WO2000045443A1

WO2000045443A1 - High performance light emitting diodes

Info

Publication number: WO2000045443A1
Application number: PCT/US2000/001943
Authority: WO
Inventors: Yu-Hwa Lo; Tuoh-Bin Ng
Original assignee: Nova Crystals, Inc.
Priority date: 1999-01-28
Filing date: 2000-01-27
Publication date: 2000-08-03

Abstract

A super-bright LED is achieved by substantially enhancing the output coupling of the device. This is achieved in several ways. First, a single LED die is divided into a group of mesas, each of which functions as a light emitting unit. This ensures that the light from the four edge emitting cones of the LED is utilized. Second, each mesa is designed so that the ratio of the radius of the pumped region to the radius of the periphery of the device is less than one-third. Third, using circular mesas reduces reflection loss. Combining these features with a reflective layer between the LED epilayer and the opaque substrate of the LED die further optimizes the optical coupling of the LED. The radius of the pumped region is reduced by a variety of methods: current confinement by forming a lateral oxidation layer that defines an aperture; current confinement by forming a central trench; current confinement by a high-resistivity layer formed by ion implantation, regrowth, or deposition of transparent layers, or material joining; current confinement by diffusion or implantation of dopant impurities creating a high-conductivity region inside a non-conductive region; and by extending a non-conductive mesa region around a conductive light emitting region.

Description

HIGH PERFORMANCE LIGHT EMITTING DIODES

FIELD OF THE INVENTION

The invention pertains to the field of light emitting diodes (LEDs). More particularly, the invention pertains to high-brightness and super-bright LEDs.

BACKGROUND OF THE INVENTION

High-brightness LEDs, particularly in the visible regime, have very high commercial values for their wide-range applications in display, traffic light, signs, automobile rear lights, and many other illuminating devices. High-brightness LEDs are different from low- and intermediate-brightness LEDs because they emit up to 6 times more photons to the outside world under the same injection current (typically 20 to 70 mA). The performance improvement comes from the superior epitaxial material quality and device structure. In the red/orange/yellow color regimes, high brightness LEDs use AlInGaP instead of AlGaAs quantum wells as active regions to achieve high "internal" quantun efficiency, defined as the ratio of the number of photons generated by the quantum wells and the number of injected carriers. However, as many intermediate- brightness LEDs are also made of AlInGaP quantum wells, the difference between high- brightness LEDs and intermediate-brightness LEDs is mostly in their "external" coupling efficiency, defined as the ratio of the optical power emitting to the outside world and the optical power generated internally.

Because semiconductor normally has a much higher refractive index than air

(about 3 : 1 ratio), light generated inside semiconductor can only be coupled to the outside world if its incident angle to the semiconductor/air boundary is less than the total reflection angle θ_t defined by Snell's law

θ_t = sin^"' (nJn.) [1]

where ri_o is the index of the air and n, is the index of the semiconductor. Assuming tio= 1 and n_s = 3, θ_t is about 20 degrees. Figs. 1 A-ID show schematically 4 designs of conventional low and intermediate brightness LEDs. Also shown in Figs. 1 A- ID are full and half light cones within which the light will not be totally reflected according to Eq. 1. Each full cone has an angle of 2θ_t. For LED die 501 in Fig. 1 A, only one full cone 30 ultimately contributes to the light output and the rest is absorbed by opaque substrate 10 before it reaches the device/air boundary. With the introduction of a bottom reflector 80 (Fig. 1C), reflected bottom emitting full cone 70 can also be coupled out so the brightness of the LED is almost doubled. When a thick transparent layer 40 is grown on top of light emitting layer 20 as shown in Figs. IB and ID, each of the four edge emitting cones will have its upper half coupled to the outside world although its bottom half will still be absorbed by substrate 10. To summarize, the four LED dies 501, 502, 503, and 504 in Figs. 1 A-ID have their light output of 1 cone (Fig. 1 A), 3 cones (Fig. IB), 2 cones (Fig.

