Increasing the brightness of III-nitride light emitting devices

Group III nitride light-emitting device with enhanced brightness

The present invention is related to increasing the brightness of III-nitride light-emitting diodes.

Light-emitting diodes ("LEDs") are extremely durable solid-state light sources that can achieve high brightness and can be used in displays, lighting, indicator lights, printers, and other types of optical pickups. Direct band gap semiconductors can emit light by electricity, which is the material choice for manufacturing LEDs. An important category of light-emitting systems is based on compound alloys formed by group III atoms (especially indium, gallium, and aluminum) and nitrogen, which are generally abbreviated as "III-nitride". One of the III-nitride families has (In_xGa_1-x)_yAl_1-yThe general composition of N, where 0<img file="TWI246779B_D0001.tif" />(x, y)<img file="TWI246779B_D0002.tif" />1. III-nitride can emit most of the light in the electromagnetic spectrum from visible light to near-ultraviolet light, including ultraviolet, blue, green, yellow and red light. Improving brightness and other LED optical characteristics is an important technical goal.

A part of a typical prior art LED structure is depicted in FIG. 1. Other known LEDs components (electrodes, window materials, etc.) in this technology are ignored for clarity.

Generally, one or more epitaxial layers are epitaxially formed on the surface of the substrate before forming the light-emitting active area of the LED. These epitaxial layers form a "base layer" capable of n-type conduction. Figure 1 depicts an example of a base layer with a gallium nitride layer under the n-type gallium nitride layer.

The radiation recombination of electrons and holes occurs in the light-emitting active area on the top of the base layer. Although the light-emitting active area can be used with single and double heterostructures and homogenous junctions, it generally exists in the form of at least one quantum well. There is a p-type conduction injection and restriction area above the active area. (Ignored in Figure 1) positive and negative contacts are also provided.

LED still needs to improve its optical performance, including higher LED brightness and higher quantum efficiency.

The present invention is related to the structure of light-emitting elements, especially LEDs in which the III-nitride light-emitting active region is deposited on an n-type conductive base layer. The substrate covered by the base layer is cut at an angle deliberately from a main crystal plane. Except for deliberately used substrates with angular deviations, the thickness of the base layer must be greater than 3.5 microns. In some specific examples of the present invention, the existence of a thick base layer provides sufficient mechanical support for the device, so that the substrate can be completely removed from the light-emitting system, thereby further improving the performance of the device.

Examples are provided to illustrate that a thick base layer is deposited on a pair of sapphire substrates with a c-axis deviation at a certain angle, and the angular deviation ranges from 0.05° to about 0.50. The present invention also has a base layer with a thickness of more than 3.5 microns, preferably in the range of 7 microns to 10 microns. The thickness of the base layer is greater than 3.5 microns, and it grows on a substrate with an angular deviation, which can lead to an amazing improvement in luminescence.

The present invention is related to the thickness and crystal orientation of the epitaxial layer of light-emitting diodes ("LEDs"). In particular, the directionality of the substrate surface and the thickness of the base layer between the substrate and the light-emitting active region can improve the light-emitting characteristics. The base layer is one or more epitaxial layers between the substrate and the active area, which includes several layers close to the substrate, such as a buffer layer or a nucleation layer, and several layers close to the active area, such as a conversion layer. Special examples related to LEDs with double heterostructures and multi-quantum well light-emitting active regions are also included, and the light-emitting active regions are made of indium gallium nitride ( InGaN). These examples are intended for illustration only. The present invention can be applied to specific examples of single-substance junctions, single- and double-heterostructures, and single- or multiple-quantum wells.

According to the present invention, the substrate is cut by a deliberately offset angle to the main crystal plane, and a thick base layer is coated on the angularly deviated substrate. The higher brightness and the improvement of efficiency are two of the improvement items which have been confirmed by the concrete examples of the off-axis growth of multiple thick base layers. One specific example is to cover a thick n-type gallium nitride base layer on an off-axis sapphire substrate.

Substrate directionality and base layer

The substrates used to manufacture LEDs include sapphire, silicon carbide, gallium nitride, gallium arsenide and gallium phosphide. Taking the special case of the off-angle base layer on the sapphire substrate as an example is also included. But sapphire is used as an example to illustrate, not to specifically limit different specific examples.

