WO2005112210A1

WO2005112210A1 - Semiconductor devices including gratings formed using quantum dots and method of manufacture

Info

Publication number: WO2005112210A1
Application number: PCT/IB2005/001325
Authority: WO
Inventors: John Kenton White
Original assignee: Bookham Technology Plc
Priority date: 2004-05-14
Filing date: 2005-05-16
Publication date: 2005-11-24

Abstract

The present invention provides semiconductors devices comprising gratings formed with aquantum dots. The invention further provides a method for the manufacture of the gratings of self-aligned quantum dot (QD) structures in these devices by using a template (23). These optical devices can be semiconductor lasers, waveguide devices, optical filters, DFB lasers, DBR devices, band-pass filters, photonic bandgap structures and the like. Furthermore, the invention provides a method of manufacture of gain-coupled gratings without the need to etch the active layer of the laser.

Description

SEMICONDUCTOR DEVICES INCLUDING GRATINGS FORMED USING QUANTUM DOTS AND METHOD OF MANUFACTURE

FIELD OF INVENTION

The present invention pertains to the field of semiconductor devices having gratings formed from quantum dots and the manufacture thereof.

BACKGROUND

Since their initial invention in 1962, semiconductor lasers have become a significant part of the electronics infrastructure. These devices provide integral elements of communication and computer systems, and are key components in a host of widely used technological products, including compact disk players and laser printers. Designs have evolved from basic light emitting diodes to multi-layer heterostructure lasers with wideband performance.

Lasers, as their name suggests, produce light amplification by stimulated emission of radiation. The emitted radiation is highly directional, monochromatic and coherent. Amplification of the emitted radiation occurs by providing an optical resonant cavity in which a photon density of a particular frequency can build up to a large value through multiple internal reflections between an arrangement of parallel plates or mirrors. In semiconductor lasers, a large forward bias is applied across a semiconductor junction to stimulate the emission of radiation. For applications such as telecommunications, where the wavelength must be closely controlled, the reflective mirrors are replaced by a periodic grating structure such as Distributed Bragg Reflectors (DBRs), which are essentially partial reflectors spaced to reflect light constructively. These are formed by laying down alternating layers of semiconductor or dielectric material with a difference in refractive index. Another type of well established grating structure is a Distributed Feedback (DFB) structure. These periodic grating structures can be above, below, or within the active region, that is, the region in which the stimulated emission occurs. There are two types of DFB lasers, a gain-coupled laser and an index-coupled laser. A gain-coupled laser has a regular reflecting pattern where the corrugations extend into the active region of the laser, whereas an index-coupled laser has corrugations that are adjacent to the active region of the laser. The gratings are normally first order gratings where the spacing of the grating or pitch, is equal to half the optical wavelength in the waveguide. Higher order gratings would be multiples of this first order grating spacing. Thus a second order grating would have twice the spacing and third order gratings would be three times the spacing, when compared to the spacing of a first order grating. Different orders of gratings can be combined in a single laser structure design.

A key part of the progress in the design of semiconductor lasers has been the development of the associated manufacturing methods. There has been a continuous effort to find ways to improve the techniques for building lasers with gratings and methods of improving the accuracy of the final constructed laser structure. The performance of a laser is highly dependent on the final physical accuracy of the layers, gratings and electrical contacts as well as the consistency of the composition of the materials used. Standard semiconductor techniques are generally used in the manufacture of semiconductor lasers. The manufacturing process requires multiple steps of deposition and etching of layers of materials on substrates and the need for these materials to have a high level of purity. Patterning of the various layers is done typically using a photolithography process in which geometrical shapes on a mask are transferred to the semiconductor layers thereby producing the required structures, such as the gratings in the laser structure. The availability of metallorganic chemical vapour deposition (MOCVD), molecular beam epitaxy (MBE), advanced deposition techniques for dielectric and metal films, stepper photolithography and electron beam lithography has allowed significant steps for the improvement in production capability of these forms of lasers. Fabrication generally requires a number of steps and can result in significant costs and potential low yield of acceptable semiconductor lasers.

Lasers, known as QD lasers, have also been made where the laser structure incorporates layers of material, containing quantum dots (QDs) that act in the same way as semiconductors. QDs resemble tiny cones of semiconductor material typically 20 or 30 nm in diameter. These QDs are capable of confining the electron wavefunction within a discrete spatial location. This confinement leads to many benefits, the most important one being that the energy levels of QDs are discrete, compared to a continuum of levels in more traditional structures, such as quantum wells and bulk semiconductors. This discrete energy level property has led some to describe QDs as artificial atoms. Unlike atoms, the energy spectrum of the QDs can be engineered by controlling the size and shape of the dot. Using QDs, materials can be manufactured with energy signatures not found in nature. Since the energy levels define how the material interacts with light, control of the energy levels can lead to specialty photonic materials. Applications range from nano-sized markers for biophotonics, to sources and receivers for traditional telecommunication systems. Untapping the potential benefits of QDs can require control over the size and position of the QDs. The "purer" the QD material, i.e. the more homogeneous the distribution of QD properties in the material, the more possible it is to take advantage of the atom-like nature of QDs.

