Semiconductor element Number:7,436,066 from the United States Patent and Trademark Office (PTO) owispatent

Title: Semiconductor element

Abstract: It is an object of the present invention to provide a highly reliable and high-quality semiconductor element by effectively preventing the migration of silver to a nitride semiconductor when an electrode main entirely or mostly of silver having high reflection efficiency is formed in contact with a nitride semiconductor layer. A semiconductor element comprises a nitride semiconductor layer, an electrode connected to said nitride semiconductor layer, and an insulating film covering at least part of said electrode, wherein the electrode comprises: a first metal film including silver or a silver alloy and in contact with the nitride semiconductor layer; and a second metal film completely covering the first metal film, and the insulating film comprises a nitride film.

Patent Number: 7,436,066 Issued on 10/14/2008 to Sonobe,   et al.

---

Inventors:	Sonobe; Shinya (Anan, JP), Tomonari; Masakatsu (Anan, JP), Inoue; Yoshiki (Anan, JP)
Assignee:	Nichia Corporation (Tokushima, JP)
Appl. No.:	10/563,100
Filed:	October 6, 2005
PCT Filed:	October 06, 2005
PCT No.:	PCT/JP2005/018510
371(c)(1),(2),(4) Date:	December 30, 2005
PCT Pub. No.:	WO2006/043422
PCT Pub. Date:	April 27, 2006

---

Foreign Application Priority Data

---

Oct 19, 2004 [JP]			2004-304763

Current U.S. Class:	257/767 ; 257/762; 257/E33.063; 438/650; 438/686
Current International Class:	H01L 29/45 (20060101); H01L 21/441 (20060101)
Field of Search:	257/103,762,E33.063,767,E33.062,E29.143 438/650,686,927

---

References Cited [Referenced By]

---

U.S. Patent Documents

6194743	February 2001	Kondoh et al.
6580870	June 2003	Kanazawa et al.
6794690	September 2004	Uemura
6900472	May 2005	Kondoh et al.
6936859	August 2005	Uemura et al.
2001/0015442	August 2001	Kondoh et al.
2002/0136932	September 2002	Yoshida
2003/0052328	March 2003	Uemura
2004/0222434	November 2004	Uemura et al.
2005/0179051	August 2005	Kondoh et al.

Foreign Patent Documents

	5-54465		Mar., 1993		JP
	8-298341		Nov., 1996		JP
	10-69756		Mar., 1998		JP
	10-93905		Apr., 1998		JP
	11-87771		Mar., 1999		JP
	11-161663		Jun., 1999		JP
	11-191641		Jul., 1999		JP
	11-220171		Aug., 1999		JP
	2001-217461		Aug., 2001		JP
	2002-140882		May., 2002		JP
	2003-17741		Jan., 2003		JP
	2003-189197		Jul., 2003		JP
	2003-243705		Aug., 2003		JP
	2004-71655		Mar., 2004		JP

Other References

Kim, Jong Kyu et al., Microstructural study of Pt contact on p-type GaN, Journal of Vauum Science and Technology B, vol. 21, Issue 1, pp. 87-90, Jan. 2003. cited by examiner.

Primary Examiner: Vu; David
Assistant Examiner: Taylor; Earl N
Attorney, Agent or Firm: Birch, Stewart, Kolasch & Birch, LLP

---

Claims

---

The invention claimed is:

1. A semiconductor element, comprising a nitride semiconductor layer, an electrode connected to said nitride semiconductor layer, and an insulating film covering at least part of said electrode, wherein the electrode comprises: a first metal film in contact with the nitride semiconductor layer; and a second metal film completely covering the first metal film, wherein the first metal film is a multilayer film comprising a silver or a silver alloy film, and further comprising one or more metal films that inhibit a reaction with silver and are disposed over the silver or silver alloy film, and the insulating film comprises a nitride film, wherein at least one of the one or more films disposed over the silver or silver alloy film in the first metal film is formed such that the thickness of a portion disposed on a side face of the silver or silver alloy film is less than the thickness of the portion disposed over the silver or silver alloy film, and wherein the second metal film is formed such that the portions disposed on sides of the first metal film are thicker than the portion disposed above said first metal film.

2. A semiconductor element, comprising a nitride semiconductor layer, an electrode connected to said nitride semiconductor layer, and an insulating film covering at least part of said electrode, wherein the electrode comprises: a first metal film including silver or a silver alloy and in contact with the nitride semiconductor layer; and a second metal film formed so as to prevent the silver from moving across the surface of the nitride semiconductor layer, and the insulating film comprises a nitride film, wherein a metal film disposed over the silver or silver alloy film in the first metal film is formed such that the thickness of a portion disposed on a side face of the silver or silver ahoy film is less than the thickness of the portion disposed over the silver or silver alloy film, and wherein the second metal film is formed such that the portions disposed on sides of the, first metal film are thicker than the portion disposed above said first metal film.

3. The semiconductor element according to claim 1, wherein the nitride film is formed from either silicon nitride or silicon oxynitride.

4. The semiconductor element according to claim 1, wherein the first metal film is a single crystal at least at the interface with the nitride semiconductor layer.

5. The semiconductor element according to claim 1, wherein the first metal film comprises, in addition to a silver or a silver alloy film, a nickel film disposed in a partial area between said silver or silver alloy film and the nitride semiconductor layer.

6. The semiconductor element according to claim 1, wherein the second metal film comprises a metal that inhibits a reaction with silver, at least in the region in contact with the first metal film.

7. The semiconductor element according to claim 1, wherein the second metal film comprises a metal selected from the group consisting of nickel (Ni), ruthenium (Ru), osmium (Os), iridium (Ir), titanium (Ti), vanadium (V), niobium (Nb), tantalum (Ta), cobalt (Co), iron (Fe), chromium (Cr), and tungsten (W) and disposed at least in the region in contact with the first metal film.

8. The semiconductor element according to claim 6, wherein at least the region of the second metal film that is in contact with the first metal film is formed from nickel.

