Integrated silicon test structure for the characterization of the PDL of a fiber / silicon optical coupler with two-dimensional diffraction grating

Abstract

A silicon-integrated test structure for characterizing the PDL of a two-dimensional diffraction grating (2DGC) "1 to 2" type fiber / silicon optical coupler, comprises the 2DGC to be tested (61) configured as an output coupler, mirror-mounted with another "1 to 2" type fiber / silicon optical coupler configured as an input coupler, via planar waveguides (64,65) coupling their respective guided optical terminals. An adjustable phase shifter (63) is arranged at one of these waveguides (65). A p or s polarized optical signal is inserted via the input coupler (62). The adjustable phase shifter is driven to apply a pure phase shift between 0 and π to the optical signal propagating in this waveguide (65). The input signal of the 2DGC under test (61) then scans all mixed polarization states. The variation of the optical transmission in the 2DGC under test (61) during this scan gives the PDL.Figure to be published with the abstract: Figure 6A.

Description

Title of the invention: Integrated test structure on silicon for the characterization of the PDL of a fiber / silicon optical coupler with two-dimensional diffraction grating Technical field of the invention

[0001] The present invention relates generally to photonic integrated circuits (or PICs) intended to be coupled to one or more optical fibers, and more particularly relates to an optical coupler with a diffraction grating for achieving optical coupling between the optical fiber(s) and such a photonic integrated circuit.

[0002] The invention finds applications, in particular, in the optical telecommunications sector and in data transmission networks (in English "Data transmission network") on single-mode or multi-mode optical fiber, for example for high-speed links in passive optical networks (PON, from the English "Passive Optical Network") such as FTTH networks (from the English "Fiber-to-the-Home"). Technological background

[0003] Fiber optic transmission systems use data links in which each fiber link includes a transmitter at one end of a fiber and a receiver at the other end of the fiber. Because fiber optics are relatively inexpensive, most systems operate in full duplex mode by transmitting in one direction on one fiber and in the opposite direction on another fiber. However, some systems, such as FTTH passive optical networks (PON), use bidirectional transmission on a single fiber, and for this purpose use transmitter-receiver components called "transceivers", which include an optical coupling device, or grating optical coupler ("Grating" in English).

[0004] The invention is more particularly concerned with optical couplers used to achieve optical coupling between the optical fiber(s) and photonic circuits, when the grating optical coupler is integrated with the photonic circuits on the same substrate, for example a silicon (Si) substrate. More particularly, the type of coupler in question is a two-dimensional grating coupler, often called either 2DGC (from the English "two-dimensional Grating Coupler") or PSGC (from the English "Polarization Splitting Grating Coupler"). A 2DGC is a polarization diversity fiber coupler. This component is typically (but not exclusively) used in a transceiver, in the receiving circuit where the light arrives via an optical fiber with an arbitrary polarization state (and which can vary as a function of time), and must be coupled to the photonic circuit waveguides (produced using semiconductor circuit manufacturing technologies, for example on silicon substrate).

[0005] The ability of a fiber optic system to correctly transmit data is generally assessed from the bit error rate (BER) for the transmitted binary information. BER is the inverse of the signal-to-noise ratio (SNR): for example, a high BER means a low SNR, and vice versa. BER ultimately depends on the optical power of the optical signal received at the receiver. In both cases, received optical power that is too high or received optical power that is not high enough will result in high bit error rates. Because in the first case, the receiving amplifier saturates, and in the second case, noise becomes a problem because it interferes with the useful signal.

[0006] The optical power at the receiver depends on two basic factors: how much power is sent into the fiber by the transmitter and how much is lost through attenuation in the fiber optic cable(s) installation that connects the transmitter to the receiver. Therefore, all manufacturers of data link components must specify the receiver sensitivity of their components and the minimum power coupled into the fiber from the source (this may be a minimum power requirement), which any data link system designer or manufacturer must be aware of, as well as the test conditions. For data link components, these include, but are not limited to, the data input frequency or bit rate and duty cycle, the supply voltages, and the type of fiber coupled to the source.

[0007] In summary, both for the development and for the operational deployment of a transceiver model, it is important to accurately characterize the insertion losses (or IL, from the English "Insertion Loss") as well as the variation of the insertion losses as a function of the incident polarization (PDL, from the English "Polarization Dependent Loss").

[0008] The present invention provides a test device which makes it possible to characterize the IL and PDL of a 2DGC type coupler in a more practical and precise manner than known methods.

[0009] The most widespread method for measuring the PDL of an optical component is to measure its transmission as a function of time by scanning all polarization states. This method, applied to a 2DGC is the one used in the scientific article by F. Van Laere et al., "Efficient Polarization Diversity Grating Couplers in Bonded InP -Membrane" IEEE Photonics Technol. Lett., vol. 20, no. 4, pp. 318-320, Feb. 2008, doi: 10.1109 / LPT.2007.915587. It has the advantage that at no time is it necessary to know the incident polarization state on the device to be tested. In the case of a 2DGC is the sum of the intensity of the two outputs that must be considered. The test structure typically consists of a 2DGC at the input, on the one hand, and either an identical 2DGC or any other type of fiber coupler or photodiodes integrated within the photonic circuit at the output, on the other hand. This method, although simple in principle, has many drawbacks in practice. In particular: • polarization must be scanned for each wavelength point. A spectrum (typically between 10 and 10,000 wavelength points) of IL and PDL can take minutes per component, which is not compatible with large-scale testing; and, • it must be ensured that the injected power does not vary according to the polarization state and the wavelength.

[0010] A second characterization method is known from the scientific article by Mekis et al., “A Grating-Coupler-Enabled CMOS Photonics Platform” IEEE J. Sel. Top. Quantum Electron., vol. 17, no. 3, pp. 597-608, May 2011, doi: 10.1109 / JSTQE.2010.2086049. This depends on the symmetry of the component which makes the incident light can be considered as a linear superposition of 5-polarized light (electric field perpendicular to the plane of incidence) and p-polarized light (electric field parallel to the plane of incidence). These two polarization states represent two extreme cases, which for each wavelength, determine the total range of transmission values ​​as a function of polarization (the PDL). Thus, the PDL of the component can be evaluated by measuring two transmission spectra, for the incident polarizations, respectively 5 and p. In principle, therefore, this method solves the problem of measurement duration.However, as with the first method, the accuracy with which the PDL can be evaluated depends on the equality of the power of the injected light at p and s, as well as the purity of the incident polarization. In practice, it is difficult to implement a polarization switch that is satisfactory in this respect. Summary of the invention

[0011] The invention aims to propose an alternative to the test methods presented above, which overcomes all or part of the aforementioned drawbacks.

[0012] This aim is achieved, according to a first aspect of the invention, by means of a silicon-integrated test structure for characterizing the polarization-dependent loss, PDL, of a "1 to 2" type fiber / silicon optical coupler with a two-dimensional diffraction grating, 2DGC, having a fiber optic terminal as well as a first guided optical terminal and a second guided optical terminal, said test structure comprising: • a first planar waveguide having two ends;

[0013]

[0014]

[0015]

[0016]

