Coupling module for a light source to a waveguide

Abstract

The invention relates to a coupling module, comprising a light source capable of emitting electromagnetic radiation at a first wavelength λ1, a coupler facing the light source comprising a conversion portion capable of absorbing electromagnetic radiation at the first wavelength λ1 to emit multidirectional electromagnetic radiation at a second wavelength λ2 strictly greater than the first wavelength λ1, a waveguide optically coupled to the coupler.

Description

Description Title: Coupling module of a light source to a waveguide TECHNICAL FIELD [ooi] The field of the invention is that of optical coupling modules of a light source to a waveguide, in particular for optical interconnections between electronic chips or for display devices. PREVIOUS STATE OF THE ART

[0002] Optical coupling modules for a light source to a waveguide are essential to many optoelectronic devices. For example, they have enabled the replacement of electrical interconnections between electronic chips with faster optical interconnections. These optical interconnections are used, for instance, as communication links between, on one side, graphics processing units (GPUs) or microprocessors, and on the other, high-capacity memories. Among these memories, some are high-bandwidth memory (HBM). This can consist of a stack of several DRAMs, typically a number of DRAMs equal to a power of 2, mounted on a control chip.

[0003] Optical interconnects enable the rapid exchange of large amounts of data. They are generally arranged, at least partially, in an interposer on which electronic or optoelectronic chips are mounted to communicate with each other. They allow for parallel optical links that are low-cost, have moderate power consumption, and low latency.

[0004] While optical connections have long since proven their advantages over long distances, they have been slow to penetrate the realm of short-distance applications, such as chip-to-chip connections. Their adoption in this market segment has long been hampered by the low performance of coupling modules, their high cost, or their bulkiness. In particular, it has been necessary to find suitable methods for generating the light signal and propagating it efficiently.

[0005] The light signal is emitted by a light source. In integrated photonics, there are several types of light sources. Lasers, however, are energy-intensive. Among these, vertical-cavity surface-emitting lasers (VCSELs) are the most interesting because they are compact, However, they have the disadvantage of being sensitive to temperature variations. Furthermore, the use of modulators optically coupled to a laser only increases an already significant current consumption.

[0006] Recently, coupling modules for optical interconnects have been proposed that use a light-emitting diode (LED) as the light source. LEDs are compact and easy to manufacture. They consume less energy than lasers. The light output of an LED can easily be modulated in amplitude by directly modulating the electric current applied to its terminals.

[0007] US11515356 B2 describes an optical interconnect system between a central processing unit (CPU) and a memory chip. The CPU and memory are two electronic chips mounted on an interposer. A waveguide array extends through the interposer. One waveguide connects a gallium nitride LED on one chip to a photodetector on the other chip. The LED emits at a wavelength of 430 nm. Each waveguide in the array is made of a material transparent to the LED's wavelength, such as silicon oxide, silicon nitride, or silicon oxynitride.

[0008] Each LED is optically coupled to a waveguide by total internal reflection of the light emitted by the LED onto a face at 45° to the interposer or waveguide. In a particularly advantageous alternative embodiment, the LED is optically coupled to the waveguide by a microlens and a rotating mirror. This significantly improves the coupling.

[0009] However, it appears necessary to increase the coupling efficiency of the solutions proposed in this document and to make them less sensitive to the angular, spectral and alignment characteristics of the light beam emitted by the LED. DESCRIPTION OF THE INVENTION [ooio] The invention aims to remedy, at least in part, the drawbacks of the prior art, and more particularly to provide a coupling module comprising a light source, which is relatively insensitive to one or more properties of electromagnetic radiation emitted by the light source. The radiation emitted by the light source can be a useful light signal for chip-to-chip communication. The invention also relates to a method for manufacturing such a coupling module. [ooii] To this end, the object of the invention is a coupling module, comprising a light source capable of emitting electromagnetic radiation at a first wavelength λi, a coupler opposite the light source including a conversion portion capable of absorbing electromagnetic radiation at the first wavelength λi to emit multidirectional electromagnetic radiation at a second wavelength λ2 strictly greater than the first wavelength λi, and a waveguide optically coupled to the coupler. The coupling module further comprises a mirror formed of a metallic portion having a vertical part arranged opposite the waveguide so as to reflect a portion of the multidirectional electromagnetic radiation towards the waveguide. [ooi2] Some preferred but not limiting aspects of this coupling module are as follows. [ooi3] The waveguide can extend in a principal plane. The mirror can surround the conversion portion in the principal plane, over an angular opening of at least 270 degrees.

[0014] The light source can be a gallium nitride light-emitting diode. The first wavelength Δi can be within a spectral range from ultraviolet to blue. [ooi5] The second wavelength A2 can be included in the visible spectrum or in the infrared spectrum.

[0016] The waveguide and coupler can be optically coupled by butt coupling. The coupler and waveguide can all have refractive indices that are equal to within 10%.

[0017] The conversion section can occupy the entire volume of the coupler. The waveguide and the conversion section can be made of the same material.

[0018] The waveguide may have an extinction coefficient less than or equal to 1 E-3 at the second wavelength A2.

[0019] The mirror can be arranged opposite one end of the coupler, opposite the waveguide.

[0020] The coupling module may further include a substrate, in which the waveguide can extend along the substrate parallel to one of its upper surfaces. The waveguide may be a vertical multimode waveguide at the second wavelength Å2.

[0021] The invention also relates to a multi-chip module comprising a coupling module according to any one of the preceding characteristics, an interposer, and a The transmitting chip. The coupler and waveguide of the coupling module can be arranged in the interposer. The light source of the coupling module can be arranged in the transmitting electronic chip.

[0022] The multi-chip module may further include a photodiode optically coupled end-to-end with the waveguide by an input face of the photodiode.

[0023] The photodiode can be made of silicon. The input face can be oriented along a crystal plane (111) of the silicon.

[0024] The invention also relates to a method for manufacturing a coupling module according to any one of the preceding characteristics, comprising the following steps: formation of the waveguide and the coupler on a substrate, formation of the conversion portion in the coupler.

[0025] The waveguide and coupler formation step may include a substep of etching a trench that may have a proximal end for defining the sides and bottom of the coupler, with the portion adjacent to the proximal end of the trench for defining the sides and bottom of the waveguide. The waveguide and coupler formation step may further include a substep of depositing a layer of coupler and waveguide material to fill the trench, thus forming part of the coupler and part of the waveguide.

[0026] The manufacturing process may further include a step of forming a photodiode prior to the engraving of the trench.

[0027] The photodiode can be made of silicon, and the trench etching can include a substep of anisotropic wet etching of one input face of the photodiode. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Other aspects, objectives, advantages and features of the invention will become more apparent from the following detailed description of preferred embodiments thereof, given by way of non-limiting example, and made with reference to the accompanying drawings in which: Figure 1A is a schematic cross-sectional view of a first embodiment; Figure 1B is a schematic cross-sectional view of a second embodiment; Figure 2A is a schematic cross-sectional view of a third embodiment; Figure 2B is a schematic cross-sectional view of a fourth embodiment; Figure 2C is a schematic cross-sectional view of a fifth embodiment; Figures 3A to 3F are schematic top views of examples of couplers suitable for the invention, optically coupled to a waveguide; Figure 4 is a schematic cross-sectional view of a particularly advantageous waveguide in a coupling module according to the invention; Figures 5A to 5E are schematic cross-sectional views of steps in a first manufacturing process; Figures 6A to 6D are schematic cross-sectional views of steps in a second manufacturing process; Figures 7A to 7E are schematic cross-sectional views of a sequence of process steps common to a third, fourth, and fifth manufacturing process of a coupling module according to the invention integrated into a multi-chip module; Figures 8A to 8D are schematic cross-sectional views of steps in the third manufacturing process, subsequent to the sequence of process steps in Figures 7A to 7E;Figures 9A to 9D are schematic cross-sectional views of steps in the fourth manufacturing process, subsequent to the sequence of process steps in Figures 7A to 7E; Figures 10A to 10E are schematic cross-sectional views of steps in the fifth manufacturing process, subsequent to the sequence of process steps in Figures 7A to 7E. DETAILED DESCRIPTION OF SPECIFIC METHODS OF IMPLEMENTATION

[0029] In the figures and throughout the description, the same reference numerals represent identical or similar elements. Furthermore, the various elements are not drawn to scale to ensure clarity. Moreover, the different embodiments and variants are not mutually exclusive and may be combined. Unless otherwise stated, the terms "approximately," "around," and "in the order of" mean within 10%, and preferably within 5%. Furthermore, the terms "between ... and ..." and equivalents mean inclusive of the bounds, unless otherwise specified.

