Optical device

Abstract

An optical device includes: an input electrode to which an electric signal is input; a conversion unit that nonlinearly converts the electric signal and the optical signal to each other; a waveguide part for transmitting an optical signal generated by converting the electrical signal inputted to the input electrode by the conversion part; a plurality of output electrodes each outputting an electrical signal generated by converting the optical signal transmitted through the waveguide unit by the conversion unit; and a reflecting portion configured to reflect the optical signal transmitted through the waveguide portion toward the inside of the waveguide portion.

Description

Technical Field

[0001] The present invention relates to optical devices. Background Art

[0002] In recent years, a method called reservoir computing has been gaining attention as a hardware-based method for machine learning using neural networks. For example, research is underway on time-delay reservoir computing using a time-delay system using semiconductor lasers (see Non-Patent Document 1).

[0003] Prior art literature

[0004] Non-patent literature

[0005] Non-patent literature 1: Larger, L., Soriano, M., Brunner, D., Appeltant, L., Gutierrez, J., Pesquera, L., Mirasso, C. and Fischer, I.: Photonic information processing beyond Turing: An optoelectronic implementation of reservoir computing, Opt. Express, Vol. 20, pp. 3241-3249 (2012) Summary of the Invention

[0006] Technical problem to be solved by the invention

[0007] However, in time-delayed reservoir calculation, since the signal needs to be fed back with a time delay, there are time constraints in the process. Therefore, the application range of the reservoir calculation in the technology described in Non-Patent Document 1 is limited.

[0008] The present invention has been made in view of such circumstances, and one of its exemplary objects is to provide a technology that can flexibly utilize reserve pool calculations in various applications.

[0009] Methods used to solve technical problems

[0010] An optical device according to one embodiment of the present invention includes: an input electrode to which an electrical signal is input; a conversion portion that nonlinearly converts the electrical signal and the optical signal into each other; a waveguide portion that internally transmits an optical signal generated by converting the electrical signal input into the input electrode through the conversion portion; a plurality of output electrodes that respectively output electrical signals generated by converting the optical signal transmitted in the waveguide portion through the conversion portion; and a reflection portion that is configured to reflect the optical signal transmitted in the waveguide portion toward the interior of the waveguide portion.

[0011] It should be noted that any combination of the above-mentioned constituent elements and any form of expression of the present invention in the form of a method, an apparatus, a system, a recording medium, a computer program, etc. are also effective as aspects of the present invention.

[0012] Effects of the Invention

[0013] According to the present invention, it is possible to provide a technology that can widely and flexibly utilize reservoir calculations in various applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a schematic diagram of the information processing system according to the first embodiment.

[0015] Figure 2 It is a schematic three-dimensional view of the storage tank in the same embodiment.

[0016] Figure 3 This is a top view of a reservoir according to the same embodiment.

[0017] Figure 4 It is a cross-sectional view perpendicular to the Y-axis of the reservoir according to the same embodiment.

[0018] Figure 5 This is a cross-sectional view perpendicular to the Y axis of the input / output portion of the same embodiment.

[0019] Figure 6 This is a diagram showing the relationship between the current flowing in the active layer and the amplification and absorption of the optical signal in the active layer.

[0020] Figure 7 It is a cross-sectional view perpendicular to the Y axis for explaining the configuration of the reflection portion of the first embodiment.

[0021] Figure 8 This is a diagram for explaining an example of the transmission of photons in the reservoir cell according to the same embodiment.

[0022] Figure 9 This is a diagram for explaining the difference in the amplification factor of photons based on the adjustment electrode corresponding to the propagation direction of photons when the substrate is oriented in a plane orientation other than (100) and (111).

[0023] Figure 10 It is a cross-sectional view perpendicular to the Y-axis of the reservoir of the second embodiment.

[0024] Figure 11 It is a cross-sectional view perpendicular to the Y-axis of the reservoir of the third embodiment. DETAILED DESCRIPTION

[0025] (Implementation Method)

[0026] Below, with reference to the accompanying drawings, the scheme for implementing the present invention is described in detail. It should be noted that in the description of the accompanying drawings, the same reference numerals are used for the same elements, and repeated descriptions are appropriately omitted. In addition, the configuration described below is for illustration only and does not limit the scope of the present invention in any way.

[0027] (First embodiment)

[0028] Figure 1 1 is a schematic diagram of the configuration of the information processing system 1 according to the first embodiment. The information processing system 1 according to this embodiment includes a signal input device 15, a reservoir 10 (optical device), and an information processing device 18. The information processing system 1 is configured to execute reservoir calculations.

[0029] Reservoir computing, for example, can identify sequential inputs using simple algorithms based on the dynamic patterns generated by the reservoir. A reservoir is similar to a water reservoir. Throwing a stone into the water creates ripples that vary depending on the stone's size, shape, and speed. Therefore, the ripples reflect information about the stone's placement. Throwing multiple stones sequentially into the water can produce more complex ripples than with a single stone. The dynamic pattern of the ripples depends on the size and shape of the stones, as well as the order in which they were thrown. Therefore, the reservoir can be viewed as a device that converts sequential inputs into spatiotemporal patterns of ripples.

[0030] The reserve pool 10 is configured to function as a reserve pool in reserve pool calculation. The reserve pool 10 receives various electrical signals Sin from the signal input device 15 and outputs electrical signals Sout corresponding to the input electrical signals Sin to the information processing device 18 .

