Magnonic signal amplification device

The device uses a resonant RF cavity and passive blocks to amplify spin waves in multiple zones of a magnonic waveguide, addressing attenuation issues and enabling efficient, low-complexity signal propagation for magnonic logic circuits.

FR3144488B1Active Publication Date: 2024-11-29THALES SA +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
FR2022014202
Authority / Receiving Office
FR · FR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2022-12-22
Publication Date
2024-11-29
Estimated Expiration
2042-12-22

AI Technical Summary

Technical Problem

Existing magnonic logic circuits face challenges in maintaining signal amplification over long distances due to spin wave attenuation, requiring complex and energy-intensive individual current supply for each amplification zone.

Method used

A device utilizing a resonant RF cavity and passive parametric amplification blocks within a magnonic waveguide, enabling localized spin wave amplification in multiple zones with a single RF wave source, without the need for individual current supply.

Benefits of technology

Enables efficient and cost-effective spin wave amplification over extended distances with minimal energy and complexity, facilitating the creation of all-magnon logic circuits.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000015_0000
    Figure 00000015_0000
  • Figure 00000015_0001
    Figure 00000015_0001
  • Figure 00000015_0002
    Figure 00000015_0002
Patent Text Reader

Abstract

Magnonic signal amplification device (1) comprising: a resonant RF cavity (C) at a frequency; a substrate (Sub); a magnonic waveguide (WG) of direction, deposited on the substrate; an excitation device (SRF) adapted to excite, from an RF magnetic field called excitation (HRF) of frequency, a spin wave (OS) at a frequency propagating within the waveguide; an RF wave source called amplification source (Samp), the amplification source and the RF cavity being adapted to emit an additional magnetic field (Hadd) at a frequency, according to the direction;a plurality of passive parametric amplification blocks (Amp), arranged above said waveguide, or arranged between the waveguide and the substrate, or formed by parts of the waveguide, said amplification source and said blocks being adapted so that, locally in a plurality of distinct zones (ZA) of said waveguide, a power of the additional magnetic field is greater than a parametric amplification threshold, called threshold power (Ps), forming an amplification magnetic field (Hamp) thus generating a parallel parametric amplification (APP) of the spin wave. [Figure 3];
Need to check novelty before this filing date? Find Prior Art

Description

Title of the invention: Device for amplifying the magnetic signal Technical field

[0001] The invention relates to the field of microwave electronics. It relates to a device for amplifying the magnetic signal. Prior art

[0002] The field of study dealing with magnetic phenomena related to spin waves is magnonics, named after the excitation quanta of spin waves: magnons. This field aims to study the generation, propagation, manipulation and detection of spin waves. If in recent years there has been a resurgence of efforts around magnonics, it is largely due to the fact that it is seen as one of the most promising ways to achieve ICT technologies (integrated circuit technologies). The advantage of magnonics lies in the frequency range of spin waves which extends from gigahertz to terahertz (500MHz-a few THz) and the wavelengths (1 / ^2-1077777) corresponding to the scales of modern electronics and spintronics.Thus, extremely small and integrated components can be envisaged for analog signal processing, directly in the radio frequency band, bringing many advantages compared to existing semiconductor and photonic components: compactness, agility, selectivity, radiation resistance. It is also possible to realize logic gates based on magnon transistors with the advantage that the output signal of the gate can be sent directly as an input signal to the next logic gate, thus allowing the design of all-magnon circuits similar to CMOS integrated circuits.

[0003] A spin wave can be schematically defined as an inhomogeneous precession of the magnetization for which the spins oscillate at the same frequency but with a certain phase shift between them. The Walker equation makes it possible to define the equation for the propagation of spin waves in a ferromagnetic material: P»04! ¢[^(1 + ^)7^=(),

[0005] with the vacuum permeability, the microwave susceptibility of the material and V is a scalar potential of the magnetic field h as such h — (p.

