JP2000106454A - Device for emitting radiation with high efficiency and fabrication thereof - Google Patents

Abstract

PROBLEM TO BE SOLVED: To emit radiation with high emission efficiency by a structure wherein the device includes a cavity having an active layer emitting radiation, the cavity has a space for confining radiation and the device includes a one or more edge having a substantially random diffraction grating structure. SOLUTION: A region (active layer) 10 where charge carriers encounter to emit radiation has irregular surface structure 122 of substantially random diffraction grating structure where surface recoupling may take place. More specifically, the radiation emitting device has a cavity including a radiation confining space for confining charge carriers in a sub-space in the cavity smaller than the radiation confining space. The device has a cavity including at least one mesa edge of substantially random diffraction grating structure for defining the radiation confining space.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

FIELD OF THE INVENTION The present invention relates to the field of radiation emitting devices. More specifically, a semiconductor device that emits light of a predetermined wavelength with high efficiency is disclosed. Also disclosed are methods of making such devices, as well as uses for the devices.

[0002]

2. Description of the Related Art Semiconductor devices capable of emitting non-coherent or coherent light are known in the art. Many publications on semiconductor-based light emitting devices relate to light emitting diodes (LEDs), microcavity LEDs or microcavity lasers, or vertical cavity surface emitting lasers. Examples of such publications are given below: De Neve, J.M. Blondelle,
R. Baets, P.M. Demeester, P .; V
an Daele, G .; Borghs, IEEE P
hoton. Technol. Lett. 7 287 (1995); F. Schubert, N .; E. FIG. J.
Hunt, R .; J. Malik, M .; Mikovi
c, D. L. Miller, “Temperature and Modulation Characteristics of Cavity Light-Emitting Diodes (Temperatur
e and Modulation Characte
risks of Resonant-Cavit
y Light-Emitting Diode
s) ", Journal of Lightwave
Technology, 14 (7), 1721-17
P. 29 (1996); Yamauchi and Y .; Arakaw
a, Enhanced and Inhibited Spontaneous Emission in GaAs / AlGaAs Vertical Microcavity Laser with Two Quantum Wells (Enhanced and Inhibit)
ed spontaneous emission i
nGaAs / AlGaAs vertical mi
crocavity lasers with two
Kindsof quantum wells), A
ppl. Phys. Lett. 58 (21), 2nd
339 (1991); J. de Lyon, J.M. M. Wood
all, D.E. T. McInturff, R.A. J.
S. Bates, J.M. A. Kash, P .;
D. Kirchner, and F.M. Cardon
e, "Doping concentration dependence of radiance and light modulation bandwidth in carbon doped Ga _0.1 In _0.49 P / GaAs light emitting diodes grown by gas source molecular beam epitaxy.
tion dependency ofradianc
e and optical modulation
bandwidth in carbon-doped
Ga _0.1 In _0.49 P / GaAs lighgt-emi
ting diodes green by gas
source molecular beam epi
taxy) ", Appl. Phys. Lett.
60 (3), 353-355 (1992); G. FIG. Deppe, J .; C. Campbe
11, R.I. Kuchibhotla, T .; J. R
ogers, B.A. G. FIG. Streetman, "Optically Coupled Mirror Quantum Well InGaAs-GaAs.
Light-emitting diodes (optically-coupled)
mirror-quantum well InGa
As-GaAs light emitting di
ode) ", Electron. Lett. 26
(20), p. 1665 (1990); Ettenberg, M .; G. FIG. Harve
y, D. R. Patterson, “Linear, high-speed, high-power strained quantum well LEDs (Linear,
High-Speed, High-Power St
rained Quantum-Well LED '
s) ", IEEE Photon. Technol.
Lett. 4 (1), page 27 (1992); U.S. Pat. No. 5,089,860 (Deppe
Et al., Feb. 18, 1992) "Quantum well device with control over spontaneous photon emission and method of manufacturing the same. (Quantum well device with)
controloft spontanous ph
oton emission, andmethod
of manufacturing same) ".

It is known in the art that light emission from an electroluminescent device or light emitting semiconductor diode (LED) is limited by total internal reflection that occurs at the interface between the semiconductor substrate on which the device is fabricated and its surrounding medium. Are known. In most cases, it is intended to emit light into air at a refractive index of one. Semiconductors typically have a refractive index n _{s of} 3-4. For example, GaAs
Has a refractive index n _s = 3.65. Snell's law is a semiconductor at an angle smaller than the critical angle _{θc = arcsin (1 / n s} ) - Only photons reaching the air interface is defines that can be emitted into the air. All other photons are totally internally reflected at the semiconductor-air interface, thus
Finally, the photons remain on the semiconductor substrate until they are re-absorbed. For GaAs, the critical angle for total internal reflection is 16 degrees. Thus, total internal reflection limits the number of photons exiting the semiconductor substrate to those that reach the semiconductor-air interface at an angle less than 16 degrees. Only about 2% of the photons generated inside the semiconductor satisfy this condition.

Some prior inventions have proposed increasing the probability of emitting photons generated in the LED. For example, US-A-5 226 05 by Cho et al.
In the microcavity type light emitting diode as described in 3, the active layer of the light emitting device is arranged in the microcavity. The cavity affects the emission of photons: more photons are produced at angles smaller than the critical angle θ _c . In this method, 1
Efficiencies of 5% or more have been achieved.

A second way to increase the efficiency of an LED is
To reabsorb photons that cannot be emitted from the semiconductor.
If reabsorption occurs in the active layer of the LED, the electron-hole pairs generated during the reabsorption will again recombine radiatively (or radiatively recombine), giving the opportunity to re-emit photons. Again, 2% of these photons are emitted and the rest can be reabsorbed. Multiple reabsorption and re-emission phenomena
Standard LEDs have been shown to provide efficiencies on the order of 10%, and have been shown to increase the efficiency of certain microcavity LEDs to 23%. The problem with this technique is that it has to wait for multiple re-absorptions and re-releases, so the technique is necessarily slow.

A third approach is to shape the semiconductor surface of the light emitting device such that more of the generated radiation reaches the semiconductor-air interface within a critical angle. Semiconductor
The optimal shape for the air interface is a hemispherical surface, in which case the emission area is limited to a small spot at the center of the virtual full sphere from which the hemisphere is cut. Other shapes have been proposed. In U.S. Pat. No. 5,087,949, Haitz describes a more practical structure, ie, V
-Proposes a set of V-groves formed in the substrate such that the normal to the facets of the gloves is oriented substantially perpendicular to the light-emitting area.
In US-A-5 349 211 Kato proposes a structure in which the side walls or edges of the substrate are shaped such that some of the photons reflected from the normal light output interface are emitted through these side walls. are doing. Egalon and Rogowski propose a shape of the sidewall relative to the substrate (rather than just the mesas) that redirects some of the photons to an angle that can exit through a regular light output surface. All these proposed structures assume that the LED substrate is fairly thick and transparent to the photons emitted by the diode.

US-A-5 by Yamanaka et al.
In accordance with the teachings of US Pat. The lateral edges or facets of the device preferably have a 45 degree angle. Photons generated in a direction parallel to the light output surface (edge) are reflected by the mesa edges of such cavities in a direction substantially perpendicular to the light output surface, so that more photons are generated. It can exit from the cavity.

A fourth way to increase LED efficiency is to provide a device structure that can redirect photons multiple times before the photons are reabsorbed. This object is achieved by providing a surface edge of the device that includes a surface portion having a different angle from the main semiconductor substrate-air interface. Each time a photon strikes such a surface, it is redirected to a new propagation angle. In this manner, photons traveling in a direction that is not favorable for emission to air have a certain probability of being redirected to a convenient angle after multiple reflections at such a surface. US-
A-3 379217 Bergh and Sau
1 proposes to form morphological irregularities (or irregularities) on the light emitting surface or the opposite surface (light refracting surface) of an LED with a transparent substrate. Noguchi et al. EP
-A-0404565 proposes texturing (or texturing) the side walls or edges of the substrate on which the light-emitting device is made.

[0009] The method described above applies to light emitting devices in which light is emitted through a substrate. Therefore, it is necessary to manufacture a light emitting device on a transparent substrate. Noguchi et al., E
The teachings of the Galon et al., as well as Kato et al. patents are applicable only to single light emitting devices and not to arrays of light emitting devices. The invention disclosed by Haitz et al. (US Pat. No. 5,087,949) is applicable to arrays, but requires the presence of a rather thick substrate, and the spacing of the light emitting devices in the array is on the order of the thickness of the substrate. Must.

A method for fabricating a light emitting device whose substrate need not be transparent is described in US Pat.
358 880. The present invention provides an original substrate, which may not be transparent, includes a transparent conductive layer such as indium tin oxide and a transparent epoxy plus glass-like host.
-Like host). Further, the disclosed invention aims to form a closed cavity by etching the active layer of a light emitting device and covering the etched mesa sidewalls with a dielectric material and metal. In this way, photons that cannot exit through the transparent window are retained in the cavity by multiple reflections, re-absorptions, and eventual re-emissions.

For example, GaAs, InAs, or Al
When applied to material systems such as As and combinations thereof, the above invention can only be used for light emitting devices with large diameters. This is because these conventional inventions rely on etching the active layer of the light emitting device. Such etching of a material having a high surface recombination rate, such as GaAs, results in significant parasitic surface recombination currents. Surface recombination is a phenomenon in which two charge carriers (electrons and holes) of the opposite type recombine without emitting light in traps such as the surface of a material. The recombination rate is proportional to the number of traps available as well as the density of charge carriers available for recombination. The surface of a III-V semiconductor such as GaAs has a large number of traps (about 10 ¹⁴ cm ⁻² ). Thus, the surface recombination rate is very high. Thus, there is a problem associated with fabricating high efficiency light emitting devices, where the loss of charge carriers due to recombination reduces the overall light emission efficiency of the device. One solution is to inject electrons and holes from electrodes that are physically far from the free surface (edge), which often coincides with the cleavage plane of the device. Due to the large distance between the edge of the device and the contacts, the resulting device is necessarily large, typically at least several hundred microns in diameter. Therefore, such devices are not suitable for integration in large arrays. In addition, the large area increases the capacitance, and thus slows down the operation.

