Ultrasensitive optical detector having a high temporal resolution and using a grating.

TECHNICAL AREA

The present invention relates to an optical detector sensitivity, high temporal resolution, and in particular a superconducting single-photon detector (English, <i>superconducting single photon detector</i>), Simply called SDSS.

The object detector of the invention uses a network coupling (English, grating) and is operable under normal incidence.

The invention applies in particular to the detection and localization of malfunctions of a semiconductor integrated circuit, by detecting the emission of specific photons by defective transistors.

It also applies to the reception circuits and routing for telecommunications at very high speed, especially those intended for telecommunications satellites, because of the very low thermal dissipation of the object of the invention detector.

The invention applies also to coding and detection of the quantum code key in a cryptographic system.

It also applies to the manufacture of detector arrays for imaging very high sensitivity, as well as a photon tomography or correlated photons.

STATE OF THE ART

Emerging technologies such as computing and quantum cryptography, the test failure of integrated circuits in microelectronics, medical imaging photonic detection, detection of biological objects or the detection of very weak signals for telecommunications or astronomy, require radiation detectors, especially in the visible or near-infrared range, which are very fast, have a jitter (in English, jitter) very low, make very little noise and are extremely sensitive they must be able to detect very low flow, even a single photon.

On this subject, please refer to document [Romestain 04] which, like the other documents cited below, is mentioned at the end of this description.

SSPDs potentially have all the qualities mentioned above and are good candidates to replace avalanche photodiodes and current photomultiplier tubes, whose performance is limited, especially in the infrared range.

SSPDs exist as STJ is to say superconducting tunnel junctions (in English, <i>superconducting tunnel junctions)</i> and as bolometers, which include bolometers called HEB, namely bolometers hot electron (in English, <i>hot electron bolometers)</i> which are the fastest.

The HEB use ultrathin superconducting films, whose thickness is less than 10 nm, to obtain very short characteristic times of the order of 10 ps (see [<patcit id="pcit0001" dnum="FR2812455"><text>FR 2812455</text></patcit>]). For these films, the preferred material is niobium nitride (NbN) cubic phase B1.

In these HEB, NbN film is grown epitaxially on a substrate which is typically made of sapphire, and the orientation is 1<o>1</o>02 (Plan R); and after structuring, the film meanders whose width is of the order of 100 nm, the film thus having the form of a coil in the active part of the detector (see [Villégier 04]).

1A is a schematic sectional view of such a detector. In this figure, the NbN film and substrate respectively have the references 2 and 4. Figure 1B is a schematic top view of the detector.

A light beam I enters the latter, at normal incidence, and it is coupled via an optical fiber (not shown). References R and T correspond to the light beams reflected and transmitted. The projection of the incident beam I on the detector is shown in dotted lines in Figure 1B where it has the reference 1.

Manufacture, operation and characterization of these SDSS detectors are described for example in [Goltsman 03]. The efficiency of detection, or photon conversion efficiency in an electric signal, is a key parameter that must be optimized and which would be 100% for an ideal detector.

It depends on the filling ratio (recovery rate of the incident beam and the active area of ​​the detector), the optical absorption in the layer of NbN and the hot spot capacity (in English, <i>hot spot)</i> which is formed as a result of photon absorption, to create a transient resistive barrier over the entire width of the superconductive track. We distinguish the detection efficiency of the quantum efficiency which is defined for a recovery rate of 100%.

The serpentine structure was designed to increase the fill rate compared to a simple straight track of NbN deposited on a sapphire substrate and illuminated at normal incidence. In recent achievements (see [Yang 05]), the occupancy rate reached 75%.

Optical absorption by NbN is of the order of 30% for a 4 nm thick, the wavelength of 1,55μm. The only way to increase would be to increase the thickness of the layer of NbN, but the ratio between the section of the hot spot and the section of the superconducting tape would decrease, since the width of this band is limited to about 50 nm and this will cause a drop in overall efficiency.

The detection efficiency can not exceed, in theory, 20 to 25% in 1,55μm (for a 75% fill factor) with this architecture is mainly limited by the very small thickness of the illuminated NbN layer under normal incidence.

It would probably be difficult to increase substantially detection efficiency without changing the design of the optical coupling. Sensors made with a 50% fill factor also have efficiencies of 5 to 10% to 1,55μm (see [Korneev 05]).

Several solutions have been proposed to increase the coupling of light in the absorbent region of the detector.

It has for example proposed to use a mirror and anti-reflection layer. The light passed through the NbN without having been absorbed is then returned to the NbN by a concave mirror (see [<patcit id="pcit0002" dnum="FR2812455"><text>FR 2812455</text></patcit>]) Or a mirror plane (see [The Coupanec 03] and [06 Rosfjord]). An anti-reflection layer, which is deposited on the rear face of the substrate if the impact is through the rear, further reduces the reflection losses of the incident wave. Detection efficiency reaches values ​​of the order of 50%.

