WO2000058999A2

WO2000058999A2 - Semiconductor structures having a strain compensated layer and method of fabrication

Info

Publication number: WO2000058999A2
Application number: PCT/IB2000/000892
Authority: WO
Inventors: Toru Takayama; Takaaki Baba; James S. Harris, Jr.
Original assignee: Matsushita Electric Industrial Co., Ltd.
Priority date: 1999-03-26
Filing date: 2000-03-01
Publication date: 2000-10-05
Also published as: WO2000058999B1; WO2000058999A9; EP1183761A2; WO2000058999A3; JP2002540618A; CN1347581A

Abstract

The present invention provides a semiconductor structure which includes a strain compensated superlattice layer comprising a plurality of pairs of constituent layers, with the first constituent layer comprising a material under tensile stress, and the second constituent layer comprising a material under compressive stress, such that the stresses of the adjacent layer compensate one another and lead to reduced defect generation. Appropriate selection of materials provides increased band gap and optical confinement in at least some implementations. The structure is particularly suited to the construction of laser diodes, photodiodes, phototransistors, and heterojunction field effect and bipolar transistors.

Description

SEMICONDUCTOR STRUCTURES HAVING A STRAIN COMPENSATED LAYER AND METHOD OF FABRICATION

SPECIFICATION

FIELD OF THE INVENTION

This invention relates to semiconductor structures and processes, and particularly relates to the use of strain compensated layers in group Ill-nitride materials systems and methods to minimize the occurrence of lattice defects.

BACKGROUND OF THE INVENTION

The development of the blue laser light source has heralded the next generation of high density optical devices, including disc memories, DVDs, and so on. Figure 1 shows a cross sectional illustration of a prior art semiconductor laser devices. (S. Nakamura, MRS BULLETIN, Vol. 23, No. 5, pp. 37-43, 1998.) On a sapphire substrate 5, a gallium nitride (GaN) buffer layer 10 is formed, followed by an n-type GaN layer 15, and a 0.1 mm thick silicon dioxide (SiO₂) layer 20 which is patterned to form 4 mm wide stripe windows 25 with a periodicity of 12 mm in the GaN<1 -100> direction. Thereafter, an n-type GaN layer 30, an n-type indium gallium nitride (in_{0 1}Ga₀₉N) layer 35, an n-type aluminum gallium nitride (Al_{0 14}Ga₀₈₆ N)/GaN MD-SLS (Modulation Doped Strained-Layer Superiattices) cladding layer 0, and an n-type GaN cladding layer 45 are formed. Next, an ln₀₀₂Ga _o.g₈N/ln_{0 15}Ga₀₈₅N MQW (Multiple Quantum Well) active layer 50 is formed followed by a p-type AI₀₂Ga₀₈N cladding layer 55, a p-type GaN cladding layer 60, a p-type Al₀., ₄Ga₀₈₆N/GaN MD- SLS cladding layer 65, and a p-type GaN cladding layer 70. A ridge stripe structure is formed in the p-type Al_{0 1 4}Ga₀₈₆N/GaN MD-SLS cladding layer 55 to confine the optical field which propagates in the ridge waveguide structure in the lateral direction. Electrodes are formed on the p-type GaN cladding layer 70 and n-type GaN cladding layer 30 to provide current injection.

In the structure shown in Figure 1 , the n-type GaN cladding layer 45 and the p-type GaN 60 cladding layer are light-guiding layers. The n-type Al₀ , ₄Ga₀₈₆N/GaN MD-SLS cladding layer 40 and the p-type AI_0.14Ga₀₈₆N/GaN MD-SLS cladding layer 65 act as cladding layers for confinement of the carriers and the light emitted from the active region of the InGaN MQW layer 50. The n-type

layer 35 serves as a buffer layer for the thick AIGaN film growth to prevent cracking. By using the structure shown in Figure 1 , carriers are injected into the InGaN

MQW active layer 50 through the electrodes, leading to emission of light in the wavelength region of 400 nm. The optical field is confined in the active layer in the lateral direction due to the ridge waveguide structure formed in the p-type Al_{0 1} ₄Ga₀₈₆N/GaN MD-SLS cladding layer 65 because the effective refractive index under the ridge stripe region is larger than that outside the ridge stripe region. On the other hand, the optical field is confined in the active layer in the transverse direction by the n-type GaN cladding layer 45, the n-type Al_{0 14}Ga₀₈₆N/GaN MD-SLS cladding layers 40, the p-type GaN cladding layer 60, and the p-type Al_{0 14}Ga₀₈₆N/GaN MD-SLS cladding layer 55 because the refractive index of the of the active layer is larger than that of the n-type GaN cladding layer 45 and the p-type GaN cladding layer 60, the n-type Al_{0 14}Ga₀₈₆N/GaN MD-SLS Iayer40, and the p-type Al_{0 14}Ga₀₈₆N/GaN MD-SLS cladding layer 60. Therefore, fundamental transverse mode operation is obtained.

However, for the structure shown in Figure 1 , it is difficult to reduce the defect density to the order of less than 10⁸ cm^'2, because the lattice constants of AIGaN, InGaN, and GaN differ sufficiently different from each other that defects are generated in the structure as a way to release the strain energy whenever the total thickness of the n-type ln_{0 1}Ga₀₉N layer 35^, the ln₀₀₂Ga ₀₉₈N/ln_{0 15}Ga₀₈₅N MQW active layer 50, the n-type AI_{0 14}Ga₀₈₆N/GaN MD-SLS cladding layer 40, the p-type Al_{0 14}Ga₀₈₆N/GaN MD-SLS cladding layer 65, and the p-type AI₀₂Ga₀₈N cladding layer 55 exceeds the critical thickness. The defects result from phase separation and act as absoφtion centers for the lasing light, causing decreased light emission efficiency and increased threshold current. The result is that the operating current becomes large, which in turn causes reliability to suffer.

Moreover, the ternary alloy system of InGaN is used as an active layer in the structure shown in Figure 1. In this case, the band gap energy changes from 1.9 eV for InN to 3.5 eV for GaN. Therefore, ultraviolet light which has an energy level higher than 3.5 eV cannot be obtained by using an InGaN active layer. This presents difficulties, since ultraviolet light is attractive as a light source for the optical pick up device in, for example, higher density optical disc memory systems and other devices.

To better understand the defects which result from phase separation in conventional ternary materials systems, the mismatch of lattice constants between InN, GaN, and AIN must be understood. The lattice mismatch between InN and GaN, between InN and AIN, and between GaN and AIN, are 1 1.3%, 13.9%, and 2.3%, respectively. Therefore, an internal strain energy is accumulates in an InGaAIN layer, even if the equivalent lattice constant is the same as that of the substrate due to the fact that equivalent bond lengths are different from each other between InN, GaN, and AIN. In order to reduce the internal strain energy, there is a compositional range which phase separates in the InGaAIN lattice mismatched material system, where In atoms, Ga atoms, and Al atoms are inhomogeneously distributed in the layer. The result of phase separation is that In atoms, Ga atoms, and Al atoms in the InGaAIN layers are not distributed uniformly according to the atomic mole fraction in each constituent layer. In turn, this. means the band gap energy distribution of any layerwhich includes phase separation also becomes inhomogeneous. The band gap region of the phase separated portion acts disproportionately as an optical absorption center or causes optical scattering for the waveguided light. As noted above, a typical prior art solution to these problems has been to increase drive current, thus reducing the life of the semiconductor device.

