Low resistance contact semiconductor diode

[0002] This is a continuation of application Ser. No. 09/068,943, filed May 21, 1998, now issued as U.S. Pat. No. 6,133,590, the entire content of which is hereby incorporated by reference in this application which is a continuation of PCT/GB 96/02914, filed Nov. 27, 1996.

[0003] 1. Field of the Invention

[0004] The current invention relates to devices made from narrow-gap (and more conventional) semiconductors for operation at infrared wavelengths.

[0005] 2. Discussion of Prior Art

[0006] Many prior art devices include a diode junction, hence metal contacts need to be formed to a region of n-type and a region of p-type material. Ideally the metal contact should be ohmic and have a low resistance, i.e. very much less than that of the junction. The surfaces of the semiconductors tend to comprise a native oxide, which is normally insulating to some extent and must be removed or disrupted to permit contact to the semiconductor underneath. The nature of the band-bending which often occurs at the metal-semiconductor interface means that it is relatively easy to obtain the desired contact to n-type material, but it can be difficult to achieve an ohmic, low resistance contact to the p-type semiconductor.

[0007] Alloying using a metal with a small proportion of a like-type dopant can be used with some semiconductors. When annealed, this gives a heavily doped region to which the metal is more likely to form an ohmic contact, for example Au/Zn for a p⁺ p contact to GaAs. This does not work well, however, with the narrow-gap semiconductors, such as indium antimonide (InSb) and cadmium mercury telluride (CMT) because of their low melting temperatures and the high diffusion rates of some of the dopants. Ion milling is often used for the narrow-gap materials to remove the surface oxide. This tends to result in near surface (typically tens of nanometers) damage which is heavily n-type in nature plus additional damage which can propagate significantly further into the material and degrade the junction properties. As a consequence the thickness of the contact regions in diodes made from the narrow-gap semiconductors needs to be greater than, or of the order of, 10⁻⁶m.

[0008] In devices such as light emitting diodes and diode lasers where optical radiation passes through one or both contact regions, this thickness of material can lead to significant optical losses which will degrade the device performance, in some cases catastrophically.

[0009] Degenerate doping of the material, i.e. doping to a sufficiently high level that the Fermi energy lies within the appropriate band, leads to a Moss-Burstein shift whereby the absorption coefficient for radiation of energy just greater than the band-gap decreases so that it becomes transparent. Owing to the band structure of the narrow-gap semiconductors, however, it is extremely difficult to induce degenerate doping levels in p-type material, so that a sufficient Moss-Burstein shift is not observable. In contrast, it is comparatively easy to produce degenerate doping in n-type material and large Moss-Burstein shifts can be achieved. Hence the n-type contact regions of diodes can be made transparent to radiation generated or absorbed in the active region but the p-type contact region can not.

[0010] Also relevant to the present invention is U.S. Pat. No. 5,338,944. This document describes a light emitting diode for emitting visible radiation in the blue region of the visible spectrum, comprising an n-type silicon carbide substrate, an n-type silicon carbide top layer and a light emitting p-n junction structure between the n-type substrate and the n-type top layer. The device also includes, between the n-type top layer and the n-type substrate, means for coupling the n-type top layer to the light emitting p-n junction structure while preventing n-p-n behaviour between the n-type top layer, the p-type layer in the junction structure and the n-type substrate. This is achieved by means of a degenerate junction structure comprising a p-type portion and an n-type portion of silicon carbide. The p-type portion and the n-type portion are very thin, of the order of 250-1000 angstroms, with a very high doping concentration of at least 1×10⁻¹⁹cm⁻³. This invention relates to light emitting diodes operating in the blue region of the visible wavelength spectrum. It therefore relates to devices formed from material having a bandgap of at least 2.6 eV, such as silicon carbide.

[0011] Applied Physics Letts, 62(1993) 17 May, No. 20, New York, USA (A. R. Sugg et al.) relates to n-p-(p⁻-n⁻)-n Al_yGa_1-y-As-GaAs-In_xGa_1-xAs quantum-well laser with ⁺-n⁺ GaAs-InGaAs tuned contact on n-GaAs. Hence, this document relates to laser devices formed on GaAs substrates.

[0012] According to the current invention, a semiconductor device comprises an active layer of p-type or n-type material forming a junction with a first layer of doped n-type material, a second layer of doped n-type material adjacent to a layer of doped p-type material, which may be adjacent to, or separated from by other layers, the active layer of p-type or n-type material, and means for providing electrical contact with the device including means for providing electrical contact, via the second layer of doped n-type material, with the adjacent layer of doped p-type material.