It was also proposed to use a waveguide by coupling (see [Jackson 03]). According to this approach, which is schematically illustrated by Figure 2, the light 6 after a single mode optical fiber 8 is focused along a line by a diffractive optical element 10 and injected by the latter into a multimode waveguide plane silicon 12 which is formed on a silica substrate 14. this waveguide is specified that is monomode vertically but multimode laterally.

A coil NbN 16 is deposited over the silicon guide. The ends of this coil are respectively provided with contact pads 18 and 20. The coil 16 absorbs the evanescent wave associated with the guided mode.

A photon, that is to be detected is likely to be absorbed by a branch formed by the meander coil, at each passage of the evanescent wave. The absorption of light energy by NbN can in principle approaching 100% if the meanders are enough. But, thus, the detection efficiency is limited by the long length of NbN.

In addition, the feasibility of growing monocrystalline NbN on silicon has not been clearly demonstrated: the epitaxial growth of NbN on Si requires delicate silicon surface treatments and the study of the deposit of interface layers by epitaxy.

It may also refer to the document [Wang 04]. In the latter, we study the absorption of a layer of NbN, placed in a stack of planar layers, wherein a guided wave propagates.

An example of such a structure is schematically shown in Figure 3A which shows a substrate 22 covered with a sapphire layer 24, serving as a waveguide, and an NbN layer 26 covering this layer 24. in one embodiment, which is schematically shown in Figure 3B, the layer of NbN is between the substrate and the sapphire layer.

If the waveguide is such a layer of sapphire appropriate thickness, length NbN necessary for absorbing light is typically a few tens of micrometers (the calculations being made in one dimension, for guiding planes).

However, neither the structuring of the waveguide and NbN in the plane of the layers or the practical implementation of the sensor are not considered in this document [Wang 04]: we are given no technological solution including material selection, a method and conditions for injecting light into the structure.

PRESENTATION OF THE INVENTION

The present invention aims to solve the problem of the design of an optical detector sensitivity, high temporal resolution, in particular an SDSS type detector, the detector having a significantly higher detection efficiency than those obtained in the prior art, this detector being able to operate at normal incidence and further being achievable by existing technology.

The normal incidence operation allows to consider a simple positioning system and compact in a cryostat, and the possibility of extending the principle of coupling to a matrix of detectors used, for example in imaging.

Specifically, the present invention relates to an optical detector for detecting at least one photon, this detector comprising:<ul><li>a dielectric substrate having a refractive index n_oand</li><li>a detecting element forming a serpentine, the detecting element being placed on the dielectric substrate and provided for generating an electrical signal from the energy of the photons received,</li></ul>this optical detector being characterized in that it further comprises<ul><li>a network (in English, <i>grating</i>) Dielectric dimensional, having parallel lines, namely first lines that are made of a first dielectric material having a refractive index n_H, And second lines that alternate with the first lines and which are made of a dielectric medium having a refractive index n_B, The dielectric dimensional network being placed above the detecting element, the assembly constituted by the dielectric-dimensional network and the sensor element having a resonant absorption of the photon, under a given incidence and for a given polarization of this or these photons, n_H being greater than n_o and n_Band</li><li>a superstrate having a refractive index n, this superstratum being placed above the one-dimensional dielectric grating, n_H further being greater than or.</li></ul>

The structure of the optical sensor, object of the invention, considerably improves the detection efficiency using the dielectric dimensional network located above the sensor element forming a coil.

Note that the detector, object of the invention, must operate with a very low photon flux. The probability of absorption of a photon is assumed to be proportional to the local intensity of the field that could have an important and continuing with photon flux (same mappings).

Preferably, the given incidence is a normal incidence.

The superstrate may be a gas, for example air, a liquid, solid or even empty (in which case neither is 1).

According to a preferred embodiment of the invention, the pitch (in English pitch) A one-dimensional dielectric grating is less than λ_o/not_o, Where λ_o represents the wavelength in vacuum of the photon, and the pair (w_r/ Λ, d) is selected to yield the resonance absorption, where w_r represents the width of the first lines of the one-dimensional dielectric grating and d represents the thickness of this dimensional dielectric grating, and A is furthermore less than λ_o/not_i.

The difference n_H-not_B is preferably greater than 1.

According to a preferred embodiment of the optical sensor object of the invention, the sensing element is made of a superconducting material.

This superconducting material may be a superconducting nitride phase. This phase may be niobium nitride.

According to a particular embodiment of the invention, the serpentine includes parallel lines, connected to each other, and each line of the serpentine is positioned facing one of the lines of the one-dimensional dielectric grating.

