WO2005106982A1

WO2005106982A1 - Group iii nitride semiconductor light-emitting device

Info

Publication number: WO2005106982A1
Application number: PCT/JP2005/008552
Authority: WO
Inventors: Hisayuki Miki; Akira Bando; Takashi Udagawa
Original assignee: Showa Denko KK
Current assignee: Resonac Holdings Corp
Priority date: 2004-04-28
Filing date: 2005-04-28
Publication date: 2005-11-10
Anticipated expiration: 2006-10-28

Abstract

A Group III nitride semiconductor light-emitting device includes a crystal substrate, an n-type and a p-type semiconductor of AlXGaYInZN1-aMa, wherein 0 ≤ X ≤ 1, 0 ≤ Y ≤ 1, 0 ≤ Z ≤ 1, X + Y + Z = 1, M denotes a Group V element other than N, and 0 ≤ a < 1, formed on the crystal substrate, and a light-emitting layer including a region doped with Ge.

Description

DESCRIPTION

GROUP HI NITRIDE SEMICONDUCTOR LIGHT-EMITTING DEVICE

Cross Reference to Related Applications: This application is an application filed under 35 U.S.C. § 111(a) claiming the benefit pursuant to 35 U.S.C. § 119(e)(1) of the filing date of Provisional Application No. - 60/570,499 filed May 13, 2004 pursuant to 35 U.S.C. § 111 (b).

Technical Field: This invention relates to a Group HI nitride semiconductor light-emitting device provided in the light-emitting layer thereof with a region doped with Ge.

Background Art: The Group HI nitride semiconductors have been heretofore finding utility as a functional material for configuring Group HI nitride semiconductor light-emitting devices of the pn junction structure, such as light-emitting diodes (LEDs) and laser diodes (LDs), which emit a visible light of a short wavelength (refer, for example, to JP-A 2000-332364). In the configuration of an LED assuming emission in the near ultraviolet band, the blue color band or the green color band, for example, the n-type or the p-type aluminum gallium nitride (AlχGaγN: 0 < X, Y < 1, X + Y = 1) is utilized for configuring a clad layer (refer, for example, to JP-A 2003-229645). Then, gallium indium nitride (GaγJ-n_zN: 0 < Y, Z < 1, Y + Z = 1) is utilized for forming an active layer (light-emitting layer) (refer, for example to JP-B SHO 55-3834). In the conventional Group HI nitride semiconductor light-emitting devices, the light- emitting layers thereof generally have an n-type or a p-type Group HI nitride semiconductor layer joined thereto. These layers are intended to configure a light-emitting part in a hetero- junction structure with the object of acquiring emission of high intensity. For the purpose of configuring a light-emitting part in a double-hetero (DH) junction structure, it has been heretofore customary to form light-emitting layers of GaylnzN (0 < Y, Z < 1, Y + Z = 1), for example, and have an n-type or a p-type Group HI nitride semiconductor layer joined thereto as a clad layer (refer, for example, to "Group HI-N Compound Semiconductors" written by Isamu Akasaki and issued by Baifukan (K.K.), May 20, 1995, chapter 13). The contact layer which is intended to form an n-type electrode has been heretofore formed solely of a Group IH nitride semiconductor having silicon (Si) added thereto. An n- type AlχGaγΝ (0 < X, Y < 1 and X + Y = 1) layer, for example, which has the resistance thereof controlled by adjusting the doping amount of silicon has been available (refer, for example, to Japanese Patent No. 3383242). Similarly, Si has been used also as a donor element for doping an active layer. The so-called codoped configuration which uses an InGaN layer of a comparatively large film thickness for a light-emitting layer is doped with zinc (Zn) forming the center of emission and additionally doped with Si (refer, for example, to JP-A HEI 8-316528). Further, in the case of using a quantum well structure, it has been proposed to have a well layer doped and a barrier layer also doped as well (refer, for example, to JP-A HEI 8-264831 and JP-A HEI 9- 365422). As an n-type dopant in the GaN-based semiconductors, generally germanium (Ge) and other elements have been known besides Si (refer, for example, to JP-A HEI 9-36423). These n-type dopants, however, are deficient in doping efficiency as compared with Si (refer, for example, to Jpn. J. Appl. Phys., 31 (9 A) (1992), 2883) and, therefore, are rated as disadvantageous for producing an n-type Group IH nitride semiconductor layer of low resistance. Further, it is held that when an n-type Group HE nitride semiconductor layer is doped with Ge to a high concentration, the doping is at a disadvantage in inducing the surface of the layer to sustain pits which impair flatness (refer, for example, to "Group HI Nitride Semiconductor Compounds," Clarendon Press (Oxford), 1998, page 104). Thus, the products heretofore obtained by doping their light-emitting layers (active layers) with Ge have not succeeded. The LED having the light-emitting layer thereof doped with Si has been at a disadvantage in suffering the emission output thereof to be degraded by aging owing to the migration of Si in the crystal in consequence of the protracted conduction of electricity. At times, it has shown a loss in the form of a lowered peak reverse voltage. The LED may be manufactured by having the light-emitting layer thereof kept undoped, i.e. not doped with a dopant. In this case, the omission of a doping treatment results in inevitably adding to the drive voltage. Preferably, therefore, the light-emitting layer is doped with an n-type dopant of some sort. This invention is aimed at providing a Group HI nitride semiconductor light-emitting device including in a light-emitting layer thereof a region doped with Ge, namely a light- emitting device exhibiting excellent emission output without impairing the flatness of the light-emitting layer or inducing no loss of crystallinity.