Sapphire or alpha alumina has a hexagonal structure belonging to space group R3c. This basic structure consists of inserting the hexagonal closest packed surface of oxygen atoms into the aluminum atom surface. Figure 2 depicts the structure of a sapphire single crystal lattice and the faces usually labeled a, c, m, and r. The {0001) plane is referred to as the "c-plane", and the "c-axis" is perpendicular to the c-plane. However, the sapphire and III-nitride crystal lattices are quite different. Therefore, before covering the III-nitride epitaxial layer on the sapphire substrate, a thin III-nitride nucleation layer, also called a buffer layer, needs to be plated first, and then the remaining base layer is grown. The base layer provides carrier transport to the light-emitting layer. The base layer generally contains one or more III-nitride materials (undoped, n-type or p-type).

Let us consider an example of an n-type doped gallium nitride base layer, including doped, lightly doped, undoped and/or non-intentionally doped gallium nitride sublayers. Note that other materials can also be used as the base layer. A graded doped base layer can also be used. The doping concentration of the base layer can be lower in the direction of the substrate and higher in the direction of the active area, but the area close to the active area and the substrate may not necessarily follow this doping trend. Typical n-type base layer plating, MQW active area formation, and p-type layer plating steps are described in several standard references, including "Nitride Semiconductor Blue Laser and Light Emitting Two" by S. Nakamura and SF Chichibu. Introduction to Polar Body" (Taylor and Francis, 2000), S Nakamura's "Indium Gallium Nitride Light Emitting Diode with Quantum Well Structure" in Materials Research Society Symposium Proceedings No. 395" and "Gallium Nitride and Related Materials" by FA Ponce, RD Dupuis, S. Nakamura and JA Edmond (Materials Research Society, 1996), pages 879-887.

The conventional technology for manufacturing LEDs is related to a base layer formed by covering one or more epitaxial layers, and the base layer serves as the conversion from the substrate to the active region. In the previous technology, the LED uses sapphire as the substrate. The base layer is conventionally grown on the sapphire substrate along the c-axis, which is generally called "coaxial" or "in-position" growth. "Coaxial" emphasizes that the sapphire crystal is cut along the c crystal plane (or other main crystal planes) as accurately as possible, and the base layer is grown substantially along the vertical c-axis direction.

In the present invention, the sapphire (or other) substrate on which the base layer is grown is not cut in the direction perpendicular to the c-axis (or other main crystal axis) in FIG. Therefore, according to these specific examples, there is a deviation between the substrate surface on which the base layer grows and the c-plane described in FIG. 2. We use "deviation angle" or "tilt angle" to indicate the angle between the vertical to the growth surface and the c-axis. Therefore, angular deviation, off-axis growth means that the growth of the base layer on the substrate surface has an angular deviation from the main crystal plane.

The direction of the angular deviation relative to the specified crystal axis can also be specified. For the example of growth on a sapphire substrate considered here, it is more convenient to refer to the direction of the c-axis deviating to the m-plane as "m-plane oblique", or the direction deviating the c-axis to the a-plane to refer to Do "a-face is inclined". However, the off-axis growth surface can have any directionality, and it is not limited to the simple m-plane or the simple a-plane is oblique. The off-axis growth of other substrates can be defined in a completely similar manner in terms of the angular deviation of a major crystal axis.

Experiments show that its optical performance improves with the inclination angle greater than about 0.05° and the thickened base layer. It is generally found that growing a thick base layer on a substrate with a deviation angle greater than 0.05° will improve the brightness and other optical properties of the LED, but there is no clear upper limit of the angle deviation. This may be due to the improvement in LED brightness caused by the tilt angle and the strain state of the III-nitride base layer. Strain represents the deviation of an epitaxial layer from the lattice constant inside the crystal. The III-nitride layer growing on the c-plane of the sapphire substrate is generally "compressed" (the transverse lattice constant is smaller than that inside the crystal). However, adding silicon to III-nitride crystals can reduce the degree of compression, and at high doping concentrations, the crystal is "stretched" (the transverse lattice constant is larger than the inside of the crystal). Figure 3 shows the relative light output power of LEDs with strained state "A" and strained state "B" (3a and 3b, respectively). When the deviation angle of 0.3° is a better choice for the metamorphosis "A", the deviation angle of 0.5° is required for the metamorphosis "B" to improve the brightness. Strained state "B" has a higher silicon doping concentration, so it has more tension than strained state "A". A higher silicon doping concentration and or thicker base layer tilt angle &gt;0.5° is expected to be more advantageous. , And the improvement in brightness has been observed when the tilt angle is 1°.

