New method for oriented growth of a ferroelectric thin film of a rhombohedral ferroelectric alloy

Abstract

The invention relates to a method for growing a ferroelectric layer (11) of a rhombohedral ferroelectric alloy on a substrate (10), wherein a growth sublayer (21) formed of a stack of crystalline sheets linked together by van der Waals bonds is arranged on the substrate, and then the rhombohedral crystalline ferroelectric alloy in the form of crystalline grains (12) is deposited onto the growth sublayer at a suitable temperature with a component that segregates between the crystalline grains. Figure for the abstract: Fig. 1.

Description

Title of the invention: New method for oriented growth of a ferroelectric thin film of a rhombohedral ferroelectric alloy technical field

[0001] The present invention relates to the technical field of thin film deposition processes. In particular, the invention relates to a growth method applicable to any type of substrate, which makes it possible to obtain the growth of a crystalline ferroelectric thin film of a rhombohedral ferroelectric alloy having ferroelectric domains organized along the direction normal to the surface of the substrate on which the growth is carried out.

[0002] Technological background of the invention and technical problem

[0003] The rhombohedral α-GeTe crystalline phase (space group R3m) is a ferroelectric semiconductor with high spin-charge conversion capacity, i.e., it is a semiconductor with strong Rashba coupling. The ferroelectric GeTe alloy is also denoted α-GeTe. The rhombohedral α-GeTe crystalline phase is particularly interesting for all applications exploiting the ferroelectricity of a thin-film material. The ability to control the orientation of the ferroelectric domains is very advantageous for applications requiring the reversal of the orientation of the ferroelectric domains, for example, by applying an electric field when the GeTe layer is placed between two electrodes, as notably in FESO (FerroElectric SpinOrbit devices) [1].

[0004] Spin-charge interconversion induced by spin-orbit (SO) phenomena has opened new avenues for the design of low-power spin-logic architectures. Recently, Intel proposed a device called MESO [2] (Magneto-Electric Spin-Orbit). The write process requires magnetization reversal, while the read component relies on the spin-charge interconversion signal resulting from SO phenomena.

[0005] Recent studies in spintronics have led to the discovery of spin-charge interconversion phenomena in ferroelectric materials [1,3], as well as the possibility of controlling spin-charge interconversion at room temperature by switching the ferroelectric state in Rashba ferroelectric semiconductors (FERSCs) such as rhombohedral GeTe [1]. These materials exhibit a change of sign in the spin-to-charge conversion by reversing the ferroelectric polarization of the material, which has led to the proposal of the FerroElectric device. Spin-Orbit (FESO). This device shares the same important characteristics as the MESO device, such as non-volatile information control for in-memory computing. The output signal is electrically controlled, but there is no magnetization switching, thus eliminating the need for an efficient magnetoelectric material like that found in the MESO write block. In fact, the output signal is controlled directly by the ferroelectric state and does not require inverting the magnetic state of a spin injection layer. Its overall simplicity makes the FESO a viable alternative to the MESO as a device based on information and spin logic that is "Beyond-CMOS."

[0006] To date, room-temperature proofs of concept for the use of this type of film in FESO-type devices have been carried out using ferroelectric GeTe films, i.e., rhombohedral α-GeTe films deposited on small samples by molecular beam epitaxy (or MBE for "Molecular Beam Epitaxy") [4]. Thus, in this context, the large-scale fabrication of such a device remains a challenge, particularly for the industrial fabrication of the active ferroelectric GeTe layer with ferroelectric domains organized to allow their reversal by the application of an electric field.

[0007] Moreover, thin films of crystalline rhombohedral a-GeTe, in which the c-axis in hexagonal indexing, an axis coinciding with the longest diagonal of the rhombohedron and corresponding to the pseudo-cubic direction

[111] , has an out-of-plane orientation, are very promising for spintronic and ferroelectric devices. Spontaneous polarization of a-GeTe occurs along the pseudo-cubic axis, <111> leading to the formation of four ferroelastic variants and three possible polarization switching between domains at 71°, 109°, or 180° [4(b)].Thus, in GeTe crystalline grains, the alignment of all the major diagonals (pseudo-cubic indexing

[111] or hexagonal indexing

[001] of the rhombohedral structure) perpendicular to the substrate surface therefore maximizes the polarization potential of the rhombohedral a-GeTe thin film, since the applied electric field is always perpendicular to the substrate surface on which the thin film is deposited, in the applications envisaged.

[0008] Non-volatile nanometric ferroelectric domains have been observed in oriented rhombohedral α-GeTe films 23 nm thick deposited by molecular beam epitaxy (MBE) using the conductive tip of an atomic force microscope in PFM (piezo-response force microscopy) mode [5]. Reversible writing of (111) and (-1-1-1) ferroelectric domains, illustrated by ferroelectric hysteresis in such cases, has been successfully achieved, revealing the potential of rhombohedral α-GeTe for ferroelectric memories. Although in the films deposited by MBE, there are only 5 possible orientations in the plane of the a and b axes of GeTe, it would be possible to obtain a similar ferroelectric potential with only rhombohedral a-GeTe crystalline grains having their c axis oriented out of the plane, and more precisely perpendicular to the plane of the substrate on which these crystalline grains are deposited, and exhibiting a random texture in the ab plane, since the ferroelectric potential of rhombohedral a-GeTe thin films corresponds only to the physical displacement of Ge atoms along the c axis.

[0009] Therefore, a new process compatible with industrial applications allowing the manufacture of thin films, with control of the growth of crystalline grains of a rhombohedral ferroelectric alloy oriented with their c-axis, perpendicular to the surface of the substrate on which the growth is carried out, would constitute a major technological advance for spintronic and ferroelectric applications. Summary of the invention

[0010] In this context, the present invention aims to provide a new method for achieving controlled growth of crystalline grains of a rhombohedral ferroelectric alloy in the desired orientation, that is, with their c-axis, and therefore their pseudo-cubic axis (111), oriented perpendicularly or substantially perpendicularly to the surface of the substrate on which the growth is carried out. The invention also provides such a method that is suitable for industrial implementation and for the deposition of large-area thin films. The method according to the invention is also perfectly suited to the controlled growth, in this orientation, of crystalline grains of a ferroelectric alloy, for example, rhombohedral a-GeTe, rhombohedral GeixSnxTe, or rhombohedral GeSeï xTex.

[0011] The invention relates to a method for growing a ferroelectric layer formed of crystalline grains of a rhombohedral ferroelectric alloy, optionally doped, on a substrate. The method comprises the following steps: a) providing a substrate; b) depositing on the substrate a growth sublayer formed of a stack of crystalline sheets bonded together by van der Waals forces; c) depositing the rhombohedral crystalline ferroelectric alloy in the form of crystalline grains onto the growth sublayer present in step b), the ferroelectric alloy optionally being doped, and the deposition being carried out in the presence of a component that does not integrate into the crystalline grains of the ferroelectric alloy. The deposition in step c) is carried out at a temperature suitable for obtaining the deposit. and the crystallization of the crystalline grains of the possibly doped rhombohedral ferroelectric alloy, with the component segregated between the crystalline grains.

[0012] Within the scope of the invention, a method for producing a ferroelectric layer formed of crystalline grains of a rhombohedral ferroelectric alloy, optionally doped, with growth along the desired orientation on any type of substrate is proposed. Advantageously, in the method according to the invention, during step c), the growth of each crystalline grain occurs along a preferred orientation of the crystallographic axis c, perpendicular or substantially perpendicular to the surface of the substrate on which the growth sublayer is deposited, and more particularly perpendicular or substantially perpendicular to the surface of the growth sublayer. By segregated, it is understood that the component that does not fit within the crystalline grains of the rhombohedral ferroelectric alloy fits between the crystalline grains of the optionally doped rhombohedral ferroelectric alloy.In other words, this component accumulates at the grain boundaries of the rhombohedral ferroelectric alloy, possibly doped, depending on the composition of the resulting ferroelectric layer.

