GB2439595A

GB2439595A - Quantum memory device

Info

Publication number: GB2439595A
Application number: GB0612841A
Authority: GB
Inventors: Robert Young; Andrew James Shields
Original assignee: Toshiba Research Europe Ltd
Current assignee: Toshiba Europe Ltd
Priority date: 2006-06-28
Filing date: 2006-06-28
Publication date: 2008-01-02
Anticipated expiration: 2026-06-28
Also published as: GB0612841D0; GB2439595B

Abstract

A photon is absorbed by a quantum dot memory device thus generating an exciton trapped by the quantum dot. One of the carriers of the exciton is removed from the quantum dot by applying a suitable bias whereas the other carrier of the exciton is prevented from leaving the quantum dot by a barrier. Removing one of the carriers from the quantum well prevents recombination and allows the spin state of the carrier remaining in the quantum dot, which is related to the polarisation state of the photon, to be preserved. The quantum dot memory device can be read by transferring a carrier back into the quantum dot for recombination to emit a photon. The polarisation state of the absorbed photon is recreated in the emitted photon. An optical cavity in the form of a pair of mirrors or a photonic crystal is used to couple photons into and out of the quantum dot memory device.

Description

1 2439595 A Quantum Memory Device The present invention relates to the

field of quantum memory devices, more specifically, the present relates to a memory structure where information is stored in the form of a quantum state encoded on a particle.

There has been considerable interest recently in the fields of quantum communication and quantum computing. In both fields, information is encoded onto single particles such as photons by setting the quantum state of the photon. Quantum communication relates to the field where encoded particles are used to transmit information between two sites, quantum computing refers to the manipulation of such encoded particles in order to perform certain logical operations. Quantum communication promises more secure transmission of information than its classical counterpart since interception of a signal by an eavesdropped is detectable. Quantum computation, in addition to high security, promises high efficiency when solving some classically difficult problems.

Important to both quantum communication and quantum computation is a memory that allows an encoded particle to be stored so that its quantum state and hence the information carried by the particle may be retrieved.

The present invention addresses this problem, and, in a first aspect provides a memory device comprising: an exciton trapping centre configured to trap at least one exciton; an optical cavity configured to couple to said trapping centre; and biasing means configured to apply an electric field to said trapping centre, wherein said device and said biasing means are configurable such that said device may be switched between a first state where a carrier of a first type has a higher probability of tunnelling out of said trapping centre before recombination of the exciton and a second state where at least one carrier is injected into said trapping centre to create an exciton. *

The carrier of the first type may be an electron or a hole.

The device of the present invention is capable of storing an arbitrary superposition state.

A quantum state is defined as a superposition of two, or more, classical states.

By preserving the spin state of the carrier which remains in the trapping centre, it is possible to recover the original polarisation state of the photon which initially excited the exciton within the trapping centre. There will be a known relationship between the polarisation of the original photon and that of the recovered photon. For example, the original photon may have the same polarisation as the recovered photon. Alternately, operations within the device may predictably affect the polarisation state of the recovered photon.

In a preferred embodiment, the trapping centre is a quantum dot, nano-crystal or impurity atom. A quantum dot may be a pure quantum dot comprising a "dot" of material or an interface fluctuation. An interface fluctuation is fonned when a few monolayers of one semiconductor with a narrow bandgap is embedded in-between layers of semiconductor with a wider bandgap, forming a two dimensional potential well. Fluctuations in this well's thickness are treated as quantum dots and can confine excitons.

Examples of quantum dot systems are InAs dots embedded in GaAs, GaAs quantum well fluctuations in A1GaAs, CdSe nano-crystals in Zn(S,Se) and hydrogen impurities in GaAs.

In general, the quantum dots may comprise most combinations of 3:5, 2:6 or 4:4 materials as long as they are embedded in a material with a higher bandgap. Further possible examples are (Al, Ga,, In etc.) with (N, P, As, Sb etc.); (Zn, Cd etc.) with (Se, Te etc.); or Ge with Si. .4

Examples of impurity atoms are H, Mn, Al, In, Si, Te, Zn, Be, C, Na, Mg, N, P, Sb, 0 impurities in GaAs. Also, other impurity atoms may be used in GaAs. Further impurities are Nitrogen vacancies in diamond, 9F in ZnSe and 31P in Si.

Preferably, the device comprises materials with zero or small nuclear spins, andIor small variations in the nuclear spins of the differing elements in the materials, to limit interactions between the spin-state stored by the memory and the surrounding nuclear spins in the material.

For the present invention to work, it is important that one carrier in the trapping centre has a higher probability of tunnelling out of the trapping centre before electron-hole recombination occurs. The band structure of the device may be configured in a number of ways which allows a bias to be applied which satisfies this requirement. A particularly useful structure comprises a barrier layer, wherein said barrier is configured to prevent tunnelling of carriers of a second type out of said trapping centre. A barrier layer is a layer with a larger bandgap than its adjacent layers.

Preferably, the trapping centre and the optical cavity are strongly coupled to allow for deterministic absorption and emission of photons. Strong coupling between a cavity and an emitter is achieved when the emitter-photon interaction is larger than the combined atomic dipole decay rate and the cavity field decay rate. When this is achieved the spontaneous emission process of the emitter is replaced is replace by a coherent periodic energy exchange between the emitter and the photon.

In summary, strong coupling is preferable because it allows higher operational efficiency of the device since the probability of photons being absorbed by the trapping centre and emitted from the trapping centre and the cavity is theoretically 100%.

However, the device will operated if the cavity and trapping centre are weakly coupled.

