EP1002252A1

EP1002252A1 - Connecting assembly of optical fibres with optical or optoelectronic components

Info

Publication number: EP1002252A1
Application number: EP99939787A
Authority: EP
Inventors: Guy Parat; Patrice Caillat; Christiane Puget
Original assignee: Commissariat a lEnergie Atomique CEA
Current assignee: Commissariat a lEnergie Atomique et aux Energies Alternatives CEA
Priority date: 1998-06-09
Filing date: 1999-06-08
Publication date: 2000-05-24
Also published as: FR2779536B1; FR2779536A1; WO1999064907A1; US6435733B1

Abstract

The invention concerns an assembly for connecting optical fibres with optical or optoelectronic components and a method for making said assembly. The invention is characterised in that it consists in forming at least a fibre support (4, 6) comprising, for each fibre (2), a V-shaped housing; fixing the support and the component (32) to a substrate (12) via fusible links (weld nuggets) (16, 38) and positioning the fibre in the housing so as to obtain a fibre alignment vertical with respect to the component, the horizontal alignment being obtained by means of the melted elements. The invention is applicable in microelectronics.

Description

CONNECTION ASSEMBLY OF OPTICAL FIBERS WITH OPTICAL OR OPTOELEC ^¬ TRONIC COMPONENTS

DESCRIPTION

5 TECHNICAL AREA

The present invention relates to an assembly allowing the connection of optical fibers with optical or optoelectronic components and to a method of manufacturing this assembly.

The invention finds applications in the field of microelectronics, in particular in all cases where it is a question of connecting optical fibers to laser sources or to optical modules (for example dividers, multiplexers or

15 sensors) mounted on optoelectronic substrates.

In particular, in the case of telecommunications by optical transmission, the invention can be used when a diode-laser (or a strip of diode-lasers with lateral emission) must

20 be connected to an optical fiber (or to several optical fibers).

PRIOR STATE OF THE ART

In the microelectronics field, the increase in the operating frequency of the 25 electronic systems requires: - the design of new data transmission principles and in particular the paralleling of electric buses allowing the simultaneous transmission of several signals at the same time and / or - the use of light thanks to optical waveguides (guides of integrated waves or optical fibers) to increase the information rate.

These optical waveguides make it possible to obtain very good immunity to electromagnetic disturbances.

In addition, optical transmission requires the production of modules for transmitting, receiving and processing light signals. To do this, techniques are developed in particular on glass or silicon to ensure:

- the coupling of optical fibers

- optical and electrical connections of optoelectronic components

- the electrical connection of electronic interface components.

We will consider the following documents:

[1] "Soldering technology for optoelectronic packaging", 1996 Electronic Component and Technology Conférence, p.26 to 36

[2] "Passive alignment for optoelectronic components, Advances in electronic packaging, EFP-vcl. 19-1, 1997, vol.l, p.753 to 758

[3] "Silicon motherboards fer ultichannel optical module", IEEE transactions on components, packaging, and manufactuπng technology, Part A, vol.19, n ° l, March 1996, p.34 to 40

[4] "Flip-chip optical fiber attachment to a monolithic optical receiver chip", SPIE, vol.2613, p.53 to 58.

Each of the assemblies known from documents [1] to [4] is generally in the form of a substrate on which: - optical fibers are connected so t opposite optical waveguides formed in the substrate either viewing of laser diodes and / or photodetectors,

- optoelectronic components fitted into this substrate or placed on the surface thereof are coupled to optical waveguides and / or to optical fibers,

- electronic components addressing or collecting information coming from optoelectronic components are positioned. The optical fibers and the optical or optoelectronic components must be perfectly aligned with each other to minimize optical losses.

The target details may be less than 0.5 μm.

To do this, two techniques are mainly used.

1) There is known a technique for active alignment of an optical fiber with a diode-laser, the objective of which is to ascertain in real time the alignment performance by an electrical measurement with a photodiode. To do this, the diode-laser is supplied and a measurement of the light power at the output of the fiber gives an indication of the relative alignment of the latter and of the diode-laser. Optimization of the alignment is ensured by small displacements of this laser diode by means of mechanical or piezoelectric micromanipulators. An assembly can then be obtained by gluing.

This active alignment technique has drawbacks:

- length of the alignment process,

- Need for a mechanical blocking, by means of an adhesive for example, of the diode-laser after alignment and - necessity that this blocking does not cause mechanical stresses likely to modify 1 alignment.

2 ⁾ A passive fiber optic alignment technique is also known, the main objective of which is to reduce costs.

When the fiber has to be connected parallel to a substrate, for example made of silicon, the most widespread connection method currently consists in producing a V-shaped cavity in the silicon substrate which has the function of an optical micro-bench, for example according to the principle of etching according to preferential crystal planes (100).

The fiber is wedged and glued to the bottom of the cavity opposite an optoelectronic component. This optoelectronic component, if attached to the substrate, is generally mounted upside down by the flip-chip technique on metal links ensuring electrical continuity, mechanical strength and thermal evacuation to the substrate. The alignment of the optoelectronic component in front of the fiber must be an absolute alignment in the three directions of space. For this we can use:

- very precise equipment which makes it possible to position and weld the component, while maintaining it on the substrate, with precision at the angle of 1 μm,

- solders with positioning shims made in the substrate and / or in the component to be assembled,

- brazing elements without shims using the self-positioning effect linked to the wettability forces of the solder in the liquid phase, the self-positioning taking place parallel to the substrate by wettability on metal pads and perpendicular to the substrate by volume control of the soldering elements.

