WO1999031738A2

WO1999031738A2 - Aiii-nitride channeled led

Info

Publication number: WO1999031738A2
Application number: PCT/IB1998/001933
Authority: WO
Inventors: Nikhil Taskar; Piotr M. Mensz
Original assignee: Koninklijke Philips Electronics N.V.; Philips Ab
Priority date: 1997-12-16
Filing date: 1998-12-03
Publication date: 1999-06-24
Also published as: JP2001511956A; EP0966766A2; WO1999031738A3

Abstract

A channeled semiconductor LED is disclosed that is comprised of a substrate, upper and lower contact layers, the lower contact layer adjacent the substrate, and a light emitting region interposed between the upper and lower contact layers. One or more channels extend through the upper contact layer and the light emitting layer, the one or more channels defining LED sub-structures, and improving the luminous efficiency of the LED.

Description

A^πι-Nitride channeled LED

BACKGROUND OF THE INVENTION

This invention relates to semiconductor light emitting diodes ("LEDs") and, in particular, to the light output of such devices.

A '-nitride LEDs are known. For example, such LEDs having GalnN or GaN active regions have been fabricated on transparent substrates, such as sapphire or 6H-SiC. A "-nitride LEDs having a high luminous efficiency are desirable. To be commercially viable, such LEDs must have luminous efficiency that compete with conventional incandescent or halogen light sources and thus must be on the order of 30 Lumens Watt or higher. Efficiencies on the order of 30 and even 40 Lumens/Watt have been achieved for yellow, amber and red LEDs using an AlInGaP material system, by designing the LED to enhance the extraction of light generated inside the device.

Such enhanced light extraction for yellow, amber and red LEDs using an AlInGaP material system has been accomplished by designing the LED in a particular manner and, in particular, by using thick GaP contact layers.

In general, the refractive indices of the layers in these known LEDs are larger than the epoxy packaging. Thus, the internally generated light incident on a surface of the LED at angles smaller than the critical angle ("P_c") for total internal reflection ("TIR") may be extracted from the LED chip. The value of the critical angle, of course, is determined by the refractive indices of the contact layers of the device and the epoxy surrounding the contact layers.

For the known (yellow, amber and red) LEDs, extraction of sufficient external light from the AlInGaP material system is achieved by using thick GaP contact layers. The GaP layers have an index of refraction close to the other (cladding and active) layers, and are also transparent at the emission wavelength. These thick contact layers act as "window" layers by permitting light within a critical angle cone to pass through the sides of the chip.

An example of the such a known LED with an AlInGaP material system and thick GaP contact layers is shown in Fig. 1. The AlInGaP material system is comprised of a light generating region 10, which is comprised of a InGaP or AlInGaP active region 12, surrounded by an upper p-doped AllnP cladding layer 14 and a lower n-doped AllnP cladding layer 16. (The AlInGaP active layer 12 and the AllnP 14, 16 cladding layers are each on the order of one micron in thickness.) The light generating region 10 is surrounded by an upper p- doped GaP contact layer 18 and a lower n-doped GaP contact layer 20. The thicknesses of the contact layers 18, 20 are related to the extraction of light from the device, as described in more detail below. (The x and y dimensions of the figures are generally not to the same scale.) As noted above, the LED is typically surrounded by an epoxy 24.

The indices of refraction of the LED layers 12, 14, 16, 18, 20 are roughly equivalent and greater than the index of refraction of the surrounding epoxy 24. It follows from Snell's law of refraction that there are six cones 26a-e from which light emitted at the light generating region 10 pass to the outside of the LED, into the epoxy, without TLR.

