JP2004134812A - Nitride compound semiconductor element - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To obtain a nitride compound semiconductor element which has an excellent light emitting intensity. <P>SOLUTION: The nitride semiconductor light emitting element which includes an active layer consists of a group III nitride semiconductor including gallium and indium, a first clad layer consists of a first conduction form nitride semiconductor including gallium, and a second clad layer consists of a second conduction type nitride semiconductor layer including gallium on a single crystal substrate, has a structure, in which a junction layer consists of a nitride compound semiconductor doped with indium is provided between a light emitting layer and the clad layer adjoinig the light emitting layer. <P>COPYRIGHT: (C)2004,JPO

Description

The present invention relates to a gallium nitride (GaN) -based compound semiconductor device having an active layer composed of a lattice-mismatched nitride compound semiconductor layer, and more particularly to a nitride compound semiconductor having an active layer (light-emitting layer) capable of increasing light emission intensity. It relates to a light emitting element.

Blue, nitride compound semiconductor light-emitting elements such as blue-green or green, etc. short-wavelength light emitting diodes (LED) and laser diodes (LD) has the general formula _{Al x Ga y In 1-xy} N (x + y = 1,0 ≤ x ≤ 1, 0 ≤ y ≤ 1). For example, a ternary or quaternary group III nitride compound semiconductor mixed crystal containing boron (B) as a group III element can also be an active layer. In a light emitting element such as an LED or an LD, the active layer is a light emitting layer. In addition, a nitride compound semiconductor containing, for example, gallium arsenide nitride (GaAsN) or a plurality of Group III elements and a plurality of Group V elements, such as gallium arsenide nitride (GaAsN), as a constituent element other than nitrogen (N). Is also a material that can be an active layer.

FIG. 1 illustrates the structure of a blue LED using a gallium indium nitride (Ga _x In ₁ -xN: x ≠ 1, 0 ≦ x <1) mixed crystal as a light emitting layer. The conventional blue LED has a buffer layer (102) of aluminum / gallium nitride (Al _x Ga _1-x N: 0 ≦ x ≦ 1), silicon (Si) or undoped (no addition) on a sapphire substrate (101). a lower cladding layer made of n-type gallium nitride (GaN) layer (103), indium composition ratio (x) of about 0.05 to 0.06 to n-type gallium indium nitride (Ga _1-x in x n ; 0 <x ≦ 1) active layer composed of a mixed crystal layer (104), annealing the magnesium doped during deposition (heat treatment) by an electrically activated p-type aluminum gallium nitride (Al _y Ga _{1 -y} N; 0 <y <1) A laminate made of a nitride compound semiconductor in which an upper clad layer (105) made of a mixed crystal and a contact layer (106) made of p-type gallium nitride (GaN) are sequentially deposited. Or It has been constructed. In the above laminated structure, the n-type gallium-indium nitride mixed crystal is used as a light-emitting layer that emits blue light. The light emitting layer is also doped with an acceptor impurity such as zinc (Zn) and a donor impurity such as silicon at the same time, though the mechanism of the light emission and its influence on the principle have not been elucidated yet. This is always the case (for example, see Patent Document 1). A gallium nitride-indium mixed crystal layer, which needs to contain a large amount of indium in gallium nitride to form a mixed crystal, is intentionally adopted as a light emitting layer because light emission that is considered to involve zinc impurities due to this mixed crystal formation is considered. This is because it is possible to shift to a blue light emission band in a wavelength band having high visibility (for example, see Patent Document 2).

Not only blue LEDs but also green LEDs having a light emission center wavelength of about 510 nanometers (nm) made of the nitride compound semiconductor laminate shown in FIG. 2 have recently been put into practical use (for example, Non-Patent Document 1). reference.). A green light-emitting LED composed of a nitride compound semiconductor laminated body is a pseudo-indirect transition type homojunction gallium phosphide (GaP) LED that includes conventional nitrogen as an isoelectric electron (isoelectronic) trap (for example, Non-Patent Comparing the light emission luminance with reference 2), a remarkable improvement of the light emission luminance of about 60 times has been achieved. A gallium-indium nitride mixed crystal (Ga _1-x In _x N: 0 <x ≦ 1) is also used as a light-emitting layer of a green LED that emits unprecedented high-intensity green light. In addition, it is a gallium nitride-indium mixed crystal whose indium mixed crystal ratio (x) is increased to about 0.20 as compared with the case of the blue LED. Therefore, the above blue, not only the green LED, blue-green or yellow LED etc. short-wavelength LED and a blue or green short wavelength LD also gallium indium nitride mixed crystal such as the _{(Ga 1-x In x N} : 0 <X ≦ 1) has been used as a light emitting layer that provides a high light emission output.

However, it is well known that forming a gallium-indium nitride mixed crystal layer involves great technical difficulties in crystal growth. In the case of a metal organic pyrolysis vapor deposition method generally used for growing a gallium indium / indium mixed crystal layer (generally abbreviated as MOCVD, MOVPE, OMCVD, etc.), excellent crystallinity is obtained. The range of the growth temperature at which a gallium-indium nitride mixed crystal layer can be obtained stably is limited to an extremely narrow range of about 750 ° C. to about 800 ° C. In addition, the incorporation rate of indium is extremely low, and the indium composition ratio does not change uniformly according to the concentration of the indium raw material added to the growth reaction system. In the MOCVD growth method, the incorporation rate is a ratio of the indium gas source to the total amount of the group III gas source generally added to the growth reaction system (usually expressed by moles). (Molar ratio of indium raw material) and the indium composition ratio in the gallium nitride-indium mixed crystal. This is the so-called ratio of the indium solid phase ratio (= indium composition ratio) to the indium gas phase composition ratio. The higher the rate of incorporation, the more efficiently indium is incorporated from the gas phase into the solid phase (mixed crystal). Means that. For example, when the growth temperature is 800 ° C., the incorporation rate of indium starts to increase sharply when the raw material molar ratio is about 0.9 or more (for example, see Non-Patent Document 3). Further, the instability in the mixed crystal growth is enhanced by the fact that the incorporation rate of indium into the layer fluctuates extremely due to a slight change in the growth temperature of about ± 10 ° C. What further hinders stable growth is that an appropriate indium composition ratio necessarily decreases with an increase in growth temperature. Suitability of a light-emitting element as an active layer (light-emitting layer) is usually determined by the intensity of photoluminescence (PL) light emission. The “appropriate” indium composition ratio referred to here is a composition ratio that causes photoluminescence emission. This “proper” indium composition ratio is a value that is almost uniquely determined with respect to the growth temperature, and even if the value is less than or greater than this value, no photoluminescence emission is obtained or the light is attenuated. In summary, there has been a problem in mixed crystal growth that extremely precise control of the indium composition ratio is required to obtain an appropriate indium composition ratio uniquely determined for a set growth temperature.

In the MOCVD method which is known as a general method for growing gallium-indium nitride mixed crystal, this mixed crystal uses trimethyl indium ((CH ₃ ) ₃ In: TMI) or the like as an indium supply source and nitrogen. It is customary to grow using ammonia as a source. However, a trialkylindium compound such as TMI exhibits Lewis acidity (see, for example, Non-Patent Document 4), and ammonia is a hydride of another group V element such as arsine (AsH ₃ ). It is the same Lewis basic substance as described above. As described above, one raw material compound has a property of supplying outer electrons to other molecules (electron donating property), and the other has a raw material composition composed of a combination of raw material compounds having the property of accepting the electron (electron accepting property). In the system, the molecules of the raw material compounds are easily bonded due to their strong attraction to each other. As a result, a polymer material that is difficult to dissociate is formed (for example, see Non-Patent Document 5). It is known that such a polymerization reaction in the gas phase occurs particularly remarkably in MOCVD growth of an indium-based compound using a trialkyl compound and a hydride of Group V (for example, Non-Patent Documents 6 and 10). reference.). If the indium raw material supplied into the growth reaction system to obtain a mixed crystal having a certain indium composition ratio is consumed to form such a composite, naturally, an indium-based mixed material having a desired indium composition ratio is used. This causes a problem that crystals cannot be obtained. In performing such growth of an indium-based mixed crystal, a complicated gas-phase reaction between raw materials also makes it more difficult to solve the above-mentioned problems associated with the prior art, particularly, instability of the indium mixed crystal ratio.

Also disclosed is a method for growing a nitride compound semiconductor layer doped with a group III element other than a constituent element such as indium in an atomic concentration range of 1 × 10 ¹⁷ cm ⁻³ to 7 × 10 ²² cm ^−3. (See, for example, Patent Document 3). As described above, for example, a gallium-indium nitride (Ga _1-x In _x N; 0 <x ≦ 1) mixed crystal provided as a light-emitting layer in a conventional nitride compound semiconductor laminate for a light-emitting element is composed of indium. The composition ratio (x) is set to about 0.05 to about 0.15 (for example, see Patent Document 4). For example, the concentration (atomic concentration) of indium atoms in an In _0.15 Ga _0.85 N mixed crystal having an indium composition ratio of 0.15 is calculated to be 6.6 × 10 ²¹ cm ⁻³ . This calculated value falls within the doping concentration range of indium atoms and the like in the above-described conventional example (for example, see Patent Document 5). That is, the doping concentration of indium in the range shown in the conventional example is sufficient to form an indium-containing nitride mixed crystal. For example, assuming that the gallium nitride layer is doped with the highest concentration of indium of 7 × 10 ²² cm ⁻³ , the amount calculated is not sufficient to form a gallium nitride-indium mixed crystal, It is an amount sufficient to be converted into indium (InN). Accompanying the high doping of indium, which can form a mixed crystal containing indium, is a problem in the growth technique which also causes instability of the indium composition ratio as described above. Even if the indium is doped at a high concentration, the problem that the indium is located at the interstitial position and does not form a mixed crystal remains similar, and the vapor phase of the trialkylindium raw material and the nitrogen hydride raw material is similar. It is the instability of the indium solid solution concentration based on the complexation reaction. The prior art (see, for example, Patent Document 6) does not disclose a method for solving these problems associated with high-concentration indium doping.

A light-emitting layer that can stably increase the light-emitting intensity in light-emitting devices even when low-concentration indium doping is performed, bypassing the difficulty of crystal growth of the indium-based nitride compound semiconductor mixed crystal layer. However, the effective range of the doping concentration of indium and the manner of distribution of the doping concentration have not yet been elucidated. Conventionally, the optimum doping concentration range of indium is defined only from the viewpoint of electrical characteristics based on a decrease in the free electron concentration in the growth layer or, conversely, an increase in the acceptor concentration. 6). It is said that when indium is added to a concentration of 1 × 10 ²¹ cm ⁻³ , the free electron concentration becomes the lowest at 1 × 10 ¹⁶ cm ⁻³ . On the other hand, it is said that the maximum acceptor concentration is obtained when indium is added to a concentration of 1 × 10 ²¹ cm ⁻³ (for example, see Patent Document 6). This only suggests that it is most effective to dope indium with an atomic concentration of 1 × 10 ²¹ cm ⁻³ .

Furthermore, the gallium / indium mixed crystal light emitting layer that has been conventionally employed is doped with a plurality of impurities composed of a combination of zinc, silicon, and the like (for example, see Patent Documents 1 and 7). Recently, a nitride compound semiconductor light-emitting device including, as a light-emitting layer, a nitride compound semiconductor layer doped with a group II element and a group VI element simultaneously has also been disclosed (for example, see Patent Document 8). This is because of the known phenomenon that the intensity of blue photoluminescence emission associated with zinc impurities doped in the gallium nitride layer increases in the coexistence of silicon (Si) (for example, Non-Patent Document 7). Is simply applied to a gallium nitride layer containing indium. This phenomenon of increasing the light emission intensity has already been applied to a MIS (metal-insulator-semiconductor) structure type LED using a gallium nitride layer doped with zinc and silicon as a light-emitting layer (for example, Patent Documents 9, 10, 11). reference.). In other words, in view of the distribution of the doping concentration of indium atoms to be provided to increase light emission, the distribution of these doped impurity elements is again optimized in relation to the distribution of indium atomic concentration. There is a need.

The magnitude of the light emission intensity strongly depends on the crystallinity of the light emitting layer. The blue or green LED uses sapphire as a substrate (see FIGS. 1 and 2). Gallium-indium nitride mixed crystal (In _x Ga ₁ -xN; 0 <x ≦ 1) does not lattice match with sapphire in the first place. The problems that arise when constructing a light emitting layer from a nitride compound semiconductor material that does not lattice match with sapphire, which is a practical substrate material, are dislocations and defects caused by lattice mismatch with the substrate material. To the light emitting layer. The propagation of such crystal defects lowers the crystal quality of the gallium-indium nitride mixed crystal used as the light emitting layer. As described above, the light emitting layer of the light emitting element having the conventional double hetero (DH) junction structure or the light emitting portion having the single quantum well structure does not lattice match with the sapphire substrate. In other words, the structure is not sufficient to sufficiently suppress the propagation of crystal defects due to lattice mismatch to the light emitting layer. A structure that is insufficient to suppress a decrease in crystallinity of the nitride compound semiconductor layer forming the light-emitting layer naturally impairs the improvement of light-emitting characteristics and the like.

If a crystal defect caused by lattice mismatch causes deterioration of the crystal quality of the light emitting layer, using a material lattice-matched with the light emitting layer as the substrate can prevent the deterioration of the crystal quality of the light emitting layer. The way is conceived. Upon this example configuration, forming the light emitting layer of aluminum nitride gallium indium mixed crystal _{(Al x Ga y In z N} (x + y + z = 1,0 ≦ x, y, z ≦ 1), emitting a substrate crystal There is a light emitting element such as manganese oxide (MnO), zinc oxide (ZnO), or calcium oxide (CaO) that lattice-matches with the layer material (for example, see Patent Document 12). Sublimates and thermally decomposes at a high temperature generally exceeding about 600 ° C. Therefore, although there is some difference in the film forming temperature depending on the film forming method, a nitride compound requiring a high film forming temperature of about 700 to 1150 ° C. As a substrate for growing a semiconductor layer, it is not practically usable due to its thermal instability, especially a nitride compound semiconductor of aluminum-gallium nitride (Al _x Ga ₁ -xN; 0 <x ≦ 1) mixed crystal. of In the case of metal organic chemical vapor deposition (MOCVD), which is generally used as a film method, the film formation temperature is generally about 1000 ° C. to 1150 ° C. At such a high temperature, it is even more practical to use the above-mentioned substrate material. This is due to the fact that sapphire (α-Al ₂ O ₃ single crystal) is used exclusively, despite the disclosure of a substrate material that results in a lattice mismatched configuration as described above. In other words, although the emission characteristics such as an increase in emission intensity are expected to be improved by using a nitride compound semiconductor layer lattice-matched to the substrate as the light-emitting layer, the lattice matching is expected to be improved. Due to restrictions such as heat resistance on a substrate material selected to obtain a system laminate structure, in reality, improvement in light emission characteristics has not yet been stably realized.

As a practical alternative to using a substrate material having a problem with thermal resistance to form a laminate from a nitride compound semiconductor layer that lattice-matches with the substrate, a certain part of the laminate, for example, a light-emitting portion is lattice-matched. In some cases, it is composed of systems. The propagation of the light-emitting layer of crystal defects such as dislocation due to lattice mismatch with the intention to mitigate even _{slightly, Al x Ga y In z N} (x + y + z = 1,0 ≦ x, y, z ≦ 1 ) has a light-emitting layer and the composition ratio of lattice matching, and prohibited from emitting layer band width (band gap) is large _{Al x Ga y In z N (} x + y + z = 1,0 ≦ x, y, z ≦ 1) This corresponds to an example in which a light-emitting portion is formed by forming a hetero-junction with a mixed crystal layer as an upper clad layer and a lower clad layer (for example, see Patent Document 12). However, the formation of the lattice-matched light-emitting portion is based on the premise that it is formed on a thermally-modified oxide-based substrate such as the above-described zinc oxide or the like, and is not practical. .

There is also a configuration example in which a light emitting layer is bonded on a nitride compound semiconductor layer lattice-matched with the light emitting layer. This example can be seen in an LED with a MIS (metal-insulator-semiconductor) structure that includes a gallium nitride homo junction. Further, an LED having a homojunction configuration in which an In _x Ga ₁ -xN (0 ≦ x ≦ 1) layer is deposited on a gallium indium nitride (In _x Ga ₁ -xN; 0 ≦ x ≦ 1) layer is also available. (For example, see Patent Document 13). In this configuration, if the indium composition ratio (x) of both the gallium-indium nitride (In _x Ga ₁ -xN; 0 ≦ x ≦ 1) layers is the same, the nitride layer that lattice-matches the light-emitting layer is formed. And a homojunction structure formed on the compound semiconductor layer. However, it is a well-known fact that the homojunction mode is disadvantageous in principle as compared with the hetero junction mode in obtaining an LED having excellent emission intensity. In order to obtain a light emitting element such as an LED having a high light emission intensity, it is essential to employ a hetero junction structure.