[0030] The invention relates to a coupling module comprising a waveguide and a coupler optically coupled to the waveguide. The coupler has a conversion section. During operation, the conversion section receives incident electromagnetic radiation emitted by a light source at a first wavelength λi. It is absorbed, at least In part, this is achieved by converting the energy into electromagnetic radiation, called "re-emitted," at a second wavelength Δ2 strictly greater than Δi. This conversion portion is, for example, a photoluminescent material. The photoluminescent material can include quantum dots or nanoplatelets, or be halogenated perovskite.

[0031] The re-emitted electromagnetic radiation is multidirectional, propagating in directions different from those of the incident electromagnetic radiation, thus confining it within the coupler. The second wavelength λ₂ can be sufficiently large compared to the first wavelength so that the re-emitted electromagnetic radiation is minimally or not at all absorbed by the conversion section. The re-emitted electromagnetic radiation can relax one or more electrons excited by the incident electromagnetic radiation in the conversion section, thereby generating stimulated electromagnetic radiation that can propagate preferentially toward the waveguide. The difference between the second and first wavelengths can, for example, result, at least in part, from a Stokes shift occurring within the conversion section.Once confined, the emitted electromagnetic radiation propagates through the waveguide.

[0032] The physical phenomena at work within the coupler and at the transition from the coupler to the waveguide are thus independent of the illumination conditions of the conversion section. Therefore, the coupling efficiency of the light source with the waveguide depends, to a first approximation, only on the energy received by the conversion section within a solid angle of acceptance of the conversion section, which defines the possible angles of incidence of photons entering the conversion section. Insertion losses are thus only weakly dependent on the angular, spectral, or alignment characteristics of the electromagnetic radiation emitted by the light source. In the case of a diopter separating the conversion section from its surroundings, the solid angle of acceptance is defined by Snell's law.

[0033] In the description, a "layer" is defined as an extent consisting of one or more sublayers of a material whose thickness along a z-axis is less than, for example, ten times or even twenty times, its longitudinal dimensions of width and length in a plane (x, y) perpendicular to the z-axis. A layer may be structured, or a structure of substantially constant thickness extending primarily along a principal plane. When it consists of several sublayers, the sublayers may be made of different materials. The sublayer(s) extend in planes substantially parallel to the (x, y) plane. When a layer possesses a property, it is understood that when it consists of several sublayers, all the sublayers possess the same property, unless explicitly stated otherwise. For example, in the absence of further details, a layer made of metal or of a semiconductor or amorphous material may contain several sub-layers, all respectively made of metal, a semiconductor material, or amorphous materials.

[0034] Throughout this description, two optical components are said to be optically coupled when electromagnetic radiation propagating in one induces the propagation of electromagnetic radiation in the other without any external energy input. Coupling can occur in various ways, for example, via direct coupling (also called end-to-end coupling), a diffraction grating, adiabatic, evanescent, or directional coupling. It may involve one or more intermediate optical components.

[0035] It is specified that an optical element is optically coupled to a waveguide by direct coupling when they jointly form an interface that allows light to be transferred from one to the other. When light is transferred from the optical element to the waveguide, the light excites an optical mode of the waveguide. For example, the optical element could be an optical coupler or another waveguide.

[0036] Two guided optical modes are said to be optically coupled when the power of one comes entirely from the power of the other, without any external energy input. Similarly, an optical mode is optically coupled to a light source when the power of the optical mode comes entirely from the power of the light source, without any external energy input.

[0037] Throughout this description, a waveguide is defined as a single-mode or multi-mode waveguide capable of confining light, or as an optical waveguide within which light propagates by total internal reflection (sometimes called a "light pipe"). Without further specification, a waveguide can be of any type. For example, it could be a ribbon, edge, or planar waveguide. A waveguide has a core and, optionally, one or more confinement layers surrounding the core so as to be in physical contact with it. A contrast or variation in refractive indices between the core and the confinement layer(s), a gas, or a vacuum can be used to confine the light. Waveguides are identified by their cores in the diagrams.Similarly, without further specification, a refractive index of a waveguide is a refractive index of the waveguide core; a distance separating two waveguides is the distance separating the cores of the respective waveguides; the material of a waveguide is the material of the waveguide core; when a waveguide extends in a direction, it is understood that the waveguide core extends in that direction. This direction; when a waveguide is in contact with a layer, we mean that the core of the waveguide is in contact with the layer. A layer can be conformal, which implies that it extends over a surface, for example a non-planar one, and that it conforms to that surface.

[0038] For clarity, in the description, when referring to a "refractive index" of a propagation medium for electromagnetic radiation of interest, this refers to the average refractive index of the propagation medium at the wavelength of the electromagnetic radiation of interest when the propagation medium is a structurally heterogeneous medium.

[0039] Specific embodiments will be described relating to a coupling module according to the invention integrated into a multi-chip module comprising a transmitting chip, a receiving chip, and an interposer. However, these embodiments can be integrated into all types of optoelectronic or photonic devices, such as, for example, a display device comprising pixels optically coupled to a waveguide array. A display device for which the embodiments of the invention are particularly advantageous is described in document FR3144675.

[0040] An optical interposer is an intermediate chip that provides a functional interface between two components, or between a component and a photonic circuit, or between two photonic circuits. This interface can perform functions such as optical coupling or optical signal transmission. The components can be electronic, optoelectronic, or photonic components, such as a waveguide or an optical fiber. The optical interposer incorporates integrated optical structures, such as waveguides and optical couplers.

[0041] A first embodiment of coupling module 1 is illustrated in Figure 1A. It is integrated into a multi-chip module comprising an interposer 30, a transmitter chip 10 and a receiver chip 20. Figure 1A is a cross-sectional view.

[0042] The interposer 30 comprises a substrate 100. The substrate 100 can be made of any type of material. It can be glass. In this case, it is a silicon wafer or part of a silicon wafer. The substrate 100 has an upper face 100.1 and a lower face opposite and substantially parallel to the upper face 100.1. Both the upper and lower faces are planar in this example.

[0043] We define here and for the rest of the description a direct three-dimensional orthogonal frame (X, Y, Z), where the X and Y axes form a plane parallel to the upper face 100.1 of the substrate 100, and where the Z axis is oriented substantially orthogonally to the upper face 100.1 of the substrate 100, from the lower face towards the upper face 100.1. In the following description, the terms "vertical" and "vertically" refer to an orientation substantially parallel to the Z-axis, and the terms "horizontally" and "horizontally" refer to an orientation substantially parallel to the (X, Y) plane. Furthermore, the terms "lower" and "upper" refer to increasing positioning as one moves away from the substrate 100 along the +Z direction. A "upper" (respectively, "lower") face or surface of an element is the face or surface substantially parallel to the (X, Y) plane with the highest (respectively, the lowest) Z-coordinate. The terms "lateral" or "transverse" refer to an orientation substantially parallel to the Z-axis. A top view is a view along a -Z orientation.

[0044] The interposer 30 can optionally be electrically connected to a printed circuit board (or PCB) on the underside of the substrate 100, for example by wire connections and / or vias passing through the substrate 100.

[0045] The coupling module 1 comprises a light source 210, a coupler 110, and a waveguide 120. The light source 210 emits electromagnetic radiation at a first wavelength λi, hereinafter referred to as the emitted electromagnetic radiation. It can, for example, be a laser. In this case, it can be a vertical-cavity surface-emitting laser (VCSEL), or any other type of integrated laser, possibly optically coupled to a light-extraction grating. Advantageously, it can be a light-emitting diode (LED) or a resonant-cavity light-emitting diode (RCLED). The light source 210 here is a gallium nitride (GaN) light-emitting diode, emitting at a first wavelength Δi within a spectral range, for example, from ultraviolet to blue.

