Non-volatile field-effect transistor based on a two-dimensional electron gas
A non-volatile field effect transistor with a two-dimensional electron gas and a high-concentration reducing layer addresses endurance and contrast issues, achieving efficient non-volatile operation and reduced power consumption.
Patent Information
- Application Number
- FR2022000906
- Authority / Receiving Office
- FR · FR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-02-02
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2042-02-02
AI Technical Summary
Existing non-volatile field effect transistors, such as Fe-FETs, suffer from limited endurance and poor contrast between the on and off states, leading to high power consumption and inefficiency.
A non-volatile field effect transistor with a two-dimensional gas of electrons is developed, comprising a reducing layer with high metallic element concentration and a subset of remanent states, allowing non-volatile modulation of conductivity through a ferroelectric, trapped charge, or ion migration effect, eliminating the need for a ferroelectric element.
The transistor achieves improved endurance and enhanced contrast between on and off states, reducing power consumption and enabling memory and logic functions without continuous power supply.
Smart Images

Figure 00000013_0000 
Figure 00000014_0000 
Figure 00000015_0000
Abstract
Description
Title of the invention: Non-volatile field effect transistor based on two-dimensional electron gas
[0001] The present invention relates to a non-volatile two-dimensional electron gas field effect transistor.
[0002] A field-effect transistor is often referred to by the acronym FET, which refers to the corresponding English term for "Field-Effect Transistor". Such a transistor is often used as a switch or amplifier.
[0003] A field effect transistor is a unipolar device with three terminals (drain, source and gate) based on the action of an electric field on the conductivity of a channel located between the drain and the source. In the channel, the carriers, electrons or holes, are free to move between the source and the drain under the action of a voltage applied between these two terminals. The control of the conductance of the channel is done by applying a voltage to the gate. Many types of field effect transistors exist depending on the nature of the carrier (electron or hole) or the nature of the electrical control exerted on the channel.
[0004] Field effect transistors are currently constructed using various semiconductor materials, including mainly monocrystalline silicon-based materials. However, other materials are used, such as amorphous silicon, polycrystalline silicon, gallium arsenide or gallium nitride.
[0005] In a FET transistor, a power supply is used to maintain the on state (i.e., non-zero channel conductivity) or off state (zero channel conductivity) of the transistor, since this state depends on the voltage applied to the gate. This state is volatile: if the power supply is switched off, the voltage applied to the gate necessarily becomes zero, and the state is not retained. Applying a voltage constantly to maintain the memory state results in energy consumption to maintain the state of the transistor. This consumption is said to be static because it is linked to the appearance of leakage currents, and not to the dynamic process of writing or reading the memory.
[0006] The creation of non-volatile field effect transistors, therefore capable of maintaining the state of the transistor in the absence of a power supply, makes it possible to eliminate static energy consumption, and therefore to reduce energy consumption. Non-volatility can also make it possible to envisage a transistor with additional functionalities, such as a memory function or a logic calculation function.
[0007] Several non-volatile field effect transistors have been proposed, based on Semiconductor transistors. In particular, work is focused on the development of the ferroelectric field-effect transistor, more often referred to by the abbreviation Fe-FET, which refers to the corresponding English term for "Ferroelectric Field-Effect Transistor". The Fe-FET includes a ferroelectric element between the gate and the drain-source channel to maintain the on or off state in the absence of a power supply.
[0008] Known Fe-FET transistors suffer in particular from limited endurance and contrast between the on and off state (sometimes referred to as "ON / OFF contrast").
[0009] There is therefore a need for a non-volatile transistor having better endurance and improved contrast between the on state and the off state.
[0010] For this purpose, the description describes a non-volatile field effect transistor, the transistor comprising a first electrode called the gate electrode having a first contact, a second electrode called the source, the source comprising a second contact, a third electrode called the drain, the drain comprising a third contact, a channel between the drain and the source, the channel consisting of a two-dimensional gas of electrons, a remanent state subassembly, the remanent state subassembly having at least two electrically controllable remanent states, the remanent state subassembly being in contact with the channel, the remanent state subassembly comprising at least one oxide layer, and a reducing layer made of at least one reducing material of metal type, each reducing material of metal type having an atomic concentration of metallic elements greater than or equal to 50%, these metallic elements being chosen from columns 3d,4d, 5d, 5f of the periodic table. The first electrode is in contact with the remanent state subassembly and the application of a voltage between the first contact and one of the second contact and the third contact results in a non-volatile modulation of the conductivity of the two-dimensional gas constituting the channel.
