WO2001050165A1 - Optical-fibre device comprising a diffraction grating - Google Patents

Abstract

Optical-fibre device comprising an optical fibre (1) and a diffraction grating (4) made into said optical fibre, said optical fibre (1) having a core (2), a first radius (R1) and a first refractive index (n1); a first cladding area (3a) made around said core (2) and having a second radius (R2) and a second refractive index (n2) lower than said first refractive index (n1); and a second cladding area (3b) made around said first cladding area (3a) and having a third radius (R3) and a third refractive index (n3) greater than said second refractive index (n2) and lower than said first refractive index (n1); the grating (4) being made into the core (2) along a longitudinal axis of the fibre (1) and being of the variable pitch type; the first and the second radiuses (R1R2) and the first, second and third refractive indexes (n1, n2, n3) are selected so that the effective refractive index (neff) of at least one higher order mode is greater than the third refractive index (n3).

Description

Optical-fibre device comprising a diffraction grating

* * *** ** * *

DESCRIPTION

The present invention relates to an optical-fibre device comprising a diffraction grating.

As known, an optical fibre is a thread-like element suitable to convey light inside it. Generally, an optical fibre comprises a central core, typically of doped glass, which confines inside it the light to transmit; a cladding, external to the core, typically of glass as well, having a lower refractive index than that of the core so as to allow confining the light inside the core itself; and one or more external protective and strengthening coatings.

Besides conveying the light in a passive way, optical fibres are also used for producing devices which allow changing the properties of the transmitted light. A wide range of recently developed optical -fibre devices are based on the possibility of making diffraction gratings, in particular Bragg gratings, directly into the core of an optical fibre.

A Bragg grating is a typically periodical structure which extends along a predetermined direction, and which comprises longitudinal areas with a high refractive index interposed between longitudinal areas with low refractive index; said structure has the property of back-reflecting the light passing through it at a wavelength (so called Bragg wavelength which is referred to as λ_B) which depends on the reciprocal distance of the above areas and on the optical properties of the means on which the grating itself is made. If the grating is made in the core of a single- mode optical fibre, the Bragg wavelength is given by the following relationship:

λ_B = 2 -n_{eff LP01}-Λ ( 1 ) where n_effjLP01 is the effective refractive index in the fundamental mode LP₀₁ propagating into the core, and Λ is the grating periodicity.

Techniques based on the use of ultraviolet light are known for writing a Bragg grating along the axis of an optical fibre. In fact, in the realisation of these gratings, the property of silica fibres doped with germanium is used, of being capable to undergo a variation of the refractive index of the doped material when the latter is exposed to an ultraviolet radiation at predetermined wavelength.

Figure 1 shows a typical transmissivity curve of a periodical Bragg grating, in particular a constant- periodicity grating. In said figure, it can be noted that a main negative peak is present in correspondence with the Bragg wavelength (in the particular case, at about 1551 nm) , because this wavelength is almost completely reflected, and a series of secondary peaks, defining undesired loss peaks, placed at lower wavelengths than the Bragg wavelength. Said loss peaks originated from the fact that in a single-mode optical fibre, after it has been stripped of its protective coating for the purpose of writing the grating, besides the fundamental mode, also a series of cladding modes can propagate. Said cladding modes extend in the entire area taken by the cladding and, for this reason, they have a value of effective refractive index which is lower than that of the fundamental mode (which is prevalently confined into the core) . A power coupling induced by the Bragg grating may occur between the fundamental mode and the cladding modes. In the case of power coupling between the fundamental mode LP₀₁ and the generic cladding mode LP_nm, the previous relationship (2) can be rewritten as follows:

^•^Bnm ^{= n}ef f , LP01 ^{+ n}ef f , LPTim *^{1 '} ^ ( 2 ) where n_{eff LPnm} is the effective refractive index of the generic cladding mode determined by the pair of indexes n and m. Thus, in a grating having periodicity Λ, for each cladding mode determined by the generic pair of indexes n and m, there is a wavelength λ_Bnm, lower than the Bragg wavelength, fulfilling the relationship (2) and in correspondence of which there is a loss peak.

In general, the presence of these loss peaks is a disadvantage in the use of Bragg gratings. For example, when a Bragg grating is used for filtering a single channel in a multi-channel optical transmission system (that is, multi -wavelength) , for example for the purpose of adding or dropping said channel to/from the transmission system itself, the losses due to the coupling of the fundamental mode with cladding modes introduce an undesired attenuation of the signals in the transmission channels adjacent the channel to be filtered. In the following description, the power coupling between the fundamental mode and the cladding modes, substantially providing for a transfer of power from the first one to the second ones, shall be shortly referred to as (power) coupling into cladding modes.

Besides Bragg gratings at a constant periodicity, also Bragg gratings at non-constant periodicity - that is, at variable pitch - are known among fibre gratings. An important application of Bragg gratings at non-constant periodicity relates to the compensation of chromatic dispersion. Chromatic dispersion is a phenomenon of broadening of the transmitted optical pulses caused by the different velocity at which the various chromatic components (that is, the different wavelengths) of the transmitted light propagate. For this application, the gratings are realised with a periodicity Λ variable along the fibre. In this way, unlike the previous application wherein the reflection occurred at a single wavelength, it is possible to reflect the radiation on an entire spectrum band whose width can range from few nanometers up to some dozens nanometers. Moreover, different wavelengths are reflected by different portions of the grating, so that also the optical path covered by each of them is different. A suitable pattern of the grating allows using the above dependence of the optical path on the wavelength to compensate the chromatic dispersion of the light propagated through it. A grating of this type is commonly called "chirped". Thus, a chirped grating is a grating with variable pitch; in particular, it is a grating in which the areas with a high refractive index have an increasing or decreasing distance along the grating. Said variation of the grating pitch is defined "chirping factor" .

Since the gratings arranged for compensating the chromatic dispersion (that is, chirped gratings) usually reflect on a spectrum band having such a width as to include the spectrum region wherein there is a power coupling into cladding modes (with the consequent presence, in said area, of loss peaks) , the reflectivity curve of said grating is typically disequalised. Thus, here and hereinafter, the expression "wide band" refers to such a band as to include (at least partly) the above spectrum region seat of coupling phenomena into cladding modes. Figure 2 shows an example of reflectivity curve of a chirped grating.

On the basis of the above, it is clear that, for many applications of Bragg gratings in devices for treating optical signals, the entity of loss peaks due to power coupling into cladding modes must be reduced below predetermined limits.