Another conventional approach to obtaining low defect density laser diodes with a GaN material system is to use only GaN in the cladding layers. However, this has the disadvantage that the optical confinement in the active layer is less than it would be with AIGaN cladding layers, because the refractive index step between the active layer and the GaN cladding layer is smaller than if AIGaN is used in the cladding layer. As a result, the optical field spreads in the transverse direction. Less optical confinement in an active layer requires increased threshold current to yield the same luminance. Moreover, for GaN cladding layers the barrier potential is smaller than for AIGaN cladding layers; this permits the carriers to easily overflow the active layer, again leading to increased threshold current. Therefore, the operating current increases, which causes a decrease in reliability and, statistically, longevity. Thus, AIGaN cladding layers are widely used despite the fact that such cladding layers generate defects.

There has therefore been a long felt need for a semiconductor structure which reduces lattice defects and can be used to obtain laser diodes, transistors or other devices with low threshold current and long term reliability.

SUMMARY OF THE INVENTION

The present invention overcomes substantially the limitations of the prior art and provides a semiconductor structure with low defect density and, consequently, improved reliability. The invention may be used to fabricate, among other devices, biue light and other laser diodes, heterojunction field effect transistors, heterojunction bipolar transistors, and photodiodes.

In simplified terms, the present invention provides a semiconductor structure having a substrate on which is formed a first cladding layer of a first conductivity type. A first superlattice layer of the first conductivity type is then formed on the first cladding layer, where the superlattice layer has characteristics discussed further below. Then, an active layer is formed on the superlattice layer, afterwhich a second superlattice layer of a second conductivity type is formed. Finally, a second cladding layer of the second conductivity type is formed. Guide layers may also be used immediately on either side of the active layer. Electrodes are formed in a conventional manner.

The superlattice layers each form a cladding layer comprised of a plurality of layers, each below its critical thickness, of alternating ternary or quaternary materials such as, for example, AIGaN and InGaN, or InGaAIN in different mole fractions. In one exemplary embodiment, the superlattice may comprise on the order of 200 pairs of layers. If a ternary system, such as AIGaN and InGaN, is used forthe superlattice, the AIGaN layers are under tensile stress, while the InGaN layers are under compressive stress. By alternating the layers, the stress is compensated at the interface of the AIGaN/lnGaN layers, resulting in fewer defects within the layer and increased reliability. The superlattice layers are of opposite conductivity types, and sandwich the quantum well active layer, which may be implemented as either a single well or as multiple wells. By appropriate selection of the mole fractions, the lattice constant of the AIGaN layer can be arranged to be less than the lattice constant of the adjacent GaN layer, and the lattice constant of the InGaN layer can be arranged to be greater than the lattice constant of the adjacent GaN layer. The net result is a superlattice layer with balanced stresses that essentially average to the lattice constant of the adjacent GaN layer, thus substantially reducing the formation of defects due to stress.

In a first embodiment of the invention, a semiconductor structure - which may be, for example, a laser diode - comprises the following: on a GaN or other substrate, a GaN first cladding layer of a first conduction type is formed, followed by the formation of a first superlattice layer of the same conduction type as the first cladding layer. The first superlattice layer, which may be thought of as a second cladding layer, may be comprised of a plurality of pairs of layers, typically either AIGaN together with

InGaN, or InGaN together with InAIN. A guide layer, typically InGaN material having the same conduction type as the first cladding layer, is then formed, after which a quantum well active layer, typically of InGaN material, is formed. The active layer may be formed with either a single quantum well or with multiple (for example, on the order of three pairs) quantum well design. Another InGaN guide layer is typically formed above the active layer, but of a conduction type opposite the first cladding layer.

A second superlattice layer, which serves as a third cladding layer and has a conduction type opposite the first cladding layer, is then formed above the guide layer. As with the first superlattice layer, the second superlattice layer is typically comprised of a plurality of layers, for example either AIGaN in combination with InGaN, or InGaN togetherwith InAIN. The superlattice layers each may comprise on the order of 200 pairs of layers of the complementary material, although the precise number is not critical. A GaN fourth cladding layer is typically formed above the superlattice third cladding layer. Electrodes are formed in the conventional manner.

As noted above, the superlattice material pair may be selected from a group which includes the pairs Al_xa,Ga₁._xa|N/ln_xlGa₁._xlN and ln_xayGa₁._xayN/ln_xnAI₁._xnN. With the foregoing structure, in the first superlattice layer the AI^Ga^N layer is under tensile stress, while the In^Ga^N layer is under compressive stress so that the stresses are able to compensate each other at the interface of the respective constituent layers. Likewise, in the second superlattice layerthe Al_xalGa,._xalN layer is under tensile stress, and the ln_xlGa₁._xlN layer is under compressive stress so that the stresses in this supperlattice are also able to compensate each other at their interfaces. The operation is the same if the ln_xayGa₁._xayN/ln_xrtAI₁._)<nN material pair is selected.

Moreover, the

superlattice layer can be designed to confine the optical field within the active layer better than if GaN alone is used for the cladding layer. By increasing the optical confinement within the active layer in the transverse direction, the threshold current of the device can be reduced. Further, the

superlattice layers design minimizes absorption of the lasing light from the active layer. Therefore, low threshold current and low defect density laser diodes are obtained. The first embodiment may be implemented with any from a selection of material choices for the superiattice and active layer, and may also include various alternatives for the substrate and the outermost cladding layers. In particular, implementation of the first embodiment may include substrates of sapphire, silicon carbide, GaN, and so on. The superlattice layers may comprise AI^Ga^N and where xal is on the order of 0.2 and xi is on the order of 0.04 to at least as high as 0.2; or may comprise In^Ca^^N and In^AI^N where xay is on the order of 0.04 and xn is on the order of 0.13. Further, the active layer may include single or multiple quantum wells of In^Ga^N material. It is preferred that the variables xi and xa have the relationship xa > xi.

In a second embodiment of the invention, a semiconductor structure based on a ternary material system is again implemented. The second arrangement, of which a laser diode is again an exemplary implementation, comprises a suitable substrate together with a first conduction type of GaN or similar first cladding layer, a superlattice second cladding layer of the first conduction type, and a quantum well active layer of, for example, In^Ga^-N material which may be either single or multiple quantum well. Guide layers may also be implemented immediately on either side of the active layer to help confine the optical field, but are not needed in all embodiments. The superlattice second cladding layer may be, as with the first embodiment, either

or equivalent.

Then, a superlattice third cladding layer of a conduction type opposite to the conduction type of the first cladding layer is formed, but in this embodiment comprises only fifteen to fifty pairs of layers of Al_xalGa,._xalN/ln_x,Ga₁._xlN, In^_yGa^^N/ln^AI^^N or equivalent material. Thereafter, a current blocking layer is formed above the superlattice third cladding layer, and a window is then formed in the current blocking layer which exposes a portion of the superlattice third cladding layer. Thereafter, a superlattice fourth cladding layer is formed above the current blocking layer, and may be on the order of 200 pairs of layers. The window in the current blocking layer provides an interface between the superlattice fourth cladding layer and the superlattice third cladding. The superlattice fourth cladding layer is of the same conduction type as the superlattice third cladding layer. As with the first embodiment, xal defines the AIN mole fraction (using that material as an example), xi and xa define the InN mole fraction, and xi and xa have a relationship of xa > xi. Finally, a fifth cladding layer of, for example, GaN, is formed above the fourth cladding layer, and electrodes are formed in a conventional manner.

