JPH06130441A - Optical information processing method - Google Patents

Abstract

(57) [Summary] [Structure] A photoelectric conversion layer and a memory layer having a function of sustaining conductivity changed by light having a certain wavelength even after light is blocked, between at least one electrode having light transparency. The optical information processing element has optical information by light of a wavelength that changes the conductivity of the memory layer, stores the optical information, and then enters optical information by light of a wavelength for photoelectric conversion in the photoelectric conversion layer. And an optical information processing method for recognizing optical information from the obtained output signal. [Effect] Optical information processing that has a function of switching the light-receiving sensitivity in an analog manner by light irradiation, that state is stably stored and held for a long time, and reversible memory can be erased by heating. The present invention provides an optical information processing method using an element, and can be applied to various applications such as pattern recognition, arithmetic processing, visual information processing, neurocomputers, and sensors.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention can change the response sensitivity to an optical signal by control light, and can store the changed response sensitivity in the element for a long time, such as pattern recognition. The present invention relates to an optical information processing method using a novel optical information processing element suitable for visual information processing and neurocomputers.

[0002]

2. Description of the Related Art In recent years, it has been considered that information processing by a neural network enables information processing that was originally thought to be peculiar to the brain and nervous system, such as associative memory and learning, which were conventionally difficult. It's coming. In addition, many attempts have been made to realize the neural network as hardware.

In addition, the parallelism of light, that is, the ability to simultaneously transmit and process a large amount of information, the advantages in wiring, crosstalk, etc., the analog nature, and the ability to directly process images, Expectations for neural networks and optical neurocomputers are increasing. At that time, in the neuro element, which is a constituent unit of the neural network, according to the input information,
A part corresponding to the synapse of the nerve cells of the living body, which weights each input information, is required. Further, it is more desirable that the degree of weighting can be changed in an analog manner and that the weight is stored and held for a long time after the change.

As a model of a neuro element having such a synapse-corresponding part, there is a McCullough-Pitz model shown in FIG. In this model, the i-th input signal u _i is integrated with the weighting factor w _i at the synapse equivalent part, all the weighted signals are added, and the nonlinear function g at the output part is added.
The output v is calculated according to (x) (delta function, sigmoid function). v = g (Σw _i · u _i ) <1> At this time, the weighting factor w _i is variable in an analog manner, and the degree of the weight with respect to the input signal is changed and held. It is considered to be essential for achieving the processing.

In the development of the neuro element according to such a model, how to realize the synapse equivalent part of the living body is a key, and an example thereof is disclosed in Japanese Patent Laid-Open No. 4-5636. There is a method of controlling the light transmittance by a spatial light modulator. According to this method, the weighting factor wi in the above equation 1 is determined depending on the light transmittance.
Can be weighted. However, it is necessary to control from the outside when controlling the light transmittance.
The controlled transmittance value cannot be stored unless an external arithmetic device is used, and the contrast before and after learning that can be realized by the spatial light modulator is limited.

As disclosed in Japanese Patent Laid-Open No. 4-90015, a weight is obtained by applying a control voltage from the outside to a light receiving element formed of a photodiode to modulate the light receiving sensitivity, that is, the degree of electrical response to an optical signal. Attempts have been made to realize this, but since this element also does not have a memory function, it is essential to use an external arithmetic unit together.

As a device having a storage function added to this device, there is a device having a built-in storage function, which is shown in pages 25 to 27 of Electronic Technology Magazine, January 1992, pages 25 to 27. In this device, it is considered that the detection sensitivity to the signal light is increased and maintained due to the influence of the space charge formed by irradiating the junction formed by the metal-gallium arsenide junction with the control light. Weighting is performed using the change in sensitivity. However, the increase in detection sensitivity is at most about several times, and the increased detection sensitivity disappears in about 20 minutes, so it cannot be said that it is sufficient as a memory function.

As described above, in the conventional optical information processing method, the weighting coefficient for the input information is controlled by an operation from an external device, and the weighting coefficient is stored in an external arithmetic device. In most cases, it was due to an element that did not have a memory function. Further, even when an element having a memory function is used, there are drawbacks such as a very short retention time of memory and a small difference between memory and non-memory.

In view of the above situation, the present inventors have a function of analogically switching the light receiving sensitivity by light irradiation,
We have completed an optical information processing device characterized in that this state is stored and held stably for a long time, and reversible storage can be erased by heating.

[0010]

However, in using the above optical information processing element, an optical information processing method for effectively performing information processing such as learning, storing, and recognizing optical information is required. . The present invention has been made to solve the above problems, and an optical information processing element is used to effectively perform information processing such as learning, memory, and recognition on optical information. The purpose is to obtain an information processing method.