Thus, the prior art does not disclose a high efficiency light emitting device that can be integrated as a small device in a high density array of light emitting devices.

[0013] High efficiency LEDs are needed for applications such as optical communications. Optical communication, for a given energy supply,
It replaces telecommunications in many areas because it can provide longer distance interconnects.
The minimum distance over which an optical interconnect is advantageous over an electrical interconnect is very short, for example several centimeters. The energy consumption required for a given data bandwidth is an important factor in determining the minimum distance that an optical interconnect can compete with an electrical interconnect. The bandwidth transmitted by the optical interconnect system is the product of the serial bandwidth per interconnect channel and the number of parallel optical channels. The optical case has the advantage that the number of channels can be much higher for the electrical interconnect.
This potential massively parallel processing (parallelis)
One of the conditions under which m) can be achieved is to keep the power consumption per channel small: heat dissipation must be kept controllable. Therefore, there is a need for high efficiency LEDs that can be integrated in a high density array.

[0014]

The object of the present invention is to disclose a device (or element) which emits radiation (preferably light) having a high radiation emission efficiency. The invention further aims at disclosing a device that emits radiation (preferably light), which can be incorporated as a small device into an array of such devices.

According to the object of the invention, devices emitting radiation (preferably light) can be arranged in a high-density array.

According to another object of the present invention, the outcoupling (or outcoupling) efficiency of a radiation (preferably light) emitting device is improved so that for a given radiation output power Reduce power consumption.

According to yet another object of the present invention,
The speed of the device emitting the radiation (preferably light) is increased, thus increasing the serial bandwidth per optical channel. The present invention further aims to disclose a light emitting device exhibiting uniform radiation emission characteristics. Another one
In accordance with one object of the present invention, a light emitting device having a low resistance contact path is disclosed. In this way, high wall-plug efficiency is achieved. In this regard, an inventive embodiment of the invention envisages an electrical contact through at least one hole on the mirror side of the light emitting device.
In such an embodiment, the mirror is preferably not conductive.

The disclosed light emitting devices can be used in parallel optical interconnects. Light emitting devices according to the present invention can be arranged in a high density array. Optical outcoupling efficiency is greatly improved, which reduces power consumption for a given optical output power. Third, the speed of the device is increased, thus increasing the serial bandwidth per optical channel.

The features of the device (diode) of the present invention can be applied to many light-emitting devices such as high-efficiency LEDs or microcavity LEDs.

The light emitting device (diode) of the present invention comprises:
It can be used for applications such as display technology where two-dimensional LED arrays (especially low power arrays) are useful. Active matrix displays using liquid crystals (eg integrated in CMOS circuits) can be replaced by LED arrays. High-density and bright one-dimensional LED arrays are useful for printing and copying applications and the like.

[0021] For single LED applications, it is also important to maximize the number of photons emitted from the light emitting surface. First, the light intensity (brightness) per unit area is higher, which is useful in many applications. Further, packaging costs can be reduced. actually,
To achieve large overall efficiencies, many conventional LEs
D is to emit light from two or more surfaces of the LED,
Requires elaborate packages that include cavities with mirrors.

[0022]

SUMMARY OF THE INVENTION For the purposes of the present invention, devices that emit radiation (preferably light) having high radiation emission efficiencies and methods of making such devices are disclosed. In accordance with this aspect of the present invention, the outcoupling efficiency of a radiation (preferably light) emitting device is improved, thereby reducing the power consumption of the device for a given radiation output power.

In another object of the invention, a radiation (preferably light) emitting device is disclosed which can be fabricated as an array of such devices as a small device and a method of making such a device. Is done.

The radiation (preferably light) emitting devices of the present invention can have increased speed performance, thus increasing the serial bandwidth per optical channel. The radiation (preferably light) emitting devices of the present invention may exhibit uniform radiation emission characteristics.

For the purposes of this patent application, some terms are defined below. The cavity (or cavity or resonator) of the device that emits radiation is the space in the device that contains the active layer of the device, the space surrounded by the edges, wherein at least one edge has reflective properties. This space further has a predetermined wavelength,
It has at least one such window or transparent edge so that photons within a given wavelength band can exit through the window. The active layer of a device that emits radiation is the area of the device where the charge carriers (eg, electrons and holes) of the device meet to produce the radiation (light). The mesa edge is the edge of the device that emits radiation, defined by a structure etched into the substrate or a structure grown on the substrate. Typically, devices with mesa edges have a table-like or truncated pyramid-like shape. In this patent application, the term "edge" will be understood as the side wall or surface or interface of the device of the invention. Of course, a given wavelength will be understood to include a limited wavelength band around the given wavelength. The term "confined" through form and function means that without additional means or additional structural features (such as extended large isolation features in an array of devices), each device in the array of devices is electrically It means that they can be individually addressed via signal connections. Such confinement through form and function can occur, for example, when the device is integrated into a single thin-film semiconductor.

In a first aspect of the present invention, a device for emitting radiation at a predetermined wavelength is disclosed. The device is mounted on a support substrate, the support substrate being transparent to the radiation, and preferably including a fiber optic faceplate. If not, this results in a denser array, since the pitch of the array of devices is limited by the thickness of the glass. The device may have an edge with a rough (or rough) reflective surface state that is transparent to the radiation. Further, the edge that is transparent to the radiation may have an uneven surface state. In a preferred embodiment of the present invention, the uneven surface state exists as a substantially random diffraction grating structure.

In a second aspect of the invention, a device is disclosed that emits radiation at a predetermined wavelength, the device having a cavity having an active layer, wherein the radiation is generated by recombination of charge carriers. , The cavity includes at least one edge having a substantially random grating structure. The edges of the device are radiation and / or
Alternatively, it defines a region or space for confining charge carriers. The edge having a substantially random grating structure may extend as at least one edge of a waveguide forming portion of the device. The radiation emitting device of the present invention may include a cavity having a space for confining the radiation, the space containing the charge carrier in a smaller sub-space than the radiation confining space within the cavity, the confinement feature for the charge carrier. including.
According to such an embodiment of the present invention, the waveguide forming portion of the device may be a radiation confinement region of the device that is larger than the electrical confinement region. An edge of the device extending as a waveguide forming portion of the device, the edge having a substantially random, partially reflective grating structure may also be adjacent to the active layer or It may extend into the layer. Thus, a method for avoiding emission of photons from the mesa region due to the guided mode is disclosed. One of the edges of the cavity of the device may be a mesa edge. According to another preferred embodiment of the invention, the region where the charge carriers meet to generate radiation (the active layer) has a non-free (or substantially non-free) surface or edge where surface recombination can occur. Have. Accordingly, the radiation emitting device of the present invention has a cavity containing a radiation confinement space that confines the charge carriers in a smaller subspace within the cavity than the radiation confinement space.
Including a confinement feature for the charge carrier. A device according to this aspect of the invention may have a cavity including at least one mesa edge for defining the radiation confinement space, the device further comprising a ring of the dielectric material. It may be contained in the cavity. Thus, according to this preferred embodiment of the invention, the device is optically and electrically confined, the electric confinement space being a subspace of the optical confinement space. A device according to this embodiment of the invention may be confined in form and function without additional means or additional structural features, such as extended large isolation features in the device. The device can operate while significantly reducing the effects of charge carrier recombination within the device. Such confinement in form and function can be achieved, for example, when the device is integrated in a single thin-film semiconductor. In a preferred embodiment of the invention, confinement in form and function is provided by double mesa edges. Electrical confinement of the device of the present invention to a space that is a subspace of the optical confinement space can be achieved by the presence of a ring of dielectric material within the cavity of the device. The double mesa edge may include a first mesa edge above a second mesa edge where the active layer of the device is located. The device may also include an active layer located above the ring of dielectric material within the cavity. Also disclosed is a double mesa that includes an isolation plus metal coating over the mesa edge. Such is important for reducing surface recombination in small radiation emitting devices. Combinations of a single mesa and an oxidized opening (oxidized opening) are also disclosed. The surface or edge of the cavity can be textured (or textured)
Alternatively, irregularities may be formed (or may be roughened). This texturing may be shorter or longer than the wavelength of the radiation and may be a substantially random grating structure pattern. Texturing or roughening (or roughening) makes the photons more scattered inside the cavity, thus again improving the efficiency and speed of the device.

The device of the present invention does not rely on resonance for its operation, and its performance is not significantly affected by variations in growth and process. Therefore, the device of the present invention is suitable for integration in a large-scale array.
Furthermore, they exhibit high efficiency properties, especially at high speeds. In a third aspect of the invention, an array of devices is disclosed, wherein the individual devices of the array are confined in form and function. At least one edge of the devices of the array may have a substantially random grating structure, and at least one edge may extend as a waveguide forming portion of the device. The radiation emitting device of the array may have a cavity containing a radiation confinement space, wherein the radiation confinement space traps the charge carriers in a smaller sub-space than the radiation confinement space within the cavity. Includes confinement features. The grooves between the individual devices of the array may have a substantially random grating structure. An array of devices may have a single electrical anode contact and a single electrical cathode contact. The devices of the array can be thinned and attached to a support substrate. Microlenses or microlens arrays may be arranged in an array of light emitting devices to enhance light output efficiency and optimize beam profiles.