[0011] The reducing layer makes it possible to create the two-dimensional gas of electrons forming the channel at the interface between the reducing layer and the subassembly with remanent states.
[0012] According to particular embodiments, the non-volatile transistor has one or more of the following characteristics, taken in isolation or in all technically possible combinations:
[0013] - the reducing layer is a monolayer of reducing material of metal type or a multilayer of metal-type reducing materials, this or these reducing materials each having an atomic concentration of metallic elements greater than or equal to 50%, this or these metallic elements being chosen from columns 3d, 4d, 5d, 5f of the periodic table.
[0014] - the remanent state subassembly comprises at least one layer or multilayer non-volatile dielectric with electrical control, by ferroelectric effect, by trapped charge effect, by ion migration effect or by combination of several of these effects.
[0015] - each metal-type reducing material has an atomic concentration in metallic elements greater than or equal to 80%.
[0016] - each metallic element is chosen from the list consisting of aluminum, tantalum, yttrium, magnesium and ruthenium.
[0017] - the reducing layer has a thickness less than or equal to 15 nanometers and preferably less than or equal to 10 nanometers.
[0018] - the transistor comprises a substrate on which rests the drain, the source and the layer reducing layer, the reducing layer, the channel and the remanent state subassembly forming in this order a stack, and the first electrode rests on the remanent state subassembly.
[0019] - the remanent state subassembly rests on the first electrode, the drain, the channel and source lie on the remanent state subset, and the reducing layer is arranged on the channel.
[0020] - the non-volatile transistor further comprises a protective layer on the layer reductive.
[0021] - the remanent state subset comprises a layer chosen from a layer of oxide in perovskite oxides, an oxide layer in (Hfi xZrx)O2, x varying between 0 and 1, possibly doped, and a layer in poly(vinylidene fluoride).
[0022] - the transistor comprises a reading unit, the reading unit comprising a sub- unit of applying a voltage between the drain and the source and a sub-unit of measuring the current between the drain and the source.
[0023] - the source and the drain are in contact with the ends of the channel.
[0024] Characteristics and advantages of the invention will appear on reading the description which follows, given solely by way of non-limiting example, and made with reference to the appended drawings, in which:
[0025] - [Fig.l] [Fig.l] is a schematic representation of an example of a non-transistor volatile two-dimensional electron gas field effect,
[0026] - [Fig.2] [Fig.2] is a schematic representation of another example of non-volatile two-dimensional electron gas field effect transistor,
[0027] - [Fig.3] [Fig.3] is a graph showing the dependence of the resistance between two contacts depending on an applied voltage,
[0028] - [Fig.4] [Fig.4] is a graph illustrating the current-voltage characteristic of a example of a transistor in the on and off states, and
[0029] - [Fig.5] [Fig.5] is a schematic representation of an example of a transistor with its reading and writing units.
[0030] A transistor 10 is illustrated in Figures 1 and 2.
[0031] Transistor 10 is a non-volatile field effect transistor.
[0032] The transistor 10 comprises three electrodes, a first electrode called the gate electrode 12 comprising a first contact C1, a second electrode 14 called the source 14 and comprising a second contact C2 and a third electrode 16 called the drain 16 and comprising a third contact C3.
[0033] These electrodes are advantageously metallic in nature, or made of heavily doped semiconductor material.
[0034] The transistor 10 also comprises a remanent state subassembly 18, a channel 20 constituted by a two-dimensional gas of electrons and located between the drain 16 and the source 14 and a reducing layer 22.
[0035] As visible in [Fig.5], the transistor 10 also comprises a reading unit 40 and a writing unit 42.
[0036] The reading unit 40 also comprises a voltage application sub-unit 44 and a current measurement sub-unit 46.
[0037] Two distinct arrangements of the layers of the transistor 10 are proposed respectively in [Fig.l] and [Fig.2].
[0038] In the following, the layers of a stack are layers stacked in a stacking direction. Relative notions of bottom and top are also defined with respect to the stacking direction. A layer is located lower than another layer if it is lower in the representation on the sheet of [Fig.l].
[0039] In the example of [Fig.l], the non-volatile transistor 10 comprises a substrate 24 on which rests the drain 16, the source 14 and a stack 26 of layers, comprising the reducing layer 22, the channel 20 and the remanent state subassembly 18, stacked in this order when the stack 26 is traversed from the bottom to the top (the bottom corresponding to the substrate in this case).
[0040] Finally, the thickness of a layer is defined as the dimension along the stacking direction of the layer, i.e. the distance between its two faces.