[0011]

SUMMARY OF THE INVENTION The gist of the present invention is that at least one of them has a function of maintaining a photoelectric conversion layer between electrodes having a light-transmitting property and the conductivity changed by light having a certain wavelength even after the light is blocked. An optical information processing element having a memory layer is made to enter optical information of light having a wavelength that changes the conductivity of the memory layer to store the optical information, and then the photoelectric conversion layer is made to perform photoelectric conversion by light of a wavelength. Inject light information,
An optical information processing method is characterized in that the optical information is recognized from the obtained output signal.

The present invention will be described in detail below. First, the optical information processing element that is the subject of the present invention will be described. The above-described optical information processing element has a photoelectric conversion layer and a memory layer having a function of sustaining conductivity changed by light having a certain wavelength even after light is blocked, between at least one electrode having a light transmitting property.

The electrodes will be described. The electrode is formed as a conductive thin film layer on the support. As the support, a quartz or glass plate, a metal plate or a metal foil, a plastic film or a sheet, etc. are used, but a glass plate or a transparent plastic (polyester, polymethacrylate, polycarbonate) plate or the like is preferable. A metal plate or the like that can be used as an electrode can also serve as a support.

The electrode forming material is usually aluminum,
Conductivity of metals such as gold, silver, platinum, cadmium, nickel, indium, palladium, tellurium, oxides of metals such as indium and / or tin, copper iodide, carbon black, poly (3-methylthiophene), polypyrrole, etc. Resin is used. When the electrode forming material is a metal or a metal oxide, the conductive thin film layer is usually formed by a sputtering method, a vacuum deposition method or the like, but it may be formed by another method depending on the kind of the electrode forming material. Good. For example, in the case of fine particles of metal such as silver, copper iodide, carbon black, fine particles of conductive metal oxide, conductive resin powder, etc., after the electrode forming material is dispersed in an appropriate binder resin solution, the support It can be formed by a method of coating on. Further, in the case of a conductive resin, it can be directly formed on the support by electrolytic polymerization. The conductive thin film layers can be laminated with different materials.

The thickness of the conductive thin film layer is not particularly limited, but it is preferably at least 50 Å or more for uniform expression of conductivity. On the other hand, if light transmission is required,
It is necessary not to exceed the film thickness that satisfies the transmittance. Even when the coating method is used to increase the film thickness, the film thickness is usually 100 μm or less. In the present invention, at least one of the electrodes needs to be a light transmissive electrode.
The light transmittance does not necessarily have to cover the entire wavelength range, but at least the light transmittance in the wavelength range of the light absorbed by the memory layer and the photoelectric conversion layer is required. The higher the light transmittance, the more preferable in terms of the efficiency of irradiation light. The transmittance is at least 10% or more, practically 30% or more, and preferably 60% or more.
% Or more is required.

Next, the photoelectric conversion layer and the memory layer formed between the electrodes will be described. The photoelectric conversion layer is made of a photoconductive material that absorbs light and generates charge carriers by separating positive and negative charges, and if necessary, a binder resin,
It is formed with an insulating material such as a long chain alkyl fatty acid, a carrier transport material such as an aromatic amine and a hydrazone compound, and other additives.

As the photoconductive material, a photoconductive material used in electrophotography, solar cells and the like is used. Examples of inorganic photoconductive materials include amorphous selenium, selenium alloys such as selenium-tellurium and selenium-arsenic, amorphous silicon, cadmium sulfide, and zinc oxide. Examples of organic photoconductive materials include various dyes and pigments. Examples of such compounds include compounds known as photoelectric conversion dyes, charge carrier generating materials and the like in known documents.

As an example of a document, “Basics and Applications of Electrophotographic Technology”, pages 437 to 448 (edited by The Electrophotographic Society, Corona Publishing Co., Ltd.,
1988), Coloring Materials Association, Vol. 47, pp. 594-604.
Page (Katsuji Maruyama, 1974), "LB film and electronics" pages 193-204 (CMC, 1986).
), "Organic electronic materials", pp. 94-101 (edited by Japan Society of Applied Physics, Ohmsha, 1990) and the like.

Examples of such compounds include porphyrin type, cyanine type, merocyanine type, pyrylium type,
Examples thereof include thiapyrylium-based dyes, triarylmethane-based dyes, squarylium-based dyes, azurenium-based dyes, perylene-based dyes, polycyclic quinone-based dyes, pyrrolopyrrole-based and other condensed ring dyes, phthalocyanine-based dyes, and azo dyes. As the phthalocyanine-based dye, various crystal-type dyes of metal-coordinated phthalocyanines such as copper, vanadium, indium, titanium, aluminum, tin, and magnesium may be used together with the metal-free phthalocyanine-based dye. As azo dyes,
Disazo, bisazo and trisazo dyes are mainly used in the form of pigment particles.

