Angular optical filtering element for angular filtering having controlled angular selectivity

Abstract

According to one aspect, the invention relates to an angular optical filtering element (E.sub.i) optimized for angular filtering about a given operating angle of incidence (θ.sub.i, 1) in a given spectral band. The angular filtering element (E.sub.i) comprises a first nanostructured, band-pass, spectral filter (11.sub.i, 301) and a second nanostructured, band-pass, spectral filter (12.sub.i, 302). Each of the first and second spectral filters comprises, respectively, in said spectral band, a first and a second central filtering wavelength that respectively has a first and second angular dispersion curve defined depending on the angle of incidence (θ.sub.inc) on the optical filtering element (E.sub.i), the curves of angular dispersion being secant about the operating angle of incidence (θ.sub.i, 1) of the optical filtering element. The invention applies to the production of a selective angular filtering device and to a multidirectional optical detection system.

Claims

1. An angular optical filtering element (E.sub.i) optimized for the angular filtering about a given angle of incidence of operation (θ.sub.i, φ.sub.i), within a given spectral band, comprising: a first nanostructured band-pass spectral filter (11.sub.i, 301), comprising, within said spectral band, a first filtering central wavelength which exhibits a first angular dispersion curve, the angular dispersion curve representing the variation of the first filtering central wavelength determined as a function of the angle of incidence on the optical filtering element (E.sub.i); a second nanostructured band-pass spectral filter (12.sub.i, 302), comprising, within said spectral band, a second filtering central wavelength which exhibits a second angular dispersion curve representing the variation of the second filtering central wavelength determined as a function of the angle of incidence on the optical filtering element (E.sub.i), the second angular dispersion curve being secant with the first angular dispersion curve around the angle of incidence of operation (θ.sub.i, φ.sub.i).

2. The angular optical filtering element as claimed in claim 1, in which at least one of the first and second spectral filters operate in transmission mode.

3. The angular optical filtering element as claimed in claim 1, in which at least one of the first and second spectral filters operate in reflection mode.

4. The angular optical filtering element as claimed in claim 1, in which one of the first and second spectral filters operates in absorption mode.

5. The angular optical filtering element as claimed in claim 1, in which the first and second spectral filters are arranged in non-parallel planes.

6. The angular optical filtering element as claimed in claim 1, in which at least one of the first and second nanostructured spectral filters is a guided mode resonance filter, a filter with free-standing metal or dielectric gratings, or else a filter comprising a resonance of the MIM (metal-insulator-metal) type.

7. The angular optical filtering element as claimed in claim 1, in which at least one of the first and second nanostructured spectral filters is an interference filter.

8. A device for selective angular filtering at several given angles of incidence, comprising an array of angular optical filtering elements as claimed in claim 1, each optimized for angular filtering about a given angle of incidence of operation (θ.sub.i, φ.sub.i).

9. The device as claimed in claim 8, in which the first nanostructured spectral filters are arranged according to at least a first one- or two-dimensional matrix (11), comprising a first plane carrier, and the second nanostructured optical filters are arranged according to at least a second one- or two-dimensional matrix (12), comprising a second plane carrier.

10. The device as claimed in claim 9, in which the first and second carriers are arranged in non-parallel planes.

11. A multidirectional optical detection system comprising a device for selective angular filtering as claimed in claim 8 and an array of optical detection elements (15.sub.j), each optical detection element being associated with an angular optical filtering element for receiving the light flux transmitted by said angular optical filtering element at the angle of incidence of operation of said angular optical filtering element.

12. The multidirectional optical detection system as claimed in claim 11, in which one of the first or second spectral filters of an angular optical filtering element is an absorption filter formed by the optical detection element associated with said angular optical filtering element.

13. The multidirectional optical detection system as claimed in claim 11, in which the optical detection elements are bolometers.

14. The multidirectional optical detection system as claimed in claim 11, in which the optical detection elements are arranged according to a one- or two-dimensional matrix (15), comprising a plane carrier.

15. The multidirectional optical detection system as claimed in claim 11 furthermore comprising a housing (130) in which the device for selective angular filtering and the array of the optical detection elements are arranged.

