Electrooptical device and method for producing electrooptical device

Abstract

To provide a method for localized formation of an on-silicon substrate lithium niobate layer portion, which has optimized cost and / or compatibility.SOLUTION: A device (1) includes, stacked in a vertical direction (z), a silicon-based substrate (10), a nucleation layer (20) made of a nitride-based refractory material, a lithium niobate-based layer portion (30), called an LNO portion, made of mesa on the nucleation layer (20), the LNO portion being bordered by a masking layer. The device (1) further includes at least one electrode (51, 51b, 52) configured to apply an electric field to the LNO portion (30). A method for producing such a device is also disclosed including forming the LNO portion by localized epitaxy.SELECTED DRAWING: Figure 12

Description

[Technical field]

[0001] The present invention relates generally to a method for producing an electro-optical component and layers based on Pockels effect materials forming the core of said component, and more particularly to producing an optical phase modulator. [Background technology]

[0002] These phase modulators can be advantageously used in circuits of the "optical phase array" (OPA) type, for example for designing laser remote sensing systems, called LIDAR (Laser Imaging Detection and Ranging) systems.

[0003] The principle of optical phase modulators is typically based on the local modulation of the refractive index of the material in which the light wave propagates. This change in refractive index can advantageously be obtained for certain crystalline materials without axial symmetry by the Pockels effect.

[0004] That is the case in particular for lithium niobate (LiNbO3) and tantalum niobate (LiTaO3). Their intrinsic properties make it possible to consider producing modulators that operate at high frequencies with low optical losses (typically on the order of a few decibels for a phase shift of π) and low energy consumption. However, the synthesis of thin layers of these materials is complex and the requirements regarding their crystalline quality are high. A thin LiNbO3 or LiTaO3 layer must in particular have a high crystalline quality and a controlled stoichiometric composition. It must also have a controlled crystalline orientation according to the architecture and the position of the electrodes of the electro-optical component to be produced. In the following, for example and for the sake of brevity, only LiNbO3 will be mentioned. The properties of LiTaO3 and Li(Nb,Ta)O3 alloys are very close to those of LiNbO3. It is understood that LiTaO3 or Li(Nb,Ta)O3 can directly replace LiNbO3.

[0005] Solutions have been developed that allow the synthesis of LiNbO3 in a thin layer on a sapphire substrate. This type of substrate is expensive and the applications for these solutions remain limited. The production of thin LiNbO3 layers on silicon-based substrates presents considerable difficulties. This allows the consideration of a new generation of multifunctional systems with electro-optical, acoustic, microelectronic and / or quantum devices integrated together on one and the same substrate. The solution consists of transferring a thin LiNbO3 layer from a donor substrate, typically a monocrystalline LiNbO3 substrate, to a receiver substrate, typically a silicon substrate with a surface oxide layer. This solution, which implements a number of technological steps, is quite costly. The feasible range of thicknesses of the thin layers is typically greater than 200 nm and less than 1 μm, but is also limited by the technological stresses of transfer and thinning. The uniformity of the thickness of the thin layers obtained by transfer is generally not sufficient for the targeted applications. The standard deviation of the thickness of the 200 nm thin LiNbO3 layers obtained by transfer typically reaches about 50 nm. This type of transfer further results in a "solid plate" over the entire surface of the receiver substrate. It is then necessary to etch the transferred LiNbO3 layer to obtain localized LiNbO3 layer portions on the receiver substrate. The etching of LiNbO3 is complex to control, in particular to obtain a suitable roughness for optical applications. Problems of contaminating the production line with lithium can occur during etching. During the solid plate transfer, it is also necessary to manage the mechanical stresses related to the thermal expansion difference between silicon and LiNbO3. Thus, the formation of localized LiNbO3 portions on the receiver substrate is difficult and expensive.

[0006] Therefore, there is a need for locally producing lithium niobate-based layers on silicon-based substrates that have crystalline quality compatible with target applications while limiting manufacturing costs. Summary of the Invention [Problem to be solved by the invention]

[0007] It is an object of the present invention to meet, at least in part, this need.

[0008] In particular, the object of the invention is to propose a method for the localized formation of lithium niobate layer portions on a silicon substrate, with optimized costs and / or compatibility with respect to current methods, such a method being advantageously implemented to produce devices comprising lithium niobate layer portions.

[0009] Another object of the invention is to propose a device comprising a lithium niobate layer portion, typically an electro-optical device of the phase modulator type.

[0010] Other objects, features, and advantages of the present invention will become apparent from a consideration of the following description and accompanying drawings, and it will be understood that other advantages may be incorporated. [Means for solving the problem]

[0011] To this end, according to an embodiment, a method for producing a device comprising a lithium niobate or lithium tantalate layer portion is provided, said method comprising the following steps: - providing a silicon based substrate; - forming on a substrate a nucleation layer made of a nitride-based refractory material; - forming a mask layer over the nucleation layer, the mask layer comprising at least one opening exposing a portion of the nucleation layer; - forming by epitaxy a portion of a lithium niobate or lithium tantalate based layer, called LNO portion, made of a mesa on the exposed portion of the nucleation layer in at least one opening; - preferably forming two electrodes configured to apply an electric field to the LNO portion; Includes.