In a fourth aspect of the present invention, a device for emitting radiation is disclosed, wherein the device has a cavity in which the radiation originates, the device comprising at least two adjacent or adjacent edges of the cavity. Wherein the cross-section includes edges that substantially form a triangle, the angle between the edges is less than 45 degrees, and at least one of the edges has a transparent portion. In one embodiment, the two edges may be one adjacent or adjacent to another, one edge being transparent and the other being reflective. Non-transparent edges may be mesa edges. Further, the transparent edge may be a mesa edge. Thus, both edges or one of the edges may extend essentially the entire thickness of the device, the thickness of the device being measured between the mesa edge and the transparent window edge. . The edge may have an uneven surface state. In a preferred embodiment of this aspect of the invention, the mesa edges are tapered at an angle greater than 45 degrees. In this case, not only photons having a specific direction are emitted. Photons also
Exiting after fewer passes (or transit paths), less recombination and re-emission occurs. Thus, faster device performance is achieved. In a fifth aspect of the invention, a thin film light emitting device has a metal mirror, preferably covered with a dielectric, and electrical contacts are provided through the reflective edge or mirror. A support substrate may be disposed on the mirror side of the thin film semiconductor device,
Light may exit through the opposite side. The support substrate may be electrically conductive. It may also be a heat sink for the LED. Thus, according to this fifth aspect of the present invention, a light emitting device having a low resistance contact path is disclosed. In this way, high wall plug efficiency is achieved. In this regard, electrical contacts are envisaged through at least one hole on the mirror side of the light emitting device. In such embodiments,
The mirror is preferably not conductive.

The features of the above-described subject matter of the present invention may be combined in any way to obtain a very efficient device for emitting radiation at a given wavelength.

In a sixth aspect of the present invention, a method for texturing at least a portion of at least one surface of a substrate is disclosed. The method includes applying, preferably substantially randomly, an overlying material covering the portion of the surface, the overlying material having a pattern with substantially randomly distributed aperture features; Etching the surface using the upper layer material as an etching mask, whereby the mask includes a substantially random mask pattern. The pattern of overlying material with substantially randomly distributed aperture features may have any configuration or shape. The pattern may have holes, which are randomly distributed. The holes may also partially overlap each other. A substantially random mask pattern may include regular repetitions of random sub-patterns.

The step of applying the overlying material with substantially randomly distributed aperture features includes the steps of applying a layer of photoresist material and a lithographic mask having the photoresist material having a substantially random mask pattern. And a sub-step of developing (or exposing) the photoresist thereafter. The developed photoresist then forms an overlying material with substantially randomly distributed features. UV light, deep UV light,
X-rays, electron beams, or any type of lithography may be used. The mask may be a metal mask for a contact aligner (or close matcher) or a stepper. It is a specular noise pattern of a coherent light source such as a laser.
(se pattern).

The method also includes applying substantially randomly distributed particles to the surface; reducing the size of the particles; and thereafter, etching the surface while using the particles as an etching mask. A step may be included. Therefore, this may mean that the etching technique used does not attack the particles, or
It means that the particles have resistance to the etchant used or have been subjected to a treatment for imparting resistance to the particles to the etchant used.

The etching step can create irregularities on the surface so that the surface becomes diffusive to electromagnetic radiation impinging on the surface. The etching step may include the step of forming pillars in the substrate by etching. In a preferred embodiment of the present invention,
The particles are applied in the form of a monolayer, two-dimensional partial coverage (or partial coverage) of the surface. The applied partial coverage may be about 60% of the surface, and after reducing the size of the particles, obtain a coverage of about 50% of the surface.

The method includes the steps of applying a second layer of mask material to the surface; developing a pattern in the second mask material while reducing the size of the particles; The method may further include a step of etching the surface while using the surface as an etching mask.

In a seventh aspect of the invention, the method further comprises forming a device on the substrate that emits radiation at a predetermined wavelength, wherein the device has a cavity and the surface has a device. At least a portion of one edge of the cavity. According to this aspect of the invention,
The reduced size of the particles may range from 50% to 200% of the wavelength of the radiation at the substrate. The edges of the device having a rough surface state may be transmissive and may be the reflective edges of the device. The devices of the first, second and third aspects of the present invention may be textured by this method. The textured edge then has a substantially random diffractive photon structure. Any embodiment of the device or method according to the various aspects of the invention may be combined to achieve an advantageous light emitting device.

DETAILED DESCRIPTION OF THE INVENTION For the purpose of teaching the present invention, preferred embodiments of the method and device of the present invention will now be described. Other alternative or equivalent embodiments of the invention may be conceived and made without departing from the scope of the invention, which is limited only by the appended claims, and the true concept of the invention. That will be apparent to those skilled in the art.

FIG. 1a shows the layer structure of a light emitting diode (LED) according to an embodiment of the device of the invention. This layer structure assumes the device described in the detailed description of the invention. Although a layer structure is shown, the layers are formed of a semiconductor material (preferably a III-V semiconductor material) and different layers may be formed of different semiconductors. The layers may be deposited on the substrate or grown epitaxially. The substrate on which the layer is formed is removed and the layer structure is deposited on the support substrate 17. The support substrate may be transparent. In another embodiment, the layer structure is first deposited on a substrate,
Thereafter, the original substrate may be removed. Active layer 10 is surrounded by cladding layers 11 and 12 and by heavily doped contact layers 13 and 14. Layers 11 and 13 are of the same doping type, as are layers 12 and 14. Layers 11 and 13
The doping type of layers 12 and 14 is such that when a suitable bias is applied between layers 13 and 14, both types of charge carriers (electrons and holes) can be injected into the active layer. And the opposite. Layers 11 and 1
The bandgap energy of the material from which 2 is formed is usually greater than the energy of the photons emitted in the active layer. Thus, the absorption of the resulting photons is limited to the thin contact layer and to the absorption at the contact layer surfaces 15 and 16. The total thickness of the semiconductor structure consisting of layers 10, 11, 12, 13, and 14 is small, typically 0.5 to 5 microns, so these LEDs are called thin-film LEDs. To support thin semiconductor films,
The transparent support 17 is attached. Preferred supports are
It is a fiber optic face plate rather than a glass substrate. A preferred means for attachment is thermally stable and transparent epoxy 18 (EPOXY Technology).
EPO-TEK 353ND from es) or transparent polyimide. To avoid having to flow the current in the LED to the active layer only by lateral flow through layers 12 and 14 (which results in forbidden series resistance), a portion of surface 16 is metallized. Cover with 19. This metal layer functions as a contact layer to the semiconductor layer 14 and, at the same time, functions as a low-resistance conductive layer for current. The metal sheet has an opening for outputting light. A method for forming a structure including a thin semiconductor film attached to a host substrate can be referred to literatures. See, for example, co-pending patent application EP-98870164.5 (19), which is incorporated herein by reference.
(Filed July 28, 98) discloses a suitable method. US-A-5 358 880 discloses another suitable method.

The layer structure of FIG. 1a is described in patent application EP-9888709, which is incorporated herein by reference.
It can be made according to the teachings of 9.3. FIG. 1b illustrates an alternative structure of a light emitting diode according to the present invention. A support substrate is arranged on the reflective side (mirror side of the cavity of the light-emitting diode structure), which cavity contains an active layer in which photons are generated by radiative recombination of charge carriers. The mirror is preferably a mirror of metal 23 covered with dielectric 22, which is not electrically conductive. There is one or several electrical contacts 14 through the mirror to conduct current through the mirror. The mirror may be coupled to a non-transparent support 20. This support may be electrically conductive and serve as an electrode contact for a light emitting diode. It may also be thermally conductive and designed to include active cooling, for example, to function as a heat sink for the LED in high power applications. The attachment of the metal 23 to the support 20 can be performed by soldering. Alternatively, wafer bonding can also be used. Contact layers 13 and 14 may be removed from regions where there is no contact to the LED to reduce parasitic absorption (or parasitic neutron absorption). Contact 19
May be used for bonding with conventional wire bonding methods. In subsequent embodiments shown in FIGS. 2-11, the support may be either transparent, as shown in FIG. 1a, or attached to the mirror side of the LED, as shown in FIG. 1b. .

FIG. 2 illustrates an embodiment of a light emitting diode according to the present invention. The mesa edge 21 shown in FIG. 2 is inclined and covered with the dielectric layer 22 and the metal layer 23 except for the contact 24. Contact layer 13
Has been removed from most of the top surface except for the contact region 24 to reduce as much as possible the parasitic light absorption by the highly doped region 13. For the same reason, the contact layer 14 below this structure has been removed from the light output window below the mesa. Occurs in the active layer 10,
For photons emitted at angles greater than the critical angle θ _C , after multiple passes due to the tilted mesa edge,
The probability of being reflected and becoming the angle θ _C increases. For example, the photons in trajectory 25 exit after two reflections. In contrast, dashed line 26 represents a conventional LED wall having substantially vertical sidewalls. In this LED, the same photon follows the trajectory 27 and cannot be emitted. As described above, current is passed to the LED by the metal layer 19. Contact to layer 19 is achieved by metal contacts 28 located on the top side of the device.

In the gist of the present invention, a radiation emitting device is disclosed, wherein the device has a cavity in which the radiation originates, wherein the device is at least two adjacent or adjacent edges of the cavity. ,
The cross-section may include edges that form a substantially triangle, the angle between the edges being less than 45 °, and at least one of the edges may have a transparent portion. FIG.
2 shows an embodiment of this aspect of the invention. In one embodiment, the two edges may be close to or adjacent to one another, one edge being transparent and the other being reflective. Non-transparent edges may be mesa edges. Further, the transparent edge may be a mesa edge. Thus, both edges or one of the edges may extend essentially over the entire thickness of the device,
Device thickness is measured as between the mesa edge and the transparent window edge. The edge may have an uneven surface state. In a preferred embodiment of this aspect of the invention, the mesa edges are tapered at an angle greater than 45 degrees. In this case, not only photons having a specific direction are emitted. Photons also exit after fewer passes and less recombination and re-emission. Thus, faster device performance is achieved.

