GB2341723A

GB2341723A - A light emitting device

Info

Publication number: GB2341723A
Application number: GB9820373A
Authority: GB
Inventors: Stephen Gerard Ayling; Michael Barry Allenson; David Robert Wight
Original assignee: UK Secretary of State for Defence
Current assignee: UK Secretary of State for Defence
Priority date: 1998-09-19
Filing date: 1998-09-19
Publication date: 2000-03-22
Also published as: GB9820373D0

Abstract

A light emitting device comprises at least two light emitting elements, 42a and 42b, each element having an impedance and an individual quantum efficiency, which are electrically and optically connected, by metal layers, 54a and 54b, such that the device quantum efficiency is greater than the individual element quantum efficiencies. The optical elements have a common optical path 48 through which the beams of output radiation form the optical elements are transmitted. The light emitting device may comprise a plurality of diodes connected in series such that the input impedance of the light emitting device is equal to 50 L without the need for resistive impedance matching elements. The diodes may be AlGaAs, AlGaInP, AlGaInAs or AlGaInAsP laser diodes. The light emitting device may comprise plural quantum cascade lasers. The light emitting device is described being used in an optically coupled bipolar transistor, in fibre optic links and an optical repeater having optical gain.

Description

2341723 A LIGHT EMITTING DEVICE AND TRANSISTOR

The invention relates to a light emitting device and a transistor comprising the light emitting device and an optical detector. In particular, the invention has advantages in the field of fibre optic communications and transmissions, and in RF applications.

The operation of a conventional common base bipolar transistor is well know in the prior art, for example see W. Shockley, Bell Syst. Technolog, J 28, 435,1949. The crucial feature of this transistor is the transit of current from the low impedance input (base/emitter) circuit to a high impedance output (base/collector) circuit by means of minority carrier diffusion across a thin (base) semiconductor layer. The reverse biased collector output can therefore deliver current into a relatively high impedance load resistor to achieve power gain.

It has been proposed to use a Beam Of Light Transistor (BOLT) [R. Rediker et al., Proc. IEEE 51, 218 1963] in which the transfer of current is achieved by converting the minority carriers into photons before transit across a "base" layer in the device and then converting the photons back to a current at the output. However, the optical transfer cannot be achieved efficiently in such devices and large current losses are always encountered so that useful power gain cannot be achieved.

In conventional microwave amplifier technology, to increase microwave output power at high frequencies several individual transistors may be connected in parallel in order to achieve higher currents. Such small devices have a high speed of operation, but the resistance of the parallel arrangement is small. In microwave circuitry, where the majority of microwave generation, transmission, reception and cable hardware is of 50 Q impedance, it is to difficult to match the low impedance of these devices to the 50 Q hardware over a broad frequency range. As conventional electronic transistors are three terminal devices, they may only be combined in parallel. This restricts the possibilities for impedance management and hence the maximum available gain and power of electronic transistor technology.

2 Another problem associated with electrical transistors is that the displacement currents produced limit the speed of operation of the transistor. Furthermore, as the input circuit and the output circuit are electrically coupled stability problems can arise due to feedback from the output circuit to the input circuit.

In the field of fibre optic links, conventional semiconductor lasers are commonly used as transmitter elements. The current technology for achieving directly modulated, broad band, high speed fibre optic links has been developed principally for digital communications systems. Typically, such fibre optic links operate at an upper frequency limit of around 10 GHz set by the response capability of the laser to the input signal. Conventional broad band fibre optic links, however, cannot deliver signal gain unless either electronic or optical amplifiers are included in the signal path. Furthermore, for use in fixed impedance environments, such as microwave circuits, a relatively narrow band impedance transformation is need to match the low impedance laser in order to minimise signal loss.

UK patent application GB 971-35-365.6 describes a laser device and optically coupled bipolar transistor which enables intrinsic 50 broad band impedance matching for RF applications with enhanced laser quantum efficiency to reduce transmission loss or even provide RF gain in optomicrowave applications. This invention is also described in Photonics West, Optoelectronics '98, "Laser Diode Applications IV, conference no. 3285 B, San Jose, California, 24-30 January 1998. However, a disadvantage of this invention for some applications is that there are multiple optical outputs, requiring multiple optical fibre connections to transmit the outputs. This may be expensive and inconvenient.

It is an object of the present invention to overcome the problems in known electronic and optical transistor devices and, in addition, to overcome the impedance transformation problems encountered when conventional transistors are used in 50 Q microwave circuitry. It is a further object of the present invention to provide an alternative to the light emitting device and transistor described in GB 9713-365.6 which has advantages in some applications.

3 According to one aspect of the invention, a light emitting device, having an input impedance and a device quantum efficiency, for generating a beam of output radiation from an input current of electrons comprises; at least two optical elements for converting the input current of electrons into a beam of output radiation, each optical element having an impedance and an individual quantum efficiency, the optical elements being electrically connected such that the device quantum efficiency is greater than the individual quantum efficiency of one of the optical elements, characterised in that the optical elements are optically connected and have a common optical path through which the beams of output radiation from all the optical elements are transmitted.

