Solid-state light source
A solid-state light source contains light emitting diodes. The elements are an inorganic light emitting diode chip and at least one wavelength conversion chip or the elements are a plurality of light emitting diode chips and one or more optional wavelength conversion chips. Both substantially parallel and orthogonal mounting configurations are used for the light emitting diodes. Orthogonal mounting or rib emitter is a preferred embodiment. The wavelength conversion chip may include an electrical interconnection means. The light emitting diode chip may include at least one GaN-based semiconductor layer that is at least ten microns thick and that is fabricated by hydride vapor phase epitaxy. Methods, equipment, and articles can be used to fabricate the solid-state light sources, freestanding foil based light emitting diodes, and wavelength conversion materials.
This application is a continuation-in-part of prior U.S. patent application Ser. No. 12/315,482, filed on Dec. 3, 2008 and claims the benefit of U.S. Provisional Patent Application Ser. No. 61/005,258, which was filed on Dec. 3, 2007, and of U.S. Provisional Patent Application Ser. No. 61/196,439, which was filed on Oct. 17, 2008, all three of which are herein incorporated by reference.
TECHNICAL FIELDThe present invention relates to solid-state light sources. More specifically, light emitting elements in a stack including at least one inorganic light emitting diode (LED) chip and optionally a wavelength conversion chip. Even more specifically, solid state light sources containing a plurality of light emitting diodes and optionally at least one wavelength conversion chip. The wavelength conversion chip may incorporate an electrical interconnection means to allow an electrical connection to one or more of the light emitting diode chips. Parallel and orthogonal mounting of the LEDs within or on at least one wavelength conversion element is disclosed. Orthogonal mounting is referred to as a rib emitter within this filing. Methods of forming an inorganic light emitting diode (LED) chip is also disclosed along with reactor designs and packaging methods for LEDs formed on freestanding nitride foils.
BACKGROUND OF THE INVENTIONSolid-state light sources can incorporate, for example, one or more LEDs and optionally may include one or more phosphor materials. A typical conventional, solid-state light source is constructed from one or more packaged LEDs. Each LED package may contain one light emitting, multilayer semiconductor structure mounted on a substrate that includes appropriate electrical contacts. Alternatively, each package may contain one multi-layer semiconductor structure mounted on a substrate and include a wavelength conversion material consisting of phosphor particles that may be embedded in a transparent polymer. The wavelength conversion material usually covers the emitting area of the LED.
Both the LEDs and the phosphors used in conventional solid-state light sources have deficiencies that can be eliminated in order to provide less expensive light sources and to provide sources with higher optical outputs. In addition, the standard combined LED/phosphor package is bulky and is deficient in many ways. Some of the deficiencies of conventional solid-state light sources are described below.
Conventional LEDs are fabricated by epitaxially growing multiple layers of semiconductors on a growth substrate. Inorganic light-emitting diodes can be fabricated from GaN-based semiconductor materials containing gallium nitride (GaN), aluminum nitride (AlN), aluminum gallium nitride (AlGaN), indium nitride (InN), indium gallium nitride (InGaN) and aluminum indium gallium nitride (AlInGaN). Other appropriate materials for LEDs include, for example, aluminum gallium indium phosphide (AlGaInP), gallium arsenide (GaAs), indium gallium arsenide (InGaAs), indium gallium arsenide phosphide (InGaAsP), diamond, boron nitride and zinc oxide (ZnO). Especially important LEDs for this invention are GaN-based LEDs that emit light in the ultraviolet, blue, cyan and green regions of the optical spectrum and AlGaInP-based LEDs that emit in the orange and red spectral regions.
The total thickness of the semiconductor layers for a conventional GaN-based LED is only about 3 microns. The layers are fabricated by epitaxially growing a layered semiconductor structure on a growth substrate using metal organic chemical vapor deposition (MOCVD), which has a very slow growth rate of approximately 0.1 micron per hour. This results in deposition times of several hours and makes the growth of thicker layers prohibitively expensive. Crystal quality is also compromised due to hetereoepitaxial nature of the growth and the thinness of the layer. The use of exotic growth techniques and processes such at superlattices and trenching are required to prevent cracking of the multi-layered structure. The approximately 3-micron thick multilayer semiconductor structure is very fragile and will break easily if removed from the growth substrate to form a free-standing die. The semiconductor layers must therefore either remain attached to the growth substrate or, alternatively, be attached to a transfer substrate using wafer bonding techniques followed by removal of the growth substrate. The wafer bonding techniques are expensive and can be unreliable. In addition wafer bonding introduces additional thermal boundary interfaces which in high power density applications can become the main thermal impedance for a device. The added steps also increase the cost of manufacturing LEDs. Removal of the growth substrate can be done by a laser liftoff process, chemical processing or mechanical polishing.
The growth substrate for GaN-based LEDs is usually sapphire or silicon carbide and is chosen to closely match the crystallographic structure of the epitaxial layers. A transfer substrate, if utilized, can be a metal, another semiconductor material such as silicon or a ceramic material such as aluminum nitride. Such growth or transfer substrates may not be suitable for the final LED device. For example, sapphire is a poor thermal conductor and is therefore not the most effective thermal conductor to direct heat away from the semiconductor layers. Thermal considerations are very important for LEDs, which generate a significant amount of heat during operation. The heat lowers the light output and operating lifetime of the LED. As LED sizes become larger, such heating effects become more important and can seriously degrade the light-output performance and lifetime of the LEDs. In all cases the growth substrate increase the thermal mass, which dramatically effects how rapidly temperature changes can occur during epitaxial growth. It should be noted that typical MOCVD growth of multi-layered nitride devices occurs at over 1000 C with rapid fluctuations of greater than 200 C required to define the various layers. As an example In composition of 20% InGaN may be processed at temperatures less than 720 C while GaN layers are typically processed at up to 1050 C at 200 Torr. The need exists for methods, equipment, and substrates which can allow for rapid changes in the growth temperature and process gas conditions within a reactor. In existing reactor designs are limited to growth rates in the angstroms/minute range because of the thermal mass of the substrate/platen and residence time of the process gases within the reactor itself. This combined with the need for stress compensation layers such as superlattices or internal reflectors to improve extraction efficiency has lead to very complex epitaxial growths consisting of hundreds of layers lasting several hours. The solid state light source disclosed in this invention eliminates the need for these layers by providing a low thermal mass native substrate based on freestanding nitride foils and a package which extracts light from both sides of the light emitting diode. The result is a very simple and efficient LED which does not require an internal reflector layer or superlattice.
In addition, the growth or transfer substrate may absorb some of the light emitted by the LED, thereby lowering the optical output. The substrate may also trap some of the light generated by the LED, resulting in an additional loss in optical output. Light trapping is caused by the high refractive index of the substrate relative to air and results in total internal reflection of emitted light back through the substrate and back through the epitaxial layers.
It would be desirable to develop thick, rugged LED chips that do not include either growth or transfer substrates and that can be easily handled without breaking. Different growth techniques will be required to make such a structure since MOCVD is too slow to fabricate thick multi-layer semiconductor structures. It is well known in the art that thicker nitride layers offer lower defect densities. This is typically accomplished using HVPE which provides for growth rates of up to 100 microns/hour and superior crystal quality over MOCVD due to elimination of carbon impurities which are inherent to MOCVD. Unfortunately, it has proven difficult to use HVPE to grow the full device structure due to high growth rates. As such the industry has attempted to use HVPE templates which consist of thick GaN layers grown on sapphire, silicon or SiC. Templates however are susceptible to cracking during the rapid thermal changes needed to grow the active devices due to thermal expansion mismatch between the nitride or GaN layer and the underlying non-native substrate. It would be desirable to make use of freestanding nitride foils in combination with novel reactor designs that take advantage of the lower thermal inertia the foils. This would enable faster and more precise growth cycles, fast temperature ramps and result in improved materials properties while eliminating thermal mismatch issues. Using an initial prototype reactor the ability of fast temperature ramps have been demonstrated exceeding 200 C/sec on freestanding foils. Conventional nitride templates (with secondary substrates) shatter under much lower thermal transients.
In general, homoepitaxial growth is preferred over heteroepitaxial growth due to reduced stresses, higher crystal quality, and the elimination of bowing. As disclosed below much higher heating ramp rates and growth rates can be realized doing homoepitaxial growth on freestanding nitride foils versus both template and non-native substrates. In conventional nitride MOCVD, sapphire, SiC or Silicon is used as growth substrates. These materials all exhibit significantly different thermal expansion coefficients throughout the processing temperature range of the device. The thermal mismatches lead to bowing during and after growth. During growth non-native substrates can bow in excess of 200 microns. The direction of bow also can change during the temperature cycling. This results in non-uniform heating, increases in strain across the wafer which in turn creates variations in doping concentration, layer thickness ultimately lowering device yield. The bow also reduces yield for post processes such as lithography and packaging where non-flat wafers must be processed and handled. The thermal mismatch also affects device design itself. Typically nucleation layers, superlattices, trenching or other methods must be employed to prevent cracking of the final epitaxial layers making up the final device. It would be desirable to eliminate these costly additional growth and processing steps by taking advantage of flexible, native, and high crystal quality freestanding nitride foils.
In standard LED-based light source designs, the back side of the LED opposite the light emitting side is a reflective surface. The light generated within the LED is substantially isotropic in nature. As such directing light out of only one surface or in one direction is inherently less efficient than extracting light out of multiple surfaces. It would be desirable to be able to extract light from not just one surface but 2,3,4,5 or all surfaces of an LED. It would also be desirable to minimize the thermal conductive path from all of these LED surfaces to ambient. The use of thermally conductive luminescent elements allow for light to be extracted out of all surfaces of an LED while maintaining a thermal conduction path to the ambient. Alternately, thick LEDs based on freestanding nitride foils offer and opportunity to use orthogonal mountings or rib emitters. In this configuration heat is removed through the edge of rib emitter and 5 surfaces are substantially emitter surfaces. In general, eliminating the back reflecting surface reduces the average optical pathlength of the emitted light within the LED structure, thereby reducing optical absorption within the LED and increasing the external quantum efficiency.
A conventional wavelength conversion material for solid-state lighting typically consists of a phosphor powder that may be embedded in a transparent polymer. The wavelength conversion material can be deposited, for example, as a dome that covers the output surface of the LED.
The standard approach to produce wavelength conversion materials begins by making bulk solid phosphors using solid-state processing as known in the art. These phosphors are then ground down to powders in the micron size range and deposited on a surface using a variety of deposition techniques such as settling, encapsulation within a polymer matrix or spray coating. Though relatively inexpensive, the phosphors generated using these methods suffer from high levels of dislocations and lattice defects. In addition, the compositional purity is also difficult to maintain. In the majority of cases, this does not represent a major problem because of the reduced excitation levels. It has been shown in accelerated aging studies, however, that very high excitation levels can degrade the output luminescence of powdered phosphors severely and impact overall life performance. These levels of high excitation exist within solid-state lighting applications. This is mainly due to the small size and concentrated flux density of the LED die itself.
Several material characteristics such as lattice defects, out-gassing, catalytic degradation of organic binders, photochemical degradation of organic binders, and compositional purity contribute to the problems of light output degradation and/or loss in efficiency for phosphor materials. It has been shown that polycrystalline and mono-crystalline phosphor films either grown on a substrate or as single crystal boules tend to exhibit much better luminosity and life characteristics than powders. In addition, every phosphor has a thermal quenching level that can degrade the output at the temperatures created by elevated excitation levels. In the case of powdered phosphors, this can be a major issue because the phosphor particles are usually isolated from any reasonable thermal conduction path. At very high excitation levels, the energy associated with less than unity quantum efficiency and Stokes shift losses can induce a significant localized thermal rise within the phosphor particles. The need exists for the creation of an improved thermal conduction path for the luminescent material. Also, the scattering created by the use of a powder can reduce the overall light output due to the backscattering and subsequent absorption of the generated light. In addition, because phosphor particles exhibit very high scatter typically very high dopant concentrations are required to convert the excitation wavelength in a very thin layer. High dopant concentrations are typically lower efficiency than low dopant concentration wavelength converters due to concentration quenching and increased lattice defects. It would be desirable to be able to use thicker wavelength conversion materials with low dopant concentrations but with high thermal conductivity to allow for thermal conduction of heat away from the light emitting diodes. It would also be desirable to be able to use wavelength conversion chips or plates which exhibit sufficient thickness and transparency such that waveguiding modes can also be used. It would be desirable to be able to use such solid state light sources in which distributed and waveguided sources can be combined enabling adjustable color temperature sources.
Mueller-Mach et al. in U.S. Pat. No. 6,696,703 disclose the deposition of a thin film phosphor directly on the LED die. However, as-deposited thin film phosphors have relatively poor wavelength conversion efficiency. A high-temperature annealing step is required in order to properly activate the phosphor. This annealing step can damage the semiconductor layers of the LED. In addition, the absorption cross-sections of most thin film phosphors are low, especially for blue and near ultraviolet (UV) excitations typically used for solid-state lighting. It is neither economical nor practical in most cases to create a sufficiently thick layer of luminescent material grown directly on the LED. Another drawback to depositing a phosphor directly on the LED die is that a large portion of the light generated within a deposited phosphor layer can be trapped due to total internal reflectance. The need therefore exists for a method to utilize high performance phosphors within an LED package such that the best phosphor can be used efficiently (e.g. with sufficient quantity, minimal backscatter, and maximum light extraction). The need also exists for a method to fabricate high efficiency phosphors without damaging the LED semiconductor layers.
Mueller-Mach et al. in U.S. Pat. No. 6,630,691 disclose a thin single-crystal phosphor substrate onto which an LED structure is fabricated by epitaxial growth techniques. However, single-crystal phosphor substrates are expensive and finding a single crystal phosphor substrate that has the proper lattice match to allow the growth of the LED structure can be difficult.
Ng et al. in US Patent Application No. 20050006659 disclose a planar sheet of a single-crystal phosphor that is placed over the output surface of an LED as a portion of a preformed transparent cap. However, single-crystal phosphor sheets must be grown by epitaxial processes or sliced from bulk single crystals of phosphor material. Single crystal phosphor sheets are therefore too expensive for most practical applications. Planar sheets of polycrystalline phosphors are not disclosed in US Patent Application No. 20050006659. Bonding the planar sheet of a single-crystal phosphor directly to the surface of the LED to improve heat dissipation in the phosphor sheet is also not disclosed. In addition, single crystal phosphor sheets can be created in a very limited range of the dopant concentration as such additional means must be used to adjust the light output to a reasonable color temperature.
