Patent Application
20050062054
1. Field of the Invention
The present invention relates to a semiconductor light emitting device.
2. Description of Related Art
A ZnSe crystal is a direct transition type semiconductor having forbidden bandwidth (band gap energy) of 2.7 eV at a room temperature, expected to have a wide range of applications for a light emitting device in the wavelength range of blue to green. Particularly after 1990 when it was found that a p-type ZnSe film could be formed by doping plasma-excited nitrogen, ZnSe type light emitting devices have attracting attention.
The inventors have devised a white LED (Light Emitting Diode) having a novel structure using a ZnSe substrate for practical use. The white LED utilizes SA (Self-Activated) light emission of an n-type ZnSe substrate. Specific structure of a light emitting device is as shown in
When a current is introduced by conducting these electrodes to cause emission of blue light (having the wavelength of around 485 nm) at active layer 104, part of the blue light is directly emitted to the outside of the device, and another part enters the substrate side. The blue light entering ZnSe substrate 101 excites an SA center in the ZnSe substrate, and as a result, induces SA light emission. The SA light emission has a peak at around 590 nm, and when mixed with a blue light having the wavelength of 485 nm with an appropriate ratio, a light that is perceived as white by human eyes can be obtained. The ZnSe based white LED has a driving voltage as low as about 2.7 V and relatively high light emission efficiency, and therefore, applications thereof are hoped for.
The ZnSe based light emitting device, however, has a problem of short lifetime. The lifetime of the ZnSe based light emitting device will be described in the following. In a semiconductor light emitting device, an active layer that emits light is positioned between an n-type semiconductor cladding layer and a p-type semiconductor cladding layer, and has a band gap that is smaller than the band gap of these two cladding layers. At the time of light emission, electrons and holes are introduced from the n-type cladding layer and from the p-type cladding layer to the active layer to create electron-hole recombination and to cause light emission by the recombination. The electrons introduced from the n-type cladding layer to the active layer mainly follow the courses below.
(1) Re-combined with holes and emit light.
(2) Leak (overflow) to the p-type cladding layer, resulting in recombination without light emission in the p-type cladding layer.
When the ratio of electrons that follow the course (2) is large, light emission efficiency lowers, and therefore, optical output from a light emitting device (LD: Laser Diode, LED) becomes smaller. In order to solve the problem related to the course (2) above, an energy barrier (hetero barrier; ΔEc) against electrons of the p-type cladding layer on the side of the active layer may be increased, so that leakage of electrons is reduced. Specifically, ΔEc is a difference in quasi-Fermi level between energy at the bottom of a conductive band of the p-type cladding layer and electrons in the active layer. Though it is difficult to accurately calculate ΔEc, there are the following three methods of increasing the barrier.
(1) Increasing the difference ΔEg between the band gap of the p-type cladding layer and the band gap of the active layer.
(2) Lowering the Fermi level of the p-type cladding layer by increasing carrier density of the p-type cladding layer.
(3) Lowering current density to be introduced to the active layer.
Among these, method (3) is meaningless in realizing a light emitting device having high intensity. As method (1) above, by way of example, use of a ZnMgSSe layer as the cladding layer in a ZnSe based light emitting device has been proposed (see, for example, Japanese Patent Laying-Open No.5-75217). When ZnMgSSe is used as mentioned above, it becomes possible to increase the band gap to as large as about 4.4 eV, under the condition that the lattice constant thereof is adapted to match that of ZnSe.
In a ZnSe based device, however, it is impossible to apply methods (1) and (2) independent of each other, and when method (1) only is pursued, the problem cannot be solved. Methods (1) and (2) are related with each other because of doping characteristics of the ZnSe based compound semiconductor. The doping characteristics of the ZnSe based semiconductor will be described in the following.
It has been known that, when a p-type impurity is introduced, in an equilibrium state, to a II-VI group compound semiconductor to which the ZnSe based compound semiconductor belongs, sufficient p-type conductivity cannot stably be attained, and that p-type conductivity is attained only when nitrogen is introduced during low-temperature growth by MBE (Molecular Beam Epitaxy) method. This doping, however, becomes more difficult as the band gap becomes wider, and the highest possible p-type carrier density becomes smaller as the band gap becomes wider.
From the phenomenon above, it is understood that there is an optimal band gap value of the p-type cladding layer for maximizing hetero barrier ΔEc. Specifically, there is such a relation between the two as schematically shown in
Though it depends on doping technique, the optimal band gap value mentioned above is around 2.9 eV to around 3.0 eV. There will be no problem if the hetero barrier ΔEc obtained with this optimal band gap is sufficiently large and as a result the leakage of electrons decreases sufficiently. Actually, however, it has been found that even when the optimal band gap in the p-type cladding layer is realized, the hetero barrier ΔEc is not large enough and that considerable amount of electrons leak from the active layer to the p-type cladding layer.
