Edge-coupling interposer

Abstract

An interposer [200] for coupling a first light beam [300] to or from a photonic integrated circuit, PIC [100], comprising at least one attachment surface [201], that is configured, when the interposer is connected to the PIC, to be in contact with a respective attachment surface of the PIC, thereby defining a position of the interposer with respect to the PIC [100], and a first protrusion [202], the first protrusion [202] comprises a transmissive surface [203] and a first curved reflective surface [204], the curved reflective surface [100] being configured to focus a first light beam [300] along at least one axis of divergence of the light beam [300] or reduce the angle of divergence of the first light beam [300] along at least one axis of divergence of the first light beam [300], the protrusion being configured to extend below a top surface of the PIC [100] when the interposer is connected to the PIC, wherein the interposer [200] is configured to accept: · a first light beam [300] emitted from the PIC [100] and entering the interposer [200] through the first transmissive surface [203] of the first protrusion [202] and propagating to the first curved reflective surface [204] of the first protrusion [202], so that the first light beam [300] propagates from the first curved reflective surface [204] through the interposer and to outside the interposer [200], and / or · a first light beam [300], that propagates from outside the interposer [200] and through the interposer [100] to the first curved reflective surface [204] of the first protrusion [202], so that at least part of the first light beam [300] exits the interposer [200] through the first transmissive surface [203] of the first protrusion [202].

Description

[0001] Edge-coupling interposer

[0002] Field of the invention

[0003] The invention relates to optical interposers that are used for optically connecting photonic integrated circuits, PICs, to other optical elements such as light sources or downstream photonic circuits.

[0004] Background

[0005] Edge couplers, i.e. , the termination of a waveguide at an etched facet or a polished edge of a photonic integrated circuit (PIC), are a common interface to couple light in or out of a PIC. Edge coupling consists in injecting light into this edge coupler from another optical element, such as a glass fiber, or in receiving light emitted by this edge coupler with another optical element. It has several advantages, such as the relative simplicity and ease of fabrication of edge couplers and the low sensitivity to polarization and wavelength of this coupling method. Several techniques exist to transform the waveguide mode at the edge couplers. These consist in e.g. direct tapers in which the waveguide is widened as it approaches the termination of the edge coupler, inverse tapers in which the waveguide width is reduced, multi-tip edge couplers in which the waveguide is segmented into several close by tips, or tapers in which the height of the waveguide core layer is also modified.

[0006] However, the assembly of optical systems using edge coupling can be very difficult, leading to yield fall-out and cost. Typically, this requires routing the waveguide to the edge of the PIC and polishing said edge, so that it can be used as a mechanical attachment plane to attach the other optical element, that can for example be a fiber array unit (FAU) or a beam transformer interposed between the FAU and the PIC. Such edge polishing is very time consuming and costly. Attaching both the PIC and the other optical element on a common substrate instead and relying on their thicknesses to be sufficiently well controlled to guarantee optical alignment is usually not practical, due to the insufficient process control applied to the thickness of the semiconductor wafers out of which the PICs are manufactured. More recently, another approach for edge coupling PICs to fibers has been developed that consists in fabricating a V- groove array directly onto the PIC, in which the fibers can be placed. This, however, makes the PIC manufacturing process more complex and requires non-standard fabrication processes. It also requires the manipulation of individual cleaved fibers that cannot be polished as an array. This poses a significant obstacle to reliable mass-manufacturing. Moreover, these edge coupling approaches are not compatible with wafer-scale manufacturing, as facets have to be polished or fibers inserted into V-grooves of singulated PICs. Since fibers are attached directly to the PIC, it is not possible to implement a connector or a reworkable connection between the array of fibers and the PIC.

[0007] Another difficulty associated to edge coupling consists in the beam profile emitted or received by the edge coupler being substantially different and typically much smaller from that of a fiber, resulting in large insertion losses if the beam is not transformed in between. This can be achieved by progressively transforming the core of a standard single mode fiber (SMF) into a high-numerical-aperture fiber core, that features a higher refractive index contrast with the fiber cladding and can be made smaller. Alternatively, another optical element implementing this beam transformation can be interposed in between the FAU and the PIC. Typical interposers consist of waveguide-based devices made in a different technology than the main PIC that is to be coupled to. Such interposers can be complex to fabricate and are thus costly. Since the beam emitted by PIC edge couplers are typically much smaller than that of a standard SMF, alignment tolerances are particularly hard to meet at the interface between the edge coupler and the interposer.

[0008] A further drawback of edge coupling results from the interfaces usually being required to be at the edge of the PIC, or, in case of an integrated V-groove array, at a boundary behind which fibers are inserted and waveguides cannot be routed. This constrains the waveguide routing on the PIC and may increase the number of waveguide crossings required to route the waveguides where they are needed. Depending on the utilized PIC technology, such waveguide crossings can be large or introduce unwanted crosstalk between the crossing waveguides. Moreover, by constraining edge couplers to be at the edge of the PIC, edge couplers may restrict the number of fibers that can be attached and coupled to.

[0009] Objective of the invention

[0010] It is therefore an objective of the invention to provide an interposer that facilitates optical connections between a photonic integrated circuit, PIC, and the outside world via an edge coupler of the PIC.

[0011] This objective is achieved by an interposer according to a first independent claim, a photonic arrangement according to a second independent claim, and respective fabrication methods according to further independent claims. Further advantageous embodiments are detailed in the respective dependent claims. Disclosure of the invention

[0012] The invention provides an interposer for coupling a first light beam to or from a photonic integrated circuit, PIC. The PIC may, for example, be a light source, such as a laser, or any circuit that takes in an input light beam, performs any kind of processing on this light beam, and produces an output light beam that is the result of this processing. The term “integrated circuit” implies that multiple operations (e.g., modulation, filtering or multiplexing performed on the input light beam may be performed within one single PIC. In particular, a PIC may be manufactured using semiconductor manufacturing techniques known from the manufacturing of electronic integrated circuits.

[0013] The interposer comprises at least one attachment surface. This attachment surface is configured, when the interposer is connected to the PIC, to be in contact with a respective attachment surface of the PIC. When the two attachment surfaces are in contact, they define a position of the interposer with respect to the PIC.

[0014] The interposer further comprises a first protrusion. This first protrusion comprises a transmissive surface and a first curved reflective surface. Herein, the term “transmissive surface” means that the main purpose of this surface is to pass light across the boundary of the interposer in either direction, and back-reflections at this surface are undesired. The curved reflective surface is configured to focus a first light beam along at least one axis of divergence of the light beam or reduce the angle of divergence of the first light beam along at least one axis of divergence of the first light beam. Herein, “focusing” may, in particular, mean that the light converges back to a focal spot. But it suffices if the curved reflective surface collimates the light or merely reduces its angle of divergence, where “collimating” is to be understood as a special case of “reducing the angle of divergence”. The protrusion is configured to extend below a top surface of the PIC when the interposer is connected to the PIC.

[0015] Wherever reference is made to, e.g., a first light beam or a first curved reflective surface, this solely serves to distinguish between this first instance of the respective entity and future instances of this entity that will be presented in the following in connection with further embodiments. That is, the designation of such an instance as “first”, “second” etc. does not imply changes to physical properties of the respective entity.

[0016] By virtue of the combination of the transmissive surface and the first curved reflective surface, the interposer is configured to accept:

[0017] • a first light beam emitted from the PIC and entering the interposer through the first transmissive surface of the first protrusion and propagating to the first curved reflective surface of the first protrusion, so that the first light beam propagates from the first curved reflective surface through the interposer and to outside the interposer, and / or

[0018] • a first light beam that propagates from outside the interposer and through the interposer to the first curved reflective surface of the first protrusion, so that at least part of the first light beam exits the interposer through the first transmissive surface of the first protrusion.

[0019] This interposer provides solutions to the challenges described above. It can be attached to PICs at the wafer scale, before singulating I dicing PICs. It uses the top surface of PICs as an attachment surface, providing accurate vertical alignment to edge couplers without requiring V- grooves or surface polishing. A well-defined attachment surface of the interposer is put in mechanical contact or bonded to an upward facing surface of the PIC that is defined on the same side on which the integrated circuit is patterned. It allows to transform the beam shape between that emitted or ideally received by an edge coupler and that ideally received or emitted by a fiber. It can be fabricated in parallel by molding of a glass plate and is thus very cost effective. Since it is a separate optical element from the PIC, it is compatible with implementing a connector at the interface to a fiber array or fiber ferrule.

[0020] In a particularly advantageous embodiment, the interposer further comprises a first top-sided reflective surface that is configured to forward the first light beam between the first curved reflective surface and an outside of the interposer. In particular, this facilitates coupling the first light beam to or from an optical fiber that is external to the interposer.

[0021] In a further particularly advantageous embodiment, the interposer further comprises a top-sided transmissive surface between the first top-sided reflective surface and the outside of the interposer. The first light beam then exits or enters the interposer through the top-sided transmissive surface. In particular, this top-sided transmissive surface may, like the first transmissive surface, be configured to reduce reflections at the boundary between the interposer and the outside world. The top-sided transmissive surface may be planar, but it may also be shaped with the goal of modifying properties of the first light beam. In particular, it may be shaped outward facing. For example, it may be shaped to form a converging lens in at least one axis of divergence. Such a lens may, for example, help to focus the light beam onto, or accept it from, the facet of an attached optical fiber with a diameter matched to the fiber mode at the focal point.

[0022] In a further particularly advantageous embodiment, a reference plane bisecting the interposer exists such that:

[0023] • the first top-sided reflective surface and the top-sided transmissive surface are on a first side of the reference plane, and • the first protrusion extends on a second side of (e.g., below) the reference plane. In particular, this facilitates the routing of optical fibers to the top-sided transmissive surface. The reference plane need not be a tangible physical feature on the interposer. It suffices that, as described here, the plane can be drawn through the interposer, so features are on different sides of this plane.

[0024] In particular, the at least one attachment surface may be in the reference plane or on the second side of the reference plane. In this manner, the first protrusion of the interposer may be hidden inside a substrate that is carrying the PIC. This saves space and also reduces the sensitivity of the interposer to stray light.

[0025] In particular, the reference plane may be parallel to the top surface of the PIC and spaced at most 10 pm apart from this surface of the PIC. In this manner, an alignment of the attachment surface of the interposer against a corresponding attachment surface at or near the top surface of the PIC may at the same time serve to align the position of the reference plane.

[0026] In a further particularly advantageous embodiment, the first top-sided reflective surface is a curved surface and is configured to focus the first light beam along at least one axis of divergence of the first light beam or to reduce the angle of divergence of the first light beam along at least one axis of divergence of the first light beam. In particular, this facilitates the coupling of the first light beam to and from an optical fiber. For example, single mode fibers have a small acceptance area, so a broad beam resulting from a high divergence would lead to a rather high coupling loss.

[0027] In a further particularly advantageous embodiment, the interposer further comprises a mechanical guide rail. The mechanical guide rail does not interact with the first light beam. The mechanical guide rail is configured to extend below the top surface of the PIC when the interposer is connected to the PIC. Such mechanical guides can be fabricated as part of the interposer to facilitate its precise assembly with the PIC. These guiding structures take the form of additional protrusions, extending beyond the reference plane and forming a rail, but that may not have an optical function. The protrusion can act as both an optical element and a mechanical guide rail, but additional mechanical guide rails without an optical function may be implemented to facilitate alignment on other sides of the PIC cavity I trench. Such additional mechanical guide rails may for example constrain the positioning of the optical interposer at the other three sides of the trench or at a subset of such sides. Similar fabrication constraints hold for these mechanical guiding structures as for the protrusion. Mechanical guide rails may also be inserted to cavities etched into the PIC that are matched to the mechanical guide rails but distinct from the cavity by which the edge coupler is terminated and the protrusion inserted. Such cavities might be etched into the PIC at the same time.

[0028] In a further advantageous embodiment, the first light beam propagates between the first curved reflective surface of the first protrusion and the first top-sided reflective surface along an axis that is within an angle of 30 degrees from the surface normal of the reference plane. This facilitates the alignment between the first light beam on the one hand and any optics outside the photonic arrangement that comprises the PIC and the interposer on the other hand.

[0029] In a further advantageous embodiment, the first light beam propagates freely through the interposer between the first curved reflective surface of the first protrusion and the first top-sided reflective surface. This means that the first light beam propagates in an unguided manner in the material of the interposer in the sense that it is not guided inside a waveguide core bounded by a different waveguide cladding material. This obviates the need to define such a waveguide during fabrication of the interposer.

[0030] In a further particularly advantageous embodiment,

[0031] • a direction of a surface normal of the first curved reflective surface of the first protrusion at at least one point, and / or in an area, of the first curved reflective surface of the first protrusion on which the first light beam is incident and

[0032] • a direction of a surface normal of the first top-sided reflective surface at at least one point, and / or in an area, of the first top-sided reflective surface on which the first light beam is incident are within 20 degrees of each other. In this manner, the performance of the interposer is made more insensitive to a variation of the thickness of the material of the interposer between the first curved reflective surface and the first top-sided reflective surface. For example, if the interposer is produced by molding of glass, the thickness may depend on the force applied to used molds and the exact temperature of the glass during the molding process. Therefore, in the mass production of interposers, the repeatability of the thickness may be limited.

[0033] In a further particularly advantageous embodiment, the first curved reflective surface of the first protrusion has, at least in a region interacting with the first light beam, a constant cross section over a length exceeding the width of the first light beam at the first curved reflective surface of the first protrusion in a direction parallel to the reference plane. In particular, the first curved reflective surface may be curved in one direction only. That is, the first curved reflective surface of the first protrusion may be a converging reflector in only one first axis of divergence of the first light beam. This means that the first curved reflective surface may be extended in an invariant manner over a length exceeding possible misalignment between the interposer and the PIC when the interposer and the PIC are combined into a photonic arrangement. Thus, the alignment is less critical for the performance of the interposer, and pick-and-place accuracy, rather than active alignment, is sufficient.

[0034] In a further particularly advantageous embodiment, the first top-sided reflective surface is a converging reflector at least along a second axis of divergence of the first light beam perpendicular to the first axis of divergence of the first light beam. Alternatively or in combination to this, the top-sided transmissive surface is a converging lens at least along the second axis of divergence of the first light beam. Both features may be used to better focus the first light beam onto, or accept the first light beam from, a facet of an attached optical fiber with a diameter matched to the fiber mode at the focal point. Fibers can be first assembled into a fiber array unit, FAU, or into an injection molded plastic ferrule that is then attached to the interposer. Such an FAU or plastic ferrule may also comprise a lens picking up the first light beam, in which case the top-sided transmissive surface or the first top-sided reflective third surface may also be configured to generate a collimated beam as required by the lensed FAU or ferrule. The top-sided transmissive surface and the top-sided reflector may be molded on the opposite side of a glass building block than the protrusion.

[0035] In a further particularly advantageous embodiment, the interposer comprises a second protrusion and a second top-sided reflective surface. The second protrusion comprises a second transmissive surface and a second curved reflective surface. The second curved reflective surface is configured to focus a second light beam along at least one axis of divergence or reduce the angle of divergence of the second light beam along at least one axis of divergence of the second light beam, and the interposer is configured to accept:

[0036] • the second light beam that enters the interposer through the second transmissive surface of the second protrusion and propagates to the second curved reflective surface of the second protrusion, which redirects it to a second top-sided reflective surface, and / or

[0037] • the second light beam that propagates from the second top-sided reflective surface to the second curved reflective surface of the second protrusion, which redirects it to the second transmissive surface of the second protrusion, so that the second light beam exits the interposer through the second transmissive surface of the second protrusion.

[0038] This can serve to increase the number of optical fibers that are coupled to without increasing the dimension of the interposer in the lateral dimension, i.e. , the third dimension that is out of the plane of the cross section in which the beams propagate, or to implement polarization and / or wavelength multiplexing or demultiplexing.

[0039] In particular, to this end, the first top-sided reflective surface may be a beam combiner and / or a beam splitter, and be reflective for the first light beam and transmissive for the second light beam. The second light beam may then exit or enter the interposer through the top-sided transmissive surface.

[0040] Alternatively, the first top-sided reflective surface and the second top-sided reflective surface may be arranged at different heights above the reference plane. In this manner, the beams may be coupled to fibers in different rows of a 2D FAU, that are stacked vertically on top of each other, or to two cores of a multi-core fiber that are at different heights relative to the PIC surface. Here, height refers to the position relative to the axis of compression of a press that is used to mold the interposer, or to the position along the surface normal of the PIC I along the surface normal of the reference plane. The beams may be routed offset from each other in the lateral axis (along the third dimension), for example if the two rows of fibers are offset from each other in the lateral direction, or if the two cores of the multi-core fiber are both vertically and laterally offset from each other.

[0041] In particular, the interposer may be configured to accept:

[0042] • a second light beam that enters the interposer through the second transmissive surface of the second protrusion and exits the interposer through the second top-sided transmissive surface, and / or

[0043] • a second light beam that enters the interposer through the second top-sided transmissive surface and exits the interposer through the second transmissive surface of the second protrusion.

[0044] This facilitates the arrangement of multiple couplings to external fibers.

[0045] In a further particularly advantageous embodiment, the first top-sided reflective surface and the second top-sided reflective surface are configured such that the interposer is configured to accept:

[0046] • the first light beam that enters the interposer through the transmissive surface of the first protrusion, propagates to the first curved reflective surface of the first protrusion, which redirects it to the first top-sided reflective surface, as well as a second light beam that enters the interposer through the second transmissive surface of the second protrusion, propagates to the second curved reflective surface of the second protrusion, which redirects it to the second top-sided reflective surface, so that the first top-sided reflective surface reflects the first light beam and the second top-sided reflective surface reflects the second light beam such that the first light beam and second light beam merge after the respective reflections into a combined merged light beam, the merged light beam exits the interposer through the top-sided transmissive surface, and / or

[0047] • a merged light beam that enters the interposer through the top-sided transmissive surface, the merged light beam is reflected by the first top-sided reflective surface and the second top-sided reflective surface such that it is split into two spatially separated light beams, the first spatially separated first light beam propagates to the first curved reflective surface of the first protrusion, which redirects it to the transmissive surface of the first protrusion though which it exits the interposer, the second spatially separated second light beam propagates to the second curved reflective surface of the second protrusion, which redirects it to the second transmissive surface of the second protrusion, though which it exits the interposer.