In the past, the fabrication of these dots has focussed on the use of a lithographic and etching approach, however success has been limited because of the technological requirements of producing ultra small structures that are defect-free and that exhibit an abrupt carrier defining potential. Typically it is very difficult to precisely "carve out" QDs that work properly. Instead of using conventional integrated circuit techniques, many investigators have shown that they can "grow" QDs. These self-assembled or self-organized QDs emerge from the proper mixing of different thin-film semiconductors. For example if a silicon germanium (SiGe) thin film is deposited on a silicon (Si) thin film, the strain caused by the lattice mismatch between the different materials deposited on each other in the thin film causes the formation of more or less regularly spaced and sized clusters of material. Applying several layers of the SiGe thin film leads to a regular array of SiGe QD conical shaped clusters that are approximately 100 nm in diameter and approximately 3-10 nm tall. Other combinations of semiconductor thin films for creating QDs are possible, for example, including indium arsenide (InAs) quantum dots on a gallium arsenide (GaAs) substrate.

A common technique used for growing QDs in this way is the Stranski-Krastanow (SK) method, Stranski and Krastanow, Sitzungsber K Preuss Akad. Wiss., Phys. Math. Kl 146 (1937) 797. This method is consistent with modern epitaxy and fabrication techniques. Usually the aim of epitaxy is to create a single, smooth crystal with little defects. It is well known that the material deposited needs to be thicker than a critical thickness, otherwise strain caused by surface tension overpowers the crystal forces, and the layer forms instead as a lumpy distribution of crystals with many defects. Rather than avoiding this state, the SK method exploits this state to create uniformly distributed QDs. The SK method is more relevant for modern device fabrication than, creating nano-crystals through a wet chemical etching reaction or ablation. Many useful nano- structures, such as buckeyballs and carbon nanotubes, can be created using these techniques, but require new fabrication techniques, such as polymer matrixes, to hold the QDs and integrate them into useful devices. For comparison, the SK method is essentially 100% compatible with current device fabrication methodologies. Until p- and n- doped polymer manufacturing is mature, which may be many years off, the SK epitaxial method will typically be the preferred method of manufacturing QD devices. One challenge of the SK method is in controlling and refining the QD mixture. One major improvement has been to control the growth conditions of the SK method in order to optimize the formation of QDs to a particular size and shape. These techniques are often referred to as self-assembled QDs, in that the dot shape is controlled by the internal crystallization physics rather than through external control.

The above methods of creating QDs, ranging from wet chemistry to epitaxy, create a heterogeneous distribution of dot shapes and sizes. This random, broad distribution can limit the usefulness of dot mixtures. Instead of being a spectrally pure nano-material, as desired, the spectral purity can be worse than naturally occurring materials. While some methods seek to use the spectral impurity of dots as a benefit for example, tunable lasers, most of the promise of the use of QDs requires methods for refining or controlling the dot distribution.

While the purity of the self-assembled QDs can be sufficient to demonstrate lasers with properties similar to existing quantum well lasers, the purity is typically not enough to realize the benefits of the QD's discrete energy levels. The distribution of dot sizes in the laser material leads to an inhomogenous broadening of the linewidth, which is typically no better than the linewidth achieved in traditional quantum well structures. Further, since the dot size distribution is determined by the internal crystal physics, instead of external factors, small changes in the epitaxy equipment between different production runs can affect the material quality and yield. Self-assembled QD growth also has no control over the location of individual dots, which is a limitation for some potential devices, such as Qbits or single photon detectors. The basis of laser operation depends on the creation of non-equilibrium populations of electrons and holes, and the coupling of electrons and holes to an optical field, stimulating radiative emission. Calculations carried out by Dingle et al, U.S. Patent No. 3,982,207 predicted the advantages of using quantum wells as the active layer in such lasers. The carrier confinement and nature of the electronic density of states could result in more efficient devices operating at lower threshold currents than lasers with "bulk" active layers. In addition, it was determined that the use of a quantum well, with discrete transition energy levels dependent on the quantum well dimensions, provides a means of "tuning" the resulting wavelength of the material. The critical feature size, in this case the thickness of the quantum well, depends on the desired spacing between energy levels. For energy levels greater than a few tens of millielectron volts, the critical dimension is approximately a few hundred angstroms. The first QD laser, demonstrated by Van der Ziel et al, 1975 Appl. Phys. Lett. 26:463-465 was many times less efficient than a conventional laser. However the situation was reversed by Tsang 1982 Appl. Phys. Lett. 40:217-219 through the use of new material growth capabilities for example MBE, and the optimization of the heterostructure laser design.

Analysis by Sakaki et al, 1995 Appl. Phys. Lett. 67:3444 predicted that QD lasers would exhibit a performance that is less temperature dependent than existing semiconductor lasers, and that would not degrade at elevated temperatures. Other benefits of QD active layers include a further reduction in threshold currents and an increase in differential gain providing a more efficient laser operation. Stimulated recombination of electron- hole pairs takes place in the quantum well region, where the confinement of carriers and the optical mode enhance the interaction between carriers and radiation. The population inversion or the creation of electrons and holes, which is necessary for lasing, occurs more efficiently as the active layer material is scaled down from bulk (3 -dimensional) to QDs (0-dimensional). However, the advantages in operation depends not only on the absolute size of the nanostructures or QDs in the active region, but also on the uniformity of size of these QDs. It was found that a broad distribution of sizes "smears" the density of states, thereby producing behavior similar to that of a bulk material.

