GB2300300A

GB2300300A - Semiconductor optical modulator device and method for its operation

Info

Publication number: GB2300300A
Application number: GB9521401A
Authority: GB
Inventors: Andrew James Shields
Original assignee: Toshiba Cambridge Research Centre Ltd
Current assignee: Toshiba Europe Ltd
Priority date: 1995-04-26
Filing date: 1995-10-20
Publication date: 1996-10-30
Anticipated expiration: 2015-10-20
Also published as: GB9521401D0; GB2300299B; GB9508466D0; GB2300299A; GB2300300B

Description

1 1 2300300 SEMICONDUCTOR DEVICE AND METHOD FOR ITS OPERATION The present

invention relates to a novel form of semiconductor device and a new mode of operation for both the new device and for a known device. In particular, such devices are optical modulators.

An optical modulator is an electro-optic device where the intensity of a transrrdtted optical beam is switched between lower and higher values by a controlling applied voltage. An efficient optical modulator must have a large contrast between the transparent and opaque states, as well as having low loss in the transparent state.

US Patent No. US-A-4 872 744 describes and claims an optical modulator comprising a quantum well layer and means in the form of a field effect transistor for altering the density of free carriers within the well layer. The field effect transistor is in the general form of a high electron mobility transistor (HEMT) but configured to guide light through a region containing the well layer. The bias applied to the gate electrode makes the device relatively transparent or opaque to the light in transit.

The device according to the aforementioned US patent, when operated as described, still absorbs some radiation in the transparent state, even though it absorbs more in the opaque state. This has a number of disadvantages. First, it impairs the contrast between the opaque and transparent states. Second, it results in higher power being required for the optical input beam. The absorption of the transparent state also restricts the overall length of a waveguide than can be formed containing the device as an integral element, which limits its use in an integrated optical circuit.

It has now been discovered that these disadvantages can be overcome by operating the device in a new mode, in accordance with a 2 first aspect of the present invention, which provides a method of operating an optical modulator which comprises a quantum well layer in an arrangement for altering the density of free carTiers within the quantum well layer, which field effect transistor includes a gate electrode, the method comprising applying a variable bias potential between the gate electrode and another electrical contact of the device, and varying the bias potential applied to the gate electrode such that the modulator is substantially transparent to incident light when the carrier density in the quantum well layer is zero or negligible and more opaque to incident light when there is a finite carrier density in the quantum well layer.

This new method according to the present invention is applicable both to the device described in US-A-4 872 744 and to a new form of device which will be described further hereinbelow.

The device of US-A-4 872 744 has both source and drain contacts to the quantum well layer. In that case the other electrical contact of the device which is biased relative to the gate electrode may be the drain contact. The source and drain contacts may be held at the same fixed potential. However, optionally, a fixed potential difference can also be applied between the source and drain contacts.

In the new form of device to be described further hereinbelow, a "back gate" structure is employed, i.e. a gate situated to the other side of the guided region relative to the side on which is situated either the primary gate as used in the device described in US-A-4 872 744 or relative to the side on which is situated a ridge structure. In other words such a back gate is situated on the substrate side of the guided region.

The method of the present invention thus overcomes the disadvantages of the mode of operation described in US-A-4 872 744 since the absorption in the transparent state is greatly reduced. The absorption in the transparent state of thi$ device has two origins. Firstly, 3 there is the residual valence-to-conduction band absorption remaining when excess electrons are added to the quantum well. Secondly, the excess electrons in the well cause absorption due to intra- and interconduction band transitions. With the method of the first aspect of the present invention, the transparent state is formed with negligible excess electron density in the quantum well, thereby almost eliminating intra- and inter-conduction band absorption by excess electrons in the transparent state. Indeed, the intra- and inter-conduction band absorption is significant only for the opaque state, thereby enhancing the transparent/opaque contrast ratio. Furthermore, the residual valence band-to-conduction band absorption in the transparent state is also reduced since the device is operated at a photon energy below the bandedge neutral exciton line at zero electron density, where the quantum well will be nearly transparent.

Optical excitation of electrons in semiconductors from the valence to conduction band results in a continuum of optical transitions, with a threshold photon energy equal to the bandgap of the material. It is known, that the electron in the conduction band and the "hole" it leaves behind in the valence band feel a mutual Coulombic attraction due to their opposite charge. The interaction leads to an electron-hole bound state being formed (consisting of one electron and one hole), referred to as a (neutral) exciton. These neutral excitons produce sharp resonances in the absorption spectrum at a photon energy below the threshold energy of the valence band-to-conduction band continuurn. The difference in the continuum threshold and lowest-energy, discrete excitonic transition energy is approximately equal to the binding energy of the electron and hole due to their mutual attraction. For a semiconductor quantum well the interband absorption threshold (at zero electron density) is due to the (neutral) exciton formed between the first electron and hole subbands.