Similar to the first embodiment, the lattice constant of AIGaN (or equivalent) in the superlattice layer is smaller than that of the GaN cladding layer, and the lattice constant of the InGaN in the superiattice layer is larger than that of the GaN cladding layer. In this case, the AIGaN layer is under tensile stress, while the InGaN layer in under compressive stress, again causing the stress in the complementary layers to compensate one another at the interface of the AIGaN layer and the InGaN layer. Likewise, the AIGaN/lnGaN superlattice layer provides better confinement of the optical field within the active layer than if GaN had been used for the cladding layer. Again, the improved optical confinement within the active layer in the transverse direction leads to reduced threshold current. Lowered threshold current is also possible because the AIGaN/lnGaN superlattice layer does not absorb the lasing light from the InGaN single quantum well active layer since the InN mole fraction xa is larger than the InN mole fraction xi. This causes the band gap energy of the InGaN in the AIGaN/lnGaN superlattice layer to become larger than that of the InGaN single quantum well active layer. The ultimate result is that a semiconductor structure with low threshold current and low defect density can be constructed.

As will be appreciated from the foregoing, the primary difference between the first and second embodiments is the addition of the current blocking layer which, in the exemplary embodiment described above, is sandwiched between a smaller superlattice layer and a larger superlattice layer. In the exemplary arrangement described above, the semiconductor structure has an AI^Ga^N current blocking layer with a window region formed therethrough into the Al_xa|Ga₁__xa,N/ln_XiGa₁.-._iN superlattice third cladding layer, with the current blocking layer having an opposite conduction type to the

superlattice layer, wherein xb defines the AIN mole fraction, and xb and xal have a relationship of xb > xal. By using such a current blocking layer to create a window region in the superlattice layer, the effective refractive index in the window region becomes larger than that outside the window region. This helps to confine the optical field within the active layer in the lateral direction under the window region. The effective refractive index in the window region is increased because the AIN mole fraction, xb, is larger than that of the superlattice cladding layer, xal, outside the window region. Moreover, by having the conduction type of the AIGaN current blocking layer different from the AIGaN/lnGaN superlattice cladding layer, the injected current is confined within the window region. This causes the injected carrier density in the active layer under the window region to become high enough to obtain lasing oscillation. Therefore, the use of such a current blocking layer with a window region into the superlattice layer permits a laser diode with single transverse mode operation to be obtained.

A third embodiment of the invention is similar in structure to the first embodiment, but is implemented with a quaternary material system instead of the ternary material system described above. In such an embodiment, a cladding layer of a first conduction type of In^.^Ga^AI^N material is formed on a GaN or other substrate. Thereafter, a first superlattice layer of the first conduction type is formed as a second cladding layer, comprising ln₁._x2._y2Ga_x2Al_y-.N and ln₁._x3__y3Ga_X3Al_y3N material. The lattice constant of the In^-.^Ga^AI^N material is selected to be larger than that of the ln-._x1._y1Ga_x1Al_y1N material in the cladding layer, while the lattice constant of the ^{l r}iι._x3._y3Ga_x3Al_y3N material is selected to be smaller than that of ln-._x1._y1Ga_x1Al_y1N material. A quantum well active layer is then formed, for example of InGaN material which may be either single or multiple quantum wells, followed by a second superlattice layer of an opposite conduction type. The second superlattice layer may, for example, comprise In^^Ga^AI^N and In_L^sGa^Al_ysN, where the lattice constant of the In^^Ga^AI^N is largerthan that of the In^.^Ga^AI^N material and the lattice constant of said

The second superlattice layer serves as a third cladding layer. Then a fourth cladding layer of a conduction type opposite to the first cladding layer is formed, typically of In, _x6._y6Ga_x6Al_y6N material. The values x1 , x2, x3, x4, x5, and x6 define the GaN mole fraction and y1 , y2, y3, y4, y5, and y6 define the AIN mole fraction. As with the first and second embodiments, guide layers may be implemented in some embodiments to help confine the optical field, and if implemented are formed immediately on either side of the active layer.

In the third embodiment, like the first and second embodiment, in the first superlattice layer the ln.._x2._y2Ga_x2Al_y2N layer is under tensile stress, while the In, _{x3 y3}Ga_x3Al_y3N layer is under compressive stress so that the stresses are able to compensate each other at the interface of the In, _x2._y2Ga_x2Al_y2N layer and the In,.^ _y3Ga_x3Al_y3N layer. Similarly, in the second superlattice layer the ln,._x4._y4Ga_x4Al_y4N layer suffers tensile stress, and the In^s ^Ga^Al_ysN layer suffers compressive stress so that the stresses compensate each other at the interface of the ln₁._x4._y4Ga_x4Al_y4N layer and the In, _{x5 y5}Ga_x5Al_y5N layer

The InGaAIN superlattice layer is designed to confine the optical field within the active layer better than if GaN alone is used for the cladding layer. By increasing the optical confinement within the active layer in the transverse direction, the threshold current of the device can be reduced. Further, the InGaAIN superiattice layer is preferably also designed not to absorb the lasing light from the active layer. Therefore, low threshold current and low defect density laser diodes are obtained.

A fourth embodiment of the invention essentially comprises the quaternary material of the third embodiment with the general structure of the second embodiment

-- that is, the use of superlattice third and fourth cladding layers on either side of a current blocking layer, with a window in the current blocking layer which permits an interface between those superlattice layers.

The foregoing points summarizing the invention may be better appreciated from the following Detailed Description of the Invention, taken together with the appended Figures described below.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 shows a prior art laser diode.

Figure 2 shows, in cross-sectional view, a simplified version of the present invention. Figure 3 shows a simplified cross-sectional illustration of a semiconductor device of the first embodiment.

Figures 4A-4C show a simplified series of steps forfabncating a semiconductor structure in accordance with the first embodiment.

Figure 5 shows the relationship between the excess stress and the In content in the superlattice cladding layer.

Figure 6 shows the output power dependence on the injected current density of the first embodiment.

Figure 7 shows the output power dependence on the injected current density of the third embodiment. Figure 8 shows the relationship between excess stress and the In content in the superlattice cladding layer.

Figure 9 shows the relationship between excess stress and the In content of the InAIN layers in the superlattice cladding layer

Figure 10 shows in cross-sectional view a simplified illustration of a semiconductor device in accordance with a second embodiment of the invention.

Figure 1 1 shows the relationship between the effective refractive index difference (Dn) between inside and outside the window region and the thickness of the third cladding layer (dp)

Figures 12A-C show a simplified series of steps for fabricating a semiconductor laser diode in accordance with the second embodiment

Figure 13 shows the output power dependence on the injected current density of the second embodiment

Figure 14 shows the relationship between the effective refractive index difference (Dn) between inside and outside the window region and the thickness of the third cladding layer (dp)

Figure 15 shows the output power dependence on the injected current density of the fourth embodiment.

Figure 16 shows in cross-sectional view a simplified illustration of a semiconductor device in accordance with a third embodiment of the invention.

Figure 17 shows in cross-sectional view a simplified illustration of a semiconductor device in accordance with a fourth embodiment of the invention Figure 18 shows a heterojunction field effect transistor constructed in accordance with the present invention

Figure 19 shows a heterojunction bipolar transistor constructed in accordance with the present invention

Figure 20 shows a photodiode constructed in accordance with the present invention

Figure 21 shows a phototransistor constructed in accordance with the present invention

DETAILED DESCRIPTION OF THE INVENTION

Referring first to Figure 2, a generalized form of a semiconductor structure in accordance with present invention is shown therein On a substrate 100, which may be GaN, sapphire, silicon carbide or any other suitable substrate, a first cladding layer 105 is formed The first cladding layer will typically be of the same conduction type as the substrate A second cladding layer 1 10 is thereafter formed atop the first cladding layer 105, with the second cladding layer having the same conductivity type as the first cladding layer

The second cladding layer 1 10 is comprised of a plurality of pairs of layers 1 1 5, each less than its critical thickness but which together form a superlattice The use of a superlattice layer overcomes the smaller lattice constant of AIGaN (and similar materials) relative to GaN, while providing the benefit of the relatively larger band gap of AIGaN or similar materials and also its smaller refractive index relative to GaN In general, the constituent layers of the superiattice are selected so that one of the layers has a lattice constant greater than that of the active layer, for example GaN, while the other constituent layer has a lattice constant less than the active layer. This relationship between the lattice constants of the constituent layers of the superiattice may be expressed simply as SL1 > GaN > SL2. For Group ill nitride, the relationship of lattice constants is InN > GaN > AIN; this also means that the lattice constants of InGaN, GaN, and AIGaN have the relationship InGaN > GaN > AIGaN. Similarly, by proper selection of the atomic content, the relationships of the lattice constants of other materials can be set as follows: InAIN > GaN > AIGaN, and InGaN > GaN > InAIN These relationships will be discussed in greater detail hereinafter.