[0006] Figure 1A schematically illustrates a spin wave propagating along a wave vector k with an angle dk relative to the static magnetization direction Ms. There are two configurations mainly studied for the propagation of spin waves: • in a film with in-plane magnetization Ms, the spin wave propagates with a direction of the propagation wave vector k _L Ms. This configuration is called a surface magnetostatic spin wave (MsSW) or Damon-Eshbach modes. • in a film with a magnetization Ms in the plane and parallel to the direction of the propagation wave vector k^MS- This configuration is called Magnetostatic Backward Volume Wave (MsBVW) configuration,

[0007] Each of these two configurations is associated with a different spin wave dispersion relation (see AA Serga, AV Chumak, and B. Hillebrands, J. Phys. D. Appl. Phys. 43, 264002 (2010) for a description of these relations). A more precise description of the spin waves would be beyond the scope of the invention and can be found in this publication.

[0008] It is known that surface spin waves can be guided in a ferromagnetic waveguide (or spin-wave waveguide) for example in NiFe or YIG films (see Kalinikos, BA; Slavin, AN Theory of Dipole-Exchange Spin Wave Spectrum for Ferromagnetic Films with Mixed Exchange Boundary Conditions. J. Phys. C: Solid State Phys. 1986, 19, 7013-517 7033). It is also known that spin waves are subject to a damping process. Indeed, there is a dissipation of the energy of the spins towards other degrees of freedom of the system such as the crystal lattice by the phonons, and the electrons in ferromagnetic metals. Let us take the example illustrated in figure 1B of a spin wave MsSW, guided in a ferromagnetic film of width L and thickness with a static magnetization Ms along F in the transverse direction of the film, Found excited at x = 0 propagating along the direction Ox of the film.In Figure 1B, three transverse modes n=1, 2 and 3 excited in the waveguide have been represented. Due to the dissipation of the spin energy, the magnetization will relax towards its equilibrium position or along the Oy direction when x becomes large. The attenuation length can be described in terms of relaxation time T. It corresponds to the time necessary for the amplitude of the variation AM(x , t) = Ms-M(x , ?) of the magnetization with respect to the equilibrium Ms to decrease by a factor l / e. The amplitude of AM(x, t) is due to the precession of the magnetization which reduces the component of the magnetization along V by a value AM equivalent to the number of excited magnons. It can be shown that in this Damon-Eshbach configuration: .

[0009] T i = 2a(toH + A2wMkn + wM (1 + P(1-P) sin2(pn) J

[0010] with (¾ = and (¾ = / / nyAf and F the gyromagnetic ratio, Aech being the exchange constant of the material which reflects the intensity of the coupling exchange, and P = 1 - ( 1- ç-kt ) / knt the dipolar matrix element taking into account the thickness '' of the film, 2 + 2 and - arctadkxlkyj^ where ky^ = mr!L is the quantized transverse wave vector taking into account periodic boundary conditions.

[0011] From r, the characteristic attenuation length of spin waves can be defined by (EQ1):

[0012]

[0013] LaU = T.vg with vg — dü) / dk, the group velocity of spin waves.

[0014] The damping of spin waves implies that the information carried by the latter will be lost after it has traveled a distance of the order of a few Latt in the magnonic waveguide. In order to produce a magnonic logic circuit by transporting the information over distances of the order of hundreds of microns, it is therefore necessary to amplify it again in order to be able to transport, receive and process the information.

[0015] It is known from Bracher, Thomas, Philipp Pirro, and Burkard Hillebrands. "Parallel pumping for magnon spintronics: Amplification and manipulation of magnon spin currents on the micron-scale." Physics Reports 699 (2017): 1-34 (hereinafter "Bracher et al"), to realize a parallel parametric amplification (PPA) of guided spin wave in a magnonic waveguide. This PPA phenomenon is a nonlinear phenomenon of amplification of a spin wave that appears when a pump magnetic field at 2 / is in a direction parallel to the direction of the spin wave of frequency f. In order to realize the parallel parametric amplification of a spin wave of mode k, the effective field within the waveguide must be greater than a threshold of magnonic mode and V(k) the coupling parameter that describes the coupling between the pump field and the magnonic mode. The device used in the Bracher paper is illustrated in Figure 2. A NiFe magnonic waveguide WG carries a spin wave at a frequency f. Below this waveguide is deposited a gold CM metal layer, with a constriction (dotted area), in which an RF current i flows at a frequency 2 / . The constriction together with the current flowing in it makes it possible to create an RF Oersted field at a frequency 2 / parallel to the waveguide, making it possible to exceed the power threshold to allow the APP of the spin wave.However, this device is only a proof of concept of the physical phenomenon of spin wave APP and does not allow an implementation of an all-magnon logic circuit because this would require individually supplying current to each metal layer in each area where amplification is to be carried out, which is not realistic for an industrial application.