[0041] The gate electrode 12 is in contact with the remanent state subassembly 18.
[0042] The order of the layers of the stack 26 can be reversed as shown in [Fig.2]
[0043] In the case of [Fig.2], the remanent state subassembly 18 rests on the gate electrode 12. The drain 16, the channel 20 and the source 14 rest on the remanent state subassembly 18.
[0044] The channel 20 and the reducing layer 22 form in this order a stack 28 by traversing the stack 28 from the bottom to the top.
[0045] Furthermore, optionally, the stack 28 comprises a protective layer 30 resting on the reducing layer 22.
[0046] The protective layer 30 is made of insulating materials, characterized by zero or almost zero conductance, and making it possible to protect the stack 28, in particular from oxidation.
[0047] The protective layer 30 is, for example, made of SiO2.
[0048] In both cases illustrated, the source 14 and the drain 16 are in contact with a respective end of channel 20.
[0049] The remanent state subassembly 18 has at least two electrically controllable remanent states. The electrical control is obtained here by applying a voltage between the first contact C1 and the second or third contacts C2 or C3.
[0050] In other words, the remanent state subset 18 is characterized by a non-linear relationship between the applied voltage and the apparent stored charge following a hysteresis cycle, and giving rise to at least two remanent states.
[0051] For this, the remanent state subassembly 18 is a dielectric layer or a stack of non-volatile dielectric layers with electrical control by ferroelectric effect, by trapped charge effect, by ion migration effect or by combination of several of these effects. The remanent states correspond to variations in the ferroelectric polarization, in the quantity of trapped charges, or in the positions of the ions, or combinations of these variations.
[0052] More precisely, the subassembly comprises at least one oxide layer made of a material chosen from perovskite oxides, or oxides based on HfO2, possibly doped with other elements, such as (Hfi xZrx)O2 or (Hfi xGax)O2 (x varying between 0 and 1), or their alloys, or poly(vinylidene fluoride).
[0053] Perovskite oxides have an ABO3 type structure where A and B are cations. BaTiO3, PZT (i.e. PbZri xTixO3 with x varying between 0 and 1), PMN-PT (i.e. [lx]Pb(Mgi / 3Nb2 / 3)O3 - xPbTiO3 with x varying between 0 and 1), BiFeO3 (optionally doped, for example with rare earths on the Bi site, or with Mn on the Fe site), SrTiO3 (optionally doped), KTiO3 (optionally doped), Pr0j7Ca0j3MnO3 (optionally doped) or YMnO3 (optionally doped) are examples of such materials.
[0054] The use of the dielectric material to produce the remanent state subassembly 18 makes it possible to electrically control the conductivity of the channel 20 in a non-volatile manner.
[0055] In each of the preceding examples, the existence of the remanent states comes from a ferroelectric effect, a trapped charge effect, an ion migration effect or a combination of several of these effects. In practice, the predominant effect depends on the materials and the deposition conditions of the layers constituting the remanent state subassembly 18. According to the example described, the coercive electric field of the dielectric element and its thickness are sufficiently weak so that the writing device 14 can write the remanent states at voltages compatible with microelectronic technologies, i.e. voltages below 10 Volts (<10 V). A thickness of less than 100 nm and advantageously less than 50 nm in the aforementioned materials makes it possible to obtain such properties. The remanent state subassembly 18 is also cycling-resistant, typically capable of withstanding at least 104 cycles.
[0056] The channel 20 is constituted by a two-dimensional gas of electrons, that is to say a confined gas of electrons which forms at the interface between two layers, whereas it does not exist in the volume of the materials taken separately.
[0057] The channel 20 is therefore an electron gas forming at the interface between the remanent state subassembly 18 and the reducing layer 22. It has a high carrier density (typically greater than or equal to 1010 cm2). The confinement is such that it can be considered that this gas is two-dimensional, only the vicinity of the interface being conductive.
[0058] The electron gas of the channel 20 being created at the interface between the remanent state subassembly 18 and the reducing layer 22, its properties and in particular its resistance depend directly on the state of the remanent state subassembly 18. The resistance of the two-dimensional electron gas and therefore of the channel 20 is therefore electrically modulated in a non-volatile manner according to the state of the remanent state subassembly 18. It is thus possible to modulate the resistance of the channel 20 by choosing the remanent state of the remanent state subassembly 18.
[0059] The reducing layer 22 makes it possible to create the two-dimensional gas of electrons forming the channel at the interface between the reducing layer 22 and the remanent state subassembly 18.
