Enhanced micro light emitting diode for inter-chip communication

Abstract

LEDs can be optimized in a variety of ways for high speed operation in optical communication systems. The LEDs, which can be micro-LEDs, can include dopants and levels of dopants that allow for increased operating speeds, the LEDs can include cladding, and the LEDs can include other features.

Description

Background Technology

[0001] This invention generally relates to LEDs, and more specifically to LEDs in optical communication systems.

[0002] Lasers, due to their narrow linewidth, single-space-mode output, and high-speed characteristics, often dominate in optical communications. The narrow linewidth of lasers allows high-speed signals to be transmitted over long distances through dispersive media without uncompensated pulse broadening. Long-distance fiber optic links are often limited by chromatic dispersion, making narrow-linewidth lasers indispensable. Furthermore, the single-space-mode nature of lasers makes coupling to single-mode fibers relatively easy.

[0003] Stimulated emission from lasers also allows for high modulation speeds. Direct modulation optical links using lasers can operate at 25 Gb / s, and PAM4 modulation can potentially carry 50 Gb / s of information.

[0004] However, the use of lasers can present challenges for very short-range optical communications, such as chip-to-chip communications. Summary of the Invention

[0005] Some embodiments provide an LED for optical communication, the LED comprising: a p-type layer; an n-type layer; at least one quantum well layer and at least one barrier layer between the p-type layer and the n-type layer, wherein at least one of the quantum well layer and the barrier layer is doped with a dopant. In some embodiments, the at least one of the quantum well layer and the barrier layer is doped with an n-type dopant. In some embodiments, the n-type dopant comprises magnesium. In some embodiments, the magnesium doping concentration is 10. 18 / cm 3 up to 10 20 / cm 3 Within the range. The at least one quantum well layer comprises at least two quantum well layers, wherein the at least two quantum well layers are separated by one of the at least one barrier layer, and the at least one barrier layer is doped with 10. 13 / cm 3 up to 10 15 / cm 3The range is defined as silicon. In some embodiments, the quantum well layer comprises InGaN, and the barrier layer comprises GaN, wherein an external tensile strain is applied parallel to a plane defined by the interface of the quantum well layer and the barrier layer, such that the applied strain is reduced to strain attributable to lattice mismatch between the InGaN of the quantum well layer and the GaN of the barrier layer. Some embodiments further include at least one interlayer located between at least one of the at least one quantum well layer and at least one of the at least one barrier layer, the barrier layer comprising GaN, and the interlayer being subjected to tensile strain together with the GaN. In some embodiments, the interlayer comprises AlGaN. In some or all of these embodiments, the LED is a microLED.

[0006] Some embodiments provide an optical communication system for transmitting information provided by a processor to another region of the processor or another chip in a multi-chip module, comprising: an LED associated with the processor; an LED driver for modulating the optical output power of the LED such that the LED generates light based on data provided from the processor to the LED driver; a detector for performing photoelectric conversion using the light; and an optical waveguide for optically coupling light from the LED to the detector; wherein the LED comprises: a p-type layer; an n-type layer; at least one quantum well layer and at least one barrier layer between the p-type layer and the n-type layer, at least one of the quantum well layer and the barrier layer being doped with a dopant.

[0007] In some embodiments, at least one of the quantum well layer and the barrier layer is doped with an n-type dopant. In some embodiments, the n-type dopant includes magnesium. In some embodiments, the magnesium doping concentration is 10. 18 / cm 3 up to 10 20 / cm 3 Within the range of [specific parameters]. In some embodiments, at least one of the quantum well layer and the barrier layer is doped with a p-type dopant. In some embodiments, the p-type dopant comprises silicon. In some embodiments, the doping concentration of the silicon is within 10 [specific parameters]. 18 / cm 3 up to 10 20 / cm 3 Within the range. In some embodiments, the at least one quantum well layer comprises at least two quantum well layers, wherein the at least two quantum well layers are separated by one of the at least one barrier layer, and the at least one barrier layer is doped with 10 13 / cm 2 up to 10 15 / cm 2Silicon within the range.