It is well known that a light emitting section of a double hetero junction type is advantageous for obtaining a light emitting element with high light emission intensity. In addition to improving the crystallinity of the nitride compound semiconductor layer constituting the light emitting layer, the semiconductor layer forming the junction with the light emitting layer has good crystallinity if the semiconductor layer is a layer with excellent crystallinity. It is effective in obtaining a light emitting layer. In the double heterojunction structure, the light-emitting portion including the light-emitting layer and the bonding layer to be bonded to the light-emitting layer is formed of a nitride compound layer having excellent crystallinity, which is useful for providing a light-emitting element with high light-emission intensity. . In the above-mentioned conventional technique (for example, see Patent Document 6) in which improvement in crystallinity is achieved by adding a Group III element other than the constituent element having an atomic radius larger than that of the constituent element, for example, indium is nitrided. It is described that the crystallinity can be improved by including the gallium-based layer. There is an example in which a gallium nitride-indium mixed crystal layer is actually arranged in a part of a laminate for a light emitting element (for example, see Non-Patent Document 8). Specifically, in this light emitting device, a gallium-indium (Ga _0.8 In _0.2 N) mixed crystal having an indium mixed crystal ratio of 0.2 is used as a well layer, and the indium composition ratio is set to _0.1 . A laser diode having a light-emitting portion composed of a heterojunction multiple quantum well structure (abbreviated as MQW) using a gallium-indium nitride mixed crystal (Ga _0.95 In _0.05 N) as a barrier layer (barrier layer), which is referred to as “05”. It is. Here, a gallium nitride-indium mixed crystal (Ga _0.9 In _0.1 N) having an indium composition ratio of 0.1 is disposed between the n-type gallium nitride layer and the MQW light emitting portion having the double hetero junction. (For example, see Non-Patent Document 8). The gallium-indium nitride mixed crystal layer (Ga _0.9 In _0.1 N) is arranged with the intention of improving the crystallinity of a laminate composed of a nitride compound semiconductor deposited above the same layer (for example, , Non-Patent Document 9.). However, a gallium-indium nitride (Ga _0.9 In _0.1 N) mixed crystal layer, which is believed to bring about the improvement in crystallinity, is arranged as a bonding layer to be bonded to the light emitting layer in the light emitting section composed of a double hetero junction. It is not. Between the Ga _0.9 In _0.1 N mixed crystal layer and the light emitting portion, an aluminum gallium nitride mixed crystal layer (Al _0.15 Ga _0.85 N) having an aluminum composition ratio of 0.15 and an n-type gallium nitride layer are inserted. Have been. These inserted nitride compound semiconductor layers have no lattice matching with the underlying gallium nitride-indium mixed crystal (Ga _0.9 In _0.1 N) layer and the gallium nitride-indium layer constituting the MQW of the light emitting portion. . Therefore, by laying the Ga _0.9 In _0.1 N mixed crystal layer, the crystallinity of the nitride compound semiconductor layer deposited on the Ga _0.9 In _0.1 N mixed crystal layer is improved. If there is a lattice mismatch, the dislocation generated by the lattice mismatch actually propagates to the light emitting portion. As described above, in the conventional laminated body for a light-emitting element using a lattice-mismatched system, the indium-containing mixed crystal layer whose crystallinity has been improved by containing indium has a double heterojunction light emitting structure. In some parts, it is not used as a bonding layer with the light emitting layer. That is, such an arrangement of the indium-added layer found in the conventional example has a junction structure that directly affects the crystallinity of the light emitting layer or the nitride compound semiconductor layer constituting the double hetero junction light emitting portion. Has not been reached.
JP-A-6-260680 JP-B-55-3834 S. Nakamura et al., Jpn. J. Appl. Phys. , Vol. 34, Part. 2, No. 10B (1995), L1332 to L1335. Supervised by Toshiji Fukami, "Semiconductor Engineering" (Tokyo Denki University Press, March 20, 1993, 7th edition), 195 pages Appl. Phys. Lett. , 59 (18) (1991), 2251. G. FIG. E. FIG. COATES, J.A. Chem. Soc. , [1951] (1951), 2003. J. Stensen et al., Surf. Sci. , 279 (1992), 272. Etc. reference R. DICHENKO et al. Inorg. Nucl. Chem. , Vol. 14 (1960), 35-37. See JP-A-5-243614 JP-A-6-209121 JP-A-5-243614 JP-A-5-243614 JP-A-6-260681 JP-A-7-249796 SolidStateCommun. , 57 (6) (1991), 405-409. JP-A-3-10665 JP-A-3-10666 JP-A-3-10667 Japanese Patent Publication No. 6-14564 JP-A-3-203388 Jpn. J. Appl. Phys. , Vol. 35, Part 2, No. 1B (Jan. 15, 1996), L74-L76. NIKKEI ELECTRONICS 1996.1.15. no. 653, January 15, 1996, published by Nikkei BP, pages 13-15 ElectronLett. , Vol. 16, no. 6 (1980), 228.

(4) The active layer is responsible for the functions and characteristics of the semiconductor device. For example, an effective means for obtaining a light-emitting element having excellent light-emitting intensity is to employ a structure of a light-emitting layer or a light-emitting portion that directly increases light-emitting intensity. As described above, the conventional measures for improving the emission intensity have common points. That is, a nitride compound semiconductor mixed crystal to which indium is added is used as a light emitting layer or as an intermediate insertion layer which improves the crystallinity of a layer constituting a laminate.

A conventional indium-containing gallium nitride layer used as a blue light emitting layer is generally a mixed crystal layer having an indium composition ratio of about 10%. As described above, there is a problem in crystal growth in forming a gallium-indium nitride mixed crystal layer having excellent quality. In order to obtain emission in a green band having a longer wavelength than in a blue band, it is necessary to further increase the indium composition ratio. It is necessary to lower the growth temperature from the relation of the incorporation rate of indium so as to form a mixed crystal layer having a high indium composition ratio. There is a problem that the crystallinity of the In _x Ga ₁ -xN mixed crystal is impaired as the crystal is grown at a lower temperature. 27), pages 1-6).

In other words, the situation is the same when a crystal growth method of a nitride compound semiconductor layer that improves crystallinity by adding indium (see Japanese Patent Application Laid-Open No. 5-243614) is the same, and it is sufficient to form an indium-containing mixed crystal. As long as it is intended to form a semiconductor layer to which indium is added to such a degree, low reproducibility of the indium composition ratio of the obtained mixed crystal is a problem. Within the range of the amount of indium allowed in the conventional example (see Japanese Patent Application Laid-Open No. 5-243614), it is estimated that there is no need to change the growth temperature, which is approximately 10 ¹⁷ cm ^-3 to 10 ¹⁹ cm. ^It has been clarified by the present inventors that the inventors of the present invention have conducted intensive studies that doping of indium with a low atomic concentration of about ^-3 does not cause a remarkable increase in emission intensity. This is a method of growing an indium-doped nitride compound semiconductor layer, as disclosed in the prior art (see Japanese Patent Application Laid-Open No. 5-243614), which improves the characteristics of a light emitting device such as an increase in light emission intensity. This is because an optimal method of doping with indium having an effect on the above is not proposed.

As described above, in an LED made of a gallium nitride-based semiconductor material, the light emitting layer is generally doped with an impurity of Group II, Group IV, Group VI, or the like. However, insufficient control of the quantitative balance of the donor or acceptor impurity is accompanied by instability in which it is not always possible to increase the light emission from the nitride compound semiconductor layer (Japanese Patent Application Laid-Open No. Hei 4-10666 and Japanese Patent Application Laid-Open No. H10-6666). See JP-A-4-10667). In particular, a so-called electron-rich n ^{+ -type} light-emitting layer with a high carrier concentration doped with a donor impurity far exceeding the concentration of an acceptor impurity only causes weak light emission. In order to form a light-emitting layer with excellent crystallinity by doping with indium, optimization of the optimum doping concentration range and concentration distribution of indium that can contribute to an increase in light emission intensity in consideration of the quantitative balance of donor and acceptor impurities is necessary. Further, in addition to optimizing the indium doping, a bonding system between the indium-doped nitride compound semiconductor layer and the light-emitting portion is provided to prevent a decrease in crystallinity of the light-emitting layer due to lattice mismatch. Is also a problem to be solved by the present invention.

That is, the present invention provides an active layer made of a gallium (Ga) -based group III nitride compound semiconductor on a crystal substrate, a first nitride compound semiconductor layer of a first conductivity type, and an active layer of a second conductivity type. In a nitride compound semiconductor device having a junction structure composed of the second nitride compound semiconductor layer and the second nitride compound semiconductor layer, the active layer has an atomic concentration in the range of 8 × 10 ¹⁷ cm ^{−3 to} 4 × 10 ²⁰ cm ^−3. It is intended to provide a nitride compound semiconductor element which is a gallium (Ga) -based group III nitride compound semiconductor doped with indium (In) with a concentration distribution in one direction of the layer thickness. In particular, (a) a gallium-based group III nitride having a concentration distribution in which the atomic concentration of indium is increased toward a junction interface with the first or second nitride compound semiconductor having a smaller lattice constant as the active layer. Compound semiconductor layer, (ii) active layer doped with donor or acceptor impurities, one of the donor and acceptor impurities is doped to a substantially constant atomic concentration in the layer thickness direction, and the other impurity is doped. An object of the present invention is to provide a nitride compound semiconductor device provided with a gallium-based group III nitride compound semiconductor layer which is doped in a layer thickness direction with a concentration distribution substantially inversely proportional to the atomic concentration of indium. In addition, from the viewpoint of the junction mode, one of the first and second nitride compound semiconductor layers that joins the active layer and the active layer is made of the same nitride compound semiconductor material as the active layer doped with indium. A nitride compound semiconductor device is provided. In addition, the first or second nitride compound semiconductor layer doped with indium substantially uniformly is bonded to the nitride compound semiconductor layer doped with indium. An element is provided.

In a light emitting device made of a nitride compound semiconductor, there is an effect of increasing light emission intensity.

In the present invention, sapphire (α-Al ₂ O ₃ single crystal) or the like which has no lattice matching with the nitride compound semiconductor forming the active layer but has excellent heat resistance can be used as a substrate crystal. . Hexagonal and cubic crystal materials such as silicon carbide (SiC) and silicon (Si) can also be used as before. There are no particular restrictions on the specifications such as the plane orientation, off angle, and substrate thickness of these substrate crystals. In the present invention, the degree of lattice mismatch is represented by the ratio of the difference between the lattice constant of the deposited layer and the lattice constant of the crystal layer constituting the deposited layer to the lattice constant of the crystalline material constituting the deposited layer. For example, if the lattice constants of the substrate (reference crystal) and the crystal material (comparative crystal) constituting the active layer such as the light emitting layer and the electron channel layer are a ₀ and a, respectively, the lattice constant of the reference crystal (a ₀ The ratio of the absolute value of the difference between the lattice constants of the reference crystal and the comparison target crystal to the reference crystal represents the degree of lattice mismatch (δ; unit%), and is given by the following equation (1).
δ (%) = ｛| a−a ₀ | / a ₀ } × 100 (1)
This δ is calculated using the lattice constant of the basic axis of the crystal such as the a-axis. The lattice spacing determined in consideration of the orientation relationship (orientation) between the substrate and the active layer is not used for calculating δ. The lattice constant used for the calculation of δ is the original lattice constant of those materials that have been published, for example, in a distorted state that has undergone some mechanical deformation at the junction interface of a lattice-mismatched system. It is not the lattice constant when it is below. In the present invention, the case where δ between the substrate crystal and the active layer exceeds approximately 0.2% is defined as a lattice mismatched active layer. For example, a laminate made of a nitride compound semiconductor which has a lattice mismatch with the sapphire substrate shown in FIG. 1 is a typical laminate made of a lattice mismatch active layer in the present invention.

In the above description, the deposited layer forming the active layer or the like refers to a gallium-based nitride compound semiconductor layer in the present invention. The gallium-based nitride compound semiconductor refers to a nitride compound semiconductor containing a group III element other than indium such as gallium and aluminum as a constituent element. For example, gallium nitride (GaN), aluminum nitride (AlN), aluminum-gallium nitride (Al _x Ga ₁ -xN; 0 <x ≦ 1) mixed crystal, and the like are applicable. Further, the kind of nitride compound semiconductor doped with indium is not limited to gallium nitride-based or aluminum nitride-based semiconductor. The effect of indium doping according to the present invention also applies to a nitride compound semiconductor material containing a Group V element other than nitrogen, such as gallium arsenide nitride (GaNAs), aluminum arsenide nitride (AlNAs), and gallium phosphide nitride (GaNP). Is revealed. Therefore, the nitride compound semiconductor device according to the present invention is formed of a stacked body including, for example, indium-doped gallium nitride, aluminum nitride, gallium arsenide nitride, aluminum-gallium nitride mixed crystal, or the like as an active layer. The conduction type of the indium-doped nitride compound semiconductor layer forming the active layer is preferably an n-type or an insulation type. Here, in a single or multiple quantum well structure, a well layer, that is, a well layer is regarded as an active layer.

According to the present invention, even in a gallium-based group III nitride compound semiconductor layer having a lattice mismatch with a substrate, light emission from the same layer can be performed without intentionally forming a mixed crystal by doping an appropriate concentration of indium. This is based on the fact that the strength can be increased. As an example, FIG. 3 shows the dependence of photoluminescence (PL) intensity on the indium doping concentration in a zinc-doped gallium nitride (GaN) layer doped with indium. As a doping source of indium, cyclopentadienylindium having a monovalent valence having a Lewis basic property capable of avoiding a complexing reaction with ammonia gas which is a nitrogen supply source during growth of a gallium nitride layer is used. (C ₅ H ₅ In) are available (J.ofCryst.Growth., 106 (1990) , see 260. and KOKOKU No. 8-17160 Publication). As a method for quantifying the atomic concentration of doped indium, various physical analysis methods such as Auger electron spectroscopy (AES) can be considered. However, since the concentration of doped indium is in a low concentration range, the characteristic X EPMA analyzing based on the detected intensity of the line (E lectron P robe M icro- a nalysis) rather than analytical techniques such as secondary ion mass spectrometry excellent in capability of detecting low concentrations of atoms here (abbr SIMS) Quantified by As shown in FIG. 3, blue (blue-violet) emission having a wavelength of about 430 nm related to zinc impurities from a gallium nitride layer excited by a helium (He) -cadmium (Cd) laser beam having a wavelength of 325 nm. The luminescence intensity at room temperature strongly depends on the doping concentration of indium. When the atomic concentration of the doped indium is approximately 5 × 10 ¹⁷ cm ⁻³ , the effect of increasing the emission intensity due to the indium doping is unstable. In other words, the luminescence intensity does not always increase due to indium doping. When the indium doping concentration reaches 8 × 10 ¹⁷ cm ⁻³ , a stable increase in luminescence intensity is observed. For example, the blue (blue-violet) emission intensity associated with zinc impurities from a gallium nitride layer doped with zinc at a concentration of about 6 × 10 ¹⁹ cm ^-3 can be obtained by converting indium to a concentration of about 5 × 10 ¹⁸ cm ^-3 . Doping increases about 60 times. On the other hand, when the zinc-added layer is doped with silicon (Si) using disilane (Si ₂ H ₆ ) gas to form a light-emitting layer in which a conventional donor and acceptor coexist, the luminescence intensity is about 5 times at most. It just becomes. That is, the effect of indium doping in an appropriate concentration range on the increase in luminescence intensity from the gallium nitride layer is orders of magnitude greater than conventional doping of both donor and acceptor impurities. However, if the indium is excessively doped, the luminescence intensity is reduced (see FIG. 3). The luminescence intensity sharply decreases when the indium doping concentration exceeds 4 × 10 ²⁰ cm ⁻³ . Therefore, the optimum range of the doping concentration of indium when the emission intensity is intended to be increased is 8 × 10 ¹⁷ cm ⁻³ or more and 4 × 10 ²⁰ cm ⁻³ or less. This optimum concentration range is extremely limited as compared with the range of the conventional example. Indium is a Group III element, and indium doped as a homologous element in a Group III nitride compound semiconductor may possibly act as an isoelectronic trap. When the doping concentration of indium is low, it is considered that indium is not doped enough to remarkably improve the crystallinity, but it is also considered that the density of electron traps is small, such as an increase in emission intensity. Presumed. On the other hand, it is presumed that the upper limit of the indium doping concentration is rather closely related to the reduction of the electron concentration in the layer due to the indium doping. When the doping amount of indium is increased, the density of the isoelectronic trap is increased, but the electron concentration in the layer is reduced as seen in the conventional example (see Japanese Patent Application Laid-Open No. 5-243614). That is, it is considered that the upper limit of the indium doping concentration is determined by the residual electron concentration rather than from the viewpoint of improving the crystallinity.

In the atomic concentration range (doping concentration range) of indium disclosed in the present invention, the emission wavelength at the band edge inherent to the nitride compound semiconductor (corresponding to the band gap (energy gap) inherent to the semiconductor material) is changed. Absent. For example, the wavelength (approximately 365 nm) at which the band edge emission of the indium-doped gallium nitride layer according to the present invention appears coincides with that of the indium-doped gallium nitride layer. The more the mixed crystal is formed, that is, if a large amount of indium is added, a spectrum peculiar to the near ultraviolet region of about 380 to about 390 nm appears. 28a-ZB-1 (Lecture Proceedings No. 1, p. 292) and Lecture No. 28a-ZB-2 (Lecture Proceedings No. 1, p. 293) The near-ultraviolet spectrum of the near-ultraviolet spectrum was determined by adding acceptor impurities such as zinc. Hardly appear when the carrier is present, and easily appear when a donor such as silicon is added or when the carrier concentration is relatively high even when undoped, but the indium doping according to the present invention is applied. from the n ⁺ -type GaN layer of high carrier concentration, the luminescence emission in the near ultraviolet region were observed. in addition, including a donor impurity such as silicon at a high concentration In gallium indium nitride, it is clearly observed that the emission wavelength of this near-ultraviolet spectrum shifts to the longer wavelength side as the content of indium increases, however, indium in various concentrations within the concentration range of the present invention. In the added gallium nitride layer, since the near-ultraviolet spectrum did not appear, the shift to the longer wavelength side of the near-ultraviolet spectrum depending on the indium concentration was not observed, that is, doping of indium in the concentration range described in the present invention. Gallium nitride has not changed to gallium nitride-indium mixed crystal since it has not changed the forbidden band width. It is reasonable to consider that the mixed layer with indium does not occur in the doped layer. While avoiding the problem of forming an indium-based mixed crystal accompanied by difficulties in crystal growth as described above, the present invention also provides a technique for easily achieving an increase in emission intensity. This is one of the advantages of the invention.

