A SEMICONDUCTOR LASER
This invention relates to a semiconductor laser.
Semiconductor lasers require a direct band gap material in which recombination of electron-hole pairs promotes photon emission. Photons of a particular wavelength are emitted. Two important factors govern the selection of direct band gap semiconductor materials suitable for lasers. Firstly the lattice matching of materials, from which the layers of the laser are made, is important. Secondly, is the choice of materials which determine the energy of the photon of energy emitted upon electron-hole pair recombination and this in turn determines the colour of light emitted by the laser or LED.
Compounds such as zinc selenide (ZnSe) and cadmium telluride (CdTe) have been found to be suitable types of materials for semiconductor laser materials. However, these materials were only able to be doped p or n type. A problem has been to create a p-n junction capable of supplying sufficient electrons and holes (minority carriers) to create a population inversion so that lasing can occur. The doping autocompensation mechanism prevents one from trying to dope the semiconductor to the opposite type. A paper published in the Journal of Luminescence, (Vol. 16 1978 at pages 363 to 394) entitled THE NATURE OF THE PREDOMINANT ACCEPTORS IN HIGH QUALITY ZINC TELLURIDE, reported that very sharp excitation spectra in high quality melt grown Zinc Telluride were demonstrated.
A primitive type of LED capable of operation in the ultraviolet (UV) range, has been developed by Isamu Akasaki et al. of Nagoya University of Japan. Details of the device appear in APPLIED PHYSICS LETTERS 64(11) 14 March 1994 at pages 1377 to 1379. The device, however, suffered from a number of drawbacks. Firstly the device had a limited life span; secondly it had low reliability; thirdly it was not able to act as a true semiconductor laser; and fourthly the device was not very efficient - typically only 0.8% output efficiency. Most of the aforementioned disadvantages arose as a result of the very high internal resistance of the device which required a high drive current, typically around 12mA. It has been suggested that by forward biasing a p-n junction it would be possible to
inject a greater number of minority carriers into an appropriate layer of a direct band gap semiconductor material and therefore achieve efficient photon emission. A report, in the JAPANESE JOURNAL OF APPLIED PHYSICS (Vol. 29, No. 2 February 1990 at pages 205 to 206) stated that type conversion of Gallium Nitride (GaN) was achieved and fabrication of a near ultraviolet laser was possible by growing a GaN film on a Sapphire substrate. The purpose of the type conversion was to turn normally n-type GaN into p-type GaN. The GaN film was able to be grown on a Sapphire substrate by using an Aluminium Nitride (A1N) buffer layer. However, lasing was achieved using a pulsed nitrogen (N2) pump laser source, having a peak power of 250 kW, a pulse length of 10 nanoseconds (ns) and a wavelength of 337.1 nanometers (nm). The authors acknowledge that the results show the possibility of a current- injection-type UV laser diode, but do not describe how such a device may be manufactured.
An article in LASER FOCUS WORLD (1994) at pages 18 to 23 describes an LED which was also demonstrated at Nagoya University, Japan. The device comprised a double-heterostructure which was electrically driven, but again suffered from a high contact resistance. The type conversion used low energy electron irradiation after the layer was doped with magnesium. The type conversion mechanism was not known.
The device was shown to operate in laboratory environments and the lifetime of the laser was very short and could not be predicted with any degree of reliability. It is known that 3-5 nitrides, such as Indium Nitride (InN) and Gallium Nitride
(GaN), tend to be highly n-type doped. A problem has been to achieve sufficiently high hole injection into an active layer using these materials. The high n-type doping is reported to be due to a large number of nitrogen vacancies. This is possibly true and the fact that type conversion cannot be readily achieved has been attributed to an auto-compensation mechanism as found in 2-6 materials. The mechanisms which prevent type conversion are not critically important for the production of a light emitter. However, an important issue is the injection of holes into n-type doped material.
Lasing has been achieved in 3-5 nitride materials, but not by injection electroluminescence. Gallium Nitride (GaN) has been made to lase by electron injection using an electron beam. The principle problem however, with the 3-5 nitrides for
conventional semiconductor lasers, is the fact that, like the 2-6 compounds, they are naturally doped either p-type or n-type and type conversion is not easily achievable. Gallium Nitride is normally highly n-type doped. Such materials need to be able to be type converted, so that p-n junctions can be made. This has not proved to be a straightforward matter or one which is easily soluble. In a paper published in the Japanese Journal of Applied Physics (Vol. 28 No. 12, Dec 1989 at pages 2112 to 2173), entitled "P-TYPE CONDUCTION IN MAGNESIUM DOPED GaN TREATED WITH LOW ENERGY ELECTRON BEAM IRRADIATION (LEEBI)", a similar device is described in greater detail. However, because the p-type layer is poorly doped there is a tendency for the LED to deteriorate and this is one reason which has contributed to the lack of success of the device. An article in IEEE Journal of Quantum Electronics, Vol. QE-9, No. 2, entitled "ANALYSIS OF PROPOSED MIS LASER STRUCTURES" by Kameda et al. discloses a metal insulator semiconductor sandwich for use as a pulsed laser.
