The invention relates to a photonic chip comprising a buried laser source and to a process for fabricating such a photonic chip.
The photonic chips essentially lie in a plane called the “plane of the chip” below.
Known photonic chips comprise:
Such a photonic chip is for example described in the following article: J. Durel et al.: “First Demonstration of a Back-Side Integrated Heterogeneous Hybrid III-V/Si DBR Lasers for Si-Photonics Applications”, IEEE International Electron Devices Meeting (IEDM), December 2016.
This chip has the advantage of being particularly simple to fabricate. However, it would be desirable, while retaining this advantage, to improve it even further by improving the thermal dissipation of the heat generated by the laser source while it is operating. To this end, the invention relates to such a photonic chip in which the photonic chip comprises an interconnect metal network forming a via that extends, in the optical layer, from the bonding interface to the first embedded electrical contact of the waveguide, this interconnect metal network comprising metal vias that electrically connect together metal lines that extend mainly parallel to the plane of the chip, these metal lines being arranged one above the other within the optical layer.
Embodiments of this photonic chip may comprise one or more of the features of the dependent claims.
Another subject of the invention is a process for fabricating the photonic chip that is the first subject of the present patent application.
The invention will be better understood on reading the following description. It is given merely by way of non-limiting example and makes reference to the drawings, in which:
In these figures, the same references have been used to designate the same elements. In the rest of this description, features and functions that are well known to those skilled in the art will not be described in detail.
The system 2 typically takes the form of a parallelepipedal package inside of which electronic and optical chips are housed. Only the solder bumps of the system 2 protrude from a lower face of this package. To simplify the illustration, this package has not been shown in
In the remainder of this description, the horizontal is defined in the figures by the X- and Y-directions of an orthogonal coordinate system R. The Z-direction of the coordinate system R corresponds to the vertical direction. Terms such as “lower”, “upper”, “above”, “below”, “top” and “bottom” are defined with respect to this Z-direction.
In its lower portion, the system 2 comprises a carrier 14. The bumps of the array 6 are placed directly on the lower face of the carrier 14. The carrier 14 and the array 6 are together known as a BGA (acronym of “ball grid array”). The upper face of the carrier 14 comprises pads to which electronic and/or photonic chips may be soldered. In the particular case of
The electronic chip only comprises electronic components that are connected to one another with a view to performing preset functions. The electronic chip 16 is therefore devoid of optical component.
In contrast, the photonic chip 18 comprises optical components for performing preset functions. An optical component is a component that, during its use, generates or modifies or guides an optical signal. Typically, the wavelength λ of the optical signal is comprised between 1200 nm and 1700 nm. Generally, at least one of these optical components is an active optical component, i.e. an optical component:
In this embodiment, the photonic chip 18 is devoid of electronic component.
With a view to the supply of electricity or the exchange of electrical signals, the photonic chip 18 and the electronic chip 16 of the system 2 are electrically connected to pads on the upper face of the carrier 14. The chips 16 and 18 are also electrically connected to printed circuit board 4. To this end, here, the chips 16 and 18 are soldered to respective pads of the upper face of the carrier 14 by way of solder microbump arrays, 20 and 22, respectively. In
The carrier 14 comprises electrical connections that extend mainly horizontally, such as a connection 24. The connection 24 electrically connects together certain pads of the upper face of the carrier 14 in order to allow the exchange of electrical signals or power between the chips soldered to these pads. The carrier 14 is also passed through by vertical connections, such as the connections 26 and 28. Generally, these vertical connections open onto a horizontal metal line of a redistribution layer (RDL). Here, only one line 25 of this redistribution layer has been illustrated. The horizontal lines of the redistribution layer electrically connect certain of the vertical connections to corresponding solder bumps of the array 6. The vertical connections notably allow the chips 16 and 18 to be supplied with electrical power and, also, electrical signals to be exchanged with other chips soldered to the printed circuit board 4. Apart from the horizontal and vertical connections, the carrier 14 is generally devoid of any electrical or optical component.
More precisely, the layers 36 and 38 are directly bonded to each other via a bonding interface 40 that lies essentially in a horizontal plane. The upper face of the layer 38 corresponds to the upper face 32 and the lower face of the layer 36 corresponds to the lower face 34.
The layer 36 is mainly made of a non-conductive material. In this description, the expression “non-conductive material” denotes a material the conductivity of which at 20° C. is lower than 10−1 or 10−2 S/m. Thus, this non-conductive material may be a semiconductor such as undoped silicon or an electrically insulating material such as glass. An electrically insulating material is a material the conductivity of which at 20° C. is lower than 10−9 S/m and, preferably, lower than 10−15 S/m or 10−16 S/m. Below, an electrically insulating material is also referred to as a “dielectric material”. Here, the non-conductive material is undoped crystalline silicon. In contrast, in this description, the expression “conductive material” or “electrically conductive material” denotes a material the electrical conductivity of which at 20° C. is generally higher than 105 or 106 S/m.
Here, the layer 36 comprises a sublayer 42 made of undoped crystalline silicon and a surface sublayer 44 produced directly on the sublayer 42. Here the sublayer 44 is made of silicon dioxide.
The lower face of the sublayer 42 corresponds to the face 34. The thickness of the sublayer 42 is quite large in order to be quite rigid and to not exhibit too much bow. In particular, the thickness of the sublayer 42 is sufficient to allow the chip 18 to be handled. Thus, the layer 36 forms a carrier for the optical layer 38. To this end, the thickness of the sublayer 42 is larger than 50 μm or than 80 μm or than 100 μm.
