JP3221359B2 - P-type group III nitride semiconductor layer and method for forming the same - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention
Formation of p-type gallium nitride (GaN) or aluminum-gallium nitride (AlGaN) mixed crystal used in nitride compound semiconductor devices such as double hetero (DH) junction type light emitting diodes (LEDs) including junctions More particularly, the present invention relates to a method for forming a low-resistance p-type group III nitride semiconductor layer without requiring special post-processing.

[0002]

2. Description of the Related Art Compounds such as diodes are used.
In a conductor element, a pn junction is used to add rectification
Is an essential component of. General formula Al_x Ga_y In_z 
N (x + y + z = 1, 0 ≦ x, y, z ≦ 1)
Composed of aluminum nitride-gallium mixed crystal
Even in a wavelength light emitting diode (LED), the light emitting portion is p
Consisting of a double heterojunction with an n-junction
It is customary (J. Appl. Phys.,76
(12) (1994), pp. 8189-8191 and Jp.
n. J. Appl. Phys. ,Vol. 34, Par
t2, No. 7A (1995), L797-L799
reference). Gallium nitride based light emitting device with excellent light emission intensity
Metal-insulator-semiconductor (MIS) junction type
It is known that the pn junction type is superior in light emission characteristics.
(Masayoshi Koike et al., “Toyoda Gosei
Information ",Vol. 38, No. 1 (1996), 46-4
See page 9.)

[0003] Looking at the usage of the p-type nitride semiconductor layer in the conventional nitride compound light emitting device, one is in the LED as described above, and the light emitting layer is composed of gallium indium or the like. On the other hand, it is used as a p-type clad layer. In a double-heterojunction structure intended to "confine" light emitted by the recombination of carriers in the light-emitting layer and to exhibit high light-emission output (supervised by Hiroshi Iwasaki, "Optoelectronic Materials" (June 15, 1985) The p-type cladding layer has a role of supplying holes for recombining with electrons to the light emitting region. Another application example of the nitride compound semiconductor light emitting device is an example in which a p-type nitride semiconductor layer is used as a contact layer. The contact layer is a layer provided for laying electrodes and for spreading the operating current over a wide area of the lower cladding layer and the light emitting layer. It is required that there be. FIG.
Is a conventional blue LE made of gallium nitride-based material.
It is a schematic diagram of the cross-section structure of D. In the laminated structure shown in the figure, a p-type aluminum-gallium nitride mixed crystal layer as an upper cladding layer (105) bonded on a gallium-indium indium light emitting layer (104) and an upper cladding layer (1)
05) P-type gallium nitride (GaN) contact layer disposed on the p-type nitride semiconductor layer is specifically used. When a p-type crystal substrate is used, it may be used as a lower cladding layer.

[0004] As described above, the p-type nitride semiconductor layer is indispensable for constituting a pn junction type light emitting device. When obtaining a p-type gallium nitride layer, a Group II element of the periodic table such as magnesium (Mg) or beryllium (Be) is added as a p-type impurity. Among these p-type impurities, magnesium is used as a typical p-type impurity.
Conventional methods for forming a magnesium-doped gallium nitride layer include, for example, an aluminum nitride-gallium mixed crystal cladding layer (shown by reference numeral (105) in FIG. 1) and a magnesium-doped gallium nitride contact layer ((10 in FIG. 1)).
In order to observe the growth of 6)), it was customary to constitute a single layer doped with magnesium to a desired concentration. In other words, it was not formed from a multilayer structure in which thin layers doped with p-type impurities were stacked. In addition, a laminated structure of a p-type layer made of a nitride semiconductor layer doped with a p-type impurity, which is practically used, has a p-type layer made of an aluminum nitride mixed crystal and a lattice match with an aluminum nitride-gallium mixed crystal. As shown in the example of laminating p-type layers made of gallium nitride which is not used (see FIG. 1), the structure is a lattice mismatch system.

However, a magnesium-doped nitride semiconductor layer formed according to a conventional method has an as-grown state (a state in which no processing has been performed after growth and the state immediately after growth is directly reflected. ), It is generally accepted that a low-resistance p-type layer cannot be obtained unless special measures such as laying an underlayer are taken (see Japanese Patent Application Laid-Open No. Hei 6-232451). That is, it is difficult to obtain a low-resistance nitride semiconductor layer simply by doping the p-type impurity at a high concentration. Gallium nitride, which is a constituent material of the contact layer, or gallium indium nitride, which is a light emitting layer of a short-wavelength visible LED, is n-type in an undoped state (JP-A-56-80183, JP-B-55-3834, etc.). reference). Therefore, the reason that a low-resistance p-type layer cannot be easily formed is that the p-type impurity is electrically activated to become an acceptor enough to sufficiently exceed the concentration of the n-type carrier remaining in the layer. It is thought that there is not. In addition, hydrogen taken into the nitride semiconductor layer doped with the p-type impurity during film formation forms a complex with the doped p-type impurity, electrically inactivates the p-type impurity, and forms a complex with the acceptor. (Japanese Patent Laid-Open Nos. 5-183189 and 5-19884).
No. 1 and Appl. Phys. Let
t. , 68 (13) (1996), 1829-1831.
Page).

Since the hydrogen atoms form a complex with the p-type impurity, the electrical activation of the p-type impurity cannot be sufficiently achieved, and the acceptor concentration cannot be effectively increased. The phenomenon that is considered to hinder the formation of the layer is not limited to nitride compound semiconductors, and has already been observed for II-VI group compound semiconductors such as zinc selenide (ZnSe) (Japanese Patent Laid-Open No. 52-1982). No. 15,1563). A conventional technique for obtaining a p-type layer in a II-VI group compound semiconductor is, for example, to subject a ZnSe layer doped with a p-type impurity to a heat treatment (see Japanese Patent Application Laid-Open No. -155633). Following the conventional means of achieving p-type by heat treatment, even in group III nitride semiconductors, the bond between p-type impurities and hydrogen is dissociated by heat treatment to provide a low-resistance p-type nitride. Methods of forming a semiconductor layer are disclosed (Japanese Patent Application Laid-Open Nos. 5-183189 and 5-19884).
No. 1).

[0007] Various conditions for making the same idea have been proposed for a heat treatment method intended to make a group III nitride semiconductor doped with a p-type impurity p-type. (A) The group III nitride semiconductor layer added (doped) during film formation is
Once cooled and taken out of the film-forming environment, it is heated again at a certain temperature for a certain time (JP-A-5-1831).
No. 89), and (b) doping with p-type impurities
After the film formation of the group III nitride semiconductor layer is completed, heating from a film formation temperature to room temperature at a slower cooling rate than natural cooling (see Japanese Patent Application Laid-Open No. 8-32113). A processing technique is disclosed. However, a conventional means for performing a heat treatment to form a p-type is a p-type II-
It has the same concept as the means for forming a group VI compound semiconductor, and has a growth step as a step prior to forming a group III nitride semiconductor layer doped with a p-type impurity and a step for forming the layer into a p-type impurity. It is a complicated process that requires a heat treatment as a process and a two-step process.

Another problem of achieving the p-type by the heat treatment is that a loss due to thermal dissociation occurs during the heat treatment for forming the laminate as shown in FIG. It is known that gallium nitride (GaN) and indium nitride (InN) have an easy sublimation property. Gallium nitride is about 8
Sublimate at 00 ° C. It is said that indium nitride sublimates easily at lower temperatures and sublimates even in vacuum at about 620 ° C. (edited by the Japan Society for the Promotion of Industry, New Materials Committee, “Compound Semiconductor Device” (September 15, 1973, (Issued by the Japan Industrial Technology Promotion Association) (see pages 316 and 397). Therefore, in a conventional heat treatment method for forming a p-type, at a temperature of 600 ° C. or more, which is a preferable heat treatment temperature, sublimation of a gallium nitride-indium mixed crystal which is a mixed crystal of gallium nitride and indium nitride is avoided. (See JP-A-5-198841). Gallium indium nitride is a material for forming the light emitting layer. Therefore, by performing the p-type heat treatment, not only the loss of the gallium nitride layer to be p-typed on the outermost layer of the laminate (see FIG. 1) may be caused, but also the gallium nitride
There is also a problem that the indium light emitting layer is deteriorated by heat. For the purpose of suppressing the volatilization of the nitride semiconductor layer at the time of heating, a heat treatment technique of disposing a protective film on the surface of a layer doped with a p-type impurity has also been disclosed (Japanese Patent Laid-Open No. 5-198841 described above). reference). However, when a protective film such as silicon dioxide (SiO ₂ ) or silicon nitride (SiN) is applied, the difference in the coefficient of thermal expansion between the protective film and the heat-treated layer is further increased by heat treatment at a high temperature. Since the p-type impurity-doped layer becomes remarkable, it undergoes excessive strain. As a result, the crystallinity of the p-type impurity-doped layer is further deteriorated due to the occurrence of dislocation and the like, so that a low-resistance p-type nitride semiconductor layer cannot be stably obtained.

[0009] A method of forming a p-type nitride semiconductor layer without heat treatment has been proposed in order to eliminate problems in forming a low-resistance p-type nitride semiconductor layer by heat treatment.
A method for forming a p-type nitride semiconductor layer without using heating means is to deposit a nitride semiconductor layer doped with a p-type impurity on a mixed crystal layer containing indium (In),
This is a technique of forming a p-type layer in a grown state (see Japanese Patent Application Laid-Open No. 6-232451). An extremely general film forming temperature of gallium nitride / indium mixed crystal depends on a desired indium composition ratio or the like, but is practically about 800 ° C. On the other hand, the film forming temperature of the p-type gallium nitride or the p-type aluminum nitride-gallium mixed crystal used for the short-wavelength visible LED as described above is approximately 1000 from the viewpoint that a good crystalline layer is obtained. It is a high temperature of 1200C to 1200C. Therefore, if a p-type aluminum nitride-gallium mixed crystal layer is to be obtained as-grown on a gallium nitride-indium mixed crystal layer, a gallium nitride-indium mixed crystal layer is formed at about 800 ° C. Approximately 10 to obtain an aluminum nitride-gallium mixed crystal layer with excellent crystallinity
It is necessary to raise the temperature to a temperature exceeding 00 ° C. About 100
At a high temperature of 0 ° C., the gallium-indium nitride mixed crystal layer is surely sublimated and lost. That is, as-g regardless of heat treatment
Even in the method of forming a p-type nitride semiconductor layer in a row state, the method simply causes complication of the process, and does not reach a method of forming a stable p-type nitride semiconductor layer commensurate with the increase in the number of processes. Was something.

