JP2015115433A - Group iii nitride semiconductor element - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent deterioration in characteristics caused by polarization charge arising from a heterointerface.SOLUTION: A group III nitride semiconductor element has a heterointerface where group III nitride semiconductors having compositions different from each other are bonded. The heterointerface S exists in a selected region which is at least one of an n-conductive type region where an electron is a major carrier and a p-conductive region where a hole is a major carrier. A dopant for generating a carrier with a code the same with that of the major carrier in the selected region is added to minute regions which sandwich the heterointerface S which is at least one heterointerface S in the selected region and where a polarization charge with a code the same with that of the major carrier in the selected region appears at an impurity concentration higher than an impurity concentration in a region except minute regions 20 in layers 142b, 17 on both sides which form the heterointerface.

Description

  The present invention relates to a group III nitride semiconductor device that suppresses deterioration of characteristics due to charges generated at a heterojunction interface.

As a blue and white LED, a light emitting element in which a plurality of layers of group III nitride semiconductors having different lattice constants are stacked is known. In Patent Document 1, a buffer layer is formed on a sapphire substrate, and an n-contact layer made of n-GaN, a guide layer made of n-GaN, a well layer made of In _0.2 Ga _0.8 N are formed on the buffer layer. In _0.02 Ga _0.98 N, In _0.05 Ga _0.95 N, In _0.02 Ga _0.98 N three-layer barrier layer active layer with multiple quantum well structure, p-GaN first guide layer, Ga _0.8 Al _0.2 N A light emitting device is disclosed in which an electron overflow prevention layer made of p-GaN, a second guide layer made of p-GaN, and a p-contact layer made of p-GaN are formed. In this technique, the barrier layer in the active layer has a three-layer structure in which the In composition ratio is sufficiently smaller than the In composition ratio of the well layer (the band gap is sufficiently large), and the In composition ratio is different. This is a technique for improving quantum efficiency by reducing the internal electric field applied to the active layer by using a symmetric structure of a small layer, a large layer, and a small layer. That is, in this technique, InGaN has a large piezoelectric polarization, but the In composition ratio of the barrier layer is small and the In distribution is made symmetric, thereby suppressing the piezoelectric field.

Patent Document 2 discloses an n-type buffer layer made of Si-doped GaN, an n-type cladding layer made of Si-doped Al _a Ga _1-a N (0 <a ≦ 0.1) on an n-type GaN substrate, An n-type guide layer made of Si-doped GaN, an active layer made of In _b Ga _1-b N (0 <b ≦ 0.1), an electron block layer made of Mg-doped Al _0.07 Ga _0.93 N, and Mg-doped Al _0.04 Ga _A first p-type cladding layer made of _0.96 N (film thickness = 400 nm, Mg doping concentration = 1.5 × 10 ¹⁹ / cm ³ ), and a second p-type cladding layer made of Mg-doped Al _0.04 Ga _0.96 N ( A multilayer p-type cladding layer having a thickness of 7 nm and a Mg doping concentration of 2 × 10 ²⁰ / cm ³ is laminated, and a Ni / Au p-contact electrode is formed on the second p-type cladding layer. A light emitting device is disclosed. In this technique, in the case of a p-AlGaN cladding layer, a p-GaN contact layer, and a p-contact electrode structure, holes injected from the p-contact electrode into the p-GaN contact layer are blocked by the barrier of the p-AlGaN cladding layer. This is a technique in which the injection inhibition is improved when the injection into the active layer side is inhibited and the resistance increases. That is, without providing a p-GaN contact layer, a second p-type cladding layer having a very high concentration of ²⁰ nm and a Mg concentration of about 10 ²⁰ / cm ³ where the p contact electrode is directly bonded is provided. The increase in resistance is suppressed by tunneling the hole through the second p-type cladding layer.

JP 2012-69982 A JP 2009-21424 A

  In the case of a group III nitride semiconductor light emitting device, a stacked structure of semiconductor layers with different lattice constants grown in the + c-axis direction (with the group III element surface as the growth surface) is employed as in Patent Documents 1 and 2. ing. In this case, the electron block layer uses a semiconductor having a larger band gap than the layers on both sides in order to increase the conduction band barrier against electrons. For this purpose, AlGaN is often used for the electron blocking layer, but the underlying layer is GaN for the guide layer or GaInN for the active layer. In this case, an in-plane tensile strain is applied to the AlGaN electron block layer having a small lattice constant from the lower layer having a larger lattice constant than that layer, and a compressive strain is applied in the thickness direction. Originally, in the AlGaN layer, spontaneous polarization exists in the direction from the group III element plane (Ga plane) to the N plane (−c axis), but the spontaneous polarization of AlGaN is larger than that of GaInN and GaN. Further, when a strain in this state is applied to the AlGaN layer, piezoelectric polarization occurs in the same direction as the spontaneous polarization. Since the lower GaInN layer or GaN layer is subjected to compressive strain (tensile strain in the thickness direction) in the in-plane direction from the upper AlGaN layer, piezoelectric polarization occurs in the direction from the N surface to the group III element surface (+ c axis). Occur. As a result, positive polarization charges appear on the N face of the AlGaN layer, which is the interface between the AlGaN layer and the lower layer. This positive charge biases the electron blocking layer of AlGaN to a positive potential and lowers the level of the conduction band of the electron blocking layer. As a result, the height of the barrier of the electron blocking layer is lowered, and the function of confining electrons in the lower guide layer and active layer is reduced. As a result, the internal quantum efficiency is lowered and the light emission output is lowered.

  Accordingly, the present invention has been made to solve the above-described problem, and prevents an excess bias potential from being applied to the layer by shielding the charge appearing at the heterointerface, or an active layer of carriers. An object of the present invention is to prevent deterioration of element characteristics by suppressing overflow from the element.

  The first invention for solving the above-mentioned problems is a group III nitride semiconductor device having a heterointerface in which group III nitride semiconductors having different compositions are joined, wherein the heterointerface has n conduction in which electrons are majority carriers. Present in a selection region that is at least one region of a p-type region and a p-conduction region in which holes are majority carriers, and has the same sign as the charge of the majority carrier in the selection region, at least one heterointerface in the selection region A carrier having the same sign as the majority carrier in the selected region with an impurity concentration higher than the impurity concentration in the region excluding the minute region in the layers on both sides forming the heterointerface is placed in a minute region sandwiching the heterointerface where the polarized charge appears. It is a group III nitride semiconductor device characterized by adding a dopant to be generated.

