JP2011187577A - Spin conducting element - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a spin conducting element applicable to a multichannel type. <P>SOLUTION: The spin conducting element includes a spin current source A, having a pinned layer provided in a semiconductor region of a semiconductor substrate 1; a spin current adsorption electrode B, having a free layer provided in the semiconductor region, a detection electrode PB1 provided in the semiconductor region to measure a voltage between the spin current adsorption electrode B and the detection electrode PB1; and a bias-applied electrode C, positioned between the spin current source A and the spin current adsorption electrode B in the semiconductor region and including a non-magnetic material to which a positive potential is applied, with respect to the spin current source A. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a spin transport element that can be used in a magnetic sensor, an MRAM (Magnetoresistive Random Access Memory), and the like.

  Conventionally, a magnetoresistive element (MR element) widely used in a magnetic head or the like is known. The MR element utilizes electron spin. Since the spin diffusion length in which electrons can hold spin is about several tens to several hundreds of nanometers, the electron travel distance needs to be less than or equal to the spin diffusion length. Recent MR elements have a structure comprising a laminated magnetic layer / non-magnetic layer / magnetic layer, and constitute a CPP (Current Perpendicular to Plane) structure in which current flows in a direction perpendicular to the layer surface. Since a manufacturing technique for reducing the layer thickness has been established, in the case of a structure in which electrons flow in the stacking direction, the electron travel distance can be shortened by reducing the thickness of each layer.

  In recent years, not only the lamination technology but also the lateral microstructure fabrication technology has advanced, and the electron travel distance can be made shorter than the spin diffusion length even in a structure in which current flows in the plane. In such an in-plane device, it is easy to produce not only a device having a two-terminal structure but also a device having a structure having three or more terminals. For example, a gate electrode is disposed in a region between a source electrode and a drain electrode. Application to spin transistors is also possible.

  As a magnetic device in which a two-terminal MR element is formed in a plane, a magnetic sensor having a local structure and a non-local structure can be considered. FIG. 22A shows a magnetic sensor having a local structure, and FIG. 22B shows a magnetic sensor having a non-local structure.

  As shown in FIG. 22A, in the local structure magnetic sensor, a pinned layer (fixed layer) A ′ made of a ferromagnetic material is disposed at one location on the nonmagnetic layer 2 such as Cu, and is separated from the nonmagnetic layer 2. A free layer B ′ made of a ferromagnetic material is disposed at another location on the magnetic layer 2. As a result, the pinned layer A ′ and the free layer B ′ are electrically connected via the nonmagnetic layer 2. In the magnetic sensor having the local structure, the current I flowing from the current source J flows through the free layer B ', the nonmagnetic layer 2 and the pinned layer A' sequentially. The direction in which electrons flow is opposite to the current I. Thereby, the resistance value R of the MR element changes according to the external magnetic field received by the free layer B ′, and the voltage V (== between the pinned layer A ′ and the free layer B ′ constituting both terminals of the MR element. I × R) changes. By measuring this voltage V with a voltmeter V, the magnitude of the external magnetic field can be measured.

  As shown in FIG. 22B, the non-local structure magnetic sensor includes a pair of electrode pads PA1 and PB1, and the electrode pads PA1 and PB1 are connected via a nonmagnetic layer 2 such as Cu. . A pinned layer A ′ made of a ferromagnetic material is arranged at one location on the nonmagnetic layer 2, and a free layer B ′ made of a ferromagnetic material is arranged at another location on the nonmagnetic layer 2 spaced apart from the pinned layer A ′. As a result, the pinned layer A ′ and the free layer B ′ are electrically connected via the nonmagnetic layer 2. In the magnetic sensor having a non-local structure, the current path and voltage are measured. The path is different, and a spin accumulation type magnetic sensor using a spin current is configured.

  Several phenomena are known for spin current. For example, when upward spin electrons and downward spin electrons flow in the same amount in opposite directions, the flow of electrons cancels, but a spin current is generated. That is, even when there is no electron flow, a spin current is generated, and there is a phenomenon in which charge accumulation is performed in one region. Such a phenomenon can also be regarded as a spin current leaking out from a region where spin electrons are accumulated.

  Further, by causing the current I to flow through the pinned layer A ′, a spin current is generated in the pinned layer A ′. Since electrons that have passed through the pinned layer A 'do not flow into the free layer B', the current is zero in the channel region between the pinned layer A 'and the free layer B'. Since the spin current is a conserved amount, it can be understood that the spin current also leaks into the nonmagnetic layer 2 from the pinned layer A ′ / nonmagnetic layer 2 interface. In the vicinity of the pinned layer A ′ / nonmagnetic layer 2 interface, regions having different electron concentrations are formed depending on the spin direction, and this phenomenon is called spin accumulation.

  In any case, when spin electrons flow in the nonmagnetic layer 2, a spin current is generated separately from the electron current, and in the state where the spin current flows, depending on the magnetization direction of the free layer B ′, A voltage is observed. More specifically, a spin current is generated in the pinned layer A ′ by the electron current flowing in the pinned layer A ′, and spin accumulation occurs near the interface between the pinned layer A ′ / nonmagnetic layer 2. Spins are diffused from a region where spins are accumulated to generate a spin current, which is absorbed by the free layer B '. At this time, the potential of the free layer B 'varies depending on the relative angle between the magnetization directions of the free layer B' and the pinned layer A ', and a voltage change occurs between the nonmagnetic layer 2 and the free layer B'. This voltage change is detected. That is, when only the magnetization direction of the free layer B ′ is changed by external magnetization, a voltage V corresponding to the magnetization direction is generated and can be detected as a sensor output.