[0046] The interposer 30 has a lower confinement layer 105 extending in contact with the upper face 100.1 of the substrate 100, in a plane substantially parallel to the (X, Y) plane. The waveguide 120 extends in a principal plane parallel to the upper face 100.1. It is in contact with the lower confinement layer 105 on one side of the lower confinement layer 105 opposite the substrate 100. For the purposes of illustration, it is shown here as straight, with a substantially constant cross-section, and it extends parallel to the X-axis. However, it can have any shape that allows it to guide light. It can be a single-mode waveguide. Advantageously, the waveguide 120 is multi-mode.

[0047] The coupler 110 is optically coupled to the waveguide 120. In this example, the waveguide 120 is optically coupled to the coupler 110 by an end-to-end coupling. The waveguide 120 and the coupler 110 have perfectly overlapping lateral sections at their respective ends, together defining a coupling surface. The coupler 110 extends parallel to the (X, Y) plane so as to be in contact with the lower confinement layer 105.

[0048] The coupler 110 includes a conversion portion 115. The conversion portion 115 preferably passes completely through the coupler 110 in a direction parallel to the Z-axis. The conversion portion 115 is made of a conversion material capable of absorbing electromagnetic radiation at the first wavelength Δi and emitting electromagnetic radiation at a second wavelength Δ2 strictly greater than the first wavelength Δi. In this example, the conversion portion 115 is made of a photoluminescent material such as a halogenated perovskite, or quantum dots, possibly embedded in a matrix.

[0049] The conversion portion 115 can occupy the entire volume of the coupler 110. Alternatively, the coupler 110 can include a passive portion which, together with the conversion portion 115, forms a partition of the coupler 110, as shown in Figure 1A. In other words, the passive portion and the conversion portion 115 are two complementary parts of the coupler 110, together forming the entire coupler 110. In this example, the passive portion is made of the same material as the waveguide 120. The conversion portion 115 faces the light source 210 so as to receive at least a portion of the electromagnetic radiation emitted by the light source 210, hereafter referred to as incident electromagnetic radiation, when the coupling module 1 is operating. This conversion portion 115 can, for example, have all its horizontal dimensions less than or equal to 20 pm.

[0050] The waveguide 120 can be optically coupled to any type of photonic component, active or passive. In this embodiment, it includes a light extractor 125 at one end of the waveguide 120 opposite the coupler 110. The light extractor 125 is capable of extracting at least a portion of one or more optical modes guided by the waveguide 120, at the second wavelength λ2. This is a diffraction grating. The diffraction grating 125 is positioned opposite the receiver chip 20. The receiver chip 20 includes, for example, a photodiode optically coupled to the waveguide 120 by the diffraction grating 125. This photodiode can be positioned opposite the diffraction grating 125 to directly receive the light extracted by the diffraction grating 125.

[0051] In a display device comprising pixels optically coupled to a waveguide array, each pixel typically includes a controllable light extractor or group of controllable light extractors. During operation, it is important that several pixels illuminated simultaneously do not interfere with each other. The electromagnetic radiation re-emitted by the conversion portion 115 in photoluminescent material has low temporal coherence, which is particularly advantageous for preventing pixel interference.

[0052] Furthermore, the photoluminescent material can be selected to ensure that the re-emitted electromagnetic radiation has a narrow bandwidth. For example, when the 115 conversion portion incorporates quantum dots, a dispersion in the size of these dots can be used to achieve the desired bandwidth for the re-emitted electromagnetic radiation. A narrow bandwidth is particularly advantageous when used with a light-emitting diode (LED). LEDs typically have a broad spectral bandwidth, especially compared to other light sources such as lasers. A narrow bandwidth improves coupling within the waveguide. A narrow bandwidth can also offer other advantages depending on the application.For example, it can increase the distance and / or speed of communication in a data exchange application, particularly when the waveguide has chromatic dispersion. It can also be adapted for a surface diffraction grating to which the waveguide is optically coupled.

[0053] Advantageously, the conversion material is chosen to exhibit a good frequency response, particularly when the conversion portion 115 is coupled to a light source 210 modulated by a pulse-width modulation (PWM) dimmer, notably to allow a display device to display images at high frequencies. For example, the conversion material can be a photoluminescent material with a decay time sufficiently short to be compatible with the image display frequency, or a data transmission frequency in the case of a datacom application.

[0054] The interposer 30 includes a protective layer 130. The protective layer 130 rests on the waveguide 120 and the coupler 110 on one side opposite the lower confinement layer 105. This is a conformal layer following the shape of the waveguide 120 and the coupler 110. In this example, it extends over the entire length of the interposer 30.

[0055] The waveguide 120, the lower confinement layer 105, and the protective layer 130 are made of materials transparent at the second wavelength Δ2. The protective layer 130 is transparent at the first wavelength Δi. The lower confinement layer 105 and the protective layer 130 have refractive indices strictly lower than one refractive index of the waveguide 120 and one refractive index of the coupler 110, so as to confine the light within the coupler 110 and the waveguide 120. The lower confinement layer 105 has sufficient thickness to avoid optically coupling the waveguide 120 with the substrate 100. For example, its thickness is between 500 nm and 1 pm, measured parallel to the Z-axis.

[0056] The interposer 30 may further include metallic interconnects 173 for electrically connecting the transmitter chip 10 and / or the receiver chip 20. Two metallic interconnects 173 are schematically represented in Figure 1A, one electrically connecting the transmitter chip 10, the other the receiver chip 20. Here, the transmitter and receiver chips 10, 20 are hybridized on the interposer 30 via hybridization elements 175, such as indium beads. Each hybridization element 175 is for example in contact with a pad of a metallic interconnect 173 and a corresponding metallic pad of the transmitting or receiving chip 10, 20. The metallic interconnects 173 extend in and / or on the interposer 30. They can be electrically connected to vias passing through the interposer 30 or to wired electrical connections.

[0057] The emitting chip 10 comprises a substrate 200, a light-emitting diode 210, and a semiconductor stack 205 resting on the substrate 200. The semiconductor stack 205 can be made of any type of semiconductor material. Here, it is GaN-based. An active region and carrier injection layers of the light-emitting diode 210 occupy a region of the semiconductor stack 205. The light-emitting diode 210 is surrounded by one or more electrically insulating regions, extending completely through the semiconductor stack 205. The semiconductor stack 205 extends over the substrate 200 parallel to a principal plane of the substrate 200.

[0058] The transmitting chip 10 includes an interconnect stack 215. The semiconductor stack 205 is interposed between the substrate 200 and the interconnect stack 215 so as to be in contact with them. The interconnect stack 215 is located between the semiconductor stack 205 and the interposer 30. Similarly, the receiving chip 20 includes a substrate 300 and an interconnect stack 315 resting on the substrate 300, between the substrate 300 and the interposer 30. The receiving chip 20 may include electronic and / or photonic components integrated in and / or on the substrate 300. The interconnect stacks 215 and 315 comprise one or more metal lines and / or one or more vias. They participate in the power supply and / or control of electronic or optoelectronic components of the transmitting and receiving chips 10, 20, including the light-emitting diode 210. The hybridization elements 175 are each in contact with a pad or a metal line of one of the interconnection stacks 215, 315.

[0059] The light-emitting diode 210 also includes an optional microlens 220. The microlens 220 increases the overlap of the electromagnetic radiation emitted by the LED 210 with the conversion section 115. The overlap is equal to the ratio of the number of photons from the LED 210 received by the conversion section 115 to the total number of photons emitted by the LED 210. The microlens 220 is positioned opposite the conversion section 115, which is arranged between the semiconductor stack 205 and the coupler 110, so as to direct the electromagnetic radiation emitted by the LED 210 towards the conversion section 115. The arrangement of the microlens 220 relative to the LED 210 and / or the conversion section 115 may differ from that shown in the figures.

[0060] During operation, at least some of the electromagnetic radiation emitted by the light source 210 reaches the conversion portion 115, referred to in the description as incident electromagnetic radiation. The conversion portion 115 absorbs at least some of the incident electromagnetic radiation, referred to in the description as absorbed electromagnetic radiation. The absorbed electromagnetic radiation generates electromagnetic radiation at the second wavelength λ2, for example, through photoluminescence. The re-emitted electromagnetic radiation propagates in one or more directions different from the incident electromagnetic radiation. Since the refractive index of the conversion portion 115, and, where applicable, that of the passive portion, is strictly greater than the refractive indices of the protective layer 130 and the lower containment layer 105, the re-emitted electromagnetic radiation is confined within the coupler 110.Since the coupler 110 is optically coupled to the waveguide 120, at least part of the re-emitted electromagnetic radiation excites one or more propagation modes of the waveguide 120. Thus, the light source 210 is optically coupled to the waveguide 120.

