ARRANGEMENT WITH A FIRST AND A SECOND PHOTON CHIP MOUNTED ON TOP OF EACH OTHER

Description

[0001] The invention relates to an assembly comprising a first and a second photonic chip attached to each other by bonding at a bonding interface. The invention also relates to a photonic chip for producing this assembly and a method for manufacturing this assembly.

[0002] In such assemblies, it is necessary to optically connect optical components contained in the first photonic chip to optical components contained in the second photonic chip. For this, a device for optical coupling of the two photonic chips is implemented partly in the first photonic chip and partly in the second photonic chip.

[0003] To bond the first photonic chip to the second photonic chip, a placement tool must be used that automatically positions the first photonic chip at the desired location on the second photonic chip. However, the alignment errors typically observed with such placement tools are, for example, between +3 µm and -3 µm, particularly in a direction perpendicular to the direction of propagation of the optical signal. The optical coupling device must therefore be robust against such alignment errors.

[0004] Currently, known optical coupling devices, sufficiently robust against such alignment errors, comprise many optical components such as lenses and mirrors.

[0005] Also known from the prior art are US2002 / 076150A1, WO2019 / 152990A1, US10288812B1, US2016 / 349454A1, US2019 / 265415A1, US2003081902A1, EP3339923A1, US2003223675A1, WO2014176175A1 and US2016 / 327748A1. These documents describe optical coupling devices between optical fibers and waveguides of a photonic chip. These optical coupling devices are not sufficiently robust against alignment errors.

[0006] The invention aims to propose a set of two photonic chips glued to each other and comprising an optical coupling device that is simpler to produce while remaining tolerant of alignment errors at least in a given direction.

[0007] Its subject is therefore an assembly conforming to claim 1.

[0008] The invention also relates to a second photonic chip for producing the above assembly in accordance with claim 9.

[0009] Finally, the invention also relates to a method of manufacturing the above assembly in accordance with claim 10.

[0010] The invention will be better understood on reading the description which follows, given solely as a non-limiting example and made with reference to the drawings in which: there figure 1 is a schematic illustration in vertical section of an assembly comprising a first and a second photonic chip attached by gluing one onto the other; the figure 2 is a top view illustration of a coupling area of ​​a coupling device of the first and second photonic chips of the assembly figure 1 ; there figure 3 is a flowchart of a design and manufacturing process for the entire figure 1 ; there figure 4 is a schematic illustration, in top view, of the coupling area of ​​the figure 2 in a particular case; the Figure 5is a graph illustrating the energy transfer between two waveguides in the coupling zone of the figure 4 ; THE figures 6, 7 and 8 are energy transmission rate profiles between the first and second photonic chips of the entire figure 1 in different situations; and the figure 9 is a schematic illustration of an alternative embodiment of a coupling area that may be used in place of the coupling area of ​​the figure 2 .

[0011] In the remainder of this description, features and functions well known to those skilled in the art are not described in detail.

[0012] In this description, detailed examples of embodiments are first described in a chapter I with reference to the figures. Then, in a chapter II, variants of these embodiments are presented. Finally, the advantages of the different embodiments are presented in a chapter III. Chapter I: Examples of embodiment

[0013] There figure 1 represents an assembly 2 comprising a photonic chip 4 bonded to a photonic chip 6 at a bonding interface 8. The interface 8 extends in a horizontal plane parallel to directions X and Y of an orthogonal reference frame XYZ. In this reference frame, the Z direction designates the vertical direction. Subsequently, terms such as “above”, “below”, “upper”, “lower” and the like are defined with respect to the Z direction.

[0014] The bonding between chips 4 and 6 can be direct bonding, i.e. without adding material, or bonding using a layer of glue placed between the two chips 4 and 6.

[0015] Chips 4 and 6 each comprise at least one photonic component that generates, measures, transforms or transports an optical signal. Typically, the wavelength λ of the optical signal is between 1260 nm and 1625 nm and, preferably, between 1260 nm and 1360 nm (O band) or between 1530 nm and 1565 nm (C band). In this example, the wavelength λ is equal to 1310 nm.

[0016] To simplify the figure 1 , a single photonic component 12 was represented in chip 4 and a single photonic component 14 was represented in chip 6.

[0017] The assembly comprises an optical coupling device 20 which makes it possible to optically connect the chips 4 and 6. Here, the device 20 optically connects the component 12 to the component 14 through the interface 8. For this purpose, the device 20 optically connects an input waveguide 22, located inside the chip 4, to an output waveguide 24 located inside the chip 6. The waveguides 22 and 24 are optically connected, respectively, to the optical components 12 and 14.

[0018] The device 20 transfers most of the energy of the optical signal propagating in the waveguide 22 to the waveguide 24. Here, by "most of the energy" is meant at least 50% or 60% of the energy, at wavelength λ, of the optical signal propagating in the guide 22. Subsequently, this percentage is also called the transmission rate. A portion of the waveguides 22 and 24 is shown in dotted lines to indicate that only a portion of these waveguides has been shown on the figure 1 .