For example, light ray 28 from light generating region 10 is emitted at an angle less than P_c with respect to the upper planar interface between the top surface of upper GaP contact layer 18 and epoxy 24. Thus, ray 28 is refracted away from the normal, but not totally internally reflected, and passes into the epoxy 24. The epoxy 24 encasing the LED is generally made sufficiently thick and is given a rounded surface, such that ray 28 incident on the interface between the epoxy 24 and air is approximately normal to the interface, and passes into the air with little or no further refraction. By identical analysis, all rays within upper shaded critical angle cone 26a will pass into the air. A similar analysis may be performed for a ray 30 emitted at an angle less than

P_c with respect to the planar interface between a side surface of upper GaP contact layer 18 and epoxy 24. Thus, light emitted within shaded critical angle cones 26c and 26f will pass into the air. By symmetry, light emitted within lower side critical angle cones 26d and 26e is also extracted. Thus, light emitted within the six cones 26a-e and extracted from the LED contributes to the potential brightness of the LED. For light that is emitted in the downward direction, for example, through cones 26b, 26d, 26e, a reflective layer (not shown) is typically used to re-direct it upwards (for example, in the direction of rays 28, 30).

Figs. 1 and 2 demonstrate how a thick upper GaP contact layer 18 increases the luminous efficiency of the LED. In Fig. 1, the width d of upper GaP contact layer 18 is sufficiently wide so that side cones 26c, 26d, include all light emitted within critical angle P_c. Also, as noted below and seen in Fig. 1 , lower GaP contact layer 20 is usually sufficiently thick to capture lower side critical angle cones 26d, 26e. In other words, all of the light emitted to the side that can be extracted, i.e., within P_c, is incident on a side interface, and is extracted.

Fig. 2 demonstrates how the amount of light extracted is reduced for the same chip width L if the thickness d' of the upper contact layer 18' is reduced. As shown in Fig. 2, the side cones 26c',26f are circumscribed by the thickness d' (or, equivalently, the angle ^Λ), not by the critical angle P_c. For light emitted at an angle between ^Λ and P_c, there is total internal reflection: Because of the reduction in thickness d', that light is incident on top surface of upper GaP contact layer 18'. Because this light does not fall within critical angle cone 26a' for the interface with the epoxy or substrate, it is totally internally reflected. (The region where this light is incident on the top surface of upper GaP contact layer 18' is denoted "TIR.") Thus, selection of the dimension d for the thickness of the upper contact layer of a particular width chip die L allows all sideways emitted light, up to the critical angle to be extracted, as in Fig. 1. (It can be seen that if the width of lower contact layer 20' of Fig. 2 were also reduced to less than d, then a portion of the lower side cones 26d', 26e' would also be totally internally reflected. Thus, the thickness considerations apply to both contact layers. However, as noted below, the width of the lower contact layer is typically large for the known GaP based system and thus is usually not a limiting factor.)

For the same indices of refraction of the layers and epoxy, the minimum contact layer thickness necessary to extract all the light within the critical angle cone increases linearly with the width L of the chip die.

Typical dimensions for the known (blue, green, yellow and red) high efficiency LEDs shown in Fig. 1 include, for example, a width L of the chip die on the order of 300 microns. The refractive index ("n") of the LED device having the material layers described above with respect to Fig. 1 (i.e., for a GaP based system) is typically on the order of 3.5. (The indices of refraction for each particular layer is generally "matched," but may differ slightly from other layers, because the composition of the layer will also be chosen, for example, to optimize light generation.) The thick GaP lower contact layer 20 will typically have a thickness on the order of 150 microns.

If the index of refraction of the epoxy is approximately that of air, i.e., n=1.0, then the LED described has side cones 26c-f with critical angle P_c of approximately 17/. For an LED with die width L of 300 microns, the thicknesses d is thus approximately 45 microns in order to accommodate the entire side critical angle cones 26c, 26f. (As noted above, since the thickness of lower contact layer 20 is on the order of 150 microns, there is no question lower side critical angle cones 26d, 26e will be accommodate the entire cone.) For epoxy with n=1.5, the side critical angle cones 26c, 26f would have P_c of approximately 25/, and the thickness d would be approximately 71 microns.