  Measuring this voltage is the principle of the spin accumulation type magnetic sensor. In the case of such a structure, since current does not contribute in the voltage measurement path, precise measurement is expected. The application of the above principle to a magnetic sensor or the like is described in, for example, Patent Documents 1 and 2 below. These devices use the injection current as input and output the voltage change in the free layer due to the external magnetic field. If the structure is such that the direction of magnetization in the free layer is written by an external magnetic field or spin injection, this device can function as a magnetic memory. That is, in FIG. 22B, when the positive magnetization direction and the negative magnetization direction of the free layer B ′ are associated with the information “1” or the information “0”, respectively, the stored information is By flowing a current I for sensing, written information can be read as a voltage.

  According to the above-mentioned non-local structure, only the spin current is generated, and the voltage difference before and after the magnetization reversal of the free layer can be measured with high accuracy, so that it can be used as a quantitative evaluation method of the spin current. The spin current has very little noise due to the AMR (Asymmetric Magnetoresistive) effect and Joule heat, and is suitable for the transmission of high-quality spin information. As described above, an attempt has been made to apply a non-local structure spin conduction element to a device. For example, Patent Document 1 discloses an application to a spin transistor, and Patent Document 2 discloses an application to a hard disk magnetic reproducing head.

  Non-patent document 1 describes recent technical trends in spin transport devices. A device using a spin current is suitable for a device operating at a low voltage because of low noise, a small amount of heat energy generated, and low power consumption, and can be used for transmission of spin information. Non-Patent Document 1 discloses a conceptual diagram of an in-plane spintronic device having a spin current generation and detection function. In this spintronic device, a part having a function of mixing, amplifying and modulating spin information is connected to a conduction channel of spin current.

  When the spin transport device is actually used in the industry, it is preferable to provide an insulating film that forms a tunnel barrier between the magnetic body and the non-magnetic channel in order to obtain as large an output voltage as possible. As a result, resistance matching at the interface between the magnetic material and the non-magnetic material channel is improved, and the output is increased.

  This structure can be applied not only when a metal channel is used as in Patent Documents 1 and 2, but also when a semiconductor region is used. Non-Patent Document 2 shows an example using a GaAs channel, and Non-Patent Document 3 shows an example using an Si channel. However, in Non-Patent Document 2, the tunnel barrier is substantially formed in place of the insulating layer by tunneling the Schottky barrier formed at the interface between GaAs and the metal magnetic material.

Non-Patent Document 4 shows electric field-dependent spin diffusion and spin injection into a semiconductor. Non-Patent Document 4 shows that spin drifts in an electric field, and spin proceeds according to the following drift-diffusion equation (Equation (1)). n _up is the up spin electron concentration, n _down is the down spin electron concentration, λ _N = (Dτ) ^1/2 is the spin diffusion length when the electric field is zero, that is, the intrinsic spin diffusion length (D: spin diffusion coefficient, τ : Spin lifetime), e is the elementary electric charge, E is the electric field, k _B is the Boltzmann constant, and T is the absolute temperature.

Conventionally, spin diffusion itself has been found in metals and handled by the diffusion equation, but in Non-Patent Document 4, an electric field term is added to the second term. According to this theory, the influence of the electric field can be treated as a change in effective spin diffusion length, and the spin diffusion length λ _{d in} the electron drift direction (down-stream) becomes longer and the opposite direction (up− The spin diffusion length λ _u of stream is shortened. The spin diffusion length is given by equations (2-1) and (2-2). When E is infinite, Expressions (2-1) and (2-2) become Expressions (2-1-1) and (2-2-1), respectively. Note that μ is the spin mobility.

  Non-Patent Document 5 experimentally proves that spin can be transmitted by using an electric field even for a wafer having a thickness of 350 μm. However, this experiment relates to spin conduction using hot electrons.

JP 2004-186274 A JP 2007-299467 A

Satoshi Oikawa, “Spintronics Technology Research Report I”, Japan Electronics and Information Technology Industries Association, March 2009 XIAOHUA. Lou et al. , "Electrical detection of spin transport in lateral ferromagnet-semiconductor devices", Nature 7 (Nature 7), Nature 7 Month, Vol. 3, pp. 197-202 O. M.M. J. et al. van't Erve et al. , “Electrical injection and detection of spin-polarized carriers in silicon in a lateral transport geometry of spin-polarized silicon・ Geometry), Applied physics Letters (Applied Physics Letters), June 2007, 91,21210 Z. G. Yu, M .; E. Flatte, “Electric-field dependent spin diffusion and spin injection into semiconductors” (Electric Field Dependent Spin Diffusion and Spin Injection Into Semiconductor), Rephysical 200 14th, B66, 201202 (R) Biqin Huang et al. , "Coherent Spin Transport through a 350 Micron Thick Silicon Wafer" (Coherent Spin Transport Through a 350 Micron Thick Silicon Wafer), Physical Review Letters, 10 Years Physical Review Letters 10 26th, 99, 177209

  However, when integrating device functions, it is preferable to configure a multichannel spin transport device that conducts one signal to a plurality of electrodes. However, when a spin current is to be supplied from a single spin current source to a plurality of spin current absorption electrodes, the upper limit of the intrinsic spin diffusion length is several μm, and the physical property is sufficient to arrange a plurality of spin current absorption electrodes. A space cannot be secured, and a multi-channel spin conduction element cannot be formed. In particular, when a plurality of spin current absorption electrodes are arranged around one spin current source, the spin current travels by diffusing in all directions, so the magnitude of the spin current supplied to one spin current absorption electrode is There is a problem that the output becomes extremely small, and therefore the output becomes small. Therefore, conventionally, a multi-channel type spin transport device could not be constructed.