[0031] The signal input device 15 can input various electrical signals Sin into the reservoir 10. The electrical signals Sin are also transmitted to the information processing device 18. The information processing device 18 can perform learning processing based on the electrical signals Sout output from the reservoir 10 to generate a learning model, or use the learning model to identify the electrical signals Sin. The signal input device 15 and the information processing device 18 can include a CPU (Central Processing Unit), RAM (Random Access Memory), and ROM (Read Only Memory), as needed.

[0032] Figure 2This is a schematic perspective view of a reservoir cell 10 according to this embodiment. Reservoir cell 10 comprises a structure comprising stacked semiconductor layers. In this specification, the stacking direction of reservoir cell 10 is defined as the Z-axis, with the X-axis, Y-axis, and Z-axis directions being orthogonal to one another. Reservoir cell 10 according to this embodiment includes an input / output section 100 for inputting and outputting various electrical signals, and a reflector 140 disposed around input / output section 100.

[0033] Figure 3 1 is a diagram of a top view of the storage tank 10 of this embodiment. Figure 3 As shown, the upper surface of the input-output part 100 of the present embodiment is circular. Various electrodes are provided on the upper surface, which are separated from each other, such as an input electrode 118a, an adjustment electrode 128a, and output electrodes 138a, 138b, etc. The electrodes of the present embodiment are mainly rectangular, and the size of the electrodes can be, for example, a size of about 10μm to 100μm on one side. In addition, the number of input electrodes can be one or more. When the number of input electrodes is more than one, it is preferred that the output electrodes are more than the input electrodes. In addition, the upper surface of the input-output part 100 is not limited to a circle, and can also be various shapes such as a rectangle. In addition, Figure 3 Although rectangular electrodes are mainly shown in the figure, the shape of the electrodes can also be other polygonal or circular shapes.

[0034] When an electrical signal is input to the input electrode 118a, an optical signal corresponding to the input electrical signal is generated inside the reservoir 10. The generated optical signal is transmitted inside the reservoir 10. After the electrical signal is input to the adjustment electrode 128a, the optical signal is amplified inside the reservoir 10 based on the input electrical signal. The electrical signal input to the input electrode 118a can also be generated by the signal input device 15. In addition, the electrical signal input to the adjustment electrode 128a can also be generated by the information processing device 18. The optical signal inside the reservoir 10 is converted into an electrical signal, which is output to the information processing device 18 through the output electrodes 138a and 138b.

[0035] The reflector 140 reflects the optical signal transmitted inside the reservoir 10 so that the optical signal stays inside the reservoir 10 .

[0036] Figure 4 It is a cross-sectional view perpendicular to the Y-axis of the reservoir tank 10 of this embodiment. Figure 4 1 shows a cross section of the input / output unit 100 and the reflective units 140a and 140b constituting the reservoir 10. The reservoir 10 of this embodiment includes a substrate 180, a lower electrode 184 provided on the lower surface of the substrate 180, a structure in which a plurality of semiconductor layers are stacked on the substrate 180, and electrodes provided on the structure.

[0037] The input-output section 100 includes a substrate 180, a lower electrode 184, a lower cladding layer 182 provided on the substrate 180, a plurality of semiconductor layers stacked on the lower cladding layer 182, and an electrode provided thereon. Further, the reflection sections 140a, 140b each include a plurality of semiconductor layers stacked on the lower cladding layer 182, and an electrode provided thereon.

[0038] The lower electrode 184 is provided on the lower surface of the substrate 180, like the various electrodes. The lower electrode 184 of the present embodiment can also be used as a ground electrode. When an electric signal is input to the various electrodes provided on the upper surface of the input-output section 100, a current is generated between the electrode and the lower electrode 184, and an optical signal is generated from the current, amplified, or detected and output as an electric signal.

[0039] The substrate 180 is composed of various semiconductor substrates, and can be, for example, a GaAs substrate or an InP substrate. The substrate 180 is doped with impurities necessary for improving the conductivity. The thickness of the substrate 180 can be, for example, about 300 μm, or can be a thickness greater than or less than this. The materials of the various semiconductor layers stacked on the substrate 180 are appropriately selected in accordance with the material of the substrate 180.

[0040] The substrate 180 can also be, for example, a substrate oriented in a (001), (110), (112), (221), (111), (311), or the like. Each semiconductor layer stacked on the substrate 180 is affected by the orientation of the substrate 180. The substrate 180 is preferably oriented in a face orientation other than (001) and (111). When the substrate 180 is oriented in a face orientation of (001) or (111), the photons are amplified at the same amplification rate by the electric signal from the adjustment electrode 128a regardless of the direction of transmission of the photons. In this regard, it is known that when the substrate 180 is oriented in a face orientation other than (001) and (111) (for example, (311) or the like), the amplification rate can be made different depending on the direction of transmission of the photons (Reference 1). Therefore, the substrate 180 can more reliably achieve high dimensionality in the reservoir 10 by being oriented in a face orientation other than (001) and (111).

[0041] Reference 1: T.Ohtoshi, T.Kuroda, A.Niwa, and S.Tsuji: Dependence of opticalgain on crystal orientation in surface-emitting lasers with strained quantumwells,Appl.Phys.Lett.,Vol.65,No.15,pp1886-1887(1994)

[0042] Lower cladding layer 182 may also be made of a material corresponding to that of substrate 180. For example, when substrate 180 is a GaAs substrate, lower cladding layer 182 may be made of a GaAs-based semiconductor material. Furthermore, lower cladding layer 182 may be doped with impurities necessary to improve conductivity. The thickness of lower cladding layer 182 is, for example, approximately 1 to 2 μm.