[0061] Advantageously, the respective refractive indices at the second wavelength Δ2 of the conversion portion 115, the waveguide 120, and the passive portion are high relative to the refractive indices of the lower confinement layer 105 and the protective layer 130, so as to increase the confinement of the radiation emitted electromagnetic radiation. The lower containment layer 105 and the protective layer 130 may each be independently made of silicon oxide, silicon nitride, or alumina, or a combination thereof. The refractive indices of the conversion portion 115, the waveguide 120, and, where applicable, the passive portion may each be independently greater than or equal to 2, or even greater than or equal to 2.4.

[0062] Advantageously, the refractive index of the conversion portion 115 is as close as possible to the refractive index of the waveguide 120 and that of the passive portion when present, in order to increase the coupling efficiency between the coupler 110 and the waveguide 120. To this end, the refractive index of the conversion portion 115 is considered suitable when it is equal to the refractive index of the waveguide 120 and the passive portion when present, within 10%, and preferably less than or equal to the refractive index of the waveguide 120 and the passive portion when present. This is an advantageous condition whereby the coupler 110 and the waveguide 120 all have refractive indices equal to within 10%, but it is not essential to the invention.

[0063] When the conversion portion 115 contains quantum dots in a matrix, for example, a matrix made of an inorganic material or a polymer, it is possible to adapt the refractive index of the conversion portion 115 by adjusting the concentration of quantum dots in the matrix when the quantum dots and the matrix have different refractive indices. The concentration of quantum dots can be equal to the number of quantum dots per unit volume of the matrix, or to the volume occupied by the quantum dots relative to the total volume of the conversion portion 115. It can be deduced from the conversion efficiency of the conversion material.

[0064] Advantageously, the complex index of the 120 waveguide has an extinction coefficient less than or equal to 1 E-3.

[0065] The second wavelength Å2 may be within the visible spectrum. If applicable, the waveguide 120 may be made of titanium oxide (TiCh), tantalum oxide (Ta2Os), niobium oxide (Nb2Os), zinc oxide (ZnO), silicon nitride (SiN), aluminum nitride (AlN), silicon carbide (SiC), gallium nitride (GaN), or alumina (Al2O3). The conversion portion 115 may be made of halogenated perovskite, for example, CsPbBra. It may also be made of a material containing quantum dots, for example, cadmium selenide (CdSe) quantum dots. When the coupler 110 includes a passive portion, it may be made of a material selected from the list of possible materials for the waveguide 120. It is advantageously of the same material. The material is that of the 120 waveguide. A 115 conversion portion in CsPbBra has a refractive index suitable for a 120 waveguide and / or a passive portion in TiC>2 or Nb20s. A 115 conversion portion containing CdSe quantum dots in an array may have a refractive index suitable for a 120 waveguide and / or a passive portion in Ta2Os or SiN.

[0066] The second wavelength Å2 can be in the infrared range. Specifically, it can be one of the standard wavelengths used in data communications. If applicable, the 120 waveguide can be made of silicon. The 115 conversion portion can include lead sulfide (PbS) quantum dots. A 115 conversion portion incorporating PbS quantum dots has a refractive index suitable for a 120 waveguide and / or a silicon passive portion.

[0067] Figures 3A to 3F represent schematic top views of various couplers 110 adapted for this first embodiment as well as for all embodiments of the description.

[0068] In Figure 3A, the coupler 110 has a roughly parallelepiped shape. The conversion portion 115 occupies the entire volume of the coupler 110. The coupling surface is an interface between the conversion portion 115 and the waveguide 120.

[0069] In Figure 3B, the coupler 110 has the shape of a right prism with a horizontal base in the form of an isosceles trapezoid. The conversion portion 115 can occupy the entire volume of the coupler 110 or only a part of it, as shown here. In this example, the conversion portion 115 is a cylinder with an axis parallel to the Z-axis, passing completely through the coupler 110. The coupling surface is the smallest cross-sectional area of ​​the coupler 110.

[0070] Figure 3C shows a coupler 110 that is particularly advantageous when the waveguide 120 is single-mode. The coupler 110 has a flared passive portion. The width of the passive portion, measured parallel to the Y-axis, increases gradually, preferably linearly, away from the waveguide 120 along the X-axis. The conversion portion 115 has a plurality of parallelepiped sections, each optically coupled by an end-to-end coupling with a junction separate from the passive portion. The junctions merge away from the conversion portion 115 along the X-axis. The number of parallelepiped sections can be arbitrary. This example has 5 parallelepiped sections.

[0071] Figure 3D shows a coupler 110 that increases the coupling efficiency of the light source 210 with the waveguide 120. The coupler 110 has a passive portion at one end opposite the waveguide. 120. The conversion portion 115 is similar or identical to the conversion portion 115 of the embodiment in Figure 3A. The passive portion is an extension of the conversion portion 115 comprising a diffraction grating 117 configured to reflect a portion of the electromagnetic radiation re-emitted by the conversion portion 115 back to the waveguide 120. The diffraction grating 117 can be replaced by a metallization playing the same role.

[0072] Figure 3E shows a coupler 110 that increases the overlap of the electromagnetic radiation emitted by the light-emitting diode 210 with the conversion portion 115, thereby increasing the coupling efficiency of the light source 210 with the waveguide 120. The conversion portion 115 has a spiral shape in top view. Its lateral cross-section along the spiral is designed to allow the propagation of the emitted electromagnetic radiation at the second wavelength λ2.

[0073] Figure 3F shows a coupler 110 optically coupled to several waveguides 120. The coupler 110 is cylindrical with its axis parallel to the Z-axis. The conversion portion 115 occupies the entire volume of the coupler 110. The waveguides 120 are arranged in a star configuration around the coupler 110 so as to be in contact with it. The coupler 110 is coupled to each waveguide 120 by an end-to-end coupling.

[0074] A second embodiment of a coupling module 2 according to the invention will now be described with reference to Figure 1B. Only the differences with the first embodiment are explained. Figure 1B is a schematic view of the coupling module 2 integrated into a multi-chip module similar to that of Figure 1A.

[0075] In this embodiment, the conversion portion 115 occupies substantially the entire volume of the coupler 110. The waveguide 120 is made of the same material as the conversion portion 115. During operation, the difference between the second wavelength Δ2 and the first wavelength Δi is greater than or equal to a Stokes shift occurring in the conversion material. The larger the Stokes shift, the less of the re-emitted electromagnetic radiation is absorbed by the conversion material. When the conversion material is a semiconductor, it has a band gap strictly greater than the energy of photons with a wavelength equal to the second wavelength Δ2. The conversion material may include quantum wells, quantum dots, or be a halogenated perovskite.

[0076] A third embodiment of coupling module 3 will now be described in connection with Figure 2A. Figure 2A is a cross-sectional view of a multi-chip module incorporating coupling module 3. The multi-chip module includes a transmitter chip 10 and a receiver chip 20. Only the differences with the first embodiment are explained.

[0077] The coupling module 3 includes a mirror 117. Mirror 117 is reflective at wavelength λ2. It is configured to reflect a portion of the re-emitted electromagnetic radiation towards the waveguide 120. This mirror is a metallic portion that fits onto one end of the coupler 110 opposite the waveguide 120. This metallic portion has a vertical section positioned opposite the waveguide 120, reflecting the re-emitted electromagnetic radiation back towards it. This vertical section is substantially perpendicular to the principal plane of the waveguide 120. It is also substantially perpendicular to the optical axis of the waveguide 120 at the coupler 110. Mirror 117 is, for example, made of aluminum, copper, or an alloy of these materials. In this example, the conversion portion 115 rests on a lower part of mirror 117, although this is not essential. Alternatively, mirror 117 can be a Bragg mirror made of any type of material.