In order for the present invention to reliably store the spin state of the carrier which remains in the trapping centre, it is preferable if the device is configured such that there is negligible polarisation splitting of the exciton transition of the trapping centre. This is true when the two bright exciton states are degenerate in energy and can be achieved by careful fabrication of the exciton confining centre, or by applying a field which preferentially alters the energy of one of the bright exciton states with respect to the other.

In a preferred embodiment, the present invention further comprises a source of weak photon pulses, the device being configured to trap excitons created by said weak photon pulses. More preferably, the absorption linewidth of the trapping centre is matched to that of the weak pulse. A weak pulse may be generated by a single photon source or a highly attenuated source, such as a laser.

The polarisation state of the pulse may be set by a polarisation rotator or the like.

Therefore, the memory device may further comprise means to set the polarisation of the weak pulse.

However, there is an alternative method for creating a state to be stored by the device using quantum entanglement. If an entangled state is formed, for example a Bell State, a measurement of one of the entangled states will affect the other of the entangled states. Thus, if a photon shares a polarisation-entangled state with an exciton, a measurement of the polarisation of the photon will affect the spin state of the remaining exciton. Hence it is possible to encode an exciton which has its state entangled with that of another photon by performing a measurement on the photon.

One method of creating such an entangled state is based around the decay of a biexciton, since, the two bright intermediate exciton states may be degenerate in energy.

When the two bright intermediate states are degenerate in energy, there is negligible polarisation splitting of the exciton transition of the trapping centre. This may be achieved by careful fabrication of the exciton confining centre, or by applying a field which preferentially alters the energy of one of the bright exciton states with respect to the other. The field may be an external or internal electric, magnetic or strain field.

The photon which is emitted when the biexciton decays to an exciton will be entangled with the spin state of the remaining exciton.

Thus, the spin of an exciton within the trapping centre which is left after biexciton decay can be set by measuring the polarisation of the photon which was created by decay of the biexciton.

Thus, in a preferred embodiment, the memory device comprises means to create a biexciton within said trapping centre. The biexciton may be created by optical excitation of carriers or electrical injection of carriers. If the biexciton is created by electrical injection, the device is preferably configured without a tunnel barrier.

The memory device may be configured to alter the lifetime of any exciton states by tuning or detuning the cavity mode to which they are coupled. The exciton lifetime may be increased if the photon emitted by decay of the exciton is not resonant with the cavity mode. The resonance of the cavity may be set during the fabrication of the cavity, but it may also be tuned afterwards for example using temperature and/or strain.

Preferably, the cavity is resonant with the wavelength of the photon emitted due to decay of the exciton created in the second state of the device. This ensures prompt decay of the exciton created by injection into the trapping centre during the second state and prompt output of the photon created by this decay. Typically, the exciton which will be created in the second state of the device will be a charged exciton as more than one carrier of the first type will be injected back into the trapping centre. Therefore, preferably the cavity is resonant with a photon which is emitted due to decay of a charged exciton.

If the device is to operate using quantum entanglement, the device may farther comprise means to measure the polarisation of the photon to be emitted due to biexciton decay, or interfere this photon with a single photon whose polarisation is to be stored.

Interference of a single photon with an entangled photon in order to transfer information concerning the polarisation of the single photon is known as quantum teleportation.

In a preferred embodiment, the trapping centre is a quantum dot. One method of forming a quantum dot is by using strained layer epitaxy. Therefore, in a preferred embodiment, said trapping centre is a quantum dot and said quantum dot comprises a material which has a substantially different lattice constant to at least one of an adjacent layer to said quantum dot.

Another method of forming a quantum dot is to embed a few monolayers of one semiconductor with a narrow bandgap in-between layers of semiconductor with a wider bandgap, forming a two dimensional potential well. Fluctuations in this well's thickness are treated as quantum dots and can confine excitons. Therefore, in a preferred embodiment, said trapping centre is a well fluctuation quantum dot. A quantum dot formed by interface fluctuations does not have to have a significantly different lattice constant to its surrounding layers.

Preferably, the memory device will comprise a plurality of cells, where each cell comprises a trapping centre. Such a device is preferably configured so that a single trapping centre or multiple trapping centres are selectively switched between said first and second states. In a preferred arrangement a device is configured with separate cavities for each trapping centre, the device may be configured with a first contact and a second contact for applying a bias across each trapping centre. The first contact may be a common contact for all trapping centres or a selection of trapping centres within the device, the second contact may be a separate contact for each trapping centre. In an alternative arrangement, the first contact may be common to rows of trapping centres and the second contact common to columns of trapping centres, such that an individual trapping centre may be addressed.

In a further arrangement, a cavity may comprise multiple trapping centres. The multiple trapping centres preferably have differing and discrete emission/excitation energies because the trapping centres have different sizes. It is possible to optically excite an exciton or biexciton in one trapping centre in by tuning the resonance of the cavity to the trapping centre which is to be irradiated. Similarly, it is possible to select output from a particular trapping centre by tuning the cavity as required.

Alternatively, a single trapping centre may be excited by using a narrowly focussed beam of radiation.

In an embodiment of the invention, the trapping centre is located within the insulating part of a PIN structure. Alternatively, said device comprises a region doped with carriers of a first type and an insulating region, said trapping centre being located within said insulating region. Preferably, such a device may comprise a Schottky contact.

In a further preferred embodiment, the device comprises a second exciton trapping centre, said device being configured to transfer said carrier of a first type from said first exciton trapping centre to said second exciton trapping centre in said first state and from said second exciton trapping centre to said first exciton trapping centre in said second state.

