JP2014078756A - Nitride semiconductor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a high-quality nitride semiconductor which exhibits high luminous efficiency when use as a light-emitting element.SOLUTION: A nitride semiconductor comprises one conductivity type nitride semiconductor part, a quantum well active layer structure and a nitride semiconductor part having another conductivity type opposite to the one conductivity type, which are sequentially laminated on a base substrate having a principal surface of a non-polar nitride by crystal growth. The one conductivity type nitride semiconductor includes a first nitride semiconductor layer and a second nitride semiconductor layer, which are sequentially laminated. The second nitride semiconductor layer has a thickness of 400 nm-20 μm and has a non-polar outermost surface. By selecting the above-described substrate as the substrate for crystal growth, spatial separation between an electron and a hole, which contributes to light emission based on the QCSE effect is inhibited thereby to achieve effective radiation. In addition, by setting the second nitride semiconductor layer at an adequate thickness, a nitride semiconductor surface is prevented from becoming extremely and highly irregular.

Description

  The present invention relates to a high-quality nitride semiconductor excellent in surface state and optical characteristics.

  A blue light emitting element or an ultraviolet light emitting element can be made into a white light source in combination with an appropriate wavelength conversion material. Such white light sources have been extensively studied for application as backlights for liquid crystal displays, light-emitting diode illumination, automotive lighting, or general lighting instead of fluorescent lamps, and some of them have already been put into practical use. ing. At present, such blue light emitting elements and ultraviolet light emitting elements are mainly used for growing gallium nitride semiconductor crystal thin films by techniques such as metal organic chemical vapor deposition (MOCVD) and molecular beam epitaxy (MBE). Which are collectively referred to as gallium nitride based light emitting diodes or GaN based LEDs.

Conventionally, most GaN-based LED substrates are sapphire substrates. Since sapphire and GaN have greatly different lattice constants, it is inevitable that a considerable number of dislocations of about 10 ⁹ / cm ² are introduced in a GaN crystal obtained by epitaxial growth on a sapphire substrate. However, the sapphire substrate has an advantage that it is cheaper than the SiC substrate or the GaN substrate. In addition, InGaN in the blue light emitting region, which is usually used as a quantum well layer of a GaN-based LED, its light emission efficiency is not very sensitive to the dislocation density. For these reasons, the sapphire substrate is still the main substrate at present.

  However, when a gallium nitride semiconductor crystal is viewed as a device material used in a high carrier density situation, the above-described high-density dislocations result in a significant decrease in device characteristics. For example, in a device such as a high-power LED or laser, if the dislocation density is high, the element life is remarkably shortened. In addition, when the active layer structure does not contain In at all (for example, when an AlGaN layer is used as the active layer), or in order to realize short wavelength emission below the near ultraviolet region, the In composition is relatively small (for example, When an InGaN layer or an InAlGaN layer is included in the active layer structure, the dependence of internal quantum efficiency on the dislocation density increases, and when the dislocation density is high, the emission intensity itself decreases.

  That is, when the active layer structure does not contain In at all, or when the active layer structure includes an InGaN layer or InAlGaN layer with a relatively small In composition, the active layer structure includes an InGaN layer having a long emission wavelength of blue or more. Compared to the case, the demand for lowering the dislocation density becomes severe.

In the case of aiming at such a low dislocation density, it is effective to use a GaN substrate as a substrate for epitaxial growth, whereby the dislocation density found in the epitaxial layer is 10 ⁸ pieces / cm ² or less, or It is expected to be 10 ⁷ pieces / cm ² or less. Further, if dislocations of the substrate or the like are further reduced, it is expected to be 10 ⁶ pieces / cm ² or less. That is, compared to the case of using a sapphire substrate, it is expected that the dislocation density is reduced by 2 to 3 digits or more. Under such circumstances, a GaN free-standing substrate or an AlN free-standing substrate is suitable as a substrate for epitaxial growth of a gallium nitride based semiconductor crystal.

A conventional attempt to epitaxially grow a gallium nitride based semiconductor crystal on a GaN substrate, which is a nitride substrate, is that the epitaxial growth surface is on a c-plane (that is, (0001) plane) substrate, that is, on a “polar plane”. Most are related to epitaxial growth. In addition,
Examples of such reports include Patent Document 1 (Japanese Patent Laid-Open No. 2005-347494), Patent Document 2 (Japanese Patent Laid-Open No. 2005-310772), Patent Document 3 (Japanese Patent Laid-Open No. 2007-67454), and the like. is there.

  In Patent Document 1, a nitride substrate ((0001) plane GaN substrate) which is a polar surface is used as a substrate for epitaxially growing a GaN layer, and the GaN substrate is cleaned at a furnace pressure of 30 kilopascals. After that, the first n-type GaN buffer layer having a thickness of 1 μm was grown while maintaining the substrate temperature at 1050 ° C. and the furnace pressure at 30 kilopascals. A method of forming a second n-type GaN buffer layer having a thickness of 1 μm by heating to a substrate temperature of 1100 ° C. while being held at 30 kilopascals is disclosed. A semiconductor device having a buffer layer with excellent crystal quality and excellent surface flatness is provided.

  Patent Document 2 discloses a process of removing dirt and moisture such as organic substances adhering to the surface of the GaN substrate while flowing hydrogen gas, nitrogen gas and ammonia gas, and simultaneously improving the crystallinity of the substrate surface. Later, while flowing nitrogen gas and hydrogen gas, an intermediate layer having a multilayer structure composed of a GaN layer and an InGaN layer was formed on the GaN substrate, and a reflective layer, an active layer, and a gallium nitride based semiconductor layer were provided on the intermediate layer. An invention of a light emitting element is disclosed.

  Further, in Example 26 of Patent Document 3, a 3 μm-thick Si-doped n-type GaN buffer layer formed on a GaN substrate is provided, and a laminated structure is formed on the n-type GaN buffer layer. The invention is disclosed. It is described that a buffer layer of 300 mm or less formed at a low temperature of about 500 ° C. may be provided between the n-type GaN buffer layer and the GaN substrate.

However, the surface flatness and optical characteristics of nitride semiconductor crystals obtained by epitaxial growth on a substrate having a polar surface as a main surface are not necessarily sufficient, and further growth conditions need to be studied. In addition to the above, problems inevitably caused by selecting a polar plane as a growth substrate are also recognized. For example, in a quantum well active layer structure (for example, a quantum well active layer structure made of InGaN / GaN) formed on a hexagonal c ⁺ plane including a c-plane GaN substrate, the so-called quantum confined Stark effect (QCSE) As a result, there is a problem that the recombination probability of electrons and holes is reduced, and the luminous efficiency becomes lower than the ideal luminous efficiency.

Against this background, attempts have been made to fabricate a quantum well active layer structure on a substrate whose main surface is a nonpolar plane, but the nonpolar plane of a hexagonal III-V nitride crystal A
Epitaxial growth on the plane, r-plane, and m-plane is considered difficult, and in particular, a high-quality hexagonal III-V nitride semiconductor multilayer structure is not obtained on an m-plane substrate.

  For example, Non-Patent Document 1 (Proceedings of the 66th JSAP Scientific Lecture Meeting (Autumn 2005) 11p-N-4) states that “it is difficult to grow on the non-polar m-plane”. As described above, on the premise of the difficulty, a GaN substrate having a c-plane as a main surface is subjected to stripe processing such that the m-plane is exposed as a side wall, and the stripe side wall (m-plane) is nitrided Special crystal growth methods such as epitaxial growth of semiconductor crystals have been reported.

  Non-Patent Document 2 (Appl. Phys. Lett., Vol.82, No.11 (17 March 2003), p.1793-1795) describes a plasma assist on a ZnO substrate having an m-plane as a main surface. It has been reported that a GaN layer is grown by MBE (Molecular Beam Epitaxy) method. However, it is described that the obtained GaN layer is “slate-like”, and the surface irregularities are severe, and a single crystal GaN layer is not obtained.

Further, in Patent Document 4 (US Pat. No. 7,208,393), when a nitride semiconductor is crystal-grown on an m-plane substrate, defects are likely to occur at the growth start interface, and the defects are reduced. The following methods are described as means for achieving this. First, after a GaN buffer is grown on the m-plane substrate by the H-VPE (Hydride Vapor Phase Epitaxy) method, the substrate is once taken out from the reactor to form a dielectric mask, which is processed into a stripe shape. . Then, a substrate on which such a mask is formed is used as a new substrate, and an epitaxial film is planarized by lateral growth (called LEO or ELO growth) from the opening of the stripe-shaped mask by MOCVD. .
JP 2005-347494 A JP 2005-310772 A JP 2007-67454 A U.S. Patent No. 7208393 66th JSAP Scientific Lecture Proceedings (Autumn 2005) 11p-N-4). Appl. Phys. Lett., Vol.82, No.11 (17 March 2003), p.1793-1795. CJ Humphreys et al., “Applications Environmental Impact and Microstructure of Light-Emitting Diodes” MICROSCOPY AND ANALYSIS, NOVEMBER 2007, pp.5-8. K. SAITOH, “High-resolution Z-contrast Imaging by the HAADF-STEM Method”, J. Cryst. Soc. Jpn., 47 (1), 9-14 (2005). K. WATANABE; “Imaging in High Resolution HAADF-STEM”, J. Cryst. Soc. Jpn., 47 (1), 15-19 (2005). H. Matsumoto, et. Al., Material Transactions, JIM 40, 1999, pp.1436-1443.

  According to the study by the present inventors, the crystal growth method on the c-plane GaN substrate premised on the growth on the polar surface described in the above literature is applied to the crystal growth on the nonpolar surface. It was confirmed that it was difficult to obtain a nitride semiconductor with excellent surface flatness, and the optical properties of the obtained film were not sufficient.

  Further, a conventional crystal growth method on a nonpolar surface will be described. In the method described in Non-Patent Document 1, a nitride semiconductor crystal having a nonpolar surface is usually formed only on the width of several microns on the side surface of the stripe mesa. Since it can only be obtained, it is extremely difficult to increase the area, resulting in a great restriction on device fabrication. On the other hand, with the technique described in Non-Patent Document 2, although the area can be increased, the surface of the obtained nitride semiconductor film exhibits a rugged morphology, and a single crystal layer cannot be obtained. Furthermore, the method described in Patent Document 4 has a problem in that the process is complicated, and positions with many defects and positions with few defects are mixed in relation to the position of the underlying stripe.

  The present invention has been made in view of such problems. The object of the present invention is to provide a high quality with a good growth characteristic of a non-polar surface, which has good optical characteristics and high luminous efficiency when used as a light emitting device. It is an object of the present invention to provide a nitride semiconductor.

  As a result of intensive studies, the inventors of the present invention achieve the above object by laminating a specific nitride semiconductor layer to a specific thickness on the nitride main surface of a nonpolar nitride substrate. I found. Moreover, the nitride semiconductor layer having such a structure is avoided during the epitaxial growth in order to avoid a nitride semiconductor surface having extremely uneven surfaces and to improve the light extraction efficiency when it is made into an element. It has been found that even when the uneven surface is formed in a self-forming manner, good internal quantum efficiency in the quantum well structure can be realized at the same time if the unevenness is moderate. Such a technical idea overturns the conventional technical idea that was aimed at improving the surface morphology, and is a major feature of the present invention.

  That is, in the nitride semiconductor of the present invention, at least one main surface of the nitride main surface of the base is a nonpolar nitride, one conductivity type nitride semiconductor portion, quantum well active layer structure portion, A nitride semiconductor in which nitride semiconductor portions of the other conductivity type opposite to the one conductivity type are sequentially stacked, wherein the nitride semiconductor portion of the one conductivity type includes a first nitride semiconductor layer and a second nitride semiconductor layer. A nitride semiconductor layer is sequentially stacked, and the second nitride semiconductor layer has a thickness of 400 nm to 20 μm and the outermost surface is a substantially nonpolar surface.

  Preferably, the first nitride semiconductor layer and the second nitride semiconductor layer have different compositions from each other.

  The base is, for example, a self-supporting substrate made of GaN, AlN, InN, BN, or a mixed crystal thereof. The nitride of the main surface of the base is, for example, any of a sapphire substrate, a SiC substrate, a ZnO substrate, a Si substrate, a GaN substrate, an AlN substrate, an InN substrate, a BN substrate, or a free-standing substrate that is a mixed crystal thereof. A GaN film, an AlN film, an InN film, a BN film, or a mixed crystal film thereof grown on the crystal.

  Preferably, the nitride main surface of the substrate is a crystal plane of (1-100) plane (m-plane) ± 10 degrees or less.

Preferably, at least one of the first nitride semiconductor layer and the second nitride semiconductor layer is GaN, AlN, InN, BN, or a mixed crystal group III-V nitride semiconductor thereof. .

The thickness L1 of the _first nitride semiconductor layer is preferably 0.1 nm or more and 300 nm or less. The silicon (Si) concentration in the first nitride semiconductor layer is preferably 1 × 10 ²¹ cm ⁻³ or less. Furthermore, the silicon concentration in the second nitride semiconductor layer is preferably 1 × 10 ¹⁷ cm ⁻³ or more and 6 × 10 ¹⁹ cm ⁻³ or less.

In the present invention, the quantum well active layer structure may emit light having a center wavelength of 400 ± 30 nm. Moreover, it is preferable that the dislocation density in the said quantum well active layer structure part is 1 * 10 ^{< 9} > cm ^<-2> or less.

  In the present invention, the quantum well active layer structure portion may have an internal quantum efficiency of 20% or more obtained from CW-PL measurement under a low excitation density condition.

  In the present invention, the quantum well active layer structure portion may have an internal quantum efficiency of 20% or more obtained from pulsed light PL measurement under a low excitation density condition.

  Furthermore, in the present invention, the quantum well active layer structure portion has a photoluminescence lifetime (τ (PL)) of 1.0 ns or more obtained from time-resolved PL measurement at room temperature and under low excitation density conditions. It can be.

  In the present invention, when obtaining a nitride semiconductor in which one conductivity type nitride semiconductor portion, a quantum well active layer structure portion, and the other conductivity type nitride semiconductor portion opposite to one conductivity type are sequentially laminated, Crystal growth is performed on a substrate having a polar nitride main surface, while a conductive nitride semiconductor portion is formed by sequentially laminating a first nitride semiconductor layer and a second nitride semiconductor layer, and At the same time, the second nitride semiconductor layer had a thickness of 400 nm or more and 20 μm or less and the outermost surface was a substantially nonpolar surface.

  By selecting a nonpolar plane as the substrate for crystal growth, spatial separation of electrons and holes contributing to light emission based on the QCSE effect is suppressed, and efficient radiation is realized. In addition, by making the thickness of the second nitride semiconductor layer appropriate, it is avoided that the nitride semiconductor surface exhibits extremely severe unevenness, but even when the surface is moderately roughened, When the thickness is moderate, good optical characteristics can be obtained when a light emitting element is used.

  As a result of these effects, according to the present invention, the internal quantum efficiency is high and the light extraction efficiency is also good. As a result, the light emitting element has a high light emission efficiency, and the nonpolar surface is the growth surface. A high-quality nitride semiconductor is easily provided.

  As described above, the nitride semiconductor of the present invention includes a nitride semiconductor portion of one conductivity type and a quantum well active layer structure portion on a nitride main surface of a substrate whose main surface is a nonpolar nitride. A nitride semiconductor portion of the other conductivity type opposite to the one conductivity type is sequentially stacked, and the one conductivity type nitride semiconductor portion includes a first nitride semiconductor layer and a first nitride semiconductor layer. Two nitride semiconductor layers are sequentially stacked, and the second nitride semiconductor layer has a thickness of 400 nm or more and 20 μm or less and the outermost surface is a substantially nonpolar surface.

  Here, the substantially nonpolar plane means that, although the outermost crystal plane includes a nonpolar plane, appropriate minute irregularities and facet planes are formed when a light emitting element is formed. The nitride semiconductor of the present invention is formed on the nitride main surface of the substrate, which is a nonpolar nitride. As described later, in the crystal growth process, the first nitride semiconductor layer and the second nitride are formed. Appropriate irregularities may be formed in a self-forming manner and intentionally at the interface between the crystal layers of the semiconductor layer and the quantum well active layer or at the outermost surface of the nitride semiconductor layer. Therefore, the “substantially nonpolar plane” in the present invention is that the main surface of the substrate surface for epitaxial growth is a nonpolar plane, but the outermost surface of the epitaxial growth plane formed thereon is not necessarily a crystal plane. It must not be a uniform non-polar surface, or the outermost surface of the epitaxial layer must include a non-polar surface, but the entire surface should not be a non-polar surface and part of it may include a semipolar surface or a polar surface. Means.

  Hereinafter, the nitride semiconductor of the present invention will be described in detail while sequentially illustrating an example of the manufacturing process. In addition, as long as the nitride semiconductor of this invention has the above-mentioned structure, the manufacturing method is not limited to the following manufacturing method.

  There are various epitaxial growth methods. The method used for crystal growth of the nitride semiconductor of the present invention (this method is referred to as “the crystal growth method of the nitride semiconductor of the present invention” or “the epitaxial growth method of the present invention”). ) Is mainly applicable to the vapor phase growth method, and among them, it can be preferably applied to the H-VPE method and the MOCVD method, and most preferably the MOCVD method.

For example, in the MOCVD method, apparatus configurations having various configurations can be applied, and depending on each apparatus configuration, a gas constituting a main atmosphere at the time of temperature rise / fall, a gas constituting a main atmosphere at the time of growth, Efficient contact and supply of source gas, gas used as carrier for realizing supply of organic metal and some dopants, gas for diluting source, source gas intake and gas constituting atmosphere Auxiliary gas for gasification, gas for adjusting the flow such as laminarization of the entire gas flow, gas for stabilizing / extending the life of components such as heaters and various ports, and opening the reactor For this purpose, a gas or the like to be introduced is appropriately introduced.

  In the present invention, among these, the flow constituted by the auxiliary gas for improving the efficiency of the incorporation of the gas constituting the source gas / atmosphere into the substrate, the contact with the substrate and the supply thereof, or the gas flow For the sake of convenience, the flow of gas for adjusting the flow such as laminating the whole is referred to as “subflow”. In addition, for the sake of convenience, a flow composed of a gas that does not directly contribute to epitaxial growth, such as a gas for stabilizing / extending the life of components such as a heater and various viewports, a gas introduced for opening the reactor, etc. , “Non-growth flow”.

  On the other hand, in the present invention, all gas flows other than the sub-flow and the non-growth flow supplied into the crystal growth apparatus are referred to as “main flow” for convenience. Therefore, the main flow is mainly used as a carrier for realizing the supply of the gas constituting the main atmosphere at the time of temperature rise / fall, the gas constituting the main atmosphere at the time of growth, the source gas, the organic metal, and a part of the dopant. It is a general term for flows such as gas used and gas for diluting raw materials. This main flow is substantially the atmosphere itself in which the surface of the base for epitaxial growth of the nitride semiconductor or the crystal surface of the nitride semiconductor during epitaxial growth is exposed. Therefore, the main flow is indispensable for vapor phase growth, while the sub-blowing and non-growth flow are optional.

  FIG. 1A and FIG. 1B show examples of horizontal and vertical MOCVD reactors, respectively, and conceptually show the flow of the main flow. For example, in a horizontal reactor (FIG. 1A), the surface of the substrate 3 placed on the susceptor 2 housed in the quartz reaction tube 1 is exposed to the main flow MF, and the main flow MF is exposed to the main flow MF. It is an apparatus having a configuration that provides a virtual “atmosphere” for the substrate 3. The main flow MF is pressed against the surface of the base 3 by the subflow SF, and the gas flow constituting the raw material gas / atmosphere is efficiently taken into the base 3 and is contacted and supplied to the base 3, and the entire gas flow is performed. Also laminarize.

  On the other hand, in the configuration of FIG. 1B shown as an example of a vertical reactor, gas for subflow is not supplied, and the gas flowing inside the quartz reaction tube 1 is only due to the main flow MF. In both the horizontal reactor (FIG. 1 (A)) and the vertical reactor (FIG. 1 (B)), the gas by the out-of-growth flow OF is supplied by gas supply as heater purge, viewport purge, etc. The flow of is occurring.

  Furthermore, in this specification, when nitride semiconductors are epitaxially grown, including when the temperature rises / falls, a flow mainly containing a gas serving as a nitrogen source, or nitrogen desorption from the substrate / epitaxial layer surface is prevented. For the sake of convenience, the flow for forming the atmosphere for suppressing may be referred to as a “first main flow”. In this case, the gas flow mainly for supplying other raw materials and forming the atmosphere is sometimes referred to as “second main flow” for convenience. Further, since a part of the gas constituting the main flow can be used as a carrier gas for supplying the organometallic raw material, a part of the gas constituting the main flow may be described as a carrier gas.

Here, the “active gas” used in the present invention means an atomic or molecular substance that decomposes or reacts under temperature and pressure conditions in a series of epitaxial crystal growth processes such as temperature increase, temperature decrease, standby, and growth processes. It is a gas that can generate active species of hydrogen, such as atomic hydrogen radicals, atomic or molecular hydrogen ions, and atomic hydrogen, and is introduced as the main gas in the main flow. It is what is done. Further, the amount exceeds 1% in the flow rate ratio of the constituent gas species in the main flow at least at any stage of the epitaxial growth process.

Therefore, hydrogen (H ₂ ) gas or ammonia (NH ₃ ) gas (including a mixed gas thereof) is exemplified as the main active gas. Such a gas has an etching effect on the nitride crystal, and particularly H ₂ gas has a very large effect. Therefore, when the surface of the nitride crystal is excessively exposed to these gases (particularly H ₂ gas), nitrogen desorption from the nitride surface tends to occur excessively, Level defects are likely to be introduced, and may also cause excessive macro surface roughness. If it becomes like this, the internal quantum efficiency of the quantum well active layer structure part in this invention will fall too much, and it will become a nitride semiconductor which is not suitable for making into a light emitting element.

On the other hand, when used under appropriate conditions, the main active gas such as hydrogen (H ₂ ) gas or ammonia (NH ₃ ) gas (including a mixed gas thereof) can be converted into the internal quantum of the quantum well active layer structure. It is possible to form moderate irregularities on the nitride semiconductor intentionally and intentionally while maintaining the efficiency within an appropriate range. For this reason, in this invention, the light extraction efficiency from a nitride semiconductor can be made high, and it is preferable.

  On the other hand, the “inert gas” is a gas that does not generate active species of hydrogen in a series of epitaxial crystal growth processes such as temperature increase, temperature decrease, standby, and growth processes, and constitutes a main flow. Of these gases, it is introduced as the main constituent gas, and the amount thereof exceeds 1% in the flow rate ratio of the constituent gas species in the main flow at least at any stage of the epitaxial growth step.

Examples of such inert gases include nitrogen (N ₂ ), helium (He), argon (Ar), xenon (Xe), and krypton (Kr). In addition, acetonitrile, azoisobutane, hydrazine compound 1,1-dimethylhydrazine, amine compounds dimethylamine, diethylamine, trimethylamine, triethylamine, triallylamine, triisobutylamine, azide compounds methylazide, ethylazide, phenylazide , Diethylaluminum azide, diethylgallium azide, trisdimethylaminoantimony, etc. also satisfy the above conditions as an inert gas, and these mixed gases are also included in the inert gas.

The details will be described later, in the present invention, the heating process (particularly the period t _A), the first growth step,
Any step of the second growth step is optional regardless of whether the main flow is mainly composed of inert gas or the main flow is mainly composed of active gas.

Here, in order to make the internal quantum efficiency of the quantum well active layer structure part relatively high in the present invention, the main flow is mainly composed of an inert gas, and the surface of the substrate or the growing outermost surface layer has excessive activity. It is preferable to control the atmosphere so as not to be exposed to gas (in particular, hydrogen gas having a large etching effect on nitride). In particular, the temperature raising step (especially period t _A ),
In the first growth step, the gas constituting the main flow is preferably controlled so as not to contain excessive hydrogen gas.

  On the other hand, in order to increase the light extraction efficiency from the nitride semiconductor in the present invention while keeping the internal quantum efficiency of the quantum well active layer structure within an appropriate range, the main flow may be mainly composed of the active gas. preferable.

  FIG. 2G and FIG. 2A are diagrams for explaining two typical sequence examples for explaining the nitride semiconductor crystal growth method of the present invention.

  FIG. 2G is an example of a sequence in which the most preferable nitride semiconductor layer can be formed in the present invention, and the nitride semiconductor layer is maintained while keeping the internal quantum efficiency of the quantum well active layer structure in an appropriate range. This is a sequence example in which moderate irregularities can be formed on the surface of the film in a self-forming manner and intentionally. For example, since a nitride semiconductor formed by such a method has improved light extraction efficiency due to unevenness or the like, the light emission efficiency when the nitride semiconductor is made into a light emitting element can be increased.

  On the other hand, FIG. 2A is an example of a sequence that can form a preferable nitride semiconductor layer in the present invention, and is an example of a sequence that can relatively increase the internal quantum efficiency of the quantum well active layer structure. For example, a nitride semiconductor formed by such a method does not necessarily have a high light extraction efficiency when it is fabricated, but has a relatively high internal quantum efficiency. Can be high.

Here, an example is shown in which a GaN film is epitaxially grown on a GaN free-standing substrate whose main surface is the (1-100) plane (m-plane) of the nonpolar planes. Note that the reactor for epitaxial growth is, for example, a metal organic vapor phase growth apparatus, a horizontal three-layer flow quartz reactor that uses atmospheric pressure growth as a normal condition, and a self-revolving reaction that uses reduced pressure growth as a normal condition. Examples thereof include a furnace (planetary reactor), a vertical SUS reaction furnace under reduced pressure growth under normal conditions, and the like.

  First, a substrate having at least one main surface made of nitride is prepared as a substrate for epitaxial growth, and this substrate is placed on a susceptor in a reactor for epitaxial growth and heated to a predetermined temperature (step A). ). The nitride of the main surface of the base is a main surface of a free-standing substrate that is GaN, AlN, InN, BN, or a mixed crystal thereof, or a sapphire substrate, a SiC substrate, a ZnO substrate, a Si substrate, a GaN substrate, Examples thereof include a GaN film, an AlN film, an InN film, a BN film, or a mixed crystal film thereof grown on any of an AlN substrate, an InN substrate, a BN substrate, or a free-standing substrate that is a mixed crystal thereof.

  In the epitaxial growth method of the present invention, unlike the epitaxial growth method on a non-polar side wall by so-called ELO (Epiaxial Lateral Growth) or the like, mask formation or uneven processing on the substrate surface is not essential but optional. Therefore, in the present invention, it is preferable that the direction of the main surface of the substrate and the direction in which epitaxial growth proceeds substantially coincide. In the present invention in which lateral growth is not an essential requirement, extremely high density dislocations are formed in the epitaxial layer in response to problems such as difficulty in increasing the area of the epitaxial layer grown on the nonpolar side wall surface and the mask structure. Problems such as occurrence of parts do not occur. As a result, it is possible to increase the area and reduce the dislocation density of the epitaxial layer whose growth surface is a nonpolar plane (for example, m-plane).

Therefore, according to the method of the present invention, it is possible to form an epitaxial layer having excellent light emission characteristics when formed into a device on a non-polar substrate, particularly an m-plane substrate, so that nitriding perpendicular to the epitaxial growth direction is possible. The area of the outermost surface of the physical semiconductor is preferably 30 mm ² or more and 500 cm ² or less. Incidentally, more preferably a is 50 mm ² or more 225 cm ² or less, more preferably 4 cm ² or more 120 cm ² or less.

  The lower limit value of the area of the outermost surface of the nitride semiconductor is defined so that the crystal growth process (and subsequent device fabrication process) can be easily performed. On the other hand, the upper limit value is specified so that the in-plane uniformity of the obtained epitaxial layer is secured.

The above upper limit will be described. First, in forming a homogeneous epitaxial layer in the MOCVD apparatus, the range (region) in which the uniform source gas can be supplied in the MOCVD apparatus is considered to be about 500 cm ² . This is the reason why the area of the outermost surface of the nitride semiconductor is preferably 500 cm ² or less.