Under a certain compositional condition, the epitaxial growth on a substrate with a mismatched crystal lattice may crack. The limit of cracking is the maximum thickness that the (specially doped) epitaxial layer can grow without obvious cracks, for example, without adversely affecting the performance of the device. There may be a trade-off between doping and thickness. With less doping, the thickness that grows before the epitaxial crack is thicker.

For the base layer formed on the sapphire, the direction from the c-plane to the m-plane, the a-plane and between has been studied. Regarding the measurement accuracy disclosed here, no obvious change in the optical performance due to the change of the tilt direction has been observed. Most of the tilt angles given here are from the c-plane to the m-plane. The tilt angle is less than 0.05° and the coaxial is not much different. Therefore, "coaxial" as used herein refers to tilting from 0 to 0.05° in any direction.

A light-emitting element is manufactured on a substrate, and a base layer containing one or more components grows on the substrate as a transition zone between the substrate and the light-emitting active region. Although other deposition techniques can also be applied and are within the scope of the present invention, "metal organic chemical vapor deposition ("MOCVD") is generally used to generate the secondary layer including the base layer. We describe a single example of the base layer coated on a sapphire substrate , Is for the specificity of our discussion, not intentionally to exclude other substrates such as silicon carbide, gallium nitride, gallium arsenide, gallium phosphide and so on.

FIG. 4 illustrates a cross-section of a part of the LED device according to a specific example of the present invention. The device includes an aluminum indium gallium nitride (AlInGaN) base layer 7 covered on an off-axis substrate with a thickness greater than about 3.5 microns. The first layer or region of the base layer is generally a buffer layer or a nucleation layer (not shown). The last layer or area of the base layer can be a conversion layer (not shown), which can provide a conversion between the previously grown base layer and the active area 8. The light-emitting active area is covered on the base layer. The active area can be a homojunction, a single or double heterostructure, or a single or multiple quantum well structure. An aluminum indium gallium nitride (AlInGaN) confinement layer covers the active area. The aluminum indium gallium nitride layer can have any composition including gallium nitride (GaN), aluminum gallium nitride (AlGaN) and indium gallium nitride (InGaN), and can be n-type, p-type Type, undoped or have a graded doping profile. The two layers of aluminum indium gallium nitride can have different compositions from each other. The aluminum indium gallium nitride base layer can be doped gradually, generally decreasing toward the substrate and increasing toward the active area, but as mentioned earlier, other regions or layers close to the substrate or the active area do not necessarily follow this doping trend. . The aluminum indium gallium nitride base layer can be composed of several sub-layers, including a sub-layer closer to the substrate and a sub-layer closer to the active area, so that the sub-layer closer to the active area is higher than the sub-layer closer to the substrate The doping concentration. The sub-layers closer to the active area can have higher n-type or p-type doping than the sub-layers closer to the substrate. Examples of sub-layer doping include: a sub-layer closer to the active area has a higher n-type doping concentration than an n-type sub-layer closer to the substrate; a sub-layer closer to the active area has a higher n-type doping concentration; The p-type sublayer closer to the substrate has a higher p-type doping concentration; a sublayer closer to the active area has a higher p-type doping concentration than an n-type sublayer closer to the substrate; and A sub-layer closer to the active area has a higher n-type doping concentration than a p-type sub-layer closer to the substrate. The p-type base layer between the substrate and the active region can occur, for example, in tunnel junction devices and n-up devices, and the more conventional p-up (p-up) devices up) Compared to the element, there is an electric field of opposite polarity. All these elements are included in the scope of the present invention.