[0012] Within the scope of the development of the invention, specifications for the nucleation layer are established. The nucleation layer should preferably have the following properties: - have a crystal structure and / or mesh parameters that are compatible with the silicon-based substrate, in particular to allow the nucleation layer to be grown epitaxially on the substrate; - have a crystal structure and / or mesh parameters close to or compatible with lithium niobate, which allows the epitaxial growth of the LNO portion on the nucleation layer; - preferably has a coefficient of thermal expansion close to that of lithium niobate, which makes it possible to limit the appearance of cracks due to temperature fluctuations during the formation of the LNO part; - In particular, the diffusion of Li atoms in Si or SiO2 can be blocked, which makes it possible to avoid the loss of stoichiometric composition of the LNO part.

[0013] The technical bias of the prior art is that such a nucleation layer must necessarily be oxide-based in order to avoid loss of stoichiometry of the LNO part due to diffusion of oxygen from the LNO part into the nucleation layer.

[0014] Conversely, to meet these specifications, a nucleation layer is selected according to the invention made of a nitride-based refractory material. For the sake of brevity, this nucleation layer made of a nitride-based refractory material is referred to below as a refractory nitride-based nucleation layer. Within the scope of the development of the invention, it has in fact been observed that such a refractory nitride-based nucleation layer, completely unexpectedly, allows epitaxy of LNO moieties under good conditions. Surprisingly, it has also been observed that such a nucleation layer makes it possible to block both the diffusion of oxygen and the diffusion of lithium into the silicon-based substrate. This results in the epitaxially grown LNO moieties on such a nucleation layer maintaining the required stoichiometric composition.

[0015] Moreover, the different crystal structures and mesh parameters of refractory nitrides, typically III-N nitrides, are fully compatible with those of lithium niobate and silicon. Such nucleation layers can therefore be advantageously epitaxially grown on silicon-based substrates, which then allow epitaxy of the LNO portion.

[0016] According to the invention, the epitaxy of LNO is carried out in a localized manner in at least one opening of the mask layer. This makes it possible to obtain only the LNO layer part. This makes it possible to limit the mechanical stress caused by the difference in the thermal expansion coefficient between lithium niobate and the nitride-based refractory material. The LNO part is subjected to less mechanical stress than a "solid plate" layer of lithium niobate. The crystalline quality of the LNO part and its electro-optical properties are best preserved after formation by epitaxy. The localized growth also makes it possible to limit the presence of lithium niobate on the entire plate. This avoids the use of lithium niobate etching, which is complicated to carry out, to obtain an LNO part of optical quality.

[0017] Epitaxy of LNO further allows better control of the thickness of the LNO part. In particular, epitaxy with good control of the thickness over the entire substrate allows for thin thicknesses of, for example, 200 nm or less to be obtained. Layer transfer techniques do not allow for such thicknesses to be obtained and / or do not allow sufficient precision and reproducibility over the entire substrate. The standard deviation in the thickness of a 200 nm thin LNO layer obtained by transfer can reach about 50 nm. Such thickness variations over the entire substrate are not compatible with applications aimed at electro-optics. On the contrary, advantageously, epitaxy of LNO allows for a standard deviation of thickness of less than 10 nm to be obtained for a layer that is 200 nm thick.

[0018] Thus, the present invention proposes a solution that allows to synthesize a thin LNO layer section directly on a silicon-based substrate, i.e. without a transfer step, via a refractory nitride-based nucleation layer, which LNO section is stoichiometric and of high crystalline quality thanks to the refractory nitride-based nucleation layer. Such a method can furthermore be directly integrated in a foundry with standard CMOS technology (based on n-type and p-type complementary metal-oxide-semiconductor transistors).

[0019] According to another aspect of the invention, there is provided a device comprising a lithium niobate-based or lithium tantalate-based layer portion, called LNO portion, made of a silicon-based substrate, a nucleation layer on said substrate, and a mesa on said nucleation layer, stacked in a vertical direction z. The LNO portion is bounded by a mask layer. The device further comprises two electrodes configured to apply an electric field to the LNO portion.

[0020] Advantageously, the nucleation layer of the device is made of a nitride-based refractory material. The advantages mentioned above apply mutatis mutandis. Such a device can be further directly integrated into silicon technology. It can be easily integrated together with other microelectronic devices or electro-optical devices or quantum devices, for example. Preferably, the device is a phase modulator type electro-optical device. An embodiment of the invention relates to a system comprising a plurality of devices on one and the same silicon-based substrate, each comprising an LNO part according to the invention, for example at least one electro-optical device, and an electronic component, such as a superconductor detector.

[0021] The aims, objects, features and advantages of the present invention will be best apparent from the following detailed description of embodiments of the invention, as illustrated by the accompanying drawings. [Brief description of the drawings]

[0022] [Figure 1]1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Diagram 2] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Diagram 3] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Figure 4] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Diagram 5] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Figure 6] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Figure 7] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Figure 8] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Figure 9] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Figure 10] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Figure 11] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Figure 12] 1-4 are cross-sectional views illustrating method steps for producing a device according to an embodiment of the present invention. [Figure 13] FIG. 13 is a top view of a device according to the embodiment illustrated in FIG. [Figure 14] FIG. 2 is a cross-sectional view showing a simulated distribution of light intensity propagating through a device according to an embodiment of the present invention. [Figure 15] 11A-11C are cross-sectional views illustrating variations of a device according to an embodiment of the present invention. [Figure 16] 11 is a cross-sectional view showing another variation of a device according to an embodiment of the present invention. [Figure 17] 1 is a cross-sectional view illustrating a system comprising an electro-optical device and a quantum device integrated together in accordance with an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] The drawings are given as examples and are not intended to limit the present invention. The drawings constitute schematic representations of principles intended to facilitate understanding of the present invention and are not necessarily scaled to actual applications. In particular, in the principle drawings, the thicknesses of various layers and parts, as well as the dimensions of patterns, do not represent reality.