The constituent layers 1 15 of the superlattice second cladding layer 1 10 are maintained in opposing stress; thus, a first layer is maintained in tensile stress while the adjacent layer is maintained in compressive stress Each layer is less than the critical thickness for that material, and thus cracking within the material is avoided The number of pairs of layers which comprise the superlattice may vary widely, from twenty or less to more than 200, with layers of increasing thickness offering greater optical confinement but increased electrical and thermal resistance and, as a result, increased heating

After the superlattice layer 1 10 is fabricated, an active layer 120 is grown atop the superlattice layer 100, and a superlattice third cladding layer 125 of a conduction type complementary to the superlattice layer 1 10 is grown Alternatively, as will be discussed in greater detail hereinafter, a guide layer may be fabricated above the superlattice layer 1 10, after which the active layer 120 is fabricated In that event, a second guide layer is grown atop the active layer, after which the superlattice third cladding layer 125 is grown The guide layers will have the same conduction types as their adjacent superlattice layers

Following fabrication of the superlattice layer 125, a fourth cladding layer 130 is formed of the same conduction type as the layer 125 A pair of electrodes 135 and 140 may then be formed in a conventional manner, for example on the underside of the substrate 100 and atop the fourth cladding layer 130

By proper selection of the materials for the superlattice layers, the tensile and compressive stresses can be balanced in those layers to minimize defect density In addition, the optical field can be better confined within the active layer than if a single cladding layer of GaN, for example, is used because the index difference between the superlattice cladding layer and the active layer is higher than'the index difference between conventional GaN and the active layer

Referring next to Figures 3 and 4A-4C, a first embodiment of a semiconductor structure in accordance with the present invention is shown in greater detail. For simplicity in illustrating this first embodiment of the invention, a laser diode has been selected as an exemplary semiconductor structure, and is shown in simplified cross- sectional view. On an n-type GaN substrate 150, an n-type GaN first cladding layer 5 155 is formed on the order of 0.5μm thick. Thereafter, a superlattice second cladding layer 160 of an n-type material is formed. The number of pairs of layers for an exemplary implementation of the first embodiment may be on the order of 200. The material used for the superlattice layers may be any of several combinations which exhibit suitable lattice constants, conductivity, and so on. Exemplary materials are 0 Al„ ₂Ga_{0 8}N/ln₀₀₄Ga₀₉₆N, or Al„ ₂Ga₀₈N/ln₀₂AI₀₈N, or ln₀₀₄Ga₀₉₆N/ln_{0 13}AI₀₈₇N, as will be discussed in greater detail hereinafter. A typical thickness for each of the constituent layers of the superlattice is on the order of 20A, although the precise thickness may differ within a reasonable tolerance as long as the critical thickness for dislocation generation is not exceeded. 5 Once the superlattice second cladding layer 160 is fabricated, an n-type guide layer 165 of an ln₀₀₂Ga₀₉₈N material on the order of 35A thick may be formed if desired, although the use of a guide layer is not required in at least some embodiments. Thereafter, a quantum well active layer 170 is formed, which may be either single or multiple wells. If multiple well, a three pair configuration has been 0 found desirable although the exact configuration may vary with the application. For a single well implementation, the layer 170 may comprise ln_{0 15}Ga₀₈₅N material on the order of 35A thick. If a multiple well quantum layer is preferred, the layer 170 may comprise three pairs of ln_{0 15}Ga₀₈₅N/ thιck)ln₀₀₃Ga₀₉₈N (35A thick) material. If a guide layer 165 is used, a second guide layer 175 is formed, for example of ln₀₀₃Ga₀₉₇N 5 material on the order 35λ thick and of a conduction type opposite the first guide layer. Thereafter, a superlattice third cladding layer 180 of p-type material is formed. As with the layer 160, the superlattice layer 180 may comprise 200 pairs of Al_{0 2}Ga₀₈N/ln₀₀₄Ga₀₉₆N material, typically 20A thick or, alternatively, may be Al₀₂Ga₀₈N/ln_{0 2}AI_{0 8}N, or ln₀₀₄Ga₀₉₆N/ln_{0 13}AI_{0 87}N Finally, a p-type of GaN fourth 0 cladding layer 185, typically on the order of 0 5μm thick, is formed A pair of electrodes, not shown, is formed in a conventional manner

In order to emit blue light with a wavelength range of 450 nm from the active layer 170, the InN mole fraction in the active layer 170 is set to be on the order of 0 15 Basically, carriers are injected from the n-type substrate 150 and the p-type 5 GaN fourth cladding layer 185 and recombine in the active layer 170, leading to the emission of the blue light

The superlattice layers 160 and 180 serve to confine the optical field in the transverse direction betterthan conventional GaN cladding layers, because the index difference between the active layer and the cladding layer becomes larger than the index step difference between the active layer and a conventional GaN layer. The strong optical confinement in the active layer leads to a low threshold current laser diode.

To minimize the generation of the defects, strain compensated superlattice layers 160 and 180 are used for the cladding layer, instead of a conventional AIGaN cladding layer or a simple GaN cladding layer. In the superlattice structure of the present invention, the thickness of each of the constituent layers which comprise the superlattice layer is maintained below a critical thickness, or typically on the order of 20λ. This substantially reduces the stress in the cladding layer, and so minimizes the defect density in the layer. For each material pair used in the superlattice layer, one of the materials, for example Al₀₂Ga₀₈N, is under tensile stress, while the other layer, for example ln_{0 04}Ga₀₉₆N, is under compressive stress. By selecting one material (e.g., AI_{0 2}Ga_{0 8}N) with a lattice constant smaller than GaN, and another (e.g., ln₀₀₄Ga₀₉₆N) with a lattice constant greater than GaN, the stress can be compensated at the interface between the layer and the layer. This prevents the accumulation of stress and leads to lower defect densities relative to conventional GaN cladding layers. As noted previously, several material combinations can be used for the superlattice layers. Each of the exemplary combinations of materials -- AI_{0 2}Ga₀₈N/ln₀₀₄Ga₀₉₆N, AI₀₂Ga₀₈N/ln_{0 2}AI₀₈N, or ln_{0 04}Ga₀₉₆N/ln_{0 13}AI₀₈₇N - will be discussed in turn.

If an AI_{0 2}Ga_{0 8}N/ln₀₀₄Ga₀₉₆N combination is used, Figure 5 shows the relation between the excess stress in the waveguide structure and the In content of the InGaN layers of the AIGaN/lnGaN superlattice cladding layer. The other structural parameters except the In content of the InGaN layer of the AIGaN/lnGaN superlattice cladding layer are fixed. For present purposes, excess stress is defined as the difference between the maximum stress in the epiiayer in the waveguide without dislocation of the waveguide structure and the effective stress associated with the dislocation line. If the excess stress becomes positive, the strain energy becomes smaller when the dislocation is generated in the waveguide structure than it does when that dislocation is not generated in the waveguide structure. This means that the structure is energetically more stable when dislocations are generated in the waveguide structure than when they are not.