9. The semiconductor element according to claim 1, wherein the nitride semiconductor layer comprises a nitride semiconductor layer of a first conduction type, a light emitting layer, and a nitride semiconductor layer of a second conduction type that is different from that of the nitride semiconductor layer of the first conduction type, in that order, and a second electrode, connected to the nitride semiconductor layer of the second conduction type.

10. The semiconductor element according to claim 9, wherein the nitride semiconductor layer of the first conduction type is an n-type semiconductor layer, and the nitride semiconductor layer of the second conduction type is a p-type semiconductor layer.

11. The semiconductor element according to claim 2, wherein the nitride film is formed from either silicon nitride or silicon oxynitride.

12. The semiconductor element according to claim 2, wherein the first metal film is a single crystal at least at the interface with the nitride semiconductor layer.

13. The semiconductor element according to claim 2, wherein the first metal film comprises, in addition to a silver or a silver alloy film, a nickel film disposed in a partial area between the silver or silver alloy film and the nitride semiconductor layer.

14. The semiconductor element according to claim 2, wherein the second metal film comprises a metal that inhibits a reaction with silver at least in the region in contact with the first metal film.

15. The semiconductor element according to claim 2, wherein the second metal film comprises a metal selected from the group consisting of nickel (Ni), ruthenium (Ru), osmium (Os), iridium (Ir), titanium (Ti), vanadium (V), niobium (Nb), tantalum (Ta), cobalt (Go), iron (Fe), chromium (Cr), and tungsten (W) and disposed at least in the region in contact with the first metal film.

16. The semiconductor element according to claim 14, wherein at least the region of the second metal film that is in contact with the first metal film is formed from nickel.

17. The semiconductor element according to claim 2, wherein the nitride semiconductor layer comprises a nitride semiconductor layer of a first conduction type, a light emitting layer, and a nitride semiconductor layer of a second conduction type that is different from that of the nitride semiconductor layer of the first conduction type, in that order, and a second electrode connected to the nitride semiconductor layer of the second conduction type.

18. The semiconductor element according to claim 17, wherein the nitride semiconductor layer of the first conduction type is an n-type semiconductor layer, and the nitride semiconductor layer of the second conduction type is a p-type semiconductor layer.

19. The semiconductor element according to claim 1, wherein a nickel film is disposed in a partial area between the silver or silver alloy film in the first metal film and the nitride semiconductor layer.

20. The semiconductor element according to claim 1, wherein the thickness of the second metal film is greater than the thickness of the side portions of the first metal film.

21. The semiconductor element of claim 2, wherein a nickel film is disposed in a partial area between the silver or silver alloy film in the first metal film and the nitride semiconductor layer.

22. The semiconductor element according to claim 2, wherein the thickness of the second metal film is greater than the thickness of the side portions of the first metal film.

23. The semiconductor element according to claim 1, wherein the silver or silver alloy film in the first metal film is formed so that it becomes smaller in size away from a side in contact with the nitride semiconductor layer in contact with the first metal film.

24. The semiconductor element according to claim 2, wherein the silver or silver alloy film in the first metal film is formed so that it becomes smaller in size away from a side in contact with the nitride semiconductor layer in contact with the first metal film.

---

Description

---

TECHNICAL FIELD

The present invention relates to a semiconductor element comprised of a nitride semiconductor, and more particularly relates to an improved electrode in a semiconductor element.

BACKGROUND ART

In the past, the p-type electrodes in flip-chip nitride semiconductor light emitting elements have been used a structure comprised of silver or a silver alloy. Because silver very efficiently reflects the light produced by the light emitting layer of a light emitting element, such an element is able to emit light of high brightness.

When silver is used for the electrode material on the p side, however, part of the silver electrode surface has to be exposed for the purpose of connection with external components and so forth, and this causes and promotes the migration of silver ions, which changes the silver at the interface with the nitride semiconductor layer, and can lead to problems such as decreased light emission intensity due to the light produced by the light emitting layer not being reflected as efficiently by the electrode, and decreased service life due to the movement of silver to the other electrode, which causes short-circuiting.

One measure that has been taken to deal with this problem is to completely cover the silver electrode with an-electrode material that does not contain silver, and form a protective film over this, in order to prevent the silver electrode surface from being exposed. However, if a heat treatment is performed after electrode formation, or depending on the conditions of this heat treatment and so forth, it is sometimes impossible to adequately suppress the diffusion of the silver into the electrode material that contains no silver, which means that the migration of silver cannot be prevented.

In response to this problem, a method has been proposed for preventing the migration of silver by disposing an SiO.sub.2 film having a plurality of through holes between the silver electrode and the electrode material containing no silver, and electrically connecting the silver electrode and the electrode material containing no silver through this holes (see Patent Document 1, for example).

Other proposals include an LED in which a metal layer is formed over a silver layer on a p-type nitride semiconductor layer, and a dielectric layer is formed on part of the surface of the silver layer (see Patent Document 2, for example), a semiconductor light emitting element in which a p-type electrode is formed from a laminate of a contact layer and a reflecting layer made of a silvery white metal disposed over this contact layer (see Patent Document 3, for example), a semiconductor element having an-electrode formed from a first metal layer connected to a contact layer on the p-type semiconductor side, and a second metal layer that covers at least the side face of the first metal layer and the surface of the contact layer not covered by the first metal layer, in which the first metal layer is formed from silver (Ag), and the second metal layer from vanadium (V) and aluminum (Al) or titanium (Ti) and gold (Au) (see Patent Document 4, for example), a light emitting element equipped with a first thin-film metal layer comprised of cobalt (Co), nickel (Ni), or an alloy of these between a p-type nitride semiconductor layer and a positive electrode (see Patent Document 5, for example), and so forth.

Patent Document 1: JP-2003-168823-A

Patent Document 2: JP-H11-186599-A

Patent Document 3: JP-H11-191641-A

Patent Document 4: JP-H11-220171-A

Patent Document 5: JP-2000-36619-A

With Patent Document 1, however, because an SiO.sub.2 film is disposed between a silver electrode and an electrode material containing no silver, even though electrical connection is ensured by the through-holes, a problem is increased contact resistance between the two electrodes.

