• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    This method of manufacturing a modulator comprises: after the bonding of a substrate to an encapsulated semiconductor layer containing a first electrode of the modulator and before the production of a second electrode of the modulator, the method comprises the withdrawal (522 ) a support on which is deposited the encapsulated semiconductor layer so as to expose a face of a buried layer of dielectric material, located under the buried semiconductor layer, without changing the thickness of the buried layer more than 5 nm, and - the embodiment (524, 528) of the second electrode is implemented directly on this exposed face of the buried layer so that once the second electrode is made, it is the layer buried which directly forms a dielectric layer interposed between proximal ends of the modulator electrodes.
    公开号:FR3054926A1
    申请号:FR1657626
    申请日:2016-08-08
    公开日:2018-02-09
    发明作者:Sylvie Menezo；Olivier Girard
    申请人:Commissariat a lEnergie Atomique CEA；Commissariat a lEnergie Atomique et aux Energies Alternatives CEA；
    IPC主号:
    专利说明:

    Holder (s): COMMISSIONER OF ATOMIC ENERGY AND ALTERNATIVE ENERGIES Public establishment.
    Extension request (s)
    Agent (s): INNOVATION COMPETENCE GROUP.
    METHOD FOR MANUFACTURING A PROPAGATION LOSS MODULATOR AND THE OPTICAL SIGNAL PROPAGATION INDEX.
    FR 3 054 926 - A1
    This method of manufacturing a modulator comprises:
    - After the bonding of a substrate on an encapsulated semiconductor layer containing a first electrode of the modulator and before the production of a second electrode of the modulator, the method comprises the removal (522) of a support on which the semi layer is deposited conductive encapsulated so as to expose a face of a buried layer of dielectric material, situated under the buried semiconductor layer, without modifying the thickness of the buried layer by more than 5 nm, and
    - The production (524, 528) of the second electrode is implemented directly on this exposed face of the buried layer so that once the second electrode is produced, it is the buried layer which directly forms a dielectric layer interposed between proximal ends of the modulator electrodes.

    METHOD FOR MANUFACTURING A PROPAGATION LOSS MODULATOR AND THE PROPAGATION INDEX OF AN OPTICAL SIGNAL [001] The invention relates to a method for manufacturing a modulator of propagation losses and of the propagation index of a guided optical signal as well as a modulator manufactured by this process.
    By propagation loss, we mean the optical losses seen by the optical mode propagating in a waveguide in which it is guided.
    By propagation index, we mean the effective propagation index of the optical mode propagating in a waveguide in which it is guided.
    Such known modulators include a waveguide formed by stacking immediately on top of each other:
    - a proximal end of a first electrode,
    - a thin dielectric layer, and
    - a proximal end of a second electrode.
    By applying a potential difference between the first and second electrodes, the density of charge carriers at the interfaces between the dielectric layer and the proximal ends of the first and second electrodes is modified. This induces a modification of the propagation losses and of the propagation index seen by the guided optical field and propagating in the waveguide. Typically, the dielectric layer is a layer of silicon dioxide.
    The known methods of manufacturing such a modulator include:
    the supply of a stack comprising successively a support, a buried layer of dielectric material and a semiconductor layer, the thickness of the buried layer being equal to e 2in i to within plus or minus 5 nm, where e 2in i is a constant and then
    the etching of the semiconductor layer to structure a first electrode of the modulator in this semiconductor layer, this first electrode having a proximal end, a distal end and an intermediate part which extends, in a transverse direction, from the proximal end to the distal end to mechanically and electrically connect these ends, then
    - Encapsulation of the structured semiconductor layer in a dielectric material to obtain a semiconductor layer encapsulated in a dielectric material in which the dielectric material extends, in the transverse direction, until it directly touches the proximal end of the first electrode and then
    - bonding of a substrate to the encapsulated semiconductor layer, then
    - the production of a second modulator electrode having a proximal end opposite the proximal end of the first electrode, these proximal ends being only separated from each other by a dielectric layer so as to forming a waveguide capable of guiding the optical signal to be modulated.
    For example, such a manufacturing process is disclosed in application WO2011037686.
    The method of application WO 2011037686, is interesting because it makes it possible to manufacture modulators in which the thickness of the dielectric layer is small, that is to say less than 25 nm. This is advantageous because such a modulator then has both good modulation efficiency and low propagation losses.
    By cons, the implementation of known manufacturing methods described in these applications leads to modulators whose performance is dispersed. By "performance" is meant here in particular the bandwidth of the modulator and the modulation efficiency of this modulator. The modulation efficiency is the ratio between the variation in refractive index induced by volts applied between the two electrodes of the modulator.
    By "performance dispersion" means the fact that the performance of one modulator to another varies, even if all these modulators are manufactured by the same manufacturing process. Such variations are typically caused by inaccuracies during manufacturing which modify, in a random and uncontrolled manner, the capacitance of the capacitor formed by the dielectric layer interposed between the two electrodes of the modulator, and the position of the maximum field intensity. optics guided by the modulator. For example, it may be imprecision on the thickness of the dielectric layer or on the positions of the electrodes relative to each other.
    The state of the art is also known from:
    - US 2015/0055910
    - US8363986B2, and
    - FR2867898.
    The invention aims to provide a manufacturing process which reduces the dispersion of the performance of the modulators manufactured by this process. In other words, the objective is to obtain a manufacturing process which makes the performance of the modulators produced more homogeneous. Its subject is therefore such a process in which:
    - after the substrate has been bonded to the encapsulated semiconductor layer and before the second electrode has been produced, the method includes removing the support so as to expose one face of the buried layer without changing its thickness by more than 5 nm , and
    the production of the second electrode is implemented directly on this exposed face of the buried layer so that once the second electrode has been produced, it is the buried layer which directly forms the dielectric layer interposed between the proximal ends electrodes.
    In the claimed process, the dielectric layer is directly formed by the buried layer constructed before the production of the first electrode. By “directly” formed by a buried layer, is meant the fact that the dielectric layer is obtained without substantially modifying the initial thickness e 2 ini of the buried layer. By the expression “without substantially modifying the initial thickness”, is meant the fact that the initial thickness e 2in i of the buried layer which forms the dielectric layer interposed between the electrodes of the modulator is not increased or decreased further of err p with respect to its initial thickness e 2 ini, where err p is a constant equal to 5 nm or 4 nm and, preferably, equal to 3 nm or 1 nm. In particular, in the claimed process, the dielectric layer is not the result of the formation on the first electrode of a layer of dielectric material. Nor is it the result of the thinning of a thicker dielectric layer. Because of this, the thickness of the dielectric layer is controlled with increased precision compared to known methods of manufacturing modulators. In addition, the thickness of the dielectric layer is more homogeneous. As a result, the dispersion of the capacitor capacitor of the modulator is much lower when implementing the claimed process than with known manufacturing processes. This therefore makes it possible to manufacture modulators whose performances are more homogeneous. In particular, the bandwidth and the modulation efficiency of the modulators manufactured by the claimed process are much less dispersed than if they had been manufactured by a known process such as that described in application WO 2011037686 or US 20150055910.
    In addition, the claimed method allows to encapsulate the first electrode of the modulator in a dielectric material without having to leave an empty space next to this first electrode. This is particularly advantageous because, in known methods such as that of application US 20150055910 where such an empty space exists, this empty space is found under the second electrode. However, an empty space between the second electrode and the support creates a parasitic capacitance which degrades the performance of the modulator.
    The embodiments of this manufacturing process may include one or more of the following characteristics:
    the supply of a stack includes the supply of a stack in which:
    - the support is made of silicon,
    the buried layer is a layer of thermal silicon oxide, that is to say obtained by oxidation of the surface of the support at a temperature above 700 ° C., and
    the step for producing the second electrode comprises: direct bonding of a layer of semiconductor material on the exposed face of the buried layer, then localized etching of this layer of semiconductor material to form the second electrode ;
    the etching of the first electrode includes a localized etching which thins the intermediate part of the first electrode without thinning its proximal end so that this intermediate part is thinner than the proximal end of the first electrode, and the production of the second electrode includes the positioning of this second electrode relative to the proximal end of the first electrode so that the proximal end of the second electrode extends systematically, in the transverse direction, on either side of the proximal end of the first electrode;
    the production of the second electrode includes:
    a localized etching step which thins out an intermediate part of the second electrode, this intermediate part being situated between its distal and proximal ends so that this intermediate part is less thick than the proximal end of the second electrode, and
    the positioning of this second electrode with respect to the proximal end of the first electrode so that the first electrode extends systematically, in the transverse direction, on either side of the proximal end of the second electrode;
    the method includes localized doping of the semiconductor layer at the location of the distal end of the first or second electrode so as to more strongly dop a first region of this proximal end directly in contact with the dielectric layer than a second region further from the dielectric layer, the thickness of the first region being greater than or equal to 70 nm;
    the thickness of the buried layer is less than or equal to 25 nm or 15 nm;
    the etching of the semiconductor layer structures a first waveguide at the same time as the first electrode, then the method also includes the production, on the exposed face of the buried layer and facing the first waveguide, a second waveguide made of III-V material capable of amplifying an optical signal capable of being modulated by the modulator, this second waveguide being coupled to the first waveguide through the buried layer;
    removing the support involves a selective etching operation of at least one thin residual layer of the support directly in contact with the buried layer, using a chemical agent which etches the support at least 500 times faster than the buried layer.