1C), and 4 cones (Fig. ID), respectively. For the devices shown in Figs. IB and ID, the thickness h of the top layer needs to satisfy the following condition:

h > d/2 sin θ_t = (nJn_s)d/2 ≡ ά lβ [2] where d is the die size. For most commercial LEDs, the die size is typically 9 to 12 mils or 230 to 300 μm. To satisfy Eq. 2, the thickness of the top transparent layer needs to be in the order to 40 to 50 μm, which can not be easily achieved by OMCVD (Organo-Metallic Chemical Vapor Deposition) or MBE (Molecular Beam Epitaxy) techniques. To increase the device throughput, VPE (Vapor Phase Epitaxy) growth technique is usually used to grow the thick top layer after the growth of LED epitaxial layers by OMCVD. Therefore, increasing the device brightness also increases the cost.

Different from the low and intermediate brightness LEDs as shown in Figs. 1A- 1D, Kish et al. has shown a different LED process to achieve high brightness red/orange/yellow LEDs with outputs in all 6 light cones. The device fabrication process of Kish is summarized in Figs. 2A-2C. In Fig. 2A, the AlInGaP epitaxial layers 20 are first grown on a GaAs opaque substrate 10 by OMCVD, followed by a VPE growth of a thick transparent epilayer 40. In Fig. 2B, the opaque GaAs substrate 10 is then etched away, leaving the epilayers 20 and 40 free-standing or temporarily attached to a dummy substrate. In Fig. 2C, epilayers 20 and 40 are finally transferred to a GaP transparent substrate 90 using wafer fusion techniques. The fused junction has been shown to be both electrically conductive and optically transparent. Therefore, all six light cones 30, 120,

100, 110 (cones into and out of the plane of the figure not shown) propagate through transparent material and are coupled to the outside world, yielding a 6-cone high- brightness LED 505.

Table 1 summarizes the (theoretically achievable) device performance and the cost contributing factors for all the existing AlInGaP visible LEDs.

Table 1

SUMMARY OF THE INVENTION

High-brightness light emitting diodes find wide applications in display, traffic light, signs, and other illuminating devices. In the not too far future, it is possible that they will replace incandescent and fluorescent light to become the main source for lighting. Although the quality of epitaxial material for most visible LEDs, particularly in the red, orange, and yellow regime, is very high, the external efficiency of the devices is limited due to a large refractive index mismatch between air and semiconductor and due to the high absorption of the semiconductor substrate. As a result, most of the light is totally reflected at the air/semiconductor interface and eventually absorbed by the substrate. This invention shows that by optimizing the device geometry, one can greatly enhance the external device efficiency by more than 10 times even if the device sits on a non- transparent substrate. The concept can be applied not only to red/orange/yellow semiconductor LEDs which have often been used as examples for illustration purpose, but also to LEDs at other wavelengths (colors) and made of other materials including high- bandgap and organic semiconductors.

Briefly stated, a super-bright LED is achieved by substantially enhancing the output coupling of the device. This is achieved in several ways. First, a single LED die is divided into a group of mesas, each of which functions as a light emitting unit. This ensures that the light from the four edge emitting cones of the LED is utilized. Second, each mesa is designed so that the ratio of the radius of the pumped region to the radius of the periphery of the device is less than one-third. Third, using circular mesas reduces reflection loss. Combining these features with a reflective layer between the LED epilayer and the opaque substrate of the LED die further optimizes the optical coupling of the LED. The radius of the pumped region is reduced by a variety of methods: current confinement by forming a lateral oxidation layer that defines an aperture; current confinement by forming a central trench; current confinement by a high-resistivity layer formed by ion implantation, regrowth, or deposition of transparent layers, or material joining; current confinement by diffusion or implantation of dopant impurities creating a high-conductivity region inside a non-conductive region; and by extending a non-conductive mesa region around a conductive light emitting region.

According to an embodiment of the invention, a light emitting diode includes an opaque substrate; a reflective layer on the substrate; and a plurality of LED epilayer mesas on the reflective layer. According to a feature of the invention, the light emitting diode further includes restricting means for restricting a pumped region in each mesa.

According to an embodiment of the invention, a method for making a light emitting diode includes the steps of: a) growing a light emitting diode epilayer on an opaque layer such that a reflecting surface is above the opaque layer; and b) etching a plurality of mesas in the light emitting diode epilayer, wherein each of the mesas has a mesa width to mesa height ratio of less than or equal to three.