Besides the use of cavities in the form of

V in the substrate, there is a method of fixing by bonding the fibers in an intermediate silicon support, also etched in a V shape, and then transferred upside down on the substrate (see the document

[4]). The alignment and brazing of the intermediate support are carried out using precision equipment. The liquid phase self-alignment effect is not used. The stiffness of the fibers and the weight or the whole do not allow it. When a fiber is connected perpendicularly to a substrate, there is also a method of inserting the fiber into a pierced shim, previously brazed and mounted by flip-chip and using the auto effect. -alignment in liquid phase. On this subject we will consider Figure 5 of document [1].

A problem arises in the case of passive alignment of an optical fiber in front of an optoelectronic component attached to an interconnection substrate.

When the fiber is fixed parallel to the substrate in a V-shaped cavity integrated into this same substrate, the optical positioning of the optoelectronic component in front of the fiber must be absolute in the three directions of space.

In a plane parallel to the substrate, the metal studs produced, receiving a solder, are perfectly aligned with the V-shaped cavity because they are generated by the same lithographic mask. The self-alignment effect in liquid phase of the solder microbeads ensures the correct positioning of the component in front of the fiber.

On the other hand, in a direction perpendicular to the substrate, the control of the height of the optical axes requires, for the fiber, the control of the width of the V-shaped cavity (variation of the burial of the fiber) and for the component optoelectronics, either a mechanical hold or a solder volume control. These operations are dependent on technological variations in manufacturing. PRESENTATION OF 1 / INVENTION

The object of the present invention is to define an assembly and a manufacturing method thereof, allowing very precise passive alignment of one or more optical fibers with one or more optical or optoelectronic components, relatively.

This assembly is carried out using micro-beads of a fusible material (orasure) on a substrate, which can be an interconnection substrate, playing the role of optical micro-bench.

The fusible material constituting the microbeads is for example indium or a fusible alloy based on tin and lead or any alloy with low melting point.

Specifically, the present invention firstly relates to an assembly comprising a substrate and, on this, at least one optical fiber support, at least one optical fiber disposed in this support and at least one optical component or optoelectronics, the optical axis of the fiber and the optical axis of the component being aligned, this assembly being characterized in that the support and the component are fixed to the substrate by means of microbeads of fusible material, allowing the axis fiber optic and the optical axis of the component to be parallel to each other in a plane plane perpendicular to a surface to the substrate, in that the support comprises, for each fiber, a housing in the form of V having two walls inclined towards each other, the opening of the "V" being located on the face of the support which is not fixed to the substrate, and in that the fiber is positioned in the housing, the volume of each microbead and the housing being determined so that the optical axis of the fiber and the optical axis of the component are parallel to each other in the same plane parallel to the surface of the substrate.

According to a first particular embodiment of the assembly which is the subject of the invention, the housing passes through the entire support, the housing thus having two lower edges, the distance separating these two lower edges being determined, taking into account the diameter of the fiber and its support points in the housing, so that the optical axis of the fiber and the optical axis of the component are parallel to each other in the same plane parallel to the surface of the substrate .

According to a second particular embodiment, the housing does not pass through the support. The present invention also relates to a method of manufacturing an assembly comprising a substrate and, on this, at least one optical fiber support, at least one optical fiber disposed in this support and at least one optical or optoelectronic component. , the optical axis of the fiber and the optical axis of the component being aligned, this process being characterized in that the fiber support is formed, this support comprising, for each fiber, a V-shaped housing having two walls inclined towards each other, the opening of the 'V' being located on the face of ^the support which is not fixed to the substrate, and also comprising a plurality of first fastening studs, in that also forms a plurality of first attachment studs on the component, in that second attachment studs are formed on the substrate, these second studs being intended to be respectively associated with the first studs, in that one forms, on the first studs and / or the second pads, elements made of a fusible material, capable of being soldered to the first and second pads, these first and second pads being wettable by this material in the molten state while their environment is not, in that the support and the component are fixed to the substrate by means of the corresponding elements, these elements being brought in the molten state for this purpose and allowing the optical axis of the fiber and the optical axis of the component to be parallel to each other in the same plane perpendicular to a surface of the substrate, and in that the fiber is positioned in the housing, the volume of each element and the housing being determined so that the optical axis of the fiber and the optical axis of the component are parallel to each other in the same plane parallel to the surface of the substrate.

According to a first particular mode of implementation (“V through”) of the method which is the subject of the invention, the housing crosses the entire support, the housing thus having two lower edges, the distance separating these two lower edges being determined, taking into account the diameter of the fiber and its support points in the housing, so that the optical axis of the fiber and the optical axis of the component are parallel to each other in the same plane parallel to the surface of the substrate. According to a second particular implementation method (“V not traversing), the housing does not pass through the support.

The support can comprise a plurality of parallel copies of said V-shaped housing and a plurality of optical fibers are then positioned respectively in said copies.

Each fiber can be fixed with an adhesive in the corresponding housing. Preferably, for greater solidity of the assembly, the adhesive spreads between the support and the substrate, around the elements made of fusible material placed under the support.