Secondly, a substrate whose principal surface is a nonpolar surface such as an m-plane is more anisotropic than a substrate whose crystal axis in the principal surface has a polar surface such as a c-plane. For example, when a c-plane substrate is used, the a-axis is provided every 120 degrees in the plane, but in the case of an m-plane substrate, the a-axis and the c-axis are provided in the plane. The direction is completely different in nature. For this reason, when the temperature is high, the substrate tends to “distort” or “warp” inhomogeneously. Accordingly, even in such a case, in order to form a relatively uniform epitaxial layer on the substrate, the thickness is preferably 225 cm ² or less, and more preferably 120 cm ² or less.

  According to the study by the present inventors, the crystal growth method of the present invention is based on the atomic arrangement on the surface of the substrate, and the (11-20) plane (a plane), (1-102) plane (r plane), ( 1-100) plane (m-plane) and the like can be suitably used, and the crystal plane orientation of the principal plane is (1-100) plane (especially among nonpolar planes). m-plane) or a surface equivalent to these, suitable for epitaxial growth. As the substrate, for example, it is preferable to use a GaN free-standing substrate having a (1-100) principal surface, and when a GaN-based nitride semiconductor is grown on this by the crystal growth method of the present invention, a good quality epitaxial growth film is obtained. can get.

  In order to increase the light extraction efficiency from the nitride semiconductor in the present invention while keeping the internal quantum efficiency of the quantum well active layer structure in an appropriate range, it is also possible to positively use a substrate having a slightly inclined surface. preferable. In this case, the main surface has an inclination angle with respect to the just m-plane in the a-axis direction and the c-axis direction is usually ± 10.0 ° or less, preferably ± 7.0 ° or less, more preferably ± 5. It is preferable that the crystal plane is 0.0 ° or less, particularly preferably ± 3.0 ° or less, and most desirably about ± 1.5 °.

  In the present invention, when appropriate irregularities are formed on the surface of the nitride semiconductor layer in a self-forming manner, the outermost surface is a substantially nonpolar surface, and the Ra of the outermost surface is usually 100 nm or more, preferably Is 150 nm or more, usually 300 nm or less, preferably 250 nm or less.

  On the other hand, in order to relatively increase the internal quantum efficiency of the quantum well active layer structure in the present invention, the so-called “off angle” from each surface is usually ± 10.0 ° or less, preferably ± 5.0 °. Hereinafter, it is more preferably ± 3.0 ° or less, particularly preferably ± 1.0 ° or less, and most preferably ± 0.5 ° or less. With such an off angle, a nitride semiconductor having a good internal quantum efficiency, in particular, can be epitaxially grown on the substrate by the method of the present invention. Even when an m-plane substrate is used, the main surface has an inclination angle from the just m-plane of usually ± 10.0 ° or less, preferably ± 5.0 ° in both the a-axis direction and the c-axis direction. Hereinafter, the crystal plane is more preferably ± 3.0 ° or less, particularly preferably ± 1.0 ° or less, and most preferably ± 0.5 ° or less.

  The nitride semiconductor of the present invention formed using the substrate is, for example, when the outermost surface of the above-described one-conductivity-type nitride semiconductor portion is substantially m-plane, or, for example, the other-conductivity-type nitridation described later When the outermost surface of the physical semiconductor portion is substantially m-plane, the outermost surface is usually (1-100) plane ± 5.0 ° or less, preferably ± 3.0 ° or less, more preferably ± 1.0 °. Hereinafter, it is particularly preferably ± 0.5 ° or less.

The temperature increase in the step A is for setting the temperature of the base to 600 ° C. to 1350 ° C., which is the growth temperature of the first nitride semiconductor layer described later. It is executed by supplying the gas constituting the main flow so as to be 35 kilopascals to 120 kilopascals. In addition, when forming the laminated structure which is mentioned later on the nitride semiconductor layer, in order to make the optical characteristic of the laminated structure governing the optical element characteristic favorable, the preferable temperature rise temperature T in step A is preferable. _A is 600 ° C to 1350 ° C, more preferably 650 ° C to 1200 ° C, still more preferably 800 ° C to 1100 ° C, and most preferably 900 ° C to 970 ° C.

In addition, it is preferable that the temperature increase reachable temperature range of the process A and the subsequent film formation temperature range of the first nitride semiconductor layer coincide, for example, the first nitride semiconductor in a range of 800 ° C. to 1100 ° C., for example. In the case of forming a layer, the temperature rise reached in step A is also in the range of 800 ° C to 1100 ° C.

The temperature raising process illustrated in FIG. 2G includes a temperature raising stage (high temperature stage) in a period t _{A in} which the temperature of the substrate is raised to a predetermined temperature in an atmosphere constituting a main flow with an active gas, and this temperature raising period. In a relatively low temperature region before t _A , the temperature rising stage in a period t _{B in} which the temperature of the substrate is raised in an atmosphere in which the main flow is composed of a gas having a composition different from that of the main flow in the temperature rising period t _A ( (Low temperature stage). Hereinafter, for convenience, the temperature rising stage (low temperature stage) in the period t _B is referred to as a “first temperature rising process”, and the temperature rising stage (high temperature stage) in the period t _A is referred to as a “second temperature rising process” for convenience. Sometimes referred to as “warm process”.

On the other hand, the temperature raising step illustrated in FIG. 2A is performed in the period t _A where the temperature of the substrate is raised to a predetermined temperature in an atmosphere that includes an inert gas and an optional active gas. substrate and warm stage (high temperature phase), at a relatively low temperature region before the Atsushi Nobori period t _a, the main flow of the heating period t _a in an atmosphere composed of the main flow gas of a different composition This shows a case of two stages of a temperature rising stage (low temperature stage) in a period t _B during which the temperature is raised.

Note that among the gases exemplified in FIG. 2A, NH ₃ gas that is an active gas other than hydrogen gas has an arbitrary configuration, but if an appropriate amount described later is supplied, the nitride semiconductor layer is configured. It can be a good source gas of nitrogen. On the other hand, in the present invention, as a nitrogen raw material constituting the nitride semiconductor layer, an NH ₃ gas is not used and 1,1-dimethylhydrazine, which is an inert gas,
(1,1-DMHy), acetonitrile, azoisobutane, dimethylamine, diethylamine, trimethylamine, triethylamine, triallylamine, triisobutylamine, methyl azide, ethyl azide, phenyl azide, diethylaluminum azide, diethylgallium azide, trisdimethylaminoantimony, etc. It is also possible to use it.

Accordingly, the main flow in the first and second growth steps described later is a first main flow mainly including a nitrogen raw material supply gas (eg, NH ₃ gas in FIGS. 2G and 2A illustrated). And at least a second main flow including a gas for supplying an element other than nitrogen constituting the nitride semiconductor layer as a raw material.

The temperature raising stage in the period t _B in the relatively low temperature area is a temperature raising stage in a temperature range in which it is not necessary to positively suppress nitrogen desorption from the nitride constituting the substrate main surface. On the other hand, heating stage of period t _A at a relatively high temperature region is a Atsushi Nobori step in a temperature region of actively necessary to suppress the nitrogen desorption from the nitride constituting the substrate main surface.

Although it depends on the pressure inside the reactor where the substrate is placed, nitrogen desorption on the nonpolar plane is considered to occur at a lower temperature than other polar planes, considering its atomic arrangement, and is generally about 450 ° C or higher. In this temperature range, it is considered that it is necessary to positively suppress nitrogen desorption from the nitride surface. Therefore, in the region lower than this temperature, any gas of active gas and inert gas may be contained in any proportion as the gas constituting the main flow. For example, all of the main flow may be constituted by N ₂ gas which is an inert gas, or may be constituted only by NH ₃ gas which is an active gas.

In the period t _A during which the temperature is raised in the high temperature stage, consideration is necessary to appropriately suppress nitrogen desorption from the nitride surface. In this case, in order to increase the light extraction efficiency from the nitride semiconductor in the present invention while keeping the internal quantum efficiency of the quantum well active layer structure within an appropriate range, an active gas such as H ₂ gas and NH is used. It is also preferable to raise the temperature with only _three gases. This is preferable because appropriate irregularities can be formed on the surface of the nitride semiconductor layer and the like in a self-forming manner and intentionally.

On the other hand, in order to relatively increase the internal quantum efficiency of the quantum well active layer structure in the present invention, it is preferable to raise the temperature only with an inert gas. In the inert gas, for example, dimethylhydrazine is used. It is important that nitrogen source gas is included. On the other hand, in the case where the temperature is increased in an atmosphere that includes an inert gas and an appropriate amount of an active gas as an optional structure, the nitrogen source gas is included in at least one of the inert gas and the active gas. Is preferably included. For example, NH ₃ gas or the like can be a nitrogen source gas that is an active gas.

In this case, it is preferable that H ₂ is not excessively introduced as the active gas. On the other hand, NH ₃ gas is preferable because an appropriate amount of NH ₃ gas can be supplied because it can be a good nitrogen raw material for a nitride semiconductor. Although details will be described later, it is preferable that the flow ratio of the active gas is less than 0.5 among all the gas species constituting the main flow in the temperature raising step. Note that the temperature of the substrate may be raised in a constant atmosphere throughout the temperature raising process, and in this case, the period t _A coincides with the temperature raising process period.

  In addition, it is preferable that the pressure in the reactor in this temperature rising stage is adjusted to be 35 kilopascals to 120 kilopascals. The reason why the lower limit of the pressure in the reaction furnace is set to 35 kilopascals is that the optical characteristics are greatly deteriorated when the atmosphere exposed to the substrate surface is in an excessively reduced pressure state. This point will be described later.

  In the case where a laminated structure as will be described later is formed on the nitride semiconductor layer, the first and second nitride semiconductors are used in order to improve the optical characteristics of the laminated structure that governs the optical element characteristics. The film formation temperature of the layer is preferably in the range of 600 ° C to 1350 ° C, more preferably in the range of 650 ° C to 1200 ° C, still more preferably in the range of 800 ° C to 1100 ° C, and most preferably in the range of 900 ° C to 970 ° C. It is a range. In addition, it is preferable that the film formation temperature range of the nitride semiconductor layer and the temperature rise reaching temperature range of the process A coincide with each other. For example, when the nitride semiconductor layer is formed in the range of 800 ° C. to 1100 ° C. The temperature rise reached temperature is also in the range of 800 ° C to 1100 ° C.

In the sequence example shown in FIG. 2G, the first temperature rise is performed while supplying hydrogen gas as the second main flow as an active gas that becomes the main atmosphere in which the nitride main surface of the substrate is exposed in the reaction furnace. The process is started, and when the substrate temperature reaches 400 ° C., the supply of NH ₃ gas, which is an active gas, is added as a gas constituting the first main flow to start the second temperature raising process. The temperature is further raised in the gas to 1000 ° C., which is the ultimate temperature. The supply of NH ₃ gas in the second temperature raising step is for preventing nitrogen from escaping from the surface of the substrate during the temperature raising step and excessively reducing the crystallinity of the epitaxial growth surface. .

As described above, in order to increase the light extraction efficiency from the nitride semiconductor in the present invention while keeping the internal quantum efficiency of the quantum well active layer structure in an appropriate range, an active gas such as H ₂ gas and NH is used. _It is preferable to raise the temperature with only _three gases. In this way, it is possible to form moderate irregularities on the surface of the nitride semiconductor layer and the like in a self-forming manner and intentionally.

On the other hand, in the sequence example shown in FIG. 2A, the first main gas is supplied as the second main flow while supplying nitrogen gas as an inert gas that becomes the main atmosphere in which the nitride main surface of the substrate is exposed in the reaction furnace. When the temperature of the substrate reaches 400 ° C., the supply of NH ₃ gas, which is an active gas, is added as a gas constituting the first main flow to start the second temperature raising process. In this mixed gas, the temperature is further increased to 1000 ° C., which is the ultimate temperature. The supply of NH ₃ gas in the second temperature raising step is for preventing nitrogen from escaping from the surface of the substrate during the temperature raising step and excessively reducing the crystallinity of the epitaxial growth surface. .

  In this second temperature raising step, in order to make the internal quantum efficiency of the quantum well active layer structure portion in the present invention relatively high, all the gases constituting the main flow (in this case, the first main flow and the second main flow) The ratio of the components of the inert gas (nitrogen gas) to the sum of the main flows is preferably 0.5 or more and 1.0 or less in flow rate ratio. Such a mixed gas component is used when the substrate surface is heated in a relatively high temperature region, and when the active gas is excessively contained in the atmosphere in which the substrate surface is exposed. This is because defects are easily introduced into the nitride crystal on the surface. Therefore, in the second temperature raising step, it is preferable that the etching effect is particularly excessive and hydrogen gas that induces nitrogen depletion is not excessively contained in the atmosphere.

  Subsequent to the temperature raising step (step A), the process proceeds to the first nitride semiconductor layer growth step (step B). In the present invention, when the nitride main surface of the substrate is the m-plane, In the present invention, in order to increase the light extraction efficiency from the nitride semiconductor, while maintaining the internal quantum efficiency of the quantum well active layer structure portion in an appropriate range, there are appropriate irregularities. It is preferable because it can be self-forming.

In the process B shown in FIG. 2G, the supply of NH ₃ gas as the gas constituting the first main flow is continued, the supply of hydrogen gas is also continued for the gas constituting the second main flow, and the reaction After the atmosphere in the furnace is stabilized, a part of the hydrogen gas constituting the second main flow is used as a carrier gas for supplying the group III element material and the dopant material, and the epitaxial growth material is put into the reactor. Then, crystal growth of the nitride semiconductor layer is started.

On the other hand, in the process B shown in FIG. 2 (A), the supply of NH ₃ gas as the gas constituting the first main flow is continued, and the gas constituting the second main flow is also continuously supplied with the nitrogen gas. After the atmosphere in the reactor is stabilized, a part of the nitrogen gas constituting the second main flow is used as a carrier gas for supplying a group III element source and a dopant source, and an epitaxial growth source is used as the reactor. Then, crystal growth of the nitride semiconductor layer is started.

  The first nitride semiconductor layer thus formed is preferably a crystal containing no polycrystalline component, and more preferably composed of a single crystal itself. A second nitride semiconductor layer is further formed on the first nitride semiconductor layer. The layer formed in this way is a nitride semiconductor layer formed on a nonpolar plane. However, according to the present invention, the layer is crystallographically sufficiently long-range ordered.

  In step B, in order to increase the light extraction efficiency from the nitride semiconductor in the present invention while keeping the internal quantum efficiency of the quantum well active layer structure in an appropriate range, as shown in FIG. It is preferable to add Si dopant appropriately. When an appropriate amount is intentionally introduced, light irregularities are formed on the surface, so that the light extraction efficiency increases, and as a result, the light emission efficiency of the light emitting element using the nitride semiconductor of the present invention increases.

In particular, in the nitride semiconductor of the present invention, Ra is about 200 nm, and it is particularly preferable in the present invention to intentionally form a roughness equivalent to the wavelength of ultraviolet light, near ultraviolet light, and visible light. In the emission wavelength range to be realized, a light extraction effect is expected, which is preferable.
Thus, the Si concentration at which the internal quantum efficiency is not excessively lowered while expecting the light extraction effect is usually 3 × 10 ¹⁷ cm ⁻³ or more, preferably 5 × 10 ¹⁸ cm ⁻³ or more, usually 1 × 10 ²¹ cm ⁻³ or less, preferably 6 × 10 ¹⁹ or less.

  In addition, when a dopant such as Si is intentionally added, if the concentration is excessive, irregularities are extremely generated on the surface, and the internal quantum efficiency of the quantum well active layer structure portion is excessively lowered, which is not preferable. .

  On the other hand, in step B, when the internal quantum efficiency of the quantum well active layer structure in the present invention is relatively high, in an environment where silicon (Si) raw material is not intentionally supplied onto the nitride main surface of the substrate. It is preferable that the first nitride semiconductor layer is epitaxially grown.

  That is, the Si raw material is intentionally supplied onto the nitride main surface of the substrate in a state where the nitride main surface of the substrate is exposed to an atmosphere containing an inert gas and an appropriate amount of active gas as an arbitrary configuration. It is preferable to epitaxially grow the first nitride semiconductor layer.

In the case of the sequence shown in FIG. 2A, the first nitridation is performed without intentionally supplying Si raw material onto the nitride main surface of the substrate in an atmosphere including N ₂ and NH ₃ and constituting a main flow. This means that the physical semiconductor layer is epitaxially grown.

  A typical film obtained by this is ideally an i-GaN layer. However, even in the case of a GaN layer epitaxially grown in a state where a dopant such as Si is not intentionally supplied, in practice, Si or the like mixed as an impurity from the source gas or quartz present in or near the reactor Usually, Si or the like mixed from the member or the like is included.

  Furthermore, when a Si-based abrasive used in a surface polishing process of a self-supporting substrate such as a GaN substrate or an AlN substrate adheres to the substrate surface as a residue, Si is deposited on the substrate surface during epitaxial growth. May stay in the i-GaN layer as impurities.

In addition, when a substrate intentionally doped with Si, such as a GaN substrate or an AlN substrate, is used as a substrate for epitaxial growth, Si may segregate on the substrate surface at the initial stage of epitaxial growth. There is also a possibility that the i-GaN layer will take in this Si. In the case of Si, the concentration of such unintended impurities is preferably suppressed to 3 × 10 ²¹ cm ⁻³ or less.

  In addition, in order to make the internal quantum efficiency of the quantum well active layer structure in the present invention relatively high, the first nitride semiconductor layer can be epitaxially grown in an environment where no silicon (Si) source is intentionally supplied. The average Si concentration in the growth film thickness direction of the first nitride semiconductor layer is preferably lower than the average Si concentration in the growth film thickness direction of the second nitride semiconductor layer.

  This is because even if Si is mixed due to the various reasons described above, the influence is considered to be extreme when the interface is several hundred nm. Therefore, the first nitride semiconductor layer sufficiently thicker than this thickness is formed. When formed, the average Si concentration in the growth film thickness direction of the first nitride semiconductor layer epitaxially grown without intentionally supplying a dopant such as Si is defined as follows. This is because it is considered to be lower than the Si concentration of the second nitride semiconductor layer intentionally supplied and grown.

In addition, it is preferable that the reactor internal pressure of this process B is also set to about 35 kilopascals-120 kilopascals, for example. At this time, in order to relatively increase the internal quantum efficiency of the quantum well active layer structure in the present invention, the ratio of the inert gas component (Fp) in the total gas constituting the main flow (of FIG. 2A) In the sequence example, the NH ₃ gas component as the first main flow, the nitrogen gas as the second main flow, and the inert gas component ratio with respect to the sum of the TMGa components) are 0.5 to 1.0 in flow rate ratio. It is preferable to supply gas. Further, a high etching effect H ₂ gas into the inert gas constituting the main flow is preferably not included in excess.

  In step B, when epitaxial growth is performed in a reduced pressure state where the pressure in the reactor is less than 35 kilopascals, the crystallinity is lowered due to an increase in point defects, and the second nitride semiconductor layer formed thereon is further formed. In addition, the photoluminescence (PL) characteristics of the multi-quantum well active layer structure formed as an arbitrary configuration are also deteriorated. Further, at a pressure of 120 kilopascals or more, the gas phase reaction in the reaction furnace increases, and carbon is taken into the nitride semiconductor layer during the epitaxial growth and the crystallinity is lowered.

  The substrate temperature in step B is set to a predetermined temperature in the temperature range of 600 ° C. to 1350 ° C., but the lower limit is set to 600 ° C. is the thermal energy required for crystal growth of a good quality nitride semiconductor. The upper limit is set to 1350 ° C. because of limitations such as deterioration of the components of the reactor. When the film is formed at a temperature lower than 600 ° C., polycrystalline components are likely to be mixed, and as a result, the light emission characteristics are also deteriorated.

The first nitride semiconductor layer obtained under the conditions preferably has a thickness L ₁ is a relatively thin layer of 300nm or less in the range of 0.1 nm. The lower limit of the thickness of the first nitride semiconductor layer is set to 0.1 nm in order to cover the substrate surface (nitride surface) with an epitaxial layer that does not contain Si excessively. This is because it takes.

Further, according to the study by the present inventors, the excessively thick first nitride semiconductor layer has an extremely deteriorated surface morphology of the entire epitaxial layer grown thereafter, and the epitaxial layer includes an active layer structure. Is not preferable because it is assumed to cause deterioration of the optical characteristics. Specifically, L ₁ is usually 0.1 nm or more, preferably 1.0 nm or more, more preferably 5.0 nm or more, usually 300 nm or less, preferably 150 nm or less, more preferably 50 nm or less. It is preferable that

  According to studies by the present inventors, a nitride semiconductor suitable for a light-emitting element can be realized by homoepitaxially growing the second nitride semiconductor layer on such a first nitride semiconductor layer.

  A relatively thick second nitride semiconductor layer (second GaN layer) is epitaxially grown on the first nitride semiconductor layer (first GaN layer) while supplying an n-type dopant material. (Step C).

  In order to increase the light extraction efficiency from the nitride semiconductor in the present invention while keeping the internal quantum efficiency of the quantum well active layer structure in an appropriate range, as shown in FIG. In a state where the surface of the nitride semiconductor layer is exposed to an atmosphere containing an appropriate active gas, the second nitride semiconductor layer is epitaxially grown while intentionally supplying an n-type dopant material onto the first nitride semiconductor layer. It is preferable to make it. In this way, it is possible to form moderate irregularities on the surface of the nitride semiconductor layer and the like in a self-forming manner and intentionally.

  In order to make the internal quantum efficiency of the quantum well active layer structure portion relatively high in the present invention, as shown in FIG. 2A, the surface of the first nitride semiconductor layer is exposed to an atmosphere containing an inert gas. In this state, the second nitride semiconductor layer is epitaxially grown while intentionally supplying the n-type dopant material on the first nitride semiconductor layer.

  Here, examples of the n-type dopant for the nitride semiconductor layer include Si, O, C, Ge, Se, S, Te, and the like, and Si, Se, and O are particularly preferable. Is most preferably available.

  The substrate temperature in step C is also set to 600 ° C. to 1350 ° C., but the pressure in the reactor is 5 kilopascals or more and less than the pressure during epitaxial growth of the first nitride semiconductor layer. Since the generation of point defects is suppressed in the process of laminating the second nitride semiconductor layer on the first nitride semiconductor layer, the pressure in the reaction furnace can be set lower than those in step A and step B. It is. However, when the pressure is less than 5 kilopascals, nitrogen easily escapes from the surface of the second nitride semiconductor layer in the growth process, and this effect is considered to be greater in the nonpolar plane than in the normal polar plane. Kilopascal is preferred.

In the sequence example shown in FIG. 2G, NH ₃ gas that can be a nitrogen source of GaN is supplied as the first main flow that constitutes the main flow, and H ₂ is used as the second main flow that constitutes the main flow. TMGa is supplied using a part of this as a carrier gas, and further, silane (SiH ₄ ) gas is supplied as a Si source which is an n-type dopant.

In order to increase the light extraction efficiency from the nitride semiconductor in the present invention while keeping the internal quantum efficiency of the quantum well active layer structure in an appropriate range, the active gas component ratio in the total gas is set to H ₂ or the like. Is preferably supplied as the second main flow gas so that the flow rate is 0.5 to 1.0.

On the other hand, in the sequence example shown in FIG. 2A, NH ₃ gas that can be a nitrogen source of GaN is supplied as the first main flow constituting the main flow, and N ₂ is used as the second main flow constituting the main flow. In use, TMGa is supplied as a part of the carrier gas, and silane (SiH ₄ ) gas is supplied as the Si source which is an n-type dopant.

In addition, in order to make the internal quantum efficiency of the quantum well active layer structure part in this invention comparatively high, the inert gas component ratio which occupies in all the gas which comprises the main flow in this process C (like process B ( In the sequence example of FIG. 2A, the ratio of the inert gas component to the sum of the NH ₃ gas component that is the first main flow and the nitrogen gas, TMGa, and SiH ₄ gas components that are the second main flow is 0. It is preferable to supply gas so that it may be 5 or more and 1.0 or less.

Thus the second nitride semiconductor layer obtained in a relatively thick layer of that it has a thickness _{L 2} is 0.4~20μm (ie 400nm or 20μm or less), the silicon concentration is usually 1 × ^{10 17} cm ⁻³ or more, preferably 5 × 10 ¹⁷ cm ⁻³ or more, more preferably 1 × 10 ¹⁸ cm ⁻³ or more, and particularly preferably 3 × 10 ¹⁸ cm ⁻³ or more. Further, it is usually about 6 × 10 ¹⁹ cm ⁻³ or less, preferably 4 × 10 ¹⁹ cm ⁻³ or less, more preferably 1 × 10 ¹⁹ cm ⁻³ or less, and particularly preferably 7 × 10 ¹⁸ cm ⁻³ or less. Here, when the thickness L2 of the _second nitride semiconductor layer is less than 0.4 μm, it is difficult to obtain good pn characteristics when a pn junction element is manufactured. Roughness is likely to occur.

The viewpoint that the thickness L ₂ makes it easy to stabilize the electrical characteristics when the device is formed, and suppresses the slight lattice defects remaining at the epitaxial growth start interface from adversely affecting the quantum well active layer structure of the device. Therefore, the thickness is preferably 0.4 μm (400 nm) or more.

Further, the thickness of L ₂ is important for adjusting the degree of unevenness such as the surface of the nitride semiconductor layer while maintaining the internal quantum efficiency of the quantum well active layer structure in an appropriate range, and is generally appropriate. The degree of unevenness can be increased by increasing the thickness appropriately according to the epitaxial growth conditions. However, excessive formation of irregularities is not preferable, and the upper limit is 20 μm.

When the dopant concentration of the second nitride semiconductor layer is less than 1 × 10 ¹⁷ cm ⁻³ , it is difficult to obtain good pn characteristics when a pn junction element is manufactured, and the dopant concentration is 6 × 10 ^19. In a heavily doped nitride semiconductor layer exceeding cm ⁻³ , the surface tends to be excessively rough.

  In the present invention, it is preferable that the first nitride semiconductor layer and the second nitride semiconductor layer have different compositions. This is because the first nitride semiconductor layer has a function as a so-called buffer layer, while a part of the second nitride semiconductor layer has at least a function of carrier injection into the quantum well active layer structure. This is because the function can be separated. For example, when the first nitride semiconductor layer is an undoped GaN layer that is not intentionally doped with Si or the like, the second nitride semiconductor layer is an n-GaN layer that is intentionally doped with Si or the like. It is preferable. By intentionally changing the growth conditions, it is possible to separate the functions of the respective layers, and in this case, the concentration of impurities contained may be different. In this sense, the first nitride semiconductor layer and the second nitride semiconductor layer are preferably different in composition.

  Although the present invention includes these steps A, B, and C, it goes without saying that additional steps may be added. For example, a layer structure in which two or more kinds of layers having different doping concentrations are repeatedly stacked between the process B and the process C (between the first nitride semiconductor layer and the second nitride semiconductor layer) or a different material. You may make it add the process of forming the layered structure etc. which laminated | stacked two or more types of comprised nitride semiconductor layers repeatedly. Further, as described above, the most preferable temperature rise temperature range of Step A of the present invention, the film formation temperature range of the first nitride semiconductor layer, and the film formation temperature range of the second nitride semiconductor layer are all 900. It is the range of ° C to 970 ° C.

  Usually, in order to carry out thick film growth of i-GaN or n-GaN on a c-plane GaN crystal, the most preferable temperature rise temperature or growth temperature is a temperature exceeding 1000 ° C. When grown, the crystallinity decreases. However, according to the results of experiments conducted by the present inventors, when a crystal is grown on a substrate having a nonpolar main surface by the method according to the present invention, the temperature is 100 ° C. or more lower than the general conditions. It was confirmed that the internal quantum efficiency was greatly improved by epitaxial growth.