[0019] In this embodiment, the waveguides are shaped to guide and propagate the optical signal at the wavelength λ. For this purpose, each waveguide comprises a core made of a first material surrounded by a cladding made of a second material. The refractive index of the core is greater than the refractive index of the cladding. Typically, the refractive index of the core is at least 1.2 times or 1.3 times greater than the refractive index of the cladding. In this exemplary embodiment, the core of the waveguides is made of silicon and their cladding is made of silicon oxide. The cross-section of these waveguides is rectangular or “T” shaped. In the figures, only the core of the waveguides is shown.

[0020] A part of the device 20 is produced inside the chip 4. This part comprises an input 26 which receives the optical signal which propagates in the guide 22. The device 20 also comprises another part produced inside the chip 6. This other part comprises an output 28 which emits the optical signal in the waveguide 24.

[0021] Between input 26 and output 28, the device 20 successively comprises: an evanescent coupling zone 30; a phase adaptation zone 32; and a power adder 34.

[0022] Evanescent couplings are well known. For example, they are defined in the following book by RG Hunsperger: "Integrated Optics: Theory and Technology", Springer, 2002, page 154-155.

[0023] Zone 30 extends from a vertical plane P1 to a vertical plane P2. Planes P1 and P2 are parallel and orthogonal to the Y direction. Plane P1 is located after entrance 26 and plane P2 is located before exit 28.

[0024] The zone 30 comprises a waveguide 40 produced in the chip 4 and a network 42 of several waveguides produced, below, in the chip 6.

[0025] In this first embodiment, the guide 40 is a rectilinear waveguide which extends mainly parallel to the Y direction. In other words, in a horizontal plane, the angle θ between the Y direction and the axis of the guide 40 is between -1° and +1°.

[0026] The guide 40 begins at the level of the inlet 26 and ends well after the plane P2 at the level of an end 44. For this, the length of the guide 40 is greater than the distance d1 which separates the planes P1 and P2. Here, the length of the guide 40 is at least greater than the distance d1 + 3 µm and, preferably, greater than the distance d1 + 6 µm or + 10 µm.

[0027] Thanks to this, even if the chip 4 is placed on the chip 6 using a placement tool whose precision, in the Y direction, is plus or minus 3 µm, it is certain that the guide 40 completely crosses the space between the planes P1 and P2 and therefore completely crosses the evanescent coupling zone 30.

[0028] For example, the width of the cross-section of the guide 40 is equal to 400 nm and its thickness is equal to 300 nm. The guide 40 also has a lower face 46 which extends in a horizontal plane. The refractive index of the guide 40 is noted n r40 and its effective propagation index is noted n eff40 .

[0029] The effective index of a waveguide is proportional to the "mode phase constant". It is defined by the following relation ng = n eff - λdn eff / dλ, where ng is the group index and λ is the wavelength of the optical signal guided by this guide. The effective index is, for example, determined by numerical simulation or experimentally.

[0030] The array 42 begins at the plane P1 and ends at the plane P2. The plane P1 comprises at least one of the distal ends of the waveguides of the array 42. The plane P2 comprises at least one of the proximal ends of the waveguides of the array 42. This array 42 is described in more detail with reference to the figure 2 .

[0031] Each of the proximal ends of the waveguides of the network 42 is connected to a respective input of the adder 34 via respective waveguides 50. The waveguides 50 each pass through the zone 32. The zone 32 has the function of modifying the phase of the optical signals which propagate in each of the guides 50 so that the phases of these optical signals received at the inputs of the adder 34 are equal. For this purpose, the zone 32 comprises an adjustable phase shifter 54 associated with each of the waveguides 50 which pass through the zone 32. Here, each phase shifter 54 is a device known as a “heater” and which makes it possible to specifically heat the waveguide 50 with which it is associated. Heating a portion of the guide 50 modifies its optical properties and makes it possible to modify the phase of the optical signal which propagates inside this guide.Although not shown, each device 54 is connected to a control unit integrated in the chip 6. This unit controls each device 54 so that the phases of the optical signals which propagate in the guides 50 are equal at the level of the inputs of the adder 34. For example, for this, the control unit measures the energy of the optical signal at the output of the adder 34 and automatically and permanently adjusts, as a function of this measurement, the power supply of the devices 54 to maximize the energy of the optical signal on this output of the adder 34. Indeed, the energy at the output of the adder 34 is maximum when the optical signals received on its inputs are in phase.

[0032] The adder 34 has inputs optically connected to the end of each of the guides 50 to receive the optical signals to be added. It also has an output optically connected to the output 28.

[0033] The adder 34 combines the different optical signals received simultaneously on its inputs to form a single optical signal restored on its output. During this combination, the energies of the optical signals received on its inputs are added together so that the energy of the signal restored on the output is greater than the energy of each of the optical signals received on its inputs. For example, the adder 34 is produced by connecting in cascade several elementary power adders. Each of these elementary adders has only two inputs to receive two optical signals to be added and a single output on which the optical signal resulting from this addition is delivered. The “cascade connection” consists of connecting the outputs of two different elementary adders to respective inputs of another elementary adder and so on to obtain the number of inputs required for the adder 34.For example, an elementary adder is a multimode interferometer known by the acronym MMI (“MultiMode Interferometer”).