According to a particular embodiment of the method which is the subject of the invention, a cover is produced which is capable of covering each housing and each fiber is pressed against the corresponding housing by means of this cover and then advantageously fixed in this corresponding housing. This cover can be transparent, which makes it possible to observe each fiber in its housing and even to stick this fiber in this housing by hardening of an adhesive polymerizable by radiation (generally an ultraviolet radiation) which is sent towards the adhesive. through the hood.

According to a first particular embodiment of the method which is the subject of the invention, the cover comprises, for each fiber, a protuberance by means of which this fiber is pressed against the corresponding housing.

According to a second particular embodiment of the process which is the subject of the invention, the cover comprises a flat face by means of which each fiber is pressed against the corresponding housing, the fiber protruding from the face of the support which is not fixed to the substrate.

According to a particular embodiment of the invention, relating to the “V through”, the support is formed from a plate in which at least two V-shaped walls are produced by chemical and / or mechanical etching of this plate. , so as to obtain the distance determined between the two lower edges of each V, the first studs corresponding to the support are formed by an αe pnotolithography technique and said walls of the housing are separated from one another.

In the particular case where the plate has an excessively large initial thickness, the formation of the support further comprises a step of thinning the plate carried out before or after the etching of the V.

According to another particular embodiment relating to the “non-traversing V”, the support is formed from a plate in which at least two walls are formed in V oe shape by chemical and / or mechanical etching of this plate and the first studs corresponding to the support by a photolithography technique. Preferably, the first pads corresponding to the component are formed by a photolithography technique.

Also preferably, the second pads are formed on the substrate by a photolithography technique.

According to a preferred embodiment of the invention, the elements formed on the second Dlots all have the same thickness. Preferably, the elements formed on the second studs are passed in the molten state, each element then taking substantially the form of a ball, these elements are then passed in the form of a ball in the solid state and the the latter in the molten state to assemble the component and the support with the substrate.

The shape of the balls depends on the shape of the second studs and is therefore not necessarily spherical. Preferably also, to have the same precision on the height of the different balls, the elements are formed simultaneously by photolithography on the second studs.

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will be better understood on reading the description of exemplary embodiments given below, by way of purely indicative and in no way limiting, with reference to the appended drawings in which: • Figure 1 is a schematic cross-sectional view of an assembly according to the invention in which an optical fiber is pressed onto thick shims by a cover containing a protuberance, • Figure 2 is a schematic cross-sectional view of an assembly according to the invention in which a optical fiber is pressed onto thin shims by means of a flat surface cover, FIG. 3 is a schematic cross-sectional view of an assembly according to the invention in which an optical fiber is pressed against shims in front of a laser diode,

• Figure 4 is a schematic cross-sectional view of an assembly according to the invention in which optical fibers are fixed on thin shims, • Figure 5 is a schematic top view of an assembly according to invention in which an optical fiber and a laser diode are optically aligned, the cover making it possible to press the fiber into the housing defined by the shims not being shown,

FIG. 6 is the section AA in FIG. 5, the cover being shown,

FIGS. 7, 8A, 8B, 9, 10A, 10B, 11A and 11B schematically illustrate a process for manufacturing a fiber support usable in the present invention,

FIGS. 12A to 12E schematically illustrate various stages of preparation of a substrate for the implementation of a method according to the invention,

FIGS. 13A to 13D schematically illustrate steps for implementing this method according to the invention which follow the steps illustrated schematically in FIGS. 12A to 12E, • Figure 14 is a schematic cross-sectional view of another assembly according to the invention in the case of a "non-crossing V", and • Figures 15A, 15B, 16A and 16B schematically illustrate stages of a method of manufacturing a “V non-through” fiber support usable in the invention.

DETAILED PRESENTATION OF PARTICULAR EMBODIMENTS

In the examples of the invention which follow, in order to ensure optical alignment, each optical fiber is positioned and advantageously fixed in a housing defined by two bevelled wedges which form a fiber support. These two wedges are previously brazed on an interconnection substrate at the same time as the optical or optoelectronic component which it is desired to connect to the fiber.

This brazing takes place following a rough prepositioning (error less than or equal to 5 μm) of the fiber and the component.

The surface tension forces of the solder microbeads in the liquid phase (technique known from document [1]) make it possible to obtain a "horizontal" optical alignment, that is to say to render the optical axis of the fiber and the optical axis of the component parallel to each other in the same plane perpendicular to the surface of the substrate, surface on which the connection is made, when this surface is planar (or parallel to a planar surface area of the substrate when the latter comprises steps defined by planar and parallel surface areas, offset in height relative to each other). "Vertical" optical alignment, that is to say making the optical axis of the fiber and the optical axis of the component parallel to each other in the same plane parallel to the surface or substrate, is controlled by the relative volumes of the microbeads 0 and by the distance separating the two lower edges of the housing taking into account the diameter of the fiber and its support points in the housing.

In fact, thanks to such an assembly, the positioning accuracy of the fiber and or component 5 is very important because it becomes relative precision and no longer absolute precision as in the prior art. It is nevertheless necessary that the core ("core") of the optical fiber is well referenced with respect to the cylindrical external surface C of the optical cladding ("cladding") that the flora.

Figure 1 shows an assembly according to the invention between an optical fiber and a component, for example a laser diode (not shown).