Also, although the migration of silver to the electrode material containing no silver is suppressed by physically blocking off the electrode with the SiO.sub.2 film, this does not effectively prevent the migration of silver to the nitride semiconductor layer, so this still does not mitigate the decrease in the service life of the light emitting element or the decrease in light emission intensity attributable to the migration of silver.

Furthermore, the suppression of silver migration cannot be considered satisfactory in any of these publications, and there is a need for a light emitting element that better prevents this migration, which in turn affords better semiconductor element reliability and higher yield, as well as a semiconductor element with more efficient light take-off.

DISCLOSURE OF THE INVENTION

The present invention was conceived in light of the above problems, and it is an object thereof to provide a highly reliable and high-quality semiconductor element by effectively preventing the migration of silver to a nitride semiconductor when an electrode main entirely or mostly of silver having high reflection efficiency is formed in contact with a nitride semiconductor layer.

In other words, the migration of silver in an electrode comprised of silver or a silver alloy is not only caused by contact with other electrode materials or semiconductors, or heat treatment in a contact state, or the flow of current in a contact state, for example, but also tends to be caused by the action of tiny amounts of water (moisture) on the silver, so it is an object of the present invention to provide a high-quality semiconductor element with which an electrode comprised of silver or a silver alloy is isolated from moisture, thereby effectively preventing the migration of silver.

As a result of diligent research into the migration of silver in an electrode comprised of silver or a silver alloy, the inventors arrived at the present invention upon discovering that the migration of silver can be dramatically reduced by configuring a semiconductor element as follows.

The present invention is characterized in that a semiconductor element, comprising a nitride semiconductor layer, an electrode connected to said nitride semiconductor layer, and an insulating film covering at least part of said electrode, wherein the electrode comprises:

a first metal film including silver or a silver alloy and in contact with the nitride semiconductor layer; and

a second metal film completely covering the first metal film or a second metal film formed so as to prevent the silver from moving across the surface of the nitride semiconductor layer, and

the insulating film comprises a nitride film.

With the semiconductor element of the present invention, a first metal film comprised of silver or a silver alloy is completely covered by a second metal film, and therefore not exposed, so the first metal film comprised of silver or a silver alloy can be prevented from coming into contact with moisture. Even if the silver moves across the nitride semiconductor layer surface, this movement can be halted by the second metal film, and in addition, the movement of silver in a direction perpendicular to the interface between the nitride semiconductor and the electrode can be prevented by the second metal film or the insulating film. Furthermore, it is believed that because this insulating film that covers the first and second metal films is formed from a nitride, any moisture that would act on the electrode is effectively trapped by the nitrogen atoms in the insulating film disposed near the electrode of the semiconductor element. As a result, even if the electrode containing the first metal film comprised of silver or a silver alloy is subjected to a heat treatment or the flow of current, the migration of silver can still be effectively prevented, which increases the light emission intensity and extends the service life, and allows a high-quality semiconductor element of high reliability to be obtained.

Saying that "a second metal film is formed so as to prevent the silver from moving across the surface of the nitride semiconductor layer" means that the second metal film touches the first metal film at least at the surface of the nitride semiconductor layer, so that the second metal film covers all the way around the first metal film, and preferably that the surface of the first metal film away from the nitride semiconductor layer is covered by the second metal film. If the second metal film is formed so as to cover all the way around the first metal film at the surface of the nitride semiconductor layer, this will prevent the most likely cause of silver migration, that is, the phenomenon whereby silver moves across the nitride semiconductor layer surface toward the other electrode. In addition, if the surface of the first metal film away from the nitride semiconductor layer is covered, it will be possible to completely prevent the movement of silver, which is prone to migration.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a cross section of an embodiment of the semiconductor element of the present invention;

FIG. 2a is a top view, and FIG. 2b a partial cross section, of another embodiment of the semiconductor element of the present invention;

FIG. 3 is a cross section of a light emitting device in which the semiconductor element of the present invention has been mounted;

FIG. 4 is a cross section of another light emitting device in which the semiconductor element of the present invention has been mounted;

FIG. 5a is a top view of the layout of electrodes in the semiconductor element of the present invention, and FIG. 5b is a top view of the layout of pad electrodes;

FIG. 6a is a top view of a support substrate for mounting the semiconductor element of the present invention, FIG. 6b is a circuit diagram of a light emitting device in which this semiconductor element is mounted, and FIG. 6c is a cross section of a light emitting device;

FIG. 7a is a top view of a support substrate for mounting the semiconductor element of the present invention, FIG. 7b is a top view of a mounted light emitting element, and FIG. 7c is a circuit diagram of a light emitting device in which this semiconductor element is mounted;

FIG. 8 is a top view of the layout of pad electrodes in another semiconductor element of the present invention;

FIG. 9a is a top view of a support substrate for mounting the semiconductor element of the present invention, and FIG. 9b is a top view of a mounted light emitting element;

FIG. 10 is a cross section of a light emitting device in which the semiconductor element of the present invention has been mounted;

FIG. 11 is a cross section of another light emitting device in which the semiconductor element of the present invention has been mounted; and

FIGS. 12a to 12d are cross sections of other embodiments of the semiconductor element of the present invention.

DESCRIPTION OF THE REFERENCE MARK

1, 30, 70 semiconductor element 2, 11 sapphire substrate 3 n-type semiconductor layer 4 light emitting layer 5 p-type semiconductor layer 6, 65, 66, 67 first metal film 7, 21, 34 second metal film 8 p-electrode (electrode) 9, 19, 35 n-electrode 10 insulating film 12 buffer layer 13 non-dope GaN layer 14 n-type contact layer 15 n-type clad layer 16 active layer 17 p-type clad layer 18 p-type contact layer 20 convex portions 22 silver film 31 end portion 32 constricted portion 33 extended portion 36, 37 pad electrode 38 p pad electrode 39, 39a, 39b resin layer 40, 41, 50, 51, 53a to 53c, 60, 61 conductive wiring 42 support substrate 43, 45, 46 light emitting device 44 bump electrode 47 wavelength conversion member 65a, 66a, 67a silver film 65b, 65d, 66b, 67b nickel film 65c, 66c, 67c platinum film 66b laminated film of titanium/nickel 120 stem 120a recess 121 first lead 122 second lead 131, 141 sealing member 150 fluorescent material 160 sub-mount member 161, 162 electrode 200 LED chip 201 mounting substrate 202 recess 203 lead 204 adhesive layer 205 sub-mount substrate 206 reflection component 207 terrace 208 light take-off component 209 translucent sealing member 212 package

BEST MODE FOR CARRYING OUT THE INVENTION

As discussed above, the semiconductor element of the present invention comprises an electrode formed on a nitride semiconductor layer, and an insulating film that covers this electrode.