  The present invention generates carriers having the same sign as the majority carrier in a minute region including the hetero interface at a concentration higher than the impurity concentration of both layers forming the hetero interface in order to shield the polarization charge appearing at the hetero interface. It is a feature that a dopant to be added is added. When a dopant generates a carrier, it becomes an ion having the opposite sign to the charge of the generated carrier. This ion charge shields the polarization charge generated at the heterointerface having the same sign as the majority carrier, and reduces the potential generated by the polarization charge.

  The heterointerface that should shield the polarization charge may exist in the p-conduction type region, the n-conduction type region, or both. The selected region is a region where there is a heterointerface that should shield polarization charges. When the heterointerface that should shield the polarization charge exists only in the p-conduction region, the selection region is the p-conduction region, and when the heterointerface that should be shielded exists only in the n-conduction region. The selection region is an n-conduction region, and when the heterointerface to be shielded exists in the p-conduction region and the n-conduction region, the selection region is the p-conduction region and the n-conduction region. Means both. In addition, the heterointerface that should shield the polarization charge may be present in the same conductivity type region as long as it is an interface in which the polarization charge having the same sign as the majority carrier charge appears. It is also possible to shield only the polarization charge by adding impurities at a high concentration to the minute region.

  The heterointerface that shields the polarization charge is a heterointerface where a positive polarization charge appears in the p-conduction region, and a heterointerface where a negative polarization charge appears in the n-conduction region. The micro region is a partial region of two layers constituting the hetero interface across the hetero interface. In at least one of the two layers, a dopant for generating majority carriers is added at a prescribed concentration. The impurity concentration of the dopant in the micro region is higher than the prescribed concentration. is there. The dopant added to the minute region is an acceptor in the p-conduction type region and a donor in the n-conduction type region, but even if it is the same element as the dopant of a prescribed concentration added to generate majority carriers, Different elements may be used.

  The thickness of the minute region is desirably 4 nm or more and 20 nm or less. The minute region is formed across two layers on both sides of the heterointerface, and the entire width across the two layers is the thickness of the minute region. If the thickness of the micro region is larger than 20 nm, the region where the impurity is added at a high concentration becomes wide, and the crystallinity of both layers is lowered. Is not desirable because it is not performed sufficiently.

The impurity concentration of the dopant added to the minute region is desirably 5 × 10 ¹⁹ / cm ³ or more. The impurity concentration is desirably 1 × 10 ²¹ / cm ³ or less. If the impurity concentration is less than 5 × 10 ¹⁹ / cm ³ , polarization charges are not sufficiently shielded at the heterointerface, and if it is greater than 1 × 10 ²¹ / cm ³ , the crystallinity is lowered, which is not desirable. Further, it is preferably 8 × 10 ¹⁹ / cm ³ or more and 1 × 10 ²¹ / cm ³ or less, more preferably 1 × 10 ²⁰ / cm ³ or more, 1 × 10 ²¹ / cm ³ or less, and further 2 It is desirable that it is 10 × 10 ²⁰ / cm ³ or more and 1 × 10 ²¹ / cm ³ or less.

When the thickness of the minute region is dnm, the impurity concentration is n / cm ³ , the polarization charge density at the heterointerface is ω / cm ² , and the impurity activation ratio is r,
rnd = ω × 10 ¹⁷ (1)
If the impurity concentration and the thickness of the minute region are determined so as to satisfy the above, polarization charges can be completely shielded.
Therefore, if a shielding error of 10% is allowed, it is desirable that the impurity concentration n and the minute area thickness d satisfy the formula (2).
ω × 0.9 × 10 ¹⁷ ≦ rnd ≦ ω × 1.1 × 10 ¹⁷ (2)
When the dopant activation ratio r is 1, n and d are
ω × 0.9 × 10 ¹⁷ ≦ nd ≦ ω × 1.1 × 10 ¹⁷ (3)
It is desirable that the value satisfies the above.

  In the above invention, when the selected region is a p-conduction type region, the dopant is preferably magnesium or zinc. Further, when the selected region is an n-conduction type region, the dopant is preferably silicon or germanium. Since the polarization charge includes spontaneous polarization and piezo polarization, it is desirable to shield the polarization charge generated by the polarization vector that is the vector sum thereof. Of course, only one of the charges may be shielded. In the above invention, the polarization charge increases when the directions of the polarization vectors in the two layers sandwiching the heterojunction existing in the minute region are opposite to each other. Therefore, it is desirable to shield the charge at the heterointerface where the polarization charge is large.

  In the case of a group III nitride semiconductor, spontaneous polarization occurs in a direction (−c axis) from a group III element (typically Ga, hereinafter referred to as Ga) to an N element. The direction of polarization is defined as the direction from negative charge to positive charge. When compressive strain (and hence tensile strain in the thickness direction) is applied to the group III nitride semiconductor in the in-plane direction, piezoelectric polarization occurs in the direction from the N element to the Ga element (+ c axis). . Conversely, when a tensile strain (and hence a compressive strain in the thickness direction) is applied to the group III nitride semiconductor in the in-plane direction, the direction from the Ga element to the N element (−c axis) Polarization occurs. In the case of AlGaN, the absolute values of spontaneous polarization and piezo polarization are comparable. Since AlGaN with a small lattice constant generally undergoes tensile strain in the in-plane direction (compression strain in the thickness direction), the directions of spontaneous polarization and piezo polarization are both equal in the direction of the −c axis. Generates a large polarization charge. In the case of GaInN, since the lattice constant is generally large with respect to the layers on both sides, it is subjected to compressive strain in the in-plane direction (tensile strain in the thickness direction). Therefore, in the case of GaInN, the direction of piezo polarization is the + c axis, which is opposite to the direction of spontaneous polarization. However, since piezo polarization is much larger than spontaneous polarization, a large polarization charge is generated at the interface.