A harder problem of the light emitting device formed of ZnSe based compound semiconductor is that leakage of electrons to the p-type cladding layer not only decreases light emitting efficiency but also reduces the lifetime of the light emitting device. This phenomenon will be described in the following.
As described earlier, in the II-VI group compound semiconductor, to which ZnSe belongs, stability of a p-type dopant is low. Therefore, the p-type carrier density cannot be increased. In addition, donor-related defects are formed by the energy emitted when the electrons that have been leaked to the p-type cladding layer are recombined with the holes in the p-type cladding layer, decreasing the p-type carrier density. When the p-type carrier density decreases, hetero barrier ΔEc reduces, and therefore the function as a barrier against electron leakage is undermined. This results in a vicious circle of (leakage of electrons to p-type cladding layer)→(decrease of p-type carrier density in p-type cladding layer)→(decrease of hetero barrier ΔEc)→ . . . , catastrophically lowering light emission efficiency. Specifically, rapid degradation starts after a short period of operation. Because of the phenomenon described above, it has been considered that the ZnSe based light emitting device inherently has a short lifetime and that it is difficult to make the lifetime longer.
An object of the present invention is to provide a ZnSe based light emitting device that has longer lifetime.
The present invention provides a ZnSe based light emitting device of a II-VI group compound semiconductor formed on a compound semiconductor substrate and having an active layer between an n-type cladding layer and a p-type cladding layer, including a semiconductor barrier layer having a band gap larger than a band gap of the p-type cladding layer, provided between the active layer and the p-type cladding layer.
Because of this structure, electrons introduced to the active layer are prevented from moving to the p-type cladding layer because of the barrier potential attained by the band gap of the barrier layer larger than that of the p-type cladding layer. Consequently, the lifetime of the ZnSe based light emitting device can significantly be improved.
The gist of the present structure resides in that, of the two roles played by a common p-type cladding layer, that is, “supplying holes to the active layer” and “suppressing electron leakage by forming hetero barrier,” the role of “suppressing electron leakage by forming hetero barrier” is assumed by the barrier layer. The p-type cladding layer is responsible only for “supplying holes to the active layer.” A large band gap or large carrier density is not much required of the p-type cladding layer that simply bears the burden of “supplying holes to the active layer.” As to the suppression of electron leakage by the barrier layer, if the band gap of the barrier layer is sufficiently large, it is possible to ensure sufficiently large hetero barrier ΔEc with respect to the quasi-Fermi level of the active layer, even in the absence of increase of the hetero barrier ΔEc attained by increasing the carrier density.
A particular advantage of the above described structure is that efficiency of electron confinement is not much dependent on the carrier density of the cladding layer. Therefore, even when the carrier (hole) density of the p-type cladding layer decreases because of the leakage current coming over the barrier layer, the efficiency of electron confinement is almost free of any influence and maintained as it is. As a result, the accelerated increase of leakage amount caused by the vicious circle suffered by the conventional structure is not induced, and the catastrophic deterioration of the device can be prevented.
Here, the effect of confinement by the barrier layer will be described. Basically, when the difference between the quasi-Fermi level of electrons in the active layer and the energy level at the bottom of the conductive band in the barrier layer is large, the efficiency of confinement improves. In order to increase the energy difference, that is, hetero barrier ΔEc, the most basic approach is to increase the band gap of the barrier layer. When ZnMgSSe is used as the barrier layer, the band gap of the barrier layer may be increased by increasing the composition ratio of Mg and S. Here, the carrier density of the barrier layer is not important, and the conventional limitation on the carrier density of the p-type cladding layer is eliminated.
Though intentional doping of the barrier layer with a p-type impurity is unnecessary, doping to some extent does not cause any problem. The material of the barrier layer is not limited to ZnMgSSe. Any material other than ZnMgSSe may be used provided that it has a larger band gap than the cladding layer and, as a result of the larger band gap, the energy level of the bottom of the conductive band is elevated (or electron affinity lowers). It is necessary, however, that the lattice constant of the material approximately matches that of the semiconductor substrate, for example, a ZnSe substrate. An example of such material is ZnMgBeSe. Compared with ZnMgSSe, ZnMgBeSe is known to attain smaller electron affinity, and therefore, if the band gap is the same, ZnMgBeSe is more preferable as it attains higher efficiency of electron confinement.
When the barrier layer is formed of ZnMgSSe or ZnMgBeSe, the larger the band gap thereof, the higher the efficiency of electron confinement. When the band gap is made too large, however, crystal characteristic of the film serving as the barrier layer tends to deteriorate, and therefore, excessive increase should be avoided. When the band gap of the barrier layer is made too large as compared with the band gap of the p-type cladding layer, it will be a barrier against introduction of holes from the p-type cladding layer to the active layer, undesirably decreasing the light emission efficiency.