[0010]

The gist of the prior art intended to be p-type is summarized in the following two points. That is, (1) a technique requiring a post-treatment step of separating a complex such as p-type impurity and hydrogen by a heating means to form an electrically active acceptor, and (2) a mixed crystal layer containing indium. This is a technique that requires an additional growth step for improving the crystallinity of a nitride semiconductor layer doped with a p-type impurity by using as a base to achieve a p-type in an as-grown state. A problem common to the prior art is the loss or alteration due to heating of the light-emitting layer composed of the easily sublimable gallium-indium nitride mixed crystal or the underlayer. Due to heat reception in a heat treatment step intended to form a low-resistance p-type layer as a post-treatment step, the crystal structure of the light-emitting layer changes, and the quality of the crystal deteriorates. Specifically, the crystal structure inside the light emitting layer is disturbed due to aggregation of indium and the like. When a new layer is formed on the underlayer made of the indium-containing group III nitride semiconductor layer, the p-type is formed as-grown.
The same disadvantages as in the case of forming a group III nitride semiconductor layer occur. That is, alteration of the indium-containing group III nitride underlayer occurs due to generation of indium droplets, aggregation of the droplets, and the like. Modification of the underlayer affects the quality of the indium-containing group III nitride semiconductor light emitting layer. At times, loss of the light emitting layer occurs. If a partial loss of the light emitting layer occurs, the layer thickness of the light emitting layer becomes uneven. As a result, non-uniformity in light emission characteristics, particularly non-uniformity in light emission wavelength, is promoted. If there is a simple method capable of forming a low-resistance p-type nitride semiconductor layer in an as-grown state without inducing a thermal behavior such as indium aggregation which is inconvenient for improving the light emitting characteristics, The problems of the prior art, which require a heat treatment technique and an underlayer that is susceptible to deterioration, are solved.

[0011]

Means for Solving the Problems That is, the present invention relates to III.
A p-type group III nitride semiconductor layer on a group III nitride semiconductor crystal
A method of epitaxially growing
The growth is stopped when the thickness of the vertical growth layer is 50 nm or less.
Wait for 60 seconds or more at the growth temperature,
Starting with a plurality of thin layers with a single layer thickness of 50 nm or less
Layer by epitaxial growth of
A p-type group III nitride semiconductor layer composed of a constituent layer is formed.

[0012] The group III nitride compound semiconductor in which the present invention to form a p-type layer of low resistance is mainly intended, the general formula _{Al x Ga y In z N (} x + y + z = 1,0 ≦ x, y ,
z ≦ 1). In addition, Al _x Ga _{y
In z N a P 1-a} (0 <a ≦ 1) and Al _x Ga _y In
_z N _b As _1-b nitride compound semiconductor containing a group V element such as arsenic (As) or phosphorus (P) in addition to the nitrogen represented by (0 <b ≦ 1) it is also of interest. In particular, in these group III nitride compound semiconductors, gallium nitride (GaN) and aluminum-gallium nitride mixed crystal (Al _x Ga _{1 -x} N: 0 <
x ≦ 1).

In the present invention, a nitride semiconductor layer doped with a p-type impurity is divided into thin layers and epitaxially grown. The thickness of one layer doped with a p-type impurity is 50
A plurality of group III nitride semiconductor thin layers having a thickness of nm or less are stacked to form one p-type group III nitride semiconductor layer as a whole. For example, to obtain a p-type GaN layer having a desired layer thickness of 100 nm in an as-grown state,
For example, two layers of gallium nitride layers doped with p-type impurities each having a thickness of 50 nm are stacked to obtain a p-type gallium nitride layer having a total thickness of 100 nm. Also,
Seven p-type impurity-doped aluminum gallium nitride mixed crystal layers each having a layer thickness of 20 nm are stacked to form a p-type aluminum nitride gallium mixed crystal layer having a total thickness of 140 nm. This is another application example according to the present invention. The thickness of each layer to be overlaid is approximately 2 nm
The thickness is preferably from 50 to 50 nm, more preferably from 5 to 50 nm. The thickness required for the p-type nitride semiconductor layer is 200 nm when used as a cladding layer.
The extent is about 100 nm when used as a contact layer. Therefore, two to 100 layers are stacked.
If the thickness of the multilayer constituting layer is extremely thin, for example, less than 1 to 2 nm, the surface of the lower layer may not be able to be coated sufficiently uniformly, which is inconvenient. Conversely, when the thickness is increased, the p-type dopant is not uniformly distributed. The layer thickness of each of the layers to be laminated does not necessarily have to be the same. There is no particular limitation on the number of layers to be overlaid, that is, the number of layers constituting the overlaid layers.

It is not necessary that the nitride semiconductors doped with the p-type impurity to be overlaid be the same. The gist of the method for forming a p-type nitride semiconductor layer according to the present invention is to adopt an as-grown method using a divided growth method in which thin layers are stacked.
The reason is that the p-type nitride semiconductor layer is formed by the method described above, and the material of the multi-layer constituent layer is not essentially limited. That is, a method of forming a p-type nitride semiconductor layer by laminating nitride semiconductor layers having different constituent elements is also acceptable.
For example, one of the multi-layered constituent layers is formed of gallium nitride (Ga
N), and the other is aluminum nitride-gallium mixed crystal (Al _x Ga _{1 -x} N: 0 <x
≤ 1). Aluminum nitride having an aluminum composition ratio of about x / 2, resulting from the interdiffusion of atoms occurring between the layers during the layering operation at a high temperature, by alternately laminating those layers having substantially the same layer thickness. -There is a method of forming a gallium mixed crystal layer. The aluminum composition ratio (x) of the aluminum nitride / gallium mixed crystal layer may be appropriately changed according to the desired aluminum composition ratio.

When a p-type layer composed of a nitride semiconductor mixed crystal is formed by the formation method according to the present invention, it is not always necessary to make the mixed crystal ratio (composition ratio) of each layer the same.
For example, when forming a p-type layer made of an aluminum nitride-gallium mixed crystal, it is not always necessary to unify the aluminum composition ratio of the aluminum nitride-gallium layers to be stacked. For example, a layer composed of an aluminum nitride-gallium mixed crystal in which the aluminum composition ratio is increased or decreased stepwise is sequentially laminated, and the aluminum composition ratio is gradually changed in the direction of increasing the layer thickness (layer direction). An aluminum nitride-gallium mixed crystal layer may be formed. The effect of the method of the present invention is, for example, when a p-type aluminum nitride / gallium mixed crystal layer is arranged on a gallium nitride / indium light emitting layer, as seen in a conventional short wavelength visible LED composed of a nitride semiconductor. Revealed. For example, first, aluminum nitride (AlN) or a first aluminum alloy having a large aluminum composition ratio is formed on an easily sublimable gallium indium nitride layer.
Is deposited. Aluminum nitride (AlN)
Is a sublimation-resistant semiconductor material as compared with gallium nitride, indium nitride, and mixed crystals thereof. Therefore, even when the aluminum nitride-gallium mixed crystal layer is formed on the gallium nitride-indium light-emitting layer at a high temperature, the aluminum nitride layer remains without sublimation, thereby preventing loss due to thermal dissociation of the light-emitting layer. It can be a protective film that is enough to perform. That is, there is an advantage that the layer doped with the p-type impurity according to the present invention becomes a constituent layer of the p-type layer and also serves as a protective film for preventing the loss of the light emitting layer. Next, a gallium nitride layer is deposited as a second layer. Then, an aluminum nitride-gallium mixed crystal layer having a smaller aluminum composition ratio than the first layer is deposited. After that, a gallium nitride layer is deposited again, and an aluminum nitride-gallium mixed crystal layer having a reduced aluminum composition ratio is successively laminated to form a laminated p-type layer including a gradient of the aluminum composition in the layer thickness direction. This is one application example of the present invention. The formation temperatures of the respective layers to be layered need not necessarily be the same. In view of the crystallinity of the nitride semiconductor material constituting the multilayer constituent layer or the role other than the component of the p-type layer carried by the layer, even if the formation temperature is changed for each layer or for each constituent material, It does not materially change the spirit of the present invention based on the growth method.

The so-called split growth method in which the thin layer is divided and overlaid according to the present invention is based on an idea based on the following experimental facts. First, a gallium nitride single crystal layer (10
9) Aluminum nitride doped with magnesium (Mg) having an aluminum composition ratio of 0.10.
The distribution of the gallium mixed crystal layer (110) in the as-grown state near the interface (111) between the aluminum and magnesium atoms when the gallium mixed crystal layer (110) is grown to a thickness of 200 nm by the conventional method will be described. FIG. 2 shows a concentration distribution state (112) of aluminum atoms near the interface (111) obtained by a line analysis by an energy dispersive X-ray micro area analysis method (generally referred to as EDX). It shows. As- without heat treatment after film formation
In the grown state, aluminum atoms are composed of gallium nitride (109) and aluminum-gallium nitride mixed crystal (11
It is shown that a large amount is accumulated particularly on the aluminum nitride-gallium mixed crystal layer (110) side near the interface (111) with (0). FIG. 2 also shows that the concentration of aluminum atoms on the surface side of the aluminum nitride-gallium mixed crystal layer (110) deviating from the region near the interface (111) sharply decreases. . In other words, aluminum nitride grown by the conventional batch growth method
In the gallium mixed crystal, the aluminum atoms are distributed so as to decrease the concentration from the vicinity of the interface (111) with the layer to be deposited (in this case, the gallium nitride single crystal layer (109)) in the direction of increasing the thickness of the mixed crystal layer. And exist. Such non-uniform distribution of aluminum as a non-uniform constituent element,
Almost independent of the growth temperature of the aluminum nitride-gallium mixed crystal, the extremely common growth temperature of the mixed crystal is about 1000
It occurs at from about 1150C to about 1150C. Further, regardless of the growth method, for example, the normal pressure and the decompression method, a difference occurs to some extent. In particular, in a space where a film can be formed near the surface of a deposition object such as a substrate in a space where deposition can be achieved rather than the pressure of a growth environment in which a growth reaction is performed, it is more remarkable to create a condition for shortening the residence time of a source gas or the like. Tends to have a remarkable change in the concentration of aluminum atoms and the like. In other words, in a growth environment in which the source gas does not escape from the film formation space smoothly, that is, the so-called “exit” of the growth-related gas does not progress smoothly, it is difficult to achieve a steep deposition layer interface. On the other hand, changes in the concentration of aluminum atoms and the like inside the layer may be reduced. However, there is no change in the concentration of aluminum atoms in the surface layer of the aluminum nitride-gallium mixed crystal layer to any extent.

On the other hand, FIG. 3 shows the distribution of magnesium atoms in the depth direction in the sample. This concentration distribution of magnesium atoms in the depth direction was obtained by general secondary ion mass spectrometry (SIMS). Similar to the distribution state near the interface (111) of aluminum atoms, magnesium atoms are deposited on the underlying gallium nitride single crystal layer (109) and aluminum nitride-gallium mixed crystal layer (11).
In particular, a large amount is shown to be accumulated in a region near the interface (111) with (0).