Therefore, it is desirable that the heterointerface that shields the polarization charge is an interface that satisfies the following conditions. In the following description, the lower layer includes a case where the strain from the lower layer is relaxed and a case where the lower layer is not relaxed, but the lower layer is also assumed to be strained from the upper layer. In the case where the thickness of the lower layer is sufficiently thicker than the thickness of the upper layer, it is assumed that the lower layer receives a strain from the upper layer although the lower layer receives a small strain from the upper layer.
The first condition is that the selected region is a p-conduction type region, and the upper layer is a lower layer in the two layers of the + c plane grown first and the + c plane grown upper layer sandwiching the heterojunction in the minute region. Layer having a larger band gap than that. Since a large band gap and a small lattice constant are equivalent, the upper layer receives tensile strain in the in-plane direction from the lower layer (compressive strain in the thickness direction), and the lower layer compresses strain in the in-plane direction from the upper layer ( Subject to tensile strain in the thickness direction. As a result, piezoelectric polarization occurs in the direction from the Ga element to the N element (−c axis), and piezoelectric polarization occurs in the lower layer in the direction from the N element to the Ga element (+ c axis). As a result, by combining the spontaneous polarization and the piezo polarization, a large positive polarization charge is generated at the heterointerface between the lower layer and the upper layer. In spontaneous polarization, generally, the upper layer having a large band gap containing Al is smaller than the lower layer having a small band gap containing In. In addition, since the direction of the spontaneous polarization is the direction of the −c axis, that is, the direction from the upper layer to the lower layer, a positive polarization charge due to the difference in the spontaneous polarization appears at the heterointerface between the lower layer and the upper layer. As a result, by combining the spontaneous polarization and the piezo polarization, a large positive polarization charge is generated at the hetero interface between the upper layer and the lower layer. Therefore, an acceptor is added to a minute region sandwiching the heterointerface, and positive polarization charges are shielded by the negative ions of the acceptor. A negative polarization charge appears at the other interface (upper side) of the upper layer, but since a free hole generated by the added acceptor shields the negative polarization charge, no bias potential is applied to the upper layer.

  The second condition is that the selected region is an n-conductivity type region, and the upper layer is divided into two layers, a lower layer grown on the + c plane and a upper layer grown on the + c plane next to the heterojunction in the minute region. Is a layer having a smaller band gap than the lower layer. For the same reason as above, the upper layer with a small band gap is subjected to compressive strain in the in-plane direction from the lower layer (tensile strain in the thickness direction), and the lower layer is tensile strain in the in-plane direction from the upper layer (compressive strain in the thickness direction). Receive. As a result, in the upper layer, piezo polarization occurs in the direction of the + c axis, and in the lower layer, piezo polarization occurs in the direction of the -c axis. As for spontaneous polarization, as described above, the lower layer with the larger band gap is larger than the upper layer with the smaller band gap. Moreover, since the direction of the spontaneous polarization is the direction of the −c axis, that is, the direction from the upper layer to the lower layer, a negative polarization charge due to the difference in the spontaneous polarization appears at the heterointerface between the lower layer and the upper layer. As a result, the spontaneous polarization and the piezo polarization are combined to generate a large negative polarization charge at the heterointerface between the upper layer and the lower layer. Therefore, a donor is added to a minute region sandwiching the heterointerface, and negative polarization charges are shielded by the positive ions of the donor. A positive polarization charge appears at the other interface (lower side) of the lower layer, but since a free electron generated by the added donor shields this positive polarization charge, no bias potential is applied to this lower layer.

  The same applies when grown in the direction of the -c axis. That is, the third condition is that the selected region is a p-conductivity type region, and the two layers of the lower layer grown on the −c plane and the upper layer grown on the −c plane are sandwiched between the heterojunction in the minute region. The upper layer is a layer having a smaller band gap than the lower layer. The upper layer with a small band gap and therefore a large lattice constant is subjected to compressive strain (tensile strain in the thickness direction) from the lower layer in the in-plane direction, and the lower layer is subjected to tensile strain in the in-plane direction from the upper layer (compressive strain in the thickness direction). ) As a result, in the upper layer, piezo polarization occurs in the direction of the + c axis, and in the lower layer, piezo polarization occurs in the direction of the -c axis. As for spontaneous polarization, as described above, the lower layer with the larger band gap is larger than the upper layer with the smaller band gap. Moreover, since the direction of spontaneous polarization is the direction of the −c axis, that is, the direction from the lower layer to the upper layer, a positive polarization charge due to the difference in spontaneous polarization appears at the heterointerface between the lower layer and the upper layer. As a result, by combining the spontaneous polarization and the piezo polarization, a large positive polarization charge is generated at the hetero interface between the upper layer and the lower layer. Therefore, an acceptor is added to a minute region sandwiching the heterointerface, and positive polarization charges are shielded by the negative ions of the acceptor. Negative polarization charge appears at the other interface (lower side) of the lower layer, but the free hole excited by the added acceptor shields the electric field due to this negative polarization charge, so the bias potential is Not applied.

  The fourth condition is that the selected region is an n-conducting type region, and the upper layer is divided into two layers, a lower layer grown on the −c plane and a next grown upper layer on the −c plane. Is a layer having a larger band gap than the lower layer. The upper layer with a large band gap and therefore a small lattice constant is subjected to tensile strain in the in-plane direction from the lower layer (compressive strain in the thickness direction), and the lower layer is subjected to compressive strain in the in-plane direction from the upper layer (tensile strain in the thickness direction). ) As a result, in the upper layer, piezo polarization occurs in the direction of the −c axis, and in the lower layer, piezo polarization occurs in the direction of the + c axis. As for spontaneous polarization, as described above, generally, the upper layer having a larger band gap is larger than the lower layer having a smaller band gap. In addition, since the direction of the spontaneous polarization is the direction of the −c axis, that is, the direction from the lower layer to the upper layer, a negative polarization charge due to the difference in spontaneous polarization appears at the heterointerface between the lower layer and the upper layer. As a result, the spontaneous polarization and the piezo polarization are combined to generate a large negative polarization charge at the heterointerface between the upper layer and the lower layer. Therefore, a donor is added to a minute region sandwiching the heterointerface, and negative polarization charges are shielded by the positive ions of the donor. A positive polarization charge appears at the other interface (upper side) of the upper layer, but a bias potential is not applied to the upper layer because free electrons generated by the added donor shield the positive polarization charge.

In the present invention, as an example, specifically, a layer having a large band gap between an upper layer and a lower layer constituting a heterointerface that should shield polarization charges is Al _x Ga _1-x N (0 <x ≦ 1) The layer with a small band gap is Ga _y In _1-y N (0 ≦ y ≦ 1), except for the case where x = 1 and y = 1. The spontaneous polarization of Al _x Ga _1-x N (0 <x ≦ 1) is larger than the spontaneous polarization of Ga _y In _1-y N (0 ≦ y ≦ 1).