  The present invention has been made in view of such problems, and an object of the present invention is to provide a spin transport device applicable to a multi-channel type.

  In order to solve the above-described problems, the present invention provides a spin transport device according to the present invention comprising a semiconductor region, a spin current source having a pinned layer provided in the semiconductor region, and a free layer provided in the semiconductor region. A spin current absorption electrode, a detection electrode provided in the semiconductor region for measuring a voltage between the spin current absorption electrode, the spin current source and the spin current absorption electrode in the semiconductor region, And a bias application electrode including a non-magnetic material to which a positive potential is applied to the spin current source.

  According to the spin transport device according to the present invention, since the positive potential is applied to the bias application electrode with respect to the spin current source, the spin accumulated immediately below the spin current source is formed by this potential. It moves to just below the bias application electrode according to the electric field. From just below the bias application electrode, the spin is oozed out and diffused, and is absorbed into the spin current absorption electrode.

  At this time, the voltage between the spin current absorption electrode and the detection electrode differs depending on the relative direction of the magnetization direction of the spin current absorption electrode and the magnetization direction of the pinned layer provided above the spin current source. . In this spin transport device, the presence of the bias application electrode allows it to be disposed relatively far from the spin current source, so that a space for arranging a plurality of bias application electrodes can be secured, and therefore adjacent to this. It is possible to realize a multi-channel type spin transport element by arranging a spin current absorption electrode. Even when the spin transport device is not a multi-channel type, the design freedom increases due to the increase in space, so that various devices can be integrated.

  The spin transport device according to the present invention is characterized in that the number of electrode groups including the spin current absorption electrode, the detection electrode, and the bias application electrode is plural. In the case where there are a plurality of these electrode groups, a multi-channel type spin conduction element can be formed, so that a plurality of voltage detection parts can be provided for a single spin source, and the element functions are integrated. can do.

    The spin transport device according to the present invention includes a bias power supply for applying the potential to each bias application electrode, and a switch interposed between the bias power supply and each bias application electrode, The switch is selectively connected. That is, by selectively connecting the switches, it is possible to suppress a decrease in the density of the spin current that flows from the single spin current source toward each bias application electrode. Thereby, a sufficient amount of spin current can be supplied to the spin current absorption electrode, and a precise voltage output can be obtained.

  Further, the spin transport device according to the present invention includes a current source for controlling the magnetization direction of the free layer, a voltage detection means for detecting a voltage between the spin current absorption electrode and the bias application electrode, It is characterized by providing. If the positive and negative magnetization directions of the free layer are associated with 0 and 1 information, the free layer stores information. The direction of this magnetization can be determined by the magnitude of the detected voltage.

  The spin transport device according to the present invention includes a semiconductor substrate including the semiconductor region, and a resistivity of a region outside the semiconductor region in the semiconductor substrate is higher than that of the semiconductor region. In this case, since the spin current selectively flows in the semiconductor region, the size of the flow path is limited, and stable spin current can be transmitted.

  According to the present invention, it is possible to provide a spin transport device applicable to a multi-channel type.

It is a perspective view of the spin transport device according to the first embodiment. It is II-II arrow sectional drawing of the spin transport element shown in FIG. It is sectional drawing of the spin current source A as an example. It is a perspective view of the spin transport device according to the second embodiment. It is a figure explaining the current source J2 which controls the direction of magnetization of a free layer. It is a figure explaining another type of current source J2 which controls the direction of magnetization of a free layer. It is a circuit diagram of a magnetic memory using a spin transport element. It is a graph which shows the mode of spin diffusion, the horizontal axis shows the position, and the vertical axis shows the spin concentration. It is a perspective view of the spin transport element used for calculation. It is a graph which shows the relationship between an electric current and spin diffusion length. 10 is a graph showing a relationship between a drift distance L1 (μm) and a spin accumulation amount (mV) according to Comparative Example A. 10 is a graph showing a relationship between a drift distance L1 (μm) and a spin accumulation amount (mV) according to Comparative Example B. 10 is a graph showing a relationship between a drift distance L1 (μm) and a spin accumulation amount (mV) according to Comparative Example B. It is a graph which shows the relationship between the drift distance L1 (micrometer) which concerns on Example A, and spin accumulation amount (mV). It is a graph which shows the relationship between the drift distance L1 (micrometer) which concerns on Example B, and spin accumulation amount (mV). It is a graph which shows the relationship between the drift distance L1 (micrometer) which concerns on Example B, and spin accumulation amount (mV). It is a graph which shows the relationship between the drift distance L1 (micrometer) which concerns on Example C, and spin accumulation amount (mV). It is a graph which shows the relationship between the drift distance L1 (micrometer) which concerns on Example C, and spin accumulation amount (mV). It is a graph which shows the relationship for every drift distance L1 of bias voltage Vb (V) and output voltage V (mV). It is a graph which shows the relationship for every drift distance L1 of bias voltage Vb (V) and output voltage V (mV). 6 is a cross-sectional view of a spin transport device according to Comparative Example 1. FIG. It is a figure which shows typically the magnetic sensor (a) of a local structure, and the magnetic sensor (b) of a non-local structure.