[0060] The reducing layer 22 is made of a reducing material having an atomic concentration of metallic elements greater than or equal to 50%, preferably greater than or equal to 80%.
[0061] By definition, atomic concentration is the ratio of the number of atoms of metallic elements to the total number of atoms.
[0062] In this sense, the reducing layer 22 is a reducing layer that can be described as metal type.
[0063] Each metallic element of the reducing layer 22 is chosen from columns 3d, 4d, 5d, 5f of the periodic table, or a combination of metals chosen from the same columns of the periodic table.
[0064] As a particular example, each metallic element is chosen from Al, Ta, Ru, Pt, W, Ir, Mo, Ti, Y, Au, or an alloy of these elements such as PtW.
[0065] Alternatively, the reducing layer 22 is a multilayer. In this case, each layer of said multilayer has, again, an atomic concentration of elements me metal greater than or equal to 50%, preferably greater than or equal to 80%.
[0066] The reducing layer 22 allows the formation of the two-dimensional electron gas in contact with the remanent state subassembly 18. In this sense, the channel 20 formed by a two-dimensional electron gas is very different from the channel of a conventional field effect transistor, in which there is no reducing layer. Also, in the channel of a conventional field effect transistor, the carriers appear reversibly in the semiconductor material (generally doped Si), by applying a gate voltage creating an electrostatic field.
[0067] The reducing layer 22 has a thickness less than or equal to 15 nanometers (nm).
[0068] This allows the reducing layer 22, partially or completely oxidized in contact with the residual state subassembly 18, to have a lower conductivity than that of the two-dimensional electron gas. The thickness of the reducing layer 22 is preferably chosen so that it is completely oxidized by contact with the residual state subassembly 18, so that its conductivity is as low as possible.
[0069] As indicated previously, the source 14 and the drain 16 are in contact with a respective end of the channel 20, and thus electrically connected to each other via this channel 20.
[0070] Each of the electrodes 12, 14 and 16 are preferably made of doped metallic or semi-conductor conductive material, so as to have high conductivity.
[0071] The operation of transistor 10 is now described with reference to FIGS. 3 and 4.
[0072] In a first example of operation, the remanent state subassembly 18 has two states denoted A and B.
[0073] This appears in particular in [Fig. 3], which shows the dependence of the resistance between the second contact C2 and the third contact C3 as a function of the voltage applied between the first contact C1 and the second contact C2 (gate voltage Vg). It is clearly observed that the variation in conductivity of the channel 20 as a function of the gate voltage exhibits a hysteresis, corresponding to the 2 remanent states A and B of the subassembly.
[0074] According to this first example of operation, writing is carried out by the writing unit by applying a voltage between the first contact C1 and the second contact C2.
[0075] Writing could also be achieved by replacing the second contact C2 with the third contact C3.
[0076] From a physical point of view, the conductivity of channel 20 is here modulated in a manner non-volatile depending on the remanent state of the remanent state subassembly 18 by applying a voltage between the first contact Cl and the second contact C2.
[0077] In this operating example, the application of a negative voltage between the first contact C1 and the second contact C2 initializes the remanent state subassembly 18 in the remanent state A, resulting in a low conductivity of the channel 20. Conversely, the application of a positive voltage induces a high conductivity of the channel 20, corresponding to the remanent state B of the remanent state subassembly 18.
[0078] Depending on the material used, the sign of the voltages giving the high and low resistance remanent states can be reversed. Thus, it may be the application of a positive voltage between the first contact C1 and the second contact C2 which results in a remanent state with low conductivity of the channel 20. In this case, it is the application of a negative voltage which will induce a remanent state corresponding to a high conductivity of the channel 20.
[0079] In this first operating example, the reading unit 40 non-destructively reads the remanent state by applying a voltage between the second contact C2 and the third contact C3 using the voltage application sub-unit 44 and by measuring the resulting current between these two contacts C2 and C3 using the measurement sub-unit 46.
[0080] The current-voltage characteristic between the second contact C2 and the third contact C3 which corresponds to the output characteristic of the transistor 10, is thus controlled by the remanent state of the remanent state subassembly 18.
[0081] This characteristic is shown in [Fig.4].
[0082] In a second operating example, the voltage application subunit 44 applies a voltage between the first contact C1 and the second contact C2 so as to obtain any state of the hysteresis cycle. The conductivity of the channel 20 can take more than two states, and thus serve to encode analog, non-binary information, such as a synaptic weight in a neural network for artificial intelligence.