In addition, in a device having a p-conduction type region and an n-conduction type region by laminating group III nitride semiconductors, a heterointerface often exists in both regions. Therefore, the selected region is a p-conduction type region and an n-conduction type region, and a small region is set with respect to the heterointerface existing in both regions, and a dopant is added at a high concentration, thereby making the heterogeneity. You may make it shield the polarization charge which arises in an interface.
When the present invention is applied to a light emitting device or a light receiving device, a layer having a larger band gap is formed between an upper layer and a lower layer sandwiching a heterointerface that shields polarization charges, and an electron blocking layer in a p-conduction type region, In the n-conduction type region, a hole blocking layer is desirable.
The present invention is intended to be based on the above principle, and as long as the above principle is satisfied, the present invention is not limited to the composition, composition ratio, and lattice constant of the layers on both sides of the heterointerface that shields polarization charges. Further, even if only spontaneous polarization appears, polarization charges appear at the heterointerface due to the difference in the magnitude of spontaneous polarization between the two layers. Therefore, the present invention includes shielding polarization charges in such cases. For example, even if the lattice constants are equal, if the composition is different, the magnitude of the spontaneous polarization is different, so that polarization charges appear at the heterointerface. The present invention includes the case of shielding this polarization charge.

  According to the present invention, the polarization charge generated at the heterointerface can be shielded by the added and ionized dopant, and the application of an electric field due to the polarization charge to the layer can be eliminated, for example, suppression of carrier overflow The characteristics of the element such as can be improved.

The block diagram of the light emitting element which concerns on the specific Example 1 of this invention. FIG. 3 is a band diagram of a region including a heterointerface that shields polarization charges in the light-emitting element of Example 1; FIG. 6 is a characteristic diagram showing the relationship between current and injection efficiency with the Mg concentration in a minute region as a parameter in the light emitting device according to Example 1; In the light emitting element which concerns on Example 1, the band figure for demonstrating the characteristic improvement by the high Mg density | concentration of a micro area | region. FIG. 6 is a characteristic diagram showing a relationship between voltage and current density using Mg concentration in a minute region as a parameter in the light emitting element according to Example 1; FIG. 6 is a band diagram of a region including a heterointerface that shields polarization charges in the light-emitting element of Example 2. The block diagram of the light emitting element which concerns on the specific Example 3 of this invention. FIG. 6 is a band diagram of a region including a heterointerface that shields polarization charges in the light-emitting element of Example 3; FIG. 6 is a characteristic diagram showing a relationship between voltage and current density using a Mg concentration in a micro region as a parameter in the light emitting device according to Example 3; The band figure of the area | region containing the hetero interface which shields the polarization charge in the light emitting element which concerns on the specific Example 4 of this invention.

  Hereinafter, the present invention will be described based on specific examples. The present invention is not limited to the following examples.

The semiconductor device of this example is a device in which a heterointerface consisting of an AlGaN layer and a GaN layer exists in the p-conduction type region, and polarization charges generated at the heterointerface are shielded.
FIG. 1 is a conceptual diagram of the layer structure of a group III nitride semiconductor light emitting device 1 that emits blue light. An undoped GaN layer 12 having a thickness of 2 μm is grown on the sapphire substrate 10 through the low temperature deposition buffer layer 11 by using a metal organic compound vapor phase growth method. At this time, the growth surface of the GaN layer 12 is a Ga surface. That is, GaN is grown in the + c axis direction. Thereafter, the Si-doped GaN layer 13 having a thickness of 3 μm, followed by an active layer 14 having a triple quantum well structure having a GaN barrier layer 142 having a thickness of 12.5 nm and a GaInN well layer 141 having a thickness of 2.5 nm. Grow. Next, the GaN barrier layer 142b is formed to a thickness of 12.5 nm. At this time, in the GaN barrier layer 142b, a region having a thickness of 2 nm below the upper surface forms a minute region lower portion 15 of the minute region 20 to which Mg is added at a higher concentration than other regions, as will be described later. ing. Next, an AlGaN layer 17 having a thickness of 20 nm is formed on the GaN barrier layer 142b. At this time, in the AlGaN layer 17, a region having a thickness of 10 nm above the lower surface forms a microregion upper portion 16 of a microregion 20 to which Mg is added at a higher concentration than other regions as described later. The 20 nm thick AlGaN layer 17 is provided to function as an electron blocking layer that suppresses electron overflow. Next, an Mg-doped GaN layer 18 having a thickness of 60 nm and an Mg-doped GaN contact layer 19 having a thickness of 10 nm are grown on the AlGaN layer 17.

The impurity concentration is as follows. In the Si-added GaN layer 13, the Si concentration is 5 × 10 ¹⁸ / cm ³ , and in the Mg-added GaN contact layer 19, the Mg concentration is 1 × 10 ²⁰ / cm ³ . In the conventional structure, the GaN barrier layer 142b is undoped, and the Mg concentration of the AlGaN layer 17 and the GaN layer 18 is a prescribed 2 × 10 ¹⁹ / cm ³ . In this case, the reason why Mg is not added to 2 × 10 ¹⁹ / cm ³ or more is that when it is added at a concentration higher than this, the hole concentration is substantially lowered and the possibility that the crystallinity is further deteriorated increases. . In the present invention, the heterointerface S between the GaN barrier layer 142b and the AlGaN layer 17 is an interface that should shield polarization charges, and a microregion lower portion 15 having a thickness of 2 nm and a microregion having a thickness of 10 nm sandwiching the heterointerface S. A region having a total width of 12 nm composed of the upper portion 16 becomes a minute region 20. In the present invention, the Mg concentration of the minute region 20 is set to 5 × 10 ¹⁹ / cm ³ or more. The active layer 14 is free of impurities.