[0013] Compared to rhombohedral α-GeTe, a rhombohedral Gei xSn xTe alloy is potentially as, or even more, interesting than α-GeTe because this alloy is less conductive and has an even higher Rashba effect due to the presence of Sn. Compared to rhombohedral α-GeTe, in a rhombohedral Gei xSn xTe alloy, Sn replaces some of the Ge in the rhombohedral lattice. Indeed, Ge and Sn can substitute for each other in the rhombohedral phase. The advantage in this case is that Sn is heavier than Ge, which increases the spin-charge interconversion power of the material and therefore the Rashba effect. In the process of the invention, when the growth relates to a ferroelectric layer formed of crystalline grains of a rhombohedral Gei xSnxTe alloy possibly doped, the value of x is greater than 0 and, preferably, less than or equal to 0.7.This preferred value of x applies to the entire description that follows concerning the growth of a ferroelectric layer formed from crystalline grains of a rhombohedral Gei xSn xTe alloy, possibly doped. It can be added that the rhombohedral GeSeï xTex alloy is even more electrically insulating than the rhombohedral Gei_xSn xTe.

[0014] In the process according to the invention, growth is carried out on a growth sublayer formed from a stack of crystalline sheets bonded together by van der Waals bonds. This growth sublayer does not disrupt the growth of the crystalline grains of the rhombohedral ferroelectric alloy, depending on the thin film to be deposited, and acts as an orientation sublayer. of the growth layer. This sublayer is compatible with the ferroelectric alloy, particularly in terms of lattice parameters and chemical interaction. It can form a van der Waals gap or pseudo-van der Waals gap with the deposited ferroelectric alloy thin film, or even an alloy with the latter. Specifically, the growth sublayer comprises Te or Se and is typically a growth sublayer of As₂Te₃, Sb₂Te₃, Bi₂Te₃, As₂Se₃, Sb₂Se₃, or Bi₂Se₃, and typically a growth sublayer of Sb₂Te₃.

[0015] In particular, step b) is carried out by depositing the growth sublayer formed from a stack of crystalline sheets linked together by van der Waals bonds. It is also possible to deposit an amorphous sublayer which will crystallize during the temperature increase to the temperature required to carry out step c), this crystallization leading to the formation of the growth sublayer formed from the stack of crystalline sheets linked together by van der Waals bonds.

[0016] Advantageously, prior to step b), the substrate surface is passivated with Te or Se, in particular using a Te or Se flux, and this is followed by the deposition of a growth sublayer of As2Te3 or Sb2Te3 or Bi2Te3 or As2Se3 or Sb2Se3 or Bi2Se3 onto the passivated substrate surface. The growth sublayer of As2Te3 or Sb2Te3 or Bi2Te3 or As2Se3 or Sb2Se3 or Bi2Se3 deposited on the passivated substrate surface is either in an amorphous form and then crystallized to form a stack of crystalline sheets linked together by van der Waals bonds, or deposited directly as a stack of crystalline sheets linked together by van der Waals bonds.

[0017] According to some embodiments, the ferroelectric alloy comprises Sn.

[0018] According to certain embodiments which can be combined with the preceding ones, the ferroelectric alloy comprises Te.

[0019] Advantageously, the ferroelectric alloy further comprises Ge.

[0020] According to a first particular embodiment, the ferroelectric alloy corresponds to a rhombohedral GeSel-xTex alloy possibly doped.

[0021] In this case, step c) can, in particular, be carried out from one or more targets leading in the ferroelectric layer to an atomic composition in Ge, Se and Te in which Se+Te represents from 50 to 75 atomic %, preferably from 50 to 60 atomic %, and in particular from 50 to 55 atomic %.

[0022] In particular, according to a first embodiment of this first embodiment, the component that does not insert itself into the GeSel-xTex crystalline grains is Ge or Se and step c) leads to the formation of crystalline grains and GeSel- rhombohedral xTex with amorphous Ge or Se (or both) which segregate(s) between said crystalline grains.

[0023] In particular, the growth of a ferroelectric layer of a rhombohedral GeSel-xTex alloy, possibly doped, can be obtained with Ge or Se segregated between said crystalline grains, and step c) is carried out from: - of a GeSetTeu target with t+u such that the atomic % Se+Te varies from 50 to 75%, preferably from 50 to 60% atomic, and in particular from 50 to 55% atomic and typically being 50%; - of a Ge target, a Se target, and a Te target, - of a GeSel-xTex target and a Se target, or - a stoichiometric GeSe target, a stoichiometric GeTe target and a Se and / or Te target.

[0024] According to a second embodiment of this first embodiment, step c) is carried out in the presence of a target of a component other than Ge, Se, or Te that segregates between said crystalline grains. In particular, the component that segregates between the crystalline grains can be chosen from C, SiNx, SiOx, GeN, Al2O3, SiC, Si, and any other component that does not incorporate into a rhombohedral crystalline phase of GeSel-xTex.

[0025] In the first embodiment, the deposited rhombohedral GeSel-xTex alloy can be doped with a dopant, in particular As, Bi or Sb, and step c) can be carried out using a dopant target or a Ge, Se and / or Te target incorporating the dopant.

[0026] According to a second particular embodiment, the ferroelectric alloy corresponds to the rhombohedral a-GeTe possibly doped, or to a rhombohedral Gel-xSnxTe alloy possibly doped.

[0027] In this case, step c) can be carried out from one or more targets leading in the ferroelectric layer to an atomic composition of Ge and Te or of Ge, Sn and Te in which Ge or Ge+Sn, respectively, represents from 50 to 75% atomic, preferably from 50 to 60% atomic, and in particular from 50 to 55% atomic.

[0028] Thus, in the process according to the invention, in the case of the growth of a ferroelectric layer formed of rhombohedral a-GeTe crystalline grains optionally doped, step c) can be carried out from one or more targets leading in the ferroelectric layer to an atomic composition in Ge and Te in which Ge represents from 50 to 75% atomic, preferably from 50 to 60% atomic, and in particular from 50 to 55% atomic.

[0029] In the process according to the invention, in the case of the growth of a ferroelectric layer formed of crystalline grains of a rhombohedral GC|XSnxTc alloy, optionally doped, step c) can be carried out from one or more targets leading in the ferroelectric layer to an atomic composition of Ge, Sn and Te in which Ge+Sn represents 50 to 75% atomic, preferably 50 to 60% atomic, and in particular 50 to 55% atomic.

[0030] According to a first embodiment of the second embodiment of the process according to the invention, the component that does not fit into the crystalline grains of a-GeTe or GeixSnxTe is Ge or Sn and step c) leads to the formation of rhombohedral crystalline grains of a-GeTe or Gei_xSnxTe, with amorphous Ge or Sn segregated between said crystalline grains.

[0031] In this first variant, in the case of the growth of a ferroelectric layer formed of rhombohedral α-GeTe crystalline grains, possibly doped with Ge which segregates between said crystalline grains, the component that does not insert itself into the α-GeTe crystalline grains is Ge, and step c) leads to the formation of rhombohedral α-GeTe crystalline grains, with amorphous Ge at the grain boundaries. In particular, step c) can be carried out starting from: - of a GeyTe target with y such that the atomic % of Ge varies from 50 to 75%, preferably from 50 to 60% atomic, and in particular from 50 to 55% atomic and typically having a GeTe composition (so-called stoichiometric composition or with the atomic % of Ge equal to 50%), - of a Ge target and a Te target, or - of a stoichiometric GeTe target and a Ge target.

[0032] In this first variant, in the case of the growth of a ferroelectric layer formed from crystalline grains of a rhombohedral GcixSnxTc alloy, possibly doped, the component that does not insert itself into the Gei xSnxTe crystalline grains is Ge or Sn, and step c) leads to the formation of rhombohedral Gei xSnxTe crystalline grains, with amorphous Ge or Sn segregating between said crystalline grains. In particular, step c) can be carried out starting from: - of a GetSnuTe target with t+u such that the atomic % Ge+Sn varies from 50 to 75%, preferably from 50 to 60% atomic, and in particular from 50 to 55% atomic and typically being 50%, - of a Ge target, an Sn target, and a Te target, - of a Gei xSnxTe target and a Ge target, or - a stoichiometric GeTe target, a stoichiometric SnTe target and a Ge and / or Sn target.

