Patent Application
20050201437
This invention relates to semiconductor lasers, and more particularly to high power semiconductor lasers suitable for optical telecommunication applications.
Semiconductor lasers are typically formed from pn-junctions that have been enhanced to facilitate the efficient recombination of electron-hole pairs leading to the emission of radiation (light energy). A well known improvement to semiconductor lasers is the addition of a new layer of material between the P-type and N-type semiconductor layers, the new layer of material having a lower band gap energy than P-type and N-type layers. The layer formed by the material having the lower band gap energy is commonly referred to as the active region (or active layer) in a semiconductor laser.
Typically, a heterojunction refers to an interface between two different materials. Therefore, the insertion of an extra layer (active region) between the P-type and N-type layers results in what is known as a double heterostructure, as there will be a heterojunction at the interface of both the P-type and N-type materials. The doping in the active region is set at various levels depending upon the effect it is intended to have.
Thus, it is now common practice for semiconductor heterostructure lasers to be made up of three or more semiconductor layers. The simplest lasers include a P-type confinement region (P-type layer), an N-type confinement region (N-type layer) and an active region. The active region is typically made up of a number of layers and is located in the depletion region of the pn-junction between the P-type and N-type confinement regions. The optical mode is primarily confined in the active region (and the adjoining layers) because of the difference in the index of refraction between the active region, and the P-type and N-type confinement regions. The active region provides gain to the optical mode when the heterostructure is forward biased.
Light is generated within the active region once the semiconductor laser is forward biased and current is injected into the heterostructure. The active region is often composed of many layers in order to tailor the performance of the laser to meet the desired requirements (e.g. modulation bandwidth, power, sensitivity to temperature, etc.) of the laser's intended application.
The maximum optical output power of a semiconductor laser is usually limited by heating. The temperature of the active region increases with drive current, which degrades the laser performance. To achieve high optical power, one usually needs to increase the cavity length and the ridge width, which decreases the dissipated power density and keeps the laser from over heating. The power density is decreased because the electrical and thermal impedances decrease as the area where the current is injected increases.
When the cavity length is increased (typical cavity length is 2 mm for a high power laser), the efficiency (mW of optical power/mA of drive current) decreases because of internal optical loss in the cavity (that is not particular to the ridge structure, but is common in all structures). The optical loss is mainly due to the absorption of the light energy in the P-type material (region). Decreasing the overlap of the optical mode within the P-type region would then be a useful way to decrease the loss of light energy within the laser, which would enable the use of longer cavities to be used to create lasers with higher output power.
There are different structures that can be used to decrease the optical losses (i.e. losses of light energy). However, those structures usually decrease the optical mode size in the laser cavity. The drawback is that the far field of the optical mode (i.e. optical far field) gets wider and the optical power is more difficult to couple into an optical fiber. The optical far field and the optical mode in the laser cavity (the near field) are mathematically related by Fourier transform. This is a consequence of optical diffraction. Usually the optical far field is symmetric even though the near field is not. The loss in the coupling efficiency into the fiber happens only because the optical mode in the fiber and the laser far field do not have the same shape. An optical fiber can only accept a circular spot with a maximal divergence. The laser far field is usually elliptical and can have a large divergence.
For telecommunication applications it is the amount of optical power coupled into the fiber and not the raw optical power out of the laser that is significant. Thus, there is a need for a structure that simultaneously:
The active region is commonly made up of a number of layers, some of which are designed to be quantum wells (or bulk wells). A quantum well is designed to be a very thin layer, thus allowing a better localization of electrons in the conduction band and holes in the valence band that will enhance electron-hole pair recombination. When an electron-hole pair recombine the excess energy the electron had possessed is emitted as light (radiation) adding to the operation of the laser. Furthermore, reducing the band gap energy of the active region relative to the band gap energies of the two confinement layers improves the confinement of the electrons and holes to the active region; thus, the optical mode profile is guided to remain within a narrow spot. However, for lasers suitable for optical telecommunications, an optical mode profile that is too narrowly confined is difficult to couple into a fiber as it will have a wide far field. To achieve the best performance in a high-power laser, both the internal and external efficiency of the laser must be maximized. The internal efficiency of a laser is the efficiency at which electrical energy is converted into light energy (i.e. into the optical mode). The external efficiency is the efficiency at which the optical mode leaves the laser. However, there is a trade-off between the two measures of efficiency and thus far high power lasers have been limited by this trade off. Specifically, when considering semiconductor lasers, the external efficiency is largely the result of optical mode energy losses in P-type confinement layer, which tends to absorb much more optical energy than the active or N-type layers. On the other hand, internal efficiency (of semiconductor lasers) is usually dominated by current leakage which increases with temperature, and the temperature in turn increases with drive current. In other words, the electrical energy supplied to the laser is not maximally converted into optical energy within the laser as some current is dissipated through the semiconductor layers.