A need exists to maximize the efficiency of wavelength conversion materials within a solid-state lighting application and to improve the thermal conductivity properties of the materials. In addition, a need exits for low-cost phosphors that have light extraction enhancements and the ability to control the level and type of scatter within the phosphor in order to enhance the overall conversion efficiency. Also additional means are required to reach useful color temperature.
It would be desirable to replace the conventional wavelength conversion material with a solid wavelength conversion chip that could be bonded to the surface of an LED chip or an array of LED chips. This cannot be done with most types of LED devices since the wavelength conversion chip would cover up one or both of the LED electrodes and prevent attachment of electrical connections to the LED. It would be desirable, therefore, to incorporate an electrical interconnection means within the wavelength conversion chip.
A conventional LED package containing an LED and a wavelength-converting phosphor is bulky compared to the light emitting epitaxial layered structure itself. If many LED packages are used in the solid-state light source, the light source is significantly larger and thicker than necessary. In typical LED applications, localized high intensity sources create localized heat sources which require large heavy heat sinks. As such commercially available LED light sources exhibit very low light output to weight ratios or lumens/gram performance. A typical PAR 38 light source which outputs 600 lumens weighs over 200 grams and therefore has a output ratio of 3 lumens/gram. In comparison, an incandescent lamp can put out 2000 lumens and weighs 40 grams or 50 lumens/gram Fluorescent lights exhibited similar high output ratios. Therefore it would be desirable to have an LED light source that exhibited comparable light output to weight ratios to existing incandescent lights. This would help market acceptance of the more efficient solid state lighting. The solid state light source disclosed in this invention based on arrays of LEDs sandwiched between thermally conductive luminescent chips or plates exhibit greater than 50 lumens/gram performance levels. This is enabled not only by thermally conductive luminescent elements but also by the ability to cost effectively grow LEDS which emit from both sides.
It would be desirable to form stacks of light emitting chips, where each stack includes one LED chip and one or more wavelength conversion chips. Such stacks of chips can be handled individually or combined with other stacks to form larger distributed light sources. Such stacks of light emitting chips could be utilized for applications such as backlights for liquid crystal displays (LCDs) or for general lighting applications such as room lighting. It would also be desirable if the LED chip used in such a stack has a thick, rugged multilayer semiconductor structure and that the chip does not retain the growth substrate nor utilize a transfer substrate. Eliminating the transfer and growth substrates can improve the thermal conduction properties of the light emitting chips. By using freestanding nitride foils the thermal interfaces created using non-native growth substrates can be eliminated.
It would also be desirable to fabricate a solid-state light source that is a stack of two or more LED chips and optionally includes one or more wavelength conversion chips. Such stacks could be handled and used individually or formed into planar arrays to construct a distributed light source. The arrays can be created by either mounting a number of individual sources on a flexible or rigid interconnect or via multiple LEDs connected in a two dimensional array within at least one larger wavelength conversion chip and preferably two wavelength conversion chips
It would also be desirable if the stacks of LED chips and wavelength conversion chips could be electrically connected in series, parallel or anti-parallel configurations or in some combination of series, parallel or anti-parallel configurations. Anti-parallel electrical configurations are desirable if the electrical power source is an alternating current source.
U.S. Patent Publication No. 20050269582 discloses a ceramic phosphor layer bonded to a conventional LED. In one example, the LED is a flip chip device with both electrodes on the side of the LED opposite the ceramic phosphor so that no electrodes are in the way when bonding the ceramic phosphor layer to the LED. The LED includes a growth substrate that is still attached to the top surface of the semiconductor layers of the LED. The phosphor layer is also bonded to the growth substrate, but on the side opposite the semiconductor layers. In a second example, the p-contact layer of the LED is attached to a transfer (host) substrate and the ceramic phosphor layer is attached to the n-layer opposite the transfer (host) layer. The n-contact is adjacent to the ceramic phosphor layer and on the same side of the LED as the phosphor layer. The original growth substrate has been removed but the transfer substrate remains with the device.
U.S. Patent Publication No. 20050269582 does not disclose LEDs that have neither a growth substrate nor a transfer substrate as an element of the LED die. U.S. Patent Publication No. 20050269582 does not disclose LEDs where the n-type layer, the p-type layer or both the n-type layer and the p-type layers are thick enough so that the multilayer semiconductor structure of the LED is rugged and may be handled as a free-standing chip without having the growth or transfer substrate still attached. In addition, U.S. Patent Publication No. 20050269582 does not disclose wavelength conversion chips that include an electrical interconnection means. U.S. Patent Publication No. 20050269582 also does not disclose a stack of light emitting chips where one chip is an LED chip and another chip is a wavelength conversion chip that includes an electrical interconnection means. U.S. Patent Publication No. 20050269582 does not disclose a stack of light emitting chips where at least two of the chips are LED chips and where the stack optionally includes at least one wavelength conversion chip.
Another problem with conventional LEDs that are comprised of a nitride layer on a secondary foreign substrate is that processes used in packaging the die or the metallic interconnects that are used limit subsequent processing to less than 400° C. This requirement many times eliminates the use of inorganic adhesives or interconnect materials that require processing temperatures in excess of 300° C. or 400° C. Inorganic materials are desirable over organic materials that break down over time when exposed to heat, light or high operating temperatures. This can severely limit the lifetime of the LED solid state light source. It would be desirable to be able to use LEDs that could tolerate these higher processing temperatures in order that more robust materials used for interconnects and other packaging functions.
Many types of conventional solid-state light sources that emit high output lumens or high output power attempt to generate the required lumens or power from one large LED die or from an array of closely spaced LED die. Although useful for light sources requiring a small emitting area or etendue, this type of light source has two deficiencies for general lighting applications such as room lighting. One deficiency is that the high output intensity from such a concentrated source can exceed eye safety standards and can be a safety hazard. It would be desirable instead to make large-area, distributed light sources using many smaller light source chips so that light intensity safety standards are not exceeded. To accomplish this, the light source chips need to be inexpensive, have a simple low cost interconnect, and be easy to handle. It would also be desirable if the light source chips were constructed as stacks of LED chips and wavelength conversion chips.
A conventional high powered LED typically operates at high current density, for example 1-2 amperes per square millimeter of LED area, resulting in lower external quantum efficiency than would be the case if the LED were operating at lower currents. Usually the highest external quantum efficiency for the LED device is obtained at a much lower current density of about 0.1-0.2 amperes per square millimeter of LED area. Using a single LED die operating at high current density instead of several smaller LEDs operating a low current density is usually dictated by the high cost of packaging multiple LED die. It would be desirable to develop lower cost methods to manufacture LEDs so that one large LED can be replaced by several smaller light source chips operating at the most efficient current density. Freestanding nitride foils can be used to reduce cost via higher throughput, increased yield and higher efficiency. As such this invention discloses methods, equipment and articles which take advantage of freestanding nitride foils.
An additional problem with high powered light sources is that a single, high-powered LED die or a closely spaced array of LED die can produce a significant amount of heat that must be dissipated quickly to prevent the die from overheating. Metal heat sinks with large area fins are generally required for convection cooling of such devices. By combining high efficiency arrays of LEDs with thermally conductive luminescent elements that both distribute the heat over a larger area for enhanced natural convection cooling and also optically distribute the light emitted over a larger area (for eye safety reasons) novel high output solid state light sources can be constructed without the need for additional heatsinking. More specifically, the authors have demonstrated that power densities of greater than 1 watt/cm2 can be natural convectively cooled using thermally conductive luminescent chips and arrays of small LEDs sandwiched between the thermally conductive luminescent chips while maintaining a junction temperature under 80 C in a standard ambient. As such sheets of light can be constructed with very high lumen output using this approach. Solid state light sources are approaching 100 to 150 lumens/watt performance levels. As such solid state light source using this approach will be able to emit 1000 lumens with sources areas of approximately 1 to 2 square inches of area.
Conventional LED-based light sources are cooled by a heat sink in thermal contact with the LED die. The LED die includes either a growth substrate or a transfer substrate. Heat flows from the LED semiconductor layers, through the growth or transfer substrate and through the heat sink to ambient. The heat sink may include fins or other types of structures to transfer heat to an ambient fluid or gas, such as air or water. It would be desirable to develop LED-based light sources where the LED die do not include the growth substrate or a transfer substrate and where the LED die can be cooled without a special appended heat sink. In such cases, heat will flow directly from the light source to an ambient gas or fluid such as air or water. As such the growth of LEDs on freestanding nitride foils, reactor designs based on freestanding nitride foils, and articles made from freestanding nitride foils are disclosed within this invention.
Jaluria has modeled continuous horizontal CVD reactors and other authors have demonstrated continuous and semi-continuous tape based CVD reactors to fabricate Si ribbon solar cells and superconductors. It would be desirable to have methods, equipment and articles for growing LEDs based on freestanding nitride foils using continuous MOCVD. It would be desirable to eliminate the need to grow the nucleation layer, thick n contact layer, and stress compensation layers utilized in conventional LED fabrication. This would reduce the total epitaxial growth layer to less than 3000 Angstroms from 3 microns or more required in more conventional LED MOCVD approaches. This would significantly lower LED fabrication cost and speed reactor cycle time. In fact by doing this it would enable a continuous belt or tape reactor to be used in LED fabrication. The mounting, reactor chamber design, heating methods, process gas delivery methods, control, and monitoring methods are all embodiments of this invention. More specifically, the need exists for rapid thermal MOCVD reactors in which the thermal mass is sufficiently small and the process gas residency time is sufficiently short enough to enable the use of higher growth rate processing conditions. The authors have demonstrated the use of lamp based rapid thermal MOCVD on freestanding foils both with a susceptor and via direct heating of the freestanding foils. The use of other techniques to indirectly and directly rapidly heat the freestanding foils is also disclosed. Preferred methods include but are not limited to direct current heating, hot gas rapid thermal heating, and laser heating.
Semi-continuous reactor designs for superconductor ribbons and solar cells made from silicon ribbons have been developed by CVD Equipment and other reactor manufacturers. Alternately, rapid thermal processing techniques based on gas heating using atmospheric plasma jets have been developed by Bollinger WO 2001035041 for RTP of silicon wafers. It is disclosed that these techniques can be used to grow LEDs on freestanding nitride foils and bulk nitride wafers. The reactor designs, methods, and articles produced are embodiments of this invention. Semi-continuous, continuous, and sequential rapid thermal reactor designs are disclosed. In conventional LED fabrication, existing commercial reactors utilize large rotating platens on which either 42-2 inch wafers or 6-6 inch wafers are mounted. Processing time is typically around 6-8 hours. If six—6 inch wafers are used a total of approximately 108,000 1mm2 LEDS can be produced in one run, ignoring yield losses, edge effects, etc. That produces a throughput of approximately 300 1mm2 die/minute. This batch processing is not conducive to high volume high throughput production. For this reason a number of solar cell companies have been moved away from batch processing of silicon wafers and developed reactor designs based on silicon ribbons processed in continuous reactors. It would be desirable to be able to use a continuous reactor processing small 1 cm2 freestanding nitride foils that output 1 foil every second and produce 6000 1mm2 die/minute. Freestanding nitride foils and bulk nitride wafers and tapes enable the use of continuous, semicontinuous, and sequential reactors to increase the throughput of LED manufacture by orders of magnitude over batch processing of large area wafers. It should be noted that the bowing problems discussed earlier are substantially worse on larger wafers so the assumption of 100% yield above is very conservative. Yields are typically less than 20% if wavelength, Rf, and intensity binning is required.
Also known in the art are the benefits of gas micro-reactor designs. Conventional large volume MOCVD reactors used commercially exhibit significant material usage inefficiencies. Micro-reactors are inherently more efficient than large volume reactors due to higher material utilization. In addition, residency times of the process gases are significantly reduced in micro or small volume reactors. This is especially important for nitrides due to the cost of metallorganic precursors. Typically commercial reactors have four or more metallorganic sources in each reactor with associated temperature, pressure and flow rate controls. This increases reactor cost, material usage, and downtime. It would be beneficial to be able to utilize the benefits of micro-reactor designs while simultaneously increasing overall throughput. The reactor designs disclosed in this invention take advantage of centralized metallorganic sources such as offered by DOW in their VaporStation product line. These centralized metallorganic sources were developed to reduce downtime for high volume metallorganics like TMGa. In this invention these centralized sources will be used to eliminate the need for source stabilization and decrease the process gas residency time by combining centralized metallorganic sources with micro-reactor chambers. In this manner higher material utilization and higher throughput can be realized.
The many deficiencies, described above, of conventional solid-state light sources and production methods thereof can be eliminated by the various embodiments of this invention as will be shown in the summary, the figures and the detailed descriptions of the preferred embodiments below.
SUMMARY OF THE INVENTIONOne embodiment of this invention is a solid-state light source that includes at least one stack of elements. The elements in the stack include at least one inorganic LED chip and optionally at least one wavelength conversion chip. The inorganic LED chip emits internally generated light of a first wavelength range. The wavelength conversion chip converts at least a portion of the light of a first wavelength range into light of a second wavelength range, different than the first wavelength range. The wavelength conversion chip may include an electrical interconnection means to allow an electrical connection to the LED chip. The interconnect means may include a via, a via plus solder bump, an electrode embedded in the wavelength conversion chip or an electrode fabricated on the surface of the wavelength conversion chip.
The wavelength conversion chip electrodes may be optically transparent, semi-transparent, semi-opaque or opaque. The electrodes may be fabricated from metals, metal alloys, high-temperature-fusible conductive materials, semiconductors or transparent conductive oxides. An example of a high-temperature-fusible material is a conductive silver paste or ink. Examples of transparent conductive oxides include indium tin oxide, zinc oxide, indium-doped zinc oxide and aluminum-doped zinc oxide. A preferred transparent conductive oxide is aluminum-doped zinc oxide. Preferably the aluminum-doped zinc oxide is fabricated by metal-organic chemical vapor deposition and preferably the aluminum-doped zinc oxide electrode is greater than 500 nanometers thick.