The sublayer 42 comprises vias, called “primary vias” below, that allow electrical connections that pass vertically through the substrate 30 to be formed. Each primary via extends from a lower end to an upper end. The lower end lies flush with the lower face 34. The upper end lies flush with the bonding interface 40. These primary vias therefore pass right through the thickness of the interconnect layer 36. In contrast, the primary vias do not pass through and do not penetrate into the optical layer 38.
The layer 36 does not contain a layer of buried silicon oxide. The smallest thickness thereof resulting in an acceptable bow is therefore smaller than that of a layer containing such a layer of buried silicon oxide. A smaller thickness notably allows the parasitic capacitance of the primary vias, which is proportional to the height of the primary vias, to be decreased. Thus, preferably, the thickness of the sublayer 42 is also smaller than 150 μm or 100 μm.
The primary via 50 is produced in a conventional manner. For example, on this subject, the reader may consult the following article: Ken Miyairi et al.: “Full integration and electrical characterization of 3D Silicon Interposer demonstrator incorporating high density TSVs and interconnects”, 45th International Symposium on Microelectronics, 2012. Thus, below, only a few details of the via 50 are described.
In this embodiment, the via 50 is formed, from bottom to top:
The lower portion 54 extends vertically from the lower end of the via 50 to the upper portion 56. The horizontal cross-sectional area of the lower portion is substantially constant over its entire height. Its height and its diameter are denoted H1 and D1 below, respectively. The lower portion 54 passes through at least 80% and, preferably, at least 90% or 95% of the thickness of the layer 36. Since the thickness of the layer 36 is large, the diameter D1 is also large so that the aspect ratio H1/D1 is lower than 10/1 or 8/1. Specifically, it is difficult to fabricate vias having an aspect ratio higher than 10/1. Here, the height H1 is larger than 80% or 90% or 95% of the thickness of the sublayer 42. Under these conditions, the height H1 is comprised between 80 μm and 142.5 μm and, generally, comprised between 95 μm and 130 μm. The diameter D1 is then typically comprised between 10 μm and 20 μm. In the case where the sublayer 42 is made of a semiconductor such as silicon, preferably, the lower portion 54 is electrically insulated from this semiconductor by a thin coating of electrically insulating material. To simplify the figures, this coating of electrically insulating material has not been shown. The lower portion 54 is made of metal and, for example, of copper.
The upper portion 56 of the via 50 is formed by an interconnect metal network. This upper portion 56 therefore consists of:
It will be recalled that a “via” is a vertical electrical connection, i.e. one that extends mainly in a direction perpendicular to the plane of the photonic chip.
The metal lines and, preferably, the metal vias are produced in the silicon-oxide sublayer 44. Here, the upper portion 56 comprises:
In this description, the expression “metal via” denotes a via made of metal and the diameter of which is generally small, i.e. the diameter of which is smaller than 3 μm and, often, smaller than 1 μm. The height of a metal via is also small, i.e. smaller than 3 μm or 1 μm.
Here the metal is copper. The structure of such an interconnect metal network is well known to those skilled in the art. It is a question of the interconnect metal network of the redistribution layer or RDL.
The thickness of the sublayer 44 is typically smaller than 10 μm or 3 μm.
The diameter of the upper portion 56 is equal to the maximum diameter of a metal via thereof. Here, the diameter of the upper portion 56 is therefore smaller than 10 μm and, generally, smaller than 3 μm or 1 μm.
The optical layer 38 comprises at least one optical component buried in this layer. Here, the optical component is chosen from the group consisting of an optical modulator, a laser source, a waveguide, a photodetector, a demultiplexer and an optical multiplexer. Among the optical components of this group, all are active optical components with the exception of the waveguide.
In this exemplary embodiment, only two optical components 70, 72 buried in the layer 38 have been shown. The component 70 is a laser source and the component 72 is an optical modulator. The component 72 is capable of modifying, depending on an electrical control signal, the phase, the amplitude or the intensity of an optical signal passing through it. Such optical modulators are well known and hence the component 72 is not described in detail here. The component 72 comprises an electrical contact 74 intended to receive the electrical control signal. Here, this electrical contact 74 is embedded in the optical layer 38. In this embodiment, the electrical contact 74 is located on a lower face of the component 72. The electrical contact 74 therefore faces the bonding interface 40. In this configuration, the electrical contact 74 is only accessible from below the optical component 72.
The component 70 is a hybrid III-V/silicon laser source. The component 70 is for example a distributed feedback (DFB) laser or a distributed Bragg reflector (DBR) laser. Such a laser source is well known to those skilled in the art and is therefore not described in detail here. Here, the component 70 notably comprises a waveguide 76 and, stacked directly on top of one another:
The component 70 also comprises an upper contact 84 and a lower contact 86 making direct contact with the upper layer 82 and lower layer 78, respectively.
The active layer 80 generates the optical signal emitted by the component 70 when a suitable potential difference is applied across the contacts 84 and 86.
Here, the lower layer 78 is optically coupled to the waveguide 76. To achieve this, the layer 78 is placed above the waveguide 76 and separated from this waveguide 76 by a thin sublayer 90 of buried oxide. For example, the sublayer 90 is made of silicon dioxide. The thickness of the oxide sublayer 90 is smaller than 100 nm and, preferably, smaller than 50 nm or 20 nm. Thus, the optical signal generated by the active layer 80 may then be guided by the waveguide 76.
In this embodiment, the waveguide 76 comprises:
The thickness e142, in the Z-direction, of the central rib 142 is generally smaller than 1 μm and, preferably, smaller than 500 nm. Here, the cross section of the rib 142 is rectangular and hence it has two parallel vertical walls.