    The embodiments of the claimed manufacturing process can also have one or more of the following advantages:
    - The claimed method makes it possible to use a layer of thermal silicon oxide as a dielectric layer, which improves the bonding of the second electrode and therefore, ultimately, the performance of the modulator.
    - Having a thicker proximal end than the intermediate part of the same electrode makes the process less sensitive to positioning errors of one electrode relative to the other electrode. More precisely, this makes it possible to more finely control the position of the maximum intensity of the optical field guided by the modulator and therefore the efficiency of the manufactured modulator. Thanks to this, the performances of the modulators produced are even more homogeneous.
    - Doping the proximal end more strongly only in a region close to the dielectric layer makes it possible, for equal performance, to reduce the propagation losses of the modulator manufactured with equal access resistances, or to decrease the access resistance at equal optical propagation losses while keeping the distribution of the optical field unchanged.
    - The fact that the thickness of the dielectric layer is less than 25 nm or 15 nm makes it possible to obtain modulators which have good modulation efficiency.
    - The fact of producing, at the same time as the modulator, a laser source which comprises a waveguide of III-V material coupled to an optical waveguide by means of the same dielectric layer also makes it possible to homogenize the characteristics of this laser source without adding an additional step to the manufacturing process of the modulator.
    The invention also relates to a modulator manufactured by the claimed manufacturing process, this modulator comprising:
    - a substrate extending mainly in a plane called “plane of the substrate”,
    a semiconductor layer encapsulated in a dielectric material, this encapsulated semiconductor layer comprising a lower face directly facing the substrate and an upper face turned to the side opposite to the substrate, this encapsulated semiconductor layer also comprising at least a first modulator electrode made of semiconductor material, this first electrode extending along a transverse direction parallel to the plane of the substrate, from a proximal end to a distal end passing through an intermediate part and the material dielectric extending in this transverse direction until it directly touches the proximal end of the first electrode, the proximal and distal ends and the intermediate part flush with the upper face of the encapsulated semiconductor layer,
    a second electrode of semiconductor material having doping of opposite sign to that of the first electrode, this second electrode extending from a proximal end to a distal end passing through an intermediate part, the proximal end being located opposite the proximal end of the first electrode and the distal end being located on the side opposite the distal end of the first electrode with respect to a plane perpendicular to the transverse direction and passing through the ends proximal,
    a dielectric layer interposed between the proximal ends of the first and second electrodes, the superposition of the proximal ends and of the dielectric layer forming a waveguide capable of guiding the optical signal to be modulated,
    contact points directly in mechanical and electrical contact with, respectively, the distal ends of the first and second electrodes for electrically connecting these electrodes to different electrical potentials so as to modify the density of the charge carriers in the waveguide, in which at any point of the dielectric layer, its thickness is equal to e 2in i to within 10 nm, where e 2in i is a constant.
    The embodiments of the modulator may include one or more of the following characteristics:
    the layer of dielectric material is a layer of thermal oxide obtained by oxidation of a substrate at more than 700 ° C;
    the proximal end of one of the first and second electrodes, called the first proximal end, is thicker than the intermediate part of this first end, and the proximal end of the other of the first and second electrodes, called the second proximal end , extends, over a distance of at least 5 nm, in the transverse direction on either side of the first proximal end.
    The invention will be better understood on reading the description which follows, given solely by way of nonlimiting example and made with reference to the drawings in which:
    - Figure 1 is a schematic illustration of a transmitter in vertical section;
    - Figure 2 is a schematic illustration, in top view, of a modulator and a phase tuning device of the transmitter of Figure 1;
    FIG. 3 is an enlarged illustration, in vertical section, of the modulator of the transmitter of FIG. 1,
    - Figure 4 is a flow diagram of a method of manufacturing the transmitter of Figure 1;
    - Figures 5 to 16 are schematic illustrations, in vertical section, of different manufacturing states obtained during the implementation of the method of Figure 4;
    - Figures 17 to 21 are schematic illustrations, in vertical section, of other possible embodiments of the modulator of Figure 3;
    - Figure 22 is a schematic illustration, in vertical section, of another embodiment of the transmitter of Figure 1.
    In these figures, the same references are used to designate the same elements. In the remainder of this description, the characteristics and functions well known to those skilled in the art are not described in detail.
    In this text, the expressions “the layer is made of material M”, “layer of material M” or “layer M” denote a layer in which the material M represents at least 90%, and preferably at least 95% or 99% of the mass of this layer.
    Figure 1 shows a transmitter 5 of an optical signal modulated in phase and / or in amplitude to transmit bits of information to a receiver via an optical fiber for example. To this end, the transmitter 5 includes a laser source 7 which emits an optical signal whose phase and / or amplitude is then modulated by a system 6 for phase modulation and / or amplitude of this optical signal.
    For example, the wavelength A Li of the optical signal emitted by the laser source 7 is between 1250 nm and 1590 nm.
    Typically, the laser source 7 is a DBR ("distributed bragg reflector laser") or DFB ("distributed feedback laser") laser. Such a laser source is well known and only the details necessary for understanding the invention are described here. For example, for general details and the operation of such a laser source, the reader can refer to the following articles:
    - Xiankai Sun and Amnon. Yariv: “Engineering supermode silicon / lll — V hybrid waveguides for laser oscillation”, Vol. 25, No. 6 / June 2008 / Journal of the Optical. Society of America B
    - B. Ben Bakir et al., “Hybrid Si / lll-Vlasers with adiabatic coupling”, 2011.
    - B. Ben Bakir, C. Sciancalepore, A. Descos, H. Duprez, D. Bordel, L. Sanchez, C. Jany, K. Hassan, P. Brianceau, V. Carron, and S. Menezo, “Heterogeneously Integrated Ili-V on Silicon Lasers ”, Invited Talk ECS 2014.
    To simplify Figure 1 and the following figures, only a hybrid laser waveguide 200, 220 and a surface coupler 8 (known by the English term "Surface grating coupler") of the laser source 7 are shown.
    Such a coupler 8 is for example described in the following article: F. Van Laere,
    G. Roelkens, J. Schrauwen, D. Taillaert, P. Dumon, W. Bogaerts, D. Van Thourhout and R. Baets, “Compact grating couplers between optical fibers and Silicon-onInsulator photonic wire waveguides with 69% coupling efficiency”. It is produced in an encapsulated semiconductor layer 3. By design, the coupler can emit up or down. It is in this description reversed, but one can choose by design to make it emit up or down. Layer 3 here comprises structured monocrystalline silicon encapsulated in a dielectric material 116. Generally, a dielectric material has an electrical conductivity at 20 ° C of less than 10 7 S / m and preferably less than 10 9 S / m or 10 '15 S / m. In addition, in the case of the dielectric material 116, its refractive index is strictly lower than the refractive index of silicon. For example, in this embodiment, the dielectric material 116 is silicon dioxide (SiO 2 ).
    The layer 3 extends horizontally and directly on a rigid substrate 44. In the layer 3, the monocrystalline silicon is located in the same horizontal plane parallel to the plane of the substrate 44. Here, the monocrystalline silicon of the layer 3 is also mechanically and electrically isolated from the substrate 44 by a thickness of the dielectric material 116. Typically, the maximum thickness of monocrystalline silicon in the layer 3 is between 100 nm and 800 nm. In this example, the maximum thickness of the monocrystalline silicon in the layer 3 is equal to 500 nm.
    In Figure 1 and the following figures, the horizontal is represented by X and Y directions of an orthogonal coordinate system. The direction Z of this orthogonal coordinate system represents the vertical direction. Subsequently, terms such as "upper", "lower", "above", "below", "top" and "bottom" are defined with respect to this direction Z. The terms "left" and " right ”are defined with respect to direction X. The terms“ front ”and“ rear ”are defined with respect to direction Y.
    Figure 1 shows the elements of the transmitter 5 in section in a vertical plane parallel to the directions X and Z.
    The substrate 44 extends horizontally. It is formed by a successive stack of a support 441 and a layer 442 of dielectric material. The thickness of the support 441 is typically greater than 80 µm or 400 µm. For example, the support 441 is a silicon support. Layer 442 is made of silicon dioxide. The thickness of layer 442 is typically greater than 500 nm or 1 μm or more.