If the excess stress becomes negative, however, the opposite occurs: the strain energy becomes smaller when the dislocation is not generated in the waveguide structure than when the dislocation is generated in the waveguide structure. This means that the structure is energetically more stable in the case that dislocations are not generated in the waveguide structure than in the case that dislocations are generated in the waveguide structure. As shown in Figure 5, the excess stress becomes smallest when the In content equals 0.04. Therefore, in the structure of the embodiment shown in Figure 3, the AIN mole fraction of the AIGaN layers in the superiattice cladding layer and the InN mole fraction of the InGaN layers in the superlattice cladding layer are set to be 0.2 and 0.04, respectively.

Moreover, if an AI₀₂Ga_{0 8}N/ln₀₀₄Ga₀₉₆N superiattice layer is used as a cladding layer, the optical confinement in the active layer in the transverse direction becomes larger than the case only a GaN cladding layer is used as a material for the cladding layer, because the average index of the Al_{0 2}Ga₀₈N/ln₀₀₄Ga₀₉₆N superlattice cladding layer is smallerthan the GaN cladding layer, leading to obtain larger index difference between the cladding layer and the active layer. Figure 6 shows the light-current characteristics of the laser diode of the first embodiment where the superlattice materials are Al_{0 2}Ga_{0 8}N/ln₀₀₄Ga₀₉₆N and a single quantum well is used. The laser diode is driven by a pulsed current with a duty cycle of 1 %. The threshold current density is found to be 5.2 kA/cm2, which is about half the threshold current density of a laser diode with cladding layers fabricated only with GaN. Figure 7 shows the light-current characteristics of a laser diode constructed in accordance with the first embodiment, but using a multiple quantum well design. The laser diode is driven with a puised current with a duty cycle of 1 %. The threshold current density of 4.2 kA cm2 which is also about half the threshold current density of a multiple quantum well laser diode using only GaN for its cladding layers. A second exemplary combination of materials for the superlattice layers 160 and 180 is AI_{0 2}Ga₀₈N/ln₀₂AI_{0 8}N. In the event AI_{0 2}Ga₀₈N/ln_{0 2}AI₀₈N is used for the superiattice layers, the stress equations are slightly different. Figure 8 shows the relation between the excess stress in the waveguide structure and the In content of the InAIN layer of the AIGaN/lnAIN superiattice cladding layer. The other structural parameters except the In content of the InAIN layers of the AIGaN/lnAIN superiattice cladding layers are fixed to the value mentioned above. As shown in Figure 8, the excess stress becomes the smallest in the case that In content equals 0.2. Therefore, if Al_{0 2}Ga_{0 8}N/ln_{0 2}AI_{0 8}N is used for the superiattice layers, the AIN mole fraction of the AIGaN layers in the superlattice cladding layer and the InN mole fraction of the InAIN layers in the superlattice cladding layer are set to be 0.2 and 0.2, respectively, to ensure that the strain is compensated by the adjacent constituent layers.

Further, for the case where AI_{0 2}Ga_{0 8}N/ ln_{C 2}AI₀₈N is used as the superlattice cladding layer, the confinement of the optical field within the active layer in the transverse direction is better than if only a GaN cladding layer is, because the average index of the Al_{0 2}Ga₀₈N/ ln_{0 2}AI₀₈N superlattice cladding layer is smaller than the GaN cladding layer. 5 A third exemplary combination of materials for the superiattice layers 160 and

180 is ln₀₀₄Ga_{o g6}N/ln_{0 13}AI_{0 87}N. In this combination, the in_{0 13}AI₀₈₇N layers are under tensile stress, while the ln₀₀₄Ga₀₉₆N layers are under compressive stress. The stress is thus compensated at the interface between the ln_{0 13}AI₀₈₇N layer and the l n_{0 04}Ga_{0 96}N layer. The relationship of lattice constants is

10 ln₀₀₄Ga₀₉₆N > GaN > ln₀.₁₃AI₀₈₇N.

As with the prior combinations of materials, the stress equations vary because of the differences in materials. Figure 9 shows the relation between the excess stress in the waveguide structure and the In content of the InAIN layer of the InGaN/lnAIN superiattice cladding layer. The other structural parameters except the In content of

15 the InAIN layers of the InGaN/lnAIN superiattice cladding layers are fixed to the value mentioned above. As shown in Figure 9, the excess stress becomes the smallest in the case that the In content of the InAIN layer equals 0.13. Therefore, in the structure of the ninth embodiment shown in Figure 9, in order to compensate the strain, the InN mole fraction of the InGaN layers in the superlattice cladding layer and the AIN mole

20 fraction of the InAIN layers in the superiattice cladding layer are set to be 0.04 and 0.87, respectively.

Moreover, if ln₀₀₄Ga_{0 96}N/ln_{0 13}AI₀₈₇N is used for the superlattice cladding layer, the confinement of the optical field within the active layer in the transverse direction is better than if a cladding layer of only GaN is used. The average index of the

25 ln₀₀₄Ga₀₉₆N/ln_{0 13}AI₀₈₇N superlattice cladding layer is smaller than the GaN cladding layer, resulting in a larger index difference between the cladding layer and the active layer that occurs if just GaN cladding layers are used.

Referring next to Figures 10 and 11A-1 1 C, a second embodiment of the present invention may be better appreciated. As with the embodiment of Figure 3,

30 Figure 10 is a simplified cross-sectional illustration of a semiconductor laser diode of the second embodiment, while Figures 11A-11C show a simplified version of the fabrication steps for creating the structure of Figure 10. On an n-type GaN substrate 300, an n-type GaN first cladding layer 305 is formed on the order of 0.5μm thick, followed by an n-type superlattice second cladding layer 310 having on the order of

35 200 pairs of constituent layers. Then, an ln₀₀₂Ga₀₉₈N guide layer 315 on the order of 35A thick is formed, followed by a quantum well active layer 320. The quantum well active layer, which may be on the order of 35A thick, can be either single or multiple quantum wells. If a single quantum well design is implemented, the active layer typically comprises ln_{0 15}Gao.₈₅N. If a multiple quantum well design is implemented, the active layer may be implemented as three pairs of ln_{0 15}Ga₀₈₅N/ln₀₀₃Ga₀₉₈N multiple quantum wells, with each layer on the order of 35A thick. Thereafter, an ln₀₀₃Ga₀₉₇N guide layer 325 on the order of 35A thick may be implemented in some embodiments.

Then, in a significant difference from the first embodiment, a p-type superlattice third cladding layer 330 is formed. However, the layer 330 typically comprises only on the order of 25 pairs of constituent layers, each on the order of 20A thick. Then, a p-type Al_{0 22}Ga_{0 78}N current-blocking layer 335 is formed, on the order of 100A thick. A stripe-like window 340 is then formed in the current-blocking layer 335 to expose a portion of the third cladding layer 330. Then, a p-type fourth superiattice cladding layer 345 is formed, typically comprising on the order of 200 pairs of constituent layers. Finally, a p-type GaN fifth cladding layer 350 is formed on the order of 0.5μm thick. Electrodes may be formed in a conventional manner.

As with the first embodiment, several different combinations of materials may be used to fabricate the superiattice layers 310, 330, and 345, and may include Al_α2Ga₀₈N/ln₀₀₄Ga_α96N, AI_{0 2}Ga₀₈N/ln_{0 2}Al_α8N, or ln₀₀₄Ga_α96N/ln_{0 13}Al_α87N. The operation of these materials is as discussed in connection with the first embodiment, except for the operation of the current blocking layer discussed in greater detail below. As a result, the remainder of the discussion of the second embodiment will use the example of AI₀₂Ga₀₈N/ln₀₀₄Ga₀₉₆N, although it is to be understood that each of the combinations is acceptable in the same manner as with the first embodiment.