The optical fiber 2 is placed on thick and bevelled shims 5 4 and 6 forming a fiber support and a machined cover 8, provided with a protuberance 10, presses the fiber against the respective walls of the shims which form a through V when they are seen in cross section. C We also see an interconnection substrate 12, the upper surface S of which has generally metallic latching pads 1 ^Δ which are formed by etching on the substrate and on which solder microbeads 16 have been previously formed. The two bevelled wedges were produced for example in a silicon wafer, the surface of which was oriented along the plane (100), by etching along preferential crystalline planes using KOH, then metallized, to form bonding pads. 18 on the undersides of the shims, and brazed freely on the substrate after positioning of the studs 18 on the microbeads.

In the example shown, the height H3 of the shims is 250 μm and the angle β of the walls of the V-shaped housing defined by the two shims, relative to the surface S of the substrate, is 54.74 °. The positioning of the two shims for the “horizontal” optical alignment of the fiber and of the component takes place automatically by wettability of the solder on the metal attachment studs 14 and 18 which face each other. The vertical positioning, that is to say in a direction perpendicular to the surface S, depends on the distance H2 (22 μm in the example shown) between these lower faces of the wedges and the upper surface S of the substrate which is directly related to the volume of solder, and the distance separating the lower edges of the housing taking into account the diameter of the fiber and its points of support in the housing.

Indeed, for a fiber of given diameter and a given distance between edges, according to the angle β, this fiber will have a variable “vertical” positioning due to the modification of the support points. Likewise, for a given angle β and a given distance between edges, depending on the diameter of the fiber, the “vertical” positioning will also be variable. The distance between edges is determined by the precise positioning of the attachment studs on the blocks.

Thus, knowing the distance D between the attachment studs near the edges and the edges, knowing the pitch of the studs made on the substrate, makes it possible to know the distance between the edges.

The optical fiber 2, whose optical sheath 20 has, in the example shown, an outside diameter d of 120 μm (diameter of the fiber) is placed in the V-shaped housing so that the axis XI of the fiber (axis of the heart 22 thereof) is parallel to the upper surface of the substrate.

Then, for example, the V-shaped housing is filled with an adhesive 24 and the fiber, for example, is brought into contact with the shims by pressing on this fiber with the cover until the adhesive is dry.

As the angle β is perfectly known, the final dimension Hl (8 μm in the example shown), minimum distance between the external surface of the optical cladding 20 and the upper surface S of the substrate, depends as we have seen on the precision of the positioning of the attachment studs 18 formed on the lower faces of the shims and therefore of the dimension D which is the distance between the lower edge 26 of a shim and the axis of the attachment stud 18 of this shim the closer to this edge and which is for example 50 μm. In the example shown, the cover is machined or molded. It can be transparent, for example made of PMMA (polymethylmethacrylate), to check the quality of the bonding and / or to polymerize the adhesive by ultraviolet radiation sent through this transparent cover, when this adhesive is photopolymerizable.

Instead of being made of silicon, the shims can be made of another factory-produced material very precisely with a sub-micronic control of the dimensions.

Preferably, a large number of microbeads 16 are placed under the shims in order to guarantee good mechanical strength when positioning the fiber and bonding it. The glue can fill the entire volume between the cover and the shims but it can also overflow under these shims around the solder microbeads. This strengthens the mechanical strength of the assembly to prevent any tearing of the fiber when handling the finished assembly.

The thickness H3 of the beveled shims may vary depending on the application considered for the assembly. Only the contact area between the fiber and each wedge is important. The shape of the cover holding the fiber against the shims can also vary depending on the thickness of these shims (see Figure 1 and Figures 2, 3, and 4).

This cover can be manufactured by etching, for example, a silicon wafer using KOH (case of FIG. 3) or be machined or molded in another material. In all cases, the cover can advantageously press each fiber against the wedges.

The dimensions of this cover are not critical except for the dimension H4, that is to say the height of the protuberance 10 (FIG. 1) which is the distance between the planar lower wall 28 of the cover and the lower face 30 of this protuberance. ). This dimension H4 should be sufficient to press on each fiber so that it is well maintained in contact with the wedges (see Figure 1). In the example in FIG. 1, H4 is chosen> 160 μm.

In the case of mounting several fibers parallel to each other (see Figure 4) the cover is supported only on the fibers. In the case of a single fiber, the cover can bear on this fiber and possibly on one of the two wedges at the same time.

Figure 2 shows an assembly according to the invention in which a fiber 2 is pressed against two thin shims 4 and 6 by means of a cover

8, the lower face 28 of which is flat (without protuberance).

We see that the assembly of Figure 2 is comparable to that of Figure 1 but, in the case of Figure 2, the thickness or height H3 of the shims is less than the diameter d of the fiber (H3 is for example d minus about 20 μm), which allows the use of the cover 8 without protuberance.

FIG. 3 shows an assembly according to the invention in which an optical fiber 2 is held on bevelled wedges 4 and 6, opposite an optical or optoelectronic component such as for example a diode-laser 32, and optically connected to the latter.

