Light emitting diodes (LEDs) are widely accepted as light sources in many applications that require low power consumption, small size, and high reliability. Energy-efficient diodes that emit light in the yellow-green to red regions of the visible spectrum contain active layers formed of an AlGaInP alloy.
GaAs is preferred as a growth substrate because it is lattice matched to (AlxGa1-x)yIn1-yP at compositions favored for the formation of LEDs that emit light in the yellow-green to red regions of the visible spectrum, at y˜0.5. Ge is an alternative lattice-matched substrate. Since typical growth substrates are absorbing, they are often removed and replaced by a transparent substrate, as illustrated in
Transparent substrate 18 and window layer 16, also a transparent semiconductor, spread current laterally in the device and increase the side light emission. Current spreading is particularly important on the p-side of the active region, due to the low mobility of holes in AlGaInP layers. The use of thick semiconductor layers has several disadvantages over other approaches, however, due to the tradeoff between light absorption and electrical and thermal resistivity common in semiconductor materials.
An alternative to a TS AlGaInP device structure is a thin film structure where the semiconductor-semiconductor bonding is eliminated. Instead, a partially processed wafer is bonded to a handle substrate, typically Si, Ge or a metal substrate. After bonding to the handle substrate, the growth substrate is removed and the wafer processing is completed. Such devices often include an absorbing n-contact layer, such as GaAs, and a vertical injection structure where the n- and p-contacts are formed on opposite sides of the semiconductor structure, as in
In accordance with embodiments of the invention, an AlGaInP light emitting device is formed as a thin, flip chip device. The device includes a semiconductor structure comprising an AlGaInP light emitting layer disposed between an n-type region and a p-type region. N- and p-contacts electrically connected to the n- and p-type regions are both formed on the same side of the semiconductor structure. In some embodiments, the device includes a thick n-layer, to distribute current laterally, and a thinner p-layer, to conduct current mostly vertically. The semiconductor structure is connected to the mount via the contacts. The growth substrate is removed from the semiconductor structure and the thick transparent substrate described above is omitted, such that the total thickness of semiconductor layers in the device may be less than 15 μm in some embodiments, less than 10 μm in some embodiments. The top side of the semiconductor structure may be textured, roughened, or patterned.
In order to minimize the contact resistance on the p-side of the light emitting layer, the semiconductor structure may include a p-type contact layer disposed between the p-type region and the p-contact. The interface between the p-type contact layer and the p-contact may be configured such that when the device is forward biased, carriers tunnel through the interface. As a result, the contact need not be annealed, which may improve the reflectivity of the contact, as annealing generally causes alloying between the semiconductor material and the metal contact, which often reduces the reflectivity of the contact. In some embodiments, the p-contact layer is one of GaP, AlGaInP, and InGaP, doped at least in portions to a hole concentration of at least 5×1018 cm−3. The p-contact may be a full sheet of metal, which increases the optical reflectivity, minimizes the electrical contact resistance and decreases the thermal impedance of the device. A tunneling contact may permit the use of a variety of highly reflective metals for the p-contact, such as Ag.
Depending on the context, as used herein, “AlGaInP” may refer in particular to a quaternary alloy of aluminum, indium, gallium, and phosphorus, or in general to any binary, ternary, or quaternary alloy of aluminum, indium, gallium, and phosphorus. Depending on the context, as used herein, “contact” may refer in particular to a metal electrode, or in general to the combination of a semiconductor contact layer, a metal electrode, and any structures disposed between the semiconductor contact layer and the metal electrode.
As described above, AlGaInP devices have conventionally included thick layers, particularly on the p-side of the light emitting region, for current spreading, due to the low mobility of holes in p-type AlGaInP material. Thinner p-type layers have generally not been used due to the difficulty of achieving high hole concentrations in AlGaInP.
In accordance with embodiments of the invention, an AlGaInP light emitting device includes a highly doped, thin p-type contact layer. A reflective layer may be formed on the p-type contact layer such that the device may be configured as a thin film flip chip. N-type III-V layers generally have higher carrier mobility than p-type layers, therefore the thickness of the current distribution layer can be reduced by designing the device so most lateral current distribution takes place in an n-type layer, rather than a p-type layer. In such a device, the characteristics of the n-type layer are selected to provide adequate current distribution, to minimize the series resistance of the device, and to minimize absorption losses.
In some embodiments, multiple etch stop layers are included in the device. Multiple etch stop layers may be separated from each other by GaAs layers, though they need not be. In one example, a first etch stop layer is grown on the GaAs growth substrate, followed by a GaAs layer, followed by a second etch stop layer. The device layers are grown over the second etch stop layer. Any of the etch stop layers described above may be used in a device with multiple etch stop layers. The etch stop layers in a device may each have the properties (such as composition and thickness), though they need not. In a first example, an AlGaAs first etch stop layer is grown over a GaAs substrate, followed by an InGaP second etch stop layer. In a second example, an AlGaAs first etch stop layer is grown over a GaAs substrate, followed by an AlInGaP second etch stop layer.