    The hybrid laser waveguide 200, 220 consists of a waveguide 200 made of a III-V gain material and a waveguide 220 of monocrystalline silicon. Generally the waveguide 200 is used to generate and amplify an optical signal inside an optical cavity of the laser source 7. Here, for this purpose, it is produced in a layer 36 comprising a gain material III -V encapsulated in a dielectric material 117. For example, the material 117 is silicon dioxide or silicon nitride. This layer 36 extends horizontally directly over a dielectric layer 20. The layer 20 itself extends horizontally directly over an upper face of the layer 3.
    The thickness of the layer 20 is typically between 5 nm and 25 nm and, preferably, between 10 nm and 25 nm. Here, the thickness of the layer 20 is equal to 20 nm.
    The layer 36 typically comprises a doped lower sublayer 30, a stack 34 of quantum wells or quantum dots in quaternary form and an upper sublayer 35 doped with a dopant of sign opposite to that of the sublayer 30. The sublayers 30 and 35 are here made of doped InP.
    In Figure 1, only a strip 33, a stack 233 and a strip 234 produced, respectively, in the sublayer 30, the stack 34 and the sublayer 35 are shown. This superposition of the strip 33, the stack 233 and the strip 234 constitutes the waveguide 200.
    The waveguide 200 also includes:
    contact sockets 243G and 243D in mechanical and electrical contact directly with the strip 33 and located, respectively, to the left and to the right of the stack 233, and
    - a contact socket 244 in mechanical and electrical contact directly with the strip 234.
    ίο These sockets 243G, 243D and 244 make it possible to inject an electric current into the stack 233 between the sockets 243G, 243D and the socket 244.
    The waveguide 220 is produced in the monocrystalline silicon of the layer 3. This waveguide 220 extends under the strip 33. Typically, its thickness and its width vary according to Y as described in the article. previously quoted from Ben Bakir and. A1. In FIG. 1, the waveguide 220 is shown, by way of illustration, in the particular case where the direction of propagation of the optical signal inside this waveguide is parallel to the direction Y For example, for this purpose, the waveguide 220 adopts a configuration known by the English term "rib". Thus, the cross section of this waveguide, parallel to the plane XZ, has a central rib 222 from which extend on each side, parallel to the direction X, of the side arms 223G and 223D thinner. Here, the waveguide 220 is only separated from the strip 33 by a portion of the layer 20. For example, the waveguide 220 is optically connected to the waveguide 200 by an adiabatic or evanescent coupling. For a detailed description of an adiabatic coupling, the reader can refer to the previously cited article by X. Sun and A. Yariv or to the following article: Amnon Yariv et al., “Supermode Si / lll-V hybrid Lasers , optical amplifiers and modulators: proposai and analysis ”Optics Express 9147, vol. 14, No. 15, 07/23/2007.
    The characteristics of the optical coupling between the waveguide 220 and the waveguide 200 depend in particular on the dimensions of the waveguide 220 and, in particular, on the thickness of the central rib 222. It is therefore important that the the thickness of this rib 222 can be adjusted independently of the dimensions of the other photonic components produced on the same substrate 44. For example, here, the thickness of the rib 222 is equal to the maximum thickness of the monocrystalline silicon in the layer 3, that is to say here at 500 nm.
    The system 6 may be a system for modulating the phase alone, or the amplitude alone or simultaneously of the phase and the amplitude. To modulate the phase or the amplitude of the optical signal, the system 6 comprises at least one modulator of the propagation losses and of the propagation index of a guided optical signal and, often, at least one phase tuning device . For example, the system 6 is a Mach-Zehnder interferometer in which the modulator 100 and the phase tuning device 300 are arranged in one of the branches of this interferometer to modulate the amplitude and / or the phase of the signal optic generated by the laser source 7. The structure of a Mach-Zehnder interferometer is well known and is not described here. Therefore, to simplify Figure 1, only a modulator 100 and a phase tuning device 300 are shown.
    The device 300 makes it possible to adjust the phase of an optical signal propagating parallel to the direction Y inside a wave guide 320. For example, the wave guide 320 is longer in the direction Y as wide in the direction X. The waveguide 320 is produced in the monocrystalline silicon of the layer 3. Here, its thickness is for example equal to the thickness of the bulge 222. The refractive index of the silicon varies greatly depending on the temperature. Thus, by varying the temperature of the waveguide 320, it is possible to modify the propagation index of the optical signal in this waveguide and therefore to adjust the phase of the optical signal. To this end, the device 300 comprises two heaters 322G and 322D each arranged on a respective side of the waveguide 320. Here, the heater 322D is deduced from the heater 322G by symmetry with respect to a vertical plane parallel to the directions Y and Z and passing through the middle of the waveguide 320. Thus, only the heater 322G will now be described in more detail with reference to FIGS. 1 and 2.
    The heater 322G has an arm 324 which extends, parallel to the direction X, from a proximal end 56 to a distal end 58. The arm 324 also extends parallel to the direction Y. The arm 324 is produced in monocrystalline silicon of layer 3.
    The proximal end 56 is in direct mechanical contact with the waveguide 320. Here, the proximal end 56 touches a vertical flank of the waveguide 320. For this purpose, the arm 324 and the guide d wave 320 forms a single block of matter. The thickness of the proximal end 56 is less than the maximum thickness of the waveguide 320 so as to confine the optical signal in the waveguide 320. For example, the thickness of the end proximal 56 is 1.5 times or twice or three times or four times smaller than the maximum thickness of the waveguide 320. The distal end 58 is doped to make the monocrystalline silicon resistive and form a resistance electric which forms only one block of material with the waveguide 320. In FIG. 1, the doped regions of monocrystalline silicon are finely hatched and appear in dark. The shortest distance between this doped region of the arm 324 and the waveguide 320 is, for example, strictly greater than 200 nm or 400 nm.
    To circulate an electric current inside the distal end 58, the heater 322G also has two contact sockets 51G and 52G in mechanical and electrical contact directly with the distal end 58. Here, these sockets 51G and 52G are located one behind the other in the Y direction and at each end of the distal end 58 in this Y direction. The contacts of the heater 322D, visible in FIG. 2, bear, respectively, the references 51D and 52D.
    When a current, brought by the sockets 51G and 52G, passes through the distal end 58, this transforms part of the electrical energy thus received into heat which is propagated, by thermal conduction through the end proximal 56, to the waveguide 320. Thus, the heater 322G makes it possible to heat the waveguide 320 without any resistive element being implanted in the waveguide 320 or in the immediate vicinity of this waveguide. 'wave. The device 300 makes it possible to slowly adjust the phase of the optical signal in the waveguide 320. On the other hand, it does not allow the phase of the optical signal to be varied quickly.
    Conversely, the modulator 100 makes it possible to quickly modify the phase of the optical signal. To this end, it has two electrodes 120 and 130. These electrodes 120 and 130 are also visible, in top view, in FIG. 2.
    The electrode 120 is produced in the monocrystalline silicon of the layer 3. It extends, in the direction X, from a proximal end 12 to a distal end 11 passing through a thinned intermediate part 13. It also extends in direction Y. In direction Y its cross section remains constant. Parallel to the XZ plane, the cross sections of the ends 11 and 12 and of the intermediate part 13 are each rectangular. The ends 11, 12 and the intermediate part 13 are flush with the flat upper face of the layer 3. The ends 11, 12 and the intermediate part 13 are therefore directly in contact with the lower face of the layer 20.
    The intermediate part 13 mechanically and electrically connects the ends 11 and 12 to each other. Its thickness e i3 is chosen so as to confine laterally (section XZ) the distribution of intensity of the optical field in the proximal end 12. For this purpose, the thickness e i3 is less than 0.8e i2 and, preferably less than 0.5e i2 or 0.25ei 2 , where e i2 is the maximum thickness of the end 12. The thickness e i3 is typically also greater than 70 nm to reduce the electrical resistance between the ends 11 and 12 This resistance is called "access resistance". For this purpose, the thickness e i3 is often greater than 0, the i2 or 0.15ei 2 . Here, the thickness at of the distal end 11 is equal to the thickness e i2 . In this embodiment, the thickness e i2 is equal to 300 nm and the thickness e i3 is equal to 150 nm or 100 nm. The horizontal lower face of the intermediate part 13 is separated from the substrate 44 only by the dielectric material 116.
    Here, the distal end 11 is more doped than the proximal end 12. For example, the concentration of dopant in the end 11 is between 10 17 and 2 x 10 19 atoms / cm 3 . The concentration of dopant in the end 12 is for example between 10 17 and 2 x 10 18 atoms / cm 3 .