In a preferred embodiment, the optical elements may be electrically connected such that the device input impedance is substantially equal to 50 Q. This removes the requirement for resistive impedance matching though a small degree of reactive matching may still be desirable.

The light emitting device has a device modulation frequency limit. In a further preferred embodiment, the device input impedance is substantially equal to 50 Q across a frequency range substantially from DC to the device modulation frequency limit.

The light emitting device is therefore capable of operating with enhanced quantum efficiency across a broad band of modulation frequencies. In particular, this may be achieved in the fixed impedance (50 Q) microwave circuitry without the need for resistive impedance matching. The invention overcomes the narrow bandwidth limitation associated with conventional lasers in that only simple reactive impedance transformation may be necessary, if at all.

One optical element may have an end face coated with a reflective coating such that the light emitting device provides a single optical output. This is advantageous as only a single optical Z fibre connection is required to transmit the single optical output.

4 The optical elements may be quantum cascade lasers or may. be p-n junctions. The p-n junctions may be laser diodes, for example any one of A1GaAs, A1GaInP, A1GaInAs or A1GaInAsP laser diodes.

Each of the optical elements may share a common optical waveguide through which radiation C1 output from each of the optical elements is transmitted, the common optical waveguide comprising a bridging region between each of optical elements wherein the or each bridging region electrically isolates the optical elements.

The or each bridging region may be electrically insulating by means of proton implantation. Preferably, the or each bridging region is optically, transparent. For example, the common optical wa,,,,cguide has an active layer -x,,,hicli may, be rendered transparent within the bridging region.

c A- t> According to a second aspect of the invention, an optically coupled transistor for generating an output electrical signal comprises; the light emitting device as herein described and a photodetector for detecting the beam of radiation output from the light emitting device and for converting the beam of output radiation into an output electrical current, wherein the light emitting device and the photodetector are in electrical isolation, thereby inhibiting electrical feedback from the photodetector to the light emittine, device.

Typically the photodetector may be a photodiode device.

The optically coupled transistor provides current gain across a broad band of modulation frequencies, from DC up to typically 30 GHz (a conventional laser in combination with the photodiode is not capable of providing current gain due to the inherent losses of the device components). Such a device therefore overcomes the impedance transformation problems encountered when conventional transistors are used in 50 0 microwave circuitry. Its characteristics and design make it suitable for application to fibre optic communications and the optical distribution of radio-frequency, microwave, mm-wave and digital signals in electronic systems such as phased array radars.

The optically coupled transistor may comprise two photodetectors, each photodetector arranged to detect a beam of radiation output from a different end of the common optical path.

The optically coupled transistor may comprise an optical fibre for transmitting the beams of output radiation to the one or more photodetectors.

The transistor device may be integrated on a single chip or the photons may be transmitted to the collector means via an optical fibre or a lens system.

According to a third aspect of the invention, a fibre optic link comprising an optical fibre having 4n an input endface and an output endface, comprises the light emitting device as herein described, 4n wherein the light emitting device is situated at the input endface of the optical fibre such that the beam of radiation output from the light emitting device is input to the optical fibre.

6 According to a fourth aspect of the invention, an optical repeater for receiving an optical input signal and generating an optical output signals comprises; a photodetector for receiving the optical input signal and converting the optical input signal into an electrical signal and the light emitting device as herein described for receiving the electrical signal and outputting an optical signal from the common optical path.

The optical repeater may also comprise amplification means for amplifying the electrical signal output from the photodetector.

7 The invention will now be described by example only with reference to the following figures in which, Figure I shows a three dimensional view of a section of a light emitting device known in the prior art,

Figure 2 shows a schematic diagram of an optically coupled transistor (OCT) comprising the light emitting device shown in Figure 1, Figure 3 shows a diagram of a conventional common base electronic transistor, Figure 4 shows the epitaxial layer structure of the light emitting device of the present invention, Figure 5 shows a schematic diagram of the light emitting device of the present invention, Figure 6 shows a three dimensional view of two elements of the light emitting device of the present invention, Figure 7a shows- the typical dimensions of the device shown in Figures 4- 6, Figure 7b shows an enlarged diagram of a bridging region between two elements of the liaht emitting device shown in Figures 4-6, Figure 8 shows the current-voltage characteristic of a three section unimplanted light emitting device, Figure 9 shows the current-voltage characteristic of a three section implanted light emitting device, 8 Figure 10 shows an optically coupled transistor comprising the light emitting device shown in Figures 4-6, and Figure 11 shows an optical repeater of the present invention.

For the purpose of this description, references to frequency shall be taken to mean modulation frequency rather than the optical emission frequency of the laser.

9 BACKGROUND

By way of background to the invention, the light emitting device described GB 97133365.6 and in Photonics West, San Jose, California, Optoelectronics '98, "Laser Diode Applications IV", conference no. 3285 B, 24-30 January 1998 will now be described with reference to Figure 1.