As described in the article by L. Dong, L. Reekie, J. . Cruz, J. E. Caplen, J. P. de Sandro and D. N. Payne, "Optical fibers with depressed claddings for suppression of coupling into cladding modes in fiber bragg gratings", IEEE Photonics Technology Letters, Vol. 9, No. 1, January 1997, the intensity of the above peaks is proportional to the normalised overlap integral (NOI) expressed by the following formula: 2π

NOI_nm = \ X_λ{r) -X_m(r)-ex_V -i - — ._Sin{θ)-r- co^φ) r -dr -dφ (3 :

whereΨ*₀₁ (r) and Ψ*_nm(r) are the functions of the transverse electrical field of the fundamental mode and, respectively, of the cladding mode considered; θ is the tilt angle of the grating with respect to the fibre axis; r and φ are the radial and azimuthal coordinates in a polar coordinates reference system with centre on the fibre axis, and the index A_grating of the integral shows that the integral must be extended to the entire transversal section of the fibre wherein the grating has been written. In the case of gratings perfectly orthogonal to the fibre axis

(θ=0) , the integral calculated for non-symmetrical cladding modes in azimuthal direction, that is with n>0, is identically null and thus, the only coupling losses into cladding modes are the ones of the symmetrical modes LP_0m. Thanks to the precision of the current methods for writing gratings, the condition θ=0 is fulfilled with a particularly high precision, so that the power coupling on the above non-symmetrical cladding modes can be considered as being substantially null, so that the main contribution to the loss is that due to LP_0m modes. In this case, since the exponential term in the integral of the expression (3) is identically equal to 1, said expression can be rewritten as follows:

NOI_0m = Jψ (r)- Ψ₀ ,(r)- r - dr - dφ (4 )

For the purpose of reducing the grating losses, the value of the integral (4) must be as small as possible.

In addition, it must be taken into account that, by the definition itself of mode of a waveguide, the orthonormality relation between the modes expressed by the following relationship applies:

X_m{r)- r - dr - dφ = sl for n, m ≠ 0,1 (5) wherein in this case the integration extends to the entire section of the optical fibre, that is, to the outer limit of the cladding.

One of the following techniques can be used for reducing the value of the normalised overlap integral (4) .

A first technique for reducing the normalised overlap integral (4) consists in confining the field of the fundamental mode inside the fibre core, where the grating is written. By doing so, the product between fields Ψ₀₁(r)* and Ψ_0m(r)* outside the area _grai_ng tends to zero, and on the basis of the orthonormality relation (5) , the overlap integral tends to be null. For an optical fibre with step- index profile, the more fundamental mode is confined into the core the higher the value of a fibre parameter, commonly known as V number, such V number is defined as V= 2π-R_co-NA/λ, wherein NA is the numerical aperture of the fibre; R_co is the radius of the core, and λ is the wavelength of the radiation transmitted into the fibre. It is known that, for an optical fibre, it is possible to define a cut-off wavelength λ_c below which the transmission is multi-mode, and above which, it is single-mode. For step index fibres , the cut-off wavelength corresponds to a value of V equal to 2.405.

To increase the V number, fibres with high values of the numerical aperture NA and of the radius of the core R_co must be used. Increasing V corresponds to incrementing the ratio between the optical power of the fundamental mode confined into the core and its total optical power, said ratio defining the "confinement factor" of the field. Nevertheless, if the single-mode condition is to be maintained, the V number cannot be increased beyond the value of 2.405.

The Applicant has noted that similar considerations can be applied also to dispersion-shifted fibres. This is confirmed in the article by T. Komukai and M. Nakazawa, "Efficient fiber grating formed on high NA dispersion- shifted fibers", Proc. 21st European Conference on Optical Communications (ECOC '95 - Brussels), Mo.A3.3, pages 31-34, 1995, NTT Access Network Systems Laboratories, wherein it is stated that by making a Bragg grating on a dispersion- shifted fibre with high NA there are much lower losses on the cladding modes with respect to the case in which the same grating is made on a traditional dispersion- shifted fibre. This is due to a greater confinement of the optical field into the fibre core with high NA value. Moreover, in this article it is proved that a greater numerical aperture NA causes a greater amplitude of the spectrum region comprised between the highest wavelength associated to the cladding modes (herein referred to with λ_L) and the Bragg wavelength (referred to with λ_B) . Said spectrum region can be efficiently used as useful band of a chirped grating used for compensating the chromatic dispersion. In fact in said spectrum region there are no undesired contributions due to the coupling into cladding modes.

The Applicant has noted that, although the use of a fibre with high NA increments the distance between the Bragg wavelength and the maximum wavelength at which there is a coupling into the cladding modes, thus setting free an useful operating band, the width of said useful band is relatively small and much narrower than that needed for many applications.

A second technique for reducing the normalised overlap integral (4) consists in extending the writing area of the grating to the cladding area, at least up to an area wherein the field of the fundamental mode is substantially null. Also in this case, the coupling integral tends to coincide with the orthonormality integral, and thus tends to be null. To make this solution possible, it is necessary to make also the cladding area photo-refractive. For example, in the article by E. Delavaque, S. Boj , J. F. Bayon, H. Poignant, J. Le Mellot, M. Monerie, "Optical fiber design for strong grating photoimprinting with radiation mode suppression", Proc . , PD5, pp. 2-5, it is proposed that an intermediate layer be inserted between the core and the cladding of a fibre, said layer having a diameter equal to three times as much as the diameter of the core and doped with germanium so as to have the same photo-sensitivity as the core. In addition, they propose that fluorine be added as co-dopant so as to obtain the same refractive index of the cladding and obtain, thus, a step index fibre. Finally, the grating is written into the photosensitive area.

The Applicant has noted that, although it is technologically possible to make a photosensitive cladding around a photosensitive core, it is very difficult to obtain, at the same time, a high photosensitivity and a photosensitivity having the same entity in the core and the cladding .

A third technique for reducing the normalised overlap integral (4) consists in using a fibre with refractive index profile having a depressed area in correspondence with the most inner area of the cladding, that is, with the area of the cladding adjacent to the core. This type of fibre is usually called "fibre with depressed cladding" . Thus, in the above depressed area, the value of the refractive index is lower than those relating to the core and to the most outer area of the cladding. This characteristic allows reducing the amplitude of the field of the cladding modes in correspondence with the core area (where the grating is written) , thus minimising the value of the integral (4) .

This last technique has been the object of many studies, some of which are reported below.

The article by L. Dong, L.Reekie, J. L. Cruz, J. E. Caplen and D. N. Payne, "Cladding mode suppression in fiber Bragg gratings using fibers with a depressed cladding", Proc. 22nd European Conference on Optical Communications (ECOC'96 - Oslo), Mo. B.3.3, pages 53-55, 1996, ORC University of Southampton, and the subsequent article (already mentioned before) by L. Dong, L.Reekie, J. L. Cruz, J. E. Caplen, J. P. de Sandro and D. N. Payne, "Optical fibers with depressed claddings for suppression of coupling into cladding modes in fiber bragg gratings" , IEEE Photonics Technology Letters, Vol. 9, No. 1, January 1997, show that it is possible to reduce the coupling into cladding modes in a Bragg grating thanks to the introduction of a depressed cladding around the core area. The Applicant notes that the fibre used in these two articles is single- mode at the considered operating wavelength.

