceptor is higher than that of GaN by I.
It is expected that nGaN will be smaller. Experimentally, the activation energy of the acceptor obtained from the temperature characteristics of the hole concentration of InGaN is smaller than that of GaN, and a hole concentration exceeding 10 ¹⁸ cm ⁻³ has been realized at room temperature.
(K.Kumakura, T.Makimoto and N.Kobayashi, Jpn.App1.Ph
ys-39 (2000) L337.).

However, since a semiconductor having a narrow band gap is used in this method, a structure for confining light or electrons in the semiconductor, for example, an optical confinement layer of a semiconductor laser,
It cannot be used as a barrier layer for confining electrons and holes in electronic devices.

In the above method, in order to form a donor-acceptor pair, not only p-type impurities but also n-type impurities of substantially the same amount are doped into the semiconductor. For this reason,
In order to produce a high-concentration p-type layer
It is necessary to dope more than twice as many impurities. As a result, the flatness of the growth surface deteriorates, and it cannot be used near the active layer of the device.

On the other hand, it has also been reported that the structure in which a thin Mg-doped InGaN layer is grown on Mg-doped GaN reduces the contact resistance between the electrode and the semiconductor (Kumakura, Makimoto, Kobayashi, Spring 2001, No. 48). Proceedings of the Annual Joint Lecture on Applied Physics p.415-31a-K-
11). Generally, the contact resistance decreases as the carrier concentration of the semiconductor increases. In this system, thin strain InGa
The band structure sharply changes on the semiconductor surface due to the polarization electric field generated in the N layer. As a result, the width of the barrier generated at the metal-semiconductor interface becomes thin, and the current easily flows. Therefore, the contact resistance at the metal-semiconductor interface is reduced. That is, in this method, the band structure on the semiconductor surface determines the resistance regardless of the decrease in the resistance of the Mg-doped GaN itself.

[0015]

SUMMARY OF THE INVENTION An object of the invention is to solve the problem that a high hole concentration cannot be obtained at room temperature in a nitride semiconductor because the activation energy of Mg acceptor is large (170 meV in GaN). , With high hole concentration,
A p-type low resistance nitride semiconductor and a method for manufacturing the same are provided.

[0016]

The low resistance nitride semiconductor of the present invention is composed of a nitride semiconductor doped with a p-type impurity and has a lattice constant a1 and a semiconductor having a lattice constant a2 different from a1. And 2 are bonded together in a state including lattice strain, and the dopant of the semiconductor 1 is activated by 5% or more.

According to the method for producing a low-resistance nitride semiconductor of the present invention, a semiconductor 2 having a lattice constant a2 different from a1 contains a lattice strain on a semiconductor 1 having a lattice constant a1 doped with a p-type impurity. And then heat treatment for activating the dopant is performed.

[0018]

[Function] A thin Mg film is formed on the surface of GaN on which Mg is dubbed.
When InGaN doped with is grown, InGaN maintains a strained state without lattice relaxation when the film thickness is thin.
Due to the effect of strain on this surface, Mg-doped G
Mg in aN does not enter an inappropriate position such as between lattices, and lattice defects are reduced. Therefore, the concentration of Mg atoms acting as an acceptor increases, and the hole concentration increases.

It differs from the prior art in that a thin semiconductor lattice strain layer is present on the semiconductor surface. In the conventional technique, the activation rate is at most about 2% even if the activation heat treatment is performed. Therefore, in order to achieve a high concentration of hole density, the doping amount of the initial impurity is increased. There must be. However, when the doping amount is large, impurities enter between the lattices, which causes deterioration of the quality of the semiconductor layer.

In the present invention, since the activation rate is as high as 5% or more, a high hole density can be achieved with a small doping amount (thus, without causing deterioration of crystallinity).

[0021]

BEST MODE FOR CARRYING OUT THE INVENTION In the present invention, the lattice constant a1
And the semiconductor 2 having a lattice constant a2 different from that of a1 are bonded together in a state including lattice strain.

The type of substrate on which the semiconductor 1 is formed is not particularly limited. For example, a sapphire substrate is preferably used.
A buffer layer and an undoped layer are preferably interposed between the substrate and the semiconductor 1.