There is also another significant source of optical energy loss that must be taken into account when considering lasers for optical telecommunication applications. Semiconductor lasers used for optical telecommunication applications must have their outputs coupled to a fiber and as such it is common that lasers are commercially packaged with a short piece of fiber, known as a pigtail, already aligned to the output of the laser. Thus, for telecommunication applications the external efficiency of a laser should be measured to include the effects of industrial packaging. In this case that would mean that the external efficiency of a laser should be measured at the end of the pigtail so that coupling losses can be taken in account. In other words, the potential for coupling loss from the laser into the pigtail must be considered in the design of a laser to be used for optical telecommunication applications as coupling loss can be a significant contributor to the degradation of the external efficiency. Precise alignment of the laser output to the pigtail is not enough to solve this problem. Current high-power lasers have outputs that have a wide far field, due to attempt to confine the optical mode in the active region. This fact combined with the current use of small numerical aperture fibers required for reduced distortion optical transmissions create a situation where there is a significant optical mode energy loss to be accounted for when coupling the laser output into the fiber.
Semiconductor lasers following the above design characteristics are known in the art. One particular close example is disclosed in Reid U.S. Pat. No. 6,724,795 B2, assigned to the assignee of the present invention and incorporated herein by reference.
It would be desirable to have a high power semiconductor laser that was optimized to be internally efficient, experienced low optical energy losses within the laser and had an output beam with a narrow far field so that the beam could be coupled into a fiber with minimal optical coupling loss.
Gratings, often Bragg gratings, of various kinds have been implemented in such high power semiconductor lasers. The invention is related to edge-emitting high power and high reliability distributed feedback lasers. Most of such distributed feedback lasers on an n-doped substrate are designed with the grating on top of the active region in the p-clad of the waveguide. In a ridge waveguide structure, the grating is usually fabricated between a first and a second growth. Data suggests that to achieve a good reliability, a larger concentration of p-dopant is required than would be dictated solely by electrical requirements during the beginning of the second growth to compensate for residue at the interface with the first growth and the grating. A few problems exist due to this large concentration of p-dopant.
The p-dopant element that is usually used is zinc, which is a highly mobile atom and thus tends to diffuse readily in the structure. The second aspect is that active p-doping is a source of holes (lack of electrons), which can lead to a significant contribution to optical absorption in the waveguide, limiting the maximum cavity length than can be used. Use of longer cavity lengths is an important design tool for minimizing the heating in the laser. Consequently, limiting the cavity length can restrict device optical output power and reliability because these performance attributes are usually thermally or current density accelerated. In short, the p-dopant concentration imposed by placing the grating in the p-cladding limits optical output power and reliability.
It would thus be desirable to minimize the optical loss in a distributed feedback laser cavity independently from the grating process.
As mentioned, this invention is directed to semiconductor lasers and applicable to lasers of the ridge waveguide type and of the buried heterostructure type. All such semiconductor lasers have or consist of a plurality of layers. The particular gist of this invention is to optimize the position and structure of one or more gratings within the semiconductor layers forming the laser. In short, the invention provides a semiconductor edge emitting distributed feedback laser comprising a grating and a low loss optical waveguide, where the optical loss and the grating fabrication can be optimized independently.
The invention will now be described by way of example and it should be understood that modifications, for example including the invention in a buried heterostructure, should not be seen as departing from the scope of the invention.
First the growth interface problem is solved by including the grating in the n-cladding of the optical waveguide below the active region as illustrated in the drawings. It is usually desirable to have a minimum thickness of InP on top of the grating to planarize the surface when growing the active region, which comprises quaternary semiconductor alloys like InGaAsP. The planar surface is desirable to maintain the stochiometry and minimize the number of defects, which is better for reliability. It is however conceivable that the active region also be grown following the geometry of the grating. Interface residue at this grating are much less an issue than when the grating is in the p-cladding since they can be easily compensated by higher n-doping, which does not contributed significantly to optical losses.
