Heterojunction Diode

Abstract

To alleviate the trade-off between a reverse breakdown voltage and a current collapse.SOLUTION: A heterojunction diode 10 includes a channel layer 14 containing a nitride semiconductor, an electron supply layer 16 which is provided on the channel layer 14 and contains a nitride semiconductor, a cathode electrode 20 and an anode electrode 22 provided on the electron supply layer 16, a recess portion 28 provided on the upper surface 17 of the electron supply layer 16 located between the cathode electrode 20 and the anode electrode 22, a gate electrode 24 which is provided inside the recess portion 28 and electrically connected to the anode electrode 22, and an insulating layer 26 provided between the cathode electrode 20 and the gate electrode 24 and between the gate electrode 24 and the anode electrode 22 on the electron supply layer 16.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to heterojunction diodes including nitride semiconductors. [Background technology]

[0002] High electron mobility transistors (HEMTs) using nitride semiconductors such as gallium nitride (GaN) are known as field-effect transistors capable of processing microwave signals. By connecting the gate and drain of a HEMT, it is possible to operate the HEMT as a rectifying diode. [Prior art documents] [Non-patent literature]

[0003] [Non-Patent Document 1] Hidetada Takahashi et al., 81st Autumn Meeting of the Japan Society of Applied Physics, Abstracts 10a-Z04-3, August 26, 2020 Summary of the Invention [Problem to be solved by the invention]

[0004] Normally-off operation is required for HEMTs used in rectifier diodes. Normally-off HEMTs have a severe trade-off between reverse breakdown voltage and current collapse, making it difficult to operate at high voltages and large currents.

[0005] The present disclosure has been made in consideration of these problems, and one of its exemplary objectives is to provide a technique for mitigating the trade-off between reverse breakdown voltage and current collapse. [Means for solving the problem]

[0006] A heterojunction diode according to an embodiment of the present disclosure includes a channel layer containing a nitride semiconductor, an electron supply layer containing a nitride semiconductor and provided on the channel layer, a cathode electrode and an anode electrode provided on the electron supply layer, a recess provided on an upper surface of the electron supply layer located between the cathode electrode and the anode electrode, a gate electrode provided in the recess and electrically connected to the anode electrode, and insulating layers provided on the electron supply layer between the cathode electrode and the gate electrode and between the gate electrode and the anode electrode.

[0007] Any combination of the above components, or mutual substitution of the components or expressions of the present disclosure between methods, systems, etc., are also valid aspects of the present disclosure. Effect of the Invention

[0008] According to the present disclosure, the trade-off between reverse breakdown voltage and current collapse can be alleviated. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram illustrating a schematic structure of a heterojunction diode according to an embodiment. [Diagram 2] FIG. 1 is a diagram illustrating a schematic structure of a heterojunction diode according to a comparative example. [Diagram 3] 1 is a graph showing the relationship between the gate-cathode distance and the reverse withstand voltage and forward current of a heterojunction diode according to a comparative example. [Figure 4] 1 is a graph showing the relationship between the gate-cathode distance and the reverse withstand voltage and forward current of a heterojunction diode according to an example. [Diagram 5] 1 is a graph showing forward current-voltage characteristics of heterojunction diodes according to an example and a comparative example. [Figure 6] 1 is a graph showing reverse current-voltage characteristics of heterojunction diodes according to an example and a comparative example. [Figure 7] 1 is a graph showing the relationship between the product of the AlN mole fraction and thickness of the electron supply layer, and the threshold voltage.

Best Mode for Carrying Out the Invention

[0010] Hereinafter, embodiments for carrying out the present disclosure will be described in detail with reference to the drawings. In the description, the same elements are denoted by the same reference numerals, and redundant descriptions will be omitted as appropriate.

[0011] FIG. 1 is a diagram schematically showing the structure of a heterojunction diode 10 according to an embodiment. The heterojunction diode 10 includes a substrate 12, a channel layer 14, an electron supply layer 16, a cap layer 18, a cathode electrode 20, an anode electrode 22, a gate electrode 24, an insulating layer 26, a cathode terminal 30, and an anode terminal 32.

[0012] The heterojunction diode 10 is based on a high electron mobility transistor (HEMT), and is made to operate as a rectifying diode by electrically connecting the gate and drain of the HEMT. The cathode electrode 20 corresponds to the source of the HEMT, and the anode electrode 22 corresponds to the drain of the HEMT.