4 It has been argued in the aforementioned US Patent that the absorption at the neutral excitonic transition energy is reduced by adding excess electrons to the quantum well. The absorption has two main components, (1) the ground state neutral exciton and (2) the continuum transition. The reduction in the continuum absorption relies upon the fact that transitions to the bottom of the conduction band are blocked by these states being occupied by electrons. Hence as the electron Fermi energy increases, the effective bandedge threshold is shifted to higher energy (the Burstein-Moss effect).

As pointed out in the US Patent, the continuum absorption is not completely removed, leaving some residual absorption in the transparent state. In fact, this problem is much worse than anticipated in that document, since the quantum well bandgap shifts to lower energy with increasing electron density, cancelling most of the shift of the effective absorption edge to higher transition energy. Hence the absorption of the device of the US Patent in its "transparent" state would be actually quite significant.

Another source of loss in the transparent state in the device of the US patent arises from the intra- and inter-conduction band absorption processes, which are also not discussed in that document. This absorption is caused by excess electrons in the conduction band being excited to higher energy states in the same or higher conduction bands. An excess electron density excess electron density of 10 12 cm- 2 has been estimated to be required to produce the transparent state according to the mode of operation in US-A-4 872 744. That corresponds to a 3D density of 1018 cm-3 in a 100 A wide quantum well. Adachi, "GaAs and Related Materials", (World Scientific, Singapore, 1994), p441) gives a plot of the absorption coefficient due to inter-conduction band processes in doped bulk GaAs. From this plot the absorption coefficient below the 1 fundamental gap for a density of 10 18 cm-2 can be estimated to be roughly 10 cm-1.

Although not wishing to be bound by any particular theory or explanation, the applicants believe that the key differences between the mode of operation described in US-A-4 872 744 and that of the present invention are two-fold. First, the device as described in the US Patent operates at the energy of the neutral band edge exciton in amplitude modulation mode and operation according to the present invention is at an energy below this. Furthermore, for operation according to the US Patent, the transparent and opaque states are formed with finite and negligible electron density in the quantum well, respectively. With the present invention this situation is reversed.

The description hereinbelow of one embodiment of a method according to the present invention concerns a device which consists of a waveguide with a central region containing a quantum well layer whose density of free electrons can be varied by application of a voltage between the Schottky gate and drain of an integral field effect transistor. The light propagating through the waveguide is chosen to have an energy below the neutral excitonic absorption of the quantum well when depleted of excess electrons. Hence when the Schottky gate is biased such that there are no electrons in the quantum well, the light can propagate through the waveguide with relatively little attenuation. This corresponds to the transparent state of the device. When excess electrons are added to the quantum well, by changing the gate voltage, the photon energy required for excitonic absorption in the quantum well is lowered, thereby increasing the absorption of the light and turning the device into its opaque state.

There will also be some enhancement of the light absorption when excess electrons are added to the quantum well (opaque state for the present invention) due to inter- and intra-conduction band transitions.

6 Also, when excess electrons are added to the quantum well, the photoexcited electron and hole can bind a second excess electron, forming a bound complex consisting of two electrons and one hole, referred to as a negatively-charged exciton. The transition energy of the negativelycharged exciton is lower than that of the neutral exciton by an amount roughly equal to the binding energy of the second excess electron to the electron/hole core of the exciton.

Fig. 7 of the accompanying drawings shows the absorption of the quantum well with a negligibly small (solid line) and finite (dashed) electron density in the quantum well. It can be seen that the device is operated at a photon energy below the neutral excitonic energy at zero electron density. Adding excess electrons to the quantum well causes absorption strength to be removed from the neutral exciton and transferred to the negatively charged exciton, which lies to lower energy and lies close to the operating photon energy. Hence adding excess electrons increases the absorption at the operating photon energy and turns the device into its opaque state.

Hence, the method according to the first aspect of the present invention relies on the ability to increase the absorption of the quantum well by introducing excess electrons to it. The operating photon energy is chosen to be below the neutral exciton resonance with negligible electron density in the quantum well (solid line in Fig. 7). Adding.excess electrons (by changing the gate voltage) to the quantum well enhances the formation of negatively-charged excitons and thereby increases the absorption at the operating photon energy (dashed line in Fig. 7). A suitable operating photon energy is marked "A" in Fig. 7.

With the method according to the first aspect of the present invention, in order to achieve the strongest absorption in the opaque state, the device should be operated (in the opaque state) at the electron density, 7 where the absorption of the negati vely- charged exciton is strongest. This will be achieved at a moderate electron density, which can be estimated as follows. A negatively-charged exciton is formed when a photon is absorbed in the vicinity of an excess electron. Hence as the excess electron density increases, the absorption strength of the negativelycharged excitons increases at the expense of the neutral ones, as the excess electrons cover an increasingly large area of the quantum well. The electron density where the negatively-charged exciton will be formed in nearly all areas of the quantum well, and therefore have maximum absorption strength, can be roughly estimated from the reciprocal of the area of the negatively-charged exciton in the quantum well plane.

Taking the example of a GaAs quantum well and a rough estimate of the inplane radius of the negatively charged to be twice that of a 3D neutral exciton, yields an excess electron density of about 3 x 1010 cm-2. This gives a rough estimate of the optimum electron density for the opaque state, which may be improved by a more detailed theoretical analysis and by taking account of the effect of localisation of the electrons by spatial inhomogenities.