The memory layer is a low-molecular compound having a hole-transporting property having one or more nitrogen atoms in the molecule, or a compound having a memory property-imparting function for sustaining the conductivity changed by the irradiation of light even after the light is blocked. And the like are dispersed in a binder polymer. First, the low molecular weight compound having a hole transporting property will be described. The hole transporting low molecular weight compound acts as a hole transporting carrier which is a charge carrier of the memory layer. The hole transport phenomenon can be regarded as an electron transfer between molecules or a redox reaction, and an electron donating compound having a small ionization potential is suitable for effective hole transport.

In the present invention, a compound having one or more nitrogen atoms in the molecule is preferably used as the low-molecular compound having a hole-transporting property from the above point. In particular, a compound in which a nitrogen atom is bonded to a π-electron conjugated system and which has a good orientation alignment between molecules is suitable. The form of the nitrogen atom is, for example, a dialkylamino group such as a diethylamino group, an amino group directly bonded to an aromatic hydrocarbon or an aromatic heterocycle by a diarylamine group such as a diphenylamino group, and the like. Examples thereof include a hydrazo group and a hydrazono group bonded to a hydrocarbon or an aromatic heterocycle, and other examples include a nitrogen atom constituting the heterocycle. And examples of the heterocycle include carbazole, indole, pyrazole,
Examples thereof include pyrazoline and oxazole.

The above-described hole transporting low molecular weight compound is easier to manufacture than the high molecular weight compound, and the impurities can be easily removed by purification. Little deterioration in characteristics. Further, since the low molecular weight compound is generally excellent in compatibility with the binder polymer, it is easy to increase the hole mobility by increasing the content in the memory layer.

In the present invention, a hydrazone compound, particularly a hydrazone compound represented by the following chemical formula [Chemical formula 1], is preferably used as the low-molecular compound having a hole transporting property.

[0024]

[Chemical 1]

In the above chemical formula [Chemical formula 1], A represents a monovalent or divalent organic group containing at least one aromatic hydrocarbon ring or aromatic heterocycle, and these rings have a substituent. You may have. Specific examples include the organic groups described in (a) to (d) below. (A) benzene, naphthalene, anthracene, pyrene,
A monovalent or divalent organic group derived from perylene, phenanthrene, fluoranthene, acenaphthene, acenaphthylene, azulene, fluorene, indene, tetralin, naphthacene and the like. The organic group is an example containing at least one aromatic hydrocarbon ring.

(B) A monovalent or divalent organic group derived from pyrrole, thiophene, furan, indole, carbazole, pyrazole, pyridine, acridine, phenazine, benzothiophene, benzofuran and the like. The organic group is an example containing at least one aromatic heterocycle. (C) A monovalent or divalent organic group derived from a compound in which each of the above organic groups is directly bonded.

Examples of the above compounds include biphenyl, terphenyl, phenylanthracene, bithiophene, terthiophene, bifuran, thienylbenzene, thienylnaphthalene, pyrrolylthiophene, N-phenylcarbazole and the like. (D) A monovalent or divalent organic group derived from a compound in which each of the above organic groups is bonded via a bonding group.

The above-mentioned bonding group is an alkylene group which may have a substituent represented by the following chemical formula [Chemical formula 2] or a divalent group represented by the following chemical formula [Chemical formula 3]. The organic groups of Further, a bonding group in which such an alkylene group and a divalent organic group are combined can be mentioned.

[0029]

[Chemical 2]

[0030]

[Chemical 3]

Specific examples of the compound corresponding to (d) include, for example, xanthene, thioxanthene, indoline, phenothiazine represented by the following chemical formula [Chemical formula 4].

[0032]

[Chemical 4]

In addition to the above, specific examples of the compound corresponding to (d) include diphenylmethane, stilbene,
Examples thereof include tolan, 1,4-diphenylphthaldiene, diphenyl ether, diphenyl sulfide, N-methyldiphenylamine, triphenylamine and azobenzene. Furthermore, instead of the benzene ring of these compounds, compounds in which other aromatic rings or heterocycles are combined by using a bonding group can be cited.

Examples of the substituent which the aromatic hydrocarbon ring and / or the aromatic heterocycle in the above (a) to (d) may have include, for example, a methyl group, an ethyl group, a propyl group and a butyl group. A lower alkyl group such as a hexyl group, a lower alkoxy group such as a methoxy group, an ethoxy group or a butoxy group,
Aralkyl groups such as allyl group, benzyl group, naphthylmethyl group and phenethyl group, aryloxy groups such as phenoxy group and trioxy group, arylalkoxy groups such as benzyloxy group and phenethyloxy group, aryl groups such as phenyl group and naphthyl group , Aryl vinyl groups such as styryl group and naphthyl vinyl group, and dialkylamino groups such as dimethylamino group and diethylamino group. The alkyl component in these substituents may contain an ether group, an ester group, a cyano group, a sulfide group or the like.