Although the semiconductor 1 is a nitride semiconductor, for example, Al _a In _b Ga _c N _d (0 ≦ a, b, c, d ≦
1) and B _x Al _y N _z (0 ≦ x, y, z ≦ 1) and other nitride semiconductors are used.

The method of forming the semiconductor 1 is not particularly limited, but for example, a metal organic chemical vapor deposition method is preferable.

The semiconductor 1 is doped with p-type impurities.

The impurities to be doped are, for example, M
g, C, etc. can be raised. In particular, Mg is preferable because it is easily activated and achieves a high hole density with low concentration doping.

In the present invention, since the activation efficiency is extremely high, it is not necessary to dope impurities excessively, and therefore a semiconductor having a high hole concentration while maintaining good crystallinity can be manufactured. .

In the present invention, the semiconductor 2 having the lattice constant a2 different from the lattice constant a1 of the semiconductor 1 is bonded to the doped semiconductor 1 on the doped semiconductor 1 in a state including the lattice strain.

The "state including lattice strain" also includes a partial lattice relaxation state. Whether or not the semiconductor 2 is in a state including lattice strain depends on the composition / composition ratio between the semiconductor 1 and the semiconductor 2 and the film thickness of the semiconductor 2.

For example, FIG. 3 shows the case where InGaN is joined to GaN, FIG. 5 shows the case where GaN is joined to AlGaN, and FIG. 7 shows the case where InGaN is joined to AlGaN. It can be seen that it changes depending on the composition and composition ratio with respect to No. Even when other than these are used as the semiconductor 1 and the semiconductor 2, it is sufficient to obtain a diagram of the strain state due to the composition / composition ratio and the film thickness by an experiment in advance and realize the state including the lattice strain based on the figure. .

The "strained" regions in FIGS. 3, 5 and 7 are regions where the difference in lattice constant between the semiconductor 1 and the semiconductor 2 having different lattice constants after bonding is about 0. The “partial lattice relaxation” region is a region where the difference in lattice constant is −2% to + 2%.

After the semiconductor 2 is bonded in a state including the lattice strain, activation heat treatment is performed. A temperature of 600 ° C. or higher is sufficient as the temperature of the activation heat treatment. That is, the dopant is activated by 5% or more by heat treatment at 600 ° C. or more. However, it is preferably 950 ° C or lower. 950 ° C
If it exceeds, the surface of the semiconductor may be roughened.

The heat treatment time is preferably 1 to 30 minutes, more preferably 5 to 15 minutes. If the heat treatment time is too short, activation may not be performed sufficiently, and if it is too long, the effect will be saturated.

In the semiconductor 1 or the semiconductor 2, In
When GaN and AlGaN are used, the composition ratio of In and Al is preferably 0.1 to 0.2. In order to control the ratio of In and Al, the amount of In and Al in the raw material may be controlled or the growth temperature may be controlled during the metal organic chemical vapor deposition.

The film thickness of the semiconductor 2 is, for example, as shown in FIG.
From the relationship with the composition ratio shown in Nos. 5 and 7, it may be set so as to enter the region having the strain. However, if the film thickness of the semiconductor 2 is too thin, the semiconductor 2 may disappear during the heat treatment. Therefore, the thickness is preferably a monoatomic layer or more.

It is also preferable that the semiconductor 2 be doped with impurities. By doping the semiconductor 2 with impurities, ohmic electrical contact can be established. The doping amount is 10 ¹⁹ to 10 ²⁰
cm ⁻¹ is preferred.

[0037]

[Example] (Example 1) Using the metal organic chemical vapor deposition method,
After growing the undoped GaN layer on the sapphire substrate,
A Mg-doped GaN layer was grown. The doping concentration of Mg at this time is 2 × 10 ¹⁹ cm ⁻³ . Thereafter, the film thickness is changed to grow Mg-doped InGaN. In
The In composition of GaN is 0.14. FIG. 1 shows the layer structure when an Mg-doped GaN layer and an InGaN layer are grown on a sapphire substrate.

The layer structure up to the Mg-doped GaN layer and the growth conditions differ depending on the type of substrate, but the growth conditions for the Mg-doped layer do not depend on the type of substrate. Therefore, the characteristics of the Mg-doped GaN layer also do not depend on the substrate.