[0024] Before commencing a detailed discussion of embodiments of the present invention, in the following optional features are described which may optionally be used in conjunction or alternatively.

[0025] According to an example, the LNO part is based on lithium niobate LiNbO3. According to an example, the LNO part is based on lithium tantalate LiTaO3. According to an example, the LNO part is based on a Li(Nb,Ta)O3 alloy.

[0026] By way of example, the nitride-based refractory material of the nucleation layer is selected to have or form: - resistance to oxidation, - A barrier to lithium diffusion.

[0027] This makes it possible to avoid the diffusion of oxygen and lithium from the LNO portion into the nucleation layer, thus preserving the stoichiometric composition of the LNO portion.

[0028] According to an example, the nitride-based refractory material of the nucleation layer is selected to have a crystalline structure, such as a hexagonal structure, that is compatible with the substrate and with the LNO part. This makes it possible to limit the appearance of structural defects during the formation of the LNO part by epitaxy. According to an example, the difference in mesh parameters between the nitride-based refractory material and the LNO material is less than 2%. This difference in mesh parameters can be taken in the subarray of regular cells of the LNO material. For example, in the case of LiNbO3, the mesh parameter of the regular cells is 5.148 Å. However, in the plane (0001), it is possible to determine a hexagonal or mini-cell subarray with a mesh parameter of 3.056 Å. According to an example, if the nitride-based refractory material is aluminum nitride AlN with a mesh parameter of 3.112 Å, the difference in mesh parameters between the LiNbO3 mini-cells and AlN is about 1.8%. LiNbO3 epitaxy can thus be carried out, surprisingly, on AlN. To align the AlN hexagonal array with the LiNbO3 minicells, in the (0001) plane, a twist of about 30° typically appears between the LiNbO3 and the AlN array.

[0029] By way of example, the nitride-based refractory material is taken from among the III-N refractory nitrides based on elements of group III, such as gallium nitride GaN, aluminum nitride AlN, and AlGaN alloys.

[0030] According to an example, the substrate is silicon-based with a (111) orientation, the nucleation layer is aluminum nitride AlN-based with a (0001) orientation, and the LNO part is oriented with a (0001) orientation.

[0031] According to an example, the LNO portion is in direct contact with the nucleation layer, specifically, there is no intermediate oxide layer between the nucleation layer and the LNO portion.

[0032] According to an example, the LNO part is topped by an aluminum-based seal, for example aluminum nitride AlN-based or sapphire Al2O3-based. This aluminum-based seal typically forms the seal of the LNO part together with the nucleation layer. This makes it possible to avoid the diffusion of oxygen and / or lithium from the LNO part to the upper layer. Thus, the stoichiometric composition of the LNO part is preserved. Compatibility with CMOS technology is preserved. The electro-optical properties of the LNO part are preserved or even improved.

[0033] According to an example, the so-called upper electrode of the two electrodes is disposed above the LNO part and the so-called lower electrode of the two electrodes is disposed below the LNO part. This device architecture is specifically configured to use a so-called "Z-cut" oriented LNO part (0001).

[0034] According to an example, the device is configured to form an electro-optical device, such as an optical phase modulator. Specifically, phase modulation of an optical wave propagating in the device is achieved by applying an electric field to the LNO portion via the Pockels effect. Application of an electric field will cause a variation in the refractive index of the LNO portion and a phase shift of the optical wave.

[0035] According to an example, the LNO part is carrying a silicon nitride SiN-based waveguide pattern configured to form an end waveguide with the LNO part.

[0036] According to an example, the LNO part is topped by a niobium nitride (NbN)-based structure, and the device forms a superconductor detector.

[0037] According to an example, the LNO portion has a thickness e3 between 50 nm and 500 nm, for example about 100 nm or 200 nm.

[0038] According to an example, the method further comprises at least partial removal of the mask layer by keeping the LNO portion formed locally in at least one opening after epitaxy of the LNO portion. This typically allows removing deposits of LNO material on the mask layer outside the opening. This eliminates a possible source of contamination for other devices of the plate on which the device with the lithium niobate layer portion is produced. This limits the effects of thermal expansion coefficient differences with the silicon substrate during subsequent manufacturing steps.

[0039] By way of example, the removal is performed by chemical mechanical polishing.

[0040] According to an example, the method further comprises, after epitaxy of the LNO part, the formation by epitaxy of an aluminum-based seal, for example aluminum nitride AlN-based or sapphire or alumina Al2O3-based, on the LNO part.

[0041] According to an example, the method further comprises, after epitaxy of the LNO portion, forming a waveguide pattern configured to form an end waveguide together with the LNO portion. According to an example, the waveguide pattern is silicon nitride SiN or TiO2, ZrO2, Ta2O5 based.