In the present invention, the gallium-based nitride compound semiconductor layer doped with indium uniformly in the thickness direction within the above concentration range is not used as the active layer. A gallium-based nitride compound semiconductor layer having a specific indium concentration distribution while limiting the indium concentration range that causes an increase in emission intensity is used as the active layer. By imparting non-uniformity to the indium doping concentration distribution, a distribution is created in the electronic properties and the like inside the active layer. In particular, indium is present in the gallium-based nitride compound semiconductor layer that forms a junction with the active layer, and increases the doping concentration of indium toward the interface with the nitride compound semiconductor layer having a smaller lattice constant. Increasing the indium doping concentration in the layer thickness direction can be achieved by increasing the amount of the indium source supplied to the growth reaction system during the formation of the active layer with respect to the film formation time. Various concentration profiles can be created depending on how the indium source supply changes with respect to the film formation time. FIG. 4 illustrates an indium doping concentration distribution according to the present invention. The atomic concentration profile (111) in the depth direction indicates that the indium concentration in the active layer (104) is higher in the bonding layers ((108) and (109)) that are bonded to the active layer (104). 3 shows an example of a distribution in which the density monotonically increases toward the interface (110) with the bonding layer (109). Such a concentration profile can be obtained by monotonically increasing the supply amount of the indium source supplied to the growth reaction system as the thickness of the active layer increases. The atomic concentration profile (112) shows another concentration distribution of indium atoms, and the doping concentration of indium is gradually increased toward the interface (110) with the bonding layer (109) from the start of the growth of the active layer (104). It is increased. The atomic concentration profile (113) shows the distribution of the atomic concentration of indium obtained by keeping the doping amount of indium constant in the middle of the active layer (104) and thereafter increasing the doping amount stepwise. If the indium doping amount is monotonically increased rather than stepwise, a profile (114) in which the atomic concentration of indium smoothly increases toward the interface (110) is obtained. In any case, in order to obtain a light emitting layer having excellent light emission intensity, the atomic concentration of indium is gradually increased toward the interface with the bonding layer having a smaller lattice constant in the bonding layer bonded to the light emitting layer. That is no different. The reason why the distribution is given to the doping concentration is to introduce a large amount of strain into the lattice strain region existing near the interface with the bonding layer, which increases the degree of lattice mismatch (the above δ (%)). is there. Further, in the as-grown state, the region near the interface is set as a region having a lower electron concentration.

The change range of the indium doping concentration is in the range of 8 × 10 ¹⁷ cm ^{−3 to} 4 × 10 ²⁰ cm ⁻³ . The rate of change of the indium doping concentration within this range is appropriately determined in consideration of the desired emission wavelength and the thickness of the active layer. The rate of change, by the following equation (2), the initial doping concentration of indium represented by a ratio of; (unit cm ^-3 N) (delta) final indium concentration on (N ₀ units cm ^-3).
Δ = N / N ₀ Equation (2)
Specifically, referring to FIG. 5, the bonding layers ((108) and (109)) that are bonded to the active layer (104), and the bonding layer (108) that reduces the difference in lattice constant from the active layer. The atomic concentration of the doped indium at the interface (124) is the initial concentration (N ₀ ). On the other hand, the concentration at the interface (110) with the bonding layer (109) that increases the degree of lattice mismatch with the active layer is N. In the present invention, since the doping concentration of indium is increased toward the interface (110) of the two interfaces ((110) and (124)), there is always a relationship of N> N ₀ , and thus Δ is positive. Take the value of. When the thickness of the active layer is constant, the rate of change is relatively small in order to obtain emission of a short wavelength such as blue. If light emission on the long wavelength side such as green is intended, the rate of change is large. Conversely, if the rate of change is constant, the thickness of the light-emitting layer is reduced to such an extent that light emission on the long wavelength side is green and yellow. That is, the degree of change in the atomic concentration of indium with respect to the change in the layer thickness is increased by narrowing the region where the indium doping concentration is changed. For example, consider the case where the initial doping concentration (N ₀ ) of indium is 1 × 10 ¹⁸ cm ⁻³ , the final concentration (N) is 4 × 10 ²⁰ cm ⁻³ , and the rate of change (Δ) is 400. In this case, in order to obtain blue light emission having a wavelength of 430 nm to 450 nm, the thickness of the light emitting layer is generally set to 40 to 70 nm, and to achieve green band light emission having a wavelength of about 500 to 550 nm, the layer thickness of the light emitting layer is intended. Is at least about 10 nm or less, preferably about 5 nm.

(1) One advantage brought about by giving a distribution to the doping concentration of indium in the active layer is that the configuration of the light emitting section can be simplified. For example, omitting the structure of a conventional green LED, the light emitting portion has a single quantum well structure. The well layer serving as the light emitting layer is a gallium nitride / indium mixed crystal layer ((104) in FIG. 2) having an indium composition ratio of 0.20. Below the gallium nitride / indium mixed crystal light emitting layer ((104) in FIG. 2), a gallium / indium nitride mixed crystal layer ((107) in FIG. 2) having an indium composition ratio of 0.05 is disposed as a base layer. Have been. In other words, the light emitting portion has a multilayer structure of gallium nitride / indium mixed crystal layers having different indium composition ratios. Indium doping according to the present invention, that is, a gallium-based group III nitride semiconductor layer having a layer thickness suitable for green light emission and doped to increase the indium atom concentration sequentially in the layer thickness direction is used as the only light emitting layer. That's enough. As described above, in the nitride layer in which the indium doping concentration is changed, only one region having a relatively low indium concentration is formed in a conventional gallium nitride / indium underlayer, and the indium atom concentration is relatively low. The high region can carry each of the conventional gallium-indium nitride light-emitting layers. Therefore, even if gallium-indium nitride mixed crystal, in which crystal growth is difficult, is not intentionally arranged as the underlayer of the light emitting layer, the same increase in light emission intensity can be easily achieved. That is, the formation of a quantum level, which is expected to occur at a junction interface as a barrier layer of a quantum well structure or when a gallium nitride-indium mixed crystal light emitting layer and an n-type gallium nitride layer are directly joined. It is not necessary to provide a gallium / indium nitride underlayer as a layer for alleviating lattice distortion in order to reduce extreme accumulation of carriers.

According to the present invention, the active layer is doped with one of the donor and acceptor impurities at a substantially constant atomic concentration in the layer thickness direction, and the other impurity is doped in the layer thickness direction with a concentration distribution substantially inversely proportional to the indium atomic concentration. It is assumed that This is a gallium nitride in which an acceptor impurity such as zinc is doped as an active layer with a substantially uniform atomic concentration in a layer thickness direction, and a donor impurity such as silicon is doped with a concentration almost inversely proportional to an indium atomic concentration. This means that a compound semiconductor layer is used. That is, in the present invention in which the atomic concentration of indium is gradually increased in the thickness direction of the active layer, sulfur (S) of a Group IV element or a Group VI element such as silicon, germanium (Ge) or tin (Sn) is used. Conversely, the atomic concentration of donor impurities such as selenium (Se) and tellurium (Te) decreases in the direction of increasing the layer thickness. The reason for providing such a contradictory atomic concentration distribution is to create a concentration distribution of carriers, that is, electrons or holes in the active layer. However, the concentrations that are kept in a substantially inverse proportion to each other are the atomic concentrations of the dopant, not the carrier concentrations such as the electron concentration. When the atomic concentration of the donor impurity gradually decreases in the layer thickness direction of the active layer, the electron concentration increases in the direction of increasing the layer thickness. In particular, the distribution may be opposite to the distribution of the donor impurity which rapidly increases at the interface with the lattice mismatched bonding layer bonded to the active layer. Contrary to the increase in the atomic concentration of indium accompanying the increase in the thickness of the active layer, the decrease in the atomic concentration of the donor impurity to be doped is caused by the fact that the lattice strain near the interface with the bonding layer deposited on the active layer is completely reduced. This is because the purpose is to increase the carrier density of electrons and the like accumulated in the region that has not been removed. In the as-grown state, the carrier density may be increased near the interface in some cases. In the case where a piezoelectric crystal such as an aluminum nitride-gallium mixed crystal is bonded to the active layer, the carrier density in the vicinity of the interface is more remarkable. In some cases. In this way, the intention is to form a so-called carrier-localized region in which carriers are confined in the vicinity of the interface with the active layer. The device achieves an increase in carrier mobility.

In the present invention, the joint structure of the light emitting unit is creatively added. The light emitting portion refers to a portion composed of a light emitting layer and a clad layer in a heterojunction configuration including a clad layer. In the quantum well structure, it refers to a portion composed of a multi-layered structure composed of a well layer and a barrier layer and a clad layer joined thereto. If only the light emitting element is mentioned, in the present invention, the gallium nitride compound semiconductor layer doped with indium is further joined to the gallium nitride compound semiconductor layer doped with indium to form the light emitting portion. Form. Other elements, such as a heat-resistant Hall element, a Schottky junction field effect transistor (abbreviated as MESFET), and an ultraviolet element, as well as a configuration including a junction between nitride compound semiconductor layers doped with indium. It can be used as an active region of a light receiving element or the like in a region or a blue band region. The reason why the indium-doped nitride compound semiconductor bonding layer is bonded to the active layer is that the bonding serves as an underlayer (deposited layer) that improves the crystallinity of the gallium-based nitride compound semiconductor layer that forms the active layer. This is because an active layer having significantly improved crystallinity can be obtained. The degree of improvement in the crystallinity of the active layer and the quality of the active layer can be determined by examining the half-width of the X-ray diffraction spectrum and the emission intensity at the band edge of the photoluminescence (PL) spectrum. In the present invention, since indium is not doped at such a high concentration as to achieve the mixed crystal of the material to be doped, the wavelength shifts to a band edge emission wavelength corresponding to the band gap inherent to the material to be doped (so-called wavelength shift). Does not occur, the intensity and the half-value width of the band edge emission can be directly used as an index for determining the quality of the crystal quality. FIG. 6 shows, as an example, a room-temperature photoluminescence spectrum (115) of a gallium nitride layer deposited on an n-type gallium nitride (GaN) junction layer doped with indium at an atomic concentration of about 8 × 10 ¹⁵ cm ^−3. . On the other hand, the spectrum (116) shows the room temperature spectrum from the gallium nitride layer deposited on the undoped n-type gallium nitride layer not doped with indium. As is evident from the comparison of the two spectra, the emission intensity of the band edge emission (indicated by the symbol BD in FIG. 6) from the gallium nitride layer on the indium-doped gallium nitride junction layer is higher than that without indium. Is clearly increasing. As described above, the crystallinity of the nitride compound semiconductor layer bonded to the nitride compound semiconductor layer doped with indium is remarkably improved.

In the gallium-based group III nitride compound semiconductor layer to be joined to the active layer, the optimum indium doping concentration range is 8 × 10 ¹⁷ cm ⁻³ or more and 4 × 10 ²⁰ cm ⁻³ or less. If the atomic concentration of indium is less than about 8 × 10 ¹⁷ cm ⁻³ , no increase is observed in the room-temperature photoluminescence intensity at the band edge of the layer to be doped. When the concentration of the doped indium atom is around 8 × 10 ¹⁷ cm ⁻³ , an increase in band edge emission intensity may be sometimes observed, but the effect is unstable. In order to stably obtain a gallium-based group III nitride compound semiconductor layer having excellent crystallinity, the indium doping concentration needs to be 8 × 10 ¹⁷ cm ⁻³ or more. Conversely, when indium doping with an atomic concentration exceeding 4 × 10 ²⁰ cm ⁻³ is performed, the probability of cracks occurring in the nitride compound semiconductor film is increased, possibly due to the hardening effect. The generation of the cracks causes deterioration of the surface state of the film, and also impairs the continuity of the film. Factors affecting the light emission intensity include the continuity of the nitride compound semiconductor film. A film lacking continuity causes an increase in inter-electrode resistance in addition to a decrease in the light emitting area in a light emitting element. This is inconvenient for obtaining a compound semiconductor light emitting device having excellent light emission intensity. Therefore, in the present invention, the upper limit of the atomic concentration of indium in the bonding layer bonded to the active layer is set to 4 × 10 ²⁰ cm ⁻³ .

There is no particular limitation on the concentration distribution of the doped indium atoms. This is because the indium-doped nitride semiconductor layer bonded to the active layer is not used as a light emitting layer. Therefore, it is not necessary to provide the distribution of the indium atom concentration for increasing the emission intensity as described above. Indium may be uniformly doped. Further, the doping may be performed with a concentration gradient in the thickness direction of the bonding layer. However, it is preferable that the doping is performed so that the atomic concentration at the interface between the bonding layer and the active layer is substantially equal to the initial concentration of indium atoms (N ₀ ) in the active layer as described above. This is in order to suppress the occurrence of distortion, crystal defects, and the like due to the difference in the atomic concentration of indium near the bonding interface between the active layer and the bonding layer.

A double heterojunction type gallium nitride blue LED adopts a configuration in which the bonding layer on the sapphire substrate side to be bonded to the active layer is n-type, and a p-type layer is disposed above the active layer. Even in a green LED having a single quantum well structure, the junction configuration is the same as viewed from the conduction type. On the other hand, when a material exhibiting p-type conductivity is used as the substrate instead of using an insulating crystal such as sapphire as the substrate, the semiconductor layer disposed on the substrate side to be bonded to the active layer is a p-type layer. Is a reasonable configuration. Therefore, the conduction type of the indium-doped nitride semiconductor layer to be joined to the active layer may be either n-type or p-type. It is essentially important that the nitride compound semiconductor layer is doped with indium in the atomic concentration range defined in the present invention and has improved crystallinity. However, when a gallium-based nitride semiconductor active layer in which indium is doped with a gradient in the atomic concentration is arranged on the indium-doped n-type gallium-based nitride semiconductor layer, the vicinity of the interface between the active layer and the n-type junction layer is provided. In order to make the density of accumulated carriers smaller than that at the junction interface between the opposing active layer and the p-type junction layer, the concentration of indium atoms at the interface between the active layer and the n-type junction layer is substantially reduced. It is important to make them equal.

接合 The number of bonding surfaces between the indium-doped nitride compound semiconductor bonding layer and the active layer may be either a single surface (one surface) or two surfaces. In other words, the bonding layer according to the present invention is allowed to be disposed immediately below, directly above, or both directly above and immediately below the active layer. If a nitride compound semiconductor bonding layer whose crystallinity is improved by doping with indium is arranged on the substrate side, the active layer of crystal defects such as dislocations that propagate to the bonding layer side due to lattice mismatch with the substrate. The effect which can suppress penetration into the water is mentioned. A nitride compound semiconductor light emitting device composed of a laminate having such a junction structure is composed of a laminate composed of constituent layers in which no measures are taken to reduce the introduction of crystal defects based on lattice mismatch. The light emission characteristics are remarkably improved as compared with the conventional LED. In the following description of the embodiments, the characteristics will be specifically compared with a comparative example.

When the number of bonding surfaces is 2, it means that indium-doped nitride compound semiconductor layers are bonded to both sides of the active layer. Examples of the configuration of the double heterojunction between the indium-doped nitride compound semiconductor layer and the active layer according to the present invention include an n-type or p-type indium-doped gallium nitride layer / n-type, p-type or insulating (high resistance). ) gallium indium nitride _{(Ga z In 1-z;} 0 ≦ z ≦ 1) active layer / p-type or n-type indium-doped aluminum gallium nitride layer junction structure. The main reason for disposing the indium-doped nitride compound semiconductor bonding layer also directly above the active layer is that it occurs due to lattice mismatch between the active layer and the bonding layer disposed immediately above the active layer. Most of the crystal defects that propagate to the bonding layer are included in the bonding layer. Thereby, the effect of suppressing the propagation of crystal defects to the current diffusion layer and the contact layer disposed on the bonding layer above the active layer can be obtained. This is because a crystal layer having few crystal defects provides an advantage of enabling a stable operation of the element, such as making the current concentration around the crystal defect or the generation of local leakage current rare.

When the indium-doped active layer is bonded to the indium-doped nitride compound semiconductor bonding layer made of the same material as the active layer, the crystallinity of the active layer is further improved. When the bonding layer and the active layer are made of the same substance (matrix), the crystallinity of the active layer on the bonding layer is further improved due to the lattice matching between the bonding layer and the active layer. Examples of such a configuration include an indium-doped n-type, p-type or an insulated gallium nitride junction layer doped with indium in the concentration range specified in the present invention. There is a configuration in which an active layer made of a gallium nitride of an insulated or insulated type is formed. As described above, the indium doping of the present invention does not cause a change in the band gap inherent to gallium nitride. Accordingly, the gallium nitride bonding layer or the active layer in the above configuration example does not become a gallium indium nitride (Ga _{1 -x} In _x N; x ≠ 0, 0 <x ≦ 1) mixed crystal. Therefore, the bonding layer and the active layer are made of the same material, that is, gallium nitride. Therefore, for example, when a bonding system having the above configuration is provided on a sapphire substrate, the substrate and the active layer have a lattice mismatching relationship, but the active layer and the bonding layer have a lattice matching relationship. Another example is a junction comprising an n-type, p-type or insulated aluminum / gallium nitride (AlGaN) mixed crystal layer doped with indium / an indium-doped n-type, p-type or insulated aluminum / gallium nitride mixed crystal active layer. A structure and a junction structure composed of an indium-doped gallium arsenide nitride (GaNAs) layer / an indium-doped gallium arsenide nitride active layer can be exemplified.

In a light-emitting element, a stacked structure provided with a band structure that further improves the emission intensity can be provided in a bonding system between the bonding layer that improves the crystallinity of the active layer and the active layer with improved crystallinity. . For example, a nitride compound semiconductor mixed crystal layer containing aluminum (Al) doped with indium is bonded to one of the active layers. On the other hand, an indium-doped nitride compound semiconductor layer made of the same material as the active layer doped with indium is joined to obtain an asymmetric heterojunction structure. When a nitride compound semiconductor layer containing aluminum having a large band gap is bonded to the active layer, a "light emission confinement effect" that contributes to an increase in light emission intensity can be obtained. Therefore, a compound semiconductor light emitting device having excellent emission intensity is provided from the laminate having such a bonding structure.