A preferred embodiment of an LED having a slanted mesa edge is an LED with an edge having an angle greater than 45 degrees with respect to the normal to the major surface of the semiconductor.
Such an LED is shown in FIG. 3a. The angle θ of the mesa edge with respect to the normal is greater than 45 ° (45 degrees).
Therefore, the supplementary angle (which is the angle of the mesa edge with respect to one face of the LED) is φ = π / 2−θ. FIG. 3 b shows a tilted mirror surface 38 and a light window surface 39. The trajectory of the photons 31 emitted at an angle α ₁ with respect to the normal in the active area of the LED can be followed. Mesa
After the first reflection at the edge, this photon has the angle α ₂ = α ₁ −
It is redirected to 2φ. If this angle is still greater than the critical angle θ _c , then a second reflection at the mesa edge redirects the photons to the angle α ₃ = α ₂ -2φ = α ₁ -4φ. If the angle φ is selected to be smaller than the critical angle θ _c , this will cause the photon to have its emission angle α
Obviously, regardless of ₁ , it will always leave the semiconductor after multiple reflections. Emission occurs without re-absorption and subsequent re-emission of photons and is solely based on the effect of tapering the mesa edge 38.

According to another aspect of the present invention, there is disclosed a device that emits radiation at a predetermined wavelength, the device having a cavity with an active layer in which radiation is generated by recombination of charge carriers, the device comprising: The cavity includes at least one edge having a substantially random grating structure. The edges of the device define an area or space for radiation and / or charge carrier confinement. The edge having a substantially random, partially reflective grating structure may extend as at least one edge of a waveguide forming portion of the device. The radiation emitting device of the present invention may have a cavity containing a radiation confinement space, wherein the radiation confinement space contains the charge carriers confining the charge carriers in a smaller subspace than the radiation confinement space. Includes containment features for the carrier.
The waveguide forming portion of the device, according to such an embodiment of the invention, may be a radiation confinement region of the device that is larger than the electrical confinement region. An edge of the device extending as a waveguide forming portion of the device, the edge having a substantially random, partially reflective grating structure may also be adjacent to the active layer or the active layer. It may extend in layers. This allows
A method is disclosed for avoiding the emission of photons from a mesa region by a guided mode. One edge of the device cavity may be a mesa edge. Embodiments of the present invention do not use a transparent conductive layer at the window edge of the device. This is replaced by a metal layer where no emission takes place.

According to a preferred embodiment of the present invention, the region where the charge carriers meet to generate radiation (active layer)
Have a non-free (or substantially non-free) surface or edge where surface recombination can occur. Accordingly, the radiation emitting device of the present invention may have a cavity containing a radiation confinement space, wherein the radiation confinement space confine the charge carriers in a sub-space smaller than the radiation confinement space, Including a confinement feature for the charge carrier. A device according to this aspect of the invention may include a cavity including at least one mesa edge for defining the radiation confinement space, the device including a ring of a dielectric material in the cavity. It may further include. Therefore,
According to this preferred embodiment of the present invention, the device comprises:
It is optically and electrically confined, and the electric confinement space is a subspace of the optical confinement space. A device according to this embodiment of the invention may be confined in form and function without additional means or without additional structural features (such as extended large isolation features in the device). The device can operate while significantly reducing the effects of charge carrier recombination within the device. Such confinement in form and function can be achieved, for example, when the device is integrated in a single thin-film semiconductor. In a preferred embodiment of the invention, confinement in form and function may be provided by double mesa edges and / or by the presence of a ring of dielectric material within the device cavity. The double mesa edge may include a first mesa edge above a second mesa edge where the active layer of the device is located. The device may also include an active layer located above or below a ring of dielectric material within the cavity.
Also disclosed is a double mesa that includes an isolation plus metal coating over the mesa edge. Such is important for reducing surface recombination in small radiation emitting devices. Combinations of a single mesa and an oxidized opening are also disclosed. The surface or edge of the cavity is textured or textured.
This texturing is shorter than the wavelength of the radiation,
Alternatively, it may be long and have a substantially random diffraction grating structure pattern. Texturing or asperity creates more scattering of photons inside the cavity, thus
Improve device efficiency and speed again.

FIGS. 4 and 5 show an embodiment of a device according to this aspect of the invention. The device is formed according to the layer structure shown in FIG. In FIG. 4, the light output window (the portion from which the contact layer 14 has been removed) has a textured or uneven surface 41. Photons (photons that would otherwise have been totally internally reflected at the surface) are scattered into the output angle set for this purpose.
Certain specific fractions of these photons can exit the device. In addition, photons scattered back to the device propagate at an angle different from the angle of incidence. After reflection at the top surface of the device, these photons reach the light output window again. Some of these exit the light output window and some are reflected into the device at random angles. This process can be performed in semiconductor media or (incomplete)
This is repeated until the photons are reabsorbed in the upper mirror.
Through the process of randomizing the angle of reflection and the opportunity for multiple exits at the light output window, a substantial fraction of the photons generated within the diode can exit the device. FIG. 5 shows an LED whose textured surface 51 is a mirror surface. The conditions for randomization of the reflection angle and multiple passes at the light output window are also satisfied in this configuration. It is also possible to combine a textured light output window with a textured mirror. The textured or textured surface or edge may have a substantially random grating.

Further, in another aspect of the present invention, there is disclosed a method of texturing or forming at least a portion of at least one surface of a substrate, the method comprising providing substantially randomly distributed particles. Applying to the surface; reducing the size of the particles; and thereafter, etching the surface using the particles as an etching mask. Therefore,
This is because the etching technique used does not attack the particles, or the particles are resistant to the etchant used, or have been treated to render the particles resistant to the etchant used. Particles.

The texturing or roughening of the surface of the device shown in FIGS. 4 and 5 is achieved using this method, which utilizes a monolayer of closely packed colloidal particles used as a mask for etching semiconductors. be able to. In a preferred embodiment of this method of the invention, the diameter of the colloidal particles attached to the semiconductor is reduced using an oxygen plasma before etching the semiconductor. In this method, the colloidal particles do not adhere to each other and the etched surface has a higher diffusivity (or diffuse reflection). The size of the colloidal particles is not critical, but advantageously they have a diameter of between 50% and 200% of the wavelength of light λ _s in the semiconductor (λ _s = λ _o / _ns , _λo is the wavelength in vacuum, and _ns is the refractive index of the semiconductor). The size reduction in the oxygen plasma is typically 10% to 50% of the original diameter.

In some cases, the colloidal particles are not resistant to the etchants used to etch the semiconductor. This is the case where, for example, wet etching is used. For such cases, alternative texturing or texture forming techniques according to another embodiment of the present invention are described below. First, a thin layer of photoresist is applied (or applied) to the surface of the light emitting device that must be textured. Thereafter, the colloidal particles are applied, preferably in the form of a single layer. Oxygen plasma is used to reduce the diameter of the colloidal particles while simultaneously etching the thin photoresist layer using the colloidal particles as a mask. Thereafter, the semiconductor surface is etched using the photoresist pattern as a mask. Finally, the colloidal particles and the resist are removed, for example, by burning in an oxygen plasma.
An advantage of this technique is that wet etching can be used to create surface texturing on the surface of the device. Wet etching results in sloping edges for the columnar structure. Depending on the angle of inclination,
As described above (FIG. 2), higher emission probabilities for colliding photons can be obtained.

An alternative to forming a mask for surface texturing is to cover the device with a photoresist and to expose the photoresist to light by texturing.
Develop the pattern. Light irradiation may be performed through a metal mask of a contact printing machine or a stepper. this is,
Particularly suitable for red, near infrared and infrared LEDs. Because these wavelengths require texturing features having an order of 200 nm or more, such resolution can be achieved directly by irradiating a suitable photoresist with UV or deep UV. The advantage of this processing method is that all the LEDs of the array can be formed in the same way, ensuring equal light output. Yet another alternative for forming a mask for surface texturing using photoresist is to apply random or pseudo-random light irradiation of the photoresist to the appropriate wavelength (deep ultraviolet, ultraviolet, and electron beam). ) By direct light irradiation using a suitable light source. Direct light irradiation can be achieved by direct writing. This can also be achieved by creating a specular pattern of a coherent light source such as a laser.

The gist of the present invention, shown in FIGS. 1, 2, 3, 4, and 5, is that after the final multiple reflections at the sides of the semiconductor medium, before the photons are reabsorbed, Increases the probability of emission from the semiconductor medium. This results in a fast light emitting diode with high quantum efficiency. It is also necessary to reduce the lateral size of the light emitting diodes in order to further increase the speed and to be able to pack the light emitting diodes in a dense array. It further shows how the present invention can reduce the size of a high efficiency light emitting diode.

To reduce surface recombination at the periphery of the mesa, a preferred embodiment of a light emitting diode according to the present invention is shown in FIG. The inclined mesa of the embodiment of FIG.
In addition to the edges as well as the textured top or bottom surface, this LED involves the use of double mesas. The first mesa 61 does not reach the active layer 10 having this structure. Preferably, the semiconductor region 62 has its free surface 63
Under such a depth that the charge carriers are depleted. Thus, even if the first mesa does not penetrate the active layer of the LED, the diode area is defined by the first mesa. To form an electrical contact to the underlying doped semiconductor layer, it may be necessary to etch a second mesa 64 that penetrates the active layer and extends into the underlying contact layer. Due to the depletion layer 62, at the free edge 65 of the second mesa 64, carriers originating from the upper electrode 24 cannot flow, and thus,
Parasitic surface recombination at this edge is suppressed.

An alternative preferred embodiment is illustrated in FIG. In this case, a ring 71 of oxide traps current in an opening 72 having a diameter smaller than the diameter of the mesa. In the region 74 below the oxide, the charge carriers are completely depleted, so that carriers injected from the top electrode and passed through the current opening 72 cannot reach the surface 75 of the mesa. This also has the effect of reducing parasitic surface recombination currents.
In addition, the upper contact is formed as a ring contact 76 and is located above the oxide ring 71. The reflection from the mirror 23 covered by the dielectric 22 is better than the reflection from the contact metal 76 because the contact metal is known to cause absorption. In the structure of FIG. 7, most of the light occurs just below the dielectric-covered mirror, so the reflection of the top mirror is better than in the structure where the light occurs below the top contact. Another method for making electrical contacts to light emitting diodes without blocking the generated light is to use a transparent conductive top contact. Such contacts can be formed, preferably using indium-tin oxide (ITO). The window of light may be partially or completely covered with such a material, which is transparent to the emitted light,
Good electrical contact can be provided.