Preferably the device of the above preferred embodiment comprises first and second barrier layers and said first and second exciton trapping centres are located between said first and second barrier layers.

Said cavity may comprise one or more Bragg mirrors. In a preferred embodiment said cavity is provided by a photonic lattice.

The photons are collected from the upper side of said device, said upper side being the side opposite the substrate of the device. Alternatively, photons may be collected from the lower (substrate) side of the device, or from the (lateral) side of the device. Photons may be collected from the side of a device by using a photonic lattice having an irregularity provided at one side of the trapping centre or a microdisk arrangement.

If a photonic lattice is used and the photonic lattice comprises holes, it is possible to tune the cavity by allowing a gas to condense onto the lattice, this will change the refractive index at the material/vacuum boundaries, the size of holes in the crystal and also strain the crystal.

Regardless of whether or not photons are collected from the upper side of the device, it is often difficult to make an electrical connection to the upper side of the device as it is difficult to electrically bond to a small area of contact metal on top of a narrow stack of layers. Thus, the device may comprise contact metal configured to make an electrical

A

connection to an upper side of the device, the device further comprising an insulator layer provided underneath said contact metal such that said contact metal may extend away from the upper side of said device.

In a second aspect, the present invention provides a method of operating a memory device, the device comprising: an exciton trapping centre configured to trap at least one exciton; an optical cavity configured to couple to said trapping centre; and biasing means configured to apply an electric field to said trapping centre, said method comprising applying an electric field such that said device may be switched between a first state where a carrier of a first type has a higher probability of tunnelling out of said trapping centre before recombination of the exciton and a second state where at least one carrier is injected into said trapping centre to create an exciton.

The present invention will now be described with reference to the following non-limiting embodiments in which: Figures Ia, Ib, Ic, id, le and lf are schematic band diagrams of a memory in accordance with a first embodiment of the present invention; Figures 2a, 2b, 2c and 2d are schematic band diagrams of a memory in accordance with a second embodiment of the present invention; Figure 3 is a schematic of a device in accordance with a further embodiment of the present invention; Figure 4 is a variation on the device of figure 3 where contact metal is provided such that bonding to the top of the mesa is not necessary Figures is a variation on the device of figure 3 where the top mirror region has been removed; Figure 6 is a variation on the device of figure 3 where the N and P regions have been interchanged; Figure 7 is a variation on the device of figure 6 where the N type region has been removed; FigureS is a variation on the device of figures 3 or 6 where the mirror regions are not doped; Figure 9 is a variation on the device of figure 4 where further insulation is provided in order to allow a small mesa to be fabricated which is still easily electrically contactable; Figure 10 is a variation on the device of figure 3 where emission from the device is collected through the substrate; Figure 11 is a variation on the device of figure 3 comprising a photonic lattice; figure 1 la is a side view of a device having a photonic lattice, figure 1 lb is a plan view of a device having a photonic lattice and figure lIc is a plan view showing a variation of the device of figure 1 Ib; Figure 12 is a device similar to figure 5 where an upper mirror is provided above the upper contact; Figures 13a, 13b, 13c and 13d are schematic band diagram of a memory in accordance with a further embodiment of the present invention demonstrating operation using a different carrier type and no tunnel barrier; Figure 14 is a device similar to figure 6 where the tunnelling barrier has been removed, the carrier type used to store the spin state is now the type with the longest tunnel time out of the centre; Figure 15 is a schematic of a device in accordance with a further embodiment of the present invention; the device contains multiple exciton centres and is capable of storing many qubits; Figure 16 is a schematic demonstrating how multiple devices may be incorporated onto a single chip, figure 16a is a side view and figure 16b is a plan view; Figure 17 is a variation on the design shown in figure 16, figure 17a is a side view and figure 1 7b is a plan view; Figures 1 8a, I 8b, 1 8c, I 8d, 1 8e and 1 8f are schematic band diagrams of a memory in accordance with a further embodiment of the present invention demonstrating operation Figure 19 is a variation of the device of figure 6 having two tunnel barriers and two rows of quantum dots.

The basic device and its operation will first be described with reference to figures la to if.

Figure Ia shows the basic band structure of the device 1. The device comprises a quantum dot 3 denoted by a sharp potential well 5 in the conduction band and a sharp potential peak 7 in the valence band.

Adjacent to quantum dot 3, barrier 9 is provided. In this particular example, the barrier 9 is provided on the right hand side of the quantum dot 3. The device is biased so that holes trapped in the valence band 7 of quantum dot 3 have a tendency to tunnel quickly out of the quantum dot 3 towards the left and electrons trapped in the conduction band of quantum dot 3 have a tendency to tunnel towards the right. However, tunnelling of electrons from the quantum dot 3 is inhibited by barrier 9.

The structure is then illuminated by weak light pulse 11. Weak light pulse 11 is polarised with a predetermined state. For example, it may be designed to encode information on the photons by denoting photons which have a vertical polarisation as bit zero and photons which have a horizontal polarisation as bit 1.

The polarised pulse 11 will excite an electron hole pair or exciton in quantum dot 3 (figure ic). Due to the biasing of the device (which has been described with reference to figure la) a hole 13 will quickly tunnel out of the valence band of quantum dot 3 before the electron hole pair can recombine to emit a photon.

The tunnelling of electron 15 of quantum dot 3 will be inhibited by barrier 9.

Therefore, electron 15 remains in quantum dot 3 as shown in figure ic and an electron is stored within the quantum dot 3 with a particular spin state. The spin state is related to the polarisation of the initial pulse of radiation and hence information concerning the polarisation state of the initial pulse of radiation is stored.