[0034] There figure 2 shows in more detail a first embodiment of the evanescent coupling zone 30. The network 42 comprises N waveguides 60 identical to each other. The guides 60 are arranged next to each other in a horizontal plane. Each guide 60 is offset relative to its nearest neighbor on the left: of a step dx - dxc in the X direction, and of a step dy in the Y direction.

[0035] In this text, the terms right and left are defined relative to the X direction.

[0036] The number N of guides 60 is chosen so that the distance d 2 between the rightmost guide 60 and the leftmost guide 60 is greater than the tolerance, in the X direction, on the position of the placement tool used to assemble chips 4 and 6. Here, the accuracy of the placement tool in the X direction is plus or minus 3 µm. Therefore, the number N is chosen so that the distance d 2 is greater than 6 µm. This number N is chosen after the pitch dx - d xc has been determined. As an illustration, on the figure 2 , N is equal to four.

[0037] Since the guides 60 are structurally identical to each other, hereinafter only one of these guides 60 is described in more detail.

[0038] The guide 60 extends mainly in the Y direction from a distal end 62 to a proximal end 64. The proximal end 64 is optically connected to a respective guide 50. Here, the cores of the guides 50 and 60 form a single block of material. Thus, the guide 60 extends, beyond the end 64, by the guide 50.

[0039] The end 62 is free and marks the beginning of the guide 60. The ends 62 and 64 of all the guides 60 are located between the planes P1 and P2. Here, the end 64 of the leftmost guide 60 is located in the plane P2 and the end 62 of the rightmost guide 60 is located in the plane P1.

[0040] The optical coupling between the guides 50 and the guide 40 must be weak and, preferably, non-existent. For this, here, each guide 50 follows a quarter-round trajectory whose radius of curvature is two or three times smaller than the radius of curvature Rmin 60 of the guide 60. This radius of curvature remains sufficiently large so that the optical losses by curvature of the guide 50 are low. For this, the radius of curvature of the guide 50 is greater than 5 µm, 15 µm or 20 µm. For example, here, the radius of curvature of the guide 50 is equal to 25 µm.

[0041] The guide 60 is symmetrical with respect to a plane P3 perpendicular to the direction Y. Consequently, subsequently, only the half, called “upstream”, of the guide 60 located between the end 62 and this plane P3 is described in detail. The part, called “downstream”, is deduced by symmetry with respect to the plane P3.

[0042] Between these two ends 62 and 64, the guide 60 is entirely comprised between a left lateral limit 66 and a right lateral limit 68. The limits 66 and 68 are rectilinear and parallel to the Y direction. The upstream half of the guide 60 is curved and touches the limit 66 at least once and the limit 68 at least once. Here, at each point of contact between the guide 60 and the limits 66 and 68, the guide 60 is tangent to the touched limit. In this embodiment, the upstream half of the guide 60 has only two points of contact, respectively, 70 and 72. Point 70 is located at the distal end 62 and point 72 is located at the plane P3.

[0043] The upstream half of the guide 60 comprises an upstream portion 74, an intermediate portion 78 and a downstream portion 76 located immediately after each other along the guide 60. The portion 74 extends from a plane P4 to a plane P5. The portion 76 extends from a plane P6 to the plane P3. The intermediate portion 78 extends from the plane P5 to the plane P6. The planes P4, P5 and P6 are parallel to the plane P1. The plane P4 contains the end 62.

[0044] The portions 74 and 76 are offset from each other in the X direction. Here, they each extend over at least 25% or 30% of the length of the upstream half of the guide 60. The portions 74, 78 and 76 are shaped to optically couple, by evanescent coupling, with, respectively, an upstream portion 80, an intermediate portion 83 and a downstream portion 82 of the guide 40 when these portions are located close to each other. The upstream portion 80 extends from the plane P4 to the plane P5. The downstream portion 82 extends from the plane P6 to the plane P3. The intermediate portion 83 extends from the plane P5 to the plane P6.

[0045] A portion of the guide 60 is considered to be in proximity to a portion of the guide 40 if the distance separating them is less than λ / 2 over the entire length of these portions.

[0046] To easily obtain evanescent optical coupling through the interface 8 between the portions 74 and 80 and between the portions 76 and 82, these portions extend mainly parallel to the Y direction. For this purpose, the smallest radius of curvature of the portions 74 and 76 is at least two or three times greater than the smallest radius of curvature Rmin 60 of the intermediate portion 78. Thus, the portions 74 and 76 are practically parallel to the Y direction.

[0047] The intermediate portion 78 is the portion most inclined relative to the Y direction. It is therefore the portion of the guide 60 which is potentially the most difficult to couple with a portion near the guide 40.