The electrode formed over the nitride semiconductor layer is directly connected to the nitride semiconductor layer and preferably ohmically connected. "Ohmically connected" is used here in the sense in which it is normally used in this field, and refers, for example, to a junction in which the current-voltage characteristics are linear or substantially linear. This also means that the voltage drop and power loss at the junction during device operation are low enough to be ignored.

This electrode is made up of at least a first metal film including silver or a silver alloy, and a second metal film.

The first metal film may be a single-layer film of silver, or a single-layer film of a silver alloy, or a multilayer film in which silver or a silver alloy is disposed as the lowermost layer. In the case of a multilayer film, the films other than the lowermost layer may be formed from silver or a silver alloy, or from an electrode material not containing silver or a silver alloy. The first metal film may also be made up of a film comprised of silver or a silver alloy, and a nickel film disposed in part of the area between said silver film and the nitride semiconductor layer.

There may also be a gradient in the composition of the first metal film, from the nitride semiconductor layer side toward the second metal film side. For instance, on the nitride semiconductor layer side it may be a silver film or a film of an alloy containing silver and about 1% of an element other than silver, and on the second metal film side it may be a film of an alloy containing silver and about 5% of an element other than silver. If the proportion of the element other than silver versus silver increases moving away from the nitride semiconductor layer, it will be possible to obtain high light reflection characteristics and at the same time suppress reaction with silver on the second metal film side.

Examples of silver alloys include alloys of silver and one or more electrode materials selected from the group consisting of Pt, Co, Au, Pd, Ti, Mn, V, Cr, Zr, Rh, Cu, Al, Mg, Bi, Sn, Ir, Ga, Nd, and Re. Nickel does not readily form an alloy with silver, that is, its reaction with silver tends to be suppressed, so the silver film may also include elemental nickel.

Examples of the films other than the lowermost layer include single-layer films of one or more metals or alloys selected from the group including these electrode materials and nickel, and multilayer films of two or more layers.

Among these, a single-layer film of silver is preferable as the first metal film, and it is even more preferable to use a two-layer structure in which the upper layer is a metal that substantially does not react with silver, or in other words, a metal whose reaction with silver is suppressed, and the lower layer is silver or a silver alloy; a two-layer structure in which the upper layer is a noble metal and the lower layer is silver or a silver alloy; a three-layer structure in which the upper layer is a noble metal, the middle layer is a metal that substantially does not react with silver, and the lower layer is silver or a silver alloy (see FIG. 12b, for example); a four-layer structure in which the upper two layers are noble metals, the middle layer is a metal that substantially does not react with silver, and the lower layer is silver or a silver alloy; or a structure of four or more layers in which the upper layer is a noble metal, the middle two or more layers are a metal that substantially does not react with silver, and the lower layer is silver or a silver alloy.

In particular, when the first metal film is formed from a multilayer film of at least a film comprised of silver or a silver alloy, and a metal film disposed over this silver film or a silver alloy, whose reaction with silver is suppressed, such as when nickel is disposed in contact with silver or a silver alloy, and a noble metal is formed over this, the movement of silver on the side across from the nitride semiconductor layer can be dramatically reduced in the film comprised of silver or a silver alloy, and migration can be further prevented. In addition, this prevents a decrease in the reflection efficiency of the electrode with respect to light produced by the light emitting layer, allowing a semiconductor element with higher light emission efficiency to be obtained. Furthermore, when a layer of titanium, tantalum, or the like is formed between nickel and a noble metal, this prevents the silver touching the nitride semiconductor layer from moving across the nitride semiconductor layer surface, and further enhances the reliability of migration prevention.

Examples of the noble metal referred to here include platinum-series metals, gold, and the like, with platinum and gold being preferable.

Examples of metals that substantially do not react with silver, that is, metals whose reaction with silver is suppressed, include metals that either substantially do not react with silver, or whose reaction with silver is suppressed, at a temperature of 1000.degree. C. or lower, and more specifically, nickel (Ni), ruthenium (Ru), osmium (Os), iridium (Ir), titanium (Ti), vanadium (V), niobium (Nb), tantalum (Ta), cobalt (Co), iron (Fe), chromium (Cr), tungsten (W), and the like. Among these, nickel is preferable.

The phrase "metals that either substantially do not react with silver, or whose reaction with silver is suppressed" refers more specifically to metals that will not mix and form a solid solution with silver, or that will do so only with difficulty, and are included in this as long as the proportion in which they are mixed into silver is less than 5 wt %.

There are no particular restrictions on the thickness of the first metal film, but in the case of a single layer of silver or a silver alloy, for instance, this thickness is one that will allow light to be effectively reflected from the light emitting layer, and more specifically about 200 .ANG. to 1 .mu.m, or about 500 to 3000 .ANG., and especially about 1000 .ANG.. In the case of a laminar structure, the overall film thickness may be about 500 .ANG. to 5 .mu.m, or about 500 .ANG. to 1 .mu.m, and the silver or silver alloy film included therein can be suitably adjusted within the above range. In the case of a laminar structure, the silver or silver alloy film and the film laminated over it may be patterned in the same step and thereby given the same shape, but it is preferable to cover the lowermost layer of silver or silver alloy film with the film laminated thereover (preferably a metal film that will not react with silver). If this is done, then no matter which electrode material is formed as part of the first metal film over the metal film that will not react with silver, it will not directly react with the silver or silver alloy film, so a reaction with the silver can be prevented.