The spontaneous polarization P _sp and the piezo polarization P _pe generated in each layer from the GaN barrier layer 142b to the GaN contact layer 19 which are p-conduction type regions will be examined. The GaN barrier layer 142b is not added with Mg except for the microregion lower portion 15, but is included in the p-conduction type region. As described above, the direction of spontaneous polarization is the direction of the −c axis, and the magnitude of spontaneous polarization increases in the order of GaN, InN, and AlN. Therefore, the magnitude of the spontaneous polarization is in a relationship of GaN, GaInN <AlGaN, AlInN. Although GaN and GaInN depend on the composition ratio of In, GaInN is smaller than GaN except for the vicinity of InN. Moreover, with AlGaN and AlInN, the higher the composition ratio of Al, the smaller the In composition ratio, the greater the spontaneous polarization. Therefore, as shown in FIG. 2, spontaneous polarization _Psp occurs in each layer in the direction from the Ga plane to the N plane (−c axis). Piezoelectric polarization depends on the magnitude of applied strain, but when comparing InN on GaN and AlN on GaN, InN on GaN is about 5 times larger in absolute value. Further, the direction of piezoelectric polarization of InN on GaN is the + c-axis direction opposite to the direction of spontaneous polarization, and AlN on GaN has the same -c-axis direction as the direction of spontaneous polarization. Accordingly, the piezoelectric polarization of the upper layer with GaN as the lower layer increases in the + c axis direction as the In composition ratio increases, and increases in the −c axis direction as the Al composition ratio increases. The increase rate of the absolute value of piezoelectric polarization with respect to the In composition ratio is larger than the increase rate with respect to the Al composition ratio. Therefore, piezoelectric polarization P _pe occurs in each layer as shown in FIG.

The AlGaN lattice constant of the AlGaN layer 17 is smaller than the GaN lattice constant of the GaN barrier layer 142b. Accordingly, since the AlGaN layer 17 is subjected to tensile strain (compression strain in the thickness direction) in the in-plane direction from the GaN barrier layer 142b, piezoelectric polarization P _pe occurs in the AlGaN layer 17 in the direction of the −c axis. doing. On the other hand, since the GaN barrier layer 142b is subjected to compressive strain (tensile strain in the thickness direction) in the in-plane direction from the upper AlGaN layer 17, the piezoelectric polarization P _{pe is} oriented in the + c axis direction in the GaN barrier layer 142b. Has occurred. Regarding the spontaneous polarization P _sp , both the GaN barrier layer 142b and the AlGaN layer 17 are oriented in the −c axis, but the AlGaN layer 17 is larger, so that the heterointerface S formed by both layers spontaneously Depending on the difference in the magnitude of the polarization _Psp , a positive polarization charge appears. Further, since the piezo polarization P _pe is in the −c axis direction for the AlGaN layer 17 and the + c axis direction for the GaN barrier layer 142b, a positive polarization charge appears for the piezo polarization P _{pe at the} heterointerface S. . Therefore, a large positive polarization charge that is the sum of the positive charge due to spontaneous polarization P _sp and the positive charge due to piezo polarization P _pe appears at the heterointerface S.

Since the lattice constant of GaN is larger than that of AlGaN, the GaN layer 18 and the GaN contact layer 19 are subjected to compressive strain (tensile strain in the thickness direction) in the in-plane direction from the lower AlGaN layer 17. Therefore, in the GaN layer 18 and the GaN contact layer 19, piezoelectric polarization _Ppe occurs in the + c axis direction. Thus, the hetero-interface T AlGaN layer 17 and GaN layer 18, since the direction of the spontaneous polarization P _sp AlGaN layer 17 is greater than the spontaneous polarization of the GaN layer 18, appear negative charge due to the spontaneous polarization, piezoelectric polarization P _Since the direction of _pe is opposite, a negative charge due to piezo polarization appears.

In the present invention, a micro-region 20 (a micro-region lower portion 15 having a thickness of 2 nm and a micro-region having a thickness of 10 nm is sandwiched between the hetero-interface S in which a positive charge having the same sign as the charge of majority carriers in the p-conduction region appears). In the upper part of the region), acceptor Mg for exciting holes is added to 5 × 10 ¹⁹ / cm ³ or more. When the Mg activation ratio is 1, the Mg anion concentration is 5 × 10 ¹⁹ / cm ³ or more. When the Mg concentration n / cm ³ is 5 × 10 ¹⁹ / cm ³ and the width d of the minute region 20 is 12 nm, the surface density of the Mg anion is 6 × 10 ¹³ / cm ² from the equation (1). . Since the polarization charge density appearing at the heterointerface S is considered to be about 10 ¹³ / cm ² , the electric field generated by the polarization charge can be shielded by the Mg anion. Although a negative charge appears due to polarization at the heterointerface T between the AlGaN layer 17 and the GaN layer 18, holes with a concentration of 5 × 10 ¹⁹ / cm ³ excited by Mg added to the minute region 20 are caused by this negative charge. Shield the electric field. Therefore, a positive potential due to polarization charge is not applied to the AlGaN layer 17. As a result, the conduction band of the AlGaN layer 17 that is the electron blocking layer is not lowered, and the barrier is not lowered. Therefore, the AlGaN layer 17 does not deteriorate the function of confining electrons to the active layer 14 side.
Note that Mg is not added to the GaN barrier layer 142b excluding the microregion lower portion 15 of the microregion 20, and the AlGaN layer 17 and the GaN layer 18 excluding the microregion upper portion 16 of the microregion 20 are conventionally provided. 2 × 10 ¹⁹ / cm ³ Mg is added. Therefore, from the lower surface of the GaN barrier layer 142b upward (to the GaN contact layer 19 side), a region of 10.5 nm from the lower surface of the GaN barrier layer 142b does not contain Mg, and from the heterointerface S in the GaN barrier layer 142b. Mg is added to a region of 2 nm (lower microregion 15) and a region 10 nm from the heterointerface S of the AlGaN layer 17 (microregion upper portion 16) at 5 × 10 ¹⁹ / cm ³ or more, and the upper surface of the AlGaN layer 17 ( Mg is added to 2 × 10 ¹⁹ / cm ³ in the GaN layer 18 in a region 10 nm below the interface T). In short, Mg is added at a concentration of 5 × 10 ¹⁹ / cm ³ or more at a higher concentration than the other regions in the minute region 20 sandwiching the heterointerface S that shields the positive polarization charge.

FIG. 3 shows the calculation result of the current density injection amount dependency in the injection efficiency into the active layer of the conventional example and the example. As an example, two cases are shown, when Mg is added to the microregion 20 at a high concentration of 5 × 10 ¹⁹ / cm ³ and when it is added to 1 × 10 ²⁰ / cm ^3. 20 shows the case where Mg is not added at a high concentration, that is, the case where GaN barrier layer 142b is not added and Mg is added to AlGaN layer 17 and GaN layer 18 at 2 × 10 ¹⁹ / cm ³ . In the structure of the comparative example, electrons overflow from the active layer and the injection efficiency is greatly reduced. However, in the structure of the example, it is understood that the electron overflow is significantly suppressed and the injection efficiency is greatly improved.