[0083] The transistor 10 which has just been described is a non-volatile field effect transistor based on a channel 20 of two-dimensional electron gas at the interface between an oxide layer and a reducing layer 22 of metal type. It is the association of an oxide layer of the remanent state subassembly 18 with the reducing layer 22 which makes it possible to achieve non-volatile control of the conductivity of the channel 20.
[0084] This avoids using an additional ferroelectric element above the structure, since the combination of the remanent state subassembly 18 and the reducing layer 22 makes it possible to create both the two-dimensional electron gas and the remanent states.
[0085] Furthermore, the presence of an assembly formed by the oxide layer, the bidi gas electron gas and the reducing layer 22 makes it possible, compared to state-of-the-art non-volatile field effect transistors, to optimize the conductivity of the channel 20, to obtain better endurance, and an improved contrast between the on state and the off state. The performance obtained thanks to the channel 20 in two-dimensional electron gas, created by using a reducing layer formed according to conventional microelectronic processes such as cathode sputtering, makes it possible to make the transistor 10 suitable for industrial exploitation.
Claims
Claims
1. Non-volatile field effect transistor (10), the transistor (10) comprising: - a first electrode (12) called the gate electrode having a first contact (Cl), - a second electrode (14) called the source (14), the source (14) having a second contact (C2), - a third electrode (16) called the drain (16), the drain (16) having a third contact (C3), - a channel (20) between the drain (16) and the source (14), the channel (20) being made of a two-dimensional gas of electrons, - a remanent state subassembly (18), the remanent state subassembly (18) having at least two electrically controllable remanent states, the remanent state subassembly (18) being in contact with the channel (20), the remanent state subassembly (18) having at least one oxide layer,and - a reducing layer (22) for creating the two-dimensional gas constituting the channel (20) at the interface between the reducing layer (22) and the residual state subassembly (18), the reducing layer (22) being made of at least one reducing material of metal type, each reducing material of metal type having an atomic concentration of metallic elements greater than or equal to 50%, the first electrode (12) being in contact with the residual state subassembly (18) and the application of a voltage between the first contact (C1) and one of the second contact (C2) and the third contact (C3) causing a non-volatile modulation of the conductivity of the two-dimensional gas constituting the channel (20).,
2. A non-volatile transistor according to claim 1, wherein the reducing layer (22) is a monolayer of metal-type reducing material or a multilayer of metal-type reducing materials, this or these reducing materials each having an atomic concentration of metallic elements greater than or equal to 50%.
3. Non-volatile transistor according to claim 1 or 2, in which the remanent state subassembly (18) comprises at least one non-volatile dielectric layer or multilayer with electrical control, by ferroelectric effect, by trapped charge effect, by ion migration effect or by combination of several of these effects.
4. A non-volatile transistor (10) according to any one of claims 1 to 3, wherein each metal-type reducing material has an atomic concentration of metallic elements greater than or equal to 80%.
5. A non-volatile transistor according to any one of claims 1 to 4, wherein each metallic element is selected from the list consisting of aluminum, tantalum, yttrium, magnesium and ruthenium.
6. Non-volatile transistor according to any one of claims 1 to 5, wherein the reducing layer (22) has a thickness less than or equal to 15 nanometers and preferably less than or equal to 10 nanometers.
7. Non-volatile transistor according to any one of claims 1 to 6, in which: - the transistor (10) comprises a substrate (24) on which rest the drain (16), the source (14) and the reducing layer (22), - the reducing layer (22), the channel (20) and the remanent state subassembly (18) forming in this order a stack, and - the first electrode (12) rests on the remanent state subassembly (18).
8. Non-volatile transistor according to any one of claims 1 to 6, wherein: - the remanent state subassembly (18) rests on the first electrode (12), - the drain (16), the channel (20) and the source (14) rest on the remanent state subassembly (18), and - the reducing layer (22) is arranged on the channel (20).
9. The non-volatile transistor of claim 8, further comprising a protective layer (30) on the reducing layer (22).
10. Non-volatile transistor according to any one of claims 1 to 9, in which the remanent state subassembly (18) comprises a layer chosen from: - an oxide layer of perovskite oxides, - an oxide layer of (Hfi xZrx)O2, x varying between 0 and 1, possibly doped, and - a layer of poly(vinylidene fluoride).
11. A non-volatile transistor according to any one of claims 1 to 10, wherein the transistor (10) comprises a reading unit, the reading unit comprising a sub-unit for applying a voltage between the drain (16) and source (14) and a subunit for measuring the current between the drain (16) and the source (14).