Also, as discussed above, it is preferable for the first metal film to include the silver or silver alloy film and a nickel film disposed in partial area between said silver film and the nitride semiconductor layer. Having a junction between the nitride semiconductor layer, the nickel, and the silver or silver alloy film increases the adhesion between the nitride semiconductor layer and the first metal film. This nickel may be present as a layer between the silver or silver alloy film and the nitride semiconductor layer, but is preferably formed as islands (see 65a in FIG. 12a). If there is a junction between the nitride semiconductor layer, the nickel, and the silver or silver alloy film, and a junction between the nitride semiconductor layer and the silver or silver alloy film, it will be possible to obtain a light emitting element with high adhesion between the nitride semiconductor layer and the electrode without decreasing the reflection efficiency. The nickel film is preferably present in a thickness of 30 angstroms or less, and especially 10 angstroms or less. In this case, good reflection characteristics can be obtained if the silver or silver alloy film has a thickness of 500 .ANG. or more, and no more than 5 .mu.m.

The adhesion at the junction between the nitride semiconductor layer, the nickel, and the silver or silver alloy film can be further increased by annealing at a temperature of 300.degree. C. or more, and 600.degree. C. or less after the formation of the first metal film or after the formation of the first metal film and the second metal film.

If the first metal film is a multilayer film, the films may all have the same size and shape (see 65a, 65b, and 65c in FIG. 12b), but the silver or silver alloy film of the lower layer is preferably covered by the film laminated thereover, not only on top surface but also on the side surface (FIGS. 12c and 12d). The purpose of the film laminated over the silver or silver alloy film is to prevent migration across the surface of the silver or silver alloy film. Accordingly, this film (or films) is preferably formed such that the thickness of the portion disposed on the side face (the thickness in the direction away from the silver or silver alloy film; the same applies hereinafter) is less than the thickness of the portion disposed over the silver or silver alloy film (p>s and q>r in FIG. 12c).

Also, in this case, the silver or silver alloy film is preferably formed so that it becomes smaller in size moving away from the semiconductor layer surface, that is, so that the sides have a sloped shape (see 67a in FIG. 12d). This is intended to ensure a stable state that strikes a good balance with silver migration on the surface and sides of the film.

Depending on the lamination state of the first metal film, such as when a nickel film is disposed directly over a single-layer film of silver, the first metal film is preferably crystallized (that is, in a crystalline state) at least at the interface with the nitride semiconductor layer. Crystallizing the first metal film ensures better ohmic characteristics with the nitride semiconductor layer, and allows the migration of silver to be prevented more effectively. The crystallization here may be either polycrystallization or monocrystallization, but a preferable state is one in which single crystals are dominant. "Crystallized state" means that the crystal grain boundaries can be recognized by observing a cross section by transmission electron microscopy (TEM), observing by scanning electron microscopy (SEM), measuring the electron diffraction pattern, observing with a super-thin film evaluation apparatus, or another such method. It is preferable here if the crystal grains can be seen to have a diameter (length, height, or width) of from 10 nm to 1 .mu.m, and preferably about 10 nm to 100 nm. For example, when the first metal film is obtained by TEM at the surface of the nitride semiconductor layer, it is preferable if crystal grains of from 10 nm to 1 .mu.m account for at least 50% of the total when the width in cross sectional view is 1.5 .mu.m.

Examples of the method for crystallizing the first metal film at the interface with the nitride semiconductor layer, or the method for forming part of the first metal film in a crystalline state, include forming the first metal film at a specific temperature by vapor deposition, sputtering, ion beam assisted vapor deposition, or another such method, and first forming the first metal film by vapor deposition, sputtering, ion beam assisted vapor deposition, or another such method, and then heat treating this product in the air or in a nitrogen atmosphere, for about 10 to 30 minutes, at a temperature range of about 300 to 600.degree. C.

There are no particular restrictions on the second metal film, but examples include single-layer and multilayer films of zinc (Zn), nickel (Ni), platinum (Pt), palladium (Pd), rhodium (Rh), ruthenium (Ru), osmium (Os), iridium (Ir), titanium (Ti), zirconium (Zr), hafnium (Hf), vanadium (V), niobium (Nb), tantalum (Ta), cobalt (Co), iron (Fe), manganese (Mn), molybdenum (Mo), chromium (Cr), tungsten (W), lanthanum (La), copper (Cu), silver (Ag), yttrium (Y), gold (Au), aluminum (Al), and other such metals and alloys, and ITO, ZnO.sub.2, SnO, and other such conductive oxide films.

Favorable examples include a platinum single-layer film, a gold single-layer film, a film with a two-layer structure in which the upper layer is gold and the lower layer is platinum, and a film with a three-layer structure in which the upper layer is platinum, the middle layer is gold, and the lower layer is platinum.

When the first metal film is a single-layer film of silver or a silver alloy, as discussed above, it is preferable for a metal that substantially does not react with silver to be disposed in at least the region of the second metal film that is in contact with the first metal film.

Particularly when the first metal film is formed as a multilayer film containing no silver or silver alloy in the uppermost layer, the second metal film preferably contains titanium, and a titanium film is preferably disposed at the lowermost layer of the second metal film.

It is also preferable to dispose a conductive material that is commonly used for connecting with other terminals, such as wire bonding over these electrodes, an example of which is disposing gold, platinum, or the like over the top surface (the connection region) of the second metal film. It is even more preferable to dispose a material with good adhesion to the insulating film (discussed below) on the top of the second metal film.

There are no particular restrictions on the thickness of the second metal film, but when a gold bump is formed on this film, for example, the second metal film may be set relatively thick, and when a eutectic (Au--Sn or the like) bump is formed, the second metal film may be set relatively thin. More specifically, the thickness is preferably adjusted as needed so that the total film thickness will be about 100 to 1000 nm.

