Semiconductor device and determination system

The semiconductor device employs oxide semiconductors and neural networks for efficient power management and determination, addressing power consumption and accuracy challenges in 5G systems, achieving reduced power usage and faster transitions.

JP7582957B2Active Publication Date: 2024-11-13SEMICON ENERGY LAB CO LTD
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
JP2021550717
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2019-10-04
Filing Date
2020-09-22
Publication Date
2024-11-13
Estimated Expiration
2040-09-22

AI Technical Summary

Technical Problem

Existing semiconductor devices face challenges in reducing power consumption, especially in hibernation states, and improving the accuracy and efficiency of determination systems, particularly in the context of 5G communication standards.

Method used

The implementation of a semiconductor device with a detection circuit, first and second determination circuits, and a power management system that utilizes oxide semiconductors and neural networks for efficient power gating and contour extraction, enabling reduced power consumption and enhanced determination accuracy.

Benefits of technology

The solution achieves reduced power consumption during hibernation states, faster transition to normal operation, and improved determination accuracy by utilizing oxide semiconductors and neural networks, enhancing the performance of semiconductor devices in 5G communication systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007582957000001
    Figure 0007582957000001
  • Figure 0007582957000002
    Figure 0007582957000002
  • Figure 0007582957000003
    Figure 0007582957000003
Patent Text Reader

Abstract

The purpose of the present invention is to reduce power consumption by a circuit for making determinations, to improve the accuracy of a system for making determinations, to improve the safety of an object monitored by a sensor element, or to provide a system for easily monitoring an object. This semiconductor device comprises a detection circuit having a function for analyzing first data to make a first determination in which a first value or a second value is selected, a first determination circuit and a second determination circuit having a function for extracting features of an image, a power circuit, and a power management device. The power management device has a function for supplying a voltage from the power circuit to the first determination circuit when the first value is selected by the first determination, the first determination circuit has a function for analyzing the first data to make a second determination, and the second determination circuit has a function for analyzing the first data to make a third determination when the occurrence of an event is detected during the second determination.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical Field]

[0001] One embodiment of the present invention relates to a semiconductor device.

[0002] Note that one aspect of the present invention is not limited to the above-mentioned technical fields. The technical fields of the inventions disclosed in this specification relate to products, methods, or manufacturing methods. Alternatively, one aspect of the present invention relates to processes, machines, manufactures, or compositions of matter.

[0003] In this specification and the like, a semiconductor device generally refers to anything that can function by utilizing semiconductor characteristics. Therefore, semiconductor elements such as transistors and diodes, and circuits including semiconductor elements are semiconductor devices. Furthermore, display devices, light-emitting devices, lighting devices, electro-optical devices, communication devices, and electronic devices may include semiconductor elements and semiconductor circuits. Therefore, display devices, light-emitting devices, lighting devices, electro-optical devices, imaging devices, communication devices, and electronic devices may also be called semiconductor devices. [Background technology]

[0004] Portable information terminals such as smartphones and tablet devices are becoming increasingly popular. As information terminals become more popular, various communication standards have been established. For example, the LTE-Advanced standard, also known as the fourth generation mobile communication system (4G), has begun operation.

[0005] In recent years, advances in information technology such as the Internet of Things (IoT) have led to an increase in the amount of data handled by information terminals. In addition, there is a demand for improved communication speeds for electronic devices such as information terminals.

[0006] To accommodate various information technologies such as IoT, a new communication standard called the fifth-generation mobile communication system (5G) is being considered, which will achieve faster communication speeds, more simultaneous connections, and shorter latency than 4G. 5G will use communication frequencies in the 3.7 GHz, 4.5 GHz, and 28 GHz bands.

[0007] 5G-compatible semiconductor devices are made using semiconductors that use a single element, such as Si, as their main component, or compound semiconductors that use multiple elements, such as Ga and As, as their main components. Furthermore, oxide semiconductors, a type of metal oxide, are attracting attention.

[0008] In oxide semiconductors, c-axis aligned crystalline (CAAC) structures and nanocrystalline (nc) structures, which are neither single crystal nor amorphous, have been found (see Non-Patent Documents 1 and 2).

[0009] Non-Patent Documents 1 and 2 disclose techniques for manufacturing a transistor using an oxide semiconductor having a CAAC structure.

[0010] Also, known techniques for reducing power consumption in semiconductor devices include, for example, power gating (PG), clock gating (CG), voltage scaling, etc. For example, Patent Document 1 describes implementing a method that is advantageous for reducing power consumption out of the DVFS (Dynamic Voltage and Frequency Scaling) method and the PG method. [Prior art documents] [Patent documents]

[0011] [Patent Document 1] International Publication No. 2009 / 078081 [Non-patent literature]

[0012] [Non-Patent Document 1] S. Yamazaki et al., “SID Symposium Digest of Technical Papers”, 2012, volume 43, issue 1, p.183-186 [Non-patent document 2] S. Yamazaki et al., “Japanese Journal of Applied Physics”, 2014, volume 53, Number 4S, p.04ED18-1-04ED18-10 Summary of the Invention [Problem to be solved by the invention]

[0013] An object of one embodiment of the present invention is to provide a novel semiconductor device or an operating method of the novel semiconductor device. Another object of one embodiment of the present invention is to provide a novel determination system or an operating method of the novel determination system. Another object of one embodiment of the present invention is to reduce power consumption, for example, to reduce power consumption in a hibernation state. Another object of one embodiment of the present invention is to shorten the time required for returning from a hibernation state to a normal state or to reduce the energy required for the process. Another object of one embodiment of the present invention is to reduce power consumption of a circuit that performs determination. Another object of one embodiment of the present invention is to improve the accuracy of a determination system. Another object of one embodiment of the present invention is to increase the safety of an object monitored by a sensor element. Another object of one embodiment of the present invention is to provide a system that easily monitors an object.

[0014] It should be noted that the description of multiple problems does not preclude the existence of each other's problems. Note that one embodiment of the present invention does not necessarily solve all of these problems. Furthermore, problems other than those listed will become apparent from the description in the specification, drawings, claims, etc., and these problems may also be problems of one embodiment of the present invention. [Means for solving the problem]

[0015] One embodiment of the present invention is a semiconductor device that includes a detection circuit, a first judgment circuit, a second judgment circuit, a power supply circuit, and a power management device. The detection circuit has a function of analyzing first data and making a first judgment in which a first value or a second value is selected. The first judgment circuit and the second judgment circuit have a function of extracting image features. If the first value is selected by the first judgment, the power management device has a function of supplying a voltage from the power supply circuit to the first judgment circuit. The first judgment circuit has a function of analyzing the first data and making a second judgment. The second judgment circuit has a function of analyzing the first data and making a third judgment if an event is detected in the second judgment.

[0016] In the above configuration, the first decision circuit preferably has a function of extracting contours.

[0017] In the above configuration, it is preferable that the first judgment circuit has a function of performing contour extraction, and the second judgment circuit has a function of executing one or more methods selected from a deep neural network, a convolutional neural network, a recurrent neural network, an autoencoder, a deep Boltzmann machine, and a deep belief network.

[0018] In the above configuration, it is preferable that the device has an antenna and has a function of transmitting the result of the third determination by wireless communication.

[0019] In the above configuration, the event is preferably human detection using contour extraction.

[0020] In the above configuration, the first determination circuit preferably has a function of performing human face authentication using contour extraction.

[0021] Alternatively, one embodiment of the present invention includes an imaging device, a detection circuit, a processing device, a second judgment circuit, and a power supply circuit. The processing device includes a first judgment circuit, a power supply management device, a processor core, and a memory circuit. The processor core has a function of issuing instructions to the first judgment circuit. The memory circuit has a function of holding data generated by the processor core. The imaging device has a function of acquiring first data. The detection circuit has a function of analyzing the first data and making a first judgment in which a first value or a second value is selected. The first judgment circuit and the second judgment circuit have a function of extracting features of an image. If the first value is selected by the first judgment, the power supply management device has a function of causing the power supply circuit to supply a voltage to the first judgment circuit. The first judgment circuit has a function of analyzing the first data and making a second judgment. If an occurrence of an event is detected in the second judgment, the second judgment circuit has a function of analyzing the first data and making a third judgment. The imaging device is a judgment system having a plurality of pixel circuits arranged in a matrix.

[0022] In the above structure, it is preferable that the memory circuit includes a first transistor and a first capacitor, one of a source and a drain of the first transistor is electrically connected to one electrode of the first capacitor, and the first transistor has an oxide semiconductor in a channel formation region.

[0023] In the above structure, it is preferable that the memory circuit includes a first transistor and a first capacitor, one of a source and a drain of the first transistor being electrically connected to one electrode of the first capacitor, the first transistor having an oxide semiconductor in a channel formation region, each of the plurality of pixel circuits including a photoelectric conversion device, a second transistor, and a second capacitor, one electrode of the photoelectric conversion device being electrically connected to one of a source and a drain of the second transistor, the other of the source and drain of the second transistor being electrically connected to one electrode of the second capacitor, and the second transistor having an oxide semiconductor in a channel formation region.

[0024] In the above configuration, the first decision circuit preferably has a function of extracting contours.

[0025] In the above configuration, it is preferable that the first judgment circuit has a function of performing contour extraction, and the second judgment circuit has a function of executing one or more methods selected from a deep neural network, a convolutional neural network, a recurrent neural network, an autoencoder, a deep Boltzmann machine, and a deep belief network. [Effects of the Invention]

[0026] According to one embodiment of the present invention, a novel semiconductor device or an operating method of the novel semiconductor device can be provided. According to one embodiment of the present invention, a novel determination system or an operating method of the novel determination system can be provided. According to one embodiment of the present invention, power consumption can be reduced, for example, power consumption in a hibernation state can be reduced. According to one embodiment of the present invention, the time required for returning from a hibernation state to a normal state can be shortened or the energy required for the process can be reduced. According to one embodiment of the present invention, power consumption of a circuit that performs determination can be reduced. According to one embodiment of the present invention, the accuracy of a determination system can be improved. According to one embodiment of the present invention, the safety of an object monitored by a sensor element can be improved. According to one embodiment of the present invention, a system that easily monitors an object can be provided.

[0027] The description of these effects does not preclude the existence of other effects. Furthermore, one embodiment of the present invention does not necessarily have all of the exemplified effects. Furthermore, problems, effects, and novel features of one embodiment of the present invention other than those described above will become apparent from the description and drawings of this specification. [Brief explanation of the drawings]

[0028] FIG. 1 is a block diagram showing an example of the configuration of a determination system. FIG. 2 is a block diagram showing an example of the configuration of the determination system. FIG. 3 is a flow chart showing an example of the operation of the semiconductor device. 4A and 4B are block diagrams showing configuration examples of a semiconductor device. 5A to 5D are diagrams illustrating an example of the operation of power management of a semiconductor device. FIG. 6 is a flowchart showing an example of the power management operation of the semiconductor device. 7A and 7B are block diagrams showing configuration examples of a semiconductor device. FIG. 8 is a block diagram showing an example of the configuration of a processor core. FIG. 9 is a circuit diagram showing an example of the configuration of the memory circuit. FIG. 10 is a timing chart illustrating an example of the operation of the memory circuit. FIG. 11 is a circuit diagram showing an example of the configuration of a memory cell of a cache. FIG. 12 is a timing chart illustrating an example of the operation of the memory cell. 13A is a functional block diagram showing an example of the configuration of a NOSRAM, and FIG. 13B is a circuit diagram showing an example of the configuration of a memory cell. Fig. 14A is a circuit diagram showing an example of the configuration of a memory cell array, and Fig. 14B and Fig. 14C are circuit diagrams showing an example of the configuration of a memory cell. Fig. 15A is a circuit diagram showing an example of the configuration of a memory cell of a DOSRAM, and Fig. 15B is a diagram showing an example of a stacked structure of a DOSRAM. 16A and 16B are diagrams showing examples of the configuration of a neural network. FIG. 17 is a diagram illustrating a configuration example of a semiconductor device. FIG. 18 is a diagram showing an example of the configuration of a memory cell. FIG. 19 is a diagram illustrating an example of the configuration of the offset circuit. FIG. 20 is a timing chart. Fig. 21A is a block diagram illustrating an imaging device, and Fig. 21B and Fig. 21C are circuit diagrams illustrating pixel circuits. Fig. 22A is a diagram illustrating the operation of a rolling shutter, and Fig. 22B is a diagram illustrating the operation of a global shutter. 23A and 23B are timing charts illustrating the operation of the pixel circuit. 24A and 24B are circuit diagrams illustrating pixel circuits. FIG. 25 is a circuit diagram and a block diagram illustrating a read circuit. 26A and 26B are cross-sectional views illustrating a pixel. 27A, 27B, and 27C are diagrams illustrating a Si transistor. 28A and 28B are cross-sectional views illustrating a pixel. 29A, 29B, 29C, and 29D illustrate OS transistors. 30A and 30B are cross-sectional views illustrating an example of an electronic device. DETAILED DESCRIPTION OF THE INVENTION

[0029] The embodiments will be described in detail with reference to the drawings. However, the present invention is not limited to the following description, and it will be readily understood by those skilled in the art that various modifications in form and detail can be made without departing from the spirit and scope of the present invention. Therefore, the present invention should not be interpreted as being limited to the description of the embodiments shown below. In the configuration of the invention described below, the same parts or parts having similar functions will be denoted by the same reference numerals in different drawings, and repeated descriptions thereof will be omitted.

[0030] Furthermore, the position, size, range, etc. of each component shown in the drawings, etc. may not represent the actual position, size, range, etc. in order to facilitate understanding of the invention. Therefore, the disclosed invention is not necessarily limited to the position, size, range, etc. disclosed in the drawings, etc. For example, in an actual manufacturing process, a resist mask, etc. may be unintentionally eroded by a process such as etching, but this may not be reflected in the drawings in order to facilitate understanding.

[0031] In addition, in top views (also called "plan views"), perspective views, and the like, some components may be omitted to make the drawings easier to understand.

[0032] Furthermore, the terms "electrode" and "wiring" used in this specification and elsewhere do not limit the functionality of these components. For example, an "electrode" may be used as part of a "wiring," and vice versa. Furthermore, the terms "electrode" and "wiring" also include cases where multiple "electrodes" or "wirings" are integrally formed.

[0033] In this specification, a "terminal" in an electric circuit refers to a portion where a current is input or output, a voltage is input or output, or a signal is received or transmitted. Therefore, a part of a wiring or an electrode may function as a terminal.

[0034] In this specification, the terms "above" and "below" do not limit the positional relationship of components to being directly above or below, and being in direct contact with each other. For example, the expression "electrode B on insulating layer A" does not necessarily mean that electrode B is formed on insulating layer A in direct contact with it, and does not exclude the inclusion of other components between insulating layer A and electrode B.

[0035] In addition, the functions of the source and drain are interchangeable depending on operating conditions, such as when transistors of different polarities are used or when the direction of current changes during circuit operation, making it difficult to define which is the source and which is the drain. For this reason, the terms source and drain can be used interchangeably in this specification.

[0036] Furthermore, in this specification, "electrically connected" includes both direct connection and connection via "something that has some kind of electrical effect." Here, "something that has some kind of electrical effect" is not particularly limited as long as it allows electrical signals to be transmitted and received between the connected objects. Therefore, even when the expression "electrically connected" is used, in an actual circuit, there may be no physical connection and only wiring may be extended.

[0037] Furthermore, in this specification and elsewhere, "parallel" refers to, for example, a state in which two straight lines are arranged at an angle of -10° or more and 10° or less. Therefore, it also includes cases in which the angle is -5° or more and 5° or less. Furthermore, "perpendicular" and "orthogonal" refer to, for example, a state in which two straight lines are arranged at an angle of 80° or more and 100° or less. Therefore, it also includes cases in which the angle is 85° or more and 95° or less.

[0038] In this specification and elsewhere, when referring to counting values ​​and measurement values, terms such as "same," "equal," or "uniform" are used, they are considered to include an error of plus or minus 20%, unless otherwise specified.

[0039] Furthermore, voltage often refers to the potential difference between a certain potential and a reference potential (for example, ground potential or source potential). Therefore, voltage and potential can often be used interchangeably. In this specification and elsewhere, unless otherwise specified, voltage and potential can be used interchangeably.

[0040] It should be noted that even when written as "semiconductor," if the conductivity is sufficiently low, it will have the properties of an "insulator." Therefore, it is also possible to use "semiconductor" instead of "insulator." In this case, the boundary between "semiconductor" and "insulator" is vague, and it is difficult to strictly distinguish between the two. Therefore, "semiconductor" and "insulator" described in this specification may be read interchangeably.

[0041] Furthermore, even when written as "semiconductor," if the conductivity is sufficiently high, it will have the properties of a "conductor." Therefore, it is also possible to use "semiconductor" instead of "conductor." In this case, the boundary between "semiconductor" and "conductor" is vague, and it is difficult to strictly distinguish between the two. Therefore, "semiconductor" and "conductor" described in this specification may be read interchangeably.

[0042] Note that ordinal numbers such as "first" and "second" used in this specification are used to avoid confusion between components, and do not indicate any order or ranking, such as the order of processes or stacking. Furthermore, even if a term does not have an ordinal number in this specification, an ordinal number may be added in the claims to avoid confusion between components. Furthermore, even if a term has an ordinal number in this specification, a different ordinal number may be added in the claims. Furthermore, even if a term has an ordinal number in this specification, the ordinal number may be omitted in the claims.