[0008] In some embodiments, the quantum well layer comprises InGaN, and the barrier layer comprises GaN, wherein an external tensile strain is applied parallel to a plane defined by the interface of the quantum well layer and the barrier layer, such that the applied strain is reduced to strain attributable to lattice mismatch between the InGaN in the quantum well layer and the GaN in the barrier layer. In some embodiments, the width of the quantum well layer is in the range of 2 nm to 3 nm. In some embodiments, the width of the quantum well layer is in the range of 3 nm to 4 nm.

[0009] Some embodiments further include at least one interlayer located between at least one of the at least one quantum well layer and at least one of the at least one barrier layer, the barrier layer comprising GaN, and the interlayer being subjected to tensile strain together with the GaN. In some embodiments, the interlayer comprises AlGaN.

[0010] In some embodiments, the LED is a micro LED.

[0011] Some embodiments further include: a further LED associated with other regions of the processor or other chips in the multi-chip module; a further LED driver for modulating the optical output power of the further LED such that the further LED generates light based on data provided to the further LED driver from the other regions of the processor or other chips in the multi-chip module; and a further detector for performing photoelectric conversion using the light from the further LED; wherein the further LED includes: a p-type layer; an n-type layer; and at least one quantum well layer and at least one barrier layer between the p-type layer and the n-type layer, at least one of the quantum well layer and the barrier layer being doped with a dopant. In some embodiments, the optical waveguide optically couples the light from the further LED to the further detector.

[0012] These and other aspects of the invention will be more fully understood upon review of this disclosure. Attached Figure Description

[0013] Figure 1 This is a block diagram illustrating an example of the use of LEDs in an optical communication system according to aspects of the present invention.

[0014] Figure 2A Showing a typical pin LED structure, Figure 2B exhibit Figure 2A The band structure of the device, and Figure 2C An optimized doping structure for an LED according to aspects of the present invention is shown, and Figure 2D exhibit Figure 2C The energy band diagram of the device.

[0015] Figure 3A Another embodiment of a micro-LED according to aspects of the present invention is shown.

[0016] Figure 3B Another embodiment of a micro LED according to aspects of the present invention is shown.

[0017] Figure 4 The band and carrier wavefunctions are shown with and without external strain compensation.

[0018] Figure 5 The band structure is shown with varying Al doping.

[0019] Figure 6 It is a curve of modulation frequency versus current density.

[0020] Figure 7 It is a graph of internal quantum efficiency versus current density.

[0021] Figure 8 This is a graph showing the time constant of a micro LED versus its driving current density.

[0022] Figure 9 A micro LED with etched through-holes according to an aspect of the present invention is shown.

[0023] Figure 10 It is a table describing the various trade-offs in the design of micro-LEDs with typical parameters. Detailed Implementation

[0024] Figure 1 Examples of using LEDs (which can be miniature LEDs) in optical communication systems are shown. Figure 1 In this process, silicon processor 111 performs various operations on or with data. For example, the silicon processor may perform calculations on the data, perform switching functions, or perform other functions. The silicon processor provides at least some data to LED driver 113, wherein the LED driver activates LED 115 to generate light, wherein the LED driver thereby modulates the output optical power of the LED to optically provide at least some data. The generated light is provided to optical coupler 117, which transmits the light into optical propagation medium 119. In some embodiments, the optical propagation medium, which may be, for example, a waveguide, may be used to transmit light from one region of the silicon processor to another region of the silicon processor. In other embodiments, the optical propagation medium may be used to transmit light from the silicon processor to another silicon processor or memory or other chip, for example in a multi-chip module (MCM). Figure 1(Not shown in the text) (where the term "chip" is generally used interchangeably with "integrated circuit" or "IC" unless the context otherwise indicates). An optical propagation medium can transmit light to another optical coupler 121, which in turn transmits the light to a detector 123 (e.g., a photodiode) for photoelectric conversion. An electrical signal containing at least some data can be amplified by an amplifier 125 and provided to a silicon processor (or other chips in a multi-chip module). In some embodiments, the LED and the detector can be individually coupled to the waveguide, and / or in some embodiments, they can be coupled in parallel as an array. In addition to transmitting light and data from one location to another, the optical waveguide can also split the light into two or more outputs, thereby allowing data fan-out. The optical waveguide or medium can also perform some switching, thereby directing the output from one receiver to another. As those skilled in the art will understand, the optical link can be full-duplex, such that when there are one or more links from the first chip to the second chip, there can also be one or more links from the second chip to the first chip.