[0042] According to an example, the formation of the nucleation layer and the formation of the LNO part are carried out by pulsed laser deposition, preferably without ventilation between these two steps. This makes it possible to avoid exposure of the surface of the nucleation layer to air. The nucleation layer is not contaminated by water and / or hydrocarbons. The growth of the LNO part is optimized and the overall performance is improved. According to an example, the formation of the LNO part and the formation of the sealing part are carried out by pulsed laser deposition successively in one and the same reactor, without ventilation between said formations. This makes it possible to avoid exposure of the surface of the LNO part to air. The LNO part is not contaminated by water and / or hydrocarbons. This makes it possible to avoid intermediate surface cleaning steps, reducing the total duration of the method and limiting the costs. The overall performance is improved.

[0043] According to an example, the formation of the nucleation layer is configured such that said nucleation layer has a thickness e2 less than or equal to 200 nm, preferably less than or equal to 50 nm, which makes it possible to limit the appearance of structural defects during the formation of the nucleation layer by epitaxy.

[0044] According to an example, the formation of the LNO portion is configured such that said LNO portion has a thickness e3 after epitaxy between 50 nm and 500 nm, for example about 200 nm. Such a thickness e3 corresponds to a thin LNO layer. Such a thickness e3 cannot be obtained with good control of the thickness over the entire substrate by transfer and thinning of layers coming from a donor substrate separate from the receiver substrate.

[0045] According to an example, the method includes forming a top electrode on the LNO part and a bottom electrode below the LNO part, which in particular makes it possible to produce a phase modulator type electro-optical device of "Z-cut" architecture.

[0046] It will be understood that, unless inconsistent, all of the optional features and / or variations shown above may be combined to form embodiments not necessarily shown or described. Such embodiments are not expressly excluded from the invention. Features of aspects of the invention, e.g., devices or methods, can be applied to other aspects of the invention, mutatis mutandis.

[0047] It is expressly stated that within the scope of the present invention, terms such as "on", "surmounts", "covers", "underlying", "opposite" and their equivalents do not necessarily mean "in contact with". Thus, for example, deposition of a first layer on a second layer does not necessarily mean that the two layers are in direct contact with each other, but rather that the first layer at least partially covers the second layer, either by being in direct contact therewith or by being separated therefrom by at least one other layer or at least one other element.

[0048] A layer may furthermore consist of several sub-layers of one and the same material or of different materials.

[0049] By a substrate, stack, layer "based" on a material M or "M-based" is meant a substrate, stack, layer comprising only this material A or optionally other materials, e.g. alloying and / or doping elements. Thus, for example, a silicon-based substrate means a Si or doped Si, or a SiGe substrate. For example, an AlN-based layer means AlN, a doped AlN-based layer, or an AlN alloy, e.g. AlGaN. For example, a silicon nitride SiN-based waveguide can comprise non-stoichiometric silicon nitride (SixNy), or stoichiometric silicon nitride (Si3N4).

[0050] Several embodiments of the invention are described below which implement successive steps of a manufacturing method. Unless otherwise specified, the adjective "successive" does not necessarily mean that the steps immediately follow one another, even if this is generally preferred, but that intermediate steps may separate them.

[0051] Moreover, the term "step" means performing a part of a method and can refer to a set of sub-steps.

[0052] Moreover, the term "step" does not necessarily imply that acts performed during a step are simultaneous or immediately consecutive. In particular, an act of a first step may be followed by an act for a different step, and then another act of the first step may be resumed. Thus, the term "step" does not necessarily imply that a single act is inseparable in time and in a sequential phase of a method.

[0053] A preferably orthonormal system with x, y and z axes is represented in the accompanying drawings. When one single system is represented on one and the same set of figures, this system applies to all the figures of this set.

[0054] In this patent application, the thickness of a layer is taken in a direction perpendicular to the main extension plane of the layer. Thus, a layer typically has a thickness along z. Relative terms such as "on", "surmounts", "under", "underlying", "inserted", "above", and "below" refer to positions taken in the z direction. This list of terms is not exhaustive. Other relative terms can be easily clarified, if necessary, by referring to the attached drawings.

[0055] The terms "vertical" and "vertically" refer to directions along z. The terms "horizontal", "horizontally", "lateral" and "laterally" refer to directions in the xy plane. Unless explicitly stated, thickness, height, and depth are measured along z.

[0056] An element that is "vertically aligned with" or "perpendicular to" another element means that these two elements are both arranged on one and the same straight line perpendicular to the plane, where the lower or upper surface of the substrate extends primarily on one and the same straight line, i.e. oriented vertically in the figure.

[0057] In the scope of the present invention, III-N nitrides are preferably nitride elements belonging to group IIIa (three valence electrons) of the periodic table, which group IIIa is grouped together with earth elements such as boron, aluminum, gallium, indium, etc. III-N nitrides considered in the present invention are specifically, for example, boron nitride BN, aluminum nitride AlN, gallium nitride GaN, indium nitride InN, and alloys thereof, as well as, without limitation, AlGaN, GaInN.