The article by C. W. Haggans, H. Singh, F. . Varner, J. S. Wang, "Narrow depressed-cladding fiber design for minimization of cladding mode losses in azimuthally asymmetric fiber Bragg gratings", J. of Lightwave Technology, vol. 16, no. 5, May 1998, considers the case of fibres in which the grating has some little azimuthal asymmetries, for example in a non-ideal case wherein the grating has a slight tilt with respect to the fibre axis, or writing asymmetries. The article shows that in this situation, by reducing the thickness of the portion of depressed cladding it is possible to reduce the losses on cladding modes. The Applicant points out that the fibre used in this article is single-mode, and on the basis of the values of the core radius and of the radius of the depressed area used, the Applicant has estimated that the ratio between the core radius and the radius of the depressed cladding area of said fibre is equal to about 0.86.

US Patent 5,852,690, assigned to Minnesota Mining and Manufacturing Company (3M) relates to a fibre with depressed cladding index profile for reducing coupling losses into cladding modes in fibre gratings, wherein the refractive indexes and the geometrical dimensions are selected so as to have a narrow depression area. In particular, the ratio between the core radius and the outer radius of the depressed cladding is comprised between 0.5 and 1. As regards to standard single-mode fibres, it is specified that the described waveguide improves the characteristics of losses at low wavelengths in case of small but not null asymmetries of the grating, for example as in the typical case of a tilted grating with an angle ranging between 0.25° and 1.5°. Moreover, said patent, in column 3, lines 19-29, specifies that the use of multi-mode fibres for housing the grating - although using the technique of confining as much as possible the core mode so as to reduce the coupling losses - is disadvantageous due to the high losses of optical power in correspondence with the junctions with the single-mode fibres typically used for transmitting signals. In addition, it is specified that further process steps must be adopted in order to have a reduction of said losses, so that the final device is more complex.

The European Patent Application EP 0831345 in the name of Sumitomo Electric Industries deals with the problem of coupling into cladding modes, and proposes an optical fibre with depressed cladding refractive index profile and with a grating photo-written into the core. As stated in column 3 of this patent, the characteristics of the fibre and of the cladding are preferably such as to meet two conditions that, together, guarantee the single-modality of the fibre. The first condition is that the parameter V must be greater than or equal to 2.4 so that the confinement factor of the fundamental mode is high; the second condition is that the refractive index n₃ of the cladding outer area (herein referred to as second cladding) is greater than the effective refractive index (N₂) of the secondary mode but lower than the refractive index (N_x) of the fundamental mode. In the section "Suppression of higher order modes" of column 6, they explain the reason for which the presence of higher order modes is regarded as undesired. In practice, in this case, there would be a coupling between the fundamental mode and the higher order modes and the quality of communication would be worse. Moreover, it is specified (col. 9, lines 2-7) that when the amplitude of the first cladding is comprised between 2 and 10 μm, the coefficient of confinement can be increased by 95%, and the conditions are such that only the fundamental mode can propagate, whereas the propagation of higher order modes is prevented.

The Applicant has noted that in the prior art, including the above mentioned documents, the use of a multi-mode fibre as fibre housing the grating is regarded as disadvantageous for several reasons. Among these reasons there is the fact that, in a real case wherein the grating is not perfectly symmetric, there are couplings between the fundamental mode and the higher order modes and the fact that the coupling between a multi-mode fibre and single- mode fibres for transmitting signals causes high losses .

The Applicant has found that in an optical device comprising an optical fibre with depressed cladding refractive index profile suitable to operate in multi-mode conditions at the wavelengths of interest and in the core of which a diffraction grating having variable periodicity

(pitch) is made, the problems arising from power coupling into cladding modes are especially reduced. Said device can advantageously be used for dispersion compensation in an optical telecommunication system. The above conditions of multi-modality (which, as already said, are obtained when the operating wavelength is lower than the cut-off wavelength) are met when the effective refractive index (π_eff) of at least one higher order mode is greater than the refractive index of the outer cladding area.

In particular, the Applicant has found that, by using a fibre with a relatively low refractive index step of the core (that is, of the difference between the refractive index of the core and that of the depressed area) , selecting the fibre parameters and, in particular, the core radius so as to make the fibre multi-mode, it is possible to obtain:

- a high suppression of the losses of cladding modes;

- a high reflectivity of the grating written into the fibre core ; and

- a low birefringence.

Moreover, the Applicant has found that an optical fibre having a refractive index profile of the depressed cladding type, a cut-off wavelength greater than 1650 and a thickness of the depressed cladding area at least equal to the core radius, is especially suitable to house a diffraction grating for making an optical device in which the losses due to power coupling to cladding modes are particularly reduced.

In addition, the Applicant has noted that:

- with a suitable selection of the fibre parameters, it is possible to make a junction between the multi-mode depressed cladding fibre and a standard single-mode fibre

(SM) with very low losses, lower than 0.3 dB;

- with a suitable technique for grating writing, the power coupling between the fundamental mode and the higher order guided modes caused by possible asymmetries of the grating can be made lower than the power coupling into the cladding modes; consequently, said effect does not significantly impair the performance of the grating;

- the effects of power coupling between guided modes due to fibre bending, for example caused by the assembly of the device into a container of practical use, are unimportant.

According to a first aspect thereof, the present invention relates to an optical-fibre device comprising an optical fibre and a diffraction grating made into said optical fibre, said optical fibre having: - a core, preferably doped with germanium, having a first radius and a first refractive index;

- a first cladding area made around said core and having a second radius and a second refractive index lower than said first refractive index; and

- a second cladding area made around said first cladding area and having a third radius and a third refractive index greater than said second refractive index and lower than said first refractive index;

said grating being made into said core along a longitudinal axis of said fibre; wherein said grating is of the variable pitch type, and said first and second radiuses and said first, second and third refractive indexes are selected so that the effective refractive index of at least one higher order mode is greater than said third refractive index.

Preferably, the ratio between said second radius and said first radius is greater than 2; more preferably, it is greater than 3.

Preferably, said diffraction grating has an operating spectrum band with width of at least 7 nm.

Advantageously, the effective refractive index for mode LP₀₂ is lower than said third refractive index.

Said optical fibre preferably hasa cut-off wavelength greater than 1650 nm; more preferably, it is comprised between 1650 nm and 2300 nm; for example, it is comprised between 1700 nm and 1900 nm.

The difference between said first refractive index and said third refractive index is preferably comprised between 0.008 and 0.015.

The difference between said second refractive index and said third refractive index is preferably comprised between -0.010 and -0.003. Said first radius is preferably comprised between 3.7 and 7 μm; more preferably, it is comprised between 4.3 and 5.8 μm.