[0025] In some embodiments, the LED is a micro LED. In some embodiments, the micro LED differs from a semiconductor laser (SL) as follows: (1) a micro LED does not have an optical resonator structure; (2) the optical output from a micro LED is almost entirely spontaneous emission, while the output from an SL is primarily stimulated emission; (3) the optical output from a micro LED is temporally and spatially incoherent, while the output from an SL is significantly temporally and spatially coherent; (4) a micro LED is designed to be driven down to zero minimum current, while an SL is designed to be driven down to a minimum threshold current, typically at least 1 mA. In some embodiments, the micro LED differs from a standard LED in that (1) it has an emission region smaller than 10 μm × 10 μm; (2) it often has positive and negative contact points on the top and bottom surfaces, while a standard LED typically has both positive and negative contact points on a single surface; and (3) it is typically used in large arrays for display and interconnect applications.

[0026] Generally, the operating speed of a microLED is limited by its capacitance and carrier recombination time (or diffusion capacitance). The capacitor and the drive output impedance form an RC circuit, which is forced to roll off at high frequencies. Carrier lifetime means that the LED needs time to turn off because even after the electrical pulse ends, a few injected carriers must recombine to stop emitting light. In these types of applications, capacitance is usually unimportant because the device size is very small (typically a few nanofarads), but carrier lifetime can be crucial. One can apply a reverse bias to the diode and electrically shape the pulse to pull out carriers, but the structure of the microLED can also be modified to improve speed.

[0027] Generally, the speed of a microLED increases with increasing current level. Carriers recombine in the LED in three ways. At low current levels, recombination is trap-mediated (SRH recombination). At higher current densities, these traps become saturated, and the quantum efficiency of the LED increases as radiative recombination becomes dominant. This radiative recombination rate accelerates with increasing carrier density, thus increasing radiative efficiency and decreasing carrier lifetime. Therefore, the more powerfully a microLED is driven (e.g., at higher current densities), the faster it operates. At high current densities, nonlinear nonradiative mechanisms (e.g., Auger recombination rate) further reduce carrier lifetime, but Auger recombination is a nonradiative process that reduces radiative quantum efficiency. For fast microLEDs with small diameters to increase current density at a given current, traps are relatively unimportant because they are saturated, and the relative significance of nonlinear nonradiative recombination to the radiative recombination rate determines the quantum efficiency of the microLED.

[0028] Typical LED structures include a p-type region, an "active" region where carriers recombine and emit light, and an n-type region, or a combination thereof. Many different LED structures exist, differing in their active region structures. In some embodiments, the active region contains one or more quantum wells (QWs).

[0029] Figure 2A This demonstrates a typical pin LED structure, and Figure 2C Demonstrating optimized doped structures, among which Figure 2B and 2D They also showcased Figure 2A and 2C The associated energy band diagram of the device. Figure 2A The device has a p-doped GaN layer 211 and an n-doped GaN layer 215 that hold an intrinsic region 213 with an InGaN quantum well. Figure 2C The device also includes a p-doped GaN layer 251 and an n-doped GaN layer 255 that clamp the intermediate region. However, Figure 2C The intermediate region of the device is p-type doped and also contains an InGaN quantum well. In some embodiments, the quantum well is positioned physically closer to the p-doped GaN layer than the n-doped GaN layer.