[0048] There are emerging application fields in which it is advantageous to route the beams such that they partially overlap after reflection by the top-sided reflective surfaces in order to synthesize a merged beam with a configurable phase and intensity distribution. This is for example the case when the beams are coupled together in a multi-mode or a few-mode fiber in which spatial diversity multiplexing (SDM) is used to increase the data throughput. Different intensity and phase profiles then serve to couple to different modes of the fiber. This is also the case when the merged light beam is intended to propagate in free space for extended distances after exiting the interposer. The interposer can then be used to facilitate the implementation of a phased array, that allows steering emitted light in free space or selectively receiving light from different regions of free space. Such phased arrays are for example used in sensor applications such as light detection and ranging (LiDAR) or microscopy. In this case, the merged beams can exit the interposer through a single transmissive surface that may be a lens, for example to facilitate coupling of the combined light beam into a few-mode fiber, or a flat surface, for example in case of the implementation of a phased array.

[0049] In a phased array, light is emitted by a closed packed array of emitters or received by a closed packed array of receivers. The phase and intensity distribution with which light is emitted across the array determines the shape and direction of the emitted beam, and, if it is coupled into a few-mode or multi-mode fiber, which modes are being excited and the phase and intensity with which these modes are being excited. In the reverse direction, the direction and shape of a received beam is mapped to the phase and intensity distribution received across the array, or, if light is received from a few-mode fiber or multi-mode fiber, the index of the modes as well as their phase and intensity is mapped to the phase and intensity distribution received across the array.

[0050] In particular, the first protrusion and the second protrusion and the first top-sided reflective surface and second top-sided reflective surface may have curved footprints with centers of curvature oriented towards the top-sided transmissive surface. That is, when looking onto the interposer from the top, the protrusions and the top-sided reflective surfaces are seen to be curved (not necessarily a circle). A curved curve has a local center of curvature, corresponding to locally approximating it as a circle and identifying the center of that circle. This center of curvature is on the side of the curve on which the transmissive surface is also located.

[0051] The invention also provides a photonic arrangement comprising the interposer described above and a photonic integrated circuit, PIC. The PIC comprises a first edge coupler and a first cavity, a facet of the first cavity defines a termination of the first edge coupler, the PIC has an attachment surface on a surface of the PIC on which a photonic circuit is fabricated. The at least one attachment surface of the interposer is attached to the attachment surface of the PIC or is in mechanical contact with the attachment surface of the PIC, and the first protrusion is inserted into the first cavity. As discussed before, a main advantage of this photonic arrangement is that it may be manufactured with less critical alignment tolerances, so that pick-and-place accuracy is sufficient or the number of axes that have to be actively aligned is reduced. Moreover, by virtue of the first protrusion of the interposer being inserted into the first cavity, space is saved.

[0052] In a particularly advantageous embodiment, the photonic arrangement is configured such that:

[0053] • the first edge coupler emits the first light beam which enters the interposer via the transmissive surface of the first protrusion and exits the interposer via the top-sided transmissive surface, and / or

[0054] • the first light beam enters the interposer via the top-sided transmissive surface and exits the interposer via the transmissive surface of the first protrusion, after which it is received by the first edge coupler.

[0055] In this manner, bringing the attachment surface of the interposer in contact with the attachment surface of the PIC allows to create a good alignment of the interposer against the PIC specifically in the dimension where this is most critical, namely in the vertical direction perpendicular to the respective attachment surfaces. This direction is most critical because the waveguide of the edge coupler is usually manufactured as a thin film and thus has a very limited height.

[0056] That is, in a particularly advantageous embodiment, the mechanical contact or attachment between the at least one attachment surface of the interposer and the top surface of the PIC determines a vertical alignment between the transmissive surface of the first protrusion and the first edge coupler.

[0057] In a further particularly advantageous embodiment, a distance between the reference plane of the interposer and a top surface of the PIC is less than 10 pm.

[0058] Attachment surfaces and the protrusion comprising both the transmissive surface and the first curved reflective surface are preferentially formed by the same mold. E.g., in the case of isothermal molding of a glass plate, they are both defined on either the top or the bottom mold. This ensures that the relative positioning between the transmissive surface, the first curved reflective surface, and an attachment surface is very well controlled and very reproducible. Moreover, since the attachment surface is in contact with or attached to the top surface of the PIC, the vertical positioning of the interposer on the PIC is very well controlled. Since there are only a few thin layers with very well-controlled thicknesses - the top waveguide cladding layers - deposited onto waveguide cores during PIC manufacturing, the vertical distance between the edge couplers terminating them and the top surface of the PIC is also very well controlled. Consequently, the vertical positioning between the transmissive and curved reflective surfaces of the interposer protrusion and the PIC edge coupler, that is fabricated in the waveguide (core) layer is also very well controlled. To couple a plurality of edge couplers to a plurality of fibers, the protrusion can be extended or fabricated in an array in a single interposer, with the direction of the extension I of the array being out of the plane of the cross-section shown in Figure 1 and referred to as the lateral direction. Such an array of protruding structures may share an attachment plane that determines the vertical positioning of the interposer. After attachment, this attachment plane of the interposer may be parallel to the plane of the top PIC surface. They may coincide if the attachment surfaces of the interposer are in direct mechanical contact with the PIC, or the attachment plane of the interposer might be slightly above the plane of the top PIC surface if a bonding material is in between. The thickness of the bonding material and thus the distance between the attachment plane of the interposer and the plane of the top PIC surface is typically not more than a few microns, and preferably below 10 pm. The attachment plane of the interposer may coincide with its reference plane.

[0059] In a further particularly advantageous embodiment, the interposer comprises at least two mechanical guide rails. The mechanical guide rails cross each other on the interposer. The PIC has a guiding trench matched to the mechanical guide rails. The interposer is attached onto the PIC such that a first mechanical guide rail is flush with a first side of the matched guiding trench for at least a portion of the first mechanical guide rail and such that the second mechanical guide rail is flush with a second side of the matched guiding trench for at least a portion of the second mechanical guide rail. The matched guiding trench is extended at a corner where the two mechanical guide rails cross, and wherein the matched guiding trench may correspond to the first cavity in which a protrusion is inserted or may be distinct from it.

[0060] If the transmissive surface of the protrusion or the outer surfaces of the mechanical guide rails merge directly with a horizontal surface, at the base of the protrusion or of the mechanical guide rail, a sharp cusp is formed in the shape of the interposer, which leads to strong strain and potentially breakage of the glass. In order to address this problem, it is beneficial to progressively change the angle of the interposer surface where the transmissive surface or the outer surfaces of the mechanical guide rails merge with a substantially horizontal surface at their base. This can be achieved by introducing an intermediate facet with an intermediate angle or by rounding this cusp, as represented as rounded corners for the transmissive surface and for the outer surfaces. This makes it difficult to use these features for mechanical alignment, since the onset of the transmissive surface and of the outer surfaces is then progressive where they merge with the rest of the interposer. This can be remedied by using pedestals in the interposer (irrespectively whether recesses are used in the PIC or not) that lift the base of the protrusion and of the mechanical guide rails above the surface of the PIC. Mechanical contacts can then be made between the upper edge of the trench I PIC cavity and a region of the transmissive surface or of the outer surfaces of the mechanical guide rails where these surfaces have a well-defined angle and position, or between the upper edge of the trench I PIC cavity and the rounded corners, where these rounded corners have a tangent that is substantially different from the horizontal direction. Preferably, the angle of the interposer surface where mechanical contact is made with the trench differs by more than 30 degrees from the horizontal direction. In another preferred configuration, this angle differs by more than 60 degrees from the horizontal direction.

[0061] The fabrication of the corresponding mold is facilitated if the mechanical guide rails are implemented as straight linear structures that extend across several interposers or glass building blocks before singulation I dicing of the molded glass plate into several such interposers or glass building blocks. This is because it is easier to precisely machine long trenches in the mold, rather than short segments with a well-defined termination. Since dicing is not very precise and the position of the cut lines can vary by a few tens of m, this results in interposer shapes where , in particular, at locations where mechanical guide rails intersect, for example if they are at a right angle, a cross-like structure occurs that cannot be completely removed by dicing. As a consequence, a rectangularly shaped trench I PIC cavity is no longer adequate to fit the guide rails. Rather the outline of the PIC cavity is determined on two sides by the position of the two crossing mechanical guiding structures, but where they cross the PIC cavity outline has to be extended beyond the corner formed by the crossing point of these two sides with an extension.

[0062] Thus, in a further particularly advantageous embodiment, the photonic arrangement further comprises a pedestal formed on a bottom side of the interposer facing the PIC. The at least one attachment surface is at the far end of the pedestal, so that the pedestal defines a position of the interposer with respect to the PIC.

[0063] In a further particularly advantageous embodiment, the first cavity does not intersect with a periphery of the PIC. The cavity may also be part of a dicing lane that is being used to singulate different instances of PICs from one another.

[0064] In a further particularly advantageous embodiment, the photonic arrangement further comprises a recess in the top surface of the PIC, a pedestal formed on the bottom side of the interposer facing the PIC, so that the at least one attachment surface is at the far end of the pedestal and in contact with a bottom surface of the recess.

[0065] Specifically, to further improve the precision of said vertical positioning and make it independent of the thickness variations of layers deposited on top of the waveguide core layer of the PIC, some or all of these layers, optionally also including the waveguide layer as well as layers below, may be fully or partially etched back where the attachment surfaces of the interposer are in mechanical contact with or bonded to the PIC. Such etching may be performed at least partially with wet etching to improve the selectivity of the etching processes with underlying layers and obtain a very precise etch stop. In such a case, places where the PIC is etched in such a manner forms recesses, while the attachment surfaces of the interposer are located at the far end of pedestals fabricated into the interposer that can be inserted into the recesses. In such a case, the vertical alignment is determined by the mechanical contact between the far end of the pedestals and the bottom of the recesses. The far end of the pedestals is opposite their base where they merge with the rest of the interposer. Recesses can also be etched by non-isotropic etches, such as KOH or TMAH etching of the silicon substrate of a silicon wafer. In such case, the boundaries of the recesses may be slanted and the attachment surfaces of the interposer also slanted accordingly. The reference plane of the interposer is defined such that the attachment surfaces of the interposer are on the reference plane or located on the side of the reference plane that is towards the PIC after attachment. The plane of the top PIC surface is the plane of the top surface of the topmost layer of the PIC, irrespectively whether this layer is still continuous or has been etched to form recesses and cavities.

[0066] In a further particularly advantageous embodiment, an optical fiber is attached to the interposer. The fiber is configured to receive the first light beam emitted by the first edge coupler and forwarded by the interposer, and / or wherein the fiber is configured to emit the first light beam that is forwarded by the interposer and received by the first edge coupler, such that the interposer transforms a profile of the first light beam between the first edge coupler and a boundary of the fiber. In this manner, a light beam profile at the boundary of the fiber is larger than a light beam profile at the termination of the first edge coupler.

[0067] The attachment may be permanent, reworkable, or implemented via a field-operatable connector. Most applications require fibers to be routed out of the package or optical subassembly in a direction that is substantially parallel to the surface of the PIC. This is facilitated by redirecting the direction of propagation of the light beam inside the interposer with a top-sided reflective surface. This surface may be a simple planar surface or may also be curved to achieve a refocusing effect. In that case, it may for example refocus the beam onto the facet of the attached fiber. The top-sided reflective surface may thus be a converging reflector in at least one angle of divergence. In a preferred embodiment, this top-sided reflective surface is defined by another mold than the one with which the attachment surface and the protrusion are defined. In a preferred embodiment in which glass plates are formed into arrays of interposers or glass building blocks, it is molded by the other of the top or bottom mold.

[0068] In a further particularly advantageous embodiment, the interposer of the photonic arrangement comprises a second protrusion and a second top-sided reflective surface. The second protrusion comprises a second transmissive surface and a second curved reflective surface configured to be a converging reflector along at least one axis of divergence of the light beam. The PIC comprises a second edge coupler.

[0069] The second edge coupler may be configured to emit a second light beam which enters the interposer via the second transmissive surface of the second protrusion and propagates to the second curved reflective surface of the second protrusion, which redirects it to the second topsided reflective surface.

[0070] Alternatively or in combination to this, a second light beam may be able to propagate from the second top-sided reflective surface to the second curved reflective surface of the second protrusion, which redirects it to the second transmissive surface of the second protrusion. The second light beam may exit the interposer through the second transmissive surface of the second protrusion and be received by the second edge coupler.

[0071] In either variation of this embodiment, the mechanical contact or attachment between the at least one attachment surface of the interposer and the respective attachment surface of the PIC determines a vertical alignment between the second transmissive surface of the second protrusion and the second edge coupler.

[0072] That is, the first and second protrusions each comprise a transmissive surface and a curved reflective surface. As in the previous configurations, the transmissive surface of the first protrusion redirects a light beam emitted from the first edge coupler to a first top-sided reflective surface, which is reflective for the light beam and further redirects it to a top-sided transmissive surface, for further coupling to an optical fiber or for coupling into free space, or vice-versa if a light beam is coupled from the optical fiber or from free space to the first edge coupler. The transmissive surface of the second protrusion redirects a second light beam emitted from the second edge coupler to a second top-sided reflective surface, for further coupling to an optical fiber or to free space, or vice-versa if a light beam is coupled from an optical fiber or from free space to the second edge coupler.

[0073] In particular, the term “top-sided” refers to elements that are above the reference plane after attachment of the interposer onto the PIC or that are molded by the opposite mold with which the protrusion is molded.

[0074] As discussed before, the advantages of having a second protrusion and a second top-sided reflective surface are particularly pronounced if light is to be coupled into a few-mode fiber. Therefore, in a further particularly advantageous embodiment, the photonic arrangement comprises a few-mode fiber, and the interposer is configured to:

[0075] • couple the first light beam and second light beam from the first edge coupler and the second edge coupler to the few-mode fiber, so that modes of the few-mode fiber are excited according to a phase and / or intensity of the first and second light beams emitted by the edge couplers, and / or

[0076] • split a merged light beam from the few-mode fiber into first and second spatially separated first and second light beams, and couple first and second spatially separated light beams to first and second edge couplers, so that modes of the few-mode fiber are mapped to relative phases and / or relative intensities of the first and second light beams received by the first and second edge couplers.

[0077] The effect is the similar in another particularly advantageous embodiment where the interposer is configured to:

[0078] • couple the first and second light beams from the first and second edge coupler to a free space beam, wherein the free space beam is shaped and directed according to the phase and intensity of the light beams emitted by the edge couplers, or

[0079] • receive a free space beam and split it into first and second spatially separated light beams, and couple first and second spatially separated light beams to first and second edge couplers, so that the shape and direction of the free space beam are mapped to relative phases and / or intensities of the light beams received by the edge couplers.

[0080] The selection of modes in the few-mode fiber and the direction and shape of the free-space beam is programmed by setting the phase and the intensity of the light beams by means of the PIC. The function of the interposer is to merge these beams in a seamless way. By doing so, the interposer solves an important problem that is very difficult to address with a PIC alone. In order to adequately shape the field front of the combined beam, it is very important that the beams emitted with a predetermined phase and intensity by individual couplers on the PIC merge, so as to generate a seamless field profile without excessive intensity drops in between the light beams. If this is not the case and beams are disjoint before being coupled to free space, the aggregate beam will not be steered in a single direction. Rather, several diffraction orders will be generated, each propagating in a different direction in free space. Similarly, in the case of coupling of the combined light beam into a few-mode fiber, it will then only be possible to excite groups of modes together. While it is relatively straightforward to solve this problem with a PIC in one direction, by emitting the light from a closed packed 1 D array of edge couplers, it is extremely difficult to generate a close packed 2D array of beams, since place has to be provided for the edge couplers and the waveguides routing light to them.

[0081] With the interposer, it is possible to emit light from a close packed array of edge couplers arranged next to each other along the lateral direction (out of the plane of the drawn crosssection), along a single protrusion, so that the functionality of a 1 D phased array as generated by a PIC is maintained. At the same time, several such rows of edge couplers, that are significantly spaced from each other, can be combined into a single seamless combined beam by means of the interposer. That transformation is afforded by the top-sided reflective surfaces, that, together, form a segmented reflector. The top-sided reflective surfaces are spaced close enough to each other in the vertical direction such that the reflected beams merge, but are spaced far enough from each other along the plane of the PIC to make space for waveguide routing on the latter.

[0082] In a further particularly advantageous embodiment, the photonic arrangement comprises a fiber, a Faraday rotator and a polarization filter. The Faraday rotator is configured to apply a nonreciprocal rotation of the light beam of 45°, wherein the first light beam propagates from the first edge coupler to the first reflective surface, from the first reflective surface to the Faraday rotator, from the Faraday rotator to the polarization filter, and from the polarization filter to the fiber. The first beam may propagate first from the first reflective surface to the top-sided reflective surface and from the top-sided reflective surface to the Faraday rotator.

[0083] In this manner, the functionality of the interposer can be augmented to that of an optical isolator.

[0084] That is, by implementing the top-sided reflective surface as a polarization selective reflector, the interposer presented above can also be converted into an isolator. This can be achieved by interposing a Faraday rotator and an additional polarization filter between the top-sided reflective surface and the optical fiber, such that the Faraday rotator is sandwiched between the top-sided reflective surface and the additional polarization filter. The Faraday rotator can be configured to apply a non-reci procal ±45° rotation (with the angle of rotation indicated for the light beam traveling from the interposer to the fiber) and the additional polarization filter to be transmissive for a polarization rotated by the same angle relative to the polarization that is reflected by the top-sided reflective surface. This then allows to couple a light beam with a predefined polarization from the PIC into the fiber, while preventing light incoming from the fiber to be coupled into the PIC. The Faraday rotator and the additional polarization filter can be interposed between the top-sided transmissive surface and the fiber, or, in case of an implementation of the interposer out of two glass building blocks, in between these two building blocks. A cavity may also be formed inside a glass building block in which the Faraday rotator and the additional polarization filter may be inserted.

[0085] Alternatively, the first beam may propagate from the first reflective surface to the Faraday rotator, and from the Faraday rotator to the top-sided reflector. The top-sided reflector may be polarization selective and play the role of the polarization filter. Light returning from the fiber to the PIC will then enter the edge coupler with the opposite polarization with which light is emitted from that edge coupler. Further polarization filtering may be provided on the PIC to fully extinguish the returning light.

[0086] In a further particularly advantageous embodiment, the PIC comprises a second edge coupler that emits or receives light from a second protrusion of the interposer. The first edge coupler and the second edge coupler, and / or the first protrusion and the second protrusion, are arranged in a lateral direction with respect to each other. The first and second edge coupler are terminated by facets of the first cavity and of a second cavity, respectively. There exists a straight extended boundary that connects the terminations of the first and the second edge coupler. The first and second edge coupler are located on the same side of the boundary. The position of the two edge couplers subdivide the extended boundary into three sections, the section between the two edge couplers, and the two outer sections. A waveguide of the PIC crosses the straight extended boundary two consecutive times, a first time in one of the sections and a second time in another one of the sections. Light is transmitted through the waveguide.