[0043] In this specification and the like, the "on state" of a transistor refers to a state in which the source and drain of the transistor are considered to be electrically short-circuited (also referred to as a "conductive state"). The "off state" of a transistor refers to a state in which the source and drain of the transistor are considered to be electrically disconnected (also referred to as a "non-conductive state").

[0044] In this specification, the term "on-state current" may refer to a current that flows between the source and drain of a transistor when the transistor is on, and the term "off-state current" may refer to a current that flows between the source and drain of a transistor when the transistor is off.

[0045] In this specification, a high power supply potential (hereinafter referred to as "power supply potential VDD," "VDD," "H potential," or "H") refers to a power supply potential that is higher than a low power supply potential (hereinafter referred to as "power supply potential VSS," "VSS," "L potential," or "L"). VSS refers to a power supply potential that is lower than VDD. Ground potential (hereinafter simply referred to as "GND" or "GND potential") can also be used as VDD or VSS. For example, when VDD is ground potential, VSS is a potential lower than ground potential, and when VSS is ground potential, VDD is a potential higher than ground potential.

[0046] In this specification and the like, a gate refers to a gate electrode and a part or all of a gate wiring, and a gate wiring refers to a wiring for electrically connecting the gate electrode of at least one transistor to another electrode or another wiring.

[0047] In this specification, the term "source" refers to a source region, a source electrode, and part or all of a source wiring. The term "source region" refers to a region of a semiconductor layer whose resistivity is equal to or less than a certain value. The term "source electrode" refers to a conductive layer connected to the source region. The term "source wiring" refers to wiring that electrically connects the source electrode of at least one transistor to another electrode or wiring.

[0048] In this specification, the term "drain" refers to a part or all of the drain region, drain electrode, and drain wiring. The term "drain region" refers to a region of the semiconductor layer whose resistivity is equal to or less than a certain value. The term "drain electrode" refers to a conductive layer connected to the drain region. The term "drain wiring" refers to wiring that electrically connects the drain electrode of at least one transistor to another electrode or wiring.

[0049] In addition, in drawings, etc., to make the potential of wiring, electrodes, etc. easier to understand, an "H" indicating an H potential or an "L" indicating an L potential may be written next to the wiring, electrode, etc. Furthermore, wiring, electrodes, etc. where a potential change has occurred may be written with "H" or "L" enclosed in letters. Furthermore, when a transistor is in an off state, an "x" symbol may be written next to the transistor.

[0050] A terminal may refer to a group of multiple terminals. For example, an independent signal is applied to each terminal of the group of multiple terminals, and one or more wires are electrically connected to each terminal.

[0051] A transistor has three terminals (nodes) called a gate, a source, and a drain. The gate is a terminal that functions as a control terminal that controls the conduction state of the transistor. Of a pair of input / output terminals (nodes) that function as a source or a drain, one becomes the source and the other becomes the drain depending on the type of transistor and the level of the potential applied to each terminal (node). Generally, in an n-type transistor, the node to which a low potential is applied is called the source, and the node to which a high potential is applied is called the drain. Conversely, in a p-type transistor, the node to which a low potential is applied is called the drain, and the node to which a high potential is applied is called the source. In this specification, the two terminals (nodes) other than the gate may be called the first terminal (node) and the second terminal (node).

[0052] In this specification, in order to facilitate understanding of the circuit configuration and its operation, one of the two input / output terminals (nodes) of a transistor may be limited to the source and the other to the drain. Of course, depending on the driving method, the magnitude relationship between the potentials applied to the three terminals of the transistor may change, and the source and drain may be interchanged. Therefore, in one embodiment of the present invention, the distinction between the source and drain of a transistor is not limited to the description in the specification and drawings.

[0053] In this specification, etc., a person skilled in the art may be able to construct an embodiment of an invention without specifying the connection destinations of all terminals of active elements (e.g., transistors, diodes, etc.) and passive elements (e.g., capacitance elements, resistance elements, etc.). In other words, it can be said that an embodiment of an invention is clear even without specifying the connection destinations. Furthermore, when an embodiment in which the connection destinations are specified is described in this specification, etc., it may be possible to determine that an embodiment of an invention in which the connection destinations are not specified is also described in this specification. In particular, when there are multiple possible connection destinations for a terminal, it is not necessary to limit the connection destination of that terminal to a specific location. Therefore, it may be possible to construct an embodiment of an invention by specifying the connection destinations of only some of the terminals of active elements (e.g., transistors, diodes, etc.) and passive elements (e.g., capacitance elements, resistance elements, etc.).

[0054] In this specification, etc., a person skilled in the art may be able to identify an invention by at least specifying the connection destination of a certain circuit. Alternatively, a person skilled in the art may be able to identify an invention by at least specifying the function of a certain circuit. In other words, if the function can be identified, it can be said that the aspect of the invention is clear. It may then be possible to determine that one aspect of the invention with a specified function is described in this specification, etc. Therefore, if the connection destination of a certain circuit is specified even without specifying the function, one aspect of the invention is disclosed and can constitute one aspect of the invention. Alternatively, if the function of a certain circuit is specified even without specifying the connection destination, one aspect of the invention is disclosed and can constitute one aspect of the invention.

[0055] (Embodiment 1) In this embodiment, a semiconductor device according to one embodiment of the present invention and application examples of the semiconductor device will be described.

[0056] <Configuration example of semiconductor device> An example of a determination system according to one embodiment of the present invention is described below. The determination system according to one embodiment of the present invention includes a semiconductor device. Preferably, the determination system according to one embodiment of the present invention includes an imaging device in addition to the semiconductor device. The semiconductor device according to one embodiment of the present invention includes a processing device. The processing device according to one embodiment of the present invention has a function of turning the power supply of some of its circuits into an off mode by power gating operation. The power supply of the circuit that has been turned into the off mode can be restored to an on mode in response to an interrupt request.

[0057] FIG. 1 shows an example of the configuration of a determination system including a semiconductor device and a device such as an imaging device. The semiconductor device 700 shown in FIG. 1 is electrically connected to an imaging device 601. The semiconductor device 700 includes a detection circuit 92, a processing unit (PU) 20, and a determination circuit 91. The PU 20 includes a preliminary determination circuit 80. The determination circuit 91 has a function of performing calculations using a neural network. The preliminary determination circuit 80 may be referred to as a first determination circuit, and the determination circuit 91 may be referred to as a second determination circuit.

[0058] The imaging device 601 has a function of capturing an image using visible light, and preferably has a function of capturing an image using infrared light.

[0059] When the imaging device 601 captures an image, a light source may be prepared and irradiated onto the subject.

[0060] 1 also includes a display device 602. The display device 602 is electrically connected to the imaging device 601. The display device 602 has a function of displaying an image captured by the imaging device 601. The display device 602 includes, for example, a display unit and a driver circuit that controls the display unit.

[0061] 1 also includes a terminal 603. The semiconductor device 700 preferably has a function of performing wireless communication with the terminal 603. The semiconductor device 700 includes, for example, an antenna for wireless communication.

[0062] The semiconductor device 700 can perform analysis in stages in the order of the detection circuit 92, the preliminary determination circuit 80, and the determination circuit 91. The more stages there are, the more accurate the analysis becomes. The more stages there are to perform analysis, the more accurate the determination can be.

[0063] On the other hand, as the accuracy of the analysis improves, the scale of the arithmetic circuit increases, which may increase the power consumption of the semiconductor device 700. Therefore, if the detection circuit 92 or the preliminary determination circuit 80 determines that the analysis is sufficient, it is not necessary to proceed to the next stage. The fewer stages there are, the more the power consumption of the semiconductor device 700 can be reduced.

[0064] The semiconductor device of one embodiment of the present invention has a function of determining whether an abnormal event has occurred. In the semiconductor device of one embodiment of the present invention, the determination is preferably performed in stages. In this case, in an initial stage, it is necessary to determine only whether an event has occurred, and not to determine whether the occurred event is abnormal.

[0065] The detection circuit 92 receives imaging data from the imaging device 601. The detection circuit 92 has a function of performing binary classification. More specifically, the detection circuit 92 analyzes the received data and determines whether an event has occurred. The detection circuit 92 does not, for example, determine whether the event is abnormal. The detection circuit 92 analyzes the imaging data and performs a first determination. The first determination is a binary classification. If a predetermined standard is exceeded, a first value is selected, and if not, a second value is selected. Here, exceeding the predetermined standard refers, for example, to analyzing the imaging data, quantifying the results, and determining that the resulting value exceeds the standard. Alternatively, for example, the imaging data is analyzed and a determination is made that an event has occurred. If the first value is selected by the first determination performed by the detection circuit 92, a signal INT is provided to the PU 20. As a next step, the imaging data is analyzed in the preliminary determination circuit 80.

[0066] As an example, consider a case where the imaging device 601 is used to identify a suspicious person. In such a case, for example, if an image presumably of a person is detected in the captured image, it may be determined that "an event has occurred." In this case, even if an event has occurred in the detection circuit 92, the captured person is not necessarily a suspicious person. The event may be a simple phenomenon. For example, the event may be "detection of movement of a subject." Or, for example, the event may be "detection of a subject." Or, for example, the event may be "a significant change in brightness."

[0067] The detection circuit 92 performs calculations that are simpler than those performed by the preliminary determination circuit 80 and the determination circuit 91, which will be described later. For example, the detection circuit 92 analyzes the spatial luminance distribution of the imaging data to detect the subject. Alternatively, the detection circuit 92 sets a threshold value and determines that an event has occurred if the obtained data exceeds the threshold value.

[0068] The detection circuit 92 has a function of sending a signal INT to the PU 20 when an event occurs. The signal INT is a signal requesting interrupt processing. The power supply of a circuit in the PU 20 that has been put into off mode, such as a processor core, can return to on mode in response to the interrupt request. The PU 20 then executes a desired command, and the preliminary determination circuit 80 of the PU 20 determines whether an abnormal event has occurred. Specifically, for example, the preliminary determination circuit 80 analyzes the captured image data and determines whether a suspicious person is present in the image.

[0069] When the PU 20 does not perform analysis in the preliminary decision circuit 80, the power consumption can be significantly reduced by the power gating operation.

[0070] The PU 20 is a circuit that has the function of executing instructions. The detailed configuration of the PU 20 will be described later.

[0071] The preliminary determination circuit 80 may determine only whether the subject is a "human" or not, without determining whether the subject is a suspicious person or not. In such a case, the determination circuit 91 may determine whether the subject is a suspicious person or not.

[0072] The preliminary determination circuit 80, for example, extracts the contours of the subject and detects a human. The contour extraction can be performed, for example, by comparing spatial brightness. Specifically, analysis can be performed using HoG (Histogram of Oriented Gradient) and SVM (Support Vector Machine).

[0073] The preliminary determination circuit 80 may also analyze, for example, temporal changes in the subject, more specifically, changes in the subject from frame to frame, and may infer whether the human behavior is suspicious by analyzing the temporal changes in the subject.

[0074] It is preferable that the preliminary judgment circuit 80 performs analysis with lower power consumption than the judgment circuit 91. It is preferable that the preliminary judgment circuit 80 performs calculations on a smaller scale than the judgment circuit 91. It is preferable that the preliminary judgment circuit 80 performs calculations in a shorter time than the judgment circuit 91.

[0075] When the preliminary decision circuit 80 uses a neural network, it is preferable that the preliminary decision circuit 80 performs calculations on a smaller scale than the decision circuit 91. For example, when the preliminary decision circuit 80 and the decision circuit 91 both perform convolutional neural networks (CNN), the number of hidden layers used in the preliminary decision circuit 80 is smaller than the number of hidden layers used in the decision circuit 91.

[0076] In the preliminary judgment circuit 80, the imaging data or the data converted from the imaging data is analyzed and a second judgment is made. If the second judgment exceeds a predetermined standard, it is preferable that, as the next step, the imaging data or the data converted from the imaging data is analyzed in the judgment circuit 91. More specifically, for example, if it is judged in the second judgment that an abnormal event has occurred, it is preferable that the judgment circuit 91 further analyzes the data and judges whether or not an abnormal event has occurred in order to improve the accuracy of the judgment. The judgment circuit 91 has a function of performing calculations using, for example, a neural network.

[0077] After the preliminary determination circuit 80 makes a determination, the determination accuracy can be improved by performing a more detailed analysis in the determination circuit 91. Furthermore, if the preliminary determination circuit 80 makes a determination in advance and determines that no abnormal event has occurred, it is not necessary to perform calculations in the determination circuit 91, and therefore the power consumption of the semiconductor device 700 can be reduced.

[0078] The decision circuit 91 can perform calculations with higher accuracy than the preliminary decision circuit 80. It is preferable that the decision circuit 91 performs analysis using artificial intelligence (AI).

[0079] The determination circuit 91 preferably analyzes the imaging data by executing one or a combination of two or more methods selected from a deep neural network (DNN), a convolutional neural network (CNN), a recurrent neural network (RNN), an autoencoder, a deep Boltzmann machine (DBM), a deep belief network (DBN), etc. More specifically, the determination circuit 91 can use a CNN to extract image features.

[0080] Furthermore, in the determination circuit 91, before extracting image features using CNN or the like, processing to increase the image resolution or colorizing a black-and-white image may be performed. For example, a method called a generative adversarial network (GAN) may be used as a colorization method.

[0081] When the determination circuit 91 determines that an abnormal event has occurred, the semiconductor device 700 transmits a signal notifying the terminal 603 of the abnormality via wireless communication. The semiconductor device 700 may also transmit image data for which it has been determined that an abnormal event has occurred, along with the signal notifying the abnormality. At this time, the image data may be subjected to processing such as increasing the resolution or colorization.

[0082] An example of the configuration of a semiconductor device will be described with reference to Fig. 2. A semiconductor device 700 shown in Fig. 1 includes a power supply circuit 10, a PU 20, a determination circuit 91, and a detection circuit 92. The semiconductor device 700 is electrically connected to an imaging device 601.

[0083] The PU 20 is a circuit that has the function of executing instructions. The PU 20 has multiple functional circuits integrated on a single chip. The PU 20 has a processor core 30, a power management unit (PMU) 60, a clock control circuit 65, a power switch (PSW) 70, a preliminary determination circuit 80, a memory 81, a circuit 82, an interface 88, and a bus line 89.

[0084] The PU 20 may also have a cache 40 and a power switch (PSW) 71. The cache 40 and the PSW 71 will be described in detail later.

[0085] The processor core 30, PMU 60, cache 40, PSW 70, PSW 71, preliminary determination circuit 80, memory 81, circuit 82, and interface 88 are electrically connected via a bus line 89. Signals can be exchanged between the circuits via the bus line 89. The processor core 30 has a function of issuing instructions to the preliminary determination circuit 80, for example.

[0086] The memory 81 has a function of holding imaging data provided from the imaging device 601 to the PU 20. As the memory 81, a memory configured using an OS transistor, which will be described later, may be used.

[0087] The interface 88 has a function of exchanging signals with devices external to the PU 20 .

[0088] The interface 88 has a function of sending an interrupt signal INT provided from the detection circuit 92 to the processor core 30 and the PMU 60 via a bus line 89 .

[0089] The interface 88 also receives signals from the processor core 30 etc. via a bus line 89, and based on the received signals, provides a signal OU1 as a control signal to a determination circuit 91. The determination circuit 91 also provides a signal IN1 as a determination result to the interface 88.

[0090] The circuit 82 includes, for example, an analog-to-digital conversion circuit (hereinafter referred to as an AD conversion circuit). The circuit 82 may also include a circuit for performing wireless communication. Examples of circuits for performing wireless communication include a modulation circuit and a demodulation circuit.

[0091] FIG. 2 shows an example in which the power supply circuit 10 is provided on a different chip from the PU 20. The power supply circuit 10 has the function of inputting a power supply potential MVDD to the PU 20. A reference clock signal CLK is provided to the clock control circuit 65 from the outside. The clock control circuit 65 has the function of providing clock signals to each circuit in the PU 20. In addition, a return sequence can be executed by inputting a signal INT, which is an interrupt signal requesting interrupt processing, from the outside to the interface 88. The PMU 60 has the function of generating a control signal and providing it to the power supply circuit 10.

[0092] In a semiconductor device according to one embodiment of the present invention, the number of bits that a processing device can handle in an arithmetic circuit or the like can be, for example, 8 bits, 16 bits, 32 bits, or 64 bits.

[0093] The clock control circuit 65 has a function of receiving the reference clock signal CLK, generating and outputting a gated clock signal.

[0094] The imaging data captured by the imaging device 601 is, for example, analog data. If the imaging device 601 has an AD conversion circuit, the analog data captured by the imaging device 601 is preferably converted by the AD conversion circuit and then provided to the semiconductor device 700.

[0095] Furthermore, analog data captured by the imaging device 601 may be provided to the semiconductor device 700 without being converted into digital data. In this case, for example, the semiconductor device 700 may perform data conversion using an AD conversion circuit included in the circuit 82.

[0096] For example, digital data converted by an AD conversion circuit is provided to the detection circuit 92. Alternatively, the detection circuit 92 may have a function of analyzing analog data.