[0016] The invention aims to overcome some of the aforementioned problems of the prior art by proposing a device for amplifying the magnonic signal by APP in which the amplification of the spin wave can be carried out in several zones with a single RF wave source at 2 / using a resonant RF cavity and a plurality of passive parametric amplification blocks. Summary of the invention

[0017] For this purpose, an object of the invention is a device for amplifying the magnetic signal comprising: • a resonant RF cavity at a frequency 2 / ; • a substrate; • a magnetic waveguide of direction x, deposited on the substrate; • an excitation device suitable for exciting, from a magnetic field RF said to be excitation frequency / , a spin wave at a frequency f propagating within the waveguide; • an RF wave source called an amplification source, the amplification source and the RF cavity being adapted to emit an additional magnetic field at a frequency 2 / , in the direction x; • a plurality of passive parametric amplification blocks, arranged above said waveguide, or arranged between the waveguide and the substrate, or formed by parts of the waveguide, said amplification source and said blocks being adapted so that, locally in a plurality of distinct zones of said waveguide, a power of the additional magnetic field is greater than a parametric amplification threshold, called threshold power, forming an amplification magnetic field thus generating a parallel parametric amplification of the spin wave.

[0018] According to a particular embodiment of the invention, two successive blocks are spaced apart by a distance less than three times an attenuation length Latt of said guided spin wave in said waveguide, and in which a distance between said excitation device and a block is less than three times the attenuation length Latt

[0019] According to a first embodiment of the invention, the amplification source is adapted so that a radiated power of said additional magnetic field within the waveguide and outside said zones is lower than the threshold power, and the blocks are metallic resonators, resonant at the frequency 2 / , arranged above said waveguide, or arranged between the waveguide and the substrate, each resonator being oriented substantially perpendicular to said waveguide, so as to locally amplify said additional magnetic field, the zones being formed by intersections between said blocks and said waveguide.

[0020] According to a second embodiment of the invention, the amplification source is adapted so that a radiated power of said additional magnetic field is greater than the threshold power, said device comprising metal patches deposited above the waveguide or between the waveguide and the substrate, said metal patches being adapted so that an eddy current within the patches generates a local Oersted magnetic field screening said additional magnetic field so that the blocks are parts of the waveguide not covered by the metal patches, forming said zones where the parallel parametric amplification is generated.

[0021] According to a third embodiment of the invention, the amplification source is adapted so that a radiated power of said additional magnetic field is greater than the threshold power, said device comprising dielectric patches deposited above the waveguide, said dielectric patches being adapted to screen said additional magnetic field, so that the blocks are parts of the waveguide not covered by the dielectric patches, forming said zones where the parallel parametric amplification is generated.

[0022] Preferably, in the third embodiment, the patches are made of a dielectric material having a dielectric constant e > 10.

[0023] Preferably, the waveguide is made of NiFe, or CoFeB or YIG.

[0024] According to a particular embodiment of the invention, the RF cavity is a 3D cavity which comprises the substrate, the excitation device, the amplification source, and the plurality of parametric amplification blocks.