According to the present invention, there is also provided a nitride compound semiconductor element comprising a stacked body including a junction structure between an indium-doped nitride compound semiconductor layer and an indium-doped nitride compound semiconductor layer forming a junction with an active layer. The layers constituting the laminated body include a low-temperature buffer layer, a buffer layer, a lower clad layer, an active layer, an upper clad layer, a current diffusion layer, a contact layer, and the like in a nitride compound semiconductor light emitting device. In the present invention, when the indium-doped nitride compound semiconductor that forms a bond with the active layer is a lower cladding layer, the lower cladding layer is bonded to a nitride compound semiconductor layer that forms a buffer layer doped with indium. is there. This is because, by this bonding method, the propagation of crystal defects caused by the lattice mismatch between the substrate and the layer constituting the laminated body to the nitride compound semiconductor layer that is bonded to the active layer can be more reliably prevented. Thereby, a ripple effect of further improving the crystallinity of the nitride compound semiconductor active layer bonded to this bonding layer is brought about.

It is not necessary that the low-temperature buffer layer immediately above the substrate crystal be an indium-doped layer. A laminate structure in which the clad layer and the contact layer are layers doped with indium is also conceivable. According to this configuration, the effect of improving the crystallinity brought about by the doping of indium can be spread over the whole of the stacked constituent layers. Specifically, there may be mentioned a laminate for a light-emitting element obtained by sequentially depositing the following nitride compound semiconductor layers on a sapphire substrate.
(1) Al _x Ga _1-x N (0 ≦ x ≦ 1) low temperature buffer layer.
(2) A buffer layer or lower cladding layer made of an n-type gallium nitride layer doped with indium.
(3) An n-type gallium nitride active layer to which a group II element and a group IV element, or an indium, a group II element, and a group VI element are added in addition to indium. For example, an n-type gallium nitride active layer to which indium, zinc and germanium are added.
(4) An upper cladding layer made of a p-type aluminum-gallium nitride mixed crystal layer to which indium is added.
(5) A p-type gallium nitride contact layer doped with indium.
The light emitting element is formed through a process of forming an n-type electrode (negative electrode) on the layer (2) and a p-type electrode (positive electrode) on the layer (5). This laminated structure includes a junction between the active layer and the nitride compound semiconductor layer doped with indium, an indium doped nitride compound semiconductor active layer composed of the same material (matrix) as the junction layer, and a low-temperature buffer. This is an example of a laminated structure in which all of the laminated body constituent layers on the layers are nitride compound semiconductor layers doped with indium.

Examples of other nitride compound semiconductor devices according to the present invention include a crystal substrate having a hexagonal or cubic crystal structure such as conductive or insulating silicon carbide (SiC) or zinc oxide (ZnO). A light-emitting element having a structure in which the following layers are sequentially stacked is exemplified.
(1) An aluminum gallium nitride (Al _x Ga ₁ -xN; 0 ≦ x ≦ 1) low-temperature buffer layer having a thickness of 2 to 50 nm.
(2) An n-type or p-type aluminum-gallium nitride mixed crystal (Al _x Ga) having an indium-doped carrier concentration of 10 ¹⁶ cm ^{−3 to} 10 ¹⁹ cm ⁻³ and a layer thickness of about 0.01 to 5 μm. _1-xN ; lower cladding layer composed of 0 <x ≦ 1).
(3) A film thickness of 2 to 10 nm, indium, cadmium (Cd) and silicon (Si) doped together, and a carrier concentration of approximately 10 ¹⁵ cm ^{−3 to} 5 × ¹⁸ cm ⁻³ . n-type or p-type aluminum gallium nitride (Al _x Ga ₁ -xN; 0 ≦ x ≦ 1) active layer.
(4) An n-type or n-type or p-type aluminum-gallium nitride / gallium mixed crystal having a carrier concentration doped with indium of 10 ¹⁶ cm ⁻³ or more and 10 ¹⁹ cm ⁻³ or less and a layer thickness of about 0.01 to 5 μm. (Al _x Ga ₁ -xN; 0 <x ≦ 1) An upper cladding layer.
(5) n-type or p-type aluminum-gallium nitride (Al _x Ga ₁ -xN; 0 ≦) in which the concentration of indium-doped carrier is about 10 ¹⁶ cm ⁻³ or more and the layer thickness is about 0.01 μm or more. x ≦ 1) Contact layer.
(6) Transparent conductive film or translucent thin film electrode.
The low-temperature buffer layer, which is a constituent layer of the laminated structure for use in the light emitting device according to the present invention, which is exemplified, may be doped with indium. If a low-temperature buffer layer doped with indium is adopted, a laminate composed of a nitride compound semiconductor doped entirely with indium is formed on the substrate.

Examples of other nitride compound semiconductor devices according to the present invention include III such as gallium phosphide (GaP), gallium arsenide (GaAs), boron nitride (BN), boron arsenide (BAs), and boron phosphide (BP). There is a light-emitting element having a structure in which the following layers are sequentially stacked on a substrate made of a group V compound semiconductor or a group IV element semiconductor such as silicon (Si) or germanium (Ge).
(1) Aluminum nitride gallium indium _{(Al x Ga y In 1-} xy N; x + y = 1,0 ≦ x ≦ 1,0 ≦ y ≦ 1) single layer or low-temperature buffer layer consisting of layer.
(2) Group IV elements or the n-type Group VI elements or both is added _{Al x Ga y In 1-xy} N (x + y = 1,0 ≦ x ≦ 1,0 ≦ y ≦ 1) single-layer or A buffer layer consisting of those layers.
(3) n-type or p-type Al _x Ga _1-x N _y P _1-y (0 ≦ x ≦ 1, 0 ≦ y <1) doped with In or n-type or p-type Al _x Ga _1-x A carrier concentration containing a Group V element other than nitrogen, such as N _y As _1-y (0 ≦ x ≦ 1, 0 ≦ y <1), is set to be about 10 ¹⁶ cm ⁻³ or more and about 10 ¹⁹ cm ⁻³ or less; lower clad layer thickness is made of a nitride compound semiconductor on the order of 0.01 to 5 [mu] m (4) Al indium and zinc and the germanium is doped simultaneously _{x Ga 1-x N y P} 1-y (0 ≦ x ≦ 1, 0 ≦ y ≦ 1) or a group V element other than nitrogen, such as Al _x Ga _1−x N _y As _1−y (0 ≦ x ≦ 1, 0 ≦ y <1). A light-emitting layer having a thickness of about 10 ¹⁵ cm ^{−3 to} 10 ¹⁸ cm ⁻³ and a thickness of about 0.01 to 2 μm;
(5) Al _x Ga _y In _1-xy N (x + y = 1, 0 ≦ x ≦ 1, 0 ≦ y ≦ 1) having a different conductivity type from the lower cladding layer or Al _x Ga to which In is added. _1-x N (0 ≦ x ≦ 1) or a group V element other than nitrogen and a carrier concentration of about 10 ¹⁶ cm ^{−3 to} about 10 ¹⁹ cm ^−3, and a layer thickness of about 0.01 to 5 μm Upper cladding layer made of a nitride compound semiconductor.
(6) similarly to the conduction type upper cladding _{layer, Al x Ga y In 1-} xy N (x + y = 1,0 ≦ x ≦ 1,0 ≦ y ≦ 1) or Al an In is added _x Ga _{1- xN} (0 ≦ x ≦ 1) or a nitride compound semiconductor containing a Group V element other than nitrogen, having a carrier concentration of about 10 ¹⁶ cm ⁻³ or more and a layer thickness of about 0.05 μm or more and about 3 μm or less Contact layer.
(7) A light-transmitting thin-film electrode layer or a transparent conductive film layer made of a metal nitride or a metal material.

Each of the above laminates is an example in which a nitride compound semiconductor layer doped with indium is provided as an active layer (light emitting layer), and a gallium indium nitride (GaInN) mixed crystal having a high indium composition ratio is conventionally activated. It is not a layer. That is, when the indium-doped nitride compound semiconductor layer according to the present invention is used as the active layer, it is not necessary to change the growth temperature for the purpose of growing the active layer. For example, in obtaining a laminate having the above structure, a light-emitting layer made of an indium-doped nitride compound semiconductor layer is grown while maintaining the same growth temperature as the lower clad layer, and then further at the same growth temperature, Can be grown. This is because in the present invention, indium is not added at a high concentration so that an indium-based mixed crystal is created.

The constituent layers of the above-described laminated structure are formed by vapor deposition such as metalorganic thermal decomposition vapor deposition (MOCVD, MOVPE, etc.), molecular beam epitaxy (MBE), and chemical beam evaporation (CBE). It can be grown by the phase growth method. In the MOCVD method, a normal pressure or a reduced pressure method may be used. That is, the pressure during film formation does not matter. A nitride compound semiconductor layer containing no aluminum may be grown by a halogen or hydride vapor phase epitaxy method, and a nitride semiconductor layer containing aluminum may be obtained by using another growth method. A physical deposition method such as a high frequency sputtering method can also be used as one film forming method. The growth raw material is not particularly limited. For example, a conventional gallium source is an organic gallium compound such as metal gallium, trimethylgallium ((CH ₃ ) ₃ Ga), or triethylgallium ((C ₂ H ₅ ) ₃ Ga). And their halogen adducts can be used. Aluminum sources include trialkyls such as metallic aluminum, trimethyl aluminum ((CH ₃ ) ₃ Al), triethyl aluminum ((C ₂ H ₅ ) ₃ Al) and triisobutyl aluminum ((i-C ₄ H ₉ ) ₃ Al) Aluminum compounds and adducts containing aluminum and a nitrogen atom can also be used. The indium-doped nitride compound semiconductor includes an organic indium compound having an aliphatic hydrocarbon group such as metal indium and a trialkyl In compound, and an alicyclic indium compound such as cyclopentadienyl I n (C ₅ H ₅ In). It can be obtained by adding indium to the nitride semiconductor layer as an addition source. No special raw materials are required to obtain the laminated constituent layers according to the present invention. As a method of adding indium, there is a method of injecting indium ions into the nitride compound semiconductor layer.

The gallium-based group III nitride compound semiconductor active layer according to the present invention is doped with indium, and the bonding method between the active layer and the indium-doped gallium group III nitride compound semiconductor bonding layer is based on the light emitting element. Increase.

(Example 1)
The nitride compound semiconductor device according to the present invention will be described in detail with reference to the drawings, taking a blue LED as an example. FIG. 7 is a schematic plan view of the LED according to the present embodiment. FIG. 8 is a schematic sectional view taken along a broken line AA ′ in FIG. In the present embodiment, a sapphire substrate, which was cut (sliced) from an ingot grown by the CZ (Czochralski) method and whose both surfaces were mirror-polished by a chemical mechanical polishing method, was used. The sapphire substrate was circular with a diameter of 50 mm, the periphery of the substrate was chamfered, and the finished thickness was about 100 μm. The plane orientation was (0001), and the substrate surface was a so-called C plane. Immediately before loading this substrate into an atmospheric pressure organometallic pyrolysis vapor deposition reactor (MOCVD reactor) for forming the constituent layers of the laminate shown in FIG. , And washed with ultrapure water having a specific resistance of about 18 megaohms (MΩ). After that, the surface of the substrate was treated with a commercially available aqueous solution of ammonium fluoride, and the surface was again washed with ultrapure water. Then, the water was evaporated using an infrared lamp and dried.

The reaction furnace and the substrate transfer system can be used for the sufficiently dried substrate (101) by using an interlock type substrate transfer system equipped with an extremely general vacuum exhaust mechanism without allowing outside air to enter the MOCVD reactor. With the inside maintained at a degree of vacuum of about 10 ⁻³ Torr, the wafer was transferred onto the surface of a cylindrical support (susceptor) made of high-purity graphite. Next, a JIS: 316L stainless steel pipe with a 1/4 inch outer diameter, which has been precisely subjected to complex electrolytic polishing so that the inner surface has a roughness of less than 0.5 μm with Rmax. , High-purity argon (Ar) gas was introduced as a purge gas. Argon gas was purified by a molecular sieve (molecular sieve) adsorption method to a dew point of about minus 90 ° C., and the flow rate as a purge gas was adjusted and controlled to 5 liters per minute by an electronic mass flowmeter (abbreviated MFC). . After continuing the flow of the argon purge gas into the reaction furnace for 20 minutes, the gas flowing through the reaction furnace was converted from argon gas to high-purity hydrogen (H ₂ ) gas at a flow rate of 8 liters per minute.

Next, while flowing the hydrogen at the above flow rate as a carrier (raw material transporting) gas into the reaction furnace, a cylindrical outer periphery made of molybdenum (Mo) material is also used as a mounting table of the high-purity graphite susceptor. The sapphire substrate was heated to 450 ° C. for growing a low-temperature buffer layer by raising the temperature of the sapphire substrate for 15 minutes from room temperature by supplying power to the semi-enclosed resistance heating heater. The temperature of the sapphire substrate is the center of the resistance heater, that is, a so-called ceramic heater which is made of a carbon material as a heating element by chemical deposition (abbreviated as CVD) on a circular boron nitride (BN) substrate. The temperature was measured using a thermocouple that penetrated the center of the element and reached a cut-out portion of a molybdenum disk constituting the upper surface of the heater. The thermocouple used was a platinum / platinum-rhodium alloy-based thermocouple (JIS: R thermocouple), and the distance between the welded tip of the thermocouple and the upper surface of the heater was 1 mm in the vertical direction. The sapphire substrate is placed at the center of a graphite circular susceptor having a thickness of 2 mm and a diameter of 120 mm. Further, the susceptor is disposed on a molybdenum disk constituting the upper surface of the heater. Accordingly, the distance between the sapphire substrate and the thermocouple is approximately 3 mm, and the existence of this distance naturally causes a difference between the temperature measured by the thermocouple and the actual temperature of the sapphire substrate, particularly the surface of the substrate. For example, when the thermoelectromotive force of the thermocouple gives a temperature indication value of 686 ° C. within the temperature range of 600 ° C. to 700 ° C., it is measured as the substrate surface temperature from the JIS-R thermocouple in contact with the sapphire substrate surface. The temperature was 660 ° C., and a difference of about 26 ° C. was observed. As described above, although there is actually a difference between the temperature indication value of the thermocouple disposed in the heater and the surface temperature of the sapphire substrate heated mainly by heat transfer via the susceptor, the present embodiment For convenience, the temperature measured by a thermocouple disposed in the heater is referred to as a growth temperature.