There has therefore been only limited success in achieving an electrically pumped semiconductor LED which has been able to provide a reliable source of radiation in the ultraviolet region (i.e. around 350 nm).
The present invention arose from a consideration of the above mentioned problems and with a view to producing a solid state semiconductor light emitting source, preferably able to operate through the whole of the visible spectrum and into the ultraviolet (UV) region.
According to one aspect of the present invention there is provided a semiconductor laser comprising a substrate layer supporting a first surface of a buffer layer, a second surface of said buffer layer contacting an active layer, said active layer being adapted to supply electrons and holes capable of recombining to produce photons, and said active layer having at least one electrical contact for the supply of minority carriers for promoting recombination so as to produce photons characterised in that the substrate layer comprises a sapphire material.
Preferably a reflective layer, formed from a material whose gain is at least 1 /t (m" * ), (where t is the thickness of the material in metres) overlays said active layer and is arranged to provide Fresnel reflection.
o-
At this point it is worth mentioning that a Schottky diode would not work. This is because the Schottky diode is a majority carrier device and does not inject minority carriers into the underlying semiconductor substrate. However, a Metal Insulator Semiconductor (MIS) structure is capable of injecting minority carriers into an underlying optically active semiconductor layer, provided a suitable MIS device is manufactured to a high quality grade.
Preferably the buffer layer comprises Aluminium Nitride; and the active layer comprises a 3-5 material. An example of a 3-5 material is Indium Gallium Nitride (InGaN). A capping layer may be provided. Preferably the capping layer comprises a plurality of holes or perforations formed therein. The capping layer might form an electrical contact layer.
The relative thicknesses of these layers is important and may be selected such that the active layer is preferably between 0.4 to 1.0 μm thick; and the Aluminium Nitride (AIN) (buffer) layer is preferably between 0.2 - 0.5 μm thick. A capping layer may be interposed between the active layer and a first metal contact, said capping layer is preferably between 0.05 to 0.15 μm thick. The capping layer is preferably Aluminium Nitride (AIN).
It will be appreciated that, although reference so far has been made to the reflective layer being formed from a material having a gain of at least 1/t (m"'), where t is the thickness of the material in metres, such as Gallium Nitride, other materials, for example Indium Tin Oxide (ITO) or Tin Oxide (TO) could be incorporated as the reflective layer into the laser. An advantage of the aforementioned two materials, namely ITO and TO, is that they are both substantially transparent to UV light. Accordingly the laser could be used in conjunction with other optical devices to act as a light valve or optical switch.
The present invention offers advantages in direct electrical driving and does not need an optical pump source. Accordingly the laser may have a direct, electrical pump source. However, an incoherent biasing light source may be provided for supplying a suitable means for enhancing electron-hole pair production. The present invention will now be described, by way of example only and with
reference to the Figures in which:
Figure 1 is a diagrammatical section through an embodiment of the invention;
Figure 2 is a graph showing the relative refractive indices of layers and waveguiding properties of the structure shown in Figure 1 ;
Figure 3 is a sectional view of an alternative embodiment of the invention;
Figure 4 is a plan view of the device in Figure 3; and
Figures 5 to 5c illustrate different types of arrays of holes which may be fabricated. Referring to the Figure 1 a laser, shown generally at 10, offers the possibility of efficient photon emission throughout the whole of the visible spectrum and into the ultraviolet (UV) region. That is at wavelengths extending from around 350 nm. The laser
10 operates by injection electroluminescence using a Metal Insulator Semiconductor and a metal insulating structure (MIS). The MIS comprises a metal layer 24a, b and c, an insulating layer 22 and an active semiconductor layer 16. The laser 10 also has a wide band gap semiconductor layer 16. The laser 10 also fulfils the requirements of a double heterostructure formed by insulating layer 22, semiconductor layer 16 and dielectric layer
18. The double heterostructure confines both carriers and photons (not shown) in an active semiconductor layer 16 (which becomes the active layer) and thereby increases photon production efficiency.
In order to fabricate an MIS laser diode the following layers need to be grown. Firstly a Sapphire substrate 20 with a first buffer layer of Aluminium Nitride (AIN) 18 of dielectric material is grown. This has already been achieved by in a number of applications, for example as demonstrated at the University of Nagoya, Japan and referred to above. Manufacture of a similar device is described in APPLIED PHYSICS LETTERS, 64(1 1) March 1994 at pages 1377 to 1379. Although the device described comprised similar layers to the present invention it could not be considered as a directly electrically driven laser and the publication is only referred to for purposes of disclosing an example of a suitable method of producing an Aluminium Nitride layer on Sapphire. Good quality single crystal AIN can be grown on sapphire by using this and similar methods.