Disclosure of the Invention: The Group HI nitride semiconductor light-emitting device according to this invention comprises a crystal substrate, an n-type and a p-type Group HI nitride semiconductor of

AlxGaylnzNi-aMa, wherein 0 < X < 1, 0 ≤Y < 1, 0 < Z < 1, X + Y + Z = 1, M denotes a

Group V element other than N, and 0 < a < 1, formed on the crystal substrate and a light- emitting layer including a region doped with Ge. In the device, the region doped with Ge has a layer having an atomic concentration of Ge varied periodically. In the second mentioned device, the region doped with Ge is formed of a structure having a Group HI nitride semiconductor layer doped with Ge and a Group HI nitride semiconductor layer undoped therewith alternately stacked periodically. In the second or third mentioned device, the region doped with Ge comprises a Group HI nitride semiconductor layer doped with Ge to a higher concentration and a Group HI nitride semiconductor layer doped with Ge to a lower concentration, and the higher- concentration layer has a smaller thickness than the lower-concentration layer. In the fourth mentioned device, the Group HI nitride semiconductor layer doped with

Ge to a higher concentration has pits having a density in the range of 1 x 10⁵ to 1 x 10¹⁰/cm³. In the fourth mentioned device, the Group HI nitride semiconductor layer doped with Ge to a lower concentration has surface flatness of 10 A or more. In any one of the first to third mentioned devices, the light-emitting layer including the region doped with Ge has a concentration of Ge atoms of 1 x 10¹⁷ cm^"3 or more and 1 x 10² cm^" or iess. In any one of the first to third mentioned devices, the Group IH nitride semiconductor layer doped with Ge has a concentration of Ge atoms of 5 x 10¹⁷ cm^"3 or more and 5 x 10¹⁹ cm^"3 or less. In any one the first to third mentioned devices, the light-emitting layer including the region doped with Ge has a multiple quantum well structure. In the device just mentioned above, the region doped with Ge in the light-emitting layer is a barrier layer of multiple quantum well structure. This invention, in a Group HI nitride semiconductor light-emitting device having a light-emitting layer thereof doped with Ge, is directed toward producing a light-emitting device showing an excellent intensity of emission without impairing the flatness and the crystallinity of the light-emitting layer. Furthermore, no degradation of the emission output by aging or no loss in peak reverse voltage will be produced.

Brief Explanation of the Drawing: Fig. 1 is a schematic cross section illustrating the configuration of a stacked structure described in Example 1. Fig. 2 is a schematic plan view of the LED described in Example 1. Fig. 3 is a schematic cross section illustrating the configuration of a stacked structure described in Example 2. Fig. 4 is a schematic plan view of the LED descried in Example 2. Fig. 5 is a cross section depicting an artist's concept of the layer structure resulting from filling the pits occurring in a Ge high concentration layer with a Ge low concentration layer. Fig. 6 is a schematic cross section illustrating the configuration of a stacked structure described in Comparative Example 1.