The reason why the injection efficiency is improved can be understood as follows. A positive fixed charge due to polarization is formed at the heterointerface S between the GaN barrier layer 142b and the AlGaN layer 17. What exists as a negative charge that shields the positive fixed charge is an ionization acceptor after generating holes. However, since the polarization fixed charge is as large as about 1 × 10 ¹³ / cm ² , the added acceptor impurity concentration is about 2 × 10 ¹⁹ / cm ³ stipulated in the prior art, or If it is added only to one side of the heterointerface S, the huge polarization fixed charge generated at the heterointerface S cannot be sufficiently shielded, and a substantially positive fixed charge remains. Therefore, the potential felt by the electrons is lowered, and the function of the AlGaN layer 17 as an electron blocking layer is lowered.

This is shown in FIG. FIG. 4 is a band diagram of a comparative example when the current density is 100 A / cm ² and this example (when the Mg concentration of the microregion 20 is 1 × 10 ²⁰ / cm ³ ). Simply applying an Mg addition amount (about 2 × 10 ¹⁹ / cm ³ ) according to the prior art to the GaN / AlGaN interface results in a significant decrease in potential because the shielding effect is insufficient. As a result, electrons pass through the active layer and reach this interface, that is, overflow, so that the injection efficiency is greatly reduced. On the other hand, in the present invention, since sufficient acceptor impurities of 5 × 10 ¹⁹ / cm ³ or more exceeding the prescribed concentration of 2 × 10 ¹⁹ / cm ³ in the prior art are added to the minute region 20 sandwiching the heterointerface S, positive Fixed charges can be shielded sufficiently, and potential drops are eliminated. As a result, the overflow of electrons can be suppressed, and the injection efficiency is greatly improved.

Further, FIG. 5 shows current-voltage characteristics. In the structure of the present embodiment, a large positive fixed charge generated at the heterointerface S can be appropriately shielded, so that a current can be passed at a low voltage, leading to an improvement in energy conversion efficiency.
In order to obtain the effects of the present invention as described above, it is necessary to add acceptor impurities of at least 5 × 10 ¹⁹ / cm ³ or more to the minute region 20 sandwiching the heterointerface S that shields positive polarization charges. Preferably it is 2 × 10 ²⁰ / cm ³ or more. The microregion 20 to be added is a region of at least 2 nm and 10 nm or less (the width d is 4 nm or more and 20 nm or less) on both sides of the heterointerface S. If the thickness is too thin, the effect cannot be obtained. The width d of the minute region sandwiching the heterointerface S is desirably 2 nm or more and 15 nm or less.

The group III nitride semiconductor device 2 of Example 2 is a case where the device 1 of Example 1 was grown in N-plane growth (−c axis direction). The configuration of each layer is the same as in FIG. In this case, the direction of the spontaneous polarization P _{sp and} the direction of the piezo polarization P _pe are opposite to those in the first embodiment. Spontaneous polarization P _sp is the direction of the −c axis in all layers, that is, the direction toward the GaN contact layer 19. The direction of the piezoelectric polarization P _pe is the + c axis direction in the GaN barrier layer 142b, the −c axis direction in the AlGaN layer 17, and the + c axis direction in the GaN layer 18. At this time, unlike the first embodiment, the heterointerface where the positive polarization charge appears is the heterointerface S between the AlGaN layer 17 and the GaN layer 18 above the AlGaN layer 17. As is apparent from FIG. 6, the heterointerface T between the AlGaN layer 17 and the GaN barrier layer 1426 has a difference in the magnitude of the spontaneous polarization described in the first embodiment and the relationship between the direction and magnitude of the piezo polarization. Negative polarization charge appears. Therefore, in the element 2 of Example 2, the heterointerface S between the AlGaN layer 17 and the GaN layer 18 becomes a heterointerface that shields positive polarization charges, and from the heterointerface S inside the AlGaN layer 17 across the heterointerface S. The lower region 171 with a thickness of 10 nm below and the upper region 181 with a thickness of 2 nm above the heterointerface S between the GaN layer 18 are added with a high concentration of Mg of 5 × 10 ¹⁹ / cm ³ or more. It becomes. Also in this case, the positive polarization charge appearing at the heterointerface S is shielded by the negative charge due to ionized Mg. Negative polarization charges appearing at the heterointerface T on the opposite side of the heterointerface S with respect to the AlGaN layer 17 are shielded by free holes excited by Mg added to the microregion 21. Therefore, the element 2 of Example 2 also exhibits the same characteristics as the element 1 of Example 1.

  Examples 1 and 2 are elements in the case where a heterointerface that shields polarization charges exists in the p-conduction type region, but Example 3 is a case in which this heterointerface is an element that exists in the n-conduction type region. It is.

FIG. 7 is a conceptual diagram showing the layer structure of the group III nitride semiconductor light emitting device 3 that emits light at 280 nm. Si-doped Al _0.7 Ga _0.3 having a thickness of 2 μm is formed on the Al-surface AlN template 30 formed on the sapphire substrate or the AlN substrate 30 whose growth main surface is the Al surface by using an organic metal compound vapor phase growth method. N layer 31 is grown. The Si-added Al _0.7 Ga _0.3 N layer 31 functions as a hole blocking layer. Thereafter, an Al _0.4 Ga _0.6 N barrier layer 342 a having a thickness of 7 nm is formed on the layer 31. Three pairs of Al _0.3 Ga _0.7 N well layer 341 having a thickness of 2 nm and an Al _0.4 Ga _0.6 N barrier layer 342 having a thickness of 5 nm are repeatedly formed on the barrier layer 342a. However, the last barrier layer is an Al _0.4 Ga _0.6 N barrier layer 342b having a thickness of 7 nm. The barrier layer 342a to the barrier layer 342b are the active layers 34 having a multiple quantum well structure. The active layer 34 is free of impurities. Subsequently, an Mg-added Al _0.4 Ga _0.6 N layer 37 having a thickness of 20 nm was formed on the barrier layer 342b. The Mg-added Al _0.8 Ga _0.2 N layer 37 is an electron block layer. Next, an Mg-added Al _0.3 Ga _0.7 N layer 38 having a thickness of 50 nm and an Mg-added Al _0.1 Ga _0.9 N contact layer 39 having a thickness of 10 nm were grown.