The protuberances 144A and 144B are located on either side of the rib 142. Here, the cross section of each of these protuberances 144A, 144B is rectangular and hence they each also have vertical walls. The upper face of each of the protuberances 144A, 144B makes direct mechanical contact with the sublayer 90. The thickness, in the Z-direction, of the protuberances 144A and 144B is equal to or smaller than the thickness of the rib 142. Here, the thickness of the protuberances 144A, 1446 is equal to the thickness of the rib 142.
The rims 146A and 146B are each located on a respective side of the rib 142. These rims extend, parallel to the X-direction, from a vertical wall of the rib 142 to a facing vertical wall of one of the protuberances 144A, 144B. The upper face of each rim 146A, 146B makes direct mechanical contact with the sublayer 90. The thickness e146 of each rim 146A, 146B is smaller than the thickness e142 of the rib 142 so as to confine the guided optical signal mainly to the rib 142. To this end, the thickness e146, in the Z-direction, of the rims 146A, 146B is generally smaller than 0.5e142 or 0.33e142. To facilitate the conduction of heat from the rib 142 to the protuberances 144A, 144B, the thickness e146 is also generally chosen to be larger than 0.1e142 or 0.2e142.
The rib 142, the protuberances 144A, 144B and the rims 146A, 146B are made of the same material so that, joined together, these various portions of the waveguide 76 form only a single piece of material. Here, the waveguide 76 is made of single-crystal silicon.
Here, waveguide 76 and optical component 72 are buried within an encapsulating sublayer 122 located directly below oxide sublayer 90. More specifically, the sublayer 122 extends from a sublayer 120 to the sublayer 90. The sublayer 122 is here divided into an upper optical-guide level 92 and a lower electrical-interconnect level 94.
The level 92 includes all optical components and portions of optical components produced under the sublayer 90. The level 92 therefore here comprises the waveguide 76, the optical component 72 and a solid dielectric material in which the optical components and portions of optical components are encapsulated. The optical components encapsulated in the sublayer 122 and, in particular their vertical walls and their lower faces, therefore make direct mechanical contact with the solid dielectric material. There is no cavity between the vertical walls of the encapsulated optical components and this solid dielectric material. For example, this level 92 is fabricated from a crystalline silicon sublayer in which the waveguide 76 and component 72 were fabricated before being encapsulated in the dielectric material.
The level 94 is devoid of optical component and of any portions of optical components. Here, as will be described below, the level 94 comprises interconnect metal networks encapsulated in a solid dielectric material. The dielectric material of the sublayer 122 has a refractive index lower than the refractive index of the material used to produce the waveguide 76 and the component 72. For example, the dielectric material is silicon dioxide.
The layers 78, 80 and 82 are for their part produced in a sublayer 128 located directly on the sublayer 90. The sublayer 128 here comprises a III-V material from which the layers 78, 80 and 82 are produced and a dielectric material that encapsulates the layers 78, 80 and 82. The sublayer 128 extends horizontally directly over the sublayer 90 and its upper face corresponds to the face 32.
To supply and/or transmit electrical signals through layer 38, this layer 38 comprises vias, which are referred to as “secondary vias”. Each secondary via extends one primary via into optical layer 38. To this end, each secondary via extends from the bonding interface 40 into the layer 38. To limit the constraints to be met with respect to the layout of the optical components with respect to these secondary vias, the diameter of the secondary vias is much smaller than the diameter of the primary vias. Here, the diameter of the secondary vias is smaller than 3 μm and, preferably, smaller than 1 μm. To facilitate the production of such secondary vias, the thickness of the layer 38 is smaller than 15 μm or 8 μm or 5 μm.
Below, the secondary vias are classified into two categories. A first category dubbed “secondary through-vias” and a second category dubbed “secondary blind vias”.
The secondary through-vias pass through the thickness of the layer 38 and therefore, in particular, the optical-guide level 92. They therefore extend vertically from the bonding interface 40 to the upper face 32. They are typically used to electrically connect electrical tracks formed on the upper face 32 to one of the primary vias. In contrast, the secondary blind via do not pass right through the thickness of the layer 38 and, in particular, do not pass through the optical-guide level 92. They therefore extend from the bonding interface 40 to an electrical contact, such as the contact 74, embedded in layer 38.
By way of illustration,
The secondary through-via 102 electrically connects the primary via 51 to an electrical track 108 produced on the face 32. The track 108 electrically connects the end of via 102 to the contact 86. The lower end of via 102 makes direct mechanical and electrical contact with the metal line of the via 51. Here, the vias 100 and 102 are used to power component 70.
The structure of via 102 is identical to that of the via 100. Thus, only the structure of via 100 is described below in more detail.
In this embodiment, the via 100 has a lower portion 110 and an upper portion 112 arranged immediately one above the other. The portions 110 and 112 are produced, during fabrication, from opposite sides of the optical layer.
The lower portion 110 is an interconnect metal network. Here, this portion 110 comprises:
The diameter of the lower portion 110 is defined in the same way as the diameter of the upper portion 58 of the primary via 50 was defined. The diameter of the portion 110 is smaller than 3 μm or 1 μm.
In this embodiment, the interconnect metal network that forms portion 110 comprises only two levels of metal lines. For example, these are the levels often designated by the references M4 and M5.
Generally, the height of this portion 110 is small, i.e. smaller than 6 μm or 3 μm or 1 μm.