In order to emit blue light with a wavelength range of 450 nm from the active layer 320, the InN mole fraction in the active layer 320 is set to be 0.15. In order to obtain a fundamental transverse mode operation, the width of the window is set to be 2 mm.

In order to have a single lateral mode oscillation, the AIN mole fraction of the current-blocking layer 335 is set to be higher than that of the p-type AI_{0 2}Ga_{0 8}N/ln₀₀₄Ga_{0 96}N superlattice fourth cladding layer 350. When the AIN mole fraction of current-blocking layer 335 is the same as that of the fourth cladding layer 345, the refractive index within the stripe will be lowered due to a plasma effect and a wave guide will be formed with a resultant failure to generate a single lateral mode oscillation. The lateral mode oscillation becomes unstable when the AIN mole fraction of the current-blocking layer 335 is lower than that of the p-type Al_{0 2}Ga_{0 8}N/ln₀₀₄Ga₀₉₅N superlattice fourth cladding layer 345. In this instance, the AIN mole fraction of the current-blocking layer 335 is set to be 0.22, which is higher than that of the p-type AI₀₂Ga₀₈N/ln₀₀₄Ga₀₉₆N superiattice fourth cladding layer 345. The thickness of the third cladding layer (dp) 330 also affects the effective index difference (Δn) inside the window region and outside the window region. When the value of dp is large, Δn becomes small. On the other hand, when the value of dp is small, Δn becomes large. In the case that Δn is large, the optical field is confined in the lateral direction more strongly, leading to the burning of a spatial hole such that the optical field deforms. The deformation of the optical field is a crucial issue for the use of such devices for optical pick-up systems. If Δn is small, the optical field spreads in the lateral direction into the active layer outside the window region. In this case, the active layer outside the window region is not highly activated by the injected carriers so that the optical field suffers optical loss, which leads to the increase of threshold current.

Figure 12 shows the relationship between Δn and dp. As shown in Figure 12, Δn becomes smaller as dp becomes large. To confine the optical field moderately inside the window region in the lateral direction, the value of Δn is set to be around 6 x 10^'3. In the second embodiment, in order to obtain the value of Δn to be 6 x 10^"3, dp is set to be 0.1 mm.

According to the structure of the second embodiment shown in Figure 10, the current injected through the p-type Al₀₂Ga₀₈N/ln₀ o_ιGa₀₉₆N superlattice fourth cladding layer 345 is confined within the window 340 and a laser oscillation of the 450 nm band is generated in the quantum well active layer 320 located under the window. As before, the use of an AIGaN constituent layer in the superlattice layer helps to confine the optical field strongly in the transverse direction. The strong optical confinement in the active layer leads to a low threshold current laser diode. Figure 13 shows the light-current characteristics of a laser diode constructed in accordance with the second embodiment with a single quantum well. The laser diode is driven with a pulsed current with a duty cycle of 1 %. The threshold current density is shown to be 4.0 kA/cm2, which is about half the threshold current density of a laser diode with only GaN cladding layers. In the event a multiple quantum well active layer is implemented, slight variations occur in the relationship between Δn and dp. As with Figure 12, the relationship shown in Figure 14 shows that Δn becomes smaller as dp becomes large. In order to confine the optical field moderately into inside the window region in the lateral direction, the value of Δn is to be around 6 x 10^'3. In the second embodiment, in order to obtain the value of Δn to be 6 x 10^"3, dp is set to be 0.08 mm.

Figure 15 shows the light-current characteristics of the laser diode of the second embodiment, but using a multiple quantum well active layer. This implementation results in improved (i.e., increased) optical confinement in the active layer in the transverse direction beyond what is possible with single quantum well active layer. The multiple quantum well impiementation thus permits further reduction of the threshold current. The plot of Figure 15 shows a laser diode driven with a pulsed current with a duty cycle of 1 %. The threshold current density is about 3.8 kA/cm2 which is also about half of the threshold current density of a laser diode with GaN cladding layers.

Referring next to Figure 16, a third embodiment of the present invention may be better appreciated. The embodiment shown in Figure 16 differs from that shown in Figure 3 in that it uses a quaternary material system ratherthan the ternary system of Figure 3. Thus, in the third embodiment of the invention, a semiconductor structure - which may be, for example, a laser diode -- comprises the following: on a GaN or other substrate 400, a cladding layer 405 of a first conduction type of In^.^Ga^AI^N material is formed. Thereafter, a superlattice second cladding layer 410 of the first conduction type is formed, comprising

material. The lattice constant of the ln₁._x2._y2Ga_x2Al_y2N material is selected to be larger than that of the ln₁._x1._y1Ga_x1Al_y1N material in the cladding layer, while the lattice constant of the ^lni-_x3-_y3Ga_x3Al_y3N material is selected to be smaller than that of In^.^Ga^AI^N material. A guide layer 415 of any suitable material may, in some embodiments, be formed about the superiattice layer 410. Thereafter, a quantum well active layer 420 is then formed, either of the single well or multiple well design, followed, in at least some implementations, by a guide layer 425. Then a superlattice third cladding layer 430 of an opposite conduction type is formed. The superlattice third cladding layer may, for example, comprise

where the lattice constant of the In^^Ga^AI^N is largerthan that of the ln,._x1._y1Ga_x1Al_y1N material and the lattice constant of said

Then a fourth cladding layer 435 of a conduction type opposite to the first cladding layer is formed, typically of

material. The values x1 , x2, x3, x4, x5, and x6 define the GaN mole fraction and y1 , y2, y3, y4, y5, and y6 define the AIN mole fraction.

With the foregoing structure, in the superlattice layerthe ln,._x2._y2Ga_x2Al_y2N layer is under compressive stress, while the ln,._x3._y3Ga_x3Al_y3N layer is under tensile stress so that the stresses compensate each other at the interface of the ln₁._x2._y2Ga_x2Al_y2N layer and the ln,._x3._y3Ga_x3Al_y3N layer. Similarly, in the second superlattice layer the ln,._x4._y4Ga_x4Al_y4N layer suffers compressive stress, and the In^.^Ga^Al_ysN layer suffers tensile stress so that the stress is able to compensate each other at the interface of the ln,._x4._y4Ga_x4Al_y4N layer and the ln,.,.₅._y5Ga,.₅Al_y5N layer. As with the prior embodiments, the InGaAIN superiattice layer can be designed to confine the optical field within the active layer better than if GaN is used for the cladding layer. By increasing the optical confinement within the active layer in the transverse direction, the threshold current of the device can be reduced. Further, the InGaAIN superiattice layer can be designed not to absorb the lasing light from the active layer. Therefore, low threshold current and low defect density laser diodes are obtained.