According to an embodiment of the invention, a method for making a light emitting diode includes the steps of: growing a reflecting layer on an opaque layer; growing a light emitting diode epilayer on the reflecting layer; and etching a plurality of mesas in the light emitting diode epilayer, wherein each of the mesas has a mesa width to mesa height ratio of less than or equal to three.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 A shows a schematic design of a conventional low-brightness LED according to the prior art.

Fig. IB shows a schematic design of a conventional intermediate-brightness LED according to the prior art.

Fig. IC shows a schematic design of a conventional low-brightness LED according to the prior art.

Fig. ID shows a schematic design of a conventional intermediate-brightness LED according to the prior art.

Fig. 2 A shows a schematic design of the steps of making a known high-brightness LED according to the prior art.

Fig. 2B shows a schematic design of the steps of making a known high-brightness LED according to the prior art. Fig. 2C shows a schematic design of the steps of making a known high-brightness LED according to the prior art.

Fig. 3 A shows a schematic design of the process and product of making a high-brightness LED according to an embodiment of the present invention.

Fig. 3B shows a cross section of a high-brightness LED according to an embodiment of the present invention.

Fig. 3C shows schematic design of the high-brightness LED of Fig. 3B.

Fig. 4 shows a schematic design of a high-brightness LED using a reflector layer.

Fig. 5A shows a schematic diagram used to explain the geometry of LED devices as it relates to the embodiments of the present invention.

Fig. 5B shows a schematic diagram used to explain the geometry of LED devices as it relates to the embodiments of the present invention.

Fig. 6A shows a cross sectional view of a super-bright LED according to an embodiment of the present invention.

Fig. 6B shows a top plan view of a super-bright LED according to an embodiment of the present invention.

Fig. 7A shows a cross sectional view of a super-bright LED according to an embodiment of the present invention.

Fig. 7B shows a top plan view of a super-bright LED according to an embodiment of the present invention.

Fig. 8 A shows a cross sectional view of a super-bright LED according to an embodiment of the present invention.

Fig. 8B shows a top plan view of a super-bright LED according to an embodiment of the present invention. Fig. 9 A shows a cross sectional view of a super-bright LED according to an embodiment of the present invention.

Fig. 9B shows a top plan view of a super-bright LED according to an embodiment of the present invention.

Fig. 10A shows a cross sectional view of a super-bright LED according to an embodiment of the present invention.

Fig. 10B shows a top plan view of a super-bright LED according to an embodiment of the present invention.

DESCRIPTION OF THE PREFERRED EMBODIMENT

This invention uses geometric effects to optimize the optical coupling of LEDs.

Two classes of device geometry design are discussed. The first design yields high brightness LEDs having similar performance to the LEDs developed by Kish et al. at Hewlett Packard, but with a much simpler process and lower cost than the HP device. The second design, which is a further improvement over the first design, creates LEDs that are 100% brighter than the state-of-the-art Hewlett Packard red/orange/yellow LEDs.

Therefore, the LEDs fabricated using the second design are called super high brightness LEDs. The second design is expected to produce devices at a fraction of the cost of today's high brightness LEDs.

First Design Consideration

From Eq. 2, it is clear that the light in four edge emitting cones can only be utilized if the transparent layer thickness on either side of the LED active layer is greater than about 1/6 of the width of the device. The practical die size of 230 to 300 μm makes the total required transparent layer thickness about 100 μm, which stretches the limit of epitaxial growth. Therefore, a transparent substrate is often used for one of the transparent layers. However, if we divide a single LED die into a group of mesas, each of which functions as a light emitting unit, then the effective width of the device can be reduced significantly. The actual number of units can be one to a few dozens. If we divide a LED die into mesa units with each mesa being 30 to 50 μm wide, the total transparent layer only needs to be 10 to 15 μm thick according to Eq. 2. This thickness can be routinely achieved in one single OMCVD growth. All these mesas may be electrically connected to a common electrode and function as a single LED die or functions individually.