The light emitting device 1, which may also be referred to as a laser device, comprises a number of discrete p-i-n laser diode elements in an array. Only two laser diodes 2a,2b are shown in the figure but in practice a greater number of laser diodes may be included in the device. Typically, each individual laser diode 2a,2b may be a p-i-n laser diode comprising a top layer of highly doped low resistance p-type material 9 (indicated but not shown in the figure), on a moderately doped upper cladding layer 4. This is formed on a waveguide layer 5 which is undoped, formed on a lower cladding layer 6, wherein the waveguide layer 5 has a lower band gap and a higher refractive index than either the upper or lower cladding layers 4,6. Embedded within the waveguide layer 5 are quantum wells (not shown) that provide optical gain. The combination of the upper and lower cladding layers 4,6 forms a double heterojunction barrier layer confining both photons and carriers to the undoped waveguide layer 5. The function of these layers would be understood by a person familiar with the art.

A highly doped n-type layer 7 forms a low resistance contact for each individual laser 2a,2b. This layer 7 is completely removed on one side of each laser element 2a,2b, leaving each laser electrically isolated from its neighbour by the semi -insulating substrate layer 11. The laser diodes ?-I 2a,2b are connected in series by means of the metal layers 10.

Current, 1, injected into the first laser 2a in the series, passes down through the p-i-n junction (layers 4,5 and 6 respectively) emitting light from the quantum wells within the intrinsic region 5. The current then leaves the first laser 2a through the bottom layer 7, and passes over the mesa of the next element in the series by means of the plated metal layer 10. The current is injected into this next laser element 2b and passes down through this p-i-n junction, as before.

A feature of the device shown in Figure I is that the external incremental quantum efficiency of the array is greater than that of a single laser element 2a or 2b within the array. The net quantum efficiency of the laser device I is the sum of all the individual efficiencies of each element. This allows the device to operate with a quantum efficiency of greater than 100%.

An application of the device shown in Figure 1, is to combine the device with a photodetector, such as a conventional semiconductor p-i-n photodiode under reverse bias, to produce a transistor-like device which operates on the basis of the optical transfer of photons rather than the transfer of minority carriers as in conventional transistor devices.

Figure 2 shows a transistor 15 which includes a photodetector 20. The transistor shown in Figure 2 is referred to as an optically coupled transistor device, or OCT device. The OCT device is capable of delivering current gain. The OCT device 15 comprises an input circuit 16 (the "base/emitter" circuit) and an output circuit 17 (the "base/col lector" circuit) which may have a common "'base" connection 18. The input circuit 16 comprises the laser device I shown in Figure 1. The laser device I in the figure is shown to comprise three p-n junctions 19. The junctions 19 may be lasers or light emitting diodes (LEDs) capable of efficiently converting an electrical current into photons. The output circuit 17 comprises a photodetector 20 which receives radiation 21 emitted from the input circuit 16 and converts it into an electrical current. The OCT is capable of delivering current gain (as well as power gain). The transit between the input circuit and the output circuit is optical rather than electrical which provides the advantage that no displacement currents are produced.

By way of further background to the invention, a conventional common base transistor 30 is shown in Figure 3, comprising a p-type base layer '32 between two n-type regions 34. In the transistor 30, electrons transit the p-type base layer 32 where a small fraction are lost by recombination and produce a small base current, lb Ub = Ic - U- Since the forward biased emitter input is of low (AC and DC) impedance and the reverse biased collector output can deliver current into a relatively high impedance load resistor, RL, power gain is achieved by the "transfer resistor" or "transistor" action.

THE INYENT10 The present invention relates to a light emitting device having , an enhanced quantum efficiency across a broad band of modulation frequencies. For the purpose of this description, the light emitting device may also be referred to as a laser device. The light emitting device also provides the further advantage that radiation is output from the device in a single output. In particular, enhanced quantum efficiency may be achieved in the fixed impedance (50 Q) microwave circuitry without the need for resistive impedance matching. Its characteristics and design make it suitable for application to fibre optic communications and the optical distribution of radiofrequency, microwave, mm-wave and digital signals in electronic systems such as phased array radars. It will be appreciated that some applications will be other than 50 Q and the device impedance can be matched appropriately, for example aerials are of 75 Q impedance.

Referring to Figure 4, the light emitting device may comprise a number of discrete p-i-n laser diode elements 42a,42b. Figure 4 illustrates the epitaxial layer structure of the device 40 in crosssection. The connection between the individual elements will be described in further detail later. Only two elements 42a,42b are shown in Figure 4 but in practice a greater number of elements may be included in the device. For example, each individual element 42a,42b may be a AlGaAs p-i-n laser diode comprising a top layer 44 of highly doped, low resistance p-type GaAs, on a moderately doped upper cladding layer 46. This is formed on a waveguide layer 48 which is undoped, formed on a lower cladding layer 50, wherein the waveguide layer 48 has a lower band gap and a higher refractive index than either the upper or lower cladding layers 46,50. GaAs quantum wells (not shown), typically of width 10 nin, may be embedded within the waveguide layer 48 to provide optical gain.