A second aspect of the invention is to minimize optical waveguide loss in the laser cavity. This is achieved by inserting one or more ballast layers in the n-cladding of the optical waveguide as shown in the drawings. The role of the ballast layers is to tilt the optical mode substantially towards the n-cladding, away from the p-cladding, to decrease the optical mode overlap with the lossy p-doped material. As will be understood by someone skilled in the art, although the ballast layers tilt the optical mode towards the n-cladding, the peak of the optical intensity is still substantially located in the active region.
A benefit of the ballast layers can be to increase the optical spot size, which then leads to narrower far field divergence, which helps optical coupling efficiency to optical fibers.
A benefit of the invention can be to enable the fabrication of DFB lasers with optical cavities longer than 2 mm to generate optical power larger than 100 mW at 90° C.
Another benefit of the invention can be to enable the fabrication of DFB lasers with optical cavities longer than 2 mm to operate the device at a small current density <5 kA/cm2 and low optical power <100 mW. Under those conditions, current and thermal acceleration of the degradation is lowered, which can lead to better device reliability without requiring accelerated device aging or burn-in, i.e. device stabilization under accelerated conditions.
Other aspects and features of the present invention will become apparent to those of ordinary skill in the art upon review of the following description of specific embodiments of the invention in conjunction with the accompanying figures.
The invention will now be described in greater detail with reference to the accompanying drawings, in which:
Referring to
The laser consists of the following layers illustrated in
The P-type confinement layer 8′ and the P-type contact layer 6 are etched to create trenches 14 and 14′ that define a ridge structure 15; at least one dielectric layer 4 (there can be more than one dielectric layer) is then deposited over the exposed surfaces of the laser such that the dielectric material making up the at least one dielectric layer substantially evenly covers the exposed surface including the vertical edges of the trenches 14 and 14′, the dielectric material typically being an oxide or nitride compound; and, atop the ridge structure 15 a via (opening) is etched through the at least one dielectric layer 4, exposing the P-type contact layer 6, into which a second metal contact 2 is deposited such that it is in contact with P-type contact layer 6 on the ridge structure 15.
In some embodiments, the layers composing the active region may include quantum well layers (layers that are quite thin, about 10 atomic layers) and barrier layers between the quantum layers. Both, quantum wells and barrier layers are sandwiched on both sides by the P-type and N-type confinement layers 8 and 9 or 9′, resp., of the semiconductor laser. The confinement layers aid in funnelling electrons and holes into the quantum wells where recombination occurs, and the significant effect of recombination is that light is generated or equivalently radiation is emitted. This results in the index of refraction profile of the active region 12 having a high index of refraction in the quantum well layers and a lower index of refraction in the barrier layers.
Referring to the first and second P-type confinement layers 8 and 8′ and the etch-stop layer 10 shown in
According to one embodiment of the invention, the grating 16 is placed within the N-type body, sandwiched between the confinement layers 9 and 9′. This grating now provides the desired effect that the optical loss in the laser cavity can be minimized independently from the grating process. Thus the invention provides the desired semiconductor edge-emitting distributed feedback laser in which the optical loss and the grating fabrication can be optimized independently.
The actual thickness of each of the aforementioned layers that make up the laser is found through empirical study for a particular application. However, the typical thickness or range can be provided here for the most important layers. It should be noted that the cross-sectional view shown in Figure is not to scale. The N-type substrate layer 11 is not important to the creation of and guiding of the optical mode, but it is required to provide a low electrical resistance mechanical support to the rest of the laser structure and as such it is typically 130 microns thick. The optical trap layer 1 is typically 0.05 to 0.25 microns thick. The N-type confinement layers 9 and 9′ may be slightly thicker, with a typical thickness ranging from 0.1 to 0.7 microns. The etch-stop layer 10 is also not important for the operation of the laser, but is present to protect the layer underneath it from the etching process used to create the trenches 14 and 14′.
The thickness of the grating layer is about 10 nm, i.e. 0.01 microns. It may be made from the same material or composition of materials as the ballast layer. InGaAs is a preferred material. As mentioned above, the ballast layers or optical superlattice layers have thicknesses in the range of 100 nm, i.e. 0.1 micron.