  Hereinafter, the spin transport device according to the embodiment will be described. Note that the same reference numerals are used for the same elements, and redundant description is omitted.

  FIG. 1 is a perspective view of the spin transport device according to the first embodiment.

  The spin transport device includes a plurality of electrode groups PA1, PB1, A, B, and C formed on the semiconductor substrate 1. Assuming that the surface region of the semiconductor substrate 1 is a semiconductor region, the semiconductor region includes an electrode pad PA1, an electrode pad (detection electrode) PB1, a spin current source A having a pinned layer, and a spin current having a free layer. An absorption electrode B and a bias application electrode C including a nonmagnetic material are provided. As shown in the figure, a three-dimensional orthogonal coordinate system is set, the thickness direction of the semiconductor substrate 1 is the Z-axis direction, the application direction of the electric field -E is the X-axis direction, and the direction perpendicular to both the Z-axis and the X-axis is Y Axial direction.

  The spin current source A is an element that generates a spin current. When electrons are supplied from the current source J to the electrode pad PA1 made of a nonmagnetic metal, the spin current source A enters the semiconductor region immediately below the spin current source A. Spins are accumulated. The spin current source A includes a ferromagnetic layer PA2 whose magnetization direction is fixed, and a tunnel barrier layer (insulating layer) A3 interposed between the ferromagnetic layer PA2 and the semiconductor region. The polarity of the current source J may be reversed, but the polarity as shown in FIG. 2 is preferable because the spin accumulation can be increased.

  The spin current absorption electrode B includes a ferromagnetic layer PB2 capable of changing the direction of magnetization, and a tunnel barrier layer (insulating layer) B3 interposed between the ferromagnetic layer PB2 and the semiconductor region. The voltage V between the flow absorption electrode B and the detection electrode PB1 made of a nonmagnetic metal can be measured by a voltmeter V. This measurement is a non-local measurement. In addition, the code | symbol of the voltmeter V was made the same with the output voltage V measured by this for convenience.

  The bias application electrode C includes a metal layer PC made of a non-magnetic material and a tunnel barrier C3 interposed between the metal layer PC and the semiconductor region. The metal layer PC is made of a nonmagnetic metal such as Cu or Al. That is, the metal layer PC constituting the bias application electrode C is a nonmagnetic material. When the metal layer PC is made of a magnetic material, it is not preferable because the spin injection signal from the magnetic metal layer PC is superimposed as noise on the original spin signal. The bias application electrode C is positioned between the spin current source A and the spin current absorption electrode B in the semiconductor region, and a positive potential (Vb) is applied to the spin current source A from the bias power source Vb. Note that the sign of the bias power supply Vb is the same as the voltage Vb to be output from now on for convenience.

In the above-described element, MgO is suitable as a material for the tunnel barrier layers A3, B3, and C3. In addition, Al ₂ O ₃ , AlN, SiO ₂ , HfO ₂ , Zr ₂ O ₃ , Cr ₂ O ₃ , TiO ₂ , SrTiO ₃ , ZnO, MgAlO ₂ , or the like can be used. In the figure, the tunnel barrier layers A3, B3, and C3 are physically separated, but may be integrally formed.

  As the material of the nonmagnetic metal, Al or Cu can be used.

  As a semiconductor material, Si, Ge, GaAs, InAs, InSb, or the like can be used.

  As the ferromagnetic material, Fe, CoFe, NiFe, or the like can be used. The magnetization direction DMA of the ferromagnetic layer PA2 is parallel to the Y axis in the figure, and the magnetization direction DMB of the ferromagnetic layer PB2 is also parallel or antiparallel to the Y axis. These magnetization directions can be controlled by increasing the aspect ratio of each layer and changing the shape anisotropy. However, as the ferromagnetic layer PA2 in which the magnetization direction is fixed, the antiferromagnetism is used. By exchange-coupling the body to the ferromagnetic layer, the magnetization direction can be strongly fixed in one direction. Further, the magnetization directions DMA and DMB can be parallel or antiparallel to the X axis.

  A voltage Vb is applied between the spin current source A and the bias application electrode C. The voltage Vb is a voltage when the potential of the spin current source A is set to the ground potential. By applying the voltage Vb, an electric field -E is formed in the semiconductor region between the spin current reduction A and the bias application electrode C. Electrons placed in the electric field -E receive a force in a direction from the electrode A toward the electrode C.

2 is a cross-sectional view taken along the line II-II of the spin transport device shown in FIG.
A contact layer G for obtaining electrical contact is provided on each of the electrodes described above. That is, a contact layer G made of a metal such as Al is formed on the surfaces of the electrode pads PA1, PB1, the ferromagnetic layer PA1, the metal layer PC, and the ferromagnetic layer PB2.

  An electron current flows into the semiconductor region (spin accumulation region R1) immediately below the spin current source A via the electrode pad PA1 and the semiconductor region located on the surface of the semiconductor substrate 1, and spins of one polarity are tunnel barriers. The spin of the other polarity is absorbed in the ferromagnetic layer PA2 through the layer A3, and remains in the spin accumulation region R1. The spin accumulated in the spin accumulation region R1 drifts according to the electric field -E and reaches the spin accumulation region R2 of the semiconductor substrate 1.