[0017] • a second planar waveguide having two ends; • an input fiber / silicon optical coupler having a fiber optic terminal, adapted to receive an input optical signal via said fiber optic terminal, and associated optical splitting means which are adapted to split said input optical signal into two components of equal respective optical intensities, and which are arranged to insert each of said components into the first planar waveguide and into the second planar waveguide, respectively, at a respective first end of said planar waveguides; and, • the 2DGC under test, with its two guided optical terminals which are coupled to a second end of the first planar waveguide and the second planar waveguide, respectively, and configured to deliver an output optical signal on its fiber optic terminal. The test structure further includes: • at least one first adjustable phase shifter which is arranged at the level of the first planar waveguide between the input optical coupler and the 2DGC under test, and which is configured to be controlled so as to apply a phase shift determined between 0 and ji, as a function of a phase shift adjustment signal, to the optical signal propagating in said first planar waveguide, said phase shift being a pure phase shift, i.e. without variation in optical intensity. The invention exploits the fact that the intensity of the light propagating in respective planar waveguides at each of the outputs of the 2DGC is identical when exciting the input coupling grating with purely p-polarized or purely s-polarized light, but with a phase shift of the propagation in the two guides which is either 0° or 180° (values ​​0 and ji, respectively, in radians) for an excitation electric field E polarized entirely in 5 and entirely in p, respectively, and especially on the observation that this relationship between the output phase and the input polarization is a continuous and reciprocal relationship. Some preferred but not limiting aspects of this device are as follows. In embodiments, the input optical coupler and its associated optical splitting means may comprise a "1 to 1" type fiber / silicon optical coupler followed by a single-input, dual-output optical power splitter, said outputs each being coupled to one of the first ends of the first planar waveguide and the second planar waveguide, respectively. For example, the "1 to 1" type fiber / silicon optical coupler can be a one-dimensional diffraction grating coupler, 1DGC. In other embodiments, the input optical coupler and its means of associated optical splitters may include a second 2DGC, identical to the 2DGC under test, and having its guided optical terminals which are each coupled to the other of the ends of the first planar waveguide and the second planar waveguide, respectively.

[0018] In embodiments, the test structure further comprises a second adjustable phase shifter, structurally identical to the first adjustable phase shifter, which is arranged at the second planar waveguide between the input optical coupler and the 2DGC under test, and which is configured such that at all times it does not apply any phase shift to the optical signal propagating in said second waveguide.

[0019] In embodiments, the optical coupler under test and the input optical coupler may be arranged symmetrically with respect to each other, such that an input optical fiber may be connected to the fiber optic terminal of the input optical coupler and an output optical fiber may be connected to the fiber optic terminal of the optical coupler under test, with said input optical fiber and said output optical fiber extending opposite each other.

[0020] In other embodiments, the optical coupler under test and the input optical coupler are arranged in a block such that an input optical fiber can be connected to the fiber optic terminal of the input optical coupler and an output optical fiber can be connected to the fiber optic terminal of the optical coupler under test, said input optical fiber and said output optical fiber being optical fibers of the same fiber network.

[0021] The first phase modulator can advantageously be a thermo-optical phase shifter.

[0022] In a second aspect, the invention also relates to a method for characterizing the polarization dependent loss, PDL, of a fiber / silicon optical coupler of the "1 to 2" type with a two-dimensional diffraction grating, 2DGC, using a test structure integrated on silicon according to the first aspect above, said method comprising the following steps: • inserting an input optical signal into the fiber optic terminal of the input optical coupler; • measuring and recording values ​​of the transmission through the test structure as a function of the value of the phase shift applied to the optical signal propagating in the first planar waveguide by the adjustable phase shifter, while the phase shift adjustment signal of said phase shifter is varied so that said phase shift sweeps the interval [0;ir]; • determination of the maximum and minimum values ​​of the measured and recorded transmission values; and, • obtaining the PDL as the difference between said maximum value and said minimum value.

[0023] A third and final aspect of the invention relates to a computer program product comprising one or more sequences of instructions stored on a memory medium readable by a machine comprising a processor, said sequences of instructions being adapted to carry out all the steps of the method according to the second aspect of the invention when the program is read from the memory medium and executed by the processor. Presentation of the drawings

[0024] Other characteristics and advantages of the invention will become apparent from reading the description which follows. This description is purely illustrative and must be read in conjunction with the appended drawings in which: • [Fig.l] is a schematic representation of a use case of 2DGC couplers for optical data transmission through an optical fiber; • [Fig.2] is a simplified diagram of an example of a 2DGC coupler according to a simplified isometric perspective view; • [Fig.3] shows simplified schematic representations of three classical test structures for measuring the PDL of a 2DGC; • [Fig.4] is an Illustration of the method of measuring the PDL of a 2DGC with a classic test structure of [Fig.3]; • [Fig.5A] and [Fig.5B] are illustrations of the reciprocal relationship between the polarization of the input light, on the one hand, and the phase shift between the two outputs, on the other hand, of a 2DGC type fiber / silicon coupler; • [Fig.ôA] and [Fig.ôB] are simplified diagrams of two variants of the proposed test structure, suitable for use with separate input and output fibers; • [Fig.7A] and [Fig.7B] are simplified diagrams of two variants of the test structures of [Fig.7A] and [Fig.7B], respectively, suitable for use with input and output fibers that are arranged in a unitary block (of the "fiber array" type); • [Fig.8A] and [Fig.8B] are schematic and functional representations illustrating the method of measuring the PDL of a 2DGC with the proposed test structure, and with test equipment suitable for implementing the method; • [Fig.9] is a simplified diagram of another variant of the test structure with an optical power divider at the input; and, • [Fig. 10] is a step diagram illustrating the implementation of the steps of the method of the invention. Description of the embodiments

[0025] In the following description of embodiments and in the figures of the attached drawings, the same or similar elements bear the same reference signs.

[0026] In wave (or vibrational) physics, electromagnetic ground is a model used to study electromagnetic radiation. It is important to distinguish between electromagnetic radiation, which is the phenomenon studied, and electromagnetic ground, which is one of the representations of the phenomenon. Another representation, namely the corpuscular (or quantum) representation, takes into account the existence of the photon, and is not considered here.

[0027] Like all waves, an electromagnetic wave can be analyzed using spectral analysis, that is to say that a wave having a certain spectral width (i.e., a frequency spectrum comprising several distinct wavelengths) can be decomposed into a sum of waves each having a single determined wavelength, called "monochromatic" waves. Light is thus made of electromagnetic waves, that is to say vibrations at one or more wavelengths in the visible frequency spectrum, i.e. approximately between the wavelengths 400 and 800 nm.

[0028] Physically speaking, an electromagnetic wave is a field, that is to say an area of ​​space whose properties are modified. Each point in space is then assigned a vector physical quantity, i.e., a vector which represents not only the amplitude of the field concerned but also the orientation of said field in space. This orientation is represented, for each vibration at a given wavelength, by the orientation of the associated vector relative to a reference frame {X,Y,Z}, commonly the terrestrial reference frame linked to the ground of the Earth. Light waves are vector waves, that is to say waves which can oscillate in more than one orientation.

[0029] An electromagnetic wave represents the propagation of an associated electric field and magnetic field, perpendicular to each other and to the direction of the ground propagation. Like any propagating electromagnetic wave, in fact, a light wave is defined by the local disturbance of the electric field (commonly noted "E", this letter being surmounted by an arrow to designate more specifically the vector associated with said field: E) and of the magnetic field (commonly noted "B", this letter being surmounted by an arrow to designate the vector associated with said field: 3) - The disturbance is initially produced by charged particles which are accelerated, but it can propagate through a propagation medium possibly devoid of particles (such as the vacuum of space for example). example). In the case of a plane wave, which is a good approximation of most light waves, each of the vectors and b oscillates in a single respective plane, which planes are both perpendicular to the rectilinear propagation direction (commonly represented by a vector y in the terrestrial reference frame {X,Y,Z} too). In the study of electromagnetic waves, and therefore in particular in that of light, it is customary, by convention, to ignore the magnetic field B because its variations can be determined from those of the electric field E which are linked to them through Maxwell's equations. We will therefore only consider the electric field E in what follows.