[0058] Within the scope of the present invention, a waveguide is intended to ensure the propagation of a light wave in a main propagation direction taken along the x-axis in the accompanying drawings. The light waves are preferably coherent, monochromatic and of wavelength λ. The light waves preferably propagate according to one single optical propagation mode, typically the fundamental optical mode. Thus, the waveguide can be configured to guide a single fundamental transverse electric (TE00) or transverse magnetic (TM00) mode of the light wave. Alternatively, the guide can be configured to allow the conversion of the TM mode into the TE mode (and vice versa) locally during the propagation. The light wave can have several secondary modes in addition to the fundamental mode. If only the fundamental mode can propagate therein, the guide is called single-mode. If the guide allows at least two modes, it is considered multimode.

[0059] In the examples described below, the LNO portions shown are lithium niobate LiNbO3 based. It is understood that in these examples LiTaO3 or Li(Nb,Ta)O3 can directly replace LiNbO3. The LNO portions protrude from the underlying nucleation layer. The LNO portions are shown in the form of mesas or pads with respect to the nucleation layer. The LNO portions are bounded entities. It is not an abstract and arbitrary slice of a broader layer that covers, for example, the entire plane. The LNO portions are specifically defined by the cross-section of an opening in a mask layer.

[0060] For example, X-ray diffraction analysis in the 2θ configuration or rotation about φ and / or Ω (phi and omega scans) can be performed to determine the crystalline quality of the LNO layer, the nucleation layer, and / or the sealing layer and their epitaxial relationship.

[0061] 1 to 12 show a first embodiment of an electro-optical device 1 according to the invention.

[0062] As shown in Fig. 1, a substrate 10 is first provided. The substrate 10 may be a SOI (Silicon On Insulator) type substrate. Such a substrate 10 typically comprises, stacked along z, a support layer 11, typically a so-called silicon "bulk" substrate, an insulating layer 12, typically a so-called "box" buried oxide layer, and a silicon-based semiconductor layer 13, e.g. a so-called top Si silicon layer. The insulating layer 12 is typically a 1000 Å thick layer with a thickness e of about 2 μm. 12 The top Si layer 13 typically has a thickness e of about 5 nm to 20 nm. 13 Preferably, the top Si layer 13 is made of silicon oriented along (111). Other crystal orientations are possible for the silicon layer 13, in particular (001), depending on the architecture desired for the device.

[0063] As shown in FIG. 2, the top Si layer 13 can be locally doped once or several times, so that it locally has silicon portions 13p that have a doping difference with the rest of the layer 13. According to an example, the whole layer 13 is p-doped, for example by implantation, then the layer 13 is masked with the portions 13p and a new doping by implantation is carried out. It is thus possible to obtain a p+ doped layer 13+ and a p-doped portion 13p. The p-doped portion 13p can form the bottom electrode of an electro-optical device. For example, in the range of optical applications, to form a phase modulator, it is typically preferable to use a p-type doping in order to limit the optical losses in the device. The portion of the p+ doped layer 13+ can be arranged to be in electrical contact with the portion 13p, for example during a subsequent recontact.

[0064] As shown in Fig. 3, a nucleation layer 20 is formed on the top Si layer 13 with a p-doped portion 13p. The nucleation layer 20 is preferably based on a III-N material. The nucleation layer 20 preferably has a hexagonal crystal structure. Such a III-N material structure allows epitaxy of the on-silicon nucleation layer 20, in particular oriented along (001) and / or (111). This also allows later epitaxial growth of LNO material with different crystal orientations.

[0065] The nucleation layer 20 is, for example, aluminum nitride AlN-based. AlN can be deposited and / or epitaxially grown on silicon oriented along (111), the deposition methods of which are advantageously known and controlled. Aluminum nitride further has mesh parameters close to those of lithium niobate. This facilitates the epitaxial growth of lithium niobate, which is then formed in the LNO part. The thermal expansion coefficient of AlN is further between that of LiNbO3 and that of Si. This makes it possible to limit the appearance of structural defects in the LNO part during temperature variations for the different deposition steps. AlN also has good resistance to oxidation and to LNO deposition methods (it is resistant to ionized species plasmas and chemical precursors, making it possible to synthesize LNO materials). Boron nitride and gallium nitride have properties similar to those mentioned above for aluminum nitride. They can also be advantageously selected for the nucleation layer 20.

[0066] The nucleation layer 20 can be formed by physical or chemical deposition techniques, for example by pulsed laser deposition (PLD). The nucleation layer 20 can alternatively be formed by one of the following techniques: chemical vapor deposition (CVD), preferably metal-organic chemical vapor deposition (MOCVD), PVD sputtering, plasma-enhanced atomic layer deposition (PEALD).

[0067] The nucleation layer 20 is preferably monocrystalline. The nucleation layer 20 has a preferred orientation. Typically, for an AlN nucleation layer 20 on a silicon top Si layer 13, an orientation along the growth plane (0001) can be selected (111). The nucleation layer 20 is preferably stoichiometric. The nucleation layer 20 typically has a thickness e2 between 5 nanometers and 20 nanometers, for example about 10 nm.

[0068] The use of a refractory nitride, in particular an AlN nucleation layer 20, advantageously makes it possible to carry out heteroepitaxy of LiNbO3 on a silicon-based substrate 10. In particular, epitaxy makes it possible to obtain a LiNbO3 layer of stoichiometric composition that is preferably oriented and of high crystalline quality. The thickness of this LiNbO3 layer obtained by epitaxy is furthermore perfectly controlled. The LiNbO3 layer can be formed directly on substrates 10 of different sizes without an intermediate transfer step. This advantageously makes it possible to reduce the manufacturing costs of the LiNbO / AlN / Si layer stack.