  That is, since the positive potential Vb is applied to the bias application electrode C with respect to the spin current source A, the spin accumulated in the region R1 immediately below the spin current source A is an electric field formed by this potential Vb. In accordance with −E, it moves to just below the bias application electrode C. From just below the bias application electrode C, the spin leaks and diffuses and is absorbed into the spin current absorption electrode B through the tunnel barrier layer B3.

  At this time, the spin current absorption electrode B depends on the relative angle between the magnetization direction DMB of the spin current absorption electrode B and the magnetization direction DMA of the ferromagnetic layer PA2 (pinned layer) provided above the spin current source A. And the voltage V between the detection electrodes PB1 are different.

  In this spin transport device, the presence of the bias application electrode C allows it to be disposed relatively far from the spin current source A, so that a space for arranging a plurality of bias application electrodes C can be secured, and therefore By arranging the spin current absorption electrode B adjacent to each other, it becomes possible to realize a multi-channel type spin transport device. Even when the spin transport device is not a multi-channel type, the design freedom increases due to the increase in space, so that various devices can be integrated.

Note that the distance between the electrode PA1 and the electrode A is L0, the distance between the electrode A and the electrode C is L1, the distance between the electrode C and the electrode B is L2, and the distance between the electrode B and the electrode PB1. Let the distance be L3. In this example, the distance L0 is 10 μm, L1 is 30 μm, L2 is 500 nm, and L3 is 30 μm. The distance L1 is set longer than the intrinsic spin diffusion length λ _N when the electric field is zero. That is, in the present invention, the spin diffusion length λ _N is extended to a longer spin drift distance by applying a bias.

  FIG. 3 is a cross-sectional view of a spin current source A as an example.

  A pinned layer (ferromagnetic layer PA21, nonmagnetic layer PA22, ferromagnetic layer PA23) and an antiferromagnetic layer PA24 are sequentially formed on the semiconductor substrate 1 via a tunnel barrier layer A3. A ferrimagnetic layer is formed. As an example of the ferromagnetic layers PA21 and PA23, CoFe, Ru as the nonmagnetic layer PA22, and PtMn as the antiferromagnetic layer PA24 can be raised. Cu or the like can also be used as the nonmagnetic layer PA22. As antiferromagnetic materials, IrMn, RuRhMn, PtPdMn and the like are known.

  FIG. 4 is a perspective view of the spin transport device according to the second embodiment.

  In this spin transport device, the number of electrode groups including the above-described spin current absorption electrode B, detection electrode PB1, and bias application electrode C is plural, and three spin current channels are formed in this example. That is, when there are a plurality of these electrode groups, a multi-channel type spin conduction element can be formed, and therefore, a plurality of voltage detection sites (B, PB1, V) for a single spin source A. Can be provided, and device functions can be integrated.

  In the spin transport device of this example, the semiconductor region on one surface of the semiconductor substrate is isolated from the surroundings as channels CH0 to CH3. That is, the resistivity of the region INS outside the channels CH0 to CH3 is higher than the resistivity of the semiconductor regions in the channels CH0 to CH3. Preferably, the region INS can be an insulator region formed by joining a semiconductor. The semiconductor regions in the channels CH0 to CH3 may have a lower resistivity by increasing the impurity concentration than in the region INS. In these cases, since the spin current selectively flows in the semiconductor regions of the channels CH0 to CH3, the size of the flow path is limited and stable spin current can be transmitted.

  The channel CH0 is formed in the semiconductor region between the electrode PA1 and the electrode A, and the channels CH1 to CH3 are formed in the semiconductor region between the electrode A and the electrodes C, B, and PB1. The traveling paths of the spin current from the electrode A to the electrode C are set to be substantially equal. Even when the channels CH0 to CH3 are not formed, the spin transport device operates.

  Each bias application electrode C is provided with a bias power supply Vb for applying a potential Vb, as in the first embodiment, and whether or not a bias is applied is determined by the bias power supply Vb, the bias application electrode C, and the like. It is configured to be controlled by a switch SW interposed between the two. Each switch SW is selectively connected according to control signals φ1, φ2, and φ3 from the control circuit CONT. That is, by selectively connecting the switches SW, it is possible to suppress a decrease in the density of the spin current flowing from the single spin current source A toward each bias application electrode C. Thereby, a sufficient amount of spin current can be supplied to the target spin current absorption electrode B, and a precise voltage output V can be obtained. Of course, if the amount of spin current is sufficient to obtain a plurality of outputs V, the number of switches SW that are simultaneously turned on may be increased.

  Further, the spin transport device of the second embodiment has a ferromagnetic layer (free layer) PB2 in addition to a voltmeter (voltage detection means) V that detects a voltage between the spin current absorption electrode B and the detection electrode PB1. Is provided with a current source J2 for controlling the magnetization direction DMB (see FIGS. 5 and 6). This current source J2 can also be provided in the first embodiment.

  If the direction of positive and negative magnetization of the free layer is associated with 0, 1 information, the free layer stores information. The direction of this magnetization can be determined by the magnitude of the detected voltage V.

  FIG. 5 is a diagram illustrating the current source J2 that controls the direction of magnetization of the free layer.

  The current source J2 can change the direction of the current according to the control signal φs. Of course, the control signal φs can also instruct the current source J2 to stop supplying current (electron current). The auxiliary channel CHβ is continuous with the channel CH1 (CH2, CH3), and the auxiliary electrode F is formed. The auxiliary electrode F includes a tunnel barrier layer F3 formed on the semiconductor substrate 1 in the auxiliary channel CHβ, and a ferromagnetic layer PF2 formed on the tunnel barrier layer F3. The ferromagnetic layer PF2 is a pinned layer. The direction of magnetization is fixed.