As such, the challenge in realizing QD lasers with an operation superior to that shown by quantum well lasers is the formation of high quality, uniform QDs in the active layer. Initially, the most widely followed approach to forming QDs was through electron beam lithography of suitably small featured patterns which are approximately ~30 nm and the subsequent dry-etch transfer of these dots into the substrate material. The problem realized with these QD arrays was their exceedingly low optical efficiency. The high surface-to-volume ratios of these nanostructures and the associated high surface recombination rates, together with damage introduced during the fabrication of the quantum dots, precluded the successful formation of a QD laser.

With the demonstration of the high optical efficiency, the self-assembled formation of QDs formed without the need for external processing and having the natural overgrowth of cladding material, there was an increase in QD laser research. The first demonstration of a QD laser with high threshold density was reported by Ledenlsov et al, 1994 Proc. ICPS-22 Vancouver 3:1855. Bimberg et al 1996 Jap Journal Appl. Phys. 35:1311-1219 demonstrated an improved operation by increasing the density of the QD structures, stacking successive, strain-aligned rows of QDs in addition to achieving vertical as well as lateral coupling of the QDs. In addition to utilizing quantum size effects in edge-emitting lasers, self-assembled QDs have also been incorporated within vertical cavity surface-emitting lasers.

Although the self-assembled dots have provided a stimulus to work in this field, there remains a number of critical issues involving their growth and formation, greater uniformity of size, controllable achievement of higher QD density, and closer dot-to-dot interaction range, wherein these parameters can further improve laser performance.

R.L. Williams et al, 2001, Journal of Crystal Growth, vol. 223, 321-331 have shown how a single row of self assembled InAs/InP QDs grown on triangular cross-section indium phosphide (InP) mesas have uniform spatial dimensions which are controlled by the mesa geometry.

Stinz et al, in U.S Patent Application Publication US 2002/0114367 Al has demonstrated self-assembled growth techniques for QDs in a semiconductor laser. This work covers success in selection of growth parameters to control the dot density and size distribution, the formation of a continuous optical gain spectrum and embodiments in tunable lasers and monolithic multi-wavelength laser arrays. Reithmaier et al, 2002, Journal of Selected Topics in Quantum Electronics, Vol 8, No 5 Sept/Oct pp 1035-1044 have also reported on the use of QDs in semiconductor lasers. These designs include lasers with distributed feedback, however this feedback is accomplished by means of physical corrugations external to the active region. Figure 1 shows such a laser, where the gratings have been etched in a layer, adjacent to the layer containing the QDs.

Wingreen, in U.S. Patent 5,963,571 has proposed a semiconductor cascade laser constructed from QDs. Such a laser would have all the desirable characteristics of a quantum- well laser, whilst at the same time eliminating the very undesirable non- radiative phonon decays.

While semiconductor lasers have been constructed including QDs, these designs have employed the QDs as an alternative to a semiconductor layer. In addition, all QD lasers made in the past have been made with the QDs grown with the SK method, which leads ^• to QDs of random size distributions. There is therefore a need for new semiconductor laser designs and associated fabrication techniques that incorporate the use of QDs of uniform size and predictable placement. In particular there is a need for new fabrication techniques that allow the exploitation of QD properties in optical devices, for example, in using them to make the gratings in semiconductor devices.

SUMMARY OF THE INVENTION

An object of the present invention is to provide semiconductor devices including gratings formed using quantum dots and a method of manufacture. In accordance with an aspect of the present invention, there is provided a semiconductor device having a grating of quantum dots, said semiconductor device comprising a wafer including one or more layers of material, said wafer having a major axis; a template formed on the wafer, said template defining a desired grating pattern; a layer of quantum dots on the template, said template controlling quantum dot location and size during formation thereof, thereby forming the grating of quantum dots having a depth; and an overgrowth layer of material enclosing the grating of quantum dots; wherein electrical contacts are formed on the semiconductor device thereby enabling activation thereof. In accordance with another aspect of the present invention there is provided a method for manufacturing a grating for a semiconductor device, said grating being formed on a wafer comprising one or more layers of material and said grating including a plurality of quantum dots, said method comprising the steps of: obtaining the wafer, said wafer having a major axis; depositing a template layer onto said wafer; patterning said template layer thereby forming a template defining a desired grating pattern; growing a layer of quantum dots on said template, said template controlling quantum dot location and size thereby forming a grating of quantum dots having a depth; depositing an overgrowth layer of material enclosing the grating of quantum dots. BRIEF DESCRIPTION OF THE FIGURES

Figure 1 shows a DFB laser using first order gratings in a ridge waveguide with a laser having quantum dots in the active region according to Reithmaier.

Figure 2 shows a diagram of a set of self-assembled quantum dots providing a gain- coupled feedback structure in the active region according to one embodiment of the present invention.

Figure 3A shows a quantum dot configuration wherein columns of quantum dots are regularly spaced according to one embodiment of the present invention.

Figure 3B shows a quantum dot configuration wherein the columns are formed at the top and bottom of the active layer according to one embodiment of the present invention.