According to a particular embodiment of the optical sensor object of the invention, the serpentine includes parallel lines, connected to each other, and each line of the serpentine is positioned facing one of the first lines of the one-dimensional dielectric grating.

According to a first particular embodiment of the invention, the serpentine includes parallel lines, connected to each other, and lines of the serpentine and the first lines of dimensional dielectric grating have the same width and form periodic units having the same period.

In a second embodiment of the invention, the first lines of one-dimensional dielectric grating have different widths W_rL and W_rS and the first lines whose width is w_rL alternate with those whose width is w_rS.

The second lines of the dimensional dielectric grating can also have different widths.

According to a particular embodiment of the invention, the optical detector further comprises an intermediate layer which is made of a dielectric material having a refractive index n greater than n_B and which is placed between the detecting element and the one-dimensional dielectric grating.

The present invention also relates to an optical detector, for detecting one or photons, this optical detector including a polarization splitter (English, <i>polarization splitter</i>) Having a track (in English, <i>Harbor</i>) Input for receiving the photon, and two output channels which are connected to two optical detectors, respectively, according to the invention.

The present invention further relates to an optical detector array according to the invention.

BRIEF DESCRIPTION OF DRAWINGS

The invention will be better understood from reading the description of embodiments given below, purely by way of non-limiting example, with reference to the accompanying drawings, wherein:<ul><li>1A is a schematic sectional view of a known detector type SDSS, and has already been described,</li><li>1B is a schematic top view of the sensor of FIG 1A, and has already been described,</li><li>Figure 2 is a schematic perspective view of another known detector type SDSS, and has already been described,</li><li>3A and 3B are schematic views of planar multilayer structures known to guided waves, and have already been described,</li><li>Figure 4 is a schematic sectional view of a first example of the object detector of the invention,</li><li>5 schematically illustrates the waves that are brought into play in the network that includes the detector of Figure 4,</li><li>Figures 6 to 8 are schematic sectional views of other examples of detectors in accordance with the invention, and</li><li>Figure 9 is a schematic view of an optical detector comprising a polarizing beam splitter and two detectors according to the invention.</li></ul>

DETAILED DESCRIPTION OF SPECIFIC EMBODIMENTS

Figure 4 is a schematic sectional view of an example of the object detector of the invention. This detector is intended to detect an incident light 28, constituted by a single photon or more photons.

The structure seen in Figure 4 comprises a dielectric substrate 30, whose index of refraction is denoted n_oAnd an ultra-thin superconducting layer 32, such as NbN, having a thickness of a few nanometers, on the dielectric substrate 30. The latter is suitable for deposition of the layer 32.

This layer 32 is intended to generate an electrical signal from the energy of the photons received. It forms a coil and therefore meanders as in the prior art.

Reference is made for example to Figure 1B of the accompanying drawings where the layer comprises parallel lines connected to each other by segments which are shorter than the lines and are perpendicular thereto.

Other configurations are possible for the coil, preferably configurations in which the coil covers the largest area possible in the area illuminated by the incident light, so that a maximum photon is absorbed by the superconductor which is done coil.

The detector of Figure 4 also comprises a network (in English, <i>grating</i>) Dimensional dielectric 34, consisting of parallel lines 36 which are formed over the superconductive layer 32, a dielectric material of high index n_H. These lines 36 alternate with other lines 38 that are made of a dielectric medium of low index n_B. As shown, lines 36 and 38 are parallel to the lines of the layer 32.

Furthermore, the detector comprises a superstrate 40 whose index of refraction is noted ni. This superstrate is placed above the dielectric grating 34.

The period or not (in English, <i>pitch</i>), The network 34 is marked A, the width of the high index W lines_r and the width of the low index w lines_g.

It should be noted that, in the example of Figure 4, all the lines 36 of grating 34 have the same width W_r and then said the network is symmetrical.

In addition, the superconducting lines 44 of layer 32 are separated by gaps 46 which are made of a dielectric material whose index of refraction is denoted n_F.

The operating principle of the Figure 4 sensor is based on the excitation of a resonance absorption of the incident light 28 in the segmented guide formed by the network 34 (see below), for one polarization and a length wave data, for example, a TE polarization, the incident light 28 (the polarization direction then being parallel to the grating lines 34).

In the absence of superconductive layer 32, the incident light 28 is partially reflected in the superstrate 40, partially transmitted in the substrate 30, and partially diffracted by the segmented guide which forms the network 34. The guided light is then partially re-diffracted in the superstrate 40 and the substrate 30.

The light transmitted directly in the substrate and the light coupled then re-diffracted in this substrate interfere destructively: their phase shift (in English, <i>phaseshift</i>) Is equal to π (see [Rosenblatt 97]).