Here, the metal line 116 is produced in a silicon-oxide sublayer 120 the lower face of which coincides with the bonding interface 40. The sublayer 120 is produced directly under the sublayer 122. Here, the sublayer 120 is made of silicon dioxide.
The metal vias 118 are located in level 94 of the sublayer 122.
The upper portion 112 of the secondary via 100 is formed from a coating 124 made of a conductive material. For example, the coating 124 is made of copper or of aluminum. This coating 124 continuously covers the vertical walls and the bottom of a hole 126 dug in the layer 38. More precisely, the hole 126 extends from the upper face 32 to the metal line 114. To this end, the hole 126 passes through, in succession, from top to bottom, the encapsulating layer 128, the sublayer 90, the optical-guide level 92 and a portion of the level 94 of the sublayer 122.
The maximum diameter of the hole 126 is smaller than 3 μm and, preferably, smaller than 1 μm. Thus, the maximum diameter of the upper portion 112 of the secondary via 100 is also smaller than 3 μm or 1 μm.
The upper portion of the coating 124 makes direct mechanical and electrical contact with the track 106, and its lower portion makes direct mechanical and electrical contact with the metal line 114. In the example of
The layer 38 also includes a secondary blind via 130. The via 130 electrically connects the electrical contact 74 to the upper end of the primary via 52, with a view to reception of the electrical control signal for controlling the optical component 72. To this end, the via 130 extends vertically from the bonding interface 40 to the electrical contact 74. It therefore passes right through the sublayer 120 and the level 94 of the sublayer 122. In contrast, it does not pass through the optical-guide level 92 or the sublayer 90. The via 130 does not open onto the upper face 32, unlike a secondary through-via.
The via 130 is solely fabricated from the side of the layer 38 which faces the bonding interface 40. Here, the via 130 is formed by an interconnect metal network. In this embodiment, the via 130 comprises, from top to bottom:
The metal line 138 makes direct mechanical and electrical contact with the upper end of the primary via 52. Here, the interconnect metal network comprises five different levels of metal lines. For example, these five levels of metal lines are often designated by the references M1 to M5, respectively. The diameter of the via 130 is also smaller than 3 μm or 1 μm.
The waveguide 76 of the laser source 70 comprises at least one electrical contact thermally coupled to one of the protuberances 144A, 144B or to the rib 142. Here, an electrical contact is said to be “thermally coupled” only if:
When the distance dv is non-zero, the electrical contact is mechanically isolated from the protuberance 144A, 144B or from the rib 142 by a thin film of the dielectric material of the sublayer 122. The minimum thickness of this film is equal to the distance dv.
In the embodiment shown in
Each electrical contact 148A, 148B and 148C is thermally connected to the interconnect layer 36 by way of a blind vias 158A, 158B and 158C, respectively. The vias 158A, 158B and 158C are each structurally identical, for example, to the via 130. However, unlike the via 130, the function of the vias 158A, 158B and 158C is not necessarily to electrically connect the contacts 148A, 148B and 148C to a primary via of the layer 36. Here, the lower end of each via 158A, 158B and 158C makes direct mechanical and electrical contact with a metal line located in the sublayer 44, and which lies flush with the interface 40. This metal line is here electrically insulated from all the primary vias. The vias 158A, 158B and 158C are here made from a material that is a good thermal conductor. A material that is a “good thermal conductor” is typically a material the thermal conductivity of which is higher than 1.2C122, and preferably higher than 2C122 or 3C122, where C122 is the thermal conductivity of the dielectric material of the sublayer 122. Here, the material used to produce the vias 158A, 158B and 158C is the same as that used to produce the via 130, i.e. a metal.
The layers 78 and 80 of the laser source 70 are thermally connected to the primary vias 50 and 51 by way of secondary vias 100 and 102, respectively. Thus, the heat produced by the laser source 70 is dissipated, in part, to the layer 36 by way of the secondary vias 100 and 102. The presence of the vias 158A, 158B and 158C improves the thermal dissipation of the heat produced by the laser source 70. Specifically, it is known to those skilled in the art that the waveguide 76 sees its temperature increase during the operation of the laser source 70. In the absence of vias 158A, 158B and 158C, the presence of the dielectric material of the sublayer 122 under the waveguide 76 prevents the produced heat from being removed effectively. The vias 158A, 158B and 158C each form, through the sublayer 122, a thermal bridge, here one made of metal, that thermally connects the waveguide 76 to the layer 36. This allows the heat produced by the laser source 70 to be more effectively removed.
Redistribution lines 150 (also known by the acronym RDL) are produced on the lower face 34 to electrically connect the lower ends of the primary vias to corresponding solder microbumps. To this end, the redistribution lines comprise connection pads to which the solder microbumps are directly fastened. In
A process for fabricating the system 2 will now be described with reference to
Initially, in a step 160, the stack of sublayers 90, 122 and 120 is fabricated on a handle 164 (
Step 160 also comprises:
The sublayer 90 is fastened directly to the handle 164. The handle 164 is a carrier that allows the stack of the sublayers 120, 122 and 90 to be easily handled. To this end, the thickness of the handle 164 is typically larger than 250 μm or 500 μm. For example, here, the handle 164 is a silicon substrate with a thickness of 750 μm or 775 μm.