Fortunately the 3-5 nitride system provides a good quality dielectric material. Aluminium Nitride (AIN) has a large (6.28 eV) band gap and is therefore an effective dielectric. AIN is both compatible with and well lattice-matched to the direct band gap 3-5 nitride alloys. As has already been shown, it is possible to grow AIN as a buffer layer 18 on layer 20 which is a Sapphire substrate and enables the growth of a good quality GaN layer 16 thereon. A metal layer which is deposited on a surface of the insulating layer 22 can be of any suitable metal and when etched reveals electrical contacts 24a, 24b and 24c. Although many metals are suitable Aluminium electrodes are preferred. Gallium nitride (GaN) is a direct band gap material with a band gap energy of 3.45 eV and Indium Nitride (InN) is also direct band gap with an energy gap of 1.95 eV. The alloy mixture InGaN is miscible for all alloy compositions and is also direct band gap for all compositions. Thus an InGaN alloy system is able to lase across the visible spectrum and into the ultraviolet. Furthermore judicious selection of relative amounts of materials, enables the output wavelength of the laser to be controlled. Aluminium Nitride (AIN) buffer layer 18 has three functions. Firstly, because of its larger band gap than active layer 16 it confines both carriers and photons to the active layer 16. This is known as normal heteroj unction action. However, the AIN buffer layer 18 has also been shown to be an excellent buffer layer for the growth of good quality GaN layers, as mentioned above. So that subsequently grown/desposited active layer(s) are of sufficiently good quality, the AIN layer 18 should be at least 200 nm thick, although thicker layers of order 0.5 μm have been found to produce better quality GaN layers. There is then deposited an optically active layer 16. This optically active layer 16 is around 0.4 μm to l .Oμm thick. A dielectric "capping" layer 22 (such as a "3-5" material e.g. AIN) is then deposited and this needs to be relatively thin so that only small voltages (typically 4 or 5 volts) are needed to invert the active layer surface. Thicknesses will be 100 nm or less typically. However, the capping layer 22 should be thick enough so that pinholes and other defects do not cause problems. Thus the thickness of capping layer 22 is in the region 50 - 150 nm. Finally an aluminium deposition 24 is etched to form a contact 24b of an electrode. All layers 16, 18, 20 and 22 are able to be deposited to the thicknesses shown using conventional chemical vapour deposition CVD or MOCVD techniques.
Figure 2 illustrates that the refractive indices of the relative layers 16 to 22 are such that photon confinement will occur within the InGaN (active layer) which is one criterion necessary for an efficient laser structure. If the laser 10 is now turned from reverse bias into forward bias, by reversing the bias voltage, the depletion region collapses and holes collected are injected into the underlying n-type doped direct band gap semiconductor layer 16. The laser 10 is formed by layers 18 and 22 and layer 16. A depletion region is in the same location as the inversion layer i.e. at the interface between layer 22 and layer 16. Photon emission also occurs at this interface. This results in photon emission at the semiconductor/dielectric interface. Although reference has been made to the use of direct band gap 3-5 nitride materials in the preferred embodiment, it will be appreciated that the same approach is applicable to all semiconductor materials in which type conversion has proved difficult or impossible. The largest group of direct band gap semiconductors in this category are the 2-6 compounds. From the above it is apparent that the laser or laser LED may operate in a pulsed mode. It is conceivable that the laser could also operate in a quasi Continuous Wave (C.W) mode, provided a suitable bias light is provided. Further it will be appreciated that although reference has been made to a directly electrically pumped device it may also be optically pumped if required. Figures 3 and 4 show different views of an alternative embodiment of the laser, in which one or more holes 30 are fabricated in an overlying layer such as Indium Tin Oxide (ITO). The holes 30 provide Fresnel reflection and therefore remove the need for providing separate mirrors or polished end faces.
The reason why this approach has an advantage is because the growth of GaN (InGaN) results in a "Wurtzite" type structure which does not have a simple cleavage plane which allows fabrication of end mirrors by the usual procedure of cleaving. This means that end mirrors, for a longitudinal laser, are more difficult and therefore more costly to fabricate. Cleaving however can be done by sawing end faces of the crystal and then polishing them afterwards. It might also be possible to ion etch end faces of sufficient quality for lasing.
This approach is not generally possible in thin active layers and is only possible in the present case due to the very high gain afforded by GaN/InGaN which is of order 10^ cm'l . This high gain enables a surface emitting laser to be made which has superior beam properties over longitudinal emitting lasers. In addition it is possible to have not just one, but many holes in a top gate metal layer, thereby allowing several laser beams to be emitted from a region. This allows one and two dimensional arrays of beams to be produced which would be of use in optical processing/optical communications. Examples of other multi hole devices are shown in Figures 5a and b. Figure 5a shows a square array of lasers in a 4x4 matrix and Figure 5 shows a linear array. In both arrangements holes 30 aare depicted diagrammatically.
Figure 5 c shows a honeycomb array 40 formed on the surface of a laser. This honeycombe array allows a large photon emission area with a very small area of (absorbing) metal contact. This structure allows large areas of lasers to be fabricated. These larger area arrays can be incorporated into 1 or 2 dimensional lasers as will be appreciated by a person skilled in the art. The holes 30 may be formed by wet or dry etching techniques.
It will be appreciated that the invention has been described by way of example only and that lasers falling within the scope of the invention may be used in a myriad applications such as in a projector or in a television apparatus.