Semiconductor device manufacturing: process – Semiconductor substrate dicing
Reexamination Certificate
2001-07-10
2002-10-22
Nelms, David (Department: 2818)
Semiconductor device manufacturing: process
Semiconductor substrate dicing
C438S479000
Reexamination Certificate
active
06468882
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a method of producing a single crystal gallium nitride (GaN) substrate and a gallium nitride substrate itself for making light emitting devices of light emitting diodes (LEDs) or laser diodes (LDs) built with the nitride semiconductors consisting of the group 3-5 elements. Gallium nitride and the like (GaInN, AlInN, InAlGaN) are semiconductors having wide band gaps which correspond to blue light or blue/green light.
This application claims the priority of Japanese Patent Application No. 2000-207783 filed on Jul. 10, 2000 which is incorporated herein by refernce.
2. Description of Related Art
Blue-light LEDs making use of the nitride semiconductor (GaN, GaInN, AlInN, AlInGaN) have been already put into practice on mass scale. There is no natural single crystal GaN mineral. It has been impossible to grow gallium nitride (GaN) single crystal ingots from a GaN melt, since heating converts solid phase GaN directly into vapor phase GaN without a liquid phase. Almost all of the nitride semiconductor LEDs on sale are produced upon sapphire (&agr;-Al
2
O
3
) monocrystal (single crystal) substrates. Sapphire belongs to a trigonal symmetry group.
The GaN LEDs are fabricated by piling n-type or p-type films of GaN, GaInN, AlInN, InAlGaN or so (called “GaN type films” collectively) heteroepitaxially upon single crystal sapphire substrates. The GaN type crystals have hexagonal symmetry. GaN belongs to a different symmetry group from sapphire. Single crystal sapphire, however, turned out a very stable, suitable material as a substrate which allows the GaN type films to grow heteroepitaxially. The excellency of sapphire allows device makers to produce a plenty of inexpensive blue light or blue-green light GaN/sapphire LEDs.
However, sapphire is a very rigid material. A sapphire single crystal has no cleavage plane. Cleavage is a convenient property for cutting a wide device-made wafer into individual device chips and for making resonator mirrors of laser diodes. Lack of natural cleavage compels device makers to cut (dice) mechanically sapphire wafers crosswise and lengthwise into device chips with an application of strong forces. The dicing process incurs an extra cost and decreases a yield in the production of GaN LEDs. Uncleavability of GaN crystals subjects GaN/sapphire LDs to other difficulties besides the problems of dicing. Uncleavable GaN prevents GaN laser diodes from making resonator mirrors by natural cleavage, which invites difficulties on laser oscillation performance and on production cost.
Another difficulty originates from the fact that sapphire is an insulator. Electrical insulation of sapphire prohibits the on-sapphire LEDs from taking a vertical electrode structure which allocates top and bottom surfaces for two electrodes (anode and cathode). Instead of the common vertical electrode structure, the sapphire-carried LEDs take a horizontal electrode structure by etching partially upper films, revealing a part of an n-type GaN-type film, making a cathode on the n-type GaN-type film and producing an anode upon a top p-type GaN-type film. The n-type GaN-type film should be thick enough to allow current to flow in the horizontal direction. The n-type electrode pad should be bonded to a cathode pin by wirebonding. The extra etching, the extra thick film and an extra wirebonding raise the cost by increasing the time of fabrication and the number of steps. Furthermore, the larger chip surface required for allocating two electrodes on the same surface. This point incurs an increment of the cost.