16. A method for fabrication of a multidirectional optical detection system as claimed in claim 11 comprising: the fabrication of a first matrix of first spectral elements on a first plane carrier; the fabrication of a second matrix of second spectral elements on a second plane carrier; the fabrication of a third matrix of optical detection elements on a third plane carrier; the arrangement of said matrices in a housing.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Other advantages and features of the invention will become apparent upon reading the description, illustrated by the following figures:

(2) FIG. 1, the diagram of one example of a multidirectional optical detection system according to the present description;

(3) FIGS. 2A and 2B, two examples of crossing angular dispersion curves of a first spectral filter and of a second spectral filter, for the formation of an angular optical filtering element according to the present description;

(4) FIG. 3, a table depicting various possible configurations for a multidirectional optical detection system according to the present description;

(5) FIG. 4, one example of a spectral filter of the type using guided mode resonance for the formation of a spectral filter of an angular optical filtering element according to the present description;

(6) FIGS. 5A and 5B, examples respectively of a first spectral filter and of a second spectral filter for the implementation of an angular optical filtering element according to the present description;

(7) FIGS. 6A and 6B, the curves showing the transmission spectra respectively for the first and second spectral filters shown in FIGS. 5A and 5B for various angles of incidence;

(8) FIG. 7, a curve showing the transmission of an angular optical filtering element obtained by association of the first and second filters shown in FIGS. 5A and 5B;

(9) FIG. 8, a diagram showing the transmission TM at the exit of an angular optical filtering element obtained by association of the first and second spectral filters shown in FIGS. 5A and 5B, as a function of the angle of incidence and of the azimuthal angle;

(10) FIGS. 9A and 9B, exemplary implementations of multidirectional optical detection systems according to the present description;

(11) FIG. 10, one example of a device for selective filtering at several angles of incidence, designed for an implementation of a detection system such as shown in FIG. 9A;

(12) FIG. 11, another example of a multidirectional optical detection system according to the present description;

(13) FIGS. 12A to 12I, the steps in the fabrication of a matrix of spectral filters, according to one embodiment of the present description;

(14) FIG. 13, one exemplary embodiment of a multidirectional optical detection system according to the present description.

DETAILED DESCRIPTION

(15) In the figures, the elements are shown for the purposes of illustration and the scales are not adhered to. Furthermore, identical elements are identified by the same references.

(16) FIG. 1 shows schematically one example of a multidirectional optical detection system 10 according to the present description. The detection system 10 is designed to operate within a given spectral band depending on the application, for example in the infrared for the detection of human presence or the detection of gas. More precisely, the detection system may be designed to operate for example within the atmospheric transmission bands II (3-5 μm) or III (8-12 μm).

(17) The multidirectional detection system 10 comprises an array of detection elements 15.sub.j, designed for detection within the desired spectral band of the detection system 10, and organized in this example in the form of a two-dimensional matrix 15. In the case of an operation in the infrared, the detection elements are for example microbolometers, allowing the production of a low-cost detection system. Alternatively, the detection elements may be cooled quantum detectors implementing a PN or PIN junction using semiconductors such as CdHgTe (cadmium mercury telluride or CMT), or a superlattice of InAs/GaSb (indium arsenide/gallium antimonide), or quantum wells of the type of those used in quantum well detectors or QWIP (abbreviation for Quantum Well Infrared Photodetectors) or in quantum cascade detectors (or QCDs) based on GaAs (gallium arsenide) and InP (indium phosphide), or photo-resistive or photo-capacitive detectors, or else, as will be described in more detail in the following, nanostructured detection elements incorporating a band-pass spectral filtering function in absorption mode.

(18) The multidirectional detection system 10 also comprises an array of optical filtering elements E.sub.i, each optical filtering element E.sub.i, being optimized for angular filtering about a given angle of incidence of operation, within the desired spectral band. In the example in FIG. 1, the light flux transmitted by each angular filtering element E.sub.i at the angle of incidence of operation is received by several detection elements, the detection signal being then able to be summed within each angular filtering element. Alternatively, it is possible to only provide one detection element per angular filtering element E.sub.i.

