Polarized light emission from cubic GaN quantum wires.

Abstract

The semiconductor structure includes a matrix having a first cubic III-nitride having a first bandgap and a second cubic III-nitride having a second bandgap and forming a region embedded within the matrix, the second cubic III-nitride including an alloy material that reduces the second bandgap relative to the first bandgap, quantum wires defined by portions within the region embedded within the matrix, the portions forming one-dimensional charge carrier confinement channels, and the quantum wires operable to exhibit optically polarized luminescence emission.

Description

[Technical field]

[0001] FIELD OF THE DISCLOSURE This disclosure relates generally to cubic III-nitride semiconductors that exhibit luminescence emission, and more particularly, but not exclusively, to quantum wires in cubic III-nitride semiconductor structures. [Background technology]

[0002] Cubic / zinc-blende phase GaN-related structures are known to be promising alternatives to the more widely known wurtzite / hexagonal GaN semiconductors and can be used to improve the efficiency of long wavelength (including green amber and red) LEDs. Semiconductors such as those containing GaN are known to give rise to photoluminescent and electroluminescent properties, and such semiconductors can be used in LED or photodiode devices. III-nitride semiconductors generally offer a wide range of optoelectronic applications, such as LEDs and laser diodes that emit in the blue, green, and red spectral regions. However, conventional LED light sources, when used in display technologies, require separate polarizing filters, which inevitably reduce the light transmittance and decrease the efficiency of the system. It would therefore be advantageous to have a polarized light source for use in LD or LCD displays, or other such devices that require a polarized light source.

[0003] Furthermore, semiconductors can be arranged to create so-called "quantum wires" (Q-wires) by arranging regions of larger and (relatively) smaller band gap semiconductors. A quantum wire can generally be thought of as a region of material in which charge carriers, i.e., electrons and / or holes, are confined in two orthogonal dimensions but are free to move in a third orthogonal dimension. This confinement of the charge carriers leads to the formation of a series of quantized (or substantially quantized) energy states within the quantum wire. Summary of the Invention [Problem to be solved by the invention]

[0004] Known techniques have observed that quantum wires may exhibit light emission at low temperatures, but as described below, the light emitting properties arising from various cubic III-nitride semiconductors are generally unknown or not yet fully understood in the art.

[0005] The present disclosure is directed to solving the problem of reliably producing polarized light sources from semiconductor devices, and in particular cubic III-nitride semiconductor devices. [Means for solving the problem]

[0006] Quantum wires are one-dimensional (1D) structures that exhibit quantum confinement of charge carriers in two dimensions (charge carriers can only move freely in 1D). This is in contrast to quantum wells, which are two-dimensional (2D) structures that exhibit quantum confinement of charge carriers in one dimension (charge carriers can move freely in 2D), and quantum dots, which are zero-dimensional structures (0D) in which charge carriers are confined in all possible dimensions.

[0007] In the following disclosure, it is demonstrated that it is possible to obtain polarized emission from cubic III-nitride based quantum wires in quantum wells. If this can be achieved in an LED structure, this provides an avenue for improved efficiency in displays and other applications. Currently, standard (non-polarized) LEDs are used as backlight sources for LCD displays. As these displays rely on polarized light, currently this is achieved by placing a polarizing film in front of the non-polarized light source. This means that 50% of the light emitted from a standard backlight source is wasted. If a polarized light source could be achieved, this would mean that more of the emitted light energy could be advantageously used in the display, thus resulting in a more efficient display. This would increase the battery life of portable devices. Thus, polarized emission combined with the potential for improved efficiency from cubic GaN LEDs (especially at longer wavelengths such as green, amber, and / or red) offers significant advantages in display applications. There is also a wide range of other applications in which polarized light sources are used, and thus a range of other suitable applications that would benefit from polarized cubic GaN LEDs.

[0008] According to one aspect of the present disclosure, there is provided a semiconductor structure including a matrix including a first cubic III-nitride having a first bandgap, and a second cubic III-nitride having a second bandgap forming a region embedded within the matrix, the second cubic III-nitride including an alloy material that reduces the second bandgap relative to the first bandgap, the semiconductor structure further including quantum wires defined by portions forming one-dimensional charge carrier confinement channels within the region embedded within the matrix, the quantum wires being operable to exhibit optically polarized luminescence emission.

[0009] The portion of the region embedded within the matrix that defines the quantum wire may have a portion of the alloy material that is further increased as compared to the remainder of the embedded region.

[0010] It will be understood by those skilled in the art that the alloy material may include one or more elements and may be commensurate with the structure of the matrix (i.e., cubic, perhaps having the same or substantially similar lattice constant). The alloy material may include an enrichment or actual depletion of one or more elements compared to the elements in the matrix. For example, the matrix may include a ternary material (e.g., AlGaN) and the buried region of low bandgap may include a binary alloy (GaN). Thus, the alloy material is enriched in GaN compared to the matrix.

[0011] Moreover, the alloying material is preferably isovalent to the elements that make up the matrix. It will be further understood by those skilled in the art that the alloying material may include any increased percentage of a particular element. By way of example only, the matrix may include GaN and the embedded region within the matrix may include InGaN, where In is a minimum of 1% and a maximum of 80% percentage, but is not limited to being within this range and may be as high as 99%. In this case, the alloying material is indium, provided as a concentrate in the embedded region compared to the matrix. Alternatively, the matrix may include AlN and the embedded region may include a suitable matrix material alloyed to have a lower band gap, such as AlGaN or BGaN.

[0012] It is further understood that the cubic III-nitride matrix generally includes a suitable combination consisting of (Al)(Ga)(In)(B)N in a cubic crystal structure. The buried regions within the matrix include alloy materials and therefore include suitable combinations of (Al)(Ga)(In)(B)N having different ratios of elements relative to the matrix, where the different ratios are such that the second band gap is lower than the first band gap. It is understood that the different ratios that result in a reduction in the band gap also apply to the portions within the buried regions that define the quantum wires. The following examples will discuss various suitable combinations that result in a reduction in the band gap.

[0013] The above semiconductor structure may include any of the following features, either alone or in combination.

[0014] The structure may further include a quantum well defined by a region embedded within the matrix, the region forming a layer embedded within the matrix.

[0015] In accordance with the above, the matrix may include cubic gallium nitride and the region embedded in the matrix may include indium-rich cubic gallium nitride. The indium reduces the band gap of the embedded region compared to the band gap of the matrix. The quantum wire defined by the portion within the embedded region may include a region including a more enriched proportion of indium relative to the embedded region. This more indium-rich region forms a charge carrier confinement channel in which charge carriers are confined in two dimensions, and the quantum wire is operable to exhibit optically polarized luminescent emission (e.g., photoluminescence or electroluminescence). Furthermore, the matrix may be formed in at least two layers, in which the embedded region is embedded, the embedded region being a layer defining a quantum well.

[0016] The charge carrier confinement channel confines the carriers so that they are free to move only in one dimension (1D), and the optically polarized light may be linearly polarized, where the plane of linear polarization may be in the same direction as the direction of the quantum wire. Advantageously, the confinement of the charge carriers, which may be electrons (negative charge) or holes (positive charge carriers), in one dimension gives rise to polarized light emission as a result of them being confined or substantially confined to move in one dimension.

[0017] It is understood that "cubic" refers to a Face Centered Cubic (FCC) crystal structure, in other words, a zinc blende crystal structure. The crystal structure can be formed by combining any group III and group V elements, preferably any group III element and nitrogen (i.e., group III nitrides). These group III elements are gallium, indium, aluminum, or boron, and can be combined to form binary, ternary, quaternary, or quinary alloys (e.g., any combination of (Al)(Ga)(In)(B)N). The alloy material can generally be any suitable group III element or elements, preferably such that the alloying / enriched element is not so large as to destroy the stability of the lattice. Furthermore, as mentioned above, the alloy material generally includes at least one group III element that provides a reduced second band gap compared to the matrix (having a first band gap).

[0018] The semiconductor structure may further include a quantum well defined by a region buried within the matrix, the buried region forming a layer buried within the matrix.

[0019] The buried regions may form continuous channels or inclusions within the matrix. Thus, the buried layer defining the quantum well may be a layer sandwiched between separate layers of matrix. Furthermore, the orientation of the plane of the quantum well / buried layer may be oriented in the same direction as the orientation of the quantum wire.

[0020] The alloy material of the semiconductor structure may be indium. In a preferred embodiment, the percentage of indium in the embedded region (compared to GaN in the matrix) may be between about 0.01 and 0.40 (1% to 40%).

[0021] The portion of the buried region that defines the quantum wire may include a localized increase in the concentration of the alloy material, i.e., the quantum wire may be composed of an increased proportion of the alloy material as compared to the proportion of the alloy material in the remainder of the buried region composition.

[0022] Additionally, the localized increase in concentration of the alloying material may be localized to the intersection between the stacking fault and the quantum well of the semiconductor structure. In some instances, the percentage concentration of the alloying material (which may be indium) may be about twice the average percentage of the alloying material in the buried region / layer.

[0023] The portion of the buried region that defines the quantum wire may also / alternatively include a local variation in the width of the buried layer that defines the quantum well, it being understood that said variation in the width of the buried layer causes confinement of charge carriers such that they can only move significantly in one dimension, resulting in a quantum wire that is operable to emit polarized luminescence.

[0024] The size of the local variations is preferably greater than 2 nm.

[0025] Depending on the initial quantum well width, a change in well width beyond about 2 and 3 nm results in an energy difference between the electronic ground states that exceeds the amount of energy required to thermally populate the higher electronic states, and thus the electrons are advantageously well confined such that thermal excitation is low and the intensity of the emission is more temperature independent.

[0026] In a preferred embodiment, the width of the buried layer defining the quantum well may vary between 2 nm and 14 nm in width, and the average width of the quantum well may vary between about 4 nm and 8 nm.

[0027] Furthermore, the charge carriers confined in the quantum wire's carrier confinement channel are preferably electrons, i.e., electrons are more likely to be strongly confined to move in only one dimension within the well width variations. Holes may not be strongly confined due to well width variations.