The inorganic LED chip can be, for example, a conventional LED chip that is fabricated so that the total thickness of the epitaxial semiconductor layers, also denoted as a multilayer semiconductor structure, is less than about 5 microns thick and is too fragile to form a self-supporting device. To provide structural support, a conventional LED chip retains the growth substrate upon which the multilayer semiconductor structure was fabricated or includes a transfer substrate that is bonded to the multilayer semiconductor structure opposite the growth substrate during the fabrication process. The transfer substrate, if present, provides structural support to the epitaxial layers once the growth substrate is removed.
A conventional LED chip has a first side and an opposing second side. Usually light is emitted predominately from only one side of the chip. The opposing side is substantially covered by one or more reflecting layers and emits little, if any, light.
The LED chip or chips utilized in this invention do not need to be conventional LED chips. An inorganic LED chip can also be a chip that does not include a growth or a transfer substrate. Such a substrate-free inorganic LED chip has at least one thick epitaxial layer to provide structural support to the chip. The thick epitaxial layer can be an n-doped layer, semi-insulating, or a p-doped layer with polar, non-polar, or semi-polar crystal orientation. Optionally both the n-doped layer and the p-doped layer may be thick layers. One embodiment of this invention is the use of freestanding nitride foils that are fabricated in a novel reactor. As an example, a 30 micron thick n doped GaN freestanding foil with a surface of 1 square centimeter may be suspended (by a wire through a hole in the freestanding foil) and processed through a horizontal (or optionally vertical) multi-stage MOCVD reactor wherein each layer is grown in a separate reactor stage. Direct heating of the freestanding foil may be done via one of the following means: gas heating, plasma heating, IR heating, laser heating, induction heating, direct current heating through the foil, and IR/visible/UV lamp heating. Alternately, indirect heating via a susceptor is also disclosed. More preferably the susceptor thermal mass would be minimized to reduce thermal time constant of the overall system consisting of but not limited to metal foils or graphite foils. By directly coupling the heating means to the freestanding foil the additional thermal mass of a susceptor is eliminated. The use of deposited layers on the freestanding foil to enhance the coupling of the heating means into the freestanding foil or to control where epitaxial growth occurs is also an embodiment of this invention.
Because very high temperature ramps can be realized due to the homogenous nature and low thermal mass of the freestanding foils, and the elimination of an n-layer, nucleation, and superlattice layers a new type of reactor is possible. As such this invention discloses novel reactor designs tailored specifically towards high speed epitaxial growth of epitaxial layers and devices on freestanding nitride devices for use in the solid state light sources disclosed in this invention. More specifically, rapid thermal MOCVD reactor with indirect and direct heating of the freestanding foils is disclosed.
Unlike existing large platen rotating disk MOCVD reactors this invention discloses a rapid thermal MOCVD reactor with direct or indirect heating or a combination of both direct and indirect heating of freestanding nitride foils and bulk nitride wafers. While the use of bulk nitride wafers in rapid thermal MOCVD reactors is an embodiment of this invention, freestanding nitride foils are preferred due to lower thermal masses allowing more rapid thermal temperature ramping. Even more preferred is a rapid thermal MOCVD reactor for freestanding foils with a semi-continuous or continuous multi-chamber configuration. Most preferred is a rapid thermal MOCVD reactor with a sequential multi-chamber micro reactor configuration based on direct heating of the freestanding foils via gas heating. In this approach high temperature processes gases (nitrogen, argon, ammonia, or hydrogen) impinges directly on the freestanding foils such that localized heating occurs. Alternately, the use of plasma jet heating (as described by Bollinger WO 2001035041) but in this case using the freestanding nitride foils or bulk nitride wafers is also an embodiment of this invention. In general, the use of heated process gases to heat low thermal mass freestanding nitride foils is an embodiment of this invention. Controlling the temperature via in-situ measurement of band-edge of the freestanding foils or temperature monitoring by other optical means is preferred. In the case of direct heating of freestanding foils transmissive measurement of the freestanding nitride foil bandage is preferred. Alternately, the use of the freestanding nitride foil as a micro balance to monitor process conditions is also disclosed. In a manner similar to quartz microbalance as known in the art the piezoelectric nature of the freestanding nitride foil maybe used to measure and monitor the epitaxially grown layer thickness within the reactor during growth and fed back to control the gas flow or temperature within the reactor.
The substrate-free LED of this invention includes a first doped layer, a first electrode in electrical contact with the first doped layer, a second doped layer, a second electrode in electrical contact with the second doped layer, and an active region interposed between the first doped layer and the second doped layer. The active region emits internally generated light in a first wavelength range. The first electrode and the second electrode may be optically transparent, semi-transparent, semi-opaque or opaque. The electrodes may be fabricated from metals, metal alloys, high-temperature-fusible conductive materials, semiconductors or transparent conductive oxides (TCOs). An example of a high-temperature-fusible material is a conductive silver paste or ink. Examples of TCOs include indium tin oxide (ITO), zinc oxide (ZnO), indium-doped zinc oxide (IZO) or aluminum-doped zinc oxide (AZO). A preferred TCO is aluminum-doped zinc oxide. Preferably the aluminum-doped zinc oxide is fabricated by metal-organic chemical vapor deposition (MOCVD) and preferably the aluminum-doped zinc oxide electrode is greater than 500 nanometers thick.
A major advantage of a substrate free LED as utilized in this invention is that they can tolerate subsequent high temperature processing such that more robust materials may be used for interconnects and packaging. Substrate-free LED chips can incorporate high-temperature electrodes that are interconnected and packaged using process temperatures greater than 400 degrees Centigrade. Examples of high-temperature processes that may be used include laser welding, brazing, glass encapsulation and fusion bonding. The use of conductive inks as used in the solar cell industry may also be used. A preferred low cost high volume method of deposition is printing via screen printing, inkjet, or tampo. Rapid firing formulations of conductive inks are preferred embodiments. This enables high throughput belt reactors to be used. Conductive inks with Silver and/or aluminum with high reflectivities are also preferred.
The LED chip or chips utilized in this invention may also be unconventional chips that have a first side and an opposing second side and that emit light from both the first and second sides. In such a device, neither the first side nor the second side is substantially covered by reflecting layers.
The wavelength conversion chip or chips utilized in this invention include a solid layer formed from wavelength conversion materials. The wavelength conversion materials may be, for example, phosphor materials or quantum dot materials. The phosphor materials may be in the form of powders, ceramics, thin solid films or bulk solids. Preferred forms are ceramics and thin solid films. The wavelength conversion layer may also be formed from two or more different wavelength conversion materials. The wavelength conversion layer may include optically inert host materials for the wavelength conversion phosphors or quantum dots.
Another embodiment of this invention is a solid-state light source that includes at least one stack of elements, where the elements in the stack include at least two substrate-free inorganic LED chips. Preferably, at least one of the substrate-free inorganic LED chips is substantially transparent to light emitted by the second LED chip. Optionally the stack also includes at least one wavelength conversion chip.
Another embodiment of this invention is a solid-state light source that includes at least one stack of elements, where the elements include at least one substrate-free LED chip and two wavelength conversion chips. The substrate-free LED chip has a first side and an opposing second side. Light of a first wavelength range is emitted from both the first side and the second side of the LED chip. One wavelength conversion chip is bonded to the first side of the at least one substrate-free LED chip and the other wavelength conversion chip is bonded to the opposing second side of the at least one substrate-free LED chip. The wavelength conversion chips convert a portion of the light of a first wavelength range into light of at least a second wavelength range. The area of the wavelength conversion chip may be larger than the area of the at least one LED chip. Light is emitted from substantially all sides of the solid-state source.
Another embodiment of this invention is a solid-state light source that includes a plurality of stacks arranged in a linear array or a substantially linear array. Each stack includes at least one substrate-free LED chip and at least one wavelength conversion chip. The stacks can be electrically connected in series, parallel or anti-parallel configurations or in some combination of series, parallel or anti-parallel configuration.
Another embodiment of this invention is a solid-state light source that includes a plurality of stacks arranged in a two-dimensional array, regular or irregular, to form an extended area light source. Each stack includes at least one substrate-free LED chip and at least one wavelength conversion chip. The stacks can be mounted on a reflecting substrate or a transparent substrate. The stacks can be electrically connected in series, parallel or anti-parallel configurations or in some combination of series, parallel or anti-parallel configuration. The LEDs may be substantially parallel to the wavelength conversion chips or substantially orthogonal to wavelength conversion chip. A preferred embodiment is the use of slots or cutouts within the wavelength conversion chip such that the LED die may be mounted such that a substantial portion of the light emitted by the LED die is waveguided through the wavelength conversion chip. When the majority of the LED die surface is parallel with the wavelength conversion chip much of the light emitted by the LED die passes only through the thinnest portion of the wavelength conversion chip. To convert the light from the LED die in first wavelength range to second wavelength range may then require higher doping concentrations in the wavelength converting chip. This higher doping concentration will limit the light spreading to a larger area of the wavelength converting chip before being emitted by the light source. This creates visible hot spots which are endemic of conventional LED light sources. If the LED chip is arranged to be perpendicular to the wavelength conversion chip or such that the larger emitting area of the the LED is coupled such that the light emitted from the LED chip is waveguided down the wavelength conversion chip. This provides allows for a longer optical path spreading the light out to be emitted from a larger area of the output side of the wavelength conversion chip. Lower doping concentrations may be used in the wavelength conversion chip which provides less absorption and higher efficiencies. This also eliminates hot spots in the light source and makes it safer to view directly. Combinations of these different mounting configuration may be used as means of controlling the color temperature of the overall solid state light source. LED chips mounted substantially orthogonal to the long axis or large face of the wavelength conversion chip are herein referred to as rib emitters. Rib emitters may also be used as standalone light sources where the majority of the thermal conduction out of the rib emitter is orthogonal to the majority of the optical emission. The high thermal conductivity of the freestanding nitride foil (up to 200 W/m/K) enables the mounting configuration disclosed. Another benefit of this mounting configuration is enhanced optical extraction as 5 surfaces are able to emit. In thin film LEDs the majority of the emission is off one surface only. The light generated within the active region is isotropic in nature. Through the use of internal and external reflective layers most of the light isotropically generated is directed towards the emitting surface. This however is not the most efficient method of optical extraction. The mounting configuration and LEDs disclosed in the invention extract light from more than one 2, 3, 4, or 5 surfaces of the LED.
Another embodiment of this invention is a method of making a solid-state light source that includes a stack of elements. The elements in the stack include at least one inorganic LED chip and at least one wavelength conversion chip.
A more detailed understanding of the present invention, as well as other objects and advantages thereof not enumerated herein, will become apparent upon consideration of the following detailed description and accompanying drawings, wherein:
The preferred embodiments of the present invention will be better understood by those skilled in the art by reference to the above listed figures. The preferred embodiments of this invention illustrated in the figures are not intended to be exhaustive or to limit the invention to the precise form disclosed. The figures are chosen to describe or to best explain the principles of the invention and its applicable and practical use to thereby enable others skilled in the art to best utilize the invention.
Light emitting diodes can be fabricated by epitaxially growing multiple layers of semiconductors on a growth substrate. Inorganic light-emitting diodes can be fabricated from gallium nitride (GaN) based semiconductor materials containing, for example, gallium nitride (GaN), aluminum nitride (AlN), aluminum gallium nitride (AlGaN), indium nitride (InN), indium gallium nitride (InGaN) and/or aluminum indium gallium nitride (AlInGaN). Other appropriate materials for LEDs include, for example, aluminum gallium indium phosphide (AlGaInP), gallium arsenide (GaAs), indium gallium arsenide (InGaAs), indium gallium arsenide phosphide (InGaAsP), diamond, boron nitride (BN) and zinc oxide (ZnO).
Especially important LEDs for this invention are GaN-based LEDs that utilize epitaxially-grown layers that can include GaN, AlN, AlGaN, InN, InGaN or AlInGaN. Depending on the composition of the semiconductor layers, GaN-based LEDs emit light in the ultraviolet, blue, cyan or green regions of the optical spectrum. The growth substrate for GaN-based LEDs is typically sapphire (Al2O3), silicon carbide (SiC), bulk gallium nitride or bulk aluminum nitride. Although the embodiments of this invention will be described using GaN-based LEDs, other types of LEDs including, but not limited to, AlGaInP and ZnO LEDs may also be utilized in the embodiments.
Typical epitaxial growth methods for thin semiconductor layers of gallium-nitride-based materials include chemical vapor deposition (CVD), metal-organic chemical vapor deposition (MOCVD), vapor phase epitaxy (VPE), hydride vapor phase epitaxy (HVPE) and molecular beam epitaxy (MBE). MOCVD is the most common method for conventional GaN-based LEDs where the total thickness of the epitaxial layers is less than about 5 microns. MOCVD is a relatively slow deposition method with growth rates of approximately 0.1 micron per hour. HVPE has much higher growth rates (10 microns per hour is possible). HVPE can be used to fabricate substrate-free LEDs that are described in this invention and where the thickness of one of the layers is at least 5 microns and could be as much as 30 microns or more.
The embodiments of this invention can utilize conventional GaN-based LED chips that are fabricated so that the total thickness of the epitaxial semiconductor layers, which include a first doped layer, a second doped layer and an active region interposed between the first doped layer and the second doped layer, is less than about 5 microns thick. The complete set of epitaxial semiconductor layers is also denoted in this application as the multilayer semiconductor structure. The first doped layer and the second doped layers of the multilayer semiconductor structure can be, respectively, an n-doped layer and a p-doped layer or the layers can be reversed so that the first doped layer is a p-doped layer and the second doped layer is an n-doped layer.
When the multilayer semiconductor structure is less that about 5 microns thick, it is too fragile to form a self-supporting device. To provide structural support, a conventional LED chip retains the growth substrate upon which the multilayer semiconductor structure was fabricated or includes a transfer substrate that is bonded to the multilayer semiconductor structure opposite the growth substrate during the fabrication process. The transfer substrate, if present, provides structural support to the epitaxial layers once the growth substrate is removed.
A conventional LED chip usually emits light predominately from one side of the chip. The opposing side is substantially covered by one or more reflecting layers and emits little, if any, light.