It will be appreciated that the mode of operation of the present invention can be utilised with devices exactly as described in US-A-4 872 744, choosing an appropriate gate bias to obtain a sufficiently low electron density in the well layer to allow utilisation of negatively- charged excitons. However, other appropriate structures can be envisaged.

Generally speaking, any heterostructure can be used which allows induction of a so-called two-dimensional electron gas in a quantum well or wells or active layer by virtue of carriers derived from a doped layer or doped layers, and in which light can be transmitted through a region containing the well(s). Optionally, an undoped spacer layer is located between the doped and quantum well layers. However, as the present 8 invention relies on lower electron densities in the quantum well, as compared with the mode of operation described in US-A-4 872 744, the presence of a spacer layer is preferred. Therefore, such a spacer layer may be relatively thick, e.g. 30-100 nrn.

Although this discussion of the method of the first aspect of the present invention is in terms of adding excess electrons to the quantum well, it should be noted that optical modulation can also be achieved by introducing excess holes instead. In this case the doped barrier regions close to the quantum well in the field-effect transistor contains a surplus of acceptors rather than donors. This facilitates the density of excess holes in the quantum well to be varied via the gate-drain voltage. Again the operating photon energy is chosen so to be less than the bandedge excitonic transition with negligible excess hole density in the quantum well (transparent state). Increasing the excess holes density, by altering the gate-drain voltage, causes positively-charged excitons (bound complexes of two holes and one electron) to form in the quantum well. These have a lower transition energy than the neutral excitons, thereby increasing the absorption of the quantum well at the operating photon energy and forming the opaque state of the device. If the hole effective mass is larger than that of the electron, which is the case for most semiconductors, the binding energy of the second hole in a positively-charged exciton will be larger than that of the second electron in a negatively-charged exciton. Hence a device operating with excess holes may provide a larger transparent/opaque contrast ratio than one with excess electrons.

The method of the first aspect of the present invention will be described hereinbelow with reference to a first embodiment using a device generally in accordance with that described in US-A-4 872 744, i.e. a device in the form of a field effect transistor with source and drain regions contacting the quantum well layer. A Schottky metal gate electrode

9 overlies the quantum well layer between source and drain regions such that a bias voltage applied to the gate controls switching between the transparent and opaque states. The well, spacer and doped layers are disposed between respective cladding layers.

However, as mentioned above, a new form of device has now been devised which can be operated in the mode according to US-A-872 744 or in the mode according to the method of the first aspect of the present invention. This device constitutes a second aspect of the present invention which provides an optical modulator comprising an optical guide region including a quantum well layer, a waveguide structure defining an optical path of the guide region, a gate situated below the guide region for changing the carrier density in the quantum well layer, an electrical contact for said gate and at least one other electrical contact.

The waveguide structure of this new form of device preferably comprises a ridge situated above the optical guide region but other forms of waveguide structure are also possible.

Conveniently, in the device according to the second embodiment of the present invention, the electrical contact for the gate is located in a recess penetrating below the quantum well in the guide region. Alternatively, the back gate can be grown directly on highly-doped buffer and substrate layers and an ohmic contact made to the back of the structure.

Two embodiments of device according to the present invention are described. In one of these embodiments, the said at least one other electrical contact comprises respective source and drain contacts, making electrical contact to the quantum well in the said guide region, respectively on either side of the ridge. In this device it is possible to deplete areas of the quantum well outside the ridge region, unlike for the previous configuration where a front gate is arranged on top of the ridge. The advantage of this is explained below. However, an additional gate could be formed on top of the ridge. Such a device would then be comparable to the kind of structure described in US-A-4 872 744, but with an additional back gate.

Nevertheless, if such a further "front" gate is formed on top of the ridge, there is then no need for the source and drain contacts. Such a device constitutes a second embodiment of the device according to the second aspect of the present invention. In the specific form of this embodiment described hereinbelow, the ridge is bounded by recessed regions etched down to extend into the guided region. This is done so that there are no carriers in the optical guide region, outside the edges of the ridge. This is advantageous if the device is operated in the mode of the first aspect of this invention. If the device is to be operated according to the mode of US-A-4 872 744, it will be better to terminate the etch within the upper cladding layer, so that the areas of the quantum well outside the ridge remain occupied with carriers.

The advantage of the back gated structure is that all the quantum well can be depleted of carriers, unlike for the front gated device. This is advantageous if the device is operated in the mode of the first aspect of the present invention. This is because, for the front gated device, the areas of the quantum well adjacent and outside the stripe will not be depleted and will therefore absorb a fraction of the light when the device is in the transparent state. This absorption in the transparent state constitutes loss and is undesirable. However, it can be avoided in the back gated device, where all of the quantum well above the back gate will be depleted.

As mentioned above, devices according to the second aspect of the present invention can work either in the mode of the device as described in US-A4 872 744, or in the mode defined by the method of the first aspect 11 of the present invention. However, they are especially advantageous for operation in the latter mode.