  In other words, the structure of the auxiliary electrode F is the same as that of the spin current source (electrode) A, and its constituent material is also the same as that of the spin current source A. The electron flow flowing out in one direction from the current source J2 enters the auxiliary electrode F, passes through the channel CHβ of the semiconductor substrate 1, and enters the ferromagnetic layer (free layer) PB2 made of CoFe or the like of the electrode B. The magnetization direction DMB of the free layer PB2 is aligned with the direction of the injected spin. Therefore, in this case, the magnetization direction DMF of the pinned layer is equal to the magnetization direction DMB of the free layer.

  When an electron current is output in the reverse direction from the current source J2, this is because some of the spins enter the pinned layer of the electrode F via the electrode B, and spins in a direction not entering the pinned layer are in the free layer. To remain. That is, in this case, the magnetization direction DMF of the pinned layer is opposite to the magnetization direction DMB of the free layer. Thus, by using the spin injection magnetization reversal phenomenon to the free layer, the magnetization direction DMB of the free layer can be controlled using the electron flow output from the current source J2.

  FIG. 6 is a diagram for explaining another type of current source J2 for controlling the magnetization direction of the free layer.

  The current source J2 can change the direction of the current according to the control signal φs. Of course, the control signal φs can also instruct the current source J2 to stop supplying current (electron current). When a current is output from the current source J2, an electric field E2 is formed around the wiring W that is the current transmission path. The wiring E2 extends through the vicinity of the free layer, and is arranged so that the direction of the peripheral electric field E2 coincides with the magnetization direction DMB. The direction of magnetization DMB of the free layer is aligned with the direction of the electric field E2. By changing the direction of the current, the magnetization direction DMB can be changed.

  The control of the magnetization direction DMB of the free layer described above corresponds to information writing. That is, the magnetization direction DMB is associated with the information of 0, 1 and written in the free layer, the switch SW is turned on to supply the spin current to the free layer, and the voltage V at this time is measured. The written information can be read out.

  FIG. 7 is a circuit diagram of a magnetic memory using the above-described spin transport element.

  Assuming that the above-described spin accumulation region R1 is the spin current source SS and the semiconductor region between the spin accumulation region R1 and the above-described bias application electrode C is denoted by the symbol DRIFT, when the bias voltage Vb is applied by turning on the switch SW, the semiconductor An electric field -E is formed in the region DRIFT, the spin current drifts from the spin current source SS toward the bias application electrode C, flows, and the spin current diffuses from here and flows into the spin current absorption electrode B The voltage V corresponding to the direction of magnetization in the spin current absorption electrode B is measured between the electrode B and the electrode PB1, and the measured output voltage V is input to the control circuit CONT.

  The current source J2 is controlled in accordance with the control signal φs from the control circuit CONT, and the magnetization direction (DMB: see FIGS. 5 and 6) of the spin current absorption electrode B is controlled. Further, whether or not the spin current is supplied to the bias application electrode C is controlled by inputting the control signals φ1 to φ3 from the control circuit CONT to the switch SW. Thus, in the magnetic memory, information is written by the control signal φs, and information written by the control signals φ1 to φ3 is read.

  FIG. 8 is a graph showing the state of spin diffusion, where the horizontal axis indicates the position and the vertical axis indicates the spin concentration.

  The effective spin diffusion length includes diffusion and drift distance about the center of gravity at the same time. By using this, spin conduction in the state where an electric field is applied can be performed by the conventional design method of metal elements. it can. When the electric field is zero, as time t elapses, the spin diffuses as shown in (a), and the spin concentration at the center of the normal distribution centering on the initial position decreases.

  On the other hand, in the presence of an electric field, similarly, the spin concentration at the center of the normal distribution decreases, but at the same time, the center position of the normal distribution moves by a distance x obtained by multiplying the velocity v by the time t. The spin diffusion length is extended (b).

  In order to specifically investigate the state of drift, the distribution of spin accumulation in the channel was calculated using the above-described effective spin diffusion length. FIG. 9 is a perspective view of a spin transport element used for calculation. A spin current source A was disposed at one end of the semiconductor substrate 1, a bias application electrode B was disposed at the other end, a distance between these electrodes was L1, and a bias voltage Vb was applied therebetween. As a result, an electric field -E is generated inside the semiconductor substrate 1, and a spin accumulation region R2 is formed immediately below the electrode B. The electrode width was 20 μm, and the thickness of the semiconductor substrate 1 contributing to spin conduction was 100 nm.

  FIG. 10 is a chart showing the relationship between current and spin diffusion length.

The semiconductor substrate 1 is made of Si. The resistivity ρ of Si is 1 × 10 ⁻³ Ωcm, the spin resistance is 40Ω, the intrinsic spin diffusion length λ _N is 2 μm, and the resistance ratio between the ferromagnetic material and Si (R _F / R _Si ). = 10, and the electron flow flows through the tunnel barrier. The calculation was performed with the spin polarizability set to 0.5. Such ferromagnetic materials include Fe and NiFe.

Spin diffusion length .lambda.d ([mu] m) in accordance with current (.mu.A) (voltage Vb) is increased, Nari longer, in 1mA (1000μA), will exceed 10 times the intrinsic spin diffusion length lambda _N when the non-bias . The reverse spin diffusion length λu (μm) decreases as the current (μA) (voltage Vb) increases.