[0043] Figure 5 This is a cross-sectional view perpendicular to the Y-axis of the input / output unit 100 of this embodiment. The input / output unit 100 receives an electrical signal from the signal input device 15 and outputs an electrical signal corresponding to the input signal. The input / output unit 100 of this embodiment includes an input unit 110, adjustment units 120a and 120b, and output units 130a and 130b. Hereinafter, the adjustment units 120a and 120b will be referred to as the "adjustment unit 120" unless otherwise specified, and the output units 130a and 130b will be referred to as the "output unit 130" unless otherwise specified.

[0044] The lower waveguide layer 102 is provided on the lower cladding layer 182, and the active layer 104 is provided on the lower waveguide layer 102. The lower waveguide layer 102 is configured to transmit an optical signal generated by converting an electrical signal in the active layer 104 (described later) therein. The lower waveguide layer 102 is configured to have a higher refractive index than adjacent layers, specifically, a higher refractive index than the lower cladding layer 182. This prevents the optical signal from being emitted from the lower waveguide layer 102 to the outside, and the optical signal is transmitted within the lower waveguide layer 102.

[0045] The thickness of the lower waveguide layer 102 may be, for example, approximately 0.1 to 1 μm. The lower waveguide layer 102 is formed of a material corresponding to the material forming the substrate 180. For example, when the substrate 180 is a GaAs substrate, the lower waveguide layer 102 may be formed of a GaAs-based semiconductor material, more specifically, AlGaAs or the like.

[0046] The active layer 104 is configured to convert optical signals and electrical signals into nonlinearities. Furthermore, the active layer 104 can absorb or amplify optical signals depending on the magnitude of the supplied current. In this embodiment, the active layer 104 includes, for example, quantum wells as a layer for converting photons (optical signals) and electrical signals into nonlinearities. For example, when the substrate 180 is a GaAs substrate, the quantum wells of the active layer 104 can be formed by stacking InGaAs and GaAs. The thickness of the active layer 104 can be, for example, approximately 0.1 μm.

[0047] The upper waveguide layer 105 is configured to propagate an optical signal generated by converting an electrical signal through the active layer 104. The upper waveguide layer 105 is provided on the active layer 104. The upper waveguide layer 105 is configured to have a higher refractive index than adjacent layers, more specifically, a higher refractive index than the upper cladding layers 114, 124, and 134 described later. This prevents the optical signal from being emitted from the upper waveguide layer 105, allowing the optical signal to remain within the upper waveguide layer 105, the active layer 104, and the lower waveguide layer 102 for a longer period of time.

[0048] The thickness of the upper waveguide layer 105 can be, for example, approximately 0.1 to 1 μm. The upper waveguide layer 105 is formed of a material corresponding to the material forming the substrate 180. For example, when the substrate 180 is a GaAs substrate, the upper waveguide layer 105 can be formed of a GaAs-based semiconductor material, more specifically, AlGaAs or the like.

[0049] Furthermore, impurities do not need to be added to the lower waveguide layer 102, the active layer 104, and the upper waveguide layer 105. This can suppress the loss of photons due to impurities.

[0050] The input unit 110, the adjustment units 120a, 120b, and the output units 130a, 130b include a plurality of semiconductor layers and electrodes stacked on the upper waveguide layer 105. Figure 5 In the embodiment, the output unit 130 is arranged at the end of the input / output unit 100 , but the present invention is not limited thereto. The input unit 110 or the adjustment unit 120 may be arranged at the end of the input / output unit 100 .

[0051] The input section 110 is configured to input an electrical signal from an external source (e.g., the signal input device 15). The input section 110 includes an upper cladding layer 114, a contact layer 116, and an input electrode 118, which are sequentially stacked in the Z-axis direction on the upper waveguide layer 105. Each of the upper cladding layer 114 and the contact layer 116 can be formed from a material corresponding to the material constituting the substrate 180. For example, if the substrate 180 is a GaAs substrate, each of the upper cladding layer 114 and the contact layer 116 can also be formed from a GaAs-based semiconductor material. Each of the upper cladding layer 114 and the contact layer 116 is doped with impurities necessary to improve conductivity. The thickness of the upper cladding layer 114 can be, for example, approximately 1 to 2 μm.

[0052] The input electrode 118 is provided on the upper surface of the stacked structure 101, specifically, on the contact layer 116. In this embodiment, the input electrode 118 receives an electrical signal from the signal input device 15. When the electrical signal is input to the input electrode 118, a current corresponding to the electrical signal flows through the active layer 104. In this embodiment, the active layer 104 generates photons corresponding to the electrical signal input to the input electrode 118. In this manner, the electrical signal input to the input electrode 118 is nonlinearly converted into photons by the active layer 104.

[0053] The adjustment section 120 includes an upper cladding layer 124, a contact layer 126, and an adjustment electrode 128, which are sequentially stacked along the Z-axis direction on the upper waveguide layer 105. The upper cladding layer 124 and the contact layer 126 of the adjustment section 120 may each have substantially the same configuration as the upper cladding layer 114 and the contact layer 116 of the input section 110.