[0048] In this embodiment, the upstream half of the guide 60 has a single inflection point 84 located between points 70 and 72. This point 84 is located in the middle of the intermediate portion 78. In addition, the radius of curvature Rmin 60 is observed at this inflection point 84. Here, the point 84 is also a center of symmetry of the upstream half of the guide 60.

[0049] In this embodiment, the upstream half of the guide 60 extends along an "S" or half-sinusoid shaped trajectory. For example, the equation of this trajectory is defined by the following equation in an orthonormal reference frame XB, YB whose origin is located at the intersection of the limit 66 and a vertical plane passing through the point 84 and parallel to the plane P1: y = (L / 2) cos (πx / dx), where: y is the ordinate along the YB ordinate axis coincident with the lateral limit 66, x is the abscissa on the XB abscissa axis parallel to the X direction and contained in the P4 plane, L / 2 is the distance between the P4 and P3 planes, dx is the offset, in the X direction, between the starting and arrival points of this trajectory.

[0050] The starting point of the trajectory is contained in the plane P4 and the arrival point of this trajectory is contained in the plane P3.

[0051] Under these conditions, the radius Rmin 60 at point 84 is equal to (L / 2) 2 < / (π 2 < dx ). To limit losses by optical curvature in the guide 60, this radius Rmin 60 is chosen to be greater than 5 µm, 15 µm or 20 µm. Here, the radius Rmin 60 is greater than 100 µm or 200 µm.

[0052] Subsequently, the following notations are used: n r60 is the refractive index of the core of the 60 guide, and n eff60 is the effective index of the 60 guide at wavelength λ.

[0053] The effective index n eff60 is constant over the entire length of the guide 60.

[0054] In order for evanescent optical coupling between portions in the vicinity of the guides 40 and 60 to be possible, the cross-section of the guide 60 is dimensioned so that the index n eff60 is between 0.9n eff40 and 1.1n eff40 . For this purpose, here, the cross-section of the core of the guide 60 is identical to the cross-section of the core of the guide 40. In addition, since the cores and the claddings of these guides 40 and 60 are made of the same materials, this means that the effective index n eff60 is equal to the effective index n eff40 .

[0055] The refractive index n r8 of each material that separates the guide 40 from each of the guides 60 is less than 0.7min(n r40 ; n r60 ), where the symbol “min(...)” designates the function that returns the smallest of the elements in parentheses.

[0056] Finally, the distance which separates the horizontal plane containing the lower face 46 of the guide 40 from the horizontal plane containing the upper face of the guide 60 is small. Here, this distance is between 50 nm and λ / 2 and, preferably, between 50 nm and λ / 4 or between 50 nm and λ / 6. For the digital applications presented in this application, this distance has been taken equal to 200 nm.

[0057] Subsequently, the symmetrical portions 74, 76 and 78 of the guide 60 with respect to the plane P3 bear the numerical references, respectively, 94, 96 and 98.

[0058] The design and manufacture of assembly 2 will now be described with reference to the process of the figure 3 .

[0059] The method begins with a phase 100 of designing the coupling device 20. During this phase 100, the various parameters which characterize the conformation of this device 20 are determined, for example, as explained below.

[0060] Initially, the parameter dx is chosen so that, whatever the position x of the guide 40 above the guide 60, the guide 40 couples selectively with only one of the portions of the upstream part of the guide 60. The position x is an abscissa which identifies the position of the guide 40 in the X direction. Subsequently, the term "position x" designates only a position of the guide 40 located above the grating 42. For this purpose, the parameter dx is chosen to be greater than λ / 2 and also greater than two or three times the width of the guide 60 in the X direction. The value of the parameter dx is also chosen to be small enough so that the radius Rmin 60 is large and, for example, greater than 100 µm. For this, the value of the parameter dx is often chosen to be less than 3λ or 4λ. For example, the value of the parameter dx is chosen between λ / 2 and 1.5λ. Here, the value of the parameter dx is taken equal to 1.2 µm.

[0061] Then, the value of the parameter L of the guide 60 is chosen to maximize the amount of energy transmitted between the guides 40 and 60. It has been determined that the position x of the guide 40 least favorable to the transmission of energy between the guides 40 and 60 is the position x = dx / 4. The position x = dx / 4 is the position where, in the reference frame XB , YB , the orthogonal projection of the guide 40 on a horizontal plane containing the guide 60 passes through the point with coordinates (dx / 4 ; 0). This particular position of the guide 40 relative to the guide 60 is schematically represented on the figure 4 This is explained by the fact that, in this particular position, the guide 40 is close to the most curved portion 78. Furthermore, in the particular case of the guide 60, this is also explained by the fact that destructive interference may appear between the optical signal which enters this guide 60 at the level of the portion 78 and the optical signal which enters the portion 98.