According to a further aspect thereof, the present invention relates to an optical fibre usable for writing a diffraction grating, having:

a core having a first radius and a first refractive index;

- a first outer cladding area adjacent to said core and having a second radius and a second refractive index lower than said first refractive index; and

a second outer cladding area adjacent to said first cladding area and having a third radius and a third refractive index greater than said second refractive index and lower than said first refractive index;

wherein the cut-off wavelength is greater than about 1650 nm, and the ratio between said second radius and said first radius is greater than 2. Said cut-off wavelength is preferably comprised between about 1650 nm and 2300 nm; for example, it is comprised between about 1700 nm and 1900 nm.

The difference between said first refractive index and said third refractive index is preferably comprised between 0.008 and 0.015. The difference between said second refractive index and said third refractive index is preferably comprised between -0.010 and -0.003.

Advantageously, said core is doped with germanium.

Said first radius is preferably comprised between 3.7 and 7 μm, more preferably, between about 4.3 and 5.8 μm.

The ratio between said second radius and said first radius preferably is greater than 3.

Further details will appear from the following description which refers to the attached drawings:

Figure 1 shows a typical transmissivity curve of a periodical Bragg grating;

- Figure 2 shows a typical transmissivity curve of a Bragg grating suitable for the compensation of chromatic dispersion;

- Figures 3a and 3b respectively show an optical fibre with depressed cladding in which a chirped Bragg grating is obtained and a periodical Bragg grating;

- Figure 4 schematically shows the refractive index profile in the fibre of the invention;

- Figures 5a, 5b, 5c and 6 show the results of numerical simulations carried out on a step index fibre provided with a chirped grating;

- Figures 7a-7c, 8, 9a-9c, 10, lla-llc, 12 and 13 show the results of numerical simulations carried out on a fibre with depressed cladding index profile provided with a chirped grating;

Figures 14a- 14c show, on the basis of a numerical simulation, the junction losses between a fibre according to the invention and a single-mode fibre;

- Figure 15 shows a configuration for experimental tests;

- Figures 16a and 16b respectively show the transmissivity of a device manufactured according to the invention measured with the configuration of figure 15, and simulated with a numerical model;

- Figures 17a and 17b respectively show the transmissivity of a comparison device operating in a single-mode way, measured with the configuration of figure 15, and simulated with a numerical model; Figure 18 shows a reflectivity curve for the device manufactured according to the invention, obtained from the experimental curve of figure 16a; and

- Figure 19 shows a reflectivity curve measured for the comparison device operating in a single-mode way, obtained from the experimental curve of figure 17a.

With reference to figure 3a, numeral 1 refers to an optical fibre comprising a core 2, preferably central with respect to the fibre itself, and with circular section, and a cladding 3 with an annular section, external to the core 2. Into the core 2 of the optical fibre there is a Bragg grating 4. The optical fibre 1 and the grating 4 define an optical device 5.

Grating 4 is a chirped grating comprising a plurality of areas with a high refractive index 4a and of areas with a low refractive index 4b, alternating with one another along the axis of the optical fibre 1. The distance between two consecutive areas with a high refractive index 4a (or between two consecutive areas with a low refractive index 4b) defines the grating pitch Λ, which in the present case is progressively increasing shifting from the left rightwards along the fibre axis.

The core 2 has a first radius R and a first refractive index n₁ . Cladding 3 comprises an inner cladding area 3a (also called depressed cladding area and represented with a first grey hue) adjacent to said core 2 and having a second radius R₂ and a second refractive index n₂ lower than the first refractive index , and an outer cladding area 3b (represented with a second grey hue) which is outer with respect to the inner cladding area 3a, and has a third radius R₃ and a third refractive index n₃ greater than the second refractive index n₂ and lower than the first refractive index n_x . The inner cladding area 3a is also called depressed cladding area, in relation to the pattern of the refractive index. Figure 3b shows an optical device 5' which differs from device 1 only in that the Bragg grating, herein referred to with 4', is a constant-pitch grating.

The characteristic parameters of device 5 havethe following values :

The quantity _x - n₃ is preferably comprised between 0.008 and 0.015.

The quantity n₂ - n₃ is preferably comprised between -0.010 and -0.003.

The quantity p=R₂/R₁ is greater than 2 and preferably greater than 3, for example comprised between 3 and 5.

The cut-off wavelength is greater than 1650 nm and preferably lower than 2300 nm, for example comprised between 1700 nm and 1900 nm.

The first radius R_x is preferably comprised between 3.7 and 7 μm, more preferably it is comprised between 4.3 and 5.8 μm.

The third radius R₃ is preferably comprised between 62 and 63 μm, for example equal to about 62.5 μm.

The Applicant has found that, using a fibre with a relatively low refractive index step between core and depressed cladding area ( - n₂ , hereafter briefly referred to refractive index step of the core) , selecting the fibre parameters and, in particular, the radius of core 2 so as to make the fibre single-mode, it is possible to obtain:

- a high suppression of losses of cladding modes;

- a high reflectivity of the grating written into the fibre core ; and

- a low birefringence.

The Applicant has also noted that, to obtain the desired value of the cut-off wavelength of a fibre with depressed cladding refractive index profile, it is possible to intervene, besides on the diameter of the core and on the thickness of the core depressed area, also on the refractive index step of the core through a suitable selection of the type and of the concentration of the dopants. In particular, the refractive index step of the core can be controlled by regulating the concentration of Ge into the core itself. Said concentration also determines the photo-refractivity of the core.

Nevertheless, the Applicant has noted that, in the specific case, an increase in the concentration of Ge into the core causes the negative effect of increasing the state of stress of the glass into the interface area between core and depressed cladding. It is known that this stress area significantly contributes to make the fibre birefringent , that is, to determine a diversity of the transmission characteristics for orthogonal polarisation modes, thus worsening the characteristics of the fibre in terms of polarisation mode dispersion.

Figure 4 schematically shows the refractive index profile in fibre 1 in function of radius R, measured starting from the longitudinal axis of the fibre itself. Radiuses R and R₂ of core 2 and of the inner cladding area 3a respectively, and the refractive index steps n₁-n₂ and n₂- n₃ between core 2 and inner cladding area 3a, and respectively, between inner cladding area 3a and outer cladding area 3b, are selected so that the guided propagation of at least one further core mode besides the fundamental mode LP₀₁ is possible, that is, of a higher order core mode. Each further core mode has an effective refractive index n_eff with a greater value than that of the refractive index n₃ in the outer cladding area 3b (as exemplified in figure 4) .

Additionally, radiuses R_x and R₂ and the refractive index steps n-_L-n-₂ and n₂-n₃ are selected in such a way that the only azimuthal symmetry core mode that can propagate is the mode LP₀₁, whereas the propagation of the guided mode LP₀₂ is not possible. In particular, the above parameters are selected so that besides the propagation of the fundamental mode, also the propagation of the higher order mode LP_1:L is possible, and preferably, also that of the higher order modes LP₁₂ and LP₂₁, all of which (at least for fibres with depressed refractive index profile) have an effective refractive index which is greater than that of the mode LP₀₂. In other words, the effective refractive index of mode LP_1:L and preferably, also the effective refractive index of the higher order modes LP₁₂ and LP₂₁, are greater than the third refractive index n₃.

