EP1377857A2 - Integrated high spectral resolution optical spectrometer and method for making same - Google Patents

Abstract

The invention concerns an integrated optical spectrometer with high spectral resolution, in particular for high-speed telecommunications and metrology, and a method for making same. Said spectrometer comprises at least an elementary spectrometer which includes an optical phase grating including a set of micro-guides (12) and formed on a cleaved planar optical guide (14), reflecting means (24, 30, 32, 34) adapted to successively reflect radiation derived from the set of micro-guides, so as to propagate said radiation in folded form and in free space, means (26) for photodetection of said radiation thus reflected and means (28) for focusing radiation on said photodetection means.

Description

 HIGH RESOLUTION INTEGRATED OPTICAL SPECTROMETER

SPECTRAL, ESPECIALLY FOR HIGH TELECOMMUNICATIONS

FLOW RATE AND METROLOGY, AND MANUFACTURING METHOD

DESCRIPTION

TECHNICAL AREA

The present invention relates to an optical spectrum analyzer also called a “spectral analysis device” or, more simply, an “optical spectrometer”.

This spectrometer is particularly applicable to infrared radiation, for example in the field of high speed optical telecommunications. Other applications of the invention, in particular optical metrology, will be mentioned below.

STATE OF THE PRIOR ART

Optical telecommunications have enabled a considerable increase in information rates thanks to spectral and time coding. Currently, speeds of the order of 40 Gigabits / s are obtained on a single optical fiber. Thanks to DWDM or dense wavelength multiplexing ("Dense Wavelength Division Mul tiplexing"), the information rate exceeds 1 terabit / s. The increase in speeds, which is necessary for establishing information transfer protocols (in particular on the Internet), requires simultaneously increasing the spectral width of the telecommunications band and reducing the spectral interval between canals. This approach is limited by the wavelength routing capacities, by the power available for the amplifiers and by the non-linear effects such as the stimulated Raman effect, the stimulated Brillouin effect and especially the four-wave mixing. which constitutes a limit for the wavelength separation.

It is customary to consider three spectral windows. The first is around 800 nm; it is used for local networks with multimode fibers.

The second spectral window is located around 1280 nm to 1350 nm (corresponding to a minimum of dispersion in silica); it is currently little used because the optical amplifiers with fibers doped with praseodymium (PDFA), which were developed for this window, have never reached the performance of fiber amplifiers doped with erbium (EDFA) for the band. at 1.55 μm.

The third window is located around

1550 nm (corresponding to a minimum attenuation for silica); it is currently broken down into several bands depending on the optical amplifiers used. Band C (“C-band”) is the spectral band amplified by traditional EDFA optical amplifiers; it extends from 1528 nm to 1565 nm and therefore over 37 nm. The L-band (“L-band”) extends from 1561 nm to 1620 nm, and therefore over 59 nm, and corresponds to EDFA optical amplifiers with “Raman” amplification. Currently, the transmission wavelengths in the C band are defined by the ITU (“ITU”), that is to say the International Telecommunication Union (“International Telecommunication Union”) according to an interval of 100 GHz. (0.8 nm). What is called the “ITU grid”, that is to say the set of wavelengths defined by the ITU, begins at 1528.77 nm (196.1 THz) to reach 1563.86 nm (191.7 THz). It has 45 wavelengths which extend over approximately 36 nm. The growing need for transmission capacity means that a channel interval of 0.4 nm (50 GHz) becomes likely, although the non-linear effects currently limit the transmission range.

New optical amplifiers with fibers doped with thulium (TDFA) make it possible to cover the spectral range which extends from 1470 nm to 1500 nm. This domain, which is currently called “S-band” (“S-band”) thus completes the third window (the spectral band located at 1510 nm + 10 nm comprising supervision channels).

In order to increase the spectral transmission capacity, optical amplifiers are studied using amplification mechanisms by stimulated Raman effect. In such amplifiers, the amplification is provided in a distributed and not a pointwise manner (as is the case in EDFA amplifiers). The noise factors obtained using a Raman amplification are better than those that one obtained with EDFA amplifiers, which makes it possible to reduce the transmitted optical powers and therefore to reduce the spectral interval between channels. In addition, unlike EDFA amplifiers which only allow amplification of the C band, stimulated Raman effect amplifiers make it possible to amplify a much wider spectral band by means of a suitable assembly of pumping lasers.

In theory, a Raman amplifier can thus cover all wavelengths between 1300 nm and 1660 nm, that is to say much more than the bands currently covered by doped fiber amplifiers.

Thus, the current DWDM transmission spectrum extends over a hundred nanometers (C and L bands) for a spectral separation between channels of 0.8 nm. However, it is believed that the TDFA amplifiers can be improved for the S band, which would make it possible to cover a total spectral band of approximately 150 nm (S, C and L bands).

The increase in DWDM capacity therefore makes the appearance of Raman amplifiers predictable, which are compatible with the existing optical network and for which the total spectral range reaches more than 350 nm, resulting in a total of nearly 900 channels per fiber. optical (for a separation between channels of 0.4 nm) instead of barely a hundred currently.

In order to effectively manage the distribution of telecommunications channels, it is therefore necessary to use optical spectrum analyzers to cover all spectral ranges mentioned previously and to identify precisely the channels to deduce their occupation.

In particular, it is important to have an optical spectrum analyzer in order to check the wavelength allocations and to maintain a low error rate on all the channels. It is also important to measure the intensity of optical signals with a good signal-to-noise ratio.

For telecommunications operators, it is desirable to design a compact, portable spectrometer, at an optimized cost, making it possible to analyze the spectral content of the ITU grid and to check the position in wavelength (to analyze a possible drift) as well. than the power of each channel.

In addition, it is desirable that the separating power of this spectrometer is approximately equal to or greater than 30000 and that the observed wavelength range makes it possible to satisfy the future bandwidth standard (between 120 nm and 180 nm). The total number of measurement points is therefore at least of the order of 2400 to 3600 and preferably close to 7200.

Such a spectrometer becomes vital for future DWDM networks and would be widely used for optical checks at each node of a network (“network”) comprising a wavelength add-drop device), a terminal or amplifier.

The spectral analysis devices marketed, making it possible to observe the spectrum defined according to the standards defined by the ITU, are diffraction gratings spectrometers (“diffraction gratings”) with single or double passage, Fabry-Perot interferometric cavities with free space scanning or Fourier Transform spectrometers based on Michelson interferometers.

All these known devices are expensive, bulky and fragile. In addition, their working frequencies are approximately equal to or less than 100 Hz and they do not guarantee a separating power of more than 30,000 over more than 40 nm.

In addition, certain known multiplexers / demultiplexers could be used as optical spectrum analyzers, but these devices do not make it possible to perform optical spectrum measurements and only perform the multiplexing functions compatible with the ITU 50 GHz standard.

The same is true for devices using diffraction gratings etched on planar substrates. In addition, these known devices do not allow direct determination of Bragg wavelengths with sufficiently high precision.

It would be desirable to have an optical spectrometer having the following qualities: "wavelength separation better than the interval between channels (0.4 nm), that is to say at least 0.05 nm ( spectrometer resolution power greater than 30,000),

»Large spectral analysis band for an optical spectrum (at least 120 nm or even 360 nm), the number of significant points being at least 2400 and extendable to 7200,

"possibility of integration (for example in a portable case)," low cost,

"great measurement dynamics,

"large frequency bandwidth (a few kilohertz or more),

"independence from _, temperature and atmospheric pressure (which are likely to induce variations in the refractive index), and

"independence from polarization of light.

There are already known optical components called "phasars" or optical phase arrays. About these components, also known as AWGs (for "Arrayed Waveguide

Gratings ”), reference may be made to the following document: [1] European patent application for

"Optical device with phase grating and method for manufacturing it", published under the number EP 0911660 A, invention of G. Grand et al. - see also the American patent application filed on October 26, 1998 under No. 09/179, 133.

Other phasars are also known by:

[2] M. Zirngibl, CH Joyner and JC Centanni, Size reduction of waveguide grating router through folding back the input-output fanouts, Electron. Lett. Vol. 33, No. 4, 1997, pages 295-297. STATEMENT OF THE INVENTION

The present invention aims to remedy the drawbacks mentioned above and to provide optical spectrometers having all or part of the qualities mentioned above and in particular an ability to separate wavelengths very close to each other, a wide spectral band of operation and the possibility of being compact and obtained in integrated form to be portable.