[0030] Polarization is a property that vector waves have of presenting a privileged distribution of the orientation, in space, of the vibrations that compose them. For example, sound waves do not have this property because they are longitudinal waves: they propagate via air molecules that collide with neighboring air molecules, the local movement of these molecules being in the same direction as the propagation of the energy. Conversely, light waves have polarization properties, because the orientation of the electric field E is transverse to that of the propagation of the ground. Concretely, the polarization of light corresponds to the direction of the electric field E-

[0031] Polarization is rectilinear when E is always oriented in the same direction. By convention, rectilinear polarization is said to be vertical (vertical polarization) when the electric field vector e is vertical relative to the reference frame {X,Y,Z}. Similarly, rectilinear polarization is said to be horizontal (horizontal polarization) when the electric field vector E is horizontal relative to said reference frame {X,Y,Z}. In both cases, the vector E is perpendicular to the propagation direction y. For an unpolarized, or natural, wave, £ rotates around its axis in an arbitrary and unpredictable way over time. Polarizing a light wave corresponds to giving a defined trajectory to the electric field ~E. Once given, the polarization of a light wave can be preserved or modified, depending on the conditions of the wave propagation.

[0032] Various phenomena affect the wave behavior of light during its propagation. In a homogeneous and isotropic medium (i.e. one whose physical properties are invariant as a function of direction), electromagnetic waves propagate in a straight line. On the other hand, in the event of encountering an obstacle, there is diffraction (i.e., diffusion of the wave by the various points of the object, which manifests itself by interference phenomena, i.e. combinations of two induced waves which are of the same frequency but have, between them, phase shifts, i.e. phase differences). In addition, during a change of propagation medium, there is reflection (a part of the electromagnetic waves returns to the original medium) and re fraction (another part of the wave propagates through the second medium, but with a different direction). There is also refraction if the properties of the propagation medium change depending on the location (heterogeneity).

[0033] The reflection of light on certain materials transforms its polarization. To understand and account for this, we decompose the polarization of light into two orthogonal rectilinear polarizations, denoted 5 and p. The 5 polarization is the component of the electric field which is perpendicular to the Fonde plane of incidence, and the p polarization is the component of the electric field which is contained in this plane. The light is more or less reflected depending on whether it is 5 or p polarized, and depending on the angle of incidence (i.e., the angle between the direction of propagation of the incident wave and the direction normal to the plane of the reflecting interface considered as locally flat).

[0034] The increase in data transmission rates on optical communication networks and the use of wavelength division multiplexing (or WDM) has the consequence of making systems all the more sensitive to the aforementioned phenomena such as chromatic dispersion or polarization. This implies, in general, that the characteristics of the optical components that constitute the network are controlled from the design phase, and taken into account to improve the design of these components. This concerns in particular the characteristics of optical couplers in integrated circuits on a silicon (or other) substrate at the interface between the optical fiber and one (or more) waveguide(s) produced on a silicon wafer according to microelectronics technologies on a silicon substrate.

[0035] Polarization dependent losses (or PDL loss) are defined as the maximum variation in the power transmitted by an optronic component or any other optical device, when the input state of polarization (SOP) is modified across all possible polarization states.

[0036] Fiber gratings (or fiber optic gratings or fiber grating units) are one-dimensional or two-dimensional networks of optical fibers. Often, such a grating is formed only at the end of a fiber bundle, rather than along the entire length of the fiber. The purpose of such a grating is generally to couple light from a source array to the fibers, or from the fibers to another component, such as a planar waveguide array on a photonic integrated circuit.

[0037] The simplified diagram of [Fig.l] illustrates a use case of such integrated optical couplers, namely a transmission coupler 10 (Tx) and a reception coupler 20 (Rx).

[0038] Here and for the rest of the description we define a direct three-dimensional orthogonal reference frame {X,Y,Z}, where the X and Y axes form a plane parallel to the main plane of the plate ("wafer") forming the flat substrate, and where the Z axis is oriented substantially orthogonal to the main plane of the plate, this Z axis being oriented in the direction of the axis of gravity. In the remainder of the description, the terms "vertical" and "vertically" are understood as relating to an orientation substantially parallel to the Z axis, and the terms "horizontal" and "horizontally" as relating to an orientation substantially parallel to the (X,Y) plane. Furthermore, the terms "above" and "below" and their derivatives (such as "above" and "below", or "over" and "below"), as well as the terms "lower" and "upper", used to qualify an element of the microstructure considered, are understood as relating to an increasing positioning when moving away from the wafer upwards, i.e., in the vertical +Z direction.

[0039] Each of the couplers 10 and 20 comprises a waveguide 11 or 21, respectively, made of silicon on a doped silicon-based substrate 12 or 22, respectively. It also comprises a coupling grating (i.e., a diffraction grating), made for example by a layer of silicon (Si) with etched patterns, on a layer of silicon dioxide (SiO2), at a first end 14 or 24 of the waveguide 11 or 21, respectively. This first end of the waveguide 11 or 21 of each of the couplers 10 and 20, respectively, is the threshold of the optical coupling of said coupler with a respective end of an optical fiber 100. The fiber 100, for example a fiber of the Smf28™ type or other, connects the optical couplers 10 and 20 and is suitable for the transmission of data between said couplers. Data transmission can be unidirectional or bidirectional, monochromatic (i.e., on a single wavelength) or with multiple wavelength division multiplexing (WDM). Generally, the light wave is polarized at the input of the fiber 100, at the transmission coupler 10. On the other hand, the polarization of the light at the other end of the fiber, at the reception coupler 20 is variable and unstable, and therefore unknown.

[0040] In [Fig. 1], the detail of the coupling zone 14 shows the coupling grating 13 providing the optical interface between the waveguide 11 of the optical coupler 10 and the relevant end of the optical fiber 100. Said patterns form a grating of holes (filled with SiO2) spaced along the X and Y directions which define a plane parallel to the surface of the substrates 11 and 21, respectively, with a determined pitch. This pitch, as well as the shape of the holes, may be non-uniform across the surface of the grating. The pitch and the shape of the holes determine the behavior of the grating, and in particular the characteristics of the coupling in terms of wavelength and polarization. In particular, the coupling grating at the receiving coupler 20 may be insensitive to the polarization of the incoming light. In addition, and where appropriate, several coupling gratings comparable to the grating 13 may be juxtaposed in the X,Y plane and / or stacked along the Z direction, and each have a network pitch distinct from that of the others to ensure coupling of multiple wavelengths.

[0041] Of course, the integrated circuits which embed the optical couplers 10 and 20 may comprise photonic circuits designed and adapted to process the optical signal (in transmission or reception, respectively) and carry out, for example, filtering, amplification, modulation or demodulation, multiplexing or demultiplexing, etc. In addition, the data transmitted or received by the optical couplers 10 and 20, respectively, may be processed by microelectronic devices produced on the same substrate as the latter and therefore included in the same integrated circuit package (“packaging” in English).The data may otherwise be received from or transmitted to, respectively, other integrated electronic circuits which are dedicated to this processing in any data processing system, specific to each application.

[0042] As indicated in the introduction, certain optical couplers are designed to ensure both types of coupling, in transmission (Tx) and in reception (Rx), simultaneously for data communication via a single optical fiber. They are called "transceiver" in the jargon of the person skilled in the art, and those which are of interest to the embodiments of the invention are the two-dimensional diffraction grating couplers which are polarization diversity fiber / silicon couplers known by the acronym 2DGC or PSGC (from the English "Polarization Splitting Grating Coupler"). They are generally specified for a relatively narrow wavelength range (a few tens of nanometers for the widest).This type of coupler is frequently used in photonics, and has a fiber terminal (i.e. coupled or capable of being coupled to an optical fiber) and two guided terminals (i.e. each coupled to a planar waveguide on silicon) in a "1 by 2" structure (noted 1 x 2), or "Y" structure (called "T-coupler" in English, which means tee coupler).

[0043] An example of a 2DGC coupler is shown in [Fig. 2] in a simplified isometric perspective view. The left part of the figure schematically shows an example of the embodiment of such a coupler 30 on a substrate, while the right part of the figure shows its symbolic representation as subsequently used in the other figures of the drawings.