The ratio p = R₂/Rχ between the second radius R₂ (relating to the depressed cladding area 3a) and the first radius R_x

(relating to core 2) is selected greater than 2 so as to increment, as described hereafter, the effect of reduction of losses due to power coupling into cladding modes.

Although in the present description reference is made to an "ideal" refractive index profile wherein, as shown in figure 4, the refractive index steps are "net", each consideration made is applicable also to a "real" refractive index profile (which is also affected by the effects of machining of the preform and of drawing of the fibre) wherein the refractive index steps provide for "gradual" and not "net" variations. The difference noted between a "real" refractive index profile and an "ideal" one, in fact, is not such as to introduce substantial and detectable effects for the device of the present invention.

The pattern of the transmissivity characteristic of a Bragg grating written into a depressed cladding fibre as the parameters of the refractive index profile change has been studied using a numerical model. A man skilled in the art, that is, a man skilled in the propagation of light in optical fibre, can calculate, once the refractive index profile of the fibre has been defined, the constants of propagation of the core and cladding modes and the distribution of the electromagnetic field for the different modes with azimuthal symmetry LP_0m. Consequently, a skilled man is capable of calculating the overlap integral (4) . Thus, once the characteristics of the grating are known, that is, the function describing the modulation of the refractive index, it is possible to calculate the pattern of the grating transmissivity in function of the wavelength.

At first, the model has been applied to a step index optical fibre, that is, an optical fibre having a predetermined refractive index step between the core and the cladding, comprising a chirped Bragg grating into its core. Said type of fibre can be considered an extreme case of a depressed cladding fibre wherein the depression is null. The grating considered for the model has a Bragg wavelength equal to 1.55 μm, and a reflectivity at this wavelength equal to 30 dB . The refractive index step between core and cladding has been selected equal to 0.011 (to which corresponds a numerical aperture NA equal to 0.18) and the radius of the core has been made vary so as to obtain three different cut-off wavelength λ_c respectively equal to 1200 nm, 1550 nm and 1900 nm, which are respectively lower than, coinciding with, and greater than the Bragg wavelength. Starting from these values, the model allowed calculating the pattern of the transmissivity curve of the grating and thus, in particular, the spectrum position and the entity of the loss peaks due to power coupling of mode LP₀₁ with each of the modes LP_0m with m>l, within a band of about 30 nm starting from the Bragg peak. Figures 5a, 5b and 5c show the pattern of the transmissivity losses (transmission loss) of the grating as the wavelength changed, respectively for λ_c equal to 1200 nm, 1550 nm and 1900 nm, obtained with the above numerical model.

More in detail, the peaks in the diagram are to be read as follows. Starting from the Bragg wavelength (that is, 1550 nm) and moving to lower wavelengths (that is, moving leftwards on the axis of wavelengths) , the first peak represents the loss due to the coupling of mode LP₀₁ with mode LP₀₂, the second peak represents the loss due to the coupling of mode LP₀₁ with mode LP₀₃, and so on. As it can be noted from the different curves, as the order of the mode increases, the loss are increasing at first, then they reach a maximum level, and decrease until they reach a minimum, and then they are relatively low. In particular, it can be noted that loss peaks tend to concentrate in a spectrum region close to the Bragg wavelength. In case of a chirped grating, the spectrum distribution and the entity of loss peaks determine the pattern of the integral loss curve of the grating.

By comparing the graphs of figures 5a, 5b and 5c, it can also be noted that the increase of the cut-off wavelength causes a progressive decrease of the height of loss peaks . In fact, although the increment of the cut-off wavelength beyond the Bragg wavelength causes a relatively small increase of the confinement factor of the fundamental mode LP₀₁ inside the core, there is anyway a significant decrease of the coupling integral (4) and thus, a reduction of losses.

Figure 6 shows, always with regard to the above fibre provided with a chirped grating, and operating between 1520 nm and 1550 n , reflectivity curves at the above three cutoff wavelengths . Each curve has been obtained by integrating the transmission losses of the grating by moving from the Bragg wavelength to lower wavelengths. As it can be noted, the different curves are substantially flat in the proximity of the Bragg wavelength due to the substantial absence of losses in this spectrum region and, when the loss contributions associated to modes LP₀₂, LP₀₃ , etc., start, they have a decreasing pattern. In addition, since the entity of loss peaks is relatively reduced in the spectrum region more on the left of the characteristics of figures 5a- 5c, each reflectivity curve of figure 6 tends to be substantially constant also in this spectrum region, so that the general ^■ pattern of each curve is assimilable to a "step" profile. By comparing the integral losses obtained for different cut-off wavelengths, it can be noted that an increase in the cut-off wavelength causes a reduction in the height and in the spectrum extension of the step. All of these characteristics contribute to making the pattern of the chirped grating reflectivity "flatter", that is, "more equalised"; said condition is advantageous for the purpose of having an even behaviour of the grating at all the wavelengths it must operate. Said condition especially concern wide-band chirped gratings, that is, chirped gratings wherein the operating band extends beyond the initial area without loss peaks.

The same numerical model has been successively applied to a fibre having a depressed cladding refractive index having the following parameter values:

- refractive index step n₁-n₃ between core 2 and outer cladding area 3b equal to 0.011 (equal to that of the step index fibre previously considered) ;

refractive index step n₂-n₃ between depressed cladding area 3a and outer cladding area 3b equal to -0.005.

Figures 7a, 7b and 7c show the pattern of transmissivity losses of the grating in the case of a fibre with depressed cladding calculated with the above numerical model, respectively for values of the cut-off wavelength λ_c equal to 1200 nm, 1550 nm and 1900 nm, and with a value of ratio p between the radius of the depressed cladding R₂ and the radius of core R_λ equal to 2.

Figures 9a, 9b and 9c show the pattern of transmissivity losses of the grating in the case of a fibre with depressed cladding calculated with the above numerical model, respectively for values of the cut-off wavelength λ_c equal to 1200 nm, 1550 nm and 1900 nm, and with a value of ratio p between the radius of the depressed cladding R₂ and the radius of core R equal to 3.

Figures 11a, lib and lie show the pattern of transmissivity losses of the grating in the case of a fibre with depressed cladding calculated with the above numerical model, respectively for values of the cut-off wavelength λ_c equal to 1200 nm, 1550 nm and 1900 nm, and with a value of ratio p between the radius of the depressed cladding R₂ and the radius of core R_± equal to 4.