In the light-emitting element 3 of Example 3, as in Example 1, the interface between the Al _0.4 Ga _0.6 N barrier layer 342b and the Mg-added Al _0.8 Ga _0.2 N layer 37 on the layer shields positive polarization charges. This is the heterointerface S1. The region 2 nm below the heterointerface S1 in the Al _0.4 Ga _0.6 N barrier layer 342b is the lower region 35, and the upper 10 nm from the heterointerface S1 in the Mg-added Al _0.8 Ga _0.2 N layer 37 is the upper region 36. . The minute region lower part 35 and the minute region upper part 36 constitute a minute region 22p. In addition, Mg is added to the minute region 22p at a concentration of 2 × 10 ²⁰ / cm ³ that is 5 × 10 ¹⁹ / cm ³ or higher, which is higher than that of the other layers. The region excluding the minute region lower portion 35 in the Al _0.4 Ga _0.6 N barrier layer 342b is free of Mg, and the region excluding the minute region upper portion 36 in the Al _0.8 Ga _0.2 N layer 37 and the Mg-added Al _0.3 Ga _0.7 N The Mg concentration of the layer 38 is 2 × 10 ¹⁹ / cm ³ , and the Mg concentration of the Mg-added Al _0.1 Ga _0.9 N contact layer 39 is 2 × 10 ²⁰ / cm ³ .

Furthermore, in Example 3, as shown in FIG. 8, the interface between the Si-added Al _0.7 Ga _0.3 N layer 31 and the Al _0.4 Ga _0.6 N barrier layer 342a existing in the n-conduction type region shields negative polarization charges. It becomes the hetero interface S2. A micro-region lower part 32 that is a region 10 nm below the heterointerface S2 in the Si-added Al _0.7 Ga _0.3 N layer 31 and a micro-region that is a region 2 nm above the heterointerface S2 in the Si-added Al _0.4 Ga _0.6 N barrier layer 342a. The microregion 22n is constituted by the region upper portion 33. Then, in the minute region 22n composed of the minute region lower part 32 and the minute region upper part 33, Si is added to 2 × 10 ²⁰ / cm ³ having a concentration higher than that of other layers of 5 × 10 ¹⁹ / cm ³ or more. Has been. Further, Si is added to the region of the Si-added Al _0.7 Ga _0.3 N layer 31 except for the small region lower portion 32 to a prescribed concentration of 1 × 10 ¹⁸ / cm ³ used conventionally, and Al _0.4 Ga _0.6. Si is not added to the region except the minute region upper portion 33 of the N barrier layer 342a.

In the band diagram of FIG. 8, the orientation of Si added Al _0.7 Ga _0.3 N layer 31, Al _0.4 Ga _0.6 N barrier layer 342a spontaneous polarization P _sp is the direction of -c axis. However, Si added Al _0.7 Ga _0.3 N layer 31, Al composition ratio, Al _0.4 Ga _0.6 N is larger than the barrier layer 342a, the direction of spontaneous polarization P _sp layer 31, the spontaneous polarization P _sp barrier layer 342a Bigger than. Therefore, negative polarization charges due to spontaneous polarization appear at the heterointerface S2 between the layer 32 and the barrier layer 342a.

Regarding piezo polarization, the following can be said. The Si-added Al _0.7 Ga _0.3 N layer 31 receives compressive strain (tensile strain in the thickness direction) from the AlN substrate 30 having a lattice constant smaller than that of the layer, but has a thickness of 2 μm. Thicker than the thickness, the strain is relaxed. Further, the Si-added Al _0.7 Ga _0.3 N layer 31 is sufficiently thicker than the total thickness of 108 nm of the epitaxial growth layer formed thereabove. Therefore, the layer 31 is not subjected to distortion from the layer formed above or can be ignored. Therefore, the piezoelectric polarization _Ppe in the layer 31 can be ignored. The Al _0.4 Ga _0.6 N barrier layer 342a is subjected to compressive strain (tensile strain in the thickness direction) from the layer 31 in the in-plane direction. Therefore, piezoelectric polarization P _pe occurs in the barrier layer 342a in the + c axis direction. As a result, a negative polarization charge due to piezo polarization _Ppe appears at the heterointerface S2. A negative charge having a large sum of negative charges due to spontaneous polarization P _sp and negative charges due to piezo polarization P _pe appears at the heterointerface S2.

FIG. 9 shows the current-voltage characteristics of the element 3 of Example 3. In the comparative example (conventional), in the structure of FIG. 7, the Si concentration of all layers in the n-conduction type region is 1 × 10 ¹⁸ / cm ³ , and the Mg concentration of all layers excluding the contact layer in the p-conduction type region is 2 × 10. ^{This is a case where 19} / cm ³ and the Mg concentration of the contact layer 39 are 2 × 10 ²⁰ / cm ³ . That is, the comparative example is a case where the micro region 22n to which Si is added at a high concentration and the micro region 22p to which Mg is added at a high concentration are not provided. Obviously, the threshold voltage of the light-emitting element 3 of Example 3 is greatly reduced as compared with the light-emitting element of the comparative example, and it is understood that the power conversion efficiency is greatly improved.

The light-emitting element 4 of Example 4 is a light-emitting element in which a structure similar to that of Example 3 is grown on the −c plane. However, a GaN substrate having an N plane as a growth main surface is used as the substrate, and a Si-added GaN layer 51 is grown to a thickness of 2 μm in the direction of the −c axis. Then, an Si-added Al _0.7 Ga _0.3 N layer 31 is formed on the GaN layer 51 with a thickness of 20 nm. Each layer on the layer 31 is the same as that in the third embodiment. The Si-added Al _0.7 Ga _0.3 N layer 31 is a hole blocking layer.

The spontaneous polarization P _{sp of} each layer occurs in the direction of the −c axis. The piezoelectric polarization P _pe is ignored because the strain is relaxed in the thick GaN layer 51 or is thick, so that the strain applied from the upper layer is small. Since the Al _0.7 Ga _0.3 N layer 31 receives tensile strain in the in-plane direction from the lower GaN layer 13 having a larger lattice constant, the direction of the piezoelectric polarization P _{pe is} the direction of the −c axis. Further, since the Al _0.4 Ga _0.6 N barrier layer 342a receives compressive strain in the in-plane direction from the Al _0.7 Ga _0.3 N layer 31 having a smaller lattice constant, the direction of the piezoelectric polarization P _pe is the + c axis direction. Therefore, a negative polarization charge appears at the heterointerface S2 between the GaN layer 51 and the Al _0.7 Ga _0.3 N layer 31, and is positive at the hetero interface T between the Al _0.7 Ga _0.3 N layer 31 and the Al _0.4 Ga _0.6 N barrier layer 342a. The polarization charge of appears.