  Note that the metal layer PC constituting the bias application electrode C described above needs to be a non-magnetic material as a practical application. That is, when the metal layer PC of the bias application electrode is made of a ferromagnetic material, spins are also injected into the semiconductor from this electrode, and spins accumulated by drifting are separately injected from the ferromagnetic metal layer PC. Spins are superimposed. This is not preferable because it causes noise with respect to the original signal. A comparative example in which the metal layer PC is made of a ferromagnetic material will be described with reference to FIGS.

  FIG. 11 is a graph showing the relationship between the drift distance L1 (μm) and the spin accumulation amount Vs (mV) according to Comparative Example A.

  In Comparative Example A, it is assumed that the metal layer PC (see FIG. 2) constituting the bias application electrode C is a ferromagnetic material, and the magnetization directions of the pinned layer and the metal layer PC are parallel. When L1 is 1 μm, the bias voltage Vb is changed to 0.5 mV, 1 mV, 2.5 mV, 3.5 mV, and 5 mV, and in each case, the current supplied from the current source J (see FIG. 2) is 100 μA and 200 μA. , 500 μA, 700 μA, and 1 mA. The sign of the output Vs changes when L1 is 0 μm and 1 μm. This means that a larger noise than the original signal is generated by superimposing spins injected from the ferromagnetic metal layer PC. The original signal has a positive sign. The sign of the noise component and that of the signal component are opposite, and the sign changes to a negative value indicates that the noise component is larger than the signal component.

  FIG. 12 is a graph showing the relationship between the drift distance L1 (μm) and the spin accumulation amount Vs (mV) according to Comparative Example B (when L1 is 100 μm), and FIG. 13 shows the drift distance L1 (μm). It is a graph (L1 is an enlarged graph around 100 μm) showing the relationship with the spin accumulation amount (mV).

  Similarly to the above, in Comparative Example B, it is assumed that the metal layer PC (see FIG. 2) constituting the bias application electrode C is a ferromagnetic material, and the magnetization directions of the pinned layer and the metal layer PC are parallel. When L1 is 100 μm, the bias voltage Vb is changed to 50 mV, 100 mV, 250 mV, 350 mV, and 500 mV. In each case, the current supplied from the current source J (see FIG. 2) is 100 μA, 200 μA, 500 μA, 700 μA, 1 mA. And When L1 is 0 μm and 100 μm, the sign of the output Vs changes, indicating that the noise component is larger than the signal component as described above.

  FIG. 14 is a graph showing the relationship between the drift distance L1 (μm) and the spin accumulation amount Vs (mV) according to Example A.

  In Example A, it is assumed that the metal layer PC (see FIG. 2) constituting the bias application electrode C is a nonmagnetic material (Al). When L1 is 1 μm, the bias voltage Vb is changed to 0.5 mV, 1 mV, 2.5 mV, 3.5 mV, and 5 mV. In each case, the current supplied from the current source J (see FIG. 2) is 100 μA and 200 μA. , 500 μA, 700 μA, and 1 mA, it can be seen that spin is accumulated even at a position where L1 is 1 μm. Of course, the larger the bias voltage Vb, the larger the spin accumulation amount. Since the metal layer PC is made of a non-magnetic material, there is no spin injection noise from the metal layer PC, and purely drifting spins are accumulated as signal components at the travel end point. Therefore, there is no change in the sign of the output Vs, and it is always positive, and has a signal transmission characteristic superior to that of Comparative Examples A and B shown in FIGS.

  FIG. 15 is a graph showing the relationship between the drift distance L1 (μm) and the spin accumulation amount Vs (mV) according to Example B (when L1 is 30 μm), and FIG. 16 shows the drift distance L1 (μm). It is a graph (L1 is an expansion graph around 30 μm) showing the relationship with the spin accumulation amount (mV).

  In Example B, it is assumed that the metal layer PC (see FIG. 2) constituting the bias application electrode C is a nonmagnetic material (Al). When L1 is 30 μm, the bias voltage Vb is changed to 15 mV, 30 mV, 75 mV, 105 mV, and 150 mV. In each case, the current supplied from the current source J (see FIG. 2) is 100 μA, 200 μA, 500 μA, 700 μA, 1 mA. Then, it can be seen that spin is accumulated even at a position where L1 is 30 μm. In FIG. 16, Vs is substantially zero when the current is 500 μA or less. Of course, the larger the bias voltage Vb, the larger the spin accumulation amount. Since the metal layer PC is made of a non-magnetic material, there is no spin injection noise from the metal layer PC, and purely drifting spins are accumulated as signal components at the travel end point. Therefore, there is no change in the sign of the output Vs, and it is always positive, and has a signal transmission characteristic superior to that of Comparative Examples A and B shown in FIGS.

  FIG. 17 is a graph showing the relationship between the drift distance L1 (μm) and the spin accumulation amount Vs (mV) according to Example C (when L1 is 100 μm), and FIG. 18 shows the drift distance L1 (μm). It is a graph (L1 is an enlarged graph around 100 μm) showing the relationship with the spin accumulation amount (mV).