[0028] Alternatively, the portion of the buried region that defines the quantum wire may be defined by a channel of the buried region that includes the alloy material, which extends through the matrix, in other words, the quantum wire may be defined by a portion of the buried region that has the alloy material, which forms a physical wire-like structure, or passage, or channel, through or within the matrix.

[0029] In general, the percentage of alloy material in the buried region may be greater than about 20% (especially when the buried region forms a layer buried in the matrix and defines a quantum well). In some examples, the matrix is ​​GaN, the region buried in the matrix is ​​InGaN, and the alloy material is In. As noted above, the alloy material may include an enrichment or actual depletion of one or more elements compared to the elements in the matrix. Furthermore, it is understood that only certain alloy materials are suitable depending on the matrix to reduce the band gap of the buried region in the matrix. For example, alloying GaN with In reduces the band gap, while alloying GaN with Al increases the band gap. Further examples are provided in the following discussion.

[0030] Advantageously, a higher indium content in the quantum wire reduces the temperature dependence of the polarization of the emission, i.e., the energy division between a first state, which may be the ground state, and a second state, which may be the excited state, is enlarged, thus reducing thermal excitation and reducing the thermal occupation of the excited states.

[0031] More generally, the carrier confinement channel of a quantum wire may have a first electronic state and a second electronic state, the difference in energy between said states being greater than a characteristic thermal energy, thereby reducing the likelihood of thermally induced transitions between the states, which may exceed about 25 meV in certain instances.

[0032] Also advantageously, the energy difference between these states (e.g., electronic ground state and excited state) of more than 25 meV significantly reduces thermal excitation and thus temperature dependence of the degree of polarization. It is understood that 26 meV is a typical thermal energy at room temperature and is related to the characteristic thermal energy. Alternatively, the first and second states can be the ground states of a quantum well (theoretically having infinite length) and a local fluctuation that behaves as a second quantum well with a finite length (thus acting as a charge carrier confinement channel).

[0033] In preferred embodiments, the average cross-sectional dimension of the quantum wires is less than about 10 nm and greater than about 2 nm. A quantum wire dimension of about 5 nm may be preferred. The average dimension or average cross-sectional dimension may refer to the width of one lateral dimension of the wire, the average of the two lateral widths of the wire, the length of the diagonal between the two corners of a quadrilateral defined by the cross-section of the wire, or any suitable measure of the size of the cross-section of the quantum wire. It is understood that the length of the wire will be longer than the average cross-sectional dimension of the wire.

[0034] According to the above examples, especially the preferred alloy composition and quantum well dimensions, the present disclosure also seeks to address and overcome the problem of producing a semiconductor structure or device in which the degree of polarization and / or emission intensity is substantially independent of temperature. The cross-sectional dimensions of the quantum wires may affect the two-dimensional confinement of holes; as the quantum wires become larger, the thermal excitation from the quantum wires may decrease because of the larger energy difference between the carrier ground state and the quantum well. However, the larger the quantum wire, the higher the thermal occupancy of the excited states of the quantum wires may be. This is because the energy separation between the ground state and the excited state of the wire may be lower. Thus, the increased thermal occupancy of the excited states may result in an undesirable polarization that is temperature dependent.

[0035] Thus, to ensure temperature independence of the polarization, smaller quantum wires may be preferred in some instances (e.g., when the quantum wire is the result of the intersection of stacking faults and quantum wells). In other instances, a balance of size must be struck to balance the competing effects of reduced thermal excitation from the quantum wire and increased thermal occupation of excited states.

[0036] However, a wider quantum wire reduces thermal excitation from the quantum wire due to a larger energy difference between the ground state of the carriers and the quantum well.However, a wider quantum wire also lowers the energy separation between the ground and excited states of the wire, leading to a higher thermal occupancy of the excited states in the wire and a temperature-dependent polarization.

[0037] According to a further related aspect of the present invention, there is provided a semiconductor device incorporating a semiconductor structure according to any of the preceding aspects and examples, wherein the semiconductor device is selected from the group including: a vertical cavity surface emitting laser, VCSEL; a laser; a sensor.

[0038] The semiconductor device may further include a substrate comprising cubic silicon carbide (3C SiC) or a suitable cubic semiconductor such as silicon, GaAs, or, in particular, the substrate may be silicon carbide on Si, which may be 3C-SiC on Si.

[0039] The device may further include an electron-rich layer of matrix material disposed on the surface of the substrate; an optically active region defined by the matrix and regions embedded within the matrix, disposed on the surface of the electron-rich layer; and an electron-deficient layer of matrix material disposed on the surface of the optically active region. This structure may generally represent an LED, and may include other elements or materials for strain relief, dislocation or point defect filtering, or control of carrier transport, as known to those skilled in the art. It is further understood that the above semiconductor structures in conjunction with LEDs and the like define a diode (i.e., a pn junction) that induces optical activity / luminescence from the optically active region.

[0040] Generally, electron-deficient layers (or hole-rich layers) are associated with p-type semiconductor layers, and electron-rich layers are associated with n-type semiconductor layers.

[0041] It is understood that the optically active region is defined by a quantum well, i.e., two layers of matrix material embedded with a layer of lower band gap material, and a quantum wire contained within the quantum well that exhibits polarized emission. It is further understood that multiple optically active layers / quantum wells may be present in a device such as an LED.

[0042] Furthermore, the substrate may include multiple layers, such as ceramic or single crystal silicon, on which is a silicon carbide layer, which in a preferred example may be cubic (3C) silicon carbide. Advantageously, silicon carbide on a silicon substrate, which has a (cubic) crystal structure compatible with gallium nitride, may allow for the direct fabrication of GaN structures on the substrate. The diameter of the substrate may be 50 mm, 100 mm, 150 mm, 200 mm, 300 mm, or any size consistent with the mass production capabilities available in conventional wafer processing factories.

[0043] Additionally, the semiconductor structure may further include optical confinement layers disposed on either side of the optically active region, which is commonly associated with laser diode device structures.

[0044] According to another related aspect of the invention, there is provided a method of fabricating a semiconductor device, the method including forming a matrix including a first cubic III-nitride having a first bandgap, forming a second cubic III-nitride having a second bandgap and forming a region embedded within the matrix, the second cubic III-nitride including an alloy material that reduces the second bandgap relative to the first bandgap, and forming a portion within the region embedded within the matrix that defines a quantum wire, the portion forming a one-dimensional charge carrier confinement channel.

[0045] The method may further include forming the region buried within the matrix as a buried layer within the matrix that defines a quantum well.

[0046] According to the above aspects, the semiconductor device formed may be selected from the group including: light emitting diodes (LEDs); vertical cavity surface emitting lasers, VCSELs; lasers; and sensors.

[0047] The method of manufacturing a semiconductor device may further include: forming a substrate comprising cubic silicon carbide; forming an electron-rich layer of a matrix material disposed on a surface of the substrate; forming an optically active region defined by the matrix and a region embedded within the matrix disposed on a surface of the electron-rich layer; and forming an electron-deficient layer of a matrix material disposed on a surface of the optically active region.

[0048] The method of manufacturing a semiconductor device may further include forming an optical confinement layer disposed on each of the first and second surfaces of the optically active region.

[0049] Further advantages related to the light emitting properties (i.e., photoluminescence or electroluminescence) resulting from the following structures will become apparent in the following description. For example, light emitted from the surface of the described structures / devices is polarized, which allows LEDs made from the semiconductor structures described herein to be used in a conventional (surface emitting) format, thus eliminating the need to manufacture edge emitting LED devices.

[0050] These and other aspects of the present disclosure will now be further described, by way of example only, with reference to the accompanying figures. [Brief description of the drawings]

[0051] [Figure 1] FIG. 1 shows quantum wells and quantum wires, respectively, formed from a low bandgap material enveloped in a high bandgap material. [Diagram 2] FIG. 2 shows a transmission electron microscope (TEM) image with an energy dispersive X-ray (EDX) overlay of GaN and indium gallium nitride (InGaN) structures as shown in the general example of FIG. [Diagram 3] FIG. 3 shows a model light emitting diode (LED) structure including GaN and InGaN quantum wires such as those shown in FIG. [Figure 4] FIG. 4 shows a laser diode structure including GaN and InGaN quantum wires as shown in FIG. [Figure 5a] FIG. 5a shows a schematic energy potential diagram of a quantum well with width variations forming a quantum wire. [Figure 5b] FIG. 5b shows a transmission electron microscope image of a zinc-blende InGaN / GaN quantum well sample with 8 nm wells, showing the variation in well width. [Figure 6] FIG. 6 shows a schematic energy potential of the conduction band of a quantum wire arising from the indium-rich region around the interaction of the stacking fault with the quantum well. [Figure 7]Figure 7 shows the solutions of the electron and hole wave functions solved for a model energy potential of an 8 nm quantum well with a 4 nm wide In-rich region. [Figure 8] FIG. 8 shows the simplified total energy potentials of the conduction and valence bands of an 8 nm quantum well with a 4 nm wide indium-rich region corresponding to the wavefunctions in FIG. [Figure 9] FIG. 9 shows tabulated results of the electronic properties of a set of quantum wires with various dimensions and indium fractions. [Figure 10] FIG. 10 is a schematic diagram of a quantum wire produced by etching a quantum well. [Figure 11] FIG. 11 shows a graph of the calculated hole ground state energies (squares) and second excited hole state energies (circles) of rectangular quantum wires created by etching InGaN / GaN quantum wells of different cross-sectional sizes (defined by the diagonal lengths) and indium contents. [Figure 12] FIG. 12 shows a graph of an experimental photoluminescence spectrum obtained at 10 Kelvin of a cubic GaN containing InGaN sample with five quantum wells, each 2 nm thick. [Figure 13] 13a and 13b show the temperature dependence of the normalized intensity and degree of polarization, respectively, corresponding to the peaks shown in FIG. [Figure 14] FIG. 14 shows a graph displaying four room temperature photoluminescence spectra of quantum wells of different widths grown from a GaN crystal. [Figure 15] FIG. 15 shows a graph displaying the photoluminescence (PL) spectrum (solid line) generated at 10 K of a 6 nm wide quantum well and the corresponding PL excitation (PLE) spectrum (dashed line) showing two distinct absorption edges. [Figure 16] FIG. 16 shows the photoluminescence time decay from a GaN / InGaN crystal at 10 K produced from a 2 nm wide quantum well. [Figure 17]FIG. 17 shows a graph displaying the luminescence decay times of four cubic InGaN / GaN quantum wells of different widths. [Figure 18] FIG. 18 shows the photoluminescence spectra of four cubic InGaN / GaN quantum wells with different widths. [Figure 19] FIG. 19 shows a calculated 2D map of the ratio of the degree of linear polarisation (DOLP) of the emission from the quantum well at 300K versus 10K for cubic InGaN / GaN quantum wells with varying indium content and well dimensions. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0052] The present disclosure and certain embodiments of the disclosure will now be described with reference to the following non-limiting examples.