Examples of conventional LED chips of the prior art that can be utilized in embodiments of this invention are illustrated in
The first electrode 102 is in electrical contact with the first doped layer 108 and the second electrode 114 is in electrical contact with the second doped layer 112. The first electrode and the second electrode may be fabricated from metals or metal alloys. For example, the first electrode and the second electrode may be formed from one or more metals or metal alloys containing, but not limited to, silver, aluminum, gold, nickel, titanium, chromium, platinum, palladium, rhodium, rhenium, ruthenium and tungsten.
The multilayer semiconductor structure 104 of the LED chip 100 can be fabricated from GaN-based semiconductor materials containing GaN, AlN, AlGaN, InN, InGaN and/or AlInGaN. Alternatively, the multilayer semiconductor structure can be fabricated from any appropriate light-emitting semiconductor material.
The active region 110 of the multilayer semiconductor structure 104 is a p-n homojunction, a p-n heterojunction, a single quantum well or a multiple quantum well of the appropriate semiconductor material for the LED.
For purposes of illustration, LED chip 100 is assumed to be a GaN-based LED chip. The important fabrication steps for this GaN-based, illustrative example will be briefly summarized.
First a multilayer semiconductor structure 104 is fabricated on a growth substrate 106. The growth substrate is sapphire. The multilayer semiconductor structure includes a first doped layer 108, an active region 110 and a second doped layer 112. The growth substrate 106 has a first surface 120 and a second surface 122 opposite the first surface.
The first doped layer 108 is an n-doped GaN layer, which is epitaxially deposited or otherwise conventionally fabricated on the second surface 122 of the growth substrate. The first doped layer 108 has a first surface 124 and a second surface 126 opposite the first surface. The first surface 124 of the first doped layer is in contact with surface 122 of the growth substrate.
The active region 110 is a GaN-based multiple quantum well structure, which is epitaxially deposited or otherwise conventionally fabricated on the second surface 126 of the first doped layer 108. The active region 110 has a first surface 128 and a second surface 130 opposite the first surface. The first surface 128 of the active region is in electrical contact with the second surface 126 of the first doped layer.
The second doped layer 112 is a p-doped GaN layer, which is epitaxially deposited or otherwise conventionally fabricated on the second surface 130 of the active region 110. The second doped layer has a first surface 132 and a second surface 134 opposite the first surface. The first surface 132 of the second doped layer is in electrical contact with the second surface 130 of the active region.
A portion 116 of the second doped layer 112 and the active region 110 is removed to expose a portion 116 of the second surface 126 of the first doped layer. The first electrode 102 and the second electrode 114 are fabricated from aluminum. An aluminum layer is deposited on the second surface 134 of the second doped layer and the exposed portion 116 of the second surface 126 of the first doped layer. The aluminum layer is patterned by standard photolithographic techniques to form the first electrode 102 and the second electrode 114. First electrode 102 has a first surface 136 and a second surface 138. The first surface 136 of the first electrode is in electrical contact with the second surface 126 of the first doped layer. Second electrode 114 has a first surface 140 and a second surface 142. First surface 140 of the second electrode is in electrical contact with the second surface 134 of the second doped layer.
The first electrode 102 only partially covers the exposed portion 116 of the second surface 126 of the first doped layer. The second electrode 114 only partially covers the second surface 134 of the second doped layer. The remaining portion of the exposed portion 116 of the second surface 126 of the first doped layer and the second surface 134 of the second doped layer are an output or exit surface for the light emitted by the LED 100. To form a back reflector 115, a layer of silver is deposited on the first surface 120 of the growth substrate.
In summary, LED chip 100 has a first electrode 102, a multilayer semiconductor structure 104 that includes first-doped, active and second-doped layers, a growth substrate 106 and a second electrode 114. LED chip 100 has a first side 152 and a second side 154. The first side 152 is substantially adjacent to the first doped layer 108. The second side 154 is substantially adjacent to the second doped layer 112. The active region 110 emits internally generated light in a first wavelength range when a current is applied through the first electrode 102 and the second electrode 114. The light is emitted from the second side 154 of the LED.
The total thickness 150 of the multilayer semiconductor structure 104 for LED chip 100 is less than 5 microns. In this illustrative example, the thickness of the first doped layer (the n-doped layer) is approximately 3 microns, the thickness of the active region (a multi-quantum well structure) is approximately 0.5 microns and the thickness of the second doped layer (the p-doped layer) is approximately 0.5 microns, resulting in a total thickness of 4 microns. In this example, all the semiconductor layers are grown by MOCVD.
When utilized as in a light source, LED chip 100 is normally attached to a submount or leadframe (neither is shown). The submount or leadframe acts as a heat transfer element or heatsink to remove heat generated by the device during operation.
Example light rays 160, 162 and 164 illustrate internally generated light that is emitted by the active region 110 of the LED. Internally generated light ray 160 is emitted by active region 110 toward output surface 134 of the LED chip. Internally generated light ray 160 is directed at an angle to surface 134 that is less than the critical angle, which allows light ray 160 to exit the LED chip through surface 134.
Internally generated light ray 162 is emitted by active region 110 toward the rear reflector 115 of the LED. Internally generated light ray 162 is reflected by reflector 115 and directed to the output surface 134 at an angle less than the critical angle. Internally generated light ray 162 exits the LED chip through surface 134.
Internally generated light ray 164 is directed to surface 134 at an angle that is greater than the critical angle. Internally generated light ray 164 is reflected by total internal reflection and is redirected toward the rear reflector 115 of the LED chip.
In the illustrative example in
It is well known by those skilled in the art that the multilayer semiconductor structure 104 may include additional layers in order to adjust and improve the operation of the LED chip 100. For example, a current spreading layer may be inserted between surface 136 of the first electrode 102 and surface 126 the first doped layer 108. Such a current spreading layer will have the same conductivity type as the first doped layer and will improve the uniformity of current injection across the entire active region. In addition, a current spreading layer may be inserted between surface 134 of the second doped layer and surface 140 of the second electrode 114. The latter current spreading layer will have the same conductivity type as the second doped layer. As another example, an electron blocking layer or a hole blocking layer may inserted either between surface 126 of the first doped layer 108 and surface 128 of the active region 110 or between surface 130 of the active region 110 and surface 132 of the second doped layer 112. An electron blocking layer reduces the escape of electrons from the active region. A hole blocking layer reduces the transfer of holes through the layer.
Another conventional LED design is illustrated in
Except for the back reflecting surfaces, most of the elements of LED chip 200 are the same as LED chip 100. Conventional LED chip 200 includes a first electrode 102, a multilayer semiconductor structure 104, a second electrode 114 and a growth substrate 106. The multilayer semiconductor structure 104 includes a first doped layer 108, an active region 110 and a second doped layer 112, which is on the opposite side of the active region 110 from the first doped semiconductor layer 108.
The first electrode 102 is in electrical contact with the first doped layer 108 and the second electrode 114 is in electrical contact with the second doped layer 112. For LED chip 200, the second electrode is a reflecting electrode and covers substantially all of surface 134 of the second doped layer 112. The first electrode and the second electrode may be fabricated from reflecting metals.
The multilayer semiconductor structure 104 of LED chip 200 can be fabricated from GaN-based semiconductor materials containing GaN, AlN, AlGaN, InN, InGaN and/or AlInGaN. Alternatively, the multilayer semiconductor structure can be fabricated from any appropriate light-emitting semiconductor material.
The active region 110 of the multilayer semiconductor structure 104 is a p-n homojunction, a p-n heterojunction, a single quantum well or a multiple quantum well of the appropriate semiconductor material for the LED chip 200.
For purposes of illustration, LED chip 200 is assumed to be a GaN-based LED. The important fabrication steps for this GaN-based, illustrative example will be briefly summarized. Many of the fabrication steps are identical to the steps for LED chip 100 and will not be repeated.
First, a multilayer semiconductor structure 104 of LED chip 200 is fabricated on a sapphire growth substrate 106 using the same methods that are described above for LED 100. The growth substrate 106 has a first surface 120 and a second surface 122 opposite the first surface. The multilayer semiconductor structure includes a first doped layer 108 that is n-doped GaN, an active region 110 that is a GaN-based multiple quantum well structure and a second doped layer 112 that is p-doped GaN.
A portion 116 of the second doped layer 112 and the active region 110 is removed to expose a portion 116 of the second surface 126 of the first doped layer. The first electrode 102 and the second electrode 114 are fabricated from aluminum. An aluminum layer is deposited on the second surface 134 of the second doped layer and the exposed portion 116 of the second surface 126 of the first doped layer. The aluminum layer is patterned by standard photolithographic techniques to form the first electrode 102 and the second electrode 114. The first surface 136 of the first electrode is in electrical contact with the second surface 126 of the first doped layer. First surface 140 of the second electrode is in electrical contact with the second surface 134 of the second doped layer.
The first electrode 102 partially covers the exposed portion 116 of the second surface 126 of the first doped layer. The second electrode 114 substantially covers the second surface 134 of the second doped layer. Surface 136 of the first electrode and surface 140 of the second electrode form the back reflector for LED chip 200.
In summary, LED chip 200 has a first electrode 102, a multilayer semiconductor structure 104 that includes first-doped, active and second-doped layers, a growth substrate 106 and a second electrode 114. LED chip 200 has a first side 252 and a second side 254. The first side 252 is substantially adjacent to the first doped layer 108. The second side 254 is substantially adjacent to the second doped layer 112. The active region 110 emits internally generated light in a first wavelength range when a current is applied through the first electrode 102 and the second electrode 114. The light is emitted from the first side 252 of the LED.
The total thickness 150 of the multilayer semiconductor structure 104 for LED chip 200 is less than 5 microns. In this illustrative example, the thickness of the first doped layer (the n-doped layer) is approximately 3 microns, the thickness of the active region (a multi-quantum well structure) is approximately 0.5 microns and the thickness of the second doped layer (the p-doped layer) is approximately 0.5 microns, resulting in a total thickness of 4 microns. In this example, all the semiconductor layers are grown by MOCVD.
When utilized as in a light source, LED chip 200 is normally attached to a submount or a leadframe (neither is shown). The submount or leadframe acts as a heat transfer element or heatsink to remove heat generated by the device during operation. The submount or leadframe also includes electrical interconnections that attach to the first electrode and second electrode.
Example light rays 260 and 262 illustrate internally generated light that is emitted by the active region 110 of the LED. Internally generated light ray 260 is emitted by active region 110 toward the first surface 120 of the growth substrate. Surface 120 is also the output surface of the LED. Internally generated light ray 260 is directed at an angle to surface 120 that is less than the critical angle, which allows light ray 260 to exit the LED chip through surface 120.
Internally generated light ray 262 is emitted by active region 110 toward the second electrode 114 of the LED. Internally generated light ray 262 is reflected by the surface 140 of the second electrode and is directed to the output surface 120 at an angle less than the critical angle. Internally generated light ray 262 exits the LED chip through surface 120.
Another conventional LED design is illustrated in
Except for the removal of the growth substrate and the addition of a transfer substrate 302, most of the elements of LED chip 300 are the same as LED chip 200. Conventional LED chip 300 includes a first electrode 102, a multilayer semiconductor structure 104, a second electrode 114 and a transfer substrate 302. The multilayer semiconductor structure 104 includes a first doped layer 108, an active region 110 and a second doped layer 112, which is on the opposite side of the active region 110 from the first doped semiconductor layer 108.
The first electrode 102 is in electrical contact with the first doped layer 108 and the second electrode 114 is in electrical contact with the second doped layer 112. For LED 300, the second electrode is a reflecting electrode and covers substantially all of surface 134 of the second doped layer 112. The first electrode and the second electrode may be fabricated from reflecting metals.
The multilayer semiconductor structure 104 of LED chip 300 can be fabricated from GaN-based semiconductor materials containing GaN, AlN, AlGaN, InN, InGaN and/or AlInGaN. Alternatively, the multilayer semiconductor structure can be fabricated from any appropriate light-emitting semiconductor material.
The active region 110 of the multilayer semiconductor structure 104 is a p-n homojunction, a p-n heterojunction, a single quantum well or a multiple quantum well of the appropriate semiconductor material for the LED chip 300.
For purposes of illustration, LED chip 300 is assumed to be a GaN-based LED. The important fabrication steps for this GaN-based, illustrative example will be briefly summarized. Many of the fabrication steps are identical to the steps for LED chip 100 and LED chip 200 and will not be repeated.
First a multilayer semiconductor structure 104 of LED chip 300 is fabricated on a sapphire growth substrate (not shown) using the same methods that are described above for LED 100. The multilayer semiconductor structure includes a first doped layer 108 that is n-doped GaN, an active region 110 that is a GaN-based multiple quantum well structure and a second doped layer 112 that is p-doped GaN.
A portion 116 of the second doped layer 112 and the active region 110 is removed to expose a portion 116 of the second surface 126 of the first doped layer. The first electrode 102 and the second electrode 114 are fabricated from aluminum. An aluminum layer is deposited on the second surface 134 of the second doped layer and the exposed portion 116 of the second surface 126 of the first doped layer. The aluminum layer is patterned by standard photolithographic techniques to form the first electrode 102 and the second electrode 114. The first surface 136 of the first electrode is in electrical contact with the second surface 126 of the first doped layer. First surface 140 of the second electrode is in electrical contact with the second surface 134 of the second doped layer.
The first electrode 102 partially covers the exposed portion 116 of the second surface 126 of the first doped layer. The second electrode 114 substantially covers the second surface 134 of the second doped layer. Surface 136 of the first electrode and surface 140 of the second electrode form the back reflector for LED chip 300.
A transfer substrate 302 is bonded to surface 138 of the first electrode and surface 142 of the second electrode. The transfer substrate is usually chosen to have good thermal conductivity and a thermal expansion coefficient that is similar to GaN. The transfer substrate may be any solid material such as a composite that is chosen for its thermal expansion properties or a ceramic material. In this example, the transfer substrate will have electrical interconnections (not shown) that attach to electrodes 102 and 114.
After the transfer substrate is attached, the growth substrate is removed by standard processing steps. For example, the growth substrate can be removed by a laser liftoff process, a chemical process or by mechanical polishing.