After the temperature of the sapphire substrate reaches 450 ° C., the temperature indication value obtained by converting the thermoelectromotive force signal of the JIS: R thermocouple is monitored, and the transient response (overshoot) state of the substrate temperature due to the heating is eliminated. It waited for a while until the temperature reading stabilized. After the sapphire substrate is maintained at the same temperature for about 10 hours, a seal material in a gas contact portion (a portion that comes into contact with gas) or the like is replaced with a fluororesin-based resin o-ring such as Kalrez. The ammonia gas whose flow rate was precisely controlled to 1 liter per minute by an electronic mass flowmeter (MFC) dedicated to ammonia (NH ₃ ) gas, which was converted into the above, was mixed with the above hydrogen carrier before reaching the reaction furnace. The hydrogen carrier gas accompanying the ammonia gas was introduced into the reaction furnace from above the sapphire substrate surface almost vertically above the precision-polished stainless steel pipe. Ammonia gas was supplied by decompressing and vaporizing liquefied ammonia contained in a pressure-resistant cylinder having an internal volume of approximately 10 liters made of an aluminum (Al) alloy whose inner surface was mirror-polished. Exactly 5 minutes after the start of the addition of the ammonia gas to the hydrogen carrier gas, a liquid of trimethylgallium ((CH ₃ ) ₃ Ga) as a gallium (Ga) raw material source is bubbled (foamed) to produce trimethylgallium. Hydrogen gas accompanying steam was mixed with a hydrogen carrier gas immediately before the reactor and supplied to the surface of the sapphire substrate in the reactor. Trimethylgallium is initially filled in a bubbler (foaming container) made of JIS: 304 and 316L stainless steel whose inner surface is mirror-polished and having an inner volume of about 150 milliliters (ml). The temperature of the container, that is, the temperature of trimethylgallium, was maintained at 0 ° C. in an electronic thermostat using the Peltier effect. Trimethylgallium is a liquid at a temperature of 0 ° C., and high-purity hydrogen gas having a flow rate of 5 ml / min by an electronic mass flow meter (MFC) was passed through the liquid as a bubbling gas. Addition of a hydrogen bubbling gas accompanied by trimethylgallium into the growth furnace was used to start the growth of a low-temperature buffer (buffer) layer (102) made of gallium nitride (GaN). The supply of the hydrogen bubbling gas accompanying the ammonia gas and the vapor of trimethylgallium into the reactor was continued for exactly 15 minutes. Thus, an undoped gallium nitride low-temperature buffer layer (102) having a layer thickness of about 7 nanometers (nm) was formed. After exactly 15 minutes had elapsed, the mixing of the hydrogen bubbling gas accompanied by the vapor of trimethylgallium into the hydrogen carrier gas was instantaneously stopped by opening and closing a valve provided in the piping system of the raw material supply system.
At the same time, the gas type constituting the carrier gas was first switched instantaneously from hydrogen to argon gas at the same flow rate in order to suppress fluctuations in the pressure in the reactor by opening and closing a valve provided in the piping system. Thereafter, the flow rate of the argon gas was gradually reduced from 8 liters per minute to 5 liters per minute by an electronic mass flow meter (MFC) for about 2 minutes. The ammonia gas flow rate was 1 liter per minute and was not changed. After changing the carrier gas to argon and adjusting its flow rate, the amount of electric power supplied to the above-mentioned resistance heating type heater is controlled through a thyristor-type electric circuit to increase the temperature of the sapphire substrate to gallium nitride growth. The temperature was increased to 1100 ° C., the growth temperature of the layer. After reaching 1100 ° C., 5 minutes after it is determined from the temperature reading that the temperature transient has attenuated, the flow rate of the argon gas is reduced from 5 liters per minute to 3 liters per minute, and at the same time, the hydrogen gas is discharged. Feeding into the reactor was started again at a flow rate of 3 liters per minute. That is, the carrier gas was composed of hydrogen and argon having the same flow rate. At the same time, the flow rate of ammonia gas was increased from 1 liter per minute to 4 liters per minute. The changes in the flow rates of these gases were all completed concurrently within about 2 minutes. In the time zone where these flow rates were changed, disilane (Si ₂ H ₆ ) gas diluted to a volume concentration of about 1.2 ppm with high-purity hydrogen gas at a flow rate of 5 ml / min was used. Before supplying the gallium source described above, it was previously supplied into the reaction furnace as a doping gas (doping source) of silicon (Si). Exactly three minutes after the change of the flow rate, the growth of the aluminum / gallium nitride growth layer was started by mixing trimethylgallium and a hydrogen bubbling gas accompanied by trimethylaluminum vapor in the carrier gas having the above composition. At this time, the temperature of the stainless steel bubbler container containing trimethylgallium was set to 13 ° C. using commercially available ethylene glycol as a heat transfer (cooling) medium by the electronic constant temperature layer. The flow rate of hydrogen for bubbling liquid trimethylgallium was 5 ml / min. The temperature of the stainless steel bubbler container containing trimethylaluminum and having an inner surface mirror-polished was set to 20 ° C. using commercially available ethylene glycol as a heat transfer (cooling) medium by the same electronic thermostat as described above. The flow rate of hydrogen for bubbling the liquefied trimethylaluminum was 5 ml / min. While maintaining the flow rate conditions of these various gases, mixing of a hydrogen bubbling gas accompanied by a vapor of trimethylgallium into a carrier gas having the above composition is continued for 90 minutes to form a silicon (thickness of about 3.5 μm) layer. A Si) -doped n-type aluminum-gallium nitride mixed crystal layer (117) was grown. Exactly 90 minutes later, the addition of hydrogen bubbling gas accompanied by the vapors of trimethylgallium and trimethylaluminum and disilane gas, which is a silicon doping gas, to the hydrogen / argon mixed carrier gas is interrupted to prevent silicon-doped n-type aluminum nitride. -Gallium layer growth is completed. The carrier concentration of the silicon-doped n-type aluminum-gallium nitride mixed crystal layer deposited on the gallium nitride buffer layer grown under the same conditions as above was not substantially constant in the layer thickness direction. According to the usual stepping hole (Hall) measurement method, the carrier concentration in the region from the vicinity of the interface with the gallium nitride buffer layer to a thickness of about 1 μm is about 3 × 10 ¹⁸ cm ⁻³ , The carrier concentration tends to decrease slightly as the layer thickness increases from the substantially central portion toward the surface, and the carrier concentration at the surface decreases to about 9 × 10 ¹⁷ cm ⁻³ from the layer depth (substrate side). Was. While the flow rate of the ammonia gas was kept as it was, the supply of the hydrogen gas, which constituted the carrier gas, to the growth reaction system was cut off, and two types of gas, argon and ammonia, were introduced into the growth reaction system. This gas switching was performed within about 2 minutes after shutting off the supply of the above-described hydrogen bubbling gas accompanied by the vapors of trimethylgallium and trimethylaluminum to the growth reaction system. Thereafter, the substrate temperature was decreased from 1100 ° C. to 850 ° C. at a rate of 25 ° C./min. After the temperature hunting is eliminated in about 3 minutes, the supply of hydrogen bubbling gas accompanied by the vapor of trimethylgallium and the supply of doping sources of indium (In) and zinc (Zn) are started. An active layer (118) consisting of a gallium nitride (GaN) layer doped with indium and zinc was formed. The above stainless steel bubbler container containing trimethylgallium was kept at 0 ° C. by an electronic thermostat. The flow rate of high-purity hydrogen gas for bubbling liquid trimethylgallium was set to 5 ml / min by an electronic mass flow meter (MFC). For the indium source to avoid conjugation reactions with electron donating ammonia gas, cyclopentadienyl indium to monovalent valencies _{(cyclopentadienylindium: C 5 H 5 In} ) was used (JP-A-1 -94613 and J. Crystal Growth, 107 (1991), pages 360-364). This indium source was housed in a cylindrical bottle made of 316L-stainless steel or the like whose inner surface was mirror-polished, and the bottle was kept at 60 ° C. in the same electronic thermostat as above. At the beginning of the stainless steel bottle containing the indium source, high-purity hydrogen gas for entraining the sublimated indium source into the growth reaction system was initially supplied via an electronic mass flow meter (MFC) at 20 ml / min. It circulated at the flow rate. Zinc was doped by adding high-purity hydrogen gas containing about 100 ppm by volume of diethyl zinc ((C ₂ H ₅ ) ₂ Zn) to the growth reaction system. The supply amount of the diethyl zinc / high-purity hydrogen mixed gas to the growth reaction system was set at 40 ml / min by an electronic mass flow meter and kept constant until the growth of the indium and zinc-doped gallium nitride layers was completed. On the other hand, the supply amount of the high-purity hydrogen gas accompanying the sublimated indium source to the growth reaction system was set to 20 ml at the start of the growth of the same layer, and as the growth time elapsed, that is, as the layer thickness of the same layer increased, It was increased linearly to 220 millilitres at the end of the same layer. Since it took 5 minutes to grow the indium and zinc doped gallium nitride layers, the amount of supply of the indium source over time was 40 ml / min. Further, since the layer thickness of the indium and zinc-doped gallium nitride layer obtained by the growth for 5 minutes was 60 nm, the increase in the supply amount of the indium source per unit layer thickness (1 nm) was about 3.3 ml. became. The supply of the vapor of the gallium source and the hydrogen gas and the diethylzinc / high-purity hydrogen gas accompanying the sublimated indium source to the growth reaction system were stopped to terminate the growth of the indium and zinc-doped gallium nitride layers. At the same time as the supply of the raw material gas and the doping gas is stopped, the supply amount of the argon carrier gas is reduced from 3 liters per minute to 1 liter per minute, and the flow ratio of the argon gas to the ammonia gas (4 liters per minute) / Argon flow ratio) was relatively reduced.

In this state, the temperature of the substrate was raised from 850 ° C. to 1150 ° C. at a rate of 60 ° C./min for 5 minutes. The ammonia flow rate was increased from 4 liters per minute to 6 liters per minute while reaching 1150 ° C. Five minutes after reaching 1150 ° C., the flow rate of the argon gas was immediately increased from 1 liter per minute to 3 liters per minute, and the supply of hydrogen gas was restarted. The flow rate of hydrogen constituting the carrier gas together with argon was adjusted to 3 liters per minute by an electronic mass flow meter (MFC). After exactly one minute by adjusting the flow rate, a hydrogen bubbling gas accompanied by trimethylgallium vapor and a hydrogen bubbling gas accompanied by trimethylaluminum ((CH ₃ ) ₃ Al) vapor were added to the argon / hydrogen mixed carrier gas. . The doping gas of zinc and magnesium (Mg) was added at the same time as the addition of the group III side material system gas to the growth reaction system. As a result, aluminum and gallium nitride mixed crystal (Al _x Ga ₁ -xN; x is an Al composition ratio having a layer thickness of 120 nm, doped with both group II acceptor impurities, zinc and magnesium, represents an Al composition ratio, An upper cladding layer (119) consisting of 0 <x <1) was grown. As a doping source of magnesium, bis-methylcyclopentadienylmagnesium (bis-methylcyclopentadienylmagnesium: (CH ₃ C ₅ H ₅ ) ₂ Mg) was used. The magnesium source was kept at 50 ° C., and the sublimated magnesium source was accompanied by a high-purity hydrogen gas to the growth reaction system in the same manner as the indium source described above. During the formation of the aluminum nitride-gallium mixed crystal for 10 minutes, while the thickness of the layer reaches about 60 nm (about 5 minutes in growth time), the supply rate of the magnesium source is 40 minutes per minute. It was kept constant at milliliters. In the latter half of the growth of the same layer (in the growth time, from about 5 minutes after the start of the growth of the same layer until reaching 10 minutes), the supply amount of the magnesium source was increased from 40 ml / min to 20 ml / min. Gradually reduced. On the other hand, the supply amount of the diethyl zinc / hydrogen mixed gas into the growth chamber was maintained at a constant flow rate of 40 ml / min with respect to the growth time.

Thereafter, the supply of the aluminum source to the reaction system was cut off, and the process shifted to the formation of a contact layer (106) made of gallium nitride doped with zinc and magnesium. The supply amount of the magnesium source was 20 ml / min at the beginning of the growth, but was gradually increased linearly to 40 ml / min as the layer thickness increased. The growth of the zinc and magnesium doped gallium nitride contact layer intended to have a layer thickness of 10 nm was completed over 5 minutes. FIG. 9 shows the change in the substrate temperature in the process of obtaining the laminated body described in the present embodiment. FIGS. 10 to 12 show the flow rates of argon, hydrogen carrier gas, and ammonia used as a nitrogen source that follow the change in the substrate temperature. Summarize the changes. FIGS. 13 and 14 show the change in the flow rate of the group III raw material system corresponding to the change in the substrate temperature. In some cases, the temperature of the supply source of the group III raw material is changed in each growth process (the temperature in parentheses in FIGS. 13 and 14 is the raw material holding temperature). Therefore, what is shown in FIGS. 13 and 14 is a change in the flow rate in each growth process, and does not indicate the absolute value of the molecular concentration (molar concentration). The flow rate to the molar concentration can be converted by the vapor pressure or sublimation pressure value of the group III element raw material at the holding temperature. Incidentally, the vapor pressures of trimethylgallium at 0 ° C. and 13 ° C. are 64.4 Torr (Torr) and 128.2 Torr, respectively. The sublimation pressure of cyclopentadienyl indium at 65 ° C. is 1.8 Torr. The vapor pressure of trimethylaluminum at a holding temperature of 25 ° C. is 11.9 Torr. The pressure of the hydrogen gas circulated in the container for storing the group III raw material is accurately adjusted to 1 atm by a gauge pressure by a low pressure reducing valve for semiconductor industry in which a gas contact portion is mirror-polished.

(4) At the stage when the growth of the gallium nitride layer doped with both zinc and magnesium was completed, supply of all the group III element raw materials to the growth reaction system was stopped. Immediately after the supply of the group III element raw material was stopped, the power supply to the resistance heater was stopped to start cooling the substrate. In particular, no measures were taken to forcibly cool the substrate, but it was observed that the substrate temperature dropped at a rate of about 30 ° C./min from 1150 ° C. to a temperature near 750 ° C. Cooling was observed at a rate of about 20 ° C. to about 25 ° C. per minute from near 750 ° C. to near 450 ° C. When the substrate temperature dropped to around 450 ° C., the supply of ammonia gas to the growth reaction system was stopped. It took about 40 minutes to decrease from around 450 ° C. to room temperature.

The supply of hydrogen gas to the reactor was stopped, and argon was supplied instead. After the supply of argon to the reaction furnace was continued for a while, the supply of argon was stopped, and the inside of the reaction furnace was evacuated with a usual rotary vane pump until the pressure reached about 2 × 10 ⁻³ Torr. Vacuum degassing in the furnace and purging with argon gas were repeated several times, and finally the pressure in the reactor was returned to almost atmospheric pressure by purging with argon gas. Then, after the wafer is taken out of the reactor, the tungsten surface is brought into contact with two tungsten probes (probes) at an interval of about 1 mm at this time to determine whether there is continuity between the probes. Inspection revealed no conductivity. Next, the wafer was placed on a metal susceptor in an implantation station of an ion implantation apparatus having a maximum acceleration voltage of 200 kilovolts (KV). After performing a vacuum evacuation operation before the start of ion implantation, magnesium ions having a mass number of 24 and an acceleration voltage of 180 KV and a dose of 7 × 10 ¹² cm ⁻² were implanted by a solid source method. Magnesium ions were perpendicularly incident on the (0001) plane used as the substrate, and no particular attention was paid to plane channeling and axial channeling in ion implantation into the face-centered cubic crystal. The plane orientation processing accuracy of the substrate used in this embodiment is within the range of (0001) ± 0.5 °, and therefore, the incident angle of the incident ion beam with respect to the sapphire C plane (the c-axis direction of the hexagonal crystal) ) Is in the range of ± 0.5 ° at the maximum. The wafer into which magnesium ions had been implanted was subjected to a surface treatment with an organic solvent such as isopropyl alcohol or high-purity hydrochloric acid (HCl), and then washed with ultrapure water and dried. Thereafter, the wafer was subjected to lamp heating by a halogen lamp in a nitrogen atmosphere. The wafer was heated from room temperature to 1050 ° C. in about 30 seconds, held at 1050 ° C. for 10 minutes, and then cooled to room temperature in about 20 minutes using both forced cooling by water cooling and natural cooling. After cooling, two probes were brought into contact with the surface at an interval of 1 mm, and the conductivity was measured. As a result, a tendency was found that the current increased linearly substantially in proportion to the voltage applied between the probes. As an example, when the imprint voltage between the probes is 10 volts (V), the flowing current reaches about 300 milliamps (mA). It is simply a current-voltage characteristic between the probes, and an accurate resistivity or the like cannot be immediately obtained from these values due to diffusion of an electric field around the probe. FIG. 16 compares current-voltage characteristics as a curve (120) immediately after wafer growth and a curve (121) after annealing after ion implantation.

Observation of the surface of the wafer with a general differential interference microscope showed that there was an extremely gentle "swell" of the surface under a low magnification of about 50 times. In the surface observation using a scanning electron microscope (English abbreviation: SEM) with a magnification of 5 × 10 ³ or 10 ^{4, the} observation target is limited to a small area due to the high magnification. Was no longer clearly observed, and pits (pores) having a substantially hexagonal opening were hardly observed in regions other than the peripheral portion of the wafer. A substantially regular hexagonal thin portion having a side length of less than 1 μm is formed in a peripheral portion of the wafer, particularly, in an annular zone of about 3 to 6 mm inward from the edge of the 2-inch diameter substrate toward the center of the wafer. A large number of pores existed at a density of 5 × 10 ³ cm ⁻² . For this reason, the central portion of the wafer was recognized as being transparent to natural light, but the peripheral annular zone did not need to dare to irradiate high-intensity white light, and was turbid due to light scattering by pores. It became something to do.

Concentration distribution of constituent elements and dopants in the depth direction from the surface of the wafer analyzed by secondary ion mass spectrometry (abbreviated to SIMS) using cesium (Cs) or oxygen ions as primary ion beams (incident ion beams) Is shown in FIG. First, when the laminated structure of the wafer is identified from the concentration distributions of the constituent elements and the dopants, a contact layer (106) made of gallium nitride doped with zinc and magnesium at the outermost surface and ion-implanted with magnesium after the growth is completed. From there on, if the layered structure is traced sequentially in the depth direction, zinc and magnesium are doped at the time of film formation, and after the growth is completed, magnesium ion is implanted. An upper clad layer (119), a gallium nitride light emitting layer (118) doped with indium and zinc and having a thickness of about 60 nm, and an aluminum-gallium nitride lower clad layer (117) doped with silicon. Was determined. However, the existence of the interface between the silicon-doped aluminum gallium nitride layer and the low-temperature buffer layer made of gallium nitride was not clearly recognized from the elemental profile by SIMS analysis. The concentrations of gallium and nitrogen are rapidly increasing near the outermost surface, and it appears that these constituent elements have accumulated on the surface. This is presumed to be due to chemical interference in SIMS analysis. Was done. The intended thickness of the gallium nitride contact layer (106) as the outermost layer is 100 nm, whereas the thickness obtained from the depth-profile of gallium by SIMS is about 97 nm. . The aluminum composition ratio (x) calculated from the atomic concentration ratio of aluminum and gallium constituting the upper cladding layer (105) made of aluminum nitride / gallium mixed crystal was approximately 0.12.

Similarly, the analysis result by SIMS showed that the concentration and distribution of the dopant in each layer were as follows. As in the case of the above matrix element, doping impurities of zinc and magnesium accumulate near the surface side of the gallium nitride contact layer (106) as the outermost layer, as shown in FIG. It was presumed to be due to chemical interference. If the data in the area near the surface where the surface condition slightly affects the detection sensitivity of the secondary ions is omitted, the concentration distribution of zinc in the gallium nitride contact layer (106) is substantially constant at 9 × 10 ¹⁸ c ^{Was -3} . On the other hand, the concentration of magnesium atoms in the same layer (106) is as high as about 8 × 10 ¹⁸ cm ⁻³ near the surface, and once the atomic concentration is reduced to 6 × 10 ¹⁸ cm ⁻³ in the depth direction, ion The distribution of implanted atoms at the time of implantation was distributed in the form of a general normal distribution curve, and increased to 2 × 10 ¹⁹ cm ⁻³ near the interface between the same layer (106) and the lower layer (119). In the active layer (118) made of an aluminum gallium nitride mixed crystal (Al _0.12 Ga _0.88 N) having an aluminum composition ratio (x) of 12%, the atomic concentration of magnesium is increased from the gallium nitride contact layer (106) side to the depth. The density also increased in the direction of a normal distribution curve, and the density rapidly increased stepwise in a region deeper than about half of the layer thickness to about 5 × 10 ¹⁹ cm ⁻³ . In particular, the increase in the atomic concentration of magnesium near the interface with the active layer (118) is due to the increase or decrease in the amount of the magnesium source added to the growth reaction system during the growth of the upper clad layer (119). Was determined to be. That is, in the implantation of magnesium ions, the concentration distribution of the implanted atoms becomes a normal distribution (Gaussian distribution) curve in principle, and the concentration of the implanted atoms cannot be changed steeply particularly at the interface side deeper than the surface side. In addition, the fact that a region doped with magnesium at the time of growth (in this embodiment, a region of about 60 nm from the interface with the active layer (118)) is arranged in a region near the interface in advance, because the region exceeds the projection range. It was interpreted to be effective in avoiding a decrease in the concentration of magnesium atoms in the deep region. Here, zinc was distributed almost uniformly in the depth direction, and its concentration was 8 × 10 ¹⁸ cm ⁻³ .