[0062] There Figure 5 is a graph comprising two curves 106 and 108. Curves 106 and 108 represent the energy of the optical signal propagating in, respectively, the guide 40 and the guide 60 as the signal propagates in the Y direction. On this graph, the abscissa axis represents the amount of energy of the optical signal in a guide as a percentage of the initial energy of the optical signal in the guide 40. The ordinate axis represents the distance traveled by the optical signal in the Y direction. This graph is obtained for the relative position of the guides 40 and 60 shown in the figure 4 . Initially, 100% of the optical signal energy is in the guide 40. At a first crossing 110 ( Fig. 4 ) between the guides 40 and 60, a first part of the energy of the optical signal passes into the guide 60. Then, at the level of a second crossing 112 ( Fig. 4) between the guides 40 and 60, a second part of the energy of the optical signal passes into the guide 60.

[0063] The amount of energy at the end 64, for different values ​​of the parameter L of the guide 60, is constructed by numerical simulation. For example, the 3D-BPM software ("Beam Propagation Method") from the company Synopsys ®< is used for this.

[0064] There is a smallest value of the parameter L for which the energy of the optical signal at the end 64 is maximum. It is this value of the parameter L which is chosen. For example, in the particular case described here, the value of the parameter L is chosen equal to 57.5 µm. For this value, 93% of the energy of the optical signal initially in the guide 40 is transmitted in the guides 60 for the position x = dx / 4. For example, this value of the parameter L is sought between 0.7N t and 1.5Nt, where N t is equal to λ / (2|n effa - n effs |), where n effa and n effs are the effective indices of the asymmetric and symmetric modes of propagation, respectively. This relationship follows from equation (48) on page 6 of the following article: Henry F. TAYLOR and AI: "Guided Wave Optics", Proceedings of the IEEE, vol. 62, n08, 08 / 1974.

[0065] Then, the pitch dy is fixed. Here, the pitch dy is chosen so that the distance separating two immediately consecutive waveguides 50 is greater than λ / 2 and preferably greater than λ. This makes it possible to avoid optical coupling of the waveguides 50 between them.

[0066] It has been established that the value of the pitch dya has very little influence on the rate of transmission of energy through the interface 8 and this regardless of the position x of the guide 40. Thus, there is a large possible choice of acceptable values ​​for the pitch dy . To minimize the size of the zone 30, the pitch dy is chosen to be less than L / 10 or L / 20. Here, the pitch dy is chosen to be equal to dx .

[0067] Next, the parameter d xc is determined. Here, the parameter d xc is chosen to verify a first condition, namely that the guides 60 must not be optically coupled to each other. For this purpose, the value of the parameter d xc is such that the pitch dx - d xc is greater than λ / 2 or greater than 0.7 λ. Furthermore, the value of the parameter d xc is chosen to respect the following second condition: that whatever the position x of the guide 40, the transmission rate through the interface 8 is greater than 50% or 60%. This transmission rate is calculated by taking into account the sum of the energies of the optical signals at the proximal ends 64 of the guides 60.Thus, if for a given position x, a first portion of the guide 40 is coupled to a portion of a first guide 60 and a second portion of the guide 40 is optically coupled to a second portion of a second guide 60, the transmission rate is calculated taking into account that the transmitted energy is equal to the sum of the energies of the optical signals which propagate in these first and second waveguides.

[0068] It has been observed that there are many possible values ​​of the parameter d xc that satisfy the two previous conditions. Therefore, here, the value of the parameter d xc chosen is the one that satisfies the following third condition: the value of the parameter d xc maximizes, in addition, the average of the energy transmission rates across the interface 8 for each possible position x. For this, here, an energy transmission profile across the interface 8 is constructed for several possible values ​​of the parameter d xc . An example of such an energy profile constructed for the value d xc = -0.4 µm is shown in the figure 6 The abscissa axis represents the x position of the guide 40. The ordinate axis represents the energy transmission rate through the interface 8 expressed as a percentage. On the figure 7, the symbol "1" denotes the value 100%. Such a profile is constructed by numerical simulation using, for example, the 3D-BPM software. The average of the values ​​of the curve 110 ( Fig. 7 ) corresponds to the average of the transmission rates across interface 8 for each possible x position. The value of the parameter d xc that maximizes this average is retained. Here, the value that maximizes this average is obtained for d xc = -0.4 µm. For this value, the transmission rate is greater than 60% for all possible x positions.

[0069] Finally, the number N of guides 60 is chosen so that the length of the network 42 in the X direction is greater than the tolerance of the placement tool in this direction. Thus, it is certain that the placement tool will manage to place the chip 4 on the chip 6 in such a way that the guide 40 is located above the network 42. Under these conditions, the position x necessarily corresponds to a position where the transmission rate is greater than 60%.

[0070] The placement tool also has an angular tolerance which means that the angle θ is not systematically equal to 0° but, on the contrary, varies between θ min and θ max . For conventional placement tools, this angular tolerance is often between -0.1° and +0.1°.

