Patent Application
20240413614
A semiconductor laser diode emits electromagnetic radiation that is comprised primarily of stimulated emission. The laser diode is comprised of a gain medium that functions to provide emission through the recombination of electron-hole pairs and a cavity region that functions as a resonator for the emission of the gain medium. When a suitable voltage is applied to sufficiently pump the gain medium, the cavity losses are overcome by the gain and the laser diode reaches the so-called threshold condition, where a steep increase in the light output versus current input characteristic is observed. Unlike light emitting diodes (LEDs), laser diodes emit directional light and have orders of magnitude higher spatial brightness.
The feedback mechanism in Fabry-Perot cavity diode lasers is based on reflectivity off cleaved or etched facets at both ends of the laser cavity. In general, because the reflectivity of the facets does not vary strongly with wavelength over the range of wavelengths emitted spontaneously by the laser diode, and because cavities are relatively long compared to the lasing wavelength of the laser diode, there are multiple longitudinal optical modes supported by the cavity that can reach lasing threshold. The spectral width of a Fabry-Perot cavity diode laser is therefore relatively large, and for the wavelength range of 400 to 600 nm, is typically on the order of a nanometer in width.
To achieve a narrow lasing spectrum, the optical feedback mechanism must be restricted to be efficient for a narrower range of wavelengths. One mechanism for achieving this is use of a distributed feedback (DFB) structure such as a grating in the laser cavity. Such DFB gratings are periodic structures that result in a periodic variation in the optical index of material overlapping the optical mode. The period is chosen such that light scattered from the grating interferes constructively only for wavelengths that lie within the spontaneous emission spectrum of the laser diode. Typically, the period is chosen as half the wavelength of light in the laser medium, as this so-called first order grating only allows for reflections into the forward and backward traveling waves. Higher order gratings, where the grating period is some whole number multiple of the first order grating, are advantageous for visible lasers because the wavelength of visible light can be quite small in the laser media where effective optical indices can be significantly larger than 1. Higher order gratings are, therefore, easier to fabricate due to their larger critical dimensions. Higher order gratings have the downside that light can be constructively scattered out of the plane of the laser cavity. While this property can be used to outcouple light from the cavity, it is also a source of optical loss.
Typically, in GaN-based laser diodes, DFB lasers are challenging. For one, buried p-type layers are a challenge due to the difficulty in electrical activation of buried p-type doped layers. This results in epitaxial structures typically being grown with the p-contact layer as the final layer of the epitaxial stack such that the p-layers are exposed. A DFB grating etched into the p-layers of a GaN laser has two negative effects. First, p-type GaN is relatively sensitive to the introduction of point defects, and plasma-based etching of p-type GaN, such as would be used to produce DFB features with sub-micron critical dimensions, can render the p-type GaN resistive and can damage p-type contacts. Next, p-type contacts in GaN are often relatively resistive compared to other semiconductor systems. Removing a large fraction of the p-type contact layer to form a grating structure will increase the p-type contact resistance and increase the forward voltage of the device.
It is possible to produce so called regrown DFB gratings, where the epitaxial device wafer is removed from the growth process, patterned and etched, and then returned for a second regrowth step where a material with a contrasting optical index is grown over the grating layer. Such regrowth processes are challenging in GaN and are less preferred as the lower optical index contrast between two semiconductor layers is less ideal than the larger contrast between a semiconductor layer and an air or a low-index dielectric. The grating can also be etched only in the side of the laser cavity ridge, leaving the p-cladding and contact layers intact. This lowers the optical overlap with the grating, however, and leads to weaker feedback.
In other semiconductor material systems, a viable approach is to etch relatively deep gratings into only a portion of the cavity, and only the portion of the cavity that is free from the gratings is electrically injected. One drawback of this design is that the passive grating section would normally be highly absorbing of the lasing wavelength leading to large optical losses. Frequently steps are taken to make the passive section less or not-absorbing using regrown layers with lower absorption at the lasing wavelength or using layers that have been intermixed to shift the absorption band edge via a process such as ion implantation followed by annealing. In GaN devices, this configuration is challenging due to the difficulty of the regrowth processes on the one hand and the high bond energies that impede effective anneals to remove implantation damage on the other hand.
Another option to achieve a narrow lasing spectrum involves using complex optoelectronic devices that include active and passive device integration. These structures can be difficult to produce on gallium nitride (GaN). Optical elements with a large footprint and low density are cost-prohibitive due to small, expensive substrates, and they typically need a separate growth of low-loss material in addition to the high-gain active material to produce efficient passive devices. Growth of GaN on large, non-native substrates has been pursued for GaN devices, but it requires buffer layers and generally has poorer material quality than native substrates.
The present invention provides methods and structures suitable for the fabrication of high-performance semiconductor devices. In an embodiment, large area substrates are used to manufacture gallium and nitrogen containing devices based on an epitaxial transfer process where as-grown epitaxial layers are transferred from the substrate to a carrier wafer for processing. Methods and material processes for fabricating laser diodes or other gallium and nitrogen containing devices using transfer technologies are described, for example, in U.S. Pat. Nos. 9,666,677 and 9,379,525, the entire contents of which are incorporated herein by reference for all purposes. In an embodiment, the gallium and nitrogen containing devices are laser diode devices operating in the 390 nm to 425 nm range, the 425 nm to 480 nm range, the 480 nm to 510 nm range, the 510 to 550 nm range, or greater than 550 nm.
Rather than attempt the active and passive integration scheme by epitaxial growth, some embodiments described herein use dielectric layers that are optically coupled to the active layers to provide high-efficient optical elements at a lower cost. Merely by example, various embodiments provide cost effective methods and structures to manufacture laser diodes for applications such as quantum sensors, display systems, laser based lighting systems including remote and integrated smart laser lighting devices and methods, spatially dynamic lighting devices and methods, LIDAR, LiFi, and visible light communication devices and methods, and various combinations of above in applications of sensing, measurement, imaging, general lighting, commercial lighting and display, automotive lighting and communication, defense and security, industrial processing such as welding, cutting, and 3D printing of various materials, internet communications, modulators, demodulators, and photodiodes for communication and de-speckle, and waveguide couplers for beam shaping, filtering, signal processing, and routing. These devices can also be coupled to on-wafer driving electronics for higher-level system integration such as read-out integrated circuits (ROICs), trans-impedance amplifiers (TIAs), analog-to-digital converters (ADCs), driver electronics, and others.
Some embodiments described herein include gallium and nitrogen containing laser device layers that are removed from their original epitaxial growth substrate and are transferred to a second carrier substrate of arbitrary composition. In some embodiments, the epitaxial layer stack is inverted during transfer to the carrier wafer, such that the n-type cladding layers are exposed on the top of the epitaxial device stack. This may be advantageous since n-type gallium and nitrogen containing materials are generally more robust than p-type gallium and nitrogen containing materials and support electrical contacts that are better than electrical contacts to p-type gallium and nitrogen containing materials. Etching DFB grating features into the n-cladding of an epi-transferred laser device may have limited impact on the conductivity of the n-cladding layers and the resistance of the n-contact.
Some embodiments enable coupling multiple cavities together on the same carrier wafer via heterogeneous integration with high coupling efficiency. In accordance with an embodiment, for example, a method for manufacturing an optical device includes providing a carrier wafer and providing a first substrate having an array of first mesas. Each of the first mesas may include a gallium and nitrogen containing epitaxial material, where the gallium and nitrogen containing epitaxial material comprises a first release material overlying the first substrate and one or more n-type gallium and nitrogen containing layers, one or more light emitting gallium and nitrogen containing layers comprising an active region configured to emit electromagnetic radiation at a first wavelength, and one or more p-type gallium and nitrogen containing layers overlying the first release material. Each of the first mesas may also include a first interface region overlying the gallium and nitrogen containing epitaxial material. The method also includes bonding the first interface region of at least a fraction of the first mesas to the carrier wafer to form first bonded structures; releasing the first bonded structures to transfer the fraction of the first mesas to the carrier wafer to form first mesa regions on the carrier wafer; and providing a second substrate having an array of passive waveguide mesas. Each of the passive waveguide mesas may include a second release material overlying the second substrate; a passive waveguide region overlying the second release material; and a second interface region overlying the passive waveguide region. The method also includes bonding the second interface region of at least a fraction of the passive waveguide mesas to the carrier wafer to form second bonded structures; and releasing the second bonded structures to transfer the fraction of the passive waveguide mesas to the carrier wafer to form second mesa regions on the carrier wafer. Each of the first mesa regions may be arranged adjacent to a corresponding one of the second mesa regions on the carrier wafer in an end-to-end configuration.
In an embodiment, the method also includes processing the first mesa regions to form active laser devices each having first facets on each end; and processing the second mesa regions to form passive waveguide structures having second facets on each end, wherein one of the first facets on each of the active laser devices is aligned with one of the second facets on the corresponding passive waveguide structure.
In another embodiment, adjacent facets on the active laser device and the corresponding passive waveguide structure have anti-reflective coatings and are arranged so that electromagnetic radiation emitted by the active laser device enters the corresponding passive waveguide structure.
In another embodiment, the method also includes forming grating features on each of the second bonded structures, wherein the grating features are configured to provide optical feedback to electromagnetic radiation received by the second bonded structures.
In another embodiment, the grating features form a distributed feedback (DFB) structure.
In another embodiment, the end-to-end configuration includes a gap between closest ends of the first mesa region and the corresponding second mesa region.
In another embodiment, the end-to-end configuration includes abutting ends of the first mesa region and the corresponding second mesa region.
In another embodiment, the method also includes processing the first bonded structures to form active laser devices each having first facets on each end; and processing the second bonded structures to form passive waveguide structures having second facets on each end. Each of the active laser devices and each of the passive waveguide structures may have vertically angled turning mirrors, and one of the first facets on each active laser device may be at an angle relative to a direction of propagation of light within a cavity of the active laser device and may be aligned with one of the second facets on the corresponding passive waveguide structure.
In another embodiment, the method also includes processing the first bonded structures to form active laser devices each having first facets on each end; and processing the second bonded structures to form passive waveguide structures having second facets on each end, wherein each of the active laser devices and each of the passive waveguide structures have horizontally angled turning mirrors, and a first primary axis of each active laser device is substantially parallel to a second primary axis of the corresponding passive waveguide structure.
In yet another embodiment, the method also includes processing the first bonded structures to form active laser devices each having first facets on each end; and processing the second bonded structures to form passive waveguide structures having second facets on each end, wherein each of the active laser devices and each of the passive waveguide structures have horizontally angled turning mirrors, and a first primary axis of each active laser device is rotated relative to a second primary axis of the corresponding passive waveguide structure.
In accordance with another embodiment, an optical device includes a carrier wafer; a laser device comprising a gallium and nitrogen containing epitaxial material disposed on the carrier wafer, the gallium and nitrogen containing epitaxial material comprising a first interface region bonded to the carrier wafer, one or more p-type gallium and nitrogen containing layers overlying the first interface region, one or more light emitting gallium and nitrogen containing layers overlying the one or more p-type gallium and nitrogen containing layers, and one or more n-type gallium and nitrogen containing layers overlying the one or more light emitting gallium and nitrogen containing layers; and a dielectric waveguide overlying the one or more n-type gallium and nitrogen containing layers of the gallium and nitrogen containing epitaxial material, the dielectric waveguide comprising a dielectric core sandwiched between dielectric cladding regions, wherein the laser device and the dielectric waveguide are arranged and configured so that electromagnetic radiation generated by the laser device is coupled upwards into the dielectric waveguide and emitted by the dielectric waveguide.
In accordance with another embodiment, a method for manufacturing an optical device includes providing a carrier wafer; providing a first substrate having an array of first mesas, each of the first mesas comprising: a gallium and nitrogen containing epitaxial material, the gallium and nitrogen containing epitaxial material comprising a first release material overlying the first substrate and one or more n-type gallium and nitrogen containing layers, one or more light emitting gallium and nitrogen containing layers comprising an active region configured to emit electromagnetic radiation at a first wavelength, and one or more p-type gallium and nitrogen containing layers overlying the first release material; and a first interface region overlying the gallium and nitrogen containing epitaxial material; bonding the first interface region of at least a fraction of the first mesas to the carrier wafer to form first bonded structures; releasing the first bonded structures to transfer the fraction of the first mesas to the carrier wafer to form first mesa regions on the carrier wafer; processing the first mesa regions to form laser devices or gain sections having first facets on each end; forming a dielectric region overlying each of the first mesa regions on the carrier wafer, wherein the dielectric region comprises a dielectric core sandwiched between dielectric cladding regions; and processing the dielectric region on each of the first mesa regions to form dielectric waveguides having second facets on each end, wherein each laser device and corresponding dielectric waveguide are arranged and configured so that electromagnetic radiation generated by the laser device is coupled upwards into the dielectric waveguide.
In an embodiment, each of the dielectric waveguides is adjacent to the one or more n-type gallium and nitrogen containing layers of a corresponding one of the laser devices or gain sections.
In another embodiment, the dielectric core comprises silicon nitride and the dielectric cladding regions comprise silicon oxide.
In another embodiment, the dielectric waveguide is configured to emit the electromagnetic radiation generated by the laser device or gain section.
In another embodiment, the laser device and the dielectric waveguide are configured so that optical modes of the laser device or gain section and the dielectric waveguide overlap.
In another embodiment, n-contacts for the laser device are formed on each side of the dielectric waveguide.
In another embodiment, the laser devices or gain sections are tapered or grating coupled, and each of the dielectric waveguides extends over a corresponding one of the laser devices or gain sections.
In another embodiment, forming the dielectric region overlying each of the first mesa regions comprises depositing a dielectric layer and polishing the dielectric layer to planarize the dielectric layer.
In yet another embodiment, processing the dielectric region on each of the first mesa regions to form the dielectric waveguides includes forming a diffraction grating on each of the dielectric waveguides. In some embodiments, processing the dielectric region on each of the first mesa regions to form the dielectric waveguides includes forming a modulator on each of the dielectric waveguides.
In accordance with yet another embodiment, a method for manufacturing an optical device includes providing a carrier wafer; forming dielectric regions on the carrier wafer, wherein the dielectric regions each comprise a dielectric core sandwiched between dielectric cladding regions; processing the dielectric regions to form dielectric waveguides having facets on each end; providing a substrate having an array of mesas, each of the mesas comprising: a gallium and nitrogen containing epitaxial material, the gallium and nitrogen containing epitaxial material comprising a release material overlying the substrate and one or more n-type gallium and nitrogen containing layers, one or more light emitting gallium and nitrogen containing layers comprising an active region configured to emit electromagnetic radiation at a first wavelength, and one or more p-type gallium and nitrogen containing layers overlying the release material; and an interface region overlying the gallium and nitrogen containing epitaxial material; bonding the interface region of at least a fraction of the mesas to corresponding ones of the dielectric waveguides on the carrier wafer to form first bonded structures; releasing the first bonded structures to transfer the fraction of the mesas to the carrier wafer to form first mesa regions on the carrier wafer; and processing the first mesa regions to form laser devices or gain sections having facets on each end.
The following drawings are merely examples for illustrative purposes according to various disclosed embodiments and are not intended to limit the scope of the present invention.