In the indium-doped gallium nitride active layer (118) immediately below the upper cladding layer (119) made of aluminum nitride / gallium mixed crystal, SIMS analysis results show that the concentration of indium atoms is lower than the same layer (118). From the interface (lower interface) with the silicon-doped aluminum nitride gallium lower cladding layer (117) to the interface (upper interface) with the aluminum cladding layer (119) on the aluminum nitride / gallium mixed crystal layer, the indium source supply amount is reduced. It was shown that it increased steadily linearly with the increase. The concentration of indium atoms was 9 × 10 ¹⁷ cm ^-3 at the lower interface and 3 × 10 ²⁰ cm ^-3 at the upper interface. On the other hand, the concentration of zinc atoms in the active layer (118) was approximately constant 8 × 10 ¹⁸ cm ⁻³ corresponding to the fact that the flow rate of the diethyl zinc / high-purity hydrogen mixed gas supplied during film formation was kept constant. became. Further, elements other than indium and zinc were not intentionally added to the indium-doped gallium nitride active layer (118), but silicon (Si) was detected at a concentration of about 4 × 10 ¹⁷ cm ⁻³ . The atomic concentration of silicon was determined to be sufficient to increase the intensity of luminescence emission from zinc (SolidStateCommun., 57 (6) (1991), pp. 405-409). See FIG.

Another sample in which the surface layer of the gallium nitride layer doped with indium and zinc was exposed was manufactured by etching the surface of the wafer using a general ion sputtering method. In the surface observation by SEM at a high magnification of about 10 ⁴ times, the macroscopic morphology was not clearly observed because the visual field became narrower at a higher magnification, but the differential interference optical microscope at a low magnification of about 50 times. According to the surface observation by, substantially circular or elliptical protrusions were observed on the surface of the indium and zinc doped gallium nitride layers. In a PL measurement at room temperature using a helium (He) -cadmium (Cd) laser beam having a wavelength of 325 nm as an incident beam, a near-ultraviolet light whose wavelength is considered to be about 380 nm when a gallium nitride-indium mixed crystal is formed is formed. No band spectrum was observed, and only band edge emission of gallium nitride was observed around a wavelength of 365 nm. That is, in the doping of indium in this example, no result was proved that gallium nitride-indium mixed crystal was formed.

The lower layer of the gallium nitride active layer (118) doped with zinc and indium is a lower clad layer (117) made of aluminum nitride / gallium mixed crystal having an aluminum mixed crystal ratio (x) of 0.01 based on the analysis result by SIMS. Was determined to be. The atomic concentration of silicon in the lower cladding layer (117) was determined to be 4 × 10 ¹⁸ cm ^{−3 by} SIMS analysis. Therefore, an Al _0.12 Ga _0.88 N mixed crystal layer of x = 0.12 is joined on the upper surface to the gallium nitride layer (118) doped with both indium and zinc to be the active layer (light emitting layer), and the lower surface is joined on the lower surface. The structure was such that an n-type Al _0.01 Ga _0.99 N mixed crystal layer of x = 0.01 was joined. Assuming the magnitudes of the lattice constants of both aluminum nitride and gallium mixed crystals, the lattice constant of aluminum nitride is 3.111 angstroms (A) and the lattice constant of gallium nitride is 3.180 (A). Is also based on Table 7.1 on page 148 of "III-V Group Compound Semiconductor" (published May 20, 1994), edited by Isamu Akasaki. Assuming that the Vegard's rule is satisfied (see FIG. 1 of JP-A-59-228776 and FIG. 8 of JP-A-4-192585, from the lower left column in the upper left column of page (2)). Line), the lattice constants of both mixed crystals were calculated to be 3.172 (A) for the Al _0.12 Ga _0.88 N mixed crystal and 3.179 (A) for the Al _0.01 Ga _0.99 N mixed crystal layer. Is calculated. Based on gallium nitride doped with indium and zinc, which constitutes the light emitting layer, the degree of lattice mismatch between the light emitting layer and the layer joined to the light emitting layer is represented by δ (see the text), and Al _0.12 Ga _0.88 Δ for the N mixed crystal is 0.264%, and δ for the Al _0.01 Ga _0.99 N mixed crystal layer is 2.2 × 10 ⁻² %. Looking at the correspondence between the degree of lattice mismatch (δ) with the light emitting layer and the above-mentioned indium atom concentration in the active layer, as shown in FIG. 17, the degree of lattice mismatch (δ) between indium atoms and the light emitting layer is shown. _Are distributed so that the atomic concentration increases toward the junction interface with the Al _0.12 Ga _0.88 N mixed crystal layer in which the ratio is increased. In the evaluation by electron beam EBIC (E lectron B eam I nduced C urrent) method to check the presence or absence of the occurrence of current induced can confirm the presence of the pn junction by irradiation, pn junction Al _0.12 Ga _0.88 N mixed crystal It was confirmed that it existed near the interface between the layer and the light emitting layer. Therefore, it can be said that indium in the light emitting layer is distributed so as to increase the degree of lattice mismatch (δ) and increase the atomic concentration toward the interface with the p-type bonding layer. That is, in the wafer of this embodiment, indium is doped with distribution in the layer thickness direction, and indium is doped with distribution such that the concentration of indium increases toward the junction interface with the active layer having a higher degree of lattice matching. The active (light emitting) layer was provided.
(Example 2)
In the present embodiment, the present invention will be described by taking as an example a wafer in which gallium nitride having a donor impurity concentration distribution corresponding to the indium atom concentration distribution in the layer is used as a light emitting layer. As the substrate, (0001) (c-plane) sapphire having the same mechanical specifications as in the first embodiment was used. The sapphire used in this example had been previously surface-treated at a temperature of about 1200 ° C. in a hydrogen stream, but the gallium nitride buffer layer described in Example 1 and the gallium nitride layer doped only with silicon were used. Before forming under the same conditions as described in Example 1, the substrate surface was cleaned at 1100 ° C. for 40 minutes in a MOCVD reactor with approximately 8 liters of high-purity hydrogen flowing per minute. Heat treatment.

After the heat treatment of the sapphire substrate was completed, the kind and flow rate of the gas flowing in the MOCVD furnace were instantaneously reset to those used when growing the gallium nitride buffer layer described in Example 1. After the change in the gas flow rate caused by the switching was recovered, a gallium nitride low-temperature buffer layer and a silicon-doped gallium nitride layer were sequentially deposited on the sapphire substrate under exactly the same conditions as in Example 1.

On the silicon-doped gallium nitride layer having the same layer thickness as in Example 1 and slightly reducing the carrier concentration toward the surface side, indium having basically the same layer thickness as in Example 1 is provided. A gallium nitride layer doped with zinc was deposited as a light emitting layer. The conditions were gradually changed with the elapse of the growth time of the light emitting layer, with the intention that indium had a distribution increasing in the direction of increasing the layer thickness of the light emitting layer as in Example 1. That is, in the first embodiment, the distribution of the atomic concentration was created by linearly increasing the flow rate of the hydrogen gas accompanied by the vapor of cyclopentadienylindium used as the indium source as the growth time increased. In the example, the distribution of the indium atomic concentration was given by changing the substrate temperature with time from the start to the end of the growth of the light emitting layer. The substrate temperature was reduced at a rate of 16 ° C./min for 5 minutes from 850 ° C. to 770 ° C. The initial temperature at which the temperature of the substrate is started to fall is set to 850 ° C., and the temperature at which the temperature reaches the drop is set to 770 ° C. when the supply amount of the high-purity hydrogen gas accompanying the sublimated indium source is set to 20 ml / min. The atomic concentration of indium doped in the gallium nitride layer at these substrate temperatures is about 9 to 10 × 10 ¹⁷ cm ^-3 and about 2 to 3 × 10 ²⁰ cm ^-3 , which has been previously determined by the present inventor. Because it was known from The amount of change (decrease) in the substrate temperature with respect to the growth time was precisely controlled by a PID type programmable temperature controller. Furthermore, the difference from the first embodiment is that the light emitting layer is intentionally doped with silicon (Si) as a donor impurity. This silicon doping is performed by using high-purity hydrogen gas containing 1.2 ppm (parts per million) of disilane (Si ₂ H ₆ ) in volume concentration and manganese (in which the inner volume is polished to 10 liters). Mn) Fulfilled by feeding from a steel high pressure cylinder. The addition amount (flow rate) of the disilane / high purity hydrogen mixed gas into the growth reaction system was 8 ml / min at the start of the growth of the light emitting layer, and linearly decreased to 1 ml / min after exactly 5 minutes. I let it. That is, as for the concentration distribution of indium atoms, the flow rate was reduced at a rate of 1.4 ml / min for the purpose of providing the distribution of the silicon element in the layer thickness direction. On the other hand, the supply flow rate of the mixed gas of diethyl zinc (volume concentration: 100 ppm) / high-purity hydrogen into the growth reaction system is kept constant, and the concentration distribution of zinc atoms in the light emitting layer is made almost constant as in Example 1. Intended. FIGS. 18 to 20 schematically show changes over time in the substrate temperature, the raw material system, and the flow rate of the doping gas during the growth of the light emitting layer. FIGS. 19 and 20 show changes in the flow rate, but not changes in the absolute value of the actual molecular concentration (molar concentration). The molar concentration of the dopant supplied to the growth reaction system can be calculated from the volume concentration of the dopant in the mixed gas and the molar (mol) volume at the temperature of the supply system of the doping gas.

ア ル ミ ニ ウ ム An aluminum nitride-gallium mixed crystal layer having an aluminum composition ratio of 0.12 and a gallium nitride layer were sequentially deposited on the light-emitting layer according to the method described in Example 1. Both layers were doped with zinc and magnesium during growth as described in Example 1. Further, after growing both layers, magnesium ions were implanted under the conditions of Example 1. After performing the ion implantation, a so-called ion at 1050 ° C. for 40 minutes in an ammonia-argon mixed gas stream composed of an ammonia gas having a flow rate of 2 liters per minute and an argon gas having a flow rate of 2 liters per minute. Implant annealing was performed. In the probe method in which the presence or absence of conductivity is easily evaluated based on the degree of current flow between tungsten (W) probes (probes), the conductivity was confirmed only after the annealing after the ion implantation. . In the as-grown state after the wafer was grown, no clear conductivity was obtained between the probes, and it was determined from the current-voltage characteristics that the resistance was rather high.

Cesium ions (Cs ⁺⁾ or oxygen ions (O ₂ ⁺⁾ flight type to the primary beam: by (T ime o f F light TOF ) SIMS, was measured in particular indium emitting layer portion, the concentration distribution of silicon and zinc . FIG. 21 shows the concentration distribution of these dopants in the light emitting layer portion in the thickness direction of the light emitting layer. The layer thickness of the light emitting layer was 57 nm when determined from the depth profile of the dopant. The lower cladding layer (117) made of an n-type aluminum nitride-gallium mixed crystal (Al _0.01 Ga _0.99 N) having an aluminum composition ratio (x) of 1% in the lower layer (substrate side) of the active layer (118) is indium atom concentration. At the interface with, it was 1 × 10 ¹⁸ cm ⁻³ . On the other hand, at the interface between the upper clad layer made of p-type Al _0.12 Ga _0.88 N mixed crystal to be joined above the active layer (118) (119) was monotonically increased to 2 × 10 ²⁰ cm ^-3. Thus, the distribution of the indium atomic concentration in the light emitting layer in the increasing direction of the layer thickness is not limited to the means for changing the supply amount (flow rate) of the indium source described in the first embodiment. It has been found that this can also be achieved by means that change over time. However, it can be said that the distribution of the atomic concentration of indium resulting from the means for changing the growth temperature of the substrate over time tends to monotonously increase in the atomic concentration of indium in the direction of increasing the layer thickness. The concentration distribution of indium atoms is increased as shown by a slightly downwardly convex distribution curve (profile) as compared with the case of using the means for increasing the supply flow rate of the indium source over time as in the first embodiment. (See FIGS. 17 and 21 for comparison). It is not known at this time whether this is a unique distribution caused by the means for changing the substrate temperature over time, or whether it depends on mechanical factors such as the structure of the MOCVD furnace.

The silicon inside the active layer (118) exhibited a tendency opposite to that of indium atoms, as evident in the concentration distribution (profile) of the atoms. That is, from the concentration of 1 × 10 ¹⁸ cm ⁻³ at the interface with the lower cladding layer (117) made of n-type Al _0.01 Ga _0.99 N) bonded below the active layer (118), the upper part of the active layer (118) Of the p-type Al _0.12 Ga _0.88 N mixed crystal _{monocrystallinely} decreased to 1 × 10 ¹⁸ cm ⁻³ toward the interface with the upper cladding layer (119) (see FIG. 21). On the other hand, zinc was uniformly distributed at a concentration of 8 × 10 ¹⁸ cm ⁻³ substantially in the thickness direction of the light emitting layer regardless of the mode of atomic concentration distribution of indium and silicon.

In a sputtering method using argon (Ar) that provides an etching effect similar to the ion sputtering method in SIMS analysis, zinc and magnesium are doped together during film formation above the light emitting layer, and ion implantation of magnesium is performed after film formation. The p-type gallium nitride layer and the p-type Al _0.12 Ga _0.88 N mixed crystal layer thus formed were removed by etching to expose the surface of the indium, silicon and zinc-doped gallium nitride light emitting layer. The surface of the exposed light-emitting layer was very slightly, but slightly yellowish. Furthermore, when the surface was polarized and observed with a low-power differential interference optical microscope, circular or elliptical hill-like projections (undulations) were observed. In photoluminescence (PL) measurement at room temperature, only a band edge of gallium nitride near 365 nm and a spectrum near 430 nm estimated to be caused by a level related to zinc were observed. No PL emission in the near ultraviolet region near 380 to 390 nm, which is peculiar to the gallium-indium nitride mixed crystal, which is considered to easily appear when silicon is intentionally doped, was not observed.

In summary, in the present embodiment, gallium nitride doped with indium at a concentration that does not achieve formation of a gallium nitride-indium mixed crystal, and provided with a concentration distribution of indium atoms by means different from that of the first embodiment, A wafer having the layer as a light emitting layer was formed.
(Example 3)
In the present embodiment, the same substrate, the light emitting layer and the layer structure above the light emitting layer as in Example 1 were provided, but the indium-doped gallium nitride layer was provided as a layer to be bonded below the light emitting layer. 2 shows a configuration example of a wafer. First, a buffer layer made of gallium nitride was deposited on the sapphire substrate described in Example 1 under the conditions of Example 1. After the growth of the buffer layer for exactly 15 minutes, mixing of the hydrogen bubbling gas accompanying the vapor of trimethylgallium into the hydrogen carrier gas is performed by opening and closing a valve installed in the piping system of the group III source supply system. Stopped instantly.

Next, in order to suppress fluctuations in the pressure in the reactor, the gas type constituting the carrier gas was first instantaneously switched from hydrogen to argon gas at the same flow rate by opening and closing a valve provided in the piping system. Thereafter, in about 2 minutes, the flow rate of the argon gas was gradually reduced from 8 liters per minute to 5 liters per minute by an electronic mass flow meter (MFC). The ammonia gas flow rate was 1 liter per minute and was not changed. After changing the carrier gas to argon and adjusting its flow rate, the amount of electric power supplied to the above-mentioned resistance heating type heater is controlled through a thyristor type electric circuit to control the temperature of the sapphire substrate to silicon-doped gallium nitride. The temperature was raised to 1100 ° C., which is the growth temperature of the growth layer. After reaching 1100 ° C., it is determined from the temperature reading that the transient of the temperature has attenuated. After about 5 minutes, the flow rate of the argon gas is reduced from 5 liters per minute to 3 liters per minute, and at the same time hydrogen gas is removed. The reactor was again fed into the reactor at a flow rate of 3 liters per minute. That is, the carrier gas was composed of hydrogen and argon having the same flow rate. At the same time, the flow rate of ammonia gas was increased from 1 liter per minute to 4 liters per minute. Changes in the flow rates of these gases were all completed in parallel within about 2 minutes. In the time zone in which these flow rates were changed, disilane (Si ₂ H ₆ ) gas diluted to a volume concentration of about 1.2 ppm with high-purity hydrogen gas at a flow rate of 5 ml / min. Before the gallium source was supplied, the doping gas (doping source) of silicon (Si) was previously supplied into the reactor. Exactly 3 minutes after the change of the flow rate, the growth of the gallium nitride growth layer was started by mixing trimethylgallium and a hydrogen bubbling gas accompanied by trimethylaluminum vapor in the carrier gas having the above composition. At this time, the temperature of the stainless steel bubbler container containing trimethylgallium was 13 ° C. using commercially available ethylene glycol as a heat transfer (cooling) medium by the electronic constant temperature layer. The flow rate of hydrogen for bubbling liquid trimethylgallium was 5 ml / min. The temperature of the stainless steel bubbler container containing trimethylaluminum and having an inner surface mirror-polished was set to 25 ° C. using commercially available ethylene glycol as a heat transfer (cooling) medium by the same electronic thermostat as described above. The flow rate of hydrogen for bubbling the liquefied trimethylaluminum was 5 ml / min. While maintaining the flow conditions of these various gases, the mixing of the hydrogen bubbling gas accompanied by the vapor of trimethylgallium into the carrier gas having the above composition is continued for 90 minutes to obtain a silicon gas having a layer thickness of about 3.5 μm. A (Si) -doped n-type aluminum nitride / gallium nitride mixed crystal layer was grown. After exactly 90 minutes, the addition of hydrogen bubbling gas accompanied by vapors of trimethylgallium and trimethylaluminum and disilane gas, which is a silicon doping gas, to the hydrogen / argon mixed carrier gas is stopped, and silicon-doped n-type aluminum nitride is removed. -Gallium layer growth is completed. Under the same conditions, the carrier concentration of the silicon-doped n-type aluminum-gallium nitride mixed crystal layer deposited on the above-mentioned gallium nitride buffer layer was constant at about 3 × 10 ¹⁸ cm ⁻³ . .