FIG. 5 is a specific example of the present invention, showing a cross-section of a part of the LED device, showing a sapphire substrate 1 and a base layer 3 between the substrate and the light-emitting active region 5. The n-type gallium nitride 2 base layer includes a sub-layer of the base layer 3, which is deposited on a buffer layer (not shown) on the sapphire substrate 1 with a deviation angle from the main crystal plane. The nitride buffer layer deposited on sapphire (on-axis or off-axis) is considered to be imprecise epitaxial due to lattice mismatch. More precisely, the initial stage of nitride growth on sapphire seems to start from the solid phase crystallization of gallium nitride in the amorphous phase on sapphire. Then the base layer is on the buffer layer. In a specific example mentioned below, the base layer is n-type gallium nitride. If no special precautions are made, the deposited gallium nitride tends to become n-type conduction. That is to say, when depositing long gallium nitride, dopants are not specially added, and n-type materials are generally formed. This "unintentional" n-type doping may result from the incorporation of n-type impurities (such as silicon and oxygen) in the background gas into the gallium nitride. However, n-type doped gallium nitride can also be used as the base layer 2, in which an appropriate amount of dopants are specifically added to the gallium nitride. Especially adding an appropriate amount of dopants can produce LED structures that are easier to control than non-intentional doping and have higher reproducibility. In the example presented here, the sub-layer 2 is a non-intentionally doped n-type.

In fact, the degree of doping has an effect on the thickness of the sub-layer 2 that can grow before reaching the crack limit. We use the term "slightly doped gallium nitride" to indicate an unintentional or low enough doping level (generally less than about 5x10 per cubic centimeter).¹⁸Dopant atoms) gallium nitride layer. Using a slightly doped secondary layer allows the base layer to grow to the desired thickness before reaching the crack limit. When "light doping" is applied to other non-gallium nitride materials used as the secondary layer 2, it also means the doping level that allows the base layer to grow to the desired thickness before reaching the level of cracking. In a specific example, the slightly doped gallium nitride 2 secondary layer is 4.5 microns thick.

In this specific example, a sub-layer of n-doped gallium nitride 4 is grown on lightly doped gallium nitride 2. In this specific example, the n-doped gallium nitride 4 is 2 microns thick. The general doping concentration range of N-gallium nitride 4 is about 10 per cubic centimeter¹⁸-10²⁰Dopant atoms. In this specific example, the n-doped gallium nitride 4 has a doping concentration of about 10 per cubic centimeter.¹⁹The number of dopant atoms. In the example presented here, silicon is used as the dopant, but this is not intended to limit the scope of the present invention. Silicon, germanium, tin, and oxygen are also used as dopant atom selection for doping III-nitride into n-type. P-type dopants include magnesium, zinc, beryllium, carbon and cadmium. Another layer, layers or areas of the base layer 3 may be before the active area 5. This conversion layer or region (not shown) is a part of the base layer 3 and can be lightly doped as a conversion between the previous base layer part and the active region 5.

On the base layer 3 is a light-emitting active area 5. In some specific examples, a multi-quantum well ("MQW") includes several quantum well layers separated by a buffer layer made of a larger bandgap material. For indium gallium nitride quantum wells, typical buffer layers include higher energy gap indium gallium nitride, gallium nitride, aluminum gallium nitride, and aluminum indium gallium nitride. Although the present invention is described in terms of an indium gallium nitride MQW with an n-type gallium nitride buffer on a sapphire substrate, the present invention is not inherently limited to this LED structure.

The p-type conductive layer (described as 6 in FIG. 4) known as the "confinement layer" and the "injection layer" is in the opposite direction of the active region to the n-type base layer. The typical material, size, and dopant concentration of the p-layer is known as a technique and is given in the previously cited reference materials, which may be, for example, 100-1000 angstroms (<img file="TWI246779B_D0003.tif" />) P-type Al_xGa_1-xN (0&lt;x&lt;0.25), followed by 100-3000 angstroms of p-type gallium nitride. A higher doped p-type layer can be formed on the p-gallium nitride to ensure good ohmic contact of the p-electrode.