In one embodiment, an AlGaAs etch stop layer 20 is grown on a GaAs growth substrate 10. An n-type AlGaInP layer, part of n-type region 22, is grown in direct contact with AlGaAs etch stop layer 20.
The device layers, including at least one light emitting layer in a light emitting region sandwiched between an n-type region and a p-type region, are grown over etch stop layer 20, starting with n-type region 22. The thickness and doping concentration of n-type region 22 are selected for low electrical resistance and good current distribution. For example, n-type region 22 may be an AlGaInP layer 4 to 10 μm thick and doped with Te to a concentration of about 1×1018 cm−3. An AlGaInP n-type region 22 is usually lattice-matched to GaAs. At higher dopant concentrations, the same current distribution may be achievable with a thinner layer; however, undesirable free carrier absorption may increase at higher dopant concentrations. N-type region 22 may therefore include a non-uniform doping concentration, such as one or more thick regions doped at 1×1018 cm−3, and one or more thin regions that are doped more heavily, up to, for example, 1×1019 cm−3. These highly doped regions may be doped with Te, Si, S, or other suitable dopants, and the high doping concentration can be achieved either by epitaxial growth, by dopant diffusion, or both.
The composition of n-type region 22 is selected to minimize the step in index of refraction at the interface with the light emitting region, to avoid waveguiding light at that interface. In one example, the composition of n-type region 22 in a device with a light emitting region configured to emit red light is (Al0.40GA0.60)0.5In0.5P, approximately the same as the average composition in the light emitting region.
A light emitting or active region 24 is grown over n-type region 22. Examples of suitable light emitting regions include a single light emitting layer, and a multiple well light emitting region, in which multiple thick or thin light emitting wells are separated by barrier layers. In one example, the light emitting region 26 of a device configured to emit red light includes (Al0.06Ga0.94)0.5In0.5P light emitting layers separated by (Al0.65Ga0.35)0.5In0.5P barriers. The light emitting layers and the barriers may each have a thickness between, for example, 20 and 200 Å. The total thickness of the light emitting region may be, for example, between 500 Å and 3 μm.
A p-type region 26 is grown over light emitting region 24. P-type region 26 is configured to confine carriers in light emitting region 24. In one example, p-type region 26 is (Al0.65Ga0.35)0.5In0.5P and includes an extra thin layer of higher Al composition to help in the confinement of electrons. Since current injection from the p-side of light emitting region 24 is mostly vertical, the thickness of p-type region 26 may be on the order of microns; for example, between 0.5 and 3 μm. The proximity of the light emitting layers of the light emitting region to the p-contact through a thin p-type region 26 may also reduce the thermal impedance of the device.
A p-type contact layer 28 is grown over p-type region 26. P-type contact layer 28 is highly doped and transparent to light emitted by the light emitting region 24. For example, p-tvpe contact layer 28 may be doped to a hole concentration of at least 5×1018 cm−3 in some embodiments, and at least 1×1019 cm−3 in some embodiments. P-type contact layer 28 may have a thickness between 100 Å and 1000 Å. in some embodiments, a reflective layer is formed over p-contact layer 28 to form a non-alloyed contact. Electrical contact between p-type contact layer 28 and the reflective layer is achieved by tunneling of carriers through the surface depletion region of the interface.
In some embodiments, p-type contact layer 28 is highly doped GaP. For example, a GaP contact layer 28 grown by metal organic chemical vapor deposition may be doped with Mg or Zn, activated to a hole concentration of at least 8×1018 cm−3. The GaP layer may be grown at low growth temperature and low growth rate; for example, at growth temperatures approximately 50 to 200° C. below typical Gal) growth temperatures of ˜850° C., and at growth rates of approximately 1% to 10% of typical GaP growth rates of ˜5 μm/hr. A GaP contact grown by molecular beam epitaxy may be doped with C to a concentration of at least 1×1019 cm−3.
As an alternative to incorporating dopants during growth, the p-type contact layer may be grown, then the dopants are diffused into the p-type contact layer from a vapor source after growth, for example by providing a high pressure dopant source in a diffusion furnace or in the growth reactor, as is known in the art. Dopants may be diffused from a vapor source into the entire area of the surface of p-type contact layer 28, or in discrete regions of p-type contact layer 28, for example by masking parts of p-type contact layer 28 with, for example, a dielectric layer, prior to dopant diffusion,
In some embodiments, p-type contact layer 28 is a highly doped GaP or lattice-matched AlGaInP layer. The layer is doped by growing the semiconductor material, then depositing a layer, including a dopant source, over the grown layer, For example, the dopant source layer may be elemental Zn, a AuZn alloy, or a doped dielectric layer. The layer including the dopant source may optionally be capped with a diffusion blocking layer. The structure is annealed such that the dopants diffuse into the semiconductor from the dopant source layer, The diffusion blocking layer and remaining dopant source layer are then stripped off. In one example, 3000 Å to 5000 Å of a AuZn alloy containing 4% Zn is deposited over a GaP layer, followed by a TiW diffusion blocking layer. The structure is heated, then the remaining TiW and AuZn are stripped.