Specifically, the subject of the present invention is an optical spectrometer comprising at least one elementary optical spectrometer, this elementary optical spectrometer being characterized in that it comprises: an optical phase network comprising a set of micro-guides, this network of optical phase being formed on a planar optical guide which is cleaved, - reflection means capable of successively reflecting radiation from all of the microguides, with a view to propagation of this radiation in folded form and in free space,

- means for photodetecting the radiation thus reflected, and means for focusing the radiation on these photodetection means.

Preferably, ^'the reflecting means are adapted to enable propagation of radiation in folded form, first in the planar optical guide then in free space, above this planar optical guide, in a plane which is parallel to the latter.

According to a first particular embodiment of the optical spectrometer which is the subject of the invention, the optical phase grating is designed to operate by reflection and the planar optical guide has a plurality of cleaved sides, made reflective with respect to radiation from the set of microguides and vis-à-vis radiation intended to penetrate this set.

In this case, according to a preferred embodiment, the set of microguides ends at one of the cleaved sides and the optical phase grating comprises a focusing zone which leads to at least one of these cleaved sides.

According to a variant of the first particular embodiment, the optical phase network is designed to operate by reflection and the planar optical guide has a cleaved side, which is made to reflect vis-à-vis radiation from the set of microguides and vis-à-vis the radiation intended to penetrate this assembly and to which all of the microguides ends, as well as other cleaved sides capable of reflecting this radiation, this radiation being intended to arrive on these other cleaved sides with angles of 'incidence large enough to lead to a total reflection of these radiations.

One of the advantages of this variant lies in the fact that these other sides, devoid of reflective treatment, do not induce any polarization of the light they reflect.

In the optical phase network, the microguides form, for example, arcs of concentric circles.

According to a second particular embodiment, the optical phase network is designed to operate by transmission.

According to a preferred embodiment of the invention, the reflection means comprise:

- a prism, which is provided for reflecting the radiation from the set of microguides in a plane parallel to the planar optical guide on which the optical phase network is formed, and - at least one mirror provided for reflecting the radiation propagating in this plan towards the photodetection means.

The optical spectrometer which is the subject of the invention may further comprise a support on which the optical phase grating, the reflection means and the photodetection means are positioned in relation to each other.

Preferably, this support is obtained by molding or hot stamping of a plastic material, from a mold obtained by a molding technique by lithography and electro-forming.

Preferably, the optical spectrometer which is the subject of the invention further comprises means for compensating for modifications undergone by the optical phase network due to variations in temperature. In the case of the preferred embodiment using at least one mirror, these compensating means preferably comprise a bar having a coefficient of preferably high thermal expansion, this bar being made integral with the mirror to generate, by thermal expansion, changes in the orientation of the mirror, capable of compensating for the changes undergone by the optical phase grating. Preferably, the planar optical guide is obtained by an integrated optical technique on glass or on a semiconductor, in particular silicon or indium phosphide.

The present invention also relates to an optical spectrometer comprising a plurality of elementary optical spectrometers according to the invention, intended to cover in a modular manner a determined spectral range and optically coupled to an input optical fiber by means of wavelength separation.

According to a first particular embodiment, this spectrometer further comprises polarization separation means which connect these wavelength separation means to the elementary optical spectrometers.

According to a second particular embodiment, this spectrometer further comprises power separation means and polarization means which connect these wavelength separation means to the elementary optical spectrometers. The present invention also relates to an advantageous method of manufacturing the spectrometer which is the subject of the invention, in which the optical phase grating is of the folded type, to operate by reflection, and manufactured in several copies, by head-to-tail pairs, according to the techniques. integrated optics, from the same substrate which is then cleaved to obtain the various optical phase networks thus manufactured and to form an elementary optical spectrometer with each of these.

To manufacture the spectrometer which is the subject of the invention, it is possible to form each optical phase network from a substrate and to form cleavage marks at the same time as the optical phase network microguides on this substrate.

The present invention further relates to a spectral analysis device for high speed optical telecommunications using dense wavelength division multiplexing, this device comprising the spectrometer object of the invention, which comprises the plurality of elementary optical spectrometers, for provide a real-time indication of channel placement in the range from 1528.77 nm to 1563.86 nm, in a modular fashion and adaptable to the needs of users.

The present invention also relates to a Bragg grating optical metrology device, this device comprising the spectrometer which is the subject of the invention, which comprises the plurality of elementary optical spectrometers, for measuring Bragg wavelengths. This spectrometer is for example intended to detect optical signals coming from at least one Bragg grating sensor.

According to various aspects of the invention: - the dispersive device that the elementary optical spectrometer comprises is a component of integrated micro-optics called an optical phase network, or also phasar, which comprises a plurality of waveguides, each of which introduces a phase shift compared to the one preceding it; such a phasar generates a “mono-dimensional” set of electrical signals

(forming a "vector"), this set being representative of the optical spectrum of the light injected into the input fiber of the elementary spectrometer; - A spectrometer according to one invention may include one or more elementary spectrometers mounted in parallel, arranged in multiple stages; the optical spectrum sought is then obtained by concatenating all the elementary spectra coming from each of the phasars; this parallel assembly makes it possible to resolve the physical contradiction between obtaining a good spectral resolution (that is to say a high dispersion), a large spectral range and a minimum bulk for the spectrometer ; - in addition, to minimize congestion, each phasar can be a phasar operating in transmission or a half-phasar "folded" (to operate in reflection) thanks to an interface forming a mirror, located in the area of phase shifting microguides; moreover, the adaptation of the phasar dispersion properties to the dimensions of the photo-detector arrays leads to “folding” the focusing area a plurality of times so that the size of the optical circuit formed remains compatible with known manufacturing techniques. and optimized to reduce its cost in the context of ^' collective manufacturing;

- Self-compensation means for temperature variations, using the "folding" of the treated light beam, can be provided.

Note the originalities of various aspects of the invention compared to known optical spectrometers: • according to one of its designs, the optical spectrometer of the invention comprises half-phasars "folded" at the level of the micro-guides by an interface forming a mirror and the focusing area is therefore identical for entry and exit (unlike conventional phasars);

• the focusing zone can also be "folded" in part on the substrate where it is formed and also in free space, by optical means assembled on a support preferably obtained by pre-forming according to the technique called LIGA, this support being capable of including temperature self-compensation means;

• it is possible to use several phasars or i-phasars formed on planar substrates mounted in parallel, arranged in ulti-stages, allowing to increase the spectral range covered by concatenation of the vector-spectra obtained.

Such a multi-stage optical spectrometer with “folding” makes it possible to solve the problems posed in the field of DWDM telecommunications.

The following advantages are likely to be obtained thanks to the invention:

• resolving power greater than 30,000 (resolution of 50 p which can be improved by centroid processing),

• spectral range adjustable in multiples of 40 nm, from 40 nm to 360 nm,

• measurement dynamics (crosstalk) of the order of -30 dB (thanks to the phase network), • significant signal-to-noise ratio (thanks to parallel spectral analysis),

• significant bandwidth (greater than 1 kHz),

• independence from polarization, • independence from temperature

(thanks to mechanical self-compensation or thermal stabilization),

• miniaturization (made possible by planar substrates and also by arrangement in the form of multi-stages),

• spectrometer suitable for use in the field,

• optimized manufacturing cost (made possible by collective manufacturing in integrated optics) and mass production of the support box (using the LIGA technique).