[0030] Figure 2B The band structure diagram shows the conduction band 231 above the valence band 233 across the n-region 221, the intrinsic / depletion region 223, and the p-region 225. The band gap between the conduction and valence bands is generally constant across these regions, with energy levels typically increasing in the intrinsic / depletion region between the n and p regions, making the energy levels in the p-region higher than those in the n-region. Electrons are injected from the n-region 235 into the intrinsic / depletion region, and holes are injected from the p-region 237 into the intrinsic / depletion region, where recombination occurs 239.

[0031] Figure 2DThe band structure diagram also shows the conduction band 261 above the valence band 263, spanning the n-region 221, depletion / p-regions 265a and b, and the p-region 225, respectively. Figure 2B Compared to the band structure diagram, in Figure 2D As can be seen, the depletion / p-region replaces the intrinsic / depletion region, with depletion region 265a adjacent to the n-region and p-region 265b adjacent to p-region 225. The band gap between the conduction band and the valence band is generally constant across these regions, with energy levels typically increasing in the depletion / p-region (mainly in the depletion region) and higher in p-region 225 than in n-region 221.

[0032] and Figure 2B compared to, Figure 2D It also demonstrates that electrons are injected 275 into the p-region above a much thinner depletion region, where recombination typically occurs. In GaN material systems, increased background doping can reduce radiative recombination time by at least one or several orders of magnitude.

[0033] Although Figure 2C The p, p-, and n structures are shown, but one can also dope n-type quantum wells instead of p-type ones. This also improves the speed of micro-LEDs compared to pin structures. The advantage of n-doping over p-doping is that n-doping does not increase defects and does not reduce radiation efficiency. The doping level can be further increased to reduce carrier recombination time at the cost of higher capacitance.

[0034] Some embodiments include... Figure 2C Further modifications to the doped structure can further improve performance. For example, some embodiments use an AlGaN barrier in the n-region to further enhance carrier injection into the p-doped recombination region and prevent hole injection into the n-type region. Some embodiments optimize the InGaN quantum wells in the p-region in terms of number, width, and strain to reduce recombination time. For example, a lower In concentration that pushes the wavelength to shorter wavelengths also increases the speed. Thus, a microLED with a wavelength between 380 nm and 430 nm can inherently be faster than a microLED at a longer wavelength. For a given current, fewer quantum wells also increase the carrier density in the quantum wells. As the carrier density increases, the carrier recombination time decreases more rapidly. Therefore, in some embodiments, the microLED has only one or a few quantum wells. In some embodiments, the quantum well width also becomes smaller. A smaller quantum well width brings electrons and holes closer together, increases the overlap integral, and reduces the radiative recombination time. Some embodiments use a suitable GaN substrate for growth to reduce the built-in electric field in the quantum well, thereby increasing the overlap integral between electrons and holes and thus further reducing the carrier recombination time. One can also reduce the built-in field by achieving a smaller molar fraction of indium, thus again obtaining a faster response in the short wavelength range. A smaller indium concentration also reduces the Auger recombination rate, thereby increasing the quantum efficiency of the LED.

[0035] In some embodiments, the structure optimized for high-speed operation has a small size, with a diameter less than about two micrometers, to increase current density and carrier density. In some embodiments, the structure optimized for high-speed operation has few quantum wells, possibly only one, to maximize carrier density for a given current density. In some embodiments, the indium concentration of the quantum well is low, so the microLED will emit at a shorter wavelength, such as blue or ultraviolet wavelength, because the lower indium concentration will produce a lower piezoelectric field, which increases the overlap integral of the hole-electron wavefunction and thus increases the recombination rate. In some embodiments, the quantum well is very small, typically 2 nm or less, to increase the overlap between electrons and holes. In some embodiments, the quantum well is doped with p-type or n-type to increase the background carrier density.