In summary, LED chip 300 has a first electrode 102, a multilayer semiconductor structure 104 that includes first-doped, active and second-doped layers, a transfer substrate 302 and a second electrode 114. LED chip 300 has a first side 352 and a second side 354. The first side 352 is substantially adjacent to the first doped layer 108. The second side 354 is substantially adjacent to the second doped layer 112. The active region 110 emits internally generated light in a first wavelength range when a current is applied through the first electrode 102 and the second electrode 114. The light is emitted from the first side 352 of the LED.
The total thickness 150 of the multilayer semiconductor structure 104 for LED chip 300 is less than 5 microns. In this illustrative example, the thickness of the first doped layer (the n-doped layer) is approximately 3 microns, the thickness of the active region (a multi-quantum well structure) is approximately 0.5 microns and the thickness of the second doped layer (the p-doped layer) is approximately 0.5 microns, resulting in a total thickness of 4 microns. In this example, all the semiconductor layers are grown by MOCVD.
When utilized as in a light source, LED chip 300 may also include a submount (not shown), which acts as a heat transfer element or heatsink to remove heat generated by the device during operation.
Example light rays 360 and 362 illustrate internally generated light that is emitted by the active region 110 of the LED. Internally generated light ray 360 is emitted by active region 110 toward the first surface 124 of the first doped layer. Surface 124 is also the output surface of the LED. Internally generated light ray 360 is directed at an angle to surface 124 that is less than the critical angle, which allows light ray 360 to exit the LED chip through surface 124.
Internally generated light ray 362 is emitted by active region 110 toward the second electrode 114 of the LED. Internally generated light ray 362 is reflected by the surface 140 of the second electrode and is directed to the output surface 124 at an angle less than the critical angle. Internally generated light ray 362 exits the LED chip through surface 124.
Another type of conventional LED that can be used in this invention has one electrode on the “top” side of the device and one electrode on the “bottom” side. LED chip 400 illustrated in
Except for the arrangement of the electrodes, most of the elements of LED chip 400 are the same as for LED chip 300. Conventional LED chip 400 includes a first electrode 102 on the first side 452 of the device, a multilayer semiconductor structure 104, a second electrode 114 on the second side 454 and a transfer substrate 402. The multilayer semiconductor structure 104 includes a first doped layer 108, an active region 110 and a second doped layer 112, which is on the opposite side of the active region 110 from the first doped semiconductor layer 108.
The first electrode 102 is in electrical contact with the first doped layer 108 and the second electrode 114 is in electrical contact with the second doped layer 112. For LED chip 400, the second electrode is a reflecting electrode and covers substantially all of surface 134 of the second doped layer 112. The first electrode and the second electrode may be fabricated from reflecting metals.
The multilayer semiconductor structure 104 of LED chip 400 can be fabricated from GaN-based semiconductor materials containing GaN, AlN, AlGaN, InN, InGaN and/or AlInGaN. Alternatively, the multilayer semiconductor structure can be fabricated from any appropriate light-emitting semiconductor material.
The active region 110 of the multilayer semiconductor structure 104 is a p-n homojunction, a p-n heterojunction, a single quantum well or a multiple quantum well of the appropriate semiconductor material for the LED chip 400.
For purposes of illustration, LED chip 400 is assumed to be a GaN-based LED. The important fabrication steps for this GaN-based, illustrative example will be briefly summarized. Many of the fabrication steps are identical to the steps for LED chip 100, LED chip 200 and LED chip 300 and will not be repeated.
First a multilayer semiconductor structure 104 of LED chip 400 is fabricated on a sapphire growth substrate (not shown) using the same methods that are described above for LED 100. The multilayer semiconductor structure includes a first doped layer 108 that is n-doped GaN, an active region 110 that is a GaN-based multiple quantum well structure and a second doped layer 112 that is p-doped GaN.
The second electrode 114 is fabricated from aluminum. An aluminum layer is deposited on the second surface 134 of the second doped layer. The first surface 140 of the second electrode is in electrical contact with the second surface 134 of the second doped layer. Surface 140 of the second electrode also forms the back reflector for LED chip 400.
A transfer substrate 402, and in particular surface 404 of the transfer substrate, is bonded to surface 142 of the second electrode. The transfer substrate is usually chosen to have good thermal conductivity and a thermal expansion coefficient that is similar to GaN. The transfer substrate may be any solid material such as a metal composite that is chosen for its thermal expansion properties or a ceramic material. The transfer substrate 402 is also either an electrical conductor or includes and electrical interconnect to the second electrode.
After the transfer substrate is attached, the growth substrate is removed by standard processing steps, exposing the first surface 124 of the first doped layer. For example, the growth substrate can be removed by a laser liftoff process, a chemical process or by mechanical polishing.
The first electrode 102 is fabricated from aluminum. An aluminum layer is deposited on the previously exposed first surface 124 of the first doped layer. The aluminum layer is patterned by standard photolithographic techniques to form the first electrode 102. The first surface 136 of the first electrode is in electrical contact with the first surface 124 of the first doped layer. The first electrode 102 partially covers surface 124 of the first doped layer.
In summary, LED chip 400 has a first electrode 102, a multilayer semiconductor structure 104 that includes first-doped, active and second-doped layers, a transfer substrate 402 and a second electrode 114. LED chip 400 has a first side 452 and a second side 454. The first side 452 is substantially adjacent to the first doped layer 108. The second side 454 is substantially adjacent to the second doped layer 112. The active region 110 emits internally generated light in a first wavelength range when a current is applied through the first electrode 102 and the second electrode 114. The light is emitted from the first side 452 of the LED.
The total thickness 150 of the multilayer semiconductor structure 104 for LED chip 400 is less than 5 microns. In this illustrative example, the thickness of the first doped layer (the n-doped layer) is approximately 3 microns, the thickness of the active region (a multi-quantum well structure) is approximately 0.5 microns and the thickness of the second doped layer (the p-doped layer) is approximately 0.5 microns, resulting in a total thickness of 4 microns. In this example, all the semiconductor layers are grown by MOCVD.
Example light rays 460 and 462 illustrate internally generated light that is emitted by the active region 110 of the LED. Internally generated light ray 460 is emitted by active region 110 toward the first surface 124 of the first doped layer. Surface 124 is also the output surface of LED 400. Internally generated light ray 460 is directed at an angle to surface 124 that is less than the critical angle, which allows light ray 460 to exit the LED chip through surface 124.
Internally generated light ray 462 is emitted by active region 110 toward the second electrode 114 of the LED. Internally generated light ray 462 is reflected by the surface 140 of the second electrode and is directed to the output surface 124 at an angle less than the critical angle. Internally generated light ray 462 exits the LED chip through surface 124.
When utilized as in a light source, LED chip 400 may also include a submount or heatsink (neither are shown) that acts as a heat transfer element to remove heat generated by the device during operation.
The heat flow from the multilayer semiconductor structure of LED chip 400 is illustrated schematically in
Non-conventional substrate-free LED chips may also be utilized in embodiments of this invention. Substrate-free LED chips have multilayer semiconductor structures that are at least 10 microns thick and, furthermore, the chips do not include either a growth substrate or a transfer substrate. The multilayer semiconductor structures utilized for the substrate-free LED chips are thick enough so that the LED chips can be handled as free-standing structures without breaking.
For the substrate-free LED chips that are utilized this invention, either the first doped layer or the second doped layer is at least 10 microns thick. Preferably either the first doped layer or the second doped layer is at least 15 microns thick. More preferably, either the first doped layer or the second doped layer is at least 20 microns thick. Most preferably, either the first doped layer or the second doped layer is at least 25 microns thick. Alternatively, both the first doped layer and the second doped layer are each at least 5 microns thick, preferably each of the two layers is at least 10 microns thick and more preferably each of the two layers is at least 15 microns thick.
The total thickness of the multilayer semiconductor structure for the substrate-free LED chips is at least 10 microns thick. Preferably the total thickness of the multilayer semiconductor structure is at least 20 microns thick. More preferably the total thickness of the multilayer semiconductor structure is at least 30 microns thick.
Since thicker semiconductor layers are utilized for the substrate-free LED chip, the optical absorption coefficients for the various layers must be low in order to prevent the absorption of a significant fraction of the internally generated light that is emitted by the active region of the chip. Lower optical absorption within the LED chip will result in higher light extraction from the chip and higher external quantum efficiency.
In some applications, it is also important that the LED chip be highly reflective to any externally incident light that comes from other light sources and is directed at the chip. The optical absorption coefficients for the various semiconductor layers must also be low in this latter case so that any externally incident light that enters the chip will not undergo significant absorption by the semiconductor layers before exiting the chip. Lowering the optical absorption coefficients of the semiconductor layers will increase the reflectivity of the LED chip to externally incident light.
The multilayer semiconductor structure of the LED chip can absorb light and has an absorption coefficient that depends on wavelength. In many cases, the absorption coefficient is not uniform across the different semiconductor layers of the multilayer semiconductor structure. If the different semiconductor layers that make up the multilayer semiconductor structure have different absorption coefficients, the absorption coefficient for the multilayer semiconductor structure is defined in this specification as the thickness-weighted average absorption coefficient. The weighting function is the fractional thickness of each semiconductor layer in the multilayer semiconductor structure. For example, if 100% of the thickness of the multilayer semiconductor structure has a uniform absorption coefficient of 50 per centimeter in the emitting wavelength range, then the thickness-weighted average absorption coefficient is 50 per centimeter. If 50% of the thickness of the multilayer semiconductor structure has an absorption coefficient of 25 per centimeter and 50% of the thickness of the multilayer semiconductor structure has an absorption coefficient of 75 per centimeter, then the thickness-weighted average absorption coefficient is also 50 per centimeter.
In order to improve the light extraction efficiency and external quantum efficiency of an LED chip and to improve the reflectivity of LED chip to externally incident light, the absorption coefficient (i.e. the thickness-weighted average absorption coefficient) of the multilayer semiconductor structure in the emitting wavelength range of the internally generated light should be less than 20 per centimeter. Preferably the absorption coefficient of the multilayer semiconductor structure in the emitting wavelength range of the internally generated light is less than 10 per centimeter. More preferably, the absorption coefficient of the multilayer semiconductor structure in the emitting wavelength range is less than 5 per centimeter. Most preferably, the absorption coefficient of the multilayer semiconductor structure in the emitting wavelength range is less than 2 per centimeter.
Minimizing the absorption coefficient of the multilayer semiconductor structure in the emitting wavelength range of the internally generated light can be accomplished by improving the deposition processes for the different semiconductor layers in order to reduce impurities or defects and to improve the crystalline structure of the layers.
Thick semiconductor layers can be grown by methods including, but not limited to, CVD, MOCVD, VPE, HVPE and MBE. MOCVD is the most common method for conventional GaN-based LEDs but it has relatively slow deposition rates of approximately 0.1 micron per hour. MOCVD deposited layers also have relatively high optical absorption coefficients due to impurities and defects. HVPE has much faster growth rates and is the preferred method for growing GaN layers that are more than 5 microns thick. HVPE can have growth rates of up to 10 microns per hour or more and can produce GaN-based LED layers that have optical absorption coefficients significantly less than 25 per centimeter.
For example, HVPE can be used to epitaxially grow the first doped layer or the second doped layer or both the first and the second doped layers or the entire multilayer semiconductor structure of the LED. HVPE does not have the carbon impurities that can be present in the MOCVD processes normally used in GaN LED fabrication. Alternatively, if MOCVD is used to deposit the semiconductor layers, a higher deposition temperature can be used to reduce carbon impurities and crystalline defects in the layers. MOCVD may optionally be used to grow active regions that are single- or multiple quantum wells. If the active region of the LED chip is a p-n heterojunction, preferably the entire multilayer semiconductor structure is fabricated by HVPE.
A substrate-free LED chip may have two electrodes on one side of the chip, either the “top” side or the “bottom” side. Alternatively, the substrate-free LED chip may have one electrode on the top side of the chip and one electrode on the bottom side of the chip.
The top and bottom electrodes may be optically transparent, semi-transparent, semi-opaque or opaque. The electrodes may be fabricated from metals, metal alloys, high-temperature-fusible conductive materials, semiconductors or transparent conductive oxides (TCOs). Examples of metals include silver, aluminum, gold, nickel, titanium, chromium, platinum, palladium, rhodium, rhenium, ruthenium and tungsten. Preferred metals are silver and aluminum. An example of a high-temperature-fusible material is a conductive silver paste or ink. Examples of TCOs include indium tin oxide (ITO), zinc oxide (ZnO), indium-doped zinc oxide (IZO) or aluminum-doped zinc oxide (AZO). A preferred TCO is aluminum-doped zinc oxide. Preferably the aluminum-doped zinc oxide is fabricated by metal-organic chemical vapor deposition (MOCVD) and preferably the aluminum-doped zinc oxide electrode is greater than 500 nanometers thick. The electrodes may also be omni-directional reflectors that include a dielectric layer, a metal layer and have electrically conducting pathways through the metal layer.
Both electrodes should be highly reflective to internally generated light to prevent excessive light absorption inside the chip. In addition, making the external surfaces of the electrodes highly reflective will result in an LED chip that has higher reflectivity to externally incident light. Preferably the reflectivity of the electrodes is greater than 90 percent to both internally generated light and externally incident light in the emitting wavelength range. More preferably, the reflectivity of the electrodes is greater than 95 percent in the emitting wavelength range. Most preferably, the reflectivity of the electrodes is greater than 98 percent in the emitting wavelength range.
Examples of substrate-free LED chips for this invention that have at least one thick epitaxial layer and that do not have either a growth substrate or a transfer substrate are illustrated in
First, examples of substrate-free LED chips that have two electrodes on the top side of the chip are now described. The chips are illustrated in
Substrate-free LED chip 500 illustrated in
Substrate-free LED chip 500 includes a first electrode 102, a multilayer semiconductor structure 504, a second electrode 114 and a back reflector 115. The multilayer semiconductor structure 504 includes a first doped layer 108, an active region 110 and a second doped layer 112, which is on the opposite side of the active region 110 from the first doped semiconductor layer 108. Consequently, the active region is interposed between the first doped layer and the second doped layer. The active region is in electrical contact with the first doped layer and the second doped layer and the active region emits light in a first wavelength range when a current is applied through the first and second electrodes.