[0033] According to a second embodiment of the second method of the invention, step c) is carried out in the presence of a target of a component other than Ge, Sn or Te, which segregates between said crystalline grains. In particular, the component which segregates between said grains is chosen from C, SiNx, SiOx, GeNx, GeOx, A1OX, MgO, Pt, SiC, Si and any other component that does not incorporate into a rhombohedral crystalline phase a-GeTe or Gei_xSnxTe.

[0034] Regardless of the embodiment or variant of implementation, in the process according to the invention, the deposition in step c) can be carried out using any suitable deposition technique. In particular, step c) is carried out by a vaporization technique, especially sputtering. Specifically, during this vaporization step, the desired ferroelectric layer is deposited onto the growth sublayer by vaporizing one or more targets adapted to the composition of the ferroelectric layer to be deposited, which correspond in particular to the targets described above.

[0035] Advantageously, in step c), the substrate bearing the growth sublayer is heated to a temperature sufficient to obtain the formation of crystalline grains of the rhombohedral ferroelectric alloy (for example, of the rhombohedral a-GeTe alloy or of the rhombohedral Gei xSnxTe alloy, or of the GcSC|XTcx alloy possibly doped), but which remains below 400°C, in particular the substrate bearing the growth sublayer is heated to a temperature in the range of 180 to 300 °C.

[0036] In the process according to the invention, when the deposited rhombohedral ferroelectric alloy is doped with a dopant, in particular As, Bi, or Sb, then step c) is carried out using a dopant target or a Ge, Sn, Se, and / or Te target incorporating the dopant. Thus, the targets described above are modified or supplemented by those skilled in the art to obtain the desired doping. By Ge, Sn, Se, and / or Te target, we mean a target that contains one or more of these elements, or even all three. Of course, the incorporation of the dopant into a target made up of or containing Sn or Se relates to the case of the growth of a ferroelectric layer formed of rhombohedral GcixSnxTc or rhombohedral GeSei xTex crystalline grains.

[0037] The process according to the invention may include one or more steps subsequent to step c). In particular, the process according to the invention includes a step d), subsequent to step c), of depositing on the ferroelectric layer an oxidation protection layer, said protection layer preferably being a dielectric layer.

[0038] It is also possible that the process according to the invention comprises a step d), subsequent to step c), of depositing on the ferroelectric layer a dielectric protection layer, followed by a step e) of depositing an electrically conductive layer.

[0039] The process according to the invention can be implemented on any type of substrate. For example, the substrate may be based on or composed of crystalline silicon, amorphous silicon, silicon dioxide, mica, CaF2, SiNx, SiOC, SiCN, MgO, aluminum oxide, magnesium oxide, TiNx, WSi, W, TiSiN, TaN, CoFe(Pt), CoFe, Au, Al, a ferromagnetic alloy, carbon, or a polymer resistant to the heating temperature applied during step c). In the case where the substrate is a conductive layer, for example TiN, WSi, W, TiSiN, TaN, etc., the substrate can act as an electrode in the applications described. Detailed description of the invention

[0040] Other features, details and advantages of the invention will become apparent from the description given with reference to the accompanying figures provided by way of illustration, which represent, respectively:

[0041] Fig. 1 is a schematic illustration, without regard to the respective dimensions of the different layers for greater visibility, showing the orientation of the c-axis, corresponding to the pseudo-cubic axis. <111> , rhombohedral a-GeTe grains 12 in a thin layer 11 produced on a growth sublayer 21, itself deposited on the surface of a substrate 10, according to the process of the invention. The x-axis, perpendicular to the plane of the substrate 10, is parallel to the c-axis, corresponding to the pseudo-cubic axis <111> , of rhombohedral a-GeTe grains 12 of the thin layer 11. The y and z axes are parallel or in the plane of the substrate 10 and the growth sublayer 21. For clarity, grain boundaries are not shown. An identical figure illustrates the growth of a GeixSnxTe thin layer 11 formed of rhombohedral Gei_xSnxTe grains 12.

[0042] Fig. 2 is a schematic illustration, without regard to the respective dimensions of the different layers for greater visibility, showing different stages of a process according to the invention.

[0043] Figure 3 shows the XRD diagram of the rhombohedral a-GeTe layer obtained on the single-crystal Si (100) substrate known as "Monitor" covered with an Sb2Te3 growth layer, in Example 1.

[0044] Fig. 4a and Fig. 4b present the rocking curve acquired on one of the Bragg diffraction peaks 001 of the rhombohedral a-GeTe layer obtained in Example 1, respectively for the Si substrate “Monitor” and the Si substrate covered with a layer of SiO2 (500 nm), substrates themselves covered with an Sb2Te3 growth layer.

[0045] Fig. 4c shows the Bragg diffraction peak curves 006 of the rhombohedral a-GeTe layer obtained in Example 1, for a SiO2 substrate (100 nm) covered with an Sb2Te3 growth sublayer (solid line curve) or without an Sb2Te3 growth sublayer (dotted line curve).

[0046] Fig. 5 presents the STEM-HAADF analysis of the rhombohedral a-GeTe layer obtained in Example 1, in the case of the Si substrate “Monitor” with Sb2Te3 sublayer.

[0047] Fig. 6 is an image obtained by piezo-response force microscopy (PFM, over an X range shown in abscissa of 1.50 pm) in the case of an amorphous silicon substrate, covered with an Sb2Te3 growth layer, on which the growth of a rhombohedral a-GeTe thin layer was carried out, and then covered with a SiN protective layer.

[0048] Fig. 7a and Fig. 7b present the XRD diagram of the rhombohedral a-GeTe layer obtained in Example 2, respectively for the Si Monitor substrate and the Si substrate coated with a SiO2 layer (100 nm), using a complementary Te target.

[0049] [Fig.8] presents the XRD diagram of the rhombohedral ferroelectric GeTe and ferroelectric GeSel-xTex layers obtained on the single-crystal Si (100) substrate known as "Monitor" covered with an Sb2Te3 growth layer, in example 1.

[0050] Figs. 9a, 9b, 9c and 9d show the rocking curves acquired on one of the Bragg 006 diffraction peaks of the rhombohedral GeTe layer and the rhombohedral GeSel-xTex layers obtained in Example 1, for the Si (100) substrate deoxidized by spraying and covered with an Sb2Te3 growth layer.