[0036] Figure 3A Another embodiment of a micro LED is shown. Figure 3A The micro-LED comprises a p-doped GAN top layer 311 and an n-doped GAN bottom layer 313. A multiple quantum well (MQW) stack or structure is located between the p-doped top layer and the n-doped bottom layer. Figure 3A In some embodiments, the MQW stack includes a first GaN barrier layer 315a on top of an n-doped bottom layer, a first InGaN quantum well (QW) layer 317a on top of the first barrier layer, a second GaN barrier layer 315b on top of the first QW layer, a second InGaN QW layer 317b on top of the second barrier layer, and a third GaN barrier layer 315c on top of the second QW layer and below the p-doped top layer. In some embodiments, only a single QW layer may be used, wherein there is only one barrier layer below the single QW layer, or wherein the single QW layer is surrounded by a barrier layer. In some embodiments, more than two QW layers may be used, wherein the barrier layer is located below or surrounds the QW layer.

[0037] In micro-LEDs, the active region where charge carriers recombine and generate light typically comprises a multiple quantum well (MQW) structure, and in some embodiments, is composed of multiple quantum well (MQW) structures. In the case of GaN micro-LEDs, the MQW stack typically comprises a stack of alternating GaN "barrier" layers and InGaN quantum well (QW) layers, and in some embodiments, is composed of alternating GaN "barrier" layers and InGaN quantum well (QW) layers. Injected charge carriers "fall" into the quantum wells (QWs), which serve as recombination regions. The doping in the QWs and barriers is typically at a low density, which tends to minimize nonradiative recombination and maximize DC quantum efficiency.

[0038] In some embodiments, the GaN / InGaN QW and / or barrier are doped with p-type dopants. In some embodiments, the GaN / InGaN QW and / or barrier are doped with n-type dopants. These dopants can generate additional charge carriers, which increases the recombination rate. Because the QW is typically very narrow, the carrier wavefunction extends significantly beyond the QW into the barrier. Therefore, doping that enhances the radiative recombination rate can be introduced into the well itself or into the barrier.

[0039] If the dopant levels are very high, they can lead to defects that increase the nonradiative recombination rate (reflected by the coefficients A, C1, and C2 in Equation 1 below). Because the increased nonradiative recombination rate increases the modulation bandwidth, in some embodiments, the increased nonradiative recombination can be acceptable or even desirable.

[0040] In some embodiments, the GaN / InGaN QW and / or barrier are doped with magnesium (Mg), which acts as a p-type dopant. In another embodiment, the Mg doping concentration is 10. 18 / cm 3 up to 10 20 / cm 3 Within the range.

[0041] In some embodiments, GaN / InGaN QW is doped with silicon (Si), which serves as an n-type dopant. In another embodiment, the Si doping concentration is 10. 18 / cm 3 up to 10 20 / cm 3 Within the range. In some embodiments, the barrier between QW contains highly doped material at 10... 13 up to 10 15 / cm 2 A thin layer of Si within a certain range; this thin layer is sometimes referred to as a "δ-doped" layer.

[0042] In GaN, quantum wells (QWs) are typically generated by sandwiching a thin InGaN QW layer between two GaN "barrier" layers. The greater the In fraction in the InGaN layer, the deeper the quantum well. InGaN QWs grown on GaN are compressively strained relative to GaN, and increasing the Al fraction increases the lattice mismatch between InGaN and GaN. These materials exhibit an embedded electric field associated with a strong piezoelectric effect attributed to their lack of lattice inversion symmetry. An embedded electric field as high as 2.45 MV / cm is generated in the In0.2Ga0.8N / GaN quantum well due to the internal strain along the c-axis caused by the large GaN / InGaN lattice mismatch.

[0043] In InGaN / GaN QW LEDs, the radiative recombination rate is proportional to the square of the electron-hole spatial coupling strength, which in turn is proportional to the overlap of the electron and hole wave functions. The built-in electric field, associated with the piezoelectric effect, reduces the overlap of the electron and hole wave functions, which lowers the recombination rate and thus reduces the modulation bandwidth of the microLED.