Hexagonal wurtzite
A gallium nitride underlayer having an a-axis lattice constant of 3.180 angstroms (Å) in the structure, and an a-axis lattice constant of 3.111 in the same structure depending on the aluminum composition ratio.
(Å) The upper layer of aluminum nitride-gallium mixed crystal of not less than 3.180 (Å) does not lattice match in the first place (for the lattice constant, see, for example, Yasuo Okuno, “Light Emitting Diode” (January 1993) (Published by Sangyo Tosho Co., Ltd. on the 20th), page 191 “Table of physical properties of compound semiconductors”). Therefore, at the interface between gallium nitride and the aluminum-gallium nitride mixed crystal layer, a strain is generated due to lattice mismatch, and in the "field" where the strain stress propagates, aluminum and magnesium as a p-type impurity are generated. It is considered that the incorporation of atoms is particularly pronounced. In addition, aluminum and magnesium are observed during the period in which the lateral growth (two-dimensional growth) occurring along the surface of the underlying layer in the early stage of growth, which is observed in the growth mechanism of the nitride semiconductor layer, is predominantly progressing. It is obtained from the experimental results by the present inventor that atoms are taken into the layer with a high probability (FIG. 3).
reference). The period during which the lateral growth is dominant can be known, for example, from the relationship between the growth (film formation) time and the layer thickness. FIG.
Sets the aluminum composition ratio to 0.15 on the gallium nitride layer.
Aluminum gallium mixed crystal (Al _0.15 Ga
FIG. 4 is a diagram showing the growth time dependence of the layer thickness when _0.85 N) is vapor-phase grown at 1100 ° C. by a general atmospheric pressure MOCVD method. As shown in the figure, the mode of change in the layer thickness differs depending on the growth time zone. In the growth time zone (113) in the early stage of growth, the increase in the layer thickness is not observed, but the time after the growth time has passed a certain time (115) (tentatively referred to as critical time). In band (114), the layer thickness monotonically increases with increasing growth time. In this time zone (114), the three-dimensional growth, that is, the growth in the vertical (vertical) direction is predominant, so that it is interpreted that the increase in the layer thickness is remarkable. On the other hand, in the time zone (113) in the early stage of growth, two-dimensional growth, that is, growth in the lateral (horizontal) direction is proceeding preferentially, and therefore, the increase in the layer thickness does not appear significantly. . The range of time during which the lateral growth according to the present invention preferentially proceeds before reaching the critical time (115) can be inferred from the correspondence between the growth time and the layer thickness as described above. . The fact that the layer thickness is too small also leaves the possibility that a laterally continuous film has not grown.

The proportion occupied by the region to which the stress caused by the lattice strain is occupied increases as the layer thickness decreases. This means that if the layer is thin, the above-mentioned constituent element or p-type impurity element is easily taken in. Means increase. In other words, in the aluminum nitride-gallium mixed crystal thin layer doped with magnesium, the thinner the layer, the more the region containing both aluminum, which is one of the constituent elements, and magnesium, which is a p-type impurity, in a high concentration. Will be present. In the nitride semiconductor layer formed by stacking such thin layers, a decrease in the concentration of the p-type impurity in the thickness direction is naturally avoided.
It is fundamentally advantageous to form a p-type nitride semiconductor layer in an as-grown state. This is the reason that the present invention forms a p-type layer by laminating thin layers having specified thicknesses.

In a layer having a layer thickness exceeding the value specified in the present invention, the area where the stress strain exists decreases as the layer thickness increases. Thereby, the constituent element or p in the layer
This results in a distribution in the concentration of the impurity. FIG. 5 schematically shows the concentration distribution of the p-type impurity in the single layer in the case where the layer thickness is out of the range of the present invention and is 100 nm, which is caused by lattice mismatch with the underlayer. This indicates that the ratio of the region to which the stress strain is applied decreases with respect to the total thickness, so that the atomic concentration decreases toward the surface. FIG.
FIG. 5 schematically shows the concentration distribution in the depth direction that results when a layer (116) having a non-uniform p-type impurity concentration distribution in the layer thickness direction as shown in FIG. is there.
Since the layer (116) to be layered has a non-uniform concentration distribution in the first place, the distribution of the atomic concentration in the layer consequently formed by layering is also non-uniform. A layer in which the resulting p-type impurity concentration inside the layer does not become constant in the depth direction but has a distribution that increases and decreases is difficult to become an as-grown p-type layer. In addition, as can be clearly seen in an aluminum nitride-gallium mixed crystal layer having an unusually large layer thickness, the incorporation rate of aluminum atoms is extremely reduced as the layer thickness increases,
The surface layer tends not to be a mixed crystal of aluminum nitride and gallium but to be a layer mainly composed of gallium nitride. That is, in regions other than the region near the interface where lattice strain is present or where lateral growth is dominant, aluminum tends to be hardly taken in and mixed crystal tends to be hardly achieved. The heat of formation (enthalpy) of aluminum nitride is 76.0 (kcal / mo)
l), whereas that of gallium nitride has about 1
/ 3 is 26.4 (kcal / mol) (see “Light-Emitting Diode” (published by Sangyo Tosho), page 192).
This remarkable difference in generation energy also supports that gallium nitride is formed more easily than aluminum nitride / gallium mixed crystal or aluminum nitride. In summary, if the thickness of the layer to be overlaid (overlying layer) is made larger than necessary, the concentration of the constituent elements and the p-type impurity in the layer thickness direction is reduced, and the depth of the p-type impurity in the depth direction is increased. As-grown due to the non-uniform distribution of
It is disadvantageous to form a p-type layer in the n state. This is the reason why in the present invention, the thicknesses of the layers to be laminated, that is, the layer thicknesses of the laminated constituent layers are added. In order to obtain a nitride semiconductor layer in which the concentration distribution of the p-type impurity is substantially uniform, the amount of the dopant to be added is gradually increased with the elapse of the growth time so as to match the decrease in the concentration of the p-type impurity with the increase in the layer thickness. According to the present invention, a nitride semiconductor having excellent uniformity of p-type impurity concentration can be obtained by simply limiting the layer thickness without using a complicated conventional growth technique. Layers can be formed. Therefore, a complicated operation of increasing the doping amount of the p-type impurity and increasing the growth time as well as the concentration is not required, and a layer having a uniform concentration distribution can be formed only by a simple operation of controlling the layer thickness, that is, the growth time. According to the present invention, it is not necessary to set the doping concentration of the p-type impurity to a particularly high concentration, and a very general doping concentration of about 10 ¹⁷ cm ⁻³ to about 10 10
^{A value of 20} cm ^-3 is sufficient.

FIG. 7 shows a p-type upper cladding layer (105) and a p-type contact layer (106) in an as-grown state, formed by laminating nitride semiconductor layers having a thickness satisfying the requirements of the present invention. SIMS in the depth direction of constituent elements and dopants for a laminate for blue LED applications, comprising:
Post the analysis results. The upper cladding layer (105) having a total layer thickness of 100 nm has a thickness of about 380 to about 390 n at room temperature.
photoluminescence (PL) in the near ultraviolet wavelength range
It was formed on a gallium-indium nitride light-emitting layer (104) emitting a spectrum. This upper cladding layer (10
5) is formed by laminating two laminated layers of aluminum-gallium mixed crystal doped with magnesium having a thickness of 50 nm each. The contact layer (106) is composed of two layers of magnesium-doped gallium nitride having a thickness of 50 nm each having a thickness of 50 nm on the upper cladding layer (105) made of the p-type aluminum nitride-gallium mixed crystal layer. The p-type gallium nitride layer was formed in an as-grown state with a total layer thickness of 100 nm. In the upper cladding layer (105) shown in FIG. 7, it is recognized that aluminum atoms are uniformly distributed while maintaining a substantially constant concentration. Also, the magnesium atom, which is a dopant,
Upper cladding layer (105) and contact layer (106)
It can be seen that they are almost evenly distributed within. That is,
According to the method of forming a p-type layer of the present invention, the surface concentration of the upper cladding layer (105) and the contact layer (106) does not cause a decrease in the atomic concentration.
The p-type layer is easily formed in the row state.

On the other hand, FIG. 8 shows one example of the multilayer means according to the present invention.
On the upper cladding layer (105) made of a p-type aluminum-gallium nitride mixed crystal layer formed at 100 ° C., a contact layer (106) made of magnesium-doped gallium nitride having a layer thickness of 100 nm formed by a conventional technique at the same temperature.
12 shows a SIMS analysis result of the laminate having the above structure. The light emitting layer (1) below the upper cladding layer (105)
04) is the same as the configuration shown in FIG. The atomic concentration of aluminum and magnesium in the aluminum / gallium nitride layer serving as the upper cladding layer (105) is substantially constant as described above, and the concentration of both atoms is in the thickness direction of the same layer (105). There is no tendency to decrease. However, the concentration of magnesium atoms in the gallium nitride contact layer (106), which is a conventional forming technique, decreases rapidly toward the surface at a surface layer portion of about 50 nm from the surface of the same layer (106). . Same layer (10
This p-type impurity (magnesium) in the surface layer of 6)
It is considered that the rapid decrease in the amount is further promoted by the evaporation and volatilization of magnesium impurities in a high-temperature film-forming environment in addition to the above-described causes. Whatever the main cause, if the p-type impurity which is electrically activated and becomes an acceptor impurity is too small, it is certain that it will hinder the formation of a p-type layer in an as-grown state. If such a situation is known to occur, the surface layer of the magnesium-doped gallium nitride contact layer (106) in which the p-type impurity has been rapidly reduced may be removed by a process technique such as a plasma etching method. In addition, the deep portion of the layer where the reduction of the p-type impurity has not occurred can be exposed on the surface.
However, this method requires a special step of etching and removing the surface layer portion in addition to the step of forming the nitride semiconductor layer doped with the p-type impurity. As a result, as compared with the method of the present invention in which the p-type layer is provided in the as-grown state, the method of forming the p-type nitride semiconductor layer includes a redundant and complicated process.

In the present invention, as-gr is formed by laminating thin layers.
In the case where a low resistance p-type layer is formed in the down state and a base layer is specially laid, it is recommended that a group III nitride semiconductor layer containing no indium as a constituent element be used as the base layer. . In particular, it is recommended to use a group III nitride semiconductor containing an element constituting the multilayer and containing indium as a constituent element as the underlayer. If the underlying layer is a nitride semiconductor layer containing an element constituting the layer to be overlaid, the adsorption and adhesion to the surface of the underlying layer will be more reliable. This results in a uniform concentration distribution of the constituent elements in the multi-layer constituent layer with high concentration and uniform in the depth direction. Also,
In the case of a group III nitride semiconductor layer not containing indium as a constituent element, damage to the indium-containing group III nitride light emitting layer can be suppressed in the underlayer due to the inconvenient thermal behavior of indium as described above. In the configuration example of the underlayer and the multi-layer constituent layer according to the present invention, the base layer is made of an aluminum nitride-gallium mixed crystal (Al _x Ga ₁ -xN: 0 ≦ x ≦ 1), and the multi-layer constituent layer is made of aluminum nitride.・ Gallium mixed crystal (A
_{l y Ga 1-y N:} 0 ≦ y ≦ 1) is. In this configuration example, a favorable result is obtained when the aluminum composition ratio (x and y) of the base layer and the multilayer constituent layer, particularly, the aluminum constituent layer to be bonded to the base layer is matched. Because, in the aluminum nitride-gallium mixed crystal, if the aluminum composition ratio is the same (here, x = y), the lattice constants of the layers match each other, and therefore, at the junction interface of both layers. This is because the occurrence of misfit dislocation is suppressed. In another combination example, the base layer is gallium nitride, and the multi-layer constituent layer is an aluminum nitride-gallium mixed crystal (Al _x Ga ₁ -xN: 0 ≦ x ≦
An example of 1) is given. The effect of maintaining the atomic concentration of the constituent elements in the overlying layer constant by the provision of the underlayer is also observed when the overlying layer composed of the aluminum nitride-gallium mixed crystal is formed on the aluminum nitride underlayer. .