In the above chemical formula [Chemical formula 1], R ¹ , R ² and R
³ , R ⁴ and R ⁵ are each a hydrogen atom or an alkyl group which may have a substituent, an aralkyl group, an aromatic hydrocarbon group,
Represents a heterocyclic group. Specific examples of R ^{1 to} R ⁵ include lower alkyl groups such as methyl group, ethyl group, propyl group, butyl group and hexyl group, aralkyl groups such as benzyl group and phenethyl group, phenyl group, naphthyl group, acenaphthyl group, An aromatic hydrocarbon group similar to A such as an anthryl group and a pyrenyl group, a thienyl group, a bithienyl group, a carbazole group, an indolyl group, a furyl group, an indoline group and the like heterocyclic group similar to that in A can be mentioned. .

The substituents that each organic group of R ^{1 to} R ⁵ may have include a methyl group, an ethyl group,
Lower alkyl group such as propyl group, butyl group and hexyl group, lower alkoxy group such as methoxy group, ethoxy group and butoxy group, aryloxy group such as phenoxy group and trioxy group, arylalkoxy such as benzyloxy group and phenethyloxy group Groups, aryl groups such as phenyl group and naphthyl group, and substituted amino groups such as dimethylamino group, diethylamino group, phenylmethylamino group and diphenylamino group.

However, R ¹ may form a ring together with A. As such an example, an organic group represented by the following chemical formula [Formula 5] can be given.

[0038]

[Chemical 5]

In the above chemical formula [Chemical formula 1], R ⁶ and R
⁷ represents an optionally substituted alkyl group, aralkyl group, allyl group, aromatic hydrocarbon group or heterocyclic group. Specifically, a lower alkyl group such as a methyl group, an ethyl group, a propyl group and a butyne group, an aralkyl group such as a benzyl group, a phenethyl group and a naphthylmethyl group, an aromatic hydrocarbon such as an allyl group, a phenyl group and a naphthyl group. Represents a heterocyclic group such as a group, a pyridyl group, a thienyl group, a furyl group and a pyrrolyl group. Examples of the substituent which these may have include the same substituents as those in R ¹ , R ² , R ³ , R ⁴ and R ⁵ .

However, R ⁶ and R ⁷ may be combined to form a ring, and examples thereof include an organic group represented by the following chemical formula [Chemical Formula 6].

[0041]

[Chemical 6]

In the above chemical formula [Chemical formula 1], l is 0 or 1,
m represents an integer of 0, 1 or 2, and n represents an integer of 1 or 2. Note that n
Is 1 when A is a monovalent group and 2 when A is a divalent group. Among the hydrazone compounds represented by the chemical formula [Formula 1], a hydrazone compound in which A is a carbazole ring is particularly preferable. The following chemical formula [Chemical formula 7] illustrates some of the hydrazone compounds.

[0043]

[Chemical 7]

Next, the compound having a memory property-imparting function will be described. Memory properties can be achieved by many compounds. Representative compounds include, for example, protic acids such as chloroacetic acid and orthobenzoylbenzoic acid, aromatic diazonium salts, triarylmethanes such as leuco crystal violet and leucomalachite green, halogenated methylene iodide, hexachloroethane and the like. Hydrocarbon, 1,3,5-tribromobenzene,
Aromatic halogen compounds such as 9,10-dichloroanthracene and 9,10-dibromoanthracene, halogenated ketone compounds such as benzamide, nitrophenol, nitroaniline, hexachloroacetone and bromoacetophenone, acetyl chloride, acetyl bromide and chlorobenzoyl chloride. And the like, acid anhydrides such as phthalic anhydride, and thioketones such as thiomichelers ketone.

Particularly, an aromatic halogen compound in which two or more chlorine atoms and / or bromine atoms are substituted or a thioketone represented by the following chemical formula [Formula 8] is preferable.

[0046]

[Chemical 8]

In the above chemical formula [Chemical formula 8], Ar ¹ and Ar
² represents an aromatic hydrocarbon group or an aromatic heterocyclic group which may have a substituent, and specifically, the above-mentioned chemical formula [Chemical formula 1]
The same aromatic hydrocarbon group or aromatic heterocyclic group as in A in the above can be mentioned. The following chemical formula [Chemical Formula 9] illustrates preferable thioketones.