[0097] Furthermore, digital data converted by, for example, an AD conversion circuit is provided to the preliminary determination circuit 80. Alternatively, the preliminary determination circuit 80 may have a function of analyzing analog data.

[0098] Furthermore, digital data converted by, for example, an AD conversion circuit is provided to the decision circuit 91. Alternatively, the decision circuit 91 may have a function of analyzing analog data.

[0099] <Example of semiconductor device operation> An operation example of a semiconductor device of one embodiment of the present invention will be described with reference to the flowchart in FIG.

[0100] In step S000, the process starts.

[0101] In step S001, imaging data of the imaging device 601 is provided to the detection circuit 92. The imaging data is, for example, a still image or a moving image of the location to be monitored.

[0102] In step S002, the detection circuit 92 analyzes the imaging data. If the detection circuit 92 determines as a result of the analysis that "an event has occurred," the process proceeds to step S003. If the detection circuit 92 determines that "an event has not occurred," the process returns to step S001. Here, the occurrence of an event refers to, for example, the detection of a "person," or the occurrence of an even simpler event. Examples of simple events include the aforementioned "detection of subject movement," "detection of a subject," and "significant change in brightness."

[0103] In step S003, the detection circuit 92 sends a signal INT to the PU 20, which then executes a recovery sequence. After the PU 20 has recovered, the image data from the image capture device 601, or data obtained by processing the image data from the image capture device 601, is sent to the PU 20. The processing performed on the image data refers to one or a combination of processes selected from, for example, signal amplification, analog-to-digital conversion of the signal, digital-to-analog conversion, signal compression, signal restoration, and the like.

[0104] In step S004, the imaging data is analyzed by the preliminary determination circuit 80 of the PU 20. One example of the imaging data analysis is the detection of a "human" by contour extraction using HoG and SVM.

[0105] In step S005, if the preliminary determination circuit 80 determines that "an event has occurred" as a result of the analysis in step S004, the process proceeds to step S006. If it determines that "an event has not occurred," the process proceeds to step S009.

[0106] Here, the preliminary determination circuit 80 may determine not only whether an event has occurred but also whether the event that has occurred is abnormal. For example, after detecting a "person" by contour extraction, face authentication of the detected "person" may be performed.

[0107] In step S009, the PU 20 transitions to the sleep state, and then the process returns to step S001.

[0108] In step S006, the imaging data of the imaging device 601 or data obtained by processing the imaging data of the imaging device 601 is provided to the determination circuit 91, which analyzes the imaging data. Specifically, for example, feature extraction is performed using CNN to recognize a "human" face. The recognized face is then compared with a database to determine whether or not the person is a dangerous person. The database is, for example, included in the PU 20 or the determination circuit 91. Alternatively, the database may be included in an external server. It is preferable that the external server and the PU 20 communicate with each other wirelessly, for example, and data is transferred between the server and the PU 20.

[0109] In step S007, if the determination circuit 91 determines, as a result of the analysis in step S006, that an abnormal event has occurred, the process proceeds to step S008. If it determines that no abnormal event has occurred, the process proceeds to step S009. Here, the occurrence of an abnormal event refers, for example, to the detection of an unexpected subject. Alternatively, the occurrence of an abnormal event refers, for example, to the detection of an object, subject, person, etc. that matches or is likely to match a pre-registered database. Alternatively, the occurrence of an abnormal event refers, for example, to the detection of an object, subject, person, etc. that is determined to be dangerous. Alternatively, the occurrence of an abnormal event refers, for example, to the determination circuit 91 determining, as a result of analysis of the imaging data, that the face of a recognized "human" is likely to match that of a dangerous person.

[0110] In step S008, the semiconductor device 700 sends a signal notifying the terminal 603 of an abnormality. Here, in addition to the signal notifying the abnormality, the semiconductor device 700 may send to the terminal 603 an image of the face of the "person" determined to be highly likely to match the dangerous person.

[0111] The processor core, memory circuitry and cache are described in more detail below.

[0112] <Processor core 30, memory circuit 31> FIG. 4A shows an example of the configuration of a processor core 30. The processor core 30 is a circuit capable of processing instructions and can be called an arithmetic processing circuit. It includes a memory circuit 31 and multiple combinational circuits 32, which constitute various functional circuits. For example, the memory circuit 31 is included in a register. In FIG. 4A and other figures, the input signal of the memory circuit 31 is denoted as signal D, and the output signal of the memory circuit 31 is denoted as signal Q.

[0113] As shown in FIG. 4B, the memory circuit 31 includes a circuit MemC1 and a circuit BKC1. The circuit MemC1 has the function of retaining data generated by the processor core 30 and can be configured, for example, with a flip-flop circuit (FF), a latch circuit, etc. The circuit BKC1 functions as a backup circuit for the circuit MemC1 and is capable of retaining data for a long period of time even when the power supply or clock signal is interrupted. The inclusion of this memory circuit 31 enables power gating of the processor core 30. This is because, by saving the data in the circuit MemC1 to the circuit BKC1 in the memory circuit 31 before power is interrupted, the state of the processor core 30 at the time of power interruption can be maintained. When power supply is resumed, the data retained in the circuit BKC1 is written to the circuit MemC1, allowing the processor core 30 to return to the state it was in at the time of power interruption. Therefore, the PU 20 can immediately perform normal processing operations after power supply is resumed.

[0114] Circuit BKC1 includes at least a retention circuit having one transistor (MW1) and one capacitor (CB1). The retention circuit shown in FIG. 4B has a circuit configuration similar to that of a 1T1C (one transistor, one capacitor) memory cell of a standard DRAM (dynamic random access memory), and it can perform write and read operations in the same way. The charging and discharging of capacitor CB1 are controlled by controlling the conduction state of transistor MW1. Turning transistor MW1 off places node FN1 in an electrically floating state. By minimizing the drain current (off current) of transistor MW1 in the off state, fluctuations in the potential of node FN1 can be suppressed, thereby extending the data retention time of circuit BKC1. The data retention time of circuit BKC1 is determined by the leakage current of transistor MW1 and the capacitance of capacitor CB1. By using transistor MW1 as a transistor with an extremely small off current, circuit BKC1 does not need to be refreshed while PU20 is operating. This enables circuit BKC1 to be used as a nonvolatile memory circuit.

[0115] It is preferable to use a transistor (also referred to as an "OS transistor" or "OS-FET") that includes an oxide semiconductor (OS), which is a type of metal oxide, in a semiconductor layer where a channel is formed as the transistor MW1. The band gap of an oxide semiconductor is 2 eV or more, and therefore the off-state current is extremely small. In an OS transistor, the normalized off-state current per 1 μm of channel width when the source-drain voltage is 10 V is 10×10 -21 A (10 zepto-A) or less. By using an OS transistor for the transistor MW1, the circuit BKC1 can function substantially as a nonvolatile memory circuit while the PU 20 is operating.

[0116] The oxide semiconductor film used in the semiconductor layer where the channel is formed may be formed as a single-layer oxide semiconductor film or as a stacked oxide semiconductor film. The oxide semiconductor constituting the semiconductor layer where the channel is formed is preferably an oxide containing at least one element selected from the group consisting of In, Ga, Sn, and Zn. Examples of such oxides include In-Sn-Ga-Zn oxide, In-Ga-Zn oxide, In-Sn-Zn oxide, In-Al-Zn oxide, Sn-Ga-Zn oxide, Al-Ga-Zn oxide, Sn-Al-Zn oxide, In-Zn oxide, Sn-Zn oxide, Al-Zn oxide, Zn-Mg oxide, Sn-Mg oxide, In-Mg oxide, In-Ga oxide, indium oxide, tin oxide, and zinc oxide.

[0117] In addition, metal oxides such as In-M-Zn oxide (element M is one or more elements selected from aluminum, gallium, yttrium, copper, vanadium, beryllium, boron, titanium, iron, nickel, germanium, zirconium, molybdenum, lanthanum, cerium, neodymium, hafnium, tantalum, tungsten, magnesium, etc.) can be used as oxide semiconductors.

[0118] Circuit BKC1 writes data using voltage, which reduces the write power required compared to MRAM (Magnetoresistive RAM), which writes data using current. Also, because data is held by the load capacitance of node FN1, there is no limit to the number of times data can be rewritten, as there is with flash memory.

[0119] In circuit BKC1, the energy required to write data corresponds to the energy required to charge and discharge capacitive element CB1. On the other hand, in a storage circuit using a two-terminal storage element such as MRAM, the energy required to write data corresponds to the energy consumed when a current flows through the storage element. In MRAM, a current continues to flow during the data write period, so the energy required to write data is high. Compared to such MRAM, circuit BKC1 can reduce the energy consumed when writing data. Therefore, compared to a storage circuit whose backup circuit is configured with MRAM, storage circuit 31 has more opportunities to perform voltage scaling and power gating, which can reduce energy consumption, thereby reducing the power consumption of PU20.

[0120] <Power management> The PMU 60 has the function of controlling power gating operation, clock gating operation, voltage scaling operation, etc. More specifically, the PMU 60 has the function of controlling the power supply circuit 10, the function of controlling the memory circuit 31, the function of controlling the clock control circuit 65, and the function of controlling the PSW 70. Therefore, the PMU 60 has the function of generating control signals to control these circuits (the power supply circuit 10, the memory circuit 31, the clock control circuit 65, and the PSW 70). The PMU 60 has a circuit 61. The circuit 61 has the function of measuring time. The PMU 60 has the function of managing the power supply based on time-related data obtained by the circuit 61.

[0121] The PSW70 has a function of controlling the supply of a power supply potential MVDD to the PU20 in accordance with a control signal from the PMU60. Here, the power supply potential supplied to the PU20 via the PSW70 is referred to as the power supply potential VDD. The processor core 30 may have multiple power domains. In this case, the PSW70 may be able to independently control the power supply to the multiple power domains. The processor core 30 may also have a power domain that does not require power gating. In this case, the power supply potential may be supplied to this power domain without going through the PSW70.

[0122] The clock control circuit 65 has a function of being able to cut off the clock signal to the processor core 30 in accordance with a control signal from the PMU 60. The power supply circuit 10 has a function of being able to change the magnitude of the power supply potential VDD in accordance with a control signal from the PMU 60.

[0123] The signal SLP output from the processor core 30 to the PMU 60 triggers the transition of the processor core 30 to a sleep state. Upon receiving the signal SLP, the PMU 60 generates a control signal for transitioning to the sleep state and outputs it to the functional circuit being controlled. The power supply circuit 10 lowers the power supply potential MVDD to a value lower than that during normal operation based on the control signal from the PMU 60. After a certain period of time has passed in the sleep state, the PMU 60 controls the PSW 70 to cut off the power supply to the processor core 30. When the processor core 30 transitions from the normal state to the sleep state, the PMU 60 performs a voltage scaling operation to lower the power supply potential VDD of the processor core 30. If the duration of the sleep state exceeds a set time, the PMU 60 performs a power gating operation to stop the supply of the power supply potential VDD to the processor core 30 to further reduce the power consumption of the processor core 30. Power management for the semiconductor device shown in FIG. 2 will now be described with reference to FIGS. 5 and 6.

[0124] FIG. 5 shows a schematic diagram of the change in potential of the power line 35. The power line 35 is a wiring to which a power supply potential VDD is supplied via a PSW 70. The horizontal axis of the diagram represents the elapsed time (time) from the normal state to the sleep state, with t0, t1, etc. representing time. The vertical axis of the diagram represents the potential of the power line. FIG. 5A shows an example in which only power gating is performed in the sleep state, and FIG. 5B shows an example in which only voltage scaling is performed in the sleep state. FIGS. 5C and 5D show examples in which both voltage scaling and power gating are performed. In the normal state, the magnitude of the power supply potential MVDD supplied from the power supply circuit 10 is VH1.

[0125] In the following, the power supply modes of the PU 20 are classified into three modes: a power on mode, a power off mode, and a low power mode. The power on mode is a mode in which a power supply potential VDD that enables normal processing is supplied to the PU 20. The power off mode is a mode in which the supply of the power supply potential VDD is stopped by the PSW 70. The low power mode is a mode in which a power supply potential VDD lower than that in the power on mode is supplied.

[0126] An example of FIG. 5A will be described. At time t0, the processor core 30 starts processing to transition to a sleep state. For example, the memory circuit 31 is backed up. The PMU 60 controls the PSW 70 to cut off the power supply to the processor core 30 at time t1. The power line 35 naturally discharges, and its potential drops to 0V. This significantly reduces the leakage current of the processor core 30 in the sleep state, thereby reducing power consumption in the sleep state (hereinafter sometimes referred to as standby power). When returning to the normal state due to an external interrupt request or the like, the PMU 60 controls the PSW 70 to resume the supply of the power supply potential VDD. Here, the supply of the power supply potential VDD is resumed at time t4. The potential of the power line 35 rises and becomes VH1 at time t6.

[0127] In the example of Figure 5B, to perform voltage scaling, at time t1, PMU 60 controls power supply circuit 10, and the power supply potential MVDD drops to VH2. The potential of power supply line 35 eventually reaches VH2. At time t4, power supply potential MVDD returns from VH2 to VH1, and the potential of power supply line 35 rises, reaching VH1 at time t5.

[0128] In the example of Figure 5A, the time required to return from the sleep state to the normal state (overhead time) is the time required for the potential of the power line 35 to rise from 0V to VH1, and the energy overhead required for return is the energy required to charge the load capacitance of the power line 35 from 0V to VH1. If the period (t1-t4) of the power-off mode is sufficiently long, power gating is effective in reducing the standby power of the PU 20. On the other hand, if the period (t1-t4) is short, the power required to return to the normal state becomes greater than the power that can be reduced by cutting off the power supply, and the effect of power gating cannot be obtained.

[0129] In the voltage scaling example shown in FIG. 5B, the potential of the power supply line 35 is VH2 in the sleep state, so the amount of standby power reduction is smaller than in the power gating example shown in FIG. 5A. On the other hand, in the example shown in FIG. 5B, the potential fluctuation of the power supply line 35 is small, so the time required to return to the normal state is shorter and the energy required for return is less than in the example shown in FIG. 5A. Therefore, the semiconductor device shown in FIG. 2 enables power management that combines power gating and voltage scaling to more efficiently reduce the standby power of PU 20. Examples of power management are shown in FIGS. 5C and 5D.

[0130] As shown in Figure 5C, first, voltage scaling operation is performed in the hibernation state, and the system transitions from power-on mode to low-power mode. As in Figure 5B, at time t1, the PMU 60 controls the power supply circuit 10 to lower the power supply potential MVDD to VH2, so that the potential of the power line 35 eventually reaches VH2. After a certain period (t1-t3) has elapsed since the transition to low-power mode, the PMU 60 controls the PSW 70 to enter power-off mode. During the period (t3-t4), it is possible to reduce power by shutting off the power to the PU 20 using power gating, even if this includes the power consumed to return to the normal state, rather than supplying VH2 to the PU 20.

[0131] For example, assume that potential VH2 is a power supply potential large enough to hold data in circuit MemC1 of memory circuit 31, and potential VH3 is a potential at which the data in circuit MemC1 is lost. In PU20 of FIG. 2, circuit BKC1 is a circuit that can hold data even when power supply is stopped. By saving the data in memory circuit 31 to circuit BKC1 during period (t0-t1), it is possible to lower VDD to potential VH3 in low power mode, at which the data in circuit MemC1 is lost. This further reduces the standby power of PU20.

[0132] The PMU 60 has a function that can return the PU 20 to a normal state based on an interrupt request or the like. The PMU 60 controls the power supply circuit 10 to boost the magnitude of MVDD to VH1, and also controls the PSW 70 to resume the supply of VDD to the PU 20. From time t4 onwards, the system is in power-on mode. At time t6, the potential of the power supply line 35 stabilizes, allowing the PU 20 to operate normally from time t6 onwards.

[0133] FIG. 5D shows an example in which an interrupt request to return to normal operation occurs before time t3. From time t2 onwards, the power-on mode is active. At time t2, the PMU 60 controls the power supply circuit 10 to change the magnitude of MVDD to the power-on mode potential VH1. At time t3, the potential of the power line 35 rises to VH1.

[0134] 5C and 5D, in the hibernation state, the time required to return the potential of the power line 35 to VH1 is longer when returning from the power-off mode to the power-on mode than when returning from the low power mode to the power-on mode. Therefore, the PMU 60 has a function that can adjust the timing of the operation to return the processor core 30 from the hibernation state to the normal state depending on the power mode. This enables the processor core 30 to return from the hibernation state to the normal state in the shortest time.

[0135] Furthermore, in the hibernation state, transition from low power mode to power off mode is possible by measuring time using a circuit 61 provided in the PMU 60. When a signal SLP is input from the PU 20, the PMU 60 starts measuring time using the circuit 61. After a predetermined time has elapsed since switching to low power mode, the PMU 60 transitions to power off mode. A control signal from the PMU 60 turns off the PSW 70, cutting off the supply of VDD. In this way, transition from low power mode to power off mode is possible using an interrupt request based on measurement data from the circuit 61. An example of the power management operation of the PMU 60 will be described below with reference to FIG. 6.