When the underlayer is provided, the thickness of the underlayer is 5
It is limited to 0 nm or less. If the layer thickness is excessive, for example, the purpose of forming an aluminum nitride-gallium mixed crystal layer, the phenomenon that the surface layer portion is easily converted to gallium nitride is as follows.
It occurs irrespective of the underlying layer or the multi-layered layer. Therefore,
If the thickness of the underlayer made of a mixed crystal of aluminum nitride and gallium is excessively large, such as 50 nm or more, the area density of aluminum atoms exposed on the surface of the underlayer may be reduced. For example, a multilayer structure layer containing aluminum as one constituent element is made of an aluminum nitride-gallium mixed crystal layer having a thickness of 1 layer.
When formed on a base layer having a thickness of more than 00 nm, the concentration of aluminum atoms in the multilayer constituting layer cannot always be increased and made uniform in the depth direction, but merely decreases the concentration of aluminum atoms on the surface. FIG. 9 shows that the layer thickness is 50 n.
In the case where a multi-layered structure layer (116) made of a mixed crystal of aluminum nitride and gallium having a layer thickness of 50 nm is formed on a base layer (117) made of aluminum nitride having a thickness of m, aluminum atoms in the multi-layered structure layer (116) It is a figure showing a mode of distribution in the depth direction. Due to the presence of the base layer (117) containing aluminum, the aluminum atoms in the multilayer constituting layer (116) are distributed at a higher concentration and substantially uniformly without decreasing the concentration at the surface layer. It is obvious. In this manner, a multilayer constituent layer containing aluminum atoms at a higher concentration can be formed on the multilayer constituent layer in which aluminum atoms are present at a high concentration not only in the layer but also in the surface region. In other words, the presence of the underlayer is beneficial not only for the multilayer constituent layer laminated immediately thereon but also for the multilayer constituent layer newly formed on the multilayer constituent layer to provide a high concentration and substantially uniform distribution of the constituent elements. is there. The effect of the arrangement of the underlayer on the concentration and distribution of the elements inside the multilayer constituent layer is exerted not only for the constituent elements. For example, if the underlayer is doped with a high concentration of magnesium to be doped into the overlying layer, magnesium impurities are evaporated from the surface of the overlying layer when the overlying layer is grown at a high temperature on the underlayer. This is because magnesium sufficient to supplement the decrease in magnesium concentration in the same layer due to volatilization is supplied from the underlayer by diffusion.

In order to further facilitate the formation of a p-type layer as-grown, in addition to a p-type impurity which forms a relatively shallow level in the multi-layer structure layer, an n-type carrier which is not a shallow impurity is used. There is a means for coexisting a deep p-type impurity which forms a relatively deep level which can be electrically compensated. Shallow group II elements that form relatively shallow levels in group III nitride semiconductors include magnesium and beryllium. On the other hand, zinc (Zn) is a typical deep impurity. De such as zinc
Ep impurities can electrically compensate for n-type carriers in gallium nitride. In a situation where the residual donor impurities are reduced by the electric compensation, the shallow of magnesium or the like is reduced.
The coexistence of an appropriate p-type impurity can increase the concentration of the electrically activated acceptor. That is, since the residual donor concentration can be reduced by the action of deep impurities, it is possible to obtain a multilayer structure in which the acceptor exists at a relatively high concentration. Since such an overlying layer has a high acceptor concentration, it provides further superiority in obtaining a low-resistance p-type nitride semiconductor layer in an as-grown state. If the amount of the deep impurity to be added is equal to the residual donor concentration, it can theoretically be electrically compensated to form a high-resistance layer, but in order to achieve reliable compensation, it exceeds the residual donor concentration. It is preferable to dope a deep impurity. For example, when zinc is selected as a deep impurity, it is preferable to add zinc to about 10 ¹⁷ cm ^-3 to about 10 ²⁰ cm ^-3 in consideration of the residual donor concentration. Further, preferably, about 1 × 1
Preferably, zinc is doped in a concentration range from 0 ¹⁸ cm ⁻³ to about 5 × ¹⁹ cm ⁻³ .

The multilayer constituent layer and the underlayer according to the present invention may be formed by a general metal-organic chemical vapor deposition (MOCVD) method, a general vapor-phase growth (VPE) method of a halogen method or a hydride method, or a molecular method. It can be formed by using a line epitaxial (MBE) method or the like. In these vapor phase growth methods, the layer thickness can be easily controlled by controlling the film formation time. Conventionally, a general growth method is to form a nitride semiconductor layer doped with a p-type impurity continuously and collectively by switching the source of a group III element to be introduced into a growth system. In the present invention contrast, the initial growth state lateral growth is dominant in order to create for each multilayer structure layer between deposition time between zones of each layer constituting the multilayer structure, 6
The deposition of more than 0 seconds to adopt growth hand stage of growth interruption method of providing a time interval interruption. 60 seconds or more, growth temperature
Layer structure layer that has already grown
This has the effect of making the concentration distribution of the p-type impurity in the inside uniform. When doping a p-type impurity into the underlayer and the multilayer constituent layer, the dopant temporarily protrudes into the growth system at an early stage of film formation due to pressure fluctuations in the dopant supply piping system when the p-type dopant is introduced. In some cases. In the present invention, a phenomenon in which p-type impurities and constituent elements are easily taken in at the initial stage of film formation (growth) is found, and this is based on the phenomenon. Therefore, a change in flow rate hinders the uniformity of the concentration distribution of p-type impurities. If so, it is necessary to take measures such as better maintaining the pressure balance in the piping system. Other means of avoiding the accumulation of excessive p-type impurities during the early stages of growth include modulation doping.
e) Delta (δ) doping technique commonly used to grow modulation-doped semiconductor layers for field effect transistor (MODFET) applications (a method of instantaneously changing the supply amount in a delta function), or a similar method. There is also a method of adjusting the concentration distribution of the p-type impurity in the depth direction by utilizing the above method.

The method of forming a p-type layer according to the present invention can also be applied to forming a p-type clad layer for a light emitting portion having a quantum well structure. For example, examples of arranging a p-type nitride semiconductor layer formed by the multilayer method of the present invention include the following. (A) Gallium-indium nitride layer is formed in a well
A p-type layer disposed as a p-type cladding layer on a light emitting portion having a single or multiple quantum well structure using a gallium nitride layer as a barrier layer. (B) A gallium indium nitride layer is formed in a well.
A p-type layer disposed as a p-type cladding layer on a light-emitting portion having a single or multiple quantum well structure as a layer and a barrier layer made of an aluminum nitride-gallium mixed crystal layer. (C) In a single or multiple quantum well structure composed of a gallium indium well layer and an aluminum gallium mixed crystal barrier layer, in order to avoid a steep change between the forbidden bands bonded to the barrier layer. A p-type layer provided on a layer for adjusting a provided band gap;
These laminated structures are made of sapphire, silicon carbide (SiC)
A known material for film formation of a nitride compound semiconductor such as zinc or zinc oxide (ZnO) can be deposited as a substrate.
The substrate is made of a metal material such as hafnium (Hf), a III-V compound semiconductor crystal having a face-centered cubic lattice structure such as gallium arsenide (GaAs) or gallium phosphide (GaP), or an element such as silicon (Si). (Single) semiconductor crystal can also be used. The conductivity type of any of the semiconductor crystal substrates does not matter. The plane orientation and off-angle (off-angle) of the crystal plane forming the substrate surface
angle (mis-orientation angle) and the like, the degree of lattice mismatch with gallium nitride (GaN) is less than 0.5% with gallium nitride (GaN) within a range that can be appropriately selected by the parties in view of the growth method and growth conditions of the low-temperature buffer layer. Lithium (Li) and gallium (Ga) or lithium and aluminum (Al)
Lithium oxide (Li ₂ Ga) which is a composite oxide with
O ₃ ) and lithium aluminate (Li ₂ AlO ₃ ) can also be used as the substrate.

Further, as specific compound semiconductor devices according to the present invention, the following devices can be exemplified. (A) The carrier concentration is about 1 to 10 × 10 ¹⁷ cm ⁻³ , the layer thickness is about 20 to 80 nm, and the indium composition ratio is about 5% (0.
05) to about 10% (0.10), a gallium / indium nitride light emitting layer, and a carrier concentration of about 1 × 10 ¹⁶ cm ⁻³ to about 5 × 10 ¹⁸ formed by the multilayer method according to the present invention.
A gallium nitride-based light emitting diode comprising: a p-type layer in which a magnesium-doped p-type gallium nitride layer having a total layer thickness of about 5 nm to 15 nm is set to be cm ⁻³ . (A) The carrier concentration of an n-type dopant such as silicon (Si) or sulfur (S) is increased to about 1 to 80 × 10 ¹⁷ c
and m ^-3, a light emitting layer composed of n-type gallium nitride comprising grains of single crystals such as made of indium nitride or gallium indium nitride mixed crystal layer thickness is about 2 to 80 nm, a low carrier concentration than the light-emitting layer About 5% aluminum composition
A short-wavelength light emitting diode including a p-type layer in which a p-type aluminum-gallium nitride mixed crystal is formed by the formation method according to the present invention to about 20%. (C) Without intentionally adding a p-type impurity such as zinc or magnesium, the concentration of a carrier doped with an n-type dopant such as silicon (Si), tin (Sn), sulfur (S) or selenium (Se) is increased. A particle of about 0.1 to 10 × 10 ¹⁷ cm ⁻³ and a layer thickness of about 1 to 20 nm, including a single crystal made of indium nitride or a mixed crystal of gallium nitride and indium, and having a layer thickness of about 380 to 390 nm at room temperature PL in the near ultraviolet wavelength band
A light emitting layer or well layer made of gallium indium nitride, which emits light; and a p-type layer made of a p-type aluminum nitride-gallium mixed crystal or gallium nitride formed by the forming method according to the present invention. A short wavelength light emitting diode or a short wavelength laser diode comprising: (D) Indium composition ratio of about 2% (0.02) to 10%
In the gallium indium nitride layer to be (0.10), a light emitting layer (or well) containing a gallium indium nitride mixed crystal or a microcrystal mainly composed of indium nitride in the layer due to phase separation or a phenomenon similar thereto. Layers),
An intervening layer provided to avoid a sharp change in the band gap between a multi-layer made of gallium nitride or an aluminum nitride-gallium mixed crystal and a cladding layer or the like bonded to the barrier layer; ~ 50 × 10 ¹⁸ c
A p-type aluminum nitride-gallium mixed crystal layer doped with a p-type impurity such as zinc, magnesium, and beryllium having a layer thickness of about m- ³ and a thickness of about 5 to 50 nm, or a p-type gallium nitride p-type layer is provided. Short wavelength light emitting diode or laser diode.