Alternatively, the waveguide layer 48 may be a bulk active layer providing optical gain. The combination of the upper and lower cladding layers 46,50 forms a double heterojunction barrier layer confining both photons and carriers to the undoped waveguide layer 48.

12 Typically, the contact layer 44 may be p-type GaAs (typical thickness 0. 1 pm, typical doping concentration 5 x 10 19 CM-3), the upper cladding layer 46 may be AlGaAs (40%), (typical 7 -3 thickness 1.5 tm, typical doping concentration 5 x 101 CM) and the lower cladding layer 50 may by n-type AlGaAs (40%) (typical thickness 1.5 pm, typical doping concentration 5 x 10 17 -3 CM). The waveguide layer 48 may typically be undoped AlGaAs (20%) (typical thickness 0.233 The cladding layer 50 of each element 42a,42b is situated on a highly doped n-type layer 52 which forms a low resistance GaAs contact for the elements 42a,42b (i.e. an n-type contact). Typically, this contact layer 52 may be a GaAs layer having a thickness of between 0.5 and 1.0 [tm and a doping concentration of 2 x 10 18 cm-3. The percentage indications refer to the 4n percentage of Al in the layers. Each element 42a,42b has an insulating layer 53, for example polyimide, covering the mesa. A plated electrical layer 54a, 54b covers the insulating layer 53. Part of a plated electrical layer 54c is also shown on element 42b.

Referring to element 42a, the p-type layer 44a is in contact with the plated electrical layer 54a via a p-type metal contact 56a. The n-type layer 52 is in contact with the plated electrical layer 54b via an n-type metal contact 59a. The plated electrical layer 54b is a single layer. The plated electrical layer 54b connects the n-ty e metal contact 59a of element 42a to the p-type metal I P contact 56b of element 42b at point X. The two elements 42a,42b are therefore in electrical connection.

13 The connection between the elements 42a,42b is now explained with reference to Figure 5 which shows a schematic plan view of the device 40 connections. Each element 42a,42b (sho-.vn within dashed lines) comprises a forward biased p-n junction. The junctions are in electrical and optical series and each produces light and converts an input current into photons. In practice more than two junctions may be used. The plated electrical layer 54a of element 42a is connected to the ptype metal contact (56 in Figure 4) of this element. The plated electrical layer 54b is connected to the n-type metal contact (59 in Figure 4) of this element. The n-type contact of element 42a is electrically connected to the p- type contact of the adjacent element 42b by means of the plated electrical layer 54b such that the elements are connected in series.

The junctions share a common optical axis, referred to as the "ridge" waveguide, 58. Light emissions from each of the elements are summed into this ridge waveguide 58 to form an optical output 66. The optical output 66 is shown to emerge from both ends of the ridge waveguide 58, although in practice suitable reflection means may be used so that only a single output 66 emerges from one end. For example, a highly reflective coating may be applied to one end of the ridge waveguide 58, thus forcing all the output radiation to be emitted from the other end. This maximises the useful light output. A combination of high and low reflectivity coatings may be used to direct output radiation. The device 40 may be constructed using integrated photonics r> technology, techniques which would be conventional to one familiar with the art. The reflectivity coatings may be applied using evaporation or sputtering techniques which would be familiar to those skilled in the art.

Figure 6 shows a three dimensional view of two elements 42a,42b of the light emitting device 40 based on the structure shown in Figures 4 and 5. Figure 6 shows the two elements 42a,42b sharing a common n-type base layer 52. The plated electrical layer 54a (in connection with the ptype metal contact 56a in Figure 4) of element 422a and the plated electrical layer 54c (in connection with the n-type metal contact 59b in Figure 4) of element 42b is shown. The plated electrical layer 54b is shown between the two elements 42a,42b which connects the n-type side of element 42a to the p-type side of element 42b, as described previously. The ridge waveguide 58 through which the optical signal is transmitted is also shown. Between the two sections 42a,42b is a bridge region 72 which will be discussed in further detail later.

14 The typical dimensions of an element 42a,42b having single mode optical waveguide operation are shown in Figure 7a. In this example, the ridge waveguide 58 (layer 44 and the upper region of layer 46) has a width of 3. 5 [im and will support only one mode when lasing. The ridge is supported on a mesa (waveguide layer 48, layer 50 and the lower region of 46), typically having a width of 25 [tm. The ridge waveguide 58 must be wide enough for the optical mode supported not to be influenced by the edges of the mesa which would otherwise support higher modes. Typically, layer 52 extends beyond the width of the mesa, on one side only, by around 50 [tm. Although narrow waveguide single mode lasers are necessary for high frequency operation, due to mode noise, for lower frequencies wider waveguide multi mode lasers may be used in the device 40. These wider lasers have the advantage of a lower resistance and therefore more elements may be connected in series. Alternatively, the waveguide layer 48 may include a distributed feedback element (e.g. a grating) as in a DFB laser.

As the elements 42a,42b, share a common optical axis 58, light emissions are summed into a single optical output 66. The same current is reused in each light emitter junction, thereby enabling the optical efficiencies to be summed and a quantum efficiency of greater than 100% to be achieved. This quantum efficiency determines the current transfer efficiency of an optical link or transistor with 50 0 resistive impedance matching.