[0054] The adjustment electrode 128 is provided on the upper surface of the stacked structure 101, specifically, on the contact layer 126. In this embodiment, an electrical signal is input to the adjustment electrode 128 from the information processing device 18. When the electrical signal is input to the adjustment electrode 128, a current corresponding to the input electrical signal is supplied to the active layer 104, thereby adjusting the amplification or absorption of the optical signal by the active layer 104.

[0055] Figure 6 Graph showing the relationship between the current flowing in the active layer 104 and the amplification and absorption of the optical signal by the active layer 104. Figure 6 In FIG. 1 , the horizontal axis shows the current flowing in the active layer 104, and the vertical axis shows the ratio of amplification and absorption of the optical signal of the active layer 104. Figure 6As shown, when the current is less than I1, active layer 104 absorbs the light signal. On the other hand, when the current is greater than I1, active layer 104 amplifies the light signal. When the current is I1, active layer 104 neither amplifies nor absorbs the light signal. In this specification, the phrase "active layer 104 neither amplifies nor absorbs the light signal" is referred to as "making active layer 104 transparent."

[0056] By adjusting the magnitude of the current flowing through the active layer 104, the behavior of the active layer 104 can be controlled. For example, by inputting an electrical signal to the adjustment electrode 128 so that a current greater than I1 flows through the active layer 104, the optical signal can be nonlinearly amplified in the active layer 104. Furthermore, by inputting an electrical signal to the adjustment electrode 128 so that a current equal to I1 flows through the active layer 104, the active layer 104 can be made transparent. Furthermore, by inputting an electrical signal to the adjustment electrode 128 so that a current less than I1 flows through the active layer 104, the active layer 104 can be caused to absorb the optical signal nonlinearly.

[0057] return Figure 5 , the configuration of the output section 130 is described. The output section 130 includes an upper cladding layer 134, a contact layer 136, and an output electrode 138, which are sequentially stacked along the Z-axis direction on the upper waveguide layer 105. The upper cladding layer 134 and the contact layer 136 of the output section 130 can each have substantially the same configuration as the upper cladding layer 114 and the contact layer 116 of the input section 110.

[0058] The output electrode 138 is provided on the upper surface of the stacked structure 101, specifically, on the contact layer 136. The output electrode 138 outputs the electrical signal generated by the nonlinear conversion of the optical signal by the active layer 104 to the outside (e.g., the information processing device 18). For example, when photons propagating through the upper waveguide layer 105 and the lower waveguide layer 102 are incident on the active layer 104 below the output electrode 138, electrons and holes are generated in the active layer 104. The output electrode 138 outputs an electrical signal corresponding to these electrons and holes.

[0059] In this embodiment, the input section 110, adjustment sections 120a and 120b, and output sections 130a and 130b are constructed by stacking multiple semiconductor layers and electrodes, separated from each other by gaps 106 to 109. For example, gap 108 is formed between the input section 110 and the adjustment section 120a, gap 106 is formed between the input section 110 and the adjustment section 120b, gap 109 is formed between the adjustment section 120a and the output section 130a, and gap 107 is formed between the adjustment section 120b and the output section 130b. This suppresses the effects of the current generated by the electrical signal input to the input electrode 118 and the current generated by the electrical signal input to the adjustment electrode 128 on the electrical signal output from the output electrode 138.

[0060] Gaps 106-109 can be formed, for example, by stacking various semiconductor layers on the active layer 104, forming electrodes thereon to integrally form a stacked structure for the input section 110, adjustment sections 120a, 120b, and output sections 130a, 130b, and then etching this stacked structure. Electrodes (e.g., input electrode 118, adjustment electrode 128, output electrode 138, etc.) can be formed after or before etching the gaps 106-109. In this embodiment, the depth of gaps 106-109 reaches the upper surface of the upper waveguide layer 105. It should be noted that the depth of gaps 106-109 can also reach the interior of the upper waveguide layer 105.

[0061] Furthermore, the aforementioned lower cladding layer 182 , lower waveguide layer 102 , active layer 104 , upper waveguide layer 105 , upper cladding layers 114 , 124 , 134 , and contact layers 116 , 126 , 136 constitute a stacked structure 101 .

[0062] Figure 7 This is a cross-sectional view perpendicular to the Y-axis, used to illustrate the structure of the reflector 140 of this embodiment. The reflector 140 is configured to reflect the optical signal transmitted through the waveguide of the input / output unit 100 back into the interior of the waveguide. The reflector 140 of this embodiment includes a first mirror portion 150a, a second mirror portion 150b, and an edge portion 160. While this embodiment illustrates an example in which the reflector 140 includes two mirror portions, the number of mirror portions may be three or more.

[0063] A gap 170 is formed between the output portion 130b and the first mirror portion 150a, a gap 172 is formed between the first mirror portion 150a and the second mirror portion 150b, and a gap 174 is formed between the second mirror portion 150b and the edge portion 160. The width of these gaps 170, 172, and 174 is d. Width d is one-quarter the wavelength of light, or an odd-integer multiple thereof. Furthermore, the width of the first mirror portion 150a and the second mirror portion 150b is d'. Width d' is the value obtained by dividing one-quarter the wavelength of light by the refractive index of the semiconductor, or an odd-integer multiple thereof. These gaps 170, 172, and 174 can be formed, for example, by stacking various semiconductor layers on the lower cladding layer 182, forming electrodes thereon to integrally form a stacked structure for the output portion 130b, the first mirror portion 150a, the second mirror portion 150b, and the edge portion 160, and then etching this stacked structure. The electrodes (specifically, the sixth mirror layers 156a, 156b, 166) may be formed after the gaps 170, 172, 174 are formed by etching, or may be formed before etching.