[0013] characterised in that the second layer of doped n-type material has a doping concentration of between 1×10¹⁸cm⁻³and less than 1×10¹⁹cm⁻³and in that semiconductor energy band-gap of the active layer is less than 0.5 eV.

[0014] The invention should not be regarded as being restricted to a particular set of doping levels.

[0015] Preferably one or both of the first and second layers of doped n-type material is transparent to radiation, of energy greater than that of the band-gap, which is emitted or absorbed by the device.

[0016] In a specific embodiment the device comprises a front surface emitting positive LED, negative LED or detector.

[0017] In another specific embodiment the device comprises a back surface emitting positive LED, negative LED or detector wherein the means for providing electrical contact, via the second layer of doped n-type material, with the adjacent layer of doped p-type material comprises a metallic contact and the second layer of doped n-type material provides a transparent front contact region to facilitate the use of the metallic contact as a mirror.

[0018] In another specific embodiment the device comprises a laser diode and the first and second layers of doped n-type material provide optical confinement within the active layer.

[0019] In another specific embodiment the second layer of doped n-type material provides an electrical path to an excluding or extracting contact. In this embodiment the device may comprise a field effect transistor or a bipolar transistor.

[0020] In another embodiment of the invention, the device may be a front surface emitting negative LED wherein the doping concentration of the doped n-type layer extends beyond 1×10¹⁹cm⁻³.

[0021] In another embodiment of the invention, the device may be a back surface emitting negative LED, wherein the doping concentration of the doped n-type layer extends beyond 1×10¹⁹cm⁻³, and wherein the means for providing electrical contact, via the second layer of doped n-type material, with the adjacent layer of doped p-type material comprises a metallic contact, and the second layer of doped n-type material provides a transparent front contact region to facilitate the use of the metallic contact as a mirror.

[0022] The invention will now be described, by way of example only, with reference to the following figures in which:

[0023] FIGS. 1aand 1brespectively show generic devices of the prior art and of the invention;

[0024] FIGS. 2aand 2brespectively show the band diagrams, at zero bias, of the devices shown in FIGS. 1aand 1b;

[0025] FIGS. 3aand 3brespectively show light emitting diodes of the prior art and the current invention;

[0026] FIGS. 4aand 4brespectively show back surface emitting devices of the prior art and the current invention;

[0027] FIGS. 5aand 5brespectively show semiconductor laser devices of the prior art and the current invention;

[0028] FIGS. 6a-6cshow various stages of construction of a semiconductor laser device according to the current invention and

[0029] FIG. 7 shows the performance of the device of FIG. 6c.

[0030] Throughout this specification p and n indicate the type of doping in a material, the superscripts − and + indicate light and heavy doping respectively, the underline indicates material of high band gap, v denotes an active layer of near intrinsic n-type material and π denotes an active layer of near intrinsic p-type material. The term “highly doped” when applied to layers within devices of the current invention should be construed as a relative term; thus layers to which the term is applied are understood to be more highly doped than those to which it is not.

[0031] The invention is described in relation to light emitting diodes and lasers fabricated in the indium aluminum antimonide (In_1-xAl_xSb) material system, but is not limited to this system. The generic concept is applicable to both types of device, and will be described first, but its detailed design criteria and implementation are slightly different for each case and these will be described subsequently.

[0032] By the current invention the majority of the highly doped p-type contact region immediately adjacent to the contact in a diode of the prior art is replaced with tightly doped, degenerate n-type material, leaving only a thin (of the order of 0.1×10⁻⁶m thick) highly doped (but not necessarily degenerate) p-type region adjacent to the remainder of the diode. The thickness of the remaining p⁺ region is sufficiently small that it does not result in significant absorption of the optical radiation.

[0033] The electrical junction formed between these n⁺ and ⁺ regions will be biased in the opposite sense to the active diode so, for example, if the active diode was forward biased the ‘contact’ diode would be reverse biased and might be expected to introduce a large series resistance. It is a property of narrow-gap semiconductors, however, that when heavily doped regions of opposite carrier type are in immediate proximity to each other, the probability of carrier tunnelling between the regions is very high even with reduced potential across the ⁺ junction. Hence a very large tunnelling current can pass through the n⁺ p⁺ diode and it does not present a significant series resistance. Therefore the requirement for transparent contact regions on both sides of the active region can be fulfilled and efficient device operation can be achieved.

[0034] Referring to FIG. 1a,a typical 4-layer structure 1, of the prior art comprises a layer of p⁺ material 2, a layer of p⁺ material 3 such as In_1-xAl_xSb, an active layer of π-type material 4 and a layer of n⁺ material 5. Electrical contacts 6 are made with layers 2 and 5. FIG. 2aqualitatively shows the relative energies of the valence band 7, conduction band 8 and Fermi level 9 of the device shown in FIG. 1a.