[0078] The waveguide 120 and the coupler 110 are optically coupled by direct optical coupling. They are inserted into a lateral confinement layer 107 of the interposer 30. The lateral confinement layer 107 extends over the lower confinement layer 105 so as to be in contact with it. The coupler 110, the waveguide 120, and the lateral confinement layer 107 have substantially equal heights, measured parallel to the Z-axis. The conversion portion 115 extends deep into the coupler 110 from its upper face until it reaches a horizontal part of the mirror 117 in contact with the lower confinement layer 105.

[0079] When the interposer 30 includes one or more photonic components, the mirror 117 may also have a geometric shape that blocks the re-emitted electromagnetic radiation and can be optically coupled to one of these photonic components, for example, via the lateral confinement layer 107, which can guide the light. For this purpose, regardless of the geometric shape of the coupler 110, the mirror 117 can surround the conversion portion 115 in the principal plane of the waveguide 120, over a continuous angular aperture of at least 270 degrees, measured in this plane. More precisely, the vertical portion of the mirror 117 can extend parallel to the principal plane so as to surround the conversion portion 115 along this angular aperture. The angular aperture can cover all angles in the principal plane that are not directly on the axis of the waveguide 120.

[0080] This particular geometric shape surrounding the conversion portion 115 is especially advantageous when the interposer 30 has a high local density of 110 couplers, and / or when a waveguide passes close to a 110 coupler. This situation is common for massively parallel inter-chip communication applications.

[0081] The lateral confinement layer 107 here comprises two sub-layers, including an upper insulating layer 109. The sub-layers of the lateral confinement layer 107 have refractive indices strictly lower than each of the refractive indices of the waveguide 120, the conversion portion 115 and the passive portion.

[0082] The protective layer 130 is substantially flat and parallel to the upper face 100.1 of the substrate 100. It extends over the waveguide 120, the conversion portion 115, the passive portion, the mirror 117, and the upper insulating layer 109, so as to be in contact with these elements. The waveguide 120 and the coupler 110 extend in depth, in the direction of the Z-axis, from the protective layer 130 to the lower confinement layer 105.

[0083] The interposer 30 further includes a photodiode 135 and an interconnect stack 145. The interconnect stack 145 comprises a network of metal interconnects 146 embedded in one or more electrically insulating layers. It has two opposite faces parallel to the upper face 100.1 of the substrate 100. It is in contact with the protective layer 130. The interconnect stack 145 includes a region transparent at the first wavelength Δi extending through the interconnect stack 145, between the light source 210 and the conversion portion 115. This is a region of the interconnect stack 145 devoid of metallic interconnects, comprising the electrically insulating layer(s).

[0084] The photodiode 135 is electrically connected to the interconnection network 146 by electrical contacts 136. The emitting chip 10 and the receiving chip 20 are electrically connected to the interconnection network 146 by the hybridization elements 175.

[0085] The photodiode 135 has an input face 135.1 in contact with an end of the waveguide 120 opposite the coupler 110, so as to couple the photodiode 135 to the waveguide 120 by butt coupling. A normal to the input face 135.1 may make a non-zero angle with an optical axis of the waveguide 120 located at the input face 135.1. In this example, the photodiode 135 is made of silicon. The input face 135.1 may be a crystalline face (111) of the silicon, making an angle of 54.74° with the (X, Y) plane.

[0086] Figure 2B shows a cross-sectional view of a coupling module 4 according to a fourth embodiment. The coupling module 4 is integrated into a module multi-chip. Only the differences with the third embodiment are described below.

[0087] The interposer 30 includes a planarizing inclusion 121 opposite the coupler 110 and the waveguide 120. The planarizing inclusion 121 extends between the coupler 110 and an end of the waveguide 120 opposite the coupler, so as to be in contact with these elements. It has a flat upper face that is coplanar with the respective upper faces of the upper insulating layer 109, the waveguide 120, the conversion portion 115, the passive portion, and the mirror 117. In top view, it is centered on the coupler 110 and the waveguide 120.

[0088] The planarizing inclusion 121 is made of a material transparent at the second wavelength Å2. It has a refractive index less than or equal to the refractive indices of the waveguide 120, the passive portion and the conversion portion 115. The planarizing inclusion 121 is for example made of silicon oxide.

[0089] Figure 2C shows a cross-sectional view of a coupling module 5 according to a fifth embodiment. The coupling module 5 is integrated into a multi-chip module. Only the differences with the fourth embodiment are described below.

[0090] In this embodiment, the planarizing inclusion 121 comprises two sublayers, each including an upper surface that is flush with the upper faces of the upper insulating layer 109, the waveguide 120, and the coupler 110. One of the two sublayers is in contact with the waveguide 120 and the coupler 110. It completely surrounds the other sublayer in a plane parallel to the (X, Y) plane. The conversion portion 115 extends deep into the coupler 110 from the planarizing inclusion 121 to the horizontal part of the mirror 117 in contact with the lower confinement layer 105. The two sublayers can be made independently of each other of silicon oxide or silicon nitride.

[0091] For all embodiments, it is preferable to increase the height of the conversion portion 115, parallel to the Z-axis, in order to increase the absorption of the incident electromagnetic radiation. It is then preferable to also increase the heights of the coupler 110 and the waveguide 120 accordingly, so as not to lose coupling efficiency between the coupler 110 and the waveguide 120. The gain in coupling efficiency provided by the optional mirror 117 increases with the respective heights of the coupler 110 and the waveguide 120. Thus, the waveguide 120 can have transverse dimensions such that it is vertically multimode at the second wavelength λ2, that is to say, all optical modes that can propagate in the Waveguides 120 have no more than one intensity lobe centered on an axis parallel to the upper face 100.1 of the substrate 100, and one of these optical modes has at least two intensity lobes centered on an axis orthogonal to the upper face 100.1. Like horizontal multimode waveguides, such a waveguide 120 has low propagation losses. It offers the additional advantage of being able to be curved in a plane parallel to the (X, Y) plane with a reduced radius of curvature, for example, less than or equal to 5 pm. It also allows for increased coupling efficiency. In the case of a vertical multimode waveguide 120, the mirror 117 is advantageously metallic.

[0092] Figure 4 shows a representation of a vertical multimode 120 waveguide capable of propagating two optical modes, including the fundamental optical mode. The two modes are schematically represented at a cross-section of the 120 waveguide. The second wavelength λ₂ is 530 nm. The cross-section has a width of 250 nm and a height of 1 pm.

[0093] A first method for realizing a coupling module 1 according to the first embodiment is now described in connection with figures 5A to 5E.

[0094] In Figure 5A, a substrate 100 is provided, and the lower confinement layer 105 and an upper layer 410 are deposited. The upper layer 410 is made of the same material as the waveguide 120 and the passive portion. The lower confinement layer 105 is sandwiched between and in contact with the substrate 100 and the upper layer 410. The lower confinement layer 105 is made of silicon dioxide, and the upper layer 410 is made of TiCh. The upper layer 410 has, for example, a thickness between 300 nm and 1 pm, measured parallel to the Z-axis.

[0095] Alternatively, the top layer 410 can be transferred onto the lower confinement layer 105 using a thin-film transfer technique employing direct bonding. The top layer 410 can be made of a crystalline semiconductor material such as silicon or gallium nitride. The step in Figure 5A can then be a step for supplying a silicon-on-insulator (SOI) or GaN-on-insulator (GAI) wafer. Active components, such as a photodiode, modulator, or light source, can thus be fabricated in and / or on the top layer 410 during process steps concurrent with and / or subsequent to those of Figures 5A to 5E. The active components can be optically coupled to the waveguide 120.In the case where the interposer 30 is a passive optical interposer, the active components can be powered and / or controlled from one or more hybrid electronic chips on the interposer 30 and / or from a PCB electrically connected to the interposer 30.

[0096] In Figure 5B, the upper layer 410 is locally etched along its entire height to define the waveguide 120 and the coupler 110. Preferably, the lower confinement layer 105 is not etched, as shown here. The etching can be an anisotropic dry etching, preferably selective with respect to the lower confinement layer 105. At the end of the step in Figure 5B, a residual portion 411 of the upper layer 410 rests on the lower confinement layer 105.

[0097] In Figure 5C, an opening extending deep into the residual part 411 is made at an end 411.1 of the residual part 411 defining the geometry of the coupler 110. Preferably, the opening goes all the way through the residual part 411. Here, the opening does not extend into the lower confinement layer 105. It is possible to use a photolithography substep, followed by an anisotropic dry etching substep.