[0069] As shown in Fig. 4, a mask layer 300 is then formed on the nucleation layer 20 and then structured. The mask layer 300 may typically comprise different layers 301, 302, 303 stacked along z, for example a SiO2-based layer 301, a SiN-based layer 302, a SiO2-based layer 303. The SiN-based layer 302, which is inserted in the mask layer, typically serves as a stop layer, for example during a subsequent chemical mechanical polishing (CMP) step. The SiO2-based layer 301 is typically formed with a thickness e of about 100 nm to 500 nm. 301 The SiN-based layer 302 typically has a thickness e of 5 nm to 50 nm. 302 The SiO2-based layer 303 typically has a thickness of about 100 nm to 500 nm. 303 The thickness of the layer 303 is e 303 is preferably thicker than the thickness provided for the LNO portion, for example twice as thick.

[0070] This mask layer 300 is configured, for example by standard lithography and etching, to form one or more openings 3a that open onto the nucleation layer 20. The openings 3a typically have a lateral dimension l3 of a few microns, for example between 1 μm and 100 μm.

[0071] 5, a lithium niobate LiNbO3-based portion 30 (LNO) is then formed in the opening 3a on the exposed portion of the nucleation layer 20. This LNO of portion 30 can be formed by a physical or chemical deposition technique, for example by pulsed laser deposition (PLD). This portion 30 can alternatively be formed by one of the following techniques: chemical vapor deposition (CVD), preferably metal-organic chemical vapor deposition (MOCVD), PVD sputtering, molecular beam epitaxy (MBE).

[0072] The LNO portion 30 is advantageously epitaxially grown on the nucleation layer 20. The LNO portion 30 is preferably monocrystalline. The LNO portion 30 has a preferred orientation. Typically, an orientation along the growth plane (0001) can be chosen. This makes it possible to maximize the Pockels effect in the LNO portion 30 with electrodes located respectively below and above the LNO portion 30 along z. Other orientations can be chosen according to the desired device architecture and / or the desired application.

[0073] The LNO portion 30 formed on the nucleation layer 20 is preferably of stoichiometric composition, for example Li1Nb1O3. The atomic percentage of lithium is ideally close to that of niobium. By way of example, the atomic percentage of lithium and niobium is about 20%. The LNO portion 30 has a thickness e3 between 50 nanometers and 500 nanometers, for example 200 nm. The thickness e3 of the LNO portion 30 can be selected according to the desired application.

[0074] After the formation of the LNO portion 30 in the opening 3a, a sealing portion 31, typically based on AlN, is preferably formed by epitaxy on the LNO portion 30. The portion 31 has a thickness e between 5 nanometers and 20 nanometers, for example about 10 nm. 31 The sum of the thicknesses of the portions 30, 31 is typically less than the thickness e of the layer 301 underlying the layer 302. 301 Layer 302 therefore remains above portions 30, 31 along z. This allows layer 302 to be used as a layer to detect the end of polishing during the subsequent CMP polishing.

[0075] According to the option, the sealing portion 31 and the LNO portion 30 are both produced in situ by PLD in one and the same growth reactor. The growth of the sealing portion 31 can thus be produced immediately after the end of the growth of the LNO portion 30. This makes it possible to avoid venting the LNO portion 30 before the epitaxy of the sealing portion 31. The surface of the LNO portion 30 therefore remains clean. This avoids an intermediate cleaning step. The duration of the method is thus reduced. This also makes it possible to limit the appearance of roughness during the formation of the sealing portion 31. The surface condition of the latter is thus optimized.

[0076] During epitaxy of the portions 30, 31 in the opening 3a, growth residues 30r, 31r may be deposited around the opening 3 on the mask layer 300.

[0077] As shown in figure 6, the opening 3 can then be filled with a dielectric filling material 32, typically SiO2, which makes it possible to planarize the structure.

[0078] As shown in Fig. 7, after filling, a chemical mechanical polishing (CMP) is performed to remove the growth residues 30r, 31r. The SiN-based layer 302 is advantageously used as a stop layer or as a layer for detecting the end of the polishing. The polishing is typically stopped at the layer 301 immediately after the layer 302 has been removed. After CMP, the portions 30, 31 remain bounded by the portion 301 of the layer and are surmounted by the material filling portion 32.

[0079] As shown in Figure 8, after polishing, a new opening 3b is made above the portions 30, 31 to expose the surface 310 of the sealing portion 31. This new opening 3b typically has a lateral dimension l3 that is smaller than or approximately equal to that of the opening 3a. This new opening 3b is typically centered on the portions 30, 31. The opening 3b can be made typically by lithography and etching.

[0080] According to a preferred option, as shown in FIG. 9, a SiO2 based spacer layer 41 and then a layer 40 are then formed on the encapsulation part 31. In this case, the layer 40 is intended to be configured in the form of a waveguide pattern in order to guide the propagation of light in the device along x. The layer 40 typically has a thickness e4 of about 100 nm to 800 nm according to the targeted application, for example according to the wavelength of the light to be guided and the materials used. The layer 40 may for example be based on SiN, TiO2, ZrO2, Ta2O5. Alternatively, this layer 40 may be based on amorphous silicon Si-a. The spacer layer 41 may be conformally deposited in the opening 3b. The spacer layer 41 typically has a thickness e4 of about 200 nm. 41 This spacer layer 41 makes it possible to control the optical coupling between the waveguide pattern and the LNO portion, in particular to limit the optical coupling losses during the passage between a zone only comprising the waveguide pattern without the LNO portion and a hybrid zone comprising both the LNO portion and the waveguide pattern.