[0053] FIG. 1 illustrates a quantum well 100 (Qwell) and a quantum wire 102 (Qwire), respectively, contained within a semiconductor structure. In general, a quantum well can be formed by sandwiching a low bandgap material between two high bandgap materials. In FIG. 1, the quantum well 100 is created by encasing a layer of a low bandgap material 104, such as gallium nitride (GaN), further including the alloying element / material indium, InGaN, between a high bandgap matrix 106, such as gallium nitride. The quantum well 100 is generally formed by the higher bandgap matrix material 106 embedding / surrounding a layer of a lower bandgap material 104. The higher bandgap matrix material surrounding the embedded region may alternatively be referred to as a barrier.

[0054] In general, any suitable group III element (e.g., indium when the matrix material 106 surrounding the quantum well is GaN) can be used to enrich the central layer 104 to create a quantum well in a GaN semiconductor. The larger In atoms have the effect of reducing the band gap of the central layer.

[0055] For example, other options for the quantum well layer 104 (buried region) and surrounding layer 106 (matrix) are: Ga-rich AlGaN (104) buried in Ga-deficient AlGaN (106); AlGaN (104) buried in BGaN (106); and Indium-rich InGaN (104) buried in Indium-deficient InGaN (106). The composition of the low band gap material (e.g., InGaN) used in the quantum well layer 104 can vary between 2% to 40% indium, and in some instances even more. Preferred examples of indium fractions for GaN semiconductors are detailed below.

[0056] Quantum wires can then be fabricated by removing lengths or portions of the low bandgap material 104 to create transverse channels 102 in which electrons are confined in two dimensions and can only move freely in one dimension.

[0057] Quantum wires are understood to conform to a variety of shapes and configurations and are not limited to the quantum wires shown in the figures, for example, there are other suitable quantum wire-like structures: truncated quantum wires, quantum dashes, stubby quantum wires, elongated quantum dots, or extended quantum dots.

[0058] FIG. 2 shows an experimental transmission electron microscope (TEM) image with an energy dispersive X-ray (EDX) overlay of GaN 106 and InGaN 104 structures as shown in the general example of FIG. 1. Quantum wells 200 containing indium (In) rich GaN 104 are aligned along the 0110 direction of the GaN crystal as shown on the axis of the graph. Quantum wires 202 are formed in the image shown where the quantum wells 200 intersect with pre-existing stacking faults (SFs) 204 in the underlying GaN crystal. The highlighted regions 202 in the EDX overlay indicate further enriched In regions. That is, at the intersection of the quantum wells 200 and the stacking faults 204, the InGaN quantum wells may be further enriched in indium, resulting in the quantum wires 202.

[0059] The GaN crystals shown here are cubic or zinc-blende GaN (zb-GaN). Typically, GaN crystal structures naturally adopt the hexagonal or wurtzite structure, as this is the thermodynamically more stable state. Zb-GaN is a metastable state of GaN, but can be fabricated under certain conditions. For example, the inventors have determined that zb-GaN can be grown on the

[0001] surface of silicon carbide on a silicon substrate using (for example) metal organic vapour phase epitaxy (MOVPE) (also known as metal organic chemical vapour deposition (MOCVD)). Alternatively, other growth techniques such as molecular beam epitaxy (MBE) or hydride vapour phase epitaxy (HVPE) can be used.

[0060] The TEM image in Figure 2 shows significant distortion of the quantum wells caused by underlying roughness of the crystals (separate from stacking faults). In this example, the quantum wells have an average width of 8 nm, but this width can vary between about 4 nm and 12 nm along the length of the quantum well. These width distortions occur in only one dimension (i.e., across the length of the wire, in the

[0001] direction as shown), potentially further confining carriers. This brings a quantum wire (Qwire)-like structure into the picture (in the [1-10] direction). Again, emission from the ground state of the quantum wire produces polarized emission along the length of the wire. Modeling results from this width distortion phenomenon are discussed in more detail below with respect to Figure 5.

[0061] Thus, generally, and as described in more detail below, quantum wires can form from underlying crystal roughness, which creates localized distortions or variations in the width of the quantum wells. Specifically, the temperature and annealing parameters used to fabricate cubic GaN crystals can be tailored to take advantage of the anisotropic diffusion inherent to the crystal's constituents. As a result, the crystals can form ridges or corrugations at the molecular level, which can cause variations and / or stacking faults that contribute to the formation of quantum wires.

[0062] Alternatively, in FIG. 2, stacking faults (SFs) 204 intersect the quantum well 200, locally increasing the indium content of the well by up to two-fold. These result in quantum wire-like structures in the In-rich regions highlighted as 202. The direction of the resulting quantum wires is perpendicular to the surface of the crystal shown, i.e., entering the figure in the [1-10] direction. The quantum wires therefore confine electrons in one dimension in the [1-10] direction, producing polarized light emission (i.e., photoluminescence (PL) or electroluminescence emission). Modeling results from this stacking fault phenomenon are discussed in more detail below with respect to FIGS. 6, 7, and 8.

[0063] Furthermore, SFs can exist in both the

[0110] and [1-10] directions. This means that quantum wires can exist in both directions. Thus, light emerging from a

[0110] quantum wire is polarized in a particular direction, while quantum wires in the

[0110] orthogonal direction polarize light in the orthogonal direction. Thus, if there are an equal number of quantum wires in both directions, no net polarization occurs. However, in the present example, the anisotropy in the SF density is related to the offcut of the substrate, which beneficially results in a net polarization.

[0064] Thus, in another general set of examples, quantum wires may be formed from In-rich regions, for example, In-rich regions caused by the intersection of quantum wells with stacking faults.

[0065] Structure example: LED 3 shows a model of a light emitting diode (LED) device 300 having low band gap strands 104 sandwiched between high band gap material 106 (such as InGaN / GaN) forming a quantum wire 104 to constitute an active region capable of emitting polarized light. Electrical contacts 301 are disposed on a p-type semiconductor 302, such as p-type GaN, and an n-type semiconductor 304, such as n-type GaN. By way of example only, p-type GaN can be fabricated by doping with Mg, where the amount of p-type dopant is, for example, 10 17 ~10 20 cm -3 For n-type, the range is about 10 18 ~10 20 cm -3 The GaN layer can be doped with Si or Ge in amounts of 0.1 to 1.0 μm. The Zb-GaN (i.e., cubic GaN) 106 constitutes the majority of the active region including the intersecting InGaN 104. The LED structure 300 is disposed on a suitable substrate, for example silicon carbide 306 such as 3C-SiC (which is the only cubic polytype of SiC), pure Si 308, or GaAs. Thus, cubic GaN crystals can be grown directly on cubic SiC substrates.

[0066] The LED structures described herein, including quantum wires and / or quantum wells in the optically active region, have other advantages beyond simply providing polarized or temperature-independent polarized emission. Not only can the emission be polarized, but it can also be spectrally broad. Furthermore, this broad emission can be tuned across the visible spectrum by varying the width of the quantum wells. This is advantageous, for example, for use in white LED backlights, where the broad spectrum allows for a broad color gamut (i.e., for whiter whites).

[0067] Structure example: Laser diode 4 shows a model of a laser diode device 400 in which quantum wires constitute the active light emitting active region, also made from low band gap strands 104 sandwiched between high band gap materials 106 (such as InGaN / GaN). Electrical contacts 301 are also disposed on the p-type and n-type semiconductors 302 and 304, respectively. The active regions 104, 106, which may include zb-GaN / InGaN, are sandwiched between reflective layers 402 that provide the laser function. Again, the diode structure is disposed or deposited on a suitable substrate, for example silicon carbide 306, such as 3C-SiC (a cubic polytype of SiC) and pure Si 308.

[0068] Quantum wires formed from quantum well width variations Figure 5a shows a two-dimensional confinement potential 500 caused by a localized increase in quantum well width, such as that produced by quantum well width distortions / variations as seen in the quantum well channel 104 of Figure 2. These width distortions create quantum wires in the z-direction (corresponding to the [1-10] crystallographic direction in Figure 2), thus producing polarized emission. In this example, the quantum well potential energy depth 506 is 80 meV, the overall quantum well width 504 is 4 nm (in the y-direction), and the width variations are 6 nm in the x-direction 502 and 10 nm in the y-direction.

[0069] To solve the potential for this example, the time independent Schrodinger equation (TISE) was solved in one dimension to find the ground state energies of the heavy holes and electrons for quantum wells of various thicknesses. These quantum wells were treated as infinite-range quantum well width distortions. Comparing the ground state energies of these different quantum wells gave an estimate of the energy difference between the ground states inside and outside the well width distortion as a measure of how confined the carriers are within the distortion.

[0070] TISE calculations of the hole potential give a ground state energy difference of up to about 6 meV, which indicates that holes are not well confined in the width-varying quantum wires. The large effective mass of holes makes the width-varying wires less influential on the holes. However, conversely, holes are more likely to localize in the indium-rich wires, which is not taken into account in this calculation.

[0071] The energy difference of the electron ground state can be as much as 100 meV for a change to a quantum well of 4 nm to 12 nm. Thus, advantageously, the electrons can be confined by the width variation structure, whereby the high energy difference of 100 meV can significantly reduce the rate of thermionic emission from the quantum wire at room temperature. Thus, further advantageously, such a width variation structure in a cubic InGaN / GaN quantum wire structure can exhibit a temperature-independent emission intensity (i.e., can function well at room temperature).