Figure 3 is comparable to Figure 1. The cover 8 of Figure 3 simply has a different shape. The diode-laser is brazed on the substrate 12 advantageously at the same time as the shims. We see the attachment pads 34 formed on the substrate 12 and corresponding to this laser diode 32, the attachment pads 36 formed on this laser diode and the solder microbeads 38 connecting the pads 34 respectively to the pads 36. The microbeads 38 being preferably all made on the substrate 12, an overall variation in the height of the solder deposit generates a vertical offset (that is to say an offset perpendicular to the upper surface S of the substrate) of the fiber and the diode-laser but the vertical alignment of the optical axis XI of the fiber and the optical axis of the diode-laser is preserved.

Figure 4 is a schematic view of an assembly according to the invention, comprising a set of N identical and parallel optical fibers (only two fibers are shown in Figure 4) on thin shims 40, 42, 44 (N> 2).

Figure 4 is comparable to Figure 2 (the underside 28 of the cover 8 is planar). The positioning of the optical fibers has been shown with a standard pitch (distance between the optical axes XI of adjacent fibers) P of 250 μm. In this example we used thin shims. It is then possible to solder, for example, a strip of N laser diodes (not shown) in front of this sheet of optical fibers to connect each fiber to one of the diodes of the bar. In the case of FIG. 4, the fiber support comprises N + 1 wedges among which N-2 are bevelled on two opposite sides (see wedge 42 in FIG. 4). Figure 5 is a schematic top view of an assembly according to the invention, allowing the optical alignment of an optical fiber 2 and a diode-laser 32, the cover of this assembly

(not shown) holding the fiber against its V-shaped housing formed by two bevelled shims

4 and 6. We see the attachment pads 18 and 36 of the solder microbeads 16 and 38 respectively on the shims and on the laser diode. The optical axis XI of the fiber and the optical axis X of the diode-laser are merged. The length of the shims, counted parallel to the optical axis of the fiber, is denoted L1 and is 23 mm in the example shown. The distance between the outer edge of a wedge parallel to the fiber and the axis XI of the fiber is denoted L2 and is 1.5 mm in the example shown.

Figure 6 is the section AA of Figure 5 (the fiber is not cut) with the cover 8 of the assembly.

The distance Yl between the face of the fiber and the face of the diode which are opposite one another is adjustable (before immobilization of the fiber in its V).

Figures 7 to 11B schematically illustrate a particular embodiment of a fiber support consisting of two wedges.

These two wedges can be produced by chemical etching of a silicon wafer. To do this, two solutions are possible: - either the initial plate is thicker than the depth of the V that we want to obtain and this plate is then thinned during the production of the shims

- either the initial plate is selected so that its thickness corresponds to this depth of the V

(precision of the order of 1 μm) and there is therefore no thinning to be done thereafter.

A particular embodiment corresponding to the first case is described below. A monocrystalline silicon plate 46 (FIG. 7) is used, the surface of which is parallel to the plane (100) and deposits of silicon nitride 48 and 50 are respectively made on the upper and lower faces of this plate. Then a photolithography using a layer of photosensitive resin 52 followed by etching of the upper layer 48 makes it possible to remove the silicon nitride on a strip oriented parallel to the family of planes (100). The assembly thus obtained is immersed in an etching bath consisting of hot KOH until the formation, in the plate 46, of a V-shaped groove 54 according to the planes (111) as seen in FIG. 8A and in FIG. 8B which is the enlarged section AA of FIG. 8A. The silicon nitride which served as a mask for the etching of the silicon is then removed (FIG. 9). The plate 46 is thinned by its rear face, by mechanical polishing, up to the point of the V. This step is critical because it determines the dimension D (see FIG. 1). This mechanical polishing must be stopped very precisely when the line corresponding to the bottom of the V appears. To do this it is possible to use optical detection by illuminating one of the two faces of the plate. The appearance of a strip of light on the opposite side indicates the end of the polishing. A thinning error of 1 μm in the shims causes a variation in the positioning height of the fiber by 1 μm.

A deposit, for example by cathodic sputtering of a layer of metal capable of being brazed, such as for example the TiNi bilayer, is then formed on the rear face of the plate 46. A lithography defining the hooking pads 18 of the micro - balls are then made and the metal deposit is engraved

(see Figure 10A and Figure 10B which is the section

AA enlarged in Figure 10A). This lithograph is aligned either on the point of the V on the side of the rear face of the plate 46 or on the front face of the latter with very precise double-sided alignment and exposure equipment.

(front-rear / front-panel alignment offset of less than 1 μm). Then delimits the area of the plate carrying the studs with cne ms αe cut U and W provided for this purpose.

Then a diamond saw cut allows "to separate two shims 4 and 6 from each other (see FIG. 11A and FIG. 11B which is the enlarged section AA of FIG. 11A).

FIGS. 12A to 12E schematically illustrate various stages of preparation of a substrate with a view to forming an assembly in accordance with the invention.

FIGS. 13A to 13D schematically illustrate various steps of a method assembly of an optical fiber and a component according to the invention.

We therefore seek to align optically on the flat surface S of the substrate 12 (FIG. 12A) an optical fiber 2 (FIG. 13D) and a component 32 which is for example a diode-laser.