On the other hand, if the die width L is 500 microns, then the thickness d would have to be approximately 75 microns for epoxy having n=l .0. If the epoxy has n=l .5, then the thickness d would have to be approximately 118 microns. (Again, a thick lower contact layer 20 on the order of 150 microns would accommodate the entire lower side cones for these cases.)

Thus, a disadvantage of the design of the known high efficiency GaP based LEDs is that the thickness of the contact layers must be relatively large, on the order of 100 microns, for a typically dimensioned LED chip die. Growing such thick window contact layers requires Hydride Vapor Phase Epitaxy ("HVPE"), which gives growth rates in the range of 10 to 100 microns per hour. Thus, fabrication of such thick contact window layers is relatively slow.

Moreover, there are impediments to such a thick contact layers system when a different material system is considered. Consider an analogous hypothetical GaN based system, for example. Such a system might have a AlInGaN active layer, AlGaN cladding layers, and GaN contact layers. The substrate for the system might be transparent SiC or Sapphire. A nominal value of the index of refraction for the GaN based layers is approximately 2.5. For such a GaN based system, if the index of refraction of the surrounding epoxy is 1.0, then the LED will have side cones (analogous to those shown as 26c-f for the GaP system of Fig. 1) with critical angle of approximately 24/. For a chip width of 300 microns, the thicknesses of the upper and lower contact layers would have to be approximately 65 microns (or more) in order to entirely accommodate the light emitted within the side critical angle cones. If the refractive index of the epoxy is 1.5, then the thicknesses would have to be approximately 110 microns. For a chip width of 500 microns, then the thickness would have to be 110 and 180 microns for epoxy having indices of refraction of 1.0 and 1.5, respectively. Thus, for a GaN based system, both upper and lower GaN contact layers would also have to be on the order of hundreds of microns. However, the growth of such thick p- doped GaN layers has never been achieved using HVPE. Moreover, growth of n-doped GaN layers with thicknesses that exceed 100 microns has never been achieved on a Sapphire substrate (one of the common substrates for such LEDs) using HVPE. Thus, the thick contact layer design described above for the GaP based system has yet to be implemented for a AUnGaN-based LED on a Sapphire substrate. Moreover, even if such thick layers could be fabricated for a GaN based system, there would be difficulties regarding the electrical contacts to the device. Unlike the GaP device, where the n-doped GaP contact layer provides a direct electrical contact, the Sapphire or SiC substrate of the GaN based system is not electrically conductive. Thus, a side metal contact would have to be used. For such thick contact layers, fabrication would be difficult and the operational efficiency would be poor.

SUMMARY OF THE INVENTION

It is an objective of this invention to overcome the disadvantages and limitations of material systems presented by a high efficiency output LED that uses thick layered contact layers.

This present invention is a ultra-violet ("UV") or visible LED structure having relatively thin transparent contact layers. The LED structure is "channeled": Within the LED that is cast on an LED chip die, there are one or more channels that extend into the LED. The channels create a multiplicity of LED sub-structures within the dimensions of the chip die. The widths of the LED sub-structures are, accordingly, less than the width of the chip die. As noted above, the requisite thickness of the contact layer for capturing the entire side critical angle cone scales linearly with the width of LED. By creating such LED islands, their width is effectively reduced. Thus, the thicknesses of the contact layer(s) necessary to capture the entire critical angle cone of the sideways emission is reduced proportionally. Feasible widths of the islands may be, for example, as low as 10 microns. For such islands, the thicknesses of the contact layers would be on the order of 4 to 5 microns to capture the entire side critical angle cones. Side electrical contacts to such LED islands could be readily fabricated and the reduced widths of the contact layers would result in high operational efficiency.