[0029]

The p-type is formed by laminating a thin layer having a thickness of 50 nm or less.
The formation of the group III nitride semiconductor layer leads to a uniform concentration distribution of the constituent elements and the p-type impurities in the layer. In growing the thin layer in a divided manner, the arrangement of the group III nitride semiconductor layer containing the constituent elements of the thin layer as an underlayer makes the distribution of the constituent elements inside the overlying thin layer more uniform and evenly distributed. It is useful.

[0030]

【Example】

(Example 1) The details of the present invention were described in the as-grown state by p
An example in which a multi-layered p-type layer is formed by a multi-layered gallium nitride layer will be described. First, a substrate (1) made of (0001) plane (so-called C plane) sapphire (alumina single crystal)
01), normal pressure metal organic chemical vapor deposition (MOCVD)
As a result, a low-temperature buffer layer (102) made of undoped gallium nitride was grown at a temperature of 430 ° C. The low-temperature buffer layer (102) is made of trimethylgallium ((CH ₃ ) ₃ Ga: TMG) as a gallium (Ga) raw material and nitrogen (N).
The growth was carried out using ammonia (NH ₃ ) as a raw material. The gallium raw material is TMG maintained at 0 ° C. and 20 g / min.
The mixture was bubbled (foamed) with cc of hydrogen gas, and supplied by introducing hydrogen gas for bubbling accompanied with vapor of TMG into the growth system. The flow rate of ammonia is 1.
It was 0 l (liter). Hydrogen gas was used as the carrier gas, and the flow rate was 8.0 l / min. Layer thickness is about 7n
m.

On the low-temperature buffer layer (102), using the same material, an n-type gallium nitride layer (10) doped with silicon (Si) having a carrier concentration of 2 × 10 ¹⁸ cm ^-3.
3) was grown at a time without forming a thin layer.
TMG was maintained at 13 ° C., and the hydrogen flow rate for bubbling was 2
0 cc / min. The flow rate of ammonia gas is 6.0 l /
Set to minutes. The carrier gas has a volume mixing ratio of 1: 1.
A mixed gas of hydrogen and argon (Ar) was used, and the flow rate was 6.0 l / min. At the time of growing the n-type gallium nitride layer (103) at a temperature of 1100 ° C., about 1.2 ppm by volume of disilane (Si ₂ H) is contained in the growth reaction system.
₆ ) containing disilane-hydrogen (H ₂ ) mixed gas containing 5
cc. n-type gallium nitride layer (103)
Was about 3.2 μm.

After the growth of the n-type gallium nitride layer (103) is terminated by stopping the supply of the gallium source (TMG) into the growth system, the introduction of the nitrogen source into the system is continued for a while. (5 minutes) Wait. After waiting for 5 minutes without changing the growth temperature, the layer thickness is reduced to 100 nm.
(= 0.1 μm) In order to form a p-type gallium nitride layer (116) doped with magnesium by the split growth method, a first multilayer structure having a layer thickness of 50 nm is formed on the n-type gallium nitride layer (103). The formation of the layer (116-1) was started. The formation of the first multilayer constituent layer (116-1) involves the vapor of a gallium raw material accompanying a carrier gas composed of a hydrogen-argon mixed gas (volume mixing ratio = 1: 1) at a flow rate of 6.0 l / min. Started by adding hydrogen bubbling gas and nitrogen source. The conditions for supplying TMG and ammonia are as described above for the n-type gallium nitride layer (10
Same as 3). At the time of formation, magnesium (Mg) was added so as to have an atomic concentration of about 2 × 10 ²⁰ cm ⁻³ . Magnesium is bis-methylcyclopentadienyl magnesium (bis-((CH ₃ ) C
₅ H ₄ ) ₂ Mg). This magnesium source keeps the temperature sublimated at 30 ° C.
Hydrogen gas at a flow rate of 0 cc was added by accompanying the inside of the growth system. First multilayer constituent layer (116-1) in which the supply of TMG, ammonia raw material and magnesium source to the growth reaction system by adding to the carrier gas is continued for exactly 7.5 minutes and the layer thickness is 50 nm. Was formed. The formation of the first multilayer component (116-1) is stopped by stopping the introduction of the gallium raw material into the growth system.
The formation of 1) was completed. The supply of the gallium raw material was stopped at the same time as the supply of the magnesium source was stopped, so that only the ammonia gas carried by the hydrogen-argon mixed gas was supplied into the growth system.

After 3 minutes have passed in this situation, the supply of the source gas and the magnesium source into the growth system is resumed, and the magnesium doped under the same growth conditions as the first multi-layered layer (116-1). 2 multi-layered constituent layers (116
-2) was laminated on the first multilayer component layer (116-1). The time interval is provided between the formation of the first and second multilayer constituent layers, because the growth is interrupted each time the multilayer constituent layer is grown, so that each of the growth layers is renewed in the horizontal (horizontal) direction. This is to provide an early growth stage in which growth becomes dominant. As described above, the first and second growth layers were stacked by the formation method of the present invention to form a magnesium-doped gallium nitride layer (116) having a total thickness of 100 nm. The obtained laminate was naturally cooled to a temperature near room temperature.

General transmission electron microscope (abbreviation: TEM)
, The cross-sectional structure of the laminate was observed. Cross section T imaged
From the EM image, the low-temperature buffer layer (1
02) was confirmed to be composed mainly of a single crystal layer. In particular, it was clearly determined that the area near the surface of the sapphire substrate (101) was covered with the single crystal layer.
Low temperature buffer layer (102) and n-type gallium nitride layer (103)
Dislocations that appear to originate from the interface with
It propagated in the n-type gallium nitride layer (103) in a direction substantially perpendicular to the surface (C-plane) of the substrate (101). However, most of the dislocations escaped in the horizontal direction in a substantially central region of the n-type gallium nitride layer (103). For this reason, the surface portion of the n-type gallium nitride layer (103) has a density of about 10 ⁴ cm in comparison with the interface region between the low-temperature buffer layer (102) and the n-type gallium nitride layer (103).
It was an area with few dislocations, about ^-2 . The interface between the n-type gallium nitride layer (103) and the first multilayer component layer (116-1) overlaid thereon, and the first and second multilayer component layers ((116-1) and (116- At the interface with 2)), a very narrow linear black contrast of less than 10 nm was observed on the TEM image. This contrast was interpreted to be due to the crystal strain in the region near the interface inside each multilayer constituent layer. Further, it was not recognized that new dislocations were particularly remarkably generated inside the first and second multilayer constituent layers ((116-1) and (116-2)).

FIG. 10 is a SIMS analysis result showing the concentration distribution in the depth direction of the magnesium (Mg) dopant in the above-mentioned laminate. The atomic concentration of magnesium inside the two stacked constituent layers was about 1.7 × 10 ²⁰ cm ⁻³ . In addition, magnesium atoms are present in the vicinity of the interface between the n-type gallium nitride layer (103) and the first multilayer constituent layer (116-1) and in the first and second multilayer constituent layers ((116-1)
And (116-2)) in the region near the interface, which was present at a slightly high concentration. However, when they were averaged in the layer thickness direction, they were distributed almost uniformly in the depth direction of the multilayer component layer (see FIG. 10).

The second multi-layered structure layer (116) is formed from magnesium-doped gallium nitride forming the surface layer of the laminate.
When the surface concentration in -2) was measured by a general electrolytic CV method, the hole (hole) concentration was set to about 7 × 10 ¹⁷ cm ^-3.
It was measured as a stratum. The surface resistance was as low as 20 milliohm-cm (mΩ · cm). The surface of the stacked body was subjected to patterning to expose the surface of the n-type gallium nitride layer (103) in a partial region by using a normal photolithography technique. Next, the p-type multilayer constituent layers ((116-1) and (116-1) in the patterned region
-2)) argon (Ar) / hydrogen / methane (CH ₄₎
The n-type gallium nitride layer (103) was removed by plasma etching using a mixed gas until the surface portion was exposed. Thereafter, aluminum (Al) is applied by a normal vacuum deposition method to a part of the region on the second multilayer constituent layer (116-2) and on the exposed n-type gallium nitride layer (103). A p-type electrode (positive electrode) (107) and an n-type electrode (negative electrode) (108) were formed. FIG. 11 shows the electrodes ((107)
And (108)) show a schematic cross-sectional view of the laminate after installation. The current-voltage (IV) characteristic between both electrodes is shown in FIG.
Rectification as shown in FIG. 2 was observed.

(Comparative Example 1) According to the conditions of Example 1 above,
A low-temperature gallium nitride buffer layer and an n-type gallium nitride layer were sequentially laminated on the surface of the C-plane sapphire substrate. On the n-type gallium nitride layer (103), a magnesium-doped gallium nitride layer (118) having a layer thickness of 100 nm was deposited by a conventional batch growth method. That is, the magnesium-doped gallium nitride layer was formed collectively instead of the divisional growth technique of stacking thin layers described in Example 1. FIG.
Third, an n-type gallium nitride layer (10
3) shows the concentration distribution in the depth direction of magnesium atoms inside the magnesium-doped gallium nitride layer (118) deposited thereon. This concentration distribution is obtained by a general SIMS analysis method. As shown in the figure, the distribution of the atomic concentration of magnesium was clearly non-uniform, and was found to decrease sharply toward the surface.

According to the measurement result of the Hall effect method, it was determined that the magnesium-doped gallium nitride layer (118) formed by the batch growth method was n-type as a whole. In particular, the surface layer was determined to be a high resistance layer because the current flow in the stylus method using a tungsten (W) probe was less than 10 μA with respect to an applied voltage of 40 V between the probes. . In addition, even in consideration of the measurement results by the electrolytic CV method, the gallium nitride layer grown by the conventional method and having a thickness of 100 nm doped with magnesium is recognized as a p-type layer at least in the as-grown state. Did not. Incidentally, when the laminate according to this comparative example was heat-treated at 800 ° C. for 20 minutes in an argon (Ar) stream, a magnesium-doped gallium nitride layer (1
In 18), it was confirmed that conductivity was slightly imparted as compared with the as-grown state. However, it was not recognized that the resistance was lower than that of the magnesium-doped gallium nitride layer described in Example 1, and it was determined that the surface resistance exceeded about 1 kohm (kΩ). The same number as that of the magnesium-doped p-type gallium nitride layer formed by the split growth method described in the first embodiment is about several to several tens of milliohm-cm (m).
(.OMEGA..cm) was measured on the surface where the surface layer was removed over a thickness of about 50 nm corresponding to about 1/2 of the layer thickness.