Preferably, the sublayer 90 is the buried oxide layer of a silicon-on-insulator (SOI) substrate. In this case, the waveguide 76 and the component 72 are produced, typically by etching, in the single-crystal silicon layer of this SOI substrate. Once the waveguide 76 and the optical component 72 have been produced on the sublayer 90, they are buried in the dielectric material of the sublayer 122 by depositing, in succession, a plurality of oxide layers one after the other. During the deposition of these successive oxide layers, the interconnect metal networks forming the contacts 148A, 148B, 148C, the secondary via 130, the vias 158A, 158B and 158C and the lower portions 110 of the secondary vias 100 and 102 are produced. Thus, these interconnect metal networks are produced from the side opposite the handle 164. Processes for producing the optical components on the sublayer 90, processes for embedding the components in the sublayer 122 and processes for producing interconnect metal networks are well known and are therefore not described in more detail here. It will be noted that, at this stage of the fabricating process, production of the upper portions 112 of the secondary vias 100 and 102 is impossible. The sublayer 122 therefore does not comprise, at this stage, the upper portions 112 of the vias 100 and 102.
In parallel, in a step 162, the interconnect layer 36 is fabricated on a handle 170 (
Step 162 also comprises producing the primary vias 50 to 52 in the layer 36. Thus, at the end of step 162, the layer 36 comprises these primary vias 50 to 52. The production of such a layer 36 on the handle 170 and the production of the vias 50 to 52 in this layer 36 are conventional operations. For example, the portions 54 of the vias 50 to 52 are produced using a dielectric coating that is covered with a barrier layer of titanium or tantalum and filled with copper. A process for fabricating these portions 54 is for example described in the article by Ken Miyairi cited above. Next, the metal vias 60 are produced, so as to make electrical contact with the portions 54 of the primary vias. Lastly, metal lines, such as the metal line 58, are produced and then encapsulated in the oxide sublayer 44. The outer face of the sublayer 44 is then prepared for its direct bonding to the face 166.
In a step 174, the faces 166 and 172 are bonded, by direct bonding, to each other. This is shown in
Next, in a step 176, the handle 164 is removed in order to expose the sublayer 90 (
In a step 178, the light-amplifying portion of the component 70 is produced on the sublayer 90, on the side opposite the optical-guide level 92 (
In a step 180, the contacts 84, 86, the electrical tracks 106 and 108 and the upper portions 112 of the secondary vias 100 and 102 are produced (
In a step 182, the handle 170 is then removed in turn to expose the lower face 34 (
In a step 184, the redistribution lines 150, the solder microbumps 152 to 154 and the encapsulating layer 156 are produced on the exposed lower face 34. The fabrication of the photonic chip 18 is thus completed.
In a step 186, the electronic chip 16 and the photonic chip 18 are soldered to pads on the upper face of the carrier 14. The chips 16 and 18 are then electrically connected to each other by way of horizontal connections on the carrier 14.
In a step 190, the carrier 14 and the chips 16 and 18 are encapsulated in an electrically non-conductive material that is a good thermal conductor, i.e. a material such as an epoxy resin. In addition, preferably, a cover on which a heat sink, such as a radiator, is located is fastened to the upper faces of the chips 16 and 18. This cover is for example adhesively bonded, using a thermally conductive adhesive, directly to the upper face 32 of the photonic chip 18 and the upper face of the chip 16. Thus, these upper faces may make direct contact with the thermal adhesive, which makes contact with the cover. This is made possible by the fact that electrical power and the electrical signals are all routed via the lower faces of chips 16 and 18.
Next, in a step 192, the system 2 is for example soldered to the printed circuit board 4.
The optical layer 202 is identical to optical layer 38 except that it in addition comprises an active electronic component 210 buried in level 92 of the sublayer 122. This electronic component 210 is qualified active because, to operate, it needs to be supplied with electricity, and it emits and/or receives an electrical signal. To this end, the component 210 comprises an electrical contact 212 on its lower face. This electrical contact 212 is therefore embedded in the sublayer 122 and faces the bonding interface 40.
Here, the secondary via 204 electrically connects the electrical contact 212 of the electronic component 210 to the primary via 52, solely passing through the lower level 94 of the sublayer 122 and the sublayer 120 to do so. The electronic component 210 is, for example, intended to supply, from an electrical signal that it receives via the primary via 52, an electrical control signal to the optical component 72. To this end, the component 210 comprises, on its lower face, another electrical contact connected to the contact 74 of the component 72. To achieve this, the interconnect metal network located in the level 94 of the sublayer 122 is modified to create a path that electrically connects together this other electrical contact and the contact 74. To simplify
The via 206 is a via that electrically connects a metal line 135, embedded in the level 94 of the sublayer 122, to an electrical track 216 produced on the upper face 32. The via 206 extends vertically from the face 32 to the embedded metal line 135, and notably passes through the sublayer 90 and the optical-guide level 92 to do so.
The electrical track 216 comprises a test pad that allows the operation of the components 210 and/or 72 to be tested during the fabricating process of the photonic chip 200 and, in particular, before the removal of the handle 170. To this end, though not shown in
The structure of via 206 is here identical to that of via 100 except that it is devoid of lower portion 110. Typically, the via 206 is fabricated in the same way as the upper portion 112 of the via 100 and generally at the same time as this upper portion 112.
The secondary through-vias 252 and 254 are identical to the vias 100 and 102, respectively, except that their upper portions 270 are produced differently. More precisely, the upper portion 270 is here a solid via. In other words, the upper portion 270 is identical to the upper portion 112, except that the central hollow of the hole 126 is filled with a conductive material. Generally, in this case, the maximum diameter of the upper portion 270 of the vias 252 and 254 is smaller than 1 μm.
The laser source 256 is identical to the component 70, except that:
The contact 274 is a metal via that electrically connects the layer 78 to the electrical track 108.