BRIEF DESCRIPTION OF THE DRAWINGS

The features of the present invention will be better understood and become readily apparent by referring to the following detailed description, in conjunction with the accompanying drawings, in which:

Fig. 1 is a cross-sectional view of a prior art, thick contact layer LED structure;

Fig. 2 is a cross-sectional view of the device of Fig. 1 with a reduction of thickness in the contact layers; Fig. 3 is a cross-sectional view of an embodiment of a channeled LED structure in accordance with the present invention, with an enlarged view of a number of the LED substructures and channels;

Fig. 4 is a cross-sectional view of one of the LED substructures shown in Fig. 3 and the adjacent channels;

Fig. 4a is a cross-sectional view of the LED sub-structure of Fig. 4, taken across line 4a-4a in Fig. 4;

Fig. 5 is a cross-sectional view of another embodiment of a channeled LED structure in accordance with the present invention, showing a number of the LED sub- structures and channels;

Fig. 5a is a cross-sectional view of one of the LED sub-structures of Fig. 5, taken across line 5a-5a in Fig. 5;

Fig. 6 is a partial top view of a channeled LED structure in accordance with the present invention; Fig. 7 is an enlarged view of a portion of the LED sub-structure and adjacent channels shown in Fig. 6 within the dashed lines;

Fig. 8 is an alternative configuration of the channeled LED structure of Figs. 6 and 7; and

Fig. 9 is an alternative configuration of the channeled LED structure of Figs. 6 and 7.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

As shown in Fig. 3, LED chip 98 on substrate 100 has width L (determined by the chip die) and a multiplicity of sub-structures 102. The enlarged portion of Fig. 3 shows the structure of each sub-structure in greater detail. For the purposes of this description, it is presumed that all sub-structures have the same characteristics.

The sub-structures 102 may each be comprised of a AlInGaN material system and GaN contact layers, analogous to that described above with respect to Fig. 1. The AlInGaN material system is comprised of a light generating region 104, which is comprised of an InGaN active layer, surrounded by an upper p-doped AlGaN cladding layer and a lower n- doped AlGaN or GaN cladding layer. The active layer is typically on the order of 30 to 1000 angstroms and the cladding layers are typically on the order of 1000 angstroms. The light generating region 104 for each sub-structure 102 is surrounded by an upper p-doped GaN contact layer 106 and a lower n-doped GaN contact layer 108. The substrate 100 supporting the sub-structures 102 may be Sapphire.

The thickness of upper GaN contact layer 106 is shown to be t_p and the thickness of lower GaN contact layer 108 is t_n. The width of each sub-structure 102 is w and each sub-structure 102 is separated by a channel of width s.

Referring to Fig. 4, the critical angle cones are shown shaded for one of the substructures 102. As noted above, with respect to the description of the prior art device of Fig. 1, the thicknesses of the contact layers necessary to capture the entire side critical angle cones is linearly proportional to the width of the LED chip. By a similar analysis, the thicknesses t_p and t„ of upper and lower GaN contact layers 106, 108 required to capture the side critical angle cones is proportional to the width w of LED sub-structure 102. However, as described in more detail below, because the width w of the sub-structure 102 is considerably less than the width of the LED chip 98 (or, equivalently, the width of the LED chip die L), the thickness of the contact layers 106, 108 will be significantly less than the thick layers of the device of Fig. 1.

Referring back to the enlarged portion of Fig. 3, metal mirror layers 110 in the channels between the LED sub-structures 102 reflect light emitted in the downward direction from the side critical angle cones upward toward the epoxy-air interface. As described in more detail below, it is preferable for the width s of the channels to allow a significant amount of the downward extracted light to be reflected without being blocked by the side of the adjacent LED sub-structure. However, as also described below, the relative widths w,s of the sub-structures 102 and channels should be such that the operating current density in the active region remains in the linear regime of the light output versus current density characteristic at the rated operating current value. For example, particular dimensions of the LED sub- structures and channels for the material system given above, and the resulting light output of the LED, may be as follows:

Example:

Referring again to Fig. 4, the minimum thicknesses t_p and t_n of the upper and lower GaN contact layers 106, 108 required to capture the side critical angle cones is determined from Snell's law and geometric considerations:

%o_w_ tan arcsin — ))

KLEDGaN where ri_o is the index of refraction of the LED package (i.e., epoxy), nLED_GaN is the index of refraction of the GaN contact layers 106, 108 and the thicknesses of the contact layers 106 are much greater than the light emitting region 104.