Best Mode of Embodying the Invention The Group HI nitride semiconductor light-emitting device according to this invention has an n-type and a p-type Group HI nitride semiconductor (AlχGaγInzN₁.aMa: 0 < X < 1, 0 < Y < 1, 0 ≤ Z < 1, X + Y + Z = 1, M denotes a Group N element other than nitrogen (Ν), and 0 < a < 1) formed on a crystal substrate and has a light-emitting layer including a region doped with germanium (Ge). The inventors' experiment has yielded a result indicating that when Ge is used as a dopant for a light-emitting layer, the phenomenon that the emission output and the peak reverse voltage of the LED are degraded by aging does not appear. In the Group HI nitride semiconductor, Group IN elements, such as Si and Ge, exhibiting n-type conductivity, when doped, are thought to be substituted for a Group HI element in the crystal to secure their existence. Gallium nitride, for example, is substituted for Ge. Ge has an atomic radius larger than Si which is generally used as a dopant and nearly equal to Ga. It is, therefore, thought to induce no change in the lattice constant of a crystal even when it is made to dope a Group HI nitride semiconductor crystal. This fact seems to be the cause for keeping even an active layer doped with Ge from inducing any decline of crystallinity and from inducing the diffusion due to a protracted conduction of electricity as observed in Si. The stacked structure formed of a Group HE nitride semiconductor layers and constituting a light-emitting device which possesses a light-emitting layer including a region doped with germanium according to this invention is configured on a substrate which is made of sapphire (α-Al₂O₃ single crystal) having a relatively high melting point and having thermal resistance, an oxide single crystal material (such as zinc oxide (ZnO) or gallium lithium oxide (GaLiO₂)), silicon (Si) single crystal or a Group IN semiconductor single crystal (such as cubic or hexagonal silicon carbide (SiC)). As the material for the substrate, Group πi-N compound semiconductor single crystal material, such as gallium phosphide (GaP) or gallium arsenide (GaAs), may be also usable. A single crystal substrate which is formed of gallium nitride crystal is included in the group of substrates which are available here. An optically transparent single crystal material which is capable of passing the emission from a light-emitting layer can be effectively utilized as a substrate. For the purpose of stacking a gallium nitride-based compound semiconductor on the aforementioned substrate which excludes a GaΝ substrate and forms no lattice matching with a gallium nitride compound, the low-temperature buffering method disclosed in Japanese Patent No. 3026087 and JP-A HEI 4-297023 and the lattice-unmatchable crystal epitaxial growth technique called a seeding process (SP) disclosed in JP-A 2003-243302 can be used. Particularly, the SP process which manufactures an A1N crystal film at a high temperature enough to permit manufacture of a GaN-based crystal serves as a lattice- unmatchable crystal epitaxial growth technique which is excellent from the viewpoint of exalting productivity. When the lattice unmatchable crystal epitaxial growth technique of the low- temperature buffer method or the SP method is used, the gallium nitride-based semiconductor to be laid as the base is preferred to be GaN which is left undoped or doped meagerly to about 5 x 10¹⁷ cm^"3. The film thickness of the base layer is preferably in the range of 1 to 20 μm and more preferably in the range of 5 to 15 μm. The Group HI nitride semiconductor light-emitting device possessing an active layer including a region doped with germanium atom according to this invention can be formed by means of gas phase growth, such as the metal organic chemical vapor deposition method (abbreviated as "MOCND," "MONPE," or "OMNPE"), the molecular beam epitaxial (MBE) method, the halogen gas phase growth method, and the hydride gas phase growth method. As the source for germanium to be added, organic germanium compounds, such as german gas (GeHi), tetramethyl germanium ((CH₃)₄Ge) and tetraethyl germanium ((C₂H5)₄Ge), can be used. The MBE method can use germanium in the elemental force as the source for doping. The MOCND method, for example, forms an n-type gallium nitride layer on the substrate of sapphire by using (CH₃)₄Ge. The active layer to be doped with Ge may be formed as a single layer of a thick film of 50 nm or formed in a quantum well structure. When it is formed in the quantum well structure, it may be in the single quantum well structure having only one well layer or in the multiple quantum well structure having a plurality of well layers. In these structures, the multiple quantum well structure is used advantageously for the device using a Group HI gallium nitride-based compound semiconductor because it is enabled to combine high output and low drive voltage. Incidentally, in the case of the multiple quantum well structure, the total of the well layer (active layer) and the barrier layer will be referred to as a "light- emitting layer" in the present specification. When the active layer is to be doped with Ge, the active layer may have the total volume thereof doped with Ge or it may have part of the region thereof doped with Ge. Particularly, when the quantum well structure is contemplated, the case of doping only the well layer, the case of doping only the barrier layer, or the case of doping both of the layers is conceivable. Among other cases enumerated above, the case of doping the barrier layer proves particularly advantageous because this doping can lower the drive voltage without degrading the output of emission. Also in the case of doping the barrier layer, the doping effected only in part of the region in the barrier layer proves effective. In the case of forming the barrier layer in the region growing at a high temperature and the region growing at a low temperature, for example, the doping effected in the region growing at the low temperature can be expected to bring a more prominent decline of the drive voltage. In the case of configuring a structure possessing a plurality of barrier layers, the amounts of Ge doping given to the individual barrier layers do not need to be equalized. The decrease in the concentration of Ge particularly in the barrier layer approximating closely to the region contiguous to the p-layer is effective in exalting the output and lowering the drive voltage. This exaltation of the output can be promoted particularly by keeping the barrier layer approximating most closely to the p-layer from undergoing a doping treatment. When the multiple quantum well structure is adopted, the number of layers to be stacked is preferably in the range from three to ten and more preferably in the range from three to six. In the case of the multiple quantum well structure, all the well layers (active layers) do not need to be furnished with a thick film part and a thin film part. Further, the dimensions and the area ratios of the thick film part and the thin film part may be varied in the individual layers. The film thickness of the barrier layer is preferably 70 A or more and more preferably 140 A or more. I-f the film thickness of the barrier layer is unduly small, the shortage will result in obstructing the flattening of the upper surface of the barrier layer and inducing a decrease in the efficiency of emission and a decrease in the aging characteristics. If the film thickness is unduly large, the overage will result in inducing an increase in the drive voltage and a decrease in the emission. Thus, the film thickness of the barrier layer is preferred to be 500 A or less. The active layer is preferably formed of an In-containing gallium nitride-based compound semiconductor. The In-containing gallium nitride-based compound semiconductor can emit light in a blue color wavelength region with high intensity. In the case of the multiple quantum structure, the barrier layer can be formed of InGaN which has a smaller In ratio than the InGaN forming the well layer (active layer) besides GaN and AlGaN. Among other materials mentioned above, GaN proves particularly advantageous. In the active layer mentioned above, the region doped with Ge may be given a structure having the concentration of germanium atom periodically varied with the object of securing the flatness of the surface. This region is formed by periodically varying with time during the gas phase growth of the Group HI nitride semiconductor layer the amount of the source for doping Ge to be supplied to the gas phase growth reaction system. A thin layer containing Ge atom in a high concentration is formed, for example, by instantaneously supplying a large amount of the source for doping Ge to the gas phase growth region after a thin undoped layer has been formed without supplying the source for doping Ge to the gas phase growth region. Through the fluctuation of the amount of the source for Ge doping to be supplied to the gas phase growth reaction system, it is made possible to form a region having the concentration of germanium atom varied periodically. This formation may be otherwise attained by growing the thin layer having a low concentration of Ge atom and subsequently suspending the growth till the conditions for growth such as the V/IH ratio are adjusted to suit the addition of Ge atom to a high concentration, thereby enabling a thin layer containing Ge atom in a high concentration to be joined thereto. The Group HI nitride semiconductor layer doped with Ge to a high concentration and used in the technique proposed by this invention (Ge atom high concentration layer), when it uses Ge as a dopant, inherently has such a high concentration as to produce pits in the surface thereof. By filling these pits with a Group HI nitride semiconductor layer doped with Ge to a low concentration (Ge atom low concentration layer), it is made possible to materialize a flat surface as compared with the case of doping the relevant layer with Ge to a high concentration in accordance with the conventional method. In the interface between the high concentration layer and the low concentration layer contemplated by this invention, the surface on the high concentration layer side (the opposite side from the substrate) contains pits of a convex shape while the surface on the low concentration layer side (the opposite side from the substrate) assumes a flat surface. The cross section depicting an artist's concept of the layer structure resulting from filling the pits occurring in a Ge atom high concentration layer with a Ge atom low concentration layer is shown in Fig. 5. In the diagram, 4a denotes a Ge atom high concentration layer, 4c denotes pits, and 4b denotes a Ge atom low concentration layer. The surface of the low concentration layer 4b is flattened in consequence of filling the pits 4c occurring in the surface of the high concentration layer 4a with the low concentration layer 4b. The pits occurring in the Ge atom high concentration layer of this invention are considered to occur at the positions of the so-called threading dislocation originating from the interface between the substrate and the Group HI nitride semiconductor layer. More often than not, therefore, the density of the pits occurring in the high concentration layer approximately coincides with the density of the threading dislocation in the base. In the GaN crystal on the ordinary sapphire substrate, the density of threading dislocation in the base falls in the range of 1 x 10⁷ to 1 x 10¹⁰/cm². The products having a density of pits of 1 x 10⁷/cm² or less are not realized at present and the products having a density of pits of 1 x 10¹⁰/cm² or more, even when they are used in a substrate for an electron device, cannot show a fully satisfactory function. The density of pits is in the range of 1 x 10⁶ to 1 x 10¹⁰/cm², though depending on the density of threading dislocations in the base. Generally, it is in the range of 1 x 10 to 1 x 10⁹/cm² When the high concentration layer alone is manufactured in a film thickness of about 10 nm or more, these pits can be observed by using such means as an atomic force microscope (AFM). When the film thickness is increased to about 500 nm, they can be observed by the use of an optical microscope. When the high