(19) According to the present description, each angular filtering element E.sub.i comprises at least a first nanostructured band-pass spectral filter 11 and a second nanostructured band-pass spectral filter 12.sub.i. Thus, in the example in FIG. 1, the angular filtering element E.sub.1 comprises an entry spectral filter 11.sub.1 designed to receive the light incident on the angular filtering element E.sub.1 and an exit spectral filter of 12.sub.1 designed to receive the light flux transmitted by the first spectral filter 11.sub.1. In this example, the first and second spectral filters operate in transmission mode. In the example in FIG. 1, the first spectral filters 11, are organized in the form of a first, substantially plane, two-dimensional matrix 11 and the second spectral filters 12.sub.i are organized in the form of a second two-dimensional matrix 12, being substantially plane and parallel to the first matrix. In the example in FIG. 1, the patchwork layout consisting of a rectangular pattern is only one example. The disposition of the spectral filters in this plane is of any kind.

(20) According to the present description, each of the first and second spectral filters 11.sub.i, 12.sub.i composing the angular filtering element E.sub.i, respectively comprises, in the spectral band of operation of the multidirectional detection system, a first filtering central wavelength and a second filtering central wavelength, the first and second filtering central wavelengths having secant angular dispersion curves around the desired angle of operation for the angular filtering element In the present description, ‘angular dispersion curve of a spectral filter 11.sub.i, 12.sub.i’ refers to the curve representing the variation of the filtering central wavelength (or of the wave number, inverse of the wavelength) of the spectral filter as a function of the angle of incidence on the angular filtering element to which the spectral filter belongs. In the case of two spectral filters 11.sub.i, 12.sub.i operating in transmission mode and arranged so as to be substantially parallel, as in the example in FIG. 1, it will be understood that the angle of incidence on the angular filtering element is also the angle of incidence on each of the spectral filters 11.sub.i, 12.sub.i. In the case of spectral filters operating in reflection, or arranged in a non-parallel manner, the angle of incidence on the angular filtering element could differ from the angle of incidence on the second spectral filter, as will be explained in more detail in the following.

(21) Generally speaking, a nanostructured band-pass spectral filter exhibits, at a given azimuthal angle φ.sub.i, a filtering function (in transmission, in reflection, or in absorption mode) F.sub.i(λ, θ.sub.inc) depending both on the angle of incidence θ.sub.inc and on the wavelength A. Thus, by combining two spectral filters having different angular dispersions, the filtering function F of the whole assembly, equal to the product of the filtering functions of the two filters (F=F.sub.1×F.sub.2), will only be significant within a cone in which the filtering function through the two filters is significant. The axis of this cone is given by the angle (θ.sub.i, φ.sub.i) for which the dispersion curves of the two spectral filters intersect, in the present description referred to as angle of operation of the angular filtering element. The angles at the apex (2θ.sub.i, 2Δφ.sub.i) of this cone, characterized by the intervals [θ.sub.1−Δθ.sub.i, θ.sub.i+Δθ.sub.i] and [φ.sub.i−Δφ.sub.i, φ.sub.i+Δφ.sub.i], define the cone in which the filtering function is significant, in other words higher than a given percentage of the maximum value F.sub.max, of the filtering function. The angular intervals [θ.sub.1−Δθ.sub.i, θ.sub.i+Δθ.sub.i] and [φ.sub.i−Δφ.sub.i, φ.sub.i+Δφ.sub.i] are respectively referred to as zenithal angular bandwidth (2Δθ.sub.i) and azimuthal angular bandwidth (2Δφ.sub.i). For example, the cone of operation of the angular filtering element may be defined by the cone within which the filtering function is greater than F.sub.max/5.

(22) FIGS. 2A and 2B illustrate dispersion curves 21, 22 of two spectral filters, according to two examples. Here, the dispersion curves show the wave number σ.sub.c, inverse of the centering wavelength λ.sub.c of each band-pass spectral filter, as a function of the angle of incidence θ.sub.inc on the angular filtering element, for a given azimuthal angle. These figures show the angle of incidence θ.sub.i for which the two angular dispersion curves intersect and the angular bandwidth 2Δθ.sub.i. From these figures, it can be observed that the angular selectivity of the angular filtering element E.sub.i according to the present description depends on the differences between the angular dispersions of the first and second spectral filters composing it, and more precisely, on the difference in slope in the neighborhood of the angle of operation. Thus, in order to obtain a low angular selectivity (Δθ.sub.i large), filters may be chosen that have dispersion relations close to one another (FIG. 2A). In contrast, an angular filter with a high selectivity (Δθ.sub.i small) will be obtainable by assembling spectral filters having greatly contrasting dispersion relations, for example two filters with a large dispersion whose slopes are of opposite sign (FIG. 2B). It is furthermore observed that the angular selectivity of the angular filtering element E.sub.i of the invention is also dependent of the spectral pass bands of the first and second filters composing it. Indeed, the narrower spectral pass band of the two filters, the higher the angular selectivity (see FIGS. 2A and 2B: the narrower the bands 21 and 22, the smaller Δθ.sub.i) will be. Indeed, the angular sector in which the pass bands of the two spectral filters 11.sub.i and 12.sub.i overlap has a smaller zenithal angle at the apex (2Δθ.sub.i).