For a certain wavelength in a vacuum (in English, <i>vacuum)</i> λ₀ of the incident light 28 and a certain orientation of this incident light from the direction of the grating lines, the two waves have the same amplitude and destructive interference are complete. The transmission is zero and all the incident light is reflected. The described structure is then a resonant reflection structure (see [Nevière 73]).

5 schematically illustrates the waves which are brought into play in a network 34 which has a resonant reflection, as in the example that was described above.

In Figure 5, we still see the incident light 28 (normal incidence in the example shown) and references 48, 50, 52, 54 and 56 respectively represent the light directly reflected, light guided by the segmented guide, light directly transmitted light guided by the segmented guide and then radiated in the superstrate and the light guided by the segmented guide and then radiated into the substrate.

The π letter symbolizes the fact that the light transmitted directly in the substrate and the light coupled then re-diffracted in the substrate interfere destructively.

The principle of operation that has just been described applies to a normal incidence of the light 28, the light 28 then falling perpendicularly to the surface of the superstrate 40.

However, it can be applied to any angle of incidence of that light, from 0 ° to 90 °.

Moreover, this principle applies to the case of Figure 4 where the layer lines 36 and 38 has the dual function of network and guide. It is then said that the guide is segmented. But this principle can also be applied in case the network and the guide are two layers.

In the literature, the segmented guides have first studied with low index contrast, as spectral filters of narrow bandwidth (less than 1 nm).

He was then shown (see [Brundrett 00]) with a contrast of important clue, and with normal incidence, certain geometrical configurations of networks provided access to much broader resonances in reflection from the perspective of the length wave (100 to 200 nm), which is favorable for sensing applications as is the case for the present invention.

The angular acceptance of the device is also much wider and allows to couple a focused beam. Finally, the tolerance on the thickness of the network is substantially increased.

It is specified that the superstrate may be simply formed by air (or then being equal to 1).

Similarly, the network 38 of lines 34 may be air.

In the present invention, a high index contrast is preferably used, which is typically such that the difference n_H-not_B is greater than 1.

The superconducting layer structured and very thin (thickness of a few nanometers) perturbs the resonance by absorbing part of the light. The transmission is not strictly zero at resonance but remains low, and reflection may decrease substantially. Typically, 80% of the incident light can be absorbed.

Absorption may be optimized by slightly modifying the geometry of the network 34, as regards its fill factor and its thickness.

The requirements to obtain a resonance absorption are:<ul><li>guide provided: No_H&gt; n_i, not_H&gt; n_o, not_H&gt; n_B, Preferably n_H-not_B &gt; 1,</li><li>no diffracted order other than the specular order Λ &lt;λ₀/not_i and Λ &lt;λ₀/not_o (The network 34 and having a pitch smaller than the wavelength in a vacuum),</li><li>selecting a pair (f, d) which gives a resonance, where f is the network of the filling factor, such that f = w_r/ Λ, and d is the thickness of the grating 34.</li></ul>

In a purely indicative and non-limiting, the following values ​​can be used for example as described above with reference to Figure 4:<ul><li>λ₀ = 1554nm;</li><li>not_i = 1 (constitutive mid superstrat: air); not_o = 1.746 (the material of the substrate: sapphire); not_B = 1 (constitutive mid lines 38: air); not_H = 3.48 (the material lines 36: silicon); A = 800nm;</li><li>Λ_NbN = 400 nm; f_NbN = 0.5; d_NbN = 4 nm; not_F = 1.45 (A_NbN : Not coil, d_NbN : Thickness of the coil, the coil filling factor f_NbNW =_NbN/ Λ_NbN with W_NbN: Width of the coil 44 lines).</li></ul>

The absorption of the structure which is schematically shown in Figure 4 was calculated by RCWA that is to say by rigorous analysis of coupled waves (in English, <i>Rigorous coupled wave analysis)</i> according to the parameters f and d. Several resonances appear; they are associated with different modes of the segmented guide.

Maximum absorption A_max 78% is reached for a line width w_r equal to 660nm, from where W_g = 140 nm, and a thickness d equal to 460nm.

The AW tolerances_rFWHM and AW_gFWHM on the parameters W_r and W_g, Expressed as FWHM that is to say, full width at half-height (in English, <i>full width at half maximum</i>)<i>,</i> worth 42nm W_r and W_gAnd tolerance Dd_FWHM d is also on 42nm. These tolerances are not very critical from the point of view of technological realization of the structure.

The RCWA method was also used to calculate the reflectance spectra, transmission and absorption of the structure of Figure 4 as a function of the wavelength, as well as the evolution of absoption of this structure as a function of the incidence angle of the light.

It is found that the width Δλ_FWHM of the resonant wavelength is equal to 60 nm. It is relatively wide and therefore practical for use.