[0025] Alternatively, according to another embodiment, the RF cavity is a 1D cavity which is a resonant antenna at frequency 2 / arranged in the vicinity of the waveguide. Preferably, the antenna is deposited on the substrate. Brief description of the drawings

[0026] Other characteristics, details and advantages of the invention will emerge from reading the description given with reference to the appended drawings given by way of example and which represent, respectively

[0027] [Fig.lA], a schematic view of a spin wave propagating along a wave vector k with an angle 9k relative to the static magnetization direction Ms,

[0028] [Fig.lB], a schematic view of a spin wave MsSW, guided in a ferromagnetic film of width L and thickness l, with a static magnetization along 2 in the transverse direction of the film,

[0029] [Fig.2], a schematic view of a prior art para-amplification device parallel metric (APP) of guided spin base in a magnonic waveguide,

[0030] [Fig.3], a schematic view of a magnonic signal amplification device according to the invention,

[0031] [Fig.4], a schematic view of a device for amplifying the magnonic signal according to the first embodiment Ml of the invention,

[0032] [Fig.5], a schematic view of a device for amplifying the magnetic signal according to the second embodiment M2 of the invention,

[0033] [Fig.6], a schematic view of a device for amplifying the magnonic signal according to the third embodiment M3 of the invention,

[0034] References to figures, when identical, correspond to the same elements.

[0035] In the figures, unless otherwise indicated, the elements are not to scale. Description of the embodiments

[0036] Figure 3 schematically illustrates a device 1 for amplifying the magnetic signal according to the invention. The device comprises an RF cavity C resonating at a frequency 2 / and a substrate Sub made of a semiconductor material such as Silicon or an insulator such as a garnet or sapphire.

[0037] A magnonic waveguide WG (Spin-wave Waveguide in English) of direction x is structured on the substrate by conventional techniques, for example by optical, or laser and electronic lithography. The waveguide WG can be made of any ferromagnetic material known to those skilled in the art such as NiFe, or CoFeB or YIG. The waveguide has dimensions and a material both adapted to present a resonant excitation of a spin wave mode at the frequency f.

[0038] In the device 1, the waveguide guides a surface spin wave OS of frequency / with a wave vector k in the direction x excited for example by an excitation RF magnetic field Hrf of frequency / generated by an excitation device SRF.

[0039] The excitation device is for example a resonant metal antenna at the frequency / which makes it possible to radiate the HRP excitation field in the waveguide, thus generating spin waves. Alternatively, the excitation device is a metal layer in which a current i circulates making it possible to excite one or more spin wave modes. The device of the invention allows a “local” generation of surface spin waves OS spin waves, that is to say in a precise zone of the waveguide, and not a “homogeneous” generation in any portion of the waveguide.

[0040] In the embodiment of Figure 3, the excitation field Hrf and the static magnetization Ms of the waveguide (magnetization at its equilibrium position) are along the direction F and perpendicular to the main propagation axis of the k-wave vector spin base (MsSW configuration). This configuration illustrated in [Fig.3] is a non-limiting example and it is understood that the invention could be implemented with a different configuration, for example MsBVW.

[0041] As explained previously, the spin wave OS guided in the WG guide undergoes a damping process. The characteristic attenuation length Latt is determined by the equation EQ1. It is fixed by the material and the geometry of the waveguide, the magnonic mode (i.e. the frequency / ) and the magnetic field HRF. In order to transport the information carried by the spin wave over distances much greater than LatF, the device 1 performs a parallel parametric amplification (PPA) of the spin wave in the WG waveguide. Unlike the prior art, the spin wave is amplified in several zones of the waveguide in order to allow propagation in the waveguide over a significant distance, while minimizing the complexity of the implementation and the overall energy cost required.

[0042] For this purpose, the device comprises an RF wave source called an amplification source Samp. The source, the amplification and the RF cavity are adapted to emit an additional magnetic field Hadd at a frequency 2f, in the direction A, throughout the waveguide. The use of the RF cavity coupled to the amplification source makes it possible to radiate an effective RF field within the relatively homogeneous guide and therefore to easily control the areas where the APP is carried out.