(23) The current technology of nanostructured band-pass spectral filters allows filters to be produced with controlled dispersion curves, allowing the formation of angular filtering elements with angles of incidence of operation and with zenithal and azimuthal angular bandwidths determined according to the application.

(24) For this purpose, a known solution is to employ algorithms referred to as “genetic algorithms”. The genetic algorithm derives its name from the analogy which is made with the natural selection within a species. It consists in exploring a population of objects as a function of their properties, and in relation with a targeted property. For this purpose, the variables to be optimized are defined (for example the dimensions of the structurings), a property with a target value (for example the transmission at a given angle) and a convergence criterion (for example the difference from the targeted transmission value). Selection and crossing operations allow new generations of objects to be created whose properties tend, in the course of the iterations, toward the target values. Such genetic algorithms are for example described in R. L. Haupt et al. (“Practical genetic algorithms”, Wiley-Interscience (2004)).

(25) By grouping a set of these filtering elements, as is illustrated for example in FIG. 1, it is possible to produce a device for selective angular filtering at several given angles of incidence. Furthermore, by arranging the first and second spectral filters in the form of matrices, it is possible to use the known techniques of microelectronics for a massively parallel, low-cost, production.

(26) In the illustrative example in FIG. 1, the first and second spectral filters 11.sub.i, 12.sub.i are filters operating in transmission mode. However, the current technologies of nanostructured filters allow spectral filters to be produced with a controlled dispersion curve in transmission, in reflection and in absorption mode.

(27) FIG. 3 shows a table schematically illustrating various possible arrangements, referenced 31 to 36, for a multidirectional optical detection system according to the present description. In each of the configurations, the spectral filter receiving the incident light flux (“entry spectral filter”) is referenced 301, the spectral filter receiving the light transmitted or reflected by the entry spectral filter is referenced 302 (“exit spectral filter”) and the detection element is referenced 303. The angle of incidence on the entry spectral filter is denoted θ.sub.1, this is also the angle of incidence θ.sub.inc on the angular filtering element composed of the first and second spectral filter. The angle of incidence on the exit spectral filter is denoted θ.sub.2, it is equal to the angle of incidence θ.sub.inc on the angular filtering element when the entry and exit spectral filters are in substantially parallel planes. When the entry and exit spectral filters are not in parallel planes, the angle of incidence θ.sub.2 on the exit spectral filter may differ from the angle of incidence θ.sub.inc on the filtering element.

(28) In practice, the determination of the angular filtering element composed of the first and the second spectral filter and giving, within a given spectral band, an angular filtering function about a given angle of incidence of operation, will be able to be achieved in the following manner. For the desired angle of operation, an exit spectral filter could be chosen with a known angular dispersion curve which will give, as a function of the angle of incidence θ.sub.inc on the filtering element, the filtering central wavelength of the exit spectral filter by a simple change of variable. The entry spectral filter could then be chosen with a known angular dispersion curve, which will filter at the angle of incidence θ.sub.inc this same filtering central wavelength. The configuration shown in the case 31 of the table thus shows two spectral filters 301 and 302 operating in transmission mode and which have between them a non-zero angle α, such that θ.sub.2=θ.sub.1−α. In order to determine more easily the desired region of operation of the angular filtering element formed from the two spectral filters, λ.sub.c1(θ.sub.1), angular dispersion curve of the entry spectral filter 301, and λ.sub.c2(θ.sub.2)=λ.sub.e2(θ.sub.1−α), angular dispersion curve of the exit spectral filter 302, could be made to coincide in such a manner that these dispersion curves intersect in the region of operation sought. In order to determine the appropriate angular dispersion curves, genetic algorithms such as previously described will for example be able to be implemented. Thus, reference could for example be made to S. Tibuleac et al. for the optimization of an all-dielectric GMR transmission filter (see S. Tibuleac et al. “Narrow-linewidth band-pass filters with diffractive thin-film layers”, Opt. Lett. 26.9, p. 584-586 (2001)).