In addition, it is found that the angular width Δθ_FWHM resonance is equal to 23 °. It is quite sufficient for effectively coupling a light beam with a diameter of 10 .mu.m and which is derived from an optical fiber.

the maximum absorption is stated that 78% is reached when the grating lines are centered on NbN lines of the coil. Otherwise, the absorption may fall to about 20%. Qualitatively, the network allows to focus the light preferably in the network lines, and the absorption is maximal if NbN is located on the lobes of the optical electric field.

this result was obtained by simulating the intensity of the optical electric field at the resonant wavelength in a period of the grating by the FDTD method, that is to say, the method of finite differences in the time domain (in English, <i>finite-difference time-domain method</i>)<i>.</i>

As the period of the lobes of the optical electric field is equal to the period of the network, you can remove the NbN lines which are located under the low index intervals, where the field is very low. The FDTD simulations then give an absorption of 82% for a 1560nm resonance length.

This solution has the advantage of halving the length of NbN necessary, and therefore to reduce the noise caused by the quantum and classical fluctuations, as well as non-uniformities width of the superconducting line.

The decrease of the total length of the line and the associated kinetic inductance allows for a faster detector. This advantage is highlighted in [Kerman 06].

Furthermore, it indicates that a uniform thin layer of a dielectric material having a refractive index equal to n_F, Can be added above the coil NbN (for reasons of technological embodiment) without affecting the absorption.

In a variant of the structure shown in Figure 4 was repeated using the same period and the same line width for the NbN network and the high index network. The structure is simpler to perform because it allows to use a single standard photolithography and etching achieve simultaneous, but the precision required on the size is a bit larger.

In a purely indicative and non-limiting, the following values ​​are used in the case of this variant:<ul><li>λ₀ = 1549nm;</li><li>not_i = 1; not_o = 1, 746; not_B = 1, n_H = 3, 48; A = 640nm; w_r = 534nm (f = 0.835) where w_g = 106nm; d = 240nm;</li><li>Λ_NbN = 640nm; f_NbN = 0, 835; d_NbN = 4 nm; and N_F = 1.</li></ul>

We then obtain the following results:<ul><li>AT_max = 77%; AW_{r FMHM} AW =_{g FWHM} = 23 nm; Dd_FWHM - 37nm; Δλ_FWHM = 50 nm; and Δθ_FWHM = 18 °.</li></ul>

In another variant, which is schematically shown in Figure 6, is added a uniform dielectric layer of high index 60 between the superconducting layer 44 and the network 34. The thickness of layer 60 is denoted d_WG; its refractive index n_I is greater than n_BOr and n_o, With preferably n_I-not_B&gt; 1; not_I is for example equal to n_H.

In this case, the network 34, having a thickness d_GEnsures only the light coupling to the guide consists of the uniform layer 60.

In a purely indicative and non-limiting, the following values ​​are used in the case of this other variant:<ul><li>λ₀ = 1549nm; </li><li>or = 1; not_o = 1, 746; not_B = 1; not_H = 3, 48; A = 800nm; W_r = 388nm where W_g = 412 nm; d_G = 235 nm; d_WG = 312 nm;</li><li>Λ_NbN = 400 nm; f_NbN = 0.5; d_NbN = 4 nm; not_F = 1.45.</li></ul>

We then obtain the following results:<ul><li>AT_max = 78%; AW_{r FWHM} AW =_{g FWHM}&gt; 120 nm; Dd_{WG FWHM} = 54nm; Δλ_FWHM = 112nm; and Δ_θFWHM = 29 °.</li></ul>

The latter structure also has the advantage of being very insensitive to the network fill factor of 34.

more can increase the absorption of the optical detector according to the invention using an "asymmetric" network instead of a symmetrical network.

An example of a detector having an asymmetric network is shown schematically in Figure 7 where the network 32 of Figure 6 is replaced by an asymmetric network 62 in which lines 64 having a width w_rL alternate with lines 66 having a width w_rs different w_rL.

It specifies that lines 64 and 66 are still made of dielectric material having a refractive index n_H.

For a symmetrical network, there is a selection rule that allows a single resonance in one of the two edges of the second band (see [Vincent 79]). The asymmetry of the network raises degeneration (see [Vincent 79]) and a simulation of an asymmetric network structure RCWA can observe that the resonances are split in the (f, d) for each TE mode guide. In addition, two resonances associated with different TE modes can intersect in the plan, which can increase absorption.

In a purely indicative and non-limiting, the following values ​​are used in the example of Figure 7:<ul><li>λ₀ = 1546nm;</li><li>or = 1; not_o = 1, 746; not_B = 1; not_H = 3, 48; A = 1600nm; W_rL = 732nm; w_rS = 549nm; the space W_g between lines 64 and 66 is constant: W_g = 160nm; d = 448nm;</li><li>Λ_NbN = 400 nm; f_NbN = 0.5; d_NbN = 4 nm; not_F = 1.45.</li></ul>

It should be noted that the network A of the period 34 is equal to w_r+ w_g in the example of Figure 4 and w_rL+ w_rs2w +_g in the example of Figure 7.