Figure 3C shows a quantum dot configuration wherein the columns of quantum dots are formed in bunches according to one embodiment of the present invention.

Figure 4 shows the layers of a semiconductor laser with the active layer formed with quantum dots and in-fill according to one embodiment of the present invention. Figure 5A shows a 500 nm n-doped layer followed by a 40 nm AlGaAs SCH layer, followed by a 90 nm AlGaAs layer grown on an n-doped GaAs substrate, according to one embodiment of the present invention. Figure 5B shows a dielectric layer patterned with a gratings design, on the structure of Figure 5 A, according to one embodiment of the present invention.

Figure 5C shows the gratings pattern of Figure 5B etched into the 40 nm n-doped AlGaAs SCH layer of Figure 5 A, according to one embodiment of the present invention.

Figure 5D shows the removal of the dielectric layer used to pattern the grating structure of Figure 5C, according to one embodiment of the present invention.

Figure 5E shows QDs grown on the top surface of the gratings structure of Figure 5D, according to one embodiment of the present invention.

Figure 5F shows the growth of undoped AlGaAs on top of the structure of Figure 5E, according to one embodiment of the present invention.

Figure 5G shows the growth of a 40 nm p-doped AlGaAs SCH layer followed by a 20 nm n-doped GaAs layer on the structure of Figure 3F, according to one embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

The present invention provides a method for the manufacture of self-aligned quantum dot (QD) structures in optical devices using a template. These optical devices can be semiconductor lasers, waveguide devices, optical filters, Distributed Feedback (DFB) laser, Distributed Bragg Reflector (DBR) laser, band-pass filters and photonic bandgap structures and the like. This invention further provides a method of manufacture of semiconductor device gratings made from self-aligned QD structures using a template, whereby the QD grating structures can be substantially optimal for the operation of the device. In addition the present invention provides arrangements of these QD structures that provide distributed feedback in semiconductor lasers. Furthermore, the invention provides a method of manufacture of gain-coupled gratings without the need to etch the active layer of the laser. Depending on the design of the device, the invention provides a method of manufacture of a DFB laser, a DBR laser, a filter, and the like. In one embodiment the method of the present invention can also be used to produce QD structures in various locations within a device, for example, a QD structure in-filled with bulk or quantum well material can be produced to form the entire active region of a semiconductor laser.

When quantum dots are formed in a corrugated pattern similar to the gratings in present DFB and DBR lasers for example, they can function in a similar, but improved way. The invention provides a new method of constructing these gratings using QDs in which the size and distribution of the QDs is controlled by patterning the substrate with a template prior to growth of the QDs using the SK technique. By using a template, the QDs can be arranged in a layer such that the optical field overlaps favorably with the dots. This feature can enhance the optical interaction with QDs, and may provide substantially optimum performance of the semiconductor device.

Employing a template has many advantages over traditional self-assembled QD growth techniques. The location and size of the QDs can be controlled, since the growth of the QDs will follow the template, and the template can be formed using standard semiconductor fabrication techniques. In addition, the technique of a template for the growth of QDs has advantages in semiconductor laser applications, for example, as it can enable the generation of different types of gratings as well as various desired QD arrangements within the lasers.

In one embodiment of the present invention the photons are constrained to overlap the QDs grown on the template, making the resulting structure a DFB structure that creates a standing wave pattern in the cavity, with the photons overlapping the QDs.

According to the present invention, a semiconductor device can be fabricated in which the QD size and location may be precisely controlled using the template, without a problem of low dot density. The invention allows control of the QD size by suitable design of the template achieving accurate positioning of the QDs. In addition the invention provides a method for the building of walls of QDs through repeated epitaxial growth to form a grating in a semiconductor device.

Having particular regard to laser design, the use of QDs provides distinct advantages over other laser designs. For example, in existing semiconductor lasers, the etching of the active region to make gain-coupled gratings can reduce the reliability of the laser. The use of QDs in the active region can allow distributed reflecting structures to be provided and can obviate the need for etching, and hence can improve the reliability of the laser. In addition, some materials, such as AlGalnAs, typically do not easily lend themselves to gain-coupled grating fabrication in quantum well lasers, however with the method of fabrication according to the present invention, these types of materials can produce gain-coupled gratings for lasers.

In one embodiment, an advantage of using uniform size QDs in lasers is that they can provide a wider operating temperature. Since semiconductor lasers have a continuum of energy states that follow a Fermi-Dirac or Maxwel-Boltzman distribution, the carrier concentrations at each energy level can have a strong temperature dependence. This can cause the laser to operate differently at different temperatures. QDs however, are different in that they have discrete energy levels determined by the size of the dot and therefore may not be temperature dependent. When the distribution of dot sizes is non- uniform, the lasers incorporating them face the same problem of temperature dependence as conventional semiconductor lasers, however if the QD dot size is more uniform, as can be provided by the present invention, the laser can operate with relatively less dependence on temperature. This property therefore may provide a technology platform for InP coolerless devices for example.