  For this reason, in order to relatively increase the internal quantum efficiency of the quantum well active layer structure in the present invention, it is preferable to form the nitride semiconductor layer at a relatively low temperature. For example, it is preferable to use TEGa rather than TMGa as a Ga raw material in order to suppress contamination of impurities in the film. This is because TEGa is decomposed at a lower temperature than TMGa, so that the incorporation of carbon into the film is suppressed. For example, in the second nitride semiconductor layer, compensation by carbon is performed when doping Si or the like. This is preferable because a layer having a small amount can be formed.

  Further, as illustrated in FIG. 2G and FIG. 2A, a process of further epitaxially growing a stacked structure including an MQW layer on the second nitride semiconductor layer (process D), p-type dopant It is also possible to provide a third step including a step of crystal-growing a nitride semiconductor layer containing silane (step E), a temperature lowering step (F), and the like.

  In the step (step D) of epitaxially growing the stacked structure including the MQW layer on the second nitride semiconductor layer, the surface of the second nitride semiconductor layer is exposed to an atmosphere containing an inert gas. It is preferable to epitaxially grow the stacked structure including the MQW layer on the second nitride semiconductor layer.

In both the case of FIG. 2 (G) and FIG. 2 (A), N ₂ is contained as an inert gas, and an appropriate amount of NH ₃ gas, which is an active gas, is optionally included as a nitrogen source. In the atmosphere constituting the main flow, a multiple quantum well active layer structure composed of a stacked structure of an InGaN layer and a GaN layer is formed on the second nitride semiconductor layer.

  The quantum well layer in the multiple quantum well active layer structure of the present invention preferably contains In, Al, etc., and most preferably contains In, in order to select the emission wavelength appropriately. The In concentration of the InGaN quantum well layer is, for example, 0.04 to 0.15 in composition ratio, and the InGaN / GaN multiple quantum well active layer structure has a center wavelength of 400 ± 30 nm (that is, 370 nm to 430 nm). It is preferable to emit light.

  More preferably, the center wavelength is 380 nm or more and 420 nm or less, and the In composition ratio of the InGaN quantum well layer corresponds to about 0.05 to 0.10. Most preferably, the center wavelength is 395 nm or more and 415 nm or less, and the In composition ratio of the InGaN quantum well layer corresponds to about 0.06 to 0.09.

In general, in an InGaN / GaN multiple quantum well active layer structure formed on a c-plane sapphire substrate, multiple dislocations due to lattice constant mismatch and the like occur from the interface between the sapphire substrate and the GaN-based epitaxial layer. It is normally propagated throughout the epitaxial layer including the active layer structure, and it is known that the dislocation density is about 1 × 10 ⁹ cm ⁻² . It is also known that even if an epitaxial layer is formed on a sapphire substrate using an uneven surface and a part of dislocations is reduced, the dislocation density is only reduced to about 3 × 10 ⁸ cm ^−2. Yes.

  Here, In localizes the spatial position of the electron-success pair in the quantum well layer, and thus localizes the light emission, so that even if there are many dislocations, it is injected or generated in the quantum well layer. It is believed that the electron-hole pairs thus trapped by these dislocations and the like are restrained from non-radiative recombination.

On the other hand, the present invention uses a nitride on a nonpolar plane substrate as a base. That is, in the present invention, the dislocation density in the epitaxial layer can be made very small. According to the experiment results of the present inventors, the dislocation density existing in the epitaxial layer in the present invention is preferably 3 × 10 ⁷ (cm ⁻² ) or less, and more preferably 5.0 × 10 ^6. It can be reduced to (cm ⁻² ) or less. Even in the case where the spatial localization effect of electron-hole pairs due to In is small, the emission characteristics of the present invention are good.

  Specifically, it has been confirmed that even if the In composition ratio of the InGaN quantum well layer is, for example, 0.04 to 0.15, very good light emission characteristics are exhibited. For this reason, the light emitted from the nitride semiconductor epitaxial layer in the present invention preferably has a center wavelength of 370 nm or more, preferably 380 nm or more, and an upper limit of 430 nm or less, preferably 420 nm or more. The lower limit is defined by the minimum In composition of the InGaN layer necessary to form a band offset with the GaN layer, while the upper limit is a relatively thick InGaN quantum layer that can be preferably used in the present invention. It is defined by the upper limit In composition that can form a well layer.

A preferable substrate temperature in forming the quantum well active layer is defined for the purpose of stably forming the InGaN layer. Since In in the InGaN layer has a high vapor pressure, the quantum well active layer is preferably formed at a lower temperature than the other layers. In particular, according to the study by the present inventors, in the growth on the nonpolar plane, unlike the growth on the polar plane, the incorporation of In is reduced even under the same conditions. As a result, when a quantum well active layer structure including an InGaN layer is formed on an epitaxial layer on a flat nonpolar surface as in the present invention, the temperature range is preferably 600 ° C. to 850 ° C. Here, when the substrate temperature is 600 ° C. or lower, the decomposition efficiency of the nitrogen raw material is lowered and the incorporation of impurities is increased.

  When the temperature is higher than 850 ° C., re-evaporation of indium atoms is activated, and it becomes difficult to control the emission wavelength depending on the indium composition and to ensure in-plane uniformity. Therefore, it is preferable that the substrate temperature when forming the quantum well active layer is set to 600 ° C. to 850 ° C. When there are fewer impurities and more wavelength reproducibility is required, it is more preferable to set the substrate temperature between 700 ° C. and 760 ° C.

  The pressure in the reaction furnace when forming the quantum well active layer is preferably not less than the pressure at the time of epitaxial growth of the first nitride semiconductor layer, and preferably not more than 120 kPa.

  This is due to the following reason. In the step of laminating the quantum well active layer on the second nitride semiconductor layer, it is necessary to suppress the generation of defects in the presence of indium which is an element having a high vapor pressure. Therefore, it is preferable to set the pressure in the reactor higher than that in step C. This makes it possible to appropriately suppress excessive reevaporation of In, which is an element having a high vapor pressure. However, at a pressure of 120 kilopascals or higher, the gas phase reaction in the reaction furnace increases, and carbon is easily taken into the nitride semiconductor layer during epitaxial growth, resulting in a decrease in crystallinity. Therefore, the pressure in the reaction furnace is preferably not less than the pressure at the time of epitaxial growth of the first nitride semiconductor layer, and preferably not more than 120 kPa.

As illustrated in FIG. 2G and FIG. 2A, NH ₃ gas that can serve as a nitrogen source for InGaN and GaN is supplied as the first main flow that constitutes the main flow, and the second main that constitutes the main flow. N ₂ is used as a flow, and TMGa and TMIn are supplied as a part of the carrier gas. It should be noted that the inert gas component ratio in the total gas constituting the main flow in the process D (in the sequence example of FIG. 2G or FIG. 2A, in the quantum well layer included in the active layer structure, The inert gas component ratio with respect to the sum of NH ₃ gas, which is one main flow, and N ₂ gas, TMIn, TMGa, which is the second main flow. On the other hand, in the barrier layer included in the active layer structure, It is preferable to supply the gas so that the NH ₃ gas that is the flow, the nitrogen gas that is the second main flow, and the inert gas component ratio with respect to the sum of TMGa are 0.5 to 1.0 in terms of flow rate ratio. In particular, it is preferable not to supply H ₂ in the quantum well layer included in the active layer structure.

This is because the introduction of an excessive active gas, particularly the introduction of H ₂ gas having a large etching effect, extremely deteriorates the flatness of a material containing InN, for example, InGaN. For this reason, even if it is a good quality nitride semiconductor layer that has undergone the first growth step and the second growth step of the present invention, the crystallinity of the active layer is likely to deteriorate, so H ₂ is supplied. This is not preferable.

  However, since the nitrogen desorption is suppressed when growing under the condition that the nitrogen raw material that is the active gas is excessively present, the ratio of the inert gas component in the total gas constituting the main flow is the present invention. According to the study by those, it can be lowered to about 0.4.

  The thickness of the active layer structure, particularly the thickness of the quantum well layer in the quantum well active layer structure, that can be produced during the epitaxial growth step of the laminated structure included in the third step is increased in the present invention. Things are preferable.

Specifically, the quantum well layer in the InGaN / GaN multiple quantum well active layer structure formed on the c-plane sapphire substrate has the highest luminous efficiency when the thickness is about 1 nm to 2 nm. Are known. This is because the injected / generated electron-hole pairs are spatially separated in the multi-quantum well active layer structure formed on the polar surface. One reason is that the well layer becomes appropriate.

  On the other hand, in the present invention, as described above, the first nitride semiconductor layer and the second nitride semiconductor layer are formed with high quality on a substrate having a nonpolar surface that does not occur in principle in QCSE. Is possible. Therefore, it is possible to produce an ideal and moderately thick quantum structure.

  Furthermore, in the present invention, appropriate irregularities, interface fluctuations, and thickness fluctuations may be added to the multiple quantum well active layer (MQW). This is a preferred embodiment for the following reasons. That is, when MQW is captured as a set of electromagnetic dipoles, the addition of irregularities to MQW changes the direction of internal radiation. Assuming a GaN substrate and an outer medium, such as air, the number of radiation elements that fall within a critical angle determined by the refractive index difference at the optical interface formed by GaN / air will increase, and the surface has been roughened. This is preferable because it is expected to improve the light extraction efficiency.

  Such naturally occurring irregularities and fluctuations are added to the MQW by forming InGaN having a certain thickness. If the film is a thin film, the degree of unevenness is small, and if it is excessively thick, the internal quantum efficiency is assumed to decrease. Furthermore, the degree of such naturally occurring irregularities and fluctuations added to the MQW can be increased by increasing the number of MQW layers and the like as will be described later in Examples and the like.

  From the above two viewpoints, for example, in the case of manufacturing a light emitting device, the quantum well layer in the multiple quantum well active layer structure can increase the efficiency and output of the light emitting device by increasing the volume. Then, according to the experiments by the present inventors, the lower limit of the thickness of at least one arbitrary quantum well layer in the multiple quantum well active layer is preferably 10 nm or more, and more preferably 15 nm or more. . Furthermore, the lower limit of the thickness of each of the quantum well layers in the multiple quantum well active layer is preferably 10 nm or more, and more preferably 15 nm or more. This means that a much thicker quantum well layer and multiple quantum well active layer structure can be formed compared to the conventional method.

  On the other hand, the quantum well layer in the quantum well active layer structure is an excessively thick quantum well from the viewpoint of realizing high-efficiency recombination due to the so-called quantum size effect, and also from the viewpoint of suppressing the reduction in crystal quality shown below. Layers are not preferred. According to the study by the present inventors, 100 nm or less is preferable. More preferably, it is 50 nm or less, More preferably, it is 30 nm or less, Most preferably, it is 20 nm or less. The upper limit of the thickness of these preferable quantum well layers is also much thicker than that of the conventional method.

  Furthermore, in the present invention, the number of quantum well layers is preferably larger than usual. In the present invention, since the layer does not have an excessively low internal quantum efficiency on the nonpolar plane, the well layer itself is expanded and the thickness fluctuation is appropriately added in the growth plane. Improvement in extraction is expected. In the present invention, as shown in Examples and the like, when multiple quantum well layers are stacked, the fluctuation can be appropriately added within an appropriate range. Therefore, the number of well layers in the multiple quantum well active layer structure is preferably 2 to 100 layers, more preferably 4 to 50 layers, more preferably 6 to 25 layers, and most preferably 8 to 15 layers. preferable.

In addition, the nitride semiconductor of the present invention has a relatively low dislocation density in the epitaxial layer, which reduces the spatial localization effect of electron-hole pairs due to In, in other words,
As described above, even when an InGaN layer having a small In composition is adopted as the quantum well layer, it is possible to realize good light emission characteristics. Here, in homoepitaxial growth on a GaN substrate, when an InGaN layer having a high In concentration is formed on the GaN layer, the InGaN layer is subjected to compressive stress in the growth plane direction.

  In particular, in an InGaN quantum well layer in an InGaN / GaN quantum well active layer structure formed on a nonpolar surface, for example, an m-plane GaN substrate, an anisotropic compressive stress is applied to the growth surface, so The critical film thickness that does not impair the crystallinity of a high InGaN layer is much thinner than the critical film thickness that does not impair the crystallinity of an InGaN layer with a low In concentration. That is, in the conventional InGaN / GaN quantum well active layer structure formed by a method that can form only a structure having many point defects, etc., which has extremely poor morphology on the nonpolar surface as in the prior art, the same as on the sapphire substrate It is necessary to increase the light emission intensity by increasing the In concentration and the resulting spatial localization effect of electron-hole pairs. As a result, the thickness of the InGaN layer cannot be increased due to a critical film thickness limitation.

  On the other hand, in the range of 0.04 to 0.15 in which the In composition of the InGaN quantum well layer, which is preferably used in the present invention, is relatively low, the thickening is particularly preferable. As a result, the preferable condition of the thickness of the quantum well layer is that the upper limit and the lower limit thereof are relatively thicker than the conventional one.

  From the above, the thickness of the active layer structure, particularly the thickness of the quantum well layer in the quantum well active layer structure, which can be produced during the epitaxial growth process of the laminated structure included in the third step is increased in the present invention. It is preferable to aim at. Further, it is preferable to increase the number of quantum wells than usual.

As shown in FIG. 2 (G) and FIG. 2 (A), as long as it is on the second nitride semiconductor layer, the layer containing a material that can be a p-type dopant is appropriately selected regardless of the position. Although possible, it is preferable to have a stacked structure including an MQW layer on the second nitride semiconductor layer, and further include a layer including a material that can be a p-type dopant thereon (step E). In this step, it is preferable to use Mg as the p-type dopant, and the concentration is preferably in the range of 1 × 10 ¹⁹ cm ^{−3 to} 8 × 10 ¹⁹ cm ⁻³ . The reason is as follows.

  Mg is difficult to be taken into the nitride crystal and its concentration is rate-limiting. However, the way it is taken in greatly depends on the flatness of the surface. Therefore, if the surface flatness of the epitaxial layer on the nonpolar surface is excessively poor as in the prior art, it is difficult to control the Mg concentration on the surface of the substrate, resulting in an unintentionally low concentration. A very high concentration layer may be accidentally formed. On the other hand, the moderately good structure of the surface state shown in the present invention and the structure with appropriate irregularities can control the Mg concentration stably and with good reproducibility. Can be selected as desired within a wide range.

That is, in order to avoid accidentally becoming a high concentration layer as in the conventional method, the target value of Mg concentration is intentionally lowered and epitaxial growth is performed, resulting in an extremely low concentration. There is nothing. For this reason, the dopant concentration in the layer containing a material that can be a p-type dopant formed on a nonpolar surface with excellent surface flatness can be set in a range that is considered to be relatively appropriate for an AlGaN-based nitride semiconductor layer. Is possible. The concentration is usually 1 × 10 ¹⁹ cm ⁻³ or more, preferably 2 × 10 ¹⁹ cm ⁻³ or more, and usually 8 × 10 ¹⁹ cm ⁻³ or less, preferably 6 × 10 ¹⁹ cm ⁻³ or less.

The layer containing a material that can be a p-type dopant is preferably a layer containing Al _x Ga _1-x N (0 ≦ x ≦ 1). In particular, when the center wavelength of light emitted from the InGaN / GaN quantum well active layer structure having a low In composition suitably used in the present invention is 370 nm to 430 nm, the light emitted from the active layer structure is p-type. In particular, Al _x Ga _1-x N (x ≠ 0) is desirable in order to suppress absorption in a layer including a material that can be a dopant.

According to the present invention, the first nitride semiconductor layer, the second nitride semiconductor layer formed thereon, and the active layer structure that can be formed thereon are epitaxial layers formed on a nonpolar plane. However, an appropriate flatness can be realized. For this reason, even in a layer containing a material that can be preferably formed as a p-type dopant, the Al composition is higher than usual, and even when the layer thickness is thicker than usual, good Al _x Ga _{1 A −xN} (x ≠ 0) layer can be formed.

  In general, an AlGaN layer on a GaN substrate is subjected to tensile stress in the layer. Furthermore, this stress increases as the Al composition increases and the film thickness increases, so that cracks and the like are generated and defects are easily introduced. However, the degree is relaxed on the epitaxial layer having moderate flatness and having few defects realized in the present invention. As a result, even if the Al composition is relatively high and the film thickness is relatively large, a high-quality AlGaN layer can be grown.

  According to the study by the present inventors, the range of Al composition that can be preferably used is usually 0.02 or more, preferably 0.03 or more. Moreover, it is 0.20 or less, Preferably it is 0.15 or less. The film thickness is usually 0.05 μm or more, preferably 0.10 μm or more, more preferably 0.12 μm or more, and usually 0.25 μm or less, preferably 0.20 μm or less, more preferably 0.18 μm or less.

From the viewpoint of forming the p-side electrode in the semiconductor light emitting device of the present invention, in order to reduce the contact resistance with the electrode, the interface layer with the electrode has a small Al composition, for example, Al _0.025 Ga _0.975 N As described above, Al _0.10 Ga _0.90 N is formed on the quantum well active layer structure side from the viewpoint of suppressing light absorption, and p-type is inserted. A layer including a material that can serve as a dopant has a two-layer structure, which is effective and preferable for achieving both optical properties and electrical properties.

The growth atmosphere in forming a layer containing a material that can become a p-type dopant, for example, an Al _x Ga _1-x N (0 ≦ x ≦ 1) layer places importance on flatness for the same reason as in Steps B and C. In some cases, an inert gas is preferred. However, even in an atmosphere mainly composed of H ₂ which is an active gas, it is possible to form a layer without any problem. In particular, when the layer containing a material that can be a p-type dopant, for example, an Al _x Ga _1-x N (0 ≦ x ≦ 1) layer is thin and the growth time is short, H ₂ that is an active gas. Even in an atmosphere mainly composed of, growth is possible. Therefore, a growth atmosphere in forming a layer containing a material that can be a p-type dopant, for example, an Al _x Ga _1-x N (0 ≦ x ≦ 1) layer can be appropriately selected. For example, from the viewpoint of suppressing the contamination of impurities such as carbon, it is preferable to grow in an atmosphere mainly composed of H ₂ which is an active gas. On the other hand, when importance is attached to surface flatness, N ₂ is mainly used. It is preferable to grow in an atmosphere.

The substrate temperature during growth when forming a layer containing a material that can be a p-type dopant, for example, an Al _x Ga _1-x N (0 ≦ x ≦ 1) layer, is 600 ° C. to 1350 similarly to the other layers. It is preferably in the range of ° C, more preferably in the range of 650 ° C to 1200 ° C, still more preferably in the range of 800 ° C to 1100 ° C, and most preferably in the range of 900 ° C to 970 ° C.

  Furthermore, a preferable pressure when forming an epitaxial layer containing a material that can be a p-type dopant is preferably 30 kPa or more from the viewpoint of suppressing the generation of defects due to the introduction of Mg, and 120 kPa or less from the suppression of the gas phase reaction described above. Is preferred.

  The temperature lowering step after the growth relating to the semiconductor nitride semiconductor layer of the present invention can be carried out by an arbitrary procedure, but the temperature lowering condition is preferably as follows. That is, this is a case where an activation process for making an epitaxial layer portion containing a material that can be a p-type dopant into a p-type layer is performed at the time of cooling (activation process during the cooling process).

  In this case, according to the study by the present inventors, when a flat epitaxial layer containing a material that can be a p-type dopant is formed on a nonpolar plane substrate, the pattern shown in FIG. The material that can be a p-type dopant can be activated by the following temperature-lowering process.

Specifically, immediately after the supply of the group III material and the dopant material is stopped, the substrate temperature is allowed to cool naturally, or the substrate temperature is lowered by cooling with temperature control, cooling with a supply gas, or the like. . In the temperature lowering step, N ₂ is continuously supplied, an inert gas is supplied, or another inert gas is supplied in addition to the continuous supply of N ₂ . When H ₂ is supplied in the growth process, this is sufficiently reduced or cut off, the NH ₃ flow rate is reduced from that during the growth, and then NH ₃ is supplied to an appropriate temperature. Thereafter, the substrate temperature is further lowered by using only N ₂ gas, only inert gas, or only mixed gas of N ₂ gas and inert gas. The present inventors have found that by such a procedure, a layer containing a material that can be a p-type dopant formed on a nonpolar surface having a surface having appropriate flatness can be made p-type. It was.

In the temperature lowering process, reducing the NH ₃ flow rate from the time of growth, and then stopping the supply after supplying NH ₃ to an appropriate temperature suppresses excessive desorption of nitrogen, which is a constituent element of the epitaxial layer, from the surface. Because. This is similar to the technical idea during the temperature raising step of the present invention.

  Furthermore, according to the study by the present inventors, the optimum p-type dopant activation sequence depends on the material constituting the outermost surface.

For example, when the outermost surface is a GaN layer, after the crystal growth sequence is completed, that is, in this case, the supply of Ga raw materials such as TMGa and TEGa and Mg raw materials such as Cp ₂ Mg is stopped, and the temperature lowering process is started. Thereafter, the flow rate of NH ₃ is preferably in the range of 100 cc / min (sccm) to 1 L / min (slm). On the other hand, when the outermost surface is an AlGaN layer, it is difficult for N to escape from the surface, so a group III raw material such as TMGa, TEGa, TMAl, etc. and Cp ₂
After the supply of Mg raw material such as Mg is stopped and the temperature lowering process is started, the flow rate of NH ₃ is preferably 30 (sccm) or more and 100 (sccm) or less.

Furthermore, in any case, the temperature at which NH ₃ is continuously introduced in the temperature lowering step is preferably continued at least up to 965 ° C., and is preferably cut off up to 450 ° C. at the longest. In the temperature lowering process, if NH ₃ supply is stopped at an excessively high temperature, excessive surface roughness may be caused unintentionally. On the other hand, if NH ₃ supply is continued to an excessively low temperature, H atoms from NH ₃ are crystallized. It will be fixed inside, and the activation rate of Mg will fall. Therefore, it is most preferable to stop the NH ₃ supply between 950 ° C. and 750 ° C.

In addition, the pressure range during the temperature lowering process can be arbitrarily set, but in the study by the present inventors, it can be performed under reduced pressure, under normal pressure, or further under pressure. It is also possible, and the preferable range of the pressure is preferably in the range of about 13 kPa to 203 kPa. In particular, when carried out under reduced pressure, desorption of H atoms is promoted, and on the other hand, when carried out under increased pressure, a layer containing a material that can be a p-type dopant on a nonpolar surface can be easily made p-type. And suitable flatness of the surface is ensured. In addition, when it is similar to the pressure at the time of epitaxial growth of a laminated structure, it is preferable when productivity etc. are considered.

  On the other hand, after the temperature lowering process, the temperature of the crystal growth apparatus is increased again, the annealing after the thermal temperature lowering process is performed again, or the electron beam irradiation is performed after the temperature lowering process, and the p-type It is also possible to separately perform a step of converting the epitaxial layer portion containing a material that can become a dopant into a p-type (an activation step after the temperature lowering step). In addition, it is arbitrary to perform the activation process after a temperature-falling process further to the epitaxial layer formed on the nonpolar surface which passed the activation process during the temperature-falling process by this invention.

  According to the study by the present inventors, the first nitride semiconductor layer, the second nitride semiconductor layer formed thereon, and the active layer structure that can be formed thereon are formed on the nonpolar surface. A similar layer formed on a c-plane sapphire substrate, even when an epitaxial layer further comprising an epitaxial layer containing a material that can be a p-type dopant is formed with appropriate flatness. It has been found that the epitaxial layer formed on the non-polar surface is relatively damaged by the activation process (for example, thermal annealing) after the temperature lowering process. In some cases, if the activation process is performed after the temperature lowering process to sufficiently realize the activation of Mg, the optical characteristics may be deteriorated.

  However, according to the present inventors, in the case of having the first and second nitride semiconductor layers grown at a relatively low temperature and having a quantum well active layer structure grown at a relatively low temperature, It was confirmed that this problem could be overcome. The difference in the degree of deterioration in the activation process after the temperature lowering process due to the difference in the growth temperature of the epitaxial layer formed on the nonpolar plane is characteristic of the epitaxial layer formed on the nonpolar plane. It is thought that.

  Here, when the activation process after the temperature lowering process is performed by so-called thermal annealing on the epitaxial layer formed on the nonpolar surface realized by the present invention, the process may be performed between 650 ° C. and 750 ° C. Preferably it is carried out between 680 ° C and 720 ° C. The time is preferably about 1 to 30 minutes, and most preferably 3 to 10 minutes. The atmosphere is preferably an oxygen atmosphere, a nitrogen atmosphere, or a mixed atmosphere thereof. Further, the activation process after the temperature lowering process can be performed as an electron beam irradiation process.

  In the present invention, the first nitride semiconductor layer, the second nitride semiconductor layer formed thereon, and the active layer structure that can be formed thereon are epitaxial layers formed on a nonpolar plane. In addition, in the case where a layer including a material that can be a p-type dopant is preferably formed thereon, the step of converting the epitaxial layer including the material that can be a p-type dopant into a p-type, Since the activation process after the temperature lowering process is simpler and damage is less likely to be introduced, it is more preferable to perform the activation process during the temperature lowering process.

As described above, the nitride semiconductor of the present invention is based on the nitride on the nonpolar plane substrate, but the dislocation density in the epitaxial layer can be reduced. That is, in the nitride semiconductor of the present invention, the dislocation density existing in the epitaxial layer is preferably 3 × 10 ⁷ (cm ⁻² ) or less, more preferably 5.0 × 10 ⁶ (cm ^−2). )

FIG. 3A is a schematic cross-sectional view for explaining an example of the nitride semiconductor of the present invention thus obtained, and is a free-standing substrate having a (1-100) plane (m plane) as a main surface. A GaN layer 11 that is not intentionally doped with Si and an n-type GaN layer 12 that is doped with Si are stacked on the main surface of the GaN substrate 10, and an InGaN layer is formed on the n-type GaN layer 12. An InGaN / GaN multiple quantum well active layer structure 13 in which quantum well layers and GaN barrier layers are alternately stacked is provided, and an Mg-doped AlGaN layer 14 and a GaN layer are provided on the multiple quantum well active layer structure 13. 15 is formed.

  The In concentration of the InGaN quantum well layer illustrated in FIG. 3A is, for example, 0.04 to 0.15 in terms of composition ratio, and the InGaN / GaN multiple quantum well active layer structure 13 has a wavelength of 400 ± 30 nm. Can emit light.

  FIG. 4 shows the PL emission characteristics of the sample (A) formed as an LED having the structure shown in FIG. 3A using an m-plane nitride semiconductor grown by the crystal growth method of the present invention. In the figure, for comparison with the PL characteristics of a c-plane nitride semiconductor LED, a c-plane nitride semiconductor LED sample grown on a c-plane GaN free-standing substrate, which is a comparative c-plane nitride semiconductor LED. The results for (B) and the c-plane nitride semiconductor LED sample (C) grown on the sapphire substrate are also shown. Each of the LED samples is a light-emitting element having a stacked structure shown in FIG. 3A, and the crystal growth sequence of each sample was performed by an optimum method in accordance with the characteristics of each substrate. The surface morphology of all the samples was extremely good.

  From the results shown in this figure, PL light emission (A) from the light emitting layer (MQW layer) of the LED sample of the present invention is PL light emission from the light emitting layer (MQW layer) of the c-plane nitride semiconductor LED sample (B, Compared to C), the strength is significantly higher. The difference in PL intensity is due to the difference in crystallinity of the light emitting layer of each LED sample. The main reason for this is that the main factor is that the nonpolar plane (especially the m plane is the outermost surface) in which the QCSE effect is suppressed. A first nitride semiconductor layer and a second nitride semiconductor layer appropriately formed on a GaN free-standing substrate having a thickness of 400 nm or more and 20 μm or less. Therefore, it is considered that the luminous efficiency of the light emitting layer (MQW layer) formed on the nitride semiconductor crystal layer is remarkably improved.

  The present inventors believe that such a result is obtained for the first time by the crystal growth method of the present invention.