The preferred embodiments of the second aspect of the present invention are based on a GaAs/AlxGa I -xAs heterostructure and the doped layer is ntype. However, the layers could be formed from a number of different materials. Nevertheless, those with strong bandedge excitonic resonances are preferred, e.g. InP/InxGa I -xAs, InxGa I -xAs/InyAl I -yAs, InP/InxAl 1 -xAs, InxGal-xAs/GaAs, GaInP/AlGaInP, CdTe/CdxZnl-xTe, CdTe/CdxMn I -xTe, ZnSe/Zn I -xMnxSe, Zn I -xCdxSe/ZnSe, ZnSYSe I -y/Zn I -xCdxSe, CdTe/Cd I -xZnxTe, GaN/AIN, GaN/AlxGa I -XN, InxGa I -xN/GaN, InxGa I.xN/AlyGa 1 -yN etc.

Various modifications to this basic structure are possible, for example:(1) The device described in the embodiments produces amplitude modulation where the absorption coefficient is changed at the operating photon energy. However, the principles outlined here can also be employed to produce a device working on the principle of phase modulation, relying on a small change in the refractive index. In this case a smaller operating photon energy is used which lies within the forbidden gap of the quantum well so reducing the absorption losses even further. On the other hand, such a device has the potential disadvantage that longer device lengths are required to produce comparable modulation. This applies both to devices according to the second aspect of the present invention and my devici intended for operation by the method of the first aspect of the present invention.

(2) For those embodiments which employ a metal Schottky gate on top of a ridge structure, the Schottky gate could be replaced by a doped semiconductor layer grown epitaxially along with the rest of the structure.

(3) The dopant atoms could be placed in the AIXGa I _xAs layer below, rather than, or in addition to, those above, the quantum well.

(4) The undoped spacer layer can be omitted.

(5) p-type dopants could be used instead of n-type. In this case excess holes are supplied to the quantum well, leading to the formation of positively-charged excitons.

(6) The upper cladding layer can be omitted or replaced by a dielectric with smaller refractive index than the guided region. This may be preferable if an AlxGal_xAs cladding layer is used, having a large background impurity concentration.

(7) A short-period superlattice structure (e.g. 50 periods in GaAs(25 A)/AlO.3GaO.7As (25 A) for the GaAs structure outlined here) can be grown before the quantum well layer, in order to improve the quality of the quantum well layer.

The modulator in use may be incorporated into an external waveguide structure for use as a discrete device or integrated into an optical circuit. It can be combined with other circuitry to form optical logic elements, as known to those skilled in the art.

The semiconductor layers of the device can be produced by a number of different methods, although epitaxial growth methods such as MOVPE, MOCVD, MBE, etc., are preferred.

The present invention will now be explained in more detail by way of the following description of preferred embodiments and with reference to the accompanying drawings, in which:-

Figure I shows a schematic partial cross-sectional view of a first embodiment of an optical modulator for operation in accordance with the first aspect of the present invention; Figure 2 shows a schematic cross-sectional view of the guided region waveguide structure of the modulator shown in Figure 1; 13 Figure 3 shows details of the layer structure of the modulator shown in Figures 1 and 2; Figure 4 shows a schematic plan view of a second embodiment of an optical modulator for operation in accordance with the first aspect of the present invention; Figure 5 shows a cross-sectional view of the modulator shown in Figure 4; Figure 6 shows details of the layer structure of the modulator shown in Figures 4 and 5; Figure 7 shows the absorption profile of a quantum well with negligible (solid line) and finite (dashed line) excess electron density in the quantum well; Figure 8 shows photoluminescence spectra obtained using the device shown in Figures 4-6 at different gate bias voltages; Figure 9 shows a schematic partial cross-sectional view of a third embodiment of an optical modulator which is in accordance with the second aspect of the present invention; Figure 10 shows a schematic cross-sectional view of the guided region waveguide structure of the modulator shown in Figure 9; Figure I I shows details of the layer structure of the modulator shown in Figures 9 and 10 and also of the fourth embodiment of the modulator shown in Figures 12 and 13; Figure 12 shows a schematic partial cross-sectional view of a fourth embodiment of an optical modulator which is in accordance with the second aspect of the present invention; and Figure 13 shows a schematic cross-sectional view of the guided region waveguide structure of the modulator shown in Figure 12.

Figure 1 shows the basic structure of a first embodiment of an optical modulator I for operation according to the first aspect of the 14 present invention. The modulator I comprises a lower cladding layer 3 above which is situated a waveguide region 5. Above the waveguide region 5 is an upper cladding layer 7.

The upper cladding layer 7 is selectively etched to produce a raised ridge 9 and a metal Schottky gate electrode 13 is formed over the ridge 9.

Respectively on either side of the ridge 9 are formed by conventional techniques, a source ohmic contact 15 and a drain ohmic contact 17 which penetrate the upper cladding layer 7 to permit electrical contact with the quantum well in the guided region 5 of the waveguide.