Figures 8, 10 and 12 show reflectivity curves at the above cut-off wavelengths respectively associated to the patterns of figures 7, 9 and 11 and obtained in a similar way as the curve in figure 6 (that is, by integrating the transmission losses of the grating moving from the Bragg wavelength to lower wavelengths) .

As it can be noted by comparing figures 7, 9 and 11, and figures 5, in a depressed cladding fibre the loss peaks have - difference of refractive index between core and outer cladding area and of cut-off wavelength being equal - a lower height with respect to the case of a step index fibre .

By comparing figures 8, 10 and 12, and the curves of figure 6, it can also be noted that, as a consequence of the lower entity of loss peaks, the transmissivity losses of a depressed cladding fibre are lower than those of a step index fibre .

Moreover, it can be noted that, as it happens for step index fibres, the increment of the cut-off wavelength determines a progressive concentration of the highest loss peaks into an increasingly reduced spectrum region.

The Applicant has noted that it is not possible to establish a general rule for which the loss peak relevant to the coupling with the generic order mode m (LP_0τn) always decreases in a monotonic way as the cut-off wavelength increases, nor it is possible to establish a general rule for which the loss peak of the generic order mode m decreases as the width of the depressed cladding area increases, other conditions being equal (or, similarly, as the ratio p between radius R₂ of the inner cladding area 3a and radius R_x of core 2 increases) . On the contrary, it can be said that, as the cut-off wavelength and the width of the depressed cladding area increase, there is an overall improvement of the characteristics of the integral loss curve. In the case of applications providing for the use of wide-band gratings (through a chirping process) , such as for example for the compensation of dispersion, integral losses are the most significant data. In addition, since there is an overall reduction of all loss peaks with the exception of a very limited number of them (possibly only one in case of relatively high cut-off wavelengths) , in the applications with periodical gratings the overall performances of the grating are improved on condition that operating wavelengths are not used in correspondence with the above peaks with a non-reduced entity.

Similarly to the case of a step index optical fibre, the increment of the cut-off wavelength determines, upon ratio p being equal, a reduction in the height and in the spectrum extension of the step in the reflectivity curves, so that the response curve of the grating is more equalised in the spectrum band of interest. The Applicant has found that, for value of the ratio p greater than 2, the curve is particularly equalised.

Figure 13 shows curves relating to the dependence of the reflectivity of the chirped grating on the refractive index step n₁-n₃ with equal cut-off wavelength (in this case, equal to 1550 nm) , equal ratio p (in this case equal to 3) and equal index step n₃-n₂ (in this case, equal to -0.005) . From the pattern of the curves of figure 13 , it can be deduced that it is not possible to determine a single selection criterion of the refractive index step n-_L*-^ to improve the response of the grating. In fact, by increasing the value of the refractive index step n₁-n₃, there is a shifting to lower wavelengths but at the same time, there is an increase in the height and in the spectrum extension of the step itself. Thus, the selection of the value of the refractive index step -!*,_-n₃ must be based also on other considerations described hereinafter.

For a correct operation of a device according to the invention, the grating must substantially reflect only the fundamental mode LP₀₁. Thus, it is necessary to prevent the optical power propagating into the grating itself from being coupled to one or more guided modes of higher order. A first reason for this coupling may be the noise effect generated by the junction between the multi-mode fibre comprising the grating and a single-mode fibre transmitting the radiation, usually present when the above device is used for optical telecommunications. A second reason for the above coupling may be an asymmetry of the grating and, more in particular, a defect in the grating orientation

(which, in the case considered, should be orthogonal to the fibre axis) and/or a defect in the uniformity of its refractive index in an orthogonal section with respect to the fibre axis.

The coupling into higher order guided modes caused by the junction with single-mode fibre can be ideally suppressed if:

- the depressed cladding multi-mode fibre and the single- mode one are jointed with the respective axes being perfectly aligned. Said condition, as it is know from A. W. Snyder, Y. D. Love, "Optical Waveguide Theory", Chapman & Hall, New York, 1983, prevents the excitation of odd order guided modes; and

- the distribution of the electromagnetic field of the fundamental mode in the two fibres is the same.

In a real situation, to minimise the power coupling into higher order guided modes when the depressed cladding multi-mode fibre is jointed with a single-mode fibre, it is necessary to both align as much as possible the axes of the two fibres, and to select the parameters of the refractive index profile of the multi-mode fibre {R_λ , R₂ , n-_L-n-_j and n₂- n₃) in such a way that the distribution of the electromagnetic field of the fundamental mode of the multi- mode fibre itself is as similar as possible to the distribution of the electromagnetic field of the fundamental mode in the single-mode fibre. In this way, it is possible to maximise the power coupling between the LP₀₁ modes of the two fibers with the consequent effect of minimising the insertion losses of the multi-mode fibre and minimisingthe excitation of even order higher modes. Moreover, the adaptation between the distributions of the fundamental mode fields in the two fibres allows increasing the tolerance to the axial disalignment of the two fibres.

Moreover, the entity of junction losses (for coupling into higher order guided modes) between the fibre object of the invention and a standard SM single-mode fibre (the parameter values of which have been assumed as being equal to the typical ones guaranteed by the manufacturers) has been assessed. This evaluation has been carried out using a numerical model based on the normalised overlap integral of the fundamental modes of the two fibres, and assuming that at the interface between the two fibres, there are no phenomena of local variation of the index profiles induced by the junction process. Losses have been calculated as the cut-of wavelength of the multi-mode fibre changed, for different values of ratio p and different values of the index step n₁-n₃. The results of these simulations are reported into figures 14a, 14b and 14c.

From these curves, it can be noted that junction losses decrease as the cut-off wavelength increases, and they are especially reduced for cut-off wavelengths greater than those typically used in the third window of optical fibre telecommunications (about 1550 nm) , where the depressed cladding fibre is multi-modal. In addition, always as regards to losses, in said multi-mode area no significant differences are noted as the ratio p changes, other parameters being equal. On the contrary, losses are sensitive to variations of the index step n₁-n₃; in particular, a decrease in the refractive index step n₁-n₃ causes a decrease of losses. Indicatively, for a smaller index step n*,_-n₃ than 0.011 and a greater cut-off wavelength than 1650 nm, junction losses are less than 0.2 dB, and these values are totally acceptable for the use in telecommunication systems.

Thus, since the shape of the step of the reflectivity curves of figure 13 varies in function of the index step n₁-n₃ (being the cut-off wavelength and the ratio p fixed) in such a way that it is not possible to identify an optimal value of the index step n-_L-n**, in relation to the problem of power coupling into cladding modes, the selection of the index step n₁-n₃ can be carried out so as to minimise junction losses.

As said before, a second reason for power coupling into higher order guided modes may be the possible asymmetry of the grating. Nevertheless, the Applicant has noted that, using a fibre having a core radius (R_x) comprised between about 4 and 5 μm, and such a content of Ge in the core as to obtain a index step n₁-n₃ comprised between about 0.011 and 0.0146, it is possible, by using the process and the parameters of writing specified below, to obtain chirped gratings with such a degree of uniformity of the refractive index in a plane orthogonal to the fibre axis so as to not introduce substantial effects of coupling into core or cladding modes that are not azimuthally symmetric.