Some embodiments described herein enable coupling multiple cavities together on the same carrier wafer via heterogeneous integration with high coupling efficiency. In accordance with an embodiment, for example, a first and second epitaxial substrate are provided. In one epitaxial growth operation(s), the epitaxial device layers of a laser diode are grown on the first epitaxial substrate, and in another epitaxial growth operation(s), the epitaxial device layers for a passive waveguide are grown on the second epitaxial substrate. The passive waveguide may comprise any material or crystal orientation. As an example, the passive waveguide may comprise a c-plane gallium and nitrogen containing material. The passive waveguide may have an absorption edge corresponding to shorter wavelengths than the lasing wavelength of the laser diode. The epitaxial device layers of the first and second substrates may be fabricated using standard lithographic and etching techniques to form a plurality of mesas. A bond media is deposited on the top surfaces of the mesas. A common carrier wafer is provided where a first surface of the carrier wafer is overlaid by regions of bond media. Using transfer processes, one or more mesas are transferred from each of the first epitaxial substrate and the second epitaxial substrate to the common carrier wafer. In some embodiments, the mesas having active device layers are transferred from the first epitaxial substrate and the mesas having the passive waveguide layers are transferred from the second epitaxial substrate to the common carrier wafer such that light generated in each of the active device mesas is coupled into one of the passive waveguide structures.
An exemplary transfer process is described herein. Embodiments are not limited to the described transfer process, however, and any transfer process may be used. The exemplary transfer process provides a method and structure for producing large area gallium and nitrogen engineered substrate members configured for the epitaxial growth of layer structures suitable for the fabrication of high-performance semiconductor devices. In a specific example, the engineered substrates are used to manufacture gallium and nitrogen containing devices based on an epitaxial transfer process wherein as-grown epitaxial layers are transferred from the engineered substrate to a carrier wafer for processing. The gallium and nitrogen containing devices may be laser diode devices operating in the 390 nm to 425 nm range, the 425 nm to 480 nm range, the 480 nm to 510 nm range, the 510 nm to 550 nm range, or greater than 550 nm. Merely by way of example, various embodiments provide cost effective methods and structures to manufacture laser diodes in applications such as quantum sensors, display systems, laser based lighting systems including remote and integrated smart laser lighting devices and methods, spatially dynamic lighting devices and methods, LIDAR, LiFi, and visible light communication devices and methods, and various combinations of above in applications of sensing, measurement, imaging, general lighting, commercial lighting and display, automotive lighting and communication, defense and security, industrial processing such as welding, cutting, and 3D printing of various materials, and internet communications, and others.
Merely by way of example, embodiments described herein can be applied to applications such as white or infrared lighting, white or infrared spot lighting, general lighting, specialty lighting, dynamic lighting, smart lighting, flash lights, automobile headlights, automobile interior lighting, automobile position lighting and any lighting function, mobile machine lighting such as autonomous machine lighting and drone lighting, all-terrain vehicle lighting, light, sensing or communication systems, navigation systems, advanced driver assistance systems (ADAS), autonomous or semi-autonomous mobile machines and robots, sources used in recreational sports such as biking, surfing, running, racing, boating, light sources used for drones, planes, robots, other mobile or robotic applications, safety, counter measures in defense applications, multi-colored lighting, lighting for flat panels, medical applications including cancer treatment or ablation or cosmetic surgery, metrology and measurement applications, beam projectors and other display devices and systems, frequency doubling systems such as second harmonic generation (SHG) systems, SHG systems combined with nonlinear crystals like barium borate (BBO) for producing wavelengths in the 200 nm to 400 nm range, wearable displays, augmented reality systems, mixed reality systems, virtual reality systems, high intensity lamps, spectroscopy, entertainment, theater, music, and concerts, analysis fraud detection and/or authenticating, tools, purification, sterilization, anti-virus, anti-bacterial, water treatment, security systems, laser dazzlers, targeting, communications, LiFi, visible light communications (VLC), sensing, detecting, distance detecting, Light Detection And Ranging (LIDAR), smart infrastructure such as smart factories or smart homes, transformations, transportations, leveling, curing and other chemical treatments, heating, cutting and/or ablating, welding, marking, laser direct imaging, pumping other optical devices, other optoelectronic devices and related applications, storage systems, quantum computing, quantum cryptography, quantum storage, and source lighting and the like.
As background, while LED-based light sources offer great advantages over incandescent based sources, there are still challenges and limitations associated with LED device physics. The first limitation is the so called “droop” phenomenon that plagues GaN based LEDs. The droop effect leads to power rollover with increased current density, which forces LEDs to hit peak external quantum efficiency at very low current densities in the 10-200 A/cm2 range.
A second limitation of LEDs is also related to their brightness, more specifically it is related to their spatial brightness. A conventional high brightness LED emits ˜1 W per mm2 of epi area. With some advances and breakthroughs, perhaps this can be increased up to 5-10× or 5-10 W per mm2 of epi area. Finally, LEDs fabricated on conventional c-plane GaN suffer from strong internal polarization fields, which spatially separate the electron and hole wave functions and lead to poor radiative recombination efficiency. Since this phenomenon becomes more pronounced in InGaN layers with increased indium content for increased wavelength emission, extending the performance of UV or blue GaN-based LEDs to the blue-green or green regime has been difficult.
An exciting new class of solid-state lighting based on laser diodes is rapidly emerging. Like an LED, a laser diode is a two-lead semiconductor light source that that emits electromagnetic radiation. However, unlike the output from an LED that is primarily spontaneous emission, the output of a laser diode is comprised primarily of stimulated emission.
The laser diode contains a gain medium that functions to provide emission through the recombination of electron-hole pairs and a cavity region that functions as a resonator for the emission from the gain medium. When a suitable voltage is applied to the leads to sufficiently pump the gain medium, the cavity losses are overcome by the gain and the laser diode reaches the so-called threshold condition, wherein a steep increase in the light output versus current input characteristic is observed. At the threshold condition, the carrier density clamps and stimulated emission dominate the emission. Since the droop phenomenon that plagues LEDs is dependent on carrier density, the clamped carrier density within laser diodes provides a solution to the droop challenge. Further, laser diodes emit highly directional and coherent light with orders of magnitude higher spatial brightness than LEDs. For example, a commercially available edge emitting GaN-based laser diode can reliably produce about 2 W of power in an aperture that is 15 μm wide by about 0.5 μm tall, which equates to over 250,000 W/mm2. This spatial brightness is over 5 orders of magnitude higher than LEDs or put another way, 10,000 times brighter than an LED.
Based on essentially all the pioneering work on GaN LEDs, visible laser diodes based on GaN technology have rapidly emerged over the past 20 years. Currently the only viable direct blue and green laser diode structures are fabricated from the wurtzite AlGaInN material system. The manufacturing of light emitting diodes from GaN related materials is dominated by the heteroepitaxial growth of GaN on foreign substrates such as Si, SiC and sapphire. Laser diode devices operate at such high current densities that the crystalline defects associated with heteroepitaxial growth are not acceptable. Because of this, very low defect-density, free-standing GaN substrates have become the substrate of choice for GaN laser diode manufacturing. Unfortunately, such bulk GaN substrates are costly and not widely available in large diameters. For example, 2″ diameter is the most common laser-quality bulk GaN c-plane substrate size today with recent progress enabling 4″ diameter LED quality, which are still relatively small compared to the 6″ and greater diameters that are commercially available for mature substrate technologies. Further details of the present invention can be found throughout the present specification and more particularly below.
Beyond the c-plane bulk GaN substrate technology, an alternative class of substrate orientations are emerging based on the promise and demonstration of improved performance, improved reliability, and potential for lower cost manufacturing. Specifically, these nonpolar and semipolar orientations offer eliminated or reduced piezo and spontaneous electric fields within the crystal that plagues conventional c-plane (i.e., polar) gallium nitride orientations. The reduced internal electric fields improve the electron-hole overlap and increases the radiative recombination efficiency in light emitting devices such as laser diodes and LEDs, which in turn, can improve the overall performance and efficiency of the device. Specifically, in laser diode devices, the improved radiative efficiency can increase the overall gain of the laser diode to enable lower threshold current densities for reduced operating current and voltage. Moreover, with the elimination or mitigation of these internal fields present on c-plane GaN, a greater design space is afforded for the epitaxial layer structure comprising the claddings, waveguide, and active regions of the laser diodes, which can further enable improved performance, reliability, and manufacturing throughput. Some example nonpolar and semipolar orientations include the nonpolar m-plane or a-plane, and semipolar orientations such as the {30-31}, {20-21}, {30-32}, {11-22}, {10-11}, {30-3-1}, {20-2-1}, {30-3-2}, or offcuts of any of these polar, nonpolar, and semipolar planes within +/−10 degrees towards a c-plane, and/or +/−10 degrees towards an a-plane, and/or +/−10 degrees towards an m-plane.
Although these semipolar/nonpolar gallium nitride orientations can offer certain improvements with respect to performance, reliability, and throughput/cost, the widespread development and deployment is constrained by the availability of substrates with large area, high quality, and reasonable cost. Specifically, high quality nonpolar and semipolar bulk GaN substrates today are only available through the growth of large boules of c-direction oriented gallium nitride, which is then cross cut into the nonpolar or semipolar orientations. Bulk boule growth is typically achieved with hydride vapor phase epitaxy (HVPE) or variants of ammonothermal growth, which tend to perform better when growth is performed in the c-direction. Since high-quality boule growth is limited from about 2 mm to about 10 mm in the c-direction, the resulting cross-cut nonpolar or semipolar substrates will have a projection of c-direction size constraint of about 2 mm to about 10 mm. The net outcome is high quality nonpolar and semipolar substrates that are limited in size from about 1 cm2 to about 4 cm2, which is drastically less area than that of a 2″ substrate with about 20 cm2 of area. High volume manufacturing of low-cost semiconductor devices such as laser diodes and LEDs on small substrates is challenging and even prohibitive due to the high amount of edge/exclusion zone, the difficulty handling small pieces, the challenge associated with moving 10-20 more wafers through the fabrication process, the incompatibility of small wafers with high volume process equipment and tooling, and the generally high per area cost for the specialized substrates.
Although approaches to large area nonpolar and semipolar substrate growth are under development, they presently suffer from severe drawbacks that keep them from commercial availability and high demand. One such drawback is reduced quality in the form a high density of crystal defects such as basal plane stacking faults that can be harmful to device performance. In other approaches, bulk growth techniques are combined with the tiling of high-quality small-area nonpolar or semipolar substrate components to create a large-area mosaic substrate. This approach leads to boundary regions, or seams, between the high-quality, small-area tiled substrate regions. One drawback to this method is the often irregular spacing and placement of the seams. This poses difficulty with placement of devices during the fabrication process and can lead to excessive yield losses. Another challenge associated with this process is that the subsequent bulk growth on the large area mosaic substrate often leads to crystal defects such as basal plane stacking faults.
As used herein, the term GaN substrate is associated with Group III-nitride based materials including GaN, InGaN, AlGaN, or other Group III containing alloys or compositions that are used as starting materials. Such starting materials include polar GaN substrates (i.e., substrate where the largest area surface is nominally an (h k 1) plane wherein h=k=0, and 1 is non-zero), non-polar GaN substrates (i.e., substrate material where the largest area surface is oriented at an angle ranging from about 80-100 degrees from the polar orientation described above towards an (h k 1) plane wherein 1=0, and at least one of h and k is non-zero), or semi-polar GaN substrates (i.e., substrate material where the largest area surface is oriented at an angle ranging from about +0.1 to 80 degrees or 110-179.9 degrees from the polar orientation described above towards an (h k 1) plane wherein 1=0, and at least one of h and k is non-zero). Of course, there can be other variations, modifications, and alternatives.
As used herein, the term engineered substrate and composite substrate are used interchangeably. In general, epitaxial substrates are monolithic wafers comprised by a single orientation and crystal structure of a single composition of a crystalline material. The substrate surface on which epitaxial growth is performed provides an exposed crystalline plane of the substrate material that acts as a template for determining the orientation and crystal structure of the epitaxially grown layers. In many cases for group III-Nitride crystal growth the substrate has a different composition and crystal structure from the III-Nitride epitaxial film, but similar crystal symmetry and lattice constants that enable growth of low defect density group III-Nitride epitaxial films. Even in the case of epitaxial growth on bulk group III-Nitride substrates, most of the substrate is not directly involved with epitaxial growth and only ensures the substrate is sufficiently stiff and strong enough to survive the epitaxial growth process. An engineered or composite substrate is composed of a plurality of layers characterized by difference in one or more of composition, crystal structure, and crystal orientation. Epitaxial growth on an engineered substrate is enabled by the inclusion of a high-quality, crystalline seed layer upon which epitaxial layers grow and from which they take their crystal structure and orientation. As used herein, the terms crystalline seed layer, epitaxial seed layer and seed layer are interchangeable. The remainder of the engineered substrate is comprised by one or more layers of materials with different composition or microstructure from the seed layer, and which can ensure the substrate is stiff and strong while providing some advantage over the seed layer material such as reduced cost, large available crystal sizes, improved electrical properties, and the like.
In an example, the seed layer of the engineered substrate is a thin GaN layer of thickness between 10 and 1000 nanometers. This GaN seed layer is comprised by a surface region oriented in either a polar, semipolar, or non-polar configuration. In some examples, the GaN seed layer is derived from a low defectivity, single crystal GaN crystal with a threading dislocation density below 5e-6 dislocations per square centimeter. In some examples, the seed layer is ≤10 microns thick. In other examples, the seed layer is ≤100 microns thick. In some examples, the seed layer is comprised by a plurality of crystalline film regions.
In another example, the seed layer of the engineered substrate is a thin silicon layer of thickness between 10 and 1000 nanometers. This silicon layer is comprised by a surface region oriented in a {111} direction, potentially with some intentional misorientation away from {111} by an angle less than 10 degrees.
In another example, the seed layer of the engineered substrate is a thin layer of a hexagonal polytype of SiC with a thickness between 10 and 1000 nanometers. This SiC layer is comprised by a surface region oriented in a {0001} direction, potentially with some intentional misorientation away from {0001} by an angle less than 10 degrees.
In another example, the seed layer of the engineered substrate is a thin layer of a sapphire with a thickness between 10 and 1000 nanometers. This sapphire layer is comprised by a surface region oriented in a {0001} direction, potentially with some intentional misorientation away from {0001} by an angle less than 10 degrees.
In another example, the seed layer of the engineered substrate is a thin layer of III-V semiconductor alloy with a composition of AlxInyGa(1-x-y)AszP(1-z) where 0≤x+y≤1 and 0≤z≤1.
The seed layer is disposed on a handle wafer that is, herein, also called a handle substrate or mechanical handle wafer or substrate. The purpose of the handle substrate is to provide physical support for the crystalline seed layer, which is too thin to be ridged. The handle wafer material is chosen to provide a ridged support for the thin sacrificial and seed layers. It need not have the same composition nor crystal structure as the epitaxially grown device layers, nor does it need to be single crystalline nor even crystalline. Preferred properties of the handle wafer are that it is solid and ridged at temperatures associated with growth of the device layers. For MBE this would be up to 900° C. or more, and for MOCVD this would be up to 1200° C. or more. The handle wafer should have similar coefficient of thermal expansion as the sacrificial, seed and device layers to prevent excessive bowing or cracking during or after device layer growth. In some examples the handle wafer material has a coefficient of thermal expansion of between 1E-6 and 1E-5. The handle wafer should neither significantly decompose in nor react with the growth environment. In MOCVD growth of GaN related alloys is frequently carried out in the presence of chemical species such as hydrogen, ammonia, hydrocarbons such as methane and ethane and with reactive metals among others.
In an example, the engineered substrate handle wafer material is chosen from sapphire, AlN, poly-crystalline AlN, boron nitride, SiC, polycrystalline SiC and sintered alumina. In another example, the engineered substrate handle wafer material is chosen from quartz, silicon, poly-crystalline silicon, molybdenum and tungsten among others.