[0051] Conventionally, within the framework of the invention, a grain is defined as a crystalline region having a periodic and uniform atomic organization. As can be seen in [Fig. 1], the process according to the invention leads, in particular, to the formation of a ferroelectric thin film 11 made up of crystalline grains 12 of a rhombohedral ferroelectric alloy (for example, of rhombohedral α-GeTe), schematically represented by blocks in this figure, in which the crystallographic orientation corresponds to the axis c perpendicular to the plane of the substrate 10 (called the growth surface), itself parallel to the surface of the growth sublayer 21. It should be noted that although a slight misorientation is possible, the surface of the growth sublayer 21, on which the ferroelectric thin film 11 is deposited, is generally parallel to the plane of the substrate 10 (called the growth surface) on which it is itself deposited.The c-axis of the crystalline grains 12 is therefore parallel to the x-axis shown. Within the framework of the invention, this refers to a preferred orientation for the growth of the grains. crystalline grains 12 corresponding to the orientation of the crystallographic axis c perpendicular to the growth surface or substantially perpendicular to the growth surface. This notion of preferred orientation means that there is a certain tolerance regarding the orientation of the crystallographic axis c of a crystalline grain 12 of a rhombohedral ferroelectric alloy (for example of rhombohedral a-GeTe or of a rhombohedral GeixSnxTe alloy, or of a GeSeï xTex alloy, .. .)in the thin layer 11 with respect to the surface of the substrate 10: for each crystalline grain 12 of the rhombohedral ferroelectric alloy, the crystallographic axis c of the unit cell of the crystalline grain 12 forms an angle with the surface of the substrate 10, on which the thin layer of the rhombohedral ferroelectric alloy is deposited, which is constant and is substantially equal to 90°, that is to say that it is equal to 90°+ / - 10° at most, or even to 90°+ / - 5° at most, or even to 90°+ / -3° at most, or even is equal to 90°. In other words, the crystallographic axis c of each crystalline grain 12 is parallel to the x-axis shown in [Fig. 1] or forms an angle with it of at most + / - 10°, or even + / - 5°, or even + / - 3°. However, there may be variability between the angle formed by the crystallographic axis c and the surface of the substrate 10, from one crystalline grain 12 to another.This variability will generally be at most + / - 10°, or even at most + / - 5°, or even + / - 3°. The orientation and possible distribution can be obtained by X-ray diffraction (XRD). The c-axis of the crystalline grains 12 of the rhombohedral ferroelectric alloy thin film 11 therefore has a preferred orientation. On the other hand, for the rest, the crystallographic orientation can change from one crystalline grain 12 to another, that is to say that the a and b axes can have different orientations, as is the case in [Fig. 1], with the c-axis maintaining this same orientation parallel or close to the x-axis. In the thin film 11, the grain boundaries, not shown in [Fig. 1], are not shown in [Fig. 1].[l] have a different composition from the rhombohedral a-GeTe crystalline grains 12 (in the case of a ferroelectric layer formed from rhombohedral a-GeTe crystalline grains 12) or from Gei xSn xTe (in the case of a ferroelectric layer formed from rhombohedral Gei_xSn xTe crystalline grains 12). However, the thin film predominantly consists of a composition that is the rhombohedral ferroelectric alloy or the doped rhombohedral ferroelectric alloy, the grain boundaries representing less than 25 atomic percent and generally from 1 to 20 atomic percent, typically from 1 to 5 atomic percent, of the thin film 11 obtained, which can therefore be properly described as ferroelectric. For short, despite the presence of a component other than the rhombohedral ferroelectric alloy (e.g. GeTe or GC|XSnxTc or GcSc, xTex depending on the case) at the grain boundaries, in the rest of the description, the thin film 11 can simply be called the thin film of the ferroelectric alloy. rhombohedral (GeTe (or GC|XSnxTc respectively)) or ferroelectric layer of (GeTe or Gei_xSnxTe respectively). For example, the crystalline grains 12 of rhombohedral a-GeTe can be composed of GeTe or doped GeTe. The crystalline grains 12 of rhombohedral Gei_xSnxTe can be composed of an alloy of Gei_xSnxTe or an alloy of doped Gci_xSnxTc. For the sake of simplicity, in the following description, when referring to a rhombohedral alloy (such as rhombohedral a-GeTe), this will encompass both rhombohedral alloys (such as rhombohedral a-GeTe) and doped rhombohedral alloys (such as doped rhombohedral a-GeTe), unless explicitly stated otherwise. Similarly, in the following description, when referring to rhombohedral Gei xSn xTe, this will encompass both rhombohedral Gei xSn xTe alloys and doped rhombohedral Gei xSn xTe alloys, unless explicitly stated otherwise.

[0052] For example, the rhombohedral ferroelectric alloy may comprise Te. For example, the rhombohedral ferroelectric alloy may be SiTe, PbTe, MnTe. The ferroelectric alloy may comprise Sn. For example, the rhombohedral ferroelectric alloy may be SnSe, SnS, SnSei xTex. For example, the rhombohedral ferroelectric alloy may be GeSei xTex, GcixSixTc, Gc, xPbxTe, GC| xMnsTc, GC| VSnvSC| XTcs. The rhombohedral ferroelectric alloys mentioned above may be doped with As, Bi, or Sb.

[0053] The detailed description below is given in the case of the growth of a rhombohedral α-GeTe ferroelectric layer. But it is equally suitable for the growth of a rhombohedral Gei xSn xTe alloy ferroelectric layer, or of a ferroelectric alloy comprising Sn, or of a ferroelectric alloy comprising Te, or of a ferroelectric alloy comprising GeSei xTex.

[0054] In this case, it is not rhombohedral α-GeTe crystalline grains that are deposited, but grains of a rhombohedral Gei xSn xTe alloy, or of a ferroelectric alloy comprising Sn, or of a ferroelectric alloy comprising Te, or of a ferroelectric alloy comprising GeSei xTex. Of course, the targets and deposition conditions detailed later in the description will be adapted by those skilled in the art to obtain the desired composition. Within the framework of the process of the invention, the growth of the rhombohedral α-GeTe crystalline grains 12 of the thin film 11 in the desired orientation can be achieved on any type of substrate 10 using a growth sublayer 21 located between the substrate 10 and the thin film 11.It is on this growth sublayer 21 composed of crystalline sheets 22 linked together by van der Waals bonds that the growth of rhombohedral a-GeTe is carried out and it is this sublayer. growth layer 21 will contribute to giving the desired orientation to the rhombohedral α-GeTe crystalline grains 12 of the thin film 11. In the context of the invention, "crystalline sheet" means a crystalline layer whose thickness corresponds to a single plane of atoms, parallel to the surface of the substrate on which the sublayer is deposited, or to several parallel planes of atoms forming a crystalline block, which are also parallel to the surface of the substrate on which the growth sublayer is deposited. In the case of crystalline sheets corresponding to a single plane of atoms, these will be called two-dimensional sheets. Such two-dimensional sheets are notably present when the subshell is composed of MoS2, for example. In the case of crystalline sheets formed of several atomic planes, called crystal blocks, each atomic plane contains atoms of the same element, as is the case, for example, with Sb2Te3 where each block is composed of 5 atomic planes. Within the growth subshell 21, the crystalline sheets 22, whether or not in the form of crystal blocks, are linked together by van der Waals forces between two planes of atoms belonging to each of the adjacent crystalline sheets, said crystalline sheets being able to exist in the form of two-dimensional crystalline sheets or crystal blocks. Van der Waals bonds can be, for example, Te-Te, Se-Se, or SS bonds in chalcogenides. The large spacing between the crystal sheets linked by these van der Waals bonds is generally attributed to the existence of a plane of van der Waals vacancies, which is called a van der Waals gap or pseudo-gap (vdW). For the purposes of this description, the vdW gap and pseudo-gap are considered equivalent and separate two crystal sheets linked by vdW bonds. This particular structural form is observed especially in phase-change materials (PCMs), chalcogenides, and more generally in materials exhibiting a chemical bonding mechanism known as "metavity." These materials are frequently described as 2D materials linked by van der Waals bonds.This description can nevertheless be expanded, as the distance between planes across the van der Waals gap is generally shorter than twice the van der Waals radius of the atoms constituting those planes. For example, in the case of tellurium, the Te-Te distance across the van der Waals gap is about 10% shorter than twice the van der Waals radius of Te. Thus, in the stacking that constitutes subshell 21, the crystalline sheets can be separated by van der Waals gaps or pseudo-van der Waals gaps, which occur, in particular, as a spacing between two pure planes of the same type of atom. The bonds between atoms across the pseudo-van der Waals gap are typically shorter and stronger than ordinary van der Waals bonds. At the interface with the rhombohedral a-GeTe layer, the growth sublayer 21 can form an alloy or be separated from it by a gap. or van der Waals pseudo-gap. The materials in the following non-exhaustive list typically adopt a crystalline sheet structure with gaps or pseudo-gaps of vdW described above: tellurium, antimony, binary alloys Ge-Sb, Ga-Sb, Ge-Te, Ge-Se, GeSe2, Sb-Te, notably Sb2Te3, Sb-Se, notably Sb2Se3, As-Se, notably As2Se3, As-Te, notably As2Te3, Sn-Se, Sn-Te, Bi-Se, notably Bi2Se3, Bi-Te, notably Bi2Te3, Ga-Te, Ga-Se, Pb-Te, In-Se, notably In2Se3, Pb-Se, Zr-Te, Zr-Se, Ti-Te, notably TiTe2, Ti-Se, notably TiSe2, Si-Te, ALTe, Mo-Te, notably MoTe2, Mo-Se, notably MoSe2, Mo-S, notably MoS2, W-Te, notably WTe2, WS, notably WS2, W-Se, notably WSe2, the ternary alloys Ge-Sb-Te, Ge-Se-Te, Ag-Sb-Te, Ge-Bi-Te, Ge-Sb-Se, quaternary Ge-Sb-Se-Te or even Ge-Bi-Sb-Te-Se.A person skilled in the art will choose such a growth sublayer formed of crystalline sheets linked together by van der Waals bonds (vdW gap or pseudo-gap sheet structure) that does not disrupt the growth of the crystalline grains of a-GeTe or the Gei_xSnxTe alloy and acts as a growth orientation sublayer. This growth sublayer 21 is compatible with the rhombohedral a-GeTe or GC|XSnxTc alloy depending on the desired ferroelectric layer, particularly in terms of lattice parameters and chemical interaction. Preferably, the growth sublayer 21 is a crystalline sheet layer corresponding to a stacking of chalcogenide crystalline blocks, in particular As2Te3, Sb2Te3, or Bi2Te3.