[0013] The heterojunction diode 10 is a so-called gallium nitride (GaN)-based semiconductor device, and a GaN-based nitride semiconductor is used for the channel layer 14, the electron supply layer 16, and the cap layer 18. Here, the GaN-based nitride semiconductor is a compound represented by Al x Ga y In 1-x-y N (0 ≦ x < 1, 0 < y ≦ 1, 0 < x + y ≦ 1), and is a compound containing at least GaN.

[0014] A two-dimensional electron gas (2DEG) 34 is generated near the interface between the channel layer 14 and the electron supply layer 16. The heterojunction diode 10 controls the concentration of the 2DEG 34 by the gate voltage applied to the gate electrode 24, thereby controlling the anode current flowing from the anode electrode 22 to the cathode electrode 20. The heterojunction diode 10 needs to cut off the anode current when the anode voltage is 0 V. Therefore, the HEMT that is the base of the heterojunction diode 10 is required to operate in a normally-off mode.

[0015] The substrate 12 is made of a material suitable for crystal growth of a nitride semiconductor, such as silicon carbide (SiC), sapphire (Al2O3), silicon (Si), or gallium nitride (GaN). In one embodiment, the substrate 12 is a semi-insulating SiC substrate.

[0016] The channel layer 14 is a GaN-based nitride semiconductor layer provided on the substrate 12. The channel layer 14 has a thickness of 100 nm or more and 10 μm or less. In one embodiment, the channel layer 14 is a GaN layer, and the channel layer 14 has a thickness of 900 nm. A nucleation layer made of AlN may be inserted between the substrate 12 and the channel layer 14, or a GaN buffer layer doped with carbon (C) or iron (Fe) may be inserted.

[0017] The electron supply layer 16 is a GaN-based nitride semiconductor layer provided on the channel layer 14. The electron supply layer 16 is, for example, an AlGaN layer, and the AlN mole fraction x of the electron supply layer 16 is 0.1 or more and 0.4 or less. In one embodiment, the AlN mole fraction x of the electron supply layer 16 is 0.22. A 2DEG 34 is formed in the vicinity of the junction interface between the channel layer 14 and the electron supply layer 16.

[0018] A recess 28 for burying the gate electrode 24 is provided in the upper surface 17 of the electron supply layer 16. The electron supply layer 16 has a first portion 16a between the channel layer 14 and the recess 28, and a second portion 16b between the channel layer 14 and the insulating layer 26. The first portion 16a has a smaller thickness than the remaining second portion 16b due to the recess 28 being formed. The thickness t1 of the first portion 16a is 1.5 nm or more and 14 nm or less, and the thickness t2 of the second portion 16b is 5 nm or more and 100 nm or less. In one embodiment, the thickness t1 of the first portion 16a is 4.5 nm, and the thickness t2 of the second portion 16b is 17.5 nm.

[0019] The cap layer 18 is a GaN-based nitride semiconductor layer provided on the electron supply layer 16. The cap layer 18 has a thickness of 1 nm or more and 100 nm or less. In one embodiment, the cap layer 18 is a GaN layer, and the cap layer 18 has a thickness of 3 nm. The cap layer 18 may be an n-type GaN layer doped with Si. Note that the cap layer 18 is not an essential component, and the cap layer 18 may not be provided.

[0020] The channel layer 14, the electron supply layer 16 and the cap layer 18 can be formed using a known epitaxial growth method such as metalorganic chemical vapor deposition (MOVPE).

[0021] The cathode electrode 20 and the anode electrode 22 are provided on the cap layer 18. The cathode electrode 20 and the anode electrode 22 are made of a metal material capable of making ohmic contact with the cap layer 18. The cathode electrode 20 and the anode electrode 22 are formed by forming a laminated structure of, for example, Ti / Al / Ni / Au, Ti / Al / Nb / Au, or Mo / Al / Mo / Au, and then alloying the laminated structure by high-temperature treatment at 600° C. to 900° C. In the case of a configuration without the cap layer 18, the cathode electrode 20 and the anode electrode 22 are provided on the electron supply layer 16 and make ohmic contact with the electron supply layer 16.

[0022] The gate electrode 24 is provided between the cathode electrode 20 and the anode electrode 22 on the cap layer 18. The gate electrode 24 has a field plate structure and includes a buried portion 24a and a field plate 24b. The buried portion 24a penetrates the insulating layer 26 and the cap layer 18 and is provided in a recess 28 carved into the electron supply layer 16. The gate electrode 24 is made of a metal material that is in Schottky contact with the electron supply layer 16 and the cap layer 18, and is, for example, made of a Ni / Au laminated structure.