As will be explained in more detail hereinbelow, a light beam indicated by arrow 19 travels through a propagation region 21 of the waveguide region 5, directly under the ridge 9 and Schottky gate, parallel to and between the source and drain ohmic contacts 15,17. The intensity and/or phase of the beam 19 exiting from the face (not shown) remote from the entry face 23 is modulated according to means of applied potentials.

The layer structure of the waveguide region 5 is shown in Figure 2. It comprises a lower undoped barrier 25 below a quantum well or active layer 27. A middle undoped barrier or spacer layer 29 (which can be omitted) is situated above the quantum well layer 27 and a doped (p-type or n-type) barrier layer 31 is formed over the middle undoped barrier layer 29. Finally, an upper undoped barrier layer 33 (which can be omitted) is situated above the doped barrier layer 3 1.

Figure 3 shows the full layer arrangement of the structure shown in Figure 1, including the structure details of the waveguide region as shown in Figure 2. However, Figure 3 also shows that the structure of Figure I is formed on a substrate 35, a buffer layer 37 is interposed between the substrate 35 and the lower cladding layer 3, and that a GaAs layer (11) caps the growth after the upper cladding layer (7). The composition and is thickness of each layer shown in Figure 3, is as listed in Table 1 hereinbelow.

TABLE 1

STRUCTURE OF DEVICE qHO" IN FIG. 3 LAYER Capping layer I I Upper cladding layer 7 Upper barrier layer 33 Doped barrier layer 31 Middle barrier (spacer) layer 29 Quantum well layer 27 Lower barrier layer 25 Lower cladding layer 3 Buffer layer 37 Substrate 35 IMPURITY MATERIAL THICKNESS CONC (CM-3) GaAs l0nm AlyGl-yAs 0.5gm AIxGal-xAs 40nm AIxGal-xAs 0.2pm AIxGal-xAs GaAs AIxGal-xAs AlyGal-yAs GaAs GaAs 60run 2Onm, 0.3gm 2gm lgm unspec.

In Table 1 x = 0.3; y = 0.5 Si 1017 The complete structure is fabricated using a conventional epitaxial growth method. The thickness of the quantum well layer 27 is chosen to provide sharp excitonic resonances and with this particular heterostructure, could be any thickness from 3nm - 50nm. The carriers in the quantum well layer are of course induced by the doped barrier layer 33. Therefore, the thickness of the spacer layer 29 and of the doped barrier layer 33, as well as the doping concentration of the latter, are chosen to allow the electron 16 density to be varied between zero and the value for maximum absorption in the opaque state.

Referring again to Figure 2, the waveguide structure 5 consists of a central guided region sandwiched above and below by cladding layers 3,7 with a smaller refractive index than the average refractive of the guided region. For the GaAs/AIxGal-xAs heterostructure illustrated here, this can be achieved by composing the cladding regions 3,7 of AlxGa 1 -xAs having a larger average AI concentration (e.g. x--0.5) than in the guided centralregion. For optimum performance the thickness of the central region should be kept sufficiently narrow that only the fundamental optical mode is supported and with the quantum well layer 27 at the maximum intensity of the optical mode. The cladding layers 3,7 should be sufficiently thick so as to prevent significant leakage of the optical mode from the guided region.

The cladding layers 3,7 on top of and below the guided region confine the optical mode in the direction nomial to the layer planes. Figure 2 illustrates just one possible method of confining the optical mode in the direction perpendicular to the growth direction, i.e. in a structure commonly referred to as a stripe waveguide. The processing steps required to form a stripe waveguide are well known and need not be detailed here. However, other waveguide structures, known to those skilled in the art, are also suitable.

The stripe is etched into the upper cladding layer 7 using dry or wetetching through a photolithographical mask. Since the refractive index within the stripe region is larger than that in the adjacent etched region, the optical mode will be confined under the stripe region. The stripe can be about 1 - 5 gm wide and the etch depth is such that about 0. 3 gm of the upper cladding layer 7 remains in the etched region.

17 The Schottky metal contact 13 is evaporated on top of the stripe using standard techniques in order to act as the gate for the field effect transistor. The ohmic source and drain contacts 15,17 to the electron gas in the quantum well layer 27 are made using standard techniques on either side of the stripe forming the source and the drain of the field effect transistor. The electron density in the quantum well is varied by applying a voltage between the Schottky and drain contacts of the device. A second (optional) voltage is also supplied between the source and drain ohmic contacts so as to enhance the drift velocity of electrons travelling in the quantum well channel. This reduces the transit time for the electrons from the quantum well to the ohmic contact when a Schottky bias is applied so as to deplete the quantum well and thereby enhances the speed of the device. Maximum drift velocity is achieved in GaAs for an electric field of about 3.5 kV/cm

Figures 4 and 5 show a general schematic of a second embodiment of an optical modulator 41 for operation according to the present invention. It is generally analogous to the device shown in Figures I and 2.

A Schottky metal gate 43 is formed over the central area of a mesa region 45 and source 47 and drain 49 are formed at either end of the mesa 45 outside the gate 43 as ohmic contacts to a buried quantum well layer which will be described in more detail hereinbelow.