The waveguide 276 is identical to the waveguide 76 except that the protuberances 144A and 144B have been omitted. Likewise, the contacts 148A, 148B and vias 158A and 158B have been omitted. In addition, the thickness of the central rib 142 is then chosen so that the optical signal guided by the laser source 256 is mainly located in the layer 80. For example, to this end, the thickness e142 is smaller than 300 nm or 200 nm. Lastly, in this embodiment, the contact 148C has been replaced by a contact 264 for which the distance dv between this contact 264 and the lower face of the rib 142 is zero. The via 158C has been replaced by a via 266 the structure of which is, for example, identical to that of the via 158A or 158B described above.
The secondary through-via 258 electrically connects the primary via 52 directly to an electrical track 276 produced on the upper face 32. The track 276 comprises a pad to which a first electrical contact of the electronic chip 260 is soldered. The structure of the via 258 is here identical to that of the via 252.
The via 262 electrically directly connects an electrical track 280 produced on the upper face 32 to the metal line 135 embedded in the level 94 of the sublayer 122. To this end, for example, the structure of the via 262 is identical to the structure of the via 252 except that the lower portion 110 is omitted. The electrical track 280 comprises a pad to which a second electrical contact of the electronic chip 260 is soldered.
Here, the metal line 135 is electrically connected to the electrical contact 74 of the optical component 72 by way of an interconnect metal network similar to that described with reference to
Section II.1: Variants of the Vias
The horizontal cross section of a via is not necessarily circular. For example, the cross section of a via may be square or rectangular. In this case, by “diameter” what is meant is the hydraulic diameter of this horizontal cross section.
As illustrated in the case of the secondary via in the preceding embodiments, the diameter of the via is not necessarily constant over its entire height. In this case, by “diameter” of the via, what is meant is the largest diameter of this via along its height.
As a variant, the upper portion 56 of the primary via is omitted. In this case, the upper end of the portion 54 of the primary via lies flush with the bonding interface 40 directly.
In another variant, the metal vias 60 are omitted. In this case, the upper end of the portion 54 makes direct mechanical and electrical contact with the metal line 58.
In another embodiment, the upper portion 56 of a primary via may comprise a plurality of additional metal lines produced at various depths in the sublayer 44. The various levels of metal lines of the portion 56 are then electrically connected together by metal vias. These additional metal lines then form a redistribution layer or RDL that allows, for example, a plurality of metal lines that lie flush with the interface 40 to be electrically connected together.
The interconnect metal network used to form the lower portion of a secondary through-via may also, as a variant, comprise more than three levels of metal lines.
As a variant, the lower portion 110 of a secondary through-via is omitted. In this case, the bottom of the hole 126 then opens directly onto the metal line 58 of the primary via. Such a secondary through-via then comprises only a single portion, namely the portion 112.
The center of the hole 126 may also be filled with a dielectric material different from the dielectric material of the sublayer 122. For example, it is filled with an organic material.
In another embodiment, the upper portion 112 of a secondary via is produced using one or more metal vias that each extend from the face 32 to the metal line 114.
If it is not necessary to relax the constraints to be met with respect to the layout of the optical components with respect to the secondary vias, the diameter of the secondary vias may be larger than 3 μm and, in particular, equal to or larger than the diameter of the primary vias.
As a variant, the distance dv between the contacts 148A, 148B and the protuberances 144A, 144B is non-zero. In this case, the distance dv is for example chosen as described in the case of the contact 148C.
In one particular embodiment, the via 148C is omitted. In this case, the heat from the waveguide 76 is dissipated only by way of the vias 158A and 158B.
Section II.2: Other Variations of the Structure of the Photonic Chip
As a variant, the interconnect layer 36 may be made from materials other than silicon. For example, it may be made of glass.
As another variant, the layer 36 comprises one or more electronic components.
The sublayers of the optical layer 38 may also be made from other materials. In particular, the sublayer 122 may be produced using another dielectric material such as, for example, amorphous silicon, silicon nitride (SiN), or silicon oxynitride (SiON).
The various oxide sublayers of the layer 38 are not necessarily all made from the same oxide. As a variant, the sublayer 120 may be made from an oxide different to that used to make the sublayer 90. It is also possible to use, to produce the sublayer 122, an oxide that is different from that of the sublayer 90.
The optical layer may in addition be passed through by additional vias the diameter of which is larger than 3 μm and, for example, larger than 10 μm or 20 μm. In this case, an exclusion zone of large area is provided around each of these additional vias and no optical component is produced in this exclusion zone. This exclusion zone is dimensioned according to the teaching given in the following article: Yan Yang et al.: “Though—Si—via (TSV), Keep—Out—Zone (KOZ) in SOI photonics interposer: A study of impact of TSV-Induced stress on Si ring resonators”, IEEE Photonics Journal, volume 5, Number 6, December 2013. Below, this article will be referred to as “Yang2013”. However, even if the photonic chip comprises a few of these additional vias, the arrangement of the optical components in the optical layer is still simplified because the presence of the secondary vias still limits the constraints on the layout of the optical components.
Any number of optical components may be buried in level 92 of sublayer 122. In addition, the various optical components are not necessarily buried at the same depth in level 92. This is notably the case if the optical components or portions of optical components are fabricated from various sublayers of material stacked on top of one another. These material sublayers may then be chosen from the group consisting of a crystalline-silicon sublayer, an amorphous-silicon sublayer, a sublayer of silicon nitride (SiN) and a sublayer of silicon oxynitride (SiON).