Referring to Figs. 3 A-3C, one example of the innovative LED design and process, which consists of 5 light emitting units, is shown. An LED epilayer 20 is grown on an opaque layer 10. Five mesas 25, 26, 27, 28, 29 are etched in layer 20, with each mesa having a mesa width and height ratio of less than or approximately equal to 3, with a typical mesa height of 6-15μm and corresponding mesa width of 20-50μm.

The light emitting from the edges of each mesa 25, 26, 27, 28, 29 hits a reflecting surface 200 after it leaves the mesa. This reflecting surface 200, made of metal or multilayers of semiconductors or dielectric materials, reflects the light outward. Assuming a total bias current of LED die 506 is the same as a conventional LED die, i.e., typically between 20 and 70 mA, the net output power is equivalent to a single LED die without mesas that emits 5 full cones of light. Mesa 25 is shown with a top emitting cone 30 and two side emitting cones 100, 110. Mesa 26 is shown with a top emitting cone 31 and two side emitting cones 101, 111. Cones extending into and out of the plane of the figure are not shown.

Referring to Fig. 4, if a reflector 80 is grown on top of a substrate 10, bottom cones 70 and 71 from mesas 25 and 26, respectively, can also contribute to the output power so the net power output would be equal to today's high brightness LED as shown in Fig. 2.

However, the new design in Fig. 4 only requires one OMCVD growth, one GaAs substrate, and an additional mesa etch process over the prior art. This is much simpler to fabricate than the current state-of-the-art device that requires two epitaxial growths, two substrates (GaAs and GaP), GaAs substrate removal, and wafer fusion (Fig. 2).

Although we use AlInGaP red/orange/yellow LED as an example because such a

LED is one of the most popular LEDs, the same design concept can be readily applied to other LED materials. For organic LEDs, the LED layers are organic semiconductors that may be deposited on a nontransparent substrate such as silicon. If mesas are etched following the design parameters of Eq. 2, the problem of substrate absorption can also be solved, yielding high-brightness organic LEDs at any colors. For (In)GaN-based green, blue, and purple LEDs, sapphire is the most commonly used substrate in spite of its large lattice and thermal mismatch with (In)GaN. Although sapphire is a transparent substrate so the substrate absorption problem mentioned before does not exist, major research efforts have been made to use Si or other lower cost and conductive substrates to replace sapphire. Our design assures that the new substrates that replace sapphire do not suffer from the substrate absorption while maintaining other advantages of lower substrate cost and more uniform current injection.

Second Design Consideration

Referring to Figs. 5A-5B, the second design further optimizes the geometry of the LED devices. For a uniformly pumped LED, light can be generated at any position in the pumped area with different distances from the boundaries of the device. Assuming that the device has a square shape, it can be shown mathematically that the output coupling of a uniformly pumped device is approximately the same as a device with all the injected current concentrated at a small central region of the device as shown in Fig. 5 A. In addition to the top and bottom cones (not shown in Fig. 5 A), four cones about 40-degrees each contribute to the output from the sides of the device. However, if we keep the pumping area and position the same but make the device a circular shape as shown in Fig. 5B, the light propagating in all directions within the plane can contribute to the output. This means that the light output from the sides of the device is enhanced by a factor of 360/(4x40) = 2.25 because of the geometric effect.

Two additional points are worth mentioning here. First, this enhancement factor of 2.25 can not be realized if the device is uniformly pumped. The best configuration as far as the geometric enhancement factor is concerned is that all the pump current is confined at the center point of the device circle. The real device, of course, cannot have such a structure. Further calculations show that as long as the pumped region is remote from the periphery of the device, substantial output coupling enhancement can be achieved. In fact, substantial enhancement in output coupling can be achieved if the ratio of the radius between the pumped region and the outer device circle is less than 1/3. Second, the new device structure also benefits from reduced Fresnel loss. Even if the light beam falls within one of the cones in Fig. 5 A, part of the light is still reflected back at the device/air interface, referred to as Fresnel loss. A beam in a cone can be decomposed into many plane waves of different incident angles using the Fourier expansion theory. The plane wave normal to the device/air interface experiences the minimum reflection loss. Beams with a greater incident angle experience a higher reflectivity at the interface, thereby so suffering from a higher reflection loss until the angle reaches the Snell's angle where the beam is totally reflected (100% reflection loss). In the new geometry in Figure 5B, all light is perpendicular to the boundary, so the in-plane reflection loss is minimized although the out-of-plane reflection loss remains the same as in the conventional design. This factor approximately contributes to an additional 40% enhancement of the output coupling from the device peripheral. In summary, combining the first and second designs of the device geometry, the total output efficiency of the device can be described as