The external incremental quantum efficiency (TIEXT) of a laser is conventionally measured in photons per electron and is a key measure of laser efficiency. For the purpose of this description, the term "quantum efficiency" shall be taken to mean the external incremental quantum efficiency. The net quantum efficiency of the laser device is not simply the sum of all the individual efficiencies of each element, as the efficiency is reduced as the length of the device increases, as explained below.

The external quantum efficiency,, TIEKry of the device is given by the following relationship; 1 1 1 + EXT N,,T where IIINT is the internal quantum efficiency of a single element in the device, N is the number of elements in the device, R] and R2 are the reflectivities of the facets at the end of the ridge waveguide 58, a is the internal loss of a single element and L is the total length of the device. For a device having N elements, each of length 1, it therefore follows that; 2ccNI 11 Err:-' Nll NT 1 + - ln(R, R2)) The external quantum efficiency of the device therefore increases as the number of elements, N, is increased, but it eventually becomes saturated at some maximum efficiency, even if the number of elements is increased; r,-XT p,'T In(R, R2) 2al For a fixed total length, A7, however, the external quantum efficiency will scale with the number of elements.

16 The total length of the optical device (i.e. the length of the number of elements 42a,42b included in the device 40) determines the external quantum efficiency, longer devices absorb more of their light internally and are therefore less efficient. There is therefore a practical limit to the longest desirable length for a laser device 40 and hence the number of element 42a,42b included in the device 40. For the device 40 subdivided into elements 42a,42b the total series resistance increases as the square of the number of elements 42a,42b for a fixed total device length. Thus, if the nominal resistance is 5 Q for a 500 Lrn device, subdividing into three elements or sections, 0 X 2 increases the resistance to 45 Q (i.e. 5 3). The quantum efficiency, however, is only increased by a factor of 3.

It may be possible to use more elements to provide a greater efficiency if the total length of the laser is increased, thereby keeping the final resistance to 50 n. However, if a longer laser is used, this will reduce the highest modulation speed achievable. However, for moderate modulation rates of about 1 GHz the device length is not the limiting factor. Preferably, the device may be configured such that the device input impedance is substantially equal to 50 0 across a frequency range from DC to the modulation frequency limit of the total device.

The device resistance can be tailored by careful choice of the layer thicknesses, composition and doping levels. The dominant resistance is due to the p-metal contact interface with the semiconductor elements and this too can be reduced by careful design (e.g. by higher p-doping, by the choice of p-metal contact, by widening the waveguide, by using T-gate transistor technology). If this resistance can be reduced to an arbitrary low value the device becomes limited by the dynamic impedance of the p-n junction which can be engineered to an extremely low value. In this demonstration the contacts are simple large area blocks for ease of testing as shown in Figures 5 & 6. In a high frequency RF design these contacts might be reduced in width, and therefore area, to further reduce capacitance.

The primary advantage of the laser device is that the impedance is intrinsically matched to the 50 Q fixed impedance of microwave circuitry, without the need for resistive matching elements. This is because the laser has substantially a 50 ? impedance with possibly a small reactive component that is simpler to match. Coupling the device to optical fibre can be achieved more easily than the prior art device as there is only one optical output from the device, whereas the prior art device provides several optical outputs. Connecting the device of the present invention does not therefore require a multi-fibre ribbon. This is inconvenient in many applications and is expensive, especially for long distance transmission.

Since the individual laser elements 42a,42b are physically connectedthrough the common optical ridge waveguide 58, complete electrical isolation between subsections 42a,42b is not normally possible. If the epitaxial wafer is designed for high electrical resistance to limit this leakage then the laser resistance will greatly exceed 5 Q and only two elements would be practical. It is therefore necessary to design a low resistance and low optical loss epitaxial structure. However, this can cause a large leakage current which may be eliminated as described below.

The optical bridge 72 between two adjacent elements 42a,42b is shown in Figure 7b and has the top p-type layer 44 (present in Figure 7a) removed. This has the effect of isolating layer 44 either side of the bridge 72. Also, the mesa width is reduced from 25 gm (Figure 7a) to 10 pm (Figure 7b). The leakage problem still occurs mainly through the remainder of the n-type layer 52 underneath the optical bridge 72 and so this must be proton isolated by implanting through the whole of the optical bridge. In the process of proton isolating the bottom layer 52, the other layers (46, 48,50) are also isolated. If the electrical isolation were perfect and optically lossless one would expect the threshold current to decrease in proportion to the number of elements 42a,42b since this relates to the number of times the current is reused. The forward bias voltage of the device (applied across the elements 42a,42b) should increase in proportion to the number of elements 42a,42b as this is the number of p-n junctions that need to be forward biased. The bias voltage is applied across the device by connecting a voltage supply to the first and last elements in the arrangement.