The optical fiber is applied against the V-shaped walls of a housing defined by two shims (only one 6 of these is shown) by means, for example, of a cover 8 and fixed in this housing by an adhesive 24. The shims are fixed to the flat surface of the substrate by means of solder microbeads

16. Each microbead is fixed, on one side, to an attachment stud 14 formed on the planar surface S of the substrate 12 and, on the other side, to another attachment stud 18 which is provided with the lower surface wedges which are located opposite this surface of the substrate. Likewise, each solder microbead 38 of the laser diode 32 is fixed, on one side, to an attachment stud 34 formed on the flat surface of the substrate and, on the other side, to another attachment stud 36 formed on the surface of the laser diode which is located opposite this planar surface of the substrate. Of course, each of the pads is wettable by the solder constituting the microbeads while the environment of these pads is not.

We begin (FIG. 12A) by forming on the flat surface S of the substrate 12 the attachment studs 14 and 34 of the various microbeads which will be subsequently formed. To do this, a full wafer metal deposit can be used by sputtering of TiNiAu type followed by a photolithography step and an etching step. These attachment studs 14 and 34 can have any shape, for example a circular, hexagonal, octagonal or square shape and even rectangular. As a result, the microbeads are not necessarily spherical.

In the example shown, these studs are discs of diameter d1 for the studs 14 and discs of diameter d2 for the studs 34.

The dimensions of these studs (the diameters dl and d2 in the example shown), counted parallel to the surface of the substrate, are determined from the heights desired for the microbeads.

In addition, electrical interconnection lines can be provided, which are integrated into the substrate or located on the surface thereof, for electrically supplying (via the pads and microbeads subsequently formed) the diode-laser and possibly other components that would require it.

Advantageously, in the case where these interconnection lines (not shown) are disposed on the surface of the substrate, in order to maintain an identical height of all of the microbead attachment pads, a surface delimited from the material used for these electrical interconnections can be placed under the hooking pads which are not electrically connected.

Then (FIG. 12B), a lithography step makes it possible to define the volumes of the microbeads from the circular openings of respective diameters Dl and D2, of the resin photosensitive used, of the thickness E of solder chosen and of the dimensions dl and d2.

To do this, a layer of photosensitive resin 56 (FIG. 12B) is deposited on the flat surface S of the substrate 12 and this resin is exposed to define the circular openings of diameter Dl (corresponding to the shims) and the other circular openings of diameter D2 (corresponding to the diode-laser). Then (FIG. 12C), the fusible material (brazing) is deposited by evaporation, intended for the subsequent formation of the microbeads, through the openings of diameters Dl and D2 thus obtained until a thickness E of material is obtained. fuse in these openings and on the resin layer.

As can be seen in FIG. 12C, this gives discs 60 of brazing, of diameter D1, and discs 62 of brazing of diameter D2 respectively formed above the studs 14 and 34. Knowing the thickness E common to all the discs 60 and 62, the respective diameters dl and d2 of the studs 14 and 34 and the respective heights hl and h2 of the microbeads 16 and 38 that we want to form, we deduce the diameters Dl and D2 of the openings that we must form in the resin layer.

After having formed the brazing discs, the resin layer 56 is removed by the technique called "lift-off". The fusible material 64 deposited on this layer 56 is thus eliminated (FIG. 12D). Then (FIG. 12E), the temperature of the substrate 12 is raised above the melting temperature of the material of which the disks 60 and 62 are made, and microbeads 16 and microbeads 38, these microbeads being respectively attached to studs 14 and 34.

We then carry out the complete assembly. We first proceed (Figure 13A) to the hybridization of the shims and the laser diode. To do this, we start by roughly aligning these shims and the laser diode (error approximately equal to ± 5 μm): we properly position the shims and the laser diode so that their respective attachment pads 18 and 36 rest on the corresponding microbeads 16 and 38.

Then the temperature of the microbeads is raised above their melting temperature. It should be noted that the wedges and the laser diode can be hybridized by means of the discs 60 and 62 (FIG. 12D) on which these wedges and the laser diode are suitably positioned. Then the temperature of the discs is raised above their melting temperature.

In both cases we arrive at the structure of Figure 13B.

The solder is then allowed to cool to room temperature, hence the hybridization of the shims and of the laser diode on the substrate.

The optical fiber 2 (FIG. 13C) is then positioned in the V-shaped housing defined by the shims before or after having deposited the adhesive 24 in this housing then the cover 8 (FIG. 13D) is pressed against the optical fiber during the phase. for drying or polymerizing the glue. In the example shown, we deliberately let the glue run under the wedges to reinforce the mechanical strength of these on the substrate. This technique is known in the field of flip-chip under the name of "underfill resin technique". Of course it is possible to simultaneously position several parallel fibers and several components by using for this purpose several shims (FIG. 4) obtained simultaneously by a technique comparable to that which is illustrated in FIGS. 7 to 11B (it is then necessary to form several parallel Vs in the silicon plate).

Figure 14 is a schematic view of another assembly according to the invention. This other assembly is to be compared to that of FIG. 1 and, in these FIGS. 1 and 14, the same elements have the same references. FIG. 14 illustrates the fact that it is possible to use a support 66 (single wedge) having a V-shaped housing which does not open out at the bottom of the support (V “not traversing”) to align a fiber 2 opposite an optical component (not shown). The shim and the optical component are aligned relatively along the axis XI. The position of the core of the fiber along the axis XI is known precisely if one precisely controls the dimensions H2 of the brazing height 16, H7 between the bottom of the V and the underside of the wedge, H8 between the fiber and the bottom of the V and the diameter d of the fiber. The dimension H7 depends on both the thickness H3 of the wedge, the opening H11 of the V and the angle β of the walls of the V. The dimension H8 depends on the angle β of the walls of the V, and of the fiber diameter. So the critical dimensions for vertical alignment are H2, H3, Hll, d and the angle β. The excess thickness H4 of the cover (protuberance 10) will be necessary to press the fiber against the walls of the V if it does not exceed the surface of the wedge, that is to say if H8 + d≤H9. This non-emerging wedge 66 can be hollowed out by several Vs in parallel to position several fibers.