(4) While the flow rate of the ammonia gas was kept as it was, the supply of the hydrogen gas, which constituted the carrier gas, to the growth reaction system was cut off, and two types of gas, argon and ammonia, were introduced into the growth reaction system. This gas switching was performed within about 2 minutes after the supply of the above-described hydrogen bubbling gas accompanying the vapor of trimethylgallium to the growth reaction system was cut off. Thereafter, the substrate temperature was lowered from 1100 ° C. to 850 ° C. at a rate of 20 ° C./min. After the temperature dropped to 850 ° C., the hunting of the temperature was eliminated in about 3 minutes and the substrate temperature was stabilized. After that, the supply of hydrogen bubbling gas accompanied by the vapor of trimethylgallium, the indium source and the silicon doping source was started to supply indium and indium. A gallium nitride layer doped with silicon was formed. At this time, the container of the above-mentioned stainless steel bubbler containing trimethylgallium was kept at 0 ° C. by an electronic thermostat. The flow rate of the high-purity hydrogen gas for bubbling the liquid trimethylgallium was 3.2 ml / min by an electronic mass flow meter (MFC). As the indium source, cyclopentadienyl indium, which hardly causes a gas-phase reaction with electron-donating ammonia, was used. The indium source is housed in a stainless steel bottle whose inner surface is mirror-polished as in the case of the first and second embodiments. The bottle was kept at 60 ° C. by an electronically controlled thermostat. A high-purity hydrogen gas was supplied at a flow rate of 20 per minute to supply the sublimated indium source to the growth reaction system along with the sublimated indium source in the bottle containing the indium source. The n-type gallium nitride layer doped with indium and silicon was grown for 60 minutes. The silicon doping conditions were the same as those for the growth of the silicon-doped n-type gallium nitride. Thus, an n-type gallium nitride layer doped with both indium and silicon was deposited on the silicon-doped n-type gallium nitride layer as a bonding layer to be bonded below the light emitting layer (substrate side).

Thereafter, the gallium nitride light emitting layer doped with indium, silicon, and zinc, and the aluminum-gallium nitride doped with zinc and magnesium are mixed with the conditions of the flow rate of the carrier gas and the substrate temperature described in Example 1 as they are. A crystal layer and a gallium nitride layer doped with zinc and magnesium were sequentially formed on the gallium nitride junction layer doped with indium and silicon. For the aluminum-gallium nitride mixed crystal layer and the gallium nitride layer doped with zinc and magnesium, magnesium ions having a mass number of 24 were implanted under the same conditions as in Example 1. The wafer subjected to the ion implantation was subjected to ion implantation annealing under the conditions described in Example 1.

After the ion implantation annealing, the concentration distribution of the constituent elements and the dopant in the depth direction from the surface of the wafer to the deep portion on the substrate side was measured by SIMS. The lamination structure determined by SIMS analysis was as follows. That is, first, a gallium nitride layer having a layer thickness of about 100 nm, an aluminum nitride-gallium mixed crystal layer having an aluminum composition ratio of 0.12, indium, silicon, and zinc are formed from the front side of the wafer. The structure was composed of a gallium nitride layer having a doped layer thickness of about 60 nm, a gallium nitride layer having a layer thickness of about 200 nm doped with indium and silicon, and a gallium nitride layer doped only with silicon. The presence of the gallium nitride low temperature buffer layer was not clearly recognized from the SIMS analysis. On the other hand, no difference was observed between the dopant and the distribution form described in Example 1. In particular, it was recognized that the distribution of indium, silicon and zinc in the gallium nitride layer doped with indium, zinc and silicon grown as the light emitting layer was almost the same in the direction of increasing the layer thickness. The indium and silicon atoms in the indium and silicon-doped gallium nitride layer junction, which is disposed below the light emitting layer and is in contact with the light emitting layer, are substantially distributed within the layer due to their distribution in the depth direction (concentration profile). Uniform distribution was observed. The atomic concentration of indium in the bonding layer was measured to be about 6 × 10 ¹⁹ cm ⁻³ . In the bonding layer, silicon was distributed almost uniformly in the thickness direction of the same layer at a substantially constant concentration (about 6 × 10 ¹⁹ cm ⁻³ ). Although the atomic concentration of silicon in the bonding layer is larger than the concentration of silicon atoms (about 3 × 10 ¹⁹ cm ⁻³ ) in the silicon-doped n-type gallium nitride layer disposed below the bonding layer, The carrier concentration of the bonding layer was measured to be about 10 to 15% lower than that of the silicon-doped n-type gallium nitride layer (9 × 10 ¹⁷ cm ⁻³ ) disposed below the bonding layer. . In this comparison, a silicon-doped n-type gallium nitride layer and a gallium nitride layer doped with indium and silicon were separately formed on the above-mentioned gallium nitride buffer layer separately under the conditions described in this example. However, each was measured under a zero (0) bias by a general electrolytic CV method, that is, a comparison on a donor concentration.

Separately, a wafer produced under the above-described conditions and confirmed by SIMS analysis to have the same lamination structure and substantially the same dopant distribution was etched from the surface layer by an argon ion (Ar ⁺ ) sputtering method. One sample was etched by sputter etching until the surface of the gallium nitride light emitting layer doped with indium, silicon and zinc was exposed. In other samples, the surface of the indium- and silicon-doped gallium nitride layer bonded to the lower part of the light emitting layer was exposed by further etching. The sample whose surface of the light emitting layer was exposed was irradiated with He-Cd laser light having a wavelength of 325 nm to observe a PL spectrum. In the PL spectrum at room temperature, band edge emission of the gallium nitride crystal was observed at a wavelength of about 365 nm, and light emission presumed to be caused by zinc was observed at about 430 nm. The wavelength position of band edge emission obtained from another undoped gallium nitride crystal was also about 365 nm. The intensity of blue light emission near a wavelength of about 430 nm observed in an n-type gallium nitride layer doped only with zinc and silicon having substantially the same thickness as that of the light emitting layer is about 1/50 of that of the present embodiment. It was about. From this, it was interpreted that the effect of indium doping is that high-intensity blue luminescence emission in a 430 nm wavelength band was obtained as described in this example. On the other hand, even at room temperature PL from the indium-doped junction layer disposed below the light-emitting layer, the band of the gallium nitride crystal whose wavelength is about 365 nm is slightly weaker in light emission intensity than that of the light-emitting layer. Only edge emission appeared. As in the PL spectrum from the surface of the light-emitting layer, no light was observed in the near-ultraviolet band near the wavelength of about 380 to 390 nm, which is likely to appear due to silicon doping. In addition, since the bonding layer was not doped with zinc, no blue luminescence light having a wavelength of about 430 nm appeared. Therefore, although the gallium nitride light-emitting layer and the gallium nitride layer bonded to the light-emitting layer are both doped with indium, the position (wavelength) at which the band edge emission appears coincides with that of the gallium nitride crystal. It was determined that it was not gallium indium nitride converted, but rather gallium nitride doped with indium. In addition, the coincidence of the wavelengths at which the band edge emission appears in both layers means that both layers have the same band gap (forbidden band width). That is, from the viewpoint of the junction structure, there are a total of two nitride compound semiconductor layers forming a junction with the light emitting layer, and one of the nitride semiconductor layers is made of the same material as the light emitting layer, that is, gallium nitride. Was formed.

A portion of the wafer obtained in this example was cut in the vertical direction to produce a strip-shaped thin piece. After the preparation of the thin section, a part of the cross section of the thin section was further thinned by argon (Ar) ion etching, and was subjected to cross-sectional TEM (transmission electron microscope) observation. The presence of the gallium nitride buffer layer, which was unclear by the SIMS method as the elemental analysis, was clearly determined by the cross-sectional TEM method. The thickness of the gallium nitride buffer layer determined by measuring the length from the captured bright-field image was about 6.5 nm. In a region corresponding to the gallium nitride buffer layer imaged by the bright field method, at least one dark line image (contrast image) supposed to be caused by the existence of dislocations has a width of 100 angstroms (A). It was present in a large amount at a density of about 10 ¹² cm ^-2 . However, observation under a high magnification such that a lattice image of the interface between the sapphire c-plane substrate and the gallium nitride buffer layer can be confirmed shows that at least several layers of lattice images are arranged on the sapphire substrate surface. Was done. The dark line that linearly penetrates upward through the gallium nitride buffer layer, that is, the dislocation is substantially at the center of the aluminum nitride-gallium mixed crystal layer having a thickness of 3.5 μm on the gallium nitride buffer layer, and is vertical due to stacking faults and the like. There were signs that propagation upwards was suppressed. The dislocation propagating to the indium-doped gallium nitride junction layer and the light emitting layer, which is almost halfway up the thickness of the aluminum nitride-gallium mixed crystal layer, clearly reduced its density. In addition, almost no new occurrence of dark contrast due to misfit dislocation was observed at the interface between the indium-doped gallium nitride bonding layer and the indium-doped light-emitting layer. Therefore, in the bright-field cross-sectional TEM image of the indium-doped light-emitting layer, it was confirmed that a region where a dark line image was not observed was enlarged. It was interpreted that this was because the light emitting layer and the bonding layer had a lattice matching relationship, and the density of dislocations and crystal defects generated at the bonding portion between the two layers was suppressed. When the observation region width reached about 1 μm in total, dark lines were observed concentrated in one or two extremely limited regions inside the light emitting layer and the bonding layer. The dark contrast was due to threading dislocations originating at or near the interface with the gallium nitride buffer layer. In some regions, these threading dislocations reached the surface of the above-described aluminum-gallium nitride mixed crystal and the outermost gallium nitride layer above the light-emitting layer. Pits were present on the surface of the gallium nitride layer corresponding to positions where threading dislocations penetrated.
(Example 4)
In this embodiment, in addition to the first bonding layer that forms a bond with the above-described light emitting layer, an example in which the second bonding layer that further bonds to the first bonding layer is formed of a group III nitride semiconductor doped with indium. Show. In short, the indium-doped gallium nitride bonding layer (first bonding layer) described in Example 3 above was provided on the indium-doped gallium nitride layer (second bonding layer) as well. An example of a wafer having a configuration will be described.

An n-type aluminum-gallium nitride mixed crystal layer doped with both indium and silicon was grown on the gallium nitride buffer layer described in Example 3 above. The mixed crystal layer was grown at 1100 ° C. in a mixed gas flow (carrier gas) composed of argon gas and hydrogen gas set at a flow rate of 3 liters per minute and ammonia gas set at a flow rate of 4 liters per minute. did. Before the start of mixed crystal growth, disilane (Si ₂ H ₆ ) gas diluted to a volume concentration of about 1.2 ppm with high-purity hydrogen gas was added at a flow rate of 5 ml / min. Next, the growth of the gallium nitride growth layer was started by mixing the carrier gas having the above composition with a hydrogen bubbling gas accompanied by trimethylgallium and trimethylaluminum vapor. At this time, the temperature of the stainless steel bubbler container containing trimethylgallium was set to 13 ° C. using commercially available ethylene glycol as a heat transfer (cooling) medium by the electronic constant temperature layer. The flow rate of hydrogen for bubbling liquid trimethylgallium was 5 ml / min. The temperature of the stainless steel bubbler container containing trimethylaluminum and having an inner surface mirror-polished was set to 25 ° C. using a commercially available ethylene glycol as a heat transfer (cooling) medium by the same electronic thermostat as described above. The flow rate of hydrogen for bubbling the liquefied trimethylaluminum was 5 ml / min.

In this example, the addition of the hydrogen bubbling gas accompanying trimethylgallium and trimethylaluminum to the carrier gas was started, and at the same time, the addition of the high-purity hydrogen gas accompanying the indium source was started. This is for obtaining an aluminum-gallium nitride mixed crystal layer doped with indium. As the indium source, Lewis acidic cyclopentadienyl indium which hardly causes a gas phase reaction with Lewis basic ammonia was used. The indium source was housed in a stainless steel bottle whose inner surface was mirror-polished as in Examples 1 to 3. The bottle was kept at 60 ° C. by an electronically controlled thermostat. A high-purity hydrogen gas was supplied at a flow rate of 20 ml / min to supply the growth reaction system with the sublimated indium source in the bottle containing the indium source. The growth of the n-type gallium nitride layer doped with indium and silicon continued for 90 minutes with a time intended for a layer thickness of about 3.5 μm. Silicon is obtained by adding a high-purity hydrogen-disilane (Si ₂ H ₆ ) mixed gas containing disilane (Si ₂ H ₆ ) in a volume concentration of about 1.2 ppm to the above carrier gas at a flow rate of 5 ml / min. Doped. Thus, an n-type gallium nitride layer doped with both indium and silicon was deposited as a second bonding layer on the gallium nitride buffer layer.

On the second bonding layer, an n-type gallium nitride layer serving as a first bonding layer doped with indium and silicon according to the conditions described in Example 3, and a light emitting layer doped with indium, silicon, and zinc Gallium nitride layer, an aluminum / gallium mixed crystal layer doped with zinc and magnesium (upper cladding layer), and a gallium nitride layer doped with zinc and magnesium (contact layer). On the aluminum-gallium nitride layer and the gallium nitride layer doped with zinc and magnesium, magnesium ions were implanted under the implantation conditions described in Example 3 after film formation. Accordingly, a light emitting layer, a first bonding layer bonded to the light emitting layer, and a second bonding layer bonded to the first bonding layer are each provided with a bonding structure in which a gallium nitride based semiconductor layer doped with indium is used. A wafer was prepared.

The obtained wafer was sliced by an ion thinning method. Next, based on the bright-field image taken by the cross-sectional TEM method, observation was made mainly on the state of generation of dislocations near the interface between the first and second bonding layers and the interface between the first bonding layer and the light emitting layer. The number of dark contrast lines in the aluminum gallium nitride layer was reduced by about 15% as compared with the case of Example 3. Although a new dislocation was rarely found near the interface between the first and second bonding layers, the occurrence of dislocations was observed at the bonding interface between the first bonding layer and the light emitting layer. The number was clearly lower than in the case of No. 3. The dislocation density of the light emitting layer was generally estimated to be 5 to 10 × 10 ⁴ cm ⁻² . In summary, although dislocations penetrating the light emitting layer almost linearly from the gallium nitride buffer layer to reach the surface of the outermost gallium nitride contact layer still exist, both the first and second bonding layers are indium doped layers. Was found to be effective in reducing the dislocation density of the light emitting layer.
(Comparative Example 1)
The same indium source, silicon source and zinc source as described in Example 1 were used on an aluminum nitride-gallium mixed crystal layer (Al _0.01 Ga _0.99 N) having the same aluminum composition ratio as that of Example 1 as 0.01. Then, a gallium nitride layer doped with indium, silicon, and zinc was stacked as a light emitting layer. However, in the light emitting layer in this comparative example, the distribution of the atomic concentrations of silicon and zinc was the same as in Example 1, but the distribution of the atomic concentration of indium was given in the thickness direction of the light emitting layer. And constant. That is, the light emitting layer in this comparative example is composed of gallium nitride as in Example 1, but differs from Example 1 in the atomic concentration distribution of indium.

During the growth of the light emitting layer, the amount of supply of indium source (cyclopentadienyl indium (I)) into the growth reaction system (sublimation) in order to provide a substantially constant concentration distribution of indium atoms in the increasing direction of the thickness of the light emitting layer. The flow rate of hydrogen gas accompanying the indium source was kept constant at 20 ml / min. The holding temperature of the stainless steel storage container that stores the indium source was a constant temperature of 60 ° C. An aluminum nitride-gallium mixed crystal layer and a gallium nitride layer were formed on the light-emitting layer having a substantially constant indium concentration distribution in the layer thickness direction under the same conditions and operations as described in Example 1, And
(Comparative Example 2)
As in Comparative Example 1, a gallium nitride layer having a substantially constant distribution of indium atom concentration was used as the light emitting layer. In Comparative Example 2, doping was performed so that the concentration distribution of indium atoms in the gallium nitride light emitting layer was substantially constant, and the atomic concentration of silicon (Si) was also constant in the direction of increasing the layer thickness in the light emitting layer. did.

In order to keep the indium atom concentration in the gallium nitride light emitting layer substantially constant in the thickness direction of the light emitting layer, the supply amount of the indium source into the growth system was kept constant without changing over time. The indium source consistently, using a cyclopentadienyl indium to monovalent valencies (C ₅ H ₅ In). In growing a gallium nitride layer doped with indium, silicon, and zinc as a light-emitting layer on an aluminum nitride-gallium mixed crystal layer having an aluminum composition ratio of 0.01, hydrogen gas accompanying the sublimated indium source was used. The supply rate to the growth reaction system was fixed at 20 ml / min. Exactly 5 minutes later, the addition of hydrogen gas accompanying the indium source into the growth reaction system was stopped. Zinc doping followed the method and conditions described in Example 1. On this light emitting layer, an upper clad layer made of aluminum nitride-gallium mixed crystal and a gallium nitride contact layer similar to those described in Example 1 were laminated.

The concentration distribution of the dopant in the layer thickness direction in the light emitting layer was analyzed by the SIMS method, and compared with the concentration distribution in Comparative Example 1. As a result, indium in Comparative Example 2 was distributed in the thickness direction of the light emitting layer almost in the same manner as in Comparative Example 1. The indium atom concentration at the junction interface between the gallium nitride light-emitting layer doped with indium and the like and the aluminum nitride-gallium mixed crystal layer with the aluminum composition ratio of the lower layer being 0.01 is about 9 × 10 ¹⁷ cm ⁻³ . there were. On the other hand, the indium concentration at the joint interface with the aluminum nitride-gallium mixed crystal (upper cladding layer) having an aluminum composition ratio of 0.12 was also about 1 × 10 ¹⁸ cm ⁻³ . That is, the indium atom concentration was almost constant as in Comparative Example 1. The atomic concentrations of silicon and zinc were also substantially constant in the thickness direction of the light emitting layer. The thickness of the light emitting layer determined from the concentration profile obtained by SIMS analysis was about 56 nm. In summary, the gallium nitride light emitting layer obtained in Comparative Example 2 was doped with dopants of indium, silicon, and zinc so as to be substantially constant in the direction of increasing the layer thickness. In the gallium nitride light emitting layer of Comparative Example 2, neither the donor impurity (silicon) nor the zinc (acceptor impurity) was distributed so as to be substantially in inverse proportion to the concentration distribution of indium atoms. The concentration distribution was almost the same as that of atoms.
(Comparative Example 3)
Except for the gallium nitride light emitting layer, the configuration was the same as that of Example 1. That is, an aluminum nitride / gallium mixed crystal layer having an aluminum composition ratio of 0.01 is provided below the gallium nitride light emitting layer on the substrate side, and an aluminum monium nitride / gallium mixed crystal having an aluminum composition ratio of 0.12 is provided above the light emitting layer. The layers were joined. Therefore, comparing the aluminum nitride-gallium mixed crystal layer bonded to the gallium nitride light emitting layer with the degree of lattice mismatch (defined by δ in the text) with the gallium nitride forming the light emitting layer, The smaller the degree of lattice mismatch (δ) is the aluminum-gallium nitride bonding layer having an aluminum composition ratio of 0.01.