    The electrode 130 is made of semiconductor material doped with doping of sign opposite to that of the electrode 120. Here, it is made of InP in the sublayer 30. The dopant concentration of the electrode 130 is, for example, between 10 17 and 2 x 10 18 atoms / cm 3 or between 10 17 and 5 x 10 18 atoms / cm 3 .
    The electrode 130 extends, parallel to the direction X, from a proximal end 32 to a distal end 31. The electrode 130 also extends in the direction Y. It is directly located on the layer 20. Parallel to the XZ plane, its cross section is rectangular. In the Y direction, this cross section is constant.
    The proximal end 32 is located opposite the proximal end 12 and extends beyond this end 12, in the direction X, so as to have an overhang 32d (Figure 3) facing the intermediate part 13. Typically, the protrusion 32d is at least 5 nm or 10 nm or 25 nm long in the direction X. The proximal end 32 is separated from the proximal end 12 and the intermediate part 13 only by a portion of the layer 20 interposed between these proximal ends.
    Relative to a vertical plane parallel to directions Y and Z and passing through the ends 12 and 32, the distal end 31 is located on one side of this plane while the distal end 11 is located from the other side. The ends 11 and 31 are therefore not opposite.
    In the embodiment of Figure 1 and the following, the area 34, which extends vertically from the end 31 to the substrate 44, comprises only solid dielectric materials. Here, these are the dielectric materials 116 and the layer 20. In FIG. 1, the area 34 has been highlighted by filling it with circles. However, there is no discontinuity between the dielectric materials situated inside the zone 34 and those situated outside this zone 34.
    The superposition, in the direction Z, of the end 12, of a portion of the layer 20 and of the end 32 is dimensioned to form a waveguide 70, capable of guiding, in the direction Y , the optical signal generated by the laser source 7. The waveguides 70 and 320 are for example optically connected to each other by an adiabatic coupler not shown here.
    The maximum thickness of the proximal ends 12 and 32 is chosen so that the point M, where the maximum intensity of the optical field of the optical signal which propagates in the waveguide 70 is located, is the closest layer 20 possible. Preferably, point M is located in the center of the portion of this layer 20 interposed between the ends 12 and 32. In fact, it is at the interfaces between the ends 12, 32 and the layer 20 that the density of charge carriers is maximum when a potential difference is present between these proximal ends. Thus, by placing the point M there, the efficiency of the modulator 100 is improved. The maximum thickness e 32 of the proximal end 32 is generally between 50 nm and 300 nm. In this embodiment, the refractive indices of the ends 12 and 32 are close to each other. Consequently, the maximum thicknesses of the ends 12 and 32 are chosen to be substantially equal so that the point M is situated inside the layer 20. For example, the maximum thickness e i2 of the proximal end 12 is between 0 , 5th 32 and l, 5th 32 , and preferably between 0.7e 32 and l, 3rd 32 . Here, the thicknesses e i2 and e 32 are each chosen to be equal to 300 nm.
    The intermediate part 13 allows better control of the position of point M in the direction X and therefore limits the dispersion of the performance of the modulators 100 during their manufacture. More precisely, the position of the point M in the direction X is essentially fixed by the width W i2 (FIG. 3) of the proximal end 12 in the direction X. In fact, the small thickness e i3 of the intermediate part 13 confines the optical signal inside the end 12. The width W i2 is, by etching, defined to within +/- δ, where δ is an error equal, typically, to +/- 5 nm, or +/- 10 nm. Conversely, if the intermediate part 13 has the same thickness as the end 12, the width W ï2 is defined by the width of the overlap of the electrodes 130 and 120. However, the positioning of the electrode 130 is done in turn by lithographic alignment, with an accuracy typically equal to +/- 25 nm, or +/- 50 nm. Therefore, in the absence of a thinned intermediate part, the error over the width W i2 is +/- ôal and the performance dispersion of the modulators 100 is greater. This conformation of the electrode 120 therefore makes it possible to be less sensitive to positioning errors of the electrode 130. In particular, the modulation efficiency depends directly on the width W i2 . Consequently, the modulation efficiency is less dispersed thanks to the intermediate part 13 and the overshoot 32d.
    Furthermore, this embodiment also allows better control of the capacity of the modulator.
    The modulator 100 also includes two contact sockets 21 and 22, in mechanical and electrical contact directly with, respectively, the distal ends 11 and 31. These sockets 21 and 22 are connected to a voltage source controllable according to the bit or information bits to be transmitted by the transmitter 5.
    A possible operation of the transmitter 5 is as follows. The laser source 7 generates an optical signal. At least part of this optical signal is directed to a Mach Zehnder interferometer, at least one of the branches of which successively comprises the modulator 100 and the phase tuning device 300. This part of the optical signal is therefore successively guided by the waveguide 70 then the waveguide 320 before being recombined with another part of the optical signal guided by the other branch of the Mach Zehnder interferometer to form the modulated optical signal. For example, the waveguides 70 and 320 are optically coupled to each other by an adiabatic coupler. At the output of the Mach Zehnder interferometer, the optical field can be coupled to an optical fiber via a waveguide similar to waveguide 320, then by the surface coupler 8. [0061] A method of manufacturing the transmitter 100 will now be described with reference to FIGS. 4 to 16. FIGS. 5 to 16 show different states of manufacture of the transmitter 5 in vertical section parallel to the directions X and Z. [0062] During a step 500, the process begins with the supply of a substrate 4 (Figure 5). Here, this substrate 4 is an SOI (“Silicon on insulator”) substrate. The substrate 4 has directly stacked one above the other in the direction Z:
    a silicon support 1, of thickness greater than 400 μm or 700 μm conventionally,
    a layer 2 buried in thermal silicon dioxide of thickness e 2in i, and
    a layer 43 of monocrystalline silicon which, at this stage, has not yet been etched or encapsulated in a dielectric material.
    Thermal silicon dioxide is a silicon oxide obtained by oxidation of the support 1 at a high temperature, that is to say greater than 650 ° C or 800 ° C. Thanks to the nature of this oxide, layer 2 has two remarkable properties:
    1) its thickness, even when it is small, remains uniform, and
    2) it makes it possible to obtain a direct bonding of better quality.
    The “homogeneous” thickness of layer 2 means that at any point in layer 2 its thickness is between e 2in i -err 2ini nm and e 2in i + err 2ini nm, where:
    - e 2i ni is a constant, typically equal to the average thickness of layer 2, and
    - err 2in i is a constant less than or equal to 5 nm and, preferably, equal to 3 nm or 1 nm.
    Direct bonding is better known by the English term "direct bonding". This is a bonding process in which two wafers or "wafers" in English are glued directly to each other without adding an intermediate layer of glue. Bonding results from the appearance of chemical bonds directly between the two faces of these wafers. Generally, after being brought into mechanical contact with each other, the wafers undergo a heat treatment to strengthen the bonding.
    Generally, the thickness ea ™ is greater than 7 nm or 10 nm and, typically, less than 100 nm or 50 nm. Here, the initial thickness e 2 ini of layer 2 is equal to 20 nm at +/- 1 nm and the thickness of layer 43 is equal to 500 nm.
    During a step 502, a localized doping of the layer 43 is carried out. Here, a first localized doping operation 504 is first carried out, during which doped regions 506 (FIG. 6) likewise doping are carried out in layer 43. These regions 506 are only produced at the locations of the future arms of the tuning device 300 and of the electrode 120 of the modulator 100. These regions 506 have a doping equal to that of the distal end 58 and from the proximal end 12.
    Then, a second operation 508 of doping the layer 43 is carried out so as to obtain a region 510 (FIG. 7) more heavily doped than the regions 506. The region 510 is here partially superimposed on one of the regions 506 For example, the region 510 is obtained by applying a new implantation on a part of one of the regions 506. The region 510 is produced at the location of the future distal end 11 of the electrode 120. The doping of the region 510 is here equal to the doping of the distal end 11.
    During a step 514, the layer 43 undergoes a first partial localized etching (FIG. 8) to thin the thickness of the silicon at the locations of the electrode 120 and of the arms 324 of the heaters 322G and 322D. At the end of step 514, the regions 506 and 510 are thinned and have a thickness less than the initial thickness of the layer 43. Here, the thickness of the thinned regions 506 and 510 is equal to the thickness of the electrode 120 and arms 324, that is to say 300 nm.
    During this first partial localized etching, the thickness of the layer 43 is also thinned in undoped regions, for example, to form the patterns of the future surface coupler 8 and the lateral arms 223G and 223D of the guide wave 220. Conversely, during this step 514, other regions known as “non-thinned”, are not etched and retain their initial thickness. In particular, these non-thinned regions are located at the location of the rib 222 of the waveguide 220 and at the location of the waveguide 320.
    Still during this step 514, the layer 43 then undergoes a second partial localized etching in order to thin the thickness of the silicon only at the location of the future intermediate part 13. At the end of this second partial localized etching, the thickness of the layer 43 at the location of the future intermediate part 13 is 150 nm.