[0043]

[0044] Furthermore, to carry out the amplification in specific areas of the waveguide, the device comprises a plurality of parametric amplification blocks Ampl, Amp2, Amp3. According to different embodiments, the blocks are arranged above said waveguide, or arranged between the waveguide and the substrate, or formed by parts of the waveguide (see below). In the invention, the APP is not carried out homogeneously throughout the waveguide to optimize the relative phase of the Hadd signal at 2 / and the phase of the OS spin Fonde at f. The blocks constitute means of non-homogeneous distribution of the radiated power of the additional magnetic field Hadd in order to exceed, at least locally in the waveguide, the parametric amplification threshold Ps (or "threshold power Ps") of Fonde of spin of wave vector k according to x and of frequency / . As seen above, the threshold Ps can be determined by the equation Ps = More precisely, the amplification source Samp and the blocks are adapted so that, locally in a plurality of distinct zones ZA of the waveguide, the power of the additional magnetic field Hadd is greater than the threshold power Ps. In these zones ZA, the additional magnetic field Hadd then forms an amplification magnetic field Hamp which induces a parallel parametric amplification of Fonde de spin.

[0045] Critically, in the invention, the blocks are passive and do not require an individual RF current supply. They are simply coupled to the additional magnetic field Hadd which is homogeneous in the waveguide thanks to the RF cavity. Thus, the energy cost and complexity required to amplify the spin wave in several ZA zones is very minimal because it is then simply necessary to arrange on the substrate a number of blocks adapted to the distance over which the information must be transported, and this without adding an additional current source or RF source and without increasing the impedance of the device. Thus, the realization of an all-magnon analog processing circuit using the device of the invention is possible. Conversely, in the prior art Bracher et al-, the APP is realized within the waveguide but above a metallic constriction powered by an RF current.In the prior art device, repeating the amplification in several distinct zones would therefore require increasing the impedance of the device by supplying each zone with a source of the same intensity, which would make it impossible to produce a complex all-magnon logic circuit.

[0046] In the description of [Fig. 3], the blocks have been described functionally and in the following, the structure of these blocks will be specified. In summary, the invention proposes three distinct embodiments for the parametric amplification blocks: • a first embodiment M1 illustrated in figure 4 in which the blocks are formed by 2 / resonant metal resonators making it possible to locally amplify the excitation field, • a second mode M2 ​​illustrated in [Fig.5] in which the blocks are parts of the waveguide not covered by metal patches, and • a third mode M3 illustrated in [Fig.6], in which the blocks are parts of the waveguide not covered by dielectric patches.

[0047] Preferably, in order not to lose the information carried by the spin wave OS, two successive blocks are spaced apart by a distance DiD2 ... less than three times the attenuation length Latt. For the same reasons, a distance Do between the excitation device and the first block is preferably less than three times the attenuation length Latt. Even more preferably, Do and D; are approximately equal to Latt.

[0048] According to one embodiment of the invention, the RF cavity is a 3D resonant cavity which comprises the substrate Sub, the excitation device Srf, the amplification source S amp, and the plurality of parametric amplification blocks Amp. By 3D cavity is meant here a resonant “box” at the frequency 2 / . This embodiment is preferred because it allows for a homogeneous distribution of the power of the additional magnetic field Hadd within the cavity and therefore in the waveguide. It is thus easier to precisely control the ZA zones where the APP is carried out.

[0049] Alternatively, the RF cavity is a 1D cavity, i.e. it is a metal antenna arranged in the vicinity of the waveguide, resonant at the frequency 2 / only in the x direction. By "in the vicinity", it is meant here that the antenna is at a distance from the waveguide adapted so that the magnetic field induced by the antenna in the waveguide makes it possible to induce the APP in the ZA zones as seen previously. The RF cavity then has a characteristic dimension c / 2 / (or a multiple of this dimension) in the x direction. This mode has the advantage of being more compact because the dimension of the antenna can be smaller than that of a 3D cavity. However, this mode has the disadvantage of having a less homogeneous distribution of the power of the additional magnetic field Hadd given the very nature of the emission of the RF field of the antenna. In order to make the device even more compact, the antenna can be deposited directly on the substrate Sub.