[0048]

[Chemical 9]

Next, the binder polymer will be described. The binder polymer is preferably a polymer which has good compatibility with the above-mentioned compounds and does not adversely affect the movement of charge carriers in the layer. For example, polymers and copolymers of vinyl compounds such as styrene, vinyl acetate, vinyl chloride, acrylic acid ester, methacrylic acid ester and butadiene, polyvinyl acetal,
Examples thereof include polycarbonate, polyester, polysulfone, polyphenylene oxide, polyurethane, cellulose ester, cellulose ether, alkyd resin, phenoxy resin, silicon resin and epoxy resin. Among these, polyester resin, polycarbonate resin, methacrylic resin, acrylic resin and phenoxy resin are preferable, and polycarbonate resin and methacrylic resin are particularly preferable.

The amount of the binder polymer used is usually in the range of 0.1 to 30 times by weight, preferably 0.3 to 10 times by weight, of the hole transporting low molecular weight compound. The memory layer in the optical information processing element of the present invention is constituted by dispersing the above-described hole transporting low molecular weight compound, a compound having a memory property imparting function in a binder, and further, if necessary, a plasticizer, Additives such as a surfactant, an ultraviolet absorber, an antioxidant, and an electron-withdrawing compound can be included.

Next, a method of manufacturing the optical information processing element of the present invention will be described. In the optical information processing device of the present invention, first, each component of the memory layer and various additive components used as necessary are dissolved in a solvent to prepare a coating liquid, and the coating liquid is coated on an electrode. After that, it is dried to form a memory layer. The film thickness of the memory layer is determined by the electric field strength required for the operation of the optical information processing element and the range of the power supply voltage, but is usually 100 μm or less, preferably 30 μm or less. Further, the lower limit of the film thickness is 0.01 μm, preferably 0. 0, from the viewpoint of ensuring the uniformity of the coating film and preventing pinholes.
The thickness is preferably 1 μm or more.

As the method for forming the photoelectric conversion layer on the memory layer, in addition to the coating method, the vacuum deposition method, the CVD method, the Langmuir-Blodgett (LB) method (LB film and electronics, pages 1 to 15) , Pp. 33-46, CMC, 1986), etc., and the like, and a method for laminating monomolecular layers. The shape of the photoconductive material also varies depending on the above manufacturing method. In addition to amorphous thin layers such as selenium and silicon, there are thin layers made of fine particles, particles dispersed in a binder resin, dissolved particles, and the like. There is also a shape of an LB film in which layers are laminated. In the case of the LB film, in addition to the photoconductive dye alone, a mixed film with an insulating long-chain fatty acid such as arachidic acid or a different dye, or a monomolecular film having a different mixing ratio can be laminated.

In the photoelectric conversion layer in which the photoconductive material is dispersed as fine particles in the binder resin, the binder resin may be butyral resin, phenoxy resin, phenol resin, etc., which is a binder resin having a hydroxyl group with good dispersibility of the fine particles. Besides, polyester, polycarbonate,
Methacrylic resin or the like is used. When the photoelectric conversion layer is formed by vapor deposition or the like, an operation of changing the crystal form by exposing it to a solvent vapor to further improve the conversion efficiency is performed, if necessary.

Further, additives for improving coatability, dispersion stability and storage stability can be added to the photoelectric conversion layer. Further, a photoconductive material having a carrier transporting ability can be added. Although the thickness of the photoelectric conversion layer varies depending on the manufacturing method, it is several tens of to several μm, and generally 1 μm.
A thickness of m or less is desirable. The order of stacking the memory layer and the photoelectric conversion layer is not necessarily the same, and the memory layer may be formed on the photoelectric conversion layer by the above method. Further, a manufacturing method in which the memory layer and the photoelectric conversion layer are formed on separate electrodes and then integrated by pressure bonding or the like can be employed. The electrodes may be laminated by a pressure bonding method as well as a sputtering method or a vapor deposition method.

Next, an optical information processing method using the optical information processing element will be described. The optical information processing element is used by applying a voltage between the electrodes. At that time, it is insulating in a dark state where light is not irradiated, and the dark current has a very small value, but during irradiation with light in the absorption wavelength region of the photoelectric conversion layer (input light),
The conductivity increases and a bright current is observed. When the light in the absorption wavelength region of the memory layer (control light) is irradiated for a certain period of time and then observed, the dark current value hardly changes, but when the input light is irradiated, it is observed before the control light is irradiated for a certain period of time. It increases or decreases compared to the bright current value. That is, the response sensitivity of the element to the input light is changed by the irradiation of the control light, and the changed light receiving sensitivity is stably maintained in the state in which the voltage is applied, so this element stores information in the form of a change in light receiving sensitivity. Will be. The light receiving sensitivity of the element can be controlled in an analog manner by the light irradiation amount, the number of times of irradiation, the applied voltage and the like.