[0087] That is, for a pair of edge couplers, we define a boundary on the PIC surface consisting in a straight line that connects the emitting ends of these two edge couplers together and extends beyond them. The two edge couplers are arranged laterally relative to the frame of reference of the interposer and are on the same side of the boundary. In conventional edge coupling, with edge couplers emitting light beams at the edge of the PIC or when a V-groove array is integrated on the PIC, waveguides that cross this boundary from the edge coupler side (the inner part of the PIC on the side in which the edge couplers are fabricated by patterning of the waveguide core layer) to the other side (the outer part of the PIC) between the two edge couplers need to cross the boundary again in between the two edge couplers in order to get back from the outer part of the PIC to the inner part of the PIC. Waveguides cannot be routed outside of the PIC to loop around or across V-grooves. This constraint does not hold when coupling is implemented with the disclosed interposer. This is possible, since waveguides can go through the two trenches via the bridge defined in between, but return by going behind and around the trench before crossing the boundary again. This makes it topologically possible to reorder waveguide arrays without waveguide crossings.

[0088] The invention also provides a method for manufacturing the interposer described above. The interposer comprises at least a glass building block of a first type, and a plurality of glass building blocks of the first type is manufactured in parallel by molding a glass plate in between a first mold and a second mold. To this end,

[0089] • one or more structures corresponding to the glass building block of the first type are defined on each of the first mold and the second mold,

[0090] • the first mold and second mold are pressed together along a compression axis with the glass plate in between the first mold and the second mold, and

[0091] • a plurality of glass building blocks of the first type molded in this manner are separated from one another by dicing of the molded glass plate.

[0092] In this manner, the interposer can be manufactured in a cost-effective manner at a wafer scale. That is, any interposers may be manufactured at once with one single molding process. Both the relative positioning of features located on the same mold and the overlay accuracy of the two molds are well controllable. Dicing as such is also a well-established and efficient process.

[0093] In a particularly advantageous embodiment, all surfaces of the interposer are angled by at least 20 degrees relative to the compression axis. This facilitates both the molding itself and the insertion of the interposer into a trench or similar structure of the PIC.

[0094] In a further advantageous embodiment, a surface normal of the reference plane is within 30 degrees of the compression axis. In this manner, variations in the thickness of the compressed glass plate will cause only little, if any, distortions of the structures of the interposer. This improves the repeatability and the yield of the manufacturing process.

[0095] In a further advantageous embodiment, a thickness of the molded glass plate remains within 20% deviation of a thickness of the initial, unmolded glass plate when averaged inside a circle with a diameter of 10 mm. In this manner, the amount of glass to be laterally displaced during the molding is kept at a manageable level. In glass molding processes, the glass is typically heated up to a temperature close to the glass transition temperature, at which the glass becomes malleable, but maintains a high level of viscosity. At such temperatures, it is possible to locally shape the glass and transfer material. However, fabrication is greatly facilitated if substantial amounts of material do not have to be transferred laterally within the plane of the glass plate. Ideally, this means that the local thickness of the glass plate remains the same before and after molding, and that the top and bottom molds have complementary shapes with an inverse profile. This is, however, much too strong a constraint to allow the fabrication of the glass building blocks required to fulfill the targeted optical functionalities. Instead, the glass building blocks have to be designed such that the average thickness of the glass, before and after molding, does not change substantially when averaged across the largest area of the glass plate in which the viscous material can flow during the molding process. This contrasts with plastic injection molding, in which such mass transfer is much less of a problem prior to hardening of the material. In a preferred embodiment, the thickness of the molded glass plate remains within 20% of the initial, unmolded glass plate when averaged over a disk with a diameter of 10 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 10% of the initial, unmolded glass plate when averaged over a disk with a diameter of 10 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 20% of the initial, unmolded glass plate when averaged over a disk with a diameter of 6 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 10% of the initial, unmolded glass plate when averaged over a disk with a diameter of 6 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 20% of the initial, unmolded glass plate when averaged over a disk with a diameter of 4 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 10% of the initial, unmolded glass plate when averaged over a disk with a diameter of 4 mm.

[0096] In one example, the glass building block of the first type comprises the at least one attachment surface and a reference plane, the first protrusion, as well as a transmissive surface, which may in particular be a top-sided transmissive surface. The transmissive surface may be the outward facing top-sided transmissive surface of the interposer or an internal transmissive surface of the interposer where the first light beam transitions from one glass building block to the other. The first transmissive surface of the first protrusion, the first curved reflective surface of the first protrusion and the at least one attachment surface are molded by a first mold. The transmissive surface of the glass building block of the first type is molded by a second mold.

[0097] In particular, the first mold and the second mold may be arranged such that the first light beam propagates between the first curved reflective surface of the first protrusion and the first top- sided reflective surface of the interposer along an axis that is within an angle of 30 degrees from the compression axis of the glass building block of the first type. This makes the direction of propagation of the first light beam more insensitive against variations of the thickness of the glass building block of the first type. That is, if this thickness changes, the path that the first light beam travels in the glass building block will be lengthened or shortened, but the first light beam will still emerge from the interposer in substantially the same direction.

[0098] In a further particularly advantageous embodiment, the glass building block of the first type comprises a pocket devoid of glass, the pocket is formed by the same mold as the first protrusion. The first light beam does not interact with the pocket. The pocket extends to the same side of the reference plane as the top-sided transmissive surface of the glass building block of the first type. The pocket is configured to compensate for material of the glass plate required for forming structures of the glass building block of the first type in the molding step such that an averaged thickness of the molded glass plate does not deviate substantially from a thickness of the unmolded glass plate. That is, for a given set of structures to be fabricated, the pockets reduce the total amount of displacement work of glass that is measured in amount of material times the distance this material has to be displaced. This improves the repeatability of the molding process.

[0099] The same effect can be achieved in a further particularly advantageous embodiment where the glass building block of the first type comprises an additive volume of glass. The additive volume of glass is formed by the same mold as the top-sided transmissive surface of the glass building block of the first type. The first light beam does not interact with the additive volume of glass. The additive volume of glass is on the same side of the reference plane as the top-sided transmissive surface of the glass building block of the first type. The additive volume is configured to compensate for material of the glass plate required for forming optical elements of the glass building block of the first type in the molding step such that an averaged thickness of the molded glass plate does not deviate substantially from a thickness of the unmolded glass plate.

[0100] In a further particularly advantageous embodiment, the top-sided transmissive surface of the interposer is curved and configured to implement a converging lens along at least one axis of divergence of the light beam. The top-sided transmissive surface of the interposer is fabricated on a glass building block of a second type. The top-sided transmissive surface of the glass building block of the first type is a first internal surface of the interposer. The glass building block of the second type has a further transmissive surface that is a second internal surface of the interposer. The glass building blocks of the first and second type are attached together such that the first light beam propagates inside the interposer from the first internal surface to the second internal surface, or vice-versa.

[0101] It is easiest to fabricate lenses when their optical axis is parallel to the direction in which the two molds are being pressed together, i.e. , the axis of compression. This ensures the highest level of control on the lens shape and the best repeatability. The optical axis of the lens is defined as the direction with which the beam is incident on it from the glass side in the assembled interposer. In a preferred embodiment, the top-sided transmissive surface forms a lens and the optical axis of the lens is within 15° of the compression axis of the press applied to the glass building block in which the lens is fabricated. In another preferred embodiment, the optical axis of the lens is within 25° of the compression axis of the press. In a further preferred embodiment, the optical axis of the lens is within 35° of the compression axis of the press.

[0102] At the same time, the best level of control between the relative position of the edge coupler and the protrusion is obtained when the reference plane is substantially perpendicular to the axis of compression, respectively when the reference plane is substantially parallel to the initial surface of the glass plate. In a preferred embodiment, the surface normal of the reference plane is within 10° of the compression axis of the press. In another preferred embodiment, the surface normal of the reference plane is within 20° of the compression axis of the press. In a further preferred embodiment, the surface normal of the reference plane is within 30° of the compression axis of the press. In a preferred embodiment, the surface normal of the reference plane coincides or is substantially parallel with the surface normal of the PIC surface.

[0103] When fibers are routed out from the side of the package, satisfying these requirements together may, however, be incompatible with forming an interposer as a single glass building block from a glass plate. In particular, if the top-sided transmissive surface is formed as a lens, its optical axis needs to be substantially parallel to the top surface of the PIC and to the reference plane (and perpendicular to their surface normals), while the surface normal of the reference plane is substantially parallel to the surface normal of the PIC. Consequently, the surface normal of the reference plane and the optical axis of the top-sided transmissive surface cannot be both substantially parallel to the axis of compression if molded in a single glass building block.

[0104] To satisfy both requirements on the compression direction jointly, in a process that only allows a single compression axis, it may be preferable to build an interposer by bonding two glass building blocks together, one comprising the protrusion as well as attachment surfaces and the other comprising the top-sided transmissive surface. These can be fabricated together on the same molded glass plate, or fabricated on separate glass plates. They can then be assembled by using a bonding material such as UV or thermally curable epoxy or solder. The epoxy can be index matched to the glass and dispensed in the optical path to minimize reflections. It can also be dispensed only out of the optical path in order to maximize the optical power damage threshold of the interposer. In this case it may be beneficial to coat the glass building blocks with an anti-reflection coating (ARC) at the surfaces at which the light beam transits from one glass building block to the other. Since silica glass is transparent in the UV, in a preferred embodiment the epoxy can be pre-cured by UV exposure, followed by complete thermal curing in an oven. Pockets can be provided between the glass building blocks at places away from the light beam to collect excess epoxy during the assembly process.

[0105] The glass building block of the first type preferably comprises attachment surfaces and a protrusion. When light enters this building block through the first transmissive surface of the first protrusion, it exits the building block through a transmissive surface. The transmissive surface is preferably a top-sided surface of the building block of the first type. The direction in which light travels can be inverted. A light beam may, but does not have to, change its direction of propagation as it crosses the adjacent transmissive surfaces of the two building blocks. In particular, the interposer can be configured such that the beam does not change its direction of propagation as it crosses from one glass building block to the other, when the two glass building blocks are being brought closely together and they are molded out of the same material, as the index of refraction is then the same on either side of these surfaces.

[0106] Therefore, to further improve the repeatability in terms of the optical performance, in a further particularly advantageous embodiment, the glass building block of the second type is manufactured by molding a glass plate with a first mold and a second mold that are pressed together along a compression axis of the glass building block of the second type. An optical axis of the converging lens formed by the top-sided transmissive surface of the interposer is within 35 degrees of the compression axis of the glass building block of the second type.

[0107] In a further particularly advantageous embodiment, a reflective layer is selectively deposited in regions of a molded glass plate corresponding to the curved reflective surfaces of protrusions, prior to dicing the molded glass plate. In this manner, the reflective properties of the regions that are meant to be reflective are enhanced beyond what can be achieved by providing a mere interface between glass and air.

[0108] In a further particularly advantageous embodiment, an antireflective coating is applied to the same side of the molded glass plate onto which the reflective layer has been selectively deposited or onto which the reflective layer is to be selectively deposited, the antireflective coating is deposited prior to dicing the molded glass plate and before or after the selective deposition of the reflective layer. Thus, it may also be advantageous to coat the first transmissive surface with an antireflective coating (ARC). This is for example the case when the gap between the first surface and the edge coupler is not filled with index matched epoxy, so that another means has to be found to suppress reflections. A prior coating of the curved reflective surface with a reflective coating then allows depositing the ARC on the rest of the structure, on a given side of the molded glass plate, without depositing it directly onto the curved reflective surface, so that it stays reflective. This has the advantage of facilitating the deposition of the ARC without requiring selective deposition. In a preferred embodiment, a reflective coating is first selectively deposited onto the curved reflective surface, for example by means of inkjet printing of a metal precursor, after which the ARC coating is deposited over the entire side of the glass plate comprising the first transmissive surface prior to singulation I dicing. The order in which the reflective layer and the ARC are deposited can also be inverted without compromising the reflection at the curved reflective surface of the protrusion.

[0109] In a further particularly advantageous embodiment, a dielectric thin film stack is applied prior to dicing to an entire side of a molded glass plate that comprises protrusions. The dielectric thin film stack is configured to be antireflective at a first angle of incidence of the first light beam onto the transmissive surface of the first protrusion and to be reflective at a second angle of incidence of the first light beam onto the reflective surface of the first protrusion.

[0110] A particularly cost-effective solution to obtain the same function consists in coating the entire side of the glass interposer with a dielectric thin-film stack engineered to be transmissive for the first transmissive surface and reflective for the curved reflective surface. This can be achieved by taking into account the different incidence angle of the beam onto these two surfaces and their respective coatings. Thin film layers can for example be engineered to be an odd multiple of quarter wavelengths (relative to the wavelength in the material) at one angle and an even multiple of quarter wavelengths at the other angle, so that it changes from reflective to antireflective (assuming an alternating dielectric stack). Another effect that can be used is that directional coating methods will result in different layer thicknesses depending on the angle of the surface. For the first transmissive surface, the typical angle of incidence of the beam is closer to the surface normal, typically with an angle lower than 30° relative to it. For the curved reflective surface of the protrusion, a typical angle of incidence relative to the surface normal is closer to 45°.

[0111] In a further particularly advantageous embodiment, a grinding step is applied after molding the glass plate and before dicing the molded glass plate. In the grinding step, the transmissive surfaces of protrusions are shaped. It is difficult to mold protrusions whose width, defined here in the longitudinal direction as the distance between their transmissive and their first curved reflective surface, is very small, since glass has then to be pressed in a very small cavity on the mold. On the other hand, from the perspective of optical functionality, it is desirable to achieve a thin width, as this reduces the length over which the light beam propagates and broadens before being collimated by the curved reflective surface. To achieve this, the protrusion can first be molded to be wider than in the final interposer, and this width further reduced by processing of the interposer or glass building block after molding, with a grinding tool, optionally followed by polishing. The first transmissive surface is then defined by the grinding step occurring after the glass molding. This method of manufacturing is particularly advantageous when the gap between the first transmissive surface and the edge coupler is filled with index matched epoxy during assembly, as residual roughness on the first transmissive surface is then less critical to resulting insertion losses. In this case, a polishing step might not be necessary after grinding or a coarse polishing might be sufficient. When the width of the protrusions is reduced in this way, it may be advantageous to deposit a reflective layer before grinding, as it can then be deposited over the entire side of the glass plate, since it is later removed by the grinding step in the vicinity of the first transmissive surface through which a light beam needs to pass. This is advantageous since selective deposition, that is more complex and time consuming, is then not required.

[0112] Thus, in a further particularly advantageous embodiment, a reflective layer is deposited onto an entire side of the molded glass plate that comprises the transmissive surfaces of the protrusions that are being shaped by the grinding step, and the reflective layer is deposited prior to the grinding step.

[0113] The invention also provides a method for manufacturing the photonic arrangement described above. In the course of this method, the PIC is fabricated by patterning a wafer. The interposer comprises at least a first glass building block of the first type. A plurality of glass building blocks of the first type are manufactured in parallel by molding a glass plate in between a first mold and a second mold on which one or more structures corresponding to the glass building block of the first type are defined. The first mold and second mold are pressed together along a compression axis with the glass plate in between the first mold and the second mold. A plurality of glass building blocks of the first type molded in this manner are separated from one another by dicing of the molded glass plate. The glass building blocks of the first type are mounted onto the PIC.

[0114] As discussed above, the interposer is molded in a material that is preferentially glass. Glass preferentially refers to silica glass with various impurities that allow to set its coefficient of thermal expansion, its refractive index, its glass transition temperature and its adhesion to molds during the molding process, but can also refer to other amorphous materials, such as chalcogenide glasses. Generally, molding processes involve several molds that compress the material in between. The process for molding the glass is preferentially applied by compressing a glass plate that already has two surfaces with optical grade quality, i.e. , with sufficiently low roughness that has a root mean square (rms) typically below a few nanometers, between two molds, a top and a bottom mold. During the molding process, these two molds are pressed together along an axis that is typically substantially perpendicular to the surface of the glass plate, i.e., along its surface normal, and is referred to as the axis of compression of the press. Several interposers, or several glass building blocks meant to be assembled into interposers, are preferentially molded in parallel in a glass plate that can be diced after the molding and after optional additional process steps such as in-line testing, cleaning or thin-film coating. The glass molding process is preferentially isothermal glass molding, in which the molds and the glass are heated up and cooled down together, since isothermal glass molding has a higher precision than non-isothermal glass molding. A typical number of interposers or glass building blocks that can be molded in parallel in a single glass plate ranges between 30 and 500 depending on their size and the number of fibers that are coupled to. Fabricating optical interposers out of glass has the advantage that their coefficient of thermal expansion (CTE) can be made to match that of the PIC, or to be very close to it, as opposed to injection molded plastic that typically has a much larger CTE.

[0115] In a particularly advantageous embodiment, the glass building block of the first type or the interposer is attached to the PIC before the wafer is diced and an instance of the PIC is separated from other instances of the PIC. In this manner, the PIC is still at a position that is known with sufficient accuracy so that the attaching of the glass building block of the first type, or of the interposer, may be performed using pick and place tools.

[0116] In a further particularly advantageous embodiment, an etching step is applied to the wafer to define dicing lanes, and the first cavity of the PIC is etched in the same etching step as the dicing lanes. In this manner, the total number of fabrication steps can be reduced, and the overall yield of the fabrication process can be improved.

[0117] In particular, the facet of the edge couplers from which a light beam is emitted may be fabricated during PIC manufacturing by applying a sufficiently deep etch. Such etches are routinely applied to form optical quality surfaces. The requirement for edge polishing with conventional edge coupling rather arises from the fact that these etches do not go completely through the PIC and the necessity of having an attachment plane at the edge of the PIC. In the case of the interposer disclosed here, a partial etch is sufficient to introduce the protrusion and pick-up the beam from the PIC, since the reference plane along which the attachment surfaces are arranged is parallel to the top surface of the PIC.

[0118] To reduce the number of required fabrication steps, this etch may be the same as the one used for defining the dicing lanes of the wafer. The edge coupler can, but does not need to, end on an actual dicing lane. Rather, the trench made to terminate the edge coupler can be anywhere on the edges and inside of the PIC. In this case, it can be terminated by a trench etched at the same time and defined on the same mask, but through which the PIC in not diced. This trench can be truncated and does not need to reach the edges of the PIC, wherein it forms a cavity in which the protrusions can be inserted. The terms trench and cavity are used interchangeably in this disclosure to refer to the place in which the protrusion of the interposer can be inserted.

[0119] Moreover, a row of edge couplers arranged in parallel (in the lateral direction in the frame of reference of the interposer) can be terminated by a single trench common to all, or can be terminated by several trenches I PIC cavities, one for each edge coupler or one for each group of edge coupler, that are also arranged along the lateral direction. Such an array of edge couplers can still be coupled to an array of fibers by means of a single interposer, provided its protrusions are also interrupted where they do not overlay with a trench. This greatly facilitates routing of waveguides on the PIC, since the PIC area in between trenches creates a bridge through which waveguides can be routed. This allows for example to reorder the waveguides. Such reordering is not possible without waveguide crossings when using conventional edge coupler to fiber attachment methods.