Total output = 1 (top cone) + 1 (bottom cone ) + 4(side cones) 2.25 α η ≡ 12 full cones,

where 2.25 is the enhancement factor from the second design, α is the correction factor due to the deviation from the ideal "point source" geometry, and η is the contribution from reduced Fresnel loss. A practical device may have α = 0.8 and η = 1.4, so the total LED output power is equivalent to about 12 full cones, which is 100% brighter than the 6-cone high-brightness LEDs. Therefore, our new design can produce super high brightness LEDs.

Super Bright LED Designs

Several super bright LED structures are given below, although more device structures and variations of the designs given can be conceived based on the same concept.

One common feature shared by all super bright LED designs is that the light generation or current pumped region is constrained to a relatively small area compared to the die size or the light emitting unit. Otherwise, the performance enhancement is less pronounced.

A. Current confinement by lateral oxidation.

Referring to Figs. 6A-6B, it is known that certain material such as AlGaAs with high Al content can be oxidized at a much faster rate than other semiconductors in the same environment. If such fast oxidizing material as AlGaAs is compatible with the LED epitaxial material, one can form the device structure shown in Figs. 6A-6B. A super- bright LED die 508 includes an opaque substrate 10 having a reflective 80 on a surface thereof. Etched mesas 125, 126, 127, 128 are similar to mesas 25, 26, 27, 28 in Figs. 3A- 3C and Fig. 4 except that they are preferably generally circular to take advantage of the geometry as explained with respect to Fig. 5B. Laterally oxidized AlGaAs layers 150, 151 define the light generating areas 160, 161 respectively. Each mesa functions as a light emitting unit in the LED die as described with respect to Design 1 (Figs. 3 A-4). In addition, the laterally oxidized layers 150, 151 (e.g., oxidized AlGaAs) form current blocking layers that funnel all the injected current into an aperture. Since only the active layers in or near the aperture are pumped, all light is generated from light generating areas 160, 161, 162, 163 thereby satisfying the requirement in Design 2 (Fig. 5A).

For red/orange/yellow LEDs, AlGaAs happens to be compatible with the AlInGaP LED material and can be grown together with the LED layers, so one can etch the mesas and put the sample in a furnace with humidity for lateral oxidization. At around 400° C, laterally oxidized AlGaAs layers that are tens of micrometers deep form in about 30 minutes.

B. Current confinement by trench

Referring to Figs. 7A-7B, another possible design of super bright LEDs is shown that works for all LED materials including As-based, P-based, N-based, and organic

LEDs. A super-bright LED die 509 includes an opaque substrate 10 with a reflective layer 80 thereon. A plurality of mesas 125, 126, 127, 128 are etched in the LED epilayer. As shown in Fig. 7 A, a trench 164, 165 is etched at the central region of each cylindrical mesa to bring the bottom of the trench closer to the active layers of the LED. Outside the trench region is an insulating layer 180, 181 to minimize current leakage and spreading. The trench can be as shallow as the thickness of the insulating layer or very deep to almost reach the active layers. Transparent conductive layers 170, 171 are formed on top of insulating layers 180, 181.

C. Current confinement by ion implantation, regrowth or deposition of transparent layers, and material joining Referring to Fig. 8, a super bright LED die 510 is shown that can be implemented by a variety of techniques including ion implantation, layer regrowth or redeposition, and material joining. Again, this design is applicable to LEDs of any colors (including UN and infrared) and made of any materials. LED die 510 includes a reflective layer 80 formed on top of an opaque substrate 10. A plurality of mesas 125, 126, 127, 128 are etched in the LED epilayer. High-resistivity areas 190, 191 are preferably formed by ion implantation, so that all light is generated form light generating areas 160, 161.