For w=10 microns,

the above equation gives thickness of the upper GaN layer 106 t_p of approximately 2.2 microns. For no=1.5, t_p is approximately 4.3 microns. Thus, the thicknesses t_p, t_n of upper and lower GaN contact layers 106, 108 will be on the order of 2.2 to 4.3 microns. These reduced thicknesses are readily achieved with, for example, Metal Organic Vapor Phase Epitaxy ("MOVPE"). As is known in the art, MOVPE gives more uniform layer thickness (due to slower growth rates) and better control of doping levels than HNPE. Also HNPE cannot be used to grow an InGaΝ active region; therefore, use of MOVPE also avoids interruption of the growth process. Thus, because the cladding and active layers are typically fabricated using MOVPE, the entire LED structure may be fabricated using MOVPE.

The channels between sub-structures 102 may be fabricated by etching trenches using known methods, such as Reactive Ion Etching. The above described layers, once deposited on the substrate 100, would be masked so as to create the sub-structures 102 after the etching process.

As an alternative to Reactive Ion Etching, the sub-structures 102 may be fabricated using masked epitaxial growth. After first growing an n-type GaΝ layer (of thickness tς), the portions of the wafer that will be the channels are covered with a mask layer (such as SiO₂). Then, the lower (n-doped) GaΝ contact layer 108 is grown using either HVPE or MOVPE, followed by MONPE growth of the light generating region 104, followed by MOVPE growth of the upper (p-doped) GaΝ contact layer 106. As is known in the art, under appropriate conditions such growth will occur in the areas not covered by the SiO₂. Selection of the width w of the sub-structures 102 and width s of the adjacent channels is also such that L-I characteristic of the junction is still linear for the current density of the device. For an unchanneled LED chip of width L (also the width of the chip die) equal to 300 microns, the current I is on the order of 20mA. L-I curves show approximately linear behavior between 20 and 40mA. The current density of the channeled LED structure is:

_/_ (w+s)

L² - w ^' where L is also the width of the LED chip (or, equivalently, the chip die, as shown in Fig. 3).

For w=10 microns as above and also taking s=10 microns, (w+s)

2, w

and the junction operates at a current density two times that of the nominal current density of a regular LED. Thus, the doubling of the current density compensates for the reduction (by half) of the junction area of the channeled LED and, because the linearity of L-I for these values, the internal power output generated by the channeled LED is essentially identical to that from a regular LED.

Moreover, as previously noted, the width s of the channels determines the overall collection of light from the side critical angle cones. For s=w=10 microns, the geometry is such that a significant amount of light will be reflected and extracted for upper and lower contact layers 108, 106 on the order of 2.2 to 4.3 microns. The metal layer 110 used to reflect light in the channel upward may be made aluminum and also serves as the electrical contact for the lower GaN contact layer 108. As shown in Figs. 3 and 4, a thin layer of n- doped GaN of width t_c is interposed between the Sapphire substrate 100 and metal layer 110 at the bottom of each channel. This thin layer of n-doped GaN underlaying each channel is contiguous with the lower GaN contact layer 108 of each island 102.

Because metal layer 110 is a side electrical contact, the thickness t_<. determines the current spreading in the lower n-doped GaN contact layer 108 and contributes to the series resistance. Thus, in order to achieve the appropriate current densities, referred to above, metal layer 110 is relatively thick. Typical values of t_c ³ 0.5 microns and the corresponding thickness of metal layer 110 ³ 0.1 microns.