In this embodiment the thickness of the active region 12 typically does not need to exceed 0.1 microns, however can be increased to approximately 1.0 microns for exotic applications. The ridge structure 15 in which the P-type confinement layer 8′ is situated is typically 1.5 to 2.5 microns thick. The first metal contact layer 3 and the second metal contact layer 2 are designed to provide a low electrical resistance interface between connecting metals (such as gold or aluminium) to the laser. The thickness of each contact does not greatly impact the optical performance of the laser.
The primary advantage of the ridge structure is that it laterally confines the light in a single narrow optical mode that can be coupled into a telecommunication type optical fiber. There are other structures that can be used to achieve lateral confinement, for example a buried heterostructure as shown in
The following description of the preferred embodiment assumes the example material introduced above are used. However, other semiconductor materials that are suitable for lasers used in telecommunications applications may be used, for example gallium arsenide (GaAs).
With reference to an orthogonal co-ordinate system xyz indicated generally at 17, shown in
Referring to
In the laser shown in
The refractive index 3.16 is that of InP (n8, n9 and n11) and as such is fixed for a given wavelength. The other refractive indices vary with the InGaAsP composition that is used. Typically the index in the optical trap layer 1 would vary from 3.25 to 3.35. The refractive index of the active region 12 is approximately an average of the refractive indices of all layers that comprise it and generally would vary from 3.35 to 3.45.
Referring back to
As the optical mode is primarily generated in the active region 12, the active region 12 generally having the highest refractive index profile within the laser, the optical mode is substantially confined to the active region 12. The energy of the optical mode is confined in the horizontal direction to substantially a single spot by the ridge structure 15. A substantial amount of the energy of the optical mode traversing the N-type confinement layer 9 is gathered and is trapped in the optical trap layer 1. Normally without the optical trap layer, the optical mode would be evenly distributed throughout either side of the active region. Thus, the optical trap layer is breaking the symmetry of the optical mode energy distribution throughout the heterostructure as described in Reid U.S. Pat. No. 6,724,795 B2, mentioned above and incorporated herein by reference.
The actual thickness of each of the aforementioned layers that make up the laser is found through empirical study for a particular application, as before for the first embodiment described in detail above. The optical trap layers 1 and 1′ are typically 0.05 to 0.25 microns thick. Each of the N-type confinement layers 9 and 9′ has a preferred thickness ranging from 0.1 to 0.7 microns. The etch-stop layer 10 is also not important for the operation of the laser. The etch-stop layer 10 is present to protect the layer underneath it from the etching process used to create the trenches 14 and 14′.
According to this second embodiment of the invention, the grating 16 is placed within the N-type body, sandwiched between the confinement layers 9′ and 9″, just below the active region 12. The grating 16 again provides the desired effect that the optical loss in the laser cavity can be minimized independently from the grating process. Thus the invention provides the desired semiconductor edge-emitting distributed feedback laser in which the optical loss and the grating fabrication can be optimized independently.
The thickness of the grating layer 16 is about 10 nm, i.e. 0.01 microns. It may be made from the same material or composition of materials as the ballast layer. InGaAs is a preferred material. As mentioned above, the ballast layers or optical superlattice layers have thicknesses in the range of 100 nm, i.e. 0.1 micron.
Using a semiconductor heterostructure described above for a laser, laser action is achieved by cleaving the semiconductor heterostructure in two places along a crystallographic plane forming a resonating cavity with mirror facets, as previously described for the first embodiment.
Referring again to
As an optical mode is initially generated in the active region 12, the active region 12 having the highest refractive index n12 within the laser, the optical mode is substantially confined to the active region 12. The energy of optical mode is also guided away from the P-type confinement layer 8 by the ridge structure 15 such that substantially more of the optical mode energy is guided towards and into the N-type confinement layer 9 adjacent to the opposite side of the active region 12. However, a substantial amount of the energy of the optical mode traversing the N-type confinement layer is 9 pulled further away from the active region 12 by the optical trap layers 1 and 1′. Each optical trap layer 1 and 1′ gathers and traps optical energy within it as a result of having higher refractive indices n1 and n1′ relative to each of the refractive indices n9, n9 and n11 corresponding to the N-type confinement layers 9 and 9′ and N-type substrate layer 11 respectively.