[0023] The buried portion 24a is provided at a position with a predetermined distance L1 from the cathode electrode 20. The gate-cathode distance L1 from the buried portion 24a (or the recess 28) to the cathode electrode 20 affects the magnitude of the reverse breakdown voltage and the forward current of the heterojunction diode 10. For example, when the gate-cathode distance L1 is increased, the reverse breakdown voltage increases while the forward current decreases, which is a trade-off relationship. The gate-cathode distance L1 is 0.5 μm or more and 10 μm or less, and preferably 1 μm or more and 5 μm or less. In one embodiment, the gate-cathode distance L1 is 1.7 μm.

[0024] The field plate 24b is provided to extend from the buried portion 24a (or the recess 28) toward the cathode electrode 20. That is, the field plate 24b extends from the gate of the base HEMT toward the source. In a normal HEMT, the field plate extends from the gate toward the drain, and this point is different from a normal HEMT.

[0025] The length L2 of the field plate 24b is 0.1 μm or more and 2 μm or less, and preferably 0.2 μm or more and 1 μm or less. The length L2 of the field plate 24b is, for example, 1 / 10 to 1 / 2 of the gate-cathode distance L1, and preferably about 1 / 5. In one embodiment, the gate-cathode distance L1 is 0.35 μm. The gate electrode 24 may be configured without the field plate 24b.

[0026] In this embodiment, by forming the gate electrode 24 into a field plate structure, the following effects can be obtained. When a positive voltage is applied to the anode terminal 32, a high concentration 2DEG 34 is induced in the channel layer 14 directly below the field plate 24b, thereby suppressing current collapse. When a negative voltage is applied to the anode terminal 32, a depletion layer extends from directly below the buried portion 24a to directly below the field plate 24b, thereby mitigating electric field concentration at the gate end and improving reverse breakdown voltage.

[0027] The cathode electrode 20, the anode electrode 22 and the gate electrode 24 are formed by using a known technique such as a vapor deposition method or a sputtering method.

[0028] The insulating layer 26 is provided on the cap layer 18, and is formed so as to cover the exposed surface of the cap layer 18 on which the cathode electrode 20, the anode electrode 22, and the buried portion 24a of the gate electrode 24 are not provided. That is, the insulating layer 26 is provided between the cathode electrode 20 and the buried portion 24a of the gate electrode 24 on the cap layer 18, and between the buried portion 24a of the gate electrode 24 and the anode electrode 22. The insulating layer 26 is also disposed between the cap layer 18 and the field plate 24b of the gate electrode 24, and is formed so as to extend under the field plate 24b. The insulating layer 26 is provided so as to overlap at least a part of the cathode electrode 20 and the anode electrode 22. The insulating layer 26 may be provided so as not to overlap the cathode electrode 20 and the anode electrode 22.

[0029] The insulating layer 26 is made of a dielectric material containing nitrogen and silicon, and is made of a material mainly composed of silicon nitride (SiNx). The insulating layer 26 can be formed by a plasma enhanced chemical vapor deposition (PECVD) method using ammonia (NH3) or nitrogen (N2) as a nitrogen source and silane (SiH4) as a silicon source. The thickness of the insulating layer 26 is 5 nm or more and 500 nm or less, for example, about 60 nm.

[0030] The cathode terminal 30 and the anode terminal 32 are terminals for external connection. The cathode terminal 30 is electrically connected to the cathode electrode 20. The anode terminal 32 is electrically connected to the anode electrode 22 and the gate electrode 24. Therefore, the gate electrode 24 is electrically connected to the anode electrode 22.

[0031] Next, the operation of the heterojunction diode 10 will be described. When a positive voltage is applied to the anode terminal 32, a positive voltage is applied to the gate electrode 24, and 2DEG 34 is accumulated in the channel layer 14 near the first portion 16a of the electron supply layer 16 located under the buried portion 24a (or the recess 28). As a result, an anode current flows from the anode electrode 22 to the cathode electrode 20. When a negative voltage is applied to the anode terminal 32, a negative voltage is applied to the gate electrode 24, and the concentration of 2DEG 34 in the channel layer 14 near the first portion 16a of the electron supply layer 16 located under the buried portion 24a (or the recess 28) decreases and becomes depleted. As a result, the anode current is cut off between the anode electrode 22 and the cathode electrode 20.