The mesa 45 is formed on a wafer 51 by defining a masked area using standard photolithography and then etching the exposed semiconductor to a depth below the quantum well layer. Photolithography is employed to deposit NiGeAu metal at either end of the mesa 45 which is then annealed to form the ohmic contacts 47,49 to the electrons in the quantum well layer.

A serr-- transparent Au layer is evaporated onto the central region of the mesa 45 to form the Schottky gate 43.

18 As illustrated and described, this device does not include a waveguide structure. However, it is to be understood that for operation as a modulator, a waveguide structure should also be incorporated, in the manner of the first embodiment.

Details of the layer structure within the mesa 45 are shown in Figure 6. The order of growth on a substrate 51 is as follows: lower buffer layer 53, middle buffer layer 55, upper buffer layer 57, lower barrier layer 59, quantum well layer 61, undoped spacer layer 63, n-doped layer 65 and finally, capping layer 67. The composition and thickness of each layer as shown in Figure 6 is listed in Table 2 hereinbelow..1AJ1Ia2, STRUCTURE OF DEVICE SHOWN IN FIG,6 LAYER Capping layer 67 Doped barrier layer 65 Spacer layer 63 Quantum well layer 61 Lower barrier layer 59 Upper buffer layer 57 Middle buffer layer 55 Lower buffer layer 53 Substrate 51 RAPURITY MATERIAL THICKNESS CONC (CM-3) GaAs AIxGal-xAs AIxGal-xAs GaAs AIxGal-xAs GaAs Superlattice GaAs GaAs 17mn 200run 6Onm 30fum 20run O.Sgm O.Sgm 1Agm unspec.

Si 1017 In Table 2, x = 0.33. The superlattice is GaAs (2.5nm)/AlGaAs (2.5nm) 19 The device of this second embodiment can be seen to be similar to that of the first embodiment. In particular, it has similar quantum well, undoped spacer and doped layers. The layers were grown by molecular beam epitaxy on a (100) oriented GaAs substrate. None of the layers were intentionally doped, except for the 2000 A doped Alo.33Gao.67As layer. The layer thicknesses and doping concentrations in Fig. 6 are the nominal values determined from the growth flux rates.

Waveguide cladding layers are omitted from Fig. 6 but these do not affect the light absorbing properties, although they will alter the operating voltage. The thickness of the back barrier layer is thinner than in the first embodiment. However, similar performance was obtained with samples with a thicker back barrier layer.

The excess electron density in the quantum well can be varied by biasing the Schottky gate with respect to the ohmic contacts. A negative applied gate voltage is required to deplete nearly all the electrons from the quantum well.

Fig. 8 plots photoluminescence spectra measured on this structure with different voltages applied to the gate relative to the source and drain contacts which are held at the same potential. The spectra are offset vertically for clarity. The measurements were taken with the incident and emitted light propagating almost normal to the layers. This is purely for investigative purposes. Operation as a modulator would require a waveguide structure to be incorporated and the incident beam to be directed as in the first embodiment. The incident light was at an energy below the bandgap of the Alo.33Gao.67As barriers. Evidence for the formation of negatively-charged excitons and enhancement of the optical transition strength below the energy of the zero-electron density bandedge exciton can be deduced from Fig. 8.

It can be seen in Figure 8 that at the lowest gate bias (i.e. -1.24V), corresponding to the smallest excess electron density in the quantum well, the spectrum is dominated by a single peak due to the neutral exciton (marked X in Fig. 8). As the gate bias is increased, so filling the quantum well with excess electrons, the neutral exciton line declines in intensity, while concurrently a transition to distinctly lower transition energy, due to the negatively-charge exciton (X-), is seen to strengthen and eventually dominate the spectrum (at - 1. OV).

The method of the first aspect of the present invention is designed to operate near the photon energy of the negatively-charged exciton seen at larger gate biases.

The spectra in Fig. 8 demonstrate that increasing the excess electron density in the quantum well causes a quenching of the absorption strength of the bandedge neutral exciton, with concurrent strengthening of the negatively charged exciton to lower photon energy. Furthermore, electroreflectance spectra recorded on the sample showed a qualitatively similar behaviour for the neutral and negatively-charged excitons with increasing excess electron density.

The first aspect of the present invention provides operation near the photon energy of the negatively-charged exciton seen at larger gate biases. It is apparent from Fig. 8 that there is a shift of the peaks due to the neutral and negatively-charged exciton to lower energy with increasing electron density. This shift is beneficial as it will tend to enhance the contrast between the absorption in the opaque and transparent states.

Turning now to Figures 9-11, there are shown structural details of one embodiment of a new form of optical modulator 71 in accordance with the second aspect of the present invention. This device comprises a lower cladding layer 73 (Figure 9) formed above a back gate layer 75. The back gate 75 and lower cladding layer 73 are about from 0. 1 - I.Ogm and 21 0.3 - 1.0 gm thick, respectively. The back gate 75 is either n-type or ptype doped to an impurity concentration of about 101 8cm-3 and is formed of GaAs or AlwGa(l-w)As.