(29) In the configurations displayed in the first column of the table, the entry spectral filter 301 operates in transmission mode. This may for example be a spectral filter using thick free-standing metal gratings (the thickness of the grating is large compared with the width of the slopes), such as described in the aforementioned article by R. Haïdar et al. This may also be a guided mode resonance filter, dielectric or metal-dielectric, such as described in the “Prior art” section of the present application. In the configurations shown in the second column of the table, the entry spectral filter 301 operates in reflection mode. This may also be a dielectric guided mode resonance filter or a nanostructured spectral filter formed from a dielectric grating (see for example the aforementioned article by P. Ghenuche et al.). The exit spectral filter 302 may be a transmission filter (configurations 31 and 32) or reflection filter (configurations 33, 34).

(30) The configurations shown in the cases 35 and 36 of the table represent the case where the exit spectral filter is integrated into the detection element 303 which itself has a function of band-pass spectral filter. In this case, the spectral filtering operates in absorption mode (see for example the aforementioned article by P. Bouchon et al.).

(31) For each of these filters, skillful use of simulation tools enables the physical characteristics of the structures to be determined allowing the desired angular dispersion curve to be obtained.

(32) The configurations in transmission mode, such as shown for example in the cases 31 and 35 of the table, are advantageous in that they allow the size of the system to be minimized. The configurations in which a spectral filter operates in reflection mode are bigger than the preceding ones but they can allow access to be gained to very high angles of incidence (in particular in the case of the association of a reflection filter and of a transmission filter), such as will be described in the following.

(33) Although the configurations presented in FIG. 3 implement angular filtering elements each comprising two spectral filters, it is possible to produce angular filtering elements with three spectral filters or more, with the aim for example of further increasing the angular selectivity. The spectral filters are then arranged together in such a manner that the succeeding spectral filter receives the light transmitted or reflected by the preceding spectral filter. In this case, the angular dispersion curves of each of the spectral filters, determined as a function of the angle of incidence on the angular filtering element to which it belongs, will be able to be secant (non-coincident) in the neighborhood of the angle of incidence sought for the angular filtering element.

(34) The applicants have highlighted the feasibility of an angular filtering element according to the present description, both by means of numerical simulations and of experimental tests. FIGS. 7 and 8 show simulation results obtained with nanostructured band-pass spectral filters such as described by means of FIGS. 4 to 6.

(35) The band-pass spectral filter 40 shown in FIG. 4 is a filter known as a guided mode resonance or GMR filter. The filter 40 comprises a waveguide 401 made of dielectric material (dielectric permittivity ∈.sub.d) comprising on each of its faces a one-dimensional metal grating (dielectric permittivity ∈.sub.m). The structure thus shown is highly reflective over the wavelength range in question because the metal grating has very narrow slits. At the resonance, on the other hand, the transmission of the component becomes significant. The orders ±1 diffracted by the grating are coupled with the eigenmodes of the dielectric waveguide. They are trapped in the waveguide and escape in transmission in the order 0 either directly or via a coupling of the orders ±1 with the diffraction grating. The detailed study of these gratings is described for example in the article by E. Sakat (see reference hereinabove). For the simulations presented in the present description, the B-Splines method described for example in the article by P. Bouchon et al. (“Fast modal method for sub-wavelength gratings based on B-spline formulation”, J. Opt. Soc. Am. A, 27 696-702 (2010)) is used.

(36) More precisely, the simulations have been carried out with an angular filtering element comprising an entry spectral filter 51, such as described in FIG. 5A, and an exit spectral filter 52 such as described in FIG. 5B.

(37) The entry spectral filter 51 is composed of a dielectric waveguide 401 comprising, on each of its faces, a one-dimensional (1D) metal grating made of gold, the whole assembly being surrounded by air. The gratings 402, 403 are referred to as “bi-atomic” gratings. This is because they comprise two patterns per period d, a slit of width a and a slit of width a+l. A filter of this type can allow a very flat dispersion relation to be obtained for angles of incidence in the range between −25 and +25 degrees (in the plane of azimuthal angle φ=0), and a TM polarization.