The upper GaN contact layer 106 in this example is thin (on the order of 2 to 4 microns) with respect to the upper contact layer 18 of the thick contact layer devices described with respect to Fig. 1. However, a thickness on the order of 2 to 4 microns for the upper contact layer 106 is itself thick relative to "regular" LEDs (i.e., typical LEDs that do not use thick contact layers to enhance the power output, as the devices described with respect to Fig. 1). Because of this, the sheet conductivity of the upper contact layer 106 is on the order often times larger than a regular LED structure. Thus, an enhanced current spreading in this layers permits the use of a thin electrical contact layer, without complete coverage of the top surface of the upper GaN contact layer 106. Thus, upper electrical contacts 112 of Figs. 3 and 4 may be, for example, thin partially transparent strips of Ni-Au alloy that cover approximately 50% of the surface and thickness on the order of 100 to 300 angstroms. Because they do not cover the entire upper GaN contact layer 106, the optical absorption losses in the emission through the top critical angle cone of the sub-structure 102 is reduced.

A structure analogous to that shown in Fig. 3 with a lower p-doped GaN contact layer and an upper n-doped GaN contact layer could also be fabricated as a channeled LED. This structure would allow an even further reduction in the metal coverage of the upper n- doped GaN contact layer compared to an upper p-doped GaN contact layer, thus enhancing extraction of light through the upward cone. However, the lower conductivity of p-doped GaN compared to n-doped GaN (on the order of a factor of 100) would require significantly smaller values of the width w of the sub-structure in order to reduce the lateral voltage drop and ensure uniform current injection in the junction. The width w would typically have to be reduced by a factor of 10, giving width w on the order of one micron. The dimensions of t_p and t_n would also be reduced, and would be on the order of one micron. Also, accommodation would have to be made so that the structure had the desirable power conversion efficiency despite the voltage drop due to the lateral contact in the lower p-doped GaN contact. (For example, where the resistivity of the p-type GaN is 100 times larger than the n-type GaN, the disparity may be compensated by increasing the ratio tς to w by a factor of 100. Thus, t_c ³ 5 microns.

Fig. 5 shows an alternative embodiment of a channeled LED in accordance with the present invention. Fig. 5 has structure analogous to Fig. 3, with analogous structure marked with a corresponding "200" series of reference numbers.

As seen in Fig. 5, lower n-doped GaN contact layer 208 is a relatively thick layer, which extends well below metal contact layers 210. Lower contact layer 208 provides one continuous flat surface on which metal contact layers 210 and the light emitting layers 204 of the LED sub-structures 202 lie. An upper p-doped GaN contact layer 206 lies on top of each light emitting layer 204. (Upper metal contact layers 212 are shown for one sub-structure 202, but are omitted from the others for clarity.)

Thus, it is seen that the structure of Fig. 5 includes sub-structures, similar to Fig. 3, but with a thick lower contact layer 208. Equivalently, the parameter t_n of Fig. 3 has been reduced to zero, while the parameter t_c has been expanded. The thick lower contact layer 208 could be grown using HVPE, with the light emitting layers 204 and upper contact layers 206 grown using MOVPE. The sub-structures 202 would be created by etching channels to the lower contact layer 208. Fabrication of the sub-structures 202 and channels may be done by analogous methods as described above with respect to Fig. 4. Of course, the various methods described would have to be adjusted to account for an increased value of t_c and a value of t_n equal to approximately zero. Exemplary dimensions for the structure of Fig. 5 would be L (width of the LED chip or, equivalently, width of the chip die) equal to 300 microns. The widths of the substructures 202 and channels, w and s, respectively, could each be on the order of 10 microns. For typical indices of refraction for the GaN layers and epoxy, the value t_p (thickness of upper GaN contact layers 206) is on the order of 2.2 to 4.3 microns and the value of t_c (thickness of lower GaN contact layer 208) is on the order of 65 to 110 microns. (As for the device of Fig. 3, the thickness of the cladding layers of light emitting layer 204 is known in the art and is on the order of 1000 angstroms for an AlInGaN material structure. The thickness of active layer of light emitting layer 204 is likewise known in the art for an AlInGaN material structure to be on the order of 30 to 1000 angstroms.) The advantages of the structure of Fig. 5 are clear from the light cones (shaded) shown for one of the sub-structures 202. The amount of light extracted from the upper side cones and the top cone is found in the same manner as described above with respect to the structure of Fig. 3. However, for the structure of Fig. 5, the light emitted into the lower side cones is not reflected by metal layer 210 and will not be blocked by an adjacent sub-structure. Provided t_c is large enough in relation to the width of the LED (or chip die), the light from the lower side cones will be emitted completely (and thus may be completely collected) from the side of the LED.