(Example 2) Sapphire (α-Al ₂ O) having a diameter of 1 inch and a thickness of about 90 μm obtained by subjecting both surfaces (front and back) to mirror polishing using diamond abrasive grains having an average particle diameter of about 3 μm. ₃ single crystals) was used as the substrate (101). The plane orientation of the surface is (0001) plane (so-called C
Plane). The mirror-polished surface of the substrate (101) is first degreased with an organic solvent such as acetone, and then washed with ultrapure water having a specific resistance of about 18 megaohms (MΩ). Acid cleaning was performed using a high-purity aqueous solution of hydrofluoric acid (HF). Thereafter, after washing with ultrapure water again, the treated surface was irradiated with infrared rays radiated from an infrared lamp to be heated and dried.

In an environment where the atmosphere is shut off through a general interlock mechanism having a dustproof gate valve, the substrate is cleaned to a predetermined position on a substrate support (susceptor) arranged in an atmospheric pressure type MOCVD growth furnace. Sapphire substrate (10
1) Placed with one main surface facing up. Palladium (Pd)
Hydrogen (H ₂ ) gas purified to a high purity by using both the permeable membrane method and the cryogenic adsorption method was passed through the growth furnace to create an atmosphere composed of hydrogen gas. The pressure in the growth furnace was maintained at approximately atmospheric pressure. Fifteen minutes after the start of the flow of the hydrogen gas into the reaction furnace, the substrate support was covered, and the power was turned on to a resistance heating heater having a ceramic heater as an element, and the temperature of the substrate (101) was reduced. Room temperature to 1150 ° C
The temperature rose. At the same temperature, the hydrogen gas was held for 40 minutes in a state where the hydrogen gas was continuously flowed through one main surface of the substrate (101).
Well-known thermal etching was performed on the substrate surface.

Next, the amount of electric power supplied to the above-described resistance heater was reduced to lower the temperature of the substrate (101) to 420 ° C. After dropping to 420 ° C., it waited for about 25 minutes until the temperature became stable. During this time, the flow rate of hydrogen gas supplied to the reactor was adjusted to 8 liters per minute. After that, MO
Ammonia (NH ₃ ) gas as a nitrogen (N) source is supplied at a flow rate of 1 liter per minute toward the surface of the substrate placed in the CVD reactor, and the growth atmosphere is hydrogen: ammonia flow ratio = 8: 1. And a hydrogen atmosphere containing a nitrogen source. After that, while supplying ammonia gas (nitrogen source) into the MOCVD reactor, supply of gallium (Ga) source is continued for 20 minutes, and undoped gallium nitride (GaN) having a layer thickness of 15 nm is used. A low temperature buffer layer (102) was grown. The gallium source includes trimethylgallium ((CH ₃ ) ₃ Ga: TM) for semiconductor industry.
G) was used. The supply rate of TMG into the reactor is 2 per minute
× 10 ^-6 mol (mol.). Therefore, the so-called V at the time of forming the gallium nitride (GaN) low-temperature buffer layer is
/ III ratio (ammonia / TMG flow ratio) is about 2.2 × 1
0 ⁴ after. Observation by the cross-sectional TEM technique revealed that the surface of the substrate (101) was covered with several single-crystal low-temperature buffer layers. The growth of the low-temperature buffer layer (102) was terminated by stopping the supply of TMG as a gallium source to the growth reaction system at substantially atmospheric pressure.

Thereafter, the gas species supplied to the reactor was converted from hydrogen to argon gas. Argon gas flow rate of 3
The temperature of the substrate (101) was set to 1100 ° C. under the flow rate conditions of 1 liter / minute and the ammonia gas flow rate of 1 liter / minute.
Was raised. When the temperature of the substrate (101) reached 1100 ° C., the supply of hydrogen gas to the reactor was restarted at a flow rate of 3 liters per minute. In parallel, the supply of ammonia gas to the reactor was increased to 6 liters per minute. When the short-term fluctuation of the temperature of the substrate (101) due to the increase in the supply amount of the hydrogen gas or the ammonia gas was eliminated, the supply of the gallium source to the reactor was restarted. The vapor of the gallium source was kept at 13 ° C. and liquid TMG was bubbled (foamed) with 20 cc of hydrogen gas per minute to accompany the reactor. A doping gas of silicon (Si) was supplied to the reactor in synchronization with the addition of the hydrogen bubbling gas accompanying the vapor of the gallium source to the reactor. The doping gas of silicon has a volume concentration of about 5 ppm
Hydrogen diluted disilane (Si ₂ H ₆ ) gas was used. The flow rate of the disilane gas was 10 cc / min. The supply of a gallium source or the like was continued for 90 minutes to grow an n-type gallium nitride layer (103) having a carrier concentration of about 3 × 10 ¹⁸ cm ⁻³ and a layer thickness of about 3.2 μm.

Next, the temperature of the substrate (101) is set to 1100 ° C.
From about 830 ° C., which is the growth temperature of the light emitting layer (104).
Dropped in 0 minutes. While lowering the substrate temperature, supply of hydrogen gas as a carrier gas to the reaction furnace was stopped. Under the conditions of a flow rate of ammonia gas of 6 liters per minute and a flow rate of argon gas of 3 liters per minute, the supply of the gallium source and the new indium source to the reactor was started. As the indium source, cyclopentadienyl indium (C ₅ H ₅ In) having a monovalent valence is used.
(I)) was used. The storage container made of stainless steel containing the indium source was kept at 65 ° C. by an electronic thermostat utilizing the Peltier effect. The vapor of the indium source generated by the sublimation was entrained in the reactor which was maintained at approximately atmospheric pressure with hydrogen gas at a flow rate of 120 cc / min. A hydrogen gas at a flow rate of 3.2 cc / min was passed through a stainless steel foam container (bubbler) containing the above-mentioned gallium source, which was maintained at 0 ° C. by an electronic thermostat, in order to accompany the vapor of TMG. . Total supply of gallium and indium (II
The ratio of the supply amount (mole number) of indium to the supply amount (mole number) of the group I element, that is, the gas phase composition ratio of indium was calculated to be 0.5. The supply of the gallium source and the indium source is continued for 7.5 minutes, silicon (Si) having an indium composition ratio (solid phase composition ratio) of 6% (0.06) and a layer thickness of about 30 nm is produced. A light emitting layer (104) made of doped gallium indium nitride was formed. Silicon was doped using the above-mentioned disilane (volume concentration = 5 ppm) -hydrogen mixed gas. The flow rate of the mixed gas was set at 10 cc / min. The indium composition ratio is
Based on the fact that the wavelength position of the near-ultraviolet spectrum emitted from a trial sample of a gallium-indium nitride crystal obtained under the same growth conditions is about 384 nm, the already disclosed gallium-indium nitride mixed crystal has been disclosed. Between forbidden band width and indium composition ratio in Japan (Japanese Patent Publication No. 55-3)
834). The thickness of the light-emitting layer (104) was measured from a bright-field cross-sectional TEM (transmission electron microscope) image.

After the supply of the indium source to the reaction furnace was stopped to complete the formation of the gallium / indium mixed crystal light emitting layer (104), the flow rates of the gallium source, the nitrogen source and the argon gas were kept constant. At 830 ° C. to form a first multi-layer constituent layer (116-1) made of gallium nitride.
That is, only the supply of the indium source is stopped, and the supply of the other source gases to the reaction furnace is continued, and the first multilayer constituent layer (116-) made of gallium nitride is formed at the same temperature as that of the light emitting layer (104). 1) was grown. Magnesium was doped as a p-type impurity. As the doping gas of magnesium, bismethylcyclopentadienylmagnesium was used as in Example 1, and the flow conditions were the same as in Example 1. The flow of the source gas and the magnesium source into the growth system was continued for exactly 3 minutes to form a magnesium-doped gallium nitride layer (116-1) (first multilayer constituent layer) having a layer thickness of 10 nm.

After completing the formation of the first multilayer constitution layer (116-1), the temperature of the substrate (101) was raised to 1100 ° C. again. The hydrogen gas supply was resumed at a flow rate of 3 liters / minute into the growth furnace while waiting for a while until the temperature hunting (fluctuation) disappeared. After about 2 minutes after reaching 1100 ° C., it was recognized that the temperature fluctuation became minute, and then a gallium source was added again into the growth furnace, and the first multilayer constituent layer (116-1) was added. The formation of the second multilayer constituent layer (116-2) was started. TMG was kept at 13 ° C. by an electronic thermostat. The flow rate of hydrogen gas for bubbling with TMG vapor was 8.0 cc / min. On the other hand, the magnesium source was kept at 30 ° C., and the flow rate of hydrogen gas accompanying the sublimation vapor of the source was 40 cc / min. The supply of the magnesium source and the gallium raw material was continued for 7.5 minutes to form a second multilayer constituent layer (116-2) made of magnesium-doped gallium nitride having a layer thickness of 50 nm. First and second multilayer constituent layers ((116-
1) and (116-2)) are separately grown.
A multilayer p-type layer composed of a magnesium-doped gallium nitride layer having a total layer thickness of 60 nm was formed. After completion of the growth of the same layer, cooling was started with only the gas species flowing into the growth system at a flow rate of 3 liters per minute of argon and ammonia gas at a flow rate of 1 liter per minute. In the course of cooling, natural cooling was performed to a temperature near room temperature without maintaining a certain temperature for a certain period of time or adjusting the cooling rate between specific temperature zones.

FIG. 14 shows the concentration distribution of magnesium (Mg) in the depth direction of the obtained laminate. Magnesium was found to be uniformly distributed at a substantially uniform concentration inside the first and second multilayer constituent layers without reducing its concentration even in the surface portion of the second multilayer constituent layer (116-2). Was done. FIG. 15 shows a schematic view of a cross-sectional TEM image of the laminate.
It is rare that the interface between the layers constituting the multilayered structure is clearly recognized even by observation with an electron microscope, and the position of the interface could not be specified for all samples. For this reason, observation was repeated using a sample which was carefully and precisely thinned to a layer thickness capable of enhancing the contrast when capturing a cross-sectional TEM image. According to this, the first and second multilayer constituent layers ((116-1) and (116-
2)), in addition to crystal defects such as dislocations (121) propagated from the underlying n-type gallium nitride layer (not shown), newly generated and caused by dislocations, stacking faults, and the like. No contrast was observed.