[0072] However, depending on the initial well width, well width variations of less than 2 and 3 nm only result in a comparable energy difference (about 6 meV) for the holes. These electrons are therefore not well confined. Thus, well width variations of less than about 3 nm result in a higher rate of thermionic emission at room temperature. Thus, at higher temperatures close to room temperature, the intensity of the emission may decrease to about 10% of the value at low temperatures (about 10 K). These modeling results therefore suggest that the rougher the quantum wells are, the stronger the confinement of carriers, especially electrons. Rough quantum wells may therefore give rise to photoluminescence (PL) with a temperature-independent intensity than smoother quantum wells.

[0073] In summary, a well width change / variation of more than about 3 nm is required to reduce the thermionic emission of electrons from the quantum wire so that a temperature-independent emission intensity is obtained. However, the quantum well strain only confines the electrons if the spatial extent along the quantum well is comparable to or larger than the Bohr radius of the electron (about 2.5 nm).

[0074] FIG. 5b shows a TEM image of an actual zinc-blende (cubic) InGaN / GaN 104 quantum well 100 sample (with an average quantum well width of 8 nm), including well width variations 508. The image shows significant distortion / variations to the quantum wells, which in this case are caused by underlying roughness of the crystal. Therefore, the underlying crystal roughness can be intentionally manufactured (i.e., by varying the annealing and growth conditions in the MOVPE growth process), which takes advantage of the anisotropic diffusion of Ga and N atoms that is inherent to the crystal at a particular temperature.

[0075] In this example in Figure 5b, the average quantum well width is 8 nm, but this can vary between about 4 nm and 12 nm. These distortions in the width of the quantum wells occur only in one dimension and can, surprisingly, result in additional carrier confinement. This allows quantum wire (Qwire)-like structures to enter the figure, i.e., in the [1-10] direction as shown in the legend). Again, advantageously, the recombination of electrons and holes from the ground state of such quantum wires due to the width variations results in polarized emission, where the plane of linear polarization is oriented along the length of the wire.

[0076] Quantum wires formed from indium-rich regions Indium is an example used in the following examples as the alloying material that causes the formation of quantum wires and thus polarized light emission in (e.g., zb-GaN / InGaN) semiconductor structures, but it will be understood by those skilled in the art that a variety of other suitable materials, elements, or crystals can be used. In general, any suitable zinc-blende III-nitride having a suitable structure may be suitable for producing quantum wires that exhibit polarized light emission.

[0077] For example: boron-rich GaN, Ga-rich AlGaN, and generally any aluminum nitride (AlN), aluminum gallium nitride (GaAlN), aluminum indium gallium nitride (AlInGaN), aluminum indium nitride (InAlN), or GaN enriched / alloyed with an appropriate group III element or combination of elements (although said enrichment / depletion / alloying causes the bandgap material to be narrower than the surrounding matrix material, inducing carrier confinement). That is, carriers are confined in two dimensions and can only move freely in one dimension. Thus, the general formula for the matrix and buried region material is (Al)(Ga)(In)(B)N, where the buried regions defining or constituting the quantum wire can contain the same elements in different ratios or be alloyed with additional materials to produce a lower bandgap material.

[0078] 6 shows a schematic energy potential 600 of the conduction band of a quantum wire arising from an indium-rich region around the interaction of a stacking fault with a quantum well. The theoretical dimensions of the quantum wire arising from the indium-rich region are 6 nm in the y-direction 606 and 4 nm in the x-direction 608. The potential at the quantum well depth 604 is 70 meV, and the total potential depth of the system 602 is 150 meV. In the theoretical modeling of this quantum wire, the band alignment between the stacking fault and the fault-free zb-GaN has been neglected, since the spatial extent of the high indium content region is larger than the stacking fault itself.

[0079] 7 shows a graph 700 displaying the solution of the wave functions for electrons 702 and holes 704, offset by the energies of states 706, solved for a model energy potential for an 8 nm quantum well with a 4 nm wide In-rich region, which generally corresponds to the model potential 600 of FIG.

[0080] Figure 8 shows a simplified total energy potential 800 for the conduction and valence bands of an 8 nm quantum well with a 4 nm wide indium-rich region corresponding to the wave function of Figure 7. The potential shows: the full depth of the conduction band 802 (0.52 eV) from the quantum wire ground to the stacking fault free quantum well; the energy difference 803 between the conduction band of the quantum wire and the carriers confined in the quantum well (0.28 eV); the band gap 804 between the potential depth of the conduction band of the quantum wire and the valence band (2.57 eV); the potential depth of the valence band of the quantum wire 805 (0.15 eV); and the respective x-dimensions 810 and y-dimensions 806 of the quantum wire (4 nm by 8 nm, respectively).

[0081] Again, the intersection of the SF and the InGaN quantum well creates a region where the In fraction is up to twice as large, as shown in Figure 2. Overall, the indium fraction in the quantum wire can therefore be about twice the value across the quantum well. To calculate how the electrons and holes are affected when the SF locally increases the indium fraction by a factor of two, we examined the potential shown in Figure 6. The potential is assumed to be constant in the z direction and does not take into account the effects of alloy variations.

[0082] The problem was simplified to the potential shown in Figure 8 to allow for separation of variables. This assumption is valid as long as the carriers are sufficiently confined by the quantum wire. The time-independent Schrödinger equation (TISE) was then solved separately for each dimension (including the effect of the Coulomb interaction between electrons and holes, extraction of the carrier wavefunction as shown in Figure 7, and the ground state energy). This model was used to investigate the temperature dependence of the intensity and polarization of the PL emission of the quantum wire.

[0083] Temperature dependence of strength The electron and hole wave functions 700 and energies 600 were calculated while varying the dimensions of the system, i.e., the width of the quantum well and the width of the quantum wire, and the indium fraction content of the quantum well. In each case, the ground state energy was compared to the total depth 602 of the quantum wire (i.e., the energy difference between the ground state of the quantum wire and the quantum well without stacking faults) to determine whether the carriers were confined by the quantum wire at a temperature of 10 K, providing a qualitative assessment of the degree of carrier confinement at room temperature.

[0084] This model therefore allows the rate of thermionic emission to be compared for different quantum wire dimensions and indium fractions.The model used for thermionic emission from quantum wells can be shown to be proportional to exp(-ΔE / kT), where T is temperature, k is the Boltzmann constant, and ΔE is the energy difference between the confined carriers and the barrier height.

[0085] When applied to quantum wires, ΔE is the energy difference 604 between the carriers confined in the quantum wire and the quantum well (i.e., 70 meV when looking at the depth of the quantum wire 604 in FIG. 6). Thus, considering the Boltzmann distribution above, the intensity of the quantum wire-related emission may decrease upon increasing temperature if the rate of thermionic emission is high compared to other mechanisms. Thus, in a preferred example, there is a large energy difference between the confined quantum wire carriers and the quantum well. This slows down the rate of thermionic emission, advantageously resulting in a temperature-independent (polarized) emission intensity. In general, thermionic emission is used to refer to the escape of charge carriers from the quantum wire, which disadvantageously reduces the intensity of the (polarized) emission. Thermal excitation is used to refer to the thermally induced occupation of an excited state.

[0086] In the case of the quantum well with the lowest indium content (4%, corresponding to a local increase of about 8% in the quantum wire itself), charge carriers are confined at 10 K for all dimensions except for the smallest quantum wire (2 nm × 1 nm) investigated. However, since the ground state of holes is at most several tens of meV (i.e., about 10 - 50 meV) lower than the QW energy, the rate of hot electron emission may become unfavorably high.

[0087] Nevertheless, advantageously, the effect of high-rate hot electron emission can be mitigated by increasing the indium content of the quantum well. In the case of a 20% indium quantum well (corresponding to an indium portion of about 40% of the quantum wire), the energy of the carrier ground state is at most several hundred meV (i.e., about 100 - 500 meV) lower than the energy of the quantum well. Thus, the rate of hot electron emission is significantly reduced, and such quantum wires can obtain a favorable level of emission intensity that is independent of temperature (e.g., PL emission or electroluminescence emission). Narrow quantum wells may also be suitable: Experimental PL measurements at 2 nm (i.e., described later in Fig. 12) show that the emission only decreases by about 40% at room temperature. This may be because the other dimensions of the quantum wire are large, as TEM / EDX studies indicate that the region of high indium content is typically 4 nm - 8 nm wide.

[0088] Generally, as the overall dimensions of the quantum wire increase, the energy of the ground state decreases, and the rate of hot electron emission further decreases. However, the main cause of reducing the unfavorable hot electron emission between states is the increase in indium content. Thus, in summary, for minimizing the effect of hot electron emission, higher indium content and wider quantum wells are beneficial and advantageous.

[0089] Temperature dependence of polarization Emission from quantum wires at very low temperatures (<10 K) is known to be optically polarized along the length of the wire. This optical polarization can be the result of mixing of light and heavy holes due to anisotropies along and perpendicular to the wire caused by the confining potential. This generates a variety of different confined hole subbands, each with a different proportion of light hole character. Furthermore, the transition involving the hole ground state (which is energetically similar to the ground state of a pure heavy hole) and the first hole excited state is polarized along the length of the quantum wire.

[0090] However, the second hole excited state, and all subsequent excited states, are polarized perpendicular to the quantum wire. This second hole excited state is approximately halfway between the heavy hole and light hole ground states. As the temperature increases, holes can be thermally excited to excited states and recombine in orthogonally polarized transitions, reducing the net polarization observed from the quantum wire emission. This temperature-dependent polarization is therefore a general property of quantum wires. Furthermore, it has been found that the dimensions of the quantum wire determine the strength of the temperature dependence.

[0091] Advantageously, however, for narrower quantum wires, the temperature dependence is observed to be reduced or even completely eliminated. This observation can be explained by considering the energy difference between the states of the quantum wire. If the energy difference between the ground state and the excited orthogonal polarization state is large, the thermal occupation of the excited state can be reasonably small, and therefore the polarization does not decrease significantly with increasing temperature. Moreover, if there are no excited states in the quantum wire, then even more advantageously, the degree of polarization is substantially independent of temperature.

[0092] To determine the properties of the quantum wire required to achieve temperature-independent polarization, the previous model (used to model the intensity dependence on temperature) was extended to include the effect of light holes. The effective mass of these holes is approximately 1 / 100th of the effective mass of the heavy holes. Using the same method as before, the ground state energy of the light holes was calculated for various indium contents and quantum wire dimensions. The approximate energy of the second excited hole state was then estimated as midway between the energies of the ground states of the heavy hole and light hole.