[0035] The design criteria and operation of this structure in the context of LEDs, detectors and transistors has been published previously (see T. Ashley, C. T. Elliott, N. T. Gordon, R. S. Hall. A. D. Johnson and G. H. Pryce, ‘Uncooled InSb/In_1-xAl_xSb mid-infrared emitter’, App. Phys. Lett.64 2433 (1994); T. Ashley, A. B. Dean, C. T. Elliott, G. J. Pryce, A. D. Johnson and H. Willis. ‘Uncooled high-speed InSb field-effect transistors’, Appl. Phys. Lett.66 481 (1995) and T. Ashley, ‘Electronic and optoelectronic devices in narrow-gap semiconductors’, Proc.7th Int. conf. Narrow-Gap Semiconductors,Santa Fe. N. Mex., USA 9-12 January 1995.

[0036] A device of the current invention 10 (comprising a modified, tunnel contact 5-layer structure) and its band diagram are shown in FIGS. 1band 2brespectively. In all the figures of this specification, corresponding items performing a similar role are labelled with the same numeral.

[0037] In addition to the first layer of n⁺ material 5, performing a role similar to that of layer 5 in FIG. 1a,device 10 has a second layer of n⁺ material 11 through which electrical contact with p⁺ layer 2 is made. The thickness of p⁺ layer 2 is greatly reduced in comparison with that of layer 2 in device 1.

[0038] The region of the semiconductor where tunnelling occurs is shown schematically by region 12 in FIG. 2b.In the case of the device shown, the active region is π-type but this is not essential to the invention and it could have a different doping type and level (for example v type) in other devices, and could even be a multiplicity of layers.

[0039] Table 1 lists typical thicknesses for the layers of devices 1 and 10, but the invention is not generally limited to these values.

[0040] Calculation of the tunnelling probability indicates that in InSb at room temperature (294 K) with a doping of 1×10¹⁸cm⁻³on each side of the n⁺/p³⁰ junction that the maximum of the product of area and resistance in reverse bias should be less than 10⁻³Ω cm². This is about half the value for a comparable device of the prior art.

[0041] The tunnel junction effectively forms a large recombination velocity contact to the p⁺ region, which could potentially inject additional electrons over the In_1-xAl_xSb barrier (layer 3) in to the active diode and degrade its performance. Modelling indicates that for a diode at 294 K with a barrier of composition x=0.15 and a p⁺ doping of 3×10¹⁸cm⁻³, the p⁺ thickness can be reduced to 0.01×10⁻⁶m without decreasing the active diode R_oA (product of resistance at zero bias and area) by more than 1%. The value of 0.01×10⁻⁶m represents a lower figure for the accuracy of the model rather than a predicted minimum thickness. The maximum thickness of the p⁺ region 2 is determined by the acceptable optical loss or electrical series resistance, as discussed below for the specific devices. If a thickness of more than 10⁻⁶m is acceptable, it is not necessary to use the n⁺ tunnel contact simply to avoid damage to the active diode as the p⁺ region alone will suffice to absorb any contacting damage. However the n⁺ region 11 also has the benefit of a very much lower (approx. factor of 50 per unit thickness) resistance, which can be very important if the lateral resistance of a contact region is significant.

[0042] Included within this group of devices are positive luminescence diodes, negative luminescence diodes detectors and vertical cavity surface emitting lasers. In all of these devices, light is transmitted essentially through one of the contact regions of the diode.

[0043] 1. Front Surface Devices:

[0044] Referring to FIG. 3a,a diode 1 of the prior art where the radiation, absorbed or emitted in active region 4, is to be transmitted through the front or top surface, would normally have an n⁺ top contact region 5, doped sufficiently highly that it is transparent, and the p⁺ contact region 2 would be at the bottom of the structure adjacent to the substrate not shown. This can lead to a large lateral resistance (R_p) along the p-type material 2 which is comparable to the diode R_o. This would lead to extra power dissipation in LEDs and would degrade the signal to noise performance of a detector. Referring to FIG. 3b,the use of a parallel n⁺ region 11 with resistance R_nand a tunnel junction with resistance R_tsuch that R_n+R_t<<R_pwould reduce the series resistance and so reduce power dissipation in the LEDs and improve signal to noise ratio in the detectors.

[0045] Back surface devices.