  In Example C, it is assumed that the metal layer PC (see FIG. 2) constituting the bias application electrode C is a nonmagnetic material (Al). When L1 is 100 μm, the bias voltage Vb is changed to 50 mV, 100 mV, 250 mV, 350 mV, and 500 mV, and in each case, the current supplied from the current source J (see FIG. 2) is 100 μA, 200 μA, 500 μA, 700 μA, 1 mA. Then, it can be seen that spin is accumulated even at a position where L1 is 100 μm. In FIG. 18, Vs is substantially zero when the current is 500 μA or less. Of course, the larger the bias voltage Vb, the larger the spin accumulation amount. Since the metal layer PC is made of a non-magnetic material, there is no spin injection noise from the metal layer PC, and purely drifting spins are accumulated as signal components at the travel end point. Therefore, there is no change in the sign of the output Vs, and it is always positive, and has a signal transmission characteristic superior to that of Comparative Examples A and B shown in FIGS.

  As described above, when a nonmagnetic material is used for the metal layer PC (see FIG. 2) (Examples A to C), all spin accumulation is caused by drift, and a ferromagnetic material is used (Comparative Example A). , B) is less mixed with noise. Hereinafter, actual measurement was performed.

  FIG. 19 is a chart showing the relationship between the bias voltage Vb (V) and the output voltage V (mV) for each drift distance L1, and FIG. 20 is a drift distance between the bias voltage Vb (V) and the output voltage V (mV). It is a graph which shows the relationship for every L1.

In this example, the spin transport device of the second embodiment is used, and in Example 1, 5 × 10 ¹⁹ / cm ³ of phosphorus (P) is doped on the entire surface of Si to contribute n-type Si to electrical conduction. In Example 2, phosphorus (P) was doped in the same manner only in CH0 to CH3. The thickness of Si was 100 nm. The dimensions are as described above. This element was produced by photolithography and ion milling. The tunnel barrier layer was made of MgO, the ferromagnetic layer was made of Fe, and the electrodes made of the nonmagnetic layer were all made of Al, and were in ohmic contact with the Si channel. The drift distance L1 was changed to 1 μm, 30 μm, and 100 μm.

  In Example 1, an output voltage V was measured when 1 mA of electrons was injected from the current source J, spin accumulation of about 1 mV was input, and a single switch SW was turned on. An increase in output was observed with the bias voltage Vb, and spin conduction was confirmed over a long distance of 2 μm or more. Practically, an output voltage V of 0.001 mV or higher is preferable, but even at a distance of 100 μm, it is satisfied with a voltage of 5 V or higher.

  When the switches SW of the bias voltage Vb were turned on at the same time, the output decreased almost in inverse proportion to the number of switches SW whose outputs were turned on, but detection was sufficiently possible. It can also be compensated by increasing the injection current. The bias voltage Vb could be output as a voltage other than a direct current voltage (DC), that is, an alternating voltage or a pulse voltage.

  In Example 2, the bias voltage dependency of the output V was almost the same as that of Example 1, but the magnitude of the output increased by about 30%. This can be interpreted as that the electric field lines extending from the injection electrode are suppressed and the electric field is increased.

  FIG. 21 is a cross-sectional view of a spin transport device according to Comparative Example 1.

  The spin transport device according to Comparative Example 1 is obtained by removing the bias application electrode from the spin transport device according to the above-described embodiment. In the spin transport device according to Comparative Example 1, since there is no bias application electrode, the spin current from the spin accumulation region R can diffuse only a short distance, and thus a multi-channel spin transport device is configured. It is difficult. When the spin current absorbing electrode B is separated from the spin current source A by 2 μm or more, the output V is zero.

  Note that the above-described spin transport element has advantages in applications such as spin transistors and magnetic sensors using pure spin current conduction that are conventionally known. That is, by using drift, the installation location of the detection terminal can be separated from the injection electrode by more than the spin diffusion length, and detection can be performed at a plurality of locations. In addition, since drift is a faster process than the diffusion process, it is suitable for higher frequencies, and has the advantage of being able to conduct signals on the order of GHz, and can also send spin information of short pulse signals. The present invention can be used for a spin sensor used in a magnetic sensor, a spin information processing device, a spin current circuit, or an MRAM using spin current.

  The present invention can be used for a spin transport device.

DESCRIPTION OF SYMBOLS 1 ... Semiconductor substrate, A ... Spin current source, B ... Spin current absorption electrode, PB1 ... Detection electrode, C ... Bias application electrode.

Claims (5)

A semiconductor region;
A spin current source having a pinned layer provided in the semiconductor region;
A spin current absorbing electrode having a free layer provided in the semiconductor region;
A detection electrode provided in the semiconductor region for measuring a voltage between the spin current absorption electrode;
A bias application electrode including a non-magnetic material that is positioned between the spin current source and the spin current absorption electrode in the semiconductor region and to which a positive potential is applied to the spin current source;
A spin transport device comprising:   2. The spin transport device according to claim 1, wherein the number of electrode groups including the spin current absorption electrode, the detection electrode, and the bias application electrode is plural. A bias power source for applying the potential to each of the bias application electrodes;
A switch interposed between the bias power source and each bias application electrode;
With
The spin transport device according to claim 2, wherein each of the switches is selectively connected. A current source for controlling the magnetization direction of the free layer;
Voltage detection means for detecting a voltage between the spin current absorption electrode and the detection electrode;
The spin transport device according to claim 1, wherein the spin transport device is provided. A semiconductor substrate including the semiconductor region;
5. The spin transport device according to claim 1, wherein a resistivity of a region outside the semiconductor region in the semiconductor substrate is higher than that of the semiconductor region.

Images