A particular embodiment of a wedge having a non-through V is explained below. The first steps are those which have been described with reference to FIGS. 7, 8A and 8B: the wedge can be produced by anisotropic chemical etching of a monocrystalline silicon plate whose thickness is perfectly known. A measurement error of 1 μm in thickness will cause a variation in the positioning height of the fiber by 1 μm.

On this monocrystalline silicon plate whose surface is parallel to the plane (100), a deposit of Si ₃ N> is carried out on the two faces. A lithography and then an etching allow the Si ₃ N- to be released on a strip oriented parallel to the family of planes (100) (Figures 7). The whole is immersed in an etching bath made of hot KOH until the formation of the V according to the plans (111) (FIGS. 8A and 8B). The Si ₃ N ₄ which served as an etching mask is then released.

The following steps are illustrated by FIGS. 15A, 15B (enlarged section AA of FIG. 15A), 16A and 16B (enlarged section AA of FIGS. 16A) to be compared with FIGS. 10A, 10B, 11A and 11B respectively, the same elements having the same references: a deposit (for example a spray) of a layer of brazable metal (for example a TiNi bilayer) is made on the rear face, a lithography defining the attachment points for the beads follows, then the metal is etched. The lithography is aligned on the front face of the plate to ensure precise positioning between the studs and the opening of the V, with very precise double-sided alignment and insolation equipment (alignment offset face-back by front panel ratio less than 1 μm). The area of the plate carrying the studs and the “V” is then sawn using the cutting paths U and W, hence the wedge (FIGS. 16A and 16B).

Various advantages of the present invention are indicated below. 1 °) It is not necessary to precisely align a component and a fiber parallel to the plane of the substrate. Indeed, there is a self-alignment between the shim (s) and the component put in place by the microbead technique. The error on such a self-alignment is less than 1 μm parallel to the plane x, y of FIGS. 5 and 6.

In fact, during the remelting of the fusible material constituting the microbeads, the surface tension forces of this molten material and the wettability forces of the latter on the metal attachment pads make it possible to obtain self-alignment on the substrate. The axis of each attachment stud located on the substrate merges with the axis of the corresponding stud located on a shim or on the component.

The alignment accuracy of the shim (s) and the component relative to each other, in a plane parallel to the substrate, depends only the alignment accuracy of the microbead attachment pads on the component side and of the shim or shims relative to the optical axes. This alignment is preferably obtained by pnotolithography, the resulting error can easily be made less than 0.3 μm with the equipment used in microelectronics.

The same applies to the positioning of the attachment studs on the wedges with respect to the edges of the bevels (dimension D in FIG. 1).

On the substrate side, the studs which receive the microbeads of the wedge or wedges and of the component are preferably produced simultaneously, the relative error between studs is almost zero (neglecting thermal expansions).

2 °) The accuracy of vertical alignment of the fiber on the shim (s) and of the optical or optoelectronic component is very good because it is relative, unlike the alignment accuracy of a component in front of a fiber wedged in a V of a substrate which must be absolute.

This remark is very important because, thanks to the present invention, it is no longer necessary to perfectly control the volume of the microbeads in order to vertically align the elements with one another.

A positive or negative fluctuation in the thickness of the solder deposited, relative to its nominal value, causes a fluctuation in the same direction, relative to the interconnection substrate, of all the hybrid elements. The relative movement is very weak, which guarantees the good relative alignment of the elements between them. 3) The microbeads are preferably produced simultaneously on the interconnection substrate and the thickness of the solder deposited being substantially constant, the volumes respectively desired for the microbeads are achieved by suitable diameters of the solder discs.

4 °) The realization, by deep engravings, of V-shaped trenches in an interconnection substrate is no longer necessary. This step is very critical on a substrate comprising metallic and dielectric levels necessary for the electrical connections between optoelectronic components.

5) it is no longer necessary to form alignment shims in or on an optical or optoelectronic component.

6 °) The substrate no longer needs to be made of perfectly oriented monocrystalline silicon.

7 °) The wedges can be made of silicon and beveled by etching using KOH. In this case, a large number of shims can be obtained on the same silicon wafer. However, the shims can also be formed from a material other than silicon. If the shims are defective, the resulting additional cost remains low given the limited number of steps to form the shims.

8 °) The repair of an assembly is easy because the alignment can be measured in optical operation before bonding and it is therefore possible to “unsolder” the shims and replace them. In the case of a substrate comprising an integrated housing, a production defect implies the rejection of the complete substrate. 9 °) The positioning of the fiber with respect to the component along the y axis (Figure 6) is adjustable until mechanical contact (variation of dimension Yl). This would pose a problem in the case of a V-shaped housing integrated into the substrate because the fiber would then abut against the inclined face of the end of the housing or trench.