In order to recover the initial polarisation state, the device is set to a second state by biasing as shown in Figure Id. Here, the bias is set so that a hole or plurality of holes 17 may tunnel back into quantum dot 3.

Once holes 17 have tunnelled back into quantum dot 3 as shown in figure le, the electron 15 and hole 17 combine in order to emit a photon 19 as shown in figure if.

The spin state of the electron 15 sets the polarisation state of the emitted photon 19 thus, the polarisation state of the emitted photon 9 is related to the polarisation state of the weak pulse 11 which was initially used to excite the exciton (figures la and Ib).

Therefore, the polarisation state of the weak photon pulse 11 can be restored and recovered using the above system.

In order to configure the memory so that it is available to store information again, the device is biased as shown in figure Ia and the process starts again.

The device may also be set to a "middle" bias or rest state (not shown) where the bias is set between the first and second states. In the middle bias state, the tunnelling probability of electron 15 is reduced from the first bias state and the probability of holes being injected into the quantum dot 3 is lower than in the second bias state. The middle bias state may be used after the hole has tunnelled out of quantum dot 3.

Figure 2 shows a further variation on the device of figure 1. The device of figure 2 again has a quantum dot 31 which is configured with degenerate bright exciton states.

A biexciton is excited in quantum dot 31. The biexciton may be created by illumination or by electrically injecting carriers.

The biexciton lifetime i.e. the time which it takes for a single electron and hole from the biexciton to combine and emit a photon can be controlled due the Purcell effect by embedding the quantum dot in an optical cavity. After the biexciton lifetime, a single exciton 37 is formed in quantum dot 31 and the electron and hole which have recombined are emitted as a photon 39.

Photon 39 has its polarisation entangled with the exciton 37 remaining in quantum dot 31. By measuring the polarisation of the photon 39, the polarisation state of the exciton 37 can be set. Alternatively the photon 39 can be interfered at a beams splitter with a photon whose polarisation is to be stored, transferring the polarisation of the photon to the exciton by quantum teleportation.

The bias of the device is then set to the first state (of figure 1) as shown in figure 2c. In this state, the hole 41 of exciton 37 tunnels out of the valence band of quantum dot 31 leaving behind electron 43. Tunnelling of electron 43 is inhibited by barrier 33.

The electron 43 spin state is dependent on the measured polarisation of the photon 39 and can thus be set to a predetermined state. This predetermined spin state is stored in quantum dot 31 (as shown in figure 2d). When reading of the state is required, the device is set to a second bias state (as shown in figure Id) where a hole is allowed to tunnel back into quantum dot 31. The newly formed exciton in the quantum dot 31 then recombines to emit a photon with its polarisation set by the spin state of the electron 43.

A biexciton can then be excited in the quantum dot as shown in figure 2a in order for the process to start again.

Figure 3 schematically illustrates a device in accordance with an embodiment of the present invention. The device comprises lower mirror region 51. Lower mirror region 51 comprises alternating layers of having a relatively higher refractive index and having a relatively lower refractive index. In this particular embodiment, the layers of the lower mirror region are semiconductor layers as they are doped. Although all of the layers of the lower mirror region in this embodiment are doped, only the top layers of the mirror region, which have an electrical contact, need to be doped. For example, the higher refractive index layer 53 may be provided by GaAs and the lower refractive index layer may be provided by AIM layer 55. However, if the layers of the mirror region do not need to be doped, materials other than semiconductors may be used.

Each layer of lower mirror region 5! has a thickness of one quarter of the wavelength of the photons generated by the device divided by the refractive index. This lower mirror region 51 forms a distributed Bragg reflector.

In this embodiment, lower mirror region 51 is n-doped.

Above the lower mirror region 51 is formed optical cavity 57. Optical cavity 57 comprises a first cavity layer 59 of undoped relatively higher refractive index semiconductor, a barrier layer 61 provided overlying and in contact with said first cavity layer 59. The barrier layer 61 is formed from a high band-gap semiconductor barrier layer material such as InGaAs or AlGaAs.

Next, a second cavity layer 63 of higher refractive index semiconductor is formed. In this embodiment, the first and second cavity layers 59 and 63 are formed from the same material.

Next, approximately 1 to 10 monolayers of InAs 65 is formed overlying and in contact with said second cavity layer 63. Due to the lattice mismatch between InAs and GaAs, the InAs layer 65 forms as a quantum dot 67. The cavity is then finished with a third cavity layer 69 of GaAs semiconductor. The total thickness of the cavity region 57 is the wavelength of the photons which will be emitted or absorbed by the quantum dot 67 divided by the mean refractive index of the region 57.

Next, upper mirror layer 71 is provided overlying and in contact with said cavity 57.

Upper mirror layer 71 is fabricated in the same way as lower mirror layer 51. However, upper mirror layer 71 is fabricated with p-type doping. Upper mirror layer 71 is also thinner than lower mirror layer 51 in order to encourage emission through the top of the device.

The layers are then etched to form mesa 77. The mesa etch progresses to the lower doped mirror region to allow an electrical contact to be made underneath the cavity region 57.

Contact metal 73 is then provided overlying the upper surface of upper mirror 71. An aperture 75 is defined in the contact metal in order to allow emission from quantum dot 67 to be collected.

In order to operate the device of Figure 3, a bias is applied between contact metal 73 and n-type lower mirror layer 51. Applying a bias between these two points allows the bias to be switched between a first state where holes can tunnel out of the quantum dot 67 (but tunnelling of electrons is inhibited by barner 61) and a second state where holes can tunnel back from p-doped region 71 into quantum dot 67 in order to allow emission of a photon.