[0064] The first mirror 150a is provided at a position separated from the waveguide portion of the input / output unit 100 in the transmission direction of the optical signal. In this embodiment, the first mirror 150a is provided so as to form a gap 170 between the first mirror 150a and the output unit 130b (specifically, the lower waveguide layer 102, the active layer 104, and the upper waveguide layer 105). In this embodiment, the first mirror 150a is provided in the in-plane direction ( Figure 7 A position separated from the waveguide portion by a distance d in the X-axis direction).

[0065] The first mirror portion 150a includes a first mirror layer 151a, a second mirror layer 152a, a third mirror layer 153a, a fourth mirror layer 154a, a fifth mirror layer 155a, and a sixth mirror layer 156a, which are stacked in this order along the Z-axis direction on the lower cladding layer 182. Each of the first mirror layer 151a, the second mirror layer 152a, the third mirror layer 153a, the fourth mirror layer 154a, the fifth mirror layer 155a, and the sixth mirror layer 156a can have substantially the same material and thickness as the lower waveguide layer 102, the active layer 104, the upper waveguide layer 105, the upper cladding layer 134, the contact layer 136, and the output electrode 138 of the output portion 130.

[0066] The second mirror portion 150b is provided so that a gap 172 is formed between the first mirror portion 150a and the second mirror portion 150b. Specifically, the second mirror portion 150b is provided so as to be spaced from the first mirror portion 150a in the in-plane direction ( Figure 7The second mirror portion 150b includes a first mirror layer 151b, a second mirror layer 152b, a third mirror layer 153b, a fourth mirror layer 154b, a fifth mirror layer 155b, and a sixth mirror layer 156b stacked in sequence along the Z-axis on the lower cladding layer 182. Each of the first mirror layer 151b, the second mirror layer 152b, the third mirror layer 153b, the fourth mirror layer 154b, the fifth mirror layer 155b, and the sixth mirror layer 156b may have substantially the same material and thickness as each of the first mirror layer 151a, the second mirror layer 152a, the third mirror layer 153a, the fourth mirror layer 154a, the fifth mirror layer 155a, and the sixth mirror layer 156a.

[0067] The edge portion 160 is provided so that a gap 174 is formed between the second mirror portion 150b and the edge portion 160. Specifically, the edge portion 160 is provided in the in-plane direction ( Figure 7 The edge portion 160 includes a first mirror layer 161, a second mirror layer 162, a third mirror layer 163, a fourth mirror layer 164, a fifth mirror layer 165, and a sixth mirror layer 166 stacked in sequence along the Z-axis on the lower cladding layer 182. Each of the first mirror layer 161, the second mirror layer 162, the third mirror layer 163, the fourth mirror layer 164, the fifth mirror layer 165, and the sixth mirror layer 166 has substantially the same material and thickness as each of the first mirror layer 151a, the second mirror layer 152a, the third mirror layer 153a, the fourth mirror layer 154a, the fifth mirror layer 155a, and the sixth mirror layer 156a.

[0068] Thus, by providing the first mirror portion 150a, the second mirror portion 150b, and the edge portion 160, the reflective portion 140 can reflect the optical signal propagating through the lower waveguide layer 102, the active layer 104, and the upper waveguide layer 105 toward the interior of the lower waveguide layer 102 and the upper waveguide layer 105. Specifically, at the interface between the semiconductor and the gap in each layer constituting the output portion 130b, the first mirror portion 150a, or the second mirror portion 150b, photons are reflected in accordance with the difference in refractive index between the semiconductor and air.

[0069] For example, at interface A between the lower waveguide layer 102, active layer 104, upper waveguide layer 105, and gap 170, most of the photons propagating through the active layer 104 and upper waveguide layer 105 are reflected inside the output section 130b. Furthermore, most of the photons that pass through interface A and enter the first mirror section 150a are reflected toward the input / output section 100 at interface B between the first mirror section 150a and gap 172. Furthermore, most of the photons that pass through interface B and enter the second mirror section 150b are reflected toward the input / output section 100 at interface C between the second mirror section 150b and gap 174. Thus, by providing the first mirror section 150a and the second mirror section 150b, photon reflection occurs due to the difference in refractive index between the semiconductor and air, allowing the reflectivity to be set to, for example, 99%. Furthermore, these gaps 170, 172, and 174 are preferably deep, and for example, the gaps 170, 172, and 174 may intrude into the lower cladding layer 182. However, as in this embodiment, the gaps 170, 172, and 174 do not intrude into the lower cladding layer 182, or even if they are shallower in depth, can still function to reflect photons.

[0070] In addition, if Figure 4 As shown, reflective portions 140a and 140b are provided at both ends of the input / output portion 100. Therefore, photons transmitted inside the reservoir 10 are reflected by the reflective portions 140a and 140b at both ends, thereby allowing the photons to remain inside the reservoir 10.

[0071] Figure 8 1 is a diagram for explaining an example of the transmission of photons in the reservoir 10 of this embodiment. Figure 8 In the example shown, when an electrical signal is input from the signal input device 15 to the input electrode 118a, a photon P1 corresponding to the input electrical signal is generated. This photon P1 propagates through the interior of the reservoir 10 while being reflected by the reflector 140. When the photon P1 is converted into an electrical signal, the electrical signal is output by the output electrode 138a.