[0032] Next, the effects of the present embodiment will be described with reference to a comparative example. Fig. 2 is a diagram showing a schematic structure of a heterojunction diode 50 according to the comparative example. The heterojunction diode 50 includes a substrate 52, a channel layer 54, an electron supply layer 56, a cap layer 58, a cathode electrode 60, an anode electrode 62, a gate electrode 64, an insulating layer 66, a cathode terminal 70, and an anode terminal 72.

[0033] The heterojunction diode 50 according to the comparative example is configured similarly to the above-described embodiment, but differs in the configuration of the gate electrode 64. The gate electrode 64 according to the comparative example is provided in a recess 68 penetrating an insulating layer 66. The recess 68 is provided only in the insulating layer 66, and is not provided in the electron supply layer 56. In other words, the thickness of the electron supply layer 56 is uniform.

[0034] In the comparative example, in order to make the base HEMT operate in a normally-off state, it is necessary to make the thickness of the electron supply layer 56 as small as possible so that a high concentration of 2DEG 74 is not generated in the channel layer 54 when the anode voltage is 0 V. For example, when the channel layer 54 is made of GaN and the electron supply layer 56 is made of AlGaN with an AlN mole fraction of 0.15, the normally-off operation can be realized by making the thickness of the electron supply layer 56 7 nm.

[0035] In the case of the heterojunction diode 50 based on the HEMT, when a negative voltage is applied to the anode terminal 72, a gate leakage current flows between the cathode electrode 60 and the gate electrode 64. At this time, electrons are trapped in the surface state existing at the interface between the electron supply layer 56 and the insulating layer 66 located between the cathode electrode 60 and the gate electrode 64, forming a surface depletion layer 76. Next, when a positive voltage is applied to the anode terminal 72, the electrons cannot reach from the cathode electrode 60 to under the gate electrode 64 without crossing the surface depletion layer 76 located between the cathode electrode 60 and the gate electrode 64. In the case of a normal HEMT, since a surface depletion layer is formed between the gate and the drain, the electrons can reach from the source to under the gate without crossing the surface depletion layer. Due to this principle, the heterojunction diode 50 is sensitive to the response of the surface state, and is more susceptible to current collapse than the HEMT. The current collapse is more likely to occur as the distance between the surface depletion layer 76 and the channel layer 54, i.e., the thickness of the electron supply layer 56, is smaller.

[0036] In order to suppress the current collapse, it is conceivable to shorten the gate-cathode distance L1 to reduce the influence of the surface depletion layer 76. However, when the gate-cathode distance L1 is shortened, the electric field strength between the cathode electrode 60 and the gate electrode 64 increases, and the reverse breakdown voltage of the heterojunction diode 50 decreases.

[0037] FIG. 3 is a graph showing the relationship between the gate-cathode distance L1 and the reverse breakdown voltage and forward current of the heterojunction diode 50 according to the comparative example. As shown in the figure, it can be seen that when the gate-cathode distance L1 is increased, the reverse breakdown voltage improves while the forward current decreases, resulting in a trade-off. In order to obtain a reverse breakdown voltage of 100V, the gate-cathode distance L1 needs to be about 4 μm, and in this case, the forward current is less than 0.2 A / mm. In the case of the comparative example, by setting the gate-cathode distance L1=1.1 μm, the characteristics of a reverse breakdown voltage of 40 V and a forward current of 0.36 A / mm are obtained. The maximum power density during half-wave rectification operation expected from this characteristic is 3.6 W / mm.

[0038] FIG. 4 is a graph showing the relationship between the gate-cathode distance L1 and the reverse breakdown voltage and forward current of the heterojunction diode 10 according to the embodiment. In the embodiment, even if the gate-cathode distance L1 is less than 4 μm, a reverse breakdown voltage of 100 V or more can be obtained. In the embodiment, the forward current tends to decrease as the gate-cathode distance L1 increases, but the embodiment obtains a forward current larger than that of the comparative example. In the embodiment, the gate-cathode distance L1 is set to 1.7 μm, and thus a reverse breakdown voltage of 100 V or more and a forward current of 0.5 A / mm can be obtained. The maximum power density during half-wave rectification operation expected from this characteristic is 12.5 W / mm or more. According to the embodiment, a performance significantly superior to that of the comparative example can be achieved.