[0044] The coupler 30 comprises a substrate 31, for example a silicon-based substrate such as a bulk silicon substrate, or a silicon-on-insulator (SOI) substrate. It then comprises an interlayer 32, for example a layer of silicon oxide or silica (SiO2) deposited directly on the substrate 31, as well as a coupling layer 33, for example made of silicon nitride (SixNy), for example Si3N4. Finally, an encapsulation layer (not shown), for example a 1 μm thick layer of silica (SiO2), covers the aforementioned stack of the substrate 31, the interlayer and the coupling layer 33.

[0045] The material of the coupling layer 33 deposited on the substrate 31 has a refractive index substantially higher than that of the interlayer 32 of said substrate 31 so as to ensure the guidance of the light by total reflection. The silicon (Si) coupling layer 33 is doubly etched, according to conventional integrated circuit manufacturing technologies (based on photolithography), to form an optical coupling network 34 and planar waveguides 35 and 36.

[0046] On the one hand, in fact the coupling layer 33 is etched at the level of a coupling zone 34 itself, to form a two-dimensional network of patterns, in this case recesses or bowls (i.e., non-through holes) operating as elements of a diffraction grating ("grating" in English). This pattern network in fact constitutes the coupling network for the optical coupling of the coupling layer 33 to an optical fiber 100. The optical coupling interface at the coupling zone 34 at which the optical fiber 100 can be coupled, therefore constitutes the fiber terminal of the coupler 30. In one example, the coupling layer 33 has a thickness of approximately 300 nm and the patterns of the coupling network 34 are etched to a depth of 150 nm, for example, from the upper surface of the coupling layer 33. This example is not limiting.The section of these patterns in a plane orthogonal to the vertical direction Y (and therefore their shape in top view) can be circular, square, in the shape of pinched squares, diamonds, etc. The patterns are distributed in rows and columns in the X,Y plane according to a square mesh, that is to say with a periodicity of order two, and with distribution steps according to the two directions X and Y of the plane, respectively, which are identical. The person skilled in the art will appreciate that the patterns thus etched to a depth of 150 nm are then filled with the material of the encapsulation layer which covers the coupling layer 33. In the example shown, the optical coupling zone 34 has a square shape, seen from above. It can also be rectangular.

[0047] The coupling layer 33 is further etched to form two planar waveguides 35 and 36 integrated on silicon, each of which extends from respective adjacent sides of the coupling zone 34, orthogonally to each other (i.e. forming an angle of 90° or ir / 2 between them). Thus, in the example shown in [Fig. 2], the first waveguide 35 extends longitudinally in the X direction from a first side of the coupling zone 34 towards the left of the figure, while the second waveguide 36 extends transversely in the Y direction from a second side of the coupling zone 34, adjacent to said first side, towards the bottom of the figure. These waveguides are suitable for guiding light in the coupler 30, at the wavelength considered. The optical coupling between the fiber 100 and the waveguides 35 and 36 is carried out by diffraction, the light being diffused at the level of each of the grating elements etched in the coupling zone 34. As has been understood, ends of the waveguides 35 and 36 opposite the coupling zone 34 constitute the guided optical terminals of the coupler 30, by which

[0048] These couplers operate by Y-splitting: the power distribution in the two daughter arms depends on the angle they make with the mother arm and on the input polarization. Thus, if the two angles are equal, the power division is 50 / 50 for the purely s and purely p polarization states. We then say that we have a 50 / 50 coupler. These couplers can be reversible, in the sense that the behavior is entirely reciprocal. Indeed, an input optical signal can be inserted via the fiber optic terminal 39, and divided into two components guided to the guided terminals 37 and 38, respectively. In this case, the fiber terminal 39 is an input terminal and the guided terminals 37 and 38 are output terminals. We then speak of a splitter coupler.But, conversely, two separate optical signals can be inserted into the coupler via the fiber optic terminals 37 and 39, an optical signal resulting from the optical combination of said optical signals being delivered by the fiber optic terminal. In this case, the guided terminals 37 and 38 are input terminals and the fiber terminal 39 is an output terminal. This is then referred to as a mixing coupler.

[0049] In silicon photonic data link receiver circuits, two-dimensional coupling grating (2DGC) type "1 by 2" couplers as shown in the above are used to efficiently couple light entering the photonic circuit with an arbitrary polarization state.

[0050] The insertion point (i.e. the point where the input optical fiber interfaces with the coupling network) can be represented, in horizontal section seen from above, by an oval. An oval is in fact the figure of intersection between a plane (that of the network 34) and a cylinder (that of the optical fiber 100) when the longitudinal axis of said cylinder is inclined along the Z direction normal to said plane (the inclination here being 8°, in accordance with an industry standard).

[0051] Like any real passive optical component, a 2DGC type coupler has insertion losses, even very low ones. The insertion loss of an optical component quantifies the power lost by an optical signal when it passes through this component. If Pi is the initial incoming power of the optical signal and Pf the final outgoing power, then the insertion loss IL is defined as being, on a logarithmic scale: [Math 1] IL / b= 1Q x / og^ ) = 10 x log( T) Equation (1) where T is the transmission coefficient (also called "transmission" for short), given by y _ .

[0052] The transmission factor depends on the polarization of the light. The polarization-dependent loss is the physical quantity that quantifies this phenomenon. It is given by the difference between the maximum transmission coefficient Tmax and the transmission coefficient Tmin, and is defined on a logarithmic scale by: [MATH 2] PDL = K) XL () G {) Equation (2) Ci- lj \ f ' f \ x. mm f

[0053] In a 2DGC type coupler, the polarization dependent loss (PDL) comes essentially from the non-zero angle of the fiber 100, which is intended to reduce reflections at the interface between the optical fiber 100 and the silicon-etched coupling network at the coupling zone 34.

[0054] Characterizing the PDL of a 2DGC is quite critical, because the PDL manifests itself by optical noise, which we do everything to reduce at the level of the design of the photonic structures and the choice of the materials which constitute them. But it is especially difficult to evaluate in operational conditions, the insertion loss IL being easier to evaluate. In practice the level of uncertainty is quite high for direct measurements of the PDL. There is also a constraint linked to the speed of the evaluation, which must be high to satisfy the requirements of an industrial implementation.

[0055] The characterization of the polarization-dependent loss of a 1DGC type coupler intended to be used on the transmission (Tx) side in an optical transmission architecture such as shown in [Fig.l] does not present any particular difficulty, because the polarization state of the light is known and controlled. But this is not the same for a 2DGC coupler intended to be used on the reception (Rx) side, because the polarization state can vary at any time, and in an arbitrary manner.

[0056] [Fig. 3] shows three known examples of a test structure of a 2DGC type coupler, which can be used to characterize the PDL of a 2DGC type coupler. In these examples, a 2DGC type coupler 41 is considered, the polarization dependent losses (PDL) of which are to be measured for use of the coupler in reception. A known test structure typically consists of the 2DGC coupler to be tested 41, which is arranged at the input, as well as a coupler 42 identical to said coupler 41 to be tested, which is arranged at the output as in example (a) shown on the left of the figure. The two couplers are arranged back-to-back, their guided optical terminals (terminals 37 and 38 of [Fig. 2]) being coupled two-by-two by two respective planar waveguides. The fiber optic terminal (terminal 39 of [Fig.2]) of the coupler 41 to be tested is configured as an optical input, adapted to receive an input optical signal via an input optical fiber.The fiber optic terminal of the mirror coupler 42 is configured as an optical output, adapted to deliver an output optical signal via an output optical fiber. Such a . test structure was used, for example, in the research work that led to the publication of the scientific article by S. Plantier et al., "Impact of scattering element shape on polarization dependent loss in two dimensional grating couplers", IEEE 13th International Conference on Group IV Photonics (GFP), August 2016, doi: 10.1109 / GROUP4.2016.7739057.