concentration layer has an extremely small film thickness, these pits may possibly become invisible because of the resolution of the atomic force microscope. When this thickness is increased to a certain extent and the film-forming conditions are adjusted to permit observation of the pits, the pits are thought to have occurred in spite of a very small thickness falling short of 10 nm. The surface of the Ge atom low concentration layer of this invention is preferred to be flat. The flatness of the surface is preferred to be approximately 10 A or less and more preferably to be 5 A or more in the Ra value. In the case of forming a region having the concentration of Ge atom periodically varied, the total layer thickness of the region having the concentration of Ge atom periodically varied is properly 5 nm or more and 100 nm or less. It is preferably 10 nm or more and 70 nm or less and more preferably 15 nm or more and 50 nm or less. If this layer thickness falls short of 5 nm, the shortage will result in adding to the conspicuousness of the deterioration of the well layer by aging. If it exceeds 100 nm, the overage will result in inducing degradation of the emission output. The total of the film thickness of the n-type Group HI nitride semiconductor layer containing Ge in a high concentration and the film thickness of the n-type nitride semiconductor layer containing Ge in a low concentration, namely the periodic film thickness, is properly 0.5 nm or more, preferably 1 nm or more and more preferably 2 nm or more. If the total of the film thickness falls short of 0.5 nm, the shortage will render it difficult to obtain the effect of periodically stacking the Ge doped layers. Specifically, when the layer doped with Ge to a high concentration is thicker than the layer doped with Ge to a low concentration in one period, the occurrence of pits cannot be repressed and the flatness cannot be easily obtained. On the other hand, when the thickness of the layer doped with Ge to a low concentration is equal to or greater than the thickness of the layer doped with Ge to a high concentration, the flatness is good. The thickness of the layer doped with Ge to a low concentration, therefore, is preferred to have a greater layer thickness than the thin layer doped with Ge. When an undoped n-type Group HI nitride semiconductor thin layer containing Ge atom in a high concentration is formed first with a view to decreasing the concentration of Ge atom to the fullest possible extent, this formation is effective in obtaining a Ge-doped Group HI nitride semiconductor thin layer of a flat surface because it further exalt the effect of filling up the pits which are present in the surface of an n-type Group HI nitride semiconductor thin layer containing Ge atom in a high concentration. If the low concentration layer is given an unduly large thickness, however, the overage will be at a disadvantage in inevitably adding to the resistance and increasing the contact resistance of the n-electrode. Specifically, if the low concentration layer is unduly large, this overage will prove unfavorable for obtaining a Group HI nitride semiconductor light-emitting device having a low forward voltage (so-called Nf) or threshold voltage (so-called, Nth). The number of periodically stacked layer is suitably 1 or more and 200 or less, preferably 1 or more and 100 or less, and more preferably 1 or more and 50 or less. The concentration of Ge atoms in the interior of the n-type Group HI nitride semiconductor layer containing Ge at a high concentration is suitably 1 x 10¹⁷ cm^"3 or more and 1 x 10 cm^" or less, preferably 5 x 10 cm^" or more and 5 x 10 cm^" or less, and more preferably 3 x 10¹⁸ cm^"3 or more and 2 x 10¹⁹ cm^"3 or less. The concentration of Ge atoms in the interior of the n-type Group HI nitride semiconductor layer containing Ge at a high concentration does not need to be fixed but may be varied continuously or discontinuously. The concentration of Ge atom in the interior of the n-type Group HI nitride semiconductor layer containing Ge at a low concentration is lower than the concentration of Ge atom in the interior of the n-type Group HI nitride semiconductor layer containing Ge at a high concentration and is suitably the lower limit of determination or more by the following analytical method and 2 x 10¹⁹ cm^"3 or less, preferably the lower limit of determination or more and 1 x 10¹⁹ cm^"3 or less and more preferably the lower limit of determination or more and 5 x 10¹⁸ cm^"3 or less. It is rather preferable to leave out the doping treatment. The concentration of Ge atom in the interior of the n-type Group HI nitride semiconductor layer containing Ge at a low concentration does not need to be fixed but may be varied continuously or discontinuously. If the concentration of Ge atom is suffered to exceed 2 x 10¹⁹ cm³, the excess will be at a disadvantage in sharply increasing the density of pits in the surface. The concentration of Ge atom can be determined, for example, by the method of secondary ion mass spectroscopy (abbreviated as "SIMS"). This method comprises irradiating the surface of a sample with a primary ion and analyzing the mass of an element expelled consequently in an ionized state and permits observation and determination of the distribution of concentration of a specific element in the direction of depth. This method is effective with respect to the Ge element which is present in the Group HI nitride semiconductor layer. The increase of the concentration of Ge atom in a layer doped with Ge to a high concentration above 5 x 10¹⁷ cm^"3 can contribute to the configuration of an LED having a low forward voltage. When the concentration is 5 x 10¹⁹ cm^"3, the total carrier concentration in the region having the concentration of Ge atom periodically varied is approximately (3 to 4) x 10¹⁹ cm^"3. If the doping with Ge is performed in excess of this atomic concentration, the excess will be at a disadvantage in sharply increasing the density of pits in the surface. The composition may be changed between a region doped with Ge to a high concentration and a region doped with Ge to a low concentration. Particularly, the inclusion of In and Al in the composition of a layer doped with Ge constitutes an important technique for realizing the flattening of the surface. In the composition of the layer doped with Ge, the concentration of In so included therein is preferably 0.1 atom% or more and 50 atom% or less and optimally 1 atom% or more and 20 atom% or less. In the composition of the layer doped with Ge, the concentration of Al so included therein is preferably 0.1 atom% or more and 20 atom% or less and optimally 0.5 atom% or more and 10 atom% or less. Preferably an n-clad layer is interposed between the contact layer and the light- emitting layer. The n-clad layer may be formed of AlGaN, GaN or InGaN. When InGaN is selected, it goes without saying that the n-clad layer is preferred to have a larger composition than the band gap of InGaN in the active layer. The carrier concentration in the n-clad layer may be equal to the n-contact layer or may be larger or smaller than that. For the purpose of enhancing the crystallinity of the active layer to be formed thereon, it is preferable to impart a high flatness to the surface by suitably adjusting the conditions of growth, such as the speed of growth, the temperature of growth, the pressure of gross and the amount of doping. The n-clad layer may be formed by having two layers differing in composition and lattice constant alternately stacked a plurality of times. In this case, depending on the layer to be stacked, the two layers may be made to differ in the amount of dopant and the film thickness besides the composition. The p-type layer generally has a thickness in the range of 0.01 to 1 μm and is composed of a p-clad layer which is held in contact with the active layer and a p-contact layer which is intended to form a positive electrode. The p-clad layer and the p-contact layer may serve in the place of each other. The p-clad layer is formed using GaN or AlGaN, for example, and is doped with Mg as a p-dopant. For the purpose of facilitating the establishment of contact with an electrode, the outermost surface is preferably formed as a layer of a high carrier concentration while most layers may possess high resistance. That is, the amount of the dopant may be safely decreased and the inclusion of hydrogen, a substance which is held to obstruct the activation of the dopant, may be safely tolerated. These actions are rather at an advantage in enhancing the reverse blocking voltage for a configured device. The p-clad layer may likewise be formed by having two layers differing in composition and lattice constant alternately stacked a plurality of times. Depending on the layers to be stacked, in this case, the two layers may be made to differ in the amount of dopant and the film thickness besides the composition. The p-contact layer may be made of GaN, AlGaN or InGaN, for example, and may be doped with Mg as an impurity. The gallium nitride-based compound semiconductor doped with Mg generally possesses high resistance while still fresh from the reaction furnace. It is held to exhibit p-conductivity after undergoing activating treatments, such as the annealing treatment, the treatment of irradiation with an electron beam and the treatment of irradiation with a microwave. As already stated, it may be put to use without undergoing the aforementioned activating treatment. The p-contact layer which is formed of boron phosphide doped with a p-type impurity may be used. The boron phosphide doped with p-type impurity exhibits p- conductivity even when it is not subjected to the aforementioned treatment for the conversion to the p-type. The method for growing the gallium nitride-based compound semiconductor of which the n-type layer, the active layer and the p-type layer are formed does not need to be particularly restricted. For this growth, the well-known methods, such as MBE, MOCND, and HNPE may be used under well-known conditions. Among other methods enumerated above, the MOCND method proves particularly favorable. As the raw materials for the source of nitrogen, ammonia, hydrazine and azides may be used. As the Group HI organic metals, trimethyl gallium (TMGa), triethyl gallium (TEGa), trimethyl indium (TMIn) and trimethyl aluminum (TMAl) may be used. As the sources for dopant, silane, disilane, german, organic germanium raw materials and biscyclopentadienyl magnesium (Cp₂Mg) may be used. As the carrier gas, nitrogen and hydrogen may be used. The growth of an active layer containing In is preferably carried out with the substrate temperature kept in the range of 650 to 900°C. If the temperature falls short of the lower limit of this range, the shortage will not allow production of an active layer of satisfactory crystallinity. If the temperature exceeds the upper limit of the range, the overage will possibly result in decreasing the amount of In to be incorporated in the active layer and preventing the device emitting light of an intended wavelength from being manufactured. When the active layer has a multiple quantum well structure, the growth of part of the region of the barrier layer is preferably carried out at a higher substrate temperature than the growth of the well layer (active layer). The higher substrate temperature is preferred to be in the range of 700 to 1,000°C. The negative electrodes are known in various compositions and various structures. These well-known negative electrodes may be used without any restriction. As the contacting materials to be used for the negative electrodes which are destined to contact the n-contact layer, not only Al, Ti, Νi and Au but also Cr, W and N may be used. It goes without saying that the impartation of the bonding property to the negative electrode is accomplished by configuring the negative electrode wholly in a multilayer structure. Particularly, the coating of the outermost surface of the negative electrode with Au is at an advantage in facilitating the bonding. The positive electrodes are also well known in various compositions and structures. These well-known positive electrodes may be used without any restriction. The positive layer materials pervious to light may include Pt, Pd, Au, Cr, Ni, Cu and