The above device can be operated in accordance with the method described in relation to Figure 1 or Figure 2.

The device of figure 4 is based closely on that of figure 3. In figure 3, a mesa 77 is defined which also provides a boundary in the direction perpendicular to the growth direction of cavity 57. In order to allow strong coupling between the quantum dot 67 and the cavity, it is desirable to restrict the width of the mesa 77 to the order of the photon wavelength. However, making a mesa this narrow can have implications for the fabrication of the device as it is difficult to make good electrical contact to the metal which is on top of the mesa 77.

In the device of figure 4, the same reference numerals will be used todenote the same features as those of figure 3.

Once the mesa 77 has been etched, an insulating material 81 is provided on the left-hand side of the mesa and at the bottom left-hand side of the mesa 83. This insulating material and the top of the upper mirror region 71 are then coated with contact metal 85.

By providing insulating layer 81, 83, it is possible to take the contact metal 85 away from the top of the mesa 77 and to an area where a large contact pad may be provided.

Electrical contact may then be made to this large contact pad and to n-doped lower mirror region 51 in order to apply the bias as described with reference to figure 3.

This allows the mesa to be designed as small as necessary without any regard for the difficulties in bonding to the top of the mesa stack 77.

Figure 5 shows a further variation on the device of figure 3. To avoid unnecessary repetition, like reference numerals will be used to denote like features.

In figure 5, upper mirror region 71 of figure 3 is removed and instead, part 91 of the third layer 69 of cavity 57 is p-doped. This doping provides holes to tunnel into the quantum dot 67 when the device is biased in the second state.

The metallisation 73 is provided to the upper p-doped layer 91.

Although not shown in figure 5, it is also possible for the device to be configured in the same way as figure 4 in order to allow the mesa 77 to be made as small as possible without bonding problems.

Figure 6 shows a further variation on the device of figure 3. However, in this device, the order of the layers are changed. The lower mirror region 51 is a p-doped region in figure 6.

Also, the order of the cavity layers 57 are changed so that the barrier layer 61 is provided above the quantum dot 67. Explicitly, the first cavity layer 59 is undoped then a few mono-layers of InAs are deposited on this layer. Due to the variation in lattice constant between lower cavity layer 59 and InAs layer 65, quantum dot 67 is fonned.

Overlying quantum dot layer is first upper cavity layer 101 and barrier layer 61 is provided overlying first upper cavity layer 101. An upper undoped cavity layer 103 is then provided overlying and in contact with barrier layer 61.

Upper mirror layer 71 is n-doped in contrast to figure 1.

Figure 7 is a variation on the device of figure 6. The layers are the same as for figure 6.

However, upper n-doped mirror region 71 is omitted. As there is no need to inject electrons into the quantum dot, there is no need to supply an n-doped region in the device. Therefore, the device can function without any n-doped layers. A thin schottky barrier is formed between the metal contact 73 and the cavity region 57.

FigureS shows a variation on the device of figure 1.

In figure 8, the layers are arranged in the same order as those of figure 6. However, the lower mirror region 51 is undoped. Further, the cavity 57 has a p-doped lower layer 111. The p-doped lower layer is provided in order to supply holes to be selectively injected into the quantwn dot as required.

A further layer is provided at the top of the cavity 57, this is upper n-doped layer 113.

The layers between p-doped lower layer 111 and upper n-doped layer 113 are identical to those described with reference to figure 6.

Formed overlying and in contact with said upper cavity layer 113 is upper mirror region 71. Upper mirror region 71 is not doped.

Metallization 115 is provided overlying and in contact with said upper mirror region 71 in order to provide an aperture 75 for the quantum dot 67. Provision of aperture 75 minimizes the amount of stray radiation which can enter cavity 57. Biasing is performed by making contact to n-type layer 113 and p-type layer 111 and applying a bias between these two layers.

In figure 8, the thickness of the cavity is one wavelength of the emission or absorption wavelengths of photons in the quantum dot 67 divided by the refractive index of the cavity. However, it is also possible to increase the thickness of the cavity to an integer number of wavelengths (divided by the refractive index of the cavity material).

It is still preferable to keep the exciton confining centre at an anti-node to the cavity mode.

Figure 9 is a further variation of the device of Figure 3. To avoid unnecessary repetition, like reference numerals will be used to denote like features.

A mesa 77 has been etched, insulating material 121 is provided over and around said mesa 77. Insulating material 121 is then etched back to provide a flat surface from the top of the mesa 77 to the top of insulating material 121. Metallization 123 is then provided overlying both insulating material 121 and mesa 77 in order for a large area contact to be made to the top of the mesa 77.

The device is biased and operated in the same manner as described with reference to Figures 1 and 2.

Figure 10 shows a further variation of the device of Figure 3. To avoid unnecessary repetition, like reference nwnerals will be used to denote like features.

In the device of Figure 10, the device is configured so that photons are emitted through the substrate 131. The substrate 131 is present in all devices previously described.

However, it is not shown in the previous devices. The lower mirror region 51 is provided overlying said substrate 131.

The thickness of the upper mirror region 71 is larger than the thickness of the lower mirror region 51 in order to encourage emission through the lower mirror region 5!.

Also, metallization 133 which overlies and is in contact with upper mirror region 71 overlies the whole of the mesa and no aperture is left as in the devices of Figures 3 to 9.

Figure 11 shows yet a further variation of the device of Figure 1. To avoid unnecessary repetition, like reference numerals are used to denote like features.