When the surface carrier concentration of the multilayer p-type layer comprising the gallium nitride layer doped with magnesium was measured by the electrolytic CV method, the hole concentration was found to be about 8 × 10 ¹⁷ c.
It was shown to be a p-type layer with m- ³ . That is, this result showed that a low-resistance p-type layer having a high carrier concentration and a low resistance was formed as-grown. Further, this laminate was subjected to a known patterning process and an etching process for forming a light emitting diode, and a chip having a planar shape as shown in FIG. 16 was prototyped. Using this chip as a sample, pn
EBIC, the usual method for confirming the presence of a junction
The region where the pn junction was formed was confirmed by the (electron beam excitation current) method. As a result, an EBIC image was obtained from the region where the p-type electrode (107) made of silver (Ag) was laid. From this, it was confirmed that the pn junction existed almost entirely under the p-type electrode (107). When a forward current of 20 milliamperes (mA) was passed between the electrodes (107) and (108), almost the entire area immediately below the p-type electrode (107) emitted blue light. From the viewpoint of the current-voltage characteristics between the two electrodes, the leakage current component of the reverse current was as small as less than 5 μA at −10 V. As a result, an additional step such as an annealing step is not required after the film is formed on the nitride semiconductor layer doped with the p-type impurity.
-It has been found that a low resistance p-type layer which provides good pn junction properties in the grown state is provided. The forward voltage was about 5 to about 6V. N-type electrode (108)
After increasing the thickness of the aluminum electrode constituting the above to make the height substantially equal to the height of the p-type electrode (107), the chip is adhered to the base (stem) on the electrode side by the flip-flop method, and the substrate polished on both sides (101) was mounted so that light emission was emitted from the back side to the outside. Then, the outer periphery of the mounted chip is molded with an envelope mainly made of transparent epoxy resin for semiconductor encapsulation with a condenser lens that emits light emitted from the back side of the substrate at an orientation angle of about 45 degrees. did. The main emission wavelength measured using a general spectrum analyzer capable of analyzing the emission spectrum was about 445 nm. In addition, about 3
Emission in the near-ultraviolet band of 85 nm and emission in a yellow band centered at about 570 nm with a wide half width were confirmed. The emission output was about 0.3 milliwatt. As described above, it was determined that the p-type layer in the as-grown state formed by the method of the present invention has sufficient properties for a short-wavelength visible LED.

(Comparative Example 2) Double-side polished (0001)-
A gallium nitride low-temperature buffer layer, a silicon-doped n-type gallium nitride layer, and a silicon-doped n-type gallium-indium nitride light-emitting layer were sequentially laminated on a sapphire substrate in the same manner as in Example 2. After the growth of the light emitting layer was completed, the layer thickness was reduced to 1 at 830 ° C., which is the growth temperature of the light emitting layer, under the conditions described in Example 2.
A gallium nitride layer doped with magnesium to a thickness of 0 nm was grown.

Thereafter, the temperature of the substrate was raised to 1100 ° C., and the system was temporarily waited until the substrate temperature was stabilized. Thereafter, the flow condition of Example 2 is followed, and only the growth time is tripled to 2 times.
Extending to 2.5 minutes, a magnesium-doped gallium nitride layer having a layer thickness of 150 nm was grown.
This layer thickness is as thick as about 1.5 times the prescribed layer thickness of the multi-layer constituting layer according to the present invention. That is, the second embodiment
Is that the layer thickness of the magnesium-doped gallium nitride layer constituting the outermost layer is remarkably different.

When the distribution of magnesium atoms in the depth direction of the magnesium in the magnesium-doped gallium nitride layer was confirmed by SIMS analysis, the atomic concentration of magnesium decreased sharply from the depth of the layer toward the surface of the laminate. SIM
On S analysis, about 15n from the surface not strongly affected by the surface
The atomic concentration of magnesium at a depth of m is about 3 × 10
¹⁷ cm ^-3 . Incidentally, the atomic concentration of magnesium deep in the same layer was about 1.5 × 10 ²⁰ cm ⁻³ . The electrolytic CV method determined that the surface region of the same layer had high resistance, as corresponding to the decrease in the atomic concentration of magnesium at the surface. In a normal Hall (Hall) effect measurement method based on the Van der Pauw method using indium as an electrode, an electrode exhibiting a sufficiently good ohmic property is not obtained due to a high input resistance. Large variation in value, n-type
It was not possible to accurately determine whether it was a p-type or a p-type.

Further, from the bright-field transmission electron microscope image taken by the cross-sectional TEM method, the state of occurrence of crystal defects inside the obtained laminate, particularly the outermost magnesium-doped gallium nitride layer having a large thickness, is shown. Got information. As a result, it was recognized that a large number of stacking faults occurred in a region where the thickness of the innermost magnesium-doped gallium nitride layer exceeded about 100 nm. Dislocations (actually imaged as a black linear contrast) that seem to start from a region where stacking faults occur at a high density were dense in the region from the surface to the surface. . That is, it has been taught that it is necessary to define a preferable layer thickness in view of the occurrence of crystal defects in addition to the decrease in the concentration of the p-type impurity in the nitride semiconductor layer doped with the p-type impurity.

Using the laminate in the as-grown state obtained in this comparative example, an LED similar to that described in Example 2 was produced (see FIG. 16). However, in the IV characteristics, the rectifying property indicating the formation of a normal pn junction is not exhibited, and when a high voltage of about 40 V is applied between the electrodes, the current flowing in the forward direction is at most less than 1 to 2 mA. It was small. This was because the surface layer of the gallium nitride layer doped with the p-type impurity according to Comparative Example 2 had high resistance. At such a low current, light is emitted in the region where the p-type electrode is laid, but the emission color is white and the emission intensity is weak, and blue emission can be obtained at several points in the region. . This region emitting blue light was probably a region where dislocations were dense, and it was presumed that a large amount of current selectively flowed through the dislocations. In any case, p
When the thickness of the nitride semiconductor layer doped with the impurity is increased, no contribution is made to the improvement of the emission intensity of the LED.

In the laminate of the present comparative example, the magnesium-doped gallium nitride layer, which is the outermost layer, was placed about 1 mm from the surface.
And removed over a thickness of about 50 nm.
After measuring the carrier concentration on the surface by the electrolytic CV method, the value was about 5 × 10 ¹⁷ cm ⁻³ . Also, in consideration of the results of the Hall effect measurement and the like, the remaining layer was determined to be a p-type conductive layer. Thus, after removing the surface layer portion of the magnesium-doped gallium nitride layer of the outermost layer, the LED manufactured on the surface by the above method has the same rectifying property as described in Example 2. I-
V characteristics were obtained. This means that if the layer thickness of the magnesium-doped gallium nitride layer does not exceed the layer thickness specified in the present invention, a nitride semiconductor layer exhibiting p-type conductivity is easily formed. I was But,
After collectively growing a nitride semiconductor layer doped with a p-type impurity having a large thickness exceeding the specification of the present invention, measures to reduce the layer thickness to the thickness specified by the present invention by means such as etching are as follows. Since an additional post-process is required, it has been determined that the method cannot be more convenient than the forming method of the present invention in view of the complexity of the process or the economical efficiency in the production of the nitride semiconductor device. .

(Embodiment 3) According to the growth conditions of the above-described Embodiment 2, a vertical normal pressure M was placed on a sapphire C-plane substrate (101).
The gallium nitride low-temperature buffer layer (10
2), a silicon-doped n-type gallium nitride layer (103), and a silicon-doped n-type gallium indium nitride light-emitting layer (10
4) was sequentially deposited. Thereafter, at a growth temperature of 830 ° C., an undoped gallium nitride layer having a layer thickness of 10 nm was deposited as a base layer (117) at a growth temperature of 830 ° C. by the method described in Example 2. After the deposition of the same layer is completed, the temperature of the substrate (101) is increased from 830 ° C. to 1100 ° C.
The temperature rose.

After the temperature is raised, after waiting for a while until the temperature is stabilized (about 5 minutes), the carrier is formed of a hydrogen-argon mixed gas having a flow ratio of 1: 1 and a total flow of 6 liter (l) / min. A hydrogen bubbling gas accompanied by an aluminum source was added to the (carrier) gas. Aluminum source is trimethyl aluminum ((CH ₃ ) ₃ Al: general name TMA)
It was used. TMA was kept at a constant temperature of 50 ° C. by an electronic thermostat. Bubbling (foaming) the liquefied TMA,
The flow rate of the hydrogen gas accompanying the vapor was 50 cc / min. At the same time as adding the aluminum source to the carrier gas,
The magnesium source (bis-((CH ₃
C ₅ H ₄ ) ₂ Mg) was introduced into the growth reaction system under the same flow conditions as in Example 2. The introduction amount of the magnesium source was set so that the atomic concentration became about 2 × 10 ²⁰ cm ⁻³ . The introduction (supply) of the aluminum source and the magnesium source is continued for exactly 5 minutes, so that the first multilayer constituent layer (1) made of magnesium-doped aluminum nitride (AlN) is formed.
16-1) was formed on the underlayer (117). With the flow rate of ammonia as the nitrogen source maintained at 6 l / min, the supply of the aluminum source and the magnesium source was temporarily interrupted to terminate the formation of the aluminum nitride layer. After the growth of the aluminum nitride layer was completed, only the carrier gas and the ammonia gas were introduced into the growth reaction system.
After the growth was interrupted for exactly 3 minutes, the supply of TMA and the magnesium source and the growth system was started again, and the supply of TMG was started again, and the second multilayer constituent layer (116-2) having a layer thickness of 20 nm was started. ) Was formed on the first multi-layer structure layer. Since the aluminum composition ratio of the mixed crystal was intended to be 10% (0.10), the supply (addition) flow rate of the aluminum source was set to 1/10 of that in the case of forming the above-mentioned first multilayer constituent layer. Again, after the supply of the group III raw material and the magnesium source is interrupted and the formation of the second multilayer constituent layer (116-2) is completed, the second multilayer constituent layer (116 A third multilayer constituent layer (116-3) having the same aluminum composition and layer thickness as in -2) was formed on the second multilayer constituent layer (116-2). From this, the three multilayer constituent layers (116-1 to 116-1)
A p-type upper cladding layer (105) made of an aluminum-gallium nitride mixed crystal layer with a total layer thickness of 50 nm doped with magnesium consisting of 116-3)
Was formed.

The formation of the third multilayer constituent layer (116-3)
The present invention provides a magnesium (Mg) -doped gallium nitride (GaN) layer (116) having a layer thickness of 30 nm after terminating by stopping the supply of the group III raw material and suspending the growth for exactly 3 minutes. In order to form the first multilayer constituent layer (11) made of gallium nitride,
6-1) was formed. The gallium source and the magnesium source were the same as those used in the formation of the second and third aluminum nitride-gallium mixed crystal layers, and the supply flow rates were also the same. The formation of the first gallium nitride layer was interrupted by stopping the supply of the gallium source into the growth reaction system. At the same time, the supply of the magnesium source was stopped at the same time. Thereafter, the substrate temperature is maintained at 1100 ° C.
After suspending the growth for exactly 3 minutes while keeping the composition ratio and the flow rate of the carrier gas and the flow rate of the ammonia gas the same as above, the gallium source and the gallium source are again introduced into the growth reaction system maintained at approximately atmospheric pressure. A magnesium source was started to be added (introduced) to form a second gallium nitride multilayer constituent layer (116-2) doped with magnesium. Second
Has the same thickness as the first gallium nitride multilayer constituent layer (116-1).
It was set to 0 nm. The second gallium nitride multilayer constituent layer (116
At the time of forming -2), the doping amount of the magnesium source was changed to the first gallium nitride layer (11) because of fear of the magnesium evaporating from the growth surface at a high temperature (1100 ° C.).
The value was twice that in the case of forming 6-2). That is, 3
The flow rate of hydrogen gas accompanying the sublimation vapor of bismethylcyclopentadienyl magnesium maintained at a constant temperature of 0 ° C. was set to 80 cc / min. The first and second gallium nitride layers were overlaid to form a gallium nitride contact layer (106) by a split growth method. Immediately after the completion of the formation of the second gallium nitride layer, the cooling was started, and the mixture was allowed to cool naturally until the temperature dropped to around room temperature. As described above, the characteristics of the means for forming the laminate of this embodiment can be summarized as follows. Is that the layer doped with the p-type impurity disposed on the light-emitting layer is formed not as a batch but as a divided layer.