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention will be better understood on reading the description of exemplary embodiments given below, by way of purely indicative and in no way limiting, with reference to the appended drawings in which:

FIG. 1 is a block diagram of a DWDM spectrometer according to the invention,

FIG. 2 is a block diagram of another DWDM spectrometer according to the invention, using polarization splitters,

FIG. 3 is a block diagram of another DWDM spectrometer according to the invention, using optical fiber polarizers,

FIG. 4 schematically illustrates an example of a masking diagram for a half-phasar in integrated optics on glass, usable in the invention,

- Figure 5 is an ^optical diagram corresponding to half phasar shown in Figure 4,

FIG. 6A is a schematic view of an assembly of an elementary micro-spectrometer according to the invention, using the technique called LIGA for "Lithography Galvanoformung Abformung",

FIG. 6B is the section AA in FIG. 6A,

- Figure 7A is a schematic view of an assembly of another elementary micro-spectrometer according to the invention, also using the LIGA technique,

FIG. 7B is the section AA in FIG. 7A,

FIG. 8 schematically illustrates a mechanism for passive self-compensation of the thermal dependence of a phase network usable in the invention,

FIG. 9 schematically illustrates another mechanism for passive self-compensation of the thermal dependence of a phase network usable in the invention,

FIG. 10 schematically illustrates an example of a masking diagram of two half-phasars in integrated optics on glass, usable in the invention,

FIG. 11A is a schematic view of an assembly of another elementary micro-spectrometer according to the invention, using the LIGA technique,

FIG. 11B is the section AA in FIG. 11A,

FIG. 12 schematically illustrates an application of a device according to the invention to the measurement of deformations, pressures and temperatures using Bragg gratings forming transducers, and

- Figure 13 schematically illustrates a wafer of 8 inches (about 20 cm) in integrated silica on silicon optics as well as an example of a masking scheme of phasar components according to the elementary mask described with reference to the figure 10. DETAILED DESCRIPTION OF PARTICULAR EMBODIMENTS

An objective of the present invention is the production of a compact spectrometer, at optimized cost, which makes it possible to cover the useful spectrum for the third telecommunications window (conventional band, or C band, and long band or L band). The spectral band to cover to satisfy the current telecommunications market is around 120 nm. In addition, the current ITU grid is 50 GHz, which corresponds to a separation of approximately 0.4 nm between channels (the number of channels being equal to 300).

An essential field of application of this high resolution and wide range spectrometer is that of telecommunications with very high spectral density (using the DWDM technique).

In the present invention, phasars are used for reasons of dispersion (phasars having a high diffraction order), compactness and crosstalk performance. However, sizing calculations show that it is not easy to manufacture a single phasar for the entire wavelength range because the number of micro-guides required to satisfy the spectral resolution criterion is very high: it is greater than 2500. In addition, the dimensioning of such a component leads to a value of focal focal length output much greater than the size of a semiconductor wafer. This is why, in the invention, it is preferable to "fold" the beam diffracted by the micro-guides one or more times in the focusing zone and mount a plurality of phasars in parallel using a “multi-stage” approach

(stacking) to cover the entire spectrum of the C and L bands by putting end to end, thanks to a computer, the information respectively from the phasars.

We will now describe a particular embodiment of a phasar which can be used in the invention and operates at order 39 for a spectral range of 40 nm. Other particular embodiments are possible, one of these consisting for example of working at an order 78 over a spectral range of 20 nm.

A phasar behaves like a concave diffraction grating in transmission, the equation of which is written: n sin θi + sin θo) ^J = n. -dg- - n _c . dg- = n. ^ ^λ d ^" g ^λo )

In this equation, d _g is the interval between two micro-guides at the entrance to the exit focusing zone, n _s the index of the planar guide, n _c the index of the guide (which is often the same) , the angles θi and θ ₀ are the diffracted angles respectively in the input focusing zone and in that of the output, ΔL is the length offset from one guide to the other, and n is the diffraction order. The wavelength λ ₀ is the central wavelength of the phasar which is written:

For a half-phasar (folded in reflection), the equation is modified by replacing ΔL by 2.ΔL: ns sin θi + sin θo) = n. The phasar demultiplexes the wavelengths from one of the input guides to the output guides. The phasar allows you to perform this function from any input guide; the spectrum received at the output is then shifted proportionally to the location of the input guide considered.

This property can be used in order to achieve temperature self-compensation by fixing the input optical fiber on a mechanical support whose displacement induced by thermal expansion compensates for the spectral shift due to the thermal dependence of the component.

In the invention, this property is also applicable in the case of the use of a half-phasar (folded) in order to offset an input fiber from a photodetector array located in the same plane, as is the case. will see further.

The dimensioning of a phase grating for spectrometry applications firstly requires adjusting the maximum possible diffraction order (and the length offset between each microguide) by considering the spectral range observed

(without overlapping from one order to a next or previous order), then the minimum number of micro-guides is chosen to guarantee the desired resolving power. Finally, the focal length of the output is calculated according to the characteristics of the photodetector array. To remove any ambiguity, the spectral overlaps from an i order to a following order i + 1 or previous i-1 are avoided. For this, the free spectral interval, denoted ISL, is chosen at best equal (or greater) to the observed spectral range, denoted ES. This defines a condition on the phasar diffraction order:

The resolving power of the phasar is written:

- k- = n.M where n is the diffraction order and M the Δλ number of micro-guides. Δλ corresponds to the spectral width of the "Airy spot"; we try to have 0.05 nm in order to be able to distinguish the channels within the ITU grid.

The length offset between each microguide is equal to ΔL =. λo for the half-phasar and at ΔL = - .λo for the entire phasar. n _c

As an example, the micro-guides are made up of arcs of circles.

In the case of the half-phasar, we choose, by way of example, an angle of 60 ° for these arcs of circles and the relationship between the offset in length and the

^• 3 offset in radius ΔR is written: ΔR = - .ΔL; for example π, for the entire phasar (angle of 120 °), the offset in radius ΔR is written: ΔR = —— .ΔL

The dispersion relation which links the displacement Δy of the focal spot, at the output of the micro-guides, at the wavelength λ is written:

where n _s is the effective index of the focusing area and f the length of this area which corresponds to the focal length of the output.

If we consider all the spectral width observed, the focused width L corresponding to the useful width of the bar, we obtain that this width L is little different from the quantity i ¹ .ISL

itself little different from f. -τ ~ —- and the distance dgXLs, focal length f is therefore little different from ^ns : ^g . L. The interval d _g cannot be reduced as much as desired; it is limited by the crosstalk between the guides. This value being fixed, the focal distance only depends on the length of the bar. On the other hand, the focal distance in the glass (index 1.46) is greater than the focal distance in the air (index 1) due to the self-focusing at the exit of the planar guide. .

Any reduction in the spectral width observed does not therefore have an effect on the focal distance because it is compensated for by an even lower order of diffraction (the ISL being adjusted accordingly). Obtaining better spectral resolution requires a new optical design, detailed in the present description, a design which allows a larger geometric path.

As a purely indicative and in no way limitative, the characteristics of each half-phasar are given below:

• free spectral interval: 40 nm, • diffraction order: 39,

• number of micro-guides: 800,

• interval between micro-guides: 19.8 μm,

• length difference between micro-guides: 20.7 μm • width of the diffraction spot: 0.05 nm,

• pixel interval: 13 μm,

• number of pixels per strip: 1024 pixels,

• length (useful) of the bar: 13.3 mm,

• spectral resolution (bar): 40 nm / 1024 pixels, that is to say approximately 0.04 nm / pixel,

• focal length of exit: 250 mm in the glass (index 1.46), from where a path of 50 mm in the glass added to 140 mm in the air (equivalent to 200 mm in the glass) [example which s 'applies to the assembly shown in Figures 6A and 6B] or 175 mm in the air [example which applies to the assembly shown in Figures 7A and 7B].

These characteristics take into account the necessary adaptation between the width of the diffraction spot and the width of the spectrum plotted on the photo-detector array (here, approximately 50 μm per pixel). The production of a mask by the beam propagation method (“Beam Propagation Method”) takes into account the dispersion of the materials used in the precise definition of the optical dimensioning of the component to be produced.

Let us now consider the operating principle of a micro-spectrometer according to the invention. Four design modes can be distinguished depending on whether the micro-spectrometer comprises a single phasar or a plurality of phasars

(respectively ^' assigned to distinct wavelength domains):

(1). single phasar spectrometer (independent of polarization or corrected for polarization dependence),

(2). phasar pair spectrometer (polarization separation),

(3). spectrometer with a plurality of elementary phasars (independent of polarization or corrected for polarization dependence), and

(4). spectrometer with a plurality of pairs of elementary phasars (polarization separation).