[0098] The aperture is then filled with the conversion material, preferably completely. When the conversion material is halogenated perovskite, the aperture can be filled by physical vapor deposition (PVD) or pulsed laser deposition (PLD). When the conversion material contains quantum dots, the aperture can be filled by inkjet printing or spin coating. The filling step may include one or more photolithography, etching, or planarization substeps.

[0099] At the end of figure 5C, we obtain the conversion portion 115, vertically crossing the residual part 411 from one side to the other, without exceeding it. [ooioo] In Figure 5D, the protective layer 130 is conformally deposited. The protective layer 130 is here made of alumina, silicon nitride, or silicon oxide. It provides a light-confining function within the coupler 110 and the waveguide 120, and a protective function for the conversion material, for example, against moisture and oxidation. The residual portion 411 is encapsulated in the protective layer 130. This results in the coupler 110 corresponding to the end 411.1 of the residual portion 411, and the waveguide 120 corresponding to the complementary portion at the end 411.1 of the residual portion 411. [ooioi] The diffraction grating 125 is then produced by etching. The diffraction grating 125 extends deep into the protective layer 130 towards the substrate 100, to a depth that allows for light extraction. Depending on the extraction efficiency As desired, the diffraction grating 125 can pass through all or part of the protective layer 130 or the waveguide 120.

[0102] The process can be completed by a step of forming the metallic interconnections 173, followed by a step of hybridizing the transmitting and receiving chips 10, 20 to obtain the multi-chip module of Figure 1A integrating a coupling module 1 according to the first embodiment.

[0103] A second method for realizing a coupling module 2 according to the second embodiment is now described in connection with figures 6A to 6D.

[0104] The step in Figure 6A is identical to that in Figure 5A. The top layer 410 is made of the same material as the conversion material. When the conversion material is halogenated perovskite, the top layer 410 can be deposited by physical vapor deposition (PVD) or pulsed laser deposition (PLD). When the conversion material contains quantum dots, the top layer 410 can be deposited by centrifugal coating.

[0105] The step in Figure 6B is identical to that in Figure 5B. Since the material of the residual portion 411 of the upper layer 410 is identical to the conversion material, the conversion portion 115 is obtained directly after the etching in Figure 6B. The end 411.1 of the residual portion 411 constitutes the conversion portion 115. It also defines the geometry of the coupler 110.

[0106] The step in Figure 6C is identical to that in Figure 5D. We obtain the coupler 110 corresponding to the end 411.1 of the residual part 411, and the waveguide 120 corresponding to the part complementary to the end 411.1 of the residual part 411.

[0107] The step in Figure 6D is identical to that in Figure 5E.

[0108] Similar to the first process, the second process can be completed by a step of forming the metallic interconnections 173, followed by a step of hybridizing the transmitting and receiving chips 10, 20 to obtain the multi-chip module of Figure 1 B, integrating a coupling module 2 according to the second embodiment.

[0109] In connection with Figures 7A to 7E, a sequence of steps common to the manufacturing processes of coupling modules 3, 4, and 5 is now described according to the third, fourth, and fifth embodiments. In each figure, a view along section AA, identified in a top view, is shown in the upper part of the upper portion. represented on its lower part. For the sake of readability, hidden elements are not shown. Some elements may be omitted, such as background elements. The diagrams are not to scale, and a proportionality factor may exist between equal distances measured along two separate axes. [oono] The sequence of steps in Figures 7A to 7E leads to the fabrication of a photodiode extending from a trench designed to accommodate a waveguide and a coupler of a coupling module according to the invention. The trench and the photodiode can be elements of an interposer. Fabrication of the photodiode before fabrication of the waveguide and coupler allows for easy adaptation of steps from an existing photodiode manufacturing process. In particular, the conversion material and the constituent materials of the waveguide and coupler are not affected by the thermal treatments used to activate the photodiode implants. [oom] In Figure 7A, a plate is provided comprising the substrate 100, the lower confinement layer 105, and an upper layer 410. The lower confinement layer 105 is in contact with the upper layer 410 and the substrate 100. The upper layer 410 extends substantially over one face of the lower confinement layer 105 opposite the substrate 100. The upper layer 410 is made of a semiconductor material. It can be deposited or transferred onto the lower confinement layer 105. The plate is a silicon-on-insulator (SOI) plate. The substrate 100 is made of silicon in this example. The lower confinement layer 105 is made of silicon oxide. The upper layer 410 is made of crystalline silicon, preferably with a (100) orientation. For example, it has a height between 500 nm and 1 pm, measured parallel to the Z-axis.

[0112] Alternatively, the substrate 100 can be made of a dielectric material, for example glass. If so, the lower confinement layer 105 can be omitted and the upper layer 410 can be made of amorphous silicon.

[0113] Alternatively, substrate 100 can be silicon carbide or gallium nitride. If so, the top layer 410 can be amorphous silicon.

[0114] In figure 7B, the upper layer 410 is locally etched from one side to the other to create a semiconductor pad 415. The semiconductor pad 415 here has a parallelepiped shape with edges oriented along one of the axes of the direct orthogonal coordinate system (X, Y, Z).

[0115] A leveling layer is then deposited on the lower confinement layer 105 and the semiconductor pad 415. The leveling layer has a thickness greater than the height of the semiconductor pad 415 measured parallel to the Z-axis. The leveling layer is then planarized, for example by chemical mechanical polishing (CMP) to obtain a first planarizing layer 421. The first planarizing layer 421 has a top surface flush with a top surface of the semiconductor pad 415. The leveling layer is made of a material transparent at the second wavelength Å2. Here it is silicon oxide.

[0116] The semiconductor pad 415 is then implanted to create a first doped region 416 and a second doped region 417 with opposing doping types. The first and second doped regions 416 and 417 are located on either side of the AA axis. Preferably, they define an intrinsic region extending from the first doped region 416 to the second doped region 417, as viewed from above. The first and second doped regions 416 and 417 can each include overdoped regions with the same type of conductivity, for the purpose of creating electrical contact areas.

[0117] In Figure 7C, a first mask 431 is formed on the first planar layer 421 and on the semiconductor pad 415. The first mask 431 extends in a plane parallel to the upper face 100.1. It has through-holes so as to expose the first and second doped regions 416, 417. There are two through-holes per doped region 416, 417. In this example, they are positioned with respect to overdoped regions flush with the upper surface of the semiconductor pad 415. The first mask 431 is made of an electrically insulating material. Here, it is silicon nitride.

[0118] A reactive metal is then deposited with the silicon in the through-holes. The metal can be, for example, nickel, cobalt, titanium, or tungsten. A heat treatment is then carried out, for example in a furnace or by laser. The heat treatment can be rapid thermal annealing (RTA). Selective etching is then performed to remove the excess metal that has not reacted with the silicon. At the end of the step shown in Figure 7C, silicidated areas 433 are obtained at the first and second doped regions 416, 417, one opposite each through-hole.

[0119] In figure 7D, a second mask is formed on the first mask 431 comprising through openings with respect to the silicified area 433 so as to expose the silicified area 433. The second mask is made of an electrically insulating material. The first 431 mask and the second mask can be made of identical materials, as is the case here.

[0120] Electrical contacts 136 are then formed through the first mask 431 and the second mask by a Damascus process. The electrical contacts 136 can, for example, be made of tungsten (W) or copper (Cu). They may optionally be provided with one or more diffusion barriers and / or one or more adhesion sublayers.

[0121] A trench 108 is then etched through the first mask 431, the second mask, and the first planarizing layer 421. This can consist of one or more etches, preferably selective with respect to the silicon. Preferably, the trench 108 does not extend into the lower confinement layer 105. To this end, the lower confinement layer 105 can be provided with an etching stop sublayer at the interface between the lower confinement layer 105 and the first planarizing layer 421.

[0122] Once trench 108 is completed, the remaining parts of the first mask 431 and the second mask together form the upper insulating layer 109. The upper insulating layer 109 and the remaining part of the first planarizing layer 421 together form the lateral containment layer 107.