[0093] These results provide a guide to the ideal dimensions and indium content to obtain quantum wires with temperature-independent polarized emission.

[0094] 9 shows a table 900 summarizing these results, where the dimensions of the quantum wires and the relative indium fractions are shown and which combinations can lead to temperature-independent polarized emission. In the table, the legend is as follows: no second excited state (902): small excited state splitting (904); large excited state splitting (906); and significant thermionic emission (908).

[0095] For example, with an indium fraction of 0.04, there are either no excited confined states or the excited states are very close to the depth of the quantum wire. Thus, the emission from these quantum wires is highly polarized up to room temperature. However, as mentioned above, the rate of thermionic emission is high because the hole ground state is close to the quantum wire barrier energy (908). This fact is reflected in Figure 9, where we used kT at room temperature (26 meV) as a rough guide to the rate of thermionic emission.

[0096] Similarly, for narrow quantum wires with dimensions of 2 nm × 2 nm or less, the polarization is temperature independent since there are no confined excited states (902). However, if the indium content is too low, the hole ground state will be close to the height of the quantum wire barrier, resulting in high rates of thermionic emission (908). To avoid this, an indium content greater than 0.16 is necessary for narrow quantum wires to yield temperature independent polarization and intensity (i.e., 902 or 906).

[0097] For quantum wires with larger dimensions and lower indium composition, the excited hole states are confined and can therefore be thermally populated at room temperature (i.e., typically 904 and 908). However, for higher indium contents, this thermal population is smaller since the energy separation between the ground and excited states is larger, up to about 60 meV. The energy separation increases due to reduced penetration of the hole wavefunction into the barrier, which results in a polarization that does not change significantly with temperature up to room temperature.

[0098] All of the example structures discussed above are effective at producing quantum wires that result in optically polarized light emission. Furthermore, the results in Figure 9 show that certain examples with sufficiently high indium content (approximately 20% or more in some examples) and quantum wires of the correct dimensions (e.g., cross-sectional diagonal lengths of approximately 2-8 nm) can have emission with emission intensity and polarization that is substantially independent of temperature.

[0099] However, the above structures may be the result of structural defects and may be more difficult to control. Therefore, techniques to fabricate semiconductor structures that can be directly controlled and operable to generate polarized light emission would be of further advantage.

[0100] FIG. 10 shows an example structure that achieves this additional controllability. FIG. 10 shows a schematic diagram of a quantum wire 104 that includes a doped cubic GaN crystal (such as InGaN) surrounded by a region cubic GaN 106. The dimensions of these structures can be controlled by varying the thickness of the quantum well and subsequently etching away portions. That is, to produce an InGaN / GaN quantum wire, an InGaN / GaN quantum well can be grown and subsequently etched away. These quantum wires can combine both confinement effects previously studied; that is, the indium content in the quantum wire 104 is increased. Also, the thickness of the quantum wells can be different.

[0101] To determine the polarized emission properties of these structures at room temperature, a similar analysis to that described above was performed. TISE was solved in two dimensions for electrons and holes in separable potentials, in a format similar to that shown in Figures 7 and 8. This was done for indium composition / fractions ranging from 0.05 to 0.30, quantum well widths from 2 to 10 nm, and etch separations from 5 to 20 nm. The ground state energies of electrons and holes confined in the quantum wires were calculated in each case.

[0102] Figure 11 shows the hole ground state energies and orthogonally polarized excited state energies for 0.05 and 0.30 fractions of indium. As with SF-induced quantum wires, it is the hole energy that primarily determines the luminescence behavior of the system (rather than the electron energy). Figure 11 shows the calculated energies of the hole ground and excited states. These results show that the calculated states are well confined in quantum wires of all dimensions and indium contents investigated. When considering how the polarization of the luminescence depends on temperature, it is therefore informative to consider the energy splitting of the ground and excited states.

[0103] The hole energy level is compared to the depth of the quantum wire. For all combinations of quantum wire dimensions and indium content, carriers are confined in the quantum wire at low temperatures. Calculations show that the ground state of the holes is at least 24 meV below the height of the quantum wire barrier, which is the case for a 0.05 indium quantum wire with dimensions of 2 nm × 5 nm. As the quantum wire dimensions increase, the hole confinement energy decreases, and as a result, the rate of thermal excitation also decreases. The results in Figure 11 also show that increasing the indium content reduces the rate of thermal excitation. Advantageously, as mentioned before, the reduction in thermal excitation leads to a greater temperature independence of the degree of polarization. In this case, since the orientations of the hole ground state and the excited state are orthogonally polarized, the reduction in the rate of thermal excitation between the two reduces the temperature dependence of the degree of polarization of the emission. In general, thermal excitation is used to refer to the thermally induced occupation of an excited state, and thermionic emission is used to refer to the escape of charge carriers from the quantum wire.

[0104] Compared to the SF-mediated QW, the hole energy is generally even lower than the QW barrier height, which indicates a slower rate of thermionic emission in the InGaN / GaN QWs produced in this way, as shown in Figure 10. Advantageously, the recombination strength can therefore be less sensitive to temperature changes, resulting in a temperature-independent polarized emission source.

[0105] More specifically, the energy splitting of the states depends on the size and indium content of the quantum wire. For an indium content of 0.05, the energy splitting increases from 4 to 11 meV as the quantum wire dimensions decrease. Smaller quantum wires therefore advantageously reduce the thermal occupation of excited states and, consequently, the effect of temperature on the light polarization. However, these energy splittings are relatively small compared to kT (26 meV) at room temperature, suggesting that the polarization of the emission is strongly dependent on temperature at low indium content. In general, increasing indium content increases the energy splitting, e.g., a maximum of 36 meV splitting is seen for a 2 nm × 5 nm quantum wire with an indium content of 0.30.

[0106] Furthermore, increasing the indium content may also amplify the beneficial effects of modifying the quantum wire dimensions. These effects arise because increasing the indium content reduces the penetration of the carrier wave function into the barrier. Thus, the results suggest that a higher indium content advantageously reduces the effect of temperature on the polarization of the light emission. In detail, the energy splitting is larger than kT at room temperature for indium contents of 20% and above for 2 × 5 nm quantum wires, and exceeds 0.25 for 4 × 5 nm quantum wires.

[0107] In summary, to beneficially achieve temperature-independent light polarization in cubic InGaN / GaN, or more generally in cubic III-nitrides alloyed with bandgap-reducing alloy materials, thin quantum wires with high indium content are desired.

[0108] FIG. 12 shows a graph 1200 of an experimental photoluminescence spectrum acquired at 10 Kelvin for cubic GaN with InGaN with five quantum wells, each 2 nm thick. In detail, the polarized PL spectrum acquired at 10 K shows two peaks 1202, 1204, which are attributed to quantum well / wire luminescence, both polarized in the [1-10] direction (corresponding to the [1-10] direction of the quantum wire shown in FIG. 2). The near-band edge emission is unpolarized, indicating that the polarization is a property of the quantum well / wire luminescence and not an external effect. Similar results could be obtained with quantum wells of thicknesses of 4 nm, 6 nm, and 8 nm.

[0109] Figure 13a shows the temperature dependence of normalized intensity corresponding to peaks 1202 and 1204. Figure 13b shows the change in linear polarization with temperature for each peak in Figure 12 and for near band emission (NBE).

[0110] The PL results of the 2 nm quantum wire, shown in Figures 13a and 13b, show that the polarization of peak 1202 is independent of temperature, and furthermore, the intensity of peak 1202 does not change significantly with temperature. One dimension of the quantum wire is defined by the width of the quantum well (here 2 nm), while the other is unknown in the results presented here.

[0111] Modeling indicates that for quantum wires with any of the dimensions investigated, the quantum wires may recombine with the properties observed in Figure 13. This becomes more likely with increasing indium fraction in the quantum wells. In contrast, the intensity of peak 1204 decreases significantly with temperature, with the polarization approaching zero at 300 K. Thus, the results for peak 1204 are less compatible with a model of a quantum wire generated by a local increase in indium concentration (i.e., at the intersection of the SF and the quantum well). This is because the strong temperature dependence of the intensity makes the polarization independent of temperature. Thus, peak 1204 is less likely to be caused by such an In-rich quantum wire.

[0112] To summarize the above discussion, PL measurements show that the emission from the zb-InGaN / GaN quantum wires is highly polarized at 10 K (up to 80%) in the [1-10] direction perpendicular to the substrate roughness / miscut. There are also two separate emission peaks that are attributable to emission from the quantum wires. Peak 1202 is lower in energy and broader than peak 1204. As the temperature increases from 10 K to 300 K, the polarization of peak 1202 remains constant at 80% and the intensity drops to about 40% of its low-temperature value. Over the same temperature range, the polarization of peak 1204 drops from 45% to nearly zero and the intensity drops to 10% of its low-temperature value.

[0113] TEM / EDX measurements reveal the presence of SFs intersecting the quantum wires. In the region of the quantum wire around this intersection, the indium content can typically increase to twice the indium content of the InGaN quantum well region. This allows the existence of quantum wires with high indium content, where charge carriers can be confined in one dimension and polarized light can be emitted.

[0114] At low indium content and small wire dimensions, the emission intensity drops dramatically with temperature because hole carriers are much more likely to leave the wire via thermionic emission.

[0115] At high indium content and large quantum wire dimensions, excited confined states exist within the quantum wire that emit light polarized perpendicular to the ground state: as temperature increases, these excited states are created and the polarization decreases. Thus, the behavior of peak 1204 is not compatible with this modeling, as its intensity and polarization decrease with increasing temperature. However, Figure 9 shows that there are various combinations of quantum wire dimensions and indium content that may advantageously result in emission of temperature-independent intensity and polarization, corresponding to peak 1202 in Figures 13a and 13b. The low peak energy of peak 1202 can also be explained by the relatively high indium content of the quantum wire. Contributions of variations between different SF crossings may result in a broad emission spectrum. Thus, the evidence in Figures 12 and 13 suggests that peak 1202 may be the result of an indium-rich quantum wire.