Here, in a light emitting device formed of a III-V group compound semiconductor, the aforementioned formation of a barrier against holes introduced from the p-type cladding layer to the active layer does not pose a serious problem. In the ZnSe based compound semiconductor, however, different from the compound semiconductor of the III-V group, when there is the above described barrier between the barrier layer and the p-type cladding layer, p-type doping becomes instable, and deterioration is likely. Therefore, the barrier against introduction of holes from the p-type cladding layer to the active layer should desirably be small. When ZnMgSSe and ZnMgBeSe are compared with respect to the barrier, when the band gap is the same, ZnMgBeSe results in a smaller barrier against holes or the above described barrier is not formed, and therefore, ZnMgBeSe is preferred.
As can be seen from the description above, the band gap of the barrier layer has an optimal value. The optimal value depends on the material of the barrier layer, the band gap of the p-type cladding layer, stability of p-type doping of the p-type cladding layer and so on, and therefore it cannot be determined in a simple manner. It is noted, however, that the optimal value of the barrier layer band gap exists in a range 0.025 eV to 0.5 eV larger than the band gap of the p-type cladding layer. Even when the value is off from the optimal value to some extent, the above described role of the barrier layer can be expected provided that the band gap is in the range of 0.025 eV to 0.5 eV larger than the band gap of the p-type cladding layer.
According to another aspect, the present invention provides a semiconductor light emitting device formed on a compound semiconductor substrate, having an active layer sandwiched between two cladding layers, wherein one of the two cladding layers is a p-type semiconductor to which a p-type impurity is introduced, and the other cladding layer is an undoped semiconductor.
By the above described structure, Fermi level of electrons in the active layer can be decreased. Therefore, it becomes possible to reduce the margin of lowering of a conductive band portion (hereinafter referred to as a conductive band lowering boundary) of the p-type cladding layer that is adjacent to the active layer and has been bent and lowered by the electric field. Accordingly, the barrier against electrons that leak from the active layer to the p-type cladding layer is not much lowered at the conductive band lowering boundary. As a result, leakage of electrons from the active layer to the p-type cladding layer can be suppressed, and the lifetime of the light emitting device can be made longer.
An undoped semiconductor refers to a semiconductor not doped with any dopant, that is, neither with p-type dopant nor with n-type dopant. The concentration of residual dopant in the undoped semiconductor must generally be smaller than the dopant concentration attained by the doping process, no matter whether it is n-type or p-type. By way of example, when doping is performed to prepare a p-type or n-type semiconductor, it is a common practice to set the p-type or n-type impurity concentration to at least 1016/cm3. Therefore, the residual impurity concentration of the undoped semiconductor should be lower than 1016/cm3, no matter whether the impurity is p-type or n-type.
Generally, a semiconductor contains both n-type and p-type impurities, and the conductivity type of the semiconductor is defined by the impurity that is dominant. Impurity concentration of the semiconductor is determined by the amount of impurity remaining after the impurities of both types are offset with each other. The aforementioned impurity concentration of 1016/cm3 is the concentration determined by the amount of remaining impurity after the impurity concentrations of both types are offset with each other, and it represents the impurity concentration of the conductivity type of the semiconductor.
The foregoing and other objects, features, aspects and advantages of the present invention will become more apparent from the following detailed description of the present invention when taken in conjunction with the accompanying drawings.
In the following, the light emitting devices in accordance with the embodiments of the present invention will be described with reference to the figures.
(First Embodiment)
On the n-type cladding layer 3, an active layer 4 is positioned, in which a quantum well layer and a barrier layer thereof are stacked. A barrier layer (first cladding layer) 11 is further positioned thereon, and on the barrier layer, a p-type ZnMgSSe cladding layer (hereinafter referred to as a p-type cladding layer or a second cladding layer) 5 is formed.
As the barrier layer, an i-type Zn1-x-yMgxBeySe (0.01≦y≦0.1) may be used, as shown in
On the p-type cladding layer, a p-type ZnSe buffer layer 6 is positioned, a p-type ZnSe/ZnTe superlattice contact layer 7 is formed thereon, and a p-electrode 9 is further provided thereon. Though an n-electrode is formed on n-type compound semiconductor substrate 1, it is not shown. A voltage is applied between the n- and p-electrodes to introduce current to the active layer, so as to cause light emission.
Next, a method of manufacturing the light emitting device in accordance with the present embodiment will be described. First, by the MBE method, the stacked structure shown in
As to the measurement of the band gap of each layer at the room temperature, utilizing light emission wavelength of PL (Photo-Luminescent) light emission (light emission caused by recombination of excitons) near the band end at 4.2 K, the band gap at a room temperature was calculated using the following equation (I).
Eg(eV)={1240/λ4.2PL(nm)}−0.1 (I)
In the equation (I), 0.1 eV is subtracted, which corresponds to the decrease in the band gap caused by the temperature increase from 4.2K to the room temperature. Though the equation above is not necessarily accurate but involves a systematic error, the equation is simple and therefore, it is adopted as an approximate expression.