As a variant, the lower end of one or more of the vias 158A, 158B, 158C and 266 is electrically and mechanically connected to at least one primary via in order to improve the thermal dissipation of the heat produced by the laser source. Thus, in a first embodiment, the lower end of at least one of the vias 158A, 158B and 266 is electrically and mechanically connected to a primary via used to electrically connect one of the components of the chip 18 to the carrier 14. For example:
The diameter of one or more of the vias 158A, 158B, 158C and of the via 266 is not necessarily smaller than 3 μm or 1 μm. Its diameter may also be larger than these values.
The use of the vias 158A, 158B, 158C and 266 to improve the dissipation of the heat produced by a laser source applies similarly to the improvement of the dissipation of the heat produced by any type of optical or electronic component at least one portion of which is embedded in the sublayer 122. For example, a via similar to via 158C or 266 may also be used to dissipate the heat produced by the optical component 72 or the electronic component 210.
Other embodiments of the waveguide 76 are possible. For example, as a variant, the protuberance 144B and the rim 146B are omitted. In this case, the contact 148B and the via 158B are also omitted. The thickness of the protuberances 144A, 144B may be larger than the thickness e142 of the rib 142. As another variant, only the protuberances 148A and 148B are omitted and the rims 146A and 146B are retained. In the latter case, the contacts 148A and 148B may be formed under the rims 146A and 146B, respectively, in the same way as was described for the case where they are formed under the protuberances 148A, 148B.
Section II.3: Other Variants
The photonic chip may comprise, on the upper face 32, an electrical track that is not necessarily used to electrically connect an optical component to one of the primary vias. For example, as described with respect to the embodiment of
In one embodiment, the photonic chip comprises solely secondary blind vias. In this case, all the secondary blind vias are fabricated from the side opposite the handle 164. The order of removal of the handles may then be inverted. Thus, the handle 170 may be removed before the handle 164. Specifically, it is not necessary to fabricate the upper portion 112 of the secondary through-vias. Also in this case, the secondary blind via is entirely fabricated before the layer 38 is bonded to the layer 36.
In another embodiment, the photonic chip comprises solely secondary through-vias.
Alternatively, the chip 260 may correspond to the chip 16 of
Other fabricating processes are also possible.
A portion of the component 70 or of the laser source 256 may be thermally connected to the interconnect layer by way of secondary, and typically blind, vias, such as the vias 158A, 158B, 158C and 266, independently of the other features described here of the photonic chip. In particular, such a thermal connection may be made in a context in which the primary vias pass through the optical layer as described in the article Yang2013 or in a context in which all the primary vias are omitted. When the primary vias are omitted, the layer 36 is no longer an interconnect layer but simply a carrier for the optical layer 38. When the primary vias are omitted, preferably, the thickness of the surface sublayer 44 is sufficiently small to not thermally insulate the secondary vias of the sublayer 42. For example, in this case, the surface sublayer 44 is omitted or its thickness is smaller than 100 nm or 50 nm. If the thickness of the surface sublayer 44 is large, i.e. larger than 100 nm or 300 nm, the upper portion 56 may be retained to form a thermal bridge that passes through the sublayer 44 and only the lower portion 54 of the primary vias is omitted. In this case, the upper portion 56 may be electrically insulated from the silicon sublayer 42 by a thin layer of dielectric material with a thickness of smaller than 100 nm or than 50 nm.
Thermally connecting the waveguide 76 or 276 to the interconnect layer by way of a via the lower end of which makes direct mechanical contact with the layer 36, allows the thermal dissipation of the heat produced by the laser source during its operation to be improved with respect to the embodiments described in the article by J. Durel et al. cited in the introduction of this patent application. In addition, the fabrication of the photonic chip remains particularly simple. In particular, the fact that the vias 158A, 158B, 158C and 266 are formed by interconnect metal networks simplifies the fabrication of these vias and therefore of the photonic chip. The fact that these vias are solely located in the lower interconnect level 94 allows the constraints on the layout of the optical components to be limited with respect to the case where vias that in addition pass through the upper optical-guide level 92 are used to create these thermal bridges between the waveguide 76, 276 and the layer 36.
It is known to the inventors that, normally, the presence of an electrically conductive contact bearing directly against a portion of a waveguide increases the optical losses of this waveguide. However, such an increase in optical losses must, preferably, be limited as much as possible. In this context, combining, in the same embodiment, a thickness e142 of the central rib 142 of smaller than 300 nm and a zero distance dv between the contact 264 and the central rib 142 allows the dissipation of heat to be improved without however substantially increasing optical losses. Specifically, when the thickness e142 is smaller than 300 nm, most of the guided optical signal is confined to the active layer 80. In this case, the fact that the contact 264 bears mechanically directly against the rib 142 increases the optical losses very little.
The fact that the distance between the contact 148C and the lower face of the rib 142 is larger than 10 nm or 50 nm allows optical losses to be limited. In addition, the fact that this distance dv is also smaller than 500 nm allows the dissipation of heat to the layer 36 to be improved.
The use of protuberances 144A, 144B and of vias 158A, 158C the upper end of which bears directly on these protuberances 144A, 144B also allows the dissipation of heat to be improved without however substantially increasing optical losses. Specifically, because of the rims 146A, 146B, the optical signal guided in the protuberances 144A, 144B is negligible. Thus, the fact that these protuberances bear directly against the upper end of the vias 158A, 158B hardly increases the optical losses at all while nonetheless improving the dissipation of heat.