In some embodiments, p-type contact layer 28 is highly doped InGaP or AlGaInP layer that is not lattice-matched to GaAs. The layer may be between 100 Å and 300 Å thick and doped with Mg or Zn to a hole concentration of at least 1×1019 cm−3.
in some embodiments, a non-metal conductive material 31 such as indium tin oxide (ITO) or ZnO is disposed between at least a portion of p-type contact layer 28 and reflective metal 30 as illustrated in
In some embodiments, a combination of small contact regions and a dielectric mirror may be disposed between p-contact layer 28 and reflective metal 30, as illustrated in
Highly doped semiconductor dots 50 in
In some embodiments, portions of p--type contact layer 28 are removed, for example by etching, as illustrated in
Returning to
In some embodiments, to obviate the need to use an underfill between the mount and the LED die to support the die, n- and p-contacts 34 and 30 may be formed in substantially the same plane, and may cover at least 85% of the back surface of the LED structure. The mount has a corresponding layout of anode and cathode contacts substantially in the same plane. The LED die contacts and mount contacts are interconnected together such that virtually the entire surface of the LED die is supported by the contacts and submount. No underfill is necessary. Different methods for LED to submount interconnection can be used, such as ultrasonic or thermosonic metal-to-metal interdiffusion (gold-gold, copper-copper, other ductile metals, or a combination of the above), or soldering with different alloy compositions such as gold-tin, gold-germanium, tin-silver, tin-lead, or other similar alloy systems. Suitable interconnects are described in more detail in US Published Patent Application 20070096130, titled “LED Assembly Having Maximum Metal Support for Laser Lift-Off of Growth Substrate,” and incorporated herein by reference,
After connecting the device to mount 40, growth substrate 10 is removed, for example by an etch that terminates on etch stop layer 20. Etch stop layer 20 may be removed by a dry etch or an etch that terminates on n-type region 22. The exposed surface of n-type region 22 may be textured (i.e. roughened or patterned with, for example, a photonic crystal) to improve light extraction. For example, n-type region 22 may be roughened by dry etching, photochemical etching, or photoelectrochemical etching. Alternatively, etch stop layer 20 may be textured, then the pattern transferred to n-type region 22 by dry etching. In some embodiments, an additional transparent conductive oxide layer is deposited on the textured surface of n-type region 22 to improve current distribution in the device.
The total thickness of the remaining semiconductor material in the finished device may be less than 15 μm in some embodiments, less than 10 μm in some embodiments. In one example, n-type region 22 is 4 to 6 μm thick, light emitting region 24 is 1.5 μm thick, and p-type region 26 is 1.5 μm thick, for a total thickness of 7 to 9 μm.
In some embodiments, all the semiconductor layers (except the light emitting layers) in the finished device, in particular the n-type layer on which the n-contact is formed, have a band gap larger than the band gap of the light emitting layers of the light emitting region, Accordingly, in such embodiments, no semiconductor layers in the device other than the light emitting layers directly absorb the light emitted by the light emitting region.
The embodiments described herein may offer several advantages over conventional TS AlGaInP devices. For example, the extraction efficiency of embodiments of the invention may approach that of thick-window TS AlGaInP devices with shaped sidewalls, but enhanced surface emission from a thin-film device may result in better directionality and higher surface brightness. In addition, the embodiments described herein allow for simpler growth structures, inexpensive fabrication, and potentially better heat extraction from the active region. Growth problems common in conventional TS AlGaInP devices, such as diffusion of p-dopants during the growth of GaP windows by VPE, may be avoided by the embodiments described herein.
The embodiments describe herein may offer several advantages over other thin film devices. For example, in some of the above embodiments, no wafer-level bonding, neither semiconductor-semiconductor wafer-level bonding nor wafer-level bonding to a handle substrate, is required. Wafer-level bonding can damage the semiconductor structure by imparting stress due to thermal expansion mismatch between the bonded structures or layers. Also, a wafer-level bond may be damaged by subsequent processing steps. The above embodiments may also simplify manufacturing because they do not require simulation of a structure bonded to a handle substrate. Further, the above embodiments eliminate problems associated with a vertical injection structure, such as optical occlusion of the light emitting region, fragile wire bonds, and structural interference with close optics.
Having described the invention in detail, those skilled in the art will appreciate that, given the present disclosure, modifications may be made to the invention without departing from the spirit of the inventive concept described herein. Therefore, it is not intended that the scope of the invention be limited to the specific embodiments illustrated and described.
This is a division of U.S. application Ser. No. 11/956,984, filed Dec. 14, 2007, by Rafael I. Aldaz et al., titled “Contact for a Semiconductor Light Emitting Device”, and incorporated herein by reference.
Number | Date | Country | |
---|---|---|---|
Parent | 11956984 | Dec 2007 | US |
Child | 13423625 | US |