[0116] TEM measurements also reveal that the quantum wire structures are distorted in one dimension due to roughness in the underlying crystal, which can result in a change in the well width as shown in the image in Figure 5b. This results in quantum wire-like structures (and thus carrier confinement and polarized emission) in the [1-10] direction. Modeling shows that these structures do not confine holes; nevertheless, holes can instead be localized due to the variation (increase) in the relative indium content.

[0117] Advantageously for the operation of semiconductor devices, electrons can be confined up to 300 K when the size of the quantum well thickness changes / variations exceeds about 3 nm. Electrons can be well confined even up to operating temperatures of about 400 K, where a temperature-independent degree of polarization is achieved for certain combinations of quantum wire dimensions and alloy material content (e.g., various configurations in Figure 9). Thus, for coarser quantum wires, the emission intensity is less dependent on temperature due to reduced thermionic emission. These structures can be attributed to peak 1204 in the PL spectrum.

[0118] To produce quantum wires in a more controllable way, quantum wells containing InGaN / GaN can be etched. Calculations suggest that for thin quantum wires with dimensions of about 4 nm and fabricated from quantum wells with indium fractions greater than about 0.20, as shown in Figure 10, light emission with temperature-independent intensity and optical polarization is achievable in such structures.

[0119] Although the above description is directed to experimental structures and modeling results for cubic GaN and InGaN (which can be grown on cubic 3C-SiC), it will be appreciated by those skilled in the art that III-nitrides of different compositions forming the quantum well regions and the surrounding matrix may also be suitable for such semiconductor devices exhibiting polarized light emission, such as: aluminum nitride (AlN), aluminum gallium nitride (GaAlN), boron gallium nitride (BgaN), Ga-rich AlGaN, indium gallium nitride (InGaN), aluminum indium gallium nitride (AlInGaN), and similar nitride layers (wherein the layers or regions defining the quantum wells are alloyed to have a lower bandgap than the surrounding (matrix) material).

[0120] Crystal composition and structure We have confirmed that optically polarized emission can be generated from cubic InGaN / GaN quantum wells (QWs) at room temperature with a degree of linear polarization (DOLP) of up to 75%. This DOLP is similar to that achieved by wz-QWs, but the advantage of generating polarized emission from cubic InGaN / GaN quantum wells is that such structures can be obtained using standard MOCVD epilayer growth, eliminating the need for further processing steps. The emission can be tuned to cover the visible spectrum with minimal impact on recombination efficiency or dynamics. The emission is associated with indium-rich quantum wires that form within the quantum wells due to their intersection with stacking faults (SFs). In addition, the quantum wires capture carriers from the remaining quantum wells. At low temperatures, emission from the remaining quantum wells is seen.

[0121] Furthermore, the inventors have identified various structural and compositional features of cubic III-nitride based crystals that may further affect the duration, degree of polarization, and temperature dependence of the light emission from the quantum wells. Photoluminescence perpendicular to the surface of cubic InGaN / GaN quantum wells emitting in the visible spectrum can be observed to be optically polarized to the extent of 86% at 10K and up to 75% at room temperature. Scanning transmission electron microscopy and energy dispersive X-ray measurements can further demonstrate that one-dimensional nanostructures are formed due to the segregation / enrichment of indium content adjacent to stacking faults. The light emission from these nanostructures dominates the room temperature spectrum and becomes red-shifted and broadened as the quantum well (QW) dimension increases from 2nm to 8nm (this dimension can be the width). Photoluminescence excitation measurements further indicate that carriers are captured by these nanostructures from the remaining quantum wells and recombine to emit polarized light along the length of the nanostructure. At low temperatures, light emission from the remaining quantum wells is observed at higher emission energies.

[0122] Generally speaking, green LEDs are not observed in the state of the art. LEDs based on InGaN / GaN quantum wells (QWs) grown on the c-face of the wurtzite (wz, also known as hexagonal) crystal structure can have room temperature internal quantum efficiency (IQE) of up to 90% for emission in the blue spectrum. However, advantageously, the emission wavelength can be extended into the green by increasing the indium content in the quantum wells. This reduces the IQE, a phenomenon known as the green gap. A possible explanation for this decrease in efficiency is that the lower growth temperatures required to increase the indium content increase the density of point defects, enhancing the rate of non-radiative recombination.

[0123] In addition, there is a strong electric field perpendicular to the quantum well due to spontaneous and piezoelectric polarization effects. A relatively high indium content increases the strain in the quantum well, increasing the electric field strength. The electric field acts to separate electrons and holes, thereby slowing down the rate of radiative recombination in long-wavelength emitters. Thus, the IQE of green quantum wells can be improved by lowering the indium content to reduce the non-radiative recombination rate.

[0124] The inventors have determined that the so-called green gap can be overcome by growing quantum wells on zinc-blende (zb, also known as cubic) GaN, which has a bandgap 200 meV smaller than that of wz-GaN. Furthermore, zb-GaN has zero spontaneous and piezoelectric fields in the

[0001] direction, so the electric field across a quantum well grown in the

[0001] plane is zero. However, zb-GaN is thermodynamically metastable during growth, so the epilayer contains a 1×10 5 cm -1 These SFs are changes in the atomic stacking order such that the crystal structure resembles that of thin planar wz-GaN.

[0125] Further disclosure below presents structures and photoluminescence measurements of zb-InGaN / GaN quantum wells, showing that the presence of SF, and in particular the presence of SF under certain conditions, can give rise to polarized emission at temperatures between 10 K and room temperature.

[0126] It will be appreciated that polarized light emission is used commercially and industrially in many applications in addition to green LEDs, which as discussed above are not generally known in the art For example, polarized light emission is useful as a backlight for liquid crystal displays. EXAMPLES

[0127] Experimental structure and photoluminescence Samples were grown on 3C-SiC / Si

[0001] substrates with an offcut of 4° in the

[0110] direction. The structural properties of the quantum wells were studied by scanning transmission electron microscope / energy dispersive X-ray (STEM / EDX) using an FEI Tecnai Osiris operated at 200 kV and equipped with four energy dispersive X-ray analyzers. High-angle annular darkfield (HAADF) images were taken with the beam direction parallel to the [1-10] zone axis. Samples for STEM analysis were prepared using the in-situ lift-out technique in a focused ion beam (FIB; FEI Helios NanoLabTM).

[0128] These 3C-SiC / Si

[0001] substrates offer a relatively small lattice mismatch with GaN (3.4%) and are available in wafer sizes up to 150 mm. Beneficially, this makes such wafers compatible with Si factories, opening a direct path to device commercialization. The zb-GaN epilayers were grown using metal-organic chemical vapor deposition (MOCVD). Five InGaN / GaN quantum wells were subsequently grown using a quasi-two-temperature (Q2T) method. Barriers with nominal quantum well thicknesses of 2 nm, 4 nm, 6 nm, and 8 nm were studied, with a nominal thickness of 16 nm. The inventors have determined through secondary ion mass spectroscopy (SIMS) measurements that, for example, the GaN epilayers have a lattice mismatch of 1×10 19 cm -3 It has been determined that GaN epilayers grown with zero oxygen impurities can contain oxygen impurity concentrations on the order of 1000 nm. As explained below, even this small presence of oxygen can result in an increase in background electrons and can partially fill the conduction band (CB), thus increasing the full-width half-maximum (FWHM) of the polarized emission. Nevertheless, the oxygen impurity is not inherent in providing or enhancing polarized emission. Thus, it is understood that the presence of oxygen is not necessary to provide or improve the polarized emission aspect, and that GaN epilayers grown with zero oxygen impurities can have similarly advantageous polarization properties.

[0129] The optical characteristics are: wavelength 325 nm, excitation power density 10 Wcm -2 of continuous wave HeCd laser, and 0.4mWcm at each wavelength. -2 The compounds were investigated by photoluminescence (PL) and PL excitation (PLE) spectroscopy using a 300 W Xe lamp coupled to a monochromator with an excitation power density of 100 W, respectively.

[0130] The PL was focused onto the slit of a double-grating spectrometer with a spectral resolution of 24 Å. The light was detected using a GaAs photomultiplier tube (PMT) and processed with lock-in amplification technique. The spectral response of the PMT and spectrometer was measured using a calibrated blackbody source and used to correct the PL spectrum. The light polarization was analyzed with a Glan-Thompson polarizer to collect the emission in the

[0001] direction. The PL time decay was obtained by excitation with a 100 fs frequency-tripled pulsed Ti:Sapphire laser with a wavelength of 267 nm, with 3 × 10 per pulse. 12 cm -2 A carrier density of was injected.

[0131] Figure 14 shows a graph 1400 displaying four polarized photoluminescence spectra obtained at room temperature for quantum wells (QWs) with widths of 2 nm 1408, 4 nm 1406, 6 nm 1404, and 8 nm 1402, resolved in polarization in the

[0110] (lower line) and [1-10] (upper line) directions. These spectra show the Fabry-Perot interference fringes that are typically observed in GaN-based multilayer structures. The peak common to all spectra at 2.8 eV is the spontaneous emission line from the laser.

[0132] The time-correlated single-photon counting technique was used to generate the PL decay transients. The polarized PL spectrum at room temperature for each sample is shown in Figure 1. A broad emission is observed when detecting light polarized in [1-10]. With orthogonal polarization (

[0110] ), the emission is observed with reduced intensity. The degree of linear polarization (DOLP) is given by DOLP = (I max -I min ) / (I max -I min ) wherein I max , I minare the maximum and minimum integrated intensities of the emission. The DOLP is determined to be 75%, 70%, 65%, and 75% for quantum well widths of 2 nm (1408), 4 nm (1406), 6 nm (1404), and 8 nm (1402), respectively, with a 5% error, and is therefore broadly independent of quantum well width.

[0133] The normalized (X,Y) CIE color values ​​of the PL spectra vary from (0.14, 0.17) for the 2 nm sample to (0.33, 0.51) for the 8 nm sample, representing blue and yellow-green, respectively. The redshift of the emission peak with increasing quantum well width is generally consistent with a decrease in quantum confinement energy for larger quantum well widths. However, the emission continues to redshift when the quantum well width significantly exceeds the Bohr radius of the electron and hole (approximately 2.6 nm and 0.26 nm using the dielectric constant and effective mass of zb-GaN). This suggests that there is another effect influencing the redshift behavior.