[0044] In some embodiments, an external tensile strain is applied parallel to the GaN / InGaN plane such that the applied strain reduces or cancels the strain attributable to the GaN / InGaN lattice mismatch. This reduces or eliminates the piezoelectric field attributable to this strain—such as… Figure 4 The graph illustrates that it displays the band and carrier wavefunctions with and without external strain compensation—this increases the overlap, recombination rate, and modulation bandwidth of the electron-hole wavefunctions.

[0045] In some embodiments, the width of the QW is in the range of 3 nm to 4 nm. In other embodiments, the width of the QW is in the range of 2 nm to 3 nm. A thinner QW can improve the overlap between the hole and electron wave functions, thereby increasing the recombination rate and thus increasing the modulation bandwidth of the microLED. However, charge carriers may escape or jump over the narrow QW more easily, thereby reducing the quantum efficiency (QE). Therefore, there is a trade-off between recombination rate and QE when the QW becomes thinner.

[0046] The QW depth, strain, and piezoelectric field increase with increasing Al fraction. Conversely, decreasing the Al fraction decreases the QW depth, strain, and piezoelectric field, which increases the radiative recombination rate, such as... Figure 5 As can be seen, it exhibits an energy band with varying Al doping. Therefore, from the perspective of radiative recombination rate and modulation bandwidth, it is desirable to reduce the Al fraction. However, if the QW is too shallow, carriers inside the well can "leak" out of the well, resulting in a reduction in the number of carriers available for radiative recombination in the QW region.

[0047] Based on this tradeoff between strain-induced recombination rate reduction and leakage current from thermally excited carriers, there exists an optimal QW depth. In some embodiments, the In percentage in the QW results in a QW depth ranging from kT to 5kT, where k is the Boltzmann constant and T is the absolute temperature in Kelvin.

[0048] Figure 3B This demonstrates yet another embodiment of a micro LED. Figure 3B Implementation examples and Figure 3A The implementation is the same, except that Figure 3B The embodiments include a mezzanine (IL) in the MQW stack. Figure 3BThe diagram describes two interlayers, wherein the first interlayer 411a is located between the first QW layer 317a and the second barrier layer 315b, and the second interlayer 411b is located between the second QW layer 317b and the third barrier layer 315c.

[0049] In some embodiments, the IL is subjected to tensile strain together with GaN, which compensates for some or all of the compressive strain from the InGaN / GaN lattice mismatch. In some embodiments, the piezoelectric effect from the IL cancels some or all of the piezoelectric field attributable to the InGaN / GaN lattice mismatch and reduces the built-in static charge at the layer interface. Therefore, the potential well in the InGaN layer tends to be in a flat-band state and maximizes the overlap of electron-hole wave functions, thereby increasing the radiative recombination rate.

[0050] In some embodiments, IL is made of AlGaN to produce In x Ga 1-x N / Al y Ga 1-y N / GaN structure. In some embodiments, the thickness of the AlGaN IL is in the range of 0 to 3 nm. The AlGaN IL is subjected to tensile strain together with the adjacent GaN layer, while the InGaN layer is subjected to compressive strain together with the adjacent GaN layer. Therefore, the AlGaN layer placed next to the InGaN layer can compensate for the InGaN / GaN lattice mismatch strain. The AlGaN IL can reduce the outward diffusion of indium from the QW and achieve higher temperature growth of the GaN barrier, which allows for annealing of the QW and IL, improves material quality, and reduces impurity doping.

[0051] The rate equation for injected carriers in the active region of a micro LED can be described by the following equation:

[0052] dN / dt = AN + BN(N + p) + C1N(p + N) 2 + C2N 2 (p + N) (1)

[0053] Where N is the injected carrier density (carriers per unit volume), p is the carrier density in the active region, A is the Shockley-Read-Hall (SRH) recombination coefficient, B is the radiative recombination coefficient, and C1 and C2 are two Auger recombination coefficients. A, C1, and C2 reflect non-radiative recombination mechanisms, while B reflects radiative recombination mechanisms.