Of course, a combination of the use of double mesas and oxides can also be performed to reduce surface recombination and to avoid light generation below metal contacts. An example is shown in FIG. The LED shown in this figure has an oxide layer, which does not form a ring surrounding the entire mesa, but limits a portion of the mesa.

A particular fraction of the light generated by the mesas of the LED exits the mesa region and travels through the semiconductor substrate by waveguiding. The waveguide is formed between surfaces 291 and 292 in FIG. If at least one of these surfaces is textured or textured, for example to be diffusive, a more efficient LED is obtained: in this case some of the light passing through the waveguide is Coupling outside the semiconductor in reflection. FIG. 9 shows an example:
Photons 91 (photons that would have been guided and eventually absorbed by the semiconductor substrate) exit the LED by texturing or texture formation 92. Because
This is because the texturing extends beyond the mesa region. Thus, in this manner, the waveguide forms part of the device by forming an edge on the device having a substantially random grating structure extending as at least one edge of the waveguide. The uneven edge of the waveguide may be the upper or lower surface of the waveguide. This edge may be adjacent to or extend to the active layer. The waveguide may extend in one or two dimensions. Thus, the optical confinement region may ultimately be larger than the mesa region.

An alternative for coupling out light that would have been lost in the semiconductor waveguide outside the mesa region, envisions a mirror that reflects this light to the mesa region or optical confinement region. It is in. In FIG. 10, this is achieved by a double mesa structure. Deep mesa 10
1 surrounds the first mesa, penetrates the active layer,
Is etched to the lower contact.
The mesa edges are covered by a dielectric covered metal mirror, which has high reflectivity. FIG.
In the device structure of FIG. 10, the photons emitted from the mesa region are reflected back to the mesa or the optical confinement region in the device structure of FIG.

FIG. 11 is an alternative to the structure of FIG.
A structure using only one mesa is shown. The active layer is located on top of the oxide confinement ring. Since there is only one mesa, the structure of FIG. 11 can be smaller than the structure of FIG. Oxide confinement results in an electrically small diode, which results in a much higher current density at lower current levels, in other words, a faster device at lower current levels. In fact, FIGS. 10 and 1
In accordance with the teachings of the embodiment of the present invention shown in FIG. 1, a device having an edge with a substantially random grating structure extending as at least one edge of the waveguide is fabricated, wherein the waveguide comprises a portion of the device. Is formed.
A light output window surrounded by a single mesa and metal layer 19 provides a small L optically confined structure of FIG.
ED, which is important for high resolution applications. Also, for high speed applications, it is important to have an optically small source (the spot size of the light beam transmitted through the optical system should be on the order of the spot size of the light source). It is necessary for the light receiving unit (or receiver) to have a small spot size on the detecting unit (or detector) side.

A preferred embodiment of the device according to the present invention will be described below.

In the preferred embodiment of the present invention, FIG.
The example of the device structure shown below and the process flow shown below are disclosed. The device structure shown in FIG. 12 is a waveguide structure that extends in two dimensions, has a waveguide structure having an uneven surface structure 122 with a substantially random diffraction grating structure, , Active layer 1 of the device
It reaches zero. The mesa region includes an oxide ring 71. The transparent window 41 of the device has an uneven surface structure with a substantially random grating structure. The device has a lower electrode contact (124), an upper electrode contact (125), and an isolation layer (125). The device is a metal mirror (12) covered with a dielectric (127).
8) attached to the support (129) with van der Waals bonds. Advantageously, the device structure of the best embodiment of the present invention can be tied into an array of devices. In this array of devices, the individual devices of the array are shown in FIG.
Are confined in form and function. The individual devices of the array are defined by mesa edges, and the grooves between the individual devices have a substantially random grating structure. An array of devices
It may have a single electrical anode contact and a single electrical cathode contact.

1. Best Mode Structure The detailed layer structure of the best mode device is shown below.

TABLE 1 Layer structure of best mode device grown using a River 3200 molecular beam epitaxy apparatus.

2. Best Mode Processing Mesas having a diameter of 30-60 microns were converted to Al _0.98 G
Etch through the a _0.02 As layer 71 and the GaAs active layer 10. The Al _0.98 Ga _0.02 As layer is oxidized at 380 ° C. to 16-24 microns on the outside. The resulting current confinement opening 1271 in the mesa has a diameter of 12-30 microns. This current confinement localizes the light generation to a region below the optical window rather than below the top metal contact 123. Since the active layer is made of GaAs, the emission wavelength of the LED is 865 nm. The lower contact 124 is an alloy AuGe / Ni / Au pad, and the upper contact 123 is a non-alloy Ti / Au pad. The isolation below the bonding pad is a cured polyimide film 125. Following this, the top side 41 of the device as well as the waveguide are also textured. A monolayer of polystyrene spheres (or polystyrene spheres) having a diameter of 400 nm is formed on the water surface and transferred to a device surface that has been treated to be hydrophilic. In this case, the coverage of the device surface by the beads is typically 60%. This is made up to about 50% by reducing the diameter of the polystyrene sphere in an oxygen plasma. The wafer is etched using this sphere as a mask by an argon etch assisted by chlorine. The optimum etch depth was found to be 170 nm. The final step is to subject the processed device to a gold mirror 128 covered with polyimide 127 by epitaxial lift-off using a 50 nm AlAs sacrificial layer below the device.
(This is deposited on the silicon wafer 129).

The detailed process flow is shown below. 1) Main mesas • Lithographic definition of a circle with a diameter of 30 to 60 μm • H ₂ SO ₄ : H ₂ O ₂ : DI until the Al _0.98 Ga _0.02 As oxide layer (layers 6 to 8 of the layer structure) is almost reached (1: 8: 2
00) (in about 4 minutes) Complete etching down to the oxide layer in citric acid: H ₂ O ₂ (10: 1) (about 1 minute) H ₂ SO ₄ : H _{2 in 2} O ₂ : DI (1: 8: 200)
Min etching (removal of oxide layer and active layer) ・ Removal of resist in acetone 2) Oxidation ・ 375 ° C. for device mesa 60 μm in diameter
(10 minutes), and Al in wet nitrogen at 365 ° C. for 30 micron device mesas
Wet thermal oxidation of _0.98 Ga _0.02 As layer. Oxidation of about 20 μm of Al _0.98 Ga _0.02 As layer from outside to a depth of 20 μm or 8 μm, respectively. 3) n-contact (124). 60μm at a distance of
Lithography by defining · H ₂ SO openings _{4: H 2 O 2: DI} (1: 8: 200) of the metal contact of the self-aligned etch · AuGe / Ni / Au (120 /10 / 60nm) of in Vapor deposition • Lift-off in acetone • Contact annealing at 380 ° C for 30 seconds 4) Polyimide (125) • Spin of adhesion promoter PIQ coupler (Hitachi) at 5000 UPM for 80 seconds • Bake at 160 ° C for 5 minutes PIQ13HV10 at 3000 RPM for 30 seconds
(Hitachi) spin-50 ° C with ramp (or tilt) at 1 ° C per minute
Baking in oven at 120 ° C. for 30 minutes Patterning of polyimide by photolithography Slow baking in FOG up to 350 ° C. (ramp time: 35 minutes) at 350 ° C. for 10 minutes Hold 5) Upper contact (123) Upper p-contact area (device only about 10 μm)
Lithographic definition of (overlying the mesas) 2 Oxygen plasma treatment at 100 W for 2 min. Deposition of TiW / Au metal contacts (40/250 nm). Liftoff in acetone 6) Removal of upper GaAs layer. Citric acid: H Etching of upper layer in ₂ O ₂ (5: 1) for 10 seconds (selective removal of upper GaAs layer) 7) Formation of surface irregularities (see FIG. 13) ・ Device surface of polystyrene sphere having a diameter of 430 nm Adhesion to: 100 W for 1 minute to make the device surface hydrophilic
-Plasma treatment at room temperature-Formation of a single layer of spheres on the water surface (Fig. 13a)
-Flooding of the device in pure water (to cover the device with water), and subsequently-flooding of the device through a sphere layer above the water surface, then (see Figure 13a) 1) removal of the device from the water to transfer the membrane of the sphere onto the device (see FIGS. 13a and b) evaporation of the water on the device (FIGS. 13b and c; FIG. 1)
(See photo of results shown in 3d) Shrinkage of spheres in oxygen plasma to about 350 nm (3 min, 100 W) (see FIGS. 14a and b) To form cylindrical pillars Dry etching (Ar milling assisted by chlorine) to 170 nm depth (see FIG. 15) Sphere removal: using oxygen plasma (15 minutes, 100 W) or using adhesive tape Ultrasonic cleaning of samples in hot acetone (see FIG. 16) Polystyrene spheres were
Fic Corporation's Nanosphere
e (trademark) of standard size. This product is an aqueous suspension of polymer microspheres. Microspheres are made of polystyrene. Oxygen plasma treatment
This is performed in a Plasma Technology plasma system. The procedure for using this system is
C ISO-9000 Procedure: Stefan Peete
rs. 11, page 6, “Wet etches in MMIC process”
nthe MMIC process ”(document reference number 17703, from October 30, 1997, Ver.
0.1.00). The only difference from this procedure is the RF power and time (100 W for LED process for 3 minutes). 8) Epitaxial lift-off and Van der Waals bonding Cleaving the sample into small pieces of about 1.5 x 3 mm Coating of the top of the sample with Apiezon black wax Epitaxial layer completely detached from substrate For up to,
Etching of AlAs release layer in 10% HF (about 4 to
6 hours) ・ Transfer of the epitaxial layer formed as follows in water on a polyimide / Au mirror ・ Ti / Au (4) on a polished silicon wafer
0/200 nm) Deposition of polyimide PIQ1 at 10,000 rpm for 60 seconds
Spin of 3 Hard bake of polyimide as above This results in a 170 nm thick polyimide layer on the gold mirror. Oxygen plasma treatment at 100 W for 1 minute to make the mirror hydrophilic.Pressing the sample with about 2 g force for at least 2 hours.Dissolving the wax in chloroform.