That is, in order to make the internal quantum efficiency of the quantum well active layer structure relatively high in the present invention, the gas species constituting the main flow in the epitaxial growth process of the first and second nitride semiconductor layers from the time of temperature rise In addition, as an optional configuration, an epitaxial growth surface having a nonpolar surface as a main surface, or a surface exposed during epitaxial growth is not excessively exposed to H ₂ gas which is an active gas. As an arbitrary configuration, even if NH ₃ which is an active gas that can be a nitrogen raw material or H ₂ of the active gas is supplied, the main flow in the epitaxial growth process of the first and second nitride semiconductor layers from the time of temperature rise is performed. By setting the ratio of the inert gas to 0.5 to 1.0 in the flow rate ratio among all the gas species to be configured, m having high crystallinity Nitride semiconductor layer is obtained, the higher the crystallinity of the upper epitaxially grown quantum well active layer structure (MQW layer).

  In addition, since the QCSE effect is suppressed on the surface of the non-polar m-plane nitride semiconductor, the light emission efficiency from the light emitting layer (MQW layer) containing an appropriate amount of In as a composition increases.

FIG. 5A to FIG. 5C are the results of examining the constituent gas dependency (see Table 1) in the main flow during the epitaxial growth of the first and second nitride layers (step B and step C). 1 is an m-plane nitride semiconductor crystal in which any sample is grown on an m-plane GaN free-standing substrate and its surface is substantially m-plane. A nitride semiconductor formed by the above method is also preferable.

In all the samples, the first temperature raising step is performed with N ₂ gas as the main flow, and the second temperature raising step is performed with a mixed gas of N ₂ and NH ₃ (inactive with respect to all the gases constituting the main flow). Epitaxial growth was carried out with the flow ratio of gas components: Fp = 0.75) as the main flow.

Sample A (Fig. 5 (A)), in the step _B, the NH _3, H 2, TMGa, also in step C _is the result of supplying NH _3, H 2, TMGa, and SiH ₄ as a main flow .

Sample B (FIG. 5 (B)), in step _B, the _{NH 3, N 2, H 2} , TMGa, also in the step _{C, NH 3, N 2,} H 2, TMGa, and SiH ₄ as a main flow This is the result of the supply.

Furthermore, Sample C (Fig. 5 (C)), in step _B, NH _3, and N 2, TMGa, also in step _C, the supply _{NH 3, N 2, H 2} , TMGa, and SiH ₄ as a main flow It is the result.

  The flow rate ratio Fp of the inert gas in the total gas constituting the main flow in each process is as shown in Table 1.

From the results shown in FIGS. 5A to 5C, in the crystal growth process, the ratio of the inert gas in the gas constituting the main flow is small, and particularly, H ₂ gas is appropriately contained therein. In this case, it is possible to form appropriate irregularities on the surface of the nitride semiconductor layer and the like in a self-forming manner and intentionally. (FIGS. 5A and 5B). On the other hand, when the ratio of the inert gas in the gas constituting the main flow is large and N ₂ is the main constituent gas of the main flow, the surface is flattened (FIG. 5C).

According to the study by the present inventors, when the Si source is intentionally supplied in the form of, for example, SiH ₄ or Si ₂ H ₆ during the epitaxial growth of the first nitride semiconductor layer (step B), unevenness is caused. Can be formed in a self-forming manner and intentionally. In addition, when the pressure in the reactor in step A and step B is excessively reduced (less than 35 kilopascals), the crystallinity in the micro sense of the first GaN layer is reduced. The characteristics of the light emitting layer are lowered.

In step A (period t _A ), step B, and step C, in order to relatively increase the internal quantum efficiency of the quantum well active layer structure, the ratio of the inert gas in the gas constituting the main flow Required 0.5 or more.

Furthermore, the thickness of the first GaN layer has an appropriate range, and in order to obtain a nitride semiconductor film having an excellent surface state and light emission characteristics, Si is formed in a form such as SiH ₄ or Si ₂ H _6. The thickness of the first nitride semiconductor layer to be epitaxially grown without intentional excessive supply is preferably a relatively thin layer in the range of 0.1 nm to 300 nm, more preferably 1.0 nm to 150 nm. And most preferably, it is the range of 5.0 nm-50 nm.

The present inventors interpret these results as follows. First, when the first nitride semiconductor layer is an intentionally excessive Si-doped film, the product produced by the gas phase reaction of SiH ₄ or Si ₂ H ₆ as the Si supply source is the substrate. It adheres to the surface and the surface of the nitride semiconductor layer immediately after the start of growth, and excessively prevents uniform growth in the plane. That is, microscopically, the first nitride semiconductor layer in the initial stage of growth starts local crystal growth in the plane. Once such excessive in-plane crystal growth begins, a film having a relatively good surface morphology cannot be obtained.

  Therefore, in the initial stage of epitaxial growth of the nitride semiconductor layer, it is important to eliminate an appropriate amount of products that cause inhomogeneous crystal growth in the plane, and once uniform crystal growth starts in the plane. For example, even if a Si-doped nitride semiconductor is crystal-grown, its surface morphology is not significantly reduced. This is the reason why the first nitride semiconductor layer is grown without intentionally excessively doping Si in the present invention.

Of course, since the gas phase reaction of SiH ₄ can be easily suppressed by lowering the pressure in the reaction furnace, the surface condition of the sample using the first nitride semiconductor layer as the Si-doped film will be better. obtain. However, if the Si doping amount is excessive and the pressure in the reactor is excessively reduced, nitrogen desorption from the surface is induced, and defects are introduced as a result. It will drop to. Therefore, in order to moderately reduce both the introduction of defects into the first nitride semiconductor layer and the growth inhibition due to the gas phase reaction of the Si supply source, the reactor pressure is not excessively reduced, that is, It is desirable that the first nitride semiconductor layer be epitaxially grown without doping excessive Si, and at least 35 kilopascals.

  When a new nitride semiconductor layer is grown on such a first nitride semiconductor layer, it is assumed that the reactor is in a reduced pressure state or Si is doped for the purpose of increasing the growth rate. In addition, it was confirmed that the surface morphology and optical characteristics of the nitride semiconductor layer hardly deteriorated.

Therefore, it is considered that the first nitride semiconductor layer in the present invention should not contain an element that becomes a growth inhibiting factor excessively. Accordingly, there can be various modes in which GaN is intentionally not excessively doped with Si. If Si is not intentionally excessively doped, the first nitride semiconductor layer is formed of InN, Group III-V nitride semiconductors such as AlN, BN, GaInN, GaAlN, GaBN, InAlN, InBN, AlBN, GaInAlN, GaInBN, InAlBN, and GaInAlBN (hereinafter sometimes collectively referred to as GaN-based semiconductors). It is also possible. Furthermore, an aspect in which the first nitride semiconductor layer is epitaxially grown by supplying an element that can be a dopant other than Si, such as O, Mg, or Zn is also possible.

  As shown in the examples described later, even when Mg or the like is intentionally doped, when a new nitride semiconductor layer is grown on the first nitride semiconductor layer, the entire epitaxial layer to be formed thereafter is It has been confirmed that surface morphology and optical properties are hardly degraded. In particular, in GaN-based materials, the present inventors have confirmed that the Mg-doped layer is superior to the Si-doped layer and the undoped layer in terms of heat resistance related to adhesion to the base. .

  Therefore, even if Si is co-doped, the Mg-doped layer can be preferably used as the first nitride semiconductor layer when heat resistance regarding adhesion to the substrate surface is required. For example, it is preferable to use for template preparation, light emitting element production, and the like.

  On the other hand, the first nitride semiconductor layer not intentionally excessively doped with Si can be suitably used for applications such as light-emitting elements and electronic devices. When the first nitride semiconductor layer is an appropriate Si-doped layer, appropriate irregularities are formed on the surface, and as a result, when the entire epitaxial layer is made into a light emitting device, the light emission efficiency is increased. In particular, in the present invention, if the internal quantum efficiency is controlled to such an extent that it does not become excessively low, Ra is approximately 200 nm, and the same degree of roughness as the wavelength is naturally formed, which effectively acts on light extraction. preferable. Therefore, in such an aspect, it can be preferably used for a light emitting element or the like.

  As a result of these, the outermost surface of the epitaxial layer formed by stacking the first nitride semiconductor layer and the second nitride semiconductor layer formed by the method of the present invention, and further, the first nitride semiconductor layer and the second nitride semiconductor layer are formed. The outermost surface of the epitaxial layer in which a further stacked structure is formed on the stacked body with the nitride semiconductor layer can be realized by forming a moderately uneven morphology.

  The morphology of the surface of the nitride semiconductor of the present invention can be measured by determining the average roughness or the center line average roughness (Ra) as an index of the degree of unevenness using a contact step meter. it can. When the nitride semiconductor of the present invention is required to have a flat surface morphology without unevenness, the outermost surface thereof preferably has Ra of 20.0 nm or less, more preferably Ra of 10.0 nm or less, still more preferably Ra is 8.0 nm or less, and most preferably Ra is 6.0 nm or less.

  On the other hand, as described above, when appropriate unevenness is formed and light extraction is made favorable to improve the light emission efficiency as a light emitting element, Ra on the outermost surface of the nitride semiconductor of the present invention is usually It is 100 nm or more, preferably 150 nm or more, and is usually 300 nm or less, preferably 250 nm or less. This is because the scattering function works efficiently at a wavelength suitably used in the present invention, and is in a preferable range.

  The Ra described here refers to a roughness curve obtained by scanning the sample surface linearly with a needle from the center line, and the area obtained by the folded roughness curve and the center line is the length of the scan. It is the value divided by.

  In addition, when the nitride semiconductor of the present invention requires a flat morphology, when the active layer has a quantum well structure, the standard deviation in the plane of the thickness of the quantum well layer is usually 0.45 nm or less. Preferably, it is 0.4 nm or less, and more preferably 0.35 nm or less.

  When the internal quantum efficiency of the quantum well active layer structure in the present invention is relatively high, the variation coefficient of thickness variation in the plane of the quantum well layer is usually 0.10 or less, Preferably, it is 0.09 nm or less. The basis of the standard deviation in the plane of the thickness of the quantum well layer and the numerical value of the coefficient of variation of the in-plane thickness variation is empirically supported by the present inventors' numerous experimental data. This will be described in detail in Examples described later.

  Further, the nitride semiconductor of the present invention is measured under specific conditions (i) The internal quantum efficiency is extremely higher than a conventionally known value, and (ii) photoluminescence lifetime (τ (PL)) However, it is also a significant feature that (iii) the luminescence recombination lifetime (τ (R)) is extremely short compared to a conventionally known value. Furthermore, when the internal quantum efficiency is relatively high, (iv) the luminescence recombination lifetime (τ (R)) can be preferably made shorter than the non-luminescence recombination lifetime (τ (NR)). is there. Moreover, these extremely excellent values were confirmed for the first time by measuring under the low excitation density conditions defined in the present invention.

  That is, the nitride semiconductor of the present invention (the first nitride semiconductor layer, the second nitride semiconductor layer, the active layer on the nitride main surface of the base body in which at least one main surface is a nonpolar nitride) In which the internal quantum efficiency is 20% or more, preferably 30% or more, and more preferably 35% or more, which is obtained from CW-PL measurement under the low excitation density condition of the active layer.

  In addition, the nitride semiconductor of the present invention has an internal quantum efficiency of 20% or more, preferably 25% or more, which is obtained from pulsed light PL measurement of the active layer under a low excitation density condition.

  In addition, the nitride semiconductor of the present invention has a photoluminescence lifetime (τ (PL)) of 1 ns or more, preferably 1.5 ns, which is obtained from time-resolved PL measurement of the active layer at room temperature and under low excitation density conditions. That's it.

  The grounds for numerical values relating to these extremely excellent features are empirically supported by the present inventors' numerous experimental data, and will be described in detail in Examples described later.

  Hereinafter, the present invention will be described in more detail with reference to Examples and Comparative Examples.

  This embodiment is an example in which a gallium nitride based semiconductor thin film is laminated and grown by MOCVD to produce a near-ultraviolet light emitting LED. The outline of a series of crystal growth processes has already been described with reference to FIG. . The grown layer structure is schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.1 mm in the c-axis direction and 15 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 6.6 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the half-width of the rocking curve in (10-12) reflection is 34.2 arcsec, the OFF angle in the c (+) direction is 0.25 °, and the OFF angle in the a direction is 0.03 °. there were. Further, the dislocation density was 5.0 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

First, as the first temperature raising step t _B , the temperature is raised while supplying 10 L / min of N ₂ as a main flow into the furnace, and when the temperature of the substrate reaches 400 ° C., the second temperature raising step t _A is performed. Started. There, NH ₃ was supplied as a gas constituting the first main flow at 7.5 L / min, and N ₂ was supplied as a gas constituting the second main flow at 12.5 L / min.

Thereafter, the substrate temperature was further raised to 1000 ° C. while increasing NH ₃ and N ₂ to 10 L / min and 30 L / min, respectively. At this time, the sub-flow was N ₂ gas 20 L / min, and the growth non-growth gas such as for purging was N ₂ for a total of 19 L / min. In the second temperature raising step, the flow ratio Fp of the inert gas component to the total gas constituting the main flow was 0.625 at the start of the second temperature raising step and 0.75 when the growth temperature was reached. .

In the next step B, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 29.5 L / min, H ₂ is 0.5 L / min, and H ₂ (0.5 L / min) which is a part of the main flow is a carrier. T as gas
MGa (0.0018 L / min with a concentration of 100%) was supplied into the furnace. With this main flow gas supply, an undoped GaN layer (first nitride semiconductor layer) was grown to a thickness of 40 nm.

At this time, the subflow was a mixed gas (20.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (20 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . During the growth of the first nitride semiconductor layer, the flow ratio Fp of the inert gas component to the total gas constituting the main flow was 0.73747.

In the next step C, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 29.24 L / min, H ₂
The 0.76L / min, TMGa and _H 2 in the main flow (0.5 L / min) as the carrier gas (0.0055L / min. At a concentration of 100%), constituting a part of the main flow _{H 2} SiH ₄ (6 × 10 ⁻⁷ L / min with a concentration of 100%) having (0.2 L / min) as a carrier gas and H ₂ (0.06 L / min) as a diluent gas was supplied into the furnace. .

With such main flow gas supply, a Si-doped GaN layer (second nitride semiconductor layer) was grown to a thickness of 7 μm. At this time, the subflow was a mixed gas (25.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (25 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . The flow rate ratio Fp of the inert gas component to the total gas constituting the main flow at the time of growing the second nitride semiconductor layer was 0.73090.

Next, the substrate temperature (substrate temperature) was set to 740 ° C., and after the substrate temperature was sufficiently stabilized, a quantum well layer of In _0.07 Ga _0.93 N (target thickness 1.5 nm) and GaN (target thickness 13 nm) were formed. A multi-quantum well active layer structure in which barrier layers were alternately stacked for 8 periods was formed (step D). Here, in the growth of the quantum well layer of In _0.07 Ga _0.93 N, NH ₃ (10 L / min) was used as a gas constituting the first main flow.

In addition, as a gas constituting the second main flow, TM 2 (0.000015 L / min) using N ₂ (20 L / min) and N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas. Then, trimethylindium (TMIn) (0.00023 L / min) using N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas was supplied.

In the growth of the GaN barrier layer, NH ₃ (10 L / min) was used as a gas constituting the first main flow. Further, as the gas constituting the second main flow, N ₂ (18.5 L / min), H ₂ (1.5 L / min), and H ₂ (0.5 L / min) which is a part of the main flow. ) Was used as a carrier gas, and TMGa (0.000017 L / min) was supplied.

In Step D, the subflow is NH ₃ (0.5 L / min) and N ₂ (25 L / min) mixed gas at 25.5 L / min, and the growth outgassing gas is N ₂ at 19 L / min. there were. The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the multiple quantum well active layer structure was 0.66666 for the InGaN well layer and 0.61667 for the GaN barrier layer.

Subsequently, the substrate temperature was set to 1000 ° C. to form an Mg-doped Al _0.1 Ga _0.9 N layer having a thickness of 50 nm (step E). The gas constituting the first main flow at this time is NH ₃ (10 L / min).

The gas constituting the second main _flow, H 2 (80L / min), trimethylaluminum that of _H 2 in the main flow (0.5 L / min) and carrier gas (TMAl) (0. 0001 L / min), TMGa (0.0018 L / min) using H ₂ (0.5 L / min), which is also a part of the main flow, as a carrier gas, and H ₂ (0 .5 L / min) as a carrier gas, cyclopentadienyl magnesium (Cp ₂ Mg) (4 × 10 ⁻⁶ L / min).

On this Mg-doped Al _0.1 Ga _0.9 N layer, an Mg-doped GaN layer was further epitaxially grown to a thickness of 70 nm (step E). The growth of the GaN layer was performed with the supply of TMAl and H ₂ (50 L / min) out of the gas in the main flow described above.

The sub flow during the growth of the Al _0.1 Ga _0.9 N layer in the process E is 50.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (50 L / min), for purging. The growth outside gas was N ₂ (19 L / min). The flow rate ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped Al _0.1 Ga _0.9 N layer was 0.

Further, the sub-flow during the growth of the Mg-doped GaN layer in the process E is 20.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (20 L / min). Was N ₂ (19 L / min). The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped GaN layer was 0.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  The surface of the substrate thus produced was extremely flat. This surface was measured with a contact-type step meter, and the average roughness or centerline average roughness (Ra) serving as an index of the degree of unevenness was determined.

  As a result, Ra according to the present example was 4.9 nm. In addition, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength is 391 nm, the integrated intensity is 96, which is a relative value, and the standard deviation of the in-plane wavelength distribution is 0.8. % Was small.

  This embodiment is an example in which a gallium nitride based semiconductor thin film is laminated and grown by MOCVD to produce a near-ultraviolet light emitting LED. The outline of a series of crystal growth processes has already been described with reference to FIG. . The grown layer structure is schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.2 mm in the c-axis direction and 15 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 6.7 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 32.8 arcsec, the OFF angle in the c (+) direction is 0.29 °, and the OFF angle in the a direction is 0.05 °. there were. The dislocation density was 5.4 × 10 ⁶ cm ⁻² . This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

First, as the first temperature raising step t _B , the temperature is raised while supplying 10 L / min of N ₂ as a main flow into the furnace, and when the temperature of the substrate reaches 400 ° C., the second temperature raising step t _A is performed. Started. There, NH ₃ was supplied as a gas constituting the first main flow at 7.5 L / min, and N ₂ was supplied as a gas constituting the second main flow at 12.5 L / min.

Thereafter, the substrate temperature was further raised to 1000 ° C. while increasing NH ₃ and N ₂ to 10 L / min and 30 L / min, respectively. At this time, the sub-flow was N ₂ gas 20 L / min, and the growth non-growth gas such as for purging was N ₂ for a total of 19 L / min. In the second temperature raising step, the flow ratio Fp of the inert gas component to the total gas constituting the main flow was 0.625 at the start of the second temperature raising step and 0.75 when the growth temperature was reached. .

In the next step B, NH ₃ was supplied at 20 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 29.5 L / min, H ₂ is 0.5 L / min, and H ₂ (0.5 L / min) which is a part of the main flow is a carrier. T as gas
MGa (0.0018 L / min with a concentration of 100%) was supplied into the furnace. With this main flow gas supply, an undoped GaN layer (first nitride semiconductor layer) was grown to a thickness of 40 nm.

At this time, the subflow was a mixed gas (20.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (20 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . During the growth of the first nitride semiconductor layer, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0.58998.

  Subsequent steps C, D, and E were performed under the same conditions as in Example 1.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  The surface of the substrate thus fabricated had good flatness although there were slight irregularities.

  This surface was measured with a contact-type step meter, and the centerline average roughness (Ra) was determined. As a result, Ra according to the present example was 17.4 nm. In addition, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength is 396 nm, the integrated intensity is 68 as a relative value, and a standard deviation of the in-plane wavelength distribution is 0.9. % Was small.

  This embodiment is an example in which a gallium nitride based semiconductor thin film is laminated and grown by MOCVD to produce a near-ultraviolet light emitting LED. The outline of a series of crystal growth processes has already been described with reference to FIG. . The grown layer structure is schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.0 mm in the c-axis direction and 15 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 6.4 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 36.7 arcsec, the OFF angle in the c (+) direction is 3.4 °, and the OFF angle in the a direction is 0.3 °. there were. The dislocation density was 5.1 × 10 ⁶ cm ⁻² . This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

First, in the first heating step t _A, the N ₂ was heated while 10L / min supplied as a main flow into the reactor, a second heating step t _A when the temperature of the substrate became 400 ° C. Started. There, NH ₃ was supplied at 7.5 L / min as a gas constituting the first main flow, and N ₂ was supplied at 10 L / min as a gas constituting the second main flow.

Thereafter, the substrate temperature was further raised to 1000 ° C. while increasing NH ₃ and N ₂ to 15 L / min and 15 L / min, respectively. At this time, the sub-flow was N ₂ gas 20 L / min, and the growth non-growth gas such as purge was N ₂ and the total was 19 L / min. In the second temperature raising step, the flow ratio Fp of the inert gas component to the total gas constituting the main flow was 0.57143 at the start of the second temperature raising step and 0.50 when the growth temperature was reached. .

In the next step B, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 29.5 L / min, H ₂ is 0.5 L / min, and H ₂ (0.5 L / min) which is a part of the main flow is a carrier. T as gas
MGa (0.0018 L / min with a concentration of 100%) was supplied into the furnace. With this main flow gas supply, an undoped GaN layer (first nitride semiconductor layer) was grown to a thickness of 120 nm.

At this time, the subflow was a mixed gas (20.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (20 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . During the growth of the first nitride semiconductor layer, the flow ratio Fp of the inert gas component to the total gas constituting the main flow was 0.73747.

In the next step C, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 29.294 L / min, H 2
₂ is 0.706 L / min, H ₂ (0.5 L / min) which is a part of the main flow is used as a carrier gas, TMGa (0.0055 L / min when the concentration is 100%), and H which constitutes a part of the main flow ₂ and (0.2 L / min) carrier gas _and supplying H 2 and SiH ₄ to (0.006L / min) diluent gas (6 × ^{10 -8} L / min. at a concentration of 100%) into the furnace did. With such main flow gas supply, a Si-doped GaN layer (second nitride semiconductor layer) was grown to a thickness of 1 μm.

At this time, the subflow was a mixed gas (25.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (25 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . During the growth of the second nitride semiconductor layer, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0.78579.

Next, the substrate temperature (substrate temperature) was set to 740 ° C., and after the substrate temperature was sufficiently stabilized, a quantum well layer of In _0.07 Ga _0.93 N (target thickness 1.5 nm) and GaN (target thickness 13 nm) were formed. A multi-quantum well active layer structure in which barrier layers were alternately stacked for 8 periods was formed (step D). Here, in the growth of the quantum well layer of In _0.07 Ga _0.93 N, NH ₃ (10 L / min) was used as a gas constituting the first main flow.

In addition, as a gas constituting the second main flow, TM 2 (0.000015 L / min) using N ₂ (20 L / min) and N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas. TMIn (0.00023 L / min) using N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas was supplied.

In the growth of the GaN barrier layer, NH ₃ (10 L / min) was used as a gas constituting the first main flow. Further, as the gas constituting the second main flow, N ₂ (18.5 L / min), H ₂ (1.5 L / min), and H ₂ (0.5 L / min) which is a part of the main flow. ) Was used as a carrier gas, and TMGa (0.000017 L / min) was supplied.

In Step D, the subflow is NH ₃ (0.5 L / min) and N ₂ (25 L / min) mixed gas at 25.5 L / min, and the growth outgassing gas is N ₂ at 19 L / min. there were. The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the multiple quantum well active layer structure was 0.66666 for the InGaN well layer and 0.61667 for the GaN barrier layer.

Subsequently, the substrate temperature was set to 1000 ° C. to form an Mg-doped Al _0.1 Ga _0.9 N layer having a thickness of 50 nm (step E). The gas constituting the first main flow at this time is NH ₃ (10 L / min). The gas constituting the second main flow is TM2 (0.0001 L / min) using H ₂ (80 L / min) and H ₂ (0.5 L / min) which is a part of the main flow as a carrier gas. likewise TMGa (0.0018 L / min) which is part _{of H} 2 and (0.5 L / min) and carrier gas in the main flow, and, likewise _H 2 (0.5 L / min in the main flow ) As a carrier gas. Cp ₂ Mg (4 × 10 ⁻⁶ L / min).

On this Mg-doped Al _0.1 Ga _0.9 N layer, an Mg-doped GaN layer was further epitaxially grown to a thickness of 70 nm (step E). The growth of the GaN layer was performed with the supply of TMAl and H ₂ (50 L / min) out of the gas in the main flow described above.

The sub flow during the growth of the Al _0.1 Ga _0.9 N layer in the process E is 50.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (50 L / min), for purging. The growth outside gas was N ₂ (19 L / min). The flow rate ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped Al _0.1 Ga _0.9 N layer was 0.

Further, the sub-flow during the growth of the Mg-doped GaN layer in the process E is 20.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (20 L / min). Was N ₂ (19 L / min). The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped GaN layer was 0. After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  The surface of the substrate thus produced had good flatness although there were very slight irregularities. This surface was measured with a contact-type step meter, and the centerline average roughness (Ra) was determined. As a result, Ra according to the present example was 8.2 nm. In addition, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength is 390 nm, the integrated intensity is 45, which is a relative value, and a standard deviation of the in-plane wavelength distribution is 0.9. % Was small.

  This example is an example in which a near-ultraviolet LED is manufactured by stacking and growing a gallium nitride-based semiconductor thin film by MOCVD, and an outline of a series of crystal growth processes will be described with reference to FIG. The grown layer structure is schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 3.8 mm in the c-axis direction and 15 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 6.9 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 30.4 arcsec, the OFF angle in the c (+) direction is −1.35 °, and the OFF angle in the a direction is 1.01 °. Met. The dislocation density was 5.6 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

First, in the first heating step t _A, the N ₂ was heated while 10L / min supplied as a main flow into the reactor, a second heating step t _A when the temperature of the substrate became 400 ° C. Started. There, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow, and N ₂ was supplied at 17.5 L / min as a gas constituting the second main flow.

Thereafter, the substrate temperature was further raised to 700 ° C. while increasing NH ₃ and N ₂ to 12.5 L / min and 17.5 L / min, respectively. At this time, the sub-flow was N ₂ gas 20 L / min, and the growth non-growth gas such as for purging was N ₂ for a total of 19 L / min. In the second temperature raising step, the flow ratio Fp of the inert gas component to the total gas constituting the main flow was 0.50 at the start of the second temperature raising step and 0.58333 when the growth temperature was reached. .

In the next step B, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 29.5 L / min, H ₂ is 0.5 L / min, and H ₂ (0.5 L / min) which is a part of the main flow is a carrier. T as gas
MGa (0.0018 L / min with a concentration of 100%) was supplied into the furnace. With this main flow gas supply, an undoped GaN layer (first nitride semiconductor layer) was grown to a thickness of 1 nm.

At this time, the subflow was a mixed gas (20.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (20 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . During the growth of the first nitride semiconductor layer, the flow ratio Fp of the inert gas component to the total gas constituting the main flow was 0.73747.

In the next step C, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 39.24 L / min, H ₂
The 0.82L / min, TMGa and _H 2 in the main flow (0.5 L / min) as the carrier gas (0.0055L / min. At a concentration of 100%), constituting a part of the main flow _{H 2} SiH ₄ (6 × 10 ⁻⁸ L / min with a concentration of 100%) using (0.2 L / min) as a carrier gas and H ₂ (0.006 L / min) as a diluent gas was supplied into the furnace. . With such main flow gas supply, a Si-doped GaN layer (second nitride semiconductor layer) was grown to a thickness of 0.5 μm.

At this time, the subflow was a mixed gas (25.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (25 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . The flow rate ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the second nitride semiconductor layer was 0.78377.

Next, the substrate temperature (substrate temperature) was set to 740 ° C., and after the substrate temperature was sufficiently stabilized, the quantum well layer of In _0.11 Ga _0.89 N (target thickness 1.5 nm) and GaN (target thickness 13 nm) A multi-quantum well active layer structure in which barrier layers were alternately stacked for 8 periods was formed (step D).