[0054] The relationship between current density J and the rate of change of carrier density is as follows:

[0055] J = ed eff dN / dt (2)

[0056] Where e is the electron charge, and d eff It is the effective thickness of the composite region. Combining equations (1) and (2), the modulation frequency pair J can be calculated, for example, as... Figure 6 It is displayed in the middle.

[0057] The internal quantum efficiency (IQE) of a micro LED is given by the following equation:

[0058] IQE = BN(N + p) / [AN + BN(N + p) + C1N(p + N) 2 + C2N 2 (p + N)] (3)

[0059] Combining equations (2) and (3) allows for the calculation of IQE with respect to J, for example, as Figure 7 As shown in the diagram, there exists a current density that maximizes QE. In high-quality LEDs used in lighting or display applications, this typically occurs at <10 A / cm². 2 .

[0060] The speed and quantum efficiency of micro-LEDs are maximized by maximizing the value of the B coefficient in equations (1) and (3). The value of B is increased by maximizing the overlap of the hole and electron wave functions. Several embodiments described herein focus on maximizing this overlap.

[0061] Figure 8 This demonstrates the relationship between the time constant of a microLED and its driving current density. For high-speed inter-chip communication, microLEDs must be driven with sufficiently high current densities to achieve the required modulation bandwidth. In some embodiments, the microLEDs are driven at current densities such that the modulation bandwidth of the microLED is between half the bit rate and the bit rate.

[0062] In some embodiments, the microLED is driven with a sufficiently high current density such that Auger recombination leads to a significant reduction in carrier lifetime. In some embodiments, high-frequency components of the drive signal are emphasized relative to low-frequency components, and the microLED operates with a current density whose modulation bandwidth is less than half of the bit rate.

[0063] Figure 10 The table describes the various trade-offs in the design of micro-LEDs with typical parameters.

[0064] Generally, higher doping also reduces nonradiative recombination time. This further shortens carrier lifetime and increases velocity, but at the cost of reduced quantum efficiency. Similarly, in these very short-distance applications where waveguide propagation losses are virtually nonexistent, quantum efficiency may be less important. Essentially, there is a trade-off between quantum efficiency and velocity when the LED response is limited by radiative lifetime versus radiative recombination time. Therefore, in some embodiments, the LED velocity is increased (significantly in some embodiments) at the cost of lower brightness.

[0065] Rapid recombination centers can be induced in LEDs through several processes. These include lower-temperature growth of crystals in the intrinsic region, proton implantation, intentionally induced defect density using dislocations in the lattice, or even roughening the etched surface or creating a larger exposed surface.

[0066] Generally, smaller microLEDs tend to have lower quantum efficiency because charge carriers diffuse and recombine at the etched surface. This reduces carrier lifetime and therefore also increases the speed of the microLED. This effect can be increased by etching holes in the structure that expose the sidewalls and recombination centers. Figure 9 Showcasing miniature LEDs with etched holes. Figure 9 Examples show microLEDs with a generally cylindrical shape extending between a circular substrate 919 and a circular top 917. The microLED may include a substrate layer 913 (which may be, for example, an n GaN layer) extending upward from the circular substrate and a top layer 911 (which may be, for example, a p GaN layer) extending downward from the circular top. An intermediate layer 915 is located between the substrate layer and the top layer, and, as will be understood, the intermediate layer may provide an intrinsic depletion region or a p-depletion region.

[0067] Figure 9 The microLED also includes etched holes, such as etched hole 921, extending from the circular top to the circular bottom. Therefore, the etched holes provide a perforation from the top surface through the microLED to the bottom surface. Figure 9 In this process, etched holes have a circular cross-section, thereby forming cylindrical holes, which are typically arranged in a square or rhomboid pattern. Etched holes can induce nonradiative recombination at the exposed surface to reduce carrier lifetime and thus increase velocity. This provides a more controllable method than proton implantation or lower-temperature growth. In this context, various structures can be used to increase the surface area when the device is etched to form a micro-LED. These include etching multiple pillars and / or etched holes in some embodiments, as illustrated in the figures. In some embodiments, other shapes, such as star-shaped or rough-edged, may also be used or alternatively.