Matching of lattice constant can be evaluated based on a deviation in diffraction angle of (400) diffraction by X-ray. After the stacked structure of layers 2˜8 is formed by the MBE method, when the diffraction line is measured, a strong diffraction from the ZnSe substrate and a relatively weak diffraction from the cladding layer can be observed. From the difference in the diffraction angle between these two, degree of lattice matching of the cladding layer can be evaluated. It is noted, however, that the diffraction peak of the barrier layer cannot be observed. Therefore, in a growth for setting condition made beforehand, a relatively thick ZnMgBeSe or a ZnMgSSe film is formed on the ZnSe substrate, and lattice matching is evaluated by measuring the X-ray diffraction angle. Similar measurement is taken for the band gap.
Though Cl is used as the n-type impurity and N is used as the p-type impurity, selection of these is not an essential factor of the present invention, and the impurities used are not limited to those mentioned above.
Though ZnMgSSe is used for the n-type and p-type cladding layers and the band gaps of these are of the same value, such selection is not an essential factor of the present invention, either. Different ZnSe based compound semiconductors may be used as the n-type and p-type cladding layers, and the band gaps of these may be different from each other.
The LED of the present invention having the stacked structure as shown in
After respective layers of the stacked structure shown in
Before formation of electrodes, (400) diffraction of X-ray (Kα1 line of Cu) was measured, and it was confirmed that the diffraction peaks of the n-type and p-type cladding layers have the deviation of at most 400 seconds from the diffraction peak of the ZnSe substrate. As to ZnMgBeSe, in the growth for setting condition performed immediately before LED growth, (400) diffraction of X-ray (Kα1 line of Cu) was measured, and it was also confirmed that the deviation was at most 400 seconds as compared with the diffraction peak of the ZnSe substrate.
Lifetimes of the LEDs as an example of the present invention and as a comparative example fabricated in accordance with the process steps described above were measured. As a method of measurement, a constant current of 15 mA was caused to flow at 70° C. and change in luminance was measured.
The degradation is suppressed as described above, because degradation of the p-type cladding layer is prevented as leakage of electrons to the side of the p-type cladding layer is suppressed, and because the efficiency of confinement is maintained even after the p-type cladding layer is deteriorated. Practical application of the conventional ZnSe based light emitting device has been hindered as the device is prone to degradation and has short lifetime. Because of the structure described above, the light emitting device of the present invention overcomes these defects.
(Second Embodiment)
The LED shown in
The n-type cladding layer 3 and the p-type cladding layer both had the thickness of about 0.5 μm, barrier layer 11 had the thickness of about 0.02 μm, and trap layer 12 had the thickness of about 0.05 μm. As to the impurity concentration, the p-type impurity concentration of the p-type cladding layer was 3×1016/cm3, and the p-type impurity concentration of the trap layer was 3×1017/cm3. On the LED, an n-electrode and a p-electrode, not shown, are provided. On a back surface 1a of n-type ZnSe substrate 1, an n-electrode formed of Ti/Au film is provided, and on an upper surface 7a of contact layer 6,7, a p-electrode formed of a semitransparent Au film having the thickness of about 10 nm is provided. A unit area of 400 μm×400 μm of the LED described above was formed on an n-type ZnSe substrate and thereafter scribe-broken to the unit area of 400 μm×400 μm to obtain a piece. The thus provided piece as a chip was bonded on a stem to fabricate an LED (example of the present invention) for evaluating the lifetime thereof.
For comparison, an LED as a comparative example was fabricated, which had the same stacked structure as the LED described above except that the barrier layer and the trap layer were not provided. Specifically, the LED having the stacked structure shown in
The LEDs as the example of the present invention and as the comparative example were tested under the following conditions. A constant current of 15 mA was caused to flow through the LEDs at 70° C. and decrease in luminance over time was measured. The test result was as follows. In the LED as the comparative example, it took 200 to more than 500 hours until the luminance was decreased to 70.% of the initial luminance, and the average time was about 350 hours. In contrast, in the LED of the present invention, it took 350 to more than 700 hours until the luminance was decreased to 70% of the initial luminance, and the average was about 500 hours.
From the test result above, it was found that the LED in accordance with the embodiment of the present invention enables about 40% longer lifetime as compared with the prior art.