It will be seen that the basic structure is otherwise very similar to that of the device shown in Figures I to 3, although the Schottky gate 13 of that structure is omitted (but optionally, it could still be included in the device of Figures 9-11). The layer thicknesses are otherwise the same as or similar to those of the device shown in Figures 1-3.

Above the lower cladding layer 73 is formed an optical guide region 77 and above the latter is formed an upper cladding layer 79. The upper cladding layer 79 is selectively etched to produce a raised ridge 81 to complete the waveguide structure defining the optical path 83 for an incident light beam 85, directly below the ridge 81.

On either side of the ridge 81 are respectively formed a source ohmic contact 87 and a drain ohmic contact 89 which penetrate the upper cladding layer 79 to permit electrical contact with the quantum well in the guide region 77.

The layer structure of the guide region 77 is shown in Figure 10. It comprises a lower undoped barrier layer 91 below a quantum well or active layer 93. A middle undoped barrier or spacer layer 95 is situated above the quantum well layer 93. A doped (p-type or n-type) barrier layer 97 is formed over the middle barrier layer 95 and an upper undoped barrier layer 99 is situated above the doped barrier layer 97. The middle barrier layer 95 and upper undoped barrier layer 99 are optional.

Electrical contact is made to the back gate 75 by an ohmic contact 101 having a contact pad 103 located in a recess 105 through the upper cladding layer 79 and guide region 77. Recessing of the gate contact 101 avoids making a short to the quantum well layer 93 in the guide region 77. This is done by etching a hole through a lithographically defined mask in 22 the surface to a depth just below the quantum well layer 93. A suitable metal is then deposited on a region, through another smaller lithographically defined mask, inside the etched region and annealed to form the ohn-dc contact to the back gate 75. For the case a GaAs n-type back gate, a suitable back gate ohmic metal is Ni-Au-Ge.

For the shallow source and drain ohmic contacts 87, 89, it is important that they do not spike into the back gate 75 itself and thus short the device. For an electron gas within the quantum well, for instance, shallow ohn-dc contacts are formed by deposition of Ge-Au and subsequent annealing. Shallow ohn-dc contacts can also be formed by deposition and annealing of Pd-Ge, after recessing the contact to within 100 A of the quantum well layer. An alternative method of making ohmic contacts to the quantum well, while avoiding making contact to the back gate, is described in Linfield et al, J. Vac. Sci. and Tech. B 12, 1290 (1994).

Figure 11 shows the full layer arrangement of the structure shown in Figure 9, including details of the guide region 77 shown in Figure 10. The structure is formed on a substratelbuffer layer structure 107 and completed with an upper cap layer 109. The composition and thickness of each layer shown in Figure 11 is listed in Table 3 hereinbelow.

23 TABLE3

STRUCTURE OF DEVICE SHO" IN FIG. 11 LAYER IMPU MATERIAL THICKNESS CON fúM-.1 Capping layer 109 GaAs lonn) Upper cladding layer 79 (12 1) AlyGal-yAs 0.5gm Upper barrier layer 99 (137) AIxGal-xAs 40nm - Doped barrier layer 97 (135) AIxGal-xAs 0.2gm 1 x 1017 Middle barrier (spacer) layer 95 (133) AIxGal-xAs 60nm - Quantum well layer 93 (13 1) GaAs 20nm Lower barrier layer 91 (129) AIxGal-xAs 0.31im - Lower cladding layer 73 (113) AlyGal-yAs 0.511m - Back gate 75 (115) GaAs 0.3un 1 x 1018 Buffer 107 GaAs lgm - Substrate 107 GaAs unspec.

In Table 3, x = 0-3; y = 0-5 The carrier density within the quantum well layer 93 is controlled by applying a voltage between the back gate 75 and the drain ohmic contact 89. Hence, the back gate 75 is used to vary the carrier density in the quantum well layer in a similar manner to the front gate 13 in the device of Figures 1-3.

For this structure, all of the quantum well can be depleted of carriers, unlike the structure in Figures 1-3, where only the region under the stripe is depleted. This is advantageous if the device is operated at photon energy A (see Figure 7), since the areas of the quantum well adjacent to the stripe in Figure 1, which are not depleted, will absorb a 24 small fraction of the light when the device is in the transparent state. This absorption in the transparent state constitutes loss and is undesirable.

Figures 12 and 13 show structural details of another embodiment of the new form of optical modulator in accordance with the second aspect of the present invention. In this device, the overall layer structure is the same as shown in Figure I I and tabulated in Table 3. Therefore, the reference numerals used in Figures 12 and 13 are shown bracketed in Figure 11 and Table 3.

This device 111, comprises a lower cladding layer 113 (Figure 12) formed above a back gate layer 115. These layers have the same thickness as the corresponding layers in the embodiment of Figures 9-11. Again, the back gate 115 can either be n-type or p-type doped and formed of GaAs or AlwGa(l -w)As.

This device differs from that in the embodiment of Figures 9-11 in that shallow source and drain ohmic contacts are not employed but instead, a front gate 117 is employed, similar to that in the device shown in Figures 1-3.