[0071] THE figures 7 and 8 represent the energy transmission profiles obtained with the same parameter values ​​as those used to obtain the profile of the figure 7 but this time for θ = -1° ( figure 7 ) and θ = +1° ( figure 8 ). The profile of the figure 6 was obtained for the angle θ = 0°. As can be observed, the transmission rate in these profiles remains systematically greater than 55%. Thus, the zone 30 makes it possible to obtain a transmission rate of energy through the interface 8 greater than 50% regardless of the value of the angle θ between -1° and +1°. In addition, for the values ​​θ = -0.1° and θ = +0.1°, the transmission rate remains systematically greater than 60%. The coupling device 20 is therefore compatible with an assembly of the chips 4 and 6 using a placement tool having an angular tolerance of plus or minus 0.1°.

[0072] Once the design phase 100 is completed, a manufacturing phase 120 of the assembly 2 is executed. Phase 120 is essentially composed of three steps 122, 124 and 126.

[0073] Step 122 is a step of manufacturing the photonic chip 4 and therefore of the part of the device 20 produced in this chip 4. This step 122 therefore notably includes the production of the guide 40.

[0074] Step 124 is a step of manufacturing the chip 6 and therefore the part of the device 20 produced in this chip 6. Step 124 therefore comprises the production of the network 42, the phase adaptation zone 30 and the adder 34. Step 124 is carried out independently of step 122.

[0075] Step 126 is a step of bonding chip 4 onto chip 6 to obtain assembly 2. Step 126 includes in particular an operation 128 of placing, using the placement tool, chip 4 onto chip 6. During operation 128, guide 40 is placed above network 42.

[0076] There figure 9represents an optical coupling device 140 identical to the device 20 except that the grating 42 is replaced by a grating 142. The grating 142 is identical to the grating 42 except that each waveguide 60 is replaced by a waveguide 144. The guide 144 is identical to the guide 60 except that the downstream half of the guide 60 is omitted. Thus, each guide 144 is here identical to the upstream half of the guide 60 after having undergone symmetry with respect to an axis parallel to the Y direction. Therefore, in this embodiment, each guide 144 has only two points of contact with, respectively, the lateral limits 66 and 68 and therefore also a single point of inflection.

[0077] The operation of the device 140 is the same as that of the device 20. However, due to the lack of symmetry of each guide 144 with respect to a vertical plane parallel to the X direction, the tolerance with respect to an angular positioning error of the guide 40 is less symmetrical. In other words, the angular range in which the values ​​of the angle θ can be included while maintaining a transmission rate greater than 50% is not symmetrical with respect to the value 0°.

[0078] In the case of guide 144, the position x of guide 40 which is the least favorable for the transmission of energy between guides 40 and 144 is the position x = dx / 2. It is therefore this position x = dx / 2 which is used to determine the value of parameter L. Chapter II: Variants Variant of the 40, 60 and 144 waveguides :

[0079] Alternatively, the guide 40 is not straight but curved. For example, the guide 40 is replaced by a curved guide. The shape of this curved guide is, for example, symmetrical to the shape of the guide 60 with respect to a vertical plane parallel to the Y direction. In this case, both the guide 40 and the guide 60 are curved.

[0080] If the alignment error in the Y direction of the placement tool is negligible, then the guide 40 is not necessarily longer than the guide 60. In this case, the guide 40 can start at the plane P1 and end at the plane P2.

[0081] Many other embodiments of the guide 60 or 144 are possible. For example, the guide 60 or 144 may have more than three or four points of contact with the boundaries 66 and 68. In this case, the guide 60 or 144 has more than two points of inflection.

[0082] In another embodiment, portions 74 and 76 are rectilinear and parallel to the Y direction.

[0083] The 60 and 144 guides can also be straight and not curved. In this case, the guide 40 must be curved to have an upstream portion and a downstream portion offset in the X direction. For example, the guide 40 is shaped as described for the guide 60 or 144. In this embodiment, the parameters dx and L characterize the shape of the guide 40 and no longer the shape of the guide 60 or 144. The dimensioning of the guide 60 or 144 is carried out as described previously for the guide 40. In particular, the length in the Y direction of the guide 60 or 144 is chosen to be much greater than that of the guide 40. Then, the shape of the guide 40, and in particular the choice of the values ​​of the parameters dx and L for the guide 40, is carried out as described previously for the guide 60. Finally, the value of the pitch dx - d xc between the rectilinear guides 60 or 144 is determined as described previously for the network 42.Under these conditions, whatever the position x of the curved guide 40, portions of this guide are capable of coupling, by evanescent coupling, to portions in the vicinity of one or two rectilinear guides 60 or 144.

[0084] The value of the parameter dx can be chosen so that the guide 40 comprises a portion optically coupled by evanescent coupling with a first copy of the guide 60 and a second portion optically coupled with a second copy of the guide 60, a third copy of the guide 60 being located between these first and second copies of the guide 60.

[0085] The value of the parameter d xc can be chosen without taking into account the third condition that the average of the transmission rates for the different positions x is maximized. Others variants:

[0086] Optical coupling between the guide 40 and any of the guides 50 can be avoided in different ways. For example, the dimensions of the core of the guide 50 are adjusted so that its effective index is very different from the effective index n eff40 . This prohibits optical coupling of this guide 50 with the guide 40 without having to adopt a particular trajectory for the guide 50.