In an LED shown in
(Third Embodiment)
The barrier layer may be formed of ZnMgSSe, and
The LED as an example of the present invention shown in
The n-type cladding layer 3 and the p-type cladding layer both had the thickness of about 0.5 μm, and barrier layer 11 had the thickness of about 0.02 μm. As to the impurity concentration, the p-type impurity concentration of the p-type cladding layer was 3×1016/cm3. On the LED, an n-electrode and a p-electrode, not shown, are provided. On a back surface 1a of n-type ZnSe substrate 1, an n-electrode formed of Ti/Au film is provided, and on an upper surface 7a of contact layer 6,7, a p-electrode formed of a semitransparent Au film having the thickness of about 10 nm is provided. A unit area of 400 μm×400 μm of the LED described above was formed on an n-type ZnSe substrate and thereafter scribe-broken to the unit area of 400 μm×400 μm to obtain a piece. The thus provided piece as a chip was bonded on a stem to fabricate an LED (example of the present invention) for evaluating the lifetime thereof. For comparison, an LED as a comparative example having the stacked structure shown in
The LEDs as the example of the present invention and as the comparative example were tested under the following conditions. A constant current of 15 mA was caused to flow through the LEDs at 70° C. and decrease in luminance over time was measured. The test result was as shown in
From the test result above, it was found that the LED in accordance with the embodiment of the present invention enables about 40% longer lifetime as compared with the prior art.
(Fourth Embodiment)
Referring to
The impurity concentration of undoped ZnMgSSe layer 3 is controlled to a level lower than a typical level attained by doping an impurity, regardless of p-type or n-type. The impurity concentration is lower than 1016/cm3.
Next, the function of semiconductor light emitting device 10 shown in
In
The band of the conventional light emitting device as a comparative example will be described first. As can be seen from
ΔEc=Eg(p-cladding layer)−V−Ef(p-cladding layer) (1)
As described above, possible approach to increase ΔEc is to increase the band gap energy (Eg(p-cladding layer)) of the p-type cladding layer, to decrease the value V by decreasing the introduced current, or to increase p-type carrier density to reduce Ef(p-cladding layer). It is noted here that the absolute position of the quasi Fermi level (φn, φp) in the active layer does not have any influence on ΔEc. Even if the position of φn were lowered while maintaining the value V constant, ΔEc would be unchanged as the bottom position of the conductive band of the p-type cladding layer would also be lowered, drawn by the lowering of φp, as schematically shown in
The situation is different, however, when a material has instable p-type doping and is prone to deterioration, such as in the case of ZnSe. Specifically, though confinement of carriers in the active layer should be discussed in relation to ΔEc, not only ΔEc but also ΔEc′ (see
From the foregoing, from the viewpoint of suppressing leakage of electrons, the larger value ΔEc′ is preferred. In order to increase ΔEc′, possible approach is to lower the Fermi level of electrons φn to reduce bending of the band of the p-type cladding layer, as can be understood from a comparison between
The next problem is how to decrease φn. It is a natural prerequisite that the value V mentioned above is kept constant. It is generally difficult to increase the p-type carrier density in a compound semiconductor used as a material of the cladding layer in a common light emitting device. Therefore, when an n-type cladding layer and a p-type cladding layer are compared, the carrier density of the n-type cladding layer tends to be higher than the carrier density of the p-type cladding layer. The Fermi levels φn and φp of the electrons and holes in the active layer are determined by the amounts of electrons and holes introduced from the cladding layer, and therefore, if the carrier density of the n-electrode side cladding layer is high, introduction of electrons is facilitated, and as a result, the level φn becomes higher.
In view of the foregoing, an approach has been found in which the p-type cladding layer is doped with a p-type impurity in the conventional manner so that it comes to have p-type conductivity, while the n-electrode side cladding layer is undoped. By this approach, introduction of electrons to the active layer is hindered, that is, electrons are less accumulated, and the Fermi level φn in the active layer lowers. There was a concern that if the n-electrode side cladding layer were undoped and made to have high electric resistance, current would not flow through the light emitting device. It was found by actual experimental prototype that electrons diffused from the n-type buffer layer 2 to n-electrode side cladding layer 3 and the current flew.
Another concern was that as the barrier against holes of the n-electrode side cladding layer 3 became lower, leakage of holes to the n-electrode side cladding layer would be more likely. In the compound semiconductor material used for the light emitting device, however, mobility of holes is far smaller than that of electrons, and therefore, leakage of holes is inherently small. Thus, that concern proved unfounded. Even if such a problem were not negligible, it could be readily solved by simply increasing the band gap of n-electrode side cladding layer 3.
It is difficult to confirm by direct measurement that by the undoped n-electrode side cladding layer, Fermi level φn is decreased and as a result ΔEc′ is increased. It is possible, however, to confirm that the effect has been successfully attained or not, by evaluating the lifetime of the device.
The LED as an example of the present invention having the structure shown in
(Fifth Embodiment)
In the LED having the above described structure, confinement of electrons is not governed by ΔEc but determined by ΔEc″. Therefore, it is expected that decrease in φn has a direct effect on suppressing the leakage of electrons.
An LED having such a structure as shown in
As described above, the main structural characteristic of the present invention is that the n-electrode side cladding layer is undoped. As to the allowable level of the residual carrier density, at least the density must be (½) of the hole density or lower in the p-type cladding layer, and the density of ({fraction (1/10)}) or lower, if possible, is desired.