In some examples, the engineered substrate contains a selectively removable selective release layer. The selective release layer, or selective release layer, is a layer comprised of a material that is selectively removable either by chemical etching or by exposure to an energy source such as optical radiation or heat. In some examples, the selectivity of the removal process for the selective release layer relative to the other layers in the engineered substrate and the epitaxial layers is very large such that only the selective release layer is significantly removed. In other examples, the layers of materials highly resistant to the removal process are disposed above and below the selective release layer or layers. Other layers in the structure with lower selectivity to the removal process are protected by passivating coatings.
In an example, the selective release layer is composed of silicon oxide or silicon nitride and is selectively removable by chemical etching with acid mixtures containing HF acid. In another example, the selective release layer is composed of semiconductor material such as germanium and is selectively removable using chemical etching with mixtures containing hydrogen peroxide. In another example, the selective release layer is composed of a metal such as titanium which can be etched with chemical mixtures containing HF acid. Other combinations of selective release layers and removal chemistries are possible, with the key feature being that the selective release layer is removable with high selectivity relative to the overlaid epitaxial device layers. In an example, the selectivity of removal of the selective release layer would also be high relative to the handle wafer material, but this is not necessary.
In an example, the selective release layer is composed of silicon dioxide and has a thickness less than 10 micrometers. The oxide thickness may be between 100 and 1000 nm. Thinner than 100 nm and lateral etch rates with acid solutions containing HF may require long times to fully etch the silicon oxide layer. Much thicker and residual stress in the oxide layer may result in wafer bowing during or after growth of the epitaxial device layers.
In some examples, the bonding and selective release layers of the engineered substrate are the same layer. For example, a selective release layer can be comprised of one or more silicon dioxide layers used to bond the seed layer to the handle wafer. The oxide bonding layers can be selectively removed, relative to the seed and handle layers, using acid solutions containing HF acid.
In some examples, the selective release layer may be removed using application of an energy source such as a laser. In this case, the selective release layer is chosen to be a material that absorbs the laser preferentially to both the handle wafer of the engineered substrate as well as the epitaxial device layers. By making the selective release layer sufficiently thick relative to the absorption length of the laser light in the selective release layer material, most of the laser light dose would be absorbed in the selective release layer and the device layers would not be exposed to damaging intensities of laser radiation. With proper laser power and focus the selective release layer could be ablated in a process like the laser lift-off (LLO) process used to detach heteroepitaxial GaN layers from sapphire wafers.
In an example, there is a single layer that acts as both the sacrificial and seed layers and which is made from silicon that is highly absorbing at the wavelength of a laser used to ablate or melt the Si layer to selectively remove it. The absorption length for Si for wavelengths shorter than 400 nm is below 100 nm. At 350 nm the absorption length is on the order of 10 nm. Light with a wavelength of 351 nm emitted from a XeF laser would be substantially absorbed by a Si layer with a thickness of 100 nm. Other laser sources and Si thicknesses could be chosen, as should be obvious to one knowledgeable in the art. Other appropriate selective release layers would be Ge, transparent conductive oxides such as ITO or ZnO that absorb in the ultra-violet, silicon oxides or nitrides contaminated with metal species to make them absorbing, and the like. Such a laser lift-off process could be used to form anchor features by selectively ablating some regions of the selective release layer and not others.
In some examples, the engineered substrate contains a bond layer that joins the crystalline seed layer to the handle wafer. In some examples, the bond layer is comprised of two layers which are deposited separately on both the seed layer bonding surface and the handle wafer bonding surface prior to bonding. The crystal donating the seed layer is then brought into contact with the handle wafer such that the bond layers are in contact, and the bond layers are fused together through the application of one or both of elevated temperature and elevated pressure. In some examples, the bond is formed between layers of metal or oxide using thermocompression bonding. Metal which could be used for thermocompression bonding include Au, Al, and Cu among others. In some examples, the thermocompression bonding is improved through the application of plasma treatments to clean or chemically activate the surfaces of the metal or oxide bonding media. In some examples, relatively low melting point oxides (e.g., borophosphorosilicate glass, etc.) are used as the bonding media.
Thermocompression bonding involves bonding of the epitaxial semiconductor layers to the carrier wafer at elevated temperatures and pressures using a bonding media disposed between the epitaxial layers and handle wafer. The bonding media may be comprised of different layers, but typically contain at least one layer (the bonding layer) that is composed of a relatively ductile material with a high surface diffusion rate. In many cases this material is comprised of Au, Al or Cu. The bonding media may also include layers disposed between the bonding layer and the epitaxial materials or handle wafer that promote adhesion. For example, an Au bonding layer on a Si wafer may result in diffusion of Si to the bonding interface, which would reduce the bonding strength. Inclusion of a diffusion barrier such as silicon oxide or nitride would limit this effect. Relatively thin layers of a second material may be applied on the top surface of the bonding layer to promote adhesion between the bonding layers disposed on the epitaxial material and handle. Some bonding layer materials of lower ductility than gold (e.g., Al, Cu etc.) or which are deposited in a way that results in a rough film (for example electrolytic deposition) may require planarization or reduction in roughness via chemical or mechanical polishing before bonding, and reactive metals may require special cleaning steps to remove oxides or organic materials that may interfere with bonding.
Thermocompression bonding can be achieved at relatively low temperatures, typically below 500 C° and above 200 C°. Temperatures should be high enough to promote diffusivity between the bonding layers at the bonding interface, but not so high as to promote unintentional alloying of individual layers in each metal stack. Application of pressure enhances the bond rate and leads to some elastic and plastic deformation of the metal stacks that brings them into better and more uniform contact. Optimal bond temperature, time and pressure will depend on the bond material, the roughness of the surfaces forming the bonding interface and the susceptibility to fracture of the handle wafer or damage to the device layers under load.
In some examples, thermocompression bond media is deposited on the handle and crystal donating the seed layer in a lithographically defined pattern rather than a blanket deposition. This can be advantageous because it enables the inclusion of alignment marks into the bond media layer as well as increases the bonding pressure achievable for any given applied force.
In some examples, the seed layer and the handle are joined with direct wafer bonding. For example, GaN and Si can be directly bonded under vacuum at room temperature and the bond solidified by long anneals at relatively low temperature. Such bonding may include plasma treatments of the seed layer and handle surfaces to remove contaminants and native oxides. In some examples, GaN seed layers are bonded directly to sapphire, silicon or SiC handle wafers.
In an example, the engineered substrate is fabricated by deposition silicon oxide on a handle wafer and on a bulk GaN substrate followed by Chemical-Mechanical Polishing (CMP) to reduce the roughness of the oxide layers to below 10 nm RMS such that the oxide surfaces will be in intimate contact when brought together. This is followed by an ion implantation of atoms of one or more elements, such as hydrogen and helium among others, at a shallow depth of less than 1 micron into the bulk GaN substrate such that the implanted species are concentrated at a narrow range of depth in the GaN wafer. The oxide surfaces of the handle and GaN wafer are then brought into contact and bonded at elevated temperature and pressure to produce a stable bond. A second anneal at a further elevated temperature induces the implanted species to come out of solid solution in the GaN wafer, thereby initiating microfracture of the GaN at the depth of the implantation. This microfracture releases the layer of GaN above the implantation depth and results in the formation of an engineered substrate consisting of the released layer of GaN overlaying the bonding oxide, which overlays the handle wafer. In this example, the GaN layer acts as the seed layer for growth of epitaxial device layers, while the bonding oxide may be used as a selective release layer, which can be selectively etched using solutions containing HF acid.
In another example, the engineered substrate is fabricated by deposition of silicon oxide on a handle wafer and on a bulk seed wafer consisting of GaN, Si, SiC or sapphire, among other materials, and use of Chemical-Mechanical Polishing (CMP) to reduce the roughness of the oxide layers to below 10 nm RMS such that the oxide surfaces will be on the whole in intimate contact when brought together. The oxide surfaces of the handle and seed wafer are then brought into contact and bonded at elevated temperature and pressure to produce a stable bond. The seed wafer is then partially removed using an etch or lap and polish process such that only a thin seed layer of less than 10 microns and preferably less than 1 micron remains overlaying the bonding oxide. This method is less preferable, as it requires destructive removal of the bulk of the bulk seed wafer, which could be an expensive process in terms of time as well as lost seed wafer material should the seed layer be composed of an expensive material such as GaN or SiC.
In another example, the engineered substrate is fabricated by deposition of a mask material on a bulk GaN wafer or a GaN template consisting of a GaN layer grown heteroepitaxially on a foreign substrate such as Si, SiC, sapphire, gallium oxide or the like as shown in
The overgrowth mask could consist of any material on which GaN or related alloys do not readily nucleate during epitaxial growth. For example, silicon nitride and silicon oxide are frequently used as overgrowth masks because under specific growth conditions the nucleation of nitride alloys on these materials can be suppressed. Thus, new material is only deposited on exposed seed wafer material or on previously grown nitride material that is not overlaid with mask material. Other overgrowth mask materials used include metals such as Ti and W as well as metal containing ceramic films such as TiN. Overgrowth mask thicknesses are limited by a minimum value limited by that needed to ensure the mask layer is continuous upon deposition and remains continuous should the mask material decompose under typical growth conditions. The upper limit of mask thickness is set by practical aspects of growth, i.e., one would not want an excessively thick mask as it would add extra time the growth process associated with the initial vertical growth of nitride material in the non-masked regions.
The shape of the overgrowth mask regions could be adjusted to correspond to the regions needed to form the optoelectronic device. For example, a substrate intended for a multi-mode laser with stripe widths of 30 microns could have overgrowth mask regions in the shape of 40-micron wide stripes of many millimeters in length. The width of the regrowth mask in excess of the laser stripe width would be needed to accommodate errors in alignment of transfer and lithographic processes. In the example of a single mode laser, where ridge widths are on the order of 1 to 3 microns the width of the overgrowth mask region would be dominated by the excess area needed to accommodate alignment errors in transfer and lithographic processes. For example, combinations of overgrowth mask widths and laser ridge widths could respectively be 10 microns and less than 3 microns, 20 microns and less than 13 microns, 30 microns and less than 23 microns, 40 microns and less than 33 microns and so forth.
In order to produce die suitable for die expansion on such a substrate, device layers are then grown on the wafer and may be initiated with the growth of one or more so-called buffer layers that do not add functionality to the device but rather are grown under conditions or with compositions that promote growth of smooth, low defect density or otherwise higher quality device layers than would be achieved without the buffer layers. As shown in
In an example, the mask layer is composed of silicon oxide, silicon nitride or silicon oxynitride with a thickness between 10 and 10000 nm, though thicker masks would be functional if not ideal. The selective removal process includes one or more of etching in acidic solution containing HF acid, etching in HF vapor or a dry etching in a fluorine-based plasma.
In another example, the mask layer is composed of titanium oxide with a thickness between 10 and 10000 nm, though thicker masks would be functional if not ideal, and the selective removal process includes etching in one or more of solutions containing phosphoric acid, sulfuric acid and HF acid.
In another example, the mask layer is composed of aluminum oxide with a thickness between 10 and 10000 nm, though thicker masks would be functional if not ideal, and the selective removal process includes etching in one or more of solutions containing ammonium hydroxide and peroxide, bromine in alcohol and phosphoric acid.
In another example, a multi-layer mask is used. For example, a mask may consist of a sacrificial germanium layer overlaid by a silicon oxide layer acting as a selective deposition mask. The silicon oxide provides a surface with a low tendency for GaN to nucleate on during MOCVD growth and which can be more easily overgrown with GaN. The Ge layer allows for the die to be undercut using a peroxide based etch that selectively removes the Ge. This would be an example where the bond medium is also an oxide or contains an oxide, and selective etching of the oxide mask would also remove the bond medium. Other multi-layer mask combinations might include silicon nitride as the selective deposition mask and may contain sacrificial under-layers composed of titanium, gold, silicon, aluminum, chrome, tungsten, molybdenum, platinum, silver, copper and palladium among other materials. In general, the sacrificial under-layer would need to be sufficiently thick 10-1000 nm as to allow rapid removal of the selective release layer. At the same time, the sacrificial under-layer material may be one on which nitride materials nucleate easily, such that too thick an under-layer would lead to nucleation of a region of misoriented or highly defective nitride material on the sidewall of the under-layer.
In an example, a Ge under-layer is overlaid with a silicon oxide overgrowth mask layer using a photoresist mask and a lift-off process. The Ge thickness is kept between 100 and 500 nm. In another example, the Ge under-layer is deposited and then overlaid by the silicon oxide overgrowth mask layer using a photoresist mask and a lift-off process. The Ge underlayer is then selectively etched with hydrogen peroxide such that the edge of the Ge layer is recessed beneath the edge of the silicon oxide layer, thereby reducing the access of precursors to the Ge and encouraging nucleation on the exposed seed material only. In another example, the Ge under-layer and the silicon oxide overgrowth mask layer are deposited as blanket layers without patterning. Photoresist is patterned on the surface and acts as an etch mask to first etch back the silicon oxide using a wet or dry etch process. The Ge underlayer is then selectively etched with hydrogen peroxide such that the edge of the Ge layer is recessed beneath the edge of the silicon oxide layer. Acceptable dry etches for silicon oxide include reactive ion etching (RIE), inductively couple plasma (ICP) etching, chemically assisted ion beam etching (CAIBE) and the like. Dry etch reactive species for silicon oxide include fluorine containing species such as CF4, CHF3, C4F8, SF6, and the like, and may include use of inert gasses such as Ar and He along with other reactive species such as oxygen containing species like O2 and NO2 among others.
In another example, a selective release layer and overlaying seed layer are deposited on GaN substrate using an epitaxial process or directly convert a portion of the seed wafer at or near the epitaxial surface into a selective release layer. A schematic representation of the process for producing such a substrate and fabricating parts on it is shown in
In another example, a bulk GaN wafer or a GaN template comprised of a GaN layer grown epitaxially on a foreign substrate such as silicon, SiC, Sapphire or the like would be used as the mother wafer to form the engineered substrate. A first epitaxial process is used to grow a selective release layer overlaying the epitaxial surface of the seed wafer. The selective release layer is comprised of one or more sacrificial InGaN layers clad on both sides by layers of some composition of GaN, AlN, AlGaN, InAlN, or InAlGaN alloys with a wider bandgap than the composition of the sacrificial InGaN layers. A non-sacrificial seed layer of some composition of GaN, AlN, AlGaN, InAlN, or InAlGaN with a wider bandgap than the sacrificial InGaN layers is then grown overlaying the selective release layer. In an example, the InGaN selective release layers are selectively etched using a photoelectrochemical etch process. In another example, the InGaN layers are selectively ablated using a laser-lift-off process. In another example, the InGaN selective release layers are selectively decomposed into their constituent elements by the application of heat either in the form of annealing in a furnace or rapid thermal annealer or by application of a laser operating at wavelengths absorbed by the sacrificial InGaN layers.
In another example, a bulk GaN wafer is again used as a mother wafer. The selective release layer is produced via ion implantation. Ions are implanted into the GaN wafer at a shallow depth of 100-1000 nm at a high enough energy or density that the optical absorptivity of the GaN implanted layer is greatly increased. This layer then becomes susceptible to laser irradiation such that a laser lift-off process can be used to selectively remove the implanted layer prior to the die expansion process. The absorptivity of the GaN at a target laser wavelength should be increased above 1000 cm−1, and ideally would be increased above 10000 cm−1 to provide the strongest absorption of laser light.