Common to both embodiments of the lasers, shown in
Further embodiments with more than two optical trap layers are within the scope of this invention. The laser is preferably embodied using a ridge structure on the P-type side of a heterojunction, as shown in the above examples. Alternatively, the ridge structure could be on the N-type side of the heterojunction.
An active region 46 is fabricated by etching to create a ridge or mesa shape. The etching is done using chemical etching techniques such as reactive ion etching (RIE) or non-selective wet chemical etches. The active region 46 consists of a multi-quantum well (MQW) core bounded by separate confinement heterostructure (SCH) layers 41 and 41′. The MQW and SCH layers will be embedded in large bandgap semiconductor material, or cladding layer, such as indium phosphide and will consist of lower bandgap materials such as InGaAsP or InAlGaAs. The purpose of these layers is to provide optical waveguiding and gain to the optical mode.
Once the active region mesa is defined, blocking layers 45 and 45′ are grown using epitaxial crystallographic techniques such as Metal Organic Chemical Vapour Deposition (MOCVD) or Liquid Phase Epitaxy (LPE). These layers will typically be indium phosphide and will act as current blocking regions to ensure current flows through the active mesa (46) under device operation. The blocking layers 45 and 45′ can be semi-insulating, such as iron-doped InP, or grown as reverse bias pn-junctions, e.g. as successive layers of zinc and silicon doped InP.
Finally, a p-doped layer 44, typically InP and InGaAs, is grown over the active region mesa 46 and the blocking layers 45 and 45′ to provide ohmic contact to the metal contacts which are deposited after epitaxy is complete using evaporation, sputtering or electroplating processes. The ohmic contacts are not shown in the figures.
For the fabrication of distributed feedback (DFB) buried heterostructure lasers either immediately prior to or after the growth of the active region 46, a grating layer 43 is periodically etched using techniques such as holography or electron beam lithography and wet chemical etching. The grating layer 43 will consist of a material with higher refractive index compared to the cladding material, e.g. InGaAsP. Care is taken to ensure the composition of this material does not generate absorption at the operating wavelength of the laser. For a 1550 nm laser, the grating layer will typically have bandgap photoluminescence wavelengths of 1100 to 1200 nm. The periodic refractive index perturbation created by the etched grating layer provides the optical feedback necessary to generate lasing action in the device. Once etched, the grating layer is overgrown typically with an InP layer 42.
The ballast layers 41 and 40 are used to provide an independent means to optimize the laser waveguide structure and consist of material with higher refractive index compared to the cladding material, e.g. InGaAsP. As with the grating layer 43, care is taken to ensure the composition of this material does not generate absorption at the operating wavelength of the laser. For a 1550 nm laser, the ballast layers 41 and 40 will typically have bandgap photoluminescence wavelengths of 1000 to 1200 nm.
The grating layer 43 and the ballast layers 41 and 40 can be unchanged by the mesa etch process, as shown in
As previously described, the energy losses are a result of the fact that the laser emits a divergent elliptical beam, which poorly couples into a circular optical fiber that accepts only light from a particular cone. As a result of aspects of the invention disclosed it is possible to shape a far field that would have a full-width at half-maximum (FWHM) of 25 degrees in the y direction and a FWHM of 10 degrees in the x direction. The optical fiber requires that the light be within a cone of 15 degrees circular.
What has been described is merely illustrative of the application of the principles of the invention. Other arrangements and methods can be implemented by those skilled in the art without departing from the spirit and scope of the present invention. In particular should it not present a problem for those skilled in the art to apply the techniques described above to other laser designs, e.g. self-aligned stripe lasers or others. Specifically, other semiconductor optical devices, such as amplifiers and distributed feedback lasers or other devices containing gratings, can be constructed using the same semiconductor heterostructure as the embodiments of the semiconductor laser provided. The same structure maybe used to produce an amplifier by applying a low reflectivity coating to the facets.
This application is a continuation-in-part of U.S. application Ser. No. 10/800,546, filed Mar. 15, 2004, and claims priority under 35 USC §119(e) to U.S. Provisional Application No. 60/583,443, filed Jun. 28, 2004. The entire disclosure of each of these applications is hereby incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| 60583443 | Jun 2004 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 10800546 | Mar 2004 | US |
| Child | 11001736 | Dec 2004 | US |