[0123] Trench 108 has a proximal end 108.1 designed to define the sides and bottom of the coupler 110. The proximal end 108.1 is shown here as flared in top view. It can have any shape suitable for the operation of the coupling module, such as a shape complementary to that of one of the couplers 110 illustrated in Figures 3A to 3F. The part complementary to the proximal end 108.1 of trench 108 is designed to define the sides and bottom of the waveguide 120. It extends parallel to the section plane AA and the X-axis to a face of the semiconductor pad 415 intended to be the input face 135.1 of the photodiode 135.

[0124] Optionally, the etching of trench 108 may include an additional anisotropic wet etching of the semiconductor pad 415 to reveal a particular crystal face of the semiconductor pad 415 opposite trench 108. For this purpose, the plate may be a (100) oriented SOI plate, with the X-axis oriented in the direction <110> The additional wet etching can be etching in a solution of potassium hydroxide (KOH) or tetramethylammonium hydroxide (TMAH). In this case, the entrance face 135.1 forms an acute angle with the (X, Y) plane, for example, 54.74°. The entrance face 135.1 can be set back from an edge of the layer upper insulating layer 109, that is to say that part of the upper insulating layer 109 overhangs the trench 108, above the entrance face 135.1.

[0125] In Figure 7E, a metal stud 408 is formed in the trench 108, at the proximal end 108.1. The metal stud 408 covers a lateral face of the trench 108 opposite the complementary portion at the proximal end 108.1 of the trench 108. The lateral face is substantially parallel to the (Y, Z) plane. The metal stud 408 can also cover the upper face of the upper insulating layer 109 and / or a portion of the bottom of the trench 108 in contact with the lower containment layer 105 and / or other faces of the proximal end 108.1.

[0126] To form the metal stud 408, a metallic layer can be deposited conformably onto the upper insulating layer 109 and into the trench 108, and the previously masked metallic layer can be engraved through to the imprint of the metal stud 408. The metallic layer has, for example, a thickness between 80 nm and 120 nm. It can be made of any type of metal reflective at the second wavelength Å2. Here, it is aluminum.

[0127] Figures 8A to 8D, 9A to 9D, and 10A to 10E depict process steps that complement the sequence of steps shown in Figures 7A to 7E. Similar to Figures 7A to 7E, the upper portion of each of these figures shows a view along section AA, identified on a top view shown in its lower portion. Hidden elements are not shown. Some elements may be omitted, such as background elements. The diagrams are not to scale, and a proportionality factor may exist between equal distances measured along two distinct axes.

[0128] Now, a third manufacturing process will be described in relation to Figures 8A to 8D. The third manufacturing process leads to the realization of a coupling module 3 according to the third embodiment, integrated into a multi-chip module.

[0129] We begin by carrying out the sequence of steps in figures 7A to 7E.

[0130] In Figure 8A, a second planarizing layer 422 is deposited on the upper insulating layer 109, on the metal stud 408 and in the trench 108. The second planarizing layer 422 has a thickness strictly greater than the depth of the trench 108 measured parallel to the Z axis and its greatest width in a plane parallel to the (X, Y) plane. Tl

[0131] The second planarizing layer 422 may include a lower sub-layer deposited prior to the deposition of a complementary sub-layer of the second planarizing layer 422. If applicable, the lower sub-layer is conformally deposited on the upper insulating layer 109, the metal stud 408, the bottom of the trench 108, and the inlet face 135.1. The lower sub-layer has a thickness strictly less than the depth of the trench 108, the smallest width of the trench 108, and the thickness of the complementary layer. Its thickness is, for example, less than one-third of the depth of the trench 108, or even less than one-fifth, or one-tenth of the depth of the trench 108. The lower sub-layer is made of a dielectric material transparent at the second wavelength λ2. It may, for example, be silicon dioxide.The lower sublayer may be intended to be intercalated between the conversion portion 115 and the mirror 117 so as to avoid excitation of a plasmon mode of the mirror 117 when the latter is metallic.

[0132] The complementary sublayer or the second planarizing layer 422, when it does not include a sublayer, is made of the same material as the waveguide 120 and the passive portion of the coupler 110. In this example, the second planarizing layer 422 does not have a sublayer. It is made of titanium oxide (TiC>2).

[0133] In Figure 8B, the second planarizing layer 422 and the metal stud 408 are polished by one or more mechano-chemical polishing operations until the upper insulating layer 109 is reached. At the end of this step, the upper insulating layer 109 is exposed over its entire upper surface. A remaining portion of the metal stud 408 forms the mirror 117. A residual portion 426 of the second planarizing layer 422 is housed in the trench 108, so as to fill its entire volume. One upper surface of the residual portion 426 is flush with the upper surface of the upper insulating layer 109. The residual portion 426 has an end 426.1 housed within the proximal end 108.1, defining the geometry of the coupler 110.

[0134] In figure 8C, an opening extending deep into the end 426.1 is engraved. Preferably, it passes completely through the residual part 426 at the level of the end 426.1. The opening here leads to the mirror 117. In this example, it has a cylindrical shape with an axis parallel to the Z axis.

[0135] The opening is then filled with the conversion material; preferably, the opening is completely filled. When the conversion material is halogenated perovskite, the opening can be filled by physical vapor deposition (PVD) or by laser ablation. Pulsed laser deposition (PLD). When the conversion material contains quantum dots, the aperture can be filled using an inkjet printing technique or by spin coating. The filling process may include substeps of photolithography, etching, and / or planarization. The conversion material may overflow the aperture, but this is not desirable. At the end of the step shown in Figure 8C, the conversion portion 115 is obtained.

[0136] In Figure 8D, the protective layer 130 is deposited onto the upper insulating layer 109, the mirror 117, the conversion portion 115, and the residual portion 426. The protective layer 130 is made of an electrically insulating material, transparent to the first and second wavelengths Δi, Δ2. The protective layer 130 can, for example, be made of alumina or silicon nitride. It can have a thickness between 10 nm and 50 nm. In this example, the protective layer 130 has two opposite faces substantially parallel to the upper face 100.1 of the substrate 100. Openings are then etched through the protective layer 130 with respect to the electrical contacts 136. As shown in Figure 8D, the coupler 110 and the waveguide 120 are obtained.

[0137] A fourth manufacturing process will now be described in connection with Figures 9A to 9D. The fourth manufacturing process leads to the production of a coupling module 4 according to the fourth embodiment, integrated into a multi-chip module.

[0138] We begin by carrying out the sequence of steps in figures 7A to 7E.

[0139] In Figure 9A, a filler layer 425 is conformally deposited on the upper insulating layer 109, the metal pad 408, and in the trench 108, on the lower containment layer 105 and on the inlet face 135.1. The filler layer 425 has a thickness strictly less than the depth of the trench 108 and a width of the trench 108 measured parallel to the Y-axis. It is made of the same material as the waveguide 120 and the passive portion, here TiC>2. The filler layer 425 has, for example, a thickness between 300 nm and 1 pm.

[0140] A third planarizing layer 423 is then deposited on top of the filling layer 425. The combined thickness of the filling layer 425 and the third planarizing layer 423 is strictly greater than the depth of the trench 108 and its greatest width in a plane parallel to the (X, Y) plane. The third planarizing layer 423 is made of any material transparent at the second wavelength λ2 that can be etched or polished. In this case, it is silicon oxide.

[0141] In Figure 9B, the third planarizing layer 423, the filling layer 425, and the metal stud 408 are polished by one or more mechano-chemical polishing operations until the upper insulating layer 109 is reached. At the end of this step, the upper insulating layer 109 is exposed over its entire upper surface. A remaining portion of the metal stud 408 forms the mirror 117. A residual portion 426 of the filling layer 425 is placed in the trench 108, so as to partially fill it. A remaining part of the third planarizing layer 423 constitutes a planarizing inclusion 121 lodged inside the trench 108, in contact with the residual part 426 on one side of it opposite the lower containment layer 105. Preferably, the residual part 426 and the planarizing inclusion 121 together completely fill the trench 108, as shown in Figure 9B.The upper insulating layer 109, the mirror 117, the residual part 426 and the planarizing inclusion 121 together define a surface extending over all these elements, substantially planar and parallel to the upper face 100.1. Here, the planarizing inclusion 121 extends parallel to the X-axis between two opposite ends of the trench 108. The residual part 426 has an end 426.1 housed in the proximal end 108.1, defining the geometry of the coupler 110.