(The width of t_c is given by the equation

t_c = — _ tanfarcsinf — — — ))

2 riLEDGaN where n_o is the index of refraction of the material surrounding the LED, such as epoxy, and n _{ED GaN} is the index of refraction of the lower contact layer 208. Thus, tc is linearly proportional to L.)

Moreover, as shown in Fig. 5a, a cross-sectional view of one of the substructures 202 parallel to the channel, the increased thickness of the lower contact layer 208 means that light will be extracted from lower side critical angle cones at all four sides of the LED chip. Thus, the luminous efficiency of the channeled LED structure is significantly increased. (Fig. 4a gives a view analogous to Fig. 5a, but for the structure of Fig. 3, and shows that each island 102 has only two lower side critical angle cones: Because of the small thickness of lower contact layer 108, lower light emitted to the side is incident on substrate 100, not the side of the chip die. It thus lies outside the critical angle cone for the contact/substrate interface and is totally internally reflected.)

In the prior embodiments, and in particular, in Figs. 3-5a, a cross-section of the channeled LED of the present invention is shown. Referring back to Figs. 3 and 5, this is a cross-sectional view in the x-y plane.

In Fig. 6 a top view of a channeled LED structure, in the x-z plane, is shown. This top view is applicable to both of the embodiments shown in cross-section in Fig. 3 and Fig. 5, and reference numbers in the 300s are used for analogous structure. (For clarity, Fig. 6 omits the metal contact layers 110, 112, 210, 212 of Figs. 3 and 5.) As shown in Fig. 6, the "sub-structures" 302 are not separate islands, but are interdigitated among channels comprising contiguous p-type contact layer 306 on top of a contiguous n-doped contact layer 308. The interdigitated sub-structures 302 and channels allow the contiguous p-type contact layer 306 to have a common metal contact. If the "sub-structures" 302 were separate "islands", then there would have to be separate metal contacts for each p-type contact layer. (The n-type contact layer 308 is already contiguous in the x-y plane, see Figs. 3 and 5, so only one metal contact layer would be required even in the case where the p contact layer forms separate islands.)

Figs. 7-9 demonstrate how the interdigitated channel structure of Fig. 6 may be manipulated to extract even more light from the side cones for the channeled LED of the present invention. In the cross-sectional views of Figs. 3 and 5, light emitted into the side cones from the light emitting region were only shown in two dimensions (x-y plane). However, when considered from the top view, such as that shown in Fig. 6, the side cones will also be "spread" out in the x-z plane.

In Fig. 7, a closer view of a sub-structure 302 and adjacent channels of Fig. 6 is shown. Light emitted into the x-z plane from side cones of the light emitting layer is shown outlined through the p-doped contact layer 302. Geometric considerations show that each side cone will have light extracted over an angle of 2P_C in the x-z plane.