The sapphire substrates are suffering from these difficulties. Someone proposes a use of silicon carbide (SiC) as a substrate for GaN light emitting devices (LEDs and LDs). Silicon carbide belongs to the hexagonal symmetry group like GaN. Natural cleavage accompanies a silicon carbide crystal. Natural cleavage will facilitate to cut a GaN optoelectronic device-loaded SiC wafer into individual device chips and will conveniently make resonator mirrors in GaN LDs. SiC is electrically conductive. The conductive SiC allows the vertical electrode structure which allocates top and bottom surfaces to an anode and a cathode. Silicon carbide substrates favor the fabrication process of making GaN type LEDs. Silicon carbide, however, has some drawbacks. Single crystal silicon carbide is highly expensive. The difficulty of producing SiC single crystals will jeopardize a continual, stable supply of SiC substrates. The crystalline property of GaN films grown on SiC substrates are still bad at present. SiC is not deemed to be the suitable material as a substrate of GaN light emission devices. SiC is rather inferior to sapphire as a substrate for GaN LEDs. GaN/SiC LEDs are not brought onto market yet.
Sapphire or silicon carbide as a substrate induces many dislocations and other defects in GaN type films grown thereupon owing to mismatches of lattice constants and thermal expansion coefficients between the upper films and the bottom substrate. GaN/sapphire LEDs on sale have about 1×10
9
cm
−2
dislocations in the GaN epitaxial films.
It is said that GaN films heteroepitaxially grown on silicon carbide (SiC) would have about 1×10
8
cm
−2
. Plenty of dislocations induced in the GaN films cause no serious damage to the practical utility of the GaN/sapphire LEDs. The GaN/sapphire LEDs enjoy a long lifetime despite the affluence of dislocations.
However, in the case of GaN/sapphire LDs which require large current density, experiments clarify the fact that the big dislocation density forbids the on-sapphire GaN type LDs from having a long lifetime. The big current increases the dislocations and other defects. Furthermore, an LD requires resonator consisting of two parallel mirrors at both ends of a cavity. Sapphire substrates without cleavage require elaborate dicing and polishing for making flat smooth mirrors with high reflection. The fabrication of the resonator mirror would raise the cost of GaN/sapphire LDs. The high cost and short lifetime degrade sapphire as a substrate for GaN LDs. From the reasons, sapphire and silicon carbide are not the most suitable material for the substrates of GaN LDs.
The best substrate should be a gallium nitride (GaN) single crystal (monocrystal). If a wide GaN single crystal substrate were obtained, the problem of the mismatches of the lattice constant and the thermal expansion would be entirely solved. GaN has natural cleavage in {1−100} planes. GaN is a semiconductor. Impurity doped GaN substrates have enough conductivity. GaN substrates would be superior to sapphire substrates in cleavability and conductivity. Gallium nitride single crystals would be the most favorable substrates for GaN LDs. However, crystal growth technology has not been matured for GaN yet. It is difficult at present to produce gallium nitride single crystals with a large size sufficient for the substrates of GaN LDs.
Heating converts solid GaN not to liquid GaN but to vapor GaN. High pressure and high temperature are requisites for making a GaN melt. It is said that it would be possible to synthesize a gallium nitride single crystal from a GaN melt in a state of thermal equilibrium maintained by ultrahigh pressure and high temperature. However, even if it succeeded, the ultrahigh pressure method would synthesize only a small GaN crystal which would be insufficient for the substrate of GaN LDs. The inventors of the present invention are unaware of such a report of succeeding in making a GaN bulk single crystal by the ultrahigh pressure method. Such a liquid phase method is hopeless for supplying wide gallium nitride crystals on an industrial scale.
Someone suggested a method of covering a sapphire substrate with a mask having windows, piling gallium nitride molecules through the mask upon the sapphire substrate and making a GaN film on the sapphire in vapor phase. The mask having windows has an effect of reducing dislocations in the GaN film.
{circle around (1)} Akira Usui, “Thick Layer
Kasai Hitoshi
Motoki Kensaku
Okahisa Takuji
Dang Phuc T.
Nelms David
Smith , Gambrell & Russell, LLP
Sumitomo Electric Industries Ltd.
LandOfFree
Method of producing a single crystal gallium nitride... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of producing a single crystal gallium nitride..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of producing a single crystal gallium nitride... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2921254