[0120] In a further particularly advantageous embodiment, the glass building block of the first type comprises a pocket devoid of glass. The pocket is formed by the same mold as the first protrusion. The first light beam does not interact with the pocket. The pocket extends to the same side of the reference plane as the top-sided transmissive surface of the glass building block of the first type. The interposer is attached to the PIC such that the pocket is at least partially filled with epoxy. As discussed before, this allows to collect excess epoxy. In particular, this can accommodate variations as to how thick the final layer of epoxy between a glass building block of the first type and a glass building block of the second type is. Given a certain amount of epoxy, the final thickness of the epoxy layer is, inter alia, dependent on the amount of force applied when assembling the interposer, and / or the photonic arrangement.

[0121] In a further particularly advantageous embodiment, a recess is formed on the PIC by using at least one wet etching step that has an etch stop at a boundary of two layers of the PIC. As discussed before, with such an etch stop, the precision of the vertical positioning of the interposer may be improved, and it may be made independent of the thickness variations of layers deposited on top of the waveguide core layer of the PIC. In a further particularly advantageous embodiment, the first curved reflective surface of the first protrusion has, at least in regions interacting with the first light beam, a constant cross section over a length exceeding the width of the first light beam at the first curved reflective surface of the first protrusion in a direction parallel to the reference plane. The first curved reflective surface of the first protrusion is a converging reflector in only one axis of divergence of the first light beam. The interposer is attached to the PIC in a first step. In a second step, a fiber is actively aligned to the interposer and attached to the interposer.

[0122] In particular, curving the curved reflective surface in one direction only has an important advantage when combined with a top-sided reflective surface that is either flat or also only curved in the plane of the cross-sections. In that case, when the interposer is formed out of two glass building blocks, the first glass building block can maintain translation invariance in the lateral direction perpendicular to the plane of the cross-sections as far as its optical properties are concerned. In other words, the curved reflective surface of the protrusion and top-sided reflector can be extended in an invariant way along that axis over a length exceeding possible misalignment of the glass building block in the lateral direction when attached to the PIC. Such lateral misalignment can be later compensated by adjusting the position of the second glass building block, which is actively aligned and attached in a second step. This is much easier than guaranteeing sufficiently precise alignment between the edge couplers and the reflective surfaces of the protrusions in both the in-plane directions of the PIC when the interposer is already fully assembled and comprises a lens focusing in the lateral direction, or when the reflective surfaces of the protrusions provide collimation in both directions. Since the beam is much larger at the surface between the first and second glass building blocks, later placement of the second glass building block with sufficient accuracy in the lateral direction is much easier.

[0123] The same effect can be achieved in a further particularly advantageous embodiment where the first curved reflective surface of the first protrusion has, at least in regions interacting with the first light beam, a constant cross section over a length exceeding the width of the first light beam at the first curved reflective surface of the first protrusion in a direction parallel to the reference plane. The first curved reflective surface of the first protrusion is a converging reflector in only one axis of divergence of the light beam. The interposer comprises a glass building block of the first type and a glass building block of a second type. The first protrusion is part of the glass building block of the first type. The glass building block of the first type is attached to the PIC in a first step. In a second step, the glass building block of the second type is actively aligned to the glass building block of the first type and fixed relative to the glass building block of the first type. Description of the Figures

[0124] In the following, the invention is described using Figures without any intention to limit the scope of the invention.

[0125] The description along the Figures follows, in part, paths of light beams. Therefore, for the sake of clarity, it uses different names for entities mentioned in the claims. In particular,

[0126] • the first transmissive surface 203 is also termed refractive surface or first surface;

[0127] • the first curved reflective surface 204 is also termed second reflective surface or second surface;

[0128] • the first top-sided reflective surface 206 is also termed third reflective surface or third surface;

[0129] • the top-sided transmissive surface 207 is also termed fourth refractive surface or fourth surface;

[0130] • the second top-sided reflective surface 206B is also termed fifth reflective surface or fifth surface.

[0131] • The second top-sided transmissive surface 207B is also termed sixth refractive surface or sixth surface.

[0132] The term refractive surface refers to the fact that a transmissive surface may, but does not have to, change the direction of propagation of a light beam as it is transmitted through it. In other words, the refraction may be by a finite angle or by an angle of zero.

[0133] The Figures show:

[0134] Figure 1 : Exemplary embodiment of the interposer 200 with a first protrusion 202;

[0135] Figure 2: Exemplary insertion of attachment surfaces 201 into recesses 106;

[0136] Figure 3: Exemplary embodiments of the interposer 200 with top-sided reflective surface 206 and top-sided transmissive surface 207;

[0137] Figure 4: Exemplary embodiment of the interposer 200 made of a glass block 200A of a first type and a glass block 200B of a second type;

[0138] Figure 5: Optimization of the design of the interposer 200 for manufacturability with an additional volume 210 of glass and / or with a subtractive pocket 211 ; Figure 6: Exemplary embodiment of the interposer 200 with multiple protrusions 202, 202B and multiple top-sided reflective surfaces 206, 206B;

[0139] Figure 7: Exemplary embodiment of the interposer 200 with the multiple top-sided reflective surfaces 206, 206B and top-sided transmissive surfaces 207, 207B at different heights;

[0140] Figure 8: Exemplary generation of a superposition of modes in a few-mode fiber 400-FMF;

[0141] Figure 9: Exemplary generation of a free space beam 300-FS that can be steered and shaped;

[0142] Figure 10: Exemplary curvature of a curved reflective surface 204 in two or only one direction;

[0143] Figure 11: Exemplary generation of a field front by multiple edge couplers 103;

[0144] Figure 12: Exemplary embodiment of the photonic arrangement that facilitates reordering of waveguides;

[0145] Figure 13: Illustration of constraints of conventional edge coupling;

[0146] Figure 14: Example of progressively changing the angle of the interposer surface to ease strain on the glass and of the cavity extensions 109;

[0147] Figure 15: Exemplary embodiment of the interposer 200 with selective reflective coating;

[0148] Figure 16: Exemplary improvement of the attachment of the interposer 200 to the PIC 100;

[0149] Figure 17: Exemplary inclusion of a grinding step into the manufacturing process.

[0150] We disclose an interposer 200 enabling edge coupling to and from a PIC 100, PIC configurations that are particularly advantageous for using this interposer, and methods for fabricating the interposer and mounting the interposers onto the PIC.

[0151] This interposer 200 provides solutions to the challenges described above. It can be attached to PICs 100 at the wafer scale, before singulating I dicing PICs. It uses the top surface of PICs as an attachment surface, providing accurate vertical alignment to edge couplers 103 without requiring V-grooves or surface polishing. A well-defined attachment surface 201 of the interposer 200 is put in mechanical contact or bonded to an upward facing surface of the PIC 100, that is defined on the same side on which the integrated circuit is patterned. It allows to transform the beam shape between that emitted or ideally received by an edge coupler 103 and that ideally received or emitted by a fiber 400. It can be fabricated in parallel by molding of a glass plate and is thus very cost effective. Since it is a separate optical element from the PIC 100, it is compatible with implementing a connector at the interface to a fiber array or fiber ferrule.

[0152] For conciseness, the following disclosure describes the function of the interposer when light is emitted from one or several edge couplers 103 of the PIC 100 and coupled into a fiber 400 or an array of fibers by means of the interposer 200. It is, however, obvious to the person skilled in the art that the same arrangement can also serve to couple light from one or several fibers 400 into the PIC 100.

[0153] The interposer 200 comprises one or several attachment surfaces 201 arranged on a common reference plane 230 or on the side of a common reference plane 230 that is towards the PIC 100. The attachment surfaces 201 are in direct mechanical contact with the top surface of the PIC 100 or are attached to the top surface of the PIC via a thin and well-controlled bonding material, such as epoxy, which may be index matched and UV curable, or a thin solder layer, which may be a eutectic metal layer. The interposer further comprises a structure protruding beyond the reference plane 230 towards the PIC 100, that serves to pick up a light beam 300 from an edge coupler 103 of the PIC 100. During assembly, this protrusion 202 may be inserted into a cavity 105 etched into the surface of the PIC 100, wherein the etch step used to fabricate the cavity may also be used to define the optical interface of the PIC’s edge coupler 103, i.e., the PIC facet 104 terminating the edge coupler 103. The cavity 105 may be located inside the periphery of the PIC 100, or may be at its periphery and be part of a dicing lane. The beam 300 emitted by the edge coupler 103 first enters the interposer 200 via a first surface 203 of the protrusion 202 that is refractive, i.e., it may (but does not have to) deflect the direction of propagation of the light beam 300 as it enters the protrusion 202, but is configured to transmit the beam and to minimize reflections. In particular, the first refractive surface 203 does not modify the angle of propagation of the light if the index of refraction is the same on both sides of the surface. This may for example be the case if index matched epoxy is used to attach the interposer 200 to the PIC 100, as described in the following. The protrusion 202 further comprises a second surface 204, that is a curved reflective surface, that reduces the divergence of the beam 300 emitted by the edge coupler 103, collimates it, or refocuses it. The second reflective surface 204 is thus a converging reflector in at least one axis of divergence. A waveguide 101 forming the edge coupler 103 may be extended all the way to the termination of the edge coupler 103 formed by the PIC facet 104 or may end before the facet 104. In either case, the beam emitted by the edge coupler 103 exits the PIC 100 at the facet 104, which is thus referred to as the termination of the edge coupler 103. The interposer is molded in a material that is preferentially glass. Glass preferentially refers to silica glass with various impurities that allow to set its coefficient of thermal expansion, its refractive index, its glass transition temperature and its adhesion to molds during the molding process, but can also refer to other amorphous materials, such as chalcogenide glasses. Generally, molding processes involve several molds that compress the material in between. The process for molding the glass is preferentially applied by compressing a glass plate that already has two surfaces with optical grade quality, i.e. , with sufficiently low roughness that has a root mean square (rms) typically below a few nanometers, between two molds, a top and a bottom mold. During the molding process, these two molds are pressed together along an axis that is typically substantially perpendicular to the surface of the glass plate, i.e., along its surface normal, and is referred to as the axis of compression 220 of the press. Several interposers 200, or several glass building blocks 200A, 200B meant to be assembled into interposers, are preferentially molded in parallel in a glass plate that can be diced after the molding and after optional additional process steps such as in-line testing, cleaning or thin-film coating. The glass molding process is preferentially isothermal glass molding, in which the molds and the glass are heated up and cooled down together, since isothermal glass molding has a higher precision than non-isothermal glass molding. A typical number of interposers 200 or glass building blocks 200A, 200B that can be molded in parallel in a single glass plate ranges between 30 and 500 depending on their size and the number of fibers that are coupled to. Fabricating optical interposers 200 out of glass has the advantage that their coefficient of thermal expansion (CTE) can be made to match that of the PIC 100, or to be very close to it, as opposed to injection molded plastic that typically has a much larger CTE.

[0154] Attachment surfaces 201 and the protrusion 202 comprising both the first refractive surface 203 and the second reflective surface 204 are preferentially formed by the same mold. E.g., in the case of isothermal molding of a glass plate, they are both defined on either the top or the bottom mold. This ensures that the relative positioning between the first refractive surface 203, the second reflective surface 204, and an attachment surface 201 is very well controlled and very reproducible. Moreover, since the attachment surface 201 is in contact with or attached to the top surface of the PIC, the vertical positioning of the interposer on the PIC is very well controlled. Since there are only a few thin layers with very well-controlled thicknesses - the top waveguide cladding layers 102 - deposited onto waveguide cores 101 during PIC manufacturing, the vertical distance between the edge couplers 103 terminating them and the top surface of the PIC 100 is also very well controlled. Consequently, the vertical positioning between the first and second surfaces 203, 204 of the interposer protrusion 202 and the PIC edge coupler 103, that is fabricated in the waveguide (core) layer 101 , is also very well controlled. To couple a plurality of edge couplers 103 to a plurality of fibers 400, the protrusion 202 can be extended or fabricated in an array in a single interposer 200, with the direction of the extension I of the array being out of the plane of the cross-section shown in Figure 1 and referred to as the lateral direction. Such an array of protruding structures share an reference plane 230 that determines the vertical positioning of the interposer 200. After attachment, the reference plane 230 of the interposer 200 is parallel to the plane 130 of the top PIC surface. They may coincide if the attachment surfaces 201 of the interposer 200 are in direct mechanical contact with the PIC 100, or the reference plane 230 of the interposer 200 might be a slightly above the plane of the top PIC surface 130 if a bonding material is in between. The thickness of the bonding material and thus the distance between the reference plane 230 of the interposer 200 and the plane of the top PIC surface 130 is typically not more than a few microns, a preferably below 10 pm.

[0155] This is illustrated in Figure 1.

[0156] To further improve the precision of this vertical positioning and make it independent of the thickness variations of layers 102 deposited on top of the waveguide core layer 101, some or all of these layers, optionally also including the waveguide layer as well as layers below, may be fully or partially etched back where the attachment surfaces 201 of the interposer 200 are in mechanical contact with or bonded to the PIC 100. Such etching may be performed at least partially with wet etching to improve the selectivity of the etching processes with underlying layers and obtain a very precise etch stop. In such a case, places where the PIC 100 is etched in such a manner forms recesses 106, while the attachment surfaces 201 of the interposer 200 are located on top of pedestals 205 fabricated into the interposer 200, that can be inserted into the recesses 106. In such a case, the vertical alignment is determined by the mechanical contact between the top of the pedestals and the bottom of the recesses. The top of the pedestals 205 refers to their far end, opposite their base where they merge with the rest of the interposer 200. Recesses 106 can also be etched by non-isotropic etches, such as KOH or TMAH etching of the silicon substrate of a silicon wafer. In such case, the boundaries of the recesses may be slanted and the attachment surfaces 201 of the interposer 200 also slanted accordingly. The reference plane 230 of the interposer 200 is defined such that the attachment surfaces 201 of the interposer 200 are on the reference plane 230 or located on the side of the reference plane 230 that is towards the PIC 100 after attachment. The plane of the top PIC surface 130 is the plane of the top surface of the topmost layer of the PIC, irrespectively whether this layer is still continuous or has been etched to form recesses 106 and cavities 105.

[0157] This is illustrated in Figure 2, in which panel (a) represents a configuration without pedestals 205 and recesses 106 and in which panel (b) represents an exemplary configuration in which the recesses stop on top of the waveguide core layer 101. Exemplary materials I layers out of which the waveguide core layer 101 may be made are the silicon device layer of silicon-on-insulator (SOI) wafers or silicon nitride (SiN) layers deposited with plasma enhanced chemical vapor deposition (PECVD) or low-pressure chemical vapor deposition (LPCVD). An exemplary material out of which the top waveguide cladding layers 102 may be formed is silicon dioxide (SiO2).

[0158] In a preferred embodiment, a fiber 400 or an array of fibers is attached to the interposer 200. This attachment may be permanent, reworkable, or implemented via a field-operatable connector. Most applications require fibers to be routed out of the package or optical subassembly in a direction that is substantially parallel to the surface of the PIC. This is facilitated by redirecting the direction of propagation of the light beam 300 inside the interposer 200 with a third surface 206, that is also reflective. This surface may be a simple planar surface or may also be curved to achieve a refocusing effect. In that case, it may for example refocus the beam 300 onto the facet of the attached fiber 400. The third surface 206 may thus be a converging reflector in at least one angle of divergence. In a preferred embodiment, this third surface 206 is defined by another mold than the one with which the attachment surface 201 and the protrusion 202 are defined. In a preferred embodiment in which glass plates are formed into arrays of interposers or glass building blocks, it is molded by the other of the top or bottom mold. Referring to the orientation in which the figures are drawn, the third reflective surface 206 is referred to as a top-sided surface and as the first top-sided reflector of the interposer 200.

[0159] As described in the following, it may be an internal surface of the interposer 200 and the terminology top-sided surface is not meant to restrict it to be an outward facing surface of the interposer 100. It is, however, located on the side of the reference plane 230 of the interposer 200 that is above the PIC, i.e., on the top side of the reference plane. On the other hand, the protrusion 202 extends towards the other, bottom-side of the reference plane 230 and the first refractive and second reflective surfaces 203, 204 of the protrusion 202 are at least partially below the reference plane 230.

[0160] Fabrication with two molds leads to difficulties in achieving the required control and repeatability on the interposer dimensions. While the relative positioning of features located on the same mold is very well controlled, the relative positioning between features defined on different molds is much harder to control. In particular, when a glass plate is molded by a top and a bottom mold, the overlay accuracy of the two molds relative to each other is still adequate, but the repeatability of the distance between the two molds can be above ~10 m, as it is determined by the force applied to them and the exact temperature of the material during the molding process. To remedy this difficulty and make the performance of the interposer insensitive to the resulting interposer thickness variation, it is advantageous for the light beam 300 to propagate between the second and third surfaces 204, 206 in a direction that is substantially parallel to the axis of compression 220 along which the two molds are being pressed together. This ensures that the light beam 300 impinges on the third surface 206 at the same point, irrespective of this process variation, and is further deflected to further top-sided optical element without being impacted by this process variation. In a preferred embodiment, the directing in which the light beam 300 propagates between the second and the third surface 204, 206 is within 10° of this compression axis 220 or of the surface normal 209 of the attachment surface 201. In another preferred embodiment, the directing in which the light beam propagates between the second and the third surface 204, 206 is within 20° of this compression axis 220 or of the surface normal 209 of the attachment surface 201. In a further preferred embodiment, the directing in which the light beam propagates between the second and the third surface 204, 206 is within 30° of this compression axis 220 or of the surface normal 209 of the attachment surface 201.

[0161] In a preferred embodiment, the light beam propagates between the second and third surface 204, 206 of the interposer 200 as an unguided light beam 300, that is freely propagating in the material of the interposer. In another preferred embodiment, the light beam 300 is an unguided light beam inside the interposer 200, in the sense that it is not guided inside a waveguide core bounded by a different waveguide cladding material. In a preferred embodiment, this beam has a diameter, measured between points at which its intensity drops to 1 / e2of its maximum or calculated as four times the square root of the second moment of the intensity distribution (D4cr method), that is above 30 fim in between the second and the third surface 204, 206. In another preferred embodiment, this beam has a diameter that is above 60 tm in between the second and the third surface 204, 206.