[0057] In another example (b) of known test structure shown in the center of [Fig.3], instead of another 2DGC like the coupler 42 of example (a), another type of silicon / fiber coupler(s) is arranged here at the output, for example a coupler 43 and a coupler 44 each in an output branch of the 2DGC coupler to be tested 41. The couplers 43 and 44 can be single-input, single-output couplers of the 1DGC type.

[0058] In a third example (c) shown to the right of [Fig.3], photodiodes are arranged directly at the respective outputs of the coupler to be tested 41, which can for example be integrated within the test structure on the same silicon substrate.

[0059] In all cases, it is the sum of the intensity T1 and the intensity T2 of the electric field propagating from the two respective outputs of the 2DGC that must be considered. For this purpose, in the case of structures (b) and (c) of [Fig. 3], a sensor of the transmitted intensity is provided at the output of each of the branches, and the values ​​that they measure are summed. The PDLs can be measured by any known measuring devices, for example devices based on a measurement in reflection or in transmission by applying the Jones matrix method, the Mueller matrix method, etc.

[0060] With reference to the illustration given by the graph in [Fig.4], a method for measuring the PDL of a 2DGC with any of the conventional test structures in [Fig.3], consists of measuring the transmission of this optical component over a given time interval AT, during which it is ensured that all possible polarization states between the 5 and p polarizations are scanned. This is the polarization scanning method described in the scientific article F. Van Laere et al. of 2088, cited in the introduction. In the example shown in [Fig.4], AT is equal to 60 seconds, and the maximum transmission deviation that is recorded is equal to 0.79 decibels (dB).

[0061] This method has the advantage that at no time is it necessary to know the state of polarization incident on the device to be tested. But it also has many disadvantages in practice, as already mentioned in the introduction. In particular: • polarization must be scanned for each wavelength point, which can take minutes per component, and is not compatible with large-scale testing. • It is necessary to ensure that the injected power does not vary depending on the state of polarization in 5 and / or in p and of the wavelength X.

[0062] The embodiments of the test structure according to the invention use another approach, based on observation made by the inventors.

[0063] [Fig.5A] and [Fig.5B] give an illustration of the reciprocal relationship that has been observed and studied by the inventors for a coupler 50 of type 2DGC, between the polarization of the light on the coupling input 53, on the one hand, and the phase shift between the two outputs 51 and 52, on the other hand. We consider a monochromatic light wave of determined wavelength Xin introduced into the input 53, for example via an optical fiber coupled to the coupling network of the input 53 of the coupler 50.

[0064] In the case of Figure 5A, the incident electric field £ is entirely p-polarized. If the incident Fonde optical power is Tp(Xin), the light waves respectively diffused in respective waveguides to which the outputs 51 and 52 are coupled have a transmitted optical power which is equal to half of Tp(Xin). In other words, the optical power propagated in each of the output branches of the coupler is then equal to Tp(Xin) / 2. The phase shift ¢) between these two propagations is equal to ir (i.e., ¢=11). In other words, the incident wave received at input 53 is divided into two waves propagated via the outputs 51 and 52 with a power shared 50 / 50 and with a relative phase shift ¢) equal to ir (¢=11).

[0065] In the case now of Figure 5B, the incident electric field E is entirely polarized 5. If the optical power of the incident wave is Ts(Xin), the light waves respectively diffused in respective waveguides to which the outputs 51 and 52 are coupled have a transmitted optical power which is always equal to half of Ts(Xin). In other words, the optical power propagated in each of the output branches of the coupler is then equal to Ts(Xin) / 2. But the phase shift ¢) between these two propagations is this time equal to zero (i.e., ¢=0). In other words again, the incident wave received at input 53 is divided into two waves propagated via the outputs 51 and 52 always with a power shared at 50 / 50, but with a relative phase shift 0 equal to 0 (¢=0) i.e. without phase shift.

[0066] In an intermediate case between those of [Fig.5A] and [Fig.5B] in which the incident wave received at input 53 has a mixed p and s polarization, the power of the incident wave is shared unequally between outputs 51 and 52, and with a phase shift between the two light waves thus transmitted which takes an intermediate value between 0 and jt.

[0067] But what has been observed is not only the continuity of the phenomenon for any mixed s and p polarization state between the purely 5 and purely p polarization states which is mentioned above, but also and above all the reciprocity of this phenomenon. Indeed: • if we force between outputs 51 and 52 a phase shift ¢) equal to ji, then the input polarization is an entirely p polarization; • if we force between the outputs 51 and 52 a phase shift ¢) equal to 0, then the input polarization is an entirely s polarization; and, more generally, • if we impose between the outputs 51 and 52 a determined phase ¢) whose value is between 0 and ji, then the polarization of the light on the input 53 is an intermediate polarization in s and p, and the relationship between said phase and said polarization is a reciprocal relationship, that is to say that it follows a continuous and bijective function.

[0068] In other words, it has been observed that, by reciprocity, modifying the phase relationship between the two output waves of a 2DGC type silicon-fiber coupler with one input and two outputs, amounts to modifying the polarization state of the optical signal on the input. The principle of the implementation of the invention is based on the ability to generate a phase shift between the two arms of the test structure of [Fig.3], version (a) for example, in order to emulate a scan of the polarization of the injected / emitted optical signal without having to control the polarization thereof.

[0069] More particularly, the test structure 60 according to [Fig.6A] is similar to example (c) of [Fig.3], but is modified to allow the PDL of the 2DGC under test to be measured by circumventing the need to change the polarization of the input light. The fiber optic terminal (terminal 39 of [Fig.2]) of the mirror coupler 62 is configured as an optical input (this will be referred to as an input coupler to designate the coupler 62 mounted in a mirror of the coupler to be tested 61 in the embodiment of [Fig.6A]). This input terminal is adapted to receive an input optical signal via an input optical fiber. The fiber optic terminal of the coupler to be tested 61 is configured as an optical output, adapted to deliver an output optical signal via an output optical fiber. The guided optical terminals of the couplers 61 and 62 are coupled two-by-two via planar waveguides 64 and 65 made on silicon.

[0070] According to embodiments of the invention, an adjustable phase shifter 63 (or phase modulator) is further provided which is integrated in a first optical path or arm, between the two identical couplers 61 and 62 mounted back-to-back. In the example shown, the phase shifter is arranged at the arm of the structure 60 which comprises the planar waveguide 65 (right arm, in [Fig.6A]). This is a pure phase modulation, i.e. without intensity modulation (for example using a so-called "thermal" phase modulator which is known for this property). The polarization of the light at the input of the input coupler 62 can thus be switched between purely 5 polarization and purely p polarization using a phase shift adjustment voltage V which is applied to a phase shift adjustment input of the phase modulator 63, as presented above with reference to [Fig.5A] and [Fig.5B]. By varying the adjustment voltage V, the phase shift applied to the propagation of the optical signal in this first arm of the test structure 60 can be caused to sweep the range [0;ji]. This forces the polarization of the optical signal entering the coupler under test 61 to sweep all the mixed states between the fully 5 polarization state and the fully p polarization state, respectively. The polarization-induced transmission difference, and therefore the PDL of the 2DGC, can thus be simply determined.

[0071] [Fig.6B] shows a variant of the test structure of [Fig.6A] in which a phase modulator 66 is introduced into the other arm of the structure, namely the one comprising the planar waveguide 64 (left arm in the figure). This other phase modulator 66 is structurally identical to the phase modulator 63 present in the first arm of the test structure 60, namely the one comprising the waveguide 65. However, the phase modulator 66 is configured so as not to introduce any phase shift for the propagation of the optical signal in the waveguide 64. Everything happens as if it were permanently controlled in such a way that the phase shift A0 introduced is zero (A0=0). In other words, it is a fictitious phase modulator ("dummy" in English), the presence of which nevertheless makes it possible to balance, between the two arms 64 and 65 of the test structure 60, the characteristics of the propagation of light in said arms.These characteristics include the possible variation in induced polarization, the possible induced losses, a possible static phase shift, or even a possible wavelength shift, which would be introduced by the phase modulator 63 depending on the technology chosen for this modulator and depending, where applicable, on the specificities specific to each application.