    During a step 516, a total localized etching of the layer 43 is carried out (FIG. 9). Unlike partial etching, total etching completely eliminates the thickness of silicon from layer 43 in the unmasked regions where it is applied. Conversely, masked regions protect layer 43 from this total etching. This total etching is carried out so as to structure, simultaneously in the layer 43, the waveguides 220 and 320, the arms of the tuning device 300, the surface coupler 8 and the electrode 120. For this purpose, only the regions corresponding to these different elements are hidden. At the end of this step, the state shown in FIG. 9 is obtained.
    During a step 518, the layer 43 of monocrystalline silicon, which was structured during the preceding steps, is encapsulated in silicon dioxide 116 (FIG. 10). We then obtain the layer 3 comprising structured monocrystalline silicon encapsulated in the dielectric material 116. The upper face of the material 116 is then prepared for bonding, for example direct or molecular. For example, the upper face of the material 116 is polished using a method such as a CMP (Chemical-Mechanical Polishing) method.
    During a step 520, the upper face of the substrate 4, that is to say at this stage the polished face of the material 116, is then bonded to the outer face of the substrate 44 (FIG. 11), by example, by direct or molecular bonding. The substrate 44 has already been described with reference to FIG. 1.
    During a step 522 (FIG. 12), the support 1 is removed to expose one face of the layer 2 and this without substantially modifying the initial thickness ea ™ of the layer 2. For this, the support 1 is eliminated by two successive operations. A first coarse removal operation of the majority of the thickness of the support 1 so as to leave only a thin residual layer of the support 1 on the layer 2. Typically, at the end of the first operation, the thickness of this residual layer is less than 40 µm or 30 µm. This first operation is said to be “coarse” because the precision on the thickness of the thin residual layer is coarse, that is to say greater than ± 0.5 μm or ± 1 μm but, generally, nevertheless less than ± 4 pm and preferably less than ± 3 pm. Since the precision required on the thickness of the residual thin layer is coarse, a quick or inexpensive removal method can be used. Typically, the coarse elimination operation is an operation of thinning the support 1 by mechanical polishing. Here, at the end of this first operation, the thin residual layer is 20 μm thick to within ± 2 μm.
    Then, a second finishing operation is implemented to completely eliminate the thin residual layer of the support 1 without modifying the thickness of the layer 2. Typically, the second operation is a very selective chemical etching operation. By "very selective" is meant the fact that the chemical agent used during this operation etches at least 500 times and, preferably at least 1000 or 2000 times, the support 1 more quickly than layer 2. Here, the agent chemical used is TMAH (Tetramethylammonium hydroxide). TMAH etches silicon 2000 times faster than thermal silicon oxide. In this embodiment, to be sure of removing all of the thin residual layer, the very selective etching is suitable for etching 22 μm of silicon, ie 2 μm more than the theoretical thickness of the thin residual layer. This choice leads, in the worst case, to an over-etching of 2 nm (= 2 pm / 2000) in the thickness of layer 2. In fact, since the thickness of the residual thin layer is 20 pm ± 2 pm, this means that in places the thickness can be 18 pm. If we plan to engrave 22 pm, then the over-etching at the place where the thickness is 18 pm can reach 4 pm. So the layer 2 over-etching at this location is 2 nm. Thus, at the end of step 522, the layer 2 is exposed and forms the layer 20. Its thickness is equal to 20 nm to ± 3 nm, that is to say ± 1 nm due to the imprecision on the initial thickness e 2 inhibits layer 2, to which is added the imprecision of ± 2 nm due to over-etching. Compared to other process for obtaining a thin layer, this has the advantage of obtaining a thin layer whose thickness is much more homogeneous. Indeed, when the thin layer is obtained by thinning a thicker oxide layer or by growing a thin oxide layer on one side of an encapsulated structured layer (see for example WO2011037686 or US2015055910), l The thickness is much less homogeneous. Typically, with known methods, the thickness of the oxide layer is controlled, at best, to within ± 10 nm or ± 20 nm.
    Because of this, the dispersion of the performance of the modulators manufactured according to this method is much lower than that obtained with the known methods.
    Finally, advantageously, degassing wells are dug in the layer 20 outside the locations where the electrode 130 and the strip 33 are to be made. Typically, these wells pass vertically right through the layer 20. They have for the function of trapping the gaseous elements generated during the direct bonding of a layer on the layer 20. Thus, these wells make it possible to obtain a better quality bonding on the layer 20. To simplify the figures, these wells have not shown in these figures.
    At the end of step 522, a stack of the substrate 44 and of the layers 3 and 20 is obtained (FIG. 12).
    In a step 524, a layer 36A (FIG. 13) of gain material III-V is produced on the layer 20. For example, the layer 36A is bonded to the layer 20 above the waveguide 220 and of the electrode 120. The layer 36A comprises the sub 3054926 layer 30 in InP doped with doping of sign opposite to that of the electrode 120, the stack 34 and the sublayer 35.
    During a step 526, a localized etching (FIG. 14) of the sublayer 35 and of the stack 34 is carried out to structure the strip 234 in the sublayer 35 and the stack 233 in the stack 34. During this step, the sublayer 30 is not etched.
    During a step 528, a total localized etching (FIG. 15) of the underlay 30 is carried out to simultaneously structure the strip 33 and the electrode 130 in this underlay. The precise positioning of the electrode 130 relative to the electrode 120 depends on the machines and techniques used to perform this step. This ôal precision is therefore known in advance. The length of the protrusion 32d of the electrode 130 depends on this accuracy. Here, the desired position of the electrode 130 is chosen so that the target length of this protrusion 32d is greater than or equal to the absolute value of the precision ôal. Thus, whatever the positioning error which occurs during the manufacture of the modulator 100, the overshoot 32d is systematically created as long as the alignment error remains in the foreseeable range ± ôal.
    In step 528, part or all of the thickness of the layer 20 located between the electrode 130 and the strip 33 can be eliminated. However, this has no effect on the thickness of the portions of the layer 20 interposed between the electrodes 120 and 130 and between the waveguide 220 and the strip 33.
    In a step 530, the structured layer 36A is encapsulated (FIG. 16) in the dielectric material 117. This then gives the layer 36 comprising the gain material III-V encapsulated in the dielectric material 117.
    Finally, during a step 532, the contacts 21, 22, 51G, 52G, 51D, 52D, 243G and 243D are made. We then obtain the transmitter 5 as shown in FIG. 1.
    This manufacturing process has many advantages. In particular :
    - It allows to precisely control the thickness of layer 20 and to obtain a particularly flat layer 20 because it is made on the side of layer 3 which has the same level everywhere, which simplifies the bonding of layer 36A.
    - It allows the thickness of the electrode 120 to be adjusted independently of the thickness of the waveguide 220 and, more generally, independently of the thickness of the layer 43 of monocrystalline silicon. This is particularly useful because, generally, to improve the operation of the laser source 7, the thickness of the waveguide 220 must be quite large, that is to say here of the order of 500 nm and that the thickness of the strip 33 is fairly thin, that is to say here of the order of 300 nm or 150 nm. Conversely, to improve the operation of the modulator 100, as explained above, the thickness of the electrode 120 and, in particular of its proximal end 12, must be chosen as a function of the thickness of the end proximal 32. Here, the thickness of the proximal end 32 is imposed by the thickness of the sublayer 30 in crystalline InP. It is therefore 300 nm or 150nm.
    - This process does not complicate the manufacture of the transmitter 5. For example, it makes it possible to carry out in a single etching operation, the strip 33 of the waveguide 200 and the electrode 130 of the modulator 100. Similarly, the The electrode 120 and the waveguide 220 are produced simultaneously during the same etching operation.
    FIG. 17 represents a modulator 550 able to replace the modulator 100. The modulator 550 is identical to the modulator 100 except that the electrode 120 is replaced by an electrode 552. The electrode 552 is identical to the electrode 120 except that the distal end 11 is replaced by a distal end 554 whose thickness is equal to the thickness e i3 of the intermediate part 13. Thus, the distal end 554 and the intermediate part 13 are in the extension one on the other and form only a single plate of rectangular cross section.
    FIG. 18 represents a modulator 560 capable of replacing the modulator 100. The modulator 560 is identical to the modulator 550 except that the electrode 552 is replaced by an electrode 562. The electrode 562 is identical to the electrode 552 except that the proximal end 12 is replaced by a proximal end 564. The proximal end 564 is identical to the proximal end 12 except that it has a more doped region 566 and a less doped region 568 stacked one above on the other in the direction Z. The region 566 is directly in contact with the layer 20. The region 568 is located on the side opposite to the layer 20. The doping of the region 566 is the same as that described for the end proximal 12 so as to maintain the same modulation efficiency. The term “doping” of region 566 or 568 denotes the average volume density of dopants in this region. To limit the access resistance while limiting the optical losses inside the waveguide, the thickness e 5 66 of the region 566 is preferably equal to the thickness e i3 of the intermediate part 13 to ± 10% close or ± 5% close. Its thickness is also generally greater than 70 nm. Here, its thickness is equal to the thickness e i3 .