18 RIESULIS A light emitting device has been manufactured using GaAlAs integrated photonics technology capabilities using MBE growth on a semi-insulating wafer. The device comprised up to five separate elements, each element having the dimensions shown in Figure 7a. The structure of the device is as described previously; the contact layer 44 was p-type GaAs (thickness approximately 9 -3 0. 1 [im, doping concentration approximately 5 x 10 1 CM), the upper cladding layer 46 was AlGa-As (40%), (thickness approximately 1.5 Lrn, doping concentration approximately 5 x 10 17 -3 CM) and the lower cladding layer 50 was n-type AlGaAs (40%) (thickness approximately 1.5 17 -3 [tm, doping concentration approximately 5 x 10 CM). The waveguide layer 48 was undoped AlGaAs (20%) (thickness approximately 0.23) tm). The n- type layer 52 was GaAs having a thickness of approximately 1 pm and a doping concentration of 2 x 10 1 8 cm-3. The elements were connected in electrical series, as described previously, with each of their optical outputs being input to the same optical ridge waveguide 58.

The results obtained are shown graphically in Figures 8 and 9. Results were obtained by L_ I receiving light from a single output of the laser device in a large, low speed photodiode held in close proximity to one output face of the laser device 40. The laser device 40 delivered equal optical outputs from each endface, but only light from one end face was intercepted by the silicon photodiode. The single element result and the results obtained for multiple elements (i.e. comprising two or more elements 42a,42b) of the same total length (approximately 1250 tm) are shown in Table 1.

19 Number of Threshold External Forward voltage Dynamic series elements current incremental Vf (V) resistance Ith (mA) quantum efficiency il (W/A) 1 27 0.34 1.8 4 2 30 0.78 3 17 3 32 0.94 4.5 25 4 21 1.17 5.6 51 32 0.83 7.9 57 Table]: Quantum efficiency results obtained (without proton isolation) for single and m ultiple section lasers of the same total length (1250 tm).

As can be seen from Table I the threshold current does not decrease systematically with the number of elements. This is due to leakage current through the optical bridge 72. This is also shown schematically in Figure 8 which shows the typical current-voltage characteristic of a three-element, unimplanted laser device. It can be seen from this graph that the device has a number of turn-on voltages 80,81,82 and notjust one. The external quantum efficiency, however, does increase with the number of elements and this indicates that the optical bridge regions between elements add little loss to the device. A value above I is demonstrated, proving quantum gain from only one facet.

The forward bias voltage scales with the number of elements, as predicted. The series resistance does not scale with the number of elements since the measured resistance includes a variable parasitic resistance due to the probes used for contacting to the device in the test set-up. This implies that the real dynamic resistance of all the devices in a properly connected RF circuit will be lower than those shown here.

Electrical isolation of the optical bridge 72 may be achieved by deep proton implantation. Figure 9 schematically shows the typical results obtained for a device having the structure described previously, but with the optical bridge regions 72 proton implanted to provide electrical isolation. As shown in Figure 9, this improves the current-voltage characteristic of the device which now shows only one turn-on voltage 84.

Although deep proton implantation improves electrical isolation of the optical bridge or bridges 72, it is possible that in using implantation the optical bridge may suffer from additional optical losses. This may be reduced or eliminated if before electrical isolation the bridges are made optically transparent by removing the active layer 48. This may be achieved by etching and regrowth or quantum well intermixing. It is also possible to completely remove the bridge material and replace it with an insulating optical material that will transmit light across the gap.

Proton isolation is a well known technique for electrical isolation and can be achieved using a thick resist implant mask. The implant needs to be a multiple energy implant to ensure isolation throughout the p-i-n structure of the bridge, right down to the semi-insulting substrate. Quantum well intermixing may be achieved by various methods but all result in the diffusion of quantum well material out of the quantum well which is exchanged with material in the quantum well barriers. This increases the quantum well bandgap energy reducing the optical loss. This subject has been recently reviewed [E.H. Li, "Quantum Well Intermixing for Photonics", SPIE Milestone Series Volume MS 145, SPIE Optical Engineering Press, 199 8].

The laser device 40 shown schematically in Figure 6 is an illustration of an integrated version of the device which would be capable of operating at high speeds. For the purpose of this specification, the term p-n junction should be taken to include any variations of the semiconductor device structure having p-type and n-type components, including p-i-n junctions. Examples of lasers which may be used include AlGaAs, AlGaInAs, AlGaInP and AlGaInAsP devices.

21 The device 40 may be an integrated device which may be constructed on a single chip. An important feature of an integrated device is that each laser element 42a,42b is on an insulating substrate with the only electrical connection between them being a plated metal layer connecting the laser diodes 42a,42b in series. However, the invention is not intended to be limited to integrated devices and may also take the form of a series of discrete components connected in series in a circuit. An integrated device, however, does have the advantage of a higher speed of operation than a circuit manifestation. Also, an integrated device has a very low optical loss between elements without reflections.

The optical elements 42a,42b in the device 40 need not be p-n junctions. Alternatively, they may be quantum cascade lasers which do not include a p-n junction but still operate for the purposes of this invention in the same manner, i.e. they are light emitting devices which are current dependent. These quantum cascade lasers are constructed from a series of light emitting layers laid one upon another within the waveguide layer of the epitaxial structure (equivalent to layer 48 of this invention, as shown in Figure 4). Current may be injected from either direction through the series of light emitting layers as the laser is essentially Ohmic in nature. Therefore, a simpler contact design might be used for series connecting the different elements in a laser device comprising quantum cascade lasers, although electrical isolation between elements is still required.