In this Comparative Example 3, a gallium nitride layer having an atomic concentration distribution different from that of Examples 1 to 4 and Comparative Examples 1 and 2 with respect to indium was used as the light emitting layer. The gallium nitride light-emitting layer of Comparative Example 3 is a gallium nitride-based bonding layer in which the indium atomic concentration is bonded to the light-emitting layer, and is bonded to a bonding layer having a smaller degree of lattice mismatch, that is, a lower portion of the light-emitting layer. It is distributed largely in the direction of the interface with the aluminum nitride-gallium mixed crystal layer where the aluminum composition ratio is 0.12. The doping condition of silicon was the same as that in Example 2, and the doping condition of zinc was in accordance with the method described in Example 1. Therefore, the gallium nitride light emitting layer of Comparative Example 3 is one in which the indium atom concentration is gradually reduced in the thickness direction of the light emitting layer, and silicon and zinc are doped so as to have a substantially constant atomic concentration.

The distribution of the atomic concentration of indium was adjusted by changing the growth temperature over time during the growth of the gallium nitride light emitting layer. That is, after growing an aluminum-gallium nitride mixed crystal layer (Al _0.01 Ga _0.99 N) doped with silicon at 1100 ° C., the substrate temperature was immediately lowered to 770 ° C. After the substrate temperature was stabilized, the amount (flow rate) of cyclopentadienyl indium (I) (C ₅ H ₅ In (I)) supplied to the growth reaction system as an indium source during the growth of the light-emitting layer was changed every time. The growth of the light emitting layer was started in a state where the light emitting layer was fixed at a constant value of 220 ml. From the start of the growth, the growth temperature was linearly increased to the growth temperature with the elapse of the growth time. The growth temperature was changed to 850 ° C. when the growth of the light emitting layer was completed exactly 5 minutes later. Therefore, the growth temperature change rate per unit time was 16 ° C./min. After the growth of the light emitting layer doped with indium, silicon, and zinc is completed, the growth temperature is instantaneously increased to 1100 ° C. again, and the aluminum composition ratio is adjusted to 0.12 according to the growth method and conditions described in Example 1. An aluminum nitride-gallium mixed crystal (upper cladding layer) and a gallium nitride contact layer were sequentially laminated.

FIG. 22 shows the atomic concentration distributions of the indium, silicon, and zinc dopants in the gallium nitride light emitting layer obtained by the SISM analysis. The concentration of indium atoms (indicated by (In) in FIG. 22) gradually decreases in the light emitting layer from the vicinity of the interface with the Al _0.01 Ga _0.99 N layer in the direction of the interface with Al _0.12 Ga _0.88 N. Was something. The concentration of indium atoms decreased almost linearly from about 2 × 10 ²⁰ cm ⁻³ to 1 × 10 ¹⁸ cm ⁻³ . On the other hand, the concentration of the dopant which affects the conductivity type of the concentration distribution of silicon and zinc atoms (indicated by (Si) and (Zn) in FIG. 22 respectively) is almost constant in the layer thickness direction inside the light emitting layer. Was. The characteristic of the gallium nitride light emitting layer of Comparative Example 3 from the viewpoint of the distribution state of the dopant is that the atomic concentration of indium increases in the direction of reducing the degree of lattice mismatch as opposed to Example 1, ie, the layer thickness Are distributed so that the atomic concentration gradually becomes smaller in the direction of increasing. In summary, the concentration distribution of indium atoms in the layer thickness direction is reversed from that of the light emitting layer described in the first embodiment.

The wafers formed by the methods described in Examples 1 to 4 and Comparative Examples 1 and 2 were processed under the same conditions to produce blue light emitting diodes (LEDs). First, a titanium nitride (TiN) film having a thickness of about 100 nm was formed on the entire surface of each wafer by a general high frequency sputtering method. A silicon nitride (SiN) film having a thickness of about 150 nm was deposited on the titanium nitride film by plasma CVD using silane gas (SiH ₄ ) and ammonia gas (NH ₃ ) as raw materials. Next, a positive type AZ-1350 photoresist was spin-coated on the silicon nitride film. Thereafter, the resist material was patterned using a known ultraviolet photolithography technique. The chip shape of the LED drawn by patterning was common to each wafer, and was a square having a side of about 350 μm.

After removing only the resist material in the area where the scribe line and the n-type electrode (positive electrode) on the outer periphery of the square chip are to be arranged with a peeling material mainly composed of an aromatic compound, the silicon nitride exposed in the area is removed. The membrane was dissolved and removed with buffered hydrofluoric acid (HF). As a result, zinc and magnesium, which are the outermost layers of the wafer, are doped only in the area where the scribe line and the n-type electrode are to be formed, and the titanium nitride film on the surface of the gallium nitride contact layer into which magnesium is ion-implanted is exposed. Was. Next, the titanium nitride film exposed by a general plasma etching method using a mixed gas of argon / methane / hydrogen or the like and the constituent layers of the wafer were removed by etching only in regions where the n-type electrodes and scribe lines were to be formed. In the four wafers described in Examples 1 and 2 and Comparative Examples 1 and 2, the plasma etching was allowed to proceed until the surface portion of the n-type aluminum nitride / gallium mixed crystal deposited on the gallium nitride low temperature buffer layer was exposed. Was. The plasma etching was performed on the wafers described in Examples 3 and 4 until the surface portion of the n-type gallium nitride layer was exposed. Due to this etching, a step was formed between the exposed surface of the region where the n-type electrode and the scribe line were to be formed and the surface of the outermost gallium nitride layer not to be etched. After completion of the etching, the photoresist material on the silicon nitride film was peeled off with an organic solvent mixed solution, leaving only the silicon nitride film. Again, after the above photoresist material was spin-coated on the entire surface of the wafer, the resist material was exposed and developed according to the shapes of the n-type and p-type electrodes (negative electrodes) to be formed and patterned. After this patterning, the titanium nitride film was exposed in a region where the resist material was stripped in the shape of the p-type electrode. On the other hand, the surface of the n-type gallium nitride layer or the n-type aluminum / gallium nitride layer was exposed in the region where the resist material was stripped in the shape of the n-type electrode. Thereafter, rod-shaped high-purity aluminum (Al) having a nominal purity of 6N (99.9999%) was applied to the entire surface of the wafer by vacuum evaporation. After taking out the wafer from the vacuum deposition apparatus, the aluminum-deposited wafer was immersed in a resist stripping solution, and the aluminum-deposited film deposited on the resist material was peeled off from the resist material by a normal lift-off method. By this lift-off step, rectangular opposing n-type and p-type electrodes each having a width of about 100 μm and a length of about 300 μm were formed.

(4) A diamond blade (cutting blade) was run along a scribe line provided so as to intersect the outer periphery of each chip in a grid pattern, thereby achieving isolation (insulation) between the LED chips. After confirming that the insulation between the chips was maintained, the main optical and electrical characteristics of the LED chips were sampled and inspected using a general prober. Table 1 summarizes the measurement results for each example and each comparative example. The light emission intensity was measured using a photodetector (abbreviated as PD) in a chip state without applying a mold made of epoxy resin. Both the forward and reverse voltages are voltage values in volts (V) at a current value of 10 microamps (μA).

く As summarized in Table 1, there is no significant difference in the emission center wavelength. It is presumed that the subtle difference in the emission center wavelength is due to the subtle variation in the doping concentration of zinc for each growth. It was the emission intensity that caused a particularly significant difference in characteristics. From the comparison between Example 1 and Comparative Example 2, when the atomic concentration distributions of silicon and zinc were similar, the emission intensity was clearly increased when the distribution of indium atomic concentration was given. In the case where the concentration of indium atoms in the gallium nitride light emitting layer is gradually increased toward the bonding interface with the gallium nitride based bonding layer, which increases the degree of lattice mismatch with the light emitting layer, the atomic concentration of silicon is reduced. The emission intensity was increased by making the opposite to the increasing / decreasing tendency of the indium atom concentration (as compared with Examples 1 and 2). On the other hand, when the atomic concentration distribution of silicon was the same, the emission intensity was increased by assigning the distribution to the indium atom concentration (by comparing Example 2 with Comparative Example 1). Even if the indium atom concentration is the same, the obtained emission intensity increases when indium is increased toward the interface with the gallium nitride-based bonding layer, which has a higher degree of lattice mismatch (δ in the text). Was achieved (by comparison between Example 1 and Comparative Example 3).

LED chips produced from the wafers described in Examples 1 to 4 were cut into single chips along a scribe line with a diamond scriber. Thereafter, the n-type and p-type electrodes were subjected to wire bonding so that the current between the two electrodes was measured via a lead wire. Typical electron beam induced excitation current method in this state, it was observed LED chip section in a so-called abbreviated EBIC (E lectron B aam I nduced C urrennt) method. As a result, a band-like contrast was observed in the region near the interface between the aluminum-gallium nitride mixed crystal to which zinc and magnesium were added and the n-type gallium nitride light-emitting layer to which indium, silicon, and zinc were doped. As a result, formation of a pn junction between the two layers became clear. Since this EBIC measurement used a general SEM (scanning electron microscope), it was not possible to obtain a sufficiently high magnification so that a light emitting layer having a thin layer thickness of approximately 50 to 60 nm was imaged at a high resolution. Therefore, it is clearly determined whether the region where the contrast due to the pn junction is maximum is located at the interface between the upper cladding layer made of the aluminum nitride-gallium mixed crystal and the gallium nitride light emitting layer or inside the upper cladding layer. I could not do it.

A region near the interface between the gallium nitride light-emitting layer and the aluminum-gallium nitride mixed crystal layer bonded to the gallium nitride light-emitting layer was observed by a cathodoluminescence (CL) method using a high-resolution analytical electron microscope (abbreviation: analytical electron microscope). As the sample, a chip obtained by thinning a chip used in the above EBIC method by an ion etching method using argon ions was used. As described in Example 1, the atomic concentration of indium was increased toward the interface with the Al _0.12 Ga _0.88 N mixed crystal layer which increased the degree of lattice mismatch (δ in the text) with the gallium nitride light emitting layer. In some cases, the CL intensity near the interface between the upper cladding layer and the light emitting layer was found to be stronger than in other regions. In the light emitting layer in which indium atoms exist at a substantially constant concentration as in Comparative Examples 1 and 2, there was no tendency to increase the CL intensity particularly near the interface. In a chip in which indium atoms are increased toward an Al _0.01 Ga _0.99 N mixed crystal layer which _{reduces the} degree of lattice mismatch with gallium nitride constituting the light emitting layer (see Comparative Example 3), Al _0.01 Ga _0.99 It was also seen that the CL intensity at the interface between the N mixed crystal layer and the gallium nitride light emitting layer was slightly increased as compared with other regions. Regardless of the distribution of the CL intensity in the light-emitting layer, it is a qualitative judgment for visual observation, but the overall intensity from the light-emitting layer is expressed as δ which describes the concentration distribution of indium atoms in the light-emitting layer in the text. It was the largest when increasing toward the larger bonding layer. Conversely, when the concentration distribution of indium atoms is increased in the direction of the interface with the bonding layer where δ is small, the CL intensity is rather weak as compared with the case where the atomic concentration distribution of indium in the light emitting layer is substantially constant. Met. When the atomic concentration distribution of indium is the same, the distribution of the doping concentration is given to either the donor or the acceptor impurity (in the present embodiment and the comparative example, the donor impurity is given a concentration distribution) to increase the CL intensity. An increasing trend was observed. In particular, the CL intensity increased when the atomic concentration distribution of doped silicon was substantially opposite to that of indium atoms. The magnitude of the CL intensity coincided with the magnitude of the emission intensity (see the emission intensity of Examples 1 and 2 in Table 1). When the indium atom concentration is increased in the interface direction of the bonding layer where δ is large, and the silicon atom concentration distribution is substantially reversed, the CL intensity observed from the light emitting layer portion is obtained by doping the light emitting layer with indium. It further increased by joining the gallium nitride based compound semiconductor layer. Since the CL intensity is arranged below the light-emitting layer, it seems that the CL intensity tends to increase as the number of indium-doped gallium nitride-based compound semiconductor layers increases. This increase in CL intensity was considered to correspond to the tendency of increase in light emission intensity of the LED chips of Examples 3 and 4 shown in Table 1.

(4) The magnitude of the CL intensity generally corresponds to the intensity of the light emission intensity of the LED chip. Here, the point deduced from the CL measurement results is simply described. The concentration distribution of indium atoms increases in the direction of the interface between the nitride compound semiconductor forming the light emitting layer and the bonding layer having a large degree of lattice mismatch. In this case, the CL intensity increases. Although the cause cannot be conclusively determined, doping indium more in the junction region where lattice distortion exists due to lattice mismatch in the first place than in other regions causes more indium to accumulate in that region. Is also presumed to be relevant. It is presumed that even if the concentration of indium atoms is distributed so as to increase in the direction of the interface having a low degree of lattice mismatch, the lattice strain is small, so that accumulation of indium at a very low concentration does not occur. It is considered that an appropriate doping concentration range of indium atoms for increasing the emission intensity is determined depending on the amount of indium accumulated at this interface. It is uncertain whether indium accumulated toward the junction interface causes further local distortion near the interface and influences on the electronic state such as emission center impurities near the interface.

In a light emitting device made of a nitride compound semiconductor, there is an effect of increasing light emission intensity.

It is a cross section showing the structure of the conventional blue LED. It is a cross section showing the structure of the conventional green LED. FIG. 4 is a diagram illustrating a relationship between indium concentration and photoluminescence intensity. FIG. 3 is a diagram illustrating an example of an indium concentration distribution inside an active layer according to the present invention. FIG. 3 is a diagram for explaining an initial concentration (N ₀ ) and a final concentration (N) of indium. It is a figure which shows the difference of luminescence intensity in comparison. FIG. 2 is a schematic plan view of a blue LED according to the first embodiment. FIG. 8 is a schematic cross-sectional view of the LED shown in FIG. 7 along the broken line A-A ′. FIG. 5 is a diagram showing a change over time in the temperature of the substrate during the growth process of the laminate described in Example 1. FIG. 4 is a diagram showing a change over time in a hydrogen gas flow rate in Example 1. FIG. 4 is a diagram illustrating a change over time in the flow rate of an argon gas in Example 1. FIG. 4 is a diagram showing a change over time in the flow rate of ammonia gas in Example 1. FIG. 4 is a diagram showing a change in the supply amount of a group III raw material during a growth process of the laminate described in Example 1. FIG. 4 is a diagram illustrating a change in the supply amount of trimethylgallium during the growth process of the stacked body described in Example 1. FIG. 4 is a diagram illustrating a change in the supply amount of trimethylaluminum during the growth process of the laminate described in Example 1. FIG. 4 is a diagram illustrating current-voltage characteristics for a surface layer portion of the laminate described in Example 1. FIG. 3 is a diagram showing a concentration distribution in the depth direction of a constituent element and a dopant inside a laminate described in Example 1. FIG. 9 is a view showing a growth temperature profile of Example 2. FIG. 8 is a diagram showing a change over time in the flow rate of a raw material carrier gas in Example 2. FIG. 9 is a diagram showing a change over time in the flow rate of a dopant gas in Example 2. FIG. 9 is a diagram illustrating a concentration distribution of a dopant in a depth direction in Example 2. FIG. 9 is a diagram illustrating a concentration distribution in a depth direction of a dopant of Comparative Example 3.

Explanation of reference numerals

(101) Substrate (102) Buffer layer (103) Lower cladding layer (104) Active layer (light emitting layer)
(105) upper cladding layer (106) contact layer (107) intermediate layer (108) bonding layer bonded to the active layer (109) bonding layer bonded to the active layer (110) degree of lattice mismatch between the active layer and the active layer Is a curve showing the concentration distribution of (111) indium atoms at a bonding interface with a bonding layer in which is larger.
(112) Distribution curve indicating that the concentration of indium atoms is increasing stepwise (113) Other distribution curve indicating that the concentration of indium atoms is increasing stepwise (114) FIG. 11 is a diagram showing a room temperature photoluminescence spectrum of Example 1 with a distribution curve (115) indicating that the increase is monotonic.
(116) A diagram showing a room temperature photoluminescence spectrum of Comparative Example 1. FIG.
(117) Lower cladding layer (118) Active layer (light emitting layer)
(119) Upper cladding layer (120) Curve showing current-voltage characteristics after nitride compound semiconductor laminate formation (121) Current-voltage characteristics after ion implantation and annealing in nitride compound semiconductor laminate Curve (122): p-type electrode (123) n-type electrode (124): junction interface between the active layer and the bonding layer that reduces the degree of lattice mismatch between the active layer and the active layer

Claims (1)

An active layer made of a group III nitride semiconductor containing gallium and indium, a first cladding layer made of a nitride semiconductor of a first conductivity type containing gallium, and a second conductive layer containing gallium on a single crystal substrate Semiconductor light-emitting device including a second clad layer made of a nitride semiconductor layer having a shape of a nitride, wherein a nitride compound in which indium is added between the light-emitting layer and the clad layer in contact with the light-emitting layer A nitride semiconductor light-emitting device comprising a bonding layer made of a semiconductor.

Images