    The doping of the region 568 is at least twice less, and preferably 4 or 10 times less, than the doping of the region 566. Typically, the region 568 is not doped or very little doped.
    The thickness of the region 568 is equal to e i2 - e 5 66. This arrangement of doping of the proximal end 564 makes it possible to reduce the propagation losses in the modulator 560 without substantially modifying the other performances such as modulation efficiency and modulation speed. It also does not modify the resistance of access to the proximal end 564. Such an arrangement of the doping of the proximal end 564 is, for example, carried out during step 502, that is to say in doping at the location of the proximal electrode 564 in the silicon layer 43 such that only the region 566 is doped. For example, regions 566 and 568 are obtained by varying the implantation energy of the dopant and the implanted dopant dose to adjust both the density of dopants and the depth at which the maximum density of dopants is located.
    FIG. 19 represents a modulator 570 capable of replacing the modulator 100. The modulator 570 is identical to the modulator 100 except that the electrode 120 is replaced by an electrode 572. The electrode 572 is identical to the electrode 120 except that the intermediate part 13 is replaced by an intermediate part 574. The thickness of the intermediate part 574 is equal to the thickness e i2 . In this embodiment, the homogenization of the performances of the modulator 570 is therefore only obtained thanks to better control of the thickness of the layer 20.
    FIG. 20 represents a modulator 580 capable of replacing the modulator 100. This modulator 580 is identical to the modulator 100 except that the electrode 130 is replaced by an electrode 582. The electrode 582 is identical to the electrode 130 except that a thinned intermediate part 584 is introduced between the proximal 32 and distal 31 ends. The intermediate part 584 is, for example, structurally identical to the intermediate part 13.
    In addition, in this embodiment, the proximal end 32 is less doped than the distal end 31. Such a different doping between the ends 31 and 32 can be obtained by carrying out a doping step located on the end 31 just after step 528 and before step 530.
    In this embodiment, as in the previous embodiments, the distal end 32 includes the protrusion 32d which is located above the intermediate part 13 of the electrode 120. Under these conditions, the position of the maximum d the intensity of the optical field in the direction X is always controlled by the width W 2 of the proximal end 12.
    FIG. 21 represents a modulator 590 capable of replacing the modulator 100. The modulator 590 is identical to the modulator 580 except that the electrode 120 is replaced by an electrode 592. The electrode 592 is identical to the electrode 572 of the modulator 570. The electrode 592 has a protrusion 12d which extends, in the direction X, below the intermediate part 584. This protrusion 12d is shaped like the protrusion 32d. Thus, in this embodiment, the position of the maximum intensity of the optical field guided by the modulator is controlled by the width W 32 of the proximal electrode 32 and no longer by the width of the proximal end 12. This provides the same advantages in terms of homogenizing the performance of the modulators manufactured as that which has already been explained in the case of the embodiment of FIG. 1 and of the proximal end 12.
    The method of manufacturing the modulator 590 is for example identical to that of FIG. 4 except that:
    during step 514, the second localized partial etching intended to thin the intermediate part 13 is omitted, and
    - During step 528, a second localized partial etching capable of thinning the intermediate part 584 is implemented in addition to the total localized etching.
    FIG. 22 represents a transmitter 600 identical to the transmitter 5 except that an encapsulated semiconductor layer 602 is interposed between layer 20 and layer 36. Layer 602 comprises a structured semiconductor layer 604 encapsulated in l 'silicon oxide. The layer 604 is directly in contact with the layer 20. Here, the layer 604 is a layer of polycrystalline silicon. This layer is structured, for example by localized total etchings such as those previously described, to form an electrode 608 of the modulator and a rib 610 of the laser source. The electrode 608 is for example identical to the electrode 130 except that it is made of polycrystalline silicon. The electrode 608 is doped by implantation after etching. The rib 610 is not doped.
    The rib 610 is located above the wave guide 220 and optically coupled to this wave guide 220 through the layer 20 to form a bi-material wave guide 612. Here, the guide bi-material wave 612 is formed from monocrystalline silicon and polycrystalline silicon.
    The waveguide 200 of III-V material is directly deposited or bonded to the layer 602 above the bi-material waveguide 612 and optically coupled to this waveguide 612. In this mode of embodiment, the electrode 608 is not made of the same material as that of the strip 33 of the waveguide 200.
    Variants of the modulator:
    The modulator 100 can be a ring modulator. To this end, the waveguide 70 closes on itself to form an annular waveguide in which the density of the charge carriers can be modified as a function of the difference in potentials applied between the taps 21 and 22. Typically, this annular waveguide is connected to a waveguide in which the optical signal to be modulated is propagated by an evanescent coupling. In this case, the tuning device
    300 of phase can be omitted. The waveguide 70 can also constitute only a limited portion of the annular waveguide.
    In another embodiment, the modulator is used to modulate the intensity of the optical signal passing through it. Indeed, a change in the density of the charge carriers in the waveguide 70 also changes the intensity of the optical signal passing through it.
    Alternatively, the thickness of the end 11 is equal to the thickness of the layer 43 of monocrystalline silicon. Indeed, to center the point M, where the maximum of the optical field intensity of the optical signal is located, at the center of the layer 20, it is the thickness of the ends 12 and 32 which is important. The thickness of the distal ends 11 and 31 is of no significant importance on this point.
    The thickness of the layer 20 can be greater than 25 nm or 40 nm.
    Similar to what has been described for the proximal end 564, the proximal end 32 of the electrode 130 can be replaced by one end with a more doped region than another. In another variant, only the end 32 has two different doping regions and the doping of the end 12 is uniform.
    Other embodiments of regions 566 and 568 of different doping are possible. For example, in a variant, the doping of the proximal end 564 decreases progressively as one moves away from the layer 20. This creates a doping gradient. There is then no clear modification of the volume density of dopants when passing from region 566 to region 568. On the other hand, the average density of dopants in region 566 remains much higher than the average density dopants in region 568.
    The doped region of the electrode 120 may extend beyond the proximal end 32 in the X direction or below.
    As a variant, the width W 32 of the proximal end 32 of the modulator 580 is smaller than the width W i2 of the proximal end 12. In this case, the position of the maximum intensity of the optical field at l the interior of the waveguide 70 is controlled by the width W 32 and no longer by the width W i2 .
    Other semiconductor materials are possible for producing the electrode 120 or 130. For example, the two electrodes are made of InP or of polycrystalline or monocrystalline silicon.
    Other dielectric materials are possible for the material 116 and the layer 20. For example, it may be silicon nitride, aluminum nitride, an electrically insulating polymer, AI 2 O 3 . In addition, in the case of layer 20, its refractive index is not necessarily lower than that of silicon.
    In another embodiment, the electrode 130 is made of a semiconductor material different from that used to make the strip 33. In this case, the electrode 130 and the strip 33 are not structured in a even underlay of lll-V material.
    Whatever the embodiment, it is possible to invert the N and P doped regions.
    Variants of the laser source:
    Other gain materials III-V are possible for producing the layer 36. For example, the layer 36 is formed from the following stack going from the bottom to the top:
    - a lower N-doped GaAs sublayer,
    - AIGaAs quantum dot sublayers, or AIGaAs quantum wells, and
    - an upper P-doped GaAs sublayer
    The lll-V material used to make the sub-layer 30 may be different. For example, it may be N or P doped AsGa. It should also be noted that P doped ΙΊηΡ has more optical loss than N doped ΙΊηΡ, and that it is therefore preferable to use at the level of the modulator for the electrode. 130 of N doped ΙΊηΡ The waveguide 220 can take a so-called “strip” configuration, that is to say that the lateral arms 223G and 223D are omitted or any other configuration capable of guiding an optical signal.
    In another variant, the layer 20 is completely eliminated where it is not essential for the operation of the transmitter. For example, it is completely eliminated except between the proximal ends 12 and 32.
    Variants of the manufacturing process:
    The withdrawal of the support 1 can be achieved differently. For example, as a variant, the support 1 is removed by implementing only the finishing operation without implementing the coarse elimination operation. In another variant, the coarse elimination is carried out using a coarse etching operation different from that implemented during the finishing operation.
    The degassing wells dug in the layer 20 can be omitted, especially if the layer 20 is thicker. Indeed, if the layer 20 is thicker then the use of degassing wells is unnecessary.