The first electrode 102 is in electrical contact with the first doped layer 108 and the second electrode 114 is in electrical contact with the second doped layer 112. The first electrode 102 and the second electrode 114 may be optically transparent, semi-transparent, semi-opaque or opaque. The electrodes may be fabricated from metals, metal alloys, high-temperature-fusible conductive materials, semiconductors or transparent conductive oxides (TCOs). Examples of metals include silver, aluminum, gold, nickel, titanium, chromium, platinum, palladium, rhodium, rhenium, ruthenium and tungsten. Preferred metals are silver and aluminum. An example of a high-temperature-fusible material is a conductive silver paste or ink. Examples of TCOs include indium tin oxide (ITO), zinc oxide (ZnO), indium-doped zinc oxide (IZO) or aluminum-doped zinc oxide (AZO). A preferred TCO is aluminum-doped zinc oxide.
Preferably the aluminum-doped zinc oxide is fabricated by metal-organic chemical vapor deposition (MOCVD) and preferably the aluminum-doped zinc oxide electrode is greater than 500 nanometers thick. The electrodes may also be omni-directional reflectors that include a dielectric layer, a metal layer and have electrically conducting pathways through the metal layer.
The multilayer semiconductor structure 504 of the LED chip 500 can be fabricated from GaN-based semiconductor materials containing GaN, AlN, AlGaN, InN, InGaN and/or AlInGaN. Alternatively, the multilayer semiconductor structure can be fabricated from any appropriate light-emitting semiconductor material.
The active region 110 of the multilayer semiconductor structure 504 is a p-n homojunction, a p-n heterojunction, a single quantum well or a multiple quantum well of the appropriate semiconductor material for the LED.
For purposes of illustration, substrate-free LED chip 500 is assumed to be a GaN-based LED chip. The important fabrication steps for this GaN-based, illustrative example will be briefly summarized.
First a multilayer semiconductor structure 504 is fabricated on a sapphire growth substrate (not shown). The multilayer semiconductor structure includes a first doped layer 108, an active region 110 and a second doped layer 112.
The first doped layer 108 is an n-doped GaN layer, which is epitaxially deposited or otherwise conventionally fabricated on a growth substrate. The first doped layer 108 has a first surface 124 and a second surface 126 opposite the first surface. The first doped layer is at least 10 microns thick. Preferably the first doped layer is at least 15 microns thick. More preferably, the first doped layer is at least 20 microns thick. Most preferably, the first doped layer is at least 25 microns thick. The first doped layer may be deposited by any standard GaN growth technique. Preferably, the first doped layer is deposited by HVPE.
The active region 110 is a GaN-based multiple quantum well structure, which is epitaxially deposited or otherwise conventionally fabricated on the second surface 126 of the first doped layer 108. The active region 110 has a first surface 128 and a second surface 130 opposite the first surface. The first surface 128 of the active region is in electrical contact with the second surface 126 of the first doped layer.
The second doped layer 112 is a p-doped GaN layer, which is epitaxially deposited or otherwise conventionally fabricated on the second surface 130 of the active region 110. The second doped layer has a first surface 132 and a second surface 134 opposite the first surface. The first surface 132 of the second doped layer is in electrical contact with the second surface 130 of the active region.
A portion 116 of the second doped layer 112 and the active region 110 is removed to expose a portion 116 of the second surface 126 of the first doped layer. The first electrode 102 and the second electrode 114 are fabricated from aluminum. An aluminum layer is deposited on the second surface 134 of the second doped layer and the exposed portion 116 of the second surface 126 of the first doped layer. The aluminum layer is patterned by standard photolithographic techniques to form the first electrode 102 and the second electrode 114. First electrode 102 has a first surface 136 and a second surface 138. The first surface 136 of the first electrode is in electrical contact with the second surface 126 of the first doped layer. Second electrode 114 has a first surface 140 and a second surface 142. First surface 140 of the second electrode is in electrical contact with the second surface 134 of the second doped layer.
The first electrode 102 only partially covers the exposed portion 116 of the second surface 126 of the first doped layer. The second electrode 114 only partially covers the second surface 134 of the second doped layer. The remaining portion of the exposed portion 116 of the second surface 126 of the first doped layer and the second surface 134 of the second doped layer are an output or exit surface for the light emitted by the LED chip 500.
To form a substrate-free LED chip, the growth substrate is removed by any conventional process including laser liftoff, chemical processes and mechanical polishing. Removing the growth substrate exposes first surface 124 of the first doped layer.
To form a back reflector 115, a layer of silver is deposited on the first surface 124 of the first doped layer. Optionally, the back reflector may also be an omni-directional reflector that includes a dielectric layer (not shown) and a metal layer.
In summary, substrate-free LED chip 500 has a first electrode 102, a multilayer semiconductor structure 504 that includes first-doped, active and second-doped layers and a second electrode 114. LED chip 500 has neither a growth substrate nor a transfer substrate. LED chip 500 has a first side 552 and a second side 554. The first side 552 is substantially adjacent to the first doped layer 108. The second side 554 is substantially adjacent to the second doped layer 112. The active region 110 emits internally generated light in a first wavelength range when a current is applied through the first electrode 102 and the second electrode 114. The light is emitted from the second side 554 of the LED.
The total thickness 550 of the multilayer semiconductor structure 504 for LED chip 500 is at least 10 microns. Preferably the total thickness 550 of the multilayer semiconductor structure is 20 microns. More preferably, the total thickness 550 of the multilayer semiconductor structure is 30 microns. As an illustrative example, the thickness of the first doped layer (the n-doped layer) is approximately 20 microns, the thickness of the active region (a multi-quantum well structure) is approximately 0.5 microns and the thickness of the second doped layer (the p-doped layer) is approximately 0.5 microns, resulting in a total thickness of 21 microns. In this example, the first doped layer is grown by HVPE and the remainder of the semiconductor layers is grown by MOCVD.
Example light rays 560, 562 and 564 illustrate internally generated light that is emitted by the active region 110 of the LED. Internally generated light ray 560 is emitted by active region 110 toward output surface 134 of the LED chip. Internally generated light ray 560 is directed at an angle to surface 134 that is less than the critical angle, which allows light ray 560 to exit the LED chip through surface 134.
Internally generated light ray 562 is emitted by active region 110 toward the rear reflector 115 of the LED. Internally generated light ray 562 is reflected by reflector 115 and directed to the output surface 134 at an angle less than the critical angle. Internally generated light ray 562 exits the LED chip through surface 134.
Internally generated light ray 564 is directed to surface 134 at an angle that is greater than the critical angle. Internally generated light ray 564 is reflected by total internal reflection and is redirected toward the rear reflector 115 of the LED chip.
Substantially all of the first side 552 of LED 500 is covered by lower reflector 115. Due to the reflectivity of reflector 115, the reflectivity of surface 138 of first electrode 102 and the reflectivity of surface 142 of second electrode 114, substrate-free LED chip 500 can reflect externally incident light. Externally incident light is light that is directed to the light emitting side of the LED from another light source or light that is emitted by the LED and is reflected back to the light emitting side of the LED as recycled light. For some applications, for example applications utilizing light recycling to increase the effective brightness of the LED, it is important that the LED have high reflectivity to externally incident light. High reflectivity to externally incident light will exist if the reflecting layers of the LED have high reflectivity (e.g. greater than 70%) and if the absorption coefficient of the multilayer semiconductor structure is low (e.g. less than 20 per centimeter). Preferably LED 500 reflects at least 60 percent of externally incident light directed to the light emitting side (second side 554) of the LED. More preferably, LED 500 reflects at least 70 percent of externally incident light. Most preferably, LED 500 reflects at least 80 of externally incident light.
Example light rays 570, 572 and 574 in
In the illustrative example in
It is well known by those skilled in the art that the multilayer semiconductor structure 504 may include additional layers in order to adjust and improve the operation of the LED chip 500. For example, a current spreading layer may be inserted between surface 136 of the first electrode 102 and surface 126 the first doped layer 108. Such a current spreading layer will have the same conductivity type as the first doped layer and will improve the uniformity of current injection across the entire active region. In addition, a current spreading layer may be inserted between surface 134 of the second doped layer and surface 140 of the second electrode 114. The latter current spreading layer will have the same conductivity type as the second doped layer. As another example, an electron blocking layer or a hole blocking layer may inserted either between surface 126 of the first doped layer 108 and surface 128 of the active region 110 or between surface 130 of the active region 110 and surface 132 of the second doped layer 112. An electron blocking layer reduces the escape of electrons from the active region. A hole blocking layer reduces the transfer of holes through the layer.
The substrate-free LED chips of this invention, including LED chip 500, preferably include light extraction elements (not shown) to aid in extracting internally generated light from the chips. The light extraction elements may be fabricated by any means, including chemical means, mechanical means such as grinding, or optical means such as laser ablation.
Substrate-free LED chip 500 does not have a growth substrate or a transfer substrate that can retard heat flow from the chip. If LED chip 500 is bonded to a surface, a submount, a heat sink or a leadframe, the thermal resistance for heat transfer is illustrated in
Substrate-free LED chip 600 in
No portion of the first side 652 of LED 600 is covered by a reflecting layer. Only a portion of the second side 654 of LED 600 is covered by the first electrode and the second electrode. Both sides of the multilayer semiconductor structure are light emitting sides and will emit internally generated light. At least a portion of the first side of the multilayer semiconductor structure and at least a portion of the second side of the multilayer semiconductor structure will also transmit externally incident light. Externally incident light is light that is directed to a light emitting side of the LED from another light source or light that is emitted by an LED and is reflected back to the light emitting side of the LED as recycled light. For some applications, where it is desirable for light from a phosphor or light from another LED to pass through the LED, the LED should transmit a large portion of externally incident light. High transmissivity will exist if the absorption coefficient of the multilayer semiconductor structure is low (e.g. less than 20 per centimeter). Preferably LED 600 transmits at least 60 percent of externally incident light directed to a light emitting side (either the first side 652 or the second side 654) of the LED. More preferably, LED 600 transmits at least 70 percent of externally incident light. Most preferably, LED 600 transmits at least 80 of externally incident light.
In substrate-free LED chip 600, the thick epitaxial layer is the first doped layer 108. Alternatively, the second doped layer may be a thick epitaxial layer.
LED chip 700 in
Example light rays 760, 762 and 764 illustrate internally generated light that is emitted by the active region 110 of the LED chip 700. Internally generated light ray 760 is emitted by active region 110 toward output surface 134 of the LED chip. Internally generated light ray 760 is directed at an angle to surface 134 that is less than the critical angle, which allows light ray 760 to exit the LED chip through surface 134.
Internally generated light ray 762 is emitted by active region 110 toward the lower reflector 115 of the LED. Internally generated light ray 762 is reflected by reflector 115 and directed to the output surface 134 at an angle less than the critical angle. Internally generated light ray 762 exits the LED chip through surface 134.
Internally generated light ray 764 is directed to surface 134 at an angle that is greater than the critical angle. Internally generated light ray 764 is reflected by total internal reflection and is redirected toward the rear reflector 115 of the LED chip.
For the remainder of this specification, the substrate-free LED chips will be illustrated as having a thick first doped layer. However, it will be apparent from the above discussion that any of the non-conventional chips illustrated in the following diagrams may instead have a thick second doped layer or may have both a thick first doped layer and a thick second doped layer.
LED chip 900 in
The first electrode 102 of LED chip 900 is fabricated on a portion 116 of the second surface 126 of the first doped layer that was previously exposed by an etching process. The second electrode 114 is fabricated on the second surface 134 of the second doped layer. The second electrode substantially covers the second surface 134. Substantially all of the light emitted by LED chip 900 is emitted through the top or first side 952 of the chip. For example, light ray 960 is emitted through surface 124. Light ray 962 is initially directed to the second side 954 but is reflected by surface 140 of the second electrode. Light ray 962 exits LED 900 through the top or first side 952.
LED chip 1100 in
The first electrode 102 and the second electrode 114 may be optically transparent, semi-transparent, semi-opaque or opaque. The electrodes may be fabricated from, for example, metals, metal alloys, high-temperature-fusible conductive materials, semiconductors or transparent conductive oxides (TCOs). Examples of metals include silver, aluminum, gold, nickel, titanium, chromium, platinum, palladium, rhodium, rhenium, ruthenium and tungsten. Preferred metals are silver and aluminum. An example of a high-temperature-fusible material is a conductive silver paste or ink. Examples of transparent conductive oxides include indium tin oxide (ITO), zinc oxide, indium-doped zinc oxide or aluminum-doped zinc oxide. A preferred transparent conductive oxides is aluminum-doped zinc oxide. Preferably the aluminum-doped zinc oxide is fabricated by metal-organic chemical vapor deposition (MOCVD) and preferably the aluminum-doped zinc oxide electrode is greater than 500 nanometers thick. The electrodes may also be omni-directional reflectors that include a dielectric layer, a metal layer and have electrically conducting pathways through the metal layer.
The first electrode 102 of LED chip 1100 is fabricated on the first surface 124 of the first doped layer. The first surface 124 was previously exposed by removing the growth substrate. In this illustrative example, the first electrode is fabricated by depositing a metal layer and patterning the layer using standard photolithographic techniques. The second electrode 114 is fabricated on the second surface 134 of the second doped layer. The second electrode substantially covers the second surface 134. Substantially all of the light emitted by LED chip 1100 is emitted through the top or first side 1152 of the chip. For example, light ray 1160 is emitted through surface 124. Light ray 1162 is initially directed to the second side 1154 but is reflected by surface 140 of the second electrode. Light ray 1162 exits LED 1100 through the top or first side 1152.
One embodiment of substrate-free LED chip 1200 illustrated in
Another embodiment of substrate-free LED chip 1200 is shown in
In addition to LED chips, embodiments of this invention usually include wavelength conversion chips. Wavelength conversion chips can absorb light of a first wavelength range emitted by the LED chip and convert the light into light of a second wavelength range, different than the first wavelength range. Wavelength conversion chips are very useful for converting ultraviolet or blue LED light into longer wavelengths such as blue (in the case of ultraviolet LEDs), cyan, green, yellow, orange, red or infrared.
Wavelength conversion chips can be fabricated separately from the LED chips and then subsequently bonded onto the light output surfaces of the LED chips. The process for fabricating wavelength conversion chips can be a batch process or a continuous web process. The resulting wavelength conversion chips are substantially planar in order to facilitate bonding to the LEDs. The length and width dimensions of the wavelength conversion chips can be greater than, equal to or smaller than the length and width dimensions of the LED chips onto which the wavelength conversion chip are attached.