The design mode (1) (respectively (2)) is a special case of the design mode (3)

(respectively (4)). This is why only the design modes (3) and (4) will be detailed. The polarization dependence of a phasar results in the superposition of two spectra at the output of this phasar, corresponding to the two polarization states (electric transverse, denoted TE and magnetic transverse, denoted TM). If the polarization state is not maintained at the output of the fiber optic link that we want to analyze (which is the case with conventional optical telecommunications links), the observed spectrum is therefore blurred by the superposition of these two states. This problem can be solved in three different ways: • by designing a dispersive device independent of polarization (for example by means of an ion exchange technique on glass),

• by designing a dispersive device where one compensates for the effects of polarization, for example by inserting a half-wave plate in the middle of the zone of the phase shift guides and, on this subject, one will refer for example to the following document: [ 3] US 5937113A, • by separating the two polarization states before dispersion by two dispersive components, each being calibrated for a given polarization state (TE or TM).

The latter approach (design modes (2) and (4)) has the advantage of using phasars of simpler design (without polarization compensation), which are for example formed by means of an integrated optics technique on silicon, and allows to realize a polarization control spectrometer. However, it requires manufacturing twice as many substrates and controlling the state of polarization at the input of these substrates, which greatly complicates the implementation of this approach and increases the cost of the final device. The design modes (1) and (2) are respectively identical to the design modes (3) and (4) but without a wavelength separator.

The purpose of this wavelength separator is to perform a first separation of the bands observed by each of the elementary micro-spectrometers. It can be a network demultiplexer of the kind that is marketed under the STIMAX brand by the Jobin-Yvon Company or else an integrated optical device such as a commercial phasar.

Consider, by way of example, a micro-spectrometer in accordance with the invention, ensuring measurement over three spectral ranges.

FIG. 1 represents the optical block diagram of such a micro-spectrometer, applied to the DWDM technique and comprising a wavelength separator 2 as well as three blocks, namely three elementary micro-spectrometers whose references are Ml respectively , M2 and M3, depending on the design mode (3).

We see on the right side of FIG. 1 the reconstituted spectrum S with, on the abscissa, the wavelengths λ and, on the ordinate, the light intensities I.

FIGS. 2 and 3 represent optical block diagrams of a micro-spectrometer made up of six blocks, namely six elementary micro-spectrometers whose references are Mlx and Mly,

M2x and M2y and M3x and M3y, depending on the design mode (4).

This design mode may involve a wavelength separator 4 (for example, a phasar having a bandwidth of 40 nm) and three polarization separation couplers C1, C2, C3 (see Figure 2) which use preferably birefringent crystals to effect the separation of the polarizations because the crosstalk is then low. Another solution (see Figure 3) is to use the wavelength separator 2 of the FIG. 1, three couplers with power separation Pi, P2, P3 and six polarizers integrated on fibers, namely three pairs of polarizers Px and Py, Px allowing polarization in a direction x and Py in a direction y perpendicular to x.

In both cases (FIGS. 2 and 3), the optical fibers for connection between the polarization separation couplers or the polarizers and the elementary microspectrometers are fibers with polarization maintenance 6.

The first neutral axis of each polarization-maintaining fiber 6 is placed parallel to the surface of the substrate on which the associated microspectrometer is formed (this surface being for example parallel to the direction x).

The second neutral axis of this fiber 6 is placed perpendicular to this surface (and therefore parallel to the y axis in the example considered).

With reference to FIGS. 1 to 3, it is specified that the light signal to be analyzed by the microspectrometer reaches it by an input optical fiber FE; the wavelength separator 2

(respectively 4) is connected to the elementary microspectrometers Ml, M2, M3 (respectively to couplers C1, C2, C3 or Pi, P2, P3) by optical fibers fl, f2, f3; each of the couplers PI, P2, P3 is connected to the two associated polarizers Px and Py by two optical fibers fx and fy.

Three possible techniques for manufacturing a phasar which can be used in the invention are: the integrated optical technique on semiconductor or the integrated optical technique on silicon - or the ion exchange technique on glass.

We will only consider techniques using a silicon substrate or a glass substrate.

Let us first consider a phasar produced by an integrated optical technique on silicon.

Techniques of the “Si0 ₂ / Si” type (guiding layers in Si0 ₂ , SiON or Si ₃ N ₄ ) are perfectly suited to the manufacture of such a component. The processes used in this case are based on vapor deposition (essentially chemical vapor deposition) or hydrolysis with a flame and on reactive ion etching for the production of the patterns.

Take the example of the technique using guides on silica on silicon. On this subject, reference is made to the following documents:

[4] S. Valette et al., Si-based integrated optics technologies, Solid State Tech., 1989, pages 69-74 [5] S. Valette et al., Silicon-based integrated optics technology for optical sensor applications, Sensors and Act. A, 1990, pages 1087-1091.

In this case, the optical substrate is a layer of silica of sufficient thickness to isolate the light from the silicon (6 μm for a wavelength of 0.8 μm and

12 μm for a wavelength of the order of 1.3 μm at 1.55 μm), the guiding layer is a phosphorus doped silica layer (with a thickness of 2 μm to 5 μm depending on the wavelength) and the covering layer, or superstrate, is equivalent, from the point of view from the optical index, to the substrate, with a thickness of 6 μm to 10 μm.

Typically, for the component described, the dimensions of the channels are of the order of 4 μm x 4 μm with doping with 1 germanium oxide making it possible to achieve an index jump of the order of 2 × 10 ⁻² in order to ensure a strong confinement of the mode and to limit crosstalk between guides.

We know how to make a phasar of high density of integration with this technique. On this subject, we refer to the following document:

[6] Y. Hibino et al., Fabrication of silica-on-

Si waveguide with higher index difference and its application to 256 channel arrayed waveguide multi / demultiplexer, Optical Fiber Communications (OFC) 2000, Baltimore, WH-2-1, page 127.

A larger index jump, of the order of 3 × 10 ^-2 , can be achieved by depositing SiON.

An important advantage of integrated optics on silicon is to be able to simultaneously engrave V-grooves or V-grooves for positioning single-mode optical fibers.

Regarding the U-shaped grooves, we can refer to the following document:

[7] G. Grand et al., New method for low-cost and efficient optical connection between single-mode fibers and silica guides, Electron. Lett., Vol. 27, No. 1, 1991, pages 16-17.

Another advantage of this technique lies in controlling the slope of the etching flanks (to limit parasitic reflections at the ends of the guides, reflections which generate crosstalk).

Now consider a phasar made by an integrated optical technique on glass. This technique is well suited for producing the component shown in FIG. 4. The technique used is that of the thermal exchange of ions such as Na ⁺ , K ⁺ or Cs ⁺ , possibly assisted by an electric field. This well-known technique consists in exchanging alkaline ions (for example Na ⁺ ions), already present in the glass, with other ions such as Ag ⁺ or Tl ⁺ which have the effect of locally increasing the refractive index. glass.

The optical losses due to the fiber / guide connection and attenuation in the guide have been considerably reduced thanks to the technique of buried guides. The latter consists in diffusing the first doping into the substrate (under an electric field) or else in performing a second thermal diffusion of sodium ions. Guides are thus obtained characterized by quasi-circular doping sections, having a mode conforming to that of a single-mode optical fiber -on optimizes modal overlap- and having much lower linear attenuations due to the virtual disappearance of diffusion surface: it is typically less than 0.1 dB / cm.

Another advantage of this technique is to be able to produce guides having a very low dependence on polarization, and thus having a less costly design: there is no longer any need to compensate for the effects of birefringence by means of '' a half-wave blade inserted in the middle of the microguides area for example. Let us now consider phasar masking schemes which can be used in the invention.

We distinguish at least two masking schemes for phasars.

The first masking scheme corresponds to a folded half-phasar (therefore operating in reflection), with a focusing zone of the input beam in a guiding layer, this input beam coming from an optical fiber (see FIGS. 4 and 5).

The second masking scheme corresponds to a phasar operating in transmission (and therefore not folded), without focusing area in integrated optics: this focusing takes place in free space.

The first (respectively second) masking scheme corresponds to a suitable housing which is shown in FIGS. 6A and 6B (respectively 7A and 7B).