In the device as shown in Figures 3 to 10, confinement in the horizontal direction (i.e. perpendicular to the growth direction) has been achieved by etching mesa 77 and the sidewalls of the mesa providing the confinement.

Figure ha and b show a more sophisticated form of confinement in the horizontal dimension where confinement is provided by a photonic lattice. A photonic lattice is created by etching a pattern as shown in Figure lib. The pattern is created by etching a pattern of holes 141 through the layers of the device. In the embodiment, the etch is stopped in the lower mirror region.

The 3D confinement is provided as an irregularity 143 in the pattern of Figure 1 lb. Explicitly, at the irregularity 143, no holes are etched.

Contact to the top of the structure is made via top contact layer 145 which has holes 141 etched through it. There is an aperture in the contact layer 145 at irregularity 143 which allows photons to be emitted through the top of the structure and for photons to enter the structure.

A contact to the lower doped mirror 51 is made by etching through the layers to expose the lower mirror 51 at region 147.

Figure 1 ic shows a plan view of a photonic crystal in accordance with a further embodiment of the present invention. The plan view is similar to that shown in figure lib with two main differences. The first difference is that a line of holes 141 is missing along row 149. By removing this row of holes, photons are encouraged to leave the cavity along row 149. Thus, this design acts as a waveguide to allow for emission/excitation from the side of the structure.

The second difference is that the top contact 145 extends over the whole of the structure and there is no aperture directly above the quantum dot (not shown) at irregularity 143 since emission is collected from the side of the structure and not the top.

Figure 12 shows a further variation of the design. The device is similar to that described with reference to Figure 5. In this particular example, an upper mirror region 151 is formed overlying and in contact with the top of contacts 85 Upper mirror region 151 may be made from a different material to the bottom mirrors, for example dielectrics such as SiO2iTiO2.

Figures 1 to 12 have illustrated embodiments of the present invention where tunnelling out of the quantum dot in inhibited by using a barrier. However, the device may be configured so that a tunnel barrier is not necessary to inhibit tunnelling out of the quantum dot by carriers of one type.

Figures 13a to d show the schematic band structure of a device without a tunnel barrier and how the device may be operated. Without the tunnel barrier in the band structure the carrier which is least likely to tunnel from the quantum dot must be used to store the spin state.

Also, the device of figure 13 is configured so that the hole is the carrier type which stores the information and not the electron. However, the device of figure 13 may be configured so that an electron may be used to store information.

The device of figure 13 has a quantum dot 231 which is configured with degenerate bright exciton states. A biexciton is excited in quantum dot 231 as shown in figure 13a.

The biexciton may be created by illumination or by electrically injecting carriers. An electron reservoir 234 is present. In order to electrically inject holes, a large bias is required.

The biexciton lifetime i.e. the time which it takes for a single electron and hole from the biexciton to combine and emit a photon can be controlled due the Purcell effect by embedding the quantum dot in an optical cavity. After the biexciton lifetime, a single exciton 237 is formed in quantum dot 231 and the electron and hole which have recombined are emitted as a photon 239 as shown in figure 1 3b.

Photon 239 has its polarisation entangled with the exciton 237 remaining in quantum dot 231. By measuring the polarisation of the photon 239, the polarisation state of the exciton 237 can be set. Alternatively the photon 239 can be interfered at a beam splitter with a photon whose polarisation is to be stored, transferring the polarisation of the photon to the exciton by quantum teleportation (not shown).

The bias of the device is then set to the first state (of figure 1) as shown in figure 13c.

In this state, the electron 243 of exciton 237 tunnels out of the valence band of quantum dot 231 leaving behind hole 241. The electron 243 tunnels first since it has a shorter tunnel time than the hole 241.

The hole 241 spin state is dependent on the measured polarisation of the photon 239 and can thus be set to a predetermined state. This predetermined spin state is stored in quantum dot 231. When reading of the state is required, the device is set to a second bias state as shown in figure 13d where electrons are injected back into quantum dot 131. The newly formed exciton in the quantum dot 131 then recombines to emit a photon with its polarisation set by the spin state of the hole 141.

A biexciton can then be excited in the quantum dot as shown in figure 1 3a in order for the process to start again.

The device of figure 13 may be controlled via a Schottky gate (not shown).

Figure 14 shows a schematic layer structure of a device which operates as explained with reference to figure 13. The device has a similar layer structure to that of figure 7.

Therefore to avoid unnecessary repetition, like reference numerals will be used to denote like features.

The lower mirror region 51 of the device of figure 14 remains identical to that of figure 7, except that the lower mirror region in figure l4is n-doped The optical cavity 57 is configured differently. Explicitly, the first cavity layer 59 is undoped then a few mono-layers of InAs are deposited on this layer. Due to the variation in lattice constant between lower cavity layer 59 and InAs layer 65, quantum dot 67 is formed. Overlying quantum dot layer is a single upper cavity layer 201. The contact layer 73 is formed overlying and in contact with said upper cavity layer 201.

Thus, there is no tunnel barrier between the doped mirror region 51 and the quantum dot 67. The carrier type used to store the spin must be the type which has the least probability of tunnelling from the quantum dot. A thin Schottky barner between the metal contact 73 and the cavity region 57 should allow for a limited number of the opposite carrier to those abundant in the doped mirror 51 to be injected into the quantum dot given sufficient bias. In this way this design can be used to populate the quantum dot with a biexciton by electrical injection.

The device is operated as explained with reference to figure 13.