[0136] The PU 20 is operating normally. The power supply mode is the power-on mode, and the PMU 60 is in an idle state (step S10). The PMU 60 remains in the idle state until the signal SLP is input. The input of the signal SLP triggers the execution of a save sequence (step S11). In the example of the save sequence shown in FIG. 6, the PMU 60 first outputs a control signal to the clock control circuit 65 to stop the output of the clock signal (step S12). Next, it outputs a control signal to the memory circuit 31 to save data (step S13). In the memory circuit 31, the data held in the circuit MemC1 is saved to the circuit BKC1 in accordance with the control signal from the PMU 60. Finally, the PMU 60 controls the power supply circuit 10 to reduce MVDD. These operations transition the power supply mode to the low power mode (step S14). When the signal SLP is input, the PMU 60 controls the built-in circuit 61 to measure the time Ta in the low power mode (step S15). The timing for operating the circuit 61 can be any timing while the backup sequence is being executed, such as when the signal SLP is input, when a control signal is output to the clock control circuit 65, when data backup is started, when data backup is completed, or when a control signal is output to the power supply circuit 10.

[0137] After the execution of the save sequence, the PMU 60 goes into an idle state (step S16), monitors the input of the signal INT, and monitors the time Ta, which is the measurement time of the clock control circuit 65. When the signal INT is input, the PMU 60 moves to the restore sequence (step S17). When the time Ta reaches the set time T VS The PMU 60 determines whether the time Ta exceeds the time T VS If the time T exceeds the time T, the power supply mode is switched to the power-off mode (step S19), and if the time T does not exceed the time T, the idle state is maintained (step S16). VS The time may be set so that the standby power consumption of the processor core 30 can be reduced more effectively by putting the processor core 30 into the power-off mode than by putting the processor core 30 into the low power mode.

[0138] In step S19, the PMU 60 outputs a control signal to the PSW 70 to cut off the power supply to the processor core 30. After entering the power-off mode, the PMU 60 again enters the idle state (step S20) and monitors the input of the signal INT (step S21). When the signal INT is input, the PMU 60 executes a recovery sequence.

[0139] In the recovery sequence, first, the PMU 60 transitions from the power-off mode to the power-on mode (step S22). The PMU 60 controls the power supply circuit 10 to output the power supply potential for normal operation. The PMU 60 also controls the PSW 70 to resume the supply of VDD to the processor core 30. Next, a control signal is output to the memory circuit 31 to restore the data in the memory circuit 31 (step S23). In response to the control signal from the PMU 60, the memory circuit 31 writes the data held in the circuit BKC1 back to the circuit MemC1. The PMU 60 outputs a control signal to the clock control circuit 65 to output a clock signal (step S24). In response to the control signal from the PMU 60, the clock control circuit 65 resumes outputting the clock signal.

[0140] When the recovery sequence is executed from the determination process of step S17, the system returns from the low power mode to the power-on mode, and the potential of the power line 35 can be stabilized more quickly than when the recovery sequence is executed from the determination process of step S21. Therefore, when transitioning from step S17 to the recovery sequence, the PMU 60 executes step S23 earlier than when transitioning from step S21 to the recovery sequence. This shortens the time it takes for the processor core 30 to return from the sleep state to the normal state.

[0141] 2, when the PU 20 enters a sleep state, the power management of the semiconductor device shown in FIG. 2 first performs a voltage scaling operation to lower the power supply potential supplied to the processor core 30, thereby reducing leakage current and minimizing the time and energy overhead required for returning from the sleep state to the normal state. If the sleep state continues for a certain period of time, a power gating operation is performed to minimize leakage current from the processor core 30. This makes it possible to reduce the power consumption of the PU 20 in the sleep state without reducing the processing performance of the PU 20.

[0142] The processing unit (PU) 20 may also have a cache 40 and a power switch (PSW) 71. The cache 40 is capable of power gating and voltage scaling, and the power mode of the cache 40 changes in conjunction with the power mode of the PU 20. The PSW 71 is a circuit that controls the supply of a power supply potential MVDD to the cache 40 and is controlled by the PMU 60. Here, the power supply potential input to the cache 40 via the PSW 71 is referred to as VDD_M. Similar to the processor core 30, the cache 40 receives a control signal from the PMU 60 and a gated clock signal from the clock control circuit 65.

[0143] <Cache 40> The cache 40 is a storage device that temporarily stores frequently used data. The cache 40 shown in FIG. 7A includes a memory array 41, a peripheral circuit 42, and a control circuit 43. The memory array 41 includes a plurality of memory cells 45. The control circuit 43 controls the operation of the cache 40 in accordance with requests from the processor core 30. For example, it controls the write and read operations of the memory array 41. The peripheral circuit 42 generates signals to drive the memory array 41 in accordance with control signals from the control circuit 43. The memory array 41 includes memory cells 45 that store data.

[0144] As shown in FIG. 7B, the memory cell 45 includes a circuit MemC2 and a circuit BKC2. The circuit MemC2 is a memory cell that is accessed during normal operation. For example, a static random access memory (SRAM) memory cell may be used. The circuit BKC2 functions as a backup circuit for the circuit MemC2 and is capable of retaining data for a long period of time even when the power supply or clock signal is cut off. The provision of such a memory cell 45 enables power gating of the cache 40. Before power is cut off, the memory cell 45 saves data from the circuit MemC2 to BKC2. After power supply is resumed, the data stored in the circuit BKC2 is written back to the circuit MemC2, allowing the PU 20 to quickly return to the state before power was cut off.

[0145] Like circuit BKC1, circuit BKC2 of memory cell 45 also includes a retention circuit having one transistor (MW2) and one capacitor (CB2). That is, circuit BKC2 also has a retention circuit with a configuration similar to that of a standard 1T1C memory cell of a DRAM. Transistor MW2 has an extremely low off-state current. Like transistor MW1, transistor MW2 may be an OS transistor. This configuration allows circuit BKC2 to suppress fluctuations in the potential of node FN2, which is electrically floating, and therefore allows circuit BKC2 to retain data for a long period of time. The data retention time of circuit BKC2 is determined by the leakage current of transistor MW2, the capacitance of capacitor CB2, and other factors. Using transistor MW2 with an extremely low off-state current allows circuit BKC2 to be used as a nonvolatile memory circuit that does not require refresh operations.

[0146] 6, data is saved from the memory circuit 31 and the cache 40. In step S19, the PSW 70 and PSW 71 are controlled to stop the power supply to the processor core 30 and the cache 40. In step S22, the PSW 70 and PSW 71 are controlled to resume the power supply to the processor core 30 and the cache 40. In step S23, the data is restored from the memory circuit 31 and the cache 40.

[0147] <<Processor core configuration example>> An example of the configuration of a processor core is shown in Fig. 8. The processor core 130 shown in Fig. 8 has a control unit 131, a program counter 132, a pipeline register 133, a pipeline register 134, a register file 135, an arithmetic logic unit (ALU) 136, and a data bus 137. Data is exchanged between the processor core 130 and peripheral circuits such as a PMU and a cache via the data bus 137.

[0148] The control device 131 has the function of decoding and executing instructions included in an input program such as an application by comprehensively controlling the operations of the program counter 132, pipeline register 133, pipeline register 134, register file 135, ALU 136, and data bus 137. The ALU 136 has the function of performing various arithmetic operations such as arithmetic operations and logical operations. The program counter 132 is a register that has the function of storing the address of the next instruction to be executed.

[0149] The pipeline register 133 is a register that has the function of temporarily storing instruction data. The register file 135 has multiple registers including general-purpose registers, and can store data read from the main memory, data obtained as a result of the arithmetic processing of the ALU 136, etc. The pipeline register 134 is a register that has the function of temporarily storing data used in the arithmetic processing of the ALU 136, data obtained by the arithmetic processing of the ALU 136, etc.

[0150] The storage circuit 31 in FIG. 4B is used for a register included in the processor core 130.

[0151] <Example of memory circuit configuration> A more specific configuration example of the memory circuit 31 shown in Fig. 4B will be described. Fig. 9 is a circuit diagram showing an example of the configuration of the memory circuit. The memory circuit 100 shown in Fig. 9 functions as a flip-flop circuit. In Fig. 9, the input signal of the memory circuit 100 is denoted as signal D, and the output signal is denoted as signal Q.

[0152] A standard flip-flop circuit (FF) can be applied to the circuit MemC1, such as a master-slave FF. An example of such a configuration is shown in FIG. 9. The FF 110 includes transmission gates (TG1, TG2, TG3, TG4, TG5), inverter circuits (INV1, INV2, INV3, INV4), and NAND circuits (NAND1, NAND2). The signals RESET and OSR are control signals output from the PMU 60. The signal OSR and its inverted signal are input to TG5. The clock signal CLK and its inverted signal are input to TG1-TG4. A single clocked inverter circuit may be provided instead of TG1 and INV1. A single clocked NAND circuit may be provided instead of TG2 and NAND2. A clocked inverter circuit may be provided instead of TG3 and INV3. TG5 functions as a switch that controls the conduction state between the output node of NAND1 and node NR1. The node NB1 is electrically connected to the input node of the circuit BKC10, and the node NR1 is electrically connected to the output node of the circuit BKC10.

[0153] The circuit BKC10 shown in FIG. 9 functions as a backup circuit for the FF110. The circuit BKC10 includes a circuit RTC10 and a circuit PCC10. The signals (OSG, OSC, OSR) input to the circuit BKC10 are control signals output from the PMU60. The power supply potential VSS is a low power supply potential, which may be, for example, the ground potential (GND) or 0V. The power supply potential VSS and the power supply potential VDD are input to the FF110 as well, similar to the BKC10. In the memory circuit 100, the supply of VDD is managed by the PMU60.

[0154] The circuit RTC10 includes transistors MW1, MA1, MR1, a capacitor CB1, and nodes FN1 and NK1. The circuit RTC10 has the function of retaining data, and is configured here as a memory circuit with a 3T-type gain cell structure. The transistor MW1 is a write transistor and an OS transistor. The transistor MR1 is a read transistor, and the transistor MA1 is an amplification transistor and a read transistor. Data is retained at node FN1. The node NK1 is a data input node. The node NR1 is a data output node of the circuit RTC10.

[0155] FIG. 9 shows an example of a configuration in which the circuit BKC10 reads data from the slave latch circuit of the FF110 during the save operation, and writes the held data back to the master latch circuit during the restore operation. The data to be saved may be data from the master latch circuit. Alternatively, the data may be restored to the slave latch circuit. In this case, TG5 may be provided in the slave latch circuit.

[0156] Furthermore, transistors MR1 ​​and MA1 of circuit RTC10 may be either n-type or p-type, and the potential of signal OSR and the level of the power supply potential supplied to transistor MA1 can be changed depending on the conductivity type of transistors MR1 ​​and MA1. The logic circuit of FF110 can also be configured appropriately. For example, if transistors MR1 ​​and MA1 are p-type transistors, NAND1 and INV3 can be swapped in the master latch circuit, and INV2 and NAND2 can be swapped in the slave latch circuit. Also, VDD can be input to transistor MA1 instead of VSS.

[0157] Because circuit BKC10 writes data using voltage, it can reduce the write power consumption compared to MRAM, which writes data using current. Also, because data is held by the load capacitance of node FN1, there is no limit to the number of times data can be rewritten, as there is in flash memory.

[0158] In the circuit RTC10, the energy required to write data corresponds to the energy required to charge and discharge the capacitance element CB1. On the other hand, in a memory circuit using a two-terminal memory element such as MRAM, the energy required to write data corresponds to the energy consumed when a current flows through the memory element. Therefore, compared to a case where an MRAM or the like is used, in which a current continues to flow during the data write period, the circuit BKC10 can reduce the energy consumed by saving data. Therefore, by providing the circuit BKC10 in the backup circuit, the break-even time (BET) can be shortened compared to a case where an MRAM is provided. As a result, there are more opportunities to perform power gating, which can reduce energy consumption, and the power consumption of the semiconductor device can be reduced.

[0159] The circuit PCC10 includes a transistor MC1 and a transistor MC2. The circuit PCC10 has a function of precharging the node FN1. The circuit PCC10 does not necessarily have to be provided. As will be described later, providing the circuit PCC10 can shorten the data backup time of the circuit BKC10.

[0160] <Memory circuit operation example> FIG. 10 is a timing chart showing an example of the operation of the memory circuit 100, illustrating the waveforms of the control signals (signal SLP, signal RESET, clock signal CLK, signal OSG, signal OSC, signal OSR) and the changes in the power supply potential VDD and the potentials of node FN1 and node NR1.

[0161] [Normal operation] A normal operation period will be described. A power supply potential VDD and a signal CLK are supplied to the memory circuit 100. The FF110 functions as a sequential circuit. Since the signal RESET is maintained at a high level, the NAND1 and NAND2 function as inverter circuits. In the circuit BKC1, the transistor MC1 is in an off state, and the transistors MC2 and MW1 are in an on state, so the potential of the node FN1 is precharged to a high level.

[0162] [Data evacuation] Next, the backup period will be described. First, the clock signal CLK is stopped. This stops rewriting data at node NB1. In the example of FIG. 10, the potential level of node NB1 is low ("0") when the potential of node NR1 is high ("1"), and is high ("1") when the potential is low ("0"). While the signal OSC is high, the data at node NB1 is saved to node FN1. Specifically, since transistors MC1 and MW1 are on, node FN1 and node NB1 are electrically connected. By setting the signal OSG to low and turning off transistor MW1, node FN1 becomes electrically floating, and circuit BKC10 becomes a data retention state. The potential of node FN1 is high when node NR1 is low ("0"), and is low when it is high ("1").

[0163] Since the data backup ends when signal OSG is set to low, the voltage scaling operation of PU20 can be performed immediately after signal OSG is set to low. Also, because transistor MC2 precharges node FN1 to high during normal operation, the data backup operation of setting node FN1 to high does not involve the transfer of charge from node FN1. This allows circuit BKC10 to complete the backup operation in a short time.

[0164] In the data saving operation, it is sufficient that the clock signal CLK is inactive. In the example of FIG. 10, the potential of the clock signal CLK is set to a low level, but it may also be set to a high level.

[0165] [Voltage scaling, low power mode] Next, the low power period will be described. In conjunction with the falling edge of the signal OSC, the PMU 60 performs a voltage scaling operation, which causes the storage circuit 100 to transition to a low power mode.

[0166] [Power gating, power off mode] Next, the power-off period will be described. After a certain period has elapsed since the transition to the low power mode, the PMU 60 performs a power gating operation to put the memory circuit 100 into the power-off mode.

[0167] [Power on mode] Next, the power-on period will be described. In response to an interrupt request, the PMU 60 returns the memory circuit 100 to the power-on mode. In the example of Fig. 10, when the potential of the power supply line that supplies VDD becomes stable, the clock signal CLK becomes high level.

[0168] [Data recovery] A data restoration operation is performed while the signal OSR is at a high level. By setting the signal RESET to a high level, the potential of the node NR1 is precharged to a high level ("1"). By setting the signal OSR to a high level, TG5 enters a high impedance state and the transistor MR1 enters a conductive state. The conductive state of the transistor MA1 is determined by the potential of the node FN1. If the node FN1 is at a high level, the transistor MA1 is in a conductive state, so the potential of the node NR1 drops to a low level ("0"). If the node FN1 is at a low level, the potential of the node NR1 is maintained at a high level. In other words, the state of FF110 is restored to the state before the transition to the sleep state.

[0169] As described above, the rising edges of the signals RESET and OSR allow high-level data to be written back (restored) to the node NR1, thereby enabling the storage circuit 100 to shorten the restoration operation period.

[0170] Figure 10 shows an example of returning from power-off mode to power-on mode. When returning from low-power mode to power-on mode, the period Ton until the potential of the power line supplying VDD stabilizes is shorter. In this case, it is better to make the rise of the signal OSR earlier than when returning from power-off mode. In Figure 10, the period from the start of the backup period to before the start of the next normal operation period is referred to as the sleep period.

[0171] [Normal operation] Next, the normal operation period will be described. By resuming the supply of the clock signal CLK, the state where normal operation is possible is restored. By setting the signal OSG to high level, the node FN1 is precharged by the circuit PCC10 and becomes high level.

[0172] <<Cache>> An example in which the cache 40 is configured using an SRAM will be described below.

[0173] <Memory cell configuration example> FIG. 11 shows an example of the configuration of a cache memory cell. The memory cell 120 shown in FIG. 11 has a circuit SMC20 and a circuit BKC20. The circuit SMC20 may have a circuit configuration similar to that of a standard SRAM memory cell. The circuit SMC20 shown in FIG. 11 has an inverter circuit INV11, an inverter circuit INV12, a transistor M11, and a transistor M12.

[0174] Circuit BKC20 functions as a backup circuit for circuit SMC20. Circuit BKC20 has transistors MW11, MW12, and capacitors CB11 and CB12. Transistors MW11 and MW12 are OS transistors. Circuit BKC20 has two 1T1C type holding circuits, and data is held at nodes SN1 and SN2, respectively. The holding circuit consisting of transistor MW11 and capacitor CB11 has the function of backing up data at node NET1. The holding circuit consisting of transistor MW12 and capacitor CB12 has the function of backing up data at node NET2.