[0134] As mentioned above, SIMS measurements have shown that in some cases the oxygen impurity concentration in the samples is as low as 1 × 10 19 cm -1 It has been shown that the FWHM is on the order of 100 nm. These oxygen impurities may act as shallow donors in the wz-GaN epilayer, resulting in a relatively high density of background electrons in the sample, which may partially fill the conduction band (CB). Thus, the CB filling may simply be partially responsible for the large FWHM of the emission in some arbitrary examples. Nevertheless, as mentioned above, the larger FWHM is not an inherent property of oxygen impurities. Preferably, the sample is free of oxygen impurities and still capable of producing polarized emission.

[0135] 15 shows a graph 1500 displaying a photoluminescence (PL) spectrum 1506 (solid line) of a 6 nm thick quantum well sample at 10 K temperature, and the corresponding PL excitation (PLE) spectrum (dashed line; 1502b, 1504b) detecting emission at an excitation photon energy of 2.53 eV. The graph further displays the characteristic energy of the GaN absorption edge for each of the PL spectrum 1502a and the PLE spectrum 1502b, and the absorption edge of the quantum well (i.e., InGaN) for each of the PL spectrum 1504a and the PLE spectrum 1504b.

[0136] When detecting emission at any wavelength across the PL peak, photon absorption edges can be observed at energies of 3.26 eV and 3 eV. The first absorption edge (3.26 eV) corresponds to the photogeneration of carriers in zb-GaN, which are captured by the quantum well and recombined.

[0137] The second absorption edge is due to direct absorption of photons in the quantum well, where a sigmoidal fit is applied to the PLE spectra 1502b, 1504b to extract characteristic energies. As seen in data point 1502a of graph 1500, the GaN edge is at a constant energy (3.26 eV) close to the band gap of unstrained zb-GaN (3.3 eV), and this discrepancy (i.e., between 3.26 eV and 3.3 eV) suggests that zb-GaN is under tensile strain. Nevertheless, it is understood that tensile strain in zb-GaN is not a required property to produce the advantageous results described herein. In contrast, the quantum well absorption edge (1504a) is not at a constant energy. When detected at a higher energy peak, the absorption edge shifts with the emission energy, and the energy difference between absorption and emission is approximately 250 meV. This is consistent with direct absorption of photons into the quantum well, creating electron-hole pairs that cool to the ground state before recombining.

[0138] For the lower energy peaks, the absorption edge shifts at a decreasing rate with emission energy, and the absorption edge drops to near zero before the onset of emission. This suggests that there are distinct regions of different energy levels in the quantum well. For example, these separate regions can be created with different indium fractions, consistent with the STEM / EDX measurements shown in Figure 2. Because these regions have small volumes and low density of states, there is little direct photon absorption into these regions. Instead, carriers are trapped in these regions of high indium content from the rest of the quantum well: this results in an energy difference of up to 800 meV between absorption and emission. Filling of the conduction band (CB) (e.g., by oxygen impurities) may also contribute to this large energy difference by preventing absorption into filled, low-energy CB states, while carriers in these filled states can still radiatively recombine. In summary, the large FWHM of the low-energy emission (between 420 meV and 510 meV, as seen in the subplot of Graph 1400Figure 14) is partly due to the CB filling and partly due to the variations in quantum wire size and indium content.

[0139] The PL spectrum of FIG. 12, detailed above, shows that two emission bands (i.e., high-energy and low-energy emission bands) are present at a temperature of 10 K. This is consistent with all samples investigated and displayed in FIG. 18, described below. For the 2 nm quantum well, there is a low-energy peak centered at 2.67 eV and a high-energy peak centered at 2.86 eV. These energies are higher than the room temperature values ​​due to the increase in the band gap with decreasing temperature. For the 2 nm quantum well, with respect to subplot 1408 of graph 1400, the low-energy peak (e.g., 1202 in FIG. 12) is polarized in the [1-10] direction with a DOLP of 86%, while the high-energy peak has a DOLP of 37%. As the sample temperature increases, the high-energy peak is quenched at a faster rate than the low-energy peak, and therefore the polarized low-energy peak dominates in the room-temperature PL spectrum shown in FIG. 14.

[0140] FIG. 16 shows a graph 1600 displaying PL time decays measured at a temperature of 10 K, which provides insight into the recombination mechanisms underlying the two emission bands at low temperatures. In particular, FIG. 16 shows photoluminescence time decays from a GaN / InGaN crystal at 10 K produced from a 2 nm wide quantum well. These decays are measured near the peak of the high energy peak (peak 1204 in FIG. 12), 2.82 eV, and at energies of 2.43 eV, where the largest component of the PL intensity is from the lower energy peak.

[0141] Another non-exponential decay component is also observed when detecting the emission at the low energy peak. This indicates that different recombination mechanisms contribute to the PL decay at this energy. This non-exponential form may indicate a distribution of different recombination rates at the same emission energy due to the recombination of carriers from different local environments. The decay times are discussed in more detail below with respect to Figure 17. Structural measurements of the quantum wells provide an explanation for this low energy emission peak. As seen in Figure 2 above, the SF 204 intersects with the quantum well 104 on its way to the GaN surface. EDX measurements from the sample shown in Figure 2 indicate that the indium content within a few nm of the SF is approximately twice that of the remaining quantum wells.

[0142] Increasing the indium content reduces the band gap in these regions. Because SFs are planar defects, these regions stretch perpendicular to the plane of the image (in the [1-10] direction), resulting in quantum wire (Qwire) 1D nanostructures. Figure 2 shows that the quantum wells are grown on a rough zb-GaN surface. This results in a serpentine profile when viewed in the [1-10] direction, but not in the

[0110] direction. This serpentine may also provide an additional contribution to the confinement of carriers in the quantum wire and the remaining quantum well. Recombination in these indium-rich regions will be lower in energy due to the reduced band gap compared to the rest of the quantum well.

[0143] Furthermore, it has been observed that the SFs corresponding to the thin faces of the wz-GaN crystal deposits may further introduce electric fields into the structure due to differences in spontaneous polarization between the zb (cubic) and wz (hexagonal) phases. The separation of the SFs affects the magnitude of these fields and therefore the recombination energy of carriers in the indium-rich regions, which may contribute to the large FWHM of the light emission in the presently described quantum wells.

[0144] Recombination in quantum wires can also explain the light polarization of low energy emission, since emission from similar structures can be polarized along the length of the quantum wire. The DOLP can be altered by modifying the cross-sectional dimensions of the wire. Thus, by fabricating quantum wires of specific cross-sectional dimensions, the DOLP can be kept relatively constant for different temperatures. The simple calculations discussed in connection with FIG. 19 illustrate an example where the dimensions of the quantum wire can achieve a DOLP that is substantially invariant to temperature. Even more advantageously, it is possible to further maximize the DOLP by increasing the anisotropy of the SF distribution.

[0145] SFs are present in the

[0111] planes of zb-GaN, generating nanostructures along both the

[0110] and [1-10] directions of the quantum wells. If the distribution of SFs were uniform in all four planes, the net light polarization perpendicular to the growth direction would be zero. However, as mentioned before, the density of SFs depends on the orientation of the substrate edge. This results in a higher density of SFs in one direction, resulting in a net light polarization. Furthermore, the meandering profile of the quantum wells is preferentially seen in the [1-10] direction.

[0146] FIG. 17 shows a graph 1700 displaying four experimental emission 1 / e decay times for cubic InGaN / GaN quantum wells with thicknesses of 2, 4, 6, and 8 nm produced at a temperature of 10 K. Also included in each subplot is a Gaussian peak representing the peak energy and FWHM of peaks 1 (curve on the left side of each subplot) and 2 (graph on the right side of each subplot), respectively. PL decay curves were measured for each quantum well sample (i.e., with thicknesses of 2, 4, 6, and 8 nm) at a temperature of 10 K. Here, at a temperature of 10 K, the dynamics is assumed to be purely radiative. From these decay curves, the time it takes for the intensity to drop to 1 / e of the peak intensity was recorded, and this measurement was repeated for various emission energies across the spectrum.

[0147] Figure 18 shows the PL spectra measured at 10 K for samples with QW widths of 2 nm, 4 nm, 6 nm, and 8 nm. Two emission peaks are observed in each spectrum, corresponding to emission from the indium-rich region of the quantum well at low energy and emission from the remaining quantum well at high energy. There is no trend in the integrated intensity of the emission with quantum well width.

[0148] At the high energy end of the PL emission, the decay times are 320 ps, ​​200 ps, ​​210 ps, ​​and 310 ps with increasing quantum well width, with an error of 10 ps. The decay times measured for the zb-QWs do not show the large variation with quantum well width expected for polarized quantum wells. This indicates that the electric field is significantly reduced compared to the corresponding c-plane wurtzite quantum wells. This is therefore in contrast to the results obtained from polarized c-plane wurtzite InGaN / GaN quantum wells, where the radiative lifetime is significantly increased as the quantum wells become thicker, due to the electric field separating the electrons and holes traversing the quantum wells.

[0149] In heterostructures where the electric field is reduced, the recombination may involve excitons. Even in the absence of localization effects, the radiative lifetime depends on the width of the quantum well due to the change in the exciton binding energy. Another possibility is that the recombination involves a hole and an excess electron, and the excess electron limits the variation of the electron-hole wave function that overlaps with the width of the quantum well. Furthermore, as shown in Figure 17, the decay shape is monoexponential, which is consistent with exciton recombination.

[0150] At the lowest energy, there is also no clear trend in the decay times; these are 375 ps, 365 ps, 330 ps, ​​and 410 ps as the quantum well width increases. Thus, the recombination lifetimes of the low energy peaks also do not change significantly with quantum well width, despite the effect of the SFs generating electric fields in these regions. However, these electric fields are not across the quantum wells, but rather in the four possible 0111 directions, which means that the quantum well width has little effect on the carrier separation due to these fields, and therefore has minimal effect on the decay times. Instead, it is seen that the distance separating the SFs determines the electric fields, which in turn determine the decay times, as plotted in graph 1700 of FIG. 17.