Referring next to Figure 17, a fourth embodiment of the present invention may be better appreciated. The fourth embodiment uses the quaternary material system of the third embodiment, but with what is otherwise the structure of the second embodiment shown in Figures 10 and 11A-1 1 C. On a GaN or other substrate 500, a cladding layer 505 of a first conduction type of In^.^Ga^AI^N material is formed. Thereafter, a superlattice second cladding layer 510 of the first conduction type is formed, comprising ln₁-_x2._y2Ga_x2Al_y2N and In^^Ga^AI^N material. The lattice constant of the ln₁._x2._y2Ga_x2Al_y2N material is selected to be larger than that of the in,.,.,. _y1Ga_x1Al_y1N material in the cladding layer, while the lattice constant of the ln₁._x3- _y3Ga_x3Al_y3N material is selected to be smaller than that of In^.^Ga^AI^N material. A guide layer 515 of any suitable material may, in some embodiments, be formed about the superlattice layer 510. Thereafter, a quantum well active layer 520 is then formed, either of the single well or multiple well design, followed, in at least some implementations, by another guide layer 525 of the opposite conduction type, as with the earlier-described embodiments. Then a superlattice third cladding layer 530 of an opposite conduction type is formed. The superlattice third cladding layer may, for example, comprise In^^Ga^AI^N and In_L^_ysGa^Al_ysN, where the lattice constant of the In^-^Ga^AI^N is larger than that of the In^.^Ga^AI^N material and the lattice constant of said ln-._x5._y5Ga_x5Al_y5N is smaller than that of In^.^Ga^AI^N. The superlattice layer 530 may be on the order of twenty-five pairs of constituent layers. Next, a current blocking layer 532 is formed of a p-type AI_{0 22}Ga_{0 78}N material on the order of 10θA thick. A stripe-like window 534 is then formed in the current-blocking layer 532 to expose the superlattice layer 530. A superiattice fourth cladding layer 535 is then formed, of the same material as the superlattice layer 330 but with on the order of 200 pairs of constituent layers Finally, a fifth cladding layer 540 is formed in the same manner as described previously Likewise a pair of electrodes may be formed in a conventional manner Referring next to Figure 18, a heterojunction field effect transistor formed from the method and structure of the present invention is shown On a GaN substrate 600, a i-GaN cladding layer 605 of about 0 5 μm thickness is formed, above which is formed an n-GaN channel layer 610 about 10θA thick. Above that, a superiattice layer 615 is formed from on the order of five pairs of constituent layers, each on the order of 2θA thick, of AI₀₂Ga₀₈N (6 layers)/ n-type ln₀₀₄Ga₀₉₆N(5 layers). Source, drain and gate electrodes 620, 625 and 630 may then be formed on the superlattice layer 615. Group-Ill nitride materials, especially GaN and AIN, are promising materials for hard electronic devices which can operate under high-power and high- temperature conditions, since GaN and AIN have wider band gaps (3.5 eV for GaN, 6.2 eV for AIN), resulting in higher breakdown electronic field, and higher saturation velocity. This is compared to the band gaps of AIAs, GaAs, and Si: 2.16 eV, 1.42 eV, and 1.12 eV, respectively. Therefore, field effect transistors (FETs) using AIGaN/GaN materials are widely researched for application in the field of microwave power transistors.

Referring next to Figure 19, a heterojunction bipolar transistor formed in accordance with the present invention is shown. A GaN substrate 650 provides the foundation on which a superiattice collector layer 655 is formed. A p-type GaN base layer 660 is then formed, after which a superlattice emitter layer 665 is then formed. Collector, base and emitter electrodes 670, 675 and 680 may be formed thereafter. Figure 19 shows an embodiment of a heterojunction bipolar transistor(HBT). On the GaN substrate 650, on the order of one hundred pairs of n-type 2θA thick AI₀₂Ga₀₈N(101 layers )/n-type 2θA thick ln₀₀₄Ga₀₉₆N(100 layers) superiattice collector layer are formed, followed by a 50 nm thick p-type GaN base layer. Then, on the order of 80 pairs of n-type 2θA thick AI₀₂Ga₀₈N(81 layers )/n-type 2θA thick ln₀₀₄Ga₀₉₆N (80 layers) superlattice layer are formed as an emitter. The stress between the AIGaN layer and InGaN layer compensate each other at the interface, so that generation of the defects can be reduced, leading to a high quality heterojunction of AIGaN/GaN. The band gap of the AIGaN/lnGaN superlattice emitter layer is larger than that of GaN base layer so that holes generated in the p-type base layer are well confined in the base layer because of the larger valence band discontinuity between GaN and the AIGaN/lnGaN superlattice layer compared to that in the GaN homojunction bipolar transistor. Therefore, large current amplification between base current and collector current is obtained. Moreover, as mentioned above, the bandgap of the AIGaN/lnGaN superlattice layer and the GaN layer are large so that the transistor can be used as a high-temperature transistor. It will further be appreciated that, while the embodiment described above uses a superlattice layer for both the emitter and the collector, it is not necessary in all instances that both layers be of the superlattice type, and a single superlattice layer may be implemented as either the collector or the emitter. Referring next to Figure 20, an embodiment of the present invention implemented as a photodiode can be better appreciated. On an n-type GaN substrate 700, an n-type GaN first cladding layer 705 is formed on the order of 0.5μm thick, followed by an n-type superlattice second cladding layer 710 having on the order of 200 pairs of constituent layers. Then, an ln₀₀₂Ga₀₉₈N guide layer 715 on the order of 35A thick is formed, followed by a quantum well active layer 720. The active layer typically comprises ln_{0 15}Ga₀₈₅N. Thereafter, an ln₀₀₃Ga_{o g7}N guide layer 325 on the order of 35A thick may be implemented in some embodiments.

Then, in a significant difference from the first embodiment, a p-type superlattice third cladding layer 330 is formed. However, the layer 330 typically comprises only on the order of 25 pairs of constituent layers, each on the order of 2θA thick. Then, a p-type Al_{0 2}_>Ga_{0 78}N current-blocking layer 335 is formed, on the order of 10OA thick. A stripe-like window 340 is then formed in the current-blocking layer 335 to expose a portion of the third cladding layer 330. Electrodes may be formed in a conventional manner.

As with the first embodiment, several different combinations of materials may be used to fabricate the superlattice layers 710 and 730, and may include AI₀₂Ga₀₈N/ln_{0 04}Ga₀₉₆N, AI₀₂Ga₀₈N/ln₀₂AI₀₈N, or ln₀₀₄Ga₀₉₆N/ln_{0 13}AI₀₈₇N. The operation of these materials is as discussed in connection with the second embodiment, except for the removal of the upper cladding layer and the third superlattice layer. In a presently preferred arrangement, the window 340 may be shaped as a small outer ring.

Referring next to Figure 21 , an embodiment of the semiconductor device of the present invention is shown implemented as a heterojunction phototransistor. The device is particularly suited to operation in the ultraviolet (UV) range, although other frequencies, including blue light, may also be detected with only slight modification.

GaN and AIGaN are attractive as the materials for photodetectors in ultraviolet(UV) range, since GaN and AIN have a wide band gap (3.5 eV for GaN which corresponds to a light wavelength of 200 nm, 6.2 eV for AIN which corresponds to a light wavelenght of 350 nm). Due to the direct band gap and the availability of AIGaN in the entire AIN alloy composition range, AIGaN/GaN based UV photo detectors have the advantage of both high quantum efficiency and tunabi ty of high cut-off wavelength However, as with the embodiments described previously, the lattice constant of AIGaN is different from GaN, so that defects tend to occur, leading to increased leakage current

As discussed hereinabove, the strain compensated superlattice structure of the present invention can reduce defects which occur in the prior art by compensating for 2Z the stress at the interface of the AIGaN and the InGaN layer, while keeping the effective band gap of the superlattice layer larger than GaN by itself. Referring still to the heterojunction photo transistor(HPT) shown in Figure 21 , on a GaN substrate 800, a superiattice collector layer 805 is formed, comprising on the order of 120 pairs of n-type 2θA thick AI₀₂Ga₀₈N(101 layers) and n-type 2θA thick ln₀₀₄Ga₀₉₆N(100 layers) constituent layers 805A and 805B. Next, a p-type GaN base layer 820 on the order of 200 nm thick is formed, followed by the formation of a superlattice emitter layer 825 comprising on the order of 80 pairs of constituent layers of n-type 20A thick AI₀₂Ga₀₈N(81 layers) and n-type 2 nm thick ln₀₀₄Ga_α96N (80 layers). As with the superlattice layer of each embodiment, the stress between the constituent layers, in this case the AIGaN layer and and the InGaN layer, compensate each other at their interface. These strain-compensated layers significantly reduce defect generation, yielding a high quality heterojunction of AIGaN/GaN. Electrodes 830 and 835 are formed in a conventional manner. As noted above, the band gap of the AIGaN/lnGaN superlattice emitter layer is larger than that of GaN base layer. In operation, light impinges from the emitter side. If the photon energy of the impinging light is larger than the band gap energy of GaN base layer, but smaller than the band gap energy of the AIGaN/lnGaN superlattice emitter layer, the impinging light is transparent to the emitter layer so that the light is absorbed in the GaN base layer and generates electron and hole pairs. The holes generated by the optical absorption in the p-type GaN base layer is better confined within the base layer because a larger valence band discontinuity exists between the GaN layer and the AIGaN/lnGaN superlattice layer than would exist in the case of a GaN homojunction photo transistor. This, in turn, leads to larger emitter currents and better neutralization in the base region than would be the case of for a conventional homojunction photo transistor. Therefore, UV photo detectors can be obtained with high quantum efficiency and high sensitivity, which means high conversion efficiency from the input light to the collector current. In the event it is desired to detect other frequencies, for example blue light, the GaN base layer may simply be replaced by InGaN.