According to the SIMS analysis, aluminum and magnesium are distributed almost uniformly in each of the first to third aluminum / gallium mixed crystal layers and the first and second gallium multi-layered constituent layers. Was recognized. The atomic concentration of magnesium is highest inside the second gallium nitride layer, and its atomic concentration is 3.2 × 10 ²⁰ cm.
It was measured as ^-3 . Also, no decrease in the concentration of magnesium atoms on the surface of the outermost layer was observed. On the other hand, the carrier concentration near the surface of the second gallium nitride layer is about 6 × 10 ¹⁷
cm ⁻³ and the surface of the layer was determined to be p-type. By the general step etching method, the multilayer constituent layer is sequentially etched and removed, and the carrier concentration and the conductivity type of the lower multilayer constituent layer are sequentially measured and determined. As a result, the first gallium nitride layer is p-type, Carrier concentration is about 2
× 10 ¹⁷ cm ^-3 , second and third aluminum nitride
Each gallium mixed crystal layer also has a carrier concentration of about 5 × 10 ¹⁶
The p-type layer had a ^size of cm ⁻³ . The carrier concentration and the conductivity type of the aluminum nitride layer, which is the first multilayer constituent layer, are determined by electrolytic C
Judgment by the -V method or the Hall effect measurement method does not allow accurate measurement due to the small thickness of the layer. Magnesium-doped nitride with a 50 nm layer thickness separately deposited on gallium indium nitride for measurement An aluminum layer was used. With respect to this layer, the results of the electrolytic CV method and the Hall effect measurement method are summarized as follows.
It was determined to be a p-type layer having a size of × 10 ¹⁶ cm ^-3 .

A known patterning technique for the laminate
The LED was constructed using a plasma etching technique or the like. FIG. 17 shows a schematic sectional view of the manufactured LED. The layer thickness of the n-type electrode (108) was substantially equal to the total layer thickness of the etching depth by the above-described plasma etching and the layer thickness of the p-type electrode (107). The thickness of the silver (Ag) film of the p-type electrode (107) was about 1 μm. This thickness was sufficient to shield light emitted from the light emitting layer (104) from the contact layer (106) side. That is, a metal film having a thickness sufficient to prevent the light-emitting layer (104) from being taken out to the outside via the contact layer (106) was disposed as the p-type electrode (107). Thereafter, the surface on which the electrodes are formed is applied for about 200 n using a general plasma CVD method.
An m-thick silicon nitride (SiN) film (120) was deposited. Thereafter, n-type and p-type electrodes ((107) and (10)
8)) The silicon nitride protective film deposited on the region above was selectively removed with buffered hydrofluoric acid to expose the surfaces of both electrodes.

After the so-called "window opening" of the protective film for exposing the electrode surface, the diamond scriber is repeated several times along concave scribe lines (scribe grooves) provided in a grid around each element. Square chip (ch having a side of about 350 μm)
ip). The chip was mounted on a very common pedestal for LED chips. The mounting method is a so-called flip-flop method, in which the above-mentioned two electrodes are arranged to face the electrode energized portion of the pedestal and are electrically connected to the respective electrodes of the pedestal. That is, instead of taking out light emission from the light emitting layer from the contact layer side, light emission is taken out from the back surface side of the sapphire substrate that has been subjected to double-side polishing, and a surface mounting mounting method different from the conventional one is used. After sealing with a general semiconductor element sealing epoxy resin, 10 LEDs manufactured from each of the above-described laminates were arbitrarily sampled, and luminescence characteristics and the like were measured. Table 1 summarizes and compares the characteristic values of the present embodiment and the same number of conventional products (currently commercially available LEDs). The LED of the present embodiment is a LE of a type in which light is emitted from the back side of the substrate.
D. The conventional product is an LED provided with a p-type aluminum-gallium nitride mixed crystal layer and a p-type gallium nitride layer grown on the light-emitting layer at one time. The emission wavelength and half-wave value in Table 1 were measured at a forward current of 20 mA, and the emission output was measured by applying a voltage of 4 V in the forward direction.

As shown in Table 1, it is judged that the LED provided with the p-type nitride semiconductor layer formed according to the present invention has excellent uniformity in light emission characteristics. In particular, for an LED having a p-type nitride semiconductor layer formed according to the present invention,
It was also characterized in that the forward voltage of the LED described in the comparative example was about 10 V or more, whereas the forward voltage was about 3 V as a whole. This is mainly due to the fact that the carrier concentration of the p-type layer formed according to the present invention is almost uniform in the layer, and particularly in the region near the surface of the p-type layer, the high resistance region due to the decrease in the concentration of the p-type impurity. It was determined that the contact resistance of the p-type electrode was reduced because no was formed. That is, when the forward voltage is low, the current flowing when the same voltage is applied (4 V in the present embodiment) is increased, which is interpreted as causing an increase in light emission output.

The ten L values for which the characteristic values listed in Table 1 were measured
Not only the ED, but also the other 100 LEs produced in Example 3
D was sampled arbitrarily to evaluate the uniformity of the characteristics. For example, regarding the uniformity of the forward voltage (so-called Vf), about two out of 100 LEDs have a forward voltage of two.
There was an abnormally low LED below volts, but other LE
The forward voltage of D was about 3V. It was about 2.9 V as indicated by the central value of the distribution. The distribution width of the light emission output was approximately 1.2 to 1.4 mW, and there were also about two LEDs having a low light emission output (less than 0.5 mW). The center value of the light emission output was about 1.3 mW. The LED with abnormal forward voltage and the LED with low light output were not the same. In any case, the method of forming a p-type nitride semiconductor layer according to the present invention has a ripple effect that results in an LED having substantially uniform characteristics.

[0062]

[Table 1]

[0063]

According to the present invention, a low-resistance p-type nitride semiconductor layer can be formed in an as-grown state without providing a dedicated process for p-type formation. Further, the method for forming a p-type nitride compound semiconductor layer of the present invention has a ripple effect that can provide a light-emitting element having excellent uniformity of characteristics.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a cross-sectional structure of a conventional blue LE.

FIG. 2 is a diagram showing a concentration distribution of aluminum atoms in an aluminum nitride-gallium mixed crystal layer.

FIG. 3 is a diagram showing a concentration distribution of magnesium atoms in an aluminum nitride-gallium mixed crystal layer.

FIG. 4 is a diagram schematically showing a relationship between a growth time and a layer thickness.

FIG. 5 is a diagram schematically showing the concentration distribution of a p-type impurity in a mixed aluminum-gallium nitride crystal having a large layer thickness.

FIG. 6 is a diagram showing an atomic concentration distribution of a p-type impurity in a depth direction of a layer formed by stacking thick layers.

FIG. 7 is a diagram showing a result of a SIMS analysis in a depth direction of a p-type nitride semiconductor layer formed by stacking layers having a small thickness according to the present invention.

FIG. 8 is a diagram showing the result of SIMS analysis of a laminate having a gallium nitride layer formed by conventional means on an aluminum-gallium nitride mixed crystal layer according to the present invention.

FIG. 9 is a view showing the distribution of aluminum atoms in the depth direction when the aluminum nitride-gallium mixed crystal layer of the present invention is overlaid on an underlayer.

FIG. 10 is a graph showing a magnesium concentration distribution of the laminate of Example 1.

FIG. 11 is a schematic diagram showing the structure of a pn junction diode manufactured from the laminate of Example 1.

FIG. 12 is a diagram showing IV characteristics of the diode of Example 1.

FIG. 13 is a diagram showing a concentration distribution of magnesium atoms inside a conventional magnesium-doped gallium nitride layer.

FIG. 14 is a graph showing the concentration distribution of magnesium atoms inside a p-type gallium nitride layer of Example 2.

FIG. 15 is a cross section T inside the p-type gallium nitride layer of Example 2.
It is the figure which copied the EM image.

FIG. 16 is a schematic diagram illustrating a planar shape of a light emitting diode chip according to a second embodiment.

FIG. 17 is a schematic sectional view of a light-emitting diode according to a third embodiment.

[Explanation of symbols]

(101) substrate (102) low-temperature buffer layer (103) n-type gallium nitride layer (104) light-emitting layer (105) upper cladding layer (106) contact layer (107) p-type electrode (negative electrode) (108) n-type electrode (Positive electrode) (109) Gallium nitride single crystal layer (110) Aluminum nitride-gallium mixed crystal layer doped with p-type impurity (111) Aluminum nitride-gallium mixed crystal layer doped with p-type impurity and gallium nitride single crystal layer (112) Concentration distribution profile of aluminum atoms (113) 2 in the horizontal (horizontal) direction rather than the vertical (vertical) direction
Time period of initial growth in which dimensional growth predominates (114) 3 in the vertical (vertical) direction rather than the horizontal (horizontal) direction
Time period of initial growth in which dimensional growth predominates (115) Point where two-dimensional growth shifts to three-dimensional growth (critical point) (116) Multi-layered layer doped with p-type impurities (117) Underlayer (118) Gallium nitride layer grown by conventional batch growth method (120) Silicon nitride film (121) Dislocation

Claims (5)

(57) [Claims] 1. A p-type I semiconductor on a group III nitride semiconductor crystal.
A method for epitaxially growing a group II nitride semiconductor layer, wherein the growth is interrupted when the thickness of the epitaxially grown layer is 50 nm or less, and is kept at a growth temperature for 60 seconds or more, and then the growth is started again, and 50n layer thickness
A method for forming a p-type group III nitride semiconductor layer, comprising forming a laminate of a plurality of thin layers of m or less. 2. A p-type group III nitride semiconductor layer formed by the formation method according to claim 1, comprising a laminate of a plurality of thin layers having a layer thickness of 50 nm or less. 3. The p-type group III nitride semiconductor layer according to claim 2, wherein the concentration distribution of the p-type dopant in the p-type group III nitride semiconductor layer is substantially constant within the layer thickness. 4. A semiconductor light emitting device using the p-type group III nitride semiconductor layer according to claim 2 as a cladding layer. 5. A semiconductor light-emitting device using the p-type group III nitride semiconductor layer according to claim 2 as a contact layer.