[0055] As illustrated in [Fig. 2], the process according to the invention comprises a step a) of providing a substrate 10, followed by a step b) in which the growth sublayer 21 is deposited onto the substrate. The growth sublayer 21 can be obtained in one or more steps. It is possible to form an amorphous deposit on the upper surface of the substrate 10, which will then crystallize upon subsequent heating to form the stack of crystalline sheets linked together by van der Waals bonds. It is also possible, and simpler, to directly deposit the growth sublayer 21 in the form of the stack of crystalline sheets linked together by van der Waals bonds.According to a preferred embodiment, the process of the invention further comprises, before step b), i.e. before the formation of the growth sublayer 21, a step of passivating the substrate 10 with tellurium or antimony of the surface on which the growth sublayer 21 is formed. This makes it easier to form the growth sublayer 21 by so-called vdW epitaxy, where growth takes place in the form of crystalline sheets 22 oriented parallel to the surface of the substrate and separated by vdW gaps or pseudo-gaps, to form a stack.

[0056] As for the substrate 10, it can be made of any type of material. It can be monolayer or multilayer. It can be a substrate 10 based on or made of crystalline silicon, amorphous silicon, silicon oxide, glass, mica, CaF2 or any standard deposition substrate... The substrate can also be made of or coated with or based on another type of material, in particular chosen from TiN, SiN, WSi, W, TiSiN, TaN, SiOC, SiCN, Pt, CoFe(Pt), Au, Cu, Al, other metallic or magnetic deposits, carbon, or a polymer, as well as any material resistant to the heating temperature applied to the substrate 10 during the deposition step c) of the growth process according to the invention, in particular vaporization techniques, such as physical vapor deposition.The substrate can also be a stack of several layers of these types of material, notably a layer of amorphous silicon covered by a layer of amorphous silicon oxide, which will then correspond to the surface on which the growth sublayer 21, and then the ferroelectric thin layer 11, will be deposited. The substrate 10 typically has a thickness of 100 to 1000 pm, and for example 100, 150, 200, or 300 pm. The substrate 10 is generally in the form of a disk with a diameter commonly used in microelectronics.

[0057] The substrate may have a flat surface or a surface with step or pattern-like features, which will ultimately result in a planar polarized stack rather than a vertically polarized one. The growth sublayer 21 is deposited on the so-called upper surface of the substrate 10, as illustrated in Figures 1 and 2. When the substrate 10 includes steps or patterns, the orientation of the crystalline grains of the ferroelectric layer 11 is determined locally with respect to the surface of the substrate on which it is deposited, which itself carries the intermediate growth sublayer 11.

[0058] For the implementation of step c) of deposition, particularly by vaporization, the substrate 10 carrying the growth sublayer 21 and the target(s) necessary to obtain the desired deposit are conventionally positioned in a deposition chamber, in particular a vaporization chamber. Within the scope of the invention, the vaporization carried out in step c) of the target(s) used can be performed using any suitable technique, particularly for industrial-scale implementation. This may be a chemical vapor deposition (CVD) technique or a physical vapor deposition (PVD) technique. Sputtering and pulsed laser deposition (PLD) techniques are particularly suitable.Within the framework of the invention, preference will be given to the use of the magnetron sputtering technique, and in particular, direct current (DC) or pulsed DC sputtering. Radio-frequency magnetron sputtering (RFMS) is a technique where a cold plasma is created in a vacuum chamber that serves as the vaporization chamber. This chamber contains the substrate 10, which carries the growth sublayer 21, and the target(s), which are the source(s) of GeTe (GeTe) to be deposited, possibly doped, and the source of the grain boundary component that will accumulate between the GeTe grains. Ions, usually argon, are bombarded onto the target(s), which act as the cathode. This allows atoms to be vaporized from the target and deposited onto the substrate to form a thin layer. In magnetron sputtering, a magnetic field is created near the target(s) to increase the plasma density and thus the sputtering efficiency.In the case of the radio frequency version, the continuous supply to the cathode is replaced by a high-frequency supply, typically at 13.56 MHz, and the anode-cathode bias is reversed at high frequency.

[0059] In the context of the invention, it has been shown that three parameters are important for obtaining controlled growth in the desired orientation: the composition of the deposited film 11 and therefore of the vaporized target(s), the use of a growth sublayer 21 on the surface of the substrate 10 on which growth is carried out, and the temperature at which the substrate 10 bearing the growth sublayer 21 is heated. Indeed, it has been observed that if, during step c), the heating is not carried out at a sufficient temperature, it is not possible to obtain a rhombohedral crystalline α-GeTe phase, but rather an amorphous or disordered crystalline phase of GeTe is obtained. As is known, this temperature must not, however, be too high to obtain a deposition.However, simply choosing the heating temperature of substrate 10 is not sufficient to obtain the desired crystallographic orientation, as will be shown in the following examples. The origin of the preferential orientation of the rhombohedral α-GeTe crystal grains, with their c-axis perpendicular or nearly perpendicular to the deposition surface (which in the example illustrated in [Fig. 1] is also perpendicular or nearly perpendicular to the surface of substrate 10), is also linked to the composition of the films actually deposited and therefore to that of the initial target(s). It has been observed that when a target with a stoichiometric GeTe composition is used, the overall composition of the ferroelectric thin film obtained in step c), particularly by sputtering, has a slightly Te-deficient composition.Without being linked by any specific mechanism of action, it is assumed that this deficit results from the greater volatility of Te during vaporization or from its desorption from the substrate surface when the latter is heated for the growth of the rhombohedral oriented α-GeTe thin film. This slight excess of Ge causes, during growth, the... segregation of an excess phase of amorphous Ge at the boundaries of the rhombohedral a-GeTe grains which then induces by effect of stress the orientation of the ferroelectric domains of the crystalline grains 12 of rhombohedral a-GeTe with their c-axis perpendicular or substantially perpendicular to the surface of the substrate 10 and the growth sublayer 21. The inventors also demonstrated that the use of an additional pure Te target to compensate for the loss of Te, in addition to the stoichiometric GeTe target, which is classically done to maintain stoichiometry, when these two targets were vaporized at the same time, caused the disappearance of the excess Ge phase and thus a loss of the correct orientation of the ferroelectric domains of the rhombohedral a-GeTe grains along the c-axis perpendicular or substantially perpendicular to the surface of the growth sublayer 21 and the substrate 10 and the rotation of the ferroelectric domains.The effect of stresses and the orientation of the ferroelectric domains of the rhombohedral a-GeTe crystalline grains 12 with their c-axis perpendicular or substantially perpendicular to the surface of the growth sublayer 21 can be obtained with a component other than Ge that does not insert into the rhombohedral a-GeTe crystalline phase.

[0060] In this context, several variations in the implementation of the process according to the invention can be envisaged. In a first variation, amorphous Ge segregates at the boundaries between the oriented rhombohedral α-GeTe grains that are deposited on the growth sublayer 21. In this case, the deposition on the growth sublayer 21 in step c) can be carried out as follows: - carry out the deposition in step c), in particular by vaporization, from a single target of stoichiometric GeTe composition, or more generally from a target composed of Ge and Te, having a composition GevTe with v such that Ge represents at least 50% atomic, in particular more than 50% atomic and at most 75% atomic, preferably more than 50% atomic and at most 60% atomic, and in particular more than 50% atomic and at most 55% atomic, or carry out the deposition in step c), in particular by vaporization, from a target of stoichiometric GeTe composition, and a Ge target, or - carry out the deposition during step c), in particular by vaporization, from a Ge target and a Te target.