Claims

1. An assembly comprising a substrate (12) and, on this, at least one support (4, 6; 40, 42, 44; 66) of optical fiber, at least one optical fiber (2) disposed in this support and at least one optical or optoelectronic component (32), the optical axis of the fiber and the optical axis of the component being aligned, this assembly being characterized in that the support and the component are fixed to the substrate by means of microbeads (16, 38) of fusible material, allowing the optical axis (XI) of the fiber and the optical axis (X) of the component to be parallel to each other in the same plane perpendicular to a surface of the substrate, in that the support comprises, for each fiber, a V-shaped housing having two walls inclined towards one another, the opening of the V being located on the face of the support which is not fixed to the substrate, and in that the fiber is positioned in the housing, the volume of each microbead and the housing being determined in such a way the optical axis of the fiber and the optical axis of the component are parallel to one another in the same plane parallel to the surface of the substrate.

2. The assembly of claim 1, wherein the housing passes through the entire support

(4, 6), the housing thus having two lower edges, the distance separating these two lower edges being determined, taking into account the diameter of the fiber and the points of support thereof in the housing, so that the axis fiber optic and the optical axis of the component are parallel to each other in the same plane parallel to the surface of the substrate.

3. The assembly of claim 1, wherein the housing does not pass through the support (66).

4. Method for manufacturing an assembly comprising a substrate (12) and, on this, at least one support (4, 6; 40, 42, 44; 66) of optical fiber, at least one optical fiber (2 ) disposed in this support and at least one optical or optoelectronic component (32), the optical axis of the fiber and the optical axis of the component being aligned, this process being characterized in that the fiber support is formed, this support comprising, for each fiber, a V-shaped housing having two walls inclined towards one another, the opening of the V being located on the face of the support which is not fixed to the substrate, and also comprising a plurality of first attachment studs (18), in that a plurality of first attachment studs (36) are also formed on the component, in that second attachment studs (14, 34) are formed on the substrate, these second studs being intended to be respectively associated with the first studs, in that one forms, on the first studs and / or the second pads, elements (60, 62) made of a fusible material, capable of being soldered to the first and second pads, these first and second pads being wettable by this material in the molten state while their environment does not is not, in that the support and the component are fixed on the substrate by means of the corresponding elements, these elements being brought to the molten state for this purpose and allowing the optical axis (XI) of the fiber and the optical axis (X) of the component to be parallel to each other in the same plane perpendicular to a surface of the substrate, and in that the fiber is positioned in the housing, the volume of each element and the housing being determined so that the optical axis of the fiber and the optical axis of the component are parallel to each other in the same plane parallel to the surface of the substrate.

5. Method according to claim 4, in which the housing crosses the entire support, the housing thus having two lower edges, the distance separating these two lower edges being determined, taking into account the diameter of the fiber and the support points. of the latter in the housing, so that the optical axis of the fiber and the optical axis of the component are parallel to each other in the same plane parallel to the surface of the substrate.

6. Method according to claim 4, wherein the housing does not pass through the support.

7. The method of claim 4, wherein the support (40, 42, 44) comprises a plurality of parallel copies of said V-shaped housing and in that respectively positions a plurality of optical fibers (2) in said copies.

8. Method according to any one of claims 4 to 7, wherein each fiber (2) is fixed by means of an adhesive (24) in the corresponding housing.

9. The method of claim 8, wherein the adhesive (24) extends between the support (4, 6) and the substrate (12), around the elements of fusible material disposed under the support.

10. Method according to any one of claims 4 to 9, in which a cover (8) is produced capable of covering each housing and in which each fiber (2) is pressed against the corresponding housing by means of this cover.

11. The method of claim 10, wherein the cover (8) is transparent.

12. Method according to any one of claims 10 and 11, wherein the cover comprises for each fiber, a protuberance (10) by means of which this fiber is pressed against the corresponding housing.

13. Method according to any one of claims 10 and 11, wherein the cover comprises a flat face (28) by means of which each fiber is pressed against the corresponding housing, the fiber projecting from the face of the support which n is not attached to the substrate.

14. The method of claim 5, wherein the support is formed from a plate (46) in which at least two V-shaped walls are produced by chemical and / or mechanical etching of this plate, so as to obtain the distance determined between the two lower edges of each V, the first studs (18) corresponding to the support are formed by a photolithography technique and said walls of the housing are separated from one another.

15. The method of claim 14, wherein, when the plate has a too large initial thickness, the formation of the support further comprises a step of thinning the plate performed before or after the etching of the V.

16. The method of claim 6, in which the support is formed from a plate in which at least two V-shaped walls are produced by chemical and / or mechanical etching of this plate and the first studs corresponding to the support are formed by a photolithography technique.

17. Method according to any one of claims 4 to 16, in which the first studs (34) corresponding to the component are formed by a photolithography technique.

18. Method according to any one of claims 4 to 17, in which the second studs (14, 34) are formed on the substrate (12) by a photolithography technique.

19. Method according to any one of claims 4 to 18, in which the elements (60, 62) formed on the second studs all have the same thickness.

20. The method of claim 19, in which the elements formed on the second studs are passed in the molten state, each element then taking substantially the form of a ball (16, 38), these elements are then passed into ball form in the solid state and these are passed in the molten state to assemble the component and the support with the substrate.

21. Method according to any one of claims 19 and 20, wherein the elements (60, 62) are formed simultaneously by photolithography on the second pads.