[0046] Referring to FIG. 4a,radiation, absorbed or emitted in active region 4, is intended to be transmitted through the substrate of prior art device 1 and the back surface would normally be designed with the p⁺ region 2 uppermost, so that the light passes through a transparent n⁺ region 5 and a transparent n⁺ substrate (not shown). In this case the top metal contact 6 can be placed over the whole of the upper surface to form a mirror to reflect light back through the active region 4. A single thick p⁺ layer 2 attenuates this reflected light, negating the benefit of the mirror. Referring to FIG. 4bthe use of an n⁺ region 11 to form a tunnelling junction to thin p⁺ material 2 enables it to be transparent so that a higher proportion of the light is reflected.

[0047] Referring to FIG. 5a,a typical semiconductor laser 13 requires an active gain region 4 surrounded by cladding layers 2, 5 which provide electrical confinement of the carriers and optical confinement and guiding of the emission. The prior art structure described above in the context of LEDs and detectors provides excellent electrical confinement, however it does not give good enough optical confinement. Optical confinement is obtained through a change (decrease) in refractive index at the interface with the cladding regions which guides the radiation back into the active region. A reduction in the refractive index can be achieved in the InSb material system through high doping, however only highly doped n-type layer 5 also becomes transparent to radiation just above the band-gap: the highly doped p-type layer 2 remains absorbing. Hence, even though some guiding is achieved by the p⁺ cladding layer 2 it strongly attenuates a wave propagating down the laser leading to the requirement for a very much higher gain in the material before stimulated emission is obtained—effectively preventing the laser from operating.

[0048] Referring to FIG. 5b,a laser device 14 of the current invention includes an n⁺ layer 11 which forms a tunnel junction with p⁺ layer 2 and permits guiding with low modal loss to be obtained on both sides of the active gain region 4 thus enabling the laser to operate.

[0049] Devices 13 and 14 would typically be fabricated on an n⁺ substrate 15.

[0050] Table 2 lists typical thicknesses for the layers of devices 13 and 14, but the invention is not generally limited to these values.

[0051] In order to produce an actual embodiment of the laser described generically in FIG. 5blayer structures as shown in therein were grown on highly doped n-type InSb substrates 15 doped within the range 10¹⁸cm⁻³to 3×10¹⁸cm⁻³. The layer thicknesses were as indicated in table 2.

[0052] Referring to FIG. 6a,the substrates were generally circular with a diameter of 1.5 or 2 inch, and were sawn or cleaved into chips 16 of approximately 6 mm square after growth. The thickness of the chips was reduced, by polishing from the substrate side, from an initial thickness of 5×10⁻⁴m to 1.5-2×10⁻⁴m.

[0053] Conventional photolithographic means were used to etch down to the lower epitaxial n⁺ layer (item 5 of FIG. 5b) to form mesa structures 17 of width 50×10⁻⁶m or 100×10⁻⁶m on a pitch of approximately 1 mm across the chip. Referring to FIG. 6b,similar photolithographic means were used to deposit chromium/gold metallic contacts 6 on to the top of each mesa 17 and, in some cases, to the common lower n⁺ layer (item 5 of FIG. 5b).

[0054] Each chip was then further cleaved, in a direction perpendicular to the mesa length, into pieces approximately 5×10⁻⁴m wide to form bars 18 each forming individual lasers which emitted radiation via the facets on the long sides of the bars as represented by arrows 19 in FIG. 6c.

[0055] Each bar was mounted separately on a standard sample package and gold wire bonds were applied to the top contact of each laser and to the common contact for subsequent assessment and use.

[0056] Modelling indicates that for the structure 14 shown in FIG. 5b,with the dimensions indicated in table 2, the absorption coefficient along the laser is only −0.0037/m×10⁻⁶, compared with −0.06/m×10⁻⁶in the structure 13 of FIG. 5a.Fabrication of this structure has allowed the first demonstration of an electrically pumped laser in InSb at a temperature of 77 K, see the output spectrum in FIG. 7. Previously, devices required cooling to substantially lower temperature and/or the use of high magnetic fields.

[0057] The use of exclusion/extraction to enable room temperature operation of InSb based transistors has been patented (GB 2 266 183). Several types of these transistors, in particular the field effect devices, require the use of a p⁺ region to form an excluding contact, which carries a significant current in a lateral direction. The use of an n⁺ tunnel junction would reduce the voltage loss along this region in a similar manner to that described for the front surface LEDs.

[0058] For the purpose of this specification, in the context of optical devices (e.g. light receiving or light emitting devices) the phrase “action region” or “active layer” shall be taken to mean that region of the device in which substantial regeneration or recombination of carriers occurs. In the context of electrical devices (e.g. transistors) the phrase “active region” or “active layer” shall be taken to refer to the region contacted by the excluding or extracting junction, in which the exclusion or extraction occurs i.e. the region in which there is a significant change in carrier concentration.