[0072] The two variants of the test structure according to [Fig.6A] and [Fig.6B] are adapted for use of the test structure with optical fibers, at the input and output, which are opposite each other, that is to say and facing each other (in the vertical direction in the schematic views of [Fig.6A] and [Fig.6B]).

[0073] In [Fig.7A] and [Fig.7B] two further embodiments of a test structure proposed by the invention are shown, which are variations of the test structures of [Fig.6A] and [Fig.6B], respectively. The test devices of [Fig.7A] and [Fig.7B] are adapted for use with respective input and output optical fibers that are arranged side-by-side (along the horizontal direction in the schematic views of [Fig.7A] and [Fig.7B]) in a unit block, also called a fiber array unit, or fiber array. Fiber arrays are one- or two-dimensional networks of optical fibers. Often, such an array is formed only at the end of a bundle of fibers, rather than along the entire length of the fibers in question. The advantage of such a fiber network is generally to allow light from an array of adjacent optical sources to be introduced into the fibers in question, or to allow separate fibers to be coupled to another component via an array of adjacent planar waveguides on a photonic integrated circuit, as is the case here.

[0074] With reference to the diagrams of [Fig.8A] and [Fig.8B], an example method for characterizing the PDL of a 2DGC with the test structure proposed above with reference to the diagrams of [Fig.6A] and [Fig.6B], or their variants of [Fig.7A] and [Fig.7B], respectively, will now be described. For the description of this example, the test structure 60 of [Fig.6A] will be considered more particularly, but the person skilled in the art will appreciate that the method for measuring the PDL is the same for the other test structures presented with reference to [Fig.6B], to [Fig.7A] or to [Fig.7B].

[0075] In one use case, a certain quantity of couplers to be tested is manufactured on the same wafer, for example a series of around a hundred couplers, each in a test device in accordance with the invention, but with small variations in design and / or technological manufacturing parameters between each of these couplers. The tests make it possible to assess which version(s) of the couplers in the series gives the smallest PDL. Such operations are common practice, in themselves, in the semiconductor industry in order to design high-performance photonic components.

[0076] A test equipment comprises a test controller 80, at least one source 81 of polarized light, and a light sensor or photodetector 82 for each of the test structures such as the structure 60 shown in [Fig.8A] and [Fig.8B] which can be tested simultaneously during the same test. The test controller 80 comprises a computer 83, such as a microprocessor, and a memory 84 comprising random access memory (RAM).

[0077] The test equipment 80 (or "probing station" in English), is used to test unpackaged chips on which the transmitted power is measured with optical reading probes respectively associated with each test structure. For example, the equipment 80 can be coupled to a photodiode 82 at the output of the test structure 60 shown in [Fig.8A] and [Fig.8B]. This photodiode 82 can be integrated into the test structure itself, i.e. it can be made on the same silicon substrate, to measure the optical power Pt at the output of the coupler under test 61, which is the power transmitted by the test structure 60 from the light power Pin injected at the input at the input coupler 62. Alternatively, the light can be output from the test structure 60 by an output optical fiber coupled to the coupling network of the coupler under test 61, in order to measure the light power with a remote external photodiode. In both cases, the photodiode converts the optical signal it receives from the coupler 61 into an electrical signal representative of the light intensity transmitted through the test structure 60. It is this electrical signal which therefore constitutes the measurement of the transmitted light power Pt.

[0078] In the equipment 80, there is software which, when loaded into the memory 84 and executed by the processor 83, records the data corresponding to the graphs of the curves shown on the right of [Fig.8A] and [Fig.8B], and which is configured to extract the PDL from said recorded data. The PDL is given by the difference between the value T_max and the value T_min of the optical transmission T of the test structure 60. This is determined on the basis of the power Pin of the input optical signal Vin, and on the basis of the transmitted power Pt transmitted by the structure under test 60 which is measured during the scanning of the range of values ​​[0;jt] of the phase shift A0.

[0079] The source 81 of polarized light can be produced directly on the silicon substrate (wafer) on which the test structures are produced, or be a source external to the wafer of the test structures to which it is optically coupled by at least one fiber, generally by a fiber network ("Fiber Array"). For example, the source 81 can comprise at least one semiconductor laser, such as a laser diode emitting at at least one wavelength in the spectrum of interest. In embodiments, the laser diode can be adjustable (also called "tunable"), that is to say that the length of the optical cavity of the laser can be modified in a controlled manner, which allows them to be tuned continuously over a relatively large range of wavelengths.For example, a distributed feedback (DFB) semiconductor laser or a vertical cavity surface emitting laser (VCSEL) uses periodic distributed Bragg reflector (DBR) structures to form the mirrors in the optical cavity. The laser temperature can also be varied, as the temperature-related change in the optical index of the DBR structure shifts its maximum reflection wavelength and thus the laser wavelength. The tuning range of these lasers is typically a few nanometers (nm), up to a maximum of about 6 nm when the laser temperature is varied by about 50 degrees Kelvin (K).Typically, the wavelength is tuned by 0.08 nm / K for DFB lasers operating in the 1550 nm wavelength regime. These lasers are commonly used in optical communications applications such as Dense Wavelength Division Multiplexing (DWDM) systems to enable tuning of the signal wavelength, so that the skilled person knows how to implement them without being born. It is necessary to provide further guidance here. To achieve even broader bandwidth tuning, using this technique, one can provide an array of such lasers on a single chip and concatenate the laser wavelength tuning ranges.

[0080] The monochromatic polarized light, produced at a wavelength Xin determined by the polarized light source 81, can be introduced into each of the test structures which are tested simultaneously, for example by a bundle of respective optical fibers, or by a fiber array. This ensures that each test structure is excited by light having the same characteristics (incident wavelength(s) Xin, input light power Pin, polarization 5 or p, phase 4>in) as the other test structures. This is not a condition specific to the measurement of the PDL according to embodiments according to the invention, but makes it possible to have comparable results of measurements of the PDL for a plurality of optical couplers which are carried out using a plurality of respective test structures.For example, the incident wavelength may be in the so-called "C" and "L" bands from 450 nm to 1650 nm, for example between 1510 nm and 1590 nm, or in the so-called "O" band between 1270 nm and 1360 nm, depending on the applications envisaged for the optical coupler model under test 61. However, what is described in this particular case also works for other wavelengths.

[0081] An example of a method for characterizing the PDL of a 2DGC type coupler under test 61 will now be explained. This explanation is given with reference again to [Fig.8A] and [Fig.8B], and with further reference to the step diagram of [Fig. 10]

[0082] In order to excite the coupling network of the input coupler 62 of the test structure 60 of [Fig.8A] and [Fig.8B], an incident optical signal Sin at the wavelength Xin, polarized p, and having an intensity (optical power) Pin, is produced by the source 81 and inserted at step 101, into the fiber optic terminal of the input coupler 62, for example via a polarization-maintaining input optical fiber. The polarization of the incident light can be either entirely p as in the example shown, or entirely 5. For these two states of incident polarization, in fact, the two components of the incident optical signal Sin propagate with the same intensity in each of the planar waveguides on silicon 65 and 64 coupled to the guided optical terminals of the input 2DGC 62. This intensity will be equal to half of Pin.In other words, the input optical signal Sin is divided into two components of respective intensities equal to each other, which are each inserted into one of the two arms of the test structure 60 which comprise the planar waveguide 65 and the planar waveguide 64, respectively. It is repeated here that, in the example shown, the polarization of the incident light Sin is entirely p (as illustrated by [Fig.5A] already described in the above). But there is no difference in the process. implemented elsewhere, if the polarization of the incident light is entirely s (as illustrated by [Fig.5B] also already described).