A biasing network is needed so that the light emitting device 40 may be DC biased above laser threshold current and the RF signal introduced to the laser. The details of a suitable biasing network would be conventional to one familiar with the art and are not described in this specification.

22 The device 40 may be included in a transistor-like device 73, as shown in Figure 10. The device 73 is referred to as an optically coupled transistor (OCT) device. An input current le is passed through a number of elements 74a,74b,74c (three are shown), such as forward biased emitter p-n junctions, in electrical and optical series. These emitter junctions are lasers, for example as described earlier. Each junction converts the input current, Ie 5 to photons. The photons are then collected by a reverse biased collector 76, such as a photodiode, which converts them back to an electrical current ICI If the conversion and transfer processes are efficient the circuit can provide DC and AC current gain. This is a new extension to the functionality of the common base transistor.

The device 73 has an input circuit 78 (the "base/emitter" circuit) and an output circuit 80 (the "base/col lector" circuit), each circuit 78,80 having a base ten-ninal 82,84 respectively. The base terminals 82,84 may be connected to ground to remain equivalent to a conventional transistor (as shown in Figure.3 3) but total isolation is enabled between the input circuit 78 and the output circuit 80 (the "base/col lector" circuit) as in the operation of opto-couplers. This provides another functional advantage.

The transistor 73 shown in Figure 10 also provides an advantage over the prior art device shown t in Figure 2, as there is only one optical signal passing between the input and output circuits, 78,80 respectively.

23 If the optical conversion and transfer functions are efficient, the device 73 is capable of delivering current gain. The incremental current transfer efficiency, 71CT.- for a single p-n junction may be defined by; 71 CT = 11 EXT '11 OP '11 PD where il_,xT is the external incremental quantum efficiency (in photons per electron) of the light emitting device 40, -nop is the optical transfer efficiency between the light emitting device 40 and the reverse biased p-i-n photodiode 76 and 71PD is the conversion efficiency (in electrons per photon) of the photodiode 76. For example, if each p-n junction 40 has a quantum efficiency (TlExT) of, say, 50%, and the optical transfer efficiency and photodiode efficiency are both 100%, then the three-junction OCT device 73 generates an incremental output current, AIC, from the photodiode 76 of 3 11CT Ale, that is Alc = 1.5. Ale, where I, is the current in the laser. Hence the current output from the OCT device is greater than the input current. This in an advantage over conventional common base transistors where the output current is always less than the input current.

The primary advantage of the OCT device 73 is its ability to deliver current gain (as well as power gain). Furthermore, this may be achieved across a broad frequency band in a fixed impedance circuit. The impedance matching is achieved from DC to the operating limit of the laser.

It is advantageous that the transit between the input circuit 78 and the output circuit 80 is optical rather than electrical as no displacement currents are produced. Furthermore, as the input circuit 78 and the output circuit 80 are decoupled, this avoids any feedback from the output circuit 80 to the input circuit 78 which is a problem in conventional electronic transistors.

24 It is a further advantage that the device 73 is two physically separated and electrically independent two-terminal circuits, unlike the conventional transistor (Figure 2) which is a three terminal device. Each of the two-terminal circuits therefore has more design degrees of freedom. The circuits 78,80 may be more easily be distributed to alter the impedance of the input circuit 78 and the output circuit 80 using series or parallel circuit configurations with current combining and splitting being achieved in the optical domain, hence having no electrical penalty. The ability to distribute the transistor as two two-terminal circuits 78, 80 also allows improved power outputs and thermal management.

In an uncoated OCT device the optical coupling efficiency (Tjop) is less the 50% since only light from one end of the device is collected by the photodiode in Figure 10. In another embodiment of the OCT device 73, a highly reflective coating may be applied to one end of the series of elements (i.e. to one end of the common optical ridge waveguide), e.g. 74a of device 40, thus forcing all or most of the output radiation to be emitted from the other end of the waveguide (74c of device 40), as described previously. This maximises the light received at the photodiode 76. Alternatively, two photodiodes may be included in the device 73, with one photodiode situated at each end of the series of elements 74a-74c. The currents output from the photodiodes may then be combined in parallel to give twice the current achieved in a single-photodiode device.

In order to achieve a greater current gain than is possible with a single OCT, a number of OCTs may be connected in parallel or in series-parallel combinations. For example, an impedance of 50 Q may be achieved by the parallel combination of loads greater than 50 0. The parallel combination of the series laser devices increases the net threshold current and drive voltage, but satisfies the 50 n impedance and increases the available current gain of the OCT. There are an infinite number of possible series and parallel combinations that might be used to achieve 50 Q compliance, so it will be possible to find one that best matches a given RF power while optimising the current gain.