As an additional effect of not doping the n-electrode side cladding layer with any impurity, the purity of the active layer can be increased. When the n-electrode side cladding layer is doped with an n-type impurity, the n-type impurity remaining in the growth furnace tends to enter and mixed in the active layer, lowering the purity of the active layer. The lowering of the purity of the active layer may possibly decrease efficiency of light emission in the active layer, though it depends on the degree of purity lowering and the material type.
In the foregoing, ZnSe based LEDs have been described as examples. The present invention, however, is not limited to ZnSe based devices only, and preferable effects such as lower leakage current are expected to some degree or another, also in the light emitting devices using III-V group compound semiconductors, such GaAs or GaN. Further, the present structure is also effective not only in LEDs but also in LDs.
An LED having the n-electrode side cladding layer doped to the n-type (comparative example) and an LED having the n-electrode side cladding layer undoped, that is not doped, (example of the invention) were fabricated as ZnSe based light emitting devices having the structure shown in
The carrier density of the n-electrode side cladding layer doped to the n-type (comparative example) was 2˜3×1017 cm−3, and the carrier density of the undoped n-electrode side cladding layer (present invention) was at most 2×1015 cm−3. The carrier density of the p-type cladding layer was 3×1016 cm−3, both in the example of the invention and in the comparative example.
Though not shown, after the barrier layer was formed, an n-electrode of Ti/Au was formed on the back side of the ZnSe substrate, and a semitransparent Au electrode having the thickness of about 100 Å was formed on the contact layer. Thereafter, the structure was scribe-broken to 400 μm×400 μm, bonded on a stem, and an LED for lifetime evaluation was prepared.
Lifetimes of the LEDs as the example of the invention and as the comparative example fabricated in the above described manner were measured. As a method of measurement, a constant current of 15 mA was caused to flow at 70° C. and decrease in luminance was measured. The result was as follows.
In the LED as the comparative example, it took 200 to more than 500 hours until the luminance was decreased to 70% of the initial luminance, and the average time was about 350 hours. In contrast, in the LED of the present invention, it took 350 to more than 650 hours until the luminance was decreased to 70% of the initial luminance, and the average was about 450 hours. In other words, the LED as the example of the present invention has its lifetime made longer by about 30% than the comparative example.
In the following, characteristics of these and other embodiments of the present invention will be described in a comprehensive manner.
The light emitting device of II-VI group compound semiconductor described above may be a ZnSe based light emitting device, in which the n-type cladding layer may be an n-type Zn1-xMgxSySe1-y (0<x<1, 0<y<1) layer, and the p-type cladding layer may be a p-type Zn1-xMgxSySe1-y (0<x<1, 0<y<1) layer. The aforementioned p-type Zn1-xMgxSySe1-y is a compound semiconductor having a large band gap, and therefore, it can form a barrier, though not very effective, against the electrons that are about to enter from the trap layer to the p-type cladding layer. Therefore, the effect of elongating lifetime to some extent can be attained.
The magnitude of the band gap of the above described barrier layer should preferably be made larger by 0.025 eV to 0.5 eV than the p-type band gap.
When the band gap of the barrier layer is not larger by at least 0.025 eV than that of the p-type cladding layer, it is difficult to sufficiently suppress movement of electrons introduced to the active layer to the p-type cladding layer. When the band gap of the barrier layer is made larger by more than 0.5 eV, crystal becomes instable, affecting device characteristics.
In connection with the band gap of the barrier layer described above, the energy of the valence band thereof may be made approximately the same as that of the p-type cladding layer, and the energy of the conductive band thereof may be made larger than that of the p-type cladding layer.
By this structure, only the leakage of the electrons from the active layer to the p-type cladding layer is suppressed, and there is almost no influence on the holes in the valence band. Therefore, the lifetime can be made longer while not affecting the device characteristic.
The above described barrier layer may be formed of a II-VI group compound semiconductor including Be. By this structure, the lifetime of a ZnSe based light emitting device can be made longer without lowering light emission characteristics.
The above described barrier layer may be formed of Zn1-x-yMgxBeySe (0≧x+y≧1, 0<x, 0<y). In the II-VI group compound semiconductor including Be, particularly Zn1-x-yMgxBeySe, it is possible to elevate the bottom of the conduction band without much changing the top of the valence band. Therefore, it is possible to form a barrier against electrons that tend to leak from the active layer to the p-type cladding layer while not presenting any barrier against holes moving from the side of the p-type cladding layer to the active layer and not hindering light emission caused by the introduction of holes. As a result, it becomes possible to elongate the lifetime of the device without degrading the light emission characteristics. Further, it is possible to form epitaxial barrier layer and p-type cladding layer, and therefore, high light emission efficiency can be attained. Though it is very difficult to introduce a p-type impurity to the Zn1-x-yMgxBeySe (0.01≦y≦0.1) layer mentioned above, if introduction of p-type impurity is possible, it may contain a p-type impurity, provided that it has a band gap larger than that of the p-type cladding layer.
The above described barrier layer may be formed of Zn1-xMgxSfySe1-y (x, y are in the range of 0˜1).