Measurements on the Device According to the Best Mode of the Invention Measurement and Setup Description of Calibration (Calibration) Probe the top and bottom contacts of the device with a needle,
Apply a forward bias. Measure the current through the device. Calibrated infrared detector (Newport 818-
SL), the light emission is detected, and the wavelength (87
0 nm). During the measurement, place the detector approximately 5 cm above the device. The total light output power (or total light output intensity) P _tot of the LED is given by the formula

(Equation 1) Where P _meas is a measure of intensity, d
Is the distance between the device and the detector, and r is the radius of the active area of the detector (r = 5.65 mm). This equation is LE
The Lambertian output of D (or the fully diffusive output) is assumed, which has been confirmed by separate measurements. The exact distance between the detector and the device is determined as follows: The distance between the sample holder and the detector housing is less than 0.1 mm using calipers (or calipers). Can also be measured with high accuracy. The offset between this measurement length and the correct distance can be determined by varying the distance between the sample and the detector and measuring the output power. For exact offsets,
The above equation gives the same result for any height.

The external quantum efficiency is given by the equation

(Equation 2) _Where E _phot is the energy of the emitted photon, q
Is the elementary charge.

Quasi-Static Characteristics FIG. 17 shows the above-described external quantum efficiency measurements for the current through the LED. The LED has a 120 nm thick active layer and the oxidized openings have a diameter of 12 microns. 4
A comparison is made between the two types of devices: Curve A: LED without steps 9 and 10 Curve B: LED without step 9, with step 10 Curve C: with steps 9 and 10, variation a
Curve D without steps a and b: LED with steps 9 and 10 and variation a, without curve b Curve E: LEDs with steps 9 and 10 and variants a and b: very high efficiencies are obtained.

In FIG. 18, the total light output power of an LED processed according to the present invention is compared to the light output power of several 850-nm and 980-nm VCSELs having oxidized openings in the range of 3-9 microns. (IEEE Photonics Technology
oggy Letters 8, no. 8, 971-
973 (1996): Weigl, M .; G
labherr, R.A. Michaelzik, G .; R
einer, and K.E. J. "High-power single-mode selective oxidation vertical cavity surface emitting lasers (High-Power Single-M) by Ebeling.
Ode Selectively Oxidized
Vertical-Cavity Surface-E
Mitting Laser); Applied P
physics Letters 70, no. 14, pp. 1781-1783 (1997): L. Hu
ffaker and D.M. G. FIG. "Low threshold vertical cavity surface emitting lasers based on high contrast distributed Bragg reflectors (Low threshold) by Deppe.
d vertical-cavity surface
-Emitting lasers based on
high contrast distribute
d bragg reflectors) "; and IEEE Electronics Letters 3
1, no. 11, pages 886-888 (1995)
Year): G. M. Yang, M .; H. MacDo
ugal, and P.U. D. Dapkus, "Ultra-low threshold surface emitting laser with ultra-low threshold current obtained by selective oxidation (Ultralow threshold)."
current vertical-cavity
surface-emitting lasers
btained with selective ox
id))). It will be appreciated that the output power of the LED of the present invention is greater than the output power of the VCSEL at low currents, especially at currents below the threshold current of each VCSEL. It will also be seen that VCSELs with very low threshold current have lower output power than the LEDs of the present invention over the entire current range. This is because VCSE with very low threshold current
L is due to having a highly reflective mirror and therefore having a lower efficiency. In an optimized short-range optical interconnect system, the required light intensity level of the light source is on the order of 10-100 μW. It is clear from FIG. 18 that the LEDs of the present invention are compatible with low power optical interconnects using this range of light intensities.

The quantum efficiency of the LED of the present invention is proportional to the thickness of the active layer. FIG. 19 illustrates the external quantum efficiency as a function of current for the LED of FIG.
Has an active layer thickness of 120 nm and shows a maximum quantum efficiency of 31% at 1.7 to 2 mA (compared to the maximum quantum efficiency of an LED of the present invention having an active layer thickness of 30 nm). For the latter LED, the maximum external quantum efficiency is also 1.7.
It is 18.7% at a current of ２2 mA. The maximum wall plug efficiency of the LED of the present invention is 11%. The opening of the device was 22 μm. It is presently believed that the reason that the quantum efficiency depends on the active layer thickness is due to carrier spill-over that occurs in thin active layers. This can be eliminated by using a p-doped active layer and a material in the cladding with a larger band gap.

The operation of the LED of the present invention does not depend on the balance between emission wavelength and cavity length. Therefore, such devices are best suited for applications where it is important to withstand large temperature fluctuations. CMOS chips
It must operate over a wide temperature range. In particular,
Processor chips reach temperatures in excess of 100 ° C. Therefore, the LED of the present invention is well suited for optical interconnection on a chip. FIG. 20 shows the decrease in light output power versus temperature of the LED of the present invention for various bias currents. A 0.36% reduction per degree is observed.

The dynamic characteristics of the device of the present invention are measured. A square voltage pulse was applied to a 50 ohm terminated coplanar probe (5
0-Ohm terminated coplanar
probe) was added to the LED of the present invention. The rise and fall times of 10% to 90% of the voltage pulse are less than 100 ns. The shape of the optical response of the LED is observed using a high frequency diode and an oscilloscope with a bandwidth of 2 GHz.

When applying a voltage pulse, the current is first
It rises quickly to a level determined by the total series resistance of the circuit. Thereafter, the current drops to its quasi-static level with a time constant RC _j. Here, C _j is the junction capacitance of the LED. This time constant is actually several hundred picoseconds. At this point, the device is still dark. Thereafter, the concentration of minority carriers in the active layer is gradually increased, essentially by the quasi-static current as a current source. This enhancement produces light, and thus the light output can be used as a direct measure of the minority carrier charge content in the active layer. The greater the quasi-static current flowing through the device, the faster the rise time of the light output.

From the above, it is clear that thinner active layers result in faster devices at a given current density. Also, a higher drive voltage (which corresponds to a higher quasi-static current) results in a shorter optical rise time. In FIG. 21, black circles indicate the active layer (30 nm) of the LED of the present invention having a current opening of 22 μm as 0%.
10% to 90% rise time measurements when switching between V and various voltages.
The horizontal axis is the quasi-static current density through the current openings corresponding to various voltages. 1 of the optical signal when the voltage is returned to 0V
Fall times from 0% to 90% are shown in FIG. 21 as open circles. This is smaller than the rise time. This is due to the fact that minority carriers are extracted from the active layer by the reverse current flowing through the LED rather than by recombination (IEEE Journal Light).
wave Technology, vol. 14, pp. 1721 to 1729 (1996): F. S
chubert, N .; E. FIG. J. Hunt, R .;
J. Malik, M .; Mivicic, and D.M.
L. Miller, “Temperature and Modulation Characteristics of Cavity Light-Emitting Diodes (Temperature)
and modulation characteri
stics of resonance-cavity
light-emitting diodes) ”).

When measuring the eye diagram of an LED, the above asymmetry between rise time and fall time manifests itself as a collapsed eye. However,
It is desirable to have a somewhat symmetrical eye at the light receiving end of the optical interconnect line, in other words, to have roughly equal rise and fall times. This is a non-zero "low state" and a larger "high state"
This can be achieved by switching the LED between and. In fact, a non-zero low state means that a certain offset carrier density exists low in the diode. This offset
Due to the carrier density, the rise time of the diode is smaller than if starting from zero. On the other hand, when switching the diode from a high state to a non-zero low state, the draw current is smaller than when switching the diode to zero. Therefore, when switching to a non-zero off state, the fall time increases (IEEE Journal of Lightw).
ave Technology, Vol. 14. First
721-1729 (1996): E.I. F. Sc
Hubert, N .; E. FIG. J. Hunt, R .;
J. Malik, M .; Mivicic, and D.M.
L. Miller, “Temperature and Modulation Characteristics of Cavity Light-Emitting Diodes (Temperature)
and modulation characteri
stics of resonance-cavity
light-emitting Diodes "). As a result, for each high state voltage (quasi-static current density), the low state voltage (quasi-static current density) is such that the optical rise and fall times observed when switching between states are equal. Density). The resulting rise and fall times are shown as solid lines in FIG. 21 as a function of the quasi-static current density also corresponding to the high state voltage. As described above, the rise time for this switching mode is considerably shorter than the rise time when switching from 0V, but the fall time is somewhat longer than when switching to 0V. The figure shows the _low state current density J _low used to achieve equal rise and fall times.

From FIG. 19, it can be seen that external quantum efficiencies in excess of 18% are obtained for currents between 1.3 mA and 3.2 mA. This is 390 mA / cm ² and 840 mA
/ Cm ² . From FIG.
The rise and fall time corresponding to 90 mA / cm ² is 2.2 ns and the rise and fall time obtained at 840 mA / cm ² is 1.9 ns. Therefore, in the current range corresponding to the maximum external quantum efficiency, the rise and fall times of the diode are about 2 ns.

When considering producing a large-scale array with a high yield, it is advantageous that the high-state and low-state currents do not need to be set with high precision. The sensitivity of the rise and fall times to high state current density variations is shown in FIG. FIG. 22 shows the influence of the fluctuation in the current density in the low state. Dotted lines are chosen to make the rise and fall times equal,
10% to 90% rise and fall times for lower current densities (ie, the solid curve in FIG. 21). The black and white circles in FIG. 22 indicate the rise and fall times, respectively, observed when the lower current density was reduced to the value shown next to the white circle. The black and white triangles indicate the rise and fall times observed when the lower current density is increased to the level shown next to the white triangle, respectively. If the low-state current density is reduced to at least one-third as compared to the low-state current density that equalizes the rise and fall times, the rise time changes little while the fall time Can be concluded to decrease. Thus, setting the current density in the low state is not important.