[0162] The light beam 300 is coupled out of the interposer 200 via a fourth surface 207, that is also refractive and configured to induce little reflections. For example, the light beam 300 is coupled from the interposer 200 to a fiber 400 via the fourth surface 207. In the simplest case, this refractive surface 207 can be planar in the region intersected by the beam. It may also be shaped, for example to form a convex lens that is a converging lens in at least one axis of divergence. In a preferred embodiment, this lens helps to focus the light beam 300 onto the facet of the attached fiber 400 with a diameter matched to the fiber mode at the focal point. Fibers 400 can be first assembled into an FAU or into an injection molded plastic ferrule, that is then attached to the interposer 200. Such an FAU or plastic ferrule may also comprise a lens picking up the light beam 300, in which case the refractive fourth surface 207 or the reflective third surface 206 may also be configured to generate a collimated beam as required by the lensed FAU or ferrule. The fourth refractive surface 207 is also preferably a top-sided optical element. The fourth refractive surface 207 serves as an interface for the beam 300 to enter or to exit the interposer 200. As such, it is an outward facing refractive surface of the interposer 200 and is also referred to as the first outward facing refractive surface of the interposer 200. The light beam 300 may, but does not have to, change its direction of propagation as it crosses the fourth refractive surface 207. In particular, the beam 300 does not change its direction of propagation if the index of refraction is the same on both sides of the surface. This is for example the case if a glass fiber is brought in direct contact with the fourth refractive surface 207 and if the index of refraction of the interposer 200 is matched to that of the fiber 400.

[0163] The second reflective surface 204 of the protrusion 202, together with the third reflective surface 206, if it is curved and a converging reflector, or together with the fourth refractive surface 207, if it is curved and a converging lens, can serve to resize the profile of the light beam 300 along both axes of divergence between the termination of the edge coupler 103 and the boundary I facet of the fiber 400 at which the light beam enters or exits the fiber. This resizing can be implemented with independent scaling factors along the two axes of divergence. Typically, the beam dimensions are larger at the fiber boundary than at the edge coupler termination. A typical mode field diameter (MDF) at the edge coupler 103 is between 2 and 5 pm. The MFD of the fiber is typically between 8 and 12 pm. At the edge coupler, the MFD is typically larger in the axis parallel to the plane of the PIC than in the axis normal to the plane of the PIC. Typically, the beam is expanded between a factor 2 to 4 in between the edge coupler 103 and the fiber 400. The MFD is defined as the distance between the points where the intensity of the light drops to 1 / e2of its maximum or as four times the square root of the second moment of the field intensity distribution.

[0164] Configurations of the third and fourth surfaces 206, 207 are illustrated in Figure 3. In panel (a), the reflective third surface 206 has a focusing effect. In panel (b), the refractive fourth surface 207 has a focusing effect. In panel (c), both the third and the fourth surfaces are flat in the region of the beam 300, which may be a preferred configuration when used with a lensed FAU or ferrule.

[0165] It is easiest to fabricate lenses when their optical axis is parallel to the direction in which the two molds are being pressed together, i.e. , the axis of compression 220. This ensures the highest level of control on the lens shape and the best repeatability. The optical axis 208 of the lens is defined as the direction with which the beam is incident on it from the glass side in the assembled interposer. In a preferred embodiment, the refractive fourth surface 207 forms a lens and the optical axis 208 of the lens is within 15° of the compression axis 220 of the press applied to the glass building block in which the lens is fabricated. In another preferred embodiment, the optical axis 208 of the lens 207 is within 25° of the compression axis 220 of the press. In a further preferred embodiment, the optical axis 208 of the lens 207 is within 35° of the compression axis 220 of the press.

[0166] At the same time, the best level of control between the relative position of the edge coupler 103 and the protrusion 202 is obtained when the reference plane 230 is substantially perpendicular to the axis of compression 220, respectively when the reference plane 230 is substantially parallel to the initial surface of the glass plate. In a preferred embodiment, the surface normal 209 of the reference plane 230 is within 10° of the compression axis 220 of the press. In another preferred embodiment, the surface normal 209 of the reference plane 230 is within 20° of the compression axis 220 of the press. In a further preferred embodiment, the surface normal 209 of the reference plane 230 is within 30° of the compression axis 220 of the press. In a preferred embodiment, the surface normal 209 of the reference plane 230 coincides or is substantially parallel with the surface normal 108 of the PIC surface.

[0167] When fibers are routed out from the side of the package, satisfying these requirements together may, however, be incompatible with forming an interposer 200 as a single glass building block from a glass plate. In particular, if the fourth refractive surface 207 is formed as a lens, its optical axis needs to be substantially parallel to the top surface of the PIC 100 and to the reference plane 230 (and perpendicular to their surface normals), while the surface normal 209 of the reference plane 230 is substantially parallel to the surface normal of the PIC 100. Consequently, the surface normal 209 of the reference plane 230 and the optical axis 208 of the fourth surface 207 cannot be both substantially parallel to the axis of compression 220 if molded in a single glass building block.

[0168] To satisfy both requirements on the compression direction jointly, in a process that only allows a single compression axis, it may be preferable to build an interposer 200 by bonding two glass building blocks 200A and 200B together, one comprising the protrusion 202 as well as attachment surfaces 201 and the other comprising the fourth surface 207. These can be fabricated together on the same molded glass plate, or fabricated on separate glass plates. They can then be assembled by using a bonding material such as UV or thermally curable epoxy or solder. The epoxy can be index matched to the glass and dispensed in the optical path to minimize reflections. It can also be dispensed only out of the optical path in order to maximize the optical power damage threshold of the interposer. In this case it may be beneficial to coat the glass building blocks 200A, 200B with an anti-reflection coating (ARC) at the surfaces 212A, 212B at which the light beam 300 transits from one glass building block to the other. Since silica glass is transparent in the UV, in a preferred embodiment the epoxy can be pre-cured by UV exposure, followed by complete thermal curing in an oven. Pockets can be provided between the glass building blocks 200A and 200B at places away from the light beam 300 to collect excess epoxy during the assembly process.

[0169] The glass building block 200A preferably comprises attachment surfaces 201 and a protrusion 202. When light enters the building block 200A through the first refractive surface 203 of the protrusion 202, it exits the building block 200A through a refractive surface 212A. The refractive surface 212A is preferably a top-sided surface of the building block 200A. The direction in which light travels can be inverted. A light beam 300 may, but does not have to, change its direction of propagation as it crosses the refractive surfaces 212A and 212B. In particular, the interposer can be configured such that the beam does not change its direction of propagation as it crosses from one glass building block to the other, when the two glass building blocks 200A and 200B are being brought closely together and they are molded out of the same material, as the index of refraction is then the same on either side of these surfaces.

[0170] This is illustrated in Figure 4.

[0171] In glass molding processes, the glass is typically heated up to a temperature close to the glass transition temperature, at which the glass becomes malleable, but maintains a high level of viscosity. At such temperatures, it is possible to locally shape the glass and transfer material. However, fabrication is greatly facilitated if substantial amounts of material do not have to be transferred laterally within the plane of the glass plate. Ideally, this means that the local thickness of the glass plate remains the same before and after molding, and that the top and bottom molds have complementary shapes with an inverse profile. This is, however, much too strong a constraint to allow the fabrication of the glass building blocks 200A, 200B required to fulfill the targeted optical functionalities. Instead, the glass building blocks have to be designed such that the average thickness of the glass, before and after molding, does not change substantially when averaged across the largest area of the glass plate in which the viscous material can flow during the molding process. This contrasts with plastic injection molding, in which such mass transfer is much less of a problem prior to hardening of the material. In a preferred embodiment, the thickness of the molded glass plate remains within 20% of the initial, unmolded glass plate when averaged over a disk with a diameter of 10 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 10% of the initial, unmolded glass plate when averaged over a disk with a diameter of 10 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 20% of the initial, unmolded glass plate when averaged over a disk with a diameter of 6 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 10% of the initial, unmolded glass plate when averaged over a disk with a diameter of 6 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 20% of the initial, unmolded glass plate when averaged over a disk with a diameter of 4 mm. In another preferred embodiment, the thickness of the molded glass plate remains within 10% of the initial, unmolded glass plate when averaged over a disk with a diameter of 4 mm.

[0172] As described above, in a preferred embodiment the light beam 300 propagates from the protrusion 202 to the third surface 206 along the surface normal of the glass plate, which means that these two structures face each other on the two sides of the molded glass plate. From an optical perspective, there are two possible angles for the reflective surface 206 to deflect the light beam 300 such that it propagates in a direction parallel to the surface of the PIC, within the cross-sections represented in Figures 1-9: The third reflective surface 206 can be oriented to have substantially the same angle as the second reflective surface 204 at the center positions of the light beam 300, as drawn in Figures 1-9, or it can be configured to have substantially the opposite angle. Since the first configuration reduces local variations in the molded glass plate thickness, it is a preferred configuration the improves manufacturability. In a preferred configuration, the angle of the second reflective surface 204 and the angle of the third reflective surface 206 are within 20° of each other as determined at the positions at which the center axis of the light beam 300 intersects with the respective surfaces. The center axis of the light beam is defined as the axis where the beam has its maximum intensity, as determined by fitting it with a Gaussian beam approximation. In another preferred configuration, the angle of the second reflective surface 204 and the angle of the third reflective surface 206 are within 20° of each other at some point of the second reflective surface 204 and some point of the third reflective surface 206 on which a substantial light intensity is incident. The angle between the two surfaces is determined by the angle between their surface normal at these points.

[0173] Figure 5 illustrates a design for glass building block 200A optimized for manufacturability. An additional volume of glass 210 has been added to the top side of the glass building block 200A, defined as the side facing away from the PIC 100 after assembly. This additional volume of glass has no optical functionality, but equalizes the thickness of the molded glass plate across the building block’s cross section. The reflective third surface 206 and the surface 212A through which the light beam 300 exits the glass building block 200A result in a local excess thickness of the molded glass plate unless a pocket 211 is also defined on the bottom surface of the glass building block. The surface 212A may be the fourth refractive surface 207 in an interposer 200 formed from a single glass building block, but may also be distinct from it in an interposer formed from several glass building blocks as shown in Figure 4. The pocket 211 is a region without glass above the reference plane 230 that has no optical function, but improves manufacturability by equalizing the local glass thickness. It is fabricated on the side of the glass building block 200A that faces the PIC 100 after assembly. The pocket 211 can also serve to collect excess epoxy or other bonding material when the interposer 200 is attached to the PIC 100. The pocket 211 is between the interposer 200 and the PIC 100. In a preferred embodiment, the pocket 211 improves the manufacturability of a glass building block 200A molded in a glass plate in that removing the pocket 211 from the design of the glass building block would increase the deviation of the averaged molded glass plate thickness from the unmolded glass plate thickness by at least 5%. The average is taken over a disk with a diameter of 2 mm.

[0174] The interposer can be adapted to have two or more protrusions offset from each other in the plane of the cross-sections drawn in Figures 1-5. This can serve to increase the number of optical fibers that are coupled to without increasing the dimension of the interposer in the lateral dimension, i.e., the third dimension that is out of the plane of the cross section, or to implement polarization and / or wavelength multiplexing or demultiplexing. Figures 6 and 7 illustrate these two configurations with two protrusions 202, 202B. They are arranged along the longitudinal direction of the interposer, which is defined as the direction in which light beams 300 propagate between the third and the fourth surfaces 206, 207.

[0175] The first and second protrusions 202, 202B each comprise a first refractive surface 203, 203B and a second reflective surface 204, 204B. As in the previous configurations, the first refractive surface 203 of the first protrusion 202 redirects a light beam 300 emitted from the first edge coupler 103 to a third top-sided surface 206, which is reflective for the light beam 300 and further redirects it to a fourth refractive surface 207, for further coupling to an optical fiber 400 or for coupling into free space, or vice-versa if a light beam 300, 300-CO is coupled from the optical fiber 400 or from free space to the first edge coupler 103. The first refractive surface 203B of the second protrusion 202B redirects a second light beam 300B emitted from the second edge coupler 103B to a fifth reflective surface 206B, for further coupling to an optical fiber or to free space, or vice-versa if a light beam 300B, 300-CO is coupled from an optical fiber 400 or from free space to the second edge coupler 103B.

[0176] The fifth reflective surface 206B is also a top-sided surface and referred to as the second topsided reflective surface of the interposer 200. Top-sided refers to elements that are above the reference plane 230 after attachment of the interposer 200 onto the PIC 100.

[0177] Figure 6 illustrates an interposer that implements polarization and / or wavelength multiplexing or demultiplexing. The second light beam 300B has a polarization and / or a wavelength that is different from the first light beam 300. The fifth reflective surface 206B redirects the second light beam 300B to the third surface 206, which is configured to be reflective for the first light beam 300 and to be transmissive (refractive) for the second light beam 300B. The third surface 206 then acts as a polarization and / or wavelength combiner. Both the first and second beams 300, 300B are directed to the fourth refractive surface 207, for further (direct or indirect) coupling to a fiber 400 or for coupling into free space. In the opposite direction, an input light beam 300-CO is (directly or indirectly) coupled from a fiber 400 or from free space to the fourth refractive surface 207, from which it is directed to the third surface 206 that acts as a polarization and / or wavelength splitter. The input light beam 300-CO is split according to its wavelength and / or polarization into first and second beams 300 and 300B. The first beam 300 is directed to the second reflective surface 204 of the first protrusion 202, by which it is redirected to the first refractive surface 203 of the first protrusion 202 and to the first edge coupler 103. The second beam 300B is directed to the fifth reflective surface 206B, from which it is redirected to the second reflective surface 204B of the second protrusion 202B, which redirects it to the first refractive surface 203B of the second protrusion 202B and to the second edge coupler 103B. The waveguide cores are represented as being truncated in the cross-section of figure 6, as they may bend away in the plane of the PIC surface. For the avoidance of doubt, referring to the third surface 206 as being reflective is not meant to restrict it to be reflective for all wavelengths and polarizations. Rather, it is meant to indicate that it is reflective for the wavelength(s) and polarization(s) of the first light beam 300 emitted from or received by the first edge coupler 103 to or from the first protrusion 202.

[0178] The third surface 206 can be implemented as a polarization and / or wavelength combiner or splitter by applying a stack of thin dielectric films onto it that implements, for example, a dichroic mirror or a polarization selective reflector. A polarization selective reflector may be implemented in MacNeille configuration. In order to provide equal refractive indices on both sides of the thin film stack deposited onto the third surface 206 and to facilitate transmission of the second light beam 300B between the third surface 206 and the fifth surface 206B, an additional glass building block 200C] can be attached onto the glass building block 200A. This may result in an interposer configuration comprising three glass building blocks 200A, 200B, 200C] or in an interposer configuration comprising two glass building blocks 200A, 200C], for example if the fourth refractive surface 207 is not shaped as a lens. Additional glass building blocks may be incorporated as the need arises.

[0179] The configuration shown in Figure 6 can be extended to more than two protrusions, for example to combine or split more than two wavelengths. In this case, the fifth surface 206B can also be implemented as a wavelength and / or polarization splitter / combiner and a third (or more) stage(s) comprising a protrusion and a top-sided reflective surface may be appended to the left of the diagram.

[0180] In a preferred embodiment, the interposer 200 couples a first beam 300 with one polarization from the first edge coupler 103 to an optical fiber 400 and a second beam 300B with the other polarization from the second edge coupler 103B to the optical fiber 400. In a second preferred embodiment, the interposer 200 splits a light beam 300-CO from an optical fiber 400 according to polarization and couples one polarization to a first edge coupler 103 and the other polarization to a second edge coupler 103B. In a third preferred embodiment, the interposer 200 comprises at least four protrusions, each collecting a light beam 300 from an edge coupler 103, 103B, 103C and 103D. Each edge coupler emits a light beam with a wavelength different from that of the other three edge couplers. The interposer 200 combines these light beams and couple them to an optical fiber 400. In a fourth embodiment, the interposer comprises at least four protrusions, each coupling a light beam 300 to an edge coupler 103, 103B, 103C and 103D. The interposer 200 receives a light beam 300-CO from a fiber 400 with a least four different wavelengths, splits it according to wavelength, and couples it to the edge couplers 103, 103B, 103C and 103D according to wavelength.

[0181] By implementing the third surface 206 as a polarization selective reflector, the interposers shown in figures 1-5 can also be converted into an isolator. This can be achieved by interposing a Faraday rotator and an additional polarization filter between the third surface 206 and the optical fiber 400, such that the Faraday rotator is sandwiched between the third surface 206 and the additional polarization filter. The Faraday rotator can be configured to apply a non-reciprocal 45° rotation (with the angle of rotation indicated for the light beam 300 traveling from the interposer to the fiber) and the additional polarization filter to be transmissive for a polarization rotated 45° relative to the polarization that is reflected by the third surface 206. This then allows to couple a light beam 300 with a predefined polarization from the PIC 100 into the fiber 400, while preventing light incoming from the fiber to be coupled into the PIC. The Faraday rotator and the additional polarization filter can be interposed between the fourth refractive surface 207 and the fiber 400, or, in case of an implementation out of two glass building blocks 200A and 200B, in between these two building blocks. A cavity may also be formed inside a glass building block 200A in which the Faraday rotator and the additional polarization filter may be inserted.

[0182] Figure 7 illustrates an interposer that allows an increase of the density of fibers that are coupled to. The fifth reflective surface 206B redirects the second light beam 300B to a sixth refractive surface 207B, for further (direct or indirect) coupling to a second fiber 400B or to a second core of a multi-core fiber 400. As in previous configurations, the direction of the light beam can be inverted to propagate from the first fiber 400 to the first edge coupler 103 and from the second fiber 400B or second core of the multi-core fiber 400 to the second edge coupler 103B. Here too, the waveguide cores are represented as being truncated in the cross-section, as they may bend away in the plane of the PIC surface. In a preferred embodiment, the light beams 300 and 300B are routed substantially parallel to the plane of the PIC and to the plane of the unmolded glass plate, respectively between the third / fifth reflective surfaces 206, 206B and the fourth / sixth refractive surfaces 207, 207B. In another preferred embodiment, they are routed substantially orthogonally to the axis of compression 220 of the press, respectively between the third I fifth reflective surfaces 206, 206B and the fourth I sixth refractive surfaces 207, 207B. Consequentially, the third and fifth top-sided reflective surfaces 206, 206B are configured to be at different heights, so that the first and second light beams 300, 300B do not overlay. This also allows placing the fifth and sixth refractive surfaces 207 and 207B at different heights, so that the beams 300 and 300B may be coupled to fibers in different rows of a 2D FAU, that are stacked vertically on top of each other, or to two cores of a multi-core fiber that are at different heights relative to the PIC surface. Here, height refers to the position relative to the axis of compression of the press, or to the position along the surface normal of the PIC I along the surface normal 209 of the reference plane 230. The beams 300, 300B may be routed offset from each other in the lateral axis, for example if the two rows of fibers are offset from each other in the lateral direction, or if the two cores of the multi-core fiber are both vertically and laterally offset from each other.