[0175] The memory cell 120 is supplied with power supply potentials VDDMC and VSS. The memory cell 120 is electrically connected to the wirings (WL, BL, BLB, BRL). A signal SLC is input to the wiring WL. When writing data, a data signal D and a data signal DB are input to the wirings BL and BLB, respectively. Data is read by detecting the potentials of the wirings BL and BLB. A signal OSS is input to the wiring BRL. The signal OSS is a signal input from the PMU 60.

[0176] <Memory cell operation example> An example of the operation of the memory cell 120 will be described below.

[0177] [Normal operation] An access request is made to the circuit MemC2, and data is written or read. In the circuit BKC2, the signal OSS is at a low level, so the nodes SN1 and SN2 are in an electrically floating state and are in a data retention state. In the example of FIG. 12, the potential of the node SN1 is at a low level ("0"), and the potential of the other node, node SN2, is at a high level ("1").

[0178] [Data evacuation] When the signal OSS goes high, the transistors MW11 and MW12 are turned on, and the nodes SN1 and SN2 are at the same potential level as the nodes NET1 and NET2, respectively. In the example of Figure 12, the potentials of the nodes SN1 and SN2 are high and low, respectively. When the signal OSS goes low, the circuit BKC20 enters the data retention state, and the data save operation is completed.

[0179] [Voltage scaling, low power mode] In response to the falling edge of the signal OSS, the PMU 60 performs a voltage scaling operation, which causes the cache 40 to transition to a low power mode.

[0180] [Power gating, power off mode] After a certain period of time has elapsed since the transition to the low power mode, the PMU 60 performs a power gating operation to put the cache 40 into the power-off mode.

[0181] [Data recovery, power on mode] In response to the interrupt request, the PMU 60 restores the cache 40 to its normal state. The signal OSS is set to high level, and the data held in the circuit BKC 20 is written back to the circuit SMC 20. While the signal OSS is high, the PMU 60 performs voltage scaling and power gating operations to restore the memory circuit 100 to power-on mode. In the example of FIG. 10, the clock signal CLK is set to high level when the potential of the power supply line supplying VDD stabilizes. Once the potential of the power supply line supplying VDDMC stabilizes, the signal OSS is set to low level, and the data restoration operation is completed. The states of the nodes SN1 and SN2 are restored to the state they were in immediately before entering the hibernation state. In FIG. 12, the period from the start of the backup period to the start of the next normal operation period is referred to as the sleep period.

[0182] [Normal operation] When the supply of VDDMC is resumed, the circuit SMC20 returns to the normal mode in which normal operation is possible.

[0183] As described above, by using OS transistors, it is possible to configure a backup circuit that can retain data for a long period of time even when power is cut off. The inclusion of this backup circuit enables power gating of the processor core and cache. Furthermore, by combining voltage scaling and power gating in power management during hibernation, it is possible to reduce the energy and time overhead required for returning from hibernation to normal operation. This allows for efficient power reduction without reducing the processing power of the processing unit.

[0184] <Example of memory> A memory including an OS transistor of one embodiment of the present invention will be described below.

[0185] The power storage device of one embodiment of the present invention preferably includes a memory. A memory device including an OS transistor can be used as the memory. For example, NOSRAM (registered trademark), DOSRAM (registered trademark), or the like, which will be described below, can be used.

[0186] NOSRAM is a gain cell type DRAM in which the write transistor of the memory cell is composed of an OS transistor. NOSRAM is an abbreviation for Nonvolatile Oxide Semiconductor RAM. An example of a NOSRAM configuration is shown below.

[0187] 13A is a block diagram showing an example of the configuration of a NOSRAM. The NOSRAM 220 is provided with power domains 212 and 213, and power switches 215 to 217. The power domain 212 is provided with a memory cell array 230, and the power domain 213 is provided with peripheral circuits for the NOSRAM 220. The peripheral circuits include a control circuit 231, a row circuit 232, and a column circuit 233.

[0188] Voltages VDDD, VSSS, VDHW, VDHR, VBG2, a clock signal GCLK2, an address signal Address, and signals CE, WE, and PSE5 are input to the NOSRAM 220 from outside. The signals CE and WE are a chip enable signal and a write enable signal. The signal PSE5 controls the on / off of power switches 215 to 217. The power switches 215 to 217 control the input of voltages VDDD, VDHW, and VDHR to the power domain 213, respectively.

[0189] The voltages, signals, etc. input to the NOSRAM 220 are appropriately selected depending on the circuit configuration and operation method of the NOSRAM 220. For example, a power domain that is not power-gated may be provided in the NOSRAM 220, and a power gating control circuit that generates the signal PSE5 may be provided.

[0190] The memory cell array 230 includes memory cells 11, write word lines WWL, read word lines RWL, write bit lines WBL, read bit lines RBL, and source lines SL.

[0191] As shown in FIG. 13B, the memory cell 11 is a 2T1C (two transistors, one capacitor) type gain cell and includes a node SN1, transistors M1 and M2, and a capacitor C1. The transistor M1 is a write transistor and is an OS transistor having a back gate. The back gate of the transistor M1 is electrically connected to a wiring BGL2 that supplies a voltage VBG2. The transistor M2 is a read transistor and is a p-channel Si transistor. The capacitor C1 is a storage capacitor that holds the voltage of the node SN1.

[0192] The voltages VDDD and VSSS represent data "1" and "0." The high level voltages of the write word line WWL and the read word line RWL are voltages VDHW and VHDR.

[0193] Fig. 14A shows an example of the configuration of the memory cell array 230. In the memory cell array 230 shown in Fig. 14A, one source line is supplied to two adjacent rows.

[0194] In principle, the memory cell 11 has no limit to the number of times it can be rewritten, data can be rewritten with low energy, and no power is consumed to retain data. Because the transistor M1 is an OS transistor with extremely low off-state current, the memory cell 11 can retain data for a long time.

[0195] The circuit configuration of the memory cell 11 is not limited to that shown in FIG. 13B. For example, the read transistor M2 may be an OS transistor with a back gate or an n-channel Si transistor. Alternatively, the memory cell 11 may be a 3T gain cell. For example, FIGS. 14B and 14C show examples of a 3T gain cell. The memory cell 15 shown in FIG. 14B includes transistors M3 to M5, a capacitor C3, and a node SN3. The transistors M3 to M5 are write transistors, read transistors, and select transistors. The transistor M3 is an OS transistor with a back gate, and the transistors M4 and M5 are p-channel Si transistors. The transistors M4 and M5 may be n-channel Si transistors or OS transistors with back gates. In the memory cell 16 shown in FIG. 14C, the three transistors are OS transistors with back gates.

[0196] Node SN3 is a storage node. Capacitor C3 is a storage capacitor for storing the voltage at node SN3. Capacitor C3 may not be intentionally provided, and the storage capacitor may be formed using the gate capacitance of transistor M4 or the like. Wiring PDL is a wiring that replaces source line SL, and a fixed voltage, for example, voltage VDDD, is input to it.

[0197] The control circuit 231 has a function of controlling the overall operation of the NOSRAM 220. For example, the control circuit 231 performs a logical operation on the signals CE and WE to determine whether an external access is a write access or a read access.

[0198] The row circuit 232 has a function of selecting the write word line WWL and read word line RWL of the selected row specified by the address signal, and the column circuit 233 has a function of writing data to the write bit line WBL of the column specified by the address signal, and a function of reading data from the read bit line RBL of the column.

[0199] DOSRAM is a RAM having 1T1C type memory cells, and is an abbreviation for Dynamic Oxide Semiconductor RAM. DOSRAM will be described below with reference to FIG.

[0200] As shown in FIG. 15A, the memory cell 16 of the DOSRAM 341 is electrically connected to a bit line BL1 (or BLB1), a word line WL1, and wirings BGL6 and PL. The bit line BLB1 is an inverted bit line. For example, voltages VBG6 and VSSS are input to the wirings BGL6 and PL. The memory cell 16 includes a transistor M6 and a capacitance element C6. The transistor M6 is an OS transistor having a back gate.

[0201] Because data is rewritten by charging and discharging the capacitive element C6, the DOSRAM 341 is theoretically unlimited in the number of rewrites and can write and read data with low energy. Furthermore, the simple circuit configuration of the memory cell 16 makes it easy to increase the capacity. Because the write transistors of the memory cell 16 are OS transistors, the retention time of the DOSRAM 341 is much longer than that of DRAM. Therefore, the frequency of refresh operations can be reduced, or even the refresh operation can be eliminated, thereby reducing the power required for the refresh operation.

[0202] 15B, in the DOSRAM 341, the memory cell array 361 can be stacked on the peripheral circuit 365. This is because the transistor M6 of the memory cell 16 is an OS transistor.

[0203] The memory cell array 361 has a plurality of memory cells 16 arranged in a matrix, and bit lines BL1, BLB1, word lines WL1, and wirings BGL6 and PL are provided according to the arrangement of the memory cells 16. The peripheral circuit 365 is provided with a control circuit, a row circuit, and a column circuit. The row circuit selects the word line WL to be accessed, etc. The column circuit writes and reads data to and from the bit line pair consisting of BL and BLB.

[0204] Power switches 371 and 373 are provided to power gate the peripheral circuit 365. The power switches 371 and 373 respectively control the input of voltages VDDD and VDHW6 to the peripheral circuit 365. The voltage VDHW6 is a high-level voltage of the word line WL1. The on / off of the power switches 371 and 373 is controlled by a signal PSE6.

[0205] <Example of an arithmetic circuit> Next, an example of the configuration of a semiconductor device that can be used for neural network calculations will be described.

[0206] As shown in FIG. 16A, the neural network NN can be configured with an input layer ILy, an output layer OLy, and an intermediate layer (hidden layer) HLy. The input layer ILy, output layer OLy, and intermediate layer HLy each have one or more neurons (units). The intermediate layer HLy may have one layer or two or more layers. A neural network having two or more intermediate layers HLy can also be called a DNN (deep neural network), and learning using a deep neural network can also be called deep learning.

[0207] Input data is input to each neuron in the input layer ILy, an output signal from a neuron in the previous or next layer is input to each neuron in the hidden layer HLy, and an output signal from a neuron in the previous layer is input to each neuron in the output layer OLy. Note that each neuron may be connected to all neurons in the previous or next layer (fully connected), or may be connected to only a portion of the neurons in the previous or next layer.

[0208] Figure 16B shows an example of a neuron's operation. It shows neuron N and two neurons in the previous layer that output signals to neuron N. Neuron N receives the output x1 of a neuron in the previous layer and the output x2 of a neuron in the previous layer. Neuron N then multiplies the output x1 by a weight w1 (x1w1) and the output x2 by a weight w2 (x2w2), calculating the sum x1w1+x2w2. After this, a bias b is added as necessary, resulting in a value a = x1w1+x2w2+b. The value a is then transformed by the activation function h, and neuron N outputs an output signal y = h(a).

[0209] As described above, the computation performed by a neuron includes the sum of the product of the output of a neuron in the previous layer and the weight, i.e., the sum-of-products computation (x1w1+x2w2 as described above). This sum-of-products computation may be performed in software using a program, or may be performed by hardware. When performing the sum-of-products computation by hardware, a sum-of-products computation circuit can be used. This sum-of-products computation circuit may be a digital circuit or an analog circuit. When an analog circuit is used for the sum-of-products computation circuit, it is possible to reduce the circuit size of the sum-of-products computation circuit or the number of memory accesses, thereby improving processing speed and reducing power consumption.

[0210] The product-sum operation circuit may be configured using transistors containing silicon (such as single crystal silicon) in their channel formation regions (hereinafter also referred to as Si transistors) or transistors containing an oxide semiconductor in their channel formation regions (hereinafter also referred to as OS transistors). OS transistors, in particular, have extremely low off-state current and are therefore suitable as transistors constituting the memory of the product-sum operation circuit. The product-sum operation circuit may be configured using both Si transistors and OS transistors. Hereinafter, a configuration example of a semiconductor device having the function of a product-sum operation circuit will be described.

[0211] 17 shows a configuration example of a semiconductor device MAC having a function of performing neural network calculations. The semiconductor device MAC has a function of performing a product-sum operation on first data corresponding to the connection strength (weight) between neurons and second data corresponding to input data. Note that the first data and second data can each be analog data or multi-valued digital data (discrete data). The semiconductor device MAC also has a function of converting the data obtained by the product-sum operation using an activation function.

[0212] The semiconductor device MAC includes a cell array CA, a current source circuit CS, a current mirror circuit CM, a circuit WDD, a circuit WLD, a circuit CLD, an offset circuit OFST, and an activation function circuit ACTV.

[0213] The cell array CA has a plurality of memory cells MC and a plurality of memory cells MCref. FIG. 17 shows an example of a configuration in which the cell array CA has m rows and n columns (m and n are integers equal to or greater than 1) of memory cells MC (MC[1,1] to [m,n]) and m memory cells MCref (MCref[1] to [m]). The memory cells MC have a function of storing first data. The memory cells MCref also have a function of storing reference data used in a product-sum operation. The reference data can be analog data or multi-level digital data.

[0214] The memory cell MC[i,j] (i is an integer between 1 and m, and j is an integer between 1 and n) is connected to the wiring WL[i], the wiring RW[i], the wiring WD[j], and the wiring BL[j]. The memory cell MCref[i] is connected to the wiring WL[i], the wiring RW[i], the wiring WDref, and the wiring BLref. Here, the current flowing between the memory cell MC[i,j] and the wiring BL[j] is I MC[i,j] and the current flowing between the memory cell MCref[i] and the wiring BLref is I MCref[i] It is written as follows.

[0215] A specific configuration example of the memory cell MC and memory cell MCref is shown in Figure 18. Although Figure 18 shows memory cells MC[1,1], [2,1] and memory cells MCref[1], [2] as representative examples, similar configurations can be used for other memory cells MC and memory cells MCref. The memory cell MC and memory cell MCref each have transistors Tr11, Tr12, and a capacitance element C11. Here, a case will be described in which transistors Tr11 and Tr12 are n-channel transistors.

[0216] In the memory cell MC, the gate of the transistor Tr11 is connected to the wiring WL, one of the source and the drain is connected to the gate of the transistor Tr12 and the first electrode of the capacitor C11, and the other of the source and the drain is connected to the wiring WD. One of the source and the drain of the transistor Tr12 is connected to the wiring BL, and the other of the source and the drain is connected to the wiring VR. The second electrode of the capacitor C11 is connected to the wiring RW. The wiring VR is a wiring that has the function of supplying a predetermined potential. Here, as an example, a case where a low power supply potential (such as a ground potential) is supplied from the wiring VR will be described.

[0217] A node connected to one of the source or drain of the transistor Tr11, the gate of the transistor Tr12, and the first electrode of the capacitive element C11 is referred to as a node NM. The nodes NM of the memory cells MC[1,1] and MC[2,1] are referred to as nodes NM[1,1] and NM[2,1], respectively.

[0218] The memory cell MCref has the same configuration as the memory cell MC. However, the memory cell MCref is connected to a wiring WDref instead of the wiring WD, and to a wiring BLref instead of the wiring BL. In addition, in the memory cells MCref[1] and [2], the nodes connected to one of the source or drain of the transistor Tr11, the gate of the transistor Tr12, and the first electrode of the capacitance element C11 are denoted as nodes NMref[1] and [2], respectively.

[0219] The node NM and the node NMref function as the storage nodes of the memory cell MC and the memory cell MCref, respectively. The node NM stores first data, and the node NMref stores reference data. In addition, a current I flows from the wiring BL[1] to the transistors Tr12 of the memory cells MC[1,1] and MC[2,1]. MC[1,1] , I MC[2,1] In addition, a current I flows from the wiring BLref to the transistor Tr12 of the memory cells MCref[1] and MCref[2]. MCref[1] , I MCref[2] is playing.

[0220] Since the transistor Tr11 has a function of maintaining the potential of the node NM or the node NMref, it is preferable that the off-state current of the transistor Tr11 be small. Therefore, it is preferable to use an OS transistor with an extremely small off-state current as the transistor Tr11. This can suppress fluctuations in the potential of the node NM or the node NMref, thereby improving the accuracy of calculation. Furthermore, it is possible to reduce the frequency of operations to refresh the potential of the node NM or the node NMref, thereby reducing power consumption.

[0221] The transistor Tr12 is not particularly limited, and may be, for example, a Si transistor or an OS transistor. When an OS transistor is used as the transistor Tr12, the transistor Tr12 can be manufactured using the same manufacturing equipment as that of the transistor Tr11, thereby reducing manufacturing costs. The transistor Tr12 may be either an n-channel type or a p-channel type.

[0222] The current source circuit CS is connected to the wirings BL[1] to [n] and the wiring BLref. The current source circuit CS has a function of supplying current to the wirings BL[1] to [n] and the wiring BLref. Note that the current value supplied to the wirings BL[1] to [n] may be different from the current value supplied to the wiring BLref. Here, the current supplied from the current source circuit CS to the wirings BL[1] to [n] is referred to as I C, the current supplied from the current source circuit CS to the wiring BLref is I Cref It is written as follows.

[0223] The current mirror circuit CM has wiring IL[1] to [n] and wiring ILref. The wiring IL[1] to [n] are connected to wiring BL[1] to [n], respectively, and the wiring ILref is connected to wiring BLref. Here, the connection points of the wiring IL[1] to [n] and the wiring BL[1] to [n] are denoted as nodes NP[1] to [n]. Also, the connection point of the wiring ILref and the wiring BLref is denoted as node NPref.