[0072] Furthermore, photons P2 generated based on the electrical signal input to input electrode 118a are reflected by reflector 140 or amplified or absorbed by adjustment electrode 128a, while propagating within reservoir 10. Amplified or absorbed photons P2′ are reflected by reflector 140 and propagated within reservoir 10. After photons P2′ are converted into electrical signals, the electrical signals are output by output electrode 138b.

[0073] Figure 9 This is a diagram for explaining the difference in the amplification factor of photons based on the adjustment electrode corresponding to the propagation direction of photons when the substrate 180 is oriented in a plane orientation other than (100) and (111). Figure 9 In the example shown, the substrate 180 can be, for example, a substrate oriented in a (311) face orientation.

[0074] In Figure 9 In the example shown, two photons P3, P4 are generated in accordance with the electrical signal input to the input electrode 118b. The arrow direction of the photons P3, P4 (or photons P3', P4') indicates the transmission direction of the photons. The two photons P3, P4 are transmitted in different directions, respectively. The photon P3 is amplified in accordance with the electrical signal from the adjustment electrode 128c, and the photon P4 is amplified in accordance with the electrical signal from the adjustment electrode 128d. Here, the electrical signals from the adjustment electrodes 128c and 128d are the same size. At this time, the photons P3, P4 are amplified when transmitted in different directions, respectively, and thus are amplified at different amplification rates even if the electrical signals from the adjustment electrodes 128c, 128d are the same size. As a result, the photons P3', P4' amplified at different amplification rates are each converted into an electrical signal and output from the output electrodes 138c, 138d. By amplifying a plurality of photons at different amplification rates in this way, it is possible to more reliably achieve the high dimensionality of the reservoir 10.

[0075] In order to achieve reservoir computing, a reservoir is required to have short-term memory, nonlinearity, and high dimensionality. The reservoir 10 according to the present embodiment can achieve short-term memory in the reservoir 10 because the optical signal can be left inside the waveguide section for a certain time by the reflection section 140. In addition, the reservoir 10 according to the present embodiment can achieve high dimensionality of the output information because it has a plurality of output electrodes 138a, 138b, and can generate data suitable for machine learning. Furthermore, the active layer 104 performs nonlinear conversion of the optical signal and the electrical signal. In this way, the reservoir 10 according to the present embodiment can achieve the short-term memory, the high dimensionality, and the nonlinearity necessary for reservoir computing. The reservoir 10 according to the present embodiment does not have the time limitation as in time-delay reservoir computing, and can be widely and flexibly used for reservoir computing in a variety of applications, or can achieve an increase in scalability of performance.

[0076] (Second Embodiment)

[0077] In the reservoir 20 of the second embodiment, the configuration and arrangement of the ground electrode are mainly different from those of the lower electrode 184 of the first embodiment. Figure 10 is a cross-sectional view of the reservoir 20 of the second embodiment taken perpendicular to the Y axis. In Figure 10 , cross sections of the input-output section 200 and the reflection sections 140a, 140b that constitute the reservoir 20 are shown. In Figure 10 , the same reference numerals are given to structures substantially the same as those of the reservoir 10 of the first embodiment, and the description thereof is appropriately omitted.

[0078] The input / output section 200 of this embodiment includes a substrate 180, a stacked structure 201 formed by stacking a plurality of semiconductor layers on the substrate 180, an input electrode 222, ground electrodes 230 and 232, and output electrodes 256 and 258, respectively, provided on the upper surface of the stacked structure 201. In other words, the input / output section 200 of this embodiment includes the substrate 180, the lower cladding layer 182, input sections 210a and 210b, an output section 240, and ground electrodes 230 and 232.

[0079] Input sections 210a and 210b are each formed by stacking a plurality of semiconductor layers on lower cladding layer 182 and providing electrodes thereon. Input sections 210a and 210b may have substantially the same structure, and the structure of input section 210b will be described here.

[0080] The input portion 210 b is constructed by laminating a lower waveguide layer 212 , an active layer 214 , an upper waveguide layer 216 , an upper cladding layer 218 , and a contact layer 220 in this order on the lower cladding layer 182 , and providing an input electrode 222 on the contact layer 220 .

[0081] Output section 240 is constructed by laminating a lower waveguide layer 242, an active layer 244, an upper waveguide layer 246, upper cladding layers 248 and 250, and contact layers 252 and 254 in this order on lower cladding layer 182, with output electrodes 256 and 258 provided on contact layers 252 and 254. A gap 260 is provided between upper cladding layer 248, contact layer 252, and output electrode 256, and between upper cladding layer 250, contact layer 254, and output electrode 258. This gap 260 separates the signal at output electrode 256 from the signal at output electrode 258.

[0082] A gap 202 is provided between the input portion 210b and the output portion 240. The signal of the input portion 210b is separated from the signal of the output portion 240 by the gap 202. The gap 202 reaches the upper surface of the lower cladding layer 182. The depth of the gap 202 may be approximately several μm.

[0083] The lower cladding layer 182 , lower waveguide layers 212 , 242 , active layers 214 , 244 , upper waveguide layers 216 , 246 , upper cladding layers 218 , 248 , 250 , and contact layers 220 , 252 , 254 described above constitute the stacked structure 201 .

[0084] The ground electrode 230 is provided on the upper surface of the stacked structure 201, specifically, in the gap 202, on the lower cladding layer 182. Furthermore, the ground electrode 232 is provided on the upper surface of the stacked structure 201, in the gap 204 between the input portion 210a and the output portion 240, on the lower cladding layer 182.