Figure 15 is a variation of the device of figure 6, containing more than one quantum dot 67 in the anti-node of the optical cavity 57. The quantum dots have differing and discrete emission/excitation energies which can individually be tuned to be on-resonance with the cavity by for example, altering the temperature of the cavity, or

applying an external field.

The quantum dots have differing and discrete emission/excitation energies because the quantum dots have different sizes, shapes, compositions etc. The distribution of sizes of quantum dots can be controlled by the growth conditions of the quantum dot layer 65.

It is possible to optically excite an exciton or biexciton in one quantum dot by tuning the resonance of the cavity to the quantum dot which is to be irradiated. Similarly, it is possible to select output from a particular quantum dot by tuning the cavity as required.

Alternatively, a single quantum dot may be excited by using a narrowly focussed beam of radiation.

Figure 16a and b is a schematic demonstrating how a large number of the devices 301, 303, 305 show in figures 3-12, 14, 15 could be useflully incorporated on a single chip'.

Figure 16a shows a cross sectional view and figure 16b shows a plan view.

The device of figure 16a is based on the layer structure of the device described with reference to figure 6. To avoid unnecessary repetition, like reference numerals will be used to demote like features.

The mesa extends down through the cavity region 57 so that each quantum dot is enclosed in a single cavity. The mesa etch stops within the lower doped layer 51 so that a common lower electrical contact can be made to all devices 301, 303, 305.

Figure 16b shows a plan view, top contacts to each of the devices 301, 303, 305 are made individually and bottom contacts are common. The top contact to devices 301, 303 and 305 can be made by bonding to the top of each device or the contact metal may be configured as shown in figure 4 to allow bonding to occur away from the stack. By keeping the top contacts separate from one another, it is possible to individually address each device 301, 303, 305.

Figure 17 is a variation on the design shown in figure 16, top contacts of devices 301, 303, 305 in the same column are now common, as are bottom contacts of devices in the same row.

Figure 17a is a side view of devices 301, 303 and 305. The devices have the layer structure described with reference to figure 6. However, any of the variations described with reference to figures 3 to 13 and 14 to 15 may be used.

In figure 16 there is a common bottom contact formed by lower doped mirror between all devices. In figure 17, the bottom contact is again formed by lower doped mirror 51.

However, devices 301, 303 and 305 which are arranged in a row share the same bottom contact. However, an isolation etch 307 is performed between rows so that the bottom contact is only common to all devices in a row.

The top contacts are connected in column 309. This allows each device to be individually addressed by applying a suitable bias to the column and row of a selected device.

In the previous examples, the carrier which has tunnelled out of the quantum dot during the first biasing state is stored in a reservoir. However, the carrier may be stored in a second quantum dot during the first biasing state and then transferred from the second quantum dot in the second biasing state.

Figure 1 8a shows the basic band structure of the device 401. The device comprises a first quantum dot 403 denoted by a sharp potential well 405 in the conduction band and a sharp potential peak 407 in the valence band. A second quantum dot 404 is provided spaced apart from said first quantum dot 403.

Adjacent to first quantum dot 403, first barrier 409 is provided. In this particular example, first barrier 409 is provided on the left hand side of first quantum dot 403. A second barrier 410 is provided adjacent to the second quantum dot 404. In this particular example, the second barrier 410 is provided on the right hand side of the second quantum dot.

The structure is then illuminated by weak light pulse 411. Weak light pulse 411 is polarised with a predetermined state. For example, it may be designed to encode information on the photons by denoting photons which have a vertical polarisation as bit zero and photons which have a horizontal polarisation as bit 1.

The polarised pulse 411 will excite an electron hole pair or exciton in first quantum dot 403 (figure 1 8b). The photon is captured in the first quantum dot 403 because the first quantum dot is resonant with the cavity which surrounds the first and second quantum dots. The device is biased so that electrons trapped in the conduction band 407 of quantum dot 403 have a tendency to tunnel quickly out of the quantum dot 403 towards the right and holes trapped in the valence band of quantum dot 403 have a tendency to tunnel towards the left. However, tunnelling of holes from the first quantum dot 403 is inhibited by first barrier 409.

Due to the biasing of the device (figure 18c) an electron 415 will quickly tunnel out of the valence band of quantum dot 403 before the electron hole pair can recombine to emit a photon. The electron 415 and its spin will then be stored in second quantum dot 404. Further tunneling of the electron 415 from second quantum dot 404 will be inhibited by second barrier 410.

The tunnelling of hole 413 from first quantum dot 403 will be inhibited by first barrier 409. Therefore, hole 413 remains in quantum dot 403 as shown in figure 18d and is stored within the quantum dot 403 with a particular spin state. The spin state is related to the potarisation of the initial pulse of' radiation and hence information concerning the polarisation state of the initial pulse of radiation is stored by both the electron 415 and the hole 413 for an arbitrary time.

In order to recover the initial polarisation state, the device is set to a second state by biasing as shown in Figure 1 8e. Here, the bias is set so that the electron 415 may tunnel back from the second quantum dot 404 to the first quantum dot 403. The device is configured so that the electron has a quicker tunneling time than that of the hole.

Otherwise, the hole would tunnel into the second quantum dot 404 during the second bias state before the electron 415 tunneled back into the first quantum dot 403.

Once the electron 415 has tunnelled back into quantum dot 403 as shown in figure 18e, the electron 415 and hole 413 combine in order to emit a photon 419 as shown in figure 18 Once tunnelling has occurred, the bias may be maintained or set back to a neutral state as shown in figure 1 8f.