[0224] The current mirror circuit CM generates a current I according to the potential of the node NPref. CM The function of flowing this current I CM 17 shows the function of flowing current I from the wiring BLref to the wiring ILref. CM is discharged, and a current I flows from the wiring BL[1] to [n] to the wiring IL[1] to [n]. CM In addition, the current flowing from the current mirror circuit CM to the cell array CA via the wiring BL[1] to [n] is expressed as I B The current flowing from the current mirror circuit CM to the cell array CA via the wiring BLref is represented as I Bref It is written as follows.

[0225] The circuit WDD is connected to the wirings WD[1] to [n] and the wiring WDref. The circuit WDD has a function of supplying a potential corresponding to first data stored in the memory cell MC to the wirings WD[1] to [n]. The circuit WDD also has a function of supplying a potential corresponding to reference data stored in the memory cell MCref to the wiring WDref. The circuit WLD is connected to the wirings WL[1] to [m]. The circuit WLD has a function of supplying a signal for selecting the memory cell MC or the memory cell MCref to which data is to be written to the wirings WL[1] to [m]. The circuit CLD is connected to the wirings RW[1] to [m]. The circuit CLD has a function of supplying a potential corresponding to second data to the wirings RW[1] to [m].

[0226] The offset circuit OFST is connected to the wirings BL[1] to [n] and the wirings OL[1] to [n]. The offset circuit OFST has a function of detecting the amount of current flowing from the wirings BL[1] to [n] to the offset circuit OFST and / or the amount of change in the current flowing from the wirings BL[1] to [n] to the offset circuit OFST. The offset circuit OFST also has a function of outputting the detection result to the wirings OL[1] to [n]. Note that the offset circuit OFST may output a current corresponding to the detection result to the wiring OL, or may convert the current corresponding to the detection result into a voltage and output it to the wiring OL. The current flowing between the cell array CA and the offset circuit OFST is expressed as I α These are written as [1] to [n].

[0227] An example configuration of the offset circuit OFST is shown in FIG. 19. The offset circuit OFST shown in FIG. 19 includes circuits OC[1] to OC[n]. Each of the circuits OC[1] to OC[n] includes a transistor Tr21, a transistor Tr22, a transistor Tr23, a capacitance element C21, and a resistance element R1. The connection relationship of each element is as shown in FIG. 19. The node connected to the first electrode of the capacitance element C21 and the first terminal of the resistance element R1 is referred to as node Na. The node connected to the second electrode of the capacitance element C21, one of the source or drain of the transistor Tr21, and the gate of the transistor Tr22 is referred to as node Nb.

[0228] The wiring VrefL has a function of supplying a potential Vref, the wiring VaL has a function of supplying a potential Va, and the wiring VbL has a function of supplying a potential Vb. The wiring VDDL has a function of supplying a potential VDD, and the wiring VSSL has a function of supplying a potential VSS. Here, the case where the potential VDD is a high power supply potential and the potential VSS is a low power supply potential will be described. The wiring RST has a function of supplying a potential for controlling the conduction state of the transistor Tr21. The transistor Tr22, the transistor Tr23, the wiring VDDL, the wiring VSSL, and the wiring VbL form a source follower circuit.

[0229] Next, an example of the operation of the circuits OC[1] to [n] will be described. Note that, although an example of the operation of the circuit OC[1] will be described here as a representative example, the circuits OC[2] to [n] can also be operated in the same manner. First, when a first current flows through the wiring BL[1], the potential of the node Na becomes a potential corresponding to the first current and the resistance value of the resistor R1. At this time, the transistor Tr21 is in an on state, and a potential Va is supplied to the node Nb. After that, the transistor Tr21 is turned off.

[0230] Next, when a second current flows through the wiring BL[1], the potential of the node Na changes to a potential corresponding to the second current and the resistance value of the resistor R1. At this time, the transistor Tr21 is in an off state and the node Nb is in a floating state, so the potential of the node Nb changes due to capacitive coupling as the potential of the node Na changes. Here, the change in the potential of the node Na is defined as ΔV Na If the capacitance coupling coefficient is 1, the potential of node Nb is Va+ΔV Na Then, the threshold voltage of the transistor Tr22 is V th Then, the potential from the wiring OL[1] is Va+ΔV Na -V th is output, where Va=V th By doing so, the potential ΔV Na can be output.

[0231] Potential ΔV Na is determined according to the change amount from the first current to the second current, the resistance value of the resistor element R1, and the potential Vref. Here, since the resistance value of the resistor element R1 and the potential Vref are known, the potential ΔV Na From this, the amount of change in the current flowing through the wiring BL can be obtained.

[0232] A signal corresponding to the amount of current and / or the amount of change in current detected by the offset circuit OFST as described above is input to the activation function circuit ACTV via the lines OL[1] to OL[n].

[0233] The activation function circuit ACTV is connected to the wirings OL[1] to [n] and NIL[1] to [n]. The activation function circuit ACTV has a function of performing calculations to convert the signal input from the offset circuit OFST according to a predefined activation function. Examples of the activation function that can be used include a sigmoid function, a tanh function, a softmax function, a ReLU function, and a threshold function. The signal converted by the activation function circuit ACTV is output as output data to the wirings NIL[1] to [n].

[0234] Using the semiconductor device MAC described above, it is possible to perform a product-sum operation on the first data and the second data. An example of the operation of the semiconductor device MAC when performing a product-sum operation will be described below.

[0235] 20 shows a timing chart of an operation example of the semiconductor device MAC. FIG. 20 shows the transition of the potentials of the wiring WL[1], wiring WL[2], wiring WD[1], wiring WDref, node NM[1,1], node NM[2,1], node NMref[1], node NMref[2], wiring RW[1], and wiring RW[2] in FIG. 18, and the current I B [1]-I α [1], and current I Bref The graph shows the transition of the value of the current I B [1]-I α [1] corresponds to the sum of the currents flowing from the wiring BL[1] to the memory cells MC[1,1] and [2,1].

[0236] Here, the operation will be explained focusing on memory cells MC[1,1], [2,1] and memory cells MCref[1], [2] shown in Figure 18 as representative examples, but other memory cells MC and memory cells MCref can also be operated in the same way.

[0237] First, at time T01-T02, the potential of the wiring WL[1] becomes high level (High), and the potential of the wiring WD[1] becomes V higher than the ground potential (GND). PR -V W[1,1] The potential of the wiring WDref becomes V higher than the ground potential. PR The potential of the wiring RW[1] and the wiring RW[2] becomes the reference potential (REFP). W[1,1] is a potential corresponding to the first data stored in the memory cell MC[1,1]. PR is a potential corresponding to the reference data. As a result, the transistors Tr11 included in the memory cells MC[1,1] and MCref[1] are turned on, and the potential of the node NM[2,1] becomes V PR -V W[1,1] , the potential of node NMref[2] is V PR This becomes:

[0238] At this time, a current I flows from the wiring BL[1] to the transistor Tr12 of the memory cell MC[1,1]. MC[1,1],0 can be expressed by the following equation: where k is a constant determined by the channel length, channel width, mobility, and capacitance of the gate insulating film of the transistor Tr12. th is the threshold voltage of transistor Tr12.

[0239] I MC[1,1],0 =k(V PR -V W[1,1] -V th ) 2 (E1)

[0240] In addition, the current I flows from the wiring BLref to the transistor Tr12 of the memory cell MCref[1]. MCref[1],0 can be expressed by the following formula:

[0241] I MCref[1],0 =k(V PR -V th ) 2 (E2)

[0242] Next, at time T02-T03, the potential of the line WL[1] becomes low, which turns off the transistors Tr11 of the memory cells MC[1,1] and MCref[1], and maintains the potentials of the nodes NM[1,1] and NMref[1].

[0243] As described above, it is preferable to use an OS transistor as the transistor Tr11, which can suppress leakage current of the transistor Tr11 and accurately maintain the potentials of the nodes NM[1,1] and NMref[1].

[0244] Next, at time T03-T04, the potential of the wiring WL[2] becomes high level, and the potential of the wiring WD[1] becomes V higher than the ground potential. PR -V W[2,1]The potential of the wiring WDref becomes V higher than the ground potential. PR The potential V W[2,1] is a potential corresponding to the first data stored in the memory cell MC[2,1]. As a result, the transistor Tr11 included in the memory cell MC[2,1] and the memory cell MCref[2] is turned on, and the potential of the node NM[2,1] becomes V PR -V W[2,1] , the potential of node NMref[2] is V PR This becomes:

[0245] At this time, a current I flows from the wiring BL[1] to the transistor Tr12 of the memory cell MC[2,1]. MC[2,1],0 can be expressed by the following formula:

[0246] I MC[2,1],0 =k(V PR -V W[2,1] -V th ) 2 (E3)

[0247] In addition, the current I flows from the wiring BLref to the transistor Tr12 of the memory cell MCref[2]. MCref[2],0 can be expressed by the following formula:

[0248] I MCref[2],0 =k(V PR -V th ) 2 (E4)

[0249] Next, at time T04-T05, the potential of the line WL[2] becomes low, which turns off the transistors Tr11 of the memory cells MC[2,1] and MCref[2], and maintains the potentials of the nodes NM[2,1] and NMref[2].

[0250] Through the above operations, the first data is stored in the memory cells MC[1,1] and [2,1], and the reference data is stored in the memory cells MCref[1] and [2].

[0251] Here, consider the current flowing through the wiring BL[1] and the wiring BLref at time T04-T05. A current is supplied to the wiring BLref from the current source circuit CS. The current flowing through the wiring BLref is discharged to the current mirror circuit CM and the memory cells MCref[1] and [2]. The current supplied from the current source circuit CS to the wiring BLref is defined as I Cref , the current flowing from the wiring BLref to the current mirror circuit CM is I CM,0 Then, the following formula holds:

[0252] I Cref -I CM,0 =I MCref[1],0 +I MCref[2],0 (E5)

[0253] A current is supplied to the wiring BL[1] from the current source circuit CS. The current flowing through the wiring BL[1] is discharged to the current mirror circuit CM and the memory cells MC[1,1], [2,1]. A current also flows from the wiring BL[1] to the offset circuit OFST. The current supplied from the current source circuit CS to the wiring BL[1] is referred to as I C,0 , the current flowing from the wiring BL[1] to the offset circuit OFST is I α,0 Then, the following formula holds:

[0254] I C -I CM,0 =I MC[1,1],0 +I MC[2,1],0 +I α,0 (E6)

[0255] Next, at time T05-T06, the potential of the wiring RW[1] becomes V X[1] At this time, the potential V X[1] is supplied, and the potential of the gate of the transistor Tr12 rises due to capacitive coupling. X[1] is a potential corresponding to the second data supplied to the memory cell MC[1,1] and the memory cell MCref[1].

[0256] The change in the potential of the gate of transistor Tr12 is calculated by multiplying the change in the potential of the wiring RW by a capacitance coupling coefficient determined by the configuration of the memory cell. The capacitance coupling coefficient is calculated based on the capacitance of the capacitance element C11, the gate capacitance of transistor Tr12, parasitic capacitance, etc. For convenience, the following description will be given assuming that the change in the potential of the wiring RW and the change in the potential of the gate of transistor Tr12 are the same, i.e., the capacitance coupling coefficient is 1. In practice, the potential V is calculated by taking the capacitance coupling coefficient into consideration. X It is sufficient to determine the following.

[0257] The potential V X[1] is supplied, the potentials of the nodes NM[1,1] and NMref[1] are V X[1] Rise.

[0258] Here, at time T05-T06, a current I flows from the wiring BL[1] to the transistor Tr12 of the memory cell MC[1,1]. MC[1,1],1 can be expressed by the following formula:

[0259] I MC[1,1],1 =k(V PR -V W[1,1] +V X[1] -V th ) 2 (E7)

[0260] That is, the potential V X[1] By supplying the current, the current flowing from the wiring BL[1] to the transistor Tr12 of the memory cell MC[1,1] is ΔI MC[1,1] =I MC[1,1],1 -I MC[1,1],0 Increase.

[0261] Also, at time T05-T06, a current I flows from the wiring BLref to the transistor Tr12 of the memory cell MCref[1]. MCref[1],1 can be expressed by the following formula:

[0262] I MCref[1],1 =k(V PR +VX[1] -V th ) 2 (E8)

[0263] That is, the potential V X[1] By supplying the current, the current flowing from the wiring BLref to the transistor Tr12 of the memory cell MCref[1] is ΔI MCref[1] =I MCref[1],1 -I MCref[1],0 Increase.

[0264] Next, consider the current flowing through the wire BL[1] and the wire BLref. The wire BLref receives a current I Cref The current flowing through the wiring BLref is discharged to the current mirror circuit CM and the memory cells MCref[1] and [2]. The current discharged from the wiring BLref to the current mirror circuit CM is I CM,1 Then, the following formula holds:

[0265] I Cref -I CM,1 =I MCref[1],1 +I MCref[2],1 (E9)

[0266] The wire BL[1] carries the current I from the current source circuit CS. C The current flowing through the wiring BL[1] is discharged to the current mirror circuit CM and memory cells MC[1,1], [2,1]. Furthermore, a current also flows from the wiring BL[1] to the offset circuit OFST. The current flowing from the wiring BL[1] to the offset circuit OFST is referred to as I α,1 Then, the following formula holds:

[0267] I C -I CM,1 =I MC[1,1],1 +I MC[2,1],1 +I α,1 (E10)

[0268] Then, from equations (E1) to (E10), the current I α,0 and current I α,1 Difference in current (ΔI α) can be expressed as follows:

[0269] ΔI α =I α,1 -I α,0 =2kV W[1,1] V X[1] (E11)

[0270] In this way, the differential current ΔI α is the potential V W[1,1] and V X[1] The value depends on the product of these.

[0271] After that, during the period from time T06 to time T07, the potential of the wiring RW[1] becomes the reference potential, and the potentials of the nodes NM[1,1] and NMref[1] become the same as those during the period from time T04 to time T05.

[0272] Next, at time T07-T08, the potential of the wiring RW[1] becomes V X[1] The potential of the wiring RW[2] becomes V higher than the reference potential. X[2] As a result, the potential V X[1] is supplied, and the potentials of the nodes NM[1,1] and NMref[1] are V X[1] In addition, the potential V X[2] is supplied, and the potentials of the nodes NM[2,1] and NMref[2] are V X[2] Rise.

[0273] Here, at time T07-T08, a current I flows from the wiring BL[1] to the transistor Tr12 of the memory cell MC[2,1]. MC[2,1],1 can be expressed by the following formula:

[0274] I MC[2,1],1 =k(V PR -V W[2,1] +V X[2] -V th )2 (E12)

[0275] That is, the potential V X[2] By supplying the current, the current flowing from the wiring BL[1] to the transistor Tr12 of the memory cell MC[2,1] is ΔI MC[2,1] =I MC[2,1],1 -I MC[2,1],0 Increase.

[0276] Also, at time T07-T08, a current I flows from the wiring BLref to the transistor Tr12 of the memory cell MCref[2]. MCref[2],1 can be expressed by the following formula:

[0277] I MCref[2],1 =k(V PR +V X[2] -V th ) 2 (E13)

[0278] That is, the potential V X[2] By supplying the current, the current flowing from the wiring BLref to the transistor Tr12 of the memory cell MCref[2] is ΔI MCref[2] =I MCref[2],1 -I MCref[2],0 Increase.

[0279] Next, consider the current flowing through the wire BL[1] and the wire BLref. The wire BLref receives a current I Cref The current flowing through the wiring BLref is discharged to the current mirror circuit CM and the memory cells MCref[1] and [2]. The current discharged from the wiring BLref to the current mirror circuit CM is I CM,2 Then, the following formula holds:

[0280] I Cref -I CM,2 =I MCref[1],1 +I MCref[2],1 (E14)

[0281] The wire BL[1] carries the current I from the current source circuit CS. CThe current flowing through the wiring BL[1] is discharged to the current mirror circuit CM and memory cells MC[1,1], [2,1]. Furthermore, a current also flows from the wiring BL[1] to the offset circuit OFST. The current flowing from the wiring BL[1] to the offset circuit OFST is referred to as I α,2 Then, the following formula holds:

[0282] I C -I CM,2 =I MC[1,1],1 +I MC[2,1],1 +I α,2 (E15)

[0283] Then, from the equations (E1) to (E8) and the equations (E12) to (E15), the current I α,0 and current I α,2 Difference in current (ΔI α ) can be expressed as follows:

[0284] ΔI α =I α,2 -I α,0 =2k(V W[1,1] V X[1] +V W[2,1] V X[2] )(E16)

[0285] In this way, the differential current ΔI α is the potential V W[1,1] and potential V X[1] and the potential V W[2,1] and potential V X[2] The value is determined by adding the product of and .

[0286] After that, at times T08-T09, the potentials of the wirings RW[1] and RW[2] become the reference potential, and the potentials of the nodes NM[1,1] and NMref[1] and NMref[2] become the same as at times T04-T05.