Although the mechanism of the above-mentioned phenomenon is not sufficiently clear at the present time, it is clear that it is not a chemical decomposition of an organic matter component or the like because it is a reversible phenomenon. However, for the time being, we can hypothesize that: That is, a low-molecular compound having a hole-transporting property in the memory layer, a compound having a function of imparting a memory property, a binder polymer, or the like, undergoes a state change such as prototropy, isomerization, or orientation change by controlled light irradiation alone or between molecules. And lowering the energy barrier of hole injection from the photoelectric conversion layer by changing the polarization in the vicinity of the photoelectric conversion layer interface, facilitating the inflow of holes that can occur when the photoelectric conversion layer is irradiated with input light. Change occurs, and the bright current at the time of illuminating the input light increases. Then, the memory function is expressed by stably maintaining the changed state.

The optical information processing element stably holds the light receiving sensitivity changed by the irradiation of the control light, but it can be restored to the original state by heating, and can be quickly heated by heating the glass transition temperature of the memory layer or higher. Since it reversibly returns to the original state, it is possible to repeatedly store and erase. The optical information processing element is used by applying a voltage between the electrodes. The degree of change in the light receiving sensitivity can be controlled by the magnitude of the electric field strength applied between the electrodes during irradiation of the control light. At this time, the greater the electric field strength, the greater the rate of change in the light receiving sensitivity of the element. The electric field strength is required not to cause dielectric breakdown, and is generally 10 ⁷ V / cm or less, usually 10 ⁶ V / cm or less, preferably 5 × 10 ⁵ V / cm or less.

Next, the light irradiation method will be described. It is necessary that the input light is absorbed by the photoelectric conversion layer and does not affect the memory layer as much as possible, and the control light is absorbed near the interface between the input layer and the memory layer. For that purpose, the input light and the control light may have the same wavelength or different wavelengths, and the light intensities may be the same or different. The light may be monochromatic light or light having a certain wavelength width. The incident direction of light may be from the memory layer side or the photoelectric conversion layer side. For example, the following cases are listed.

(1) When the wavelength regions of the input light and the control light are different. In this case, the input light is light in the absorption wavelength region of the photoelectric conversion layer, and the control light is light in the absorption wavelength region of the memory layer.
When both the input light and the control light are incident from the memory layer side, the memory layer should not absorb the input light at all, or even if it absorbs, part of the input light should reach the photoelectric conversion layer. Is. Further, the control light is made to reach the vicinity of the bonding interface between the memory layer and the photoelectric conversion layer.

When both the incident light and the control light are emitted from the photoelectric conversion layer side, the photoelectric conversion layer does not absorb the control light at all, or even if it absorbs, a part of the control light reaches the memory layer. It is necessary to be. The incident direction is the input light,
When the control light is opposite, the input light is made to reach the photoelectric conversion layer and the control light is made to reach the junction interface between the memory layer and the photoelectric conversion layer, as in the above case.

In addition to the above conditions, it is desirable that the input light does not affect the light receiving sensitivity of the optical information processing element. In that case, the light receiving sensitivity which is changed and held by the control light is not affected and the element concerned is not affected. The light receiving sensitivity of can be monitored by illuminating the input light. That is, the control light can be used for writing information and the input light can be used for reading information.

(2) When the wavelength regions of the input light and the control light are the same. The wavelength in this case corresponds to the absorption wavelength region of the memory layer, but at the same time, it needs to be the absorption wavelength region of the photoelectric conversion layer. When both the incident light and the control light are emitted from the memory layer side, it is necessary that a part of the input light reaches the photoelectric conversion layer, and the control light is the bonding interface between the memory layer and the photoelectric conversion layer. It is necessary to reach the neighborhood. In addition, the light intensity of the input light should be smaller than that of the control light, or the input light irradiation time should be short enough not to affect the light receiving sensitivity of the device. If the influence exerted is made as small as possible, it is possible to monitor by the input light with little influence on the light receiving sensitivity of the element which is changed and held by the control light.

When both the incident light and the control light are incident from the photoelectric conversion layer side, it is necessary that the control light reaches near the junction interface between the memory layer and the photoelectric conversion layer. When the incident direction is opposite to the input light and the control light, the control light is irradiated from the memory layer side and the input light is irradiated from the photoelectric conversion layer side, and the input light is emitted near the junction interface between the memory layer and the photoelectric conversion layer. If the light is not affected, it can be monitored by the input light with little influence on the light receiving sensitivity of the element which is changed and held by the control light.