We then obtain the following results:<ul><li>AT_max = 92%; Aw_{rL FWHM} Aw =_{rS FWHM} Aw =_{g FWHM} = 40nm; Dd_FWHM = 39nm; Δλ_FWHM = 56nm; and Δθ_FWHM = 21 °.</li></ul>

In addition, A_max is equal to 88% if we suppress the NbN lines which are located under the intervals between the high index lines for W_rL = 742nm, w_rs = 556 nm and d = 444nm.

One can "combine" the detectors of Figures 6 and 7. This possibility is illustrated by the example that is shown schematically in Figure 8. In this example, the detector comprises uniformly high index layer 60 and the network 62 to profile asymmetrical, which is then between the layer 60 and the superstrate 40.

This gives more degrees of freedom to the optical detector and obtain absorption resonances that are optimized with different geometric configurations.

In a purely indicative and non-limiting, the following values ​​are used in the example of Figure 8:<ul><li>λ₀ = 1550nm;</li><li>or = 1; not_o = 1.746; not_B = 1; not_H = 3.48; A = 1600nm; w_rL = 616nm; w_rs = 431nm; space between lines 64 and 66 constant: W_g = 276nm; d_G = 440 nm; d_WG = 598nm;</li><li>Λ_NbN = 400 nm; f_NbN = 0.5; d_NbN = 4 nm; not_F = 1.45.</li></ul>

We then obtain the following results:<ul><li>AT_max = 94%; AW_{rL FWHM} AW =_{rS FWHM} AW =_{g FWHM} = 116nm; Dd_{G FWHM} = 43nm; Δλ_FWHM = 43nm; and Δθ_FWHM = 12 °.</li></ul>

In all structures considered above, the propagation distance of a guided wave is very short. The light is laterally confined in a substantially equal area to the size of the beam formed by the incident light. this was verified by FDTD simulation for a Gaussian beam having a diameter equal to 10 .mu.m.

We can therefore use a coil having a length L NbN comparable to that used in the prior art and which is a few hundred micrometers, or a length L / 2 if the deletion of the NbN lines which are located under the gaps between the high index lines.

Furthermore, it specifies that the widths W_g trunks 38 of 34 may also be uneven (as can the widths w_r lines 36). This is of interest: there may be the case where the network consists of a repeating pattern of alternating lines 36 and 38, this pattern comprising a plurality of lines unequally separated.

Moreover, exept the coil material 32, which is typically a superconducting material, the materials used in the invention preferably have a very low extinction coefficient of less than 5x10^-4, Which makes negligible absorption losses in these materials.

It should also be noted that the detectors in accordance with the invention, as just described, are designed for a given polarization (TE or TM) of the incident light. In the case of incident light of unknown polarization may be used a polarizing beam splitter and a detector as described above on each of the channels (English<i>ports</i>) Output of the separator.

This is schematically illustrated by Figure 9 which shows a polarization separator 68 of which the input path 70 receives the detection light 72 and the two output channels 74 and 76 are respectively coupled to two detectors in accordance with the invention 78 and 80 to inject a TE polarized beam and a TM polarized beam, respectively, resulting from the decomposition of the light 72 by the separator 68.

Examples of the optical sensor, object of the invention, were given in the case of a single pixel whose typical dimensions are 10x10μm and which is illuminated by a focused light beam, coming for example from an optical fiber of the same diameter (10 .mu.m).

However, the principle of the invention can be extended to the design of a pixel matrix that is illuminated by a broader light beam, for example for ultrasensitive imaging applications, in which case the angular width criterion may be relaxed.

The detection of the light energy in the superconductor can be made:<ul><li>either according to the principle of hot electron bolometers threshold, which use a current bias subcritical, including here also the PSC mechanism or phase slip centers (in English, <i>Phase-slip centers</i>)<i>,</i></li><li>or by measuring the fast variation of Josephson current (detection of a "Josephson vortex") of a superconducting junction (STJ) or a SQUID,</li><li>or by detection of the variation in impedance of a superconducting tunnel junction that is polarized in its "gap" quasi-particle according to the principle of SIS receivers (omit the Josephson current using a magnetic field ).</li></ul>

Is given in the following examples of processes for making sensors in accordance with the invention.

Such a detector is achievable by well controlled technological means (deposition of thin films, substrates carry). The layer thicknesses are not critical (tolerance of ± 5%), except for the thickness of the superconducting layer, which is known to master by epitaxy.

The methods to be described can be applied to the manufacture of a detector having a single pixel or to the manufacture of an array of such detectors.