It is desirable to fabricate an engineered substrate where the total area of seed layer on the engineered wafer is larger than the area of the mother wafer. This can be achieved by sequentially transferring seed layers to the same handle wafer using one or more mother wafers. Another method for achieving a larger area engineered substrate is to configure multiple mother wafers into a single, so-called tiled mother that can be used to simultaneously transfer the seed layers from the plurality of mothers.
In an example, a tiled engineered substrate is fabricated using serial transfer from multiple donors. The process for this is shown in
In an example, a tiled engineered substrate is fabricated using parallel transfer from multiple mother crystals. The process for this is shown in
A tiled mother wafer or crystal can be fabricated from a plurality of smaller mother crystals. This is achieved by bonding the mother crystals to a handle wafer using a bond that can survive the seed layer transfer processes. The bond can be formed either by direct wafer bonding between the mother crystal and the handle wafer at elevated temperature or pressure, or oxide-oxide bonding at elevated temperature and pressure by depositing oxide layers on the bonded surfaces of both the mother crystal and the handle wafer.
Another method for bonding mother crystals to a handle is to use thermocompression bonding of metal films deposited on the bonded surfaces of both the mother crystal and the handle wafer. Metals that can be used for thermocompression bonding include gold, silver, aluminum, and copper among others. Thermocompression bonding is advantageous in that a strong bond can be formed using a material that melts at temperatures higher than the maximum process temperature of the seed layer transfer process. Indium, aluminum, silver, gold, and copper melt at approximately 157 degrees C., 660 degrees C., 961 degrees C., 1063 degrees C., and 1084 degrees C., respectively, allowing for a wide range of thermocompression bonding materials depending on the maximum temperature of the seed transfer process. Au is the preferred metal for thermo-compression bonding as it is ductile, has a high melting point, and is resistant to oxidation which can inhibit the bonding process. Bonding of mother crystals to a handle can also be achieved with the use of solders, however these are less preferable to thermocompression bonding due to the relatively low temperatures at which these materials melt. For example, Au—Sn eutectic solder widely used in the electronics industry melts at approximately 280 degrees C. Bonding of mother crystals to a handle can also be achieved with the use of nano-particle pastes. These pastes are advantageous for this application in that the melting point of many metals is dramatically reduced when the metal particle size is reduced below 100 nm diameter. This allows for a nano-particulate metal film to be melted or sintered at relatively low temperature while the melting point of the resulting quasi-bulk metal film is similar to that of a bulk metal film.
A tiled mother crystal should comprise individual mother crystals that are bonded to a handle such that their crystal structures are oriented similarly with respect to each other. Variation in the crystal orientation of the mother wafers will be translated to any engineered substrates fabricated by transferring seed material from the tiled mother crystal. Any properties of the epitaxial films grown on the engineered substrate which depend on the crystal orientation of the seed layer with therefore vary across the full area of the engineered substrate. It is preferred that the lattice vectors of the mother wafers be parallel to each other to a tolerance of 10 degrees or less. It is most preferred that the lattice vectors of the mother wafers be parallel to each other to a tolerance of 1 degree or less.
A tiled mother crystal should comprise individual mother crystals that are bonded to a handle such that the gap between the individual mother crystals is minimized. This is advantageous because in the resulting engineered substrate any gap between individual mother crystals will result in a discontinuity of the transferred seed layer. In epitaxial growth techniques such as MOCVD, where vapor-phase and surface mobilities of precursor compounds are high, gaps in the seed layer can lead to spatial variation in epitaxial layer growth rates and compositions as precursors landing on the no-seed-layer portion of the engineered substrate diffuse to the regions with seed layer leading to a local increase in precursor concentrations near the edges of the discontinuous seed layer regions.
The process for transfer of seed layers from a tiled mother wafer is shown in
A single transferred seed layer or plurality or transferred seed layers can be configured to have various shapes as well as to be positioned in various ways on the handle wafer. In some examples, the transferred seed layers may have a variety of shapes such as square, rectangular, circular, or round shaped like various other polygons (triangles, hexagons, etc.), or be irregular in shape. It should also be understood that transferred seed layer regions may have substantial deviations from a nominal shape due to imperfect transfer of the seed layer. For example, a transferred seed layer may have irregular edges or may contain interior regions that are not transferred. A single transferred seed layer may be transferred at the center of a larger handle wafer or off-center, and it may be oriented in-plane in any way respective to the crystal structure or orienting features of the handle wafer. A plurality of seed layer regions may be transferred to a larger handle wafer with arbitrary positions and arbitrary in-plane orientations respective to the crystal structure or orienting features of the handle wafer. In an example, the plurality of seed layer regions are transferred to a larger handle wafer in a regular way that such that the seed layer locations and in-plane orientations are controlled to within some tolerance. Regular placement of transferred seed layers reduces the complexity of device fabrication after epitaxial growth.
As previously mentioned, a key element to this invention is that the boundaries (i.e., seam regions) separating each of the epitaxial seed layers in engineered substrates containing more than one crystalline seed regions are precisely spatially positioned relative to each other such that devices can be efficiently positioned with respect to the seam regions to avoid excessive yield losses or process variability that would result from seam position variation. In examples where more than one crystalline seed region is transferred to a handle wafer serially from one or more mother crystals the precise placement would be done during the seed transfer process steps. In examples where more than one crystalline seed region is transferred to a handle wafer in a single transfer process from a tiled mother crystal, the precise placement is achieved during fabrication of the tiled mother wafer.
Precision alignment of mother wafers to handle wafers can be achieved through several means. Firstly, die bonding tools can be used. Modern die bonding tools allow for micron and sub-micron scale alignment tolerances and can exert force and elevated temperatures on mother wafers and handles in the ranges needed for thermocompression bonding. After initial placement of the mother wafers, secondary bonding steps can be carried out using less precise but higher force and temperature bonding tools. For lower precision placements, precision jigs or alignment plates can be manufactured using precision machining or 3D printing. These jigs would overlay the handle wafer and constrain where mother wafers could be placed.
In another example, shown in
In some examples, precise alignment is achieved by use of die bonding equipment to align mother wafers to either the engineered wafer handle or the tiled mother handle wafer. In some examples, jigs are used to locate mother wafers prior to bonding to either the engineered wafer or tiled mother wafer handles. In some examples, lithographically defined alignment marks are produced on both the mother wafers and the handle wafers to precise alignments, especially when used in combination with die attach tools, with tolerances of a few microns to ≤100 microns. In some examples, a self-aligned process is achieved by patterning the bonding layer on the handle wafer such that it has self-aligning features. Such self-alignment features include forming pockets in the bond layer of the handle wafer that are similar in size and shape to the mother crystals.
In an example, the epitaxial seed regions of a tiled engineered substrate are configured in a pre-determined arrangement where the relative positions of the edges of the epitaxial seed regions as well as the gaps between epitaxial seed regions are controlled and known to within a tolerance of ≤1 mm.
In some examples, the mother crystals are reclaimed and reused to transfer additional seed layers to additional handle wafers. The purpose of the reclaim is to planarize the mother wafer surface, restore the surface roughness to its original value, remove any non-transferred sections of the previously transferred seed layer, and remove any sub-surface defects that may have been introduced by the seed-layer transfer process. This reclaim may include lapping, polishing and chemical-mechanical polishing (CMP) of the mother crystal to restore the transferred surface to an atomically flat, low-defect state. This reclaim may include wet chemical cleans including solvent cleans with ethanol, methanol, isopropanol, acetone, and chlorinated solvents among others. This reclaim may include wet chemical etches such as with HCl, nitric acid, aqua regia, combinations of sulfuric acid and hydrogen peroxide, HF acid, buffered oxide etchant, and the like to remove foreign elements from the transferable surface of the mother crystal. In some examples, the reclaim may involve removal of tiled mother crystals from their handle and remounting to the same or a different handle.
Crystal polishing, such as with very small diameter diamond or SiC slurries can result in optically flat surfaces and can be effective as an intermediate step in reducing the roughness of a mother crystal surface after transfer of a seed layer. However, the relatively large particle size on the order of 1 micron or larger and hardness of these media can often result in the creation of scratches on the surface of a crystal and extended defects buried beneath the surface of a crystal. Should such a treated mother crystal be used for a second transfer of a seed layer, the scratches and sub-surface defects may be transferred along with the seed layer and be replicated in epitaxial films grown on the seed layer. To ensure scratches and sub-surface damage is minimized during reclaim of mother crystals, CMP should be used. CMP utilizes a chemical process, such as oxidation, to alter the composition of a crystal surface such that it is more easily removed by a soft polishing media. In an example, the CMP GaN mother crystals are reclaimed using a CMP media comprised by colloidal silica with particle diameters less than or equal to 500 nm and preferably less than 100 nm. The CMP slurry has a pH value higher than 7, with optimal value typically being above 8 and optimal value depending on the orientation and composition of the crystal surface undergoing CMP. CMP polished reclaimed surfaces should have a root-mean-square [RMS] roughness less than 10 nm and preferably less than 2 nm. CMP polished reclaimed surfaces should in 100 nm of crystal thickness nearest the surface have ≤5E6 cm−3 dislocations introduced by the polishing process.
In an example, semi-polar GaN mother crystals are crosscut from a [0001] oriented GaN boule grown by hydride vapor phase epitaxy [HVPE] or ammonothermal growth. The resulting mother crystals have a first and second surfaces, which are substantially parallel to each other, and which are oriented as a non-polar or semi-polar plane among others. The first and second surfaces have lateral dimensions of <20 mm in a first in-plane direction parallel to the projection of the [0001] direction onto the surface of the mother crystal and <50 mm in a second in-plane direction perpendicular to the first direction. Preferably the tolerance of the lateral dimensions of the mother crystals is plus or minus ≤100 microns. More preferably, the tolerance is plus or minus ≤50 microns. Most preferably the tolerance is plus or minus ≤20 microns. The mother crystals have a thickness corresponding to the distance between the first and second surfaces of between 0.1 mm and 10 mm. The first and second surfaces are lapped and polished to achieve surface roughness of less than 50 nm RMS. The first surface of each mother crystal is patterned lithographically, and Au is deposited on the crystal surface in a pattern that includes alignment and fiducial marks. A handle wafer is provided which is also configured with a lithographically patterned Au bonding layer that includes corresponding alignment marks. A die attach tool capable of aligning the mother crystals and handle wafer using the lithographically defined marks is used to place the mother crystals on the handle wafer with alignment tolerances of ≤10 microns. The alignment marks on the handle wafer are positioned such that the mother crystals do not overlap when placed on the handle wafer and the nominal size of the gap between mother crystals is minimized. Preferably, the gap between mother crystals when mounted on the handle wafer is ≤500 microns. More preferably the gap between mother crystals when mounted on the handle wafer is ≤100 microns. Most preferably the gap between mother crystals when mounted on the handle wafer is ≤50 microns. When the mother crystals are placed on the handle wafer, they are fixed in place by bonding of the Au bond-layers using thermocompression by application of force at elevated temperatures above 100 degrees C. Once all the mother crystals are mounted to the handle wafer a secondary bonding step can be performed using higher temperatures and pressures than achievable in the die attach tool to improve the strength of the Au—Au bond interface. The tiled mother crystal can then be chemical-mechanically polished (CMP) to ensure the second surface of the individual mother wafers are at the same height above the handle wafer. Silicon oxide is deposited on a second handle wafer and on the second surface of the tiled mother crystals a bulk GaN substrate. CMP is used to reduce the roughness of the oxide layers on both the handle and the tiled mother crystal to below 10 nm RMS such that the oxide surfaces will be in intimate contact when brought together. This is followed by an ion implantation of atoms of one or more elements, such as hydrogen and helium among others, at a shallow depth of less than 500 nm into the second surface of the tiled mother wafer such that the implanted species are concentrated at a narrow range of depth in the GaN crystals. The oxide surfaces of the handle and tiled mother crystal are then brought into contact and bonded at elevated temperature and pressure to produce a stable bond. A second anneal at elevated temperature induces the implanted species to come out of solid solution in the GaN wafer, thereby initiating microfracture of the GaN at the depth of the implantation. This microfracture releases the layer of GaN above the implantation depth and results in the formation of an engineered substrate consisting of the released layer of GaN overlaying the bonding oxide, which overlays the handle wafer. In this example, the GaN layer acts as the seed layer for growth of epitaxial device layers, while the bonding oxide may be used as a selective release layer, which can be selectively etched using solutions containing HF acid.
The gallium and nitrogen containing laser diode devices can be fabricated on engineered substrates configured with a conventional orientation of a gallium and nitrogen containing film (e.g., GaN) such as the polar c-plane. In an example, the gallium and nitrogen containing laser diode devices are fabricated on engineered substrates configured with a specialized orientation of a gallium and nitrogen containing film (e.g., GaN) such as the nonpolar m-plane or a-plane orientation, or on a semipolar orientation such as the {30-31}, {20-21}, {30-32}, {11-22}, {10-11}, {30-3-1}, {20-2-1}, {30-3-2}, or offcuts of any of these polar, nonpolar, and semipolar planes within +/−10 degrees towards a c-plane, and/or +/−10 degrees towards an a-plane, and/or +/−10 degrees towards an m-plane. A gallium and nitrogen containing laser diode epitaxial layer structure is grown on a gallium and nitrogen containing engineered substrate with multiple tile regions. The substrate member may have a surface region on the polar {0001} plane (c-plane), nonpolar plane (m-plane, a-plane), and semipolar plain ({11-22}, {10-1-1}, {20-21}, {30-31}) or other planes of a gallium and nitrogen containing substrate. The laser device can be configured to emit a laser beam characterized by one or more wavelengths from about 390 nm to about 540 nm.
An example of an epitaxial structure for a laser diode device according to this invention is shown in epitaxial structure 1101 of
Another example of an epitaxial structure for a laser diode device according to this invention is shown in epitaxial structure 1102 of
The exemplary laser diode devices in
In a specific example, the method of facet formation includes subjecting the substrates to a laser for pattern formation. In an example, the pattern is configured for the formation of a pair of facets for a ridge laser. In an example, the pair of facets face each other and are in parallel alignment with each other. In an example, the method uses a UV (355 nm) laser to scribe the laser bars. In a specific example, the laser is configured on a system, which allows for accurate scribe lines configured in a different patterns and profiles. In an example, the laser scribing can be performed on the back-side, front-side, or both depending upon the application. Of course, there can be other variations, modifications, and alternatives.
It is well known that etch techniques such as chemical assisted ion beam etching (CAIBE), inductively coupled plasma (ICP) etching, or reactive ion etching (RIE) can result in smooth and vertical etched sidewall regions, which could serve as facets in etched facet laser diodes. In the etched facet process a masking layer is deposited and patterned on the surface of the wafer. The etch mask layer could be comprised of dielectrics such as silicon dioxide (SiO2), silicon nitride (SixNy), a combination thereof or other dielectric materials. Further, the mask layer could be comprised of metal layers such as Ni or Cr, but could be comprised of metal combination stacks or stacks comprising metal and dielectrics. In another approach, photoresist masks can be used either alone or in combination with dielectrics and/or metals. The etch mask layer is patterned using conventional photolithography and etch steps. The alignment lithography could be performed with a contact aligner or stepper aligner. Such lithographically defined mirrors provide a high level of control to the design engineer. After patterning of the photoresist mask on top of the etch mask is complete, the patterns in then transferred to the etch mask using a wet etch or dry etch technique. Finally, the facet pattern is then etched into the wafer using a dry etching technique selected from CAIBE, ICP, RIE and/or other techniques. The etched facet surfaces must be highly vertical of between about 87 and about 93 degrees or between about 89 and about 91 degrees from the surface plane of the wafer. The etched facet surface region must be very smooth with root mean square roughness values of less than about 50 nm, 20 nm, 5 nm, or 1 nm. Lastly, the etched must be substantially free from damage, which could act as nonradiative recombination centers and hence reduce the catastrophic optical mirror damage (COMD) threshold. CAIBE is known to provide very smooth and low damage sidewalls due to the chemical nature of the etch, while it can provide highly vertical etches due to the ability to tilt the wafer stage to compensate for any inherent angle in etch.