[0039] 5 is a graph showing forward current-voltage characteristics of heterojunction diodes 10 and 50 according to an embodiment and a comparative example. The embodiment is shown by a solid line A, and the comparative example is shown by a dashed line B. As shown in the figure, the embodiment can increase the forward current compared to the comparative example.

[0040] 6 is a graph showing reverse current-voltage characteristics of heterojunction diodes 10 and 50 according to an embodiment and a comparative example. The embodiment is shown by a solid line A, and the comparative example is shown by a dashed line B. As shown in the figure, the embodiment can increase the reverse withstand voltage compared to the comparative example.

[0041] According to this embodiment, by providing the buried portion 24a of the gate electrode 24 in the recess 28 provided in the electron supply layer 16, the thickness of the first portion 16a of the electron supply layer 16 located directly below the buried portion 24a (or the recess 28) can be reduced. As a result, the concentration of the 2DEG 34 in the channel layer 14 located directly below the first portion 16a of the electron supply layer 16 can be reduced, and a normally-off operation can be realized. According to this embodiment, by increasing the thickness of the second portion 16b of the electron supply layer 16 located directly below the surface depletion layer 36 that may occur near the interface between the cap layer 18 and the insulating layer 26, the decrease in the concentration of the 2DEG 34 caused by the surface depletion layer 36 can be mitigated. As a result, the occurrence of current collapse can be suppressed and the forward current can be increased.

[0042] In this embodiment, the first portion 16a of the electron supply layer 16 is configured so that the threshold voltage Vth at which the channel layer 14 located directly below the first portion 16a is turned on is 0.1 V or more and 0.7 V or less, preferably about 0.4 V. When the heterojunction diode 10 is used for microwave rectification, the forward voltage Vf of the heterojunction diode 10 needs to be 0.1 V or more and 0.7 V or less, and is preferably about 0.4 V. This is because, when the forward voltage Vf is lower than 0.1 V, the channel cannot be sufficiently pinched off during reverse bias, and the reverse leakage current increases. Also, when the forward voltage Vf is higher than 0.7 V, the forward voltage amplitude cannot be sufficiently obtained, and the forward current If decreases.

[0043] On the other hand, the second portion 16b of the electron supply layer 16 is configured so that the threshold voltage Vth at which the channel layer 14 located directly below the second portion 16b is turned on is less than 0 V. By making the threshold voltage Vth of the second portion 16b of the electron supply layer 16 smaller than 0 V, it is possible to suitably suppress the occurrence of current collapse and increase the forward current.

[0044] The threshold voltage Vth can be adjusted by changing the product x·t of the AlN mole fraction x and the thickness t of the electron supply layer 16. FIG. 7 is a graph showing the relationship between the product x·t of the AlN mole fraction x and the thickness t of the electron supply layer 16 and the threshold voltage Vth. According to FIG. 7, the threshold voltage Vth is approximately proportional to the product x·t of the AlN ratio x and the thickness t of the electron supply layer 16. When the product x·t=1 nm, the threshold voltage Vth=0.4 V, when the product x·t=0.6 nm, the threshold voltage Vth=0.7 V, and when the product x·t=1.4 nm, the threshold voltage Vth=0.1 V. Therefore, the product x·t1 of the AlN mole fraction x and the thickness t1 of the first portion 16a of the electron supply layer 16 must be 0.6 nm or more and 1.4 nm or less, and is preferably about 1 nm.

[0045] Furthermore, when the product x·t=2 nm or more, the threshold voltage Vth becomes sufficiently smaller than 0 V. Therefore, the product x·t2 of the AlN mole fraction x and the thickness t2 of the second portion 16b of the electron supply layer 16 is preferably 2 nm or more. If the thickness of the electron supply layer 16 is increased beyond the critical film thickness, the quality of the electron supply layer 16 is degraded, so the thickness of the electron supply layer 16 is preferably equal to or less than the critical film thickness. The critical film thickness of the AlGaN layer formed on the GaN layer is about 100 nm when the AlN mole fraction x=0.1, and about 20 nm when the AlN mole fraction x=0.4.