    Alternatively, the electrode 130 and the strip 33 are not produced at the same time in the same sublayer 30. For example, during step 528, only the strip 33 is structured. Then, a semiconductor layer is deposited or glued on the layer 20 at the location of the future electrode 130. Then, it is etched to obtain the electrode 130. In this case, the electrode 130 can be produced in a material different from that used for strip 33 such as crystalline silicon.
    The order of the partial and total etching steps can be reversed. For example, a first mask is placed on the layer 43 to delimit the periphery of the electrode 120, then a total localized etching is carried out to construct the vertical sides of this electrode 120. Then, a localized partial etching is implemented to thin the intermediate part 13 of the electrode 120. During this localized partial etching, a second mask covering at least the proximal end of the electrode 120 is deposited. This second mask leaves the intermediate part 13 bare.
    In another variant, the second localized total etching is replaced by a uniform etching of the entire surface of the layer 3 to transform the unthinned regions into thinned regions and completely eliminate the thinned regions.
    The order of the doping and etching steps can be reversed.
    The modulator and the laser source can be manufactured independently of one another. For example, the manufacturing methods described here can be easily adapted to manufacture either only a modulator or only a laser source.
    Other variants:
    Layer 442 can be made of materials other than silicon oxide. For example, in an advantageous variant, the layer 442 is made of aluminum nitride (AIN) which improves the dissipation of the heat generated by the laser source 200 towards the substrate 441.
    As a variant, part or all of the contact connections are made, not through the material 117, but through the substrate 44. In this case, with respect to what has been shown in the preceding figures, one or more electrical contact points emerge under the substrate.
    Alternatively, the waveguide 70, 220 or 320 is curved. In this case, the shape of the different elements optically coupled to these waveguides is adapted to the radius of curvature of these waveguides.
    As a variant, the phase tuning device is omitted or produced differently. The fact that a thicker proximal end than the intermediate part makes the manufacturing process more robust with regard to electrode positioning errors can also be used to improve other methods of manufacturing modulators. In particular, this can be implemented in methods other than those where the dielectric layer is directly formed by the buried layer. In particular, the production of a thicker proximal end than the intermediate part can also be implemented in known methods such as that described in applications WO 2011037686 or US2015 / 0055910. In the latter case, the intermediate part is thinned during the structuring of the electrode and before it is encapsulated in silicon oxide.
    Similarly, the greater doping of the region 566 of the proximal end directly in contact with the dielectric layer 20 can also be implemented independently of the fact that the dielectric layer is directly formed by the buried layer.
    权利要求:
    Claims (11)
    [1" id="c-fr-0001]
    1. Method for manufacturing a propagation loss modulator and the propagation index of a guided optical signal, this method comprising:
    - The supply (500) of a stack comprising successively a support (1), a buried layer (2) of dielectric material and a semiconductor layer (43), the thickness of the buried layer (2) being equal to e 2in i to within plus or minus 5 nm, where e 2 ini is a constant, then
    - The etching (514, 516) of the semiconductor layer (43) to structure a first electrode (120) of the modulator in this semiconductor layer (43), this first electrode (120) having a proximal end (12) , a distal end (11) and an intermediate part (13) which extends, in a transverse direction (X), from the proximal end (12) to the distal end (11) to connect mechanically and electrically these ends (11,12), then
    - Encapsulation (518) of the structured semiconductor layer (43) in a dielectric material (116) to obtain a semiconductor layer (3) encapsulated in a dielectric material in which the dielectric material extends, in the transverse direction, until directly touching the proximal end of the first electrode, then
    - bonding (520) of a substrate (44) on the encapsulated semiconductor layer (3), then
    - The production (524, 528) of a second electrode (130) of the modulator having a proximal end (32) facing the proximal end (12) of the first electrode (120), these proximal ends (12, 32) being only separated from each other by a dielectric layer (20) so as to form a waveguide (70) capable of guiding the optical signal to be modulated, characterized in that:
    - after bonding of the substrate (44) on the encapsulated semiconductor layer (3) and before the production of the second electrode (130), the method comprises the removal (522) of the support (1) so as to expose one face of the buried layer (2) without modifying its thickness by more than 5 nm, and
    - The production (524, 528) of the second electrode (130) is implemented directly on this exposed face of the buried layer (2) so that once the second electrode (130) is produced, it is the buried layer (2) which directly forms the dielectric layer (20) interposed between the proximal ends (12, 32) of the electrodes.
    [2" id="c-fr-0002]
    2. Method according to claim 1, in which:
    - the supply (500) of a stack includes the supply of a stack in which:
    The support (1) is made of silicon, the buried layer (2) is a layer of thermal silicon oxide, that is to say obtained by oxidation of the surface of the support at a temperature above 700 ° C., and
    the step of producing the second electrode comprises:
    • direct bonding (524) of a layer (30) of semiconductor material on the exposed face of the buried layer (2), then • localized etching (528) of this layer (30) of semi-material -conductor to form the second electrode (130).
    [3" id="c-fr-0003]
    3. Method according to any one of the preceding claims, in which:
    - The etching of the first electrode (120) includes a localized etching (514) which thins the intermediate part (13) of the first electrode without thinning its proximal end (12) so that this intermediate part is less thick than the end proximal of the first electrode, and
    - The production of the second electrode (130) includes the positioning of this second electrode relative to the proximal end (12) of the first electrode so that the proximal end (32) of the second electrode extends systematically, in the transverse direction, on either side of the proximal end (12) of the first electrode.
    [4" id="c-fr-0004]
    4. Method according to any one of claims 1 to 2, in which the embodiment (524, 528) of the second electrode comprises:
    - a localized etching step which thins an intermediate part (584) of the second electrode (582), this intermediate part (584) being located between its distal (31) and proximal (32) ends so that this intermediate part is less thicker than the proximal end (32) of the second electrode, and
    - The positioning of this second electrode (582) relative to the proximal end (12) of the first electrode (592) so that the first electrode (592) extends systematically, in the transverse direction, from and on the other side of the proximal end (32) of the second electrode (582).
    [5" id="c-fr-0005]
    5. Method according to any one of the preceding claims, in which the method comprises localized doping of the semiconductor layer (43) at the location of the distal end (564) of the first (562) or of the second electrode so as to more strongly dop a first region (566) of this proximal end directly in contact with the dielectric layer (20) than a second region (568) further from the dielectric layer (20), the thickness of the first region (566) being greater than or equal to 70 nm.
    [6" id="c-fr-0006]
    6. The method of claim 1, wherein the thickness of the buried layer (2) is less than or equal to 25 nm or 15 nm.
    [7" id="c-fr-0007]
    7. Method according to any one of the preceding claims, in which:
    - The etching (514, 516) of the semiconductor layer (43) structures a first waveguide (220) at the same time as the first electrode (120), then
    - The method further comprises the realization (524, 526, 528), on the exposed face of the buried layer (2) and opposite the first waveguide (220), a second waveguide (200) of III-V material capable of amplifying an optical signal capable of being modulated by the modulator (100), this second waveguide being coupled to the first waveguide through the buried layer ( 2).
    [8" id="c-fr-0008]
    8. Method according to any one of the preceding claims, in which the withdrawal (522) of the support (1) comprises a selective etching operation of at least one thin residual layer of the support (1) directly in contact with the buried layer. (2), using a chemical agent which etches at least 500 times faster the support (1) than the buried layer (2).
    [9" id="c-fr-0009]
    9. Modulator (100; 550; 560; 570; 580; 590) of the propagation losses and of the propagation index of an optical signal, manufactured by a manufacturing process according to claim 1, this modulator comprising:
    a substrate (44) extending mainly in a plane called "plane of the substrate",
    a semiconductor layer (3) encapsulated in a dielectric material (116), this encapsulated semiconductor layer comprising a lower face directly facing the substrate (44) and an upper face facing the side opposite the substrate (44), this encapsulated semiconductor layer (3) also comprising at least a first modulator electrode (120) produced in the semiconductor material, this first electrode (120) extending along a transverse direction parallel to the plane of the substrate, from a proximal end (12) to a distal end (11) passing through an intermediate part (13) and the dielectric material extending, in this transverse direction, until it directly touches the proximal end of the first electrode, the proximal (12) and distal (11) ends and the intermediate part (13) flush with the upper face of the encapsulated semiconductor layer (3),
    - a second electrode (130) of semiconductor material having doping of sign opposite to that of the first electrode (120), this second electrode extending from a proximal end (32) to a distal end (31) passing through an intermediate part, the proximal end (32) being located opposite the proximal end (12) of the first electrode and the distal end (31) being located on the side opposite the distal end (11) of the first electrode with respect to a plane perpendicular to the transverse direction and passing through the proximal ends,
    - a dielectric layer (20) interposed between the proximal ends (12, 32) of the first and second electrodes, the superposition of the proximal ends (12, 32) and of the dielectric layer (20) forming a waveguide (70) able to guide the optical signal to be modulated,
    - contact sockets (21, 22) directly in mechanical and electrical contact with, respectively, the distal ends (11, 31) of the first and second electrodes for electrically connecting these electrodes to different electrical potentials so as to modify the density of the charge carriers in the waveguide, characterized in that at any point of the dielectric layer (20), its thickness is equal to e 2 ini to within approximately 10 nm, where e 2 ini is a constant.