Template and Quantum Dot Fabrication

A key parameter in the design of a QD semiconductor device is the dot size distribution. In a normal SK distribution of QDs, a broad range of sizes is typically obtained. By using a template that is small enough in size (on the order of the desired dot size), the dot size can be controlled by the template. For example, if the width of the template is 10 nm, dots with a width greater than 10 nm will not be able to form. Thus the template provides a method of producing single sized QDs in a self-aligned fashion that can be useful for making active optical devices. The method of fabrication of semiconductor devices employing embodiments of the present invention follows the standard methods of fabrication of particular semiconductor device, with the exception of the layers including QDs where a repeated epitaxial process employing a template may be used to create layers of QDs to define the grating.

The template ridges are spaced by a particular amount dependent on the design of the device, for example, the spacing may be a multiple of the lasing wavelength, thus causing the resulting QD pattern to self-align into a structure with similar spacing. Present fabrication techniques allow the minimum width of the ridges to be about 10 nm, and the spacing between ridges to be about 100 nm. Further improvements in epitaxial and lithographic techniques may allow improved accuracy and capability of the templates, thereby potentially further enhancing the use and applicability of the present invention.

Template ridges can be fabricated using a number of standard semiconductor processing techniques. For example, the template material can be grown using an epitaxial process such as MBE however, other techniques for the creation of the template would be readily understood by a worker skilled in the art. The template is subsequently patterned using, for example a process involving the use of a dielectric layer as a mask, which may be patterned using electron-beam lithography or holographic lithography, stepper masks or other patterning techniques as would be readily understood. A process such as PEC D can be used for deposition of the dielectric material. Standard wet or dry techniques can then be used for etching the pattern into the template material. Other variations and methods of producing the template may also be used such as the use of a selective area growth (SAG) process. In one embodiment of the present invention, the template ridges can be patterned parallel to the major flat axis of the wafer, however alternate orientations of the template pattern may also be used as desired for example. Examples 1 and 2 provide process steps that may be used to fabricate the template according to one embodiment of the present invention.

The standard SK method of growth can then be used to grow the QDs, which can self- align into a grating structure corresponding to the template design and size of the template line width. The QDs will typically only grow on the (001) surface of the template and therefore material deposited on the other surfaces will typically migrate to the (001) surface, thereby enabling the growth of walls. For example, referring to Figure 2, if a template mesa is oriented so that the sidewall 21 is a {011} plane, that is any plane that is of the same family as the (011) plane, and the top 22 is the (001) plane, quantum dots will only grow on the template mesa top 22, not the sidewalls 21. Overgrowing the first layer of quantum dots with the same template mesa material 23 will provide a template mesa where the top 24 is the (001) plane and the sidewalls 25 will be the {011} plane. As before, dots will only grow on top 24 of the mesa. Repeating these steps can produce a vertical wall 26 of self-assembled quantum dots as illustrated in Figure 2. This process can be limited by the vertical strain that propagates with the SK method however, this strain can be minimized using techniques as would be readily understood by a worker skilled in the art. The depth of the gratings determined from typical depth and size of QDs, requires stacks of QDs in order to form gratings of sufficient depth. Since a typical grating is usually 50 to 100 nm deep for a gain coupled grating, approximately 10 layers of QDs would be required to form this size of grating. As would be readily understood, fewer layers of QDs would be required to form a grating for use with an index coupled grating as this type of grating can be in the range of 40 nm deep. For other semiconductor devices, the required depth and subsequently the number of layers of QDs required for a desired grating, would be readily understood by a worker skilled in the art.

Once the QD growth is complete, the template/QD structure is given a final overgrowth to fill in the template. The top surface can then be planarized for deposition of the subsequent layer of the semiconductor device.

Variations

The desired function of the self-aligned QD structure in a semiconductor device can vary depending on the design of the device. For example, in one embodiment, the QD structure can be designed for use in a DFB laser and in another embodiment the QD structure can be designed for use in a DBR laser, and in a further embodiment the QD structure can be designed for use in an optical filter. In one embodiment, in which the QDs grown on the template are designed to have gain at the operating wavelength, a gain-coupled grating can be formed using the template and self-aligned QDs. In such a device, the QDs may be used in conjunction with a quantum well or bulk active region and the QDs used can form a low loss, highly reliable gain-coupled grating. In another embodiment, the QDs can also form the entire active region of the laser.

In another embodiment, the QDs can be chosen such that they do not have gain at the desired wavelength but have a different iridex of refraction from the surrounding material, in this case the QDs then form an index-coupled grating. This grating can be above or below the active region depending on when during the fabrication process the template is used to grow the QDs. In both the gain-coupled and index-coupled cases, the optical properties of the QDs can be tuned by controlling their size through the lateral dimension of the ridges upon which the template is formed and can provide an extra degree of freedom to a laser designer. Controlling the index contribution of the QDs can result in stronger or weaker index-coupled gratings than are currently possible using present techniques. Controlling the gain of the QDs through current injection can make a grating that bleaches as more or less current is applied thereto, for example a grating that disappears at one current or power level, but is present at other current or power levels.

In one embodiment of the present invention, the design of the grating can be used for the control of the output power of a laser. For example, below a certain input power the output power is linear and above a critical input power the output power is limited to a particular level or even reduced.