As to verify the previously described numerical results, an optical fibre with depressed cladding refractive index profile has been manufactured, having the following characteristics :

- a refractive index step n₁-n₃ between core 2 and outer cladding area 3a equal to 0,011;

- a refractive index step n₂-n₃ between depressed cladding area 3a and outer cladding area 3b equal to -0,005;

- a ratio p equal to 3.2;

- a thickness of the depressed cladding area 3a equal to 12.3 μm; and

- a cut-off wavelength equal to 1830 nm,- and

- a birefringence equal to about 5.2-10^"7.

For this fibre, the cut-off wavelengths of higher modes than mode LP_1;L are all lower than 1280 nm; thus, the only guided modes of core are modes LP₀₁ and LP_1:L . In the above fibre, Bragg gratings were written using the technique described in the patent application UK n. 9617688.8 in the name of University of Southampton, briefly described in the following.

During the process of writing, the fibre is translated with a steady motion. The UV laser beam is sent to the fibre through an acoustic-optical modulator and a phase mask placed in the proximity of the fibre itself. The interference between the beams diffracted by the mask of order 1 and -1 generates a series of fringes which define the periodical modulation of the UV beam. Through the acoustic-optical modulator, the exposure to the UV beam is periodically interrupted and restored when the fibre has been moved by a distance equal to the period Λ of the refractive index modulation. In this way, the fringes of the UV beam are overlapped to the previously exposed areas.

Thus, the grating is made with a relatively high number of subsequent exposures. A suitable technique of apodisation (of the known type, and not described herein) has been used for suppressing the secondary lobes typically present in the grating spectrum response.

In particular, narrow-band (0.2 nm) gratings with a length equal to 30 mm have been made, with a typical transmissivity greater than 30 dB . The transmissivity spectrum of these gratings has been measured through a spectrum analyser with a resolution of 0.01 nm.

For the purpose of minimising the coupling into anti- symmetrical cladding modes, the gratings were prevented from having asymmetries .

In particular, to prevent the grating from being slanted with respect to the fibre axis (slanted grating) due to a modulation of the refractive index not parallel to the fibre axis, a careful alignment of the phase mask was carried out, checking on whether the interference fringes were orthogonal to the axis of the optical fibre. Moreover, for the purpose of preventing the distribution of the refractive index from not being even in a section orthogonal to the fibre axis due to an uneven absorption of

UV radiations (this is typically due to the fact that the side of the fibre core facing the UV source absorbs a greater fraction of the UV radiation with respect to the remaining part) , the writing was carried out at a relatively reduced translation speed of the fibre (equal to 0.3 mm/s) , so as to approach the conditions of saturation that bleach the portion of core first impinged by the UV beam, thus guaranteeing that the opposed portion of core would substantially receive the same UV intensity. Finally, for the purpose of preventing the distribution of the refractive index from not being even in a section orthogonal to the fibre axis, as a consequence of a non- perfect focusing of the UV beam into the fibre core, during the entire writing of the grating a technique for tracking the position of the UV beam with respect to the photo- refracting core of the fibre was used. Said tracking technique provides for the detection, through a photodetector, of the fluorescence induced by the UV beam and guided by the fibre itself, and the alignment of the various parts so that the intensity of the detected signal is maximum.

Figure 15 shows a configuration used for experimental tests on device 5. The characteristics of device 5 are those just described. The ends of fibre 1 were jointed with respective standard SM single-mode fibres 7a, 7b (through fusion and with junction losses equal to about 0.26 dB) . The transmissivity spectrum of device 5 was detected by feeding optical power to one of the single-mode fibres (for example, fibre 7a) and measuring the optical power in output from the other single-mode fibre (fibre 7b) . For the purpose of reducing as much as possible the possible optical power present on the higher order guided mode P_U caused by the non-perfect adjustment of the distribution of the fundamental modes in the different fibres and by the possible asymmetries in correspondence with the junctions, the multi -mode fibre was curved so as to obtain, both before and after grating 4 (schematically shown) , three fibre rings 10a, 10b with a diameter equal to about 30 mm. The Applicant has checked that these rings allow inducing bending losses on the higher order mode LP_11; whereas there are no significant effects on the fundamental mode.

Figure 16a shows the transmissivity spectrum of device 1 measured with the configuration just described. The maximum attenuation in correspondence with the Bragg wavelength is equal to about 38.1 dB . Figure 16b shows the results of the numerical simulation carried out on the same device using the numerical model already used before. It is possible to note the presence, in the experimental graph, of a loss peak at about 1550 nm, not present in the numerical simulation graph. Said peak, probably caused by the presence of residual asymmetries in the grating, has an especially reduced height (about 0.4 dB) ; thus, it is not such as to impair the substantial advantages of using a multi-mode fibre.

For the purpose of experimentally checking the advantages of using a multi-mode fibre, a second fibre with cut-off wavelength equal to 1370 nm and thus, single-mode at the typical operating wavelengths, was manufactured starting from the same preform used for the previous fibre. Besides the cut-off wavelength, said further fibre has the same characteristic values as the multi-mode fibre; in particular, it has the same values of the refractive indexes of the different core and cladding areas, as well as the same values of the relative ratios between the radiuses of the core and cladding areas. A Bragg grating was written into said fibre, equal to that made in the multi -mode fibre, for the purpose of obtaining a comparison optical device. Figures 17a and 17b show the corresponding measured and simulated transmissivity spectrums . The attenuation peak of this device is of about 39 dB, similar to that of the device according to the invention. It can be noted that the entity of loss peaks is much greater with respect to the case of the device according to the invention. Figures 18 and 19 show the reflectivity curves relating to the device with multi-mode fibre made according to the invention and to the device with single-mode fibre used for comparison, obtained by calculating the integral losses starting from the experimental spectrums of figures 16a and 17a respectively. The greater equalisation of the reflectivity curve of the device according to the invention is evident. While in figure 19 it is possible to note a plurality of steps having reduced height but such as to provide for a significant overall contribution, in the curve of figure 18 there are only a first step 8 with reduced height, due to the coupling to mode LP_X1, and a second step 9 with reduced height (although greater than that of step 8) , due to the highest loss peak.

Device 5 can be housed into a container to protect it during use. The Applicant has noted that the insertion of the device into a container wherein the fibre has, in the area taken by the grating, a maximum curvature with a radius of about 50 mm, does not cause substantial variations in the transmissivity spectrum of the grating itself .

The device of the present invention can be used in an optical telecommunication system in order to compensate chromatic dispersion. The telecommunication system considered herein is a long-distance WDM (Wavelength Division Multiplexing) telecommunication system, for example, a submarine telecommunication system, wherein the problem of chromatic dispersion is significant.