It is preferable for the second metal film to completely cover the first metal film, that is, to cover substantially all of the top and the entire side surfaces of the first metal film, but a small amount of missing coverage is acceptable, to the extent that no active measures are performed on the second metal film to expose the first metal film. The phase "the second metal film covers the first metal film" may refer to a state in which the second metal film covers the first metal film via another film, or to a state preferably in which the second metal film covers the first metal film by contact or adhering thereto.

In particular, when the first metal film is a multilayer film, the second metal film is preferably thicker than the first metal film other than the silver or silver alloy film on the sides of the first metal film (y<z in FIG. 12d). Also, the second metal film is preferably such that the portions disposed on the sides are thicker than the portion disposed above the first metal film (x<z in FIGS. 12c and 12d). Doing this effectively prevents the movement of silver on the semiconductor surface.

Thus, particularly when the first metal film is comprised of silver or a silver alloy and the second metal film is a metal that substantially does not react with silver (a metal whose reaction with silver is suppressed, such as nickel) at least in the region where it is in contact with the first metal film, there will be no reduction in the amount of silver present near the interface with the nitride semiconductor layer. That is, the silver in the first metal film can be further prevented from being alloyed through diffusion, movement, or the like to the second metal film side as a result of reaction with the second metal film, the light emitted from the light emitting layer can be reflected more efficiently near the surface of the nitride semiconductor layer, and the light emission efficiency can therefore be further improved.

The insulating film need only cover the above-mentioned electrode, but is preferably a nitride film, for example. Typical examples of nitride films include SiN, TiN, SiO.sub.xN.sub.y, and other such single-layer films and multilayer films. Of these, a nitride film whose main component is silicon is preferred, and a single-layer film of SiN or the like is particularly favorable. Because the insulating film that covers the electrodes thus involves no film with a relatively high moisture content, such as SiO.sub.2, and is instead a film containing nitrogen, in addition to being able to easily form the insulating film with just an ordinary manufacturing process, the nitrogen atoms will trap moisture, so it is possible that moisture can be effectively prevented from penetrating an electrode comprised of silver or a silver alloy, and the migration of silver can be prevented.

The insulating film does not need to cover the electrode completely, and preferably covers everything except the region of the electrode necessary for connection with other terminals. A suitable thickness for the insulating film is about 400 to 1000 nm, for example.

Nevertheless, the insulating film preferably covers the entire surface of the semiconductor layer (such as a p layer) along with the electrode. This prevents silver migration from occurring at the p layer surface.

With the semiconductor element of the present invention, the above-mentioned nitride semiconductor layer may be made up of a nitride semiconductor layer of a first conduction type, a light emitting layer, and a nitride semiconductor layer of a second conduction type, laminated in that order on a substrate, for example. The "first conduction type" here refers to either p-type or n-type, and the "second conduction type" to a conduction type that is different from the first conduction type, that is, either n-type or p-type. Preferably, the semiconductor layer of the first conduction type is an n-type semiconductor layer, and the semiconductor layer of the second conduction type is a p-type semiconductor layer. With this configuration, good ohmic contact is ensured in a p-type nitride semiconductor layer, in which electron diffusion does not readily occur, so current diffusion is increased and the efficiency at which light is reflected from the light emitting layer can be maximized. Therefore, light can be taken off with greater efficiency, and a light emitting element of high quality and high performance can be obtained.

Examples of substrates that can be used include known conductive substrates and insulating substrates, such as sapphire, spinel, SiC, GaN, and GaAs. Of these, a sapphire substrate is preferred.

An insulating substrate may be removed eventually, but need not be removed. If the insulating substrate is removed, the p-electrode and n-electrode may be formed on the same side, or may be formed on different sides. If the insulating substrate is not removed, then the p-electrode and n-electrode are usually both formed on the same side of the nitride semiconductor layer.

The substrate need not be one whose surface is flat, and may be formed with regular or irregular peaks and valleys or the like to the extent that light generated in the light emitting layer can be scattered when reflected.

There are no particular restrictions on the nitride semiconductor layer, but a gallium nitride-based semiconductor compound such as In.sub.xAl.sub.yGa.sub.1-x-yN (0.ltoreq.X, 0.ltoreq.Y, X+Y.ltoreq.1) can be used favorably. The nitride semiconductor layer may have a single-layer structure, but may also have a laminated structure such as a homo-structure, hetero-structure, or double hetero-structure having an MIS junction, a PIN junction, or a PN junction. It is also possible to use a super-lattice structure, or a single quantum well structure or multiple quantum well structure which produces a quantum effect. The layer may also be doped with either n-type or p-type impurities. This nitride semiconductor layer can be formed by a known process such as metal organic chemical vapor deposition (MOCVD), hydride vapor phase epitaxy (HVPE), molecular beam epitaxy (MBE), sputtering, ion plating, or electron shower. There are no particular restrictions on the thickness of the nitride semiconductor layer, and various thicknesses can be applied as appropriate.

As the laminated structures of the nitride semiconductor layers include described in (1) to (5) below, for example.

(1) A buffer layer composed of GaN of a thickness of 200 .ANG., an n-type contact layer composed of an Si doped n-type GaN of a thickness of 4 .mu.m, a light emitting layer having a single quantum well structure and composed of undoped In.sub.0.2Ga.sub.0.8N of a thickness of 30 .ANG., a p-type clad layer composed of an Mg doped p-type Al.sub.0.1Ga.sub.0.9N of a thickness of 0.2 .mu.m, and a p-type contact layer composed of an Mg doped p-type GaN of a thickness of 0.5 .mu.m.