Therefore, in this embodiment, a minute region 22n composed of a minute region lower part 511 2 nm below the heterointerface S2 in the GaN layer 51 and a minute region upper part 311 10 nm above the heterointerface S2 in the Al _0.7 Ga _0.3 N layer 31. In addition, Si is added to a concentration of 2 × 10 ²⁰ / cm ³ , which is higher than that of other layers and is 5 × 10 ¹⁹ / cm ³ or more. Further, in a region excluding the minute region lower portion 511 of the GaN layer 51 and a region excluding the minute region upper portion 311 of the Al _0.7 Ga _0.3 N layer 31, Si has a prescribed concentration of 1 × 10 ¹⁸ / cm 2 conventionally used. ^The active layer including Al _0.4 Ga _0.6 N barrier layer 342a added to 3 is not doped with impurities. The structure of the p-conduction type region is the same as that of Example 3, and Mg is added to 2 × 10 ²⁰ / cm ³ , which is 5 × 10 ¹⁹ / cm ³ or more, in the minute region 22p as in the example.

  As described above, the present invention can be applied to a light emitting device grown in the + c-axis direction, a light-emitting element grown in the −c-axis direction, a p-conduction type region, and an n-conduction type region. As shown in Examples 1 and 3, when grown in the + c-axis direction, the electron block layer having a large band gap and the interface on the active layer side having a small band gap of the hole block layer (close to the active layer) Side) become heterointerfaces that shield positive and negative charges, respectively. On the other hand, as shown in Examples 2 and 4, when grown in the direction of -c axis, it is opposite to an electron blocking layer having a large band gap and an active layer having a small band gap of a hole blocking layer. The side interface (the side far from the active layer) is a heterointerface that shields positive and negative charges, respectively. In any case, an electron blocking layer with a large band gap is biased to a positive potential, which lowers the conduction band barrier, and a hole blocking layer with a large band gap is biased to a negative potential, thereby reducing the valence band barrier. Is prevented. As a result, an increase in the driving voltage of the device is suppressed, and the confinement of electrons and holes in the active layer is effectively performed, so that the quantum efficiency is improved as compared with the prior art.

15,32,35,171 ... minute area under 16,33,36,181 ... micro areas top 17 ... AlGaN layer 18 ... GaN layer 20,21,22p, 22n ... minute area 37 ... Mg added Al _0.8 Ga _0.2 N layer 31 ... Si added Al _0.7 Ga _0.3 N layer 51 ... Si added GaN layer 142b ... GaN barrier layer 342a, 342b ... barrier layer S, S1, S2 ... heterointerface

Claims (14)

In a group III nitride semiconductor device having a heterointerface in which group III nitride semiconductors having different compositions are joined,
The heterointerface exists in a selected region that is at least one of an n-conduction region where electrons are majority carriers and a p-conduction region where holes are majority carriers,
At least one heterointerface in the selected region, and a minute region sandwiching a heterointerface in which a polarization charge having the same sign as the charge of multiple carriers in the selected region appears,
A dopant that generates carriers having the same sign as the majority carriers in the selected region at an impurity concentration higher than the impurity concentration in a region excluding the minute region in both layers forming the heterointerface is added. Group III nitride semiconductor device.   2. The group III nitride semiconductor device according to claim 1, wherein a thickness of the minute region is 4 nm or more and 20 nm or less. ^{3. The} group III nitride semiconductor device according to claim 1, wherein an impurity concentration of the dopant added to the minute region is 5 × 10 ¹⁹ / cm ³ or more.     4. The group III nitride semiconductor device according to claim 1, wherein the selection region is a p-conduction type region, and the dopant is magnesium or zinc. 5.     4. The group III nitride semiconductor device according to claim 1, wherein the selection region is an n-conduction type region, and the dopant is silicon or germanium. 5.   6. The group III nitride semiconductor device according to claim 1, wherein the polarization charge is a polarization charge generated by a polarization vector that is a vector sum of spontaneous polarization and piezoelectric polarization. .   The group III nitride semiconductor device according to claim 6, wherein directions of the polarization vectors in two layers sandwiching the heterojunction existing in the minute region are opposite to each other.   The selective region is a p-conductivity type region, and in the two layers, the lower layer grown on the + c plane first and the upper layer grown on the + c plane next to the heterojunction in the minute region, the upper layer is lower than the lower layer. 8. The group III nitride semiconductor device according to claim 1, wherein the layer has a large band gap.   The selection region is an n-conductivity type region, and the upper layer is lower than the lower layer in two layers, a lower layer grown on the + c plane and the upper layer grown on the + c plane next to the heterojunction in the minute region. 8. The group III nitride semiconductor device according to claim 1, wherein the group III is a layer having a small band gap.   The selection region is a p-conductivity type region, and in the two layers, a lower layer that has been grown on a −c plane and a next layer that has been grown on a −c plane, sandwiching the heterojunction in the minute region, The group III nitride semiconductor device according to any one of claims 1 to 7, wherein the group III nitride semiconductor device has a band gap smaller than that of the lower layer.   The selection region is an n-conductivity type region, and in the two layers, a lower layer that has been grown with a −c plane and a second layer that has been grown with a −c plane, and the upper layer is the above-described upper layer. 8. The group III nitride semiconductor device according to claim 1, wherein the group III nitride semiconductor device has a band gap larger than that of the lower layer. The layer having a large band gap is Al _x Ga _1-x N (0 <x ≦ 1), and the layer having a small band gap is Ga _y In _1-y N (0 ≦ y ≦ 1). However, the group III nitride semiconductor device according to any one of claims 8 to 11, excluding a case where x = 1 and y = 1. When the thickness of the microregion is dnm, the impurity concentration is n / cm ³ , the polarization charge density at the heterointerface is ω / cm ² , and the activation ratio of the impurity is r, the thickness d and the impurity concentration n are
ω × 0.9 × 10 ¹⁷ ≦ rnd ≦ ω × 1.1 × 10 ¹⁷
The group III nitride semiconductor device according to any one of claims 1 to 12, wherein the group III nitride semiconductor device has a value satisfying the following conditions.   The group III nitride semiconductor device according to any one of claims 1 to 13, wherein the selection region is two regions of a p-conduction type region and an n-conduction type region.

Images