After the growth, in order to activate the Mg acceptor,
Heat treatment was performed at 700 ° C. in a nitrogen atmosphere. The heat treatment time was 10 minutes. The heat treatment of the nitride semiconductor is 6
It is sufficient to carry out at a temperature of 00 ° C. or higher. In order to eliminate the influence of electric conduction of the Mg-doped InGaN layer in which strain is induced, InG is formed by ECR plasma etching after heat treatment.
The aN layer is removed.

To obtain electrical contact to the semiconductor, P
A d / Au electrode was vapor-deposited and the electrical characteristics were evaluated. Figure 2
In addition, when the thickness of the Mg-doped InGaN layer is changed,
4 shows a change in resistivity in a doped GaN layer. The usual method is when the thickness of the InGaN layer is zero. Mg-doped G
The hole concentration in aN is 8 × 10 ¹⁷ cm ⁻³ , but I
When the nGaN layer was grown to 2 nm or 5 nm, the hole concentration increased to 2.4 × 10 ¹⁸ cm ^−3, which is three times the normal concentration.

Further, the mobility is similarly 2.2 cm ² / V
It increased from −s to 2.5 cm ² / V−s. Since the resistivity is proportional to the reciprocal of [hole concentration × mobility], the strain I
Due to the presence of the nGaN layer, the resistivity of the GaN layer is 1/3
It became the following. On the other hand, when the thickness of the InGaN layer is increased, the hole concentration decreases, which is the same as the case without the InGaN layer.

FIG. 3 shows the calculation results of the strain / lattice relaxation state of the InGaN layer when the In composition and the film thickness in the InGaN layer are changed. When the In composition is 0.14 and the film thickness is 2 nm, the InGaN layer is in a distorted state.

When the film thickness exceeds 5 nm, the InGaN layer begins to partially relax the lattice, and when the film thickness is further increased to 15 nm, the InGaN layer is completely relaxed. Therefore, it can be seen that the dependency of the hole concentration in the Mg-doped GaN on the thickness of the InGaN layer is sensitive to the lattice strain on the surface during the heat treatment. It is considered that the presence of the strained layer on the surface increases the Mg acceptor concentration itself.

As described above, in order to increase the hole concentration in Mg-doped GaN, the InGaN layer in the strain region or the partial lattice relaxation region is shown in the strain / lattice relaxation diagram depending on the In composition and film thickness shown in FIG. Is grown on the Mg-doped GaN layer and the Mg acceptor is activated by heat treatment.

In this embodiment, the initial doping amount is
It is 2 × 10 ¹⁹ cm ⁻³ , and the hole concentration after heat treatment is 2.4 × 10 ¹⁸ cm ⁻³ . Therefore, the activation rate of Mg is {(2.4 × 10 ¹⁸ cm ⁻³ ) / (2 × 10 ¹⁹ cm
^-3 )} × 100 (%) = 12%, which is an extremely high value.

(Embodiment 2) As easily inferred from Embodiment 1, Mg-doped InGaN on Mg-doped GaN.
Instead of Mg-doped GaN on Mg-doped AlGaN
However, the same effect can be obtained. A structural diagram is shown in FIG. A1
FIG. 5 shows the strain / lattice relaxation state of the GaN layer in the case of GaN on GaN. As shown in the A1 composition and GaN film thickness of FIG. 5, it is desirable to grow a strained GaN layer on the Mg-doped AlGaN layer and activate the Mg acceptor by heat treatment.

Also in this example, 5% or more of Mg was activated.

(Embodiment 3) As is easily inferred from Embodiment 1, Mg-doped InGaN on Mg-doped GaN.
Instead of Mg, Mg-doped InG on Mg-doped AlGaN
The same effect can be obtained with aN. The structural drawing is shown in FIG.
FIG. 7 shows the strain / lattice relaxation state of the InGaN layer in the case of InGaN on A1GaN. A1 composition and In of FIG.
It is desirable to grow a strained InGaN layer on the Mg-doped AlGaN layer as shown in GaN film thickness and activate the Mg acceptor by heat treatment. Also in this example, 5% or more of Mg was activated.