(2) A buffer layer composed of AlGaN of a thickness of about 100 .ANG., an undoped GaN layer of a thickness of 1 .mu.m, an n-side contact layer composed of GaN containing 4.5.times.10.sup.18/cm.sup.3 of Si of a thickness of 5 .mu.m, an n-side first multi-layer film layer composed of three layers--a bottom layer composed of an undoped GaN of 3000 .ANG., an intermediate layer composed of GaN containing 4.5.times.10.sup.18/cm.sup.3 of Si of a thickness of 300 .ANG., and an upper layer composed of undoped GaN of a thickness of 50 .ANG. (an overall thickness of 3350 .ANG.), an n-side second multi-layer film layer having a superlattice structure in which 10 layers each of a 40 .ANG. nitride semiconductor layer composed of undoped GaN and a 20 .ANG. nitride semiconductor layer composed of undoped In.sub.0.1Ga.sub.0.9N are repeatedly alternately laminated and a 40 .ANG. nitride semiconductor layer composed of undoped GaN is formed thereon (an overall thickness of 640 .ANG.), a light emitting layer having a multiquantum well structure in which six layers each of a 250 .ANG. barrier layer composed of undoped GaN and a 30 .ANG. well layer composed of In.sub.0.3Ga.sub.0.7N are repeatedly alternately laminated and a 250 .ANG. barrier composed of undoped GaN is formed thereon (an overall thickness of 1930 .ANG.), an p-side multi-layer film layer having a superlattice structure in which 5 layers each of a 40 .ANG. nitride semiconductor layer composed of Al.sub.0.15Ga.sub.0.85N containing 5.times.10.sup.19/cm.sup.3 of Mg and a 25 .ANG. nitride semiconductor layer composed of In.sub.0.03Ga.sub.0.97N containing 5.times.10.sup.19/cm.sup.3 of Mg are repeatedly alternately laminated and a 40 .ANG. nitride semiconductor layer composed of Al0.15Ga.sub.0.85N containing 5.times.10.sup.19/cm.sup.3 of Mg is formed thereon (an overall thickness of 365 .ANG.), and a p-side contact layer composed of GaN containing 1.times.10.sup.20/cm.sup.3 of Mg of 1200 .ANG. thickness.

(3) A buffer layer composed of AlGaN of a thickness of about 100 .ANG., an undoped GaN layer of a thickness of 1 .mu.m, an n-side contact layer composed of GaN containing 4.5.times.10.sup.18/cm.sup.3 of Si of a thickness of 5 .mu.m, an n-side first multi-layer film layer composed of three layers--a bottom layer composed of an undoped GaN of 3000 .ANG., an intermediate layer composed of GaN containing 4.5.times.10.sup.18/cm.sup.3 of Si of a thickness of 300 .ANG., and an upper layer composed of undoped GaN of a thickness of 50 .ANG. (an overall thickness of 3350 .ANG.), an n-side second multi-layer film layer having a superlattice structure in which 10 layers each of a 40 .ANG. nitride semiconductor layer composed of undoped GaN and a 20 .ANG. nitride semiconductor layer composed of undoped In.sub.0.1Ga.sub.0.9N are repeatedly alternately laminated and a 40 .ANG. nitride semiconductor layer composed of undoped GaN is formed thereon (an overall thickness of 640 .ANG.), a light emitting layer having a multiquantum well structure in which a barrier layer composed of undoped GaN of 250 .ANG. thickness is first formed and then 6 layers each of a well layer composed of In.sub.0.3Ga.sub.0.7N of 30 .ANG. thickness, a first barrier layer composed of In.sub.0.02Ga.sub.0.98N of 100 .ANG. thickness and a second barrier layer composed of undoped GaN of 150 .ANG. thickness are repeatedly alternately laminated (an overall thickness of 1930 .ANG.) (3-6 layers are preferably repeatedly alternately laminated), an p-side multi-layer film layer having a superlattice structure in which 5 layers each of a 40 .ANG. nitride semiconductor layer composed of Al.sub.0.15Ga.sub.0.85N containing 5.times.10.sup.19/cm.sup.3 of Mg and a 25 .ANG. nitride semiconductor layer composed of In.sub.0.03Ga.sub.0.97N containing 5.times.10.sup.19/cm.sup.3 of Mg are repeatedly alternately laminated and a 40 .ANG. nitride semiconductor layer composed of Al.sub.0.15Ga.sub.0.85N containing 5.times.10.sup.19/cm.sup.3 of Mg is formed thereon (an overall thickness of 365 .ANG.), and a p-side contact layer composed of GaN containing 1.times.10.sup.20/cm.sup.3 of Mg of 1200 .ANG. thickness.

Furthermore, by making a bottom layer composed of undoped GaN of 3000 .ANG. arranged on the n-side the bottom layer of a three layer structure composed of a first layer composed of undoped GaN of 1500 .ANG. thickness, a second layer composed of GaN containing 5.times.10.sup.17/cm.sup.3 of Si of 100 .ANG. thickness, and a third layer composed of undoped GaN of 1500 .ANG. thickness, it will become possible to control Vf fluctuation that accompanies the elapsed drive time of a light emitting element.

Further, GaN or AlGaN layer of 2000 .ANG. thickness may be provided between the p-side laminated layer and the p-side contact layer. This layer is formed as undepoed layer, but represents p-type conductivity by means of diffusion of Mg from the adjacent layer thereto. Forming this layer improves a withstand electrostatic voltage in the light emitting element. This layer can be omitted for a light emitting device which has electrostatic protective function, but it is preferable to form this layer when the light emitting element do not have electrostatic protective means, such as an electrostatic protective element outside of the light emitting element so that a withstand electrostatic voltage can be improved.

(4) A buffer layer, an undoped GaN layer, an n-side contact layer composed of GaN containing 6.0.times.10.sup.18/cm.sup.3 of Si, an undoped GaN layer (an n-type nitride semiconductor layer of an overall thickness of 6 nm), a light emitting layer having a multiquantum well structure in which 5 layers each of a GaN barrier layer containing 2.0.times.10.sup.18/cm.sup.3 of Si and a InGaN well layer are repeatedly alternately laminated (overall thickness: 1000 .ANG.), and a p-type nitride semiconductor layer composed of GaN containing 5.0.times.10.sup.18/cm.sup.3 of Mg of a thickness of 1300 .ANG..

Further, InGaN layer may be provided on the p-type nitride semiconductor layer of a 30 to 100 .ANG. thickness, preferably a 50 .ANG. thickness, this layer can be placed in contact with a positive electrode, and will become a p-side contact layer.