[0083] Then, in step 102, the optical transmission of the 2DGC under test 61, that is to say the transmitted power Pt which is measured at the output of said coupler 61 by the photodetector 82, which is for example a photodiode. More precisely, a series of measurements are carried out while the phase shift A0 introduced into the arm of the structure which comprises the planar waveguide 65 by the phase modulator 63 is made to vary between 0 and ji, by varying the adjustment voltage V of the phase modulator between its extreme values ​​V=V_A0=0 on the one hand, and V=V_A0=7r, on the other hand. This scan, over the range of values ​​[0;ji] of the phase shift A0 which is then introduced into the arm 65 of the test structure by the phase modulator 63, is controlled by the processor 83 of the test controller 80 by the execution of ad-hoc pilot software.The phase shift A0 between the two branches 64 and 65 varies the output polarization state, i.e. at the output coupler 61 of the test structure 60, which is the device under test (DUT), for an input polarization of the test structure 60 which does not vary. The values ​​{Pt} of the power Pt transmitted by the test structure thus measured are stored in an indexed manner (as a function of the current value of the phase shift A0), for example in the form of a table of values ​​(or LUT, from the English "Lookup Table"), in the memory 84 of the test controller 80. This table of values ​​is indexed by the associated values ​​of the phase shifts A0.

[0084] Once measurements {Pt} of the transmitted power Pt have been acquired for the entire range of values ​​[0;ir] of the phase shift A0, that is to say once the values ​​of the adjustment voltage V of the phase modulator 63 have swept the interval [V_A4>=O;V_A([)=7r], the computer 83 determines in step 103 the amplitude of the variation of this transmitted power Pt, from the values ​​{Pt} stored in the memory 84. The maximum value {Pt_max} is given by the configuration shown in [Fig.8A] in which the phase ¢) at the input of the coupler 61 under test is equal to ir (as shown by the point on the curve to the right of said figure). The minimum value {Pt_min} is given by the configuration shown in [Fig.8B] in which the phase ¢) at the input of the coupler under test 61 is equal to 0 (as shown by the point on the curve to the right of said figure). The person skilled in the art will appreciate that between the initial state illustrated by [Fig.8A] and corresponding to an output polarization identical to the totally p input polarization (with a phase 4>=jr at the input of the coupler 61), on the one hand, and the final state illustrated by [Fig.8B] and corresponding to an output polarization which is a totally 5 polarization (with a phase ¢=0 at the input of the coupler 61 for an invariant phase ¢=^ at the input of the test device 60), there are mixed states of p and 5 polarization corresponding to each value of the phase shift A¢ between 0 and ir which is introduced by the phase modulator 63 under the control of the controller 80. via the variable phase shifter adjustment voltage V 63.

[0085] The data processing executed by the computer 83 on the values ​​{Pt} acquired and stored in the memory 84 of the controller 80, which are representative of the transmitted power Pt as a function of the phase shift A0, comprises the identification of the maximum value {Pt_max} and the identification of the maximum value {Pt_min} of the transmitted optical power Pt. In the example illustrated by [Fig.8A] and [Fig.8B], the maximum value {Pt_max} is obtained for a phase shift between the two arms of the test structure 60 which is equal to zero (A<])=0), which corresponds to the configuration of [Fig.8A]. Graphically, the value {Pt_max} corresponds to a peak of the curve represented on the right of [Fig.8A], which gives the shape of the phase shift response of the test structure 60, that is to say the transmitted power Pt as a function of the phase shift A0 introduced by the phase modulator 63.Furthermore, the value {Pt_min} corresponds graphically to a trough in the curve represented on the right of [Fig.8B], which gives the shape of the transmitted power Pt as a function of the phase shift A0 introduced by the phase modulator 63.

[0086] The transmission coefficient of the test structure 60 given by j1 — £l. P in The amplitude Tmax-Tmin of the variation of the transmission T is given by the difference between the two extreme values ​​{Pt_max} and {Pt_min}, at a value Pin kept constant throughout the test process. This amplitude gives, at step 104, the value of the PDL of the 2DGC type coupler 61 under test, in application of equation (2) given above. In other words, the PDL losses of the coupler 61 linked to the polarization are obtained by calculation on the basis of the maximum value {Pt_max} and the minimum value {Pt_min} of the optical power Pt measured at the output of the test structure 60.

[0087] Advantageously, it is noted that it is not useful to know the relationship between the adjustment voltage V of the variable phase shifter 63 and the phase shift value A0 that it generates, since we are only interested in the difference in light intensity between a peak and a trough only, to make the difference. In other words, the determination of the PDL is carried out in differential mode, or relative mode, by simple searches for a maximum and a minimum in a table of indexed values, which is a process that can be executed quickly by the processor 83 of the test controller 80.

[0088] The person skilled in the art will appreciate that from the measurements {Pt} of the power transmitted in the test structure which are recorded in the memory 84, it is also possible to obtain the value of the insertion losses IL of the coupler under test 61. This is conventional in itself. The insertion losses of the coupling network 61 under test can in fact be obtained by dividing by two the transmitted power Pt while the phase modulator is not actuated, that is to say with V=V_Ac[)=0, so that the phase shift A0 between the two branches of the test structure 60 is zero. In this configuration, the incoming and outgoing polarizations are the same (fully s polarization, in the example shown in the corresponding [Fig.8A]). In practice, this is true if the length of the guides between inputs and outputs is less than a few millimeters (mm). Otherwise, small random manufacturing variations mean that the relative phase at the end of the two guides may not be strictly equal to zero.

[0089] As will be understood, the test process described above for a given monochromatic input light wave can be repeated for other values ​​of the wavelength Xin within a wavelength band of interest, in order to characterize in wavelengths the PDL and IL losses of the coupler 61 under test for said band.

[0090] In principle, the values ​​{Pt_max} and {Pt_min} obtained for each wavelength do not vary from one interval [0,ir] to another, so that there is no need, for example, to average the calculation of the PDL over several successive trough-peak deviations of the phase shift response curve of the test structure 60. The values ​​{Pt_max} and / or {Pt_min} could vary if there were variable losses on the optical paths in the test structure 60, which is not the case with the proposed test structure.

[0091] In order to further avoid an uncontrolled dynamic phase difference between the two respective arms of the test structure 60 which could be introduced by the phase modulator as a function of the variation of its control voltage V, certain embodiments of the test structure may provide for the use of an optical phase shifter with a thermo-optical effect. With such an optical phase shifter, the variation of the phase A0 is obtained by a variation of the refractive index of the material forming the core of the waveguide 65 considered, by means of a modification of the temperature applied to said waveguide 65. Advantageously, such a phase shifter with a thermo-optical effect does not modify the intensity of the optical signal circulating in said waveguide 65. A possible embodiment of an optical phase shifter with a thermo-optical effect is described, for example, in the scientific article by Harris NC et al., "Efficient, Compact and Low Loss Thermo-Optic Phase Shifter in Silicon", Optics Express 22 9 (2014), doi: 10487-93. Furthermore, a comparison between different examples of thermo-optic optical phase shifters was the subject of the scientific article by A. Masood et al., "Comparison of heater architectures for thermal control of Silicon photonic circuits", 10th International Conference on Group IV Photonics, Seoul, Korea (South), 2013, pp. 83-84, doi: 10.1109 / Group4.2013.6644437. .

[0092] For example, in a particular embodiment, the phase shifter 63 may comprise: • a waveguide crossed by the optical signal to be phase shifted, which is made up of the planar waveguide 65 of the first arm of the test structure considered here; and, • a resistive metal arranged near this planar waveguide 65.

[0093] In response to the phase shift control voltage V delivered by the test controller 80, a potential difference is applied to this resistive metal, which generates heat. The resistive metal therefore varies the temperature of the waveguide 65 and therefore its refractive index. This modifies the phase of the optical signal which passes through this waveguide.