Here, in the growth of the quantum well layer of In _0.11 Ga _0.89 N, NH ₃ (10 L / min) was used as a gas constituting the first main flow. In addition, as a gas constituting the second main flow, TM 2 (0.000015 L / min) using N ₂ (20 L / min) and N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas. TMIn (0.00023 L / min) using N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas was supplied.

In the growth of the GaN barrier layer, NH ₃ (10 L / min) was used as a gas constituting the first main flow. Further, as the gas constituting the second main flow, N ₂ (18.5 L / min), H ₂ (1.5 L / min), and H ₂ (0.5 L / min) which is a part of the main flow. ) Was used as a carrier gas, and TMGa (0.000017 L / min) was supplied.

In Step D, the subflow is NH ₃ (0.5 L / min) and N ₂ (25 L / min) mixed gas at 25.5 L / min, and the growth outgassing gas is N ₂ at 19 L / min. there were. The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the multiple quantum well active layer structure was 0.66666 for the InGaN well layer and 0.61667 for the GaN barrier layer.

Subsequently, the substrate temperature was set to 1000 ° C. to form an Mg-doped Al _0.1 Ga _0.9 N layer having a thickness of 50 nm (step E). The gas constituting the first main flow at this time is NH ₃ (10 L / min). The gas constituting the second main flow is TM2 (0.0001 L / min) using H ₂ (80 L / min) and H ₂ (0.5 L / min) which is a part of the main flow as a carrier gas. likewise TMGa (0.0018 L / min) which is part _{of H} 2 and (0.5 L / min) and carrier gas in the main flow, and, likewise _H 2 (0.5 L / min in the main flow ) As a carrier gas. Cp ₂ Mg (4 × 10 ⁻⁶ L / min).

On this Mg-doped Al _0.1 Ga _0.9 N layer, an Mg-doped GaN layer was further epitaxially grown to a thickness of 70 nm (step E). The growth of the GaN layer was performed with the supply of TMAl and H ₂ (50 L / min) out of the gas in the main flow described above.

The sub flow during the growth of the Al _0.1 Ga _0.9 N layer in the process E is 50.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (50 L / min), for purging. The growth outside gas was N ₂ (19 L / min). The flow rate ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped Al _0.1 Ga _0.9 N layer was 0.

Further, the sub-flow during the growth of the Mg-doped GaN layer in the process E is 20.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (20 L / min). Was N ₂ (19 L / min). The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped GaN layer was 0.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  The surface of the substrate thus produced was extremely flat. This surface was measured with a contact-type step meter, and the centerline average roughness (Ra) was determined. As a result, Ra according to the present example was 5.2 nm. In addition, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength is 412 nm, the integrated intensity is 90, which is a relative value, and a high standard deviation of the in-plane wavelength distribution is 1.0. % Was small.

  This embodiment is an example in which a blue LED is manufactured by laminating and growing a gallium nitride-based semiconductor thin film by MOCVD. The outline of a series of crystal growth processes has already been described with reference to FIG. The grown layer structure is schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 12 mm in the c-axis direction and 20 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 1.5 × 10 ¹⁸ cm ⁻³ . According to X-ray diffraction, the half-value width of the rocking curve in (10-12) reflection is 48.1 arcsec, the OFF angle in the c (+) direction is −0.85 °, and the OFF angle in the a direction is 2.64 °. Met. The dislocation density was 4.0 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

First, as the first temperature raising step t _B , the temperature is raised while supplying 10 L / min of N ₂ as a main flow into the furnace, and when the temperature of the substrate reaches 400 ° C., the second temperature raising step t _A is performed. Started. There, NH ₃ was supplied at 7.5 L / min as a gas constituting the first main flow, and N ₂ was supplied at 10 L / min as a gas constituting the second main flow.

Thereafter, the substrate temperature was further raised to 1000 ° C. while increasing NH ₃ and N ₂ to 10 L / min and 17.5 L / min, respectively. At this time, the sub-flow was N ₂ gas 20 L / min, and the growth non-growth gas such as for purging was N ₂ for a total of 19 L / min. In the second temperature raising step, the flow ratio Fp of the inert gas component to the total gas constituting the main flow was 0.57143 at the start of the second temperature raising step, and 0.63636 when the growth temperature was reached. .

In the next step B, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 49.5 L / min, H ₂ is 0.5 L / min, and H ₂ (0.5 L / min) which is a part of the main flow is a carrier. T as gas
MGa (0.0018 L / min with a concentration of 100%) was supplied into the furnace. With this main flow gas supply, an undoped GaN layer (first nitride semiconductor layer) was grown to a thickness of 15 nm.

At this time, the subflow was a mixed gas (20.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (20 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the first nitride semiconductor layer was 0.82498.

In the next step C, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 49.24 L / min, H ₂
The 0.82L / min, TMGa and _H 2 in the main flow (0.5 L / min) as the carrier gas (0.0055L / min. At a concentration of 100%), constituting a part of the main flow _{H 2} SiH ₄ (6 × 10 ⁻⁸ L / min with a concentration of 100%) using (0.2 L / min) as a carrier gas and H ₂ (0.006 L / min) as a diluent gas was supplied into the furnace. . With such main flow gas supply, a Si-doped GaN layer (second nitride semiconductor layer) was grown to a thickness of 5 μm.

At this time, the subflow was a mixed gas (25.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (25 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . At the time of growing the second nitride semiconductor layer, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0.81977.

Next, the substrate temperature (substrate temperature) was set to 700 ° C., and after the substrate temperature was sufficiently stabilized, a quantum well layer of In _0.14 Ga _0.86 N (target thickness 1.5 nm) and GaN (target thickness 13 nm) were formed. A multi-quantum well active layer structure in which barrier layers were alternately stacked for 8 periods was formed (step D). Here, in the growth of the In _0.14 Ga _0.86 N quantum well layer, NH ₃ (10 L / min) was used as a gas constituting the first main flow.

Further, as gas constituting the second main flow, TM ₂ (0.000008 L / min) using N ₂ (20 L / min) as a carrier gas and N ₂ (0.5 L / min) which is a part of the main flow as a gas. TMIn with N ₂ (0.5 L / min), which is part of the main flow, as a carrier gas
(0.00023 L / min) was supplied.

In the growth of the GaN barrier layer, NH ₃ (10 L / min) was used as a gas constituting the first main flow. Further, as the gas constituting the second main flow, N ₂ (18.5 L / min), H ₂ (1.5 L / min), and H ₂ (0.5 L / min) which is a part of the main flow. ) Was used as a carrier gas, and TMGa (0.000017 L / min) was supplied.

In Step D, the subflow is NH ₃ (0.5 L / min) and N ₂ (25 L / min) mixed gas at 25.5 L / min, and the growth outgassing gas is N ₂ at 19 L / min. there were. The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the multiple quantum well active layer structure was 0.66666 for the InGaN well layer and 0.61667 for the GaN barrier layer.

Subsequently, the substrate temperature was set to 1000 ° C. to form an Mg-doped Al _0.1 Ga _0.9 N layer having a thickness of 50 nm (step E). The gas constituting the first main flow at this time is NH ₃ (10 L / min).

The gas constituting the second main flow is TM2 (0.0001 L / min) using H ₂ (80 L / min) and H ₂ (0.5 L / min) which is a part of the main flow as a carrier gas. likewise TMGa (0.0018 L / min) which is part _{of H} 2 and (0.5 L / min) and carrier gas in the main flow, and, likewise _H 2 (0.5 L / min in the main flow ) As a carrier gas. Cp ₂ Mg (4 × 10 ⁻⁶ L / min).

On this Mg-doped Al _0.1 Ga _0.9 N layer, an Mg-doped GaN layer was further epitaxially grown to a thickness of 70 nm (step E). The growth of the GaN layer was performed with the supply of TMAl and H ₂ (50 L / min) out of the gas in the main flow described above.

The sub flow during the growth of the Al _0.1 Ga _0.9 N layer in the process E is 50.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (50 L / min), for purging. The growth outside gas was N ₂ (19 L / min). The flow rate ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped Al _0.1 Ga _0.9 N layer was 0.

Further, the sub-flow during the growth of the Mg-doped GaN layer in the process E is 20.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (20 L / min). Was N ₂ (19 L / min). The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped GaN layer was 0.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  The surface of the substrate thus produced was extremely flat. This surface was measured with a contact-type step meter, and the centerline average roughness (Ra) was determined. As a result, Ra according to the present example was 4.8 nm. In addition, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength is 440 nm, the integrated intensity is 51, which is a relative value, and a standard deviation of the in-plane wavelength distribution is 1.1. %Met.

  This embodiment is an example in which a near-ultraviolet LED is manufactured by stacking and growing a gallium nitride-based semiconductor thin film by MOCVD. An outline of a series of crystal growth processes will be described with reference to FIG. The grown layer structure is schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 10 mm in the c-axis direction and 15 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 6.7 × 10 ¹⁷ cm ⁻³ .

According to X-ray diffraction, the half-value width of the rocking curve in (10-12) reflection is 31.1 arcsec, the OFF angle in the c (+) direction is 4.2 °, and the OFF angle in the a direction is 0.02 °. there were. The dislocation density was 5.2 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a susceptor in a vertical reaction furnace made of SUS under reduced pressure growth under normal conditions. This apparatus has no piping corresponding to the subflow.

First, pressure in the reactor was set to 40 kPa, as the first heating step t _B, the N ₂ was heated while 20L / min supplied as a main flow into the furnace, where the temperature of the substrate became 400 ° C. It was initiated the second heating step t _a.

There, 1,1-dimethylhydrazine (1,1-DMHy), which is an inert gas, is used as a gas constituting the first main flow, and N ₂ (0.5 L / min), which is a part of the main flow, is used. As a carrier gas, 0.003 L / min was supplied, and as a gas constituting the second main flow, N ₂ was supplied at 15 L / min.

Thereafter, the substrate temperature was further increased to 1040 ° C. while increasing N ₂ to 20 L / min. At this time, non-growth gases such as for purging were 6.75 L / min in total with N ₂ . In the second temperature raising step, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 1.0 from the start of the second temperature raising step to the arrival of the growth temperature.

In the next step B, the pressure in the reactor is 40 kPa, and N ₁ (0.5 L / min), which is a part of the main flow, is used as a carrier gas for the gas constituting the first main flow. DMHy was supplied at 0.0229 L / min. As the gas constituting the second main flow, N ₂ is 20 L / min, N ₂ (0.5 L / min) which is a part of the main flow is used as a carrier gas, and TMGa (0.0018 L with a concentration of 100%). / Min) was fed into the furnace. With this main flow gas supply, an undoped GaN layer (first nitride semiconductor layer) was grown to a thickness of 40 nm. At this time, the non-growth gas for purging and the like was 6.375 L / min for N ₂ . During the growth of the first nitride semiconductor layer, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0.99875.

In the next step C, the pressure in the reactor is 8 kPa, and N ₁ (2.27 L / min) which is part of the main flow is used as a gas constituting the first main flow as a carrier gas and 1,1-DMHy Of 0.104 L / min. As the gas constituting the second main flow, N ₂ is 20 L / min, N ₂ (0.5 L / min) which is a part of the main flow is used as a carrier gas, and TMGa (0.0018 L with a concentration of 100%). / Min), SiH ₄ (6 × 10 ⁻⁸ L / min as a concentration of 100%) using N ₂ (0.2 L / min) constituting a part of the main flow as a diluent gas was supplied into the furnace. With such main flow gas supply, a Si-doped GaN layer (second nitride semiconductor layer) was grown to a thickness of 4 μm.

At this time, the non-growth gas for purging and the like was 6.375 L / min for N ₂ . During the growth of the second nitride semiconductor layer, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0.99448.

Next, the pressure in the reactor was 67 kPa, the substrate temperature (substrate temperature) was 740 ° C., and In _0.
A multi-quantum well active layer structure was formed in which a quantum well layer of ₀₇ Ga _0.93 N (target thickness 1.5 nm) was alternately stacked for five periods with a barrier layer of GaN (target thickness 13 nm) at a substrate temperature of 840 ° C. (Process D). Here, in the growth of the quantum well layer of In _0.07 Ga _0.93 N, NH ₃ (12 L / min) was used as a gas constituting the first main flow. Also, as a gas constituting the second main flow, TM ₂ (0.000008 L / min) using N ₂ (10 L / min) as a carrier gas and N ₂ (0.5 L / min) which is a part of the main flow. TMIn (0.00023 L / min) using N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas was supplied.

In the growth of the GaN barrier layer, NH ₃ (12 L / min) was used as a gas constituting the first main flow. Further, as a constituent gas of the second main _flow, and N 2 (9.5 L / _min), and H 2 (0.5 L / min), _H 2 (0.5 L / min in the main flow ) Was used as a carrier gas, and TMGa (0.000017 L / min) was supplied.

Further, the growth outside gas for purging and the like was 6.375 L / min for N ₂ . The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the multiple quantum well active layer structure was 0.45452 for the InGaN well layer and 0.43179 for the GaN barrier layer.

Subsequently, the pressure in the reactor was 36 kPa, the substrate temperature was 1000 ° C., and an Mg-doped GaN layer was epitaxially grown to a thickness of 120 nm (step E). At this time, the gas constituting the first main flow was supplied with 1,2-DMHy at 0.0229 L / min using N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas. The gas constituting the second main flow is N ₂ (14 L / min) and H ₂ (1 L / min), and H ₂ (0.5 L / min) which is a part of the main flow is used as a carrier gas. It is Cp ₂ Mg (4 × 10 ⁻⁶ L / min) using TMGa (0.0018 L / min) and H ₂ (0.5 L / min) which is also a part of the main flow as a carrier gas.

Further, the gas outside the growth, such as for purging during the growth of the Mg-doped GaN layer in the process E, was N ₂ (6.375 L / min). The flow rate ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped GaN layer was 0.93180.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  The surface of the substrate thus produced had good flatness although there were very slight irregularities. This surface was measured with a contact-type step meter, and the centerline average roughness (Ra) was determined. As a result, Ra according to the present example was 9.1 nm. In addition, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength is 395 nm, the integrated intensity is 109, which is a relative value, and a standard deviation of the in-plane wavelength distribution is 0.9. % Was small.

  This example is an example in which a near-ultraviolet LED is manufactured by stacking and growing a gallium nitride-based semiconductor thin film by MOCVD, and an outline of a series of crystal growth processes will be described with reference to FIG. The grown layer structure is schematically shown in FIG.

  As the substrate, an m-plane template obtained by growing a GaN layer having an m-plane orientation on a (1-100) plane (m-plane) ZnO substrate having a 2 inch diameter size to 0.5 μm is used.

  Using this m-plane GaN template, the m-plane GaN template is placed on a susceptor in a self-revolving SUS reactor that uses reduced pressure growth under normal conditions.

First, as the first temperature raising step t _B , the pressure in the reactor is 40 kPa, the temperature is raised while supplying N ₂ 30 L / min as the main flow into the furnace, and the temperature of the substrate reaches 400 ° C. it is possible to start the second heating step t _a. There, NH ₃ is supplied at 7.5 L / min as a gas constituting the first main flow, N ₂ is supplied at 27.5 L / min and Ar is supplied at 10 L / min as gases constituting the second main flow.

Thereafter, the pressure in the reactor is 40 kPa, and the substrate temperature is further increased to 1000 ° C. while increasing NH ₃ and N ₂ to 10 L / min and 80 L / min, respectively. At this time, the sub-flow is N ₂ gas 30 L / min, and the non-growth gas for purging and the like is N ₂ for 21 L / min in total. In the second temperature raising step, the flow ratio Fp of the inert gas component to the total gas constituting the main flow is 0.83333 at the start of the second temperature raising step, and 0.90 when the growth temperature is reached.

In the next step B, the pressure in the reaction furnace is 40 kPa, and NH ₃ is supplied as a gas constituting the first main flow at 10 L / min. Further, as the gas constituting the second main flow, N ₂ is 50 L / min, Ar is 10 L / min, and N ₂ (0.5
TMGa (0.0081 L / min with a concentration of 100%) is supplied into the furnace as a carrier gas. With such main flow gas supply, an undoped GaN layer (first nitride semiconductor layer) can be grown to a thickness of 10 nm.

Meanwhile, a subflow _NH 3 (1L / min) and mixed gas of _N 2 (34L / min) (35L / min), growth-gases such as for purging and 21L / min with _{N 2.} When the first nitride semiconductor layer is grown, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow is 0.85704.

In the next step C, the pressure in the reaction furnace is 12 kPa, and NH ₃ is supplied at 10 L / min as a gas constituting the first main flow. Further, as the gas constituting the second main flow, N ₂ is 70 L / min, Ar is 10 L / min, and N ₂ (0.5
LH / min) as carrier gas, TMGa (0.026 L / min as 100% concentration), and SiH ₄ (as concentration 100%) using N ₂ (0.2 L / min) as part of the main flow as diluent gas 1.8 × 10 ⁻⁶ L / min) is fed into the furnace. With such main flow gas supply, a Si-doped GaN layer (second nitride semiconductor layer) can be grown to a thickness of 15 μm.

Meanwhile, a subflow _NH 3 (1L / min) and mixed gas of _N 2 (34L / min) (35L / min), growth-gases such as for purging and 21L / min with _{N 2.} At the time of growing the second nitride semiconductor layer, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow is 0.88864.

Next, the pressure in the reactor is 40 kPa, the substrate temperature (substrate temperature) is 740 ° C., the quantum well layer of In _0.08 Ga _0.92 N (target thickness 1.5 nm) and GaN (target thickness 13 nm). A multi-quantum well active layer structure can be formed by alternately stacking five barrier layers (step D). Here, in the growth of the quantum well layer of In _0.08 Ga _0.92 N, NH ₃ (10 L / min) is used as a gas constituting the first main flow. Also, as a gas constituting the second main flow, TM 2 (0.00013 L / min) using N ₂ (60 L / min) as a carrier gas and N ₂ (0.5 L / min) which is a part of the main flow is used. Then, TMIn (0.00023 L / min) using N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas is supplied.

In the growth of the GaN barrier layer, NH ₃ (10 L / min) is used as a gas constituting the first main flow. Further, as gas constituting the second main flow, TM 2 (0.0075 L / min) using N ₂ (60 L / min) and N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas. ).

Further, in the step D, the sub-flow is 35 L / min with a mixed gas of NH ₃ (1 L / min) and N ₂ (34 L / min), and the growth outside gas such as for purging is 21 L / min with N ₂ . The flow ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the multiple quantum well active layer structure is 0.85713 for the InGaN well layer and 0.85713 for the GaN barrier layer.

Subsequently, the pressure in the reactor is 40 kPa, the substrate temperature is 1000 ° C., and an Mg-doped GaN layer can be formed to 120 nm (step E). The gas constituting the first main flow at this time is NH ₃ (10 L / min). The gas constituting the second main flow is TM ₂ (0.0081 L / min) using N ₂ (60 L / min) and N ₂ (0.5 L / min) which is a part of the main flow as a carrier gas. And Cp ₂ Mg (6 × 10 ⁻⁵ L / min) using N ₂ (0.5 L / min) which is also a part of the main flow as a carrier gas.

Here, the subflow is 35 L / min with a mixed gas of NH ₃ (1 L / min) and N ₂ (34 L / min), and N ₂ (21 L / min) is used as a non-growth gas for purging. The flow rate ratio Fp of the inert gas component to the total gas constituting the main flow during the growth of the Mg-doped GaN layer is 0.85704.

In this manner, a near-ultraviolet LED can be fabricated on an m-plane GaN template (substrate) using a ZnO substrate.
[Reference Example 1]

  This embodiment is an example in which a gallium nitride based semiconductor thin film is laminated and grown by MOCVD to produce a near-ultraviolet light emitting LED. The outline of a series of crystal growth processes has already been described with reference to FIG. . The grown layer structure is schematically shown in FIG.

As a substrate, an undoped GaN layer of 1 μm and an n-type GaN layer (carrier concentration of 5 × 10 ¹⁸ cm ⁻³ ) of 3.5 μm were continuously stacked on c-plane sapphire which is a polar surface, and once taken out from the furnace A substrate was used. GaN on the surface of the substrate is a polar surface and is c (+)-oriented. This substrate was placed on a tray (susceptor) in a quartz horizontal reactor under normal pressure growth conditions. The pressure in the reactor here was 100 ± 2 kPa in all steps.

  All the growth conditions from the temperature raising step to the end were the same as those in Example 1.

  The surface of the substrate thus prepared was extremely rough and completely clouded. This surface was measured with a contact-type step meter, and the centerline average roughness (Ra) was determined. As a result, Ra according to the present example was 376 nm. When an attempt was made to evaluate the PL characteristics by excitation with a laser beam having a wavelength of 325 nm, no PL emission could be observed.

  In this example, the flatness of a multiple quantum well (MQW) layer of an LED that emits near-ultraviolet light produced by the crystal growth method of the present invention was confirmed by a transmission electron microscope (TEM) observation method. The outline of the series of crystal growth processes has already been described with reference to FIG. 2A, and the layer structure of the obtained LED is the same as that schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.0 mm in the c-axis direction and 20 mm in the a-axis direction. The substrate was Si-doped (Si concentration 6.2 × 10 ¹⁷ cm ⁻³ ), the electrical characteristics were n-type, and the carrier density was 6.5 × 10 ¹⁷ cm ⁻³ .

According to X-ray diffraction, the half-value width of the rocking curve in (10-12) reflection is 36.2 arcsec, the OFF angle in the c (+) direction is −0.145 °, and the OFF angle in the a direction is 0. It was 05 °. Furthermore, the dislocation density was 5.1 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate is placed on a tray (susceptor) in a quartz horizontal reactor where atmospheric pressure growth is normal, and the film thickness of the second semiconductor layer (Si-doped n-type GaN layer) LED having the layer configuration of FIG. 3A was manufactured under the same conditions as in Example 1 except that the thickness was changed to 4.2 μm.

  When the surface of the substrate thus obtained was measured with a contact-type step gauge, it was confirmed that Ra = 4.7 nm, which was a very flat surface. Further, in the PL characteristics evaluated by laser light excitation with a wavelength of 325 nm, the peak wavelength was 394 nm, and the integrated intensity thereof was 89, which is a relative value, indicating a very high PL intensity.

  The LED sample having such characteristics was thinned by the following procedure and observed with a high-angle scattering dark-field scanning transmission electron microscope (HAADF-STEM). First, using a parallel pre-polishing machine and a rotating polishing plate, mechanical polishing is performed so that the film thickness is 50 μm or less, and then using a 691 type PIPS (registered trademark) manufactured by Gatan, an acceleration voltage of 2 kV to 5 kV, milling Ar ion milling was performed at an angle of 2 to 4 ° to form a thin film (having a thickness of about 50 nm) suitable for HAADF-STEM (HAADF-STEM: high-angle annular dark-field scanning transmission electron microscopy). The thin film sample thus obtained was fixed to a Mo mesh and used as an observation sample.

The apparatus used for the HAADF-STEM observation is TECNAI G2 F20 manufactured by FEI Company equipped with a field emission electron gun (Field Emission Gun: FEG). This device has a HAA described later.
An annular dark field detector (Model 3000 manufactured by Fischione) for carrying out DF-STEM observation is mounted.

  Note that, as will be described later, the purpose of observation of this example is to clearly observe the InGaN / GaN boundary and clarify the fluctuation of the interface, the fluctuation of the film thickness, and the degree of unevenness. HAADF-STEM was applied in order to make the contrast of the layer clear and observe the sample without damaging it. The acceleration voltage of the electron beam was 200 kV.

  The above-mentioned observation sample was observed by HAADF-STEM at the a-axis incidence. The observed visual field was made into the area | region where the average thickness of each quantum well layer is calculated | required, without sacrificing the resolution which a HAADF-STEM apparatus has. Specifically, it was 190 nm in the direction perpendicular to the stacking direction of the quantum well layers.

  Here, the reason why the HAADF-STEM method is applied will be briefly described. In order to accurately observe the thickness and the like of the quantum well (QW), it is assumed that the observation sample is not damaged. This is because if the sample is damaged by electron beam irradiation, the obtained TEM image may frequently become an image (artifact) that does not accurately reflect the original state (for example, Non-Patent Document 3: CJ Humphreys et al., “Applications Environmental Impact and Microstructure of Light-Emitting Diodes” MICROSCOPY AND ANALYSIS, NOVEMBER 2007, pp.5-8).

For this reason, low dose (lower current density per unit time), low current density (lower current density per unit area), and cooling the sample and observing at low temperature Is essential. As an observation method satisfying these conditions, a low-temperature observation method using the HAADF-STEM method is most desirable.

  Regarding the HAADF-STEM method, for example, Non-Patent Document 4 (K. SAITOH, “High-resolution Z-contrast Imaging by the HAADF-STEM Method”, J. Cryst. Soc. Jpn., 47 (1), 9- 14 (2005)) and Non-Patent Document 5 (K. WATANABE; “Imaging in High Resolution HAADF-STEM”, J. Cryst. Soc. Jpn., 47 (1), 15-19 (2005)). .

  The HAADF-STEM method uses an annular detector to receive inelastically scattered electrons at high angles due to diffuse thermal scattering due to lattice vibration, among STEM images observed by scanning transmission electron microscopy (STEM). In this method, the integrated intensity of electrons is measured as a function of the probe position, and the intensity is displayed as an image. The brightness of the image obtained by this method is proportional to the square of the atomic number. When observing a sample such as a GaN / InGaN system, there is an advantage that the contrast of a layer containing a heavy atom such as In becomes clear and the sample can be observed at a low dose.

  The specific observation conditions in this example are an electron beam incident angle of 10 mrad or less, a spot diameter of 0.2 nm or less, and a detection scattering angle of 70 mrad or more. Image data collection conditions are 640,000 times magnification and pixel resolution. 0.186 nm / pix (pixel), scan speed of 64 μsec / pix, and pixel frame of 1024 pix square. The pixel resolution (0.186 nm / pix) is a condition that substantially matches the STEM resolution (0.18 nm) of the apparatus used in this experiment.

  By collecting such image data, intensity data for each pixel of 1024 × 1024 pix (about 1.05 million pix) is obtained, and a two-dimensional intensity distribution is obtained. Since 0.186 nm / pix × 1024 pix is approximately 190 nm, it corresponds to the above-described quantum well layer observation region (190 nm).

  The incident direction of the electron beam to the observation sample was made parallel to the [11-20] direction of the wurtzite crystal structure GaN. Note that the observation may be performed under an electron beam incident condition parallel to the [0001] direction. Furthermore, other electron beam incident directions can be selected.

For cooling the sample, a liquid nitrogen injection type cryo holder (Model 636DHM manufactured by GATAN) is used. The minimum sample temperature when using this cryoholder is in the temperature range of 77 to 140 K about 40 minutes after liquid nitrogen injection, and there is also a report on an experiment in which a temperature was measured by attaching a thermocouple to the observation sample. Yes (Non-Patent Document 6).

TEM observation is performed when checking the field of view and determining the incident direction of the electron beam. In this TEM observation, the current density of the electron beam is adjusted to 30 to 500 A / m ² in order to avoid damage to the sample. I went there. It has been confirmed that the morphology of the sample does not change even when the electron beam of 500 A / m ² is irradiated for 10 minutes under the cryo-condition.

  FIG. 6 is a HAADF-STEM image obtained in this way, and FIG. 7 is a result after image processing performed for calculating the average thickness of each quantum well layer from the HAADF-STEM image.

The image processing performed here includes (1) binarizing the HAADF-STEM image, (2) approximating each quantum well layer to an ellipsoid, and (3) the major axis of the ellipsoid. Horizontal direction of the image (
The image is rotated so that it is parallel to the X-axis direction. (4) A method of measuring all thicknesses in the Y-axis direction for all the pixels in the X-axis direction for each quantum well layer, and (5) obtaining “average thickness / standard deviation / variation coefficient of each quantum well layer”. It was calculated by. Here, the “variation coefficient” is a coefficient obtained by dividing the standard deviation value by the average value, and the value is referred to as “CV value”.