Alternatively, a wavelength conversion layer can be fabricated in the form of a large planar wafer. The wavelength conversion wafer can be bonded to a planar wafer of LEDs. Afterwards the bonded wafers can be diced or otherwise cut into light source chips such that each light source chip is a stack of elements that include a wavelength conversion chip bonded to an LED chip. This alternative will be discussed below for
An exemplary wavelength conversion chip is illustrated in
A wavelength conversion layer can be formed from wavelength conversion materials. The wavelength conversion materials absorb light in a first wavelength range and emit light in a second wavelength range, where the light of a second wavelength range is different than the light of the first wavelength range and has longer wavelengths than the light of a first wavelength range. The wavelength conversion materials may be, for example, phosphor materials or quantum dot materials. The phosphor materials may be in the form of powders, ceramics, thin film solids or bulk solids. Preferred forms are ceramics and thin solid films. The wavelength conversion layer may also be formed from two or more different wavelength conversion materials. In addition, the wavelength conversion layer may also include optically inert host materials for the phosphor or quantum dot wavelength conversion materials.
Phosphor materials are typically optical inorganic materials doped with ions of lanthanide (rare earth) elements or, alternatively, ions such as chromium, titanium, vanadium, cobalt, manganese or magnesium. The lanthanide elements are lanthanum, cerium, praseodymium, neodymium, promethium, samarium, europium, gadolinium, terbium, dysprosium, holmium, erbium, thulium, ytterbium and lutetium. Optical inorganic materials include, but are not limited to, sapphire (Al2O3), gallium arsenide (GaAs), beryllium aluminum oxide (BeAl2O4), magnesium fluoride (MgF2), indium phosphide (InP), gallium phosphide (GaP), yttrium aluminum garnet (YAG or Y3Al5O12), terbium-containing garnet, yttrium-aluminum-lanthanide oxide compounds, yttrium-aluminum-lanthanide-gallium oxide compounds, yttrium oxide (Y2O3), calcium or strontium or barium halophosphates (Ca,Sr,Ba)5(PO4)3(Cl,F), the compound CeMgAl11O19, lanthanum phosphate (LaPO4), lanthanide pentaborate materials ((lanthanide)(Mg,Zn)B5O10), the compound BaMgAl10O17, the compound SrGa2S4, the compounds (Sr,Mg,Ca,Ba)(Ga,Al,In)2S4, the compound SrS, the compound ZnS, doped zinc oxide (ZnO) and nitridosilicate. There are several exemplary phosphors that can be excited at 250 nm or thereabouts. An exemplary red emitting phosphor is Y2O3:Eu3+. An exemplary yellow emitting phosphor is YAG:Ce3+. Exemplary green emitting phosphors include CeMgAl11O19:Tb3+, ((lanthanide)PO4:Ce3+,Tb3+) and GdMgB5O10:Ce3+,Tb3+. Exemplary blue emitting phosphors are BaMgAl10O17:Eu2+ and (Sr,Ba,Ca)5(PO4)3Cl:Eu2+. For longer wavelength LED excitation in the 400-450 nm wavelength region or thereabouts, exemplary optical inorganic materials include yttrium aluminum garnet (YAG or Y3Al5O12), terbium-containing garnet, yttrium oxide (Y2O3), YVO4, SrGa2S4, (Sr,Mg,Ca,Ba)(Ga,Al,In)2S4, SrS, and nitridosilicate. Exemplary phosphors for LED excitation in the 400-450 nm wavelength region include YAG:Ce3+, YAG:Ho3+, YAG:Pr3+, YAG:Tb3+, YAG:Cr3+, YAG:Cr4+, SrGa2S4:Eu2+, SrGa2S4:Ce3+, SrS:Eu2+ and nitridosilicates doped with Eu2+. Other phosphor materials not listed here are also within the scope of this invention.
Quantum dot materials are small particles of inorganic semiconductors having particle sizes less than about 30 nanometers. Exemplary quantum dot materials include, but are not limited to, small particles of CdS, CdSe, ZnSe, InAs, GaAs and GaN. Quantum dot materials can absorb light at first wavelength and then emit light at a second wavelength, where the second wavelength is longer than the first wavelength. The wavelength of the emitted light depends on the particle size, the particle surface properties, and the inorganic semiconductor material.
The transparent and optically inert host materials are especially useful to process phosphor powders or to spatially separate quantum dots. Host materials include polymer materials and inorganic materials. The polymer materials include, but are not limited to, acrylates, polystyrene, polycarbonate, fluoroacrylates, chlorofluoroacrylates, perfluoroacrylates, fluorophosphinate polymers, fluorinated polyimides, polytetrafluoroethylene, fluorosilicones, sol-gels, epoxies, thermoplastics, thermosetting plastics and silicones. Fluorinated polymers are especially useful at ultraviolet wavelengths less than 400 nanometers and infrared wavelengths greater than 700 nanometers owing to their low light absorption in those wavelength ranges. Exemplary inorganic materials include, but are not limited to, silicon dioxide, optical glasses and chalcogenide glasses.
A general process for forming wavelength conversion layers and wavelength conversion chips is now described. A wavelength conversion layer is formed by depositing phosphor materials using any one of a variety of techniques. The deposition is usually done on an inert substrate. The techniques include, but are not limited to, chemical vapor deposition (CVD), metal-organic chemical vapor deposition (MOCVD), sputtering, electron beam evaporation, laser deposition, sol-gel deposition, molecular beam epitaxy (MBE), liquid phase epitaxy (LPE), spin coating, slip casting, doctor blading and tape casting. Preferred techniques include slip casting, doctor blading, tape casting, CVD, MOCVD and sputtering. More preferred techniques include slip casting and tape casting. When the wavelength conversion layer is formed from quantum dot materials and inert host materials, deposition techniques include spin coating, slip casting, doctor blading, tape casting, self assembly, lithography, and nanoimprinting.
The thickness of the wavelength conversion layer can range from about 0.1 micron to about 2000 microns or more. Preferred thicknesses range from about 10 microns to about 500 microns.
Once the wavelength conversion layer is formed, it is optionally thermally annealed or radiation annealed in order to increase the wavelength conversion efficiency of the layer or, in the case of a phosphor powder, to sinter the powder to form a ceramic layer. This step is especially important for thin film phosphors, since as-deposited thin film phosphors may have low wavelength conversion efficiency if the deposited layer is not properly annealed. The annealing step can be any heat treatment or any radiation treatment of the wavelength conversion material in the wavelength conversion layer that anneals the phosphor. Heating the wavelength conversion material in the wavelength conversion layer to, for example, 600 degrees Celsius for one hour can result in thermal annealing of the wavelength conversion material. Appropriate annealing temperatures and times may vary for different wavelength conversion materials. Example radiation annealing treatments include subjecting the wavelength conversion material in the wavelength conversion layer to infrared, visible or ultraviolet light or subjecting the wavelength conversion material in the wavelength conversion layer to electron beam, atomic beam or ion beam bombardment. The radiation sources may be pulsed or continuous. The light sources may be incoherent or coherent (e.g. laser) sources.
If the wavelength conversion material is a phosphor powder mixed with an organic binder, the annealing step may be done in two or more parts. For example, a low temperature (less than 300 degrees Celcius) anneal can be done to remove the organic binder material. Following the removal of the organic material, a high temperature (greater than 500 degrees Celcius) anneal can then be done to sinter the phosphor powder into a ceramic solid.
The annealing step can occur after the deposition of the wavelength conversion layer on a substrate and before segmentation of the layer. However, the annealing step may also be done later in the process sequence, including after the wavelength conversion layer is removed from any substrate that is used to form the layer. Doing the annealing step after the wavelength conversion layer is removed from the substrate is necessary if the substrate cannot withstand high temperature or high radiation processing. The annealing step may be done in air, in an inert atmosphere such as nitrogen or argon or in a partial vacuum.
At this point in the process, the wavelength conversion material is in the form of an extended layer or wafer of material. The wavelength conversion wafer can be bonded in one piece to a wafer of LEDs or the wavelength conversion wafer may be segmented into wavelength conversion chips that later will be bonded to individual LED chips.
If one wishes to form wavelength conversion chips, the next process step is to segment the wavelength conversion layer into a plurality of wavelength conversion chips. Grooves or streets are formed through the wavelength conversion layer. The streets are fabricated in two directions to form a plurality of wavelength conversion chips that can be square, rectangular or any other planar geometric shape. The streets can be formed by techniques that include, but are not limited to, laser scribing, mechanical scribing or optical lithography accompanied by wet chemical etching, sputter etching or ion beam etching.
Wavelength conversion chip 1300 in
Other desirable features for the wavelength conversion chip include the incorporation of interconnection means to facilitate the making of electrical connections to an LED chip. Examples of interconnection means are shown in
The wavelength conversion chip 1600 illustrated in
The wavelength conversion chip 1700 illustrated in
The wavelength conversion chip 1800 illustrated in
The wavelength conversion chip 1900 illustrated in
The wavelength conversion chip 2000 illustrated in
The wavelength conversion chip electrodes may be optically transparent, semi-transparent, semi-opaque or opaque. The electrodes may be fabricated from metals, metal alloys, high-temperature-fusible conductive materials, semiconductors or transparent conductive oxides. An example of a high-temperature-fusible material is a conductive silver paste or ink. Examples of transparent conductive oxides include indium tin oxide, zinc oxide, indium-doped zinc oxide and aluminum-doped zinc oxide. A preferred transparent conductive oxide is aluminum-doped zinc oxide. Preferably the aluminum-doped zinc oxide is fabricated by metal-organic chemical vapor deposition and preferably the aluminum-doped zinc oxide electrode is greater than 500 nanometers thick.
Wavelength conversion chip 2050 is illustrated in
The wavelength conversion chip 2100 illustrated in
The wavelength conversion chip 2200 illustrated in
The building blocks for this invention have been described above. The building blocks are LED chips (either conventional or non-conventional substrate-free) and wavelength conversion chips.
One embodiment of this invention is a solid-state light source that includes at least one stack of elements. The elements of the stack include a conventional inorganic LED chip and at least one wavelength conversion chip, where the wavelength conversion chip incorporates an electrical interconnection means. The conventional LED chip is defined in this specification as an LED chip that includes either a growth substrate or a transfer substrate and that has a multilayer semiconductor structure that is less than 5 microns thick. Examples of conventional LED chips are illustrated above in
Example solid-state light source 2300 in
The bond between the LED chip 100 and the wavelength conversion chip 2100 is formed by a transparent bonding layer 2302. Example bonding materials for bonding layer 2302 include ceramic adhesives, sol-gel compositions, low melting point glasses and transparent polymers. Preferred bonding materials include ceramic adhesives made by Aremco Products, Inc. The ceramic adhesives may be filled with alumina or silica powders or the ceramic adhesives may be unfilled.
The LED chip 100 in solid-state light source 2300 emits, for example, blue light. Wavelength conversion chip 2100 converts a portion of the blue light to, for example, yellow light.
Example solid-state light source 2400 in
Example solid-state light source 2600 in
Example solid-state light source 2700 in
Another embodiment of this invention is a solid-state light source that is comprised of at least one stack of elements. The elements include at least one non-conventional substrate-free LED chip and a wavelength conversion chip. The wavelength conversion chip may optionally include an electrical interconnection means. The substrate-free LED chip is defined in this specification as an LED chip that includes neither a growth substrate nor a transfer substrate. In addition, a substrate-free LED chip has at least one thick semiconductor layer (either the first doped layer or the second doped layer or both the first and second doped layers) that is at least 10 microns thick, preferably at least 15 microns thick, more preferably at least 20 microns thick and most preferably at least 25 microns thick. The multilayer semiconductor structure of the substrate-free LED is at least 10 microns thick, preferably at least 20 microns thick, more preferably at least 30 microns thick. The thick layer or layers can be grown by any standard technique, but preferably the one or more thick semiconductor layers are grown by HVPE. Examples of this embodiment that include just one stack of elements are illustrated in
First, the examples in
Example solid-state light source 2800 in
Example solid-state light source 2900 in
Next, the examples in
Example solid-state light source 3000 in
Solid-state light source 3100 in
The following examples in
Example solid-state light source 3200 in
Another embodiment of this invention is a solid-state light source that includes at least one stack of elements. The elements include a substrate-free LED chip and at least two wavelength conversion chips. The LED chip has a first side and an opposing second side. The first wavelength conversion chip is bonded to the first side of the LED chip. The second wavelength conversion chip is bonded to the second side of the LED chip so that the stack of elements includes an LED chip interposed between the two wavelength conversion chips. Each wavelength conversion chip may optionally include an electrical interconnection means. The substrate-free LED chip is defined in this specification as an LED chip that includes neither a growth substrate nor a transfer substrate. In addition, a substrate-free LED chip has at least one thick semiconductor layer (either the first doped layer or the second doped layer or both the first and second doped layers) that is at least 10 microns thick, preferably at least 15 microns thick, more preferably at least 20 microns thick and most preferably at least 25 microns thick. The multilayer semiconductor structure of the substrate-free LED is at least 10 microns thick, preferably at least 20 microns thick and more preferably at least 30 microns thick. The thick layer or layers can be grown by any standard technique, but preferably the one or more thick semiconductor layers are grown by HVPE.
Examples of solid-state light sources are illustrated in
Solid-state light source 3500 illustrated in a side cross-sectional view in
Examples of solid-state light sources that have transparent electrodes are illustrated in
Solid-state light sources may also be constructed with two or more LEDs bonded between two wavelength conversion chips. Solid-state light source 3570 having two LED chips is illustrated in a side cross-sectional view in
Arrays of solid-state light sources that include stacks of elements can be electrically connected in series, parallel or anti-parallel configurations. Anti-parallel configurations are desirable if the current source is an alternating current (AC) source. Arrays of three or more solid-state light sources can also be connected in combinations of series, parallel or anti-parallel configurations. For example, with an array of four light sources, a first pair of sources can be connected in series, a second pair can also be connected in series and the two pairs can be connected in parallel.
Another embodiment of this invention is a solid-state light source that includes at least one stack of elements enclosed in a sealed transparent envelope. Each stack includes at least one LED chip and at least one wavelength conversion chip. The stack is substantially cooled by convection utilizing direct contact with an optically transparent fluid. The fluid can be either a gas or a liquid and can be either a single chemical element or compound or can be a mixture of chemical elements or compounds.