In the LEDs in the general prior art described in FIG. 1, the base layer is grown on the coaxial substrate 1a, and its total thickness is less than 3.5 microns. As a specific example of the present invention, FIG. 5 shows the base layer 3 growing on the off-axis substrate, and the thickness of the base layer 3 is thicker than that in FIG. 1. In other words, the base layer 3 used in the present invention, which is long on the off-axis substrate, has a thickness greater than about 3.5 microns. The preferred thickness of the base layer 3 growing on the off-axis substrate is from about 6.5 to about 9.5 microns. The improvement in brightness comes from one of the good results produced by using a thick base layer 3 plus off-axis epitaxial growth. To simplify the language, we refer to "thin" and "thick" base layers to indicate that the general range is "thin" if it is lower than 3.5 microns, and "thick" if it is higher than 3.5 microns.

In the example mentioned here, the thickness of the N-gallium nitride layer 4 is maintained at about 2 microns, and the base layer 3 is thickened by thickening the lightly doped gallium nitride layer 2. This is just an example, and the thickening of the base layer described herein can be achieved by thickening any sub-layers or any combination of sub-layers contained in the base layer.

Although the thickness of the long base layer 3 on the off-axis substrate is greater than about 3.5 microns, and in the application of the present invention, the thickness range from about 6.5 microns to about 9.5 microns has satisfactory results, but the off-axis growth is up to about 200 microns Thick epitaxial layers are also feasible. The brightness increases as the thickness of the off-axis growth base layer increases, and there is no upper limit on the thickness in the present invention. However, the increase in the thickness of the entire LED structure tends to increase the complexity of manufacturing, such as yield and device packaging isolation (singulation).

Substrate removal

The main function of the substrate is to provide a platform on which different layers of the entire light-emitting element can be manufactured. Therefore, the substrate provides mechanical strength and stability during manufacturing and operation. However, during the operation of the light-emitting element, the optical characteristics of the substrate may interfere with the luminous efficiency (among other characteristics), thereby hindering the performance of the element. In some cases, the thick base layer used here provides sufficient mechanical stability to allow the substrate to be separated from the rest of the device after the thick base layer is present.

Several examples compare the brightness and other optical characteristics of LEDs with different off-axis tilt angles and different n-type base layer thicknesses. The data relates to the indium gallium nitride MQW LEDs approximately described in Figure 5.

Several batches of LEDs with different main emission wavelengths are manufactured. Figure 6 depicts the relationship between the main wavelengths of LEDs with thin and thick base layers on the brightness of the LED in lumens. In Figure 6, the "thin" base layer is about 3.5 microns thick, and the "thick" base layer is about 6.5 microns thick. It is surprising that the increase in light emission through the combination of thick and off-axis base layers far exceeds the sum of the individual increases caused by individual changes. For example, Figure 6 depicts that the lift from the thin positive axis to the thick positive axis is a line segment 100 at about 510 nanometers. The lift from the thin normal axis to the thin off axis is described as 100+101. The increase obtained from the combination of thick and off-axis base layer is 100+101+102, which obviously exceeds the sum of the individual increases caused by individual changes in thickness and off-axis (100+101+100=flux level 200). Therefore, the combination of the thick base layer and the growth on the off-axis substrate achieves the improvement of luminescence, which is unexpected and significantly exceeds the sum of the improvement brought by individual changes.

Figure 7 describes the relationship between the forward current driving the LED and the relative luminous efficiency. From the point of view of absolute light output (lumens), the data is not corrected. Instead, the LED drive current is compared with the current generated by the individual light detectors used to measure all emitted light. Therefore, the relative change of the emitted light efficiency from data point to data point and curve-to-curve can be seen from FIG. 7.

The data in Figure 7 are given two deviation angles from the c-axis to the m-plane, which are related to the thick base layer (layer 3, about 6.5 microns) approximately described in Figure 5. The upper curve value is measured on a device with a deviation angle of about 0.39° on a substrate and a base layer thickness of about 6.5 microns. The bottom curve value is measured on a component that grows on a coaxial substrate and has a base layer thickness of about 6.5 microns. We see in Fig. 7 that two elements with a thick base layer of similar thickness have a higher efficiency curve maximum for off-axis deposition than for coaxial deposition. In addition, off-axis deposition reaches the highest efficiency peak at a lower current than coaxial deposition, 7.9 milliamperes (mA) versus 12.6 milliamperes (mA).