The laser stripe 110, referring to
The laser stripe 110, referring to
In a specific example, the laser device may emit red light with a center wavelength between 600 nm and 750 nm. Such a device may comprise layers of varying compositions of AlxInyGa1-x-yAszP1-z, where x+y≤1 and z≤1. The red laser device comprises at least an n-type and p-type cladding layer, an n-type SCH of higher refractive index than the n-type cladding, a p-type SCH of higher refractive index than the p-type cladding and an active region where light is emitted. In a specific example, the laser stripe is provided by an etching process selected from dry etching or wet etching. In a preferred example, the etching process is dry, but can be others. The device also has an overlying dielectric region, which exposes the contact region. In a specific example, the dielectric region is an oxide such as silicon dioxide but can be others. Of course, there can be other variations, modifications, and alternatives. The laser stripe is characterized by a length and width. The length ranges from about 50 μm to about 3000 μm, but is preferably between 10 μm and 400 μm, between about 400 μm and 800 μm, or about 800 μm and 1600 μm, but could be others such as greater than 1600 μm. The stripe also has a width ranging from about 0.5 μm to about 80 μm, but is preferably between 0.8 μm and 2.5 μm for single lateral mode operation or between 2.5 μm and 60 μm for multi-lateral mode operation, but can be other dimensions. The laser strip region has a first end and a second end having a pair of cleaved or etched mirror structures, which face each other. The first facet comprises a reflective coating and the second facet comprises no coating, an antireflective coating, or exposes gallium and nitrogen containing material. The first facet is substantially parallel with the second cleaved or etched facet.
Given the high gallium and nitrogen containing substrate costs, difficulty in scaling up gallium and nitrogen containing substrate size, the inefficiencies inherent in the processing of small wafers, and potential supply limitations it becomes extremely desirable to maximize utilization of available gallium and nitrogen containing substrate and overlying epitaxial material. In the fabrication of lateral cavity laser diodes, it is typically the case that minimum die size is determined by device components such as the wire bonding pads or mechanical handling considerations, rather than by laser cavity widths. Minimizing die size is critical to reducing manufacturing costs as smaller die sizes allow a greater number of devices to be fabricated on a single wafer in a single processing run. The current invention is a method of maximizing the number of devices which can be fabricated from a given gallium and nitrogen containing substrate and overlying epitaxial material by spreading out the epitaxial material onto a carrier wafer via a die expansion process.
Similar to an edge emitting laser diode, a SLED is typically configured as an edge-emitting device wherein the high brightness, highly directional optical emission exits a waveguide directed outward from the side of the semiconductor chip. SLEDs are designed to have high single pass gain or amplification for the spontaneous emission generated along the waveguide. However, unlike laser diodes, they are designed to provide insufficient feedback to in the cavity to achieve the lasing condition where the gain equals the total losses in the waveguide cavity. In a typical example, at least one of the waveguide ends or facets is designed to provide very low reflectivity back into the waveguide. Several methods can be used to achieve reduced reflectivity on the waveguide end or facet. In one approach an optical coating is applied to at least one of the facets, wherein the optical coating is designed for low reflectivity such as less than 1%, less than 0.1%, less than 0.001%, or less than 0.0001% reflectivity. In another approach for reduced reflectivity the waveguide ends are designed to be tilted or angled with respect to the direction of light propagation such that the light that is reflected back into the chip does not constructively interfere with the light in the cavity to provide feedback. The tilt angle must be carefully designed around a null in the reflectivity versus angle relationship for optimum performance. The tilted or angled facet approach can be achieved in a number of ways including providing an etched facet that is designed with an optimized angle lateral angle with respect to the direction of light propagation. The angle of the tilt is pre-determined by the lithographically defined etched facet patter. Alternatively, the angled output could be achieved by curving and/or angling the waveguide with respect to a cleaved facet that forms on a pre-determined crystallographic plane in the semiconductor chip. Another approach to reduce the reflectivity is to provide a roughened or patterned surface on the facet to reduce the feedback to the cavity. The roughening could be achieved using chemical etching and/or a dry etching, or with an alternative technique. Of course, there may be other methods for reduced feedback to the cavity to form a SLED device. In many examples, a number of techniques can be used in combination to reduce the facet reflectivity including using low reflectivity coatings in combination with angled or tilted output facets with respect to the light propagation.
In a specific example, the present invention provides an alternative device structure capable of emitting 501 nm and greater light in a ridge laser example. The device is provided with a of the following epitaxially grown elements:
A gallium and nitrogen containing laser diode laser device may also include other structures, such as a surface ridge architecture, a buried heterostructure architecture, and/or a plurality of metal electrodes for selectively exciting the active region. For example, the active region may comprise first and second gallium and nitrogen containing cladding layers and an indium and gallium containing emitting layer positioned between the first and second cladding layers. A laser device may further include an n-type gallium and nitrogen containing material and an n-type cladding material overlying the n-type gallium and nitrogen containing material. In a specific example, the device also has an overlying n-type gallium nitride layer, an active region, and an overlying p-type gallium nitride layer structured as a laser stripe region. Additionally, the device may also include an n-side separate confinement hetereostructure (SCH), p-side guiding layer or SCH, p-AlGaN EBL, among other features. In a specific example, the device also has a p++ type gallium nitride material to form a contact region. In a specific example, the p++ type contact region has a suitable thickness and may range from about 10 nm 50 nm, or other thicknesses. In a specific example, the doping level can be higher than the p-type cladding region and/or bulk region. In a specific example, the p++ type region has doping concentration ranging from about 1019 to 1021 Mg/cm3, and others. The p++ type region preferably causes tunneling between the semiconductor region and overlying metal contact region. In a specific example, each of these regions is formed using at least an epitaxial deposition technique of metal organic chemical vapor deposition (MOCVD), molecular beam epitaxy (MBE), or other epitaxial growth techniques suitable for GaN growth. In a specific example, the epitaxial layer is a high-quality epitaxial layer overlying the n-type gallium nitride layer. In some examples, the high-quality layer is doped, for example, with Si or O to form n-type material, with a dopant concentration between about 1016 cm−3 and 1020 cm−3.
An n-type AluInvGa1-u-vN layer, where 0≤u, v, u+v≤1, is deposited on the substrate. The carrier concentration may lie in the range between about 1016 cm−3 and 1020 cm−3. The deposition may be performed using metalorganic chemical vapor deposition (MOCVD) or molecular beam epitaxy (MBE).
For example, the engineered GaN substrate is placed on a susceptor in an MOCVD reactor. After closing, evacuating, and back-filling the reactor (or using a load lock configuration) to atmospheric pressure, the susceptor is heated to a temperature between about 1000 and about 1200 degrees Celsius in the presence of a nitrogen-containing gas. The susceptor is heated to approximately 90° to 1200 degrees Celsius under flowing ammonia. A flow of a gallium-containing metalorganic precursor, such as trimethylgallium (TMG) or triethylgallium (TEG) is initiated, in a carrier gas, at a total rate between approximately 1 and 50 standard cubic centimeters per minute (sccm). The carrier gas may comprise hydrogen, helium, nitrogen, or argon. The ratio of the flow rate of the group V precursor (ammonia) to that of the group III precursor (trimethylgallium, triethylgallium, trimethylindium, trimethylaluminum) during growth is between about 2000 and about 12000. A flow of disilane in a carrier gas, with a total flow rate of between about 0.1 sccm and 10 sccm, is initiated.
In one example, the laser stripe region 1321 is a p-type gallium nitride layer. The laser stripe is provided by a dry etching process, but wet etching can be used. The dry etching process is an inductively coupled process using chlorine bearing species or a reactive ion etching process using similar chemistries. The chlorine bearing species are commonly derived from chlorine gas or the like. The device also has an overlying dielectric region, which exposes a contact region 1322. The dielectric region is an oxide such as silicon dioxide or silicon nitride, and a contact region is coupled to an overlying metal layer 1320. The overlying metal layer is preferably a multilayered structure containing gold and platinum (Pt/Au), palladium and gold (Pd/Au), or nickel gold (Ni/Au), or a combination thereof. In some examples, barrier layers and more complex metal stacks are included.
Active region 1318 preferably includes one to ten quantum well regions or a double heterostructure region for light emission. Following deposition of the n-type AluInvGa1-u-vN layer to achieve a desired thickness, an active layer is deposited. The quantum wells are preferably InGaN with GaN, AlGaN, InAlGaN, or InGaN barrier layers separating them. In other examples, the well layers and barrier layers comprise AlwInxGa1-w-xN and AlyInzGa1-y-zN, respectively, where 0≤w, x, y, z, w+x, y+z≤1, where w<u, y and/or x>v, z so that the bandgap of the well layer(s) is less than that of the barrier layer(s) and the n-type layer. The well layers and barrier layers each have a thickness between about 1 nm and about 20 nm. The composition and structure of the active layer are chosen to provide light emission at a preselected wavelength. The active layer may be left undoped (or unintentionally doped) or may be doped n-type or p-type.
The active region can also include an electron blocking region, and a separate confinement heterostructure. The electron-blocking layer may comprise AlsIntGa1-s-tN, where 0≤s, t, s+t≤1, with a higher bandgap than the active layer, and may be doped p-type. In one specific example, the electron blocking layer includes AlGaN. In another example, the electron blocking layer includes an AlGaN/GaN super-lattice structure, comprising alternating layers of AlGaN and GaN, each with a thickness between about 0.2 nm and about 5 nm.
As noted, the p-type gallium nitride or aluminum gallium nitride structure is deposited above the electron blocking layer and active layer(s). The p-type layer may be doped with Mg, to a level between about 1016 cm−3 and 1022 cm−3, with a thickness between about 5 nm and about 1000 nm. The outermost 1-50 nm of the p-type layer may be doped more heavily than the rest of the layer, to enable an improved electrical contact. The device also has an overlying dielectric region, for example, silicon dioxide, which exposes the contact region 1322.
The metal contact is made of suitable material such as silver, gold, aluminum, nickel, platinum, rhodium, palladium, chromium, or the like. The contact may be deposited by thermal evaporation, electron beam evaporation, electroplating, sputtering, or another suitable technique. In an example, the electrical contact serves as a p-type electrode for the optical device. In another example, the electrical contact serves as an n-type electrode for the optical device. The laser devices illustrated in
In various examples, the present invention realizes high output power from a diode laser is by widening a portion of the laser cavity member from the single lateral mode regime of 1.0-3.0 μm to the multi-lateral mode range 5.0-20 μm. In some cases, laser diodes having cavities at a width of 50 μm or greater are employed.
The laser stripe length, or cavity length ranges from 100 to 3000 μm and employs growth and fabrication techniques such as those described in U.S. patent application Ser. No. 12/759,273, filed Apr. 13, 2010, which is incorporated by reference herein. As an example, laser diodes are fabricated on nonpolar or semipolar gallium containing substrates, where the internal electric fields are substantially eliminated or mitigated relative to polar c-plane oriented devices. It is to be appreciated that reduction in internal fields often enables more efficient radiative recombination. Further, the heavy hole mass is expected to be lighter on nonpolar and semipolar substrates, such that better gain properties from the lasers can be achieved.
Optionally,
Optionally,
In multiple examples, the device layers comprise a super-luminescent light emitting diode or SLED. In all applicable examples, a SLED device can be interchanged with or combined with laser diode devices according to the methods and architectures described in this invention. A SLED is in many ways similar to an edge emitting laser diode; however, the emitting facet of the device is designed so as to have a very low reflectivity. A SLED is similar to a laser diode as it is based on an electrically driven junction that when injected with current becomes optically active and generates amplified spontaneous emission (ASE) and gain over a wide range of wavelengths.
In an example, the LD or SLED device is characterized by a ridge with non-uniform width. The ridge is comprised by a first section of uniform width and a second section of varying width. The first section has a length between 100 and 500 μm long, though it may be longer. The first section has a width of between 1 and 2.5 μm, with a width preferably between 1 and 1.5 μm. The second section of the ridge has a first end and a second end. The first end connects with the first section of the ridge and has the same width as the first section of the ridge. The second end of the second section of the ridge is wider than the first section of the ridge, with a width between 5 and 50 μm and more preferably with a width between 15 and 35 μm. The second section of the ridge waveguide varies in width between its first and second end smoothly. In some examples, the second derivative of the ridge width versus length is zero such that the taper of the ridge is linear. In some examples, the second derivative is chosen to be positive or negative. In general, the rate of width increase is chosen such that the ridge does not expand in width significantly faster than the optical mode. In specific examples, the electrically injected area is patterned such that only a part of the tapered portion of the waveguide is electrically injected.
In an example, multiple laser dice emitting at different wavelengths are transferred to the same carrier wafer in close proximity to one another; preferably within one millimeter of each other, more preferably within about 200 micrometers of each other and most preferably within about 50 μm of each other. The laser die wavelengths are chosen to be separated in wavelength by at least twice the full width at half maximum of their spectra. For example, three dice, emitting at 440 nm, 450 nm and 460 nm, respectively, are transferred to a single carrier chip with a separation between die of less than 50 μm and die widths of less than 50 μm such that the total lateral separation, center to center, of the laser light emitted by the die is less than 200 μm. The closeness of the laser die allows for their emission to be easily coupled into the same optical train or fiber optic waveguide or projected in the far field into overlapping spots. In a sense, the lasers can be operated effectively as a single laser light source.
After the laser diode chip fabrication as described above, the laser diode can be mounted to a submount. In some examples the submount is comprised of AlN, SiC, BeO, diamond, or other materials such as metals, ceramics, or composites. The submount member may be characterized by a width, length, and thickness.
A schematic diagram illustrating a CoS based laser diode formed on an engineered gallium and nitrogen containing substrate technology is shown in
In an example, the engineered substrate will contain one or more electrically insulating layers that prevent or limit electrical conduction through the engineered substrate. In this example, multiple sets of wire bonds 1404 are used to couple the electrical power from the electrodes 1403 and 1405 on the submount to electrode pads on the laser diode chip to generate a laser beam output 1406 from the laser diode.
In another example, the gallium and nitrogen containing laser diode fabrication includes an epitaxial release step from the engineered substrate to lift off the epitaxially grown gallium and nitrogen layers and prepare them for transferring to a carrier wafer which could comprise the submount after laser fabrication. The transfer step requires precise placement of the epitaxial layers on the carrier wafer to enable subsequent processing of the epitaxial layers into laser diode devices. The attachment process to the carrier wafer could include a wafer bonding step with a bond interface comprised of metal-metal, semiconductor-semiconductor, glass-glass, dielectric-dielectric, or a combination thereof.