The tolerances achieved in the creation of uniform QD lasers using epitaxial techniques can be compatible with the grating dimensions needed for 1.3 and 1.55 μm lasers, for example while applicability to other grating dimensions would be readily understood. Thus the same methods used to create uniform quantum dots epitaxially can also serve to generate gain-coupled gratings within the active region. In existing semiconductor lasers with gratings, the grating is generally etched with the reflecting members on one side of the active layer. By using QDs formed using the method according to the present invention, there can be a greater flexibility in the placement within the active layer as well as the size and shape of the gratings obtainable. For example, the QDs can be spaced at various desired intervals within the active layer. In addition, the QDs may be distributed with varying longitudinal intervals to provide chirp. The QDs can also be distributed in longitudinal groups. Different shapes, including circles, ellipses and wires are also obtainable with the use of QDs fabricated using the method of the present invention. For example, having particular regard to the formation of quantum wires, modification of the growth conditions, and in particular for example changing the growth temperature to move from the quantum dot regime to the quantum wire regime can be performed, as would be readily understood by a worker skilled in the art. Furthermore, having particular regard to the formation of quantum circles or ellipses, modification of predetermined growth conditions can be used in order to account for the varying crystallographic plane around the circle or ellipse. In particular the formation of a quantum circle can provide a means for the manufacture of photonic bandgap materials, for example.

Examples of various QD configurations are shown in Figure 3. Figure 3 A shows columns of QDs 31 with fixed intervals between each column in a cross-section 30 of a device. Figure 3B shows groups of QDs 33 in a cross-section 32 of a device where the QDs have varying spacing in the vertical axis and fixed spacing in the horizontal axis. Figure 3C shows another configuration of QD groups 35 in a cross-section 34 of a device where the QDs have fixed spacing in the vertical axis and varying spacing in the horizontal axis. It would be obvious to a worker skilled in the art that a large variety of such configurations may be implemented with variations in for example, location, size, stacking and cluster shape of QDs.

Note that, whilst the use of QDs in the construction of gratings as a mode of distributed feedback in a semiconductor laser is an important application, the invention may be applied to the use of QDs in other configurations that may be desired. Whether the desired affect is to have the QDs behave as a grating, or just to maximize the optical interaction with a layer of dots, the use of a template can be used to controlling the placement and size of the QDs according to the present invention. In addition, the use of a template to grow QDs can provide considerable flexibility to enable growth of QDs in different locations within the entire semiconductor structure. For example, in existing DFB lasers the gratings are situated on one side of the active layer, however, using the method of the present invention, the QDs can be grown in other locations as well. Figure 4 shows a semiconductor laser structure in which a layer 44 with photoluminescence (PL) of 1.3 μm InGaAsP QDs infilled with a PL of 1.1 μm InGaAsP forms the active region of the laser between a 20 nm Si doped p-type InP confinement 43 and a 20 nm Zn doped p-type with a PL of 1.1 μm InGaAsP confinement 45.

When the QDs are fabricated within the active region, the accuracy of the dot size can typically be less important compared to the position of the dots within the active region. Thus, a variation in the spacing between the walls of the QDs can degrade the performance of the laser. The epitaxial process, which can be used in a number of stages in the semiconductor laser fabrication process, is a repeatable method, and is important in obtaining QDs with minimal variation in the spacing between them.

The type of materials used for the QDs and template/active layer can include, for example, Gao. Irto_.6As QDs grown on a GaAs template, and InAs QDs embedded in a GalnAs active layer. It would be readily understood by a worker skilled in the art that the composition of the QDs can equally comprise alternate percentages of Ga and In. Another example of a material system that can be used in association with the present invention is the use of a GaAs substrate with AlGaAs barriers and InAs QDs. The choice of template used can depend on the type of semiconductor device being fabricated using QDs. Broad classes of devices may all derive from similar generic templates. For example, many laser devices, including 980 nm pump lasers, 1310 & 1550 nm modulated lasers, 1550 high power lasers and tunable lasers use a DFB grating to achieve their performance.

Regardless of the final design intent, for example the type of semiconductor device or whether a DFB grating is gain- or index-coupled, above, below or within the active region, the template generation and QD growth process can have a similar basic method. Firstly, a template material is deposited. Then a periodic pattern, using for example e- beam or holographic lithography, is transferred to the surface of the template using standard lithographic materials. This pattern is then etched into the template material in a controlled manner. The exposed surface is cleaned to create a suitable surface for epitaxial growth, and then overgrown with epitaxial material that has a difference in gain and/or refractive index from the substrate material thereby resulting in the growth of QDs at the desired locations represented by the template pattern.

For a better understanding of the invention described herein, the following examples are set forth. It should be understood that these examples are for illustrative purposes only. Therefore, they should not limit the scope of this invention in any way.