An optical telecommunication system typically comprises a transmitting station, a receiving station and an optical communication line connecting the transmitting and receiving stations.

The transmitting station comprises a plurality of optical transmitters each being suitable to transmit an optical signal at a respective wavelength. Each optical transmitter can, for example, comprise a laser source and a wavelength converter suitable to receive the signal generated by the laser and to transmit a signal at a predetermined wavelength. A wavelength multiplexer is connected in input to the transmitters for receiving the plurality of transmitted signals, and it has a single output connected to the communication line for transmitting the wavelength- multiplexed signals on the line. In addition, the transmitting station can comprise a power amplifier, connected to the multiplexer output, to impart to the transmitted signals the power needed for the transmission along the line.

The receiving station comprises a wavelength demultiplexer connected in input to the line for receiving the transmitted signals, and is provided with a plurality of outputs on which the different wavelengths transmitted are separated. The receiving station also comprises a plurality of optical receivers, each connected to a respective output of the demultiplexer for receiving a signal at a respective wavelength. Each receiver can comprise a wavelength converter suitable to convert the wavelength of the signal into a wavelength which is suitable for the reception of the signal by a photodetector optically connected to the converter itself. In addition, the receiving station can comprise a pre-amplifier, arranged upstream of the demultiplexer, for imparting to the transmitted signals the power needed for a correct reception.

The communication line comprises several spans of optical fibre (preferably, single-mode optical fibre) and a plurality of line amplifiers, spaced from one another (for example, by about a hundred kilometres) suitable to amplify the signals at a level of power suitable for the transmission on a subsequent optical fibre span.

A device according to the invention can be used along the telecommunication line and/or in transmitting and receiving stations. The use of the device according to the invention for compensating chromatic dispersion is especially advantageous when the transmission speeds are of 10 Gbit/s or higher. A dispersion compensator using the device of the invention can comprise a three-port optical circulator, suitable to receive at an input port the transmitted signals and to feed said signals to an output port after they have passed through an intermediate port . Input and output ports can be serially connected to the communication line. A portion of the fibre of device 5 is connected to the intermediate port. During use, when the signals reach the intermediate port, enter into fibre 1, are reflected by grating 4 and pass again through the intermediate port to continue to the output port. Grating 4 reflects the different wavelengths of the signals in correspondence with different areas, thus allowing the reduction of the chromatic dispersion of the signals.

Claims

1. Optical-fibre device comprising an optical fibre (1;1') and a diffraction grating (4;4') made into said optical fibre (1;1'), said optical fibre (1;1') having:

- a core (2) , having a first radius (R_x) and a first refractive index (n_x) ;

- a first cladding area (3a) made around said core (2) and having a second radius (R₂) and a second refractive index (n₂) lower than said first refractive index (n_x) ; and

- a second cladding area (3b) made around said first cladding area (3a) and having a third radius (R₃) and a third refractive index ( ) greater than said second refractive index (n₂) and lower than said first refractive index (n₂) ;

said grating being made into said core (2) along a longitudinal direction of said core (2) ;

characterised in that said grating (4;4') is of the variable pitch type, and said first and second radiuses (R-_L, R₂) and said first, second and third refractive indexes ( , n₂ , n₃) are selected so that the effective refractive index (n_eff) of at least one higher order mode is greater than said third refractive index ( ) .

2. Optical-fibre device according to claim 1, characterised in that the ratio (p= R₂/R-_L) between said second radius (R₂) and said first radius (R- is greater than 2.

3. Optical-fibre device according to claim 2, characterised in that the ratio (p=R₂/R₁) between said second radius (R₂) and said first radius {R_x) is greater than 3.

4. Optical-fibre device according to claim 1, characterised in that said diffraction grating (4;4') has an operating spectrum band having width of at least 7 nm.

5. Optical-fibre device according to claim 1, characterised in that the effective refractive index for mode LP₀₂ is lower than said third refractive index (m) .

6. Optical-fibre device according to claim 1, characterised in that said optical -fibre (5) has a cut-off wavelength

(λc) greater than 1650 nm.

7. Optical-fibre device according to claim 6, characterised in that said optical-fibre (5) has a cut-off wavelength (λc) comprised between 1650 nm and 2300 nm.

8. Optical-fibre device according to claim 6, characterised in that said optical-fibre (5) has a cut-off wavelength (λc) comprised between 1700 nm and 1900 nm.

9. Optical-fibre device according to claim 1, characterised in that the difference between said first refractive index (n_x) and said third refractive index (n₃) is comprised between 0.008 and 0.015.

10. Optical -fibre device according to claim 1, characterised in that the difference between said second refractive index (n₂) and said third refractive index (n₃) is comprised between -0.010 and -0.003.

11. Optical -fibre device according to claim 1, characterised in that said core (2) is doped with germanium.

12. Optical-fibre device according to claim 1, characterised in that said first radius (R_x) is comprised between 3.7 and 7μm.

13. Optical-fibre device according to claim 12, characterised in that said first radius (R- is comprised between 4.3 and 5.8μm.

14. Optical-fibre usable for writing a diffraction grating (4 ;4 ' ) having: a core (2) , having a first radius (R_x) and a first refractive index (n_x) ;

- a first outer cladding area (3a) adjacent to said core (2) and having a second radius (R₂) and a second refractive index (n₂) lower than said first refractive index (n_x) ; and

- a second outer cladding area (3b) adjacent to said first cladding area (3a) and having a third radius (R₃) and a third refractive index (n₃) greater than said second refractive index (n₂) and lower than said first refractive index (n_x) ;

characterised in that it has a cut-off wavelength (λc) greater than about 1650 nm and that the ratio (p) between said second radius (R₂) and said first radius (R- is greater than 2.

15. Optical-fibre according to claim 14, characterised in that said cut-off wavelength (λc) is comprised between 1650 nm and 2300 nm.

16. Optical-fibre according to claim 15, characterised in that said cut-off wavelength (λc) is comprised between 1700 nm and 1900 nm.

17. Optical -fibre according to claim 14, characterised in that the ratio (p) between said second radius (R₂) and said first radius (R_x) is greater than 3.

18. Optical-fibre according to claim 14, characterised in that the difference between said first refractive index

(]!_]_) and said third refractive index (n₃) is comprised between 0.008 and 0.015.

19. Optical-fibre according to claim 14, characterised in that the difference between said second refractive index (n₂) and said third refractive index (n₃) is comprised between -0.010 and -0.003.

20. Optical-fibre according to claim 14, characterised in that said core (2) is doped with germanium.

21. Optical-fibre according to claim 14, characterised in that said first radius (R_x) is comprised between 3.7 and 7μm.

22. Optical-fibre according to claim 21, characterised in that said first radius (R-_L) is comprised between 4.3 and 5.8μm.