[0287] As shown in equations (E11) and (E16), the differential current ΔI input to the offset circuit OFST is α is the potential V corresponding to the first data (weight) Wand the potential V corresponding to the second data (input data) X That is, the differential current ΔI α By measuring this with the offset circuit OFST, the result of the product-sum operation of the first data and the second data can be obtained.

[0288] Although the above focuses on memory cells MC[1,1], [2,1] and memory cells MCref[1], [2], the number of memory cells MC and memory cells MCref can be set arbitrarily. When the number of rows m of memory cells MC and memory cells MCref is set to an arbitrary number i, the differential current ΔIα can be expressed by the following equation.

[0289] ΔI α =2kΣ i V W[i,1] V X[i] (E17)

[0290] Moreover, by increasing the number n of columns of memory cells MC and memory cells MCref, the number of product-sum operations executed in parallel can be increased.

[0291] As described above, by using the semiconductor device MAC, it is possible to perform a product-sum operation on the first data and the second data. Note that by using the configuration shown in FIG. 18 for the memory cells MC and the memory cells MCref, it is possible to configure a product-sum operation circuit with a small number of transistors. Therefore, it is possible to reduce the circuit scale of the semiconductor device MAC.

[0292] When the semiconductor device MAC is used for calculations in a neural network, the number of rows m of the memory cells MC can be set to correspond to the number of input data supplied to one neuron, and the number of columns n of the memory cells MC can be set to correspond to the number of neurons. For example, consider a case where a product-sum calculation is performed using the semiconductor device MAC in the hidden layer HLy shown in Figure 16A. In this case, the number of rows m of the memory cells MC can be set to the number of input data supplied from the input layer ILy (the number of neurons in the input layer ILy), and the number of columns n of the memory cells MC can be set to the number of neurons in the hidden layer HL.

[0293] The structure of the neural network to which the semiconductor device MAC is applied is not particularly limited. For example, the semiconductor device MAC can be used in a convolutional neural network (CNN), a recurrent neural network (RNN), an autoencoder, a Boltzmann machine (including a restricted Boltzmann machine), etc.

[0294] As described above, by using the semiconductor device MAC, it is possible to perform product-sum operations of a neural network. Furthermore, by using the memory cells MC and MCref shown in Fig. 18 in the cell array CA, it is possible to provide an integrated circuit that can improve operation accuracy, reduce power consumption, or reduce the circuit scale.

[0295] The structures, configurations, methods, and the like described in this embodiment can be used in appropriate combination with structures, configurations, methods, and the like described in other embodiments.

[0296] (Embodiment 2) In this embodiment, an imaging device that can be applied to a determination system of one embodiment of the present invention will be described.

[0297] 21A shows a configuration example of an imaging device. Pixel circuits 331 are arranged in a matrix and electrically connected to a driver circuit 332 via wiring 351. The driver circuit 332 can control the data acquisition operation and selection operation of the pixel circuits 331. The driver circuit 332 can be, for example, a shift register.

[0298] In addition, the pixel circuit 331 is electrically connected to a readout circuit 311 (RC) via a wiring 352. The readout circuit 311 has a correlated double sampling circuit (CDS circuit) that reduces noise and an A / D converter that converts analog data into digital data.

[0299] The readout circuit 311 is electrically connected to, for example, a memory. The memory can hold the digital data output from the readout circuit 311.

[0300] The A / D converter included in the read circuit 311 outputs binary data of a predetermined number of bits in parallel. Therefore, the A / D converter is connected to memory cells of the number of bits. For example, if the output of the A / D converter is 8 bits, the A / D converter is connected to 8 memory cells.

[0301] <Pixel circuit> 21B is a circuit diagram illustrating an example of a pixel circuit 331. The pixel circuit 331 can include a photoelectric conversion device 240, a transistor 103, a transistor 104, a transistor 105, a transistor 106, and a capacitor 108. Note that a configuration without the capacitor 108 is also possible.

[0302] One electrode (cathode) of the photoelectric conversion device 240 is electrically connected to one of the source and drain of the transistor 103. The other of the source and drain of the transistor 103 is electrically connected to one of the source and drain of the transistor 104. The one of the source and drain of the transistor 104 is electrically connected to one electrode of the capacitor 108. One electrode of the capacitor 108 is electrically connected to the gate of the transistor 105. The one of the source and drain of the transistor 105 is electrically connected to one of the source and drain of the transistor 106.

[0303] Here, a wiring that connects the other of the source and the drain of the transistor 103, the other of the source and the drain of the transistor 104, one electrode of the capacitor 108, and the gate of the transistor 105 is referred to as a node FD. The node FD can function as a charge detection portion.

[0304] The other electrode (anode) of the photoelectric conversion device 240 is electrically connected to a wiring 121. The gate of the transistor 103 is electrically connected to a wiring 127. The other of the source and the drain of the transistor 104 is electrically connected to a wiring 122. The other of the source and the drain of the transistor 105 is electrically connected to a wiring 123. The gate of the transistor 104 is electrically connected to a wiring 126. The gate of the transistor 106 is electrically connected to a wiring 128. The other electrode of the capacitor 108 is electrically connected to a reference potential line such as a GND wiring. The other of the source and the drain of the transistor 106 is electrically connected to a wiring 352.

[0305] The wirings 127, 126, and 128 can function as signal lines that control the conduction of each transistor. The wiring 352 can function as an output line.

[0306] 21B, the cathode side of the photoelectric conversion device 240 is electrically connected to the transistor 103, and the node FD is reset to a high potential to operate, so the wiring 122 has a high potential (a higher potential than the wiring 121).

[0307] In Figure 21B, a configuration is shown in which the cathode of the photoelectric conversion device 240 is electrically connected to the node FD, but as shown in Figure 21C, a configuration in which the anode side of the photoelectric conversion device 240 is electrically connected to either the source or the drain of the transistor 103 may also be used.

[0308] In this configuration, the node FD is reset to a low potential for operation, and therefore the wiring 122 is set to a low potential (a potential lower than that of the wiring 121).

[0309] The transistor 103 has a function of controlling the potential of the node FD. The transistor 104 has a function of resetting the potential of the node FD. The transistor 105 functions as a source follower circuit and can output the potential of the node FD as image data to the wiring 352. The transistor 106 has a function of selecting a pixel to which image data is output.

[0310] It is preferable to use OS transistors as the transistors 103 to 106 in the pixel circuit 331. OS transistors have extremely low off-state current. In particular, by using transistors with low off-state current as the transistors 103 and 104, the period during which charge can be held at the node FD can be significantly extended. Therefore, a global shutter system in which charge is accumulated simultaneously in all pixels can be applied without complicating the circuit configuration or operation method.

[0311] <Operation method of imaging device> Fig. 22A is a diagram illustrating the operation method of the rolling shutter method, and Fig. 22B is a diagram illustrating the operation method of the global shutter method. En represents the exposure (accumulation operation) of the nth column (n is a natural number), and Rn represents the readout operation of the nth column. Figs. 22A and 22B show the operations from the 1st row to the Mth row (M is a natural number).

[0312] The rolling shutter method is an operating method in which exposure and data readout are performed sequentially, overlapping the readout period of one row with the exposure period of another row. Because the readout operation is performed immediately after exposure, imaging can be performed even with a circuit configuration in which the data retention period is relatively short. However, because one frame of image is composed of data that is not captured simultaneously, distortion occurs in the image when capturing moving objects.

[0313] On the other hand, the global shutter method is a method in which all pixels are exposed simultaneously, data is stored in each pixel, and the data is read out row by row, making it possible to obtain distortion-free images even when capturing moving objects.

[0314] When a transistor with a relatively large off-state current, such as a Si transistor, is used in a pixel circuit, a rolling shutter system is often used because charge tends to flow out of the charge detection unit. To achieve the global shutter system using a Si transistor, complex operations, such as storing data in a separate memory circuit, must be performed at high speed. On the other hand, when an OS transistor is used in a pixel circuit, the global shutter system can be easily achieved because there is almost no data potential flow from the charge detection unit. Note that the imaging device of one embodiment of the present invention can also be operated using the rolling shutter system.

[0315] Note that the pixel circuit 331 may be configured by any combination of OS transistors and Si transistors, or all the transistors may be Si transistors.

[0316] <Pixel circuit operation>

[0317] Next, an example of the operation of the pixel circuit shown in Fig. 21B will be described using the timing chart in Fig. 23A. In the explanation of the timing chart in this specification, a high potential is represented by "H" and a low potential by "L." It is assumed that "L" is always supplied to the wiring 121, and "H" is always supplied to the wirings 122 and 123.

[0318] In the period T1, when the potential of the wiring 126 is set to "H", the potential of the wiring 127 is set to "H", and the potential of the wiring 128 is set to "L", the transistors 103 and 104 are turned on, and the potential of the wiring 122, "H", is supplied to the node FD (reset operation).

[0319] In period T2, when the potential of the wiring 126 is set to "L", the potential of the wiring 127 is set to "H", and the potential of the wiring 128 is set to "L", the transistor 104 becomes non-conductive and the supply of the reset potential is cut off. Also, the potential of the node FD decreases in response to the operation of the photoelectric conversion device 240 (storage operation).

[0320] In the period T3, when the potential of the wiring 126 is set to "L", the potential of the wiring 127 is set to "L", and the potential of the wiring 128 is set to "L", the transistor 103 is turned off, and the potential of the node FD is determined and held (holding operation). At this time, by using OS transistors with low off-state current as the transistors 103 and 104 connected to the node FD, unnecessary charge outflow from the node FD can be suppressed, and the data holding time can be extended.

[0321] In the period T4, when the potential of the wiring 126 is set to "L", the potential of the wiring 127 is set to "L", and the potential of the wiring 128 is set to "H", the transistor 106 is turned on, and the potential of the node FD is read to the wiring 352 by the source follower operation of the transistor 105 (read operation).

[0322] The above is an example of the operation of the pixel circuit shown in FIG. 21B.

[0323] The pixel circuit shown in Fig. 21C can be operated according to the timing chart of Fig. 23B. It is assumed that "H" is always supplied to the wirings 121 and 123, and "L" is always supplied to the wiring 122. The basic operation is the same as that described above with reference to the timing chart of Fig. 23A.

[0324] In one embodiment of the present invention, a transistor may have a back gate, as illustrated in Fig. 24A and Fig. 24B. Fig. 24A shows a structure in which the back gate is electrically connected to the front gate, which has the effect of increasing the on-state current. Fig. 24B shows a structure in which the back gate is electrically connected to a wiring that can supply a constant potential, which allows the threshold voltage of the transistor to be controlled.

[0325] 24A and 24B may be combined to allow each transistor to operate appropriately. Also, the pixel circuit may have a transistor that is not provided with a back gate.

[0326] <Readout circuit> 25 is a diagram illustrating an example of a readout circuit 311 connected to a pixel circuit 331, and shows a circuit diagram of a CDS circuit 400 and a block diagram of an A / D converter 410 electrically connected to the CDS circuit 400. Note that the CDS circuit and A / D converter shown in FIG. 25 are merely examples, and other configurations may also be used.

[0327] The CDS circuit 400 can be configured to include a resistor 401 for voltage conversion, a capacitive element 402 for capacitive coupling, a transistor 403 for supplying a potential V0, a ​​transistor 404 for holding a potential to be supplied to an A / D converter 410, and a capacitive element 405 for holding a potential. The CDS circuit 400 has an input electrically connected to the pixel circuit 331 and an output electrically connected to a comparator circuit (COMP) of the A / D converter 410.

[0328] The potential of the wiring 352 is V resWhen the pixel circuit 331 is in a reset state, the potential of the node N (the connection point of the transistors 403 and 404 and the capacitor 402) is set to V0. Then, the node N is set to a floating state, and the potential of the wiring 352 is set to V data When the pixel circuit 331 outputs image data, the potential of the node N becomes V0+V data -V res Therefore, in the CDS circuit 400, it is possible to subtract the potential in the reset state from the potential of the imaging data output by the pixel circuit 331, thereby reducing noise components.

[0329] The A / D converter 410 can be configured to include a comparator circuit (COMP) and a counter circuit (COUNTER). In the A / D converter 410, the signal potential input to the comparator circuit (COMP) from the CDS circuit 400 is compared with a swept reference potential (RAMP). Then, the counter circuit (COUNTER) operates in response to the output of the comparator circuit (COMP), and digital signals are output to a plurality of wirings 353.

[0330] [Laminated structure 1] Next, the layered structure of the imaging device will be described with reference to a cross-sectional view.

[0331] FIG. 26A is an example of a cross-sectional view of a stack having layers 502 and 503.

[0332] <layer 502> The layer 502 includes a pixel circuit 331 formed on a silicon substrate. Here, the transistor 103, the transistor 104, the capacitor 108, and the photoelectric conversion device 240 are shown as parts of the pixel circuit 331.

[0333] The photoelectric conversion device 240 is a pn junction photodiode formed on a silicon substrate, and has a p-type region 243 and an n-type region 244. The photoelectric conversion device 240 is a buried photodiode, and the p-type region 241 provided on the surface side of the n-type region 244 can suppress dark current and reduce noise. Note that the p-type region 243 may be used as the p-type region 241. It is preferable that the p-type region 241 has a lower resistance than the p-type region 243. It is also preferable that the n-type region 244 has a lower resistance than the p-type region 243. It is also possible to interchange the p-type region and the n-type region among the p-type region 243, the p-type region 241, and the n-type region 244.

[0334] The transistors 103 and 104 are transistors formed on a silicon substrate. The transistors 103 and 104 each include a conductive layer functioning as a gate, a source and a drain formed on the silicon substrate, a channel formation region located between the source and drain, and a gate insulating layer provided between the conductive layer functioning as the gate and the channel formation region. Note that in the example shown in FIG. 26A , the source and drain regions of the transistors 103 and 104 are formed of n-type regions.

[0335] The layer 502 is provided with an insulating layer 242 and an insulating layer 245. The insulating layer 242 functions as an element isolation layer. The insulating layer 245 functions to suppress the outflow of carriers.

[0336] The silicon substrate has grooves for separating pixels, and the insulating layer 245 is provided on the upper surface of the silicon substrate and in the grooves. The insulating layer 245 can prevent carriers generated in the photoelectric conversion device 240 from flowing into adjacent pixels. The insulating layer 245 also has the function of suppressing the intrusion of stray light. For example, the insulating layer 245 may have grooves to suppress the intrusion of stray light from adjacent pixels. Therefore, the insulating layer 245 can suppress color mixing. An anti-reflection film may be provided between the upper surface of the silicon substrate and the insulating layer 245.

[0337] The element isolation layer can be formed using a LOCOS (LOCal Oxidation of Silicon) method, an STI (Shallow Trench Isolation) method, or the like. For example, an inorganic insulating film such as a silicon oxide film or silicon nitride, or an organic insulating film such as polyimide or acrylic can be used as the insulating layer 245. Note that the insulating layer 245 may have a multi-layer structure.

[0338] In the example shown in FIG. 26A, the n-type region 244 (corresponding to the cathode) of the photoelectric conversion device 240 can also function as either the source or the drain of the transistor 103.

[0339] The layer 502 also includes insulating layers 222, 223, 226, and 227, a wiring 121, and electrodes 129a and 129b. The insulating layer 222 functions as a protective film. The insulating layers 223 and 227 function as an interlayer insulating film and a planarizing film. The electrodes 129a and 129b each function as an electrode of the capacitor 108. The insulating layer 226 is sandwiched between the electrodes 129a and 129b and functions as a dielectric layer of the capacitor 108. The electrode 129a is electrically connected to the other of the source and drain of the transistor 103 via a plug provided in the insulating layer 223. The wiring 121 functions as a power supply line. The p-type region 243 (anode) is electrically connected to the wiring 121.

[0340] The protective film may be, for example, a silicon nitride film, a silicon oxide film, or an aluminum oxide film.The interlayer insulating film and the planarizing film may be, for example, an inorganic insulating film such as a silicon oxide film, or an organic insulating film such as an acrylic resin or a polyimide resin.The dielectric layer of the capacitive element may be, for example, a silicon nitride film, a silicon oxide film, or an aluminum oxide film.

[0341] The Si transistor shown in Fig. 26A is a planar type having a channel formation region in a silicon substrate. The Si transistor may also be a fin type as shown in Fig. 27A. Fig. 27B shows a cross section (cross section in the channel width direction) taken along line A1-A2 shown in Fig. 27A.

[0342] 27C, the transistor may have a silicon thin-film semiconductor layer 545. The semiconductor layer 545 may be, for example, single-crystal silicon (SOI (Silicon on Insulator)) formed on an insulating layer 546 on a silicon substrate 210.

[0343] Conductors that can be used as wiring, electrodes, and plugs for electrical connection between devices may be made of a metal element selected from aluminum, chromium, copper, silver, gold, platinum, tantalum, nickel, titanium, molybdenum, tungsten, hafnium, vanadium, niobium, manganese, magnesium, zirconium, beryllium, indium, ruthenium, iridium, strontium, lanthanum, etc., or an alloy containing the above-mentioned metal elements as a component, or an alloy combining the above-mentioned metal elements, etc. The conductor is not limited to a single layer, and may also be made of multiple layers composed of different materials. It may also contain trontium.

[0344] <layer 503> The layer 503 is formed on the layer 502. The layer 503 includes a light-shielding layer 251, an optical conversion layer 250, and a microlens array 255.