EXAMPLE 1: One example of the present invention is described in the fabrication of a 1310 nm laser and is illustrated in Figures 5 A to 5G. In this example the creation of only a single layer of dots is described, however multiple layers of dots can be fabricated by repeating the template fabrication and QD growth steps. Two methods for making the template are included. One method uses a wet chemical etch, and the other method uses selective area growth (SAG). However, other methods, such as dry etching, may also be used for template creation. 1. Starting with an n-doped GaAs (001) substrate 50, a 500 nm n-doped (lel8/cm³), GaAs buffer layer 51 is grown. The buffer layer 51 thickness is chosen to be sufficiently thick that any defects present on the substrate have been overgrown and the top surface is suitably defect free for epitaxial growth. 2. A 20 nm n-doped (2el7/cm³) layer of unstrained AlGaAs with a photoluminescence (PL) of 1.0 μm is grown followed by a 20 nm n-doped (2el7/cm³) layer of unstrained AlGaAs with a PL of 1.1 μm. Together, these layers form the bottom half of the separate confinement heterostucture (SCH) 52, which confines the charge carriers and the photons within the active region. The SCH described is suitable for lasing action in this system, but is not optimized. Suitable optimizations to the SCH, such as using thinner or thicker layers, using continual grading, or more layers with small PL steps, would be obvious to one skilled in the art. 3. Next, a 90 nm n-doped (2el7/cm³) AlGaAs layer 53 with a PL of 1.1 μm is grown. Note that this is the material etched away to form the template for the QD growth. The thickness depends on the details of the template etch process. In this example the template will be etched 90 nm deep. The above three layers deposited on a GaAs substrate 50 are illustrated in Figure 5 A.

4. A dielectric layer is then deposited and patterned using a holographic gratings process. In this example a first order grating centered at 1.3 μm (roughly 200 nm, but depends on the actual material refractive index, which could vary depending on the growth details) is created. This results in a pattern of dielectric strips 55 with a width 57 of approximately 20 nm and a separation 56 of approximately 200 nm between each strip as illustrated in Figure 5B. Therefore, the grating will be 90 nm deep, with a 20 nm flat spot on top of each tooth (a mark-to-space ratio of 0.1) once etched. The grating may not be limited to being made with a holographic grating process, for example E-beam lithography can be equally suitable, although may be more expensive and time consuming. Additionally any other method of patterning may be used to define the grating pattern.

5. Next, the grating is fabricated by removing unwanted portions of the AlGaAs layer 53 for which any standard process can be used. Since one is etching through AlGaAs, an in-situ etch technique can be used or any standard wet or dry etch plus a suitable post etch clean can be sufficient. The key however, is to etch the material along the {111} or {011} side walls. An example of one such wet etch solution is H₂0:AgNO₃:CrO₃:HF (10ml:40mg:5g:8ml), which creates {111} facets in GaAs/ AlGaAs systems (A. R. Clawson, Materials Science and Engineering, vol. 31, pp. 1-438, 2001; P. P. Demeester et al., J. Appl. Phys., vol. 63, pp. 2284-2290, 1988). The resulting structure is shown in Figure 5C which illustrates the exposed {111} facets 58.

6. The patterned dielectric layer 55 used as a mask for the fabrication of the grating is then removed as illustrated in Figure 5D to expose the clean (001) surface on top of the etched mesas followed by a native oxide removal process using an H SO₄ solution, for example, resulting in the creation of the desired template pattern.

7. The InAs quantum dots 59 are then formed as illustrated in Figure 5E using the SK method. Since the only surface suitable for growing QDs is typically on top of the templates, the quantum dots 59 can form lines that have the desired spacing for forming a gain-coupled grating, for example. In addition the width of the template can control the dot size. This results in a line of quantum dots with uniform size forming a gain-coupled distributed feedback grating structure, transverse to the direction of light propagation.

8. After the growth of the QDs 59, the region between the mesas is filled with undoped AlGaAs 591 with a PL of 1.1 μm. The depth of AlGaAs is such that the 90 nm deep trenches are filled as illustrated in Figure 5F.

9. The p-side SCH 592 is then grown as illustrated in Figure 5G. This comprises of 20 nm p-doped (2el7/cm³) AlGaAs with PL of 1.1 μm, followed by 20 nm p- doped (2el7/cm³) AlGaAs with PL of 1.0 μm. Following the p-side SCH layer 592, a 20 nm p-doped (2el7/cm³) GaAs layer 593 is grown.

10. Lastly, a thin etch stop layer can be deposited, which is not illustrated. The composition of the etch-stop layer can be dependent on the etch technique used to form the ridge, which is the subsequent component of the laser structure formed. For example, if a wet etch chemistry of HNO₃:H₂O (1:200) is used to etch the ridge through the GaAs, then a 4 nm layer of AlGaAs with PL of 1.0 μm would be a suitable etch stop layer.

11. The remainder of the laser can subsequently be fabricated using standard techniques. The remaining fabrication steps can include a dielectric deposition step, and via clear step, a p-metal deposition step, an n-side thin and polish step, and finally an n-side metal step.

Note that template size and therefore dot size may be adjusted to obtain the desired PL of 1.3 μm. A 20 nm dot size can be sufficient, however subsequent adjustment of the QD size by adjustment of the template design may be required for "dialing in" the correct PL.

EXAMPLE 2: In another embodiment, the template can be formed using selective area growth (SAG). The grating pattern can be deposited as in step 4, but with a mark-to-space-ratio (57/56) of 0.5, for example. SAG can then be used to grow AlGaAs mesas with sidewalls oriented on the {111} or {011 } plane. Growth would continue until the (001) surface on top is eliminated (P. Legray et al, J. Crystal Growth, vol. 150, pp. 394-398, 1995), however, in this case growth is be stopped before the (001) surface is eliminated, and timed so that the top of the mesa is approximately 20 nm. For this method the dielectric is not removed since the dielectric is between the SAG grown mesas and covering the exposed (001) plane in the grating trenches. The dielectric will be removed after the dots are grown. Note that there may be methods, for example using a sacrificial Q- layer, that can be used to facilitate dielectric removal from the trenches. This technique however will require additional growth steps before this step.