(38) FIG. 6A thus shows a series of transmission spectra calculated for an azimuthal angle of zero (φ=0) for angles in the range between 0 and 25° of a first spectral filter such as shown in FIG. 5A. For this simulation, the waveguide 401 is chosen to be silicon nitride (SiN.sub.x), with a dielectric permittivity ∈.sub.d=4, comprising, on each of its faces, a 1D metal grating made of gold. The parameters of the grating are: d=3 μm, t.sub.m=0.1 μm, a=0.2 μm, l=0.5 μm, t.sub.d=0.7 μm, n(SiN.sub.x)=2. In this range of angles, the maximum TM transmission at the resonance is close to 80%. It is observed that, between θ.sub.1=0° (normal incidence) and θ.sub.1=25°, the filtering central wavelength (around 4 μm) does not vary much.

(39) The second spectral filter 52 is composed, as is the first spectral filter 51, of a dielectric waveguide 401 comprising, on each of its faces, a one-dimensional (1D) metal grating made of gold. In contrast to the preceding case, the gratings 402, 403 only consist of a single pattern per period d, a slit of width a. A filter of this type can allow a very high angular dispersion in the plane of zero azimuthal angle (φ=0) to be obtained.

(40) The FIG. 6B thus shows a series of transmission spectra calculated for an azimuthal angle of zero (φ=0) for angles in the range between 0 and 25° of a second spectral filter as is shown in FIG. 5B. For this simulation, the waveguide 401 is chosen to be made of silicon nitride (SiN.sub.x), with a dielectric permittivity s.sub.d=4, comprising a 1D metal grating made of gold on each of its faces. The parameters of the grating are: d=2.5 μm, t.sub.m=0.1 μm, a=0.3 μm, t.sub.d=0.8 n(SiN.sub.x)=2. It is observed that the maximum transmission shifts by 4.23 μm at normal incidence to 3.7 μm for an incidence of 25°. The maximum TM transmission (at the resonance) also varies from 78% to 50%.

(41) Thus, the dispersion curves of the first and second spectral filters such as shown in FIGS. 5A and 5B, calculated in the plane of zero azimuthal angle (φ=0) and with TM biasing (magnetic field H parallel to the slits of the grating), with the parameters stated hereinabove, are very contrasted. The first spectral filter exhibits a virtually-zero dispersion, whereas the second shows a strong positive dispersion (the resonance is shifted toward the shorter wavelengths when the angle of incidence increases). It is therefore expected that, by combining the first and the second spectral filters in order to form the angular filtering element, a high spectral selectivity will result. It can be observed that the resonance wavelengths of the two filters coincide (λ.sub.c1=λ.sub.c2) around an angle close to 13.5°. At λ=4.005 μm, θ=13.5°, the transmissions of the filters are T.sub.1=70.73% and T.sub.2=68.04%. An angular filter is therefore expected that has a resonance for λ=4.005 μm, θ.sub.inc=13.5° with a total transmission T=T.sub.1×T.sub.2=48.1%.

(42) FIG. 7 shows transmission curves calculated with the two filters 51 and 52 previously described, separated by an air gap of 300 μm, for angles of incidence going from 0 to 25° and in the plane of azimuthal angle φ=0. In accordance with the preceding observation, a transmission resonant at λ=4.005 μm, a maximum transmission for λ=4.005 μm and θ.sub.inc=13.5° (T=48.1%), a narrow spectral width of the peak (full width at half-height FWHM of 60 nm and quality factor Q, defined by the ratio of the resonance wavelength over the width at half-height, of 67), an absence of transmission outside of the interval [3.8, 4.3] μm; as can be seen in FIG. 7, the transmission is less than 0.5% outside of this interval. The angular selectivity obtained is Δθ≈+8°.

(43) These simulations show that the transmission of the angular filter is correctly described by the simple law: T(λ, θ)=T.sub.1(λ, θ)*T.sub.2(λ, θ). It is thus possible to design the parameters of the spectral filters independently from one another and to predict the performance characteristics of the angular filtering element obtained by their combination.

(44) The results of the simulations presented hereinabove are situated within a plane of space of azimuthal angle zero, in other words φ=0 (as is indicated in FIG. 4). In practice, it may be sought to target a detection within a cone of axis (θ.sub.i, φ.sub.i) and with angles at the apex (2Δθ.sub.i, 2Δφ.sub.i).