[0094] To directly access the PDL of the 2DGC using the test method described above with reference to [Fig.8A] and [Fig.8B], it is appropriate that the light intensity of the optical signal circulating in each of the two arms of the test structure (i.e., in each of the two waveguides 64 and 65) is the same. This condition is satisfied if the incident polarization is strictly symmetrical, which is the case when the incident optical signal is fully 5-polarized or fully p-polarized, as provided in the embodiments of the method that have been described above. If this cannot be ensured, however, a variant of the test structure such as that illustrated in [Fig.9] may be used.

[0095] [Fig.9] in fact shows a variant of the test structure comprising, instead of the coupler 62 mirroring the coupler 61 under test, another type of fiber / silicon coupler 91, a waveguide 92 on silicon, and an optical power divider 93. The fiber / silicon coupler 91 is a "1 to 1" type coupler (i.e. with a fiber input and a silicon-guided output), for example, a one-dimensional coupling grating (1DGC) optical coupler as shown, or an edge-mounted coupler. It is adapted and arranged to couple incident light from a source to which it is connected by an optical fiber, into the waveguide 92. The light intensity in the waveguide 92 is then divided equally by the optical power divider 93. The latter may be a "Y" shaped junction, a multimode interference (MMI) coupler, or a directional coupler.A first output of the optical power divider 93 is coupled to the waveguide 64 which forms the first arm of the test structure, while the second output of said divider 93 is coupled to the waveguide 65 which forms the second arm of said test structure. With this variant, the PDL of the 2DGC can be determined without the need for a well-defined known polarization at the input.

[0096] The present invention has been described and illustrated in this detailed description and in the figures of the accompanying drawings, in possible embodiments. The present invention is not limited, however, to the embodiments presented. Other variations and embodiments may be deduced and implemented by those skilled in the art upon reading this description and the accompanying drawings.

[0097] In the claims, the term "comprise" or "include" does not exclude other elements or other steps. The different features presented and / or claimed may be advantageously combined. Their presence in the description or in different dependent claims does not exclude this possibility. The reference signs should not be understood as limiting the scope of the invention. List of cited documents Patent documents

[0098] None Non-patent literature

[0099] F. Van Laere et al., “Efficient Polarization Diversity Grating Couplers in Bonded InP-Membrane” IEEE Photonics Technol. Lett., vol. 20, no. 4, pp. 318-320, Feb. 2008, DOI: 10.1109 / LPT.2007.915587

[0100] Mekis et al., “A Grating-Coupler-Enabled CMOS Photonics Platform”, IEEE J. Sel. Great. Quantum Electron., vol. 17, no. 3, pp. 597-608, May 2011, doi: 10.1109 / JSTQE.2010.2086049

[0101] S. Plantier et al., "Impact of scattering élément shape on polarisation dépendent loss in two dimensional grating couplers", IEEE 13th International Conférence on Group IV Photonics (GFP), August 2016, doi:10.1109 / GRGUP4.2016.7739057

[0102] Harris, N. C., et al., "Efficient, Compact and Low Loss Thermo-Optic Phase Shifter in Silicon", Optics Express 22 9 (2014), doi: 10487-93

[0103] A. Masood et al., "Comparison ofheater architectures for thermal control of Silicon photonic circuits", lOth International Conférence on Group IV Photonics, Séoul, Korea (South), 2013, pp. 83-84, doi: 10.1109 / Group4.2013.6644437

Claims

Claims

1. Silicon integrated test structure for characterizing the polarization dependent loss, PDL, of a "1 to 2" type fiber / silicon optical coupler with two-dimensional diffraction grating, 2DGC, having a fiber optic terminal as well as a first guided optical terminal and a second guided optical terminal, said test structure comprising: • a first planar waveguide (65) having two ends; • a second planar waveguide (64) having two ends; • an input fiber / silicon optical coupler (62,91) having a fiber optic terminal, adapted to receive an input optical signal (Sin) via said fiber optic terminal, and associated optical splitting means (93) which are adapted to split said input optical signal (Sin) into two components of equal respective optical intensities, and which are arranged to insert each of said components into the first planar waveguide (65) and into the second planar waveguide (64), respectively, at a respective first end of said planar waveguides (64,65); and, • the 2DGC under test (61), with its two guided optical terminals which are coupled to a second end of the first planar waveguide (65) and the second planar waveguide (64), respectively, and configured to deliver an output optical signal (Vout) on its fiber optic terminal, characterized in that it further comprises: • at least one first adjustable phase shifter (63) which is arranged at the level of the first planar waveguide (65) between the input optical coupler (62, 91) and the 2DGC under test (61), and which is configured to be controlled so as to apply a phase shift determined between 0 and ji, as a function of a phase shift adjustment signal (V), to the optical signal propagating in said first planar waveguide (65), said phase shift being a pure phase shift, i.e. without variation in optical intensity.

2. A test structure according to claim 1, wherein the coupler input optics (62,91) and its associated optical splitting means (93) comprise a "1 to 1" type fiber / silicon optical coupler followed by an optical power splitter (93) with one input and two outputs, said outputs each being coupled to one of the first ends of the first planar waveguide (65) and the second planar waveguide (64), respectively.

3. A test structure according to claim 2, wherein the "1 to 1" type fiber / silicon optical coupler is a one-dimensional diffraction grating coupler, 1DGC.

4. A test structure according to claim 1, wherein the input optical coupler (62,91) and its associated optical splitting means (93) comprise a second 2DGC (62), identical to the 2DGC under test (61), and having its guided optical terminals which are each coupled to the other of the ends of the first planar waveguide (65) and the second planar waveguide (64), respectively.

5. Test structure according to any one of claims 1 to 4, further comprising a second adjustable phase shifter (66), structurally identical to the first adjustable phase shifter (63), which is arranged at the second planar waveguide (64) between the input optical coupler (62,91) and the 2DGC under test (61), and which is configured such that at all times it does not apply any phase shift to the optical signal propagating in said second waveguide (64).

6. A test device according to any one of claims 1 to 5, wherein the optical coupler under test (61) and the input optical coupler (62) are arranged symmetrically with respect to each other, such that an input optical fiber can be connected to the fiber optic terminal of the input optical coupler (62) and an output optical fiber can be connected to the fiber optic terminal of the optical coupler under test (61), with said input optical fiber and said output optical fiber extending opposite each other.

7. A test device according to any one of claims 1 to 5, wherein the optical coupler under test (61) and the input optical coupler (62) are arranged in a block such that an input optical fiber can be connected to the fiber optic terminal of the input optical coupler (62) and an output optical fiber can be connected to the fiber optic terminal of the optical coupler under test (61), said input optical fiber and said output optical fiber being optical fibers of the same fiber network.

8. A test device according to any one of claims 1 to 7, wherein the first phase modulator is a thermo-optic phase shifter.

9. A method for characterizing the polarization dependent loss, PDL, of a "1 to 2" type fiber / silicon optical coupler with a two-dimensional diffraction grating, 2DGC, using a silicon integrated test structure (60) according to any one of claims 1 to 8, said method comprising the following steps: • inserting (101) an input optical signal (Vin) into the fiber optic terminal of the input optical coupler; • measuring and recording (102) values ​​of the transmission (Pt) through the test structure (60) as a function of the value (A0) of the phase shift applied to the optical signal propagating in the first planar waveguide (65) by the adjustable phase shifter (63), while the phase shift adjustment signal (V) of said phase shifter is varied so that said phase shift sweeps the interval [0;ji];• determining (103) the maximum value (Pt_max) and the minimum value (Pt_min) of the measured and recorded transmission values ​​(Pt); and, • obtaining (104) the PDL as the difference between said maximum value (Pt_max) and said minimum value (Pt_min).;

10. Computer program product comprising one or more sequences of instructions stored on a memory medium (84) readable by a machine (80) comprising a processor (83), said sequences of instructions being adapted to carry out all the steps of the method according to claim 9 when the program is read from the memory medium (84) and executed by the processor (83).

Images