Next, the eye diagram obtained by pulsing the LED of the present invention between two voltages so that the rise and fall times are equivalent to a 2 ²³ -1 bit pseudo-random bit stream. To consider. For each high state voltage, there is a maximum frequency at which the eye begins to close. The acceptable limit of the open eye was defined as the eye where the sum of the amplitude scatter of the upper and lower light levels was approximately equal to the open amplitude, as shown in FIG. FIG. 24 shows the maximum bit rate (or bit rate) of operation based on this definition for a quasi-static current in the high state. As expected, the bit rate increases with increasing high state quasi-static current. The external quantum efficiency versus current for this same device is plotted against the right axis of FIG. The decrease in efficiency at currents exceeding the maximum efficiency is small. This is particularly appropriate for high speed interconnects. This means that higher speeds can be obtained without increasing the efficiency by increasing the current. For example, a bit rate of 622 Mbit / s is obtained at a high state current of about 3.9 mA, and the corresponding external quantum efficiency is still 17% or 0.24
It turns out that it is larger than mW / mA. This means that a standard LED (Applied Ph
ysics Letters Vol. 60, 35
3 (1992): T.M. J. deLyon, J.M.
M. Woodall, D.M. T. McIntur
ff, R.F. J. S. Bates, J.M. A. K
ash, p. D. Kirchner, and F.M.
"Carbon doped Ga _0.51 In _0.49 P / grown by gas source molecular beam epitaxy.
Doping concentration dependence of emission and light modulation bandwidth in GaAs light emitting diodes
translation dependency of ra
diationand optical modula
Tion bandwidth incarbon-d
oped Ga _0.51 In _0.49 P / GaAs lihgt
-Emitting diodes grown by
gas source molecular bea
epitaxy) ") or RC-L
ED (IEEE Journal of Lightw
ave Technology, Vol. 14. First
721-1729 (1996): E.I. F. Sch
ubert, N.M. E. FIG. J. Hunt, R .; J.
Malik, M .; Mivicic, and D.M.
L. Miller, “Temperature and modulation characteristics of cavity-cavity light-emitting diodes.
second modulation characters
tics of resonance-cavity l
light-emitting diodes), much higher than the efficiencies obtained at equivalent bit rates. Interestingly, FIG.
It should be noted that values greater than M bits / s can be reached at sub-mA current levels and the external quantum efficiency at 1 mA is greater than 17%.

The device of the present invention does not rely on resonance for its operation, and its performance is not significantly affected by variations in growth and process. Therefore, the device of the present invention is suitable for integration in a large-scale array.
Furthermore, they exhibit high efficiency properties, especially at high speeds.

Alternative Embodiments of the Invention In accordance with another aspect of the present invention, a large-scale (or large-area) light emitting device or diode is disclosed, which includes the advantageous aspects of the present invention. One feature of this large-scale light-emitting device is that it aims to avoid losing light guided laterally in the waveguide formed between different interfaces in the light-emitting device or diode. is there. The layer structure of many semiconductor light emitting devices or diodes actually consists of multiple layers of various semiconductor materials having different refractive indices. Thus, light can be trapped inside the waveguide formed between these materials. This light can exit the semiconductor at the side of the semiconductor crystal. By packaging the diode in a reflective cup, this light can be redirected in a direction perpendicular to the major surface of the diode crystal.

However, during propagation of light in the waveguide, a significant amount of photons can be reabsorbed in the semiconductor crystal.

In order to avoid this re-absorption, it is advantageous to have a structure in which light guided in the lateral direction can be emitted from the semiconductor on the main surface in the vicinity of where the light is generated.

The purpose of the large-area light-emitting device according to the invention is to propose a structure for a light-emitting diode which can perform a short-wave guided outcoupling (or external connection) in a semiconductor. It is in. The proposed structure results in a large area planar diode that can be fabricated over a wide range of sizes. Large area LED
Are formed from a plurality of small light emitting diodes. These smaller diodes can be arranged in any configuration, such as a one-dimensional or preferably a two-dimensional array structure. For separation into individual small light emitting diodes,
The lateral waveguiding that normally occurs in large light emitting diodes is avoided: lateral waveguiding modes can be externally coupled through the top surface of the diode. In a large area diode including a plurality of smaller diodes, the overall emitted power is simply proportional to the area of all diodes or the number of smaller diodes contained therein. Therefore, redesigning a large area diode according to the specification of the all-optical output power is simplified.

Two embodiments of this aspect of the invention are shown in FIGS. 25 and 26, respectively. In FIG. 25, an individual light emitting device or diode (250) is provided so that the transverse waveguide mode can exit the semiconductor at the mesa edge.
1) (2502) (2503) (2504) are isolated by mesa etching through the active layer (2510). To enhance outcoupling efficiency, a preferred embodiment of this diode has at least one textured major surface. In the embodiment shown in FIG. 26, the mesas that isolate the individual diodes do not reach the active layer (2610). Texturing the LED surface between the individual diodes to couple the laterally guided light to the outside (2
62).

Preferably, the large area diode light emitting device or diode of the present invention is a thin film diode. The overall thickness of the semiconductor film is small enough not to absorb more than 10% of the beam propagated perpendicular to the main surface of the diode. One face of the diode is preferably mirror-like and light is coupled out through the opposite face. The mesas of the embodiment shown in FIGS. 25 and 26 can be formed from a mirror-like surface or from a light-emitting surface. If the semiconductor substrate on which the light emitting diode layer is grown is not sufficiently transparent to light, the original substrate may be removed and a transparent support may be used. This transparent support may be provided on the side of the original substrate or on the opposite side.

In one particular embodiment, the transparent support is a foil (or thin film) of a material such as a polymeric material. In this case, a phosphor may be mixed with the foil to change the wavelength of the diode. A rear mirror may or may not be provided. In the latter case, the front side and the rear side emit different colors 2
A color diode is produced.

A preferred embodiment of a thin film large area diode according to this aspect of the present invention is shown in FIG. The figure shows a one-dimensional cross-section of a one-dimensional or two-dimensional array of smaller diodes (collectively forming one large area diode). Support made of a transparent material (27
28) is placed on top of the mesa and light is coupled through this transparent material to the outside of the diode. The semiconductor material is thinner below. The side opposite the transparent support may be covered with a mirror (2729). In certain embodiments, the top surface may be textured, for example, to enhance light outcoupling efficiency.
An AlGaAs oxide layer may be used to confine current (and thus light generation) to the central region of each of the smaller diodes.

In another embodiment of the large area light emitting device, each individual small diode is provided with a microlens. The resulting large area diode is pseudo-planar (or nearly flat), and
It has all the advantages of a large area diode with microlenses, such as more efficient light outcoupling and an optimized desired beam profile. Microlenses can be formed on small diodes at the wafer level, in contrast to current state of the art large area diode lenses that must be formed during packaging. When the transparent support is a polymeric material,
The method of forming the microlenses is to emboss (or emboss) them directly onto this polymer support (2828). This is illustrated in FIG.
This figure shows, at the front, a cross section of a two-dimensional array of small diodes that collectively form one large area diode. For large area diodes without microlenses, consisting of a plurality of smaller diodes, the total optical output power is the sum of the output power of the individual diodes, each with an individual microlens. Thus, it becomes easier to redesign a large area diode for a particular output power level.

In the embodiment shown in FIGS.
Smaller diodes that make up large area diodes
Shown electrically coupled in parallel. The voltages applied to the smaller diodes are essentially the same, and the current of the large area diode is the sum of the currents flowing through the individual smaller diodes. In another embodiment of this aspect of the invention, the large area diode may comprise an array of smaller diodes electrically coupled in series. In this case, the current flowing through the large area diode is essentially equal to the current flowing through each smaller diode, whereas the voltage applied to the large area diode is less than the voltage applied to each small area diode. Equal to the sum. This total voltage is 110V or 2
It can be equal to a voltage standard such as 30V.

[Brief description of the drawings]

FIG. 1a shows the layer structure of a light emitting diode according to an embodiment of the device of the invention. FIG. 1b shows another layer structure of a light emitting diode according to another embodiment of the device according to the invention, preferably having a metal mirror (23) covered with a dielectric (22) and passing through this mirror. 1 shows a thin-film light-emitting device with electrical contacts (14) provided by way of example.

FIG. 2 shows an embodiment of a light emitting diode according to the invention.

FIGS. 3a and 3b show a tilted mesa having an angle greater than 45 degrees with respect to the normal to the main surface of the semiconductor.
1 shows a preferred embodiment of a light-emitting diode with edges.

FIG. 4 shows an embodiment of the device according to the invention,
The device has an embodiment comprising a cavity with an active layer in which radiation is generated by recombination of charge carriers, the cavity comprising at least one edge having a substantially random grating structure.

FIG. 5 is an embodiment of the device according to the invention,
The device has an embodiment comprising a cavity with an active layer in which radiation is generated by recombination of charge carriers, the cavity comprising at least one edge having a substantially random grating structure.

FIG. 6 shows an embodiment of the device according to the invention,
The device has a cavity that includes a region that confines radiation, the region illustrating an embodiment that includes a confinement feature for the charge carrier that confines the charge carrier to a smaller subregion within the cavity than the radiation confinement region.

FIG. 7 shows an embodiment of the device according to the invention,
The device has a cavity that includes a region that confines radiation, the region illustrating an embodiment that includes a confinement feature for the charge carrier that confines the charge carrier to a smaller subregion within the cavity than the radiation confinement region.

FIG. 8 shows an embodiment of the device according to the invention,
The device has a cavity that includes a region that confines radiation, the region illustrating an embodiment that includes a confinement feature for the charge carrier that confines the charge carrier to a smaller subregion within the cavity than the radiation confinement region.