  Table 2 summarizes the average thickness (nm) of each quantum well layer (w1 to w8) obtained in this way, w1 is the average thickness of the quantum well layer on the lowermost side (substrate side), w8 is the average thickness of the uppermost (surface side) quantum well layer. In each layer, the standard deviation of thickness in the Y-axis direction (see Equation 1 described later) and the coefficient of variation (CV value) are also shown.

  The sample is an LED manufactured by stacking and growing a gallium nitride-based semiconductor thin film by the MOCVD method, and the outline of a series of crystal growth processes has been described with reference to FIG. The structure has a 5-cycle structure as shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.1 mm in the c-axis direction and 15 mm in the a-axis direction. The substrate was doped with Si, and the Si concentration was 6.2 × 10 ¹⁷ cm ⁻³ . The electrical characteristics of the substrate were n-type and the carrier density was 6.5 × 10 ¹⁷ cm ⁻³ .

According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 38.9 arcsec, the OFF angle in the c (+) direction is −0.165 °, and the OFF angle in the a direction is 0.05. °. The dislocation density was 5.8 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reactor under normal pressure growth conditions to produce an LED structure.

  The temperature raising step was performed in the same manner as in Example 8.

In the next step B, NH ₃ 10 L / min was supplied as a gas constituting the first main flow. Further, as the gas constituting the second main flow, N ₂ is 0 L / min, H ₂ (30 L / min), and H ₂ (0.5 L / min) which is a part of the main flow is used as a carrier gas.
(0.0018 L / min as 100% concentration) was fed into the furnace. With such main flow gas supply, an undoped GaN layer (first nitride semiconductor layer) was grown to a thickness of 40 nm (step B).

At this time, the subflow was a mixed gas (20.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (20 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . During the growth of the first nitride semiconductor layer, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0.0.

In the next step C, NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. Further, as gas constituting the second main flow, TM 2 (concentration of 100%) using H ₂ (40 L / min) and H ₂ (0.5 L / min) constituting a part of the main flow as a carrier gas is used. 0.0055 L / min), SiH ₄ (concentration of 100) using H ₂ (0.2 L / min) constituting a part of the main flow as a carrier gas and H ₂ (0.06 L / min) as a diluent gas. %, 6 × 10 ⁻⁷ L / min). With this main flow gas supply, a Si-doped GaN layer (second nitride semiconductor layer) was grown to a thickness of 1.7 μm.

At this time, the subflow was 25.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (25 L / min), and the growth outside gas such as for purging was 19 L / min with N ₂ . During the growth of the first nitride semiconductor layer, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0.0.

Subsequent process D was made into the completely same conditions as Example 8 except having made the period number into 5 periods.
Step E was performed under the same conditions as in Example 8.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  Some irregularities were confirmed on the surface of the substrate thus fabricated. When the surface was measured with a contact-type step meter, Ra = 210 nm, it was easy to interact with the emission wavelength, and the roughness that was expected to exhibit the scattering function was naturally formed. Further, in the PL characteristics evaluated by laser light excitation with a wavelength of 325 nm, the peak wavelength was 423 nm and the integrated intensity was 11 in relative value.

  FIG. 8 is a HAADF-STEM image obtained from the LED thus obtained by the same procedure as in Example 8, and FIG. 9 is the result after the image processing of FIG.

  Table 3 summarizes the average thickness (nm) of each quantum well layer (w1 to w5) of this comparative sample obtained as a result of image processing, and w1 is the quantum well layer on the lowermost side (substrate side). The average thickness, w5, is the average thickness of the uppermost (surface side) quantum well layer. In each layer, the standard deviation of thickness in the Y-axis direction (see Formula 1 described later) and the coefficient of variation are also shown.

FIG. 10 is a graph plotting how the standard deviation of the average thickness of each quantum well layer changes as the quantum wells shown in Example 8 and Example 9 are stacked from the substrate side. . Here, as described above, the “average thickness of each quantum well layer” is as follows: (1) Binarization processing is performed on the HAADF-STEM image, (2) Each quantum well layer is approximated by an ellipsoid, 3) The image is rotated so that the major axis of the ellipsoid is parallel to the horizontal direction (X-axis direction) of the image. (4) A method of measuring all thicknesses in the Y-axis direction for all the pixels in the X-axis direction for each quantum well layer, and (5) obtaining “average thickness / standard deviation / variation coefficient of each quantum well layer”. It was calculated by. The “variation coefficient” is a value obtained by dividing “standard deviation” by “average” (standard deviation / average).

The standard deviation of the average thickness (standard deviation of the average thickness of each quantum well layer) y (n) (nm) when n cycles were laminated from the substrate side was calculated by (Formula 1) shown below. Here, w n _(nm) is the mean thickness of the n-th quantum well layer from the substrate side.

The above equation is calculated with n> 1, and when n = 1, the standard deviation of the average thickness is 0.

  In the example 8, the standard deviation y (n) (nm) of the average thickness of each quantum well layer shows a relationship of the quantum well period number n and approximately y (n) = 0.0079 (n−1), The standard deviation of the average thickness of each quantum well layer was extremely small. On the other hand, in Example 9, the relationship is generally y (n) = 0.0735 (n−1), and the standard deviation of the average thickness of each quantum well layer increases as the quantum well layers are stacked. Showed a tendency to become.

  In Example 8, the thickness of each quantum well layer is designed to be 3.64 nm. In Example 9, the thickness of each quantum well layer is designed to be 4.80 nm. The remarkable difference in the standard deviation y (n) (nm) of the average thickness of the well layer is not caused by the difference in the design value of the average thickness, but is caused by the difference in crystallinity including the flatness of the quantum well layer. Is.

As described above, the nitride semiconductor of the present invention has a quantum well period n of the standard deviation y (n) of the average thickness of each quantum well layer as evaluated by the HAADF-STEM method under the condition that the damage to the observation sample is small. Dependency can be confirmed. As shown here, the flatness of the quantum well active layer structure in the present invention is changed by changing the growth conditions of the first semiconductor layer and the second semiconductor layer formed before the growth of the quantum well layer. It is possible to make it.

  In particular, it can be seen that the standard deviation with respect to the average thickness of the quantum well layer is increased by stacking multiple layers even in the case of having appropriate fluctuations and in the case of being flat. It is preferable that the quantum well layer has a moderate fluctuation in the above, and in this sense, it is preferable to increase the total number of quantum wells within an appropriate range.

In this example, a sample similar to that in Example 8 was prepared, and an ultra high voltage transmission electron microscope (ultra high voltage TE
The results observed in M) are shown. In general-purpose TEM, 100 to 400 kV
Although the acceleration voltage is of the order, this effective acceleration voltage has an effective electron transmissivity of about several hundred nanometers at most, and is not sufficient for observing crystal defects and dislocations in the assumed epitaxial growth layer on the GaN substrate. The reason is as follows.

Unlike homoepitaxial growth of GaN-based materials on sapphire substrates, etc., homoepitaxial growth of GaN-based materials on GaN substrates essentially reduces low dislocations if good epitaxial growth is fully studied. It is assumed that it is possible.

  Therefore, in observing crystal defects and dislocations, it is necessary to prepare a sample having a certain “thickness” so that dislocations can be included in the observation field even if the sample has a low dislocation density. In order to observe a sample having such a thickness by TEM, an electron acceleration voltage at the time of observation of 1000 kV or more is required.

  With such an electron acceleration voltage, the observable sample thickness can be increased to about 1 to 2 μm, and dislocations can be observed even with the low dislocation density sample as described above. . If dislocation lines cannot be observed even when a sample having such a thickness is observed, it is expected that the dislocation density is considerably low. In other words, it is impossible to conclude that the dislocation density is sufficiently low even if a dislocation line is not observed by observing a sample having a thickness of several hundred nm at most.

Specifically, in a sample having a thickness of about several hundreds nm at most, if dislocation is not observed by observing one visual field, it indicates that the dislocation density is about 108 (cm −2) or less. On the other hand, when dislocations are not observed by observing one field of a sample thickened to about 1 to 2 μm, the dislocation density is about 107 (cm −2) or less at the maximum. It is shown.

  Further, since dislocation may not be observed depending on the incident direction of the electron beam, the incident electron beam is preferably adjusted to be parallel to the c-axis direction.

  In view of the above, the TEM observation in this example was performed using an ultrahigh voltage TEM having an effective permeability exceeding 1 μm and an electron beam acceleration voltage of 1000 kV. The sample thickness was about 1.5 μm, and the incident electron beam was parallel to the c-axis direction. The ultra-high pressure TEM device used was JEM-ARM1000 manufactured by JEOL.

  FIG. 11 is an ultrahigh-pressure TEM observation image under the above conditions. As shown in this photograph, no crystal defects or dislocations were observed.

  In this example, the light emission characteristics of a multiple quantum well (MQW) layer manufactured by the crystal growth method of the present invention are evaluated by a photoluminescence (PL) method, and an MQW structure having excellent internal quantum efficiency and light emission lifetime is obtained. Confirmed that.

  FIG. 3D is a diagram showing a stacked structure of a sample used in this example, and FIG. 2F is a diagram for explaining a sequence of a crystal growth method of this sample. In this sample, an InGaN / GaN multiple quantum well active layer structure 13 having a period number of 5 is provided in the uppermost layer.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.1 mm in the c-axis direction and 15 mm in the a-axis direction. The substrate was doped with Si, and the Si concentration was 6.2 × 10 ¹⁷ cm ⁻³ . The electrical characteristics of the substrate were n-type, and the carrier density was 6.5 × 10 ¹⁷ cm ⁻³ .

According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 35 arcsec, the OFF angle in the c (+) direction is -0.165 °, and the OFF angle in the a direction is 0.05 °. It was. The dislocation density was 4.9 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reactor under normal pressure growth conditions to produce a quantum well structure. The temperature of the second semiconductor layer (Si-doped n-type GaN layer) was 2.5 μm, the number of periods of the quantum well grown in step D was five, and the temperature was lowered without performing step E. The crystal growth conditions were the same as in Example 1 except for the above.

  When the surface of the substrate thus prepared was measured with a contact-type step meter, Ra = 4.8 nm, which was a very flat surface. Further, in the PL characteristics evaluated by excitation with laser light having a wavelength of 325 nm at room temperature, the peak wavelength was 393 nm, and the integrated intensity thereof was 196 in relative value, indicating a very high PL intensity. The internal quantum efficiency and PL emission lifetime of such a sample were evaluated as follows.

  First, the internal quantum efficiency is evaluated by performing PL (CW-PL) measurement when the irradiation light to the sample is continuous wave (Continuous Wave: CW) and time-resolved PL measurement when irradiating pulse light. did.

FIG. 12 is a block diagram of an optical system used in CW-PL measurement. A He—Cd laser 101 having a wavelength of 325 nm is used as an excitation light source, the intensity of the laser light is adjusted by the ND filter 102, and the sample 104 placed inside the cryostat 103 is irradiated. The PL light from the sample 104 is dispersed by the spectroscope 106 through the λ / 4 wavelength plate 105, measured by the photomultiplier tube 107, and collected by the control computer 108. The sample temperature was in the range of 13K to 300K, and the laser power on the sample surface was 150 W / cm ² per unit area. The photoexcited steady excess carrier density in this condition is estimated to be in the range of 1 × 10 ¹⁶ cm ⁻³ to 1 × 10 ¹⁷ cm ⁻³ .

  As a result of this CW-PL measurement, when the integrated intensity I (13K) obtained by integrating the PL intensity from the multiple quantum well at 13K with the emission energy is 1, the same integrated intensity I (300K) at 300K is I (13K). Of 58%.

  FIG. 13 is a block diagram of an optical system used in time-resolved PL measurement (with a polarizing filter). In this time-resolved PL measurement, an optical system including a tunable pulse laser light source 109 and a streak scope (manufactured by Hamamatsu Photonics) 112 was used. The tunable pulse laser light source was composed of a mode-locked Ti: sapphire laser, a Ti: sapphire regenerative amplifier, an optical parametric amplifier, and a harmonic generation crystal. The pulse repetition frequency was 1 kHz and the pulse width was 120 fs.

  PL measurement by selective excitation is effective for directly evaluating the optical quality of a quantum well layer having a single layer or a plurality of layers inherent in the active layer structure. For example, in order to selectively excite a plurality of quantum well layers, it is necessary to select light having energy smaller than the band gap of GaN and larger than the band gap of the InGaN quantum well layer. It was 370 nm. After the pulse energy is adjusted by the ultraviolet ND filter 102, the sample 104 in the cryostat 103 that can be cooled by the liquid He is irradiated. The PL from the sample 104 passes through the polarizing filter 110 and the depolarizer 111 and is dispersed by the spectroscope 106 and then guided to the streak scope 112. The sample temperature was changed from 4K to 300K.

The polarizing filter 110 is disposed between a condenser lens (not shown) and the spectroscope 106, and is used to selectively transmit light having an electric field component perpendicular to the c-axis of the sample 104 into the spectroscope 106. ing. The depolarizer 111 is for preventing the influence of the polarization dependence of the diffraction characteristics of the spectroscope 106. The pulse energy applied to the sample 104 was determined by measuring the power with a power meter and dividing by the repetition frequency. The pulse energy density per unit area was 1400 nJ / cm ² .

Although the exact value of the absorption coefficient α at the excitation wavelength of one quantum well layer inherent in the nitride semiconductor layer prepared in this experiment is unknown, it is 1 × 10 ⁴ cm ⁻¹ to 1 × 10 ⁵ cm ⁻¹ . The range is considered to be a valid range. When the absorption coefficient α in one quantum well layer is 1 × 10 ⁴ cm ⁻¹ , the average excess carrier density excited in all the quantum well layers inherent in the nitride semiconductor layer is approximately 3 × 10 ¹⁶ cm ^−3. It is estimated. Similarly, when the absorption coefficient α is 5 × 10 ⁴ cm ⁻¹ , the average excess carrier density to be excited is estimated to be approximately 1 × 10 ¹⁷ cm ^−3, and the absorption coefficient α is 1 × 10 ⁵ cm. ^In the case of ^−1, the average excess carrier density excited is estimated to be approximately 3 × 10 ¹⁷ cm ⁻³ .

As a result of this time-resolved PL measurement, the transient response of PL emission from a multiple quantum well at 4K is time-integrated to obtain the PL intensity, and further, the integrated intensity I (4K) obtained by integrating this with the emission energy is set to 1, and at 300K Similar integrated intensity I (300K) was 93% of I (4K).

Also, at 300K, the pulse energy density per unit area is 1400 nJ / c.
The PL lifetime was determined from the transient response (attenuation curve) with respect to time of the PL intensity after pulse excitation obtained by time-resolved PL measurement (with a polarizing filter), which was m ² . When the time from the maximum intensity to 1 / e of the maximum intensity in the PL intensity decay curve was defined as the PL life, the PL life τ (PL) was 3.6 nsec. The luminescence recombination lifetime τ (R) was 4.4 ns, and the non-radiative recombination lifetime τ (NR) was 20.6 ns. The luminescence recombination lifetime τ (R) is expressed by the equation τ (R) = τ (PL) / η, where η is the internal quantum efficiency, and the non-radiative recombination lifetime τ (NR) is expressed by the equation τ (NR) = τ (PL ) / (1-η). As a result, τ (R) <τ (
NR).

  FIG. 14 is a block diagram of an optical system used in time-resolved PL measurement (without a polarizing filter), which differs from that in FIG. 13 only in that the polarizing filter 110 is omitted.

  The pulse repetition frequency was 1 kHz and the pulse width was 120 fs. The tunable pulse laser light source was composed of a mode-locked Ti: sapphire laser, a Ti: sapphire regenerative amplifier, an optical parametric amplifier, and a harmonic generation crystal.

  In order to directly evaluate the optical quality of a quantum well layer having a single layer or a plurality of layers inherent in the active layer structure, PL measurement by selective excitation is effective. For example, in order to selectively excite a plurality of quantum well layers, it is necessary to select light having energy smaller than the band gap of GaN and larger than the band gap of InGaN quantum well layers. Was 370 nm.

After the pulse energy is adjusted by the ultraviolet ND filter 102, the sample 104 in the cryostat 103 that can be cooled by the liquid He is irradiated. The PL from the sample 104 passes through the depolarizer 111 and is dispersed by the spectroscope 106 and then guided to the streak scope 112. The sample temperature was changed from 4K to 300K. The pulse energy applied to the sample 104 was determined by measuring the power with a power meter and dividing by the repetition frequency. The pulse energy density per unit area was 510 nJ / cm ² .

Although the exact value of the absorption coefficient α at the excitation wavelength of one quantum well layer inherent in the nitride semiconductor layer prepared in this experiment is unknown, it is 1 × 10 ⁴ cm ⁻¹ to 1 × 10 ⁵ cm ⁻¹ . The range is considered to be a valid range. When the absorption coefficient α in one quantum well layer is 1 × 10 ⁴ cm ⁻¹ , the average excess carrier density excited in all the quantum well layers inherent in the nitride semiconductor layer is approximately 1 × 10 ¹⁶ cm ^−3. It is estimated. Similarly, when the absorption coefficient α is 5 × 10 ⁴ cm ⁻¹ , the average excess carrier density excited is estimated to be approximately 5 × 10 ¹⁶ cm ^−3, and the absorption coefficient α is 1 × 10 ⁵ cm. ^In the case of ^−1, the average excess carrier density excited is estimated to be approximately 1 × 10 ¹⁷ cm ⁻³ .

As a result of this time-resolved PL measurement, the transient response of PL emission from a multiple quantum well at 4K is time-integrated to obtain the PL intensity, and further, the integrated intensity I (4K) obtained by integrating this with the emission energy is set to 1, and at 300K A similar integrated intensity I (300K) was 60% of I (4K).

Further, at 300 K, the pulse energy density per unit area is 510 nJ / cm.
^The PL lifetime was obtained from the transient response (attenuation curve) with respect to time of the PL intensity after pulse excitation obtained by time-resolved PL measurement (without a polarizing filter) of 2. When the time from the maximum intensity to 1 / e of the maximum intensity in the PL intensity decay curve was defined as the PL life, the PL life τ (PL) was 2.5 nsec. The luminescence recombination lifetime τ (R) was 4.1 ns, and the non-radiative recombination lifetime τ (NR) was 6.3 ns. The luminescence recombination lifetime τ (R) is expressed by the equation τ (R) = τ (PL) / η, where η is the internal quantum efficiency, and the non-radiative recombination lifetime τ (NR) is expressed by the equation τ (NR) = τ (PL ) / (1-η).

  The above-mentioned internal quantum efficiency obtained in this example is a significantly higher value than conventionally known values, and the non-radiative recombination lifetime τ (NR) and the PL lifetime τ (PL) are also significant. Long value. These support the high quality of the nitride semiconductor crystal produced by the method of the present invention. In order to clarify this, the PL principle and the dependence of the internal quantum efficiency on the excitation light intensity are described below. Add a description.

  FIG. 15 is a band diagram for explaining radiative recombination and non-radiative recombination occurring in a semiconductor crystal. As shown in this figure, the electron-hole pair generated by excitation with light of energy hν is directly recombined between the lower end of the conduction band and the upper end of the valence band to emit energy as light. Or recombination via a defect level or the like formed in the forbidden band to cause non-radiative recombination that emits energy other than light (mainly as heat) and disappears.

  The “emission recombination lifetime τ (R)” in such a process is defined as the time until the total number of generated carriers is reduced to 1 / e of the initial time when the carriers generated by the excitation light are recombined by luminescence. Similarly, the time until the total number of generated carriers is reduced to 1 / e of the initial value due to non-radiative recombination is defined as “non-radiative recombination lifetime τ (NR)”.

The internal quantum efficiency η is expressed by η = 1 / (1 + τ (R) / τ (NR)), and τ (R) is short and τ (NR)
Is longer, the internal quantum efficiency η is higher.

  The above-described time-resolved PL method is a method capable of observing the transient response of light emission of electron-hole pairs generated by light pulse excitation. Assuming that τ (PL) is the time required for PL light emission to decay to 1 / e of the maximum intensity, τ (PL) is determined by the competition between luminescent recombination and non-radiative recombination, and 1 / τ (PL) = 1 / τ (R) + 1 / τ (NR).

  When the crystallinity of the sample is poor and the amount of crystal defects is large, non-radiative recombination among the recombination tends to occur. For this reason, τ (NR) is shortened, and as a result, τ (PL) is shortened.

  On the other hand, even if the crystallinity of the sample is good and the contribution of non-radiative recombination is small, τ (PL) is short when τ (R) is short. When the crystallinity of the sample is good and the amount of crystal defects is small, τ (NR) becomes long and consequently τ (PL) becomes long.

  Here, since the non-radiative recombination process is a thermal excitation type, the higher the temperature, the greater the contribution of the non-radiative recombination process. Therefore, when the sample temperature is measured at a very low temperature, the contribution (influence) of the non-radiative recombination process is almost negligible. For these reasons, when experimentally determining the internal quantum efficiency η, the relative value of the emission intensity at room temperature when the emission intensity at an extremely low temperature is 1 is often used. Furthermore, the contribution of the non-radiative recombination process increases as the density of generated carriers decreases. In other words, the non-radiative recombination process saturates as the carrier density increases, and the contribution of the radiative recombination process relatively increases, so the internal quantum efficiency η tends to increase apparently.

FIG. 16 is a diagram showing a result of an experimental investigation of the excitation energy density dependence of the internal quantum efficiency of the multiple quantum well structure when excited by an optical pulse. From the results shown in this figure, when the excitation energy density is as low as about 500 nJ / cm ² , the internal quantum efficiency is as small as about several percent. However, even in the same sample, the internal quantum efficiency reaches 90% or more in the case of a high excitation energy density exceeding 300000 nJ / cm ² .

This result means that in order to correctly evaluate the superiority or inferiority of the crystallinity between samples by the PL method, it is necessary to carry out under a low excitation energy density condition in which the influence of the non-radiative recombination process is strongly reflected. . That is, the excitation energy density at the time of evaluating by the PL method is a low excitation energy density of about 2000 nJ / cm ² or less, and the excess carrier density generated by light is 10 ¹⁷ (cm ⁻³ ) from the second half of 10 ¹⁶ (cm ⁻³ ). ^-3 ) It is necessary to make it about the first half. In the conventional reports, the photoexcited excess carrier density generated under the high excitation energy density condition, that is, essentially in the active layer structure is not excessively high. There are many examples of PL evaluation under the conditions. However, even if the internal quantum efficiency is obtained under such conditions, a high value of “apparent” that does not reflect the crystal quality is obtained, and the crystallinity is not correctly evaluated.

Therefore, the PL measurement under the low excitation density condition in this specification means that the nitride semiconductor formed on the nitride main surface of the substrate that is a nonpolar nitride is at least an optically active layer structure. In this active layer structure, the excess carrier density generated by light is changed from the second half of 10 ¹⁶ (cm ⁻³ ) to the first half of 10 ¹⁷ (cm ⁻³ ) in this active layer structure. It is what you point to. Furthermore, when the active layer structure is a multi-quantum well active layer, the average photoexcited excess carrier density in all quantum well layers is from 10 ¹⁶ (cm ⁻³ ) to 10 ¹⁷ (cm ⁻³ ). This means that the PL measurement is performed in the first half.

  Therefore, when the structure of the object to be measured changes, the photoexcitation energy density or photoexcitation power density is appropriately changed, and in the quantum well active layer structure or in the quantum well active layer structure. It is necessary for the correct crystallinity evaluation to reduce the photoexcited excess carrier density.

In this example, PL measurement is performed under low excitation energy density conditions. In cw-PL measurement, 58% (I (300K) / I (13K) = 58%) at an excitation power density of 150 W / cm ² , In the time-resolved PL measurement, the excitation energy density is 1400 nJ / cm ² and is 93% (I (300K) / I (4K) = 93%), and the excitation energy density is 510 nJ / cm ² and 60% (I (300K) / I (4K). ) = 60%) means that the crystallinity is much higher than that of a conventional nitride semiconductor crystal.

  This embodiment is an example in which a near-ultraviolet LED is fabricated by laminating and growing a gallium nitride-based semiconductor thin film by MOCVD. An outline of a series of crystal growth processes is shown in FIG. The grown layer structure is schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.0 mm in the c-axis direction and 15 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 7.0 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 37.5 arcsec, the OFF angle in the c (+) direction is −0.28 °, and the OFF angle in the a direction is 0.03 °. Met. The dislocation density was 5.0 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

In the temperature raising step (step A), first, as the first temperature raising step t _B , the temperature is raised while supplying H ₂ at 10 L / min as the main flow into the furnace, and the temperature of the substrate reaches 400 ° C. It was initiated the second heating step t _a. There, NH ₃ was supplied at 7.5 L / min as a gas constituting the first main flow, and H ₂ was supplied at 12.5 L / min as a gas constituting the second main flow.

Thereafter, the substrate temperature was further raised to 1000 ° C. while increasing NH ₃ and H ₂ to 10 L / min and 30 L / min, respectively. At this time, the sub-flow was N ₂ gas 20 L / min, and the growth non-growth gas such as for purging was N ₂ for a total of 19 L / min. In the second temperature raising step, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0 throughout the temperature raising step.

In the first growth step (step B), NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. Further, as gas constituting the second main flow, TM 2 (concentration of 100%) using H ₂ (40 L / min) and H ₂ (0.5 L / min) constituting a part of the main flow as a carrier gas is used. 0.0055 L / min), SiH ₄ (concentration of 100) using H ₂ (0.2 L / min) constituting a part of the main flow as a carrier gas and H ₂ (0.06 L / min) as a diluent gas. %, 6 × 10 ⁻⁷ L / min). At this time, Fp was 0.0. With this main flow gas supply, a Si-doped GaN layer (first nitride semiconductor layer) was grown to a thickness of 40 nm.

In the second growth step (step C), NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. The gas constituting the second main flow is H ₂ (40 L /
Min), TMGa (0.0055 L / min as 100% concentration) using H ₂ (0.5 L / min) constituting a part of the main flow as a carrier gas, and H ₂ (0 SiH ₄ (6 × 10 ⁻⁷ L / min with a concentration of 100%) was supplied with a carrier gas of .2 L / min) and a diluent gas of H ₂ (0.06 L / min). At this time, Fp was 0.0. With such main flow gas supply, a Si-doped GaN layer (second nitride semiconductor layer) was grown to a thickness of 3.96 μm.

  Step D in the third step was performed under the same conditions as in Example 9 except that the MQW cycle number was set to 8.

  The process E in the third process was performed under the same conditions as in Example 9.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  Some irregularities were confirmed on the surface of the substrate thus fabricated. When the surface was measured with a contact-type profilometer, Ra = 244 nm, and the roughness that was expected to exhibit a scattering function that easily interacted with the emission wavelength was naturally formed. Further, in the PL characteristics evaluated by excitation with laser light having a wavelength of 325 nm, the peak wavelength was 422 nm, and the integrated intensity was 10 as a relative value.

  Table 4 shows the average thickness of each quantum well layer (w1 to w8) of the comparative sample obtained as a result of observing the HAADF-STEM image of the above sample by the same procedure as in Example 8 and Example 9. nm), w1 is the average thickness of the lowermost (substrate side) quantum well layer, and w8 is the average thickness of the uppermost (surface side) quantum well layer. In addition, the standard deviation and coefficient of variation between pixels in each layer are also shown.

[Comparative Example 1]

  This comparative example is an example in which a gallium nitride based semiconductor thin film is laminated and grown by MOCVD to produce a near-ultraviolet LED, and a series of crystal growth processes is schematically shown in FIG. The grown layer structure is schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 3.9 mm in the c-axis direction and 15 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 7.4 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 36.5 arcsec, the OFF angle in the c (+) direction is -0.31 °, and the OFF angle in the a direction is 0.02 °. Met. The dislocation density was 5.2 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

  The temperature raising step (step A) was performed in the same manner as in Example 12.