(5) A buffer layer, an undoped GaN layer, an n-side contact layer composed of GaN containing 1.times.10.sup.19/cm.sup.3 of Si, an undoped GaN layer (an n-type nitride semiconductor layer of an overall thickness of 6 nm), a light emitting layer having a multiquantum well structure in which 7 layers each of a GaN barrier layer containing 3.0.times.10.sup.18/cm.sup.3 of Si and an InGaN well layer are repeatedly alternately laminated (overall width: 800 .ANG.), and a p-type nitride semiconductor layer composed of GaN containing 2.5.times.10.sup.20/cm.sup.3 of Mg. InGaN layer may be provided on the p-type nitride semiconductor layer of a 30 to 100 .ANG. thickness, preferably a 50 .ANG. thickness.

In plan view, a semiconductor element made up of these semiconductor layers is usually square or approximately square, and the first semiconductor layer has an exposed region in which part of the second semiconductor layer and the light emitting layer are removed as desired in the depth direction of the first semiconductor layer in part of the region of one semiconductor element, so that the surface thereof is exposed. A first electrode is formed on the surface of this exposed first semiconductor layer.

With the semiconductor element of the present invention, a plurality of peaks and valleys are preferably formed in the part of the exposed region where the first electrode is not formed (including the outer edge of the semiconductor element). In other words, as will be discussed below, even though a light emitting layer is present, the plurality of peaks and valleys are preferably formed in the region where holes and electrons are not supplied, so this region does not function as a light emitting layer and does not itself emit light.

The results of thus forming peaks and valleys are believed to be that (1) the light guided through the first semiconductor layer is taken into the convex portions, and taken out from the tops of these convex portions or from the middle portions thereof, (2) the light guided through the first semiconductor layer is scattered at the bases of the convex portions and taken off, and (3) the light coming out sideways from the light emitting layer end face is reflected and scattered by the plurality of convex portions and taken off; that is, light coming out laterally (in the sideways direction of the semiconductor element) can be selectively emitted to the second semiconductor layer side or the substrate side (the vertical direction of the semiconductor element), and as a result, the light take-off efficiency can be increased by about 10 to 20%, for example, and the directionality of the light can be controlled. Particularly with a semiconductor element having a structure in which the light emitting layer is sandwiched by layers of lower refractive index (known as a double hetero-structure), because light is trapped between these layers of lower refractive index, most of the light ends up moving sideways, and providing peaks and valleys is particularly effective with light emitting elements with a structure such as this. Furthermore, providing a plurality of peaks and valleys makes it possible to take off light uniformly over the entire region on the substrate side and the second semiconductor layer side.

These peaks and valleys may be produced by performing a special step for forming convex portions, such as by growing a semiconductor layer on the exposed first semiconductor layer, but the concave and convex portions are preferably formed at the same time, such as in the course of exposing the first semiconductor layer, or in the course of making specific regions into a thin film in order to divide into individual chips. This avoids increasing the number of manufacturing steps. Thus, the peaks and valleys are made up of the same laminar structure as the semiconductor laminar structure of the semiconductor element, that is, of a plurality of layers of different materials, so the difference in the refractive index of the various layers allows the light taken in to the convex portions to be reflected more easily at the layer interfaces, the result of which is believed to be that these peaks and valleys contribute to increased light take-off to the second semiconductor layer side and the substrate side.

The convex portions of these peaks and valleys may be higher than at least the interface between the light emitting layer and the first semiconductor layer adjacent thereto in a semiconductor element cross section, but the tops thereof are preferably located more on the second semiconductor layer side than the light emitting layer, and even more preferably are substantially the same height as the second semiconductor layer. In other words, the convex portions are preferably formed so that their the tops are higher than the light emitting layer. If the convex portions are configured so as to include the second semiconductor layer, the tops thereof will be approximately the same height, so they will not be blocked by the second electrode (discussed below) or the like, and light can be taken off more effectively from the tops of the convex portions to the second semiconductor layer side (and substrate side). If the convex portions are configured so as to be higher than the second semiconductor layer, and preferably higher than the second electrode, light can be taken off even more effectively. The concave portions between the convex portions may be lower than at least the interfaces between the light emitting layer and the second semiconductor layer adjacent thereto, and are preferably formed so as to be lower than the light emitting layer.

There are no particular restrictions on the density of the peaks and valleys, which can be laid out in a number of at least 100, and preferably at least 200, and even more preferably at least 300, and better yet at least 500, per semiconductor element. Doing this further enhances the above-mentioned effects. Viewed from the electrode formation side, the proportion accounted for by the region in which the peaks and valleys are formed can be at least 20 percent, and preferably at least 30 percent, and even more preferably at least 40 percent. There are no particular restrictions on the upper limit, but 80 percent or less is preferred. The surface area of a single convex portion, as measured at its base, can be from 3 to 300 .mu.m.sup.2, and preferably 6 to 80 .mu.m.sup.2, and even more preferably 12 to 50 .mu.m.sup.2.

The vertical cross sectional shape of the convex portions may be square, trapezoidal, semicircular, or the like, but is preferably trapezoidal, in other words, a truncated cone in which the convex portion itself gradually tapers. The angle of inclination of the convex portion in this case can be from 30.degree. to 80.degree., for example, with 40.degree. to 70.degree. being preferable. In other words, if the convex portions are sloped so that they gradually taper toward their distal ends, all of the light from the light emitting layer is reflected by the convex portion surface, or light that has been guided through the first semiconductor layer is scattered, and as a result, light can be effectively taken off to the second semiconductor layer side (and substrate side). In addition, it is impossible to more easily control directivity of light as well as to take off light more uniformly as a whole.

If the convex portions have a truncated conical shape, then concave portions may further be formed on the top side of the trapezoid (the second semiconductor layer side). The result of this is that the concave portions formed at the tops of the convex portions make it easier for light to be emitted to the second semiconductor layer side (and substrate side) when light that has been guided through the first semiconductor layer penetrates into the convex portions.

Furthermore, the convex portions are preferably disposed at least partially overlapping in groups of two or more, and preferably three or mor