The spin state of the hole 413 and the electron 415 sets the polarisation state of the emitted photon 419. Thus, the polarisation state of the emitted photon 419 is related to the polarisation state of the weak pulse 411 which was initially used to excite the exciton (figure 1 8a). Therefore, the polarisation state of the weak photon pulse 411 can be restored and recovered using the above system.

Figure 19 schematically illustrates a device which can operate as explained with reference to figure 18.

The device of figure 19 is similar to that of figure 6. Therefore, to avoid unnecessary repetition, like reference numerals will be used to denote like features. The lower mirror region 51 and the upper mirror region 71 are the same as those described with reference to figure 6. However, the cavity region 57 differs in that the cavity region of fIgure 19 has two layers of quantum dots sandwiched between two tunnel barriers.

Explicitly, the cavity region 57 comprises a first cavity layer 501 of undoped relatively higher refractive index semiconductor, a first barrier layer 503 provided overlying and in contact with said first cavity layer 501. The first barrier layer 503 is formed from a high band-gap semiconductor barrier layer material such as JnGaAs or AIGaAs.

Next, a second cavity layer of 505 of higher refractive index semiconductor is formed.

In this embodiment, the first and second cavity layers 501 and 505 are formed from the same material.

Next, approximately I to 10 monolayers of InAs is formed overlying and in contact with said second cavity layer 505 as first quantum dot layer 507. Due to the lattice mismatch between InAs and GaAs, the InAs layer forms first quantum dot 509.

Third cavity layer 511 is then formed overlying and in contact with first quantum dot layer 507. Second quantum dot layer 513 is then formed overlying and in contact with the third cavity layer 511. Second quantum dot layer 513 is formed in the same manner as first quantum dot layer 507.

Fourth cavity layer 515 is then formed overlying and in contact with second quantum dot layer 513. Second barrier layer 517 is then formed overlying and in contact with said fourth cavity layer 515. Finally, fifth cavity layer 519 is formed overlying and in contact with the second barrier layer.

The size requirements of the cavity 57 are the same as those previously described. The first, second, third, fourth and fifth cavity layers preferably comprise the same material.

The first and second barrier layers may also comprise the same material as each other.

In the previous examples, the quantum dots have been fabricated using a strained growth technique. However, the dots may also be formed using interface fluctuations.

Claims

CLAIMS: I. A memory device comprising: an exciton trapping centre

configured to trap at least one exciton; an optical cavity configured to couple to said trapping centre; and biasing means configured to apply an electric field to said trapping centre, wherein said device and said biasing means are configurable such that said device may be switched between a first state where a carrier of a first type has a higher probability of tunnelling out of said trapping centre before recombination of the exciton and a second state where at least one carrier is injected into said trapping centre to create an exciton.

2. A memory device according to claim I, wherein the trapping centre is a quantum dot, interface fluctuation, nano-crystal or impurity atom.

3. A memory device according to either of claims I or 2, further comprising a barrier layer, wherein said barrier is configured to prevent tunnelling of carrier of a second type out of said trapping centre.

4. A memory device according to any preceding claim, wherein the trapping centre and the optical cavity are strongly coupled.

5. A memory device according to any preceding claim, the device being configured such that there is negligible polarisation splitting of the exciton transition of the trapping centre.

6. A memory device according to any preceding claim, further comprising a source of weak photon pulses, the device being configured to trap excitons created by excitation by said weak photon pulses.

7. A memory device according to claim 6, wherein the absorption linewidth of the trapping centre is matched to that of the weak pulse.

8. A memory device according to either of claims 6 or 7, further comprising means to set the polarisation of the weak pulse.

9. A memory device according to any of claims Ito 5, further comprising means to create a biexciton within said trapping centre.

10. A memory device according to any preceding claim, wherein the cavity is resonant with the wavelength of the photon emitted due to decay of the exciton created in the second state of the device.

11. A memory device according to either of claims 9 or 10, further comprising means to measure the polarisation of the photon to be emitted due to biexciton decay or to interfere the emitted photon with a further photon having a set polarisation.

12. A memory device according to any preceding claim, wherein said trapping centre is a quantum dot and said quantum dot comprises a material which has a substantially different lattice constant to at least one of an adjacent layer to said quantum dot.

13. A memory device according to any preceding claim, comprising a plurality of trapping centres, and wherein the device is configured so that a single trapping centre or multiple trapping centres are selectively switched between said first and second states.

14. A memory device according to any preceding claim, wherein the trapping centre is located within the insulating part of a PIN structure.

15. A memory device according to any of claims 1 to 13, wherein said device comprises a region doped with carriers of a first type and an insulating region, said trapping centre being located within said insulating region.

16. A memory device according to any preceding claim, wherein said cavity comprises one or more Bragg mirrors.

17. A memory device according to any preceding claim, wherein said cavity is provided by a photonic lattice.

18. A memory device according to any preceding claim, wherein photons are collected from the upper side of said device, said upper side being the side opposite the substrate of the device.

19. A memory device according to any preceding claim, further comprising a second excitoR trapping centre, said device being configured to transfer said carrier of a first type from said first exciton trapping centre to said second exciton trapping centre in said first state and from said second exciton trapping centre to said first exciton trapping centre in said second state.

20. A method of operating a memory device, the device comprising: an exciton trapping centre configured to trap an exciton; an optical cavity configured to couple an incident photon to said trapping centre; and biasing means configured to apply a bias to said trapping centre, said method comprising applying a bias such that said device may be switched between a first state where a carrier of a first type has a higher probability of tunnelling out of said trapping centre before recombination of the exciton and a second state where at least one carrier is injected into said trapping centre to create an exciton.