[0183] The sixth refractive surface 207B also serves as an interface of the interposer 200 through which a light beam 300B enters or exits the interposer. It is thus outward facing and is also referred to as the second outward facing refractive surface of the interposer 200.

[0184] This configuration can also be extended to more than two protrusions, in which case a top-sided reflective surface 206, 206B, 206C], etc., and a top-sided refractive surface 207, 207B, 207C], etc., is associated to each of the protrusions 202, 202B, 202C], etc., and each of the protrusions may be associated to a separate row of fibers or to fiber cores at a different height in a row of multi-core fibers.

[0185] When the top-sided refractive surfaces 207, 207B, etc., are flat, they may also be on a same plane and form a single facet, i.e., the description is not meant to restrict these surfaces to be segmented (but they may be).

[0186] The apparatus shown in Figure 7 is meant to couple light beams 300 to independent rows of fibers 400, which may be single mode fibers, or to couple light beams 300 to separate fiber cores of one or several multi-core fibers. Consequently, the beams 300 and 300B are represented as being spatially separated, in particular they are routed at different heights.

[0187] There are, however, emerging application fields in which it is advantageous to route the beams 300, 300B, 300C], etc., such that they partially overlap after reflection by the top-sided reflective surfaces 206, 206B, 206C], etc., in order to synthesize a merged beam 300-M with a configurable phase and intensity distribution. This is for example the case when the beams are coupled together in a multi-mode or a few-mode fiber in which spatial diversity multiplexing (SDM) is used to increase the data throughput. Different intensity and phase profiles then serve to couple to different modes of the fiber. This is also the case when the merged light beam 300- M is intended to propagate in free space for extended distances after exiting the interposer 200. The interposer can then be used to facilitate the implementation of a phased array, that allows steering emitted light in free space or selectively receiving light from different regions of free space. Such phased arrays are for example used in sensor applications such as light detection and ranging (LiDAR) or microscopy. In this case, the merged beams can exit the interposer through a single refractive surface 207 that may be a lens, for example to facilitate coupling of the combined light beam 300-M into a few-mode fiber 400-FMF, or a flat surface, for example in case of the implementation of a phased array.

[0188] In a phased array, light is emitted by a closed packed array of emitters or received by a closed packed array of receivers. The phase and intensity distribution with which light is emitted across the array determines the shape and direction of the emitted beam 300-FS, and, if it is coupled into a few-mode 400-FMF or multi-mode fiber, which modes are being excited and the phase and intensity with which these modes are being excited. In the reverse direction, the direction and shape of a received beam 300-FS is mapped to the phase and intensity distribution received across the array, or, if light is received from a few-mode fiber 400-FMF or multi-mode fiber, the index of the modes as well as their phase and intensity is mapped to the phase and intensity distribution received across the array.

[0189] Figure 8(a) illustrates the generation of an arbitrary superposition of modes in a few-mode fiber 400-FMF by means of the interposer 200, while Figure 9 describes the generation of a free space beam 300-FS that can be steered and shaped.

[0190] The selection of modes in the few-mode fiber and the direction and shape of the free-space beam 300-FS is programmed in both cases by setting the phase and the intensity of the light beams 300, 300B, 300C], etc., by means of the PIC 100. The function of the interposer is to merge these beams in a seamless way. By doing so, the interposer 200 solves an important problem that is very difficult to address with a PIC alone. In order to adequately shape the field front of the combined beam 300-M, it is very important that the beams 300, 300B, 300C], etc., emitted with a predetermined phase and intensity by individual couplers on the PIC merge, so as to generate a seamless field profile without excessive intensity drops in between the light beams. If this is not the case and beams are disjoint before being coupled to free space, the aggregate beam will not be steered in a single direction. Rather, several diffraction orders will be generated, each propagating in a different direction in free space. Similarly, in the case of coupling of the combined light beam 300-M into a few-mode fiber 400-FMF, it will then only be possible to excite groups of modes together. While it is relatively straightforward to solve this problem with a PIC in one direction, by emitting the light from a closed packed 1D array of edge couplers 103, it is extremely difficult to generate a close packed 2D array of beams, since place has to be provided for the edge couplers 103 and the waveguides 101 routing light to them.

[0191] With the interposer 200, it is possible to emit light from a close packed array of edge couplers 103 arranged next to each other along the lateral direction (out of the plane of the drawn crosssection), along a single protrusion, so that the functionality of a 1D phased array as generated by a PIC is maintained. At the same time, several such rows of edge couplers, 103, 103B, 103C, etc., that are significantly spaced from each other, can be combined into a single seamless combined beam 300-M by means of the interposer 200. That transformation is afforded by the top-sided reflective surfaces 206, 206B, 206C, etc., that, together, form a segmented reflector. The top-sided reflective surfaces 206, 206B, 206C, etc., are spaced close enough to each other in the vertical direction such that the reflected beams merge, but are spaced far enough from each other along the plane of the PIC 100 to make space for waveguide routing on the latter.

[0192] There exists a trade-off in the size of the top-sided reflective surfaces 206, 206B, 206C, etc. If they are too small, only the central portions of the beams 300, 300B, 300C, etc., are reflected and the beams are consequently truncated. If they are too large, there will be a significant drop in intensity in between the merged beams, leading to the excitation of side lobes I higher diffraction orders I multiple fiber modes. Consequently, the width of the top-sided reflective surfaces, as measured along the horizontal axis in the plane of the cross-sections (the longitudinal direction), should be in the order of the full width at half maximum (FWHM) of the beams and may preferably range from half a FWHM to three FWHM. The center-to-center spacing between the top-sided reflective surfaces 206, 206B, 206C, etc., as also measured along the horizontal direction of the cross-section (the longitudinal direction), is, however, much larger, and typically more than three times the width of the reflective surfaces. The reflective surfaces may be flat surfaces, typically angled 45° relative to the surface normal of the PIC, but may also be focusing reflectors. The boundary of the glass building block between the top-sided reflective surfaces does not need to be a straight line, rather, the top-sided reflective surfaces might also be individually stamped into the top of the building block, as shown in Figure 8(b). This configuration is also advantageous to equalize the local thickness of the molded glass plate.

[0193] As with the previous configurations, the direction of the light beams may be inverted. In this case, by analyzing the phase and the intensity of the light beam 300 coupled to each edge coupler 103, it can be determined from which direction(s) of free space or which mode(s) of a few-mode fiber 400-FMF the light originated from. This serves for example to determine the position of an object in a LiDAR, or to demultiplex modes in an SDM communication link.

[0194] In the embodiments shown in Figures 6-9, it is preferential to mold all the protrusions 202, 202B, 202C, etc., belonging to a given interposer by the same (first) mold in a single glass building block 200A, so that their relative positioning can be better controlled and match that of the edge couplers 103, 103B, 103C, etc., on the PIC 100. Similarly, in the embodiments shown in Figures 6-9, it is preferential to mold the third and fifth surfaces 206, 206B and additional topsided reflective surfaces 206C, etc., deflecting light beams from additional protrusions 202C, etc., by the same (second) mold, in the same glass building block as the protrusions. This allows to maintain a better relative positioning between these reflective surfaces, in particular in terms of their height. The refractive surface 207 or the top-sided refractive surfaces 207, 207B, etc., may be part of a separate glass building block attached to the first to facilitate glass molding.

[0195] While figures and the description of the interposer 200 focused mainly on the cross-section to this point, it is understood that interposers 200 may couple arrays of edge couplers 103, arranged next to each other along the axis orthogonal to the cross-sections drawn in Figures 1- 9 (the lateral direction), to arrays of fibers arranged next to each other along the same axis.

[0196] While the curvature of curved surfaces are visible only in one direction in the represented crosssections, it is understood that they may be curved in the other direction as well to achieve collimation or a focusing effect in both. Flat surfaces, or surfaces curved only in the plane of the cross-section, might be shared across the array, without visible segmentation in between. Figure 10 illustrates this by showing two designs of the protrusion 202. In Figure 10(a), the second reflective surfaces 204 are curved in both directions and consequently apply a focusing effect along both axes. In Figure 10(b), the second reflective surface 204 is curved in on direction only, corresponding to the one represented in the cross-sections 1-9.

[0197] To facilitate fabrication, the second surface 204 of the protrusions 202 might be curved along a single direction, inside the plane of the represented cross-sections, applying focusing to the beam 300 in a single direction, and the curvature of the third surface 206 or of the fourth refractive surface 207 might be adapted to obtain focusing of the beam in the other direction. Implementing the curvature that is not in the plane of the cross-sections is easier on the third surface 206 than on the second surface 204, as the beam is larger there and the structures bigger. Moreover, it is straightforward to design edge couplers 103 with a divergence angle that is smaller in the plane of the PIC 100 than in the vertical direction (with directions labeled prior to the beam being deflected by the second reflective surface 204, i.e. , as the beam 300 exits the edge coupler 103). This can be accomplished by widening the edge coupler in the plane of the PIC, which is a simple design change (optionally with a segmented structure consisting in multiple, highly coupled waveguide tips). It is, however, much harder to reduce the divergence angle in the vertical direction, as it is constrained by the thickness of the waveguide core layer 101. It is essential to apply a focusing effect to the fast divergence axis of the beam as early as possible in the beam path, while the focusing I collimation along the other axis can be done at a later stage. This is exactly what is done by the curvature of the second reflective surface 204 as shown in the cross-sections 1-9, which applies a focusing effect to the beam emitted by the edge couplers 103 in the vertical direction, which is typically the fast divergence axis.

[0198] Curving the second reflective surface 204 in one direction only, according to the cross-sections 1-9 and to Figure 10(b), has another important advantage when combined with a third reflective surface 206 that is either flat or also only curved in the plane of the cross-sections. In that case, when the interposer is formed out of two glass building blocks as shown in Figure 4, the first glass building block 200A can maintain translation invariance in the lateral direction perpendicular to the plane of the cross-sections as far as its optical properties are concerned. In other words, the second and third surfaces 204, 206 can be extended in an invariant way along that axis over a length exceeding possible misalignment of the glass building block 200A in the lateral direction when attached to the PIC 100. Such lateral misalignment can be later compensated by adjusting the position of the second glass building block 200B, which is actively aligned and attached in a second step. This is much easier than guaranteeing sufficiently precise alignment between the edge couplers 103 and the second reflective surfaces 204 in both the in-plane directions of the PIC 100 when the interposer is already fully assembled and comprises a lens focusing in the lateral direction, or when the second reflective surfaces provide collimation in both directions, as shown in Figure 10(a). Since the beam is much larger at the surface 212A, 212B between the first and second glass building blocks 200A and 200B, later placement of the second glass building block 200B with sufficient accuracy in the lateral direction is much easier.

[0199] Similarly, this can also be accomplished with a pre-assembled interposer 200 or an interposer 200 made out of a single glass building block, if the fourth surface 207 also has this translation invariance in the lateral direction, i.e., is not curved or curved only in the plane of the crosssections in the vicinity of the beams. In this case, focusing on the other axis can be accomplished by a lensed FAU or fiber ferrule, wherein the FAU or fiber ferrule can be actively aligned and attached to the interposer 200 after attachment of the interposer 200 to the PIC 100 in order to compensate for its misalignment.

[0200] In a preferred embodiment, the second reflective surface 204 is configured such that the divergence of the light beam 300 emitted by the edge coupler 103 is reduced to less than 10° (half angle, measured at the 1 / e2intensity point) along its fast divergence axis (or is converted into a focusing beam). In another preferred embodiment, the second reflective surface 204 is configured such that the divergence of the light beam 300 emitted by the edge coupler 103 is reduced to less than 5° (half angle, measured at the 1 / e2intensity point) along its fast divergence axis (or is converted into a focusing beam).

[0201] In the case of the configurations illustrated by Figures 8 and 9, that serve for beam front shaping as used in SDM or phased array applications, it can be beneficial to curve the protrusions as they extend out of the plane of the cross-sections, instead of fabricating them as straight, linear structures. The top-sided reflective surfaces 206, 206B, 206C], etc., are then configured to follow that curvature. This creates a focusing effect towards the center of curvature, which may for example be the center of a fiber core facet. This has the advantage that the fourth refractive surface 207 does not need to be fabricated as a lens and that the interposer 200 can be more straightforwardly fabricated out of a single glass building block, instead of a configuration as shown in Figure 4. The center of curvature may deviate from the fiber facet, in particular if there is a different medium such as air between the fiber 400, 400- FMF and the interposer, but preferentially remains towards the direction of the fiber facet and typically on the optical axis of the fiber.

[0202] This is illustrated in Figure 11. Edge couplers 103 are exemplarily shown, but may form much denser arrays so as to generate a seamless field front. They are preferably oriented such that their optical axis, that coincides with the direction of the waveguides at the interface and with the direction in which they emit a light beam, is orthogonal to the curve followed by the protrusions 202, 202B, 202C], etc., and by top-sided reflective surfaces 206, 206B, 206C], etc. In another embodiment, the edge couplers 103 are located on the other side of the outline of the top-sided reflective surfaces 206, 206B, 206C], etc., which in that case are oriented with the opposite angle in the plane of the cross-section of Figure 8, so as to still redirect the light beams 300 to the fourth refractive surface 207.

[0203] In the following, we disclose aspects relating to the design and fabrication of PICs on which these interposers can be mounted.

[0204] The facet 104 of the edge couplers 103 from which a light beam 300 is emitted may be fabricated during PIC manufacturing by applying a sufficiently deep etch. Such etches are routinely applied to form optical quality surfaces. The requirement for edge polishing with conventional edge coupling rather arises from the fact that these etches do not go completely through the PIC 100 and the necessity of having an attachment plane at the edge of the PIC. In the case of the interposer disclosed here, a partial etch is sufficient to introduce the protrusion and pick-up the beam from the PIC, since the reference plane 230 is parallel to the top surface of the PIC 100.

[0205] To reduce the number of required fabrication steps, this etch may be the same as the one used for defining the dicing lanes of the wafer. The edge coupler 103 can, but does not need to, end on an actual dicing lane. Rather, the trench 105 made to terminate the edge coupler 103 can be anywhere on the edges and inside of the PIC 100. In this case, it can be terminated by a trench 105 etched at the same time and defined on the same mask, but through which the PIC in not diced. This trench can be truncated and does not need to reach the edges of the PIC, wherein it forms a cavity in which the protrusions 202 can be inserted. The terms trench and cavity are used interchangeably in this disclosure to refer to the place in which the protrusion 202 of the interposer 200 can be inserted.

[0206] Moreover, a row of edge couplers 103L, 103R arranged in parallel (in the lateral direction in the frame of reference of the interposer 100) can be terminated by a single trench 105 common to all, or can be terminated by several trenches I PIC cavities 105L, 105R, one for each edge coupler or one for each group of edge coupler, that are also arranged along the lateral direction. Such an array of edge couplers can still be coupled to an array of fibers 400 by means of a single interposer 200, provided its protrusions 202L, 200R are also interrupted where they do not overlay with a trench 105. This greatly facilitates routing of waveguides 101 on the PIC 100, since the PIC area in between trenches 105L, 105R creates a bridge through which waveguides can be routed. This allows for example to reorder the waveguides, as shown in Figure 12. Such reordering is not possible without waveguide crossings when using conventional edge coupler to fiber attachment methods. In Figure 12, Tx stands for Transmitter and Rx for Receiver. The labeling is an exemplary configuration, that is advantageous since the transmitters and receivers can be implemented in separate regions of the PIC after reshuffling of the waveguide ordering, while transmitting and receiving fibers can be interleaved at the optical interface, as required to e.g. split bi-directional links among two separate fiber ribbons.

[0207] For a pair of edge couplers 103, we define a boundary 107 on the PIC surface consisting in a straight line that connects the emitting ends of these two edge couplers together and extends beyond them. The two edge couplers 103L, 103R are arranged laterally relative to the frame of reference of the interposer 200 and are on the same side of the boundary 107. In conventional edge coupling, with edge couplers 103 emitting light beams at the edge of the PIC or when a V- groove array is integrated on the PIC, waveguides that cross this boundary from the edge coupler side (the inner part of the PIC on the side in which the edge couplers are fabricated by patterning of the waveguide core layer 101) to the other side (the outer part of the PIC) between the two edge couplers need to cross the boundary again in between the two edge couplers in order to get back from the outer part of the PIC to the inner part of the PIC. This is illustrated in Figure 13(a) for edge coupling at the outer boundary of the PIC 100 and in Figure 13(b) for edge coupling with integrated V-grove arrays. Waveguides cannot be routed outside of the PIC to loop around (Figure 13(a)) or across V-grooves (Figure 13(b)). As shown in Figure 12, this constraint does not hold when coupling is implemented with the disclosed interposer 200. This is possible, since waveguides can go through the two trenches 105L, 105R via the bridge defined in between, but return by going behind and around the trench before crossing the boundary 107 again. This makes it topologically possible to reorder waveguide arrays without waveguide crossings.

[0208] In the following, we disclose aspects relating to the mounting of the interposers on PICs and aspects relating to interposer fabrication addressing requirements arising from this.

[0209] In a preferred embodiment, the first refractive surface 203 is fabricated with an angle 213 relative to the surface normal 209 of the reference plane 2301 the axis of compression 220 of the press. This facilitates fabrication, as it is very difficult to fabricate surfaces that are perfectly vertical, i.e., along the compression direction 220 of the press. This is not a general requirement with other materials and molding processes, since the fabrication of vertical surfaces is for example easier with plastic injection molding.

[0210] The angle of these surfaces can, however, be used to facilitate the insertion of the protrusion 202 in a tightly dimensioned trench 105, for better alignment between the PIC 100 and the interposer 200. The slight angle from vertical then facilitates insertion of the interposer 200 or interposer building block 200A. In order to facilitate molding of the glass piece as well as insertion of the interposer into the trench 105, the angle 213 of the first refractive surface 203 is chosen to be at least 20 degrees from the compression axis of the press 220 and / or from the surface normal 209 of the reference plane 230.

[0211] Additional mechanical guides 215 can be fabricated as part of the interposer to facilitate its precise insertion into the trench 105. These guiding structures take the form of additional protrusions, extending beyond the reference plane 230 and forming a rail, but that may not have an optical function. The protrusion 202 can act as both an optical element and a mechanical guide rail, but additional mechanical guide rails without an optical function may be implemented to facilitate alignment on other sides of the PIC cavity I trench. Such additional mechanical guide rails 215 may for example constrain the positioning of the optical interposer 200 at the other three sides of the trench 105 or at a subset of such sides. Similar fabrication constraints hold for these mechanical guiding structures 215 as for the protrusion 202. In order to facilitate molding of the glass piece as well as insertion of the interposer into the trench 105, the angle 216 of the outer surfaces 217 of the mechanical guide rails 215 (the surfaces facing the etched facet 104 of the trench 105) is chosen to be at least 20 degrees from the compression axis of the press 220 and / or from the surface normal 209 of the reference plane 230. Mechanical guide rails 215 may also be inserted to cavities 105 etched into the PIC 100 that are matched to the mechanical guide rails but distinct from the cavity 105 by which the edge coupler 103 is terminated and the protrusion 202 inserted. Such cavities 105 might be etched into the PIC 100 at the same time.