In solid-state light source 4100, light is emitted by LED chip 1200 of one of the stacks 3500, is converted by a wavelength conversion chip 1800 in the same stack, and exits solid-state light source 4100 through transparent envelope 4110. For example, internally generated light ray 4130 of a first wavelength range is emitted by LED chip 1200 in the left stack, is converted to light ray 4132 of a second wavelength range by a wavelength conversion chip 1800 in the left stack and exits solid-state light source 4100 through transparent envelope 4110.
Another embodiment of this invention is a solid-state light source that includes at least one stack of elements, where the elements in the stack include at least two inorganic LED chips. Preferably, at least one of the inorganic LED chips is substantially transparent to light emitted by the second LED chip. Optionally the stack can also include at least one wavelength conversion chip.
Another embodiment of this invention is a method for fabricating a solid-state light source that is a stack of elements, where the elements include a light emitting diode chip and a wavelength conversion chip.
One initial step is to provide a wafer of unsegmented wavelength conversion chips. A side cross-sectional view of wafer 4302 of unsegmented wavelength conversion chips is illustrated in
Another step is to provide a wafer of unsegmented light emitting diode chips that includes either a growth substrate or a transfer substrate.
The wafer 4302 of unsegmented wavelength conversion chips and the wafer 4304 of unsegmented LED chips are bonded together by transparent bonding layer 2302. This is illustrated in
Optionally, the growth substrate 106 or the transfer substrate (not shown) is removed by standard processes that include, but are not limited to, laser liftoff, chemical processes or mechanical polishing.
Finally, the bonded wafers are segmented into a plurality of solid-state light sources. The dotted lines in
The use of flexible freestanding foils 99 offers a novel substrate for semiconductor devices compared to any of this work due to elimination of polishing and dicing defect and the flexible nature of the freestanding foils 99 themselves. Any useful semiconductor device 101 relies on compositional differences to operate. As such while homoepitaxy is preferable significant stress are created even within a particular family of materials. InGaN alloys on GaN have inherently lower stresses during growth than sapphire or SiC but stresses still do exist. It is disclosed that the flexible nature of freestanding foils 99 offer novel growth conditions relative to all other substrates. The exact mechanism for this difference is not totally understood but the effect has translated in higher quality crystal growth of nitride alloys on freestanding nitride foils. The ability to direct heat the freestanding foils 99 as disclosed previously enables epitaxial growth conditions uniquely different than other growth substrates including thick GaN substrates. This coupled with the novel attributes of freestanding foils 99 to be processed on both sides, eliminate the need for wafer thinning, ease of micromachining, and through hole processing are disclosed in this invention.
For solar applications, rib emitter 105 may consist of but not limited to multiple nitride based quantum wells, hybrid solar cell designs containing nitrides, silicon, GaAs, GaP and other compound semiconductors. A preferred rib emitter 105 for solar emitters consists of a silicon solar cell grown on nitride solar cell in which the longer wavelength of the solar spectrum are absorbed and converted into electrical energy and the shorter wavelengths are converted by at least one nitride solar cell. Using this approach eliminates the issues associated with growing higher temperature nitride layers on lower temperature silicon devices. The high temperature of nitride growth (up to 1050 C) tends to degrade silicon devices due to diffusional effects. By growing the nitride solar cells on the freestanding nitride foils first this problem is eliminated.
As an example of the problem with conventional reactors, a typical InGaN quantum well will be approximately 30 angstroms thick with an indium concentration of between 5 and 30% for visible LEDs. In existing reactors, during device growth the gallium source is constantly being delivered to the reactor with the indium, aluminum, silicon, and magnesium sources being turned on or off depending on the layer being grown. As such from a process gas standpoint, the rate at which indium or the other sources reach the desired concentration and the rate at which indium or these other sources are purged from the reactor determine how sharply the boundaries of the various layers are defined. Alternately, the growth temperature is also typically ramped up and down to change the composition because the indium composition in particular is strongly dependent on growth temperature. In general, the limiting factor on layer definition is how quickly the growth conditions at the substrate can be changed both from a process gases standpoint and a temperature standpoint. More specifically, a low pressure reactor (e.g. 100 to 400 torr) operates around 1000 C for pure GaN and less than 600 C for high indium InGaN. As such over 400 C temperature differences are desirable between layers. The transient thermal time constant of the reactor is a function of the rate or level of heating provided (typically resistive heating in commercially available reactors), the gas flow rate over the platen, the thermal mass of the platen, the thermal impedance between the platen and substrate, and the thermal mass of the substrate. The thermal time constant can be estimated based on t=hA/C if h can be determined within the reactor.
Thermal time constant r=hA/C
C=thermal capacitance=specific heat×mass
h=heat transfer coefficient
A=surface area exposed to fluid
Solving for h within this reactor
h=0.00132 W/cm2
h will be close to the same for freestanding foils but the lower mass will greatly reduce the thermal time constant
Thermal time constant (PlyGaN)=0.294
There is almost 100× difference in the thermal time constant of the graphite susceptor and the freestanding foils as shown in
Based on this data it is easy to understand why very low growth rates are required in the conventional reactors. This calculation does not include the contribution of the sapphire wafer or the thermal impedance between the sapphire wafer and the susceptor. In order to drop the temperature by 200C more than 2 minutes are required in this example. This requires that the growth rate be on the order of a few Angstroms per minute to prevent washing out the 30 Angstrom well which is standard in the industry.
From the process gas standpoint, TMI vapor or the other sources must be introduced and purged from the reactor to also create well defined layers. It is also easy to calculate the residence time in the associated gas lines and within the reactor itself. These are on the order of several seconds to minutes as well. For this reason a preferred embodiment of this invention is a rapid thermal MOCVD micro reactor with a process gas residency time with the reaction chamber of less than 1 second. Assuming a total growth thickness of less than 3000 Angstroms and growth cycle of 30 minutes, an average growth rate of greater than 100 Angstroms/minute or 0.6 microns/hour is required. A reactor, heating means, control system, and a low thermal mass substrate with a thermal and gas residence time less than 10 seconds is a preferred embodiment of this invention. Even more preferred is a reactor, heating means, control system, and low thermal mass substrate with a thermal and gas residency time less than 1 second.
While some argue that material quality is the determining factor for low growth rates in MQW structures, it is obvious from the example above that reactor and substrate transient response time is the limiting factor to growth rates in conventional reactors. In addition the freestanding nitride foils used in this invention were grown at 30 microns/hour growth rates and have superior material properties based on XRD, optical, and other characterization methods to the material typical grown in MQWs in conventional MOCVD reactors. It should be restated that the quality of epitaxially grown layers on freestanding nitride foils versus sapphire, SiC and even bulk GaN is higher based on what appears to be lower stresses during growth.
It should also be pointed out that the desire to increase throughput via more wafers per run or larger wafers per run by making larger susceptor platens and larger reactor chambers only compound these problems. A typical commercially available MOCVD reactor has 45-2 inch wafers per run with a run time of 8 hours or approximately 1 cm2 of devices/minute. As such a continuous or sequential reactor designed to process even 1 cm2 freestanding foils every second would revolutionize the industry. This invention offers a 100× throughput increase over existing commercial MOCVD reactor designs.
Based on this understanding
Additional processing chambers maybe added to increase the number of layers deposited on freestanding foil 138. In this embodiment dual sided or single sided deposition can occur. The addition of sacrificial coatings on one side of the freestanding foil 138 to inhibit growth or enhance heating is also an embodiment of this invention. As an example, freestanding foils 138 suspended from moly wires and heated using halogen lamps have been epitaxially coated on both sides with greater than 45% Indium concentration InGaN with growth rates greater than 200 Angstroms/minute. HRXRD data shows very high crystal quality and uniformity. The number of chambers used will be determined by the number of layers.
While the invention has been described in conjunction with specific embodiments and examples, it is evident to those skilled in the art that many alternatives, modifications and variations will be apparent in light of the foregoing description. Accordingly, the invention is intended to embrace all such alternatives, modifications and variations as fall within the spirit and scope of the appended claims.
Claims
1. A solid-state light source comprising a stack of elements, said elements comprising:
- at least two substrate-free inorganic light emitting diode chips; and
- a first wavelength conversion chip,
- wherein each of said at least two substrate-free inorganic light emitting diode chips has a multilayer semiconductor structure that includes a first doped layer, a first electrode in electrical contact with said first doped layer, a second doped layer, a second electrode in electrical contact with said second doped layer, and an active region interposed between said first doped layer and said second doped layer, wherein said active region emits internally generated light in a first wavelength range and wherein said multilayer semiconductor structure is at least 10 microns thick; and
- wherein said first wavelength conversion chip is attached to said at least two substrate-free inorganic light emitting diode chips by a transparent bonding layer, wherein said first wavelength conversion chip converts at least a portion of said light of a first wavelength range into light of a second wavelength range, said second wavelength range being different from said first wavelength range, and, wherein said first wavelength conversion chip cools said at least two substrate-free inorganic light emitting diode chips.
2. A CVD reactor for fabricating freestanding nitride foils containing at least one of the following CVD processes, HVPE, MOCVD, MBE, ALD, EBCVD, and PECVD.
3. A CVD reactor for fabricating freestanding nitride foils as in claim 2 wherein freestanding nitride foils are directly heated via at least one of the following methods, laser heating, gas heating, RF heating, IR/Visible/UV lamp heating, induction heating, and direct current heating.
4. A naturally convection-cooled distributed light source comprising
- a distributed array of small LED light source chips operating at less than 350 ma/mm2 of said small LED light source chips surface area,
- a thermally conductive luminescent element, said distributed array of small LED light source chips thermally bonded to said thermally conductive luminescent element with a surface area at least 10× times larger than the combined area of said distributed array of small LED light sources and a thermal conductivity greater than 1 W/m/K.
5. A substrate-free inorganic light emitting diode chip comprising:
- a multilayer semiconductor structure that includes a first doped layer, a second doped layer and an active region interposed between said first doped layer and said second doped layer;
- a first electrode in electrical contact with said first doped layer, said first electrode being processed at greater than 400 degrees C.; and
- a second electrode in electrical contact with said second doped layer, said second electrode being processed at greater than 400 degrees C.;
- wherein said first doped layer or said second doped layer is at least 10 microns thick; and
- wherein said active region emits internally generated light in a first wavelength range when an electrical current is applied between said first electrode and said second electrode.
6. A substrate-free inorganic light emitting diode chip comprising:
- a multilayer semiconductor structure that includes a first doped layer, a second doped layer and an active region interposed between said first doped layer and said second doped layer;
- a first electrode in electrical contact with said first doped layer; and
- a second electrode in electrical contact with said second doped layer;
- wherein said first doped layer or said second doped layer is at least 10 microns thick;
- wherein said active region emits internally generated light in a first wavelength range when an electrical current is applied between said first electrode and said second electrode; and
- wherein the thermal extraction direction is substantially orthogonal to the majority of the optical emission direction.
7. An inorganic self cooled solid state emitter comprising:
- a multilayer semiconductor structure that includes a first doped layer, a second doped layer and an active region interposed between said first doped layer and said second doped layer;
- a first electrode in electrical contact with said first doped layer; and
- a second electrode in electrical contact with said second doped layer;
- wherein said first doped layer or said second doped layer is at least 10 microns thick;
- wherein said active region emits internally generated light in a first wavelength range when an electrical current is applied between said first electrode and said second electrode; and
- wherein said inorganic self cooled solid state emitter exhibits greater than 30 lumens/gram and an output greater than 30 lumens.
8. An inorganic self cooled solid state emitter as in claim 7 wherein said inorganic self cooled solid state emitter exhibits greater than 70 lumens/gram and an output greater than 50 lumens
9. An inorganic self cooled solid state emitter as in claim 7 wherein said inorganic self cooled solid state emitter exhibits greater than 100 lumens/gram and an output greater than 50 lumens.
10. A process for growing nitride LEDs on freestanding nitride foils in less than 1 hour.
11. A stack of elements comprising;
- at least one LED stack, and
- at least one superstrate with an area larger than the area of said at least one LED stack, said at least one superstrate containing an electrical interconnect formed on said at least one superstrate wherein said electrical interconnect provides power to said at least one LED stack and provides cooling to ambient for said at least one LED stack substantially via said at least one superstrate,
- wherein at least a portion of the light emitted by said at least one LED stack is transmitted through said at least one superstrate.
12. The stack of elements of claim 11 wherein said superstrate is Al2O3.
13. The stack of elements of claim 11 wherein said at least one LED stack comprises at least one LED die and at least one wavelength conversion element.
14. The stack of elements of claim 11 wherein said electrical interconnect is a high temperature fusible material.
15. The stack of elements of claim 11 wherein said electrical interconnect is a transparent conductive oxide.
16. The stack of elements of claim 11 wherein multiple said at least one LED stacks are arranged to form a substantially linear array.
17. The stack of elements of claim 11 wherein multiple said at least one LED stacks are arranged to form a two-dimensional array.
18. The stack of elements of claim 11 wherein said electrical interconnect connects multiple LED stacks together using at least one of the following electrical connection means; series, parallel or anti-parallel configurations or in some combination of series, parallel or anti-parallel configurations.
19. A solid-state light source comprising;
- a stack of LED elements sealed within a transparent enclosure, said transparent enclosure containing a fluid or gas which transfers the heat generated by said stack of LED elements to the transparent enclosure and from the transparent enclosure to the surrounding ambient,
- wherein said stack of LED elements contains at least one LED die and at least one wavelength conversion chip.
20. A stack of elements comprising;
- at least one LED die,
- a wavelength conversion element containing at least one electrical interconnect wherein said at least one LED die is electrically, optically and thermally bonded to said wavelength conversion element.
21. The stack of elements of claim 20 wherein said wavelength conversion element is a ceramic.
22. The stack of elements of claim 20 wherein said wavelength conversion element is Al2O3.
Type: Application
Filed: Sep 27, 2010
Publication Date: Mar 10, 2022
Inventors: Karl W. Beeson (Princeton, NJ), Scott M. Zimmerman (Basking Ridge, NJ), William R. Livesay (San Diego, CA)
Application Number: 12/924,479