Co, for example. It is known that the positive electrodes are enabled to acquire enhanced perviousness to light by having them configured in a partly oxidized structure. As reflecting positive electrode materials, Rh, Ag, Al, for example, may be used besides the aforementioned materials. These positive electrodes may be configures by methods, such as sputtering and vacuum evaporation. Particularly when the sputtering is adopted, the positive electrodes are enabled by properly controlling the conditions of sputtering to acquire ohmic contact without undergoing an annealing treatment after the formation of an electrode film. Thus, the sputtering proves favorable. The light-emitting device may be configured in a flip-chip structure which is provided with a reflecting positive electrode or in a face-up structure which is provided with a light-pervious positive electrode or a lattice or comb positive electrode.

Example 1 : First, this invention will be specifically described citing as an example the case of configuring a Group HI nitride semiconductor light-emitting diode by stacking a light- emitting layer furnished with a Ge-doped barrier layer and formed of a multiple quantum well on a Ge-doped GaN layer stacked, with the concentration periodically varied. The dopant concentrations reported in the description were invariably determined by the SIMS method described above. The film thicknesses were determined by a method using a reflectance spectrum of a white light and by the observation of a cross section transmission electron microscope (TEM). These determinations apply to Example 2 and the subsequent examples. ' Fig. 1 schematically illustrates the profile of an epitaxially stacked structure for the manufacture of an LED described in the present example. The schematic section of an LED chips manufactured in the present example is illustrated in Fig. 2. The epitaxially stacked structure was configured by the following procedure using a common reduced-pressure MOCND means. First, a (OOOl)-sapphire substrate 101 was mounted on a susceptor made of a semiconductor-grade, high-purity graphite and adapted to be heated by a radio-frequency (RF) induction heater to a film-forming temperature. After the mounting was completed, a gas phase growth reaction furnace made of stainless steel had the interior thereof displaced with a stream of nitrogen gas. After the supply of the nitrogen gas to the interior of the gas phase growth reaction furnace was continued for 8 minutes, the induction heater was set operating to elevate the temperature of the substrate 101 from room temperature to 600°C over a period of 10 minutes. While the temperature of the substrate 101 was kept at 600°C, the pressure in the gas phase growth reaction furnace was increased by supply of hydrogen gas and nitrogen gas to 1.5 x 104 pascals (unit of pressure: Pa). The substrate 101 was left standing at the temperature under the pressure mentioned above for 2 minutes to effect thermal cleaning of the surface thereof. After the thermal cleaning was completed, the supply of the nitrogen gas to the interior of the gas phase growth reaction furnace was stopped. The supply of the hydrogen gas was continued. Thereafter, in the atmosphere of hydrogen, the temperature of the substrate 101 was elevated to 1,120°C. After it was confirmed that the temperature was stabilized at 1120°C, the hydrogen gas entraining vapor of trimethyl aluminum (TMAl) was supplied for a period of 8 minutes and 30 seconds into the gas phase growth reaction furnace. Consequently, the trimethyl aluminum was made to react with the nitrogen (Ν) atom generated in consequence of the decomposition of the nitrogen (Ν)-containing deposit formerly adhering to the inner wall of the gas phase growth reaction furnace and induce deposition of an aluminum nitride (AIN) thin film (not shown) having a thickness of several nm on the sapphire substrate 101. After the growth of AIN was terminated by stopping the supply of the hydrogen gas entraining the vapor of TMAl into the gas phase growth reaction furnace, the furnace was left standing for 4 minutes and then opened to expel the residual TMAl completely from the interior of the gas phase growth reaction furnace. Subsequently, after the lapse of 4 minutes from the start of supply of the ammonia (NH₃) gas into the gas phase growth reaction furnace, the temperature of the susceptor was lowered to 1040°C while the flow of the ammonia gas was continued. After the temperature of the susceptor was confirmed to have reached 1040°C, the susceptor was left standing for a while till the temperature thereof was stabilized and the supply of trimethyl gallium (TMGa) into the gas phase growth reaction furnace was started to induce the growth of an undoped GaN layer 102 over 4 hours. The thickness of the undoped GaN layer 102 was 8 μm. Then, the temperature of the wafer was elevated to 1140°C and allowed to stabilize. A cycle of continuing flow of tetramethyl germanium ((CH₃) Ge) and then stopping the flow was carried out up to 100 repetitions to form a Ge-doped GaN layer 103 measuring 2.0 μm in thickness and having a Ge concentration varied periodically. When a sample separately taken out of the furnace subsequently to the growth of the