[0046] For example, when the AlN molar fraction x of the electron supply layer 16 is 0.1, the thickness t1 of the first portion 16a of the electron supply layer 16 must be 6 nm or more and 14 nm or less, and is preferably about 10 nm. When the AlN molar fraction x of the electron supply layer 16 is 0.4, the thickness t1 of the first portion 16a of the electron supply layer 16 must be 1.5 nm or more and 3.5 nm or less, and is preferably about 2.5 nm. Therefore, when the AlN molar fraction x of the electron supply layer 16 is 0.1 or more and 0.4 or less, the thickness t1 of the first portion 16a must be 1.5 nm or more and 14 nm or less, and is preferably 2.5 nm or more and 10 nm or less.

[0047] Furthermore, when the AlN molar fraction x of the electron supply layer 16 is 0.1, the thickness t2 of the second portion 16b of the electron supply layer 16 is preferably 20 nm or more and 100 nm or less. When the AlN molar fraction x of the electron supply layer 16 is 0.4, the thickness t2 of the second portion 16b of the electron supply layer 16 is preferably 5 nm or more and 20 nm or less. Therefore, when the AlN molar fraction x of the electron supply layer 16 is 0.1 or more and 0.4 or less, the thickness t2 of the second portion 16b is preferably 5 nm or more and 100 nm or less.

[0048] According to this embodiment, current collapse can be effectively suppressed by forming a field plate 24b that extends from the gate electrode 24 of the heterojunction diode 10 toward the cathode electrode 20. In particular, current collapse can be effectively suppressed by forming a field plate that extends from the gate of the HEMT, which serves as the base, toward the source, contrary to the case of a normal HEMT.

[0049] The heterojunction diode 10 according to the present embodiment can be used in a rectifier circuit for converting microwave power to DC power, etc. According to the present embodiment, the maximum power density during half-wave rectification can be increased.

[0050] The present disclosure has been described above based on the embodiments. It will be understood by those skilled in the art that the present disclosure is not limited to the above embodiments, that various design changes are possible, that various modifications are possible, and that such modifications are also within the scope of the present disclosure. [Explanation of symbols]

[0051] 10...heterojunction diode, 12...substrate, 14...channel layer, 16...electron supply layer, 18...cap layer, 20...cathode electrode, 22...anode electrode, 24...gate electrode, 26...insulating layer, 28...recess.

Claims

1. a channel layer including a nitride semiconductor; an electron supply layer including a nitride semiconductor provided on the channel layer; a cathode electrode and an anode electrode provided on the electron supply layer; a recess provided on an upper surface of the electron supply layer located between the cathode electrode and the anode electrode; a gate electrode provided in the recess and electrically connected to the anode electrode; an insulating layer provided on the electron supply layer between the cathode electrode and the gate electrode and between the gate electrode and the anode electrode; the electron supply layer has a first portion between the channel layer and the recess and a second portion between the channel layer and the insulating layer, the first portion having a thickness smaller than a thickness of the second portion; The heterojunction diode, wherein the electron supply layer is an AlGaN layer, and the AlN mole fraction of the first portion is the same as the AlN mole fraction of the second portion.

2. A heterojunction diode as described in claim 1, wherein the thickness of the channel layer is not less than 100 nm and not more than 10 μm, the thickness of the first portion is not less than 1.5 nm and not more than 14 nm, and the thickness of the second portion is not less than 5 nm and not more than 100 nm.

3. A channel layer including a nitride semiconductor; an electron supply layer including a nitride semiconductor provided on the channel layer; a cathode electrode and an anode electrode provided on the electron supply layer; a recess provided on an upper surface of the electron supply layer located between the cathode electrode and the anode electrode; a gate electrode provided in the recess and electrically connected to the anode electrode; an insulating layer provided on the electron supply layer between the cathode electrode and the gate electrode and between the gate electrode and the anode electrode; the channel layer is a GaN layer, the electron supply layer is an AlGaN layer, the electron supply layer has a first portion between the channel layer and the recess and a second portion between the channel layer and an insulating layer, the product of the AlN mole fraction and the thickness of the first portion is 0.6 nm or more and 1.4 nm or less, and the product of the AlN mole fraction and the thickness of the second portion is 2 nm or more.

4. 4. The heterojunction diode according to claim 1, wherein a distance from the recess to the cathode electrode is not less than 0.5 μm and not more than 10 μm.

5. The heterojunction diode according to claim 1 , wherein the gate electrode has a field plate extending from the recess on the insulating layer toward the cathode electrode.

6. 6. The heterojunction diode according to claim 5, wherein the length of the field plate is not less than 0.1 μm and not more than 2 μm.

Images