[0068] Although the invention has been discussed with respect to various embodiments, it should be recognized that the invention includes novel and non-obvious claims supported by this disclosure.

Claims

1. An optical communication system for transmitting information provided by a processor to another region of the processor or another chip in a multi-chip module, comprising: LEDs, which are associated with the processor; An LED driver is used to modulate the optical output power of the LED so that the LED will generate light based on data provided from the processor to the LED driver; A detector used to perform photoelectric conversion using the light; and An optical waveguide that optically couples light from the LED to the detector; The LEDs mentioned above include: p-type layer; n-type layer; and An intermediate region, between the p-type layer and the n-type layer, wherein at least one quantum well layer and at least one barrier layer are alternately distributed, wherein the at least one quantum well layer is entirely doped with a dopant and is positioned physically closer to the p-type layer than the n-type layer, such that the intermediate region has a depletion region adjacent to the n-type layer and a p-region adjacent to the p-type layer, wherein the depletion region is thinner than the p-region and the energy level increases more in the depletion region than in the p-region.

2. The system of claim 1, wherein the at least one barrier layer is doped with the dopant.

3. The system according to claim 1 or 2, wherein the dopant is a p-type dopant.

4. The system of claim 3, wherein the p-type dopant comprises magnesium.

5. The system according to claim 4, wherein the magnesium doping concentration is 10. 18 / cm 3 Up to 10 20 / cm 3 Within the range.

6. The system according to claim 1 or 2, wherein the dopant is an n-type dopant.

7. The system of claim 6, wherein the n-type dopant comprises silicon.

8. The system of claim 7, wherein the silicon doping concentration is 10. 18 / cm 3 Up to 10 20 / cm 3 Within the range.

9. The system of claim 1, wherein the at least one quantum well layer comprises at least two quantum well layers, wherein the at least two quantum well layers are separated by one of the at least one barrier layer, and said at least one barrier layer is doped with 10 13 / cm 2 Up to 10 15 / cm 2 Silicon within the range.

10. The system of claim 1, wherein the quantum well layer comprises InGaN and the barrier layer comprises GaN, and wherein an external tensile strain is applied parallel to a plane defined by the interface of the quantum well layer and the barrier layer, such that the applied strain is reduced to strain attributable to lattice mismatch between the InGaN of the quantum well layer and the GaN of the barrier layer.

11. The system of claim 10, wherein the width of the quantum well layer is in the range of 2 nm to 3 nm.

12. The system of claim 10, wherein the width of the quantum well layer is in the range of 3 nm to 4 nm.

13. The system of claim 1, further comprising at least one interlayer located between at least one of the at least one quantum well layers and at least one of the at least one barrier layer, the barrier layer comprising GaN, and the interlayer being subjected to tensile strain together with the GaN.

14. The system of claim 13, wherein the interlayer comprises AlGaN.

15. The system of claim 1, wherein the LED is a micro LED.

16. The system of claim 1, further comprising: Another LED, which is associated with other areas of the processor or other chips in the multi-chip module; Another LED driver is used to modulate the optical output power of the other LED, such that the other LED will generate light based on data provided to the other LED driver from other areas of the processor or other chips in the multi-chip module; and Another detector is used to perform photoelectric conversion using the light from the other LED; The other LED mentioned above includes: p-type layer; n-type layer; At least one quantum well layer and at least one barrier layer are located between the p-type layer and the n-type layer, wherein at least one of the quantum well layer and the barrier layer is doped with a dopant.

17. The system of claim 16, wherein the optical waveguide optically couples light from the other LED to the other detector.

Images