Ge high concentration layer was observed under an atomic force microscope, the number of pits formed on the surface of the high concentration layer was found to be 2 x 10⁷/cm³. After the Ge-doped GaN layer was consequently deposited, a Ge-doped n-type Ino.oδGao.₉₄N clad layer 104 was stacked thereon. The thickness of this clad layer 104 was 12.5 nm and the amount of Ga doping was 1 x 10¹⁸ cm^"3. Then, the temperature of the substrate 101 was set at 730°C and a multiple quantum well structure light-emitting layer 105 of a five-cycle structure comprising a barrier layer of GaN and a well layer of l o.₂5Gao.₇5N was formed on a Ge-doped n-type tαo.o₂Gao.₉₈N clad layer 104. In the light-emitting layer 105 of the multiple quantum well structure, first a Ge- doped GaN barrier layer was disposed as joined to a Ge-doped n-type l o.o₆Gao.₉₄N clad layer 104. The GaN barrier layer was grown by using triethyl gallium (TEGa) as the source for gallium and tetraethyl germanium (TEGe) as the source of dopant. The layer thickness was 16 nm and the concentration of Ge was 5 x 10 cm^"3. The lno.₂5Gao.₇5N well layer was grown by using triethyl gallium (TEGa) as the source for gallium and trimethyl indium (TMIn) as the source for indium. The layer thickness was 2.5 nm and the doping was left out. On the light-emitting layer 105 of the multiple quantum well structure, a p-type Alo.o₇G o.₉₃N clag layer 106 doped with magnesium (Mg) was formed. The layer thickness was 10 nm. On the p-type Alo.₀₇Gao.seN clag layer 106, a p-type GaN contact layer 107 doped with Mg was formed. Bis-cyclopentadienyl magnesium (bis-Cp₂Mg) was used as the source for Mg doping. The addition of Mg was so performed as to give the p-type GaN contact layer 107 a hole concentration of 8 x 10¹⁷ cm^'3' The thickness of the p-type GaN contact layer 107 was 100 nm. After the growth of the p-type GaN contact layer 107 was completed, the conduction of electricity to the induction heater was stopped and the temperature of the substrate 101 was lowered to room temperature over a period of about 20 minutes. During the fall of the temperature, the atmosphere inside the gas phase growth reaction furnace was formed solely of nitrogen and the flow volume of NH₃ was decreased. Thereafter, the supply of NH₃ was further stopped. After the temperature of the substrate 101 was confirmed to have fallen to room temperature, a stacked structure 11 was taken out of the gas phase growth reaction furnace. At this point of time, the p-type GaN contact layer 107 was already showing p-type conductivity without undergoing an annealing treatment electrically activating the p-type carrier (Mg). Subsequently, the surface of the GaN layer 103 doped with Ge to a high concentration was exposed exclusively in the region expected to form an n-type ohmic electrode 108 by using the known photolithographic technique and the common dry etching technique. On the exposed surface of the Ge-doped n-type GaN layer 103, an n-type ohmic electrode 108 having chromium (Cr) and gold (Au) deposited on the surface side thereof was formed. On the entire surface of the p-type GaN contact layer 107 forming the residual surface of the stacked structure 11, a reflection p-type ohmic electrode 109 having platinum (Pt), silver (Ag) and gold (Au) sequentially stacked thereon from the surface side was formed by using the common sputtering means and the known photolithographic means. Thereafter, an LED chip 10 cut in a square of 350 μm as seen in a plan view was joined to a wire connection auxiliary member called a "submount." The LED chip on the submount was mounted on a lead frame (not shown) in order that a gold wire (not shown) connected to the lead frame may advance a device drive current from the lead frame to the LED chip 10. Through the lead frame, the device drive current was advanced in the forward direction between the n-type and p-type ohmic electrodes 108 and 109. The forward voltage was 3.0 N when the forward current was set at 20 mA. While the 20 mA forward current was flowing, the central wavelength of the blue band emission to be emitted was 460 nm. The intensity of the emission determined by using the ordinary integrating sphere reached 12 mW. Thus, the Group HI nitride semiconductor LED yielding emission of high intensity was completed. After the LED thus manufactured had a current of 50 mA conducted thereto for 1000 hours, it was subjected to the same determination as mentioned above. This determination found no change in the intensity of emission and in the drive voltage. The peak reverse voltage for effecting conduction of 10 μA was not changed from 20 N.