FIG. 9 shows an embodiment of the device according to the invention,
The device has a cavity that includes a region that confines radiation, the region illustrating an embodiment that includes a confinement feature for the charge carrier that confines the charge carrier to a smaller subregion within the cavity than the radiation confinement region.

FIG. 10 is an embodiment of the device according to the present invention, wherein the device has a cavity including a region for confining radiation, the region confining charge carriers to a sub-region smaller than the region for confining radiation within the cavity. 5 shows an embodiment that includes a confinement feature for the charge carrier.

FIG. 11 is an embodiment of a device according to the present invention, wherein the device has a cavity that includes a region for confining radiation, the region confining charge carriers to a sub-region within the cavity that is smaller than the region for confining radiation. 5 shows an embodiment that includes a confinement feature for the charge carrier.

FIG. 12 illustrates a light emitting device according to a best embodiment.

FIG. 13 outlines the best mode process flow for making a device according to the present invention.

FIG. 14 outlines the best mode process flow for making a device according to the present invention.

FIG. 15 outlines the best mode process flow for making a device according to the present invention.

FIG. 16 outlines the best mode process flow for making a device according to the present invention.

FIG. 17 shows measurements for a device according to the invention.

FIG. 18 shows measurements for a device according to the invention.

FIG. 19 shows measured values for a device according to the invention.

FIG. 20 shows measured values for a device according to the invention.

FIG. 21 shows measured values for a device according to the invention.

FIG. 22 shows measurements for a device according to the invention.

FIG. 23 shows measurements for a device according to the invention.

FIG. 24 shows measurements for a device according to the invention.

FIG. 25 shows a large area LE according to one preferred embodiment.
D shows a large area LED where transverse waveguide mode outcoupling occurs at the mesa edge and the mesa is etched through the active layer.

FIG. 26 illustrates another preferred embodiment of a large area LED, wherein the textured regions between the mesas of the large area LED result in out-coupling of the transverse waveguide mode.

FIG. 27 illustrates a cross-sectional view of a large area LED having a textured surface.

FIG. 28 illustrates another preferred embodiment of a large area LED having a textured surface, wherein microlenses are combined with the diodes of the array.

 ────────────────────────────────────────────────── ─── Continued on the front page (31) Priority claim number 60/110322 (32) Priority date November 30, 1998 (November 30, 1998) (33) Priority claim country United States (US) (31) Priority claim number 60/131358 (32) Priority date April 28, 1999 (1999. April 28) (33) Priority claiming country United States (US) (71) Applicant 596099561 Breyer Universitate Brussels Le VRIJE UNIVERSITEIT BRUSSEL Belgium, 1050 Brussels, Preinlan, 2 (72) Inventor Paul Heremans Belgium, Be-3000 Louvain, Tinsestrad 99 (72) Inventor Maarten Quayck Belgium, Be-2600 Antwerp, Fredestrato 154 (72) Inventor Inner Winterhalter dish Belgium, based -3001 Leuven, Naam Se Steen wet heat 288 No. (72) inventor Gusutafu-Bors Belgium, based -3010 Kessel - Russia, Bell Fusutorato No. 70

Claims (40)

[Claims] 1. A device that emits radiation at a predetermined wavelength, the device comprising a cavity with an active layer in which the radiation is generated by recombination of charge carriers;
The device, wherein the cavity has a radiation confinement space that includes confinement features for the charge carriers, and the device includes at least one edge having a substantially random grating structure. 2. A device for emitting radiation at a predetermined wavelength, the device having a cavity with an active layer in which the radiation is generated by recombination of charge carriers,
The device, wherein the device comprises a waveguide, wherein the device comprises at least one edge having a substantially random grating structure. 3. A device that emits radiation at a predetermined wavelength, the device having a cavity with an active layer in which the radiation is generated by recombination of charge carriers;
The device, wherein the device comprises at least one edge provided on the transparent carrier having a substantially random grating structure. 4. A device that emits radiation at a predetermined wavelength, said device having a cavity with an active layer in which said radiation is generated by recombination of charge carriers, said device comprising a substantially random device. A device comprising at least one edge having a grating structure, having at least one reflective edge, wherein at least one electrical contact to the device passes through the reflective edge. 5. The device according to claim 4, wherein the device is provided on a transparent carrier. 6. A device that emits radiation at a predetermined wavelength, the device including at least two edges that form a substantially triangular shape in cross section, wherein the angle between the edges is 45 degrees. °, wherein at least one of the edges has a transparent portion. 7. The method of claim 1, wherein one of the two edges of the cavity is transparent to the radiation and one of the two edges is reflective, wherein the edges are at least close to or adjacent to one another. Item 7. The device according to Item 6. 8. The device of claim 7, wherein at least one of the two edges has a rough surface condition. 9. The device according to any of the preceding claims, wherein the substantially random grating structure is a regular repetition of a random grating. 10. The device according to any of the preceding claims, further comprising a waveguide. 11. The edge having a substantially random grating structure extends as at least one edge of the waveguide forming portion of the device, wherein the edge preferably comprises the active layer. Or the cavity,
A device according to any of the preceding claims, alternatively adjoining or extending a confinement feature for the charge carriers or the radiation confinement space. 12. The device according to any of the preceding claims, wherein the cavity comprises at least one mesa edge for defining the radiation confinement space. 13. The device according to any of the preceding claims, wherein the radiation confinement space confines the charge carriers in a smaller subspace within the cavity than the radiation confinement space. 14. The device according to claim 1, having a charge carrier confinement space made of a dielectric material. 15. The charge carrier confinement feature made of a dielectric material is a ring, wherein the device is mounted on a support substrate, the support substrate preferably being transparent to the radiation. A device according to any preceding claim, wherein the device comprises a fiber optic face plate. 16. The device includes at least two edges, the edges forming a substantially triangle in cross-section, wherein the angle between the edges is less than 45 °, and at least one of the edges is transparent. Device according to any of the preceding claims, having a portion. 17. The method of claim 17, wherein one of the two edges of the cavity is transparent to the radiation and one of the two edges is reflective, wherein the edges are at least proximate or adjacent to one another. A device according to any of the preceding claims. 18. The device according to any of the preceding claims, wherein at least one of the two edges has a rough surface condition. 19. An array of devices according to any of the preceding claims, wherein the individual devices of the array are confined in form and function. 20. The method according to claim 19, wherein the individual devices of the array are
20. An array of devices according to claim 19, wherein the grooves defined by edges, wherein the grooves between the individual devices have a substantially random grating structure. 21. An array of devices according to claim 20, having a common anode contact and a common cathode contact. 22. An array of devices according to claim 19, wherein each device has a microlens thereon. 23. An array comprising the device of claim 19 attached to a polymeric foil support. 24. The polymeric foil further has a phosphorous layer thereon for converting radiation, so that the device preferably has a different wavelength on the support side of the device compared to the front side of the device. An array comprising the device of claim 23, wherein the device emits light. 25. An array of devices that emit radiation at a predetermined wavelength, the devices including a cavity with an active layer in which the radiation is generated by recombination of charge carriers, wherein individual devices of the array are An array, confined in form and function, wherein there is a waveguide between the individual devices, the waveguide having a grating structure, preferably a substantially random grating structure. 26. A method of operating a device that emits radiation at a predetermined wavelength, the device including a cavity with an active layer in which the radiation is generated by recombination of charge carriers, wherein the cavity is located within the cavity. And confining the charge carriers in a sub-space that is smaller than the radiation confinement space with a radiation confinement space including confinement features for the charge carriers. 27. The method of claim 26, wherein the active layer is between a highly reflective edge and a transparent edge having irregularities. 28. The method of claim 26, wherein at least one of said edges is roughened. 29. A method of texturing at least a portion of at least one surface of a substrate, comprising: applying an overlying material over a portion of the surface, wherein the overlying material is substantially randomly distributed. Having a pattern with a defined opening feature (the step of applying the overlying material comprises:
Applying a substantially randomly distributed particle to the surface), reducing the size of the particle, and subsequently etching the surface using the particle as an etching mask. , Whereby the etching mask comprises a substantially random mask pattern. 30. Applying a layer of a second mask material to the surface; developing a pattern in the second mask material while reducing the size of the particles; and thereafter etching the pattern with an etching mask 30. The method of claim 29, further comprising etching the surface while using as. 31. The method of claim 30, further comprising removing the pattern and the particles. 32. The method of claim 29, wherein the etching step forms irregularities on the surface, thereby rendering the surface diffusible to electromagnetic radiation impinging on the surface. . 33. A reduction in particle size of about 50 of the surface
30. The method of claim 29, which provides a% coverage. 34. The reduced size of the particles is between 50% and 200% of the wavelength in the device material for the radiation.
30. The method of claim 29, which is in the% range. 35. A method of texturing at least a portion of at least one surface of a substrate, comprising: applying an overlying material comprising a photoresist material over a portion of the substrate, wherein the overlying material is substantially. And a step of etching the surface while using the upper layer material as an etching mask, whereby the etching mask forms a substantially random mask pattern. And irradiating the photoresist material with a lithographic mask having a substantially random mask pattern, and thereafter developing the photoresist. 36. The lithographic mask is one of a metal mask, a direct-write mask, a mask formed of metal particles, or a specular noise pattern of a coherent light source.
36. The method of claim 35, wherein: 37. The method further comprising forming a device that emits radiation at a predetermined wavelength in the substrate, the device having a cavity, the surface having at least one edge of one of the cavities of the device. 37. The method of claims 29-36, wherein the method is a part. 38. The method further comprising forming an array of devices emitting radiation at a predetermined wavelength on the substrate, wherein edges of at least two devices of the array are substantially the same random. 38. The method of claim 37, having a texturing pattern. 39. The method of claim 37, further comprising thinning the device or the devices of the array; and disposing the devices or the devices of the array on a support substrate. 40. The support substrate is transparent to the radiation and comprises a fiber optic face plate.
38. The method of claim 37.