[0061] In these different cases, the deposition conditions, and in particular the vaporization of the targets when there are several, are adjusted by a person skilled in the art to obtain a ferroelectric layer 11 of atomic composition in Ge and Te in which Ge represents at least 50% atomic, in particular more than 50% atomic and at most 75% atomic, preferably more than 50% atomic and at most 60% atomic, and in particular more than 50% atomic and at most 55% atomic.

[0062] In a second embodiment, a component (other than Ge) that does not insert itself into the GeTe crystalline grains segregates at the boundaries between the oriented rhombohedral α-GeTe grains. In this case, the deposition on the growth sublayer 21 in step c) can be carried out as follows: - carry out the deposition during step c), in particular by vaporization, from a target of composition GewTe with w such that Ge represents less than 50 atomic % and from a target of a component that does not fit into the crystalline grains of a-GeTe, or - carry out the deposition during step c), in particular by vaporization, from a Ge target, of a Te target and of a target of a component which does not fit into the crystalline grains of a-GeTe.

[0063] In these cases, the deposition conditions, and in particular the vaporization of the targets, are adjusted by those skilled in the art to obtain a ferroelectric layer 11 with an atomic composition of Ge and Te in which Ge represents 50% atomic, i.e., a stoichiometric GeTe composition. A component that does not incorporate into the crystalline grains of α-GeTe may, in particular, be C, SiNx (in particular Si3N4), SiOx (in particular SiO2), GeN, Al2O3, ... or any other material that does not incorporate into the ferroelectric rhombohedral crystalline phase of α-GeTe (in particular SiC, Si, ...). It forms an amorphous or crystalline deposit at the boundaries between the oriented rhombohedral α-GeTe grains.

[0064] In all cases, when obtaining doped α-GeTe is desired, the target(s) used above in step c) may be supplemented by a target consisting of the desired dopant element, such as As, Bi, or Sb, or alternatively, the Ge(v or w)Te target. The Te or Ge target may be doped with the desired amount of dopant (generally representing less than 10% atomic percentage of the α-GeTe). Those skilled in the art will adjust the deposition parameters, and in particular the vaporization parameters, to obtain the desired doping, including the power applied to each of the targets.

[0065] In the case where the process according to the invention is applied to the growth of rhombohedral crystalline Gei xSn xTe, all the above applies. In this case, the deposition will be carried out with a target comprising Sn in the proportions or conditions allowing the growth of the desired layer. The component that segregates between the crystalline grains and forms the grain boundaries may be amorphous Ge or Sn, or a component that does not insert itself into the crystalline grains of rhombohedral Gei xSn xTe. Again, this component may, in particular, be C, SiNx (in particular Si3N4), SiOx (in particular SiO2), GeN, Al2O3, SiC, Si, etc. It forms an amorphous or crystalline insertion at the boundaries between the deposited oriented rhombohedral crystalline Gei xSn xTe grains.

[0066] In the case where a doped material is sought, dopants such as As, Bi or Sb in ferroelectric GeTe or rhombohedral alloys Gc, xSnxTe located on the pseudo-binary GeTe-SnTe line of the ternary Ge-Sn-Te phase diagram are of interest because they will also allow the same ferroelectric properties to be obtained, but with a lower electronic conductivity in the rhombohedral Ge-Sn-Te alloys or by charge compensation via n-induced doping by As, Bi and / or Sb and compensating for the intrinsic p-doping of GeTe due to Ge vacancies in the crystal lattice of the a-GeTe phase.

[0067] In a conventional manner, a person skilled in the art will adjust the parameters of step c) to obtain the desired composition. In particular, depending on the vaporization device used, and especially the sputtering device, a person skilled in the art will adjust the pressure in the vaporization chamber, the gas flow rate used to perform vaporization, the target-substrate distance, and, in the case of magnetron sputtering, the strength of the magnetic field and, in the case of radio-frequency magnetron sputtering, the radio frequency power as well. All these parameters depend on the equipment used.The optimal heating temperature for the substrate carrying the growth sublayer during vaporization to achieve growth of rhombohedral a-GeTe or Gei xSnxTe alloy crystalline grains depends on the vaporization technique used, the vaporization setup within that technique, and the composition of the targets and grain boundaries. These boundaries will be inserted between the a-GeTe or rhombohedral Gc xSnxTe alloy crystalline grains, constraining them and favoring the orientation of the ferroelectric domains. Specifically, the optimal heating temperature for the substrate during vaporization to achieve growth of a-GeTe or rhombohedral GC|XSnxTc alloy crystalline grains will be adjusted by a person skilled in the art, based on the deposition setup, and in particular the sputtering system used.Most often, a heating temperature above 180°C and below 330°C will be suitable, in particular a temperature in the range of 180 to 300°C, advantageously a temperature in the range of 200°C to 250°C. Such a preferred temperature range is particularly suitable, especially when the grain boundaries are formed of Ge (i.e., especially when the rhombohedral α-GeTe source target is of the GevTe type as previously defined) and the vaporization technique is a sputtering technique. In any case, with knowledge of the invention and therefore of the importance of the substrate heating temperature, and depending on the composition of the selected target(s), a person skilled in the art can conduct some evaluation tests to adjust the temperature. heating of the substrate, depending on the vaporization device used, to obtain the desired crystallographic orientation and ferroelectric polarization.

[0068] Figure 2 schematically illustrates various stages of a growth process according to the invention. The substrate 10 available in step a) can correspond to any substrate as previously described. The substrate 10 may be subjected to a preliminary cleaning and / or deoxidation step (in particular by treatment with an alkaline solution, typically potassium hydroxide or sodium hydroxide) or an acidic solution (typically hydrofluoric, nitric, or sulfuric acid) or a solvent (alcohol, acetone, etc.) or by spraying its surface with Ar+ ions, depending on the nature of the substrate, followed by rinsing and then drying, a procedure conventionally carried out prior to the deposition of any thin films.

[0069] In step b), the sublayer 21, formed from a stack of crystalline sheets 22 bonded together by van der Waals forces, is deposited on the upper surface of the substrate 10, most often after prior preparation, notably by passivation with tellurium or antimony, particularly when the sublayer 21 is As2Te3, Sb2Te3, or Bi2Te3. For this purpose, it is possible to perform a flash vaporization of Te to saturate the surface of the substrate 10, resulting in an atomic plane of Te on the surface of the substrate 10 before depositing a growth sublayer 21 of As2Te3, Sb2Te3, or Bi2Te3, among others. The sublayer 21 is generally a few nanometers thick (typically 1 to 10 nm for Sb2Te3). This sub-layer 21 can be formed in step b) by a prior vaporization step from a suitable target in the same device as that used in step c) or in a different device.A method of co-spraying Sb and Te or Sb2Te3 and Te targets by adjusting the power applied to each target can be used. The use of a Te-enriched Sb2Te3 target is also conceivable. This makes it possible to compensate for a potential tellurium deficiency during the formation of sublayer 21.

[0070] When the sublayer 21 is a sublayer of As2Te3, Sb2Te3 or Bi2Te3 is present under the thin layer 11, the latter can also be used to obtain an As, Sb or Bi doping, respectively, of the thin layer 11, by carrying out an appropriate diffusion annealing after deposition of the thin layer 11, in particular at an annealing temperature greater than or equal to 300°C.

[0071] Finally, subsequent to step c), it is possible to provide for a step d) aimed at protecting the thin film 11 formed. In particular, a protective layer 30 against oxidation may be deposited during a step d), as illustrated in [Fig. 2]. As an example of such protective layers 30, a conductive layer, such as a layer of TiN, WSi, W, TiSiN, TaN, PtCoFe (with or (without Pt), Au, Al, any ferromagnetic alloy, or a dielectric or insulating layer or tunnel barrier for the injection of a spin-polarized current, such as a layer of SiN, SiOx, SiOC, SiCN, MgO, TiOx, a polymer layer, or any other protective layer commonly used in microelectronics. Again, this protective layer 30 can be formed by a vaporization step from a suitable target in the same device as that used in step c) or in a different device. A dielectric protective layer 30 is advantageous because it allows the thin film 11 to be polarized by the application of an electric field, thus preventing leakage currents through the heavily doped, pseudo-metallic GeTe thin film 11.