Example 2: First, this invention will be described citing the case of configuring a Group HI nitride semiconductor light-emitting diode by causing a light-emitting layer 111 of a multiple quantum well structure having an undoped GaΝ layer and a Ge-doped GaΝ layer alternately stacked to be deposited as a barrier layer via an n-type clad layer 104 on a Ge- doped GaΝ layer 103 stacked by periodically varying the concentration. Fig. 2 schematically illustrates the profile in cross section of an epitaxially stacked structure 12 for the manufacture of an LED described in the present example. The manufacture up to the formation of the Ge-doped n-type Ino.o6Gao.₉₄Ν clad layer 104 was carried out by following the procedure of Example 1. The doping amount of Ge was set at 1 x 10¹⁸ cm^"3 and the thickness of the clad layer was set at 50 nm. Then, after the temperature of the substrate 101 was set at 730°C, a multiple quantum well structure light-emitting layer 111 of a 5 -cycle structure comprising a barrier layer formed by stacking an undoped GaN layer 2 nm in thickness and a Ge-doped GaN layer 2 nm in thickness each up to four cycles and a well layer formed of undoped Itio.₂5Gao.₇5N was disposed on a Ge-doped n-type Ino.o₂Gao.₉₈N clad layer 104. In the light-emitting layer 111 of the multiple quantum well structure, first the GaN barrier layer was disposed as joined to the Ge-doped n-type In..o6Gao.₉₄N clad layer 104. The GaN barrier layer was grown by using triethyl gallium (TEGa) as the source for gallium and tetraethyl germanium (TEGe) as the source for germanium. The total layer thickness was 16 nm. It was configured by stacking an undoped GaN layer 2 nm in thickness and a Ge-doped GaN layer 2 nm in thickness each up to four cycles. The amount of Ge in the Ge-doped region was set at 1 x 10¹⁸ cm^"3. The Ino.₂_Gao.₇5N well layer was grown by using triethyl gallium (TEGa) as the source for gallium and trimethyl indium (TMJn) as the source for indium. The layer had a thickness of 2.5 nm and was not doped. Thereafter, the p-type contact layer 107 was stacked by following the procedure of Example 1 and the wafer was subsequently taken out of the reactor. Then, the surface of the GaN layer 103 doped with Ge to a high concentration was exposed exclusively in the region expected to form an n-type ohmic electrode 108 by using the known photolithographic technique and the common dry etching technique. On the exposed surface of the Ge-doped n-type GaN layer 103, an n-type ohmic electrode 108 having titanium (Ti) and gold (Au) stacked on the surface side thereof was formed. On the entire surface of the p-type GaN contact layer 107 forming the residual surface of the stacked structure 12, a transparent p-ohmic electrode 109 having platinum (Pt) and gold (Au) stacked sequentially from the front surface side and a bonding-grade electrode 110 were formed by using the common vacuum evaporation means and the known photolithographic means. Thereafter, an LED chip 20 cut in a square of 350 μm as seen in a plan view was mounted on a lead frame (not shown) and a gold wire (not shown) was connected to the lead frame so as to allow conduction of the device drive current from the lead frame to the LED chip 20. Through the lead frame, the device drive current was passed in the forward direction between the n-type and p-type ohmic electrodes 108 and 109. The forward voltage was 2.9 V when the forward current was set at 20 mA. The central wavelength of the blue band emission emitted during the flow of the forward current of 20 mA was 460 nm. The intensity of the emission determined by using the ordinary integration sphere reached 5.5 mW. Thus a Group HI nitride semiconductor LED yielding emission of high intensity in spite of a low drive voltage was completed. After the LED thus manufactured had a current of 50 mA conducted thereto for 1000 hours, it was subjected to the same determination as mentioned above. This determination found no change in the intensity of emission and in the drive voltage. The peak reverse voltage for effecting conduction of 10 μA was not changed from 20 N.