By way of example, FIG. 4 shows a mask of a half-phasar folded according to the first masking scheme. The substrate used is a glass disc 8 60 mm in diameter and 1.5 mm thick. The useful area, delimited by a dotted circle 10, is restricted to a disc 50 mm in diameter. The microguides 12 and the planar guide 14 are obtained by burying the guide layer.

Using “silica on silicon” technology with a 6 inch wafer

(that is to say about 15 cm) in diameter, at least four half-phasars of the kind which is shown in FIG. 4 can be produced by plate. For plates 8 inches (that is to say about 20 cm) in diameter, it becomes possible to form, on a single substrate, eight half-phasars of the kind shown in FIG. 4, and therefore to further reduce manufacturing costs.

Still by way of example, this half-phasar has five cleaved and polished sides cl, c2, c3, c4 and c5. The sides c2 and c3 make an angle of 90 ° relative to each other; the sides cl and c2 make an angle of 45 ° + 60 ° = 105 ° relative to each other; the sides c4 and c3 make an angle of 45 ° + 90 ° = 135 ° relative to each other; and side c5 is perpendicular to side c4.

By way of example, according to the specifications described above and relating to a .phasar working at a 39 order for a spectral range of 40 nm, 800 microguides 12 separated by 19.8 micrometers from each other form arcs of circles of 60 °. The minimum radius of curvature ^is' 4 mm and the maximum radius of curvature is 19.8 mm.

The focusing zone F, integrated on the planar guide 14 and delimited by the sides c, c ₃ and c ₂ (FIG. 5), makes it possible to extend all the light emitted of an optical fiber 16 (FIG. 5) on all the microguides without additional focusing optics.

The numerical aperture of the single-mode optical fibers traditionally used in optical telecommunications is of the order of 0.15 to 0.17 in air, hence a half-angle of divergence of approximately 9 ° to 10 ° in air and about 6 ° to 7 ° in the glass. The end of the fiber 16 is thus approximately 55 mm from the interface of the micro-guides. It is specified that this end of the fiber 16 is optically coupled to the planar guide by micrositioning and bonding (“pigtail” technique).

After cleavage and polishing, the sides cl, c2 and c3 must behave as reflectors. To do this, the side cl to which the respective ends of the respective microguides 12 orthogonally end forming concentric arcs of a circle, and possibly the sides c2 and c3 receive a reflective deposit R which may be a metallization or, preferably, a dielectric multilayer whose reflection spectrum is centered on the wavelength domain analyzed.

Even if the sides c2 and c3 do not receive a reflective deposit, the incident rays and their counterparts, which emerge from the phasar, arrive on these sides with a sufficient angle for there to be reflection on the dioptres which they constitute.

The side c5 can receive an anti-reflection multilayer deposit A (in the useful spectral band along the path from the glass to the air). The microguide outputs are distributed along a circle with a radius of 250 mm corresponding to the diameter of the Rowland circle on which the image points are dispersed. A small image distortion (carried on a circle with a radius of 125 mm) is therefore observed on the flat array of photodetectors, used with the half-phasar (see Figures 6A and 6B), and does not exceed half a pixel at the edge of the field, which therefore allows a flatness correction in the spectral measurement.

The sides c2, c3 and c4 allow the incident light beam 18 coming from the injection optical fiber 16 to be folded back (FIG. 5). This beam 18 is then naturally extended over all the inputs of the microguides. Fiber 16 is offset from the output field. By way of example, the fiber 16 is a single-mode optical fiber of 125 μm and digital aperture 0.16 and this fiber can be placed close to the axis of the wafer and oriented at about 13 ° relative to this axis. The incident wavefront is therefore offset from the front of the microguides and the wavelength λo defined above then corresponds to the lowest wavelength of the spectrum and no longer to the central wavelength (corresponding in case the fiber is placed in the center). In FIGS. 4 and 5, the reference 20 designates a separating zone which results, for example, from a diffusion of chromium or cobalt (absorbent at 1.55 μm).

According to a second masking scheme (not shown), the phasar is similar to that which has been described with reference to Figures 4 and 5, except that it is not folded and therefore operates in transmission. AT by way of example, it has four precisely cleaved sides which are parallel in pairs and an additional cleaved side. A first cleaved side corresponding to the entry of the micro-guides makes an angle of 120 ° with a second cleaved side corresponding to the exit of the micro-guides and the third and fourth cleaved sides are respectively parallel to the first and second cleaved sides.

The additional cleaved side connects the first cleaved side to the fourth and the cleavage of this additional side makes it possible to accommodate a cylindrical lens in the support of a microspectrometer using this phasar as seen in FIG. 7A.

As an exemplary embodiment, a spectrometer according to the invention is an assembly of several blocks of elementary micro-spectrometers as we saw above, each elementary micro-spectrometer making it possible to cover a spectral range of 40 nm. The superposition of several of these blocks makes it possible to cover a larger spectral range which can be adjusted according to the applications.

The focusing of the beam at the exit of the spectrometer can be carried out in different ways. It is possible to align several guiding layers by superimposing planar substrates connected together by a 45 ° polishing forming a reflective prism or by a ribbon of single-mode optical fibers.

Advantageously, the recommended assembly technique consists in placing the planar substrate (comprising the phasar) and the reflective prism as well as the lens and the array of photodetectors in a molded substrate or support, developed by a lithography and electro-forming molding technique, called "LIGA technique" (for Lithography Galvanoformung Abformung). On ^'this technique, one may refer to the following:

[8] J. Mohr, LIGA- A technology for fabricating microstructures and microsystems, Sensors and Materials, vol.10, n ° 6, 1998, pages 363-373 [9] J. Mohr et al., Micro-optical devices based on free space optics with LIGA micro-optical benches- examples and perspectives, SPIE 2783, 1996, pages 48-54

[10] H. Nakajima et al., Micro-optical sensors fabricated by the LIGA process, SPIE 3513, 1998 pages 106-112.

The LIGA process allows mass production, and therefore at an optimized cost, of elementary detection blocks, that is to say elementary micro-spectrometers. According to this process, a metal mold is formed by electro-forming (hence an electrolytic growth of the mold) after a high resolution lithography (of the order of lμm).

The LIGA X process, a special case of the LIGA process, consists in exposing a photosensitive resin ("photoresist"), for example PMMA, through a membrane mask carrying a layer of gold which absorbs X-rays. exposed parts, a layer of metal is deposited by electro-forming until it covers the PMMA pattern and forms a mold which is used to form the final parts by molding or hot stamping of a plastic, techniques suitable for mass production.

If, however, the plastic is molded on a metallic substrate, a second electrolytic growth can make it possible to obtain a finished metallic product. In addition, the hot stamping of the plastic can take place on a ceramic substrate in order to reduce the thermal expansion of the final component. The elementary spectrometer corresponding to a first assembly in accordance with the invention (FIGS. 6A and 6B) uses the first masking scheme (see FIGS. 4 and 5). The planar substrate 22, carrying the half-phasar, is placed guiding layer below and comprises the input focusing part (injection optical fiber 16 - planar guide 14) integrated on the planar substrate 22.

After reflection at the mirror interface (side cl provided with a reflective coating - see FIG. 4) at the exit of the micro-guides 12, the light beam diffracted by these micro-guides is recovered at the exit of the planar guide 14 by a reflective prism 24 then focused on a photodetector array 26 by a cylindrical lens 28 whose focal distance is approximately 6 mm. Three mirrors 30, 32 and 34 allow the beam to be reflected towards the bar 26, over a distance of 140 mm in free space.

More precisely, the diffracted beam coming from the prism is reflected on the mirror 30 then on the mirror 32 which is perpendicular to this mirror 30 then on the mirror 34 which is perpendicular to this mirror 32 and the mirror 34 reflects the beam towards the bar 26.

Taking into account the conjugation relation between the mode at the exit of the planar guide and the image on the bar, the edge of the substrate is at a distance of 6.27 mm from the center of the lens for a focal distance of 6 mm. The height of the image beam on the photo-detectors is approximately 120 μm.