By this structure, the band gap of the barrier layer can be made sufficiently larger than that of the p-type cladding layer, and entrance of electrons from the active layer to the p-type cladding layer can be prevented. As a result, longer lifetime of the light emitting device can be attained. The aforementioned Zn1-xMgxSySe1-y (x, y are in the range of 0˜1) layer should preferably be an i-type compound semiconductor. It may, however, contain a p-type impurity, provided that it has a band gap larger than that of the p-type cladding layer.
Further, between the above described barrier layer and the p-type cladding layer, a semiconductor trap layer may be provided that has a band gap smaller than the band gap of the p-type cladding layer.
By this structure, most of the electrons that have leaked over the barrier layer are trapped by defects in the trap layer or recombined with the holes before reaching the p-type cladding layer, and therefore, the number of electrons that reach the p-type cladding layer is significantly reduced. Even when the trap layer is of a p-type semiconductor, the band gap is smaller than that of the p-type cladding layer, and therefore, degradation proceeds slowly after the leaked electrons reach.
When the trap layer is provided, there results in a multi-stacked structure in which two such portions including the barrier layer and the trap layer are arranged between the active layer and the p-type cladding layer. By such multi-stacked structure, leakage of electrons to the p-type cladding layer can more surely be prevented, and the lifetime of the semiconductor light emitting device can further be made longer.
Further, the above described trap layer may be formed of ZnSxSe1-x (0≦x≦0.1). When ZnSxSe1-x (0≦x≦0.1) mentioned above is used, a trap layer having a band gap smaller than that of the p-type cladding layer can be epitaxially formed while maintaining good crystal characteristic. Further, this also enables good crystal characteristic of the epitaxial p-type cladding layer formed thereon. It is needless to say that ZnSxSe1-x (0≦x≦0. 1) mentioned above includes ZnSe.
The p-type cladding layer described above may be formed of (Zn1-xCdxS)1-z(MgS1-ySey)z (where x, y, z satisfy 0≦x≦1, 0≦y<1, 0≦z<1).
By adopting this structure in which a barrier layer is arranged between the active layer and the p-type cladding layer and the p-type cladding layer is formed of (Zn1-xCdxS)1-z(MgS1-ySey)z, lowering of luminance can be suppressed and the longer lifetime can be realized. When the p-type cladding layer is formed of other material such as p-type ZnMgSSe to have large band gap, the energy level at the bottom of the conductive band is elevated, while the energy level at the upper end of the valence band lowers. Therefore, though the band gap becomes large and the barrier potential against leaked electrons can be formed, a barrier potential is also formed against the holes that are to be introduced from the p-type cladding layer through the barrier layer to the active layer. This causes lowering of luminance of the light emitting device.
The composition x of the p-type cladding layer is determined such that the lattice constant of Zn1-xCdxS matches the lattice constant of the semiconductor substrate. Further, the composition y is determined such that the lattice constant of MgS1-ySey matches the lattice constant of the semiconductor substrate.
If the band gap of the ZnMgSSe layer used as the common p-type cladding layer were increased along with the increase in the band gap of the barrier layer, the barrier against holes would not be formed. When the band gap of the ZnMgSSe layer were made too large, however, p-type doping would become difficult. When Cd is contained in the p-type cladding layer as described above, the energy level at the upper end of the valence band becomes lower than when Cd is not contained, with the same band gap. Therefore, the barrier against holes is not formed but rather, introduction of holes to the side of the active layer is facilitated. Therefore, degradation in luminance can be suppressed. When Cd is contained in the material for forming the p-type cladding layer, energy level of not only the valence band but also of the conductive band is decreased, and therefore, sufficient confinement of electrons cannot be attained by the p-type cladding layer only. For this reason, a barrier layer having a band gap larger than that of the p-type cladding layer is combined as described above, so as to make the most of the effect attained by forming the p-type cladding layer of (Zn1-xCdxS)1-z(MgS1-ySey).
The thickness of the barrier layer described above may be in the range of at least 5 nm and at most the thickness of the active layer. When the thickness of the barrier layer is smaller than 5 nm, electrons in the active layer flow into the p-type cladding layer because of the tunneling effect, and the function as a barrier potential is hardly exhibited. When the thickness exceeds the thickness of the active layer, rigidity of the barrier layer increases, distortion matching would be lost and large distortion would result. As the upper limit of the thickness of the barrier layer, the thickness of the active layer may be used, or a specific value of 100 nm may be separately set as the upper limit.
As the compound semiconductor substrate described above, an n-type ZnSe single crystal substrate may be used. Using this substrate, it becomes possible to form an epitaxial film with good crystal characteristic, and to fabricate a light emitting device having good light emission efficiency and long lifetime.