Then, after the substrate temperature reached 1000 ° C., NH ₃ and H ₂ were continuously supplied at 10 L / min and 30 L / min, respectively, and in this state, the substrate was waited for 15 minutes to clean the substrate surface. At this time, the sub-flow was N ₂ gas 20 L / min, and the growth non-growth gas such as for purging was N ₂ for a total of 19 L / min.
At this time, Fp = 0.

In the first growth step (step B), NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. Further, as gas constituting the second main flow, TM 2 (concentration of 100%) using H ₂ (40 L / min) and H ₂ (0.5 L / min) constituting a part of the main flow as a carrier gas is used. 0.0055 L / min), SiH ₄ (concentration of 100) using H ₂ (0.2 L / min) constituting a part of the main flow as a carrier gas and H ₂ (0.06 L / min) as a diluent gas. %, 6 × 10 ⁻⁷ L / min). At this time, Fp was 0.0. The Si-doped GaN layer (first nitride semiconductor layer) was grown with a thickness of 5 nm by such main flow gas supply.

In the second growth step (step C), NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. The gas constituting the second main flow is H ₂ (40 L /
Min), TMGa (0.0055 L / min as 100% concentration) using H ₂ (0.5 L / min) constituting a part of the main flow as a carrier gas, and H ₂ (0 SiH ₄ (6 × 10 ⁻⁷ L / min with a concentration of 100%) was supplied with a carrier gas of .2 L / min) and a diluent gas of H ₂ (0.06 L / min). At this time, Fp was 0.0. With such main flow gas supply, a Si-doped GaN layer (second nitride semiconductor layer) was grown to a thickness of 45 nm.

  Step D in the third step was performed in the same manner as in Example 12.

  Step E in the third step was performed in the same manner as in Example 12.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  As a result, the Ra of the present example is 85 nm and the formed unevenness is small, it is difficult to interact with light in the general near-ultraviolet and visible regions, and the light extraction effect when used as an LED cannot be expected so much. It was a thing. Moreover, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength was 400 nm. Further, the PL integral intensity was as extremely low as 1.

  Table 5 shows the HAADF-STEM images of the above samples in the same procedure as in Example 8, Example 9, and Example 12, and the respective quantum well layers (w1 to w8) of this comparative sample obtained as a result. The average thickness (nm) is summarized, where w1 is the average thickness of the lowermost (substrate side) quantum well layer and w8 is the average thickness of the uppermost (surface side) quantum well layer. Moreover, the standard deviation and variation coefficient of the thickness of each layer are also shown simultaneously.

  From the standard deviation of the thickness of the quantum well layer and the CV value of Example 8, Example 9, Example 12, and Comparative Example 1, when Fp is small in Steps B and C, the standard deviation of the thickness in the Y-axis direction Obviously, and the coefficient of variation can be increased. The standard deviation of the thickness in the Y-axis direction and the average value of the coefficient of variation for each example are summarized in Table 6 below.

  From Table 6, in the sample of Example 8, when the average value is seen, the standard deviation of the thickness is 0.28 and the coefficient of variation is 0.076, and in the samples of Example 9, Example 12, and Comparative Example 1, In comparison, the film thickness fluctuation is remarkably small. Thus, when the standard deviation of thickness and a variation coefficient are small, it is preferable from a viewpoint of internal quantum efficiency. This is also clear from Table 7 described later. Therefore, in order to relatively increase the internal quantum efficiency of the quantum well active layer structure, it is preferable that the standard deviation is 0.45 nm or less and the variation coefficient is 0.1 or less.

  In this example, the light emission characteristics of a multiple quantum well (MQW) layer manufactured by various crystal growth methods were evaluated by a photoluminescence (PL) method.

  FIG. 3D is a diagram showing a stacked structure of a sample used in this example, and FIG. 2I is a diagram for explaining a sequence of a crystal growth method of this sample. In this sample, an InGaN / GaN multiple quantum well active layer structure 13 having a period number of 5 is provided in the uppermost layer.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.1 mm in the c-axis direction and 18 mm in the a-axis direction. The substrate was doped with Si, and the Si concentration was 6.2 × 10 ¹⁷ cm ⁻³ . The electrical characteristics of the substrate were n-type, and the carrier density was 6.5 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 32.2 arcsec, and the OFF angle in the c (+) direction is −
The OFF angle in the a direction was 0.12 ° and 0.02 °. The dislocation density was 5.0 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reactor under normal pressure growth conditions to produce a quantum well structure.

  The pressure in the reactor was 100 ± 2 kPa in all steps.

  The temperature raising step (step A) was the same as in Example 9.

  The first growth step (step B) was performed in the same manner as in Example 9.

  The second growth step (step C) was performed in the same manner as in Example 9 except that the thickness of the second semiconductor layer (Si-doped n-type GaN layer) was 2.5 μm.

  Step D in the third step was performed in the same manner as in Example 9.

  Step E in the third step was not performed.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  When the surface of the substrate thus prepared was measured with a contact-type step meter, Ra = 185 nm, and the roughness that is expected to exhibit a scattering function that easily interacts with the emission wavelength is naturally formed. It was. Further, in the PL characteristics evaluated by excitation with laser light having a wavelength of 325 nm at room temperature, the peak wavelength was 397 nm and the integrated intensity was 37 as a relative value. The internal quantum efficiency and PL emission lifetime of such a sample were evaluated as follows.

  The PL characteristics of the above structure were evaluated. The setup of the optical system used in the evaluation is the same as that described in Example 17. First, the internal quantum efficiency was evaluated by performing CW-PL measurement and time-resolved PL measurement when irradiated with pulsed light.

The optical system is as shown in FIG. Similar to Example 17, the sample temperature was in the range of 13 K to 300 K, and the laser power on the sample surface was 150 W / cm ² per unit area. The photoexcited steady excess carrier density in this condition is estimated to be in the range of 1 × 10 ¹⁶ cm ⁻³ to 1 × 10 ¹⁷ cm ⁻³ .

  As a result of this CW-PL measurement, when the integrated intensity I (13K) obtained by integrating the PL intensity from the multiple quantum well at 13K with the emission energy is 1, the same integrated intensity I (300K) at 300K is I (13K). Of 39%.

  Next, time-resolved PL measurement (without a polarizing filter) was performed with the setup shown in FIG.

The pulse repetition frequency was 1 kHz and the pulse width was 120 fs. The wavelength of the tunable pulse laser here was 370 nm, and the pulse energy density per unit area was 510 nJ / cm ² .

As a result of this time-resolved PL measurement, the PL intensity is obtained by time-integrating the transient response of PL emission from the multiple quantum well at 4K, and the integrated intensity I (4K) obtained by integrating this with the emission energy.
Assuming that 1 is 1, the same integrated intensity I (300K) at 300K was 27% of I (4K).

Further, at 300 K, the pulse energy density per unit area is 510 nJ / cm.
^{When the} PL life is obtained from the transient response (decay curve) with respect to time of the PL intensity after pulse excitation obtained by time-resolved PL measurement (without a polarizing filter) set to 2, the PL life τ (PL) is 1.8 nsec. there were. The luminescence recombination lifetime τ (R) was 6.7 ns, and the non-radiative recombination lifetime τ (NR) was 2.4 ns. As a result, τ (R)> τ (NR).

  In this example, the light emission characteristics of a multiple quantum well (MQW) layer manufactured by various crystal growth methods were evaluated by a photoluminescence (PL) method.

  FIG. 3F is a diagram showing a stacked structure of a sample used in this example, and FIG. 2J is a diagram for explaining a sequence of a crystal growth method of this sample. In this sample, an InGaN / GaN multiple quantum well active layer structure 13 having a period number of 5 is provided in the uppermost layer.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.5 mm in the c-axis direction and 20 mm in the a-axis direction. The substrate was doped with Si, and the Si concentration was 6.2 × 10 ¹⁷ cm ⁻³ . The electrical characteristics of the substrate were n-type, and the carrier density was 6.5 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the half-value width of the rocking curve in (10-12) reflection is 35 arcsec, the OFF angle in the c (+) direction is -0.095 °, and the OFF angle in the a direction is 0.03 °. It was. The dislocation density was 5.0 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reactor under normal pressure growth conditions to produce a quantum well structure.

  The pressure in the reactor was 100 ± 2 kPa in all steps.

  The temperature raising step (step A) was performed in the same manner as in Example 12.

  The first growth step (step B) was performed in the same manner as in Example 12.

The second growth step (Step C) was performed in the same manner as in Example 12 except that the thickness of the second semiconductor layer (Si-doped n-type GaN layer) was 2.46 μm.

  Step D in the third step was performed in the same manner as in Example 12 except that the cycle of the MQW layer was set to 5 cycles.

  Step E in the third step was not performed.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  When the surface of the substrate thus fabricated was measured with a contact-type step meter, Ra = 230 nm, and the roughness that is expected to easily exhibit interaction with the emission wavelength and to exhibit the scattering function is naturally formed. It was. Moreover, in the PL characteristic evaluated by laser beam excitation at a wavelength of 325 nm at room temperature, the peak wavelength was 455 nm and the integrated intensity was 6 in relative value.

  The PL characteristics of the above structure were evaluated. The setup of the optical system used in the evaluation is the same as that described in Example 17. First, the internal quantum efficiency was evaluated by performing CW-PL measurement and time-resolved PL measurement when irradiated with pulsed light.

The optical system shown in FIG. 12 was used. Similar to Example 16, the sample temperature was in the range of 13K to 300K, and the laser power on the sample surface was 150 W / cm ² per unit area.
The photoexcited steady excess carrier density in this condition is estimated to be in the range of 1 × 10 ¹⁶ cm ⁻³ to 1 × 10 ¹⁷ cm ⁻³ .

  As a result of this CW-PL measurement, when the integrated intensity I (13K) obtained by integrating the PL intensity from the multiple quantum well at 13K with the emission energy is 1, the same integrated intensity I (300K) at 300K is I (13K). 46%.

  Next, time-resolved PL measurement (without a polarizing filter) was performed with the setup shown in FIG.

The pulse repetition frequency was 1 kHz and the pulse width was 120 fs. The wavelength of the tunable pulse laser here was 370 nm, and the pulse energy density per unit area was 510 nJ / cm ² .

  As a result of this time-resolved PL measurement, the transient response of PL emission from a multiple quantum well at 4K is time-integrated to obtain the PL intensity. A similar integrated intensity I (300K) was 36% of I (4K).

Further, at 300 K, the pulse energy density per unit area is 510 nJ / cm.
^{When the} PL lifetime is obtained from the transient response (decay curve) with respect to time of the PL intensity after pulse excitation obtained by time-resolved PL measurement (without a polarizing filter) set to 2, the PL lifetime τ (PL) is 1.7 nsec. there were. The luminescence recombination lifetime τ (R) was 4.8 ns, and the non-radiative recombination lifetime τ (NR) was 2.7 ns. As a result, τ (R)> τ (NR).
[Comparative Example 2]

  In this comparative example, the light emission characteristics of a multiple quantum well (MQW) layer manufactured by various crystal growth techniques were evaluated by a photoluminescence (PL) method.

  FIG. 3 (F) is a diagram showing a stacked structure of the sample used in this comparative example, and FIG. 2 (K) is a diagram for explaining the sequence of the crystal growth method of this sample. In this sample, the Mg doped layer is not provided on the MQW structure, and the InGaN / GaN multiple quantum well active layer structure 13 having a period number of 5 is provided in the uppermost layer.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.0 mm in the c-axis direction and 19 mm in the a-axis direction. The substrate was doped with Si, and the Si concentration was 6.2 × 10 ¹⁷ cm ⁻³ . The electrical characteristics of the substrate were n-type, and the carrier density was 6.5 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 31 arcsec, the OFF angle in the c (+) direction is -0.085 °, and the OFF angle in the a direction is 0.05 °. It was. The dislocation density was 4.9 × 10 ⁶ cm ⁻² .

This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reactor under normal pressure growth conditions to produce a quantum well structure. Note that the pressure in the reactor is 10 in all steps.
It was set to 0 ± 2 kPa. The temperature raising step (step A) was performed in the same manner as in Comparative Example 1.

In the cleaning process, as in Comparative Example 1, after the substrate temperature reached 1000 ° C., NH ₃ and H ₂ were continuously supplied at 10 L / min and 30 L / min, respectively, and in this state, the substrate was on standby for 15 minutes. Cleaning was performed. At this time, the sub-flow was N ₂ gas 20 L / min, and the growth non-growth gas such as for purging was N ₂ for a total of 19 L / min. At this time, Fp = 0.0.

  The first growth step (Step B) was performed in the same manner as in Comparative Example 1.

  The second growth step (Step C) was performed in the same manner as Comparative Example 1.

  Step D in the third step was performed in the same manner as Comparative Example 1 except that the cycle of MQW was 5 cycles.

Step E in the third step was not performed.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

When the surface of the substrate thus prepared was measured with a contact-type step gauge, Ra = 78 nm and the formed irregularities were small, and it was difficult to interact with light up to the general near ultraviolet and visible range. The light extraction effect at the time was not expected. The PL characteristics of this substrate evaluated by laser excitation with a wavelength of 325 nm at room temperature showed a very low PL intensity with a peak wavelength of 433 nm and an integrated intensity of 3 as a relative value.

  The PL characteristics of the above structure were evaluated. The setup of the optical system used in the evaluation is the same as that described in Example 16. First, the internal quantum efficiency was evaluated by performing CW-PL measurement and time-resolved PL measurement when irradiated with pulsed light.

The optical system shown in FIG. 12 was used. Similar to Example 11, the sample temperature was in the range of 13K to 300K, and the laser power on the sample surface was 150 W / cm ² per unit area.
The photoexcited steady excess carrier density in this condition is estimated to be in the range of 1 × 10 ¹⁶ cm ⁻³ to 1 × 10 ¹⁷ cm ⁻³ .

  As a result of this CW-PL measurement, when the integrated intensity I (13K) obtained by integrating the PL intensity from the multiple quantum well at 13K with the emission energy is 1, the same integrated intensity I (300K) at 300K is I (13K). Of 10%.

  Next, time-resolved PL measurement (without a polarizing filter) was performed with the setup shown in FIG.

The pulse repetition frequency was 1 kHz and the pulse width was 120 fs. The wavelength of the tunable pulse laser here was 370 nm, and the pulse energy density per unit area was 510 nJ / cm ² .

As a result of this time-resolved PL measurement, the transient response of PL emission from a multiple quantum well at 4K is time-integrated to obtain the PL intensity, and further, the integrated intensity I (4K) obtained by integrating this with the emission energy is set to 1, and at 300K Similar integrated intensity I (300K) was 16% of I (4K).

Further, at 300 K, the pulse energy density per unit area is 510 nJ / cm ^2.
The PL lifetime τ (PL) was 0.66 nsec when the PL lifetime was determined from the transient response (decay curve) of the PL intensity after pulse excitation obtained by time-resolved PL measurement (without a polarizing filter). It was. The luminescence recombination lifetime τ (R) was 4.1 ns, and the non-radiative recombination lifetime τ (NR) was 0.78 ns. As a result, τ (R)> τ (NR).
[Reference Example 2]

  In Reference Example 2, the light emission characteristics of a multiple quantum well (MQW) layer fabricated on the conventionally used c-plane were evaluated by a photoluminescence (PL) method.

  FIG. 3D is a diagram showing a stacked structure of a sample used in this reference example, and FIG. 2L is a diagram for explaining a sequence of a crystal growth method of this sample. In this sample, the Mg doped layer is not provided on the MQW structure, and the InGaN / GaN multiple quantum well active layer structure 13 having a period number of 5 is provided in the uppermost layer. Note that the substrate used in the stacked structure illustrated in FIG. 3D is also denoted by reference numeral 10 in the same manner as the GaN substrate having the m-plane as the main surface.

As the substrate, a (0001) (c-plane) oriented GaN free-standing substrate 10 was used. The substrate size was 48 mm in diameter. The substrate was Si-doped and its Si concentration was 1.5 × 10 ¹⁷ cm ⁻³ . The electrical characteristics of the substrate were n-type, and the carrier density was 1.5 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection was 45 arcsec, the OFF angle in the m direction was 0.205 °, and the OFF angle in the a direction was 0.310 °. The dislocation density was 7.0 × 10 ⁶ cm ⁻² .

This c-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reactor under normal pressure growth conditions to produce a quantum well structure.
The temperature raising step (step A) was the same as in Example 13.

The first growth step (step B) was performed in the same manner as in Example 13.

The second growth step (step C) was performed in the same manner as in Example 13 except that the thickness of the second semiconductor layer (Si-doped n-type GaN layer) was 4 μm.

  Step D in the third step was performed in the same manner as in Example 13 except that the temperature of the substrate was 780 ° C.

  Step E in the third step was not performed.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated. Note that, according to the study by the present inventors, the growth by such a sequence is the crystal growth method showing the best characteristics in the growth on the c-plane GaN substrate.

  When the surface of the substrate thus fabricated was measured with a contact-type step meter, Ra = 12 nm, which was an extremely flat surface. Further, in the PL characteristics evaluated by excitation with laser light having a wavelength of 325 nm at room temperature, the peak wavelength was 405 nm, and the integrated intensity was 16 as a relative value, indicating a low PL intensity.

  The PL characteristics of the above structure were evaluated. The setup of the optical system used in the evaluation is the same as that described in Example 11. First, the internal quantum efficiency was evaluated by performing CW-PL measurement and time-resolved PL measurement when irradiated with pulsed light.

The optical system shown in FIG. 12 was used. Similar to Example 16, the sample temperature was in the range of 13K to 300K, and the laser power on the sample surface was 150 W / cm ² per unit area.
The photoexcited steady excess carrier density in this condition is estimated to be in the range of 1 × 10 ¹⁶ cm ⁻³ to 1 × 10 ¹⁷ cm ⁻³ .

  As a result of this CW-PL measurement, when the integrated intensity I (13K) obtained by integrating the PL intensity from the multiple quantum well at 13K with the emission energy is 1, the same integrated intensity I (300K) at 300K is I (13K). Of 9%.

  Next, time-resolved PL measurement (without a polarizing filter) was performed with the setup shown in FIG.

The pulse repetition frequency was 1 kHz and the pulse width was 120 fs. The wavelength of the tunable pulse laser here was 370 nm, and the pulse energy density per unit area was 510 nJ / cm ² .

  As a result of this time-resolved PL measurement, the transient response of PL emission from a multiple quantum well at 4K is time-integrated to obtain the PL intensity, and further, the integrated intensity I (4K) obtained by integrating this with the emission energy is set to 1, and at 300K A similar integral intensity I (300K) was 5% of I (4K).

In addition, at 300 K, the PL life from the transient response (attenuation curve) of PL intensity after pulse excitation obtained by time-resolved PL measurement (without polarization filter) with a pulse energy density per unit area of 510 nJ / cm ^2. The PL life τ (PL) was 1.2 nsec. The luminescence recombination lifetime τ (R) was 21.9 ns, and the non-radiative recombination lifetime τ (NR) was 1.3 ns. As a result, τ (R)> τ (NR).

  The measurement results of the PL samples (MQW) of Example 11, Example 13, Example 14, Comparative Example 2, and Reference Example 2 are summarized in Table 7 below. In addition, in Example 11 and Example 8, Example 13 and Example 9, Example 14 and Example 12, and Comparative Example 2 and Comparative Example 1, the former reaches the quantum well active layer structure. The latter relates to a sample in which an LED structure is produced including the other conductivity type layer, that is, a layer grown by doping Mg. Here, the sample with the quantum well active layer structure part and the sample with the LED structure made in the above description are basically similar, although there are differences in the total number of quantum well layers, etc. It is a sample manufactured under the growth conditions.

From the results shown in Table 7, in Examples 11, 13, and 14, the internal quantum efficiencies were 58%, 39%, and 46% in CW measurement, and 60%, 27%, and 36% in pulse measurement, respectively. On the other hand, Comparative Example 2 and Reference Example 2 were as low as 10% and 9%, respectively. These show that the portion related to the start of growth in the crystal growth sequence of the present invention greatly affects the characteristics of the active layer formed thereafter. In order to improve the characteristics of the LED, the higher the internal quantum efficiency, the better. If it is 20% or more, an LED having good characteristics can be realized. Similarly, the PL life (τ (PL)) of Examples 11, 13, and 14 is extremely longer than that of Comparative Example 2 and Reference Example 2. This is also a result of reflecting good optical quality, and the PL lifetime is desirably 1.5 nanoseconds (ns) or longer.

  In the m-plane examples and comparative examples, except for Example 13, τ (R) is about 4 ns, which is not very different, but c-plane τ (R) is very long. This is because electrons and holes contributing to light emission are spatially separated by the QCSE effect, which is disadvantageous in terms of light emission efficiency. That is, these are results showing that radiation is efficiently realized by using a non-polar m-plane for the substrate. That is, in order to obtain high luminous efficiency in the case of a light emitting device, it means that it is effective to grow a nitride semiconductor crystal on the nonpolar nitride main surface.

  Furthermore, in the m-plane examples and comparative examples, τ (NR) of Example 11 is extremely long. This indicates that there are few carriers that do not contribute to light emission and disappear. Further, when looking at the magnitude relationship between τ (R) and τ (NR) in each example and comparative example, it can be seen that only Example 16 satisfies τ (R) <τ (NR). This has shown that the produced | generated carrier has contributed to light emission effectively. That is, in order to realize a quantum well active layer structure with high internal quantum efficiency, it is desirable that τ (R) <τ (NR).

  Further, here, the surface roughness of the samples of Examples, Comparative Examples, Reference Examples, etc., which has been described so far, was measured with a contact-type step meter, and the center line average roughness (Ra) was obtained, and the wavelength of 325 nm The relative values of the integrated intensities of the PL characteristics evaluated by excitation with the laser beam are summarized.

  Example 8 and Example 11, Example 9 and Example 13, Example 12 and Example 14, and Comparative Example 1 and Comparative Example 2 shown here are the other conductive type layers. That is, the present invention relates to a sample in which an LED structure is manufactured including a layer grown by doping Mg, and the latter relates to a sample in which a quantum well active layer structure is manufactured. Here, the sample in which the LED structure is manufactured and the sample in which the quantum well active layer structure part is manufactured are “paired” in the above description, although there are differences in the total number of quantum well layers, etc. This is a sample produced under similar growth conditions. Moreover, the result of a reference example is also shown partially in part.

  Table 9 summarizes the results of epitaxial growth of the LED structure, and Table 10 summarizes the results of growth up to the quantum well active layer structure.

  Here, among the results of Table 9 and Table 10, each result is interpreted as follows.

  The results shown in Example 8 and Example 11 show that when the internal quantum efficiency of the quantum well active layer structure portion is relatively high, the surface flatness of the epitaxial layer is excellent, so that the nitride semiconductor layer It is understood that it is difficult to form appropriate irregularities on the surface of the film in a self-forming and intentional manner, and as a result, Ra is relatively small.

In general, as shown in FIG. 16, the internal quantum efficiency in the quantum well structure increases as the excitation density increases, that is, as the excess carrier density generated or injected into the quantum well increases, and the luminescence recombination efficiency increases. improves. Although the high internal quantum efficiency in the weakly excited state indicates that the crystal quality is remarkably superior, the current injection environment in which the LED is actually driven far exceeds 1 × 10 ²⁰ (cm ⁻³ ). Assuming that the injected carrier density is very high, the fact that the internal quantum efficiency under the low excitation density condition is remarkably excellent can be said to have a moderate influence on the improvement of the light emission efficiency of the LED. Therefore, it is considered that the necessary characteristics that the quantum well structure and the epitaxial layer should have ensure crystal quality that is at least the minimum.

  Thus, for example, when a reflective p-side electrode is formed on the surface of this epitaxial layer to produce a flip chip type LED or the like, or one electrode on the side of the nitride nonpolar surface substrate that does not have an epitaxial layer, A vertical conduction type LED having the other electrode on the side having the epitaxial layer of the nonpolar plane substrate and having a portion in which current injection is performed substantially perpendicularly to the laminated surface of the active layer structure is manufactured. In some cases, the internal quantum efficiency or PL intensity is high, so the effect is clearly recognized. However, the fact that the internal quantum efficiency is remarkably superior has a moderate effect on the luminous efficiency improvement when the epitaxial layer is converted to an LED, and the degree of the influence on the luminous efficiency of the LED is limited. Conceivable.

  On the other hand, from the viewpoint of light extraction, it is as follows. Light contained within the critical angle of light determined by the difference in refractive index between the surrounding medium and the medium containing the light-emitting portion can be extracted, but the light extraction efficiency is also low because there are few effects such as scattering on a flat surface. It is considered moderate.

  Therefore, it is considered that the overall luminous efficiency when such a nitride semiconductor is turned into an LED is moderate.

  The results shown in Example 9 and Example 13 and the results shown in Example 12 and Example 14 indicate that the internal quantum efficiency of the quantum well active layer structure portion is kept in an appropriate range, while the nitride semiconductor layer surface is maintained. It is interpreted that moderate irregularities are the result of being able to form intentionally and intentionally.

  When Table 7, Table 9, and Table 10 are viewed comprehensively, the results of Example 9, Example 12, Example 13, and Example 14 are on the nonpolar surface in terms of internal quantum efficiency or PL intensity. Compared with the formed comparative examples 1 and 2, sufficiently good values are shown. Also, it is good compared with the results on the polar surface shown in Reference Example 2. Therefore, it is understood that the crystal quality of the epitaxial layer which is required at the time of manufacturing the LED is sufficiently ensured. For example, looking at the internal quantum efficiency in Table 7, that of Example 14 is larger than that of Comparative Example 2 and Reference Example 2.

  Thus, for example, when a reflective p-side electrode is formed on the surface of this epitaxial layer to produce a flip chip type LED or the like, or one electrode on the side of the nitride nonpolar surface substrate that does not have an epitaxial layer, A vertical conduction type LED having the other electrode on the side having the epitaxial layer of the nonpolar plane substrate and having a portion in which current injection is performed substantially perpendicularly to the laminated surface of the active layer structure is manufactured. In some cases, it can be said that such a quantum well layer has sufficient quality.

  On the other hand, from the viewpoint of light extraction, it is as follows. In the nitride semiconductor layer surfaces of Example 9, Example 12, Example 13, and Example 14, the degree of unevenness is usually in the range of about 150 nm or more and 250 nm or less, and up to the general near ultraviolet and visible range. It is a rough surface that interacts well with the wavelength of light and can exhibit a scattering function. In particular, it exhibits a high scattering effect because its physical size is close to near-ultraviolet, which is a wavelength region that can be preferably used in the present invention, specifically, light having a wavelength from about 370 nm to about 430 nm. Things are expected and preferable.

  Therefore, it is considered that the overall luminous efficiency when such a nitride semiconductor is used as an LED is high.

  Finally, the results shown in Comparative Example 1 and Comparative Example 2 show that the second nitride semiconductor layer is as thin as 45 nm, so that the formation of irregularities is not sufficient, and the internal quantum efficiency and the PL intensity are low. It is interpreted as a result.

  Using such a nitride semiconductor, for example, when a reflective p-side electrode is formed on the surface of this epitaxial layer to produce a flip-chip type LED or the like, a transmissive p-side electrode is formed, and vertical conduction When manufacturing a type LED or the like, it is understood that the quality is not sufficient even if the excess carrier density generated or injected into the quantum well is high. Moreover, the unevenness | corrugation formed is small, it is hard to interact with the light to general ultraviolet, near ultraviolet, and visible region, and the light extraction effect at the time of setting it as LED cannot be expected so much.

  Therefore, it is considered that the overall luminous efficiency when such a nitride semiconductor is used as an LED is low.

  This example is an example in which a near-ultraviolet LED is fabricated by stacking and growing a gallium nitride-based semiconductor thin film by MOCVD, and the outline of a series of crystal growth processes is as shown in FIG. The grown layer structure is schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 3.8 mm in the c-axis direction and 22 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 6.5 × 10 ¹⁷ cm ⁻³ .

According to X-ray diffraction, the half-width of the rocking curve in (10-12) reflection is 34.2 arcsec, the OFF angle in the c (+) direction is 0.25 °, and the OFF angle in the a direction is 0.03 °. there were. Further, the dislocation density was 5.0 × 10 ⁶ cm ⁻² .