[0212] If the first refractive surface 203 of the protrusion or the outer surfaces 217 of the mechanical guide rails 215 merge directly with a horizontal surface, at the base of the protrusion 202 or of the mechanical guide rail 215, a sharp cusp is formed in the shape of the interposer 200, which leads to strong strain and potentially breakage of the glass. In order to address this problem, it is beneficial to progressively change the angle of the interposer surface where the first refractive surface 203 or the outer surfaces 217 of the mechanical guide rails 215 merge with a substantially horizontal surface at their base. This can be achieved by introducing an intermediate facet with an intermediate angle or by rounding this cusp, as represented as rounded corners 214 for the first surface 203 and 218 for the outer surfaces 217 in Figure 14(a). This makes it difficult to use these features for mechanical alignment, since the onset of the first surface 203 and of the outer surfaces 217 is then progressive where they merge with the rest of the interposer 100. This can be remedied by using pedestals 205 in the interposer (irrespectively whether recesses 106 are used in the PIC or not) that lift the base of the protrusion 202 and of the mechanical guide rails 215 above the surface of the PIC. Mechanical contacts can then be made between the upper edge of the trench I PIC cavity 105 and a region of the first refractive surface 203 or the outer surfaces 217 of the mechanical guide rails 215 where these surfaces have a well-defined angle and position, or between the upper edge of the trench I PIC cavity 105 and the rounded corners 214, 218, where these rounded corners have a tangent that is substantially different from the horizontal direction. Preferably, the angle of the interposer surface where mechanical contact is made with the trench 105 differs by more than 30 degrees from the horizontal direction. In another preferred configuration, this angle differs by more than 60 degrees from the horizontal direction.

[0213] The fabrication of the corresponding mold is facilitated if the mechanical guide rails 215 are implemented as straight linear structures that extend across several interposers 200 or glass building blocks 200A before singulation I dicing of the molded glass plate into several such interposers or glass building blocks. This is because it is easier to precisely machine long trenches in the mold, rather than short segments with a well-defined termination. Since dicing is not very precise and the position of the cut lines 219 can vary by a few tens of m, this results in interposer shapes as shown in Figure 14(b). In particular, at locations where mechanical guide rails 215 intersect, for example if they are at a right angle as shown in Figure 14(b), a cross-like structure occurs that cannot be completely removed by dicing. As a consequence, a rectangularly shaped trench I PIC cavity 105 is no longer adequate. Rather the outline of the PIC cavity 105 is determined on two sides by the position of the two crossing mechanical guiding structures 215, but where they cross the PIC cavity outline has to be extended beyond the corner formed by the crossing point of these two sides with an extension 109. This is shown in Figure 14(c).

[0214] When the insertion of the interposer 200 into the PIC cavity 105 is not constrained on all sides by mechanical guide rails 215, mechanical guiding can be accomplished by pushing the structure against one or several sides until a predetermined level of resistance is met.

[0215] The attachment of the interposer onto the PIC can be done with curable epoxy 500. Since in a preferred embodiment the interposer is made out of silica glass, which is not only transparent at telecommunication wavelengths, but also at short wavelengths, the epoxy can be cured with ultra-violet radiation. A UV curing step might be followed by thermal curing to fully cure the subassembly. In this case, it is advantageous for epoxy to fill the gap between the angled first surface 203 of the protrusion 202 and the etched facet 104 of the PIC 100 terminating the edge coupler 103. This is because index matched epoxy reduces back-reflections at the optical interfaces. Moreover, filling this gap prevents accumulation of moisture while operating it in the field, which could lead to absorption and scattering of the light beam 300 and thus to reliability concerns. Suppressing reflections at the glass interface of the interposer 200 is straightforward, since its index can be matched by epoxy. At the PIC side, index matched epoxy might also reduce reflections very significantly. For example, in silicon photonics edge couplers, the mode is typically strongly delocalized from the silicon core to the surrounding oxide cladding, which is easy to match.

[0216] The second surface 204, however, has to stay reflective. If this reflection is reliant on total internal reflection (TIR), the second surface 204 needs to stay free of epoxy. This can be achieved by either dispensing the epoxy 500 locally on the first surface 203 of the protrusion 202 or on the corresponding etched surface 104 of the PIC 100, such that the second surface stays free of epoxy. It may, however, result in yield fall-out if the epoxy spills to the second surface 204. Alternatively, the second surface 204 may be selectively coated with a reflective coating 221 , for example with a high-quality metal layer or a dielectric stack, to ensure that it stays reflective even when surrounded by index matched epoxy. This has the advantage of facilitating dispensing of the epoxy. This selective reflective coating can be performed particularly effectively with inkjet printing of metal ink that can be cured into a dense metal layer. Selective coating with a metal layer or with a reflective dielectric stack can also be performed with shadow masking.

[0217] This configuration is illustrated in Figure 15(a).

[0218] It may also be advantageous to coat the first surface 203 with an antireflective coating (ARC) 222. This is for example the case when the gap between the first surface 203 and the edge coupler 103 is not filled with index matched epoxy 500, so that another means has to be found to suppress reflections. A prior coating of the second surface with a reflective coating 221 then allows depositing the ARC on the rest of the structure, on a given side of the molded glass plate, without depositing it directly onto the second surface, so that it stays reflective. This has the advantage of facilitating the deposition of the ARC without requiring selective deposition. In a preferred embodiment, a reflective coating 221 is first selectively deposited onto the second reflective surface 204, for example by means of ink jet printing of a metal precursor, after which the ARC coating is deposited over the entire side of the glass plate comprising the first refractive surface 203 prior to singulation I dicing. The order in which the reflective layer 221 and the ARC 222 are deposited can also be inverted without compromising the reflection at the second reflective surface 204 of the protrusion 202.

[0219] This configuration is illustrated in Figure 15(b).

[0220] A particularly cost-effective solution to obtain the same function consists in coating the entire side of the glass interposer with a dielectric thin-film stack 223 engineered to be transmissive for the first surface 203 and reflective for the second surface 204. This can be achieved by taking into account the different incidence angle of the beam onto the first and second surfaces and their respective coatings. Thin film layers can for example be engineered to be an odd multiple of quarter wavelengths (relative to the wavelength in the material) at one angle and an even multiple of quarter wavelengths at the other angle, so that it changes from reflective to antireflective (assuming an alternating dielectric stack). Another effect that can be used is that directional coating methods will result in different layer thicknesses depending on the angle of the surface. For the first surface, the typical angle of incidence of the beam is closer to the surface normal, typically with an angle lower than 30° relative to it. For the second surface, a typical angle of incidence relative to the surface normal is closer to 45°.

[0221] This is illustrated in Figure 15(c).

[0222] As already explained above, defining the attachment surface at the top of pedestals 205 fabricated in the bottom side of the interposer 200 can be advantageous even if recesses 106 are not defined in the PIC 100. The resulting gaps (pockets) between the PIC 100 and the interposer 200 may play additional functions: If epoxy is dispensed primarily inside the trenches 105 in which the protrusions 202 of the interposer 200 are inserted, epoxy 500 might spill out and a pocket 224 may be defined inside the interposer 200 to make room for the excess epoxy. This pocket may be the same as the one introduced to facilitate molding 211, taking into account the topology of the top interposer 2001 glass building block 200A top-side, as described above. Attachment by filling the trenches 105 with epoxy 500 might also not be sufficient by itself to reliably hold the interposer 200 in the presence of mechanical strain. Thus, dispensing epoxy 500 elsewhere on the surface of the PIC 100 or on the bottom of the interposer 200 might be advantageous to increase the bonding surface. Since the presence of epoxy might change the vertical alignment between the interposer and the PIC if dispensed directly below the pedestals 205, it may be advantageous to dispense it in areas in between the pedestals. The same configuration is applicable if other bonding materials are used. This is also applicable irrespectively on whether recesses 106 have been introduced into the PIC 100 or not, so long as sufficient room is allowed for fitting the bonding material away from the pedestals 205.

[0223] This is illustrated in Figure 16.

[0224] It is difficult to mold protrusions 202 whose width, defined here in the longitudinal direction as the distance between their first refractive 203 and their second reflective 204 surface, is very small, since glass has then to be pressed in a very small cavity on the mold. On the other hand, from the perspective of optical functionality, it is desirable to achieve a thin width, as this reduces the length over which the light beam 300 propagates and broadens before being collimated by the second reflective surface 204. To achieve this, the protrusion 202 can first be molded to be wider than in the final interposer 200, and this width further reduced by processing of the interposer 200 or glass building block 200A after molding, with a grinding tool 600, optionally followed by polishing. The first refractive surface 203 is then defined by the grinding step occurring after the glass molding. This method of manufacturing is particularly advantageous when the gap between the first refractive surface 203 and the edge coupler 103 is filled with index matched epoxy 500 during assembly, as residual roughness on the first refractive surface 203 is then less critical to resulting insertion losses. In this case, a polishing step might not be necessary after grinding or a coarse polishing might be sufficient. When the width of the protrusions 202 is reduced in this way, it may be advantageous to deposit a reflective layer 221 before grinding, as it can then be deposited over the entire side of the glass plate, since it is later removed by the grinding step in the vicinity of the first refractive surface 203 through which a light beam needs to pass. This is advantageous since selective deposition, that is more complex and time consuming, is then not required. This is illustrated in Figure 17. Figure 17(a) shows the protrusion before grinding, Figure 17(b) shows the protrusion after grinding as well as the overlay with the tip of the grinding tool 600 during grinding.

[0225] List of reference signs

[0226] 100 PIC

[0227] 101 Waveguide, waveguide core layer

[0228] 102 Waveguide top cladding

[0229] 103 Edge coupler

[0230] 103B Second edge coupler

[0231] 103L Edge coupler arranged to the left along the lateral direction

[0232] 103R Edge coupler arranged to the right along the lateral direction

[0233] 104 PIC facet terminating the edge coupler

[0234] 105 PIC cavity I trench

[0235] 105B Second PIC cavity I trench

[0236] 105L PIC cavity arranged to the left along the lateral direction

[0237] 105R PIC cavity arranged to the right along the lateral direction

[0238] 106 Recess

[0239] 107 Boundary

[0240] 108 Surface normal of the PIC

[0241] 109 Extension at the corners of the PIC trench I cavity

[0242] 130 Plane of the top PIC surface

[0243] 200 Optical interposer

[0244] 200A Glass building block

[0245] 200B Second glass building block

[0246] 201 Attachment surface

[0247] 202 Protrusion

[0248] 202B Second protrusion

[0249] 202L Protrusion arranged to the left along the lateral direction

[0250] 202R Protrusion arranged to the right along the lateral direction

[0251] 203 First surface (of the first protrusion), transmissive / refractive surface

[0252] 203B First surface of the second protrusion, transmissive / refractive surface

[0253] 204 Second surface (of the first protrusion), curved reflective surface

[0254] 204B Second surface of the second protrusion, curved reflective surface

[0255] 205 Pedestal

[0256] 206 First top-sided reflective surface or beam combiner-splitter

[0257] 206B Second top-sided reflective surface

[0258] 207 First outward facing transmissive / refractive surface

[0259] 207B Second outward facing transmissive / refractive surface 208 Optical axis of a lens

[0260] 209 Surface normal of the reference plane

[0261] 210 Equalizing glass volume (additive)

[0262] 211 Equalizing glass volume (subtractive), pocket

[0263] 212A Top-sided refractive surface of glass building block 200A, internal surface of the interposer

[0264] 212B Refractive surface of glass building block 200B, internal surface of the interposer

[0265] 213 Angle of the first refractive surface of the protrusion

[0266] 214 Rounded corner at the base of the second surface of the protrusion

[0267] 215 Mechanical guide rails

[0268] 216 Angle of the outer surface of the mechanical guide rail

[0269] 217 Outer surface of the mechanical guide rail

[0270] 218 Rounded corner at the base of the mechanical guide

[0271] 219 Cut lines from dicing

[0272] 220 Compression axis of the press

[0273] 221 Reflective coating

[0274] 222 Anti-reflective coating (ARC)

[0275] 223 Dielectric thin-film stack

[0276] 224 Epoxy pocket

[0277] 230 Reference plane

[0278] 300 Light beam

[0279] 300B Second light beam

[0280] 300-FS Free-space beam

[0281] 300-CO Combined light beam

[0282] 300-M Merged light beam

[0283] 400 Optical fiber

[0284] 400B Second optical fiber

[0285] 400-FMF Few-mode fiber

[0286] 500 Epoxy

[0287] 600 Grinding tool

Claims

Claims1. An interposer [200] for coupling a first light beam [300] to or from a photonic integrated circuit, PIC [100], comprising at least one attachment surface [201], that is configured, when the interposer is connected to the PIC, to be in contact with a respective attachment surface of the PIC, thereby defining a position of the interposer with respect to the PIC [100], and a first protrusion [202], the first protrusion [202] comprises a transmissive surface [203] and a first curved reflective surface [204], the curved reflective surface [100] being configured to focus a first light beam [300] along at least one axis of divergence of the light beam [300] or reduce the angle of divergence of the first light beam [300] along at least one axis of divergence of the first light beam [300], the protrusion being configured to extend below a top surface of the PIC [100] when the interposer is connected to the PIC, wherein the interposer [200] is configured to accept:• a first light beam [300] emitted from the PIC [100] and entering the interposer [200] through the first transmissive surface [203] of the first protrusion [202] and propagating to the first curved reflective surface [204] of the first protrusion [202], so that the first light beam [300] propagates from the first curved reflective surface [204] through the interposer and to outside the interposer [200], and / or• a first light beam [300], that propagates from outside the interposer [200] and through the interposer [100] to the first curved reflective surface [204] of the first protrusion [202], so that at least part of the first light beam [300] exits the interposer [200] through the first transmissive surface [203] of the first protrusion [202],2. The interposer [200] of claim 1, further comprising a first top-sided reflective surface [206] that is configured to forward the first light beam [300] between the first curved reflective surface [204] and an outside of the interposer [200],3. The interposer [200] of claim 2, further comprising a top-sided transmissive surface [207] between the first top-sided reflective surface [206] and the outside of the interposer [200], wherein the first light beam [300] exits or enters the interposer [200] through the top-sided transmissive surface [207],4. The interposer [200] of claim 2 or 3, wherein a reference plane [230] bisecting the interposer [200] exists such that:• the first top-sided reflective surface [206] and the top-sided transmissive surface [207] are on a first side of the reference plane [230], andthe first protrusion [202] extends on a second side of the reference plane [230],5. The interposer [200] of claim 4, wherein the at least one attachment surface [201] is in the reference plane [230] or on the second side of the reference plane [230],6. The interposer [200] of any one of claims 4 to 5, wherein the reference plane [230] is parallel to the top surface of the PIC [100] and spaced at most 10 pm apart from this surface of the PIC [100],7. The interposer [200] of any one of claims 2 to 6, wherein the first top-sided reflective surface [206] is a curved surface and is configured to focus the first light beam [300] along at least one axis of divergence of the first light beam [300] or to reduce the angle of divergence of the first light beam [300] along at least one axis of divergence of the first light beam [300],8. The interposer [200] of any one of claims 1 to 7, wherein the interposer further comprises a mechanical guide rail [215], the mechanical guide rail [215] does not interact with the first light beam [300], the mechanical guide rail [215] being configured to extend below the top surface of the PIC [100] when the interposer is connected to the PIC.

9. The interposer [200] of any one of claims 4 to 8, wherein the first light beam [300] propagates between the first curved reflective surface [204] of the first protrusion [202] and the first topsided reflective surface [206] along an axis that is within an angle of 30 degrees from the surface normal [209] of the reference plane [230],10. The interposer [200] of any one of claims 2 to 9, wherein the first light beam [300] propagates freely through the interposer [200] between the first curved reflective surface [204] of the first protrusion [202] and the first top-sided reflective surface [206],11. The interposer [200] of any one of claims 2 to 10, wherein• a direction of a surface normal of the first curved reflective surface [204] of the first protrusion [202] at at least one point, and / or in an area, of the first curved reflective surface [204] of the first protrusion [202] on which the first light beam [300] is incident and• a direction of a surface normal of the first top-sided reflective surface [206] at at least one point, and / or in an area, of the first top-sided reflective surface [206] on which the first light beam [300] is incident are within 20 degrees of each other.