As the compound semiconductor substrate described above, an n-type GaAs single crystal substrate may be used. A GaAs substrate is inexpensive, and it also allows formation of a ZnSe based epitaxial film. Therefore, an inexpensive light emitting device having long lifetime and good light emitting efficiency can be obtained. Further, when the n-type GaAs single crystal substrate is used as the compound semiconductor substrate, it is possible to obtain a large number of semiconductor light emitting devices of a prescribed performance level or higher efficiently at a low cost. When an n-type GaAs single crystal substrate is used, it is preferred to use ZnSSe containing S for the trap layer, from the relation with the lattice constant of the crystal.
In the stacked structure including the compound semiconductor substrate constituting the ZnSe based light emitting device described above, deviation between the peak of X-ray diffraction of the plane orientation used as an index of distortion from the substrate and the peak of x-ray diffraction of the plane orientation from the stacked structure may be at most 1000 seconds.
By this structure, it is possible to obtain a ZnSe based light emitting device having long lifetime and superior light emitting characteristics, as the deviation mentioned above is suppressed. The plane index used as an index of distortion of a compound semiconductor substrate is, generally, (400) plane. Suppression of the deviation described above leads to suppression of distortion generated in the light emitting device.
In the embodiments above, only a p-type semiconductor layer has been described as the trap layer. The trap layer, however, may be an undoped layer substantially containing no impurity (though it may contain residual impurity, regardless of p-type or n-type). Though an impurity was not mentioned in relation with the barrier layer, the barrier layer may be an undoped layer substantially containing no impurity (though it may contain residual impurity, regardless of p-type or n-type).
Magnitude relation of the thickness of trap layer and barrier layer need not specifically be limited. Functionally, however, the trap layer should preferably be thicker than the barrier layer, as the barrier layer forms a potential barrier and the trap layer traps the electrons that are moving.
Though description of LEDs only has been made in the embodiments above, the present invention is applicable to any light emitting device that uses II-VI group compound semiconductor. By way of example, the present invention may be applied to an LD, especially green LD.
Though an impurity of the barrier layer is not mentioned in the embodiments above, the barrier layer may be an undoped layer substantially containing no impurity (though it may contain residual impurity, regardless of p-type or n-type), or it may contain an impurity to attain p-type conductivity.
In the above described semiconductor light emitting device in which the other cladding layer is of an undoped semiconductor, the impurity concentration remaining in the undoped semiconductor may be smaller than 1×1016/cm3.
By the above described structure, it becomes possible to suppress impurity concentration remaining in the undoped semiconductor to be low and to decrease Fermi level of the electrons in the active layer. As a result, leakage of electrons from the active layer to the p-type cladding layer can be suppressed.
Between the active layer described above and the cladding layer to which a p-type impurity has been introduced (p-type cladding layer), a barrier layer having a band gap larger than the band gap (forbidden band) of the p-type cladding layer may be positioned.
In the above-described structure provided with the barrier layer, the undoped semiconductor lowers Fermi level of electrons in the active layer, and the barrier layer can form a higher barrier against the electrons in the active layer.
The above described semiconductor may be a II-VI group compound semiconductor.
By this structure, in the p-type cladding layer of a light emitting device formed of the II-VI group compound semiconductor, generation of donor-related defects caused by recombination of electrons and holes can be suppressed.
The semiconductor may be a ZnSe based compound semiconductor.
By this structure, when a light emitting device is formed using the ZnSe based compound semiconductor that is highly sensitive to generation of the aforementioned donor-related defects, in the p-type cladding layer of the light emitting device, generation of donor-related defects caused by recombination of electrons and holes can be suppressed.
The above described two cladding layers may be formed of ZnMgSSe.
By this structure, the band gap of the cladding layer is surely made larger than the band gap of the active layer, and the leakage of electrons from the active layer to the p-type cladding layer can be suppressed to be not higher than a prescribed amount.
The above described barrier layer may be formed of ZnMgBeSe.
As ZnMgBeSe is used as a material of the barrier layer, the band gap becomes larger than that of the cladding layer, and as a result of this larger band gap, the energy level at the bottom of the conductive band is elevated (in other words, electron affinity lowers). Therefore, leakage of electrons to the p-type cladding layer can be suppressed. Compared with ZnMgSSe, ZnMgBeSe is known to attain smaller electron affinity, and therefore, if the band gap is the same, ZnMgBeSe is more preferable as it attains higher efficiency of electron confinement. Further, ZnMgBeSe increases energy value of the conductive band while it hardly has an influence on the valence band, and therefore, it does not prevent introduction of holes from the p-type cladding layer to the active layer.
Although the present invention has been described and illustrated in detail, it is clearly understood that the same is by way of illustration and example only and is not to be taken by way of limitation, the spirit and scope of the present invention being limited only by the terms of the appended claims.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2003-328088(P) | Sep 2003 | JP | national |
| 2003-390261(P) | Nov 2003 | JP | national |
| 2003-401557(P) | Dec 2003 | JP | national |
| 2003-401560(P) | Dec 2003 | JP | national |