In the example of FIGS. 6A and 6B, calibrated spacers 36, 38 and 40, which are respectively arranged on the cleaved sides cl, c4 and c5 of the planar substrate 22, between these sides and the support 42 molded by the LIGA technique, make it possible to position this planar substrate in height and to guarantee that it is parallel to the detection strip 26. the three mirrors 30, 32 and 34 as well as the strip 26 rest in notches, such as the notches 44 provided in the support 42. In the case of the ion exchange technique on glass, the planar guide 14 is buried at a known distance from the upper surface of the planar substrate 22, this surface serving as a reference.

The elementary spectrometer corresponding to a second assembly in accordance with the invention (FIGS. 7A and 7B) uses the second masking scheme mentioned above. The planar substrate of this second assembly is also placed as a guide layer below.

Unlike the first assembly, the focusing of the light coming from the single mode optical fiber 16 is not done in integrated optics but in free space. To do this, the fiber is aligned with passively in a groove -46. The support 48, intended to receive the planar substrate 50 is formed with the LIGA technique, provided with this groove 46 as well as all the notches which are needed. For example, at the output of the single-mode fiber 16, the divergent light beam 52 (whose half-angle of divergence is 8 °) is filtered by a circular opening 54 and then focused by a cylindrical lens 56 of equal focal length at 22 mm placed at equal distance (22 mm) from the planar guide and the fiber. The light beam is reflected towards the microguides 60 by a mirror 58 placed at 35 ° and the fiber is inclined by 10 °. The focusing parameters are defined by the Gaussian optical conjugation relationships for a guided mode having a half-width (“waist”) of 2.2 μm.

The transmitted beam, diffracted by the microguides 60, is recovered at the exit of the planar guide by a reflective prism 62 and then focused on the photodetector array 64 by a cylindrical lens 66 with a focal distance approximately equal to 8 mm. This lens 66 can be a piano-convex lens or a Fresnel lens.

Taking into account the conjugation relation between the mode at the exit of the planar guide and the image on the bar, the edge of the planar substrate 50 is at a distance of 8.5 mm from the center of the lens 66 for a focal distance 8 mm. The height of the image beam is approximately 100 μm on the photo-detectors. Three mirrors 68, 70 and 72 make it possible to reflect the beam towards the bar over a distance of 175 mm in free space. More precisely, the diffracted beam coming from the prism 62 is reflected on the mirror 68 then on the mirror 70 then on the mirror 72 and the latter reflects the beam towards the bar 64.

By way of example, taking the vertical as a reference (interface of the planar substrate), the first mirror 68 makes an angle of 15 °, the second mirror 70 makes an angle of 75 ° and the third mirror 72 makes an angle of 48 , 75 ° (its normal being oriented by 41.25 ° along the vertical).

In the case of FIGS. 7A and 7B, in the same way as for the first assembly, calibrated spacers 74, 76 and 78 which are arranged on cleaved sides of the planar substrate 50, between these sides and the molded support 48, as is seen in FIG. 7A, make it possible to position this planar substrate 50 in height and to guarantee that it is parallel not only to the detection strip 64 but also to the part of the fiber which is in the groove 46 and to the axis of the cylindrical lens 56.

An elementary micro-spectrometer according to the first assembly (FIGS. 6A and 6B) is capable of occupying a volume of 60 × 60 × 9 mm ³ .

According to the second assembly (Figures 7A and 7B), it is likely to occupy a volume of 60x60x12mm ³ . A micro-spectrometer comprising a stack of four elementary blocks makes it possible to cover a spectral range of more than 150 nm (at a wavelength of 1.55 μm) by superposition according to the following distribution:

Band Domain Spectral number (nm) of blocks

S from 1470 to 1 1500

C from 1528 to 1 1565

L from 1566 to 2 1645

Let us now consider the temperature stability of a spectrometer according to the invention.

A phasar is sensitive to temperature regardless of the technology with which it is made. This results in variations in the central wavelength λo of the phasar as a function of temperature, these variations resulting from the thermal expansion of the phasar material and ^' variations in the refractive index of this material as a function of the temperature (thermo-optical effect). The central wavelength increases with temperature.

For a half-phasar mentioned above, working at order 39 for a spectral range of 40 nm, produced in integrated optics on silicon or in integrated optics on glass, the thermal dependence of the central wavelength is order of 10 pm / ° C. We therefore end up with a spectrum shift on the array of photodetectors of the order of 3 μm / ° C, i.e. a quarter of a pixel / ° C.

Each of the two proposed assemblies (FIGS. 6A-6B and 7A-7B) can be temperature controlled by a heating resistor, in order to guarantee the stability of the spectral measurements over time, and the sealing of the assembly can be provided to guarantee the resistance of this assembly to humidity and also guarantee a constant air pressure. Another solution consists in providing a mechanical element for compensating for this angular offset.

In the case of the spectrometers in FIGS. 6A-6B and 7A-7B, a compensation solution consists in actuating one of the mirrors for "folding" the output beam, advantageously the first mirror 30 (FIG. 6A) or 68 (FIG. 7A ) although this principle is adaptable to each of the mirrors. This mirror 30 or 68 is engaged, by one of its ends, in a positioning groove 80 (Figure 8) or 82 (Figure 9) which serves as a pivot and this mirror 30 or 68 is actuated by a lever arm 84 (figure 8). or 86 (Figure 9) attached to the other end of the mirror.

By way of example, this lever arm is an aluminum bar, engaged in a groove 88 or 90, the coefficient of thermal expansion of which is approximately 23 × ¹⁰ -5 ^" / ° C. The extension by thermal expansion of this bar tilt the mirror and compensate for the angular offset induced by the temperature variation in the planar substrate. An inclination of approximately 2.5 × ¹⁰ -5 ^" rad / ° C. is necessary and performed, for example, by a 20 mm long bar placed 20 mm from the pivot according to FIGS. 8 and 9 which correspond respectively to the first and second assemblies). Regarding the sensitivity of the measurements, the total distance traveled by light within the spectrometer is around 10 cm, resulting in a loss of propagation of 1 dB (considering an attenuation of 0.1 dB / cm). Connectivity losses and losses at interfaces must be added. Let us therefore consider a total optical loss of the spectrometer of 6 dB.

In the case of the use of this spectrometer for the measurement of wavelengths of Bragg gratings in optical metrology, most of the continuous superluminescent sources used typically emit a few hundred microwatts over a few tens of nm of spectral width. This corresponds to an excitation spectral density of the order of 100 μW / 10 nm, that is to say 1 μW / λ. Knowing that the typical spectral width of a transducer Bragg grating is of the order of 1 A (100 μm), the power returned by the external optical fiber is therefore approximately equal to -30 dBm (1 μW). Thus, the power analyzed at the level of the photodetector array is estimated at -42 dBm (taking into account the optical losses and the spot / pixel recovery).

An external optical fiber is connected to the planar guide 14 (FIG. 5) so as to be subsequently welded or connected to the optical circuit incorporating one or more sensitive optical fibers. This fiber external thus constitutes the optical interface with the external environment (accessible by the end user).

Advantageously, this external optical fiber is single mode at the wavelength of use (typically 1300 nm, 1550 nm or even 820 nm).

The fiber-guide connection can be ensured by the V-groove technique.

The fibers can be connected to the guides by gluing (for example with an adhesive that can be polymerized by ultraviolet radiation) or by laser welding.

Advantageously, the detection unit is a set of photodiodes or an array of InGaAs photodiodes produced by epitaxy. Bars such as those sold by Thomson can be used. The useful detection area is approximately 5 μm; it is separated by two passivated zones of 8 μm, hence a period of 13 μm.

A high frequency bandwidth (100 kHz) can be achieved by operating the photodiodes in photoconductive mode and by inserting these photodiodes into an electronic assembly of the transimpedance type for example.

As a variant, in the case of an integrated optical technology on silicon, the photodetectors can be directly incorporated into the circuit.

A matrix of photo-detectors making it possible to form images in two dimensions can also be used instead of several linear bars. With regard to calibration, a polynomial correction between pixel and wavelength is generally implemented due to the distortion of the field observed on the bar.

We also propose a third assembly in accordance with the invention, incorporating an integrated optical component, schematically represented in FIG. 10, with a smaller surface area in order to optimize its manufacturing cost. It is a half-phasar operating in reflection.

Two identical half-phasars 91, operating in reflection, can be produced on the same disc

92 of 60 mm in diameter (the dotted circle 94 delimiting the useful area of the disc 92).