[0085] In the reservoir cell 10 of the first embodiment, the input electrode 118 and the output electrode 138 are disposed on the upper surface of the stacked structure 101, and the lower electrode 184, which constitutes the ground electrode, is disposed on the lower surface of the substrate 180. Therefore, in the reservoir cell 10 of the first embodiment, the entire reservoir cell 10 forms a capacitor. If the signals input and output to and from the input electrode 118 and the output electrode 138 are high-frequency, a large capacitance of the capacitor formed may limit the frequency of the signal that can be responded to, potentially resulting in signal degradation.

[0086] In contrast, in the reservoir cell 20 of the second embodiment, the input electrode 222, output electrodes 256, 258, and ground electrodes 230, 232 are all disposed on the upper surface of the stacked structure 201. Consequently, by reducing the distance between the input electrodes 222, output electrodes 256, 258, and ground electrodes 230, 232, the capacitance of the capacitor formed in the reservoir cell 20 (more specifically, the input / output section 200) can be reduced. As a result, a reservoir cell 20 capable of responding to high-speed signals and suppressing signal degradation can be realized. Furthermore, according to this embodiment, since a ground electrode need not be disposed beneath the substrate 180, the substrate 180 does not need to be doped with impurities.

[0087] In addition, Figure 10 Although the adjustment electrode is not shown in the figure, the adjustment electrode may be provided in addition to the input electrode 222 and the output electrodes 256 and 258 (or in place of a part of these electrodes).

[0088] Furthermore, the number of ground electrodes can be combined with the number of input electrodes and output electrodes, and multiple input electrodes or multiple output electrodes can share one ground electrode. For example, a ground electrode can be provided between two output electrodes, or the two output electrodes can share one ground electrode.

[0089] In this embodiment, an example is described in which gaps 202 and 204 reach the upper surface of lower cladding layer 182. However, this is not limiting. Gaps 202 and 204 may also be formed at a depth where the lower waveguide layers 242 and 212 are connected to the lower waveguide layer of input section 210a. In this case, a ground electrode may be provided on the integrated lower waveguide layer in each gap. In this case, the lower waveguide layer is doped with impurities necessary to improve conductivity.

[0090] (Third embodiment)

[0091] Figure 11 4 is a cross-sectional view perpendicular to the Y-axis of the reservoir tank 30 according to the third embodiment. Figure 11 3 shows a cross section of the input / output portion 300 and the reflecting portions 340a, 340b constituting the reservoir 30. Figure 11 In the drawings, substantially the same components as those of the reserve tank 10 of the first embodiment are denoted by the same reference numerals, and description thereof will be omitted as appropriate.

[0092] The input / output section 300 includes a substrate 180, a stacked structure 301 formed by stacking multiple semiconductor layers on the substrate 180, a lower electrode 184, and an upper electrode 314 provided on the upper surface of the stacked structure 301. On the substrate 180, the lower cladding layer 182, the lower reflective layer 362, the lower waveguide layer 302, the active layer 304, and the upper waveguide layer 306 are stacked in this order. Furthermore, a plurality of signal transmission sections 320 are provided on the upper waveguide layer 306, spaced apart from one another.

[0093] In the stacked structure 301 of this embodiment, a lower reflective layer 362 is provided below the lower waveguide layer 302. The lower reflective layer 362 is formed by stacking a plurality of pairs of two semiconductor layers with different refractive indices. For example, the lower reflective layer 362 may have a structure in which a pair of GaAs layers and AlAs layers are stacked. Alternatively, an AlGaAs layer may be used in place of the AlAs layer. Figure 11 , the lower reflection layer 362 is shown to be composed of two pairs (four semiconductor layers), but the number of pairs constituting the lower reflection layer 362 may be three or more.

[0094] The lower reflection layer 362 reflects photons propagating in the lower waveguide layer 302 provided between the active layer 304 and the lower reflection layer 362, and retains the photons inside the lower waveguide layer 302. Thus, the photons can stay in the lower waveguide layer 302 for a longer time.

[0095] The signal transmission unit 320 includes an upper reflective layer 308, an upper cladding layer 310, and a contact layer 312, which are stacked in this order on the upper waveguide layer 306. Furthermore, an upper electrode 314 is provided on the contact layer 312. The upper electrode 314 can also function as an input electrode, an output electrode, or an adjustment electrode.

[0096] The upper reflective layer 308 may be configured similarly to the lower reflective layer 362. The upper reflective layer 308 reflects photons propagating through the upper waveguide layer 306, which is provided between the active layer 304 and the upper reflective layer 308, and causes the photons to remain within the upper waveguide layer 306. This allows the photons to remain in the upper waveguide layer 306 for a longer period of time.

[0097] Similar to 140a and 140b of the first embodiment, the reflectors 340a and 340b are configured to retain photons propagating through the lower waveguide layer 302 and the upper waveguide layer 306. The reflectors 340a and 340b of this embodiment may include the same configuration as the upper waveguide layer 306.

[0098] (Replenish)

[0099] While the present invention has been described above based on the embodiments, those skilled in the art will appreciate that these embodiments are merely examples and that various modifications can be made to the combinations of their constituent elements or processing steps, and that such modifications are also within the scope of the present invention.