[0050] Figure 4 schematically illustrates the first embodiment M1 of the invention. In this mode M1, the amplification source is adapted so that the radiated power of the additional magnetic field Hadd within the waveguide and outside the zones ZA is lower than the threshold power Ps. This condition makes it possible to avoid the APP being carried out homogeneously throughout the waveguide. In the device of mode M1, in order to be able to locally exceed the threshold Ps in the zones ZA, the blocks are metallic resonators resonating at the frequency 2 / . These resonators have a quality factor (typically Q>10) arranged above the waveguide, or arranged between the waveguide and the substrate, each resonator being oriented substantially perpendicular to the waveguide.More precisely, the blocks are antennas or cavities of dimension of about c / 2f (or a multiple of this dimension) along 3' which are adapted to locally amplify the additional magnetic field Hadd so that it forms the field H amp in the intersections between the blocks and the waveguide. Thus, in this embodiment, the zones ZA in which the spin-Foundation APP is carried out are formed by the intersections between the blocks and the waveguide. In the waveguide but outside these zones ZA, the power of Hadd is lower than Ps and the spin wave OS at the frequency / propagates normally, without being disturbed by Hadd and by the APP in these zones.

[0051] As previously stated, thanks to the cavity C, the Amp resonators are all coupled with the additional Hadd field. Thus, in the Ml mode, adding a resonator to increase the propagation distance of the spin wave in the waveguide involves a low additional energy cost, does not increase the impedance of the device and does not complicate its implementation.

[0052] This embodiment M1 has the advantage of being easy to design and implement. Indeed, the shape and material of the Amp resonators can be easily designed by simulation as a function of the frequency f and the field H^ in order to ensure that the threshold Ps can be exceeded in the ZA zones. However, the Ml mode does not have a broadband operation because the Amp resonators have a high quality factor Q only for a fixed frequency 2 / . That is to say, for a spin wave of frequency ff and an excitation field at 2 / the amplification induced by the resonators will be strongly reduced and potentially too weak to locally exceed the threshold Ps in the ZA zones.

[0053] In embodiment M1, it is essential that the distance between resonators Amp along the direction x of propagation of the spin wave is less than or comparable to Latt to avoid too much attenuation of the spin wave at the frequency f.

[0054] Figure 5 schematically illustrates the second embodiment M2 of the invention. In this mode M2, the amplification source is adapted so that the radiated power of the additional magnetic field Hadd is greater than the threshold power Ps. Thus, unlike embodiment M1, in embodiment M2, the APP is theoretically possible at any point of the waveguide WG. In order for the APP not to destroy the spatial coherence of the spin Fonde at the frequency / , it is necessary for the APP to occur only in zones ZA of the waveguide. For this, the device comprises metal patches MP deposited above the waveguide or between the waveguide and the substrate. These patches will make it possible to reduce the power of the effective field ^^add (or screen) in the portions of the waveguide covered (or above) the patches.Indeed, the metal patches being crossed by the additional magnetic field Hadd at 2 / , variable over time, they will naturally create eddy currents at 2 / . In the M2 mode device, the geometry, the arrangement with respect to the waveguide and the material of the metal patches are adapted so that the eddy current within the patches generates a local magnetic field of Oersted HF screening the additional magnetic field. More precisely, the arrangement with respect to the waveguide is adapted so that the Oersted HF field is in the z direction so that it applies an additional torque to the spin waves at 2 / generated by the additional magnetic field Hadd- Thus, the phase matching condition between the wave at 2 / and that at / is no longer fulfilled and the APP in the portions of the waveguide covered by the patches is lower than the APP threshold.

[0055] In the M2 mode, the blocks are therefore formed by the parts of the waveguide not covered by the metal patches, these parts being the ZA zones where the parallel parametric amplification of the spin wave at the frequency / is generated.

[0056] The advantage of embodiment M2 over embodiment M1 is that it has a broadband operation because the metal patches are not resonant at a fixed frequency f. By "broadband" we mean here that it allows spin wave APP to be carried out with a wide frequency range, typically 5 GHz. Indeed, the eddy current is generated within the metal patches for a wide frequency range of the additional magnetic field Hadd.

[0057] Figure 6 schematically illustrates the third embodiment M3 of the invention. In mode M3, in a manner identical to mode M2, the amplification source is adapted so that the radiated power of the additional magnetic field is greater than the threshold power Ps. In mode M3, the device comprises dielectric patches DP deposited above the waveguide. These patches will make it possible to reduce the power of the effective field in the portions of the waveguide covered by the patches by simple dielectric impedance contrast effect. The effective field below the patch and in the waveguide will be attenuated and its power will be lower than the threshold power Ps, thus making APP impossible in these portions.