Similar to Figure 4, a set

96 (respectively 98) of 800 microguides forming arcs of circles of 60 ° is delimited, on one side, by a cleavage 100 (respectively 102), the side resulting from this cleavage being provided with a reflective deposit 104

(respectively 106), and on the other side, by an interface 108 (respectively 110) to a planar guide layer 112 (respectively 114) allowing a reflection of light beam thanks to another cleavage 116 (respectively 118).

Unlike the first assembly, a reflective deposit is not necessary on the cleavage 116 (respectively 118) because, in the example of the figure

10, the angle of incidence is sufficient for there to be total reflection for the incident and refracted beams.

The two half-phasars are first separated by a cleavage 124 before any other operation. In order to facilitate this cleavage operation, marks marking the axis of this cleavage, for example marks m at the two ends of this axis, can advantageously be photoetched at the same time as the masking for the diffusion of the guide.

An upper cleavage 126 and a lower cleavage 128 are also formed for mechanical reasons, as seen in FIG. 10.

In the same way as for the first assembly (FIGS. 6A and 6B), the third assembly schematically represented in FIGS. 11A and 11B incorporates the optical component 91 described with reference to FIG. 10, an optical fiber 130 connected to this component by traditional techniques and a mechanical support 132, molded by the LIGA technique and supporting all the optical elements (reflective prism 134, cylindrical lens 136, folding mirrors 138, 140 and 142 and array of photo-detectors 144). It is specified that, in this assembly, the light coming from the optical component 91 is reflected by the reflective prism 134 then focused by the cylindrical lens 136 on the array of photo-detectors 144 after reflection on the folding mirrors 138, 140 and 142 .

The optical principle of this assembly is identical to that of the first assembly (FIGS. 6A and 6B). The cylindrical lens 136 is however placed between the reflective prism 134 and the planar optical component 91 but the distance between the lens and this component is the same as for the first assembly. This component also rests on three spacers 146, 148 and 150.

It should be noted that, in each of the spectrometers of FIGS. 6A-6B, 7A-7B and 11A-11B, the light which one wants to analyze is guided, over part of its path, in a plane guide, comprising the microguides 12 or 60 or 96 then another part of this path takes place in free space, after folding this path thanks to the prism 24 or 62 or 134, this folding allowing the light to be substantially in a plane parallel to the plane guide, then this free space path is again folded several times in this plane thanks to the mirrors 30-32-34 or 68-70-72 or 138-140-142 before reaching the photodetectors, which makes it possible to confine the total path in a very small volume and to obtain very space-saving spectrometers.

However, it would not go beyond the scope of the invention to reduce the number of folds and to reduce, for this purpose, the number of mirrors and / or by removing the prism; we would still obtain less bulky spectrometers than in the prior art; in the case of FIGS. 6A-6B and 7A-7B, the removal of the prism would of course require placing the lens opposite the planar guide so that it can focus the light which leaves it.

Returning to FIGS. 11A and 11B, the optical fiber 130 is advantageously made of germanosilicate, its core ("core") has a very small diameter (approximately 2 micrometers) and it has a very high index jump (greater than or approximately equal to 0.05). This fiber is assembled to the planar substrate 152 of the component 191 by bonding. Fibers of this kind are used for non-linear optics or optical amplification, applications for which a very high light intensity is sought.

This optical fiber 130 is welded to another optical fiber 154, of the kind used in telecommunications in the wavelength band of 1.55 micrometers. These fibers have a heart of the order of 9 micrometers in diameter and a jump in index of approximately 5 × 10 ^-3 .

An adaptation of the mode of the fiber 130 to the mode of the fiber 154 is carried out in order to weld these fibers 130 and 154 with a minimum of losses. To do this, the fiber 130 is locally heated to diffuse the dopant (germanium in the example considered) outwards and thus create a progressive variation in the diameter of the core until it adapts to that of fiber 154. This technique is known as TEC for Thermally-diffused Expanded Core.

The optical fiber 130 therefore comprises, at one end, a portion of diffused core fiber 156 onto which the fiber 154 is traditionally welded. We now consider four examples of application of the invention.

1) Analysis of optical spectra, compatible with the ITU grid, with high capacity in DWDM.

In this field, a spectrometer according to the invention makes it possible to provide a user with a visual indication of the optical placement of the channels. DWDM. It is possible to provide feedback signals to laser diode control modules in order to correct a possible drift in the wavelengths of the transmitters. This control mode corresponds to pixel-by-pixel addressing. It allows for a dynamically reconfigurable multi-channel receiver.

Such a spectrometer is compatible with future DWDM specifications. It is characterized by a small footprint, an optimized cost (resulting from collective manufacture) and a large bandwidth, which typically ranges from a few kilohertz to a few hundred kilohertz and depends only on the photodetectors, with or without analysis of the state of polarization.

Its modular aspect (resulting from the concept of multiple stages) makes it possible to address all or part of the optical spectrum used in long-distance telecommunications, this spectrum extending from 1300 nm to 1700 nm. 2) Measurement of the signal-to-noise ratio of optical channels in DWDM telecommunications networks.

This involves measuring the power of each channel as well as the ambient noise, induced in particular by the amplified spontaneous emission (ASE) of the amplifiers. Thanks to the separating power of a micro-spectrometer according to the invention, it is possible to determine the envelope curve of the optical background noise of the channels, to interpolate the noise values for each wavelength of the channel and d '' deduce the signal-to-noise ratio for each of the channels as described in the following document: [11] WO 98/54862 (CIENA Corp.).

3) Spectral measurement of the dispersion of polarization modes.

The monomode fibers used support two polarization modes due to the birefringence of the silica. These two modes are characterized by two slightly different effective indices. The light pulse received is therefore formed of two pulses according to two polarization states, the delay of which changes over time, in particular because of the constraints in the optical cables, these constraints resulting for example from temperature variations. Chromatic dispersion has long limited time multiplexing. Currently, polarization mode dispersion (PMD) constitutes a new limitation in terms of modulation capacity.

There are two regimes of statistical behavior for PMD. In one of these regimes, the PMD is proportional to the length L of fiber while in the other regime, it is proportional to L ¹ . In practice, a 100 km line, with a transmission capacity greater than 10 Gbits, requires a PMD coefficient of less than 1 ps.km ^{1 2} and it becomes important to characterize this parameter in the field. The usual PMD values are therefore a few tens of ps over several hundred kilometers of fiber. Among all known methods for measuring PMD, there is one that uses a large source spectral band, alternately polarized in two orthogonal directions, and in which the corresponding spectra of the light beam transmitted through an analyzer are collected and the number of passages by OdB of the curve-ratio of the two spectra is counted. On this subject we will refer to the following document:

[12] CD. Poole et al., Polarization-mode dispersion measurement based on transmission spectra though a polarizer, J. of Lightwave technol. Vol. 12, No. 6, 1994, pages 917-929.

With this method, the large PMD values (corresponding to the measurement range) are defined by the spectral resolution of the spectrometer that is used, while the low PMD values are defined by the spectral range of the observed spectrum.

4) Analysis of Bragg grating spectra for Bragg grating metrology.

For such an analysis, a broadband optical source is used and the wavelengths reflected by the various Bragg gratings multiplexed in wavelengths along the measurement line are analyzed. Measurement and demultiplexing are carried out simultaneously by addressing to a photodetector array at an optimized cost and with a high frequency bandwidth, all these parameters being important in order to be able to use this type of metrology in the industrial environment.

The spectral behavior equation of a Bragg grating written in a standard germanosilicate fiber is written: ^ = 0.78 ε + 7.4 x 10 ^"6 ΔT (°) - 5.2 x 10 ^" λ

⁶ ΔP (MPa).

In this equation ε, ΔT and ΔP correspond respectively to the deformation, the temperature difference and the pressure difference.

For an excitation source whose width at half height is approximately 48 nm, this corresponds to a multiplexing of 8 Bragg transducers on the measurement line. It is then possible to design an optical micro-system for measuring deformations or temperatures, comprising photoaggregated Bragg grating transducers as shown diagrammatically in FIG. 12. Such a micro-system can also comprise a balanced four-way coupler (with 50% transmission on the two output channels) which has the reference 158 in Figure 12.