    [10" id="c-fr-0010]
    10. The modulator as claimed in claim 9, in which the layer of dielectric material is a layer of thermal oxide obtained by oxidation of a substrate at more than 700 ° C.
    [11" id="c-fr-0011]
    11. Modulator according to claim 9 or 10, in which:
    the proximal end (12) of one of the first and second electrodes, called the first proximal end, is thicker than the intermediate part (13) of this first end, and
    - the proximal end (32) of the other of the first and second electrodes, called the second proximal end, extends, over a distance of at least 5 nm, in the transverse direction on either side of the first proximal end (12).
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3206079B1|2018-07-11|Photonic transmitter
    EP3497512B1|2020-06-03|Method for producing a modulator of propagation and propagation index losses of an optical signal
    EP2988378B1|2017-04-26|Laser device and method for manufacturing such a laser device
    EP0539298B1|1996-07-24|Integrated electro-optic modulator and method of making the same
    EP3190672B1|2018-04-18|Semiconductor laser source
    EP3540878B1|2020-10-14|Photonic device including a laser optically connected to a silicon waveguide and method for manufacturing such a photonic device
    EP3168946B1|2019-08-28|Laser device and method for manufacturing such a laser device
    EP3599684B1|2021-01-13|Photonic transmitter
    EP3716345B1|2021-03-31|Method for manufacturing a photonic chip comprising an sacm-apd photodiode optically coupled with an integrated waveguide
    EP3822698B1|2022-01-12|Photonic transmitter
    FR2867898A1|2005-09-23|FABRICATION OF AN OPTICAL INTERCONNECTION LAYER ON AN ELECTRONIC CIRCUIT
    FR3053538A1|2018-01-05|LASER SOURCE WITH SEMICONDUCTOR
    WO2000065393A1|2000-11-02|Photonic integrated circuit comprising a resonant optical component and methods for making same
    EP3767376B1|2021-09-29|Capacitive optical modulator
    EP1918745A1|2008-05-07|Process for manufacturing a slot waveguide
    JP7016858B2|2022-02-07|Manufacturing method of modulator of propagation loss and propagation index of optical signal
    EP3772145A1|2021-02-03|Hybrid laser source comprising a waveguide built into an intermediate bragg network
    FR3109242A1|2021-10-15|A method of manufacturing a mixed layer comprising a silicon waveguide and a silicon nitride waveguide
    FR3098312A1|2021-01-08|active semiconductor component, passive silicon-based component, assembly of said components and method of coupling between waveguides
    FR3034879A1|2016-10-14|WAVEGUIDE FOR ELECTRO-OPTICAL MODULATOR OF CAPACITIVE TYPE.
    同族专利:
    公开号 | 公开日
    US20200301178A1|2020-09-24|
    JP2019525251A|2019-09-05|
    EP3497512A1|2019-06-19|
    FR3054926B1|2018-10-12|
    US11048106B2|2021-06-29|
    WO2018029414A1|2018-02-15|
    EP3497512B1|2020-06-03|
    EP3734353A1|2020-11-04|
    US20190243167A1|2019-08-08|
    US10705354B2|2020-07-07|
    EP3734353B1|2021-07-07|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    WO2005091057A1|2004-03-18|2005-09-29|Honeywell International Inc.|Bonded thin-film structures for optical modulators and methods of manufacture|
    US20110073989A1|2009-09-25|2011-03-31|Haisheng Rong|Optical modulator utilizing wafer bonding technology|
    WO2013155378A1|2012-04-13|2013-10-17|Skorpios Technologies, Inc.|Hybrid optical modulator|
    US20150055910A1|2012-04-30|2015-02-26|Hewlett-Packard Development Company, L.P.|Hybrid mos optical modulator|
    WO2015194002A1|2014-06-19|2015-12-23|株式会社日立製作所|Optical modulator and method for manufacturing same|EP3822698A1|2019-11-18|2021-05-19|Commissariat à l'Energie Atomique et aux Energies Alternatives|Photonic transmitter|JP4828018B2|2000-11-06|2011-11-30|三菱電機株式会社|Optical modulator, method for manufacturing the same, and optical semiconductor device|
    FR2867898B1|2004-03-17|2006-09-15|Commissariat Energie Atomique|FABRICATION OF AN OPTICAL INTERCONNECTION LAYER ON AN ELECTRONIC CIRCUIT|
    US8363986B2|2010-03-10|2013-01-29|Mark Webster|Dopant profile control for high speed silicon-based optical modulators|
    FR2967831B1|2010-11-18|2013-07-19|Commissariat Energie Atomique|HETEROGENE HIGH EFFICIENCY LASER AND PROCESS FOR PRODUCING THE LASER|
    US8620164B2|2011-01-20|2013-12-31|Intel Corporation|Hybrid III-V silicon laser formed by direct bonding|
    FR3007589B1|2013-06-24|2015-07-24|St Microelectronics Crolles 2|PHOTONIC INTEGRATED CIRCUIT AND METHOD OF MANUFACTURE|
    FR3024910A1|2014-08-18|2016-02-19|St Microelectronics Crolles 2|METHOD FOR MANUFACTURING A PHOTONIC INTEGRATED CIRCUIT OPTICALLY COUPLED TO A LASER IN A MATERIAN III-V|
    FR3047811B1|2016-02-12|2018-03-16|Commissariat A L'energie Atomique Et Aux Energies Alternatives|MODULATOR OF PROPAGATION LOSSES AND OF THE PROPAGATION INDEX OF A GUIDE OPTICAL SIGNAL|
    FR3054926B1|2016-08-08|2018-10-12|Commissariat A L'energie Atomique Et Aux Energies Alternatives|METHOD FOR MANUFACTURING PROPAGATION LOSS MODULATOR AND PROPAGATION INDEX OF OPTICAL SIGNAL|FR3054926B1|2016-08-08|2018-10-12|Commissariat A L'energie Atomique Et Aux Energies Alternatives|METHOD FOR MANUFACTURING PROPAGATION LOSS MODULATOR AND PROPAGATION INDEX OF OPTICAL SIGNAL|
    FR3084174B1|2018-07-23|2020-06-26|Commissariat A L'energie Atomique Et Aux Energies Alternatives|PHOTON TRANSMITTER|
    US10852476B2|2019-02-26|2020-12-01|Taiwan Semiconductor Manufacturing Company, Ltd.|Semiconductor package, integrated optical communication system and manufacturing method of integrated optical communication system|
    CN111596473B|2020-05-22|2021-02-12|联合微电子中心有限责任公司|Method of manufacturing semiconductor device, and semiconductor integrated circuit|
    法律状态:
    2017-08-31| PLFP| Fee payment|Year of fee payment: 2 |
    2018-02-09| PLSC| Search report ready|Effective date: 20180209 |
    2018-08-30| PLFP| Fee payment|Year of fee payment: 3 |
    2019-08-30| PLFP| Fee payment|Year of fee payment: 4 |
    2020-08-31| PLFP| Fee payment|Year of fee payment: 5 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1657626A|FR3054926B1|2016-08-08|2016-08-08|METHOD FOR MANUFACTURING PROPAGATION LOSS MODULATOR AND PROPAGATION INDEX OF OPTICAL SIGNAL|
    FR1657626|2016-08-08|FR1657626A| FR3054926B1|2016-08-08|2016-08-08|METHOD FOR MANUFACTURING PROPAGATION LOSS MODULATOR AND PROPAGATION INDEX OF OPTICAL SIGNAL|
    EP17754415.2A| EP3497512B1|2016-08-08|2017-07-27|Method for producing a modulator of propagation and propagation index losses of an optical signal|
    US16/319,902| US10705354B2|2016-08-08|2017-07-27|Method of fabricating a modulator of the propagation losses and of the index of propagation of an optical signal|
    JP2019506421A| JP7016858B2|2016-08-08|2017-07-27|Manufacturing method of modulator of propagation loss and propagation index of optical signal|
    PCT/FR2017/052100| WO2018029414A1|2016-08-08|2017-07-27|Method for producing a modulator of propagation and propagation index losses of an optical signal|
    EP20174609.6A| EP3734353B1|2016-08-08|2017-07-27|Modulator of propagation losses and of the propagation index of an optical signal|
    US16/889,860| US11048106B2|2016-08-08|2020-06-02|Method of fabricating a modulator of the propagation losses and of the index of propagation of an optical signal|
    [返回顶部]