Figs. 8 and 9 show that by contouring the interdigitated sub-structures 302 and channels of Fig. 6 into a series of polygonal shapes, the extent of the side cones in the x-z plane, and thus the light extracted from the LED, is increased. In Fig. 8, the interdigitated substructures 302 and channels are given a series of hexagonal shapes, with each side of a channel that borders a hexagon formed in the sub-structure 302 defining an angle 2P_C with respect to the center of the hexagon. Geometric considerations show that, for a point source at the center of the hexagon, light emitted from each side cone will be extracted from the LED over an angle of 4P_C in the x-z plane. Similarly, in Fig. 9, the interdigitated sub-structures 302 and channels are given a series of octagonal shapes, with each side of a channel that borders an octagon formed in the sub-structure 302 defining an angle 2P_C with respect to the center of the octagon. Geometric considerations show that, for a point source at the center of the octagon, light emitted from each side cone will be extracted from the LED over an angle of 6P_C in the x-z plane. Light will be emitted over the entire extent of the light emitting region for each of Figs. 7-9 and not just a point source. Thus, for Figs. 8 and 9 above, light will be emitted not just from the center of the sub-structures, but along the entire length of the sub-structure 302 (i.e., in the x direction). Nonetheless, more light will be extracted than in the case where the sub-structures 302 and channels are straight, shown in Fig. 7. For light emitted from a point outside of the center of the polygon, such as points a and b exactly between polygons in Figs. 8 and 9, for example, less light is internally "trapped" by total internal reflection. The angles created by the polygons at the interfaces of the sub-structures 302 and channels ensures that light that is initially internally reflected within a sub-structure 302 will eventually be incident on a surface within the critical angle, and thus extracted from the LED. It will be understood that various modifications can be made to the various embodiments of the present invention herein disclosed without departing from its spirit and scope. For example, various geometrical configurations of the channeled LED structure and various compositions of the materials used are contemplated. All such channeled LED structures would fall within the scope of the present invention. Therefore the above description should not be construed as limiting the invention but merely as presenting preferred embodiments of the invention. Those skilled in the art will envision other modifications within the spirit and scope of the present invention as defined by the claims presented below.

Claims

CLAIMS:

1. A semiconductor LED comprised of a) upper and lower contact layers, the lower contact layer adjacent the substrate, and b) a light emitting region interposed between the upper and lower contact layers, wherein one or more channels extend through the upper contact layer and the light emitting layer, the one or more channels defining LED substructures.

2. A semiconductor LED as in Claim 1, wherein the channels extend into the lower contact layer.

3. A semiconductor LED as in Claim 2, wherein the semiconductor LED is rectangular and there are a multiplicity of channels, each channel extending parallel to the other channels.

4. A semiconductor LED as in Claim 3, wherein the surfaces of the channels are covered with a reflective metallic layer.

5. A semiconductor LED as in Claim 3, wherein the light emitted from the light emitting region into the upper contact layer in the direction of a channel within a side critical angle cone is completely incident on the side surface between the upper contact layer and the channel.

6. A semiconductor LED as in Claim 3, wherein the light emitted from the light emitting region into the lower contact layer in the direction of a channel within a side critical angle cone is completely incident on the side surface between the lower contact layer and the channel.

7. A semiconductor LED as in Claim 2, wherein the upper and lower contact layers have approximately the same thickness.

8. A semiconductor LED as in Claim 1, wherein the channels extend through the upper contact layer and light emitting region to the upper surface of the lower contact layer.

9. A semiconductor LED as in Claim 8, wherein the semiconductor LED is rectangular and there are a multiplicity of channels, each channel extending parallel to the other channels.

10. A semiconductor LED as in Claim 8, wherein the light emitted from the light emitting region into the upper contact layer in the direction of a channel within a side critical angle cone is completely incident on the side surface between the upper contact layer and the channel.

11. A semiconductor LED as in Claim 8, wherein the light emitted from the light emitting region into the lower contact layer within a side critical angle cone is completely incident on the side surface of the LED.

12. A semiconductor LED as in Claim 8, wherein the lower contact layer is substantially thicker than the upper contact layer.