The following gives an example of fabrication of the structure which is schematically shown in Figure 4.<ol><li>1. It begins by forming a coil structure in a superconducting NbN layer, having a thickness of about 4 nm and whose refractive index n is about 2.5 + 4,5j where j²= -1, By epitaxy on a sapphire substrate having an index of refraction n_s is equal to 1.75 at 1550 nm as in the prior art, for example with a step of 400 nm and a 200 nm track width; and also is etched positioning patterns NbN (not shown) on the substrate. Then forming electrical contacts, gold and platinum, the thickness is about 15 nm at both ends of the serpentine structure, by a peeling technique (English, lift<i>off).</i></li><li>2. Next, a layer of SiO is deposited₂ over the coil and contacts, by spraying (in English, <i>sputtering)</i> or by PECVD at 350 ° C, this layer having a thickness of about 60nm and a refractive index n_F equal to 1, 45. Then a chemical mechanical polishing is carried out to a depth of approximately 40 nm to reach contacts.</li><li>3. then prepared an SOI (silicon on insulator), comprising firstly a thin layer of Si having a thickness of 460 nm and a refractive index n_H equal to 3.48, this thin layer comprising positioning patterns, and secondly a layer of SiO₂ between the SOI substrate and the thin layer of Si. optionally is deposited a very thin layer of SiO₂ on this SOI substrate to have better adhesion during subsequent bonding of the SOI substrate on the sapphire substrate, this very thin layer having a thickness of about 5nm to 10nm. This is followed by etching (in English,<i>etching)</i> dry or chemical deep windows in the Si substrate (part of the SOI substrate) to the layer of SiO₂To disengage the contacts and the positioning patterns.</li><li>4. This is followed by a collage "full plate" of the SOI substrate on the sapphire substrate (Al₂O₃) By molecular adhesion at room temperature, after conducting an alignment in the area visible through the silica, according to the positioning patterns, a lateral alignment accuracy of + -0,5μm being ample. Optionally, is annealed to a few hundred degrees Celsius above room temperature.</li><li>5. This is followed by a thinning of the Si substrate by chemical mechanical polishing or by planing down the layer of SiO₂ then this layer is etched. </li><li>6. then performed electron beam lithography and etching (in the example n_B is 1) of the Si layer, optionally in alignment with the lines of NbN to achieve the maximum absorption (required accuracy: + -50nm). It is specified that the superstrate is constituted by air.</li><li>7. Then, the optical fiber is put in place for the propagation of the incident light beam that is to be detected. For this implementation, we proceed to active alignment of the optical fiber on the detector. then sets the optical and electrical connections and cooled detector to its operating temperature (eg, 4K).</li></ol>

The size of the sapphire substrates can be chosen to be compatible with the silicon substrates which are used in the microelectronics industry (100 or 200mm). This makes possible the low cost production of many chips on a single substrate.

Regarding variations of the structure of Figure 4, which has been discussed above, in particular structures which are shown schematically in Figures 6, 7 and 8, the method of manufacturing the above described further comprises a step of depositing a layer of amorphous Si or polycrystalline (n_H and N_F and being equal to a number between 3 to 3.5) on the coil, before the network formation, the latter being then formed on the Si layer thus deposited amorphous or polycrystalline.

The following describes another method of manufacture, relating to the case where the silicon lines 36 follow the superconducting coil.

In this embodiment of the self-aligned structure, are implemented the steps of:<ol><li>1. Filing a superconducting NbN layer (thickness of about 4 nm) epitaxially grown on a sapphire substrate, as in the prior art.</li><li>2. Training peel (in English, lift-off) electrodes eg gold or aluminum, a thickness of about 100 nm.</li><li>3. depositing an amorphous silicon layer (180 nm thick) by magnetron sputtering.</li><li>4. coil formation (line width 250 nm, not 730 nm) by electron beam lithography and simultaneous etching silicon and NbN, by any means of a hard mask made of silica.</li><li>5. Filing a ZnS layer (thickness 400 nm) by sputtering and chemical mechanical polishing to about 90 nm above the silicon.</li><li>6. Resumption of contacts by etching ZnS.</li></ol>

The structure proposed in the present invention is achievable by well-mastered technological means (deposition of thin films, substrates carry). The layer thicknesses are not critical (tolerance of around<u>+</u>5%), except that of the superconductor, which is known to master by epitaxy.

In addition, the specified methods may be applied to the realization of a detector having a single pixel or a matrix of detectors.

In the figures, does not show the means for biasing the sensing member (superconducting coil) or the signal processing means provided by this element.

Furthermore, instead of NbN, one can use other superconducting materials to form the sensing element, for example other nitride phases (e.g. MoN or TiN) or cuprate phases (eg YBa₂Cu₃O_7-x or ReBa₂Cu₃O_7-x).

More generally, instead of being made of a superconducting material, this sensing element may be made of any material capable of detecting low light energy and to restore an electrical signal corresponding to this energy. This may be for example a semiconductor or bolometric material.