(45) FIG. 8 shows results of simulations carried out with an angular filtering element such as previously defined, for all the planes in space, for a TM polarized incident wave (magnetic field H orthogonal to the plane of incidence, i.e. parallel to the slits in the case φ=0) and a detection also according to a TM biasing. More precisely, FIG. 8 shows a polar diagram representing the behavior of the angular filtering element as a function of the angle of incidence θ (radial axis) and of the azimuthal angle φ (angle in the polar diagram). The color scale corresponds to the integral of the TM transmission of the angular filtering element over the wavelength range [3.81−4.19] μm. This figure demonstrates a maximum transmission within a cone of light corresponding to θ.sub.i=13.5±8° and φ.sub.i=±20°. The stray light patches (θ close to 90°), visible in FIG. 8, may be shielded either by the housing of the detector or by a field limiter screen.

(46) Simulations made with TE incident biasing (electric field E parallel to the slits of the grating) show that, with an angular filtering element using this type of spectral filters, the TM polarized incident light is transmitted, along the axis of angular filtering, with a high transmission coefficient. Simulations made with a TM and TE incident biasing and with a TE detection show that very little light gets through the filtering element, whatever the zenithal θ and azimuthal φ angles of incidence. The angular filtering element is therefore a selective polarizing element within a light cone.

(47) FIGS. 9A and 9B show exemplary implementations within a volume 90, for example a room in a dwelling, of multidirectional optical detection systems 10 according to the present description. In the example in FIG. 9A, the detection system 10 is situated on the ceiling, whereas in the example in FIG. 9B, the detection system 10 is situated on a wall of the room. The multidirectional detection system, in this example, comprises angular filtering elements formed from spectral filters with one-dimensional gratings. The lines thus show schematically the slits of the grating. The plane φ=0 is orthogonal to the direction of the slits (it is in this plane that the angular selection of the filter is the most efficient). Only the TM polarized light (magnetic field H parallel to the slits of the grating) is transmitted. The angular filter allows the light within two cones centered on the directions (φ=0, θ) and (φ=0, −θ) to pass. The axes of these two cones are represented by the arrows in FIGS. 9A and 9B.

(48) In the example in FIG. 9A, when the detection system 10 is placed on the ceiling of the room, the angle of incidence θ (in other words, the angle that can be tuned) is in a vertical plane; the slits themselves are in a horizontal plane.

(49) FIG. 10 shows one example of a device for selective filtering at several angles of incidence, in the shape of a rosette, designed for the implementation of a detection system such as that shown in FIG. 9A, and also allowing space to be scanned in φ. In this example, the device for selective angular filtering comprises a array of angular filtering elements E.sub.i arranged in the form of one-dimensional matrices (barrettes), themselves laid out in the shape of a rosette. Each of the angular filtering elements E.sub.1, E.sub.2 and E.sub.3 selects a different cone (θ.sub.i, φ.sub.i=±20°). It is thus the various branches of the rosette that allow the various azimuthal planes, and hence the various corners of the room, to be targeted. The overlapping of the signals detected on the matrix of detectors (for example, signals coming from two different barrettes not sufficiently separated in φ) is possible but should be limited as far as possible in order to facilitate the exploitation of the data.

(50) In the example in FIG. 9B, when the detection system 10 is situated on the wall, the slits are advantageously oriented vertically. In this case, the plane φ=0 is horizontal (parallel to the ceiling), and all the corners of the room are scanned with the various angles θ. The implementation of a rosette such as that shown in FIG. 10 is therefore unnecessary in this case. In this configuration, it suffices to construct a patchwork layout of rectilinear filters (with vertical slits), each of the filters of which selects a different angle θ, which is equivalent to a single branch of the previous rosette. The implementation is simpler and the angular selectivity is no longer limited by Δφ but by Δθ (full width at half-height of the angular cone selected along θ). A higher angular selectivity may therefore be expected in this case (it is recalled that, in the illustrative example Δθ=±8°, while Δφ=±20°). Furthermore, if the patchwork layout of filters is placed in a corner of the room, it just needs to select angles θ in the range between −45° and 45° in order to scan the whole room. Similarly, a detector placed in the angle formed by the walls and the ceiling allows a detection over the whole room for angles θ in the range between −60 and 60 degrees.

(51) FIG. 11 shows another example of a multidirectional optical detection system 110 according to the present description.