[0087] The phase shifter 54 can be implemented differently. For example, as a variant, the heater is replaced by an adjustable PiN diode phase shifter. In a very simplified embodiment, the length of each of the guides 50 is adjusted so that the optical signals received at the inputs of the adder 34 are in phase. Preferably, this adjustment of the lengths of the guides 50 is combined with the use of an adjustable phase shifter like those previously described. In another embodiment, this adjustment of the lengths of the guides 50 is followed, after bonding the chips together, by an operation of implanting impurities in the guides 50 to adjust the phase shift between the optical signals propagating in these guides and to obtain optical signals in phase at the inputs of the power adder. For example, such implantation methods for adjusting the phase of an optical signal propagating in a waveguide are described in application FR1553024.

[0088] In another embodiment, the phase shifter is made by placing a phase change material near the core of the waveguide coupled to electrodes to pass a current through it. The mode will therefore have a different effective index depending on the state of the crystalline phase of the phase change material. For example, the phase change material is GeSbTe known by the acronym GST. Chapter III: Advantages of the described embodiments

[0089] The use of evanescent coupling between the guide 40 and the guide 60 or 144 is a simple means for optically coupling the chip 4 to the chip 6 through the bonding interface 8. In particular, such evanescent coupling is simpler to implement than known devices which require the implementation of optical components such as a lens or a mirror. In addition, the coupling devices described here are tolerant at least with respect to alignment errors in the X direction.

[0090] The fact that the curved guide undulates between two lateral limits and has a single point of inflection between two successive right and left points of contact makes it possible to make the coupling device tolerant of angular positioning errors of chip 4 on chip 6.

[0091] The fact that each curved guide has a vertical plane of symmetry parallel to the X direction makes it possible to obtain a tolerance with respect to angular positioning errors which is symmetrical with respect to the angular position where this angular positioning error is zero. In other words, this tolerance is the same with respect to angular position errors in the trigonometric or anti-trigonometric direction.

[0092] The fact that the curved guide has only two inflection points makes it possible to minimize the number of intermediate portions where optical coupling between the curved guide and the guide 40 is more difficult to obtain. This therefore makes it possible to limit the length of the curved guide and therefore to reduce the size of the coupling device.

Claims

1. Assembly comprising: - first and second photonic chips (4, 6) adjoined to each other by bonding via a bonding interface, - an optical coupling device (20; 140) that optically connects together the first and second photonic chips through the bonding interface in order to allow propagation between the first and second photonic chips of an optical signal at a predetermined wavelength A, this optical coupling device comprising first and second portions produced in the first and second chips, respectively, - the first portion of the optical coupling device comprises a first waveguide (40) that mainly extends in a first direction parallel to the bonding interface, - the second portion of the optical coupling device comprises a second waveguide (60; 144) able to be optically coupled to the first waveguide, when the first waveguide is located above this second waveguide, via evanescent coupling that allows at least 50% of the energy of the optical signal, at the wavelength λ, propagating through the first waveguide to be transferred to this second waveguide, characterized in that: - the second portion of the optical coupling device comprises: - an array (42; 142) of a plurality of identical copies of the second waveguide, these second waveguides being offset with respect to one another, in a second direction, by a pitch larger than A / 2, this second direction being parallel to the bonding interface and perpendicular to the first direction, and - a power summer (34) comprising inputs that are optically connected to one end of each of the second waveguides of the array, and an output on which the summer delivers the addition of the optical signals propagating through each of the second waveguides, - each of the first and second waveguides comprises upstream (74, 80), intermediate (78, 83) and downstream (76, 82) segments placed immediately one after the other in the direction of propagation of the optical signal, the upstream and downstream segments of the first waveguide or of each second waveguide being offset with respect to each other in the second direction, and - the configurations of the first waveguide and of the second waveguides are such that, for any position of the first waveguide above the array of second waveguides, the distance between one of the segments of the first waveguide and one of the segments of one of the second waveguides is smaller than A / 2.

2. Assembly according to Claim 1, wherein the waveguide (60; 144) the upstream and downstream segments of which are offset comprises the following features: - this waveguide is entirely comprised between left and right lateral limits (66, 68), each of these lateral limits being rectilinear and parallel to the first direction, - this waveguide is tangent to the left lateral limit at at least one first left point of contact (72) and tangent to the right lateral limit at at least one right point of contact (70), and - between each pair of left and right points of contact immediately consecutive along this waveguide, the waveguide has a single point of inflection (84).

3. Assembly according to Claim 2, wherein the waveguide (60) the upstream and downstream segments of which are offset is symmetric with respect to a plane (P3) perpendicular to the first direction.

4. Assembly according to Claim 3, wherein the waveguide (60) the upstream and downstream segments of which are offset comprises solely three points of contact forming two different pairs of left and right points of contact immediately consecutive along this waveguide.