Example 4 In this example, C was doped instead of Mg. The other points were the same as in Example 1. In this example, 5% or more of C was activated.

[0050]

As described above, Mg-doped GaN
A strained Mg-doped InGaN layer is grown on the layer, or strained Mg-doped GaN is grown on the Mg-doped AlGaN layer.
By growing the layer and heat-treating it, there is an advantage that a resistivity of 1/3 or less of the usual value can be obtained.

Further, in the method of the present invention, since the effective concentration of Mg acceptor is increased, it becomes possible to substantially suppress the doping amount of Mg, and it is expected that the crystallinity of the nitride semiconductor itself is improved.

[Brief description of drawings]

FIG. 1 shows a structural diagram when a Mg-doped GaN layer and an InGaN layer are grown on a sapphire substrate.

FIG. 2 shows changes in the hole concentration in the Mg-doped GaN layer when the film thickness of the Mg-doped InGaN layer is changed.

FIG. 3 shows the calculation results of strain / lattice relaxation states of the InGaN layer when the In composition and the film thickness in the InGaN layer are changed.

FIG. 4 shows a structural diagram when an Mg-doped AlGaN layer and a GaN layer are grown on a sapphire substrate.

FIG. 5 shows the calculation results of the strain / lattice relaxation state of the GaN layer when the Al composition of the AlGaN layer and the film thickness of GaN on the AlGaN layer are changed.

FIG. 6 shows a structural diagram when an Mg-doped AlGaN layer and an InGaN layer are grown on a sapphire substrate.

FIG. 7: Al composition of AlGaN layer and InGaN thereon
2 shows the calculation results of the strain / lattice relaxation state of the InGaN layer when the film thickness is changed.

Continued front page    (72) Inventor Naoki Kobayashi             2-3-1, Otemachi, Chiyoda-ku, Tokyo Japan             Telegraph and Telephone Corporation F-term (reference) 5F041 AA21 CA34 CA40 CA46 CA49                       CA57 CA65 CA73                 5F073 CA07 CB05 CB07 CB19 DA05                       DA12 DA35 EA29

Claims (11)

[Claims] 1. A semiconductor composed of a nitride semiconductor doped with a p-type impurity and having a lattice constant a1 and a semiconductor 2 having a lattice constant a2 different from a1 are bonded together in a state containing lattice strain, 1. A low resistance nitride semiconductor, wherein the dopant 1 is activated by 5% or more. 2. The semiconductor 1 is GaN and the semiconductor 2 is InGa.
The low resistance nitride semiconductor according to claim 1, wherein the low resistance nitride semiconductor is N. 3. The semiconductor 1 is A1GaN and the semiconductor 2 is Ga.
The low resistance nitride semiconductor according to claim 1, wherein the low resistance nitride semiconductor is N. 4. The semiconductor 1 is A1GaN and the semiconductor 2 is In
The low resistance nitride semiconductor according to claim 1, which is GaN. 5. The low resistance nitride semiconductor according to claim 1, wherein the dopant is Mg or C. 6. A lattice constant a doped with a p-type impurity
On a semiconductor 1 having a lattice constant a2 different from a1
A method for manufacturing a low-resistance nitride semiconductor, comprising: bonding the semiconductor 2 having the above-mentioned condition in a state containing a lattice strain, and then performing a heat treatment for activating the dopant. 7. The method for producing a low resistance nitride semiconductor according to claim 6, wherein the heat treatment is performed at a temperature of 600 ° C. or higher. 8. The semiconductor 1 is GaN and the semiconductor 2 is InGa.
N is N, The manufacturing method of the low resistance nitride semiconductor of Claim 6 or 7 characterized by the above-mentioned. 9. The semiconductor 1 is A1GaN and the semiconductor 2 is Ga.
N is N, The manufacturing method of the low resistance nitride semiconductor of Claim 6 or 7 characterized by the above-mentioned. 10. The semiconductor 1 is A1GaN and the semiconductor 2 is I.
8. The method for producing a low resistance nitride semiconductor according to claim 6, which is nGaN. 11. The method for manufacturing a low resistance nitride semiconductor according to claim 6, wherein the dopant is Mg or C.