(52) In this example, the multidirectional detection system comprises a first matrix 11 of entry spectral filters operating in transmission mode, a second matrix 13 of entry spectral filters operating in reflection mode and a matrix 12 of exit spectral filters operating in transmission mode. With each exit spectral filter (for example 12.sub.j or 12.sub.k) is associated one entry spectral filter, either from the first matrix of entry spectral filters (for example 11.sub.k) or from the second matrix of entry spectral filters (for example 13.sub.j), in order to respectively form angular filtering elements E.sub.j, E.sub.k. In this example, the first matrix of entry spectral filters 11, formed from transmission band-pass filters, allows the angles of incidence between around −40° and 40° to be collected; the second matrix of entry spectral filters 13, composed of reflection band-pass filters, allows the angles in the range between values less than −90° and −40° to be collected. A third matrix of entry spectral filters could be added operating in reflection mode, symmetric with the second matrix of entry spectral filters 13 with respect to the first matrix of entry transmission spectral filters 11 for collecting the angles in the range between +40° and values above +90°. It is thus possible, by virtue of the multidirectional detection system according to the present description, to go beyond 90° and to aim behind the detector. It is furthermore possible to differentiate on the detector matrix 15 both the different angles of incidence θ1 and θ2 and also the angle θ1 from its opposite −θ1.

(53) Advantageously, the height h between the plane of the matrix 15 of optical detection elements and the matrix 12 of exit spectral filters is chosen to be sufficiently large in order to avoid overlapping of the detection elements. Typically, a height h greater than 6 times the distance between the two matrices of spectral filters may be chosen. Each detection element, or array of detection elements, thus only receives the light flux F(θ.sub.i) incident on the filtering element whose angle of operation is θ.sub.i.

(54) The matrices of spectral filters will be able to be fabricated according to known technologies, depending on the nature of the spectral filters used.

(55) FIGS. 12A to 12I illustrate, more particularly, the steps for fabricating a matrix of spectral filters comprising guided mode resonance structures, of the type of those described in FIGS. 4, 5A, 5B.

(56) According to one example, the fabrication method can comprise the following steps: Preparation of a double-side polished substrate 120, transparent in the spectral band sought, for example a substrate of polished sapphire (FIG. 12A), and forming the carrier for the matrix; Deposition of a layer 121 of silicon carbide (SiC) in a sputtering tool (FIG. 12B); Implementation of a first electron lithography step allowing an etch resist mask to be defined, comprising the deposition of a layer of resist 122, the deposition of a thin film of gold 123 (FIG. 12C), electron lithography so as to form a first pattern corresponding to the grating 403 (FIG. 4) for the various angular filtering elements of the matrix, the removal of the layer of gold and the development (FIG. 12D); Etching of the layer of silicon carbide (121) by RIE (according to the abbreviation for the expression “Reactive Ion Etching”) (FIG. 12E); Deposition of a gold grating (124) in the slots created by the etching of the SiC (FIG. 12F); Removal of the resist and deposition of a thick layer (125) of SiC (FIG. 12G); Second lithography step with, where necessary, an alignment of the first and second pattern (FIG. 12H); Lift-off of a gold grating (FIG. 12I).

(57) Once formed, the matrices of spectral filters may be arranged in a suitable housing for receiving a matrix of optical detection elements, as is illustrated in FIG. 13.

(58) In this example, a housing 130 is designed to receive two matrices 11, 12, respectively of entry and exit spectral filters, and a matrix 15 of optical detection elements. The matrices 11 and 12 are for example held by shims 131 and are typically separated by a few hundreds of micrometers. The matrix 15 of detection elements is positioned at a greater distance from the matrices of spectral filters, typically a few millimeters. The housing may be a cryostat, for example when the matrix 15 of optical detection elements needs to operate in a cooled or cold environment. It may be a vacuum chamber, for example when the matrix 15 of detection elements is composed of bolometers. It will be evident for those skilled in the art that the housing will be able to be adapted according to the number and to the shape of the matrices of entry and exit spectral filters, and also of the matrix of optical detection elements.

(59) Although described via a certain number of detailed exemplary embodiments, the structure and the method of fabrication of the angular optical filtering element according to the invention comprises several variants, modifications and improvements which will be obvious to those skilled in the art, it being understood that these different variants, modifications and improvements form part of the scope of the invention, such as defined by the claims that follow.