[0058] In the M3 mode, the blocks are therefore formed by the parts of the waveguide not covered by the dielectric patches, these parts being the ZA zones where the parallel parametric amplification of the spin wave at the frequency / is generated.

[0059] Embodiment M3 also exhibits broadband operation because the dielectric patches are not resonant at a fixed frequency. Indeed, they can screen a large frequency range.

[0060] Preferably, the dielectric patches are made of a dielectric material having a dielectric constant s > 10 in order to ensure sufficient screening of the additional field in the portions of the waveguide covered by the patches.

Claims

1.

2.

3. Claims Device (1) for amplifying the magnetic signal comprising: - an RF cavity (C) resonating at a frequency 2 / ; - a substrate (Sub); - a magnetic waveguide (WG) of direction A, deposited on the substrate; - an excitation device (Srf) adapted to excite, from an RF magnetic field called excitation (HRF) of frequency / , a spin wave (OS) at a frequency f propagating within the waveguide; - an RF wave source called an amplification source (Samp), the amplification source and the RF cavity being adapted to emit an additional magnetic field (Hadd) at a frequency 2f, in the direction x; - a plurality of passive parametric amplification blocks (Amp), arranged above said waveguide, or arranged between the waveguide and the substrate, or formed by parts of the waveguide, said amplification source and said blocks being adapted so that, locally in a plurality of distinct zones (ZA) of said waveguide, a power of the additional magnetic field is greater than a parametric amplification threshold, called threshold power (Ps), forming an amplification magnetic field (Hamp) thus generating a parallel parametric amplification (APP) of the spin wave. Device according to claim 1, wherein two successive blocks are spaced apart by a distance (DiD2) less than three times an attenuation length Lat} of said guided spin wave in said waveguide, and wherein a distance (Do) between said excitation device and a block is less than three times the attenuation length L^. Device according to claim 1 or 2, wherein: - said amplification source is adapted so that a radiated power of said additional magnetic field within the waveguide and outside said zones is lower than the threshold power (Ps), and, - the blocks are metallic resonators, resonating at the frequency 2 / , arranged above said waveguide, or arranged between the waveguide and the substrate, each resonator being oriented substantially perpendicular to said waveguide, so as to locally amplify said additional magnetic field, - the zones (ZA) being formed by intersections between said blocks and said waveguide.

4. Device according to claim 1 or 2, wherein said amplification source is adapted so that a radiated power of said additional magnetic field is greater than the threshold power (Ps), said device comprising metal patches deposited above the waveguide or between the waveguide and the substrate, said metal patches (MP) being adapted so that an eddy current within the patches generates a local Oersted magnetic field (HF) screening said additional magnetic field so that the blocks are parts of the waveguide not covered by the metal patches, forming said zones (ZA) where the parallel parametric amplification is generated.

5. Device according to claim 1 or 2, wherein said amplification source is adapted so that a radiated power of said additional magnetic field is greater than the threshold power (Ps), said device comprising dielectric patches (DP) deposited above the waveguide, said dielectric patches being adapted to screen said additional magnetic field, so that the blocks are parts of the waveguide not covered by the dielectric patches, forming said zones (ZA) where the parallel parametric amplification is generated.

6. Device according to the preceding claim, in which the patches are made of a dielectric material having a dielectric constant £ > 10.

7. Device according to any one of the preceding claims, in which the waveguide is made of NiFe, or of CoFeB or even of YIG.

8. A device according to any preceding claim, wherein the RF cavity is a 3D cavity which comprises the substrate (Sub), the excitation device (Srf), the amplification source (Samp), and the plurality of parametric amplification blocks (Amp).

9. Device according to any one of claims 1 to 7, in which the RF cavity is a 1D cavity which is a resonant antenna at frequency 2 / arranged in the vicinity of the waveguide.

10. Device according to the preceding claim, wherein said antenna is deposited on said substrate.