In this example, gallium and nitrogen containing epitaxial layers are grown on an engineered substrate with a gallium and nitrogen containing seed layer. The epitaxial layer stack comprises at least the laser diode device layers overlying the seed layer. Following the growth of the epitaxial layers on the engineered gallium and nitrogen containing substrate, the semiconductor device layers are separated from the engineered substrate by a selective wet etching process such as by etching of one or more dielectric layers in engineered substrate with conventional acids such as BHF or by a PEC etch configured to selectively remove an epitaxially grown selective release layer and enable release of the device layers to a carrier wafer. In one example, a bonding material is deposited on the surface overlying the semiconductor device layers. A bonding material is also deposited either as a blanket coating or patterned on the carrier wafer. Standard lithographic processes are used to selectively mask the semiconductor device layers. The wafer is then subjected to an etch process such as dry etch or wet etch processes to define via structures that expose the selective release layers on the sidewall of the mesa structure. As used herein, the term mesa region or mesa is used to describe the patterned epitaxial material on the gallium and nitrogen containing engineered substrate and prepared for transferring to the carrier wafer. The mesa region can be any shape or form including a rectangular shape, a square shape, a triangular shape, a circular shape, an elliptical shape, a polyhedron shape, or other shape. The term mesa shall not limit the scope of the present invention.
Following the definition of the mesa, a selective etch process is performed to fully or partially remove the selective release layers while leaving the semiconductor device layers intact. The resulting structure comprises undercut mesas comprised of epitaxial device layers. The undercut mesas correspond to dice from which semiconductor devices will be formed on. In some examples, a protective passivation layer can be employed on the sidewall of the mesa regions to prevent the device layers from being exposed to the selective etch when the etch selectivity is not perfect. In other examples, a protective passivation is not needed because the device layers are not sensitive to the selective etch or measures are taken to prevent etching of sensitive layers such as shorting the anode and cathode. The undercut mesas corresponding to device dice are then transferred to the carrier wafer using a bonding technique wherein the bonding material overlying the semiconductor device layers is joined with the bonding material on the carrier wafer. The resulting structure is a carrier wafer comprising gallium and nitrogen containing epitaxial device layers overlying the bonding region.
The preparation of the epitaxy wafer is shown in
In some examples, the engineered substrates contain either oxide bond layers or oxide selective release layers. These layers are selectively etched using either wet or vapor etching with acidic solutions such as hydrofluoric acid or buffered oxide etchant, or they are selectively dry etched using fluorine containing plasmas.
In some examples, PEC etching is deployed as the selective etch to remove selective release layers included in the epitaxial device structure. PEC is a photo-assisted wet etch technique that can be used to etch GaN and its alloys.
Epitaxial sacrificial layers for lift-off of the engineered substrate via photochemical etching would incorporate at a minimum a low-bandgap or doped layer that would absorb the pump light and have enhanced etch rate relative to the surrounding material. The epitaxial selective release layers 1707 can be deposited epitaxially and the alloy composition and doping of these can be selected such that hole carrier lifetime and diffusion lengths are high. Defects that reduce hole carrier lifetimes and diffusion length must can be avoided by growing the selective release layers under growth conditions that promote high material crystalline quality. An example of an epitaxial selective release layer would be InGaN layers that absorb at the wavelength of an external light source. An etch stop layer designed with very low etch rate to control the thickness of the adjacent material remaining after engineered substrate removal can also be incorporated to allow better control of the etch process. The etch properties of the etch stop layer can be controlled solely by or a combination of alloy composition and doping. A potential etch stop layer would an AlGaN or GaN layer with a bandgap higher than the external light source. Another potential etch stop layer is a highly doped n-type AlGaN or GaN layer with reduce minority carrier diffusion lengths and lifetime thereby dramatically reducing the etch rate of the etch stop material.
In some examples, PEC etching is achieved without the use of an active region protecting layer by electrically shorting the p-side of the laser diode pn-junction to the n-side. Etching in the PEC process is achieved by the dissolution of AlInGaN materials at the wafer surface when holes are transferred to the etching solution. These holes are then recombined in the solution with electrons extracted at the cathode metal interface with the etching solution. Charge neutrality is therefore achieved. Selective etching is achieved by electrically shorting the anode to the cathode. Referring to
The relative etch rates of the epitaxial sacrificial and active regions are determined by a number of factors, but primarily it is determined by the density of holes found in the active region at steady state. If the metal interconnects or anchors are very resistive, or if either the cathode or anode electrical contacts to the p-type and n-type, respectively, cladding regions are too resistive or have large Schottky barriers then it is possible for carriers to accumulate on either side of the p-n junction. These carriers will produce an electric field that acts against the field in the depletion region and will reduce the magnitude of the field in the depletion region until the rate of photo-generated carrier drift out of the active region is balanced by the recombination rate of carriers via the metal layers shorting the cathode and anode. Some recombination will take place via photochemical etching, and since it scales with the density of holes in the active region it is preferable to prevent the buildup of a photo-induced bias across the active region.
In one example, thermocompression bonding is used to transfer the gallium and nitrogen epitaxial semiconductor layers to the carrier wafer. In this example, thermocompression bonding involves bonding of the epitaxial semiconductor layers to the carrier wafer at elevated temperatures and pressures using a bonding media disposed between the epitaxial layers and handle wafer. The bonding media may be comprised of a number of different layers, but typically contain at least one layer (the bonding layer) that is composed of a relatively ductile material with a high surface diffusion rate. In many cases this material is comprised of Au, Al or Cu. The bonding media may also include layers disposed between the bonding layer and the epitaxial materials or handle wafer that promote adhesion. For example, an Au bonding layer on a Si wafer may result in diffusion of Si to the bonding interface, which would reduce the bonding strength. Inclusion of a diffusion barrier such as silicon oxide or nitride would limit this effect. Relatively thin layers of a second material may be applied on the top surface of the bonding layer to promote adhesion between the bonding layers disposed on the epitaxial material and handle. Some bonding layer materials of lower ductility than gold (e.g. Al, Cu etc.) or which are deposited in a way that results in a rough film (for example electrolytic deposition) may require planarization or reduction in roughness via chemical or mechanical polishing before bonding, and reactive metals may require special cleaning steps to remove oxides or organic materials that may interfere with bonding.
Thermocompressive bonding can be achieved at relatively low temperatures, typically below 500 C° and above 200 C°. Temperatures should be high enough to promote diffusivity between the bonding layers at the bonding interface, but not so high as to promote unintentional alloying of individual layers in each metal stack. Application of pressure enhances the bond rate and leads to some elastic and plastic deformation of the metal stacks that brings them into better and more uniform contact. Optimal bond temperature, time and pressure will depend on the bond material, the roughness of the surfaces forming the bonding interface and the susceptibility to fracture of the handle wafer or damage to the device layers under load.
The bonding interface need not be composed of the totality of the wafer surface. For example, rather than a blanket deposition of bonding metal, a lithographic process could be used to deposit metal in discontinuous areas separated by regions with no bonding metal. This may be advantageous in instances where defined regions of weak or no bonding aid later processing steps, or where an air gap is needed. One example of this would be in removal of the GaN engineered substrate using wet etching of an epitaxially grown selective release layer. To access the selective release layer, one must etch vias into either of the two surfaces of the epitaxial wafer, and preserving the wafer for re-use is most easily done if the vias are etched from the bonded side of the wafer. Once bonded, the etched vias result in channels that can conduct etching solution from the edges to the center of the bonded wafers, and therefore the areas of the engineered substrate comprising the vias are not in intimate contact with the handle wafer such that a bond would form.
The bonding media can also be an amorphous or glassy material bonded either in a reflow process or anodically. In anodic bonding the media is a glass with high ion content where mass transport of material is facilitated by the application of a large electric field. In reflow bonding the glass has a low melting point and will form contact and a good bond under moderate pressures and temperatures. All glass bonds are relatively brittle, and require the coefficient of thermal expansion of the glass to be sufficiently close to the bonding partner wafers (i.e. the GaN wafer and the handle). Glasses in both cases could be deposited via vapor deposition or with a process involving spin on glass. In both cases the bonding areas could be limited in extent and with geometry defined by lithography or silk-screening process.
Gold-gold metallic bonding is used as an example in this work, although a wide variety of oxide bonds, polymer bonds, wax bonds, etc., are potentially suitable. Submicron alignment tolerances are possible using commercially available die bonding equipment. In another example, the bonding layers can be a variety of bonding pairs including metal-metal, oxide-oxide, soldering alloys, photoresists, polymers, wax, etc. Only epitaxial die which are in contact with a bond bad on the carrier wafer will bond. Sub-micron alignment tolerances are possible on commercially available die or flip chip bonders.
In an example, an oxide is overlaid on an exposed planar n-type or p-type gallium and nitrogen containing material or over an exposed planar n-type or p-type gallium and nitrogen containing material using direct wafer bonding of the surface of the gallium and nitrogen containing material to the surface of a carrier wafer comprised primarily of an oxide or a carrier wafer with oxide layers disposed on them. In both cases the oxide surface on the carrier wafer and the exposed gallium and nitrogen containing material are cleaned to reduce the amount of hydrocarbons, metal ions and other contaminants on the bonding surfaces. The bonding surfaces are then brought into contact and bonded at elevated temperature under applied pressure. In some cases, the surfaces are treated chemically with acids, bases or plasma treatments to produce a surface that yields a weak bond when brought into contact with the oxide surface. For example, the exposed surface of the gallium containing material may be treated to form a thin layer of gallium oxide, which being chemically similar to the oxide bonding surface will bond more readily. Furthermore, the oxide and now gallium oxide terminated surface of the gallium and nitrogen containing material may be treated chemically to encourage the formation of dangling hydroxyl groups (among other chemical species) that will form temporary or weak chemical or van der Waals bonds when the surfaces are brought into contact, which are subsequently made permanent when treated at elevated temperatures and elevated pressures.
In an alternative example, an oxide material is deposited overlying the device layer mesa region to form a bond region. The carrier wafer is also prepared with an oxide layer to form a bond region. The oxide layer overlying the carrier wafer could be patterned or could be a blanket layer. The oxide surface on the carrier wafer and the oxide surface overlying the mesa device layer mesa regions are cleaned to reduce the amount of hydrocarbons, metal ions and other contaminants on the bonding surfaces. The bonding surfaces are then brought into contact and bonded at elevated temperature under applied pressure. In one example, a chemical mechanical polish (CMP) process is used to planarize the oxide surface and make them smooth to improve the resulting bond. In some cases, the surfaces are treated chemically with acids, bases or plasma treatments to produce a surface that yields a weak bond when brought into contact with the oxide surface. Bonding is performed at elevated temperatures and elevated pressures.
In another example, the bonding media could be a dielectric material such as silicon dioxide or silicon nitride. Such a bonding media may be desirable where low conductivity is desired at the bond interface to achieve properties such as reduced device capacitance to enable increased frequency operation. The bond media comprising the bond interface can be comprised of many other materials such as oxide-oxide pair, semiconductor-semiconductor pair, spin-on-glass, soldering alloys, polymers, photoresists, wax, or a combination thereof.
The carrier wafer can be chosen based on any number of criteria including but not limited to cost, thermal conductivity, thermal expansion coefficients, size, electrical conductivity, optical properties, and processing compatibility. The patterned epitaxy wafer, or donor, is prepared in such a way as to allow subsequent selective release of bonded epitaxy regions, here-in referred to as die. The patterned carrier wafer is prepared such that bond pads are arranged in order to enable the selective area bonding process. The bonding material can be a variety of media including but not limited to metals, polymers, waxes, and oxides. These wafers can be prepared by a variety of process flows, some examples of which are described below. In the first selective area bond step, the epitaxy wafer is aligned with the pre-patterned bonding pads on the carrier wafer and a combination of pressure, heat, and/or sonication is used to bond the mesas to the bonding pads.
In some examples, the carrier wafer is another semiconductor material, a metallic material, or a ceramic material. Some potential candidates include silicon, gallium arsenide, sapphire, silicon carbide, diamond, gallium nitride, AlN, polycrystalline AlN, indium phosphide, germanium, quartz, copper, copper tungsten, gold, silver, aluminum, stainless steel, or steel.
In some examples, the carrier wafer is selected based on size and cost. For example, ingle crystal silicon wafers are available in diameters up to 300 mm or 12 inch and are most cost effective. By transferring gallium and nitrogen epitaxial materials from 2″ gallium and nitrogen containing engineered substrates to large silicon substrates of 150 mm, 200 mm, or 300 mm diameter the effective area of the semiconductor device wafer can be increases by factors of up to 36 or greater. This feature of this invention allows for high quality gallium and nitrogen containing semiconductor devices to be fabricated in mass volume leveraging the established infrastructure in silicon foundries.
In some examples, the carrier wafer material is chosen such that it has similar thermal expansion properties to group-III nitrides, high thermal conductivity, and is available as large area wafers compatible with standard semiconductor device fabrication processes. The carrier wafer is then processed with structures enabling it to also act as the submount for the semiconductor devices. Singulation of the carrier wafers into individual die can be accomplished either by sawing, cleaving, or a scribing and breaking process. By combining the functions of the carrier wafer and finished semiconductor device submount the number of components and operations needed to build a packaged device is reduced, thereby lowering the cost of the final semiconductor device significantly.
In an example, the carrier wafer is a solid material with thermal conductivity greater than 100 W/m-K. In an example, the common substrate is preferably a solid material with thermal conductivity greater than 200 W/m-K. In an example, the common substrate is preferably a solid material with thermal conductivity greater than 400 W/m-K. In an example, the common substrate is preferably a solid material with electrical insulator with electrical resistivity greater than 1×106 ohm-cm. In an example, the common substrate is preferably a solid material with thin film material providing electrical 1×106 ohm-cm. In an example, the common substrate selected from one or more of Al2O3, AlN, SiC, BeO and diamond. In an example, the common substrate is preferably comprised of crystalline SiC. In an example, the common substrate is preferably comprised of crystalline SiC with a thin film of Si3N4 deposited onto the top surface. In an example, the common substrate contains metal traces providing electrically conductive connections between the one or more low-cost laser diodes. In an example, the common substrate contains metal traces providing thermally conductive connections between the one or more low-cost laser diodes and the common substrate.
In one example, the bonding of the semiconductor device epitaxial material to the carrier wafer process can be performed prior to the selective etching of the sacrificial region and subsequent release of the gallium and nitrogen containing engineered substrate.
In other examples, the engineered substrate contains a selectively removable selective release layer that is separate from the engineered substrate bond layer. In some examples, the crystalline seed layer is also the selective release layer of the engineered substrate. In some examples, there is no selective release layer in the engineered substrate and an epitaxially grown selective release layer composed of InGaN or GaN is selectively etched using a photoelectrochemical etch process.
In an example, the bonding process is performed after the selective etching of the sacrificial region. This example offers several advantages. One advantage is easier access for the selective etchant to uniformly etch the sacrificial region across the semiconductor wafer comprising a gallium and nitrogen containing engineered substrate such as GaN and bulk gallium and nitrogen containing epitaxial device layers. A second advantage is the ability to perform multiple bond steps. In one example, the “etch-then-bond” process flow can be deployed where the mesas are retained on the engineered substrate by controlling the etch process such that not all parts of the selective release layer are removed. Referring to
Other than typical GaN based laser devices, undercut AlInGaAsP based laser devices can be produced in a manner similar to GaN based laser diodes described in this invention. There are a number of wet etches that etch some AlInGaAsP alloys selectively. In one example, an AlGaAs or AlGaP selective release layer could be grown clad with GaAs etch stop layers. When the composition of AlxGa1-xAs and AlxGa1-xP is high (x>0.5) AlGaAs can be etched with almost complete selectivity (i.e., etch rate of AlGaAs>106 times that of GaAs) when etched with HF. InGaP and AlInP with high InP and AlP compositions can be etched with HCl selectively relative to GaAs. GaAs can be etched selectively relative to AlGaAs using C6H8O7:H2O2:H2O. There are a number of other combinations of selective release layer, etch-stop layer and etch chemistry which are widely known to those knowledgeable in the art of micromachining AlInGaAsP alloys.