[0081] As shown in FIG. 10, the layer 40 is typically structured by lithography and etching to form a waveguide pattern 4. The waveguide pattern 4 typically has a lateral dimension l4 along y between 200 nm and 2000 nm and a height e4 along z between 100 nm and 800 nm. The waveguide pattern 4 is configured to guide the propagation of light in the device along x in the illustrated example. The propagation can be along y or, more generally, in the plane xy, according to the design of the waveguide pattern 4, if the orientation of the LNO part is Z-cut. The waveguide pattern 4 can form an end waveguide together with the underlying LNO part 30. This allows guiding the propagation of TM optical modes in a phase modulation device based on a Z-cut LNO part.

[0082] 11, after the creation of the waveguide pattern 4, an encapsulation with a dielectric material 33, typically SiO2, is performed on and around the waveguide pattern 4. A CMP planarization step can be performed after the filling with the dielectric material 33 to obtain a flat surface 330.

[0083] As shown in FIG. 12, after planarization, interconnects 50v can be performed in the form of through vias to electrically connect the p-doped silicon parts 13p forming the bottom electrode 51 of the device 1. These interconnects 50v typically pass through the layer 301 and the nucleation layer 20 and have ends in contact with the p+ doped top Si layer 13+. A contact level can now be created on the front side of the device 1. This level typically comprises contacts 50 connected to the through vias 50v and a top electrode 52 arranged above the LNO part 30. Thus, an electro-optical device 1 is formed. Advantageously, this device 1 comprises a (0001) oriented LNO part 30 on an oriented silicon substrate (111). The LNO part 30 is typically inserted between two layers AlN 20, 31 (0001). The device 1 further comprises a bottom electrode 51 arranged below the LNO part 30 and a top electrode 52 arranged above the LNO part 30. Optionally, the device 1 comprises an end waveguide pattern 4 on which the LNO section 30 rests. This electro-optical device 1 advantageously makes it possible to modulate the phase of an optical wave propagating along x in the waveguide formed by the LNO section 30 and the end waveguide pattern 4 by the Pockels effect through the application of an electric field between the bottom electrode 51 and the top electrode 52.

[0084] Figure 13 shows the electro-optical device 1 in a top view, with cross section AA corresponding to the side view shown in Figure 12. The design of the co-contacts 50 and electrodes 52 can be done according to rules known to those skilled in the art to minimize the light propagation losses.

[0085] Figure 14 shows the simulation results showing the distribution of light waves in the waveguide formed by the LNO section 30 and the end waveguide 4 in the device 1. The light waves have a wavelength λ = 1.55 μm in this case. The top Si layer 13 has a thickness e 13= 20 nm. The AlN nucleation layer 20 has a thickness e2 = 20 nm. The LNO part 30 has a thickness e3 = 300 nm. The AlN encapsulation layer 40 has a thickness e4 = 20 nm. The waveguide pattern 4 has a height e4 = 200 nm and a width l4 = 1200 nm. The actual refractive index of the thus formed waveguide is 2.35. The coverage of the light wave on the LNO part is in this case 50%. Modulation of the refractive index of the LNO part 30 by the Pockels effect in this device therefore makes it possible to efficiently modulate the phase of the light wave. An optical phase modulator comprising an on-silicon LNO part is thus advantageously produced. This phase modulator can be integrated into an on-silicon Mach-Zehnder architecture.

[0086] 15 shows a variant of the electro-optical device 1, in which the waveguide is entirely formed by an LNO portion 30, inserted between the encapsulation portion 31 and the nucleation layer 20, without any other protruding waveguide. This type of waveguide typically has a height e between 100 and 500 nanometers. 3’ , and a width l3 along y between 500 nanometers and 2000 nanometers. This type of waveguide requires fewer manufacturing steps and is cheaper to produce.

[0087] 16 shows another variant of the electro-optical device 1, in which a bottom electrode 51b is formed on the back side of the device, below a doped silicon-based support layer 11. The top Si layer 13 is in this case undoped. This reduces the optical losses of the device 1.

[0088] Other applications are also conceivable: localized on-silicon LNO epitaxial can be advantageously implemented to integrate different devices with different functions together.

[0089] FIG. 17 shows a system according to the invention comprising an electro-optical device 1 and a niobium nitride NbN-based superconductor detector 2, as described above, on one and the same silicon-based substrate 10. It is then possible to carry out, in a second opening in the mask layer, the epitaxial formation of a second LNO portion 30b, then an NbN structure 6 on LNO or an NbN / AlN / LNO epitaxial formation. This makes it possible to envisage integrating together different devices 1, 2 formed partly by the same growth localized in different openings. The LNO portion advantageously makes it possible to epitaxially grow transition metal nitride-based superconductor materials such as NbN or TiN or NbTiN or others. The LNO portion typically forms a nucleation layer for these superconductor materials. Their mesh parameters are close and their thermal expansion coefficient difference is small. The formation of the different layers and portions can advantageously be carried out successively by epitaxy. The overall cost of the method for manufacturing a system comprising the electro-optical device 1 and the superconductor detector 2 is reduced.