In addition to the above cases (1) and (2), it is possible to use the input light and the control light without distinguishing them, and the light reaching the vicinity of the junction interface between the memory layer and the photoelectric conversion layer can be used. If used, the bright current value changes with each light irradiation. Next, application of the optical information processing element to pattern recognition will be described. In the optical information processing element, even within one electrode, the response sensitivity is selectively increased only in the portion exposed to the learning light. Therefore, it is possible to learn a certain pattern by irradiating the element with a control light through an optical mask having a two-dimensional pattern, and then to memorize the stored pattern by irradiating the two-dimensional pattern with the input light. is there. That is, when a pattern that is more similar to the learned and stored pattern, that is, a pattern with a smaller Hamming distance is input, a larger output is output. In addition, since the response sensitivity of the optical information processing element can be continuously controlled by the irradiation light amount or the applied voltage, for example, even when an optical mask having gradation of gradation is used, It is possible to memorize according to the degree and to give the responsiveness an in-plane distribution. Further, even when different patterns are stored in the same element, the information can be processed without crosstalk by being linearly integrated.

[0065]

【Example】

Example 1 <Preparation of optical information processing element> 9-ethylcarbazole-3
-Carbaldehyde diphenylhydrazone 1.0 g, 4,
4'-bis (dimethylamino) thiobenzophenone 1
A coating solution was prepared by dissolving 2.5 mg of polycarbonate and 1.25 g of polycarbonate in 14 g of dioxane.

On the ITO electrode layer formed on the glass substrate
The above coating solution is applied to the film so that the film thickness after drying is 3 μm.
It was clothed and dried to form a memory layer. Then titani
Ruphthalocyanine 2 parts by weight, polyvinyl butyral 1 layer
Weight ratio of n-propyl alcohol and methanol
The solid content ratio becomes 3.4% in the solvent mixed at 60:40.
And mixed to prepare a dispersion liquid. This dispersion is
The film thickness after drying on the memory layer is about 0.1 μm
Thus coated and dried to form a photoelectric conversion layer. Photoelectric conversion
Aluminum on the surface of the replacement layer has an electrode area of 1 cm²When
By vacuum evaporation to form the counter electrode,
A device was produced. <Evaluation of functions of optical information processing element> The above optical information processing element
And a DC voltage of 30V is applied with the aluminum electrode side as the positive electrode.
After applying, output 100μ from the transparent electrode (ITO electrode) side
W / cm²At a wavelength of 700 nm for 1 minute
However, as shown in FIG. 1, 1 × 10^-8A / cm ²The light of
An electric current was observed. After shutting off the light, again 700 nm light
When irradiated, almost the same bright current was observed with good reproducibility.
Then, 445 nm monochromatic light is irradiated from the transparent electrode side for 5 minutes.
Bright current gradually increased, and after 5 minutes, 3 × 10^{ー 7}A
/ Cm²Reached After shutting off the light, the dark current quickly disappeared.
It returned to the current value of. After 20 minutes, again 700nm monochromatic
When light is irradiated from the transparent electrode side, 5 × 10^{ー 7}A / c
m²The bright current was observed with good reproducibility. That is, 445n
The bright current response to 700 nm light is
It was confirmed that the response sensitivity increased about 50 times. This increase
The response sensitivity was stable over time at room temperature, and after 3 hours,
6%, 91% after 5 hours, 91% after 7 hours
It was Example 2 <Production of Optical Information Processing Device>
An information processing element was prepared. <Evaluation of functions of optical information processing element> The above optical information processing element
And a DC voltage of 30V is applied with the aluminum electrode side as the positive electrode.
After applying, output 100μ from the transparent electrode (ITO electrode) side
W / cm²With monochromatic light of wavelength 445 nm for 20 seconds
3 x 10 on the spot^-8A / cm²The bright current of was observed.
After the light was cut off, the original dark current value was quickly restored. Wavelength 44
Repeated irradiation with 5 nm light showed that
The bright current value gradually increases, and after irradiation of 10 times, 1 × 10
^{ー 7}A / cm²Increased to. That is, for light of 445 nm
Response sensitivity increases continuously with each irradiation of light.
And confirmed. Example 3 <Preparation of optical information processing element> 9-ethylcarbazole-3
-Carbaldehyde diphenylhydrazone 1.0 g, 4,
4'-bis (dimethylamino) thiobenzophenone 1
2.5 mg, polycarbonate 1.25 g dioxane
It was dissolved in 14 g to prepare a coating solution.