The invention being thus described, it will be obvious that the same may be varied in many ways. Such variations are not to be regarded as a departure from the spirit and scope of the invention, and all such modifications as would be obvious to one skilled in the art are intended to be included within the scope of the following claims.

Claims

I CLAIM:

1. A semiconductor device having a grating of quantum dots, said semiconductor device comprising: a) a wafer including one or more layers of material, said wafer having a major axis; b) a template formed on the wafer, said template defining a desired grating pattern; c) a layer of quantum dots on the template, said template controlling quantum dot location and size during formation thereof, thereby forming the grating of quantum dots having a depth; and d) an overgrowth layer of material enclosing the grating of quantum dots; wherein electrical contacts are formed on the semiconductor device thereby enabling activation thereof.

2. The semiconductor device according to claim 1 further comprising: a) a further template overgrown onto the grating of quantum dots; and b) a subsequent layer of quantum dots on said further template thereby increasing the depth of the grating of quantum dots.

3. The semiconductor device according to claim 1 wherein said semiconductor device further comprises one or more additional layers of material grown on the overgrowth layer.

4. The semiconductor device according to claim 2, further comprising a plurality of templates and a plurality of layers of quantum dots thereby creating a desired grating depth.

5. The semiconductor device according to claim 1, wherein said semiconductor device is selected from the group comprising a semiconductor laser, a waveguide device, an optical filter, a DFB laser, a DBR laser, a band-pass filter and a photonic bandgap structure.

6. The semiconductor device according to claim 5, wherein said semiconductor device is a semiconductor laser and further comprises a plurality of templates and a plurality of layers of quantum dots thereby creating a desired grating depth.

7. The semiconductor device according to claim 6 wherein the desired grating depth and position thereof provides an index-coupled grating.

8. The semiconductor device according to claim 6 wherein the desired grating depth and position thereof provides a gain-coupled grating.

9. The semiconductor device according to claim 1 wherein the desired grating pattern is parallel to the major axis of the wafer.

10. The semiconductor device according to claim 1, wherein the grating of quantum dots is configured as a chirp grating or a segmented grating.

11. The semiconductor device according to claim 1 wherein the desired grating pattern is off of the major axis of the wafer.

12. The semiconductor device according to claim 6 wherein the quantum dots are GalnAs and the template is GaAs.

13. The semiconductor device according to claim 6 wherein the quantum dots are InAs.

14. The semiconductor device according to claim 13 wherein the template is AlGaAs.

15. A method for manufacturing a grating for a semiconductor device, said grating being formed on a wafer comprising one or more layers of material and said grating including a plurality of quantum dots, said method comprising the steps of: a) obtaining the wafer, said wafer having a major axis; b) depositing a template layer onto said wafer; c) patterning said template layer thereby forming a template defining a desired grating pattern; d) growing a layer of quantum dots on said template, said template controlling quantum dot location and size thereby forming a grating of quantum dots having a depth; e) depositing an overgrowth layer of material enclosing the grating of quantum dots.

16. The method for manufacturing a grating for a semiconductor device according to claim 15, wherein prior to the step of depositing an overgrowth layer of material, said method further comprises a repeatable sequence of steps including: a) overgrowing said layer of quantum dots with a further template; and b) growing a subsequent layer of quantum dots on said further template thereby increasing the depth of the grating of quantum dots; wherein said repeatable sequence of steps is repeated until a desired depth is obtained.

17. The method for manufacturing a grating for a semiconductor device according to claims 15 or 16, wherein said template and said further template are formed by an epitaxial process.

18. The method for manufacturing a grating for a semiconductor device according to claim 17, wherein said epitaxial process is molecular beam epitaxy (MBE).

19. The method for manufacturing a grating for a semiconductor device according to claim 15, wherein the step of patterning said template layer comprises the steps of: a) depositing a dielectric mask onto said template layer; b) patterning said dielectric mask with the desired grating pattern using a patterning technique; and c) etching said desired grating pattern into the template layer.

20. The method for manufacturing a grating for a semiconductor device according to claim 19, wherein the step of patterning said dielectric mask is performed by a technique selected from the group comprising electron-beam lithography, holographic lithography and stepper lithography.

21. The method for manufacturing a grating for a semiconductor device according to claim 15, wherein said desired grating pattern is parallel to the major axis of the wafer.

22. The method for manufacturing a grating for a semiconductor device according to claim 15, wherein said desired grating pattern is off of the major axis of the wafer.

23. The method for manufacturing a grating for a semiconductor device according to claim 15, wherein said semiconductor device is a laser and said grating of quantum dots is configured as an index-coupled grating.

24. The method for manufacturing a grating for a semiconductor device according to claim 15, wherein said semiconductor device is a laser and said grating of quantum dots is configured as a gain-coupled grating.

25. The method for manufacturing a grating for a semiconductor device according to claim 15, wherein the grating of quantum dots is configured as a chirp grating or a segmented grating.