In an example, the AlInGaAsP device layers are exposed to the etch solution which is chosen along with the selective release layer composition such that only the selective release layers experience significant etching. The active region can be prevented from etching during the compositionally selective etch using an etch resistant protective layer, such as like silicon dioxide, silicon nitride, metals or photoresist among others, on the sidewall. This step is followed by the deposition of a protective insulating layer on the mesa sidewalls, which serves to block etching of the active region during the later sacrificial region undercut etching step. A second top down etch is then performed to expose the selective release layers and bonding metal is deposited. With the sacrificial region exposed a compositionally selective etch is used to undercut the mesas. At this point, the selective area bonding process is used to continue fabricating devices. The device layers should be separated from the selective release layers by a layer of material that is resistant to etching. This is to prevent etching into the device layers after partially removing the selective release layers.
A challenge of the etch-then-bond example is mechanically supporting the undercut epitaxial device layer mesa region from spatially shifting prior to the bonding step. If the mesas shift the ability to accurately align and arrange them to the carrier wafer will be compromised, and hence the ability to manufacture with acceptable yields. Mechanically fixing the mesa regions in place prior to bonding can be achieved in several ways. In an example, anchor regions are used to mechanically support the mesas to the gallium and nitrogen containing engineered substrate prior to the bonding step wherein they are releases from the gallium and nitrogen containing engineered substrate and transferred to the carrier wafer.
Anchor regions are special features that can be designed into the photo masks which attach the undercut device layers to the engineered substrate and which prevent fully under-cut regions of the device layers from prematurely detaching from the engineered substrate during the etch and transfer processes. In some examples, the anchors are formed as part of the mesa region, but which are designed to be too wide to themselves be undercut, or which due to the design of the mask contain regions where the selective release layers are not removed. In other examples, these features may be composed of metals or dielectrics that are resistant to the etch and which overlay part or all of each mesa and connect it to the engineered substrate. These features act as anchors, preventing the undercut device layers from detaching from the engineered substrate and prevent the device layers from spatially shifting. This anchor attachment to the engineered substrate can also be achieved by incompletely removing the selective release layer, such that there is a tenuous connection between the undercut device layers and the engineered substrate which can be broken during bonding. The surfaces of the bonding material on the carrier wafer and the device wafer are then brought into contact and a bond is formed which is stronger than the attachment of the undercut device layers to the material in the anchor regions of the selective release layers. After bonding, the separation of the carrier wafer and epitaxial device wafer breaks the anchors and results in transfer of the device layers to the carrier wafer.
In one example, the anchor region is formed by features that are wider than the device layer mesas such that the sacrificial region in these anchor regions is not fully removed during the undercut of the device layers. In one example the mesas are retained on the engineered substrate by deposition of an etch-resistant material acting as an anchor by connecting the mesas to the engineered substrate. In this example an engineered substrate wafer is overlaid by a buffer layer, a selective release layer and a collection of device layers. The bond layer is deposited along with a cathode metal that will be used to facilitate the photoelectrochemical etch process for selectively removing the selective release layer. A layer of etch resistant material, which may be composed of metal, ceramic, polymer or a glass, is deposited such that it connects to both the mesa and the engineered substrate. The selective etch process is carried out such that the selective release layer is fully removed and only the etch-resistant layer connects the mesa to the engineered substrate.
In another example of anchor techniques, the mesas are retained on the engineered substrate by use of an anchor composed of epitaxial material. In this example an engineered substrate wafer is overlaid by a buffer layer, a selective release layer and a collection of device layers. The bond layer is deposited along with a cathode metal that will be used to facilitate the photoelectrochemical etch process for selectively removing the selective release layer. The anchor is shaped such that during the etch, a small portion of the selective release layer remains unetched and creates a connection between the undercut mesa and the engineered substrate wafer. In some examples, of this example a selective release layer in the engineered substrate is used to undercut the mesas. For example, an oxide bond layer in the engineered substrate could be etched selectively with a solution containing hydrofluoric acid.
In one example, the anchors are positioned either at the ends or sides of the undercut die such that they are connected by a narrow, undercut region of material. In this example the narrow connecting material is far from the bond metal and is design such that the undercut material cleaves at the connecting material rather than across the die. This has the advantage of keeping the entire width of the die undamaged, which would be advantageous. In another example, geometric features are added to the connecting material to act as stress concentrators and the bond metal is extended onto the narrow connecting material. The bond metal reinforces the bulk of the connecting material. Adding these features increases the control over where the connection will cleave. These features can be triangles, circles, rectangles, or any deviation that provides a narrowing of the connecting material or a concave profile to the edge of the connecting material.
In another example, the anchors are of small enough lateral extent that they may be undercut, however a protective coating is used to prevent etch solution from accessing the selective release layers in the anchors. This example is advantageous in cases when the width of the die to be transferred is large. Unprotected anchors would need to be larger to prevent complete undercutting, which would reduce the density of die and reduce the utilization efficiency of epitaxial material.
In another example, the anchors are located at the ends of the die and the anchors form a continuous strip of material that connects to all or a plurality of die. This configuration is advantageous since the anchors can be patterned into the material near the edge of wafers or lithographic masks where material utilization is otherwise poor. This allows for utilization of device material at the center of the pattern to remain high even when die sizes become large.
In an example, the anchors are formed by depositing separate regions of an etch-resistant material that adheres well to the epitaxial and engineered substrate material. These regions overlay a portion of the semiconductor device layer mesa and some portion of the structure, such as the engineered substrate, that will not be undercut during the etch. These regions form a continuous connection, such that after the semiconductor device layer mesa is completely undercut, they provide a mechanical support preventing the semiconductor device layer mesa from detaching from the engineered substrate. Metal layers are then deposited on the top of semiconductor device layer mesa, the sidewall of the semiconductor device layer mesa and the bottom of the etched region surrounding the mesa such that a continuous connection is formed. As an example, the metal layers could comprise about 20 nm of titanium to provide good adhesion and be capped with about 500 nm of gold, but of course the choice of metal and the thicknesses could be others. In an example, the length of the semiconductor device die sidewall coated in metal is about 1 nm to about 40 nm, with the upper thickness being less than the width of the semiconductor device die such that the selective release layer is etched completely in the region near the metal anchor where access to the selective release layer by etchant will be limited.
The mesa regions can be formed by dry or wet chemical etching, and in one example would include at least a p++ GaN contact layer, a p-type cladding layer comprised of GaN, AlGaN, or InAlGaN, light emitting layers such as quantum wells separated by barriers, waveguiding layers such as InGaN layers, and the a n-type cladding layers comprised of GaN, AlGaN, or InAlGaN, the selective release layer (504), and a portion of the n-type GaN epitaxial layer beneath the selective release layer. A p-contact metal is first deposited on the p++ GaN contact layer in order to form a high-quality electrical contact with the p-type cladding. A second metal stack is then patterned and deposited on the mesa, overlaying the p-contact metal. The second metal stack consists of an n-contact metal, forming a good electrical contact with the n-type GaN layer beneath the selective release layer, as well as a relatively thick metal layer that acts as both the mesa bond pad 505 and the cathode metal stack 506. The bond/cathode metal also forms a thick layer overlaying the edge of the mesa and providing a continuous connection between the mesa top and the engineered substrate. After the selective release layer 504 is removed by selective photochemical etching the thick metal provides mechanical support to retain the mesa in position on the GaN wafer until the bonding to (the bond material 507 of) the carrier wafer 508 is carried out.
The use of metal anchors has several advantages over the use of anchors made from the epitaxial device material. The first is density of the transferrable mesas on the donor wafer containing the epitaxial semiconductor device layers and the gallium and nitrogen containing engineered substrate. Anchors made from the epitaxial material must be large enough to not be fully undercut by the selective etch, or they must be protected somehow with a passivation layer. The inclusion of a large feature that is not transferred will reduce the density of mesas in two dimensions on the epitaxial device wafer. The use of metal anchors is preferable because the anchors are made from a material that is resistant to etch and therefore can be made with small dimensions that do not impact mesa density. The second advantage is that it simplifies the processing of the mesas because a separate passivation layer is no longer needed to isolate the active region from the etch solution. Removing the active region protecting layer reduces the number of fabrication steps while also reducing the size of the mesa required.
Referring to
In another example, the sacrificial region is completely removed by PEC etching and the mesa remains anchored in place by any remaining defect pillars. PEC etching is known to leave intact material around defects which act as recombination centers. Additional mechanisms by which a mesa could remain in place after a complete sacrificial etch include static forces or Van der Waals forces. In one example, the undercutting process is controlled such that the selective release layer is not fully removed.
In an example, the semiconductor device epitaxy material with the underlying sacrificial region is fabricated into a dense array of mesas on the gallium and nitrogen containing engineered substrate with the overlying semiconductor device layers. The mesas are formed using a patterning and a wet or dry etching process wherein the patterning comprises a lithography step to define the size and pitch of the mesa regions. Dry etching techniques such as reactive ion etching, inductively coupled plasma etching, or chemical assisted ion beam etching are candidate methods. Alternatively, a wet etch can be used. The etch is configured to terminate at or below a sacrificial region below the device layers. This is followed by a selective etch process such as PEC to fully or partially etch the exposed sacrificial region such that the mesas are undercut. This undercut mesa pattern pitch will be referred to as the ‘first pitch’. The first pitch is often a design width that is suitable for fabricating each of the epitaxial regions on the engineered substrate, while not large enough for the desired completed semiconductor device design, which often desire larger non-active regions or regions for contacts and the like. For example, these mesas would have a first pitch ranging from about 5 μm to about 500 μm or to about 5000 μm. Each of these mesas is a ‘die’.
In an example, these dice are transferred to a carrier wafer at a second pitch using a selective bonding process such that the second pitch on the carrier wafer is greater than the first pitch on the gallium and nitrogen containing engineered substrate. In this example, the dice are on an expanded pitch for so called “die expansion”. In an example, the second pitch is configured with the dice to allow each die with a portion of the carrier wafer to be a semiconductor device, including contacts and other components. For example, the second pitch would be about 50 μm to about 1000 μm or to about 5000 μm but could be as large at about 3-10 mm or greater in the case where a large semiconductor device chip is required for the application. The larger second pitch could enable easier mechanical handling without the expense of the costly gallium and nitrogen containing substrate and epitaxial material, allow the real estate for additional features to be added to the semiconductor device chip such as bond pads that do not require the costly gallium and nitrogen containing engineered substrate and epitaxial material, and/or allow a smaller gallium and nitrogen containing epitaxial wafer containing epitaxial layers to populate a much larger carrier wafer for subsequent processing for reduced processing cost. For example, a 4 to 1 die expansion ratio would reduce the density of the gallium and nitrogen containing material by a factor of 4, and hence populate an area on the carrier wafer 4 times larger than the gallium and nitrogen containing engineered substrate. This would be equivalent to turning a 2″ gallium and nitrogen engineered substrate into a 4″ carrier wafer. In particular, the present invention increases utilization of engineered substrate wafers and epitaxy material through a selective area bonding process to transfer individual die of epitaxy material to a carrier wafer in such a way that the die pitch is increased on the carrier wafer relative to the original epitaxy wafer. The arrangement of epitaxy material allows device components which do not require the presence of the expensive gallium and nitrogen containing engineered substrate and overlying epitaxy material often fabricated on a gallium and nitrogen containing engineered substrate to be fabricated on the lower cost carrier wafer, allowing for more efficient utilization of the gallium and nitrogen containing engineered substrate and overlying epitaxy material.
In the example depicted in
In one example, the bonding between the carrier wafer and the gallium and nitrogen containing engineered substrate with epitaxial layers is performed between bonding layers that have been applied to the carrier and the gallium and nitrogen containing engineered substrate with epitaxial layers. The bonding layers can be a variety of bonding pairs including metal-metal, oxide-oxide, soldering alloys, photoresists, polymers, wax, etc. Only epitaxial dice which are in contact with a bond pad 1807 on the carrier wafer 1806 will bond. Sub-micron alignment tolerances are possible on commercial die bonders. The engineered substrate with epitaxial layers is then pulled away, breaking the epitaxy material at a weakened epitaxial release layer or anchor regions such that the desired epitaxial layers remain on the carrier wafer. Herein, a ‘selective area bonding step’ is defined as a single iteration of this process.
In one example, the carrier wafer 1806 is patterned in such a way that only selected mesas come in contact with the metallic bond pads 1807 on the carrier wafer 1806. When the engineered substrate is pulled away the bonded mesas break off at the weakened sacrificial region or anchor regions, while the un-bonded mesas remain attached to the engineered substrate. This selective area bonding process can then be repeated to transfer the remaining mesas in the desired configuration. This process can be repeated through any number of iterations and is not limited to the two iterations depicted in
In an example, the present invention provides a method for increasing the number of gallium and nitrogen containing semiconductor devices which can be fabricated from a given epitaxial surface area, where the gallium and nitrogen containing epitaxial layers overlay an engineered substrate. The gallium and nitrogen containing epitaxial material is patterned into die with a first die pitch; the die from the gallium and nitrogen containing epitaxial material with a first pitch is transferred to a carrier wafer to form a second die pitch on the carrier wafer; the second die pitch is larger than the first die pitch.
In an example, each epitaxial device die is an etched mesa with a pitch of between about 1 μm and about 100 μm wide or between about 100 μm and about 500 μm wide or between about 500 μm and about 3000 μm wide and between about 100 and about 3000 μm long. In an example, the second die pitch on the carrier wafer is between about 100 μm and about 200 μm or between about 200 μm and about 1000 μm or between about 1000 μm and about 3000 μm. In an example, the second die pitch on the carrier wafer is between about 2 times and about 50 times larger than the die pitch on the epitaxy wafer. In an example, semiconductor LED devices, laser devices, or electronic devices are fabricated on the carrier wafer after epitaxial transfer. In an example, the semiconductor devices contain GaN, AlN, InN, InGaN, AlGaN, InAlN, and/or InAlGaN. In an example, the gallium and nitrogen containing material are grown on a polar, nonpolar, or semipolar plane. In an example, one or multiple semiconductor devices are fabricated on each die of epitaxial material. In an example, device components which do not require epitaxy material are placed in the space between epitaxy die.
In one example, device dice are transferred to a carrier wafer such that the distance between die is expanded in both the transverse as well as lateral directions. This can be achieved by spacing bond pads on the carrier wafer with larger pitches than the spacing of device die on the engineered substrate.
In another example, device dice from a plurality of epitaxial wafers are transferred to the carrier wafer such that each design width on the carrier wafer contains dice from a plurality of epitaxial wafers. When transferring dice at close spacing from multiple epitaxial wafers, it is important for the un-transferred dice on the epitaxial wafer to not inadvertently contact and bond to die already transferred to the carrier wafer. To achieve this, epitaxial dice from a first epitaxial wafer are transferred to a carrier wafer using the methods described above. A second set of bond pads are then deposited on the carrier wafer and are made with a thickness such that the bonding surface of the second pads is higher than the top surface of the first set of transferred die. This is done to provide adequate clearance for bonding of the dice from the second epitaxial wafer. A second epitaxial wafer transfers a second set of dice to the carrier wafer. Finally, the semiconductor devices are fabricated, and passivation layers are deposited followed by electrical contact layers that allow each die to be individually driven. The dice transferred from the first and second engineered substrates are spaced at a pitch which is smaller than the second pitch of the carrier wafer. This process can be extended to transfer of dice from any number of epitaxial engineered substrates, and to transfer of any number of devices per dice from each epitaxial engineered substrate.