[0151] Further modelling of InGaN / GaN quantum wells To determine whether carriers can be trapped by regions of high indium content, a three-dimensional model system was developed, as described above with respect to Figure 8. The system consists of two intersecting quantum wells of the same indium content, forming a rectangular cross-section quantum wire with twice the indium content at the intersection. The dimensions of the quantum wells were varied with indium content, and the ground state energies of electrons and holes were calculated using the Numerov method. Analysis of the model observed that both holes and electrons are confined in the quantum wire, and the degree of this confinement increases with increasing indium composition and quantum well width. In most cases, the energy required for carriers to escape the quantum wire into the quantum well will be much greater than the average thermal energy (26 meV) at room temperature. It is therefore possible that these quantum wires can trap both electrons and holes.

[0152] It is known that mixing heavy and light holes in a quantum wire results in the emission of light polarized in different directions. Recombination of the electron-hole ground state emits light polarized along the length of the quantum wire. At low temperatures and low carrier densities, only the ground state is occupied, so the overall emission from the quantum wire is polarized. This is consistent with the PL measurements in Figure 12, where the degree of linear polarization (DOLP) of the quantum wire emission at 10 K is 86%. The DOLP depends on the cross-sectional area of ​​the quantum wire: the narrower the quantum wire, the higher the DOLP. Higher states confined in the quantum wire can emit light polarized orthogonally to the ground state. Thus, the occurrence of higher states, which can occur at higher temperatures, reduces the DOLP of the emission. The measurements shown in Figure 15 show that the DOLP of the quantum wire emission at 300 K (i.e., room temperature) is as high as 75% (for a 2 nm wide quantum well), a reduction of about 10% compared to the quantum wire emission at 10 K.

[0153] FIG. 19 shows a calculated 2D map 1900 of the ratio of the degree of linear polarization (DOLP) of the emission from the quantum well at 300 K to the DOLP at 10 K for cubic InGaN / GaN quantum wells with varying indium content and quantum well dimensions. The simulation considers a thermal equilibrium population of ground and excited states and assumes that the recombination rate is unaffected. It can be seen that there are no confined states when the indium content is less than 0.25 and the diagonal dimensions are less than 2 nm. In other words, when the indium content is less than 0.25 and the diagonal dimensions are less than 2 nm, the degree of linear polarization (DOLP) of the InGaN / GaN quantum well at 300 K is equal to the DOLP at 10 K, and the DOLP is independent of temperature. Legend 1902 shows the ratio of the DOLP at 300 K to the DOLP at 10 K. The negative ratio, which corresponds to the lower right portion of the graph 1900, indicates that the polarization is expected to be orthogonal at room temperature. This orthogonal polarization pertains to quantum wells with diagonal lengths greater than 7 nm for low indium content and quantum well widths greater than 11 nm for high indium content.

[0154] Thus, as shown in Figure 19, DOLP may be temperature independent if there are no excited states in the quantum wire, or it may be weakly temperature dependent if the splitting between the ground and excited states is large compared to the average thermal energy (kT). The energies of these excited states depend on the dimensions of the quantum wire. To estimate the energies of these excited states, we use the energy of the ground state of the light hole (mass 0.2m 0) was calculated as an upper limit for these excited states. Because the excited states lie between these two, a slightly improved estimate of the splitting was obtained by halving the separation between the heavy and light hole states. To estimate how this splitting varies with indium content and quantum wire dimensions, the parameters of the system were varied. For an indium content of 0.20, the energy of the first excited state of a 2 nm quantum wire is above the energy of the quantum well ground state (i.e., the energy required for a carrier to escape the quantum wire). For all quantum wires with diagonal dimensions of 3 nm or greater, the first excited state is confined below the energy of the quantum well ground state.

[0155] Thus, in agreement with Figure 19, only for the narrowest calculated dimensions the excited state is above the barrier and therefore not confined to the quantum wire. In this case the DOLP is independent of temperature. As the wire dimensions increase the splitting decreases from 60 meV to 26 meV. Thus, at room temperature the excited state is populated for wider quantum wires and the DOLP of the emission from the wider wires decreases further as the temperature increases. Assuming that thermal equilibrium was reached during the experiment, this dependence was estimated by calculating the occupancy of the ground state and the first excited state at room temperature using Boltzmann statistics.

[0156] The ratio of DOLP at 300K and 10K was then estimated for various indium contents and dimensions, assuming the recombination rate of each state is the same. The results are shown in graph 1900 of FIG. 19. A change in DOLP of less than 10% can be achieved by using quantum wires with diagonal cross-sectional lengths less than 3 nm. Thus, this model 1900 suggests that it is possible to achieve small changes in DOLP between 10K and room temperature. In general, thinner quantum wires and / or higher indium contents will make DOLP less dependent on temperature. The zigzag nature of this plot 1900 is due to separate changes in the two dimensions of the cross-section, and the data is only plotted for their orthogonal sum. If one dimension is significantly narrower than the other, the calculated ratio increases. The darker areas on the right side of FIG. 19 show that the polarization is expected to be orthogonal at room temperature for diagonal lengths greater than 7 nm for low indium content and greater than 11 nm for high indium content. However, such effects may not be seen, since DOLP decreases to zero with increasing dimensions.

[0157] Although the present disclosure has been described with respect to the above preferred embodiments, it should be understood that these embodiments are merely illustrative and the claims are not limited to these embodiments. Those skilled in the art may make modifications and substitutions in light of the present disclosure, which are intended to be included within the scope of the appended claims. Each feature disclosed or illustrated in this specification may be incorporated into the present invention, whether alone or in suitable combination with other features disclosed or illustrated in this specification.

Claims

1. 1. A semiconductor structure comprising: a matrix comprising a first cubic Group III nitride having a first band gap; a second cubic III-nitride having a second bandgap and forming an embedded region within the matrix, the second cubic III-nitride including an alloy material that reduces the second bandgap relative to the first bandgap; a quantum wire defined by a portion forming a one-dimensional charge carrier confinement channel within a region embedded within the matrix; Including, A semiconductor structure, wherein the quantum wires are operable to exhibit optically polarized luminescent emission.

2. 10. The semiconductor structure of claim 1 further comprising a quantum well defined by a region buried within said matrix, said region forming a layer buried within said matrix.

3. The semiconductor structure of claim 1 or 2, wherein the matrix comprises cubic gallium nitride.

4. The semiconductor structure of any one of claims 1 to 3, wherein the alloy material comprises indium.

5. The semiconductor structure of any one of claims 1 to 4, wherein the portion of the region embedded within the matrix that defines the quantum wire comprises a localized increase in concentration of the alloy material.

6. 6. The semiconductor structure of claim 5 when dependent on claim 2, wherein the localized increase in concentration of alloying material is localized to an intersection between a stacking fault and the quantum well of the semiconductor structure.

7. 5. The semiconductor structure of claim 2, or claim 3 or claim 4 when dependent on claim 2, wherein the portions of the buried regions within the matrix that define the quantum wires include local variations in width of the buried layers that define the quantum wells.

8. The semiconductor structure of claim 7 , wherein the local variations have a dimension greater than 2 nm.

9. 9. The semiconductor structure of claim 7 or 8, wherein the width of the buried layer defining the quantum well varies between a width of 2 nm and 14 nm, inclusive.

10. The semiconductor structure of any one of claims 7 to 9, wherein the charge carriers confined in the quantum wire carrier confinement channel are electrons.

11. 6. The semiconductor structure of claim 1, wherein the portion of the region embedded within the matrix that defines the quantum wire is defined by a channel of a region comprising the alloy material, the channel extending through the matrix.

12. The semiconductor structure of any one of claims 1 to 11, wherein the percentage of said alloy material in the region embedded within said matrix is ​​greater than 20%.

13. 13. The semiconductor structure of claim 1, wherein the one-dimensional charge carrier confinement channel of the quantum wire has a first electronic state and a second electronic state, the difference in energy between the states being greater than a characteristic thermal energy, thereby reducing the likelihood of thermally induced transitions between the states.

14. The semiconductor structure of any one of claims 1 to 13, wherein the average dimension of said quantum wires is less than 10 nm and greater than 2 nm.

15. 15. A semiconductor device incorporating the semiconductor structure of any one of claims 1 to 14, wherein the semiconductor device is selected from the group consisting of: light emitting diodes (LEDs); vertical cavity surface emitting lasers (VCSELs); lasers; and sensors.

16. moreover, a substrate comprising cubic silicon carbide; an electron-rich layer of a matrix material disposed on a surface of the substrate; an optically active region defined by the matrix and a region embedded within the matrix, disposed on a surface of the electron-rich layer; an electron-deficient layer of matrix material disposed on a surface of said optically active region; 16. The semiconductor device of claim 15, comprising:

17. 17. The semiconductor device of claim 16, wherein the semiconductor structure further comprises optical confinement layers disposed on either side of the optically active region.

18. 1. A method for manufacturing a semiconductor device, comprising: forming a matrix comprising a first cubic Group III nitride having a first band gap; forming a second cubic III-nitride having a second bandgap and forming an embedded region within the matrix, the second cubic III-nitride including an alloy material that reduces the second bandgap relative to the first bandgap; forming a portion within a region embedded within said matrix defining a quantum wire, said portion forming a one-dimensional charge carrier confinement channel, whereby said quantum wire is operable to exhibit optically polarized luminescence emission; A manufacturing method comprising:

19. The method of claim 18 further comprising forming a buried region within the matrix as a buried layer within the matrix that defines a quantum well.

20. 20. The method of claim 18 or 19, wherein the semiconductor device is selected from the group consisting of: a light emitting diode (LED); a vertical cavity surface emitting laser, VCSEL; a laser; and a sensor.

21. moreover, forming a substrate comprising cubic silicon carbide; forming an electron-rich layer of a matrix material disposed on a surface of the substrate; forming an optically active region defined by said matrix and a region embedded within said matrix disposed on a surface of said electron-rich layer; forming an electron-deficient layer of matrix material disposed on a surface of said optically active region; The method according to any one of claims 18 to 20, comprising:

22. The method of claim 21 further comprising forming an optical confinement layer disposed on each of the first and second surfaces of the optically active region.

Images