Above the lower cladding layer 113 is formed an optical guide region 119 and above the latter is formed an upper cladding layer 12 1. The front gate 117 is formed as a Schottky metal gate on top of a ridge 123. However, in this case, the etching to produce the ridge is taken through the upper cladding layer 12 1, part-way into the guided region 119, so as to deplete the region of the quantum well outside the ridge. Alternatively the etch can be terminated in the upper cladding layer, so as to leave carriers in the regions of the quantum well outside the ridge. Again, the upper ridge 123 defines the optical path 125 for an incident light beam 127, directly therebelow.

The layer structure of the guide region 119 is shown in Figure 13. It comprises a lower undoped barrier layer 129, above the lower cladding layer 113. Above the lower barrier layer 129 is formed a quantum well or active layer 131. A middle undoped barrier spacer layer 133 is situated above the quantum well layer 13 1. A doped (p-type or n-type) barrier layer 135 is formed over the middle barrier layer 133. An upper undoped barrier layer 137 is situated above the doped barrier layer 135 and above this is formed the upper cladding layer 121. Again, the middle barrier layer 133 and upper undoped barrier layer 137 are optional. It will be seen from Figure 13 that the etch to produce the ridge 123 is taken down through the upper cladding layer 121, the upper barrier layer 137 and the doped barrier layer 135, part way down into the middle barrier layer 133 so that the doped barTier layer 135 and upper barrier layer 137 are completely included in the ridge 123.

As with the device of Figures 9-11, electrical contact is made to the back gate 115 by an ohmic contact 139 having a contact pad 141 located in a recess 143 through the guided region 119. This is formed in the same way as for the back gate contact in the embodiment of Figures 9-11.

With the device operating at photon energy A (of the charged exciton) the waveguide is essentially transparent when the voltage between the front and back gate is chosen so that there are no (or very few) excess electrons in the well. At this photon energy A, the absorption is small.

By changing the voltage applied between the front and back gates excess electrons are induced in the area of the well underneath the ridge. This increases the absorption at photon energy A thus turning the device into its opaque state.

The device can also be operated at photon energy B, whereby it is opaque with no excess electrons in the well and more transparent when excess electrons are added to the well.

In the light of this disclosure, modifications of the described embodiments, as well as other embodiments, all within the scope of the

A?,:- 26 present invention as defined by the appended claims, will now be apparent to persons skilled in this art.

1^ 27

Claims

1. An optical modulator comprising an optical guide region including a quantum well layer, a waveguide structure defining an optical path of the guide region, a gate situated below the guide region for changing the carrier density in the quantum well layer, an electrical contact for said gate and at least one other electrical contact.

2. A modulator according to claim 1, wherein the waveguide structure comprises a ridge above the guide region.

3. A modulator according to claim 1 or claim 2, wherein the electrical contact for the gate is located in a recessed portion of the guide region.

4. A modulator according to claim I or claim 2, wherein the back gate is formed on a highly-doped buffer and substrate layer and the electrical contact for the gate is located on the substrate side of the structure.

5. A modulator according to any preceding claim, wherein the said at least one other electrical contact comprises respective source and drain contacts making electrical contact to at least one quantum well layer in said guide region.

6. A modulator according to claim 5 when dependent on claim 2, wherein the source and drain contacts respectively make electrical contact to at least one quantum well layer in said guide region on either side of said ridge.

28

7. A modulator according to any of claims 1 to 3, wherein the said at least one other electrical contact comprises a further gate situated above said waveguide structure.

8. A modulator according to claim 7, wherein said waveguide structure is a ridge which is bounded by recessed regions which extend down into said guided region.

9. A method of operating an optical modulator which comprises a quantum well layer in an arrangement for altering the density of free carriers within the quantum well layer, which arrangement includes a gate electrode, the method comprising applying a variable bias potential between the gate electrode and another electrical contact of the device, and varying the bias potential applied to the gate electrode such that the modulator is substantially transparent to incident light when the carrier density in the quantum well layer is zero or negligible and more opaque to incident light when there is a finite carrier density in the quantum well layer.

10. A method according to claim 9, wherein the said other electrical contact of the device is one of respective source and drain contacts.

11. A method according to claim 10, wherein the optical modulator is a modulator according to claim 4.

12. A method according to claim 11, wherein a further gate electrode is situated above said ridge.

29

13. A method according to claim 11, wherein no gate electrode is situated above said ridge.

14. A method according to any of claims 8-11, wherein the source and drain contacts are held at the same fixed potential.

15. A method according to any of claim 8-11, wherein a fixed potential difference is applied across the source and drain contacts.

16. A method according to claim 9, wherein the optical modulator is a modulator according to claim 7 or claim 8.

17. A method of operating an optical modulator which comprises a quantlim well layer in a field effect transistor arrangement for altering the density of free carriers within the quantum well layer, the method being substantially as hereinbefore described with reference to any of Figures 1 to 6 or 8 of the accompanying drawings.

18. An optical modulator substantially as hereinbefore described with reference to any of Figures 9-13 of the accompanying drawings.