Having fully described several embodiments of the invention, it will be apparent to those skilled in the that numerous alternatives and equivalents exist which do not depart from the invention. As a result, the scope of the invention is not to be limited by the foregoing description, but only by the appended claims.

Claims

We claim:

1. A semiconductor structure comprising a substrate of a first conductivity, and a first superlattice layer comprising a plurality of pairs of first and second constitutent layers, the first constituent layer comprising a material under tensile stress and the second constituent layer comprising a material under compressive stress, the compressive and tensile stresses each compensating the other at the interface therebetween.

2. The semiconductor structure of claim 1 further comprising an active layer formed above the first superlattice layer, the superiattice layer having a conductivity of a first type, and a second superiattice layer of a conductivity type complementary to the first type comprising a plurality of pairs of third and fourth constitutent layers, the third constituent layer comprising a material under tensile stress and the fourth constituent layer comprising a material under compressive stress, the compressive and tensile stresses each compensating the other at the interface therebetween.

3. The semiconductor structure of claim 1 wherein the substrate is GaN, the semiconductor structure further comprising a first i-GaN cladding layer formed between the substrate and the first superlattice layer, and an n-GaN channel layer formed between the i-GaN cladding layer and the first superlattice layer.

4. The semiconductor structure of claim 1 further including a base layer and an emitter layer and wherein the first superlattice layer comprises a collector layer.

5. The semiconductor structure of claim 1 further comprising a first cladding layer formed above and having the same conductivity type as the substrate, a second cladding layer of the same conductivity type as the substrate, an active layer, and wherein the first superlattice layer forms a third cladding layer and is of a conductivity type complementary to the conductivity type of the substrate, a blocking layer formed above the superlattice third cladding layer and having a window therein which exposes a portion of the superiattice third cladding layer.

6. The semiconductor structure of ciaiml wherein the substrate is of GaN and the first superiattice layer forms a collector layer, and further including a base layer and a superlattice emitter layer comprised of a plurality of pairs of strain-compensated constituent layers.

7. A method of fabricating a semiconductor structure comprising providing a substrate of a first conductivity type, forming a first constituent layer of less than critical thickness, the first layer being under tensile stress of a predetermined magnitude, forming a second constituent layer above the first constituent layer, the second layer being under compressive stress of approximately the same predetermined magnitude as the tensile stress of the first constituent layer, such that the tensile and compressive stresses in the constituent layers compensate for one another.

8. A transistor device comprising on a semi-insulating layer of In^.^Ga^AI^N layer, an n-type In^.^Ga^AI^N conductive channel layer, an n-type of a superlattice layer consisting of ln-._x2. _y2Ga_x2Al_y2N and In^_yaGa^Al_yaN, where the lattice constant of said \n _x2._y2Ga_x2Al_y2N is larger than that of said In^.^Ga^AI^N, and the lattice constant of said \n _x3. _y3Ga_x3Al_y3N is smallerthan thatof In^-.^Ga^AI^N, are successively formed one upon each other, whrerein x1 , x2, and x3 define the GaN mole fraction and y1 , y2, and y3 define the AIN mole fraction and the effective bandgap of said superlattice layer is larger than that of ln-._x1._y1Ga_x1Al_y1N layer.

9. A transistor device comprising a first conductivity type of superlattice collector layer consisting of In^.^Ga^AI^N and ln₁._x2._y2Ga_x2Al_y2N, an opposite conductivity type of ln-._x3. _y3Ga_x3Al_y3N base layer, a first conductivity type of type of superlattice ln,._x1._y1Ga_x1Al_y1N and ln₁._x2._y2Ga_x2Al_y2N emitter layer, wherein the lattice constant of said ln,._x2._y2Ga_x2Al_y2N is larger than that of said ln₁._x3._y3Ga_x3Al_y3N base layer, and the lattice constant of said lⁿι-χ₂-_y2G^a ₂ ^Aly₂ ^{N is} smaller than that of said In^^Ga^AI^N base layer, all successively formed one upon the other, wherein the bandgap of said lⁿi-_x3-_y_G^a _x3Al_y3N base layer is smaller than the effective bandgap of ln,._x1._y,Ga_x1Al_y,N and ln,._x2._y2Ga_x2Al_y2N superlattice layers, and x1 , x2, and x3 define the GaN mole fraction and y1 , y2, and y3 define the AIN mole fraction

10. A semiconductor laser diode comprises: on a certain conduction type of In^.^Ga^AI^N, a certain conduction type 5 of a superiattice layer consisting of ln₁._x2._y2Ga_x2Al_y2N and In^^Ga^AI^N, whrere the lattice constant of said In^._y-.Ga_j.-.A._y-.N is larger than that of said ln₁._x1_ _y1Ga_x1Al_y1N, and the lattice constant of said

is smaller than that of In^.^Ga^AI^N, an active layer, an opposite conduction type of a superiattice layer consisting of

where the lattice constant

10 of said In^^Ga^AI^N is larger than that of said materiall , and the lattice constant of said In^_ysGa^Al_ysN is smaller than that of In^.^Ga^AI^N, an opposite conduction type of In^_yeGa^Al_yeN are successively formed one upon each other, whrerein x1 , x2, x3, x4, x5, and x6 define the GaN mole fraction and y1 , y2, y3, y4, y5, and y6 define the AIN mole fraction.

15

1 1. A semiconductor laser diode comprises: a certain conduction type of GaN first cladding layer, an

_xiN superlattice second cladding layer of said certain conduction type, an In^Ga,. _xaN single quantum well active layer, an opposite conduction type of an AI^Ga^ 20

superlattice third cladding layer, an opposite conduction type of GaN fourth cladding layer, all successively formed one upon each other, wherein xal defines the AIN mole fraction, xi and xa define the InN mole fraction, and xi and xa have a relationship of xa > xi.

25 12. A semiconductor laser diode according to claim 2, wherein an AI^Ga^N current blocking layer having a window region is formed in the said AI^Ga^

superiattice third cladding layer, and having an opposite conduction type of said

third superlattice layer which contains said AI^Ga^N current blocking layer, wherein xb defines the AIN mole fraction, and xb

30 and xal have a relationship of xb > xal.

13. A semiconductor laser diode comprises: a certain conduction type of GaN first cladding layer, an Al_xalGa,._xalN/ln_xjGaι_. _XIN superlattice second layer of said certain conduction type, an InGaN multiple 35 quantum well active layer, an opposite conduction type of an AI^Ga^_atN/ln^Ga^.N superiattice third layer, an opposite conduction type of GaN fourth cladding layer, all successively formed one upon each other, wherein xal defines the AIN mole fraction, xi defines the InN mole fraction.