[0072] As illustrated in [Fig. 2], the process according to the invention may include a step d), subsequent to step c), of depositing a dielectric protective layer 30 (in particular as previously described) onto the ferroelectric layer 11, followed by a step e) of depositing an electrically conductive layer 40 (in particular as previously described). The deposition of these protective layers 30, or even of an electrically conductive layer 40, will depend on the intended application.

[0073] The growth process according to the invention makes it possible to obtain thin films 11 of ferroelectric rhombohedral a-GeTe or thin films 11 of a ferroelectric rhombohedral Gei xSnxTe alloy with the crystalline grains 12 corresponding to the ferroelectric domains aligned along the pseudo-cubic axis <111> and perpendicular or substantially perpendicular to the surface of the substrate 10, regardless of the chemical nature of the surface of the growth substrate used. Vaporization in step c) leads, on the growth sublayer 21 deposited on the substrate 10, to the segregation of an excess phase of Ge (or Ge or Sn in the case of Gei_xSnxTe) or of a phase of another component which does not fit into the crystalline phase of a-GeTe (or into the crystalline phase of the Gei xSnxTe alloy depending on the case), to the grain boundaries of rhombohedral a-GeTe (or of the Gei xSnxTe alloy depending on the case) which form during their growth.This segregation at grain boundaries constrains the crystalline grains of a-GeTe (or of the Gei xSn xTe alloy depending on the case) and thus induces their orientation and the formation of ferroelectric domains stabilized with the c-axis (corresponding to the pseudo-cubic axis). <111> ) of the crystals present in the a-GeTe grains (or of the Gci xSn xTc alloy depending on the case) rhombohedral perpendicular or substantially perpendicular to the surface of the substrate 10 and to the growth sublayer 21 on which it is deposited. By applying an electric field perpendicular to the surface of the substrate, it is possible to reversibly orient the ferroelectric domains along the

[111] or [-1-1-1] direction. .

[0074] Finally, at the end of step c), a thin ferroelectric layer 11 corresponding to a thin film, in particular with a thickness of 5 to 500 nm (typically 20 nm), is obtained. The coercive field required to reverse the orientation of the ferroelectric domains in the case of a ferroelectric layer of a Gei xSn xTe alloy will be lower than for a ferroelectric layer of a-GeTe. Thus, those skilled in the art can adapt the thickness of the ferroelectric layer 11 formed, depending on the a-GeTe or Gei xSn xTe material, to potentially adjust the required coercive field.

[0075] The ferroelectric domains corresponding to the crystalline grains 12 of rhombohedral a-GeTe or of the rhombohedral Gei xSnxTe alloy, as appropriate, generally have an average size in the plane parallel to the surface of the substrate 10 of the order of 20 to 100 nm. With the process according to the invention, which can be implemented on an industrial scale, it is possible to produce large surface area thin films, in particular with a surface area greater than 7500 mm2, and typically from 100,000 to 300,000 mm2, or even 636,173 mm2.

[0076] As will be apparent from the examples that follow, the quality of the thin films obtained with a process according to the invention was checked by X-ray diffraction (XRD). In particular, it was observed that the growth obtained with the process according to the invention led to thin films in which the rhombohedral a-GeTe grains were perfectly oriented: only the diffraction peaks 001 were observed (hexagonal indexing of the rhombohedral structure of a-GeTe), which indicates oriented growth of the a-GeTe crystals with the c-axis perpendicular to the plane of the thin film 11 formed and therefore to the surface of the substrate 10.

[0077] The manufacture, according to the processes of the invention, of substrates 10 bearing a ferroelectric layer 11 formed of rhombohedral α-GeTe crystalline grains 12 or of a rhombohedral GeixSnxTe alloy, depending on the case, and deposited on an intermediate growth sublayer 21 present on the surface of the substrate, is of interest in a large number of applications: in CMOS microelectronic technologies, the manufacture of FESO devices, spintronic, ferroelectric, ferromagnetic, and multiferroic devices for applications in logic, computing, neuromorphic computing, memory computing, resistive memory, etc.

[0078] Implementation examples

[0079] Example 1

[0080] The deposition of 100 nm thin films of a-GeTe composed of crystalline grains with their pseudo-cubic axis <111> perpendicular to the surface of the substrate (named a-GeTe(l 11) film) was carried out either on a silicon substrate of 200 mm diameter previously deoxidized by sputtering with Ar+ ions (called Si Monitor), either on a 200 mm diameter silicon substrate coated with 500 nm of SiO2 obtained by thermal oxidation of Si. In this example, vaporization was carried out using a substrate heating temperature of 250°C, the deposition of a 5 nm growth sublayer of Sb2Te3 by co-sputtering of a stoichiometric Sb2Te3 target and a pure Te target followed by the deposition of the α-GeTe(l 11) film by sputtering of a GeTe stoichiometric target. The [Fig.3] which presents the XRD diagram obtained with the Si Monitor substrate highlights that in the resulting thin film, the rhombohedral a-GeTe grains were perfectly oriented: only the 001 diffraction peaks were observed (hexagonal indexing of the rhombohedral structure of a-GeTe), indicating oriented growth of the a-GeTe crystals with the c-axis perpendicular to the plane of the thin film formed and therefore to the substrate surface. The full width at half maximum (FWHM) of the rocking curve acquired on the 001 Bragg diffraction peaks shown in Figures 4a and 4b, for both substrates, confirms the excellent out-of-plane orientation of the crystallites in the film with low mosaicity around 3°. The STEM-HAADF analyses presented in [Fig.[5], in the case of the Si Monitor substrate with Sb2Te3 sublayer, also confirm the high structural quality of the films, showing ferroelectric domains inside the a-GeTe crystalline grains and aligned along the pseudo-cubic <111> axis perpendicular to the substrate.

[0081] Finally, a further test was carried out under similar conditions, but to produce a 40 nm thin film with the same target, using a 200 mm silicon (100) substrate deoxidized on its surface by Ar+ sputtering. The resulting a-GeTe thin film was then coated in situ by sputtering a 10 nm thick SiN protective layer. Piezo-response force microscopy (PFM) demonstrated the possibility of writing ([Fig. 6]) and highlighting the polarization state in the ferroelectric domains of the material in an upward or downward orientation depending on the direction of the applied electric field and with a phase shift of 180° measured between the two programmed areas (analogously with [Fig. 5]). It was thus demonstrated that the obtained a-GeTe film is indeed ferroelectric and that the domains are reversible using an electric field.

[0082] In the absence of an Sb2Te3 layer, a deposition carried out under the same conditions and from the same stoichiometric GeTe target did not allow the growth of α-GeTe with all the crystalline grains in the desired orientation, as shown in [Fig. 4c]. In [Fig. 4c], the intensity of the Bragg diffraction peak on one of the 006 lines of the rhombohedral α-GeTe deposited on SiO2 without a sublayer is very low compared to that obtained for a film of rhombohedral α-GeTe deposited under the same conditions, but on a 5 nm Sb2Te3 growth sublayer indicating a very high proportion of a-GeTe grains not oriented along the c-axis perpendicular to the substrate in the absence of a growth sublayer.

[0083] For example, the deposition of 100 nm thin films of GeSeï xTex alloys with different proportions of Se and composed of rhombohedral crystalline grains with their pseudocubic axes <111> perpendicular to the substrate surface was carried out on a 200 mm diameter silicon substrate previously deoxidized by spraying with Ar+ ions. Vaporization was carried out using a substrate heating temperature of 200°C, deposition of a 5 nm growth sublayer of Sb2Te3 by co-spraying of a stoichiometric Sb2Te3 target and a pure Te target, followed by deposition of the Ge-Se-Te based alloy film by co-spraying of two targets respectively of GeTe and GeSe with relative deposition rates adjusted to vary the composition of the resulting alloy (with respectively 0%, 5%, 15% and 25% Se as an atomic % in the alloy).