[0345] The light-shielding layer 251 can prevent light from flowing into adjacent pixels. A metal layer such as aluminum or tungsten can be used for the light-shielding layer 251. The metal layer may also be stacked with a dielectric film that functions as an anti-reflection film.

[0346] A color filter can be used for the optical conversion layer 250. A color image can be obtained by assigning colors such as R (red), G (green), B (blue), Y (yellow), C (cyan), and M (magenta) to the color filter for each pixel.

[0347] Furthermore, if a wavelength cut filter is used in the optical conversion layer 250, an imaging device that can obtain images in various wavelength regions can be obtained.

[0348] For example, if a filter that blocks light with wavelengths equal to or shorter than visible light is used in the optical conversion layer 250, it can be used as an infrared imaging device. Also, if a filter that blocks light with wavelengths equal to or shorter than near-infrared light is used in the optical conversion layer 250, it can be used as a far-infrared imaging device. Also, if a filter that blocks light with wavelengths equal to or longer than visible light is used in the optical conversion layer 250, it can be used as an ultraviolet imaging device.

[0349] Furthermore, if a scintillator is used for the optical conversion layer 250, an imaging device can be provided that obtains an image that visualizes the intensity of radiation, such as for use in an X-ray imaging device. When radiation such as X-rays that has passed through a subject is incident on the scintillator, it is converted into light (fluorescence) such as visible light or ultraviolet light by the photoluminescence phenomenon. Then, image data is obtained by detecting this light with the photoelectric conversion device 240. An imaging device having this configuration may also be used for a radiation detector or the like.

[0350] Scintillators contain materials that absorb the energy of radiation such as X-rays or gamma rays and emit visible or ultraviolet light when irradiated with such radiation. For example, materials such as Gd2O2S:Tb, Gd2O2S:Pr, Gd2O2S:Eu, BaFCl:Eu, NaI, CsI, CaF2, BaF2, CeF3, LiF, LiI, and ZnO dispersed in resin or ceramics can be used.

[0351] A microlens array 255 is provided on the optical conversion layer 250. Light passing through each lens of the microlens array 255 passes through the optical conversion layer 250 directly below and is irradiated onto the photoelectric conversion device 240. By providing the microlens array 255, concentrated light can be incident on the photoelectric conversion device 240, thereby enabling efficient photoelectric conversion. The microlens array 255 is preferably formed from a resin or glass that is highly translucent to visible light.

[0352] Alternatively, OS transistors may be used as the transistors 103 and 104. The stacked structure shown in FIG. 26B includes layers 502 and 503. The layer 502 includes a layer 562 including an element provided on a Si substrate and a layer 563 including an OS transistor. The layer 562 shown in FIG. 26B includes a photoelectric conversion device 240 provided on a Si substrate. Although no Si transistor is shown in the layer 562 in FIG. 26B, the layer 562 may include the transistor 105 as shown in FIG. 28A. Although not shown, the layer 562 may also include the transistor 106. Alternatively, the transistors 105, 106, and the like may be OS transistors provided in the layer 563.

[0353] Since the off-state current of the OS transistor is extremely small, the charge stored in the capacitor 108 can be held for a long time.

[0354] The OS transistor can be stacked with a Si transistor or a photoelectric conversion device 240 provided on a Si substrate. This allows for circuit integration and increases the area of ​​the photoelectric conversion device 240. The OS transistor will be described in detail with reference to FIG. 29.

[0355] In the configuration shown in FIG. 28B, layer 502 includes layer 561 in addition to layers 562 and 563. Layer 561 is a configuration example in which a pn junction photodiode with a selenium-based material as a photoelectric conversion layer is used as photoelectric conversion device 240. Layer 566a serves as one electrode, layers 566b and 566c serve as photoelectric conversion layers, and layer 566d serves as the other electrode. Layer 561 can be formed directly on layer 563. Note that even when an organic photoconductive film is used as photoelectric conversion device 240 included in layer 561, the connection with the transistor is similar.

[0356] The insulating layer 541 is provided over the transistor 102 and the transistor 103 included in the layer 563. The layer 566a has a region buried in the insulating layer 541.

[0357] 28B shows an example in which fin-type transistors are used as Si transistors in a layer 562. In the example shown in FIG.

[0358] The stacked structure of the Si transistor and the OS transistor shown in FIGS. 28A and 28B can also be applied to the configuration of the Si transistor and the OS transistor included in the processing unit (PU). For example, a pixel circuit of an imaging device and a semiconductor device including the processing unit can be fabricated on the same silicon substrate. Therefore, for example, the pixel circuit and the driver circuit of the imaging device and the semiconductor device included in the determination system of one embodiment of the present invention can be provided on the same chip. By providing the configuration of the determination system of one embodiment of the present invention on a single chip, for example, the chip area can be reduced. Alternatively, for example, the pixel circuit and the driver circuit of the imaging device and the semiconductor device including the processing unit (PU) can be fabricated using the same process, thereby reducing the cost required for chip manufacturing. Furthermore, by providing the configuration of the determination system of one embodiment of the present invention on a single chip, for example, the energy required for transferring an image to a memory circuit or a determination circuit included in the semiconductor device can be reduced, thereby improving the computational efficiency.

[0359] 29A shows the details of an OS transistor. The OS transistor shown in FIG. 29A has a self-aligned structure in which an insulating layer is provided over a stack of an oxide semiconductor layer and a conductive layer, and a source electrode 705 and a drain electrode 706 are formed by providing grooves that reach the semiconductor layer.

[0360] The OS transistor can have a structure including a channel formation region, a source region 703, and a drain region 704 formed in an oxide semiconductor layer, as well as a gate electrode 701 and a gate insulating film 702. At least the gate insulating film 702 and the gate electrode 701 are provided in the trench. An oxide semiconductor layer 707 may be further provided in the trench.

[0361] As shown in FIG. 29B, the OS transistor may have a self-aligned structure in which a source region and a drain region are formed in a semiconductor layer using a gate electrode 701 as a mask.

[0362] Alternatively, as shown in FIG. 29C, it may be a non-self-aligned top-gate transistor having a region where the source electrode 705 or the drain electrode 706 overlaps with the gate electrode 701.

[0363] Although the transistors 102 and 103 are shown as having a back gate 535, they may not necessarily have a back gate. The back gate 535 may be electrically connected to the front gate of the transistor provided opposite to it, as shown in the cross-sectional view of the transistor in the channel width direction in FIG. 29D. Note that FIG. 29D illustrates the transistor in FIG. 21A as an example, but the same applies to transistors with other structures. Furthermore, a fixed potential different from that of the front gate may be supplied to the back gate 535.

[0364] An insulating layer 543 having a function of preventing hydrogen diffusion is provided between the region where the OS transistor is formed and the region where the Si transistor is formed. Hydrogen in the insulating layer provided near the channel formation regions of the transistors 105 and 106 terminates dangling bonds of silicon. Meanwhile, hydrogen in the insulating layer provided near the channel formation regions of the transistors 102 and 103 is one of the factors that generate carriers in the oxide semiconductor layer.

[0365] The insulating layer 543 confines hydrogen in one layer, thereby improving the reliability of the transistors 105 and 106. Furthermore, the insulating layer 543 suppresses diffusion of hydrogen from one layer to the other layer, thereby improving the reliability of the transistors 102 and 103.

[0366] The insulating layer 543 can be made of, for example, aluminum oxide, aluminum oxynitride, gallium oxide, gallium oxynitride, yttrium oxide, yttrium oxynitride, hafnium oxide, hafnium oxynitride, or yttria-stabilized zirconia (YSZ).

[0367] The structures, configurations, methods, and the like described in this embodiment can be used in appropriate combination with structures, configurations, methods, and the like described in other embodiments.

[0368] (Embodiment 3) In this embodiment, examples of electronic devices to which a semiconductor device of one embodiment of the present invention is applied will be described.

[0369] FIG. 30A shows a surveillance camera, which includes a support base 951, a camera unit 952, a protective cover 953, and the like. The camera unit 952 is provided with a rotation mechanism and is installed on a ceiling, enabling capturing images of the entire periphery. The camera unit 952 can be used as an imaging device included in the determination system of one embodiment of the present invention. Furthermore, when the camera unit 952 is electrically connected to the semiconductor device of one embodiment of the present invention, a suspicious person can be identified from captured image information. Note that the term "surveillance camera" is a common name and is not intended to limit the application. For example, a device having a function as a surveillance camera is also called a camera or a video camera.

[0370] Fig. 30B shows an example of an aircraft. An aircraft 6500 shown in Fig. 30B has a propeller 6501, a camera 6502, a battery 6503, and the like, and has the function of flying autonomously.

[0371] For example, image data captured by the camera 6502 is stored in the electronic component 6504. The electronic component 6504 can analyze the image data and detect the presence or absence of an obstacle when moving. The camera 6502 may be an imaging device of a plurality of types. The camera 6502 can be used as the imaging device included in the determination system of one embodiment of the present invention. Furthermore, when the camera 6502 is electrically connected to the semiconductor device of one embodiment of the present invention, a suspicious person can be identified from captured information.

[0372] The structures, configurations, methods, and the like described in this embodiment can be used in appropriate combination with structures, configurations, methods, and the like described in other embodiments. [Explanation of symbols]

[0373] :ACTV: activation function circuit, BGL2: wiring, BGL6: wiring, BKC1: circuit, BKC2: circuit, BKC10: circuit, BKC20: circuit, BL: wiring, BLref: wiring, BL1: bit line, BLB: wiring, BLB1: bit line, C1: capacitor, C3: capacitor, C6: capacitor, C11: capacitor, C21: capacitor, CA: cell array, CB1: capacitor, CB2: capacitor, CB11: capacitor, CB12: capacitor, CLD: circuit, CM: current mirror circuit, CS: current source circuit, HLy: hidden layer, IL: wiring, ILref: wiring, ILy: Input layer, INV11: inverter circuit, INV12: inverter circuit, INV2: inverter circuit, INV3: inverter circuit, INV4: inverter circuit, M1: transistor, M11: transistor, M12: transistor, M2: transistor, M3: transistor, M4: transistor, M5: transistor, M6: transistor, MA1: transistor, MAC: semiconductor device, MC: memory cell, MCref: memory cell, MC1: transistor, MC2: transistor, MemC1: circuit, MemC2: circuit, MemC20: circuit, MR1: transistor Transistor,MW1:Transistor,MW2:Transistor,MW11:Transistor,MW12:Transistor,OFST:Offset circuit,OLy:Output layer,PCC10:Circuit,PDL:Wiring,R1:Resistance element,RBL:Bit line,RTC10:Circuit,RWL:Word line,SMC20:Circuit,Tr11:Transistor,Tr12:Transistor,WBL:Bit line,WD:Wiring,WDref:Wiring,WL:Wiring,WL1:Word line,WLD:Circuit,WWL:Word line,10:Power supply circuit,11:Memory cell,15:Memory cell,16:Memory cell,20: PU, 30: processor core, 31: memory circuit, 32: circuit, 35: power supply line, 40: cache, 41: memory array, 42: peripheral circuit, 43: control circuit, 45: memory cell, 60: PMU, 61: circuit, 65: clock control circuit, 70: PSW, 71: PSW, 80: preliminary determination circuit, 81: memory, 82: circuit, 88: interface, 89: bus line, 91: determination circuit, 92: detection circuit, 100: memory circuit, 102: transistor, 103: transistor, 104: transistor, 105: transistor, 106: transistor, 108: capacitance element,110: FF, 120: memory cell, 121: wiring, 122: wiring, 123: wiring, 126: wiring, 127: wiring, 128: wiring, 129a: electrode, 129b: electrode, 130: processor core, 131: control device, 132: program counter, 133: pipeline register, 134: pipeline register, 135: register file, 136: ALU, 137: data bus, 210: silicon substrate, 212: power domain, 213: power domain, 215: power switch switch, 217: power switch, 220: NOSRAM, 222: insulating layer, 223: insulating layer, 226: insulating layer, 227: insulating layer, 230: memory cell array, 231: control circuit, 232: row circuit, 233: column circuit, 240: photoelectric conversion device, 241: p-type region, 242: insulating layer, 243: p-type region, 244: n-type region, 245: insulating layer, 250: optical conversion layer, 251: light-shielding layer, 255: microlens array, 311: circuit, 331: pixel circuit, 332: drive circuit, 341: DO SRAM, 351: wiring, 352: wiring, 353: wiring, 361: memory cell array, 365: peripheral circuit, 371: power switch, 373: power switch, 400: CDS circuit, 401: resistor, 402: capacitance element, 403: transistor, 404: transistor, 405: capacitance element, 410: A / D converter, 502: layer, 503: layer, 535: back gate, 541: insulating layer, 543: insulating layer, 545: semiconductor layer, 546: insulating layer, 561: layer, 562: layer, 563: layer, 566a: layer, 566b: layer, 566c: layer, 566d: layer, 601: imaging device, 602: display device, 603: terminal, 700: semiconductor device, 701: gate electrode, 702: gate insulating film, 703: source region, 704: drain region, 705: source electrode, 706: drain electrode, 707: oxide semiconductor layer, 951: support, 952: camera unit, 953: protective cover, 6500: flying vehicle, 6501: propeller, 6502: camera, 6503: battery, 6504: electronic component,

Claims

1. The power supply control device includes a detection circuit, a first determination circuit, a second determination circuit, a power supply circuit, and a power supply management device. the detection circuit has a function of analyzing the imaging data, and selecting a first value if an object is detected, and selecting a second value if the object is not detected; when the first value is selected by the detection circuit, the power supply management device has a function of switching the first determination circuit from an off mode to an on mode by supplying a voltage from the power supply circuit to the first determination circuit; when the second value is selected by the detection circuit, the power management device has a function of not supplying a voltage from the power supply circuit to the first determination circuit, thereby not switching the first determination circuit from an off mode to an on mode; the first determination circuit has a function of detecting a human being by extracting contours from the imaging data when the first value is selected by the detection circuit; when the first determination circuit does not detect the human being, the power supply circuit and the power supply management device have a function of transitioning to a hibernation state; the second determination circuit has a function of extracting features from the imaging data when the first determination circuit detects the human, recognizing the face of the human, and then collating the result with a database to determine whether the human is a dangerous person, and a function of sending a signal to an external terminal notifying an abnormality when the human is determined to be a dangerous person; The semiconductor device has a function of causing the power supply circuit and the power supply management device to enter a hibernation state when the second determination circuit does not determine that the person is a dangerous person.

2. 2. The semiconductor device according to claim 1, wherein the first determination circuit consumes less power than the second determination circuit.

3. The imaging device includes an imaging circuit, a detection circuit, a processing device, a second determination circuit, and a power supply circuit. the processing device includes a first determination circuit, a power management device, a processor core, and a memory circuit; the processor core has a function of giving an instruction to the first determination circuit; the memory circuit has a function of storing data generated by the processor core; The imaging device has a function of acquiring imaging data, the detection circuit has a function of analyzing the imaging data, and selecting a first value if an object is detected, and selecting a second value if the object is not detected; when the first value is selected by the detection circuit, the power supply management device has a function of switching the first determination circuit from an off mode to an on mode by supplying a voltage from the power supply circuit to the first determination circuit; when the second value is selected by the detection circuit, the power management device has a function of not supplying a voltage from the power supply circuit to the first determination circuit, thereby not switching the first determination circuit from an off mode to an on mode; the first determination circuit has a function of detecting a human being by extracting contours from the imaging data when the first value is selected by the detection circuit; when the first determination circuit does not detect the human being, the power supply circuit and the power supply management device have a function of transitioning to a hibernation state; the second determination circuit has a function of extracting features from the imaging data when the first determination circuit detects the human, recognizing the face of the human, and then collating the result with a database to determine whether the human is a dangerous person, and a function of sending a signal to an external terminal notifying an abnormality when the human is determined to be a dangerous person; When the second determination circuit does not determine that the person is a dangerous person, the power supply circuit and the power supply management device have a function of transitioning to a hibernation state, The imaging device is a determination system having a plurality of pixel circuits arranged in a matrix.

4. 4. The determination system according to claim 3, wherein the first determination circuit consumes less power than the second determination circuit.

5. In claim 3 or 4, the memory circuit includes a first transistor and a first capacitor; one of a source and a drain of the first transistor is electrically connected to one electrode of the first capacitance element; The first transistor has a channel formation region formed of an oxide semiconductor.

6. In claim 3 or 4, the memory circuit includes a first transistor and a first capacitor; one of a source and a drain of the first transistor is electrically connected to one electrode of the first capacitance element; the first transistor has a channel formation region including an oxide semiconductor; each of the pixel circuits includes a photoelectric conversion device, a second transistor, and a second capacitance element; one electrode of the photoelectric conversion device is electrically connected to one of the source and drain of the second transistor; the other of the source and the drain of the second transistor is electrically connected to one electrode of the second capacitance element; The second transistor has a channel formation region formed of an oxide semiconductor.

Citation Information

Patent Citations

  • Image recognition device

    JP2012141922A

  • Pedestrian detection device

    JP2014006700A

  • Surrounding environment recognition device

    JP2015095886A

  • Display device for vehicle

    JP2018184707A

  • Semiconductor integrated circuit

    WO2009078081A1