12. The interposer of any one of claims 4 to 11, wherein the first curved reflective surface [204] of the first protrusion [202] has, at least in a region interacting with the first light beam [300], a constant cross section over a length exceeding the width of the first light beam [300] at the first curved reflective surface [204] of the first protrusion [202] in a direction parallel to the reference plane [230],13. The interposer of claim 12, wherein the first curved reflective surface [204] of the first protrusion [202] is a converging reflector in only one first axis of divergence of the first light beam [300],14. The interposer [200] of claim 13, wherein at least one of the first top-sided reflective surface [206] is a converging reflector at least along a second axis of divergence of the first light beam [300] perpendicular to the first axis of divergence of the first light beam [300], or the top-sided transmissive surface [207] is a converging lens at least along the second axis of divergence of the first light beam [300],15. The interposer of claims 2 to 14, wherein the interposer comprises a second protrusion [202B] and a second top-sided reflective surface [206B], the second protrusion comprises a second transmissive surface [203B] and a second curved reflective surface [204B], the second curved reflective surface [204B] is configured to focus a second light beam [300B] along at least one axis of divergence or reduce the angle of divergence of the second light beam along at least one axis of divergence of the second light beam [300B], and the interposer [200] is configured to accept:• the second light beam [300B] that enters the interposer [200] through the second transmissive surface [203B] of the second protrusion [202B] and propagates to the second curved reflective surface [204B] of the second protrusion [202B], which redirects it to a second top-sided reflective surface [206B], and / or• the second light beam [300B] that propagates from the second top-sided reflective surface [206B] to the second curved reflective surface [204B] of the second protrusion [202B], which redirects it to the second transmissive surface [203B] of the second protrusion [202B], so that the second light beam [300B] exits the interposer [200] through the second transmissive surface [203B] of the second protrusion [202B],16. The interposer of claim 15, wherein the first top-sided reflective surface [206] is a beam combiner and / or a beam splitter, and is reflective for the first light beam [300] and transmissive for the second light beam [300B], and wherein the second light beam [300B] exits or enters the interposer [200] through the top-sided transmissive surface [207],17. The interposer of claim 15, wherein the first top-sided reflective surface [206] and the second top-sided reflective surface [206B] are arranged at different heights above the reference plane [230],18. The interposer of claim 17, wherein the interposer [200] is configured to accept:• a second light beam [300B] that enters the interposer through the second transmissive surface [203B] of the second protrusion [202B] and exits the interposer [200] through the second top-sided transmissive surface [207B], and / or• a second light beam [300B] that enters the interposer [200] through the second topsided transmissive surface [207B] and exits the interposer [200] through the second transmissive surface [203B] of the second protrusion [202B],19. The interposer of claim 17, wherein the first top-sided reflective surface [206] and the second top-sided reflective surface [206B] are configured such that the interposer [200] is configured to accept:• the first light beam [300] that enters the interposer [200] through the transmissive surface [203] of the first protrusion [202], propagates to the first curved reflective surface [204] of the first protrusion [202], which redirects it to the first top-sided reflective surface [206], as well as a second light beam [300B] that enters the interposer [200] through the second transmissive surface [203B] of the second protrusion [202B], propagates to the second curved reflective surface [204B] of the second protrusion [202B], which redirects it to the second top-sided reflective surface [206B], so that the first top-sided reflective surface [206] reflects the first light beam [300] and the second top-sided reflective surface [206B] reflects the second light beam [300B] such that the first light beam [300] and second light beam [300B] merge after the respective reflections into a combined merged light beam [300-M], the merged light beam [300-M] exits the interposer through the top-sided transmissive surface [207], and / or• a merged light beam [300-M] that enters the interposer [200] through the top-sided transmissive surface [207], the merged light beam [300-M] is reflected by the first topsided reflective surface [206] and the second top-sided reflective surface [206B] such that it is split into two spatially separated light beams [300], [300B], the first spatially separated first light beam [300] propagates to the first curved reflective surface [204] of the first protrusion, which redirects it to the transmissive surface [203] of the first protrusion [202], though which it exits the interposer [200], the second spatially separated second light beam [300B] propagates to the second curved reflective surface [204B] of the second protrusion [202B], which redirects it to the second transmissive surface [203B] of the second protrusion [202B], though which it exits the interposer[200],20. The interposer of claim 19, wherein the first protrusion [202] and the second protrusion [202B] and the first top-sided reflective surface [206] and second top-sided reflective surface [206B] have curved footprints with centers of curvature oriented towards the top-sided transmissive surface [207],21. A photonic arrangement comprising the interposer [200] of any one of claims 1 to 20 and a photonic integrated circuit PIC [100], wherein the PIC [100] comprises a first edge coupler [103] and a first cavity [105], a facet [104] of the first cavity [105] defines a termination of the first edge coupler [103], the PIC [100] has an attachment surface on a surface of the PIC on which a photonic circuit is fabricated, wherein the at least one attachment surface [201] of the interposer [200] is attached to the attachment surface of the PIC [100] or is in mechanical contact with the attachment surface of the PIC [100], and wherein the first protrusion [202] is inserted into the first cavity [105],22. The photonic arrangement of claim 21, configured such that:• the first edge coupler [103] emits the first light beam [300] which enters the interposer [200] via the transmissive surface [203] of the first protrusion [202] and exits the interposer [200] via the top-sided transmissive surface [207], and / or• the first light beam [300] enters the interposer [200] via the top-sided transmissive surface [207] and exits the interposer [200] via the transmissive surface [203] of the first protrusion [202], after which it is received by the first edge coupler [103],23. The photonic arrangement of claim 21 or 22, wherein the mechanical contact or attachment between the at least one attachment surface [201] of the interposer [200] and the top surface of the PIC [100] determines a vertical alignment between the transmissive surface [203] of the first protrusion [202] and the first edge coupler [103],24. The photonic arrangement of claim 21 , wherein a distance between the reference plane [230] of the interposer [200] and a top surface of the PIC is less than 10 pm.

25. The photonic arrangement of any one of claims 21 to 24, wherein the interposer [200] comprises at least two mechanical guide rails [215], the mechanical guide rails cross each other on the interposer [200], the PIC [100] has a guiding trench matched to the mechanical guide rails [215], the interposer [200] is attached onto the PIC [100] such that a first mechanical guide rail [215] is flush with a first side of the matched guiding trench for at least a portion of the first mechanical guide rail and such that the second mechanical guide rail [215] is flush with asecond side of the matched guiding trench for at least a portion of the second mechanical guide rail, the matched guiding trench is extended at a corner [109] where the two mechanical guide rails [215] cross, and wherein the matched guiding trench may correspond to the first cavity [105] or may be distinct from it.

26. The photonic arrangement of any one of claims 21 to 25, further comprising a pedestal [205] formed on a bottom side of the interposer [200] facing the PIC [100], the at least one attachment surface [201] is at the far end of the pedestal [205], so that the pedestal defines a position of the interposer [200] with respect to the PIC.

27. The photonic arrangement of any one of claims 21 to 26, wherein the first cavity [105] does not intersect with a periphery of the PIC [100],28. The photonic arrangement of any one of claims 21 to 27, further comprising a recess [106] in the top surface of the PIC [100], a pedestal [205] formed on the bottom side of the interposer [200] facing the PIC [100], so that the at least one attachment surface [201] is at the far end of the pedestal [205] and in contact with a bottom surface of the recess [106],29. The photonic arrangement of any one of claims 21 to 28, further comprising a fiber [400] attached to the interposer [200], wherein the fiber [400] is configured to receive the first light beam [300] emitted by the first edge coupler [103] and forwarded by the interposer [200], and / or wherein the fiber [400] is configured to emit the first light beam [300] that is forwarded by the interposer [200] and received by the first edge coupler [103], such that the interposer [200] transforms a profile of the first light beam [300] between the first edge coupler [103] and a boundary of the fiber [400], wherein a light beam profile at the boundary of the fiber [400] is larger than a light beam profile at the termination of the first edge coupler [103],30. The photonic arrangement of any one of claims 21 to 29, wherein the interposer [200] comprises a second protrusion [202B] and a second top-sided reflective surface [206B], the second protrusion comprises a second transmissive surface [203B] and a second curved reflective surface [204B] configured to be a converging reflector along at least one axis of divergence of the light beam [300], and wherein the PIC [100] comprises a second edge coupler [103B], wherein• the second edge coupler [103B] is configured to emit a second light beam [300B] which enters the interposer [200] via the second transmissive surface [203B] of the second protrusion [202B] and propagates to the second curved reflective surface [204B] of the second protrusion [202B], which redirects it to the second top-sided reflective surface [206B], and / or• a second light beam [300B] is able to propagate from the second top-sided reflective surface [206B] to the second curved reflective surface [204B] of the second protrusion [202B], which redirects it to the second transmissive surface [203B] of the second protrusion [202B], the second light beam [300B] exits the interposer [200] through the second transmissive surface [203B] of the second protrusion [202B] and is received by the second edge coupler [103B], and wherein the mechanical contact or attachment between the at least one attachment surface [201] of the interposer [200] and the respective attachment surface of the PIC [100] determines a vertical alignment between the second transmissive surface [203B] of the second protrusion [202B] and the second edge coupler [103B],31. The photonic arrangement of claim 30, wherein the photonic arrangement comprises a fewmode fiber [400-FMF], and wherein the interposer [200] is configured to:• couple the first light beam [300] and second light beam [300B] from the first edge coupler [103] and the second edge coupler [103B] to the few-mode fiber [400-FMF], so that modes of the few-mode fiber [400-FMF] are excited according to a phase and / or intensity of the first and second light beams [300], [300B] emitted by the edge couplers [103], [103B], and / or• split a merged light beam [300-M] from the few-mode fiber [400-FMF] into first and second spatially separated first and second light beams [300] and [300B], and couple first and second spatially separated light beams [300], [300B] to first and second edge couplers [103], [103B], so that modes of the few-mode fiber [400-FMF] are mapped to relative phases and / or relative intensities of the first and second light beams [300], [300B] received by the first and second edge couplers [103], [103B],32. The photonic arrangement of claim 30, wherein the interposer [200] is configured to:• couple the first and second light beams [300], [300B] from the first and second edge coupler [103], [103B] to a free space beam [400-FS], the free space beam [400-FS] is shaped and directed according to the phase and intensity of the light beams [300], [300B] emitted by the edge couplers [103], [103B], or• receive a free space beam [400-FS] and split it into first and second spatially separated light beams [300] and [300B], and couple first and second spatially separated light beams [300], [300B] to first and second edge couplers [103], [103B], so that the shape and direction of the free space beam [400-FS] are mapped to relative phases and / or intensities of the light beams [300], [300B] received by the edge couplers [103], [103B],33. The photonic arrangement of any one of claims 21 to 29, wherein the photonic arrangement comprises a fiber [400], a Faraday rotator and a polarization filter, the Faraday rotator isconfigured to apply a non-reciprocal rotation of the light beam [300] of 45°, wherein the first light beam [300] propagates from the first edge coupler [103] to the first curved reflective surface [204], from the first curved reflective surface [204] to the Faraday rotator, from the Faraday rotator to the polarization filter, and from the polarization filter to the fiber [400],34. The photonic arrangement of any one of claims 21 to 33, wherein the PIC [100] comprises a second edge coupler [103R] that emits or receives light from a second protrusion [202R] of the interposer [200], wherein the first edge coupler [103], [103L] and the second edge coupler [103R], and / or the first protrusion [202], [202L] and the second protrusion [202R], are arranged in a lateral direction with respect to each other, wherein the first and second edge coupler [103L], [103R] are terminated by facets [104] of the first cavity [105], [105L] and of a second cavity [105R], respectively, wherein there exists a straight extended boundary [107] that connects the terminations of the first and the second edge coupler [103L], [103R], the first and second edge coupler are located on the same side of the boundary [107], the positions of the first edge coupler [103], [103L] and of the second edge coupler [103R] subdivide the extended boundary [107] into three sections, and wherein a waveguide [101] of the PIC [100] consecutively crosses the straight extended boundary [107] through two different sections, and light is transmitted through the waveguide [101],35. A method for manufacturing the interposer [200] of claims 1 to 20, wherein the interposer [200] comprises at least a glass building block of a first type [200A], and wherein a plurality of glass building blocks of the first type [200A] is manufactured in parallel by molding a glass plate in between a first mold and a second mold, wherein• one or more structures corresponding to the glass building block of the first type [200A] are defined on each of the first mold and the second mold,• the first mold and second mold are pressed together along a compression axis [220] with the glass plate in between the first mold and the second mold, and• a plurality of glass building blocks of the first type [200A] molded in this manner are separated from one another by dicing of the molded glass plate.

36. The method of manufacturing of claim 35, wherein all surfaces of the interposer [200] are angled by at least 20 degrees relative to the compression axis [220],37. The manufacturing method of any one of claims 35 or 36, wherein a surface normal [209] of the reference plane [230] is within 30 degrees of the compression axis [220],38. The manufacturing method of any one of claims 35 to 37, wherein a thickness of the molded glass plate remains within 20% deviation of a thickness of the initial, unmolded glass plate whenaveraged inside a circle with a diameter of 10 mm.

39. The manufacturing method of any one of claims 35 to 38, the glass building block of the first type [200A] comprises the at least one attachment surface [201] and a reference plane [230], the first protrusion [202], as well as a transmissive surface [207], [212A] which may be the outward facing top-sided transmissive surface [207] of the interposer [200] or an internal transmissive surface [212A] of the interposer [200], wherein the first transmissive surface [203] of the first protrusion [202], the first curved reflective surface [204] of the first protrusion [202] and the at least one attachment surface [201] are molded by a first mold and the top-sided transmissive surface [207], [212A] of the glass building block of the first type [200A] is molded by a second mold.

40. The manufacturing method of claim 39, wherein the first mold and the second mold are arranged such that the first light beam [300] propagates between the first curved reflective surface [204] of the first protrusion [202] and the first top-sided reflective surface [206] of the interposer [200] along an axis that is within an angle of 30 degrees from the compression axis [220A] of the glass building block of the first type [200A],41. The manufacturing method of any one of claims 39 or 40, wherein the glass building block of the first type [200A] comprises a pocket [211] devoid of glass, the pocket [211] is formed by the same mold as the first protrusion [202], the first light beam [300] does not interact with the pocket [211], the pocket extends to the same side of the reference plane [230] as the top-sided transmissive surface [207], [212A] of the glass building block of the first type [200A], and the pocket is configured to compensate for material of the glass plate required for forming structures of the glass building block of the first type [200A] in the molding step such that an averaged thickness of the molded glass plate does not deviate substantially from a thickness of the unmolded glass plate.

42. The manufacturing method of any one of claims 39 to 41 , wherein the glass building block of the first type [200A] comprises an additive volume of glass [210], the additive volume of glass [210] is formed by the same mold as the top-sided transmissive surface [207], [212A] of the glass building block of the first type [200A], the first light beam [300] does not interact with the additive volume of glass [210], the additive volume of glass [210] is on the same side of the reference plane [230] as the top-sided transmissive surface [207], [212A] of the glass building block of the first type [200A], and the additive volume is configured to compensate for material of the glass plate required for forming optical elements of the glass building block of the first type [200A] in the molding step such that an averaged thickness of the molded glass plate does not deviate substantially from a thickness of the unmolded glass plate.

43. The manufacturing method of claims 39 to 42, wherein the top-sided transmissive surface [207] of the interposer [200] is curved and configured to implement a converging lens along at least one axis of divergence of the light beam, the top-sided transmissive surface [207] of the interposer [200] is fabricated on a glass building block of a second type [200B], the top-sided transmissive surface [212A] of the glass building block of the first type [200A] is a first internal surface of the interposer [200], the glass building block of the second type [200B] has a further transmissive surface [212B] that is a second internal surface of the interposer [200], the glass building blocks of the first and second type [200A], [200B] are attached together such that the first light beam [300] propagates inside the interposer [200] from the first internal surface [212A] to the second internal surface [212B], or vice-versa.

44. The manufacturing method of claim 43, wherein the glass building block of the second type [200B] is manufactured by molding a glass plate with a first mold and a second mold that are pressed together along a compression axis [220B] of the glass building block of the second type [200B], wherein an optical axis [208] of the converging lens formed by the top-sided transmissive surface [207] of the interposer [200] is within 35 degrees of the compression axis [220B] of the glass building block of the second type [200B],45. The manufacturing method of any one of claims 35 to 44, wherein a reflective layer [221] is selectively deposited in regions of a molded glass plate corresponding to the curved reflective surfaces [204] of protrusions [202], prior to dicing the molded glass plate.

46. The manufacturing method of claim 45, wherein an antireflective coating [222] is applied to the same side of the molded glass plate onto which the reflective layer [221] has been selectively deposited or onto which the reflective layer [221] is to be selectively deposited, the antireflective coating [222] is deposited prior to dicing the molded glass plate and before or after the selective deposition of the reflective layer [221],47. The manufacturing method of any one of claims 35 to 44, wherein a dielectric thin film stack [223] is applied prior to dicing to an entire side of a molded glass plate that comprises protrusions [202], the dielectric thin film stack [223] is configured to be antireflective at a first angle of incidence of the first light beam [300] onto the transmissive surface [203] of the first protrusion [202] and to be reflective at a second angle of incidence of the first light beam [300] onto the reflective surface [204] of the first protrusion [202],48. The manufacturing method of any one of claims 35 to 44, wherein a grinding step is applied after molding the glass plate and before dicing the molded glass plate, and wherein in thegrinding step the transmissive surfaces [203] of protrusions [202] are shaped.

49. The manufacturing method of claim 48, wherein a reflective layer [221] is deposited onto an entire side of the molded glass plate that comprises the transmissive surfaces [203] of the protrusions [202] that are being shaped by the grinding step, and the reflective layer [221] is deposited prior to the grinding step.

50. A method for manufacturing the photonic arrangement of claims 21 to 33, wherein the PIC [100] is fabricated by patterning a wafer, the interposer [200] comprises at least a first glass building block of the first type [200A], a plurality of glass building blocks of the first type [200A] are manufactured in parallel by molding a glass plate in between a first mold and a second mold on which one or more structures corresponding to the glass building block of the first type [200A] are defined, the first mold and second mold are pressed together along a compression axis [220] with the glass plate in between the first mold and the second mold, a plurality of glass building blocks of the first type [200A] molded in this manner are separated from one another by dicing of the molded glass plate, and the glass building blocks of the first type [200A] are mounted onto the PIC [100],51 . The manufacturing method of claim 50, wherein the glass building block of the first type [200A] or the interposer [200] is attached to the PIC [100] before the wafer is diced and an instance of the PIC [100] is separated from other instances of the PIC [100],52. The manufacturing method of any one of claims 50 or 51 , wherein an etching step is applied to the wafer to define dicing lanes, and the first cavity [105] of the PIC [100] is etched in the same etching step as the dicing lanes.

53. The manufacturing method of claims 50 to 52, wherein the glass building block of the first type [200A] comprises a pocket [224] devoid of glass, and wherein the pocket [224] is formed by the same mold as the first protrusion [202], wherein the first light beam [300] does not interact with the pocket [224], the pocket extends to the same side of the reference plane [230] as the top-sided transmissive surface [207], [212A] of the glass building block of the first type [200A], and the interposer [200] is attached to the PIC [100] such that the pocket [224] is at least partially filled with epoxy.

54. The manufacturing method of claims 50 to 53, wherein a recess [106] is formed on the PIC [100] by using at least one wet etching step that has an etch stop at a boundary of two layers of the PIC [100],55. The manufacturing method of claims 50 to 54, wherein the first curved reflective surface [204] of the first protrusion [202] has, at least in regions interacting with the first light beam [300], a constant cross section over a length exceeding the width of the first light beam [300] at the first curved reflective surface [204] of the first protrusion [202] in a direction parallel to the reference plane [230], the first curved reflective surface [204] of the first protrusion [202] is a converging reflector in only one axis of divergence of the first light beam [300], the interposer [200] is attached to the PIC [100] in a first step, and in a second step a fiber [400] is actively aligned to the interposer [200] and attached to the interposer [200],56. The manufacturing method of claims 50 to 54, wherein the first curved reflective surface [204] of the first protrusion [202] has, at least in regions interacting with the first light beam [300], a constant cross section over a length exceeding the width of the first light beam [300] at the first curved reflective surface [204] of the first protrusion [202] in a direction parallel to the reference plane [230], the first curved reflective surface [204] of the first protrusion [202] is a converging reflector in only one axis of divergence of the light beam [300], and wherein the interposer [200] comprises a glass building block of the first type [200A] and a glass building block of a second type [200B], the first protrusion [202] is part of the glass building block of the first type [200A], the glass building block of the first type [200A] is attached to the PIC [100] in a first step, and in a second step the glass building block of the second type [200B] is actively aligned to the glass building block of the first type [200A] and fixed relative to the glass building block of the first type.

Images