As the input and output circuits 78,80 are decoupled, the OCT device 73 has useful application in the field of fibre optic links. For example, referring again to Figure 10, the optical transit between the input circuit 78 and the output circuit 80 may be achieved over a large distance, up to say 100 kin. Thus, simple optical fibre links with broadband characteristics having reduced insertion loss, and possibly gain, may be realised by deploying the input and output circuits 78,80 of the transistor at either end of the optical fibre link. This may be of particular advantage in the field of fibre optic links where signal insertion losses into fibre may be reduced. This provides a significant advantage over the OCT device known in the prior art, as there is only one optical output from the input circuit 78.

The OCT device 73 has so far been described as a device for converting an input electrical signal into an optical signal and then outputting an output electrical signal. An altemative configuration for the OCT is as an optical repeater 90 with optical gain, in which an input optical signal 92 is converted into an electrical signal which is then converted back into a greater optical output signal 94. One possible configuration of the optical repeater is shown in Figure 11. In this arrangement, the optical repeater 90 comprises a photodetector 96 for receiving the input optical signal 92 and converting it into an electrical signal, and the light emitting device 40, having three elements 74a,74b,74c, for converting this electrical signal into the optical output signal 94. In practice, it may be preferable to include an amplifier and biasing network in the laser input circuitry to allow for the inefficiencies that arise in the optical conversion processes. The details of a suitable biasing network would be conventional to one familiar with the art.

Since this optical repeater device 90 does not need to be 50 n compliant, a larger number of elements may be included in the device output circuit since the photodetector is a current source capable of driving high load resistances. This further enhances the degree of signal amplification that is possible.

26

Claims

1. A light emitting device, having an input impedance and a device quantum efficiency, for generating a beam of output radiation from an input current of electrons comprising; at least two optical elements for converting the input current of electrons into a beam of output radiation, each optical element having an impedance and an individual quantum efficiency, the optical elements being electrically connected such that the device quantum efficiency is greater than the individual quantum efficiency of one of the optical elements, characterised in that the optical elements are optically, connected and have a common optical path through which the beams of output radiation from the optical elements are transmitted.

2. The light emitting device of claim 1 wherein the optical elements are electrically connected such that the device input impedance is substantially equal to 50 Q without additional resistive impedance matching.

3 i. The light emitting device of claim 2, having a device modulation frequency limit, wherein the device input impedance is substantially equal to 50 Q across a frequency range substantially from DC to the device modulation frequency limit.

4. The light emitting device of claim 1, wherein one optical element has an end face coated with a reflective coating such that the light emitting device provides a single optical output.

5. The light emitting device of clairn 1, wherein the optical elements are quantum. cascade lasers.

tl-

6. The light emitting device of claim 1 wherein the optical elements are p-n junctions.

7. The light emitting device of claim 6, wherein the p-n junctions are laser diodes.

27

8. The light emitting device of claim 7, wherein the laser diodes are any one of A1GaAs, A1GaInP, A1GaInAs or A1GaInAsP laser diodes.

9. The light emitting device of claim 5 or 6, wherein each of the optical elements share a common optical waveguide through which radiation output from each of the optical elements is transmitted, the common optical waveguide comprising a bridging region between each of c Z.> optical elements wherein the or each bridging region electrically isolates the optical elements.

10. The Ii ht emitting device of claim 9, wherein the or each bridging region is electrically 9 Z=1 insulating by means of proton implantation.

11. The light emitting device of claim 9, wherein the or each bridging region is optically transparent.

12. The light ernittino device of claim 11, the common optical liaving an active layer.

L_ t> - c whercin the active layer is rendered transparent the bridoing region.

13. An optically coupled transistor for generating an output electrical signal comprising; the light emitting device of any of claims 1- 12 for emitting a beam of output radiation and a photodetector for detecting the beam of radiation output from the light emitting device and for converting the beam of output radiation into an output electrical current, wherein the light emitting device and the photodetector are in electrical isolation, thereby inhibiting electrical feedback from the photodetector to the light emitting device.

0

14. The optically coupled transistor of claim 13 wherein the photodetector is a photodiode device.

28

15. The optically coupled transistor of claim 13, comprising two photodetectors, each photodetector arranged to detect a beam of radiation output from a different end of the common optical path.

16. The optically coupled transistor of any of claims 1-3) - 15, comprising an optical fibre for transmitting the beams of output radiation to the one or more photodetectors.

17. A fibre optic link comprising an optical fibre having an input endface and an output endface, and also comprising the light emitting device of claim 1, wherein the light emitting device is situated at the input endface of the optical fibre such that the beam of radiation output from the light emitting device is input to the optical fibre.

18. An optical repeater for receiving an optical input signal and generating an optical output signals comprising; a photodetector for receiving the optical input signal and converting the optical input signal into an electrical signal and the light emitting device of claim I for receiving the electrical signal and outputting an optical signal from the common optical path.

19. The optical repeater of claim 18 and also comprising amplification means for amplifying the electrical signal output from the photodetector.

20. The light emitting device of claim 1, arranged in series, in parallel or in series parallel combination with one or more other light emitting devices as in claim 1.