This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

  The temperature raising step (step A) was the same as in Example 1.

In the first growth step (step B), NH ₃ was supplied at 10 L / min as a gas constituting the first main flow. As the gas constituting the second main flow, N ₂ is 29.0 L / min, H ₂ is 1.0 L / min, and H ₂ (0.5 L / min) which is a part of the main flow is a carrier. TMGa (0.0018 L / min as 100% concentration) was supplied into the furnace as the gas. Further, H ₂ (0.5 L / min) which is a part of the main flow was used as a carrier gas, and Cp ₂ Mg (7 × 10 ⁻⁶ L / min as a concentration of 100%) was supplied into the furnace. With this main flow gas supply, an Mg-doped GaN layer (first nitride semiconductor layer) having a doping concentration of 2 × 10 ¹⁹ cm ⁻³ was grown to a thickness of 40 nm.

At this time, the subflow was a mixed gas (20.5 L / min) of NH ₃ (0.5 L / min) and N ₂ (20 L / min), and the growth outside gas such as for purging was 19 L / min for N ₂ . During the growth of the first nitride semiconductor layer, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0.72500.

  The second growth process (process C), the process D in the third process, and the process E in the third process were performed under the same conditions as in Example 1.

  The surface of the substrate thus produced was extremely flat. This surface was measured with a contact-type step meter, and the average roughness or centerline average roughness (Ra) serving as an index of the degree of unevenness was determined. As a result, Ra according to the present example was 4.9 nm. In addition, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength is 391 nm, the integrated intensity is 86 in relative value, and the standard deviation of the in-plane wavelength distribution is 0. It was as small as 8%.

  This example is an example in which a near-ultraviolet LED is manufactured by stacking and growing a gallium nitride-based semiconductor thin film by MOCVD, and the outline of a series of crystal growth processes is as shown in FIG. The grown layer structure is as schematically shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.0 mm in the c-axis direction and 21 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 6.4 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 41.0 arcsec, the OFF angle in the c (+) direction is 0.09 °, and the OFF angle in the a direction is 0.02 °. there were. The threading dislocation density was 5.0 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

  The temperature raising step (step A) was the same as in Example 4.

  In the subsequent first growth step (step B), second growth step (step C), step D in the third step, and step E in the third step, TMGa is changed to TEGa having the same flow rate. All the procedures were the same as in Example 4 except that.

After such a growth process was completed, the substrate temperature was lowered and the gas in the reaction furnace was completely replaced with N ₂ gas, and then the substrate was taken out and evaluated.

  The surface of the substrate thus produced was extremely flat. This surface was measured with a contact-type step meter, and the average roughness or centerline average roughness (Ra) serving as an index of the degree of unevenness was determined.

  As a result, Ra according to the present example was 3.0 nm. In addition, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength is 391 nm, the integrated intensity is 95 in relative value, and the standard deviation of the in-plane wavelength distribution is 0. It was as small as 4%.

  This example is an example in which only a single layer of a gallium nitride-based semiconductor thin film is grown by MOCVD, and is for accurately evaluating the growth temperature dependence of the flatness of an epitaxial layer. An outline of a series of crystal growth processes is as shown in FIG. The grown layer structure is schematically shown in FIG.

  As the substrate, two (1-100) plane (m-plane) oriented GaN free-standing substrates were used. These are designated as Sample X and Sample Y.

The substrate size used for producing Sample X was 4.1 mm in the c-axis direction and 12 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 6.6 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the half-width of the rocking curve in (10-12) reflection is 34.2 arcsec, the OFF angle in the c (+) direction is 0.25 °, and the OFF angle in the a direction is 0.03 °. there were. Further, the dislocation density was 5.0 × 10 ⁶ cm ⁻² .

The substrate size used for producing Sample Y was 4.1 mm in the c-axis direction and 12 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 6.6 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the half-width of the rocking curve in (10-12) reflection is 34.2 arcsec, the OFF angle in the c (+) direction is 0.25 °, and the OFF angle in the a direction is 0.03 °. there were. Further, the dislocation density was 5.0 × 10 ⁶ cm ⁻² .

  These m-plane GaN free-standing substrates were placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

  For sample X, the example was the same as the example except that the thickness of the first nitride semiconductor layer in the first growth step was set to 0.25 μm and the epitaxial growth was terminated only in the first growth step. Same as 1. For sample Y, the temperature reached in the temperature raising step was 920 ° C., the thickness of the first nitride semiconductor layer was 0.25 μm in the first growth step, and Example 1 was the same as Example 1 except that the epitaxial growth was terminated by only one growth step.

When the surfaces of Sample X and Sample Y produced in this way were observed with an optical microscope, both were extremely flat. In order to examine the surface flatness of both samples in detail, the surface morphology was evaluated by AFM (Atomic Force Microscope). The results are shown in FIGS. 17 (A) and 17 (B). 17A is an AFM image of sample X, and FIG. 17B is an AFM image of sample Y. As is clear from these AFM images, the sample Y is superior in surface flatness. The value of Ra (center line average roughness), which is an index of the degree of unevenness, obtained by AFM is very small at 0.093 nm for sample Y, while that for sample X is medium at 1.303. Further, the value of Rmax (maximum height) was as small as 0.851 nm in the sample Y while the sample X was an intermediate value of 7.469 nm.

In this example, the light emission of the multiple quantum well active layer structure in the epitaxial layer before and after performing the activation process after the temperature lowering process by thermal annealing is performed for the LED structure that emits near ultraviolet light produced by the crystal growth method of the present invention. Comparison was made by PL measurement. Specifically, after measuring the PL immediately after the completion of the third step, and after the temperature lowering step of the epitaxial wafer formed on the nonpolar surface, in order to allow a part of the doped Mg atoms to act as an acceptor. The activation process was performed by thermal annealing, and then PL measurement was performed. The annealing conditions here were 5 minutes at 700 ° C. in an N ₂ atmosphere.

  The number of samples used here was three, and the growth conditions were as in Example 1 except for the temperature rise at step A, the temperatures at step B and step C, and the thickness of the InGaN quantum well. The surface morphology of these samples was all very flat. Table 8 below shows the temperature rise arrival temperature in step A, the growth temperature in step B and step C, and the thickness of the InGaN quantum well for each sample.

  FIG. 18 shows PL spectra after the thermal activation of the activation process after the temperature lowering process of these samples X, Y, and Z by thermal annealing. Note that the PL intensity of samples X, Y, and Z is lowered by annealing. The decrease in PL intensity induced by thermal annealing treatment is considered to include both effects of band structure change and crystallinity deterioration in the vicinity of the active layer structure caused by Mg activation. It is done. In this case, since the effect of the former is comparable, it can be considered that the degree of decrease in the present PL intensity reflects the latter effect more strongly. Here, the degree is small in the order of Z, Y, and X. In particular, in sample X, PL emission from the quantum well active layer is significantly reduced after annealing, and instead, band edge emission in the GaN bulk layer near 365 nm is conspicuous, whereas in samples Y and Z, PL spectrum is observed. Good shape is maintained.

  Usually, the optimal growth temperature for thick film growth of i-GaN or n-GaN on c-plane GaN crystal is around 1000 ° C., and crystallinity is low when crystal growth is performed at a temperature lower than that. In addition, the emission intensity in the active layer structure formed thereon is also reduced. However, as described above, when a crystal is grown on a substrate having a nonpolar main surface, a thick film crystal is grown at a temperature (920 ° C.) lower than the general condition by 100 ° C. or more. In addition to the significant improvement in emission intensity, it is possible to suppress the decrease in PL intensity due to annealing. In addition, by the method of the present invention, the first nitride semiconductor layer, the second nitride semiconductor layer, the quantum well active layer structure, and the p-type dopant that are extremely flat on the nonpolar plane can be formed. It is possible to increase the thickness of the quantum well layer by forming a nitride semiconductor layer containing, and an epitaxial layer having excellent optical characteristics can be formed by a synergistic effect with annealing.

  Therefore, an annealed epitaxial growth structure formed on these nonpolar surfaces was subjected to an element process, and processed into an LED having an emission wavelength of about 410 nm. The results are summarized in Table 11.

  As shown in Table 11, the ratio of the relative total radiant flux compared by introducing the same current value was 1.4 for material Y and 2 for material Z, assuming that material X was 1. On the c-plane GaN crystal, the optimal growth temperature for thick film growth of i-GaN and n-GaN is around 1000 ° C., and crystallinity decreases when crystal growth is performed at a lower temperature. In addition, the emission intensity in the active layer structure formed thereon is also lowered.

  However, as described above, when a crystal is grown on a substrate having a nonpolar main surface, a thick film crystal is grown at a temperature (920 ° C.) lower than the general condition by 100 ° C. or more. Not only the emission intensity was greatly improved, but also the decrease in PL intensity due to annealing could be suppressed, and the LED element characteristics were also greatly improved. Furthermore, according to the present invention, when a nonpolar plane is grown on a substrate having a main surface, the In concentration is relatively low because the threading dislocation density of the epitaxial layer is overwhelmingly lower than other methods. Even in a low InGaN / GaN multiple quantum well active layer structure, good optical characteristics can be realized.

  Furthermore, since it is a non-polar surface and a region with a relatively low In concentration, it is possible to increase the thickness of the InGaN quantum well layer, and the characteristics of an LED having a film thickness exceeding 10 nm are outstanding. It is thought that it was good. Furthermore, since it is an epitaxial layer manufactured by such a method, it is considered that good activation of Mg was realized even under annealing conditions in which deterioration was noticeable in other structures.

  This example is an example in which a near-ultraviolet LED is fabricated by stacking and growing a gallium nitride-based semiconductor thin film by MOCVD, and the outline of a series of crystal growth processes is as shown in FIG. Is as shown in FIG.

As the substrate, a (1-100) plane (m-plane) oriented GaN free-standing substrate was used. The substrate size was 4.4 mm in the c-axis direction and 23 mm in the a-axis direction. The electrical characteristics of the substrate were n-type and the carrier density was 4.5 × 10 ¹⁷ cm ⁻³ . According to X-ray diffraction, the full width at half maximum of the rocking curve in (10-12) reflection is 43.3 arcsec, the OFF angle in the c (+) direction is 0.04 °, and the OFF angle in the a direction is 0.02 °. there were. The threading dislocation density was 5.0 × 10 ⁶ cm ⁻² .

  This m-plane GaN free-standing substrate was placed on a tray (susceptor) in a quartz horizontal reaction furnace under normal pressure growth conditions. The pressure in the reactor was 100 ± 2 kPa in all steps.

  The temperature raising step (step A), the first growth step (step B), the second growth step (step C), and the step D in the third step are performed in the same manner as the sample Z of Example 18. It was.

In the subsequent step E in the third step, first, the substrate temperature was set to 980 ° C., and an Mg-doped Al _0.1 Ga _0.9 N layer was formed to a thickness of 100 nm. The gas constituting the first main flow at this time is NH ₃ (10 L / min).

The gas constituting the second main _flow, H 2 (80L / min), trimethylaluminum that of _H 2 in the main flow (0.5 L / min) and carrier gas (TMAl) (0. 0001 L / min), TMGa (0.0018 L / min) using H ₂ (0.5 L / min), which is also a part of the main flow, as a carrier gas, and H ₂ (0 .5 L / min) as a carrier gas, cyclopentadienyl magnesium (Cp ₂ Mg) (4 × 10 ⁻⁶ L / min).

On this Mg-doped Al _0.1 Ga _0.9 N layer, an Mg-doped Al _0.03 Ga _0.97 N layer was further epitaxially grown to a thickness of 15 nm. The growth of the GaN layer was performed by reducing TMAl in the above-described main flow gas to 0.00003 L / min.

In step E, the subflow was 50.5 L / min with a mixed gas of NH ₃ (0.5 L / min) and N ₂ (50 L / min), and the non-growth gas for purging or the like was N ₂ (19 L / min). It was. Further, the flow rate ratio Fp of the inert gas component to the total gas constituting the main flow was 0.0.

When such a growth process was completed, TMGa, TMAl, Cp ₂ Mg and H ₂ in the main flow were shut off, and N ₂ was introduced at 20 L / min. At the same time, the flow rate of NH ₃ was reduced from 10 L / min to 50 cc / min. Further, blocks the NH ₃ in the subflow was a _{N 2} flow rate 10L / min. Non-growth flows have not changed.

Simultaneously with the switching of these gases, the substrate heater power supply was shut off and the substrate was forcedly cooled by the gas introduced. When the temperature of the substrate dropped to 930 ° C., NH ₃ in the main flow was shut off, and the substrate was subsequently cooled in an N ₂ atmosphere until it became 100 ° C. or lower. After the substrate was sufficiently cooled, the substrate was taken out and evaluated.

  The surface of the substrate thus produced had good flatness although there were very slight irregularities. This surface was measured with a contact-type step meter, and the centerline average roughness (Ra) was determined. As a result, Ra according to the present example was 8.8 nm. In addition, in the PL characteristics evaluated by excitation with a laser beam having a wavelength of 325 nm, the peak wavelength is 410 nm, the integrated intensity is 93 in relative value, and the standard deviation of the in-plane wavelength distribution is 0. It was as small as 7%.

  Thereafter, an epitaxial growth layer on the nonpolar surface prepared by such a method was processed to produce an LED. Both the light emission characteristics and the current-voltage characteristics were good, and it was confirmed that good optical characteristics and sufficient Mg activation were realized by the activation process during the temperature lowering process.

  In this example, the ultra-high pressure transmission electron microscope observation was performed many times on the nonpolar GaN substrate on the LED structure that was observed in Example 10 and produced by the crystal growth method of the present invention in the near ultraviolet emission. The threading dislocation density in the formed epitaxial layer and active layer structure was determined. At this time, the sample was prepared so that it could observe 8 fields of view. When observing, the observed sample / observation conditions were prepared in the same manner as in Example 9 except that a 12.6 μm region along the substrate surface was used and the thickness of the observation view flake was about 1.0 μm. . Observation was performed by electron beam transmission of 1000 KV.

As a result, a total of five dislocations propagating in the m-axis direction were observed in the epitaxial growth portion including the active layer portion. As a result, the average threading dislocation density according to the present invention was estimated to be 5 ÷ (12.6 μm × 1.0 μm × 8) = 4.96 × 10 ⁶ (cm ⁻² ).

In one field of view, a sample in which no dislocation was observed at all and a sample in which three dislocations were observed at the maximum were mixed. Therefore, the dislocation density realized in the epitaxial layer in the present invention is 3 ÷ (12.6 μm × 1.0 μm) even if the highest threading dislocation density is assumed.
= 2.42 × 10 ⁷ (cm ⁻² ). Therefore, the threading dislocation density present in the epitaxial layer in the present invention is preferably 3 × 10 ⁷ (cm ⁻² ) or less, more preferably 5.0 × 10 ⁶ (cm ⁻² ) or less. It can be said.

  The present invention has been described with reference to the embodiments. However, the above-described embodiments are merely examples for carrying out the present invention, and the present invention is not limited thereto. In the embodiment, what is considered to be GaN may be AlN, InN or BN, or a mixed crystal thereof. Further, the growth temperature, the supply amount of each raw material or the film thickness of each layer can be changed according to the purpose.

  Further, in the above-described examples, the surface of the nitride semiconductor layer obtained by crystal growth has irregularities, and some samples are not necessarily good as the surface morphology itself. It is considered that the degree of unevenness can improve the light extraction efficiency when it is made into an element as a light emitting device, and can improve the light emission efficiency. That is, it does not deviate from the allowable range. In this regard, the present inventor understands that the light emission efficiency depends on not only the internal quantum efficiency but also the light extraction efficiency as described above. That is, in general, the crystallinity of a nitride semiconductor layer having surface irregularities is moderate, and the internal quantum efficiency tends to decrease, while the surface irregularities effectively scatter light from the active layer region, resulting in I think that there is an effect of increasing the luminous efficiency.

  In addition, in the step of epitaxially growing the laminated structure including the active layer described above, when introducing a p-type dopant into a part thereof, the introduction location can be set as appropriate, and the entire laminated structure There may be, and it is good also as only an upper part, a lower part, or a center part. Examples of such a p-type dopant include magnesium, zinc, carbon, and beryllium.

  Furthermore, the epitaxial growth process of the stacked structure including the active layer is performed under the condition that the p-type dopant is activated, or following the epitaxial growth process of the stacked structure including the active layer, At least one treatment step of thermal annealing or electron beam irradiation for activating the p-type dopant may be further performed.

As a base used in the present invention, a nitridation having a nonpolar surface as a main surface on a sapphire substrate, ZnO substrate, Si substrate, SiC substrate, GaAs substrate, GaP substrate, Ga ₂ O ₃ substrate, Ge substrate, MgO substrate, etc. What formed the physical crystal into a film may be used.

  It is apparent from the above description that various modifications of these embodiments are within the scope of the present invention, and that various other embodiments are possible within the scope of the present invention.

  The present invention provides a high-quality nitride semiconductor crystal growth method that has good optical characteristics and high luminous efficiency when used as a light-emitting device.

It is a figure for demonstrating notionally the flow of the main flow in a horizontal type MOCVD reactor. It is a figure for demonstrating notionally the flow of the main flow in a vertical type MOCVD reactor. It is a figure for demonstrating the example of a sequence for demonstrating the crystal growth method of the nitride semiconductor of this invention. FIG. 10 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 4. FIG. 10 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 6. FIG. 10 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 7. FIG. 10 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 9. FIG. 22 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 11. It is a figure for demonstrating the example of a sequence for demonstrating the crystal growth method of the nitride semiconductor of this invention. 6 is a diagram for explaining a sequence example for explaining a nitride semiconductor crystal growth method of Comparative Example 1. FIG. FIG. 22 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 13. FIG. 20 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 14. 10 is a diagram for explaining a sequence example for explaining a nitride semiconductor crystal growth method of Comparative Example 2. FIG. 10 is a diagram for explaining a sequence example for explaining a nitride semiconductor crystal growth method of Reference Example 2. FIG. FIG. 32 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 15. FIG. 32 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 16. FIG. 17 is a diagram for explaining a sequence example for explaining the nitride semiconductor crystal growth method of Example 17. It is the cross-sectional schematic for demonstrating the nitride semiconductor example of this invention. 6 is a schematic cross-sectional view for explaining an example of a nitride semiconductor of Example 6 and Example 7. FIG. 10 is a schematic cross-sectional view for explaining an example of a nitride semiconductor of Example 9. FIG. 10 is a schematic cross-sectional view for explaining examples of nitride semiconductors of Example 11, Example 13, and Reference Example 2. FIG. 6 is a schematic cross-sectional view for explaining an example of a nitride semiconductor of Example 12 and Comparative Example 1. FIG. 14 is a schematic cross-sectional view for explaining an example of a nitride semiconductor in Example 14 and Comparative Example 2. FIG. 22 is a schematic cross-sectional view for explaining an example of a nitride semiconductor in Example 15. FIG. 22 is a schematic cross-sectional view for explaining an example of a nitride semiconductor in Example 17. FIG. It is a figure for demonstrating the PL light emission characteristic of the sample used as LED of the structure shown to FIG. 3 (A) using the m-plane nitride semiconductor grown by the crystal growth method of this invention. 4 is a differential interference microscope image of Sample A for explaining the result of finding the dependence of the main flow on the constituent gas species during the epitaxial growth of the first and second nitride layers. 4 is a differential interference microscope image of Sample B for explaining the result of finding the dependence of the main flow on the constituent gas species during the epitaxial growth of the first and second nitride layers. 4 is a differential interference microscope image of Sample C for explaining the result of finding the dependency of the main flow on the constituent gas species during the epitaxial growth of the first and second nitride layers. 10 is a STEM image obtained in Example 8. FIG. 7 is a result of image processing performed to calculate an average thickness and the like of each quantum well layer from the STEM image of FIG. 10 is a STEM image obtained from the LED of Example 9. It is the result of the image processing performed in order to calculate the average thickness of each quantum well layer, etc. from the STEM image of FIG. It is the figure which compared the average thickness uniformity (standard deviation of the thickness of a quantum well layer) of each quantum well of Example 8 and Example 9. FIG. It is the image which observed the sample created in Example 10 with the ultrahigh pressure TEM. It is a block diagram of the optical system used by CW-PL measurement. It is a block diagram of the optical system used by time-resolved PL measurement (with a polarizing filter). It is a block diagram of the optical system used by time-resolved PL measurement (without a polarizing filter). 2 is a band diagram for explaining luminescent recombination and non-radiative recombination occurring in a semiconductor crystal. It is a figure which shows the result of having investigated experimentally the excitation energy density dependence of the internal quantum efficiency of the multiple quantum well structure at the time of exciting with an optical pulse. 18 is a surface AFM (atomic force microscope) image of a nitride semiconductor sample X in Example 17. 18 is a surface AFM (atomic force microscope) image of a nitride semiconductor sample Y in Example 17. 20 shows a PL measurement result of Example 18.

1 Quartz reaction tube 2 Susceptor 3 Base 4 Stainless steel chamber 10 GaN substrate 11 First nitride semiconductor layer (GaN layer)
12 Second nitride semiconductor layer (Si-doped n-type GaN layer)
13 InGaN / GaN multiple quantum well active layer structure 14 Mg-doped AlGaN layer 15 Mg-doped GaN layer 101 He-Cd laser 102 ND filter 103 Cryostat 104 Sample 105 λ / 4 wavelength plate 106 Spectrometer 107 Photomultiplier tube 108 Control computer 109 Pulsed laser light source 110 Polarizing filter 111 Depolarizer 112 Streak scope

That is, in the nitride semiconductor of the present invention, at least one main surface of the nitride main surface of the base is a nonpolar nitride, one conductivity type nitride semiconductor portion, quantum well active layer structure portion, A nitride semiconductor in which nitride semiconductor portions of the other conductivity type opposite to the one conductivity type are sequentially stacked, wherein the nitride semiconductor portion of the one conductivity type includes a first nitride semiconductor layer and a second nitride semiconductor layer. are those nitride semiconductor layer are sequentially laminated, wherein the second nitride semiconductor layer is Ri nonpolar plane der substantially the top-surface having a 20μm thickness of not less than 400 nm, the quantum well active layer structure part The internal quantum efficiency obtained from CW-PL measurement under low excitation density conditions is 20% or more .

Claims (28)

A nitride semiconductor portion of one conductivity type, a quantum well active layer structure portion, and the other conductivity opposite to the one conductivity type are formed on a nitride principal surface of a substrate whose at least one principal surface is a nonpolar nitride. A nitride semiconductor in which nitride semiconductor parts of a type are sequentially stacked,
The one-conductivity-type nitride semiconductor portion is formed by sequentially laminating a first nitride semiconductor layer and a second nitride semiconductor layer,
The nitride semiconductor, wherein the second nitride semiconductor layer has a thickness of 400 nm or more and 20 μm or less, and an outermost surface is a substantially nonpolar surface.   The nitride semiconductor according to claim 1, wherein the first nitride semiconductor layer and the second nitride semiconductor layer have different compositions.   3. The nitride semiconductor according to claim 1, wherein the base is a free-standing substrate made of GaN, AlN, InN, BN, or a mixed crystal thereof.   The nitride of the main surface of the base is grown on a sapphire substrate, a SiC substrate, a ZnO substrate, a Si substrate, a GaN substrate, an AlN substrate, an InN substrate, a BN substrate, or a free-standing substrate that is a mixed crystal thereof. The nitride semiconductor according to claim 1, wherein the nitride semiconductor is a GaN film, an AlN film, an InN film, a BN film, or a mixed crystal film thereof.   5. The nitride semiconductor according to claim 1, wherein the nitride main surface of the base is a crystal plane of (1-100) plane (m-plane) ± 10 degrees or less. 6. At least one of the first nitride semiconductor layer and the second nitride semiconductor layer is GaN, AlN, InN, BN, or a mixed crystal group III-V nitride semiconductor thereof. The nitride semiconductor according to any one of the above. The nitride semiconductor according to any one of claims 1 to 6, wherein a thickness L1 of the _first nitride semiconductor layer is not less than 0.1 nm and not more than 300 nm. The nitride semiconductor according to claim 1, wherein a silicon (Si) concentration in the first nitride semiconductor layer is 1 × 10 ²¹ cm ⁻³ or less. 9. The nitride semiconductor according to claim 1, wherein the silicon concentration in the second nitride semiconductor layer is not less than 1 × 10 ¹⁷ cm ^{−3 and not} more than 6 × 10 ¹⁹ cm ⁻³ .   The nitride semiconductor according to claim 1, wherein the quantum well active layer structure emits light having a center wavelength of 400 ± 30 nm. 11. The nitride semiconductor according to claim 1, wherein a dislocation density in the quantum well active layer structure portion is 1 × 10 ⁹ cm ⁻² or less.   The internal quantum efficiency calculated | required from the CW-PL measurement of low excitation density conditions of the said quantum well active layer structure part is 20% or more, The any one of Claims 1 thru | or 11 characterized by the above-mentioned. Nitride semiconductor.   13. The internal quantum efficiency of the quantum well active layer structure portion, which is obtained from pulsed light PL measurement under a low excitation density condition, is 20% or more. 13. Nitride semiconductor.   The photoluminescence lifetime (τ (PL)) obtained from the time-resolved PL measurement at room temperature and under a low excitation density condition of the quantum well active layer structure is 1.0 ns or more. 14. The nitride semiconductor according to any one of 1 to 13.   15. The nitride semiconductor according to claim 1, wherein the quantum well layer included in the quantum well active layer structure has a thickness of 10 nm to 100 nm.   The nitride semiconductor according to any one of claims 1 to 15, wherein the one conductivity type is n-type and the other conductivity type is p-type. The concentration of the dopant element in the other conductivity type nitride semiconductor part is 1 × 10 ¹⁹ cm ⁻³ or more and 8 × 10 ¹⁹ cm ⁻³ or less, according to any one of claims 1 to 16. Nitride semiconductor.   18. The nitride semiconductor according to claim 1, comprising at least one element selected from magnesium, zinc, carbon, and beryllium as a dopant of the other conductivity type nitride semiconductor portion. .   19. The nitride semiconductor according to claim 1, wherein the thickness of the other conductivity type nitride semiconductor portion is 0.05 μm or more and 0.25 μm or less.   The nitride semiconductor according to any one of claims 1 to 19, wherein the other conductivity type nitride semiconductor portion is made of AlxGa1-xN (0≤X≤1).   21. The nitride semiconductor according to claim 20, wherein an Al composition x of the AlxGa1-xN is 0.02 or more and 0.20 or less. The nitride semiconductor according to any one of claims 1 to 21, wherein an area of an outermost surface of the nitride semiconductor is 30 mm ² or more and 500 cm ² or less.   The nitride semiconductor according to any one of claims 1 to 22, wherein an average surface roughness Ra of the outermost surface of the nitride semiconductor is in a range of 100 nm to 300 nm.   The first nitride semiconductor layer of the nitride semiconductor is epitaxially grown without intentional Si material supply, and is provided without interposing another layer between the nitride main surface of the substrate. The nitride semiconductor according to any one of claims 1 to 23, wherein:   The nitride semiconductor according to any one of claims 1 to 24, wherein a standard deviation in a plane of a thickness of a quantum well layer constituting the quantum well active layer structure portion is 0.45 nm or less.   The nitride semiconductor according to any one of claims 1 to 25, wherein a variation coefficient of thickness variation in a plane of the quantum well layer constituting the quantum well active layer structure portion is 0.1 or less.   The luminescence recombination lifetime (τ (R)) obtained from the time-resolved PL measurement at room temperature and under a low excitation density condition of the quantum well active layer structure is greater than the non-radiative recombination lifetime (τ (NR)). 27. The nitride semiconductor according to claim 1, wherein the nitride semiconductor is short.   The nitride semiconductor according to any one of claims 1 to 27, wherein an outermost surface of the one conductivity type nitride semiconductor portion is a crystal plane of (1-100) plane (m-plane) ± 5 degrees or less.