[0100] In the above embodiment, an example is described in which the reflector 140 uses the difference in refractive index between a semiconductor and air to reflect photons. However, this is not limiting, and the reflector can be configured to reflect photons by connecting semiconductors with different refractive indices. For example, the reflector can be arranged to contact the waveguide and be composed of a semiconductor with a lower refractive index than the waveguide. As a result, reflection of photons based on the difference in refractive index can be generated at the interface between the waveguide and the reflector. In addition, the reflector 140 can also reflect photons by using the difference in refractive index between a semiconductor and an organic material such as a polymer having a lower refractive index than the semiconductor.

[0101] In the above embodiment, the active layer 104 has been described as having a quantum well structure, but the present invention is not limited thereto, and the active layer may have various structures made of semiconductors capable of nonlinearly converting optical signals and electrical signals into each other.

[0102] Furthermore, the active layer can include various known quantum dots. Such an active layer can also nonlinearly convert optical and electrical signals. Furthermore, the active layer can also amplify or absorb optical signals depending on the magnitude of the supplied current, rendering it transparent to the optical signal.

[0103] In the above embodiment, the first mirror 150a, the second mirror 150b, and the edge 160 are provided to surround the input / output unit 100. Even without the first mirror 150a, the second mirror 150b, and the edge 160, photons are reflected inward at the end of the input / output unit 100 (e.g., interface A of the output unit 130). In this case, the reflectivity is approximately 30%, for example, and the end of the input / output unit 100 functions as a reflector.

[0104] In the above embodiment, the reflector 140 is described as comprising two mirror portions (a first mirror portion 150a and a second mirror portion 150b). The combination of gaps between a mirror portion and an adjacent component (another mirror portion or an input / output portion) is not limited to two groups as in the above embodiment, but may also be three or more groups. A greater number of mirror portion and gap combinations can improve reflectivity.

[0105] Alternatively, the end of the input / output section 100 (e.g., the end of the output section 130 on the interface A side) may be etched away, whereupon a semiconductor stack structure (or a single semiconductor layer) may be formed using a different type of semiconductor than that forming the end of the input / output section 100. Furthermore, a reflective section may be formed by forming a plurality of groups formed in the mirror section and the gap, as described above.

[0106] Furthermore, any components of the reservoir of one embodiment may be combined with the reservoir of another embodiment. For example, the same components as the reflective layer (upper reflective layer 308 or lower reflective layer 362) of the third embodiment may be incorporated into the reservoir 20 of the second embodiment.

[0107] Industrial applicability

[0108] The present invention relates to optical devices.

[0109] Description of Reference Numerals

[0110] 1 information processing system; 10, 20, 30 storage tank; 18 information processing device; 100, 200, 300 input and output units; 101, 201 stacked structure; 102, 212, 242, 302 lower waveguide layer; 104, 214, 244, 304 active layer; 105, 216, 246, 306 upper waveguide layer; 112, 124, 134, 218, 248, 250, 310 upper cladding layer; 116, 1 26, 136, 220, 252, 254, 312 contact layers; 118, 222 input electrodes; 128 adjustment electrode; 138, 256, 258 output electrodes; 140 reflection portion; 150a first mirror portion; 150b second mirror portion; 180 substrate; 182 lower cladding layer; 184 lower electrode; 230, 232 ground electrodes; 308 upper waveguide layer; 314 upper electrode; 320 signal transmission portion; 362 lower reflection layer.

Claims

1. An optical device comprising: Input electrodes, used for inputting electrical signals; a conversion unit that converts the electrical signal and the optical signal into each other nonlinearly; a waveguide portion for transmitting therein an optical signal generated by the conversion portion converting the electrical signal input to the input electrode; a plurality of output electrodes for respectively outputting electrical signals generated by the conversion portion converting the optical signal transmitted in the waveguide portion; as well as The reflecting portion is configured to reflect the optical signal transmitted through the waveguide portion toward the inside of the waveguide portion.

2. The optical device according to claim 1, The reflecting portion includes a plurality of mirror portions that are positioned apart from the waveguide portion in the transmission direction of the optical signal and are spaced apart from each other in the transmission direction.

3. The optical device according to claim 1, The conversion portion includes a quantum well or a quantum dot that nonlinearly converts an optical signal and an electrical signal into each other.

4. The optical device according to claim 3, The invention further includes a current for amplifying the optical signal transmitted in the waveguide portion in the conversion portion, or an electrode for supplying a current to the conversion portion to make the conversion portion transparent.

5. The optical device according to any one of claims 1 to 4, The optical device has a stacked structure formed by stacking a plurality of semiconductor layers including an active layer constituting the conversion portion and a waveguide layer constituting the waveguide portion. The input electrode and the output electrode are respectively provided on the upper surface of the stacked structure.

6. The optical device according to claim 5, The invention also includes a ground electrode, which is arranged on the upper surface of the stacked structure.

7. The optical device according to claim 5, The stacked structure further includes a reflective layer formed by overlapping a plurality of pairs of two semiconductor layers having different refractive indices. The waveguide layer is disposed between the active layer and the reflective layer.

8. The optical device according to claim 7, The waveguide layer is an upper waveguide layer provided on the active layer, The reflective layer is an upper reflective layer provided on the upper waveguide layer, The stacked structure further includes a lower waveguide layer provided below the active layer, and a lower reflective layer formed by overlapping a plurality of pairs of two semiconductor layers having different refractive indices.

9. The optical device according to claim 5, The stacked structure is formed by stacking the plurality of semiconductor layers on a substrate. The substrate is oriented in a plane orientation other than (100) and (111).

Images