5. Assembly according to any of Claims 2 to 4, wherein the distance between the left and right lateral limits is larger than A / 2.

6. Assembly according to any of the preceding claims, wherein: - the first waveguide (40) has an effective index neff1 at the wavelength λ and comprises a core comprising a first face (46) turned toward the bonding interface, this first face lying in a first plane parallel to the bonding interface, this core being made of a material of refractive index nr1 at the wavelength A, - each second waveguide (60; 144) has an effective index neff2 at the wavelength λ comprised between 0.9neff1 and 1.1neff1 and comprises a core comprising a second face turned toward the bonding interface, this core being made of a material of refractive index nr2 at the wavelength A, this second face lying in a second plane parallel to the bonding interface, the distance between the second plane and the first plane being comprised between 50 nm and A / 2, - the space between the first and second planes is filled with a material the refractive index of which at the wavelength λ is lower than 0.7Min(nr1; nr2), where Min(nr1; nr2) is the function that returns the lowest of the refractive indices nr1 and nr2.

7. Assembly according to any of the preceding claims, wherein: - all of the upstream and downstream segments that are offset with respect to each other are located between a first plane (P1) and a second plane (P2) that are perpendicular to the first direction, and - the waveguide (40) the upstream and downstream segments of which are not offset is a rectilinear waveguide the length of which is larger than d1+3µm, where d1 is the distance that separates the first and second planes (P1, P2).

8. Assembly according to any of the preceding claims, wherein the second portion of the optical coupling device comprises an adjustable phase shifter (54) able, after the first and second photonic chips have been bonded to each other, to adjust the phase of the optical signals received on the inputs of the power summer so that these optical signals arrive in phase on the inputs of the power summer.

9. Second photonic chip (6) for producing an assembly according to any of the preceding claims, this second photonic chip comprising: - a bonding interface intended to adjoin this second photonic chip to the first photonic chip, and - a second portion of the optical coupling device (20; 140), which optically connects together the first and second photonic chips through the bonding interface in order to allow propagation between the first and second photonic chips of an optical signal at a predetermined wavelength A, this second portion comprising a second waveguide (60; 144) able to be optically coupled to the first waveguide, when the first waveguide is located above this second waveguide, via evanescent coupling that allows at least 50% of the energy of the optical signal, at the wavelength λ, propagating through the first waveguide to be transferred to this second waveguide, the second waveguide mainly extending in a first direction, characterized in that: - the second portion of the optical coupling device comprises: - an array (42; 142) of a plurality of identical copies of the second waveguide, these second waveguides being offset with respect to one another, in a second direction, by a pitch larger than A / 2, this second direction being parallel to the bonding interface and perpendicular to the first direction, and - a power summer (34) comprising inputs that are optically connected to one end of each of the second waveguides of the array, and an output on which the summer delivers the addition of the optical signals propagating through each of the second waveguides, - each of the second waveguides comprises upstream (74, 80), intermediate (78, 83) and downstream (76, 82) segments placed immediately one after the other in the direction of propagation of the optical signal, the upstream and downstream segments of each second waveguide being offset with respect to each other in the second direction, and - the configurations of the second waveguides are such that, for any position of the first waveguide above the array of second waveguides, the distance between one of the segments of the first waveguide and one of the segments of one of the second waveguides is smaller than A / 2.

10. Process for fabricating an assembly according to Claim 1, wherein the process comprises: - bonding (126), via a bonding interface, first and second photonic chips adjoined to each other, - producing (122, 124) an optical coupling device that optically connects together the first and second photonic chips through the bonding interface in order to allow propagation between the first and second photonic chips of an optical signal at a predetermined wavelength A, this optical coupling device comprising first and second portions produced in the first and second chips, respectively, - the first portion of the optical coupling device comprising a first waveguide that mainly extends in a first direction parallel to the bonding interface, - the second portion of the optical coupling device comprising a second waveguide able to be optically coupled to the first waveguide, when the first waveguide is located above this second waveguide, via evanescent coupling that allows at least 50% of the energy of the optical signal, at the wavelength λ, propagating through the first waveguide to be transferred to this second waveguide, characterized in that production of the coupling device comprises producing (124) a coupling device in which: - the second portion of the optical coupling device comprises: - an array of a plurality of identical copies of the second waveguide, these second waveguides being offset with respect to one another, in a second direction, by a pitch larger than A / 2, this second direction being parallel to the bonding interface and perpendicular to the first direction, and - a power summer comprising inputs that are optically connected to one end of each of the second waveguides of the array, and an output on which the summer delivers the addition of the optical signals propagating through each of the second waveguides, - each of the first and second waveguides comprises upstream, intermediate and downstream segments placed immediately one after the other in the direction of propagation of the optical signal, the upstream and downstream segments of the first waveguide or of each second waveguide being offset with respect to each other in the second direction, and - the configurations of the first waveguide and of the second waveguides are such that, for any position of the first waveguide above the array of second waveguides, the distance between one of the segments of the first waveguide and one of the segments of one of the second waveguides is smaller than A / 2.