It is known that the presence of through-vias nearby an optical component causes problems. Specifically, such vias are made of an electrically conductive material the coefficient of thermal expansion of which is different from the material from which the optical component is made. Thus, in response to temperature variations, the through-vias exert, on the nearby optical component, a mechanical stress that varies as a function of temperature. This variation in mechanical stress modifies the optical properties of the optical component and causes a modification of the properties of this optical component. For example, when the optical component is an optical filter, this modifies its central wavelength. Such variations in the properties of the optical component must be avoided. In the described embodiments, the primary vias do not exert substantial mechanical stresses on the optical components. Specifically, the interconnect layer 36 is completely devoid of optical component. The optical components are solely located in the optical layer. Therefore, the optical components may be laid out, with respect to one another, without worrying about the arrangement of the primary vias. In particular, it is not necessary to respect any exclusion zone around each primary via. The secondary through-vias that pass through the optical-guide level 92 for their part have a diameter that is preferably smaller than or equal to 3 μm, or even to 1 μm. Such secondary through-vias may thus be placed 4 μm or even 2 μm from the optical components without exerting mechanical stresses on the latter. By comparison, the through-vias described in Yang2013, which had a diameter larger than 10 μm, had to be placed 40 μm from the optical components to not stress them mechanically. It is therefore possible to place the optical components much closer to the secondary vias than what would have been possible if their diameter were equal to the diameter of the primary vias. The combination, in the same photonic chip, of primary vias and secondary vias therefore allows a photonic chip to be obtained that is thick enough to achieve an acceptable bow, while limiting the constraints to be met with respect to the arrangement of the optical components relative to the vias. It will also be noted that the interconnect layer does not necessarily contain a layer of buried silicon oxide. The smallest thickness thereof resulting in an acceptable bow is therefore smaller than that of a layer containing a buried silicon oxide. Decreasing the thickness of the interconnect layer notably allows the parasitic capacitances of the primary vias to be decreased. Specifically, this parasitic capacitance is proportional to the height of the primary vias.
As regards the chips 18, 200 and 250, placing the laser source on the upper face of the oxide layer 90 facilitates its cooling from the outside, notably by virtue of the packaging of the system.
The fact that the center of the hole 126 is filled with an electrically non-conductive material the coefficient of thermal expansion of which is lower than that of the coating 124 allows the mechanical stresses exerted by the secondary via on nearby optical components to be further decreased.
Number | Date | Country | Kind |
---|---|---|---|
1855010 | Jun 2018 | FR | national |
Filing Document | Filing Date | Country | Kind |
---|---|---|---|
PCT/FR2019/051082 | 5/14/2019 | WO |
Publishing Document | Publishing Date | Country | Kind |
---|---|---|---|
WO2019/234323 | 12/12/2019 | WO | A |
Number | Name | Date | Kind |
---|---|---|---|
7639719 | Fang et al. | Dec 2009 | B2 |
20030038344 | Palmer et al. | Feb 2003 | A1 |
20040201095 | Palmer et al. | Oct 2004 | A1 |
20090168821 | Fang et al. | Jul 2009 | A1 |
20090296367 | Sekine et al. | Dec 2009 | A1 |
20110248397 | Coffy et al. | Oct 2011 | A1 |
20120155055 | Kang et al. | Jun 2012 | A1 |
20140264400 | Lipson | Sep 2014 | A1 |
20140376857 | Chantre | Dec 2014 | A1 |
20160005711 | Kang et al. | Jan 2016 | A1 |
20160047986 | Chantre et al. | Feb 2016 | A1 |
20160141467 | Li et al. | May 2016 | A1 |
20170133356 | Mercier | May 2017 | A1 |
20170309593 | Kang et al. | Oct 2017 | A1 |
20180006427 | Sciancalepore | Jan 2018 | A1 |
Number | Date | Country |
---|---|---|
3 023 066 | Jan 2016 | FR |
WO 2018200059 | Nov 2018 | WO |
Entry |
---|
International Search Report dated Oct. 4, 2019 in PCT/FR2019/051082 filed May 14, 2019. |
Durel, J., et al., “First Demonstration of a Back-Side Integrated Heterogeneous Hybrid III-V/Si DRB Lasers for Si-Photonics Applications”, IEEE International Electron Devices Meeting (IEDM), IEEE, XP033055031, 2016, 4 total pages. |
Beilliard, Y., et al., “Chip to wafer copper direct bonding electrical characterization and thermal cycling”, IEEE, 2013, 7 total pages. |
Zhang, C., et al., “Thermal management of Hybrid Silicon Ring Lasers for High Temperature Operation”, IEEE Journal of Selected Topics In Quantum Electronics, vol. 21, No. 6, 2015, 7 total pages. |
Miyairi, K., et al., “Full integration and Electrical Characterization of 3D Silicon Interposer Demonstrator Incorporating High Density TSVs and Interconnects”, 45th International Symposium on Microelectronics, 2012, pp. 000984-000990. |
Shin, D., et al., “O-band DFB laser heterogeneously integrated on a bulk-silicon platform”, Optics Express, vol. 26, No. 11, 2018, 7 total pages. |
Sysak, M., et al., “Hybrid Silicon Laser Technology: A Thermal Perspective”, IEEE Journal of Selected Topics in Quantum Electronics, vol. 17, No. 6, 2011, pp. 1490-1498. |
Yang, Y., et al., “Through-Si-via (TSV) Keep-Out-Zone (KOZ) in SOI Photonics Interposer: A Study of the Impact of TSV-Induced Stress on SI Ring Resonators”, IEEE Photonics Journal, vol. 5, No. 6, XP002788522, 2013, 12 total pages. |
Number | Date | Country | |
---|---|---|---|
20210234334 A1 | Jul 2021 | US |