[0142] In Figure 9C, an opening extending deep into the end 426.1 is engraved. Preferably, it passes completely through the residual part 426 at the level of the end 426.1. The opening here leads to the mirror 117. The opening may optionally pass completely through the planarizing inclusion 121. In this example, the opening has a cylindrical shape with an axis parallel to the Z axis.

[0143] The aperture is then filled with the conversion material, preferably completely. When the conversion material is halogenated perovskite, the aperture can be filled by physical vapor deposition (PVD) or pulsed laser deposition (PLD). When the conversion material contains quantum dots, the aperture can be filled by inkjet printing or spin coating. The filling process may include photolithography, etching, and / or planarization substeps. The conversion material may overflow the aperture, but this is not desirable. At the end of the step shown in Figure 9C, the conversion portion 115 is obtained.

[0144] Figure 9D reproduces the step of Figure 8D, with the difference that the protective layer 130 rests here, in addition, on the planarizing inclusion 121. At the end of Figure 9D, we obtain the coupler 110 and the waveguide 120.

[0145] Now, a fifth manufacturing process will be described in relation to Figures 10A to 10E. The fifth manufacturing process leads to the realization of a coupling module 5 according to the fifth embodiment, integrated into a multi-chip module.

[0146] We begin by carrying out the sequence of steps in figures 7A to 7E.

[0147] Figure 10A reproduces the step of Figure 9A except for the deposition of the third planarizing layer 423. At the end of the step in Figure 10A, the filling layer 425 has a depression 428 on one side of the layer opposite the lower containment layer 105. The depression 428 extends parallel to the X-axis between two opposite ends of the trench 108. It extends into the proximal end 108.1. In depth, it extends into the interior of the trench 108. The filling layer 425 surrounds the depression 428 in all directions in a plane parallel to the (X,Y) plane.

[0148] In Figure 10B, a region of the filler layer 425 located opposite the upper insulating layer 109 and the metal stud 408 is engraved throughout. A residual portion 427 of the filler layer 425 fills the trench 108 and completely covers its bottom. This residual portion 427 extends beyond the trench 108, forming an upper part that completely surrounds it. The residual portion 427 is in contact with the inlet face 135.1, the upper insulating layer 109, and the metal stud 408. It has a rectangular shape when viewed from above. It does not cover the electrical contacts 136.

[0149] In Figure 10C, an opening extending deep into the residual part 427 is engraved at a portion of the residual part 427 located inside the proximal end 108.1. Preferably, it extends from a flat surface parallel to the (X, Y) plane of the residual part 427, located inside the trench 108. Advantageously, it passes completely through the residual part 427. The opening here leads to the mirror 117. In this example, it has a cylindrical shape with its axis parallel to the Z axis.

[0150] The opening is then filled with the conversion material; preferably, the opening is completely filled. When the conversion material is halogenated perovskite, it is possible to fill the opening by physical vapor deposition (or PVD (Physical Vapor Deposition) or pulsed laser ablation (PLD) deposition. When the conversion material contains quantum dots, the aperture can be filled using inkjet printing or spin coating. The filling process may include photolithography, etching, and / or planarization substeps. While the conversion material may overflow the aperture, this is not desirable. The conversion portion 115 is obtained after the step shown in Figure 10C.

[0151] Figure 10D shows the step shown in Figure 8D. The deposition of the protective layer 130 is a conformal deposition. It covers the upper insulating layer 109, the metal stud 408, the residual portion 427, and the conversion portion 115.

[0152] Figure 10E shows an optional step facilitating the subsequent construction of the interconnect stack 145. A leveling layer is deposited on the protective layer 130. The leveling layer may be made of silicon oxide. It has sufficient thickness to fill the depression 428. The leveling layer, the protective layer 130, and the metal pad 408 are then polished until the upper insulating layer 109 is reached but not exceeded. This results in the planarizing inclusion 121, consisting of a residual portion of the protective layer 130 and a residual portion of the leveling layer.

[0153] The third, fourth and fifth manufacturing processes can be continued with a view to realizing the multi-chip module. If necessary, they continue with a step of forming the stack of interconnections 145 on the protective layer 130 electrically connecting the interconnection network 146 to the electrical contacts 136, followed by a step of hybridizing the transmitting and receiving chips 10, 20 to obtain the multi-chip module integrating the coupling module 3, 4, 5.

[0154] Specific embodiments have just been described. Different variations and modifications will be apparent to those skilled in the art.

Claims

DEMANDS 1. Coupling module (3, 4, 5), comprising: o a light source (210) capable of emitting electromagnetic radiation at a first wavelength Åi, o a coupler (110) opposite the light source (210) comprising a conversion portion (115) capable of absorbing electromagnetic radiation at the first wavelength Åi to emit multidirectional electromagnetic radiation at a second wavelength Å2 strictly greater than the first wavelength Åi, o a waveguide (120) optically coupled to the coupler (110), o a mirror (117) formed of a metallic portion having a vertical part arranged opposite the waveguide (120) so as to reflect a part of the multidirectional electromagnetic radiation in the direction of the waveguide (120).

2. Coupling module (3, 4, 5) according to claim 1, wherein the waveguide (120) extends in a principal plane and wherein the mirror (117) surrounds the conversion portion (115) in the principal plane, over an angular opening of at least 270 degrees.

3. Coupling module (3, 4, 5) according to claim 1 or 2, wherein the light source (210) is a gallium nitride light-emitting diode and the first wavelength Δi is within a spectral range from ultraviolet to blue.

4. Coupling module (3, 4, 5) according to claim 3, wherein the second wavelength A2 is included in the visible spectrum or in the infrared spectrum.

5. Coupling module (3, 4, 5) according to any one of claims 1 to 4, wherein the waveguide (120) and the coupler (110) are optically coupled by butt coupling, and wherein the coupler (110) and the waveguide (120) all have their refractive indices equal to within 10%.

6. Coupling module (2) according to claim 5, in which the conversion portion (115) occupies the entire volume of the coupler (110) and in which the waveguide (120) and the conversion portion (115) are made of the same material.

7. Coupling module (3, 4, 5) according to any one of claims 1 to 5, wherein the waveguide (120) has an extinction coefficient less than or equal to 1 E-3 at the second wavelength A2.

8. Coupling module (3, 4, 5) according to any one of the preceding claims, further comprising a substrate (100), in which the waveguide (120) extends over the substrate (100) parallel to an upper face (100.1) of the substrate (100), and in which the waveguide (120) is a vertical multimode waveguide at the second wavelength Å2.

9. Multi-chip module comprising a coupling module (3, 4, 5) according to any one of the preceding claims, an interposer (30) and an emitting chip (10), wherein the coupler (110) and the waveguide (120) of the coupling module (3, 4, 5) are arranged in the interposer (30), and wherein the light source (210) of the coupling module (3, 4, 5) is arranged in the emitting electronic chip.

10. Multi-chip module according to claim 9, further comprising a photodiode (135) optically coupled end-to-end with the waveguide (120) by an input face (135.1) of the photodiode (135).

11. Multi-chip module according to claim 10, wherein the photodiode (135) is made of silicon, and the input face (135.1) is oriented along a crystal plane (111).

12. Method of manufacturing a coupling module (3, 4, 5) according to any one of claims 1 to 8 comprising the following steps: o formation of the waveguide (120) and the coupler (110) on a substrate (100), o formation of the conversion portion (115) in the coupler (110).

13. Manufacturing method according to claim 11, wherein the step of forming the waveguide (120) and the coupler (110) comprises the following substeps: o engraving a trench (108) having a proximal end (108.1) intended to define sides and a bottom of the coupler (110), the part complementary to the proximal end (108.1) of the trench (108) being intended to define sides and a bottom of the waveguide (120), o deposition of a layer (422, 425) filling the trench (108), in a material of the coupler (110) and the waveguide (120), so as to form a part of the coupler (110) and a part of the waveguide (120).

14. Manufacturing method according to claim 13, further comprising a step of forming a photodiode (135) prior to the engraving of the trench (108).

15. Manufacturing method according to claim 14, wherein the photodiode (135) is made of silicon, and the trench etching (108) comprises a substep of anisotropic wet etching of an input face (135.1) of the photodiode (135).

Images