The above coating liquid was applied onto the ITO electrode layer formed on the glass substrate so that the film thickness after drying was 3 μm, and dried to form a memory layer. Then p-
1.0 mmol / (10-carboxydecyloxy) phenyltritolylporphyrin and arachidic acid
l, 5.0 mmol / l containing mixed chloroform solution was prepared, and this was mixed with Langmuir-Blodgett (L
According to the method B), after dropping the solution onto an aqueous solution containing a divalent cadmium salt to form a monomolecular film, the surface pressure is 25 mN /
The film was compressed to m, and 36 layers were accumulated on the memory layer to form a photoelectric conversion layer. Aluminum was vacuum-deposited on the surface of the photoelectric conversion layer so that the electrode area of the device was 1 cm ² to form a counter electrode, and an optical information processing device was produced. <Evaluation of Function of Optical Information Processing Element> After applying a DC voltage of 30 V to the above optical information processing element with the aluminum electrode side as a positive electrode, an output of 100 μm from the transparent electrode (ITO electrode) side.
When a monochromatic light having a wavelength of 440 nm was irradiated for 10 seconds at W / cm ² , a bright current of 5 × 10 ⁻⁹ A / cm ² was observed.
After the light was cut off, the original dark current value was quickly restored. Wavelength 44
When the light of 0 nm is repeatedly irradiated, the bright current value gradually increases as shown in FIG.
It increased to ^-8 A / cm ² . That is, it was confirmed that the response sensitivity to the light of 440 nm continuously increases each time the light irradiation is repeated. Example 4 An optical information processing element was produced in the same manner as in Example 1.

All of the patterns 1 to 9 shown in FIG. 4A consist of 8 pixels, which is half of the 4 × 4 matrix. Of these, 1 and 4 were used for learning.
The graph of FIG. 4B shows the Hamming distances of the patterns 1 to 9 from the patterns 1 and 4. This graph shows that the pattern 1 is closer to the left of the coordinate axis, and the pattern 4 is closer to the upper side. As shown in (a) of FIG. 5, before learning, no difference in the responsiveness at the time of irradiation of the input light with respect to patterns 1 to 9 is observed, but after learning pattern 1, As shown in the figure, the distribution of the output varies depending on the Hamming distance from the pattern 1. Further, after learning the pattern 4 subsequently to this, as shown in FIG. 5C, although the history of the pattern 1 remains, the output distribution substantially corresponds to the Hamming distance from the pattern 4. Taking the difference between the graph of FIG. 5B and the graph of FIG. 5C, the output distribution according to the Hamming distance from the pattern 4 is obtained as shown in FIG. It was found that the overwriting was performed linearly without being affected by the stored information of pattern 1. As described above, it was confirmed that this element can perform product-sum operation of input information and stored information, which is necessary for a neuro element, and can also perform simple pattern recognition.

[0069]

According to the present invention, the light-receiving sensitivity has a function of analog-switching by light irradiation, this state is stably stored and held for a long time, and reversible storage can be erased by heating. It provides an optical information processing method using an optical information processing element having characteristics, and can be expected to be applied to various applications such as pattern recognition, arithmetic processing, visual information processing, neurocomputers, and sensors.

[Brief description of drawings]

FIG. 1 is a diagram showing current response characteristics of an optical information processing element before and after irradiation of control light with irradiation of input light.

FIG. 2 is a diagram showing current response characteristics of the optical information processing element of Example 2 due to repeated light irradiation.

FIG. 3 is a diagram showing the current response characteristics of the optical information processing element of Example 3 upon repeated light irradiation.

4A is a diagram of a two-dimensional pattern used in Example 4. FIG. (B) The figure which shows the Hamming distance from the pattern 1 and the pattern 4 of the two-dimensional pattern used in Example 4. In the figure, ~ represents patterns 1 to 9 in Fig. 4A.

5A is a diagram showing output current responses to patterns 1 to 9 before learning. FIG. (B) The figure which showed the output current response with respect to patterns 1-9 after the pattern 1 learning. (C) The figure which showed the output current response with respect to patterns 1-9 after the pattern 4 learning. (D) The figure which took the difference of the output current of (b) of FIG. 5 and (c) of FIG.

FIG. 6 is a diagram showing a neuron model by McCullough Pitts.

[Explanation of symbols]

u ₁ ~u _n: input signal w ₁ to w _n: synaptic connection weights v: output signal g (x): Output Functions

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Tetsuro Murayama Sanyo Kasei Co., Ltd. Research Institute, 1000, Kamoshida-cho, Midori-ku, Yokohama-shi, Kanagawa

Claims (1)

[Claims] 1. An optical information processing system in which at least one electrode has a photoelectric conversion layer between electrodes having light transmissivity and a memory layer having a function of maintaining conductivity changed by light having a certain wavelength even after light is blocked. Obtained by inputting light information of light having a wavelength that changes the conductivity of the memory layer into the device to store the light information, and then making light information of light having a wavelength causing photoelectric conversion enter the photoelectric conversion layer. An optical information processing method characterized by recognizing optical information from an output signal.