[0090] From the above it becomes clear that the invention advantageously makes it possible to form localized portions of thin LNO layers of good crystalline quality on silicon-based substrates with a nucleation layer made of nitride-based refractory materials, typically based on III-N materials. These localized LNO portions can advantageously be directly integrated into electro-optical or quantum devices. Other applications are also conceivable. The invention is not limited to the embodiments described above. [Explanation of symbols]

[0091] 1. Device, electro-optical device 2. Superconductor detector 3 aperture 3a aperture 3b aperture 4 Silicon nitride (SiN) based waveguide pattern, waveguide pattern, end waveguide, end waveguide pattern 6 Structures, NbN structures 10 Silicon-based substrates 11 Doped silicon-based support layer 12 Insulating layer 13 Top Si layer, silicon-based semiconductor layer 13p p-doped silicon part 13+ p+ doped top Si layer, p+ doped layer 20 Nucleation layer 30 Lithium niobate-based layer portion, Lithium niobate-based or Lithium tantalate-based layer portion, LNO portion 30b Second LNO section 30r growth residue 31 Sealing part 31r growth residue 32 Material filling part, dielectric filling material 33 Dielectric Materials 40 layers 41 Spacer layer 50 common points 50v interconnect 51 Lower electrode 51b Lower electrode 52 Upper electrode 300 Mask Layers 301 SiO2 based layer 302 SiN-based layer 303 SiO2 based layer 310 Surface 330 surface

Claims

1. 1. A device (1) comprising a silicon-based substrate (10), a nucleation layer (20) on said substrate (10) made of a nitride-based heat-resistant material, and a mesa-shaped lithium niobate- or lithium tantalate-based layer portion (30) on said nucleation layer (20), called an LNO portion, said LNO portion being bounded by a mask layer, said device (1) further comprising two electrodes (51, 51b, 52) configured to apply an electric field to said LNO portion (30).

2. The device (1) according to claim 1, wherein the nitride-based refractory material is taken from among the III-N refractory nitrides based on elements of group III, such as gallium nitride GaN, aluminum nitride AlN, and AlGaN alloys.

3. 3. The device (1) according to claim 1 or 2, wherein the substrate (10) is silicon-based with a (111) orientation, the nucleation layer (20) is aluminum nitride AlN-based with a (0001) orientation, and the LNO portion (30) is oriented in the (0001) orientation.

4. The device (1) according to any one of claims 1 to 3, wherein the LNO part (30) carries an aluminium-based sealing part (31), for example aluminium nitride AlN-based or sapphire Al2O3-based.

5. 5. The device (1) according to claim 1, wherein the so-called upper electrode (52) of the two electrodes is arranged above the LNO part (30) and the so-called lower electrode (51, 51b) of the two electrodes is arranged below the LNO part (30).

6. The device (1) according to any one of the preceding claims, adapted to form an electro-optical device, such as an optical phase modulator.

7. The device (1) according to any one of the preceding claims, wherein the LNO part (30) carries a waveguide pattern (4) configured to form an end waveguide together with the LNO part (30).

8. 6. The device (1) according to any one of claims 1 to 5, wherein the LNO part (30) carries a structure (6) made of a transition metal nitride based superconductor material, such as niobium nitride NbN, titanium nitride TiN or a NbTiN alloy, and the device forms a superconductor detector (2).

9. A system comprising at least one electro-optical device (1) according to claim 6 or 7 and electronic components, e.g. a superconductor detector (2) according to claim 8, on one and the same silicon-based substrate (10).

10. Providing a silicon-based substrate (10); forming a nucleation layer (20) made of a nitride-based refractory material on said substrate (10); forming a mask layer (300) on said nucleation layer (20) comprising at least one opening (3a) exposing a portion of said nucleation layer (20); forming by epitaxy in said at least one opening (3a) a mesa-shaped lithium niobate or lithium tantalate based layer portion (30), called the LNO portion, on said exposed portion of said nucleation layer (20); forming two electrodes (51, 51b, 52) configured to apply an electric field to the LNO part (30); A method for producing a device (1) according to any one of claims 1 to 8, comprising:

11. 11. The method according to claim 10, further comprising, after epitaxy of the LNO portion (30), at least partial removal of the mask layer (300) by keeping the LNO portion (30) locally formed in the at least one opening (3a).

12. The method of claim 11 , wherein the removing is performed by chemical mechanical polishing.

13. 13. The method according to claim 10, further comprising forming, after epitaxy of the LNO part (30), an aluminum-based sealing part (31), for example an aluminum nitride AlN-based or sapphire Al2O3-based sealing part (31) on the LNO part (30) by epitaxy.

14. 14. The method according to any one of claims 10 to 13, further comprising, after epitaxy of the LNO part (30), the formation of a silicon nitride SiN-based waveguide pattern (4) configured to form a ridge waveguide together with the LNO part (30).

15. 15. The method of any one of claims 10 to 14, wherein the forming of the nucleation layer (20) and the forming of the LNO portion (30) are performed by pulsed laser deposition.

Images