Comparative Example 1 : A Ge-doped GaΝ layer 103 having a Ge concentration periodically varied was formed as an n-type contact layer in the same manner as in Example 1, and a multiple quantum well structure 112 of GaΝ having a barrier layer thereof doped with Si in the place of Ge of Example 1 was stacked as a light-emitting layer. Thereafter, an LED was manufactured by forming electrodes, mounting them on a lead frame and making necessary connection under the same conditions as in Example 2 on the stacked structure 13 of Fig. 6 formed of a p-type Al₀.o₇Gao.₉₃Ν clad layer 106 and a p-type GaN contact layer 107 under the same conditions as in Example 1. As a result, the forward voltage was 2.9 V when the forward current was set at 20 mA. The central wavelength of the blue band emission emitted during the conduction of the forward current of 20 mA was 460 nm. As regards the characteristics during the conduction of the forward current of 20 mA, the intensity of the emission determined by using the ordinary integration sphere was 4 mW, a magnitude lower than when an Si-doped GaN layer was used as a barrier layer. When the LED thus manufactured had a current of 50 mA supplied thereto for 1000 hours and then subjected to the same determination as described above, the intensity of emission was found to have dropped to 3 mA. The peak reverse voltage required for conduction of 10 μA fell from 20 V to 5 V.

Industrial Applicability: The light-emitting device obtained by using a stacked gallium nitride-based compound semiconductor according to this invention shows no change of characteristic properties in consequence of aging due to a protracted conduction of electric current. Thus, it has an immense commercial value.

Claims

1. A Group HI nitride semiconductor light-emitting device comprising a crystal substrate, an n-type and a p-type Group HI nitride semiconductor of AlχGaγInzNι__aMa, wherein 0 <X < 1, 0 ≤ Y≤ 1, 0 <Z < 1, X + Y + Z = 1, M denotes a Group V element other than nitrogen, and 0 < a < 1, formed on the crystal substrate, and a light-emitting layer including a region doped with Ge.

2. A device according to claim 1, wherein said region doped with Ge has a layer having an atomic concentration of Ge varied periodically.

3. A device according to claim 2, wherein said region doped with Ge is formed of a structure having a Group HI nitride semiconductor layer doped with Ge and a Group HI nitride semiconductor layer undoped therewith alternately stacked periodically.

4. A device according to claim 2 or claim 3, wherein said region doped with Ge comprises a Group HI nitride semiconductor layer doped with Ge to a higher concentration and a Group HI nitride semiconductor layer doped with Ge to a lower concentration, and the higher-concentration layer has a smaller thickness than the lower-concentration layer.

5. A device according to claim 4, wherein the Group HI nitride semiconductor layer doped with Ge to a higher concentration has pits having a density in the range of 1 x 10⁵ to 1 x 10¹⁰/cm³.

6. A device according to claim 4, wherein the Group HI nitride semiconductor layer doped with Ge to a lower concentration has surface flatness of 10 A or more.

7. A device according to any one of claims 1 to 3, wherein said light-emitting layer including said region doped with Ge has a concentration of Ge atoms of 1 x 10¹⁷ cm^"3 or more and 1 x 10²⁰ cm^"3 or less.

8. A device according to any one of claims 1 to 3, wherein said Group HI nitride semiconductor layer doped with Ge has a concentration of Ge atoms of 5 x 10¹⁷ cm^"3 or more and 5 x 10¹⁹ cm^"3 or less.

9. A device according to any one of claims 1 to 3, wherein the light-emitting layer including said region doped with Ge has a multiple quantum well structure.

10. A device according to claim 9, wherein said region doped with Ge in the light- emitting layer is a barrier layer of multiple quantum well structure.