An optical source 160 with a broad spectral band, which can be a superfluorescent source with erbium-doped fiber or else a superluminescent diode, is then connected to an input arm of the coupler 158 while the micro-spectrometer 162 conforms to the The invention is connected to the other input arm of the coupler. One of the two output arms of the coupler is, for its part, connected to the end of a sensitive optical fiber 164 on which have been photo-registered several Bragg networks transducers 166 and the other end of which has a cleavage at an angle 168.

It is recalled that the optical spectrum of superluminescent diodes has a Gaussian appearance and a typical spectral width of the order of 30 nm to 50 nm. FIG. 12 also shows the means 170 for supplying the source 160, the photodiodes array 172 which is associated with the micro-spectrometer and the electronic means 174 for detecting the signals supplied by the photodiodes array.

A device according to the invention therefore applies to the monitoring in real time, at high bandwidth (1 kHz), of several constraints or pressures applied to a sensitive optical fiber incorporated in a structure, for example made of composite material.

It can also be used to make real-time measurements of distributed temperatures.

In addition, it can be used in the field of telecommunications to demultiplex several channels and to measure information coded in wavelength.

This device also applies to demultiplexing and to the measurement of several wavelengths, for example in the field of telecommunications multiplexed in wavelength.

Finally, according to the first demultiplexing design mode, the great manufacturing flexibility of this device makes it particularly attractive for ultisectoral instrumentation, by easy adjustment of the Bragg wavelengths of the photoinscribed networks, an addressing being then carried out on the array of photo-detectors.

Let us now consider a collective manufacturing method of phasars usable in the invention, with reference to FIG. 13. Such phasars can be manufactured using integrated silica-on-silicon optics technology and a wafer 4 inches in diameter. With this technology we can also process slices of 8 inches (about 20 cm) in diameter. On such a support 175 (FIG. 13) it is possible to form 16 double-phasars 176 of the kind shown in FIG. 10 and therefore 32 phasar components simultaneously (collective manufacturing approach).

After the patterns have been produced (according to the methods described in documents [4] and [5], the assemblies 176 are separated by sawing (using a metal blade or a diamond blade) according to an automated digital control procedure The saw cut is approximately 200 μm to 300 μm wide; it is not shown in FIG. 13.

The sawing operation can for example start with a separation of bands 178 of double-phasars 176 according to the dotted lines 180 and then continue with an assembly of the bands thus separated and a recutting of these to isolate the patterns of double phasars 176 The phasars can then be reassembled and sawn along their midline (not shown).

The economic advantage of this folded phasar scheme is that the component surface is halved compared to a traditional transmission scheme. The number of phasar components manufactured per slice is therefore multiplied at least by two and the cost of an individual phasar is halved at least.

In addition, the polishing operations which follow the sawing operation can be carried out simultaneously on a very large number of substrates, which further reduces the cost. It is the same for the operation of forming a reflective deposit mentioned above in the description of Figures 4 and 5. It is also possible to form cleavage marks m as we have seen above, to define the various cleavage lines that are needed, in particular cleavage lines 180.

Claims

1. Optical spectrometer comprising at least one elementary optical spectrometer, this elementary optical spectrometer being characterized in that it comprises:

- an optical phase network comprising a set of micro-guides (12, 60, 96, 98), this optical phase network being formed on a planar optical guide which is cleaved, - reflection means (24, 30, 32 , 34; 62,

68, 70, 72; 134, 138, 140, 142) capable of successively reflecting radiation from all of the microguides, with a view to propagation of this radiation in folded form and in free space, - means (26, 64, 144) for photodetection of the radiation thus reflected, and means (28, 66, 136) for focusing the radiation on these photodetection means.

2. Spectrometer according to claim 1, in which the reflection means are capable of permitting the propagation of the radiation in folded form, first in the planar optical guide then in free space, above this planar optical guide, in a plane which is parallel to the latter.

3. Spectrometer according to any one of claims 1 and 2, in which the optical phase grating is designed to operate by reflection and the planar optical guide comprises a plurality of cleaved sides (cl, c2, c3), made reflective vis- vis-à-vis radiation from the set of microguides and vis-à-vis radiation intended to penetrate this assembly.

4. Spectrometer according to claim 3, in which the set of microguides ends at one (cl) of the cleaved sides and the optical phase grating comprises a focusing zone (F) which leads to at least one of these cleaved sides.

5. Spectrometer according to any one of claims 1 and 2, in which the optical phase grating is designed to operate by reflection and the planar optical guide comprises a cleaved side (cl), which is made reflective with respect to radiation from the set of microguides and vis-à-vis radiation intended to penetrate into this set and to which the set of microguides ends, as well as other cleaved sides (c2, c3) capable of reflecting these radiations, these radiations being planned to arrive on these other cleaved sides with angles of incidence large enough to lead to a total reflection of these radiations.

6. Spectrometer according to any one of claims 1 to 5, in which the microguides form arcs of concentric circles (12, 96, 98).

7. Spectrometer according to any one of claims 1 and 2, in which the phase network is designed to operate by transmission.

8. Spectrometer according to any one of claims 1 to 7, in which the reflection means comprise: - a prism (24, 62, 134), which is designed to reflect the radiations coming from the set of microguides in a plane parallel to the planar optical guide on which the optical phase network is formed, and

- At least one mirror (30, 32, 34; 68, 70, 72; 138, 140, 142) provided for reflecting the radiation propagating in this plane towards the photodetection means.

9. Spectrometer according to any one of claims 1 to 8, further comprising a support (42, 48, 132) on which the optical phase grating, the reflection means and the photodetection means are positioned relative to each other.

10. Spectrometer according to claim 9, in which the support (42, 48, 132) is obtained by molding or hot stamping of a plastic material, from a mold obtained by a molding technique by lithography and electroforming.

11. Spectrometer according to any one of claims 1 to 10, further comprising means (84, 86) for compensating for modifications undergone by the optical phase network due to temperature variations.

12. Spectrometer according to claim 8, further comprising means for compensating for modifications undergone by the optical phase network due to temperature variations, these compensation means comprising a bar (84, 86) having a coefficient of thermal expansion preferably high. , this bar being made integral with the mirror (30, 68) to generate, by thermal expansion, modifications of the orientation of the mirror, capable of compensate for the modifications undergone by the optical phase network.

13. Spectrometer according to any one of Claims 1 to 12, in which the planar optical guide is obtained by an integrated optical technique on glass or on a semiconductor, in particular silicon or indium phosphide.

14. Spectrometer according to any one of Claims 1 to 13, comprising a plurality of elementary optical spectrometers (Ml, M2, M3; Mlx, Mly, M2x, M2y, M3x, M3y) designed to cover in a modular fashion a determined spectral range and optically coupled to an input optical fiber (FE) via wavelength separation means (2, 4).

15. Spectrometer according to claim 14, further comprising polarization separation means (C1, C2, C3) which connect these wavelength separation means to the elementary optical spectrometers.

16. Spectrometer according to claim 14, further comprising power separation means (Pi, P2, P3) and polarization means

(Px, Py) which connect these wavelength separation means to the elementary optical spectrometers.

17. Method of manufacturing the spectrometer according to claim 1, in which the optical phase grating is of the folded type, to operate by reflection, and produced in several copies, in head-to-tail pairs, according to optical techniques integrated, from the same substrate which is then cleaved to obtain the various optical phase networks thus manufactured and to form an elementary optical spectrometer with each of these.

18. Method of manufacturing the spectrometer according to any one of Claims 1 to 16, in which each optical phase network is formed from a substrate and cleavage marks (m) are formed at the same time as the microguides of this optical phase grating on this substrate.

19. Spectral analysis device for high speed optical telecommunications using dense wavelength division multiplexing, this device comprising the spectrometer according to any one of claims 14 to 16 to provide a real time indication of the placement of channels in the '' range from 1528.77 nm to 1563.86 nm, modular and adaptable according to user needs.

20. Bragg grating optical metrology device, this device comprising the spectrometer according to any one of claims 14 to 16 for measuring Bragg wavelengths.

21. Device according to claim 20, in which the spectrometer is intended to detect optical signals coming from at least one Bragg grating sensor (166).