Once the laser diode epitaxial structure has been transferred to the carrier wafer as described in this invention, wafer level processing can be used to fabricate the dice into laser diode devices. The wafer process steps may be similar to those described in this specification for more conventional laser diodes. For example, in many instances the bonding media and dice may have a total thickness of less than about 7 μm, making it possible to use standard photoresist, photoresist dispensing technology and contact and projection lithography tools and techniques to pattern the wafers. The aspect ratios of the features are compatible with deposition of thin films, such as metal and dielectric layers, using evaporators, sputter, and CVD deposition tools.
The laser diode device may have laser stripe region formed in the transferred gallium and nitrogen containing epitaxial layers. In the case where the laser is formed on a polar c-plane, the laser diode cavity can be aligned in the m-direction with cleaved or etched mirrors. Alternatively, in the case where the laser is formed on a semipolar plane, the laser diode cavity can be aligned in a projection of a c-direction. The laser strip region has a first end and a second end and is formed on a gallium and nitrogen containing engineered substrate having a pair of cleaved mirror structures, which face each other. The first cleaved facet comprises a reflective coating and the second cleaved facet comprises no coating, an antireflective coating, or exposes gallium and nitrogen containing material. The first cleaved facet is substantially parallel with the second cleaved facet. The first and second cleaved facets are provided by a scribing and breaking process according to an example or alternatively by etching techniques using etching technologies such as reactive ion etching (RIE), inductively coupled plasma etching (ICP), or chemical assisted ion beam etching (CAIBE), or other method. Typical gases used in the etching process may include Cl and/or BCl3. The first and second mirror surfaces each comprise a reflective coating. The coating is selected from silicon dioxide, hafnia, and titania, tantalum pentoxide, zirconia, including combinations, and the like. Depending upon the design, the mirror surfaces can also comprise an anti-reflective coating.
In a specific example, the method of facet formation includes subjecting the engineered substrates to a laser for pattern formation. In an example, the pattern is configured for the formation of a pair of facets for a ridge laser. In an example, the pair of facets face each other and are in parallel alignment with each other. In an example, the method uses a UV (355 nm) laser to scribe the laser bars. In a specific example, the laser is configured on a system, which allows for accurate scribe lines configured in a different patterns and profiles. In some examples, the laser scribing can be performed on the back-side, front-side, or both depending upon the application. Of course, there can be other variations, modifications, and alternatives.
By aligning the device dice such that the intended plane of the facet is coplanar with an easily cleaved plane of the single-crystal carrier wafer. Mechanical or laser scribes can then be used, as described above, to guide and initiate cleavage in the carrier wafer such that it is located properly with respect to the laser die and carrier wafer patterns. Zincblende, cubic and diamond-lattice crystals work well for cleaved carriers with several sets of orthogonal cleavage planes (e.g. [110], [001], etc.). Singulation of the carrier wafers into individual die can be accomplished either by sawing or cleaving. In the case of singulation using cleaving the same cleavage planes and techniques can be used as described for facet formation.
In a specific example, the method uses backside laser scribing or the like. With backside laser scribing, the method preferably forms a continuous line laser scribe that is perpendicular to the laser bars on the backside of the GaN engineered substrate. In a specific example, the laser scribe is generally about 15-20 μm deep or other suitable depth. Preferably, backside scribing can be advantageous. That is, the laser scribe process does not depend on the pitch of the laser bars or other like pattern. Accordingly, backside laser scribing can lead to a higher density of laser bars on each engineered substrate according to an example. In a specific example, backside laser scribing, however, may lead to residue from the tape on the facets. In a specific example, backside laser scribe often requires that the engineered substrates face down on the tape. With front-side laser scribing, the backside of the engineered substrate is in contact with the tape. Of course, there can be other variations, modifications, and alternatives.
It is well known that etch techniques such as chemical assisted ion beam etching (CAIBE), inductively coupled plasma (ICP) etching, or reactive ion etching (RIE) can result in smooth and vertical etched sidewall regions, which could serve as facets in etched facet laser diodes. In the etched facet process a masking layer is deposited and patterned on the surface of the wafer. The etch mask layer could be comprised of dielectrics such as silicon dioxide (SiO2), silicon nitride (SixNy), a combination thereof or other dielectric materials. Further, the mask layer could be comprised of metal layers such as Ni or Cr, but could be comprised of metal combination stacks or stacks comprising metal and dielectrics. In another approach, photoresist masks can be used either alone or in combination with dielectrics and/or metals. The etch mask layer is patterned using conventional photolithography and etch steps. The alignment lithography could be performed with a contact aligner or stepper aligner. Such lithographically defined mirrors provide a high level of control to the design engineer. After patterning of the photoresist mask on top of the etch mask is complete, the patterns in then transferred to the etch mask using a wet etch or dry etch technique. Finally, the facet pattern is then etched into the wafer using a dry etching technique selected from CAIBE, ICP, RIE and/or other techniques. The etched facet surfaces must be highly vertical of between about 87 and about 93 degrees or between about 89 and about 91 degrees from the surface plane of the wafer. The etched facet surface region must be very smooth with root mean square roughness values of less than about 50 nm, 20 nm, 5 nm, or 1 nm. Lastly, the etched must be substantially free from damage, which could act as non-radiative recombination centers and hence reduce the COMD threshold. CABE is known to provide very smooth and low damage sidewalls due to the chemical nature of the etch, while it can provide highly vertical etches due to the ability to tilt the wafer stage to compensate for any inherent angle in etch.
The laser stripe is characterized by a length and width. The length ranges from about 50 μm to about 3000 μm, but is preferably between about 10 μm and about 400 μm, between about 400 μm and about 800 μm, or about 800 μm and about 1600 μm, but could be others such as greater than 1600 μm. The stripe also has a width ranging from about 0.5 μm to about 50 μm but is preferably between about 0.8 μm and about 2.5 μm for single lateral mode operation or between about 2.5 μm and about 80 μm for multi-lateral mode operation, but can be other dimensions. In a specific example, the present device has a width ranging from about 0.5 μm to about 1.5 μm, a width ranging from about 1.5 μm to about 3.0 μm, a width ranging from about 3.0 μm to about 360 μm, and others. In a specific example, the width is substantially constant in dimension, although there may be slight variations. The width and length are often formed using a masking and etching process, which are commonly used in the art.
The laser stripe is provided by an etching process selected from dry etching or wet etching. The device also has an overlying dielectric region, which exposes a p-type contact region. Overlying the contact region is a contact material, which may be metal or a conductive oxide or a combination thereof. The p-type electrical contact may be deposited by thermal evaporation, electron beam evaporation, electroplating, sputtering, or another suitable technique. Overlying the polished region of the engineered substrate is a second contact material, which may be metal or a conductive oxide or a combination thereof and which comprises the n-type electrical contact. The n-type electrical contact may be deposited by thermal evaporation, electron beam evaporation, electroplating, sputtering, or another suitable technique.
An example of a processed laser diode cross-section according to one example is shown in
Once the laser diodes have been fully processed within the gallium and nitrogen containing layers that have been transferred to the carrier wafer, the carrier wafer must be diced. Several techniques can be used to dice the carrier wafer and the optimal process will depend on the material selection for the carrier wafer. As an example, for Si, InP, or GaAs carrier wafers that cleave very easily, a cleaving process can be used wherein a scribing and breaking process using conventional diamond scribe techniques may be most suitable. For harder materials such as GaN, AlN, SiC, sapphire, or others where cleaving becomes more difficult a laser scribing and breaking technique may be most suitable. In other examples, a sawing process may be the most optimal way to dice the carrier wafer into individual laser chips. In a sawing process a rapidly rotating blade with hard cutting surfaces like diamond are used, typically in conjunction with spraying water to cool and lubricate the blade. Example saw tools used to commonly dice wafers include Disco saws and Accretech saws.
By choosing a carrier wafer material such as AlN, BeO, diamond, or SiC that is suitable as a submount between the laser chip and the mounting surface, the diced laser chip on the carrier wafer is a chip on submount (CoS). This wafer level packaging features is a strong benefit of the lifted-off and transferred gallium and nitrogen containing epitaxial layer example.
A schematic diagram illustrating a CoS based on lifted off and transferred epitaxial gallium and nitrogen containing layers is shown in
Further process and device description for this example describing laser diodes formed in gallium and nitrogen containing epitaxial layers that have been transferred from the native gallium and nitrogen containing engineered substrates are described in U.S. patent application Ser. No. 14/312,427 and U.S. Patent Publication No. 2015/0140710, which are incorporated by reference herein. As an example, this technology of GaN transfer can enable lower cost, higher performance, and a more highly manufacturable process flow.
Some embodiments introduce a waveguide layer coupled with a device layer of gallium and nitrogen containing material to produce passive optical elements in addition to the active optical elements. First, gallium and nitrogen containing layers are grown or transferred onto a non-native substrate and patterned into desired locations. Next, a layer consisting of silicon oxide (e.g., SiO2), silicon nitride (e.g., SiN), and/or other dielectrics is deposited onto the non-native substrate and patterned into waveguide structures to guide the light. Depending on the approach to light coupling between the active and passive layers, there may be an additional planarization step before the waveguide layer deposition in which a dielectric or polymer is deposited onto the non-native substrate and polished down to provide a flat plane for optical waveguiding. In one approach, the light is coupled from the gallium and nitrogen containing device through an etched or cleaved facet into a waveguide layer deposited directly in front. The adjacent interfaces may be anti-reflective coated to minimize reflections. In another approach, the light is directionally coupled into a waveguide layer located in parallel above or below the gallium and nitrogen containing layers such that the light will couple to and from the gallium and nitrogen containing waveguide.
Because the dielectric material does not have strict substrate requirements for its formation, large, cheap, readily available wafers from mature material systems such as silicon may be used. Large, complicated passive optical circuits may be added, including, but not limited to, splitters/combiners, 90° optical hybrids, arrayed waveguide gratings (AWGs), gratings, and beam shapers/expanders. Active functions may be added by depositing materials such as lithium niobate and metal electrodes for optical phase modulation, de-speckle, frequency doubling phase shifting, de-speckle, and signal attenuation. Different wavelength active material may be transferred to a common substrate and patterned into arrays for wavelength division multiplexing (WDM) or display applications. Electronic functions may be also added either by transfer or patterned into the non-native substrate itself, such as drivers, trans-impedance amplifiers, and read/write functions. Phosphor material may be added to convert blue to yellow light.
Some embodiments enable coupling multiple cavities together on the same carrier wafer via heterogeneous integration with high coupling efficiency. In accordance with an embodiment, for example, a first and second epitaxial substrate are provided. In one epitaxial growth operation, the epitaxial device layers of a laser diode are grown on the first epitaxial substrate, and in another epitaxial growth operation, the epitaxial device layers for a passive waveguide are grown on the second epitaxial substrate. The passive waveguide may comprise any material or crystal orientation. As an example, the passive waveguide may comprise a c-plane gallium and nitrogen containing material. The passive waveguide may have an absorption edge corresponding to shorter wavelengths than the lasing wavelength of the laser diode and may consist of epitaxial layers for optical confinement with a doping profile optimized to minimize propagation loss. Alternatively, the epitaxial layers within the passive waveguide may be optimized to reduce transit time for high-speed signal modulation or provide index tunability through electrical current injection. The epitaxial device layers of the first and second substrates may be fabricated using standard lithographic and etching techniques to form a plurality of mesas. A bond media is deposited on the top surfaces of the mesas. In some embodiments, the mesas have anchor regions at either end consisting of bond mesa to secure the mesas prior to transfer and etched away during facet formation. In some embodiments, the anchors are distributed evenly throughout the mesa to minimize space between the facets on the active device mesas and passive waveguide mesas. A common carrier wafer is provided where a first surface of the carrier wafer is overlaid by regions of bond media. Using transfer processes, one or more mesas are transferred from each of the first epitaxial substrate and the second epitaxial substrate to the common carrier wafer. In some embodiments, the mesas having active device layers are transferred from the first epitaxial substrate and the mesas having the passive waveguide layers are transferred from the second epitaxial substrate to the common carrier wafer such that light generated in each of the active device mesas is coupled into one of the passive waveguide mesas.
In some embodiments, cavities are formed in both the active device mesas and the passive waveguide mesas. The cavities may be formed by creating facets on each end of the mesas and/or by providing a mechanism for lateral optical confinement in directions perpendicular to the direction of light propagation within the cavity. The facets may be formed, for example, using an etch process. The facets may be formed prior to transfer of the mesas to the common carrier wafer in some embodiments and after the transfer in other embodiments. When the facets are formed after the transfer, a single etch process may be used to form the facets on both the active device mesas and the passive waveguide mesas. The etched facets may then be coated.
In some embodiments, lateral confinement may be provided by forming ridges on the active device mesas. The ridges may be formed, for example, using an etch process that is performed prior to transfer of the mesas to the common carrier in some embodiments or after transfer of the mesas to the common carrier in other embodiments. In some embodiments, the edges of the active device mesas perpendicular to the direction of the laser light propagation establish the lateral confinement of optical modes and no ridges are etched into the mesas.
In some embodiments, the active device mesa cavity and the passive waveguide mesa cavity are coupled end-to-end on the common carrier wafer.
In some embodiments, adjacent active device mesas and passive waveguide mesas may be disposed so that the butt-coupled facets are touching or nearly touching. In the embodiment that the facets are nearly touching, the facets must be etched and anti-reflective coated prior to the transfer of the mesas to the common carrier. Embodiments that include a gap between facets of the adjacent mesas may include a material with a refractive index close to the effective index of the optical mode confined within the mesas (n=2.5) and low absorption loss, such as tantalum pentoxide deposited by atomic layer deposition (ALD) in the gap. Many materials with an index higher than air and low optical absorption in the visible light region could be used to reduce the index contrast from the waveguides and air gap, such as silicon dioxide, which could be applied by plasma-enhanced chemical vapor deposition (PECVD), ALD, or as spin-on-glass.
In some embodiments, the active device mesa and the passive waveguide mesa may be coupled using vertically angled turning mirrors etched into the mesa cavities.
In some embodiments, the active device mesas and the passive waveguide mesas are coupled using horizontally angled turning mirrors formed in the mesa cavities.
In some embodiments, only one facet is angled, and the major axes of the two cavities are not parallel. Rather, the major axis of one of the cavities is rotated relative to the other such that light exiting the active device cavity would be refracted into guided modes of the passive waveguide cavity, and light reflected from the angled facet of the passive waveguide cavity would be mostly reflected away from the active device cavity.
Some embodiments provide gallium and nitrogen containing laser diodes or gain sections coupled to waveguide structures where the waveguide structures function as emitters. The waveguides may be configured as resonators and/or include diffraction gratings in some embodiments. These structures can be used, for example, to narrow linewidth, narrow bandwidth, and/or enable wavelength tuning. The waveguides can be formed from lower cost dielectric materials compared to the gallium and nitrogen containing laser diodes and gain sections. Providing a narrow linewidth and narrow emission band can be useful in a number of applications including controlling atomic transitions in quantum sensors.
A schematic diagram illustrating a gallium and nitrogen containing laser diode or gain sections coupled to a dielectric waveguide is shown in
A schematic cross section illustrating a laser diode or gain section coupled to a dielectric waveguide according to an embodiment is shown in
While the above is a full description of the specific embodiments, various modifications, alternative constructions and equivalents may be used. As an example, the packaged device can include any combination of elements described above, as well as outside of the present specification. Although the embodiments above have been described in terms of a laser diode, the methods and device structures can also be applied to other stimulated light emitting devices. Therefore, the above description and illustrations should not be taken as limiting the scope of the present invention which is defined by the appended claims.