Ion implantation offers one convenient technique to modify the conductivity of a material, as demonstrated in semiconductor laser diodes. After implantation of hydrogen, oxygen, He, or other ions into the LED layers with the central region masked, the electric conductivity of the masked region can be orders of magnitude higher than the surrounding implanted region. Hence, the injected current can be concentrated and limited to the unimplanted region to produce the desired current distribution. For the epitaxial regrowth method, the mesas are first patterned and then a nonconductive transparent layer is grown on the patterned mesas. Alternatively, the initial epitaxial surface can be first turned into an insulator except for the light emitting regions, after which a transparent conductive layer is either deposited or joined to this surface to form the final structure.

D. Diffusion or implantation of dopant impurities for current confinement

Referring to Figs. 9A-9B, the device structure for a super-bright LED die 511 shows a different design where the LED layers are normally nonconductive or poorly conductive. In most designs, the whole epitaxial layers have a doping profile that makes the device conductive under forward bias; hence, current confinement is achieved by introducing current blocking layers to the structure. In the present invention, a highly conductive region is created by implanting or diffusing dopant impurities into the central region of the device. A reflective layer 80 is formed on top of an opaque substrate 10. A plurality of mesas 125, 126, 127, 128 are etched in the LED epilayer. High conductivity regions 200, 201 are preferably formed in mesas 125, 126 by ion implantation or diffusion, thus forming light generating areas 160, 161.

E. Extending the non-conductive mesa region Referring to Figs. 10A-10B, another approach of making super bright LEDs is shown in which a central conductive light emitting region is defined first, while a nearly matched refractive index transparent region is formed afterwards. A super-bright LED die 512 includes a reflective layer 80 over an opaque substrate 10. A plurality of mesas 125, 126, 127, 128 are etched in the LED epilayer. Nearly index-matched non-conductive, transparent layers 210, 211 are formed in mesas 125, 126. For GaAs-based or InP-based semiconductors that have a very high refractive index, it is difficult to find other materials with a comparable index other than its own class of material. In this situation, regrowth of the same type of As- or P-based transparent material may be needed. Since the only requirements for the surrounding material is transparency and index matching, the material can be even polycrystalline or amorphous. For example, high Al-content polycrystalline AlGaAs may be deposited around the AlInGaP light emitting area for output coupling enhancement.

If the material is high-bandgap semiconductors such as InGaN, GaN, and AlGaN or π-VI compounds, the refractive index is much lower and the surrounding material can be a dielectric material (e.g., SiN, TiO, etc.) or polymer. Polymer is particularly attractive since it can be spin-coated to a large thickness and is very easy to pattern.

The device structure in Figs. 10A-10B is easiest to implement for organic LEDs, since many different organic polymer materials are transparent, nonconductive, easy to deposit, and index matched.

We have discussed two key geometric designs to enhance the brightness of LEDs operating at all wavelengths and colors. The first design solves fundamentally the substrate absorption problem and the requirement for a thick top transparent layer for high brightness LEDs. As a result, LEDs with a relatively thin epitaxial layer grown on an opaque substrate can deliver the same performance as the state-of-the-art high brightness

LEDs on transparent substrates. The second design almost doubles the output coupling efficiency. Because of the larger light capture angle and reduced Fresnel loss, super bright LEDs that are 100% more efficient than today's high brightness LEDs can be obtained. Above all, all device designs in this invention requires only one substrate and one epitaxial growth, thus yielding not only high performance but also low cost LEDs. Accordingly, it is to be understood that the embodiments of the invention herein described are merely illustrative of the application of the principles of the invention. Reference herein to details of the illustrated embodiments are not intended to limit the scope of the claims, which themselves recite those features regarded as essential to the invention.

Claims

What is claimed is:

1. A light emitting diode, comprising:

a) an opaque substrate;

b) a reflective layer on said substrate; and

c) a plurality of LED epilayer mesas on said reflective layer.

2. A light emitting diode according to claim 1, further comprising restricting means for restricting a pumped region in each mesa.

3. A light emitting diode according to claim 2, wherein said restricting means includes laterally oxidized material defining an aperture such that all injected current is tunneled into said aperture.