It is absolutely better to achieve the higher efficiency of off-axis deposition as described in Figure 7, that is, to have brighter LEDs at the same current. But achieving maximum efficiency at lower current values is also a proof of a better LED structure. In a certain aspect, the luminous efficiency is determined by the radiant electron-hole recombination and non-radiative loss mechanism. Non-radiation loss tends to dominate the performance of LEDs at low currents. The higher current tends to saturate the non-radiative losses, resulting in an increase in luminous efficiency at higher currents. Therefore, achieving maximum efficiency at lower currents is evidence of less non-radiative loss mechanisms, indicating fewer defects and a better LED material overall.

Figure 8 depicts four sets of experiments, in which on-axis and off-axis sapphire substrates are loaded in the same reactor, and LEDs are manufactured under the same conditions except for the angular deviation of the substrate. All experiments described in Figure 8 used a thick base layer of about 6.5 microns thick. Therefore, removing the influence of other experiment-to-experimental changes, Figure 8 shows a clear comparison of the effects of tilting and non-tilting of a substrate with a thick base layer in accordance with the present invention. Figure 8 clearly illustrates that the long and thick base layer on the inclined substrate produces a brightness increase.

The high-brightness LEDs produced according to the present invention are particularly suitable for color displays, which use red, green and blue LEDs as pixels. Such a display is well known and described in Figure 9. A display 300 has an array of red, green, and blue LEDs, which are selectively controlled by well-known circuit distribution to display images. For simplicity, only three pixels are displayed in FIG. 9. In a specific example, each main color is arranged in a column. In other specific examples, the main colors are arranged in other shapes, such as triangles. High-brightness LEDs can also be used in the backlight of liquid crystal displays (LCD).

After describing the present invention in detail, those skilled in the art will realize that within the scope of the invention, possible modifications to the present invention will not deviate from the spirit of the inventive concept described here. Therefore, it is not intended to limit the scope of the invention to the specific and preferred specific examples illustrated and described above.

<p>1. . . Substrate (off-axis)</p><p>1a. . . Substrate (coaxial)</p><p>2. . . n-doped GaN secondary layer</p><p>3. . . Base layer of n-doped gallium nitride and n-doped gallium nitride</p><p>4. . . n-doped GaN secondary layer</p><p>5. . . Luminescent active area</p><p>6. . . P-GaN "confinement layer" and P-AlGaN "implantation layer"</p><p>7. . . Al-InGa-based layer</p><p>8. . . Active area</p><p>200. . . Flux hierarchy</p><p>300. . . Display panel</p><p>301. . . Red pixel</p><p>302. . . Green pixel</p><p>303. . . Blue pixel</p>

The graphics here are not made according to the size ratio.

Figure 1: A cross-sectional schematic diagram depicting a part of the prior art LED epitaxial layer structure.

Figure 2: Schematic diagram of a sapphire single crystal lattice.

Figures 3a and 3b: The relative luminous efficiency of LEDs that are offset by 0.3 and 0.5° from the c-plane of the sapphire substrate. The data shows that the LED structure has strain states A(a) and B(b) in the base layer.

Figure 4: A cross-sectional schematic diagram of a part of the LED epitaxial layer structure according to an example of the present invention.

Figure 5: A cross-sectional schematic diagram of a part of the LED epitaxial layer structure according to another example of the present invention.

Figure 6: The brightness changes with the main wavelength of LEDs. The LEDs have a thin (3.5 micron) and thick (6.5 micron) base layer, which is covered on coaxial (angle deviation less than 0.05°) and off-axis (angle deviation range from about 0.20°) To about 0.40°).

Figure 7: For the thick base layer (6.5 micron) LED described in Figure 5, the base layer is covered on the coaxial (0.03°) and off-axis (0.39°), and its relative luminous efficiency varies with the forward current.

Figure 8: Four sets of experiments on the same device structure with a thick base layer (6.5 microns) covered on a substrate with and without angular deviation, and the brightness changes with the main wavelength of LEDs.

Figure 9: Display elements using the high-brightness LEDs of the present invention.