In the previous examples, the superstrate is air but may be another gas (all gases have a close refractive index of 1) or a liquid, for example liquid helium (whose index of refraction is 1.025) or solid, for example, silica (whose refractive index is 1.45 in the visible and near-infrared regions), or even a vacuum (in English, <i>vacuum</i>)<i>,</i> in which case the refractive index is 1 and is then worked in vacuo.

Documents cited in this specification are the following: <ul><li>[Brundrett 00] <nplcit id="ncit0001" npl-type="s"><text>Brundrett DL et al., "Effects of modulation strength in guided mode resonant gratings Subwavelength at normal incidence," J. Opt. Soc. Am. A 17, 1221 (2000</text></nplcit>)</li><li>[<patcit id="pcit0003" dnum="FR2812455"><text>FR 2812455</text></patcit>] <patcit id="pcit0004" dnum="FR2812455A"><text>FR 2812455 A</text></patcit>R. Sobolewski et al. (Schlumberger and Rochester, 2/2002)</li><li>[Gol'tsman 03] <nplcit id="ncit0002" npl-type="s"><text>Gol'tsman et al., Fabrication of Nanostructured Superconducting Single Photon Detectors, IEEE Transactions on Applied Superconductivity, 13 (2), 192, June 2003</text></nplcit></li><li>[Jackson 03] <nplcit id="ncit0003" npl-type="s"><text>D. Jackson, J. Stern, "High bandwidth, Improved quantum efficiency detector development for multi -GHz class OKD throughput," Jet Propulsion Laboratory, California Institute of Technology, Single Photon Detector Workshop, NIST Gaithersburg, 4/2003</text></nplcit></li><li>[Kerman 06] <nplcit id="ncit0004" npl-type="s"><text>Kerman AJ et al., "Kinetic inductance-limited reset time of superconducting nanowire photon counters," Appl. Phys. Letters 88, 111116 (2006</text></nplcit>) </li><li>[Korneev 05] <nplcit id="ncit0005" npl-type="s"><text>A. Korneev et al., "Quantum efficiency and noise equivalent power of nanostructured, NbN, single photon detectors in the wavelength from visible to infrared," IEEE Trans. Appl. Superconduct. 15, 571 (2005</text></nplcit>)</li><li>[LeCoupanec 03] <nplcit id="ncit0006" npl-type="s"><text>P. LeCoupanec, WK Lo, KR Wilsher, "An ultra-low dark count and jitter, superconducting, single-photon detector for issuance timing analysis of integrated circuits", Microelectronics Reliability, 43 (2003), 1621</text></nplcit></li><li>[Nevière 73] <nplcit id="ncit0007" npl-type="s"><text>Nevière P. et al., "Systematic study of the resonances of Thin Film holographic couplers", Opt. Common. 9, 48 (1973</text></nplcit>)</li><li>[Romestain 04] <nplcit id="ncit0008" npl-type="s"><text>Romestain R. et al., "Fabrication of superconducting niobium nitride hot electron bolometer for single photon counting," New Journal of Physics, 6, 129, (2004</text></nplcit>)</li><li>[Rosenblatt 97] <nplcit id="ncit0009" npl-type="s"><text>D. Rosenblatt et al., "Resonant waveguide grating structures", IEEE J. Quant. Elec. 33, 2038 (1997</text></nplcit>) </li><li>[Rosfjord 06] <nplcit id="ncit0010" npl-type="s"><text>KM Rosfjord, "Nanowire single photon detector with an integrated optical cavity and Antireflection coating", Opt. Expr. 14, 527 (2006</text></nplcit>)</li><li>[Villégier 04] <nplcit id="ncit0011" npl-type="s"><text>JC Villégier, "Fabrication of High-Speed ​​Single Photon Detectors and Analog-to-Digital Modulators in NbN Technology for Quantum Information Processing", Invited conference, workshop WEH 28 Nov.-3 Dec. 2004, Bad Honnef, D</text></nplcit></li><li>[Vincent 79] <nplcit id="ncit0012" npl-type="s"><text>Vincent P. et al. "Corrugated dielectric waveguides: a numerical study of the second-order stop-bands", Appl. Phys. 20, 345 (1979</text></nplcit>)</li><li>[Wang 04] <nplcit id="ncit0013" npl-type="s"><text>F. Wang, "superconductive Photo-detectors for photon quantum information", scientific option internship report, Ecole Polytechnique, July 2004</text></nplcit></li><li>[Yang 05] <nplcit id="ncit0014" npl-type="s"><text>JKW Yang et al., "Making development for nanowire-GHz counting-rate single photon detectors", IEEE Trans. Appl. Superconduct. 15, 626 (2005</text></nplcit>).</li></ul>