4. A light emitting diode according to claim 2, wherein said restricting means includes:

a trench in a central region of each mesa; and

an insulating layer surrounding said central region on each mesa.

5. A light emitting diode according to claim 2, wherein said restricting means includes a high-resistivity layer defining an aperture such that all injected current is tunneled into said aperture.

6. A light emitting diode according to claim 5, wherein said high-resistivity layer is formed by one of ion implantation, regrowth, deposition, and material joining.

7. A light emitting diode according to claim 2, wherein said restricting means includes forming said LED epilayer mesas of nonconductive or poorly conductive materials and forming a highly conductive region centrally located in a top of each mesa.

8. A light emitting diode according to claim 2, wherein said restricting means includes forming a centrally located conductive light emitting region in each mesa and forming a non-conductive transparent region surrounding said light emitting region, wherein said transparent region is closely index-matched to said light emitting region.

9. A light emitting diode according to claim 1, wherein each of said plurality of mesas are substantially circular.

10. A light emitting diode according to claim 9, further comprising restricting means for restricting a pumped region in each mesa so that a ratio of a radius of said pumped region to a radius of said mesa less than one-third.

11. A light emitting diode according to claim 10, wherein said restricting means includes laterally oxidized material defining an aperture such that all injected current is tunneled into said aperture.

12. A light emitting diode according to claim 10, wherein said restricting means includes:

a trench in a central region of each mesa; and

an insulating layer surrounding said central region on each mesa.

13. A light emitting diode according to claim 10, wherein said restricting means includes a high-resistivity layer defining an aperture such that all injected current is tunneled into said aperture.

14. A light emitting diode according to claim 13, wherein said high-resistivity layer is formed by one of ion implantation, regrowth, deposition, and material joining.

15. A light emitting diode according to claim 10, wherein said restricting means includes forming said LED epilayer mesas of nonconductive or poorly conductive materials and forming a highly conductive region centrally located in a top of each mesa.

16. A light emitting diode according to claim 10, wherein said restricting means includes forming a centrally located conductive light emitting region in each mesa and forming a non-conductive transparent region surrounding said light emitting region, wherein said transparent region is closely index-matched to said light emitting region

17. A method for making a light emitting diode, comprising the steps of:

a) growing a light emitting diode epilayer on an opaque layer such that a reflecting surface is above said opaque layer; and

b) etching a plurality of mesas in said light emitting diode epilayer, wherein each of said mesas has a mesa width to mesa height ratio of less than or equal to three.

18. A method according to claim 17, wherein said mesa height ranges between 6 and 15 microns and said mesa width correspondingly ranges between 20 and 15 microns.

19. A method for making a light emitting diode, comprising the steps of:

growing a reflecting layer on an opaque layer;

growing a light emitting diode epilayer on said reflecting layer; and

etching a plurality of mesas in said light emitting diode epilayer, wherein each of said mesas has a mesa width to mesa height ratio of less than or equal to three.

20. A method according to claim 19, wherein said mesa height ranges between 6 and 15 microns and said mesa width correspondingly ranges between 20 and 15 microns.

21. A method according to claim 19, wherein each of said mesas is substantially circular.

22. A method according to claim 19, further comprising the step of restricting a pumped region in each of said mesas.

23. A method according to claim 22, wherein said step of constricting includes laterally oxidizing material in each of said mesas to define an aperture such that all injected current is funneled into said aperture.

24. A method according to claim 22, wherein said step of constricting includes:

forming a trench in a central region of each of said mesas; forming an insulating layer on each of said mesas outside said trench; and

forming a transparent conductive layer on each of said mesas covering said trench and said insulating layer.

25. A method according to claim 22, wherein said step of constricting includes forming a high-resistivity layer in each of said mesas to define an aperture such that all injected current is funneled into said aperture.

26. A method according to claim 22, wherein said step of constricting includes:

forming said light emitting diode epilayer of a nonconductive or poorly conductive epilayer; and

forming a highly conductive region centrally located in each of said mesas.

27. A method according to claim 22, wherein said step of constricting includes:

forming a centrally located conductive light emitting region in each of said mesas; and

forming a non-conductive transparent region surrounding said light emitting region, wherein said transparent region is closely index- matched to said light emitting region.