[0084] The XRD diagram, illustrated in [Fig. 8], shows that in the thin films obtained, the rhombohedral grains are perfectly oriented: only the 001 diffraction peaks are observed (hexagonal indexing of the rhombohedral structure), which indicates oriented crystal growth with the c-axis perpendicular to the plane of the thin film formed and therefore to the substrate surface. The full width at half maximum (FWHM) of the rocking curve acquired on the 001 Bragg diffraction peaks confirms the excellent out-of-plane orientation of the crystallites in the film with low mosaicity around 3°.

[0085] Example 2

[0086] The use of a 5 nm Sb2Te3 sublayer deposited on an amorphous silicon substrate (according to the conditions described in [6a]) also made it possible to obtain a rhombohedral a-GeTe thin film with grains oriented along the pseudo-cubic axis <111> perpendicular to the substrate.

[0087] GeTe thin films of 50 nm thickness were able to be deposited at different substrate heating temperatures between 180°C and 300°C on a 5 nm thick Sb2Te3 sublayer as described in the literature [6b] to promote orientation.

[0088] By vaporizing a stoichiometric GeTe target without adding Te, no misoriented crystalline grain was detected by XRD for all GeTe thin films deposited on a 5 nm growth sublayer of Sb2Te3: here again, well-ordered α-GeTe ferroelectric domains inside the crystalline grains and aligned along the pseudo-cubic axis. <111> were obtained perpendicular to the substrate. On the other hand, by carrying out the vaporization by adding a target of Te, the excess Ge which used to accumulate at the grain boundaries was lost and the preferred orientation of the rhombohedral a-GeTe was also lost, as can be seen from figures 7a and 7b. List of documents cited

[0089] [1] Room-temperature ferroelectric switching of spin-to-charge conversion in germanium telluride. Nature electronics 4 (2021) 740-747.

[0090] [2] Scalable energy-efficient magnetoelectric spin-orbit logic, Nature 565 (2018) 35-42.

[0091] [3] Non-volatile electric control of spin-charge conversion in a SrTiO3 Rashba System, Nature 580 (2020) 483-486.

[0092] [4] (a) A. Giussani, K. Perumal, M. Hanke, P. Rodenbach, H. Riechert, R. Calarco, physica status solidi (b) 2012, 249, 1939; (b) B. Croes, F. Cheynis, P. Millier, S. Curiotto, F. Leroy, Polar surface of ferroelectric nanodomains in GeTe thin films, Phys. Rev. Materials 2022, 6, 064407.

[0093] [5] C. Rinaldi, S. Varotto, M. Asa, J. Slawihska, J. Fujii, G. Vinai, S. Cecchi, D. Di Santé, R. Calarco, I. Vobornik, G. Panaccione, S. Picozzi, R. Bertacco, Nano Lett. 2018, 18,2751.

[0094] [6] (a) F. Hippert, P. Kowalczyk, N. Bernier, C. Sabbione, X. Zucchi, D. Térébénec, C. Mocuta and P. Noé, J. Phys. D: Appl. Phys., 2020, 53, 154003. (b) V. Sever, N. Bernier, D. Térébénec, C. Sabbione, J. Paterson, F. Castioni, P. Quéméré, A. Jannaud, J.-L. Rouvière, H. Roussel, J.-Y. Raty, F. Hippert and P. Noé, physica status solidi (RRL) - Rapid Research Letters, 2024, 2300402.

Claims

Demands

1. A method for growing a ferroelectric layer (11) formed of crystalline grains (12) of a rhombohedral ferroelectric alloy, optionally doped, on a substrate (10), said method comprising the following steps: a) providing a substrate (10), b) providing on the substrate a growth sublayer (21) formed of a stack of crystalline sheets bonded together by van der Waals bonds, c) depositing rhombohedral crystalline ferroelectric alloy in the form of crystalline grains (12) onto the growth sublayer (21) present in step b), the ferroelectric alloy optionally being doped, and the deposition being carried out in the presence of a component that does not insert itself into the crystalline grains of the ferroelectric alloy, the deposition in step c) being carried out at a temperature suitable for obtaining the deposition and crystallization of the crystalline grains (12) of the rhombohedral ferroelectric alloy, possibly doped,with the component that segregates between the crystalline grains (12).

2. Method according to claim 1, characterized in that the growth sublayer (21) of step b) is a sublayer (21) of As2Te3, Sb2Te3, Bi2Te3, As2Se3, Sb2Se3 or Bi2Se3.

3. A method according to claim 1 or 2, characterized in that it comprises a step of passivating the substrate (10) before step b), followed by the deposition of an underlayer (21) of As2Te3, Sb2Te3, Bi2Te3, As2Se3, Sb2Se3 or Bi2Se3 on the passivated surface of the substrate (10).

4. A process according to any one of claims 1 to 3, characterized in that the ferroelectric alloy corresponds to a rhombohedral GeSeï xTex alloy optionally doped.

5. A method according to claim 4, characterized in that step c) is carried out from one or more targets leading in the ferroelectric layer (11) to an atomic composition of Ge, Se and Te in which Se+Te represents from 50 to 75 atomic %, preferably from 50 to 60 atomic %, and in particular from 50 to 55 atomic %.

6. A process according to any one of claims 1 to 3, characterized in that the ferroelectric alloy corresponds to rhombohedral α-GeTe possibly doped, or to a rhombohedral GeixSnxTe alloy possibly doped.

7. A method according to claim 6, characterized in that step c) is carried out from one or more targets leading in the ferroelectric layer (11) to an atomic composition of Ge and Te or Ge, Sn and Te in which Ge or Ge+Sn, respectively, represents from 50 to 75 atomic%, preferably from 50 to 60 atomic%, and in particular from 50 to 55 atomic%.

8. A method according to claim 6 or 7, characterized in that the component which does not insert into the crystalline grains of a-GeTe or Gei_xSnxTe is Ge or Sn and step c) leads to the formation of rhombohedral crystalline grains (12) of a-GeTe or Gei xSnxTe, with amorphous Ge or Sn (or both) which segregate(s) between said crystalline grains (12).

9. A method according to any one of claims 6 to 8, characterized in that when the rhombohedral a-GeTe or the deposited rhombohedral Gei xSn xTe alloy is doped with a dopant, in particular As, Bi or Sb, step c) is carried out using a dopant target or a Ge, Sn and / or Te target incorporating the dopant.

10. A method according to any one of claims 1 to 9, characterized in that during step c), the growth of each crystalline grain (12) takes place along a preferred orientation of the crystallographic axis c perpendicular to the surface of the substrate (10).

11. A method according to any one of claims 1 to 10, characterized in that step c) is carried out by a vaporization technique, in particular by cathodic sputtering.

12. A method according to any one of claims 1 to 11, characterized in that in step c), the substrate (10) bearing the growth sublayer (21) is heated to a temperature sufficient to obtain the formation of crystalline grains (12), with this temperature remaining below 400°C, in particular the substrate (10) bearing the growth sublayer (21) is heated to a temperature in the range of 180 to 300°C.

13. A method according to any one of claims 1 to 12, characterized in that it comprises a step (d), subsequent to step (c), of depositing a protective layer on the ferroelectric layer (11) (30) to oxidation, said protective layer (30) preferably being a dielectric layer.

14. A method according to any one of claims 1 to 13, characterized in that it comprises a step d), subsequent to step c), of depositing on the ferroelectric layer (11) a dielectric protective layer (30), followed by a step e) of depositing an electrically conductive layer (40).

15. A method according to any one of claims 1 to 14, characterized in that the substrate (10) is based on or made up of crystalline silicon, amorphous silicon, silicon oxide, mica, CaF2, SiN, SiOC, SiCN, MgO, aluminium oxide, magnesium oxide, TiN, WSi, W, TiSiN, TaN, PtCoFe, CoFe, Au, Al, a ferromagnetic alloy, carbon, or a polymer resistant to the heating temperature applied during step c).

Images