EP2517354A1

EP2517354A1 - Mineral electret-based electromechanical device and method for manufacturing same

Info

Publication number: EP2517354A1
Application number: EP10799024A
Authority: EP
Inventors: Emmanuel Defay; Sébastien Boisseau; Ghislain Despesse
Original assignee: Commissariat a lEnergie Atomique CEA; Commissariat a lEnergie Atomique et aux Energies Alternatives CEA
Current assignee: Commissariat a lEnergie Atomique et aux Energies Alternatives CEA
Priority date: 2009-12-23
Filing date: 2010-12-21
Publication date: 2012-10-31
Also published as: FR2954582A1; US20120273904A1; FR2954582B1; WO2011076806A1; JP2013516099A

Abstract

The invention relates to a mineral electret-based electromechanical device and the method for manufacturing same. Said device includes a dielectric stack comprising at least one electret layer (2E), and two electrodes (16, 20) on two opposite surfaces (18, 22) of the stack. The electret is mineral. The device is of use in particular in the telecommunications field.

Description

ELECTROMECHANICAL DEVICE BASED ON MINERAL ELECTRET, AND METHOD FOR MANUFACTURING THE SAME

DESCRIPTION

TECHNICAL AREA

The present invention relates to an electromechanical device (MEMS or NEMS) such as, for example, an electromechanical actuator or a sensor or an acoustic resonator, in particular a high quality factor, and a method of manufacturing this device.

It applies in particular to the field of telecommunications.

STATE OF THE PRIOR ART

In the field of telecommunications, it is necessary to have very precise time bases, with frequencies ranging from a few hundred megahertz to a few gigahertz.

For this purpose, an oscillator whose oscillation frequency is regulated by a piezoelectric resonator, a resonator whose electromechanical resonance frequency has an impedance variation sufficient to impose the frequency of the oscillator, is used for the most part.

It is recalled that the merit factor of a resonator, for an application to the oscillators, is the product of the quality factor Q by the resonance frequency F. The desired value of the QxF product is of the order of 10 ¹⁴ Hz for the most accurate frequency sources of nomadic systems.

The most commonly used resonators include bulk elements (in English, bulk) made of quartz.

They have the disadvantage of resonating at frequencies F of a few megahertz because they are thick: their thickness is of the order of 100 ym minimum; but they have a very high quality factor Q, of the order of 100000, hence a very good stability of the oscillators at a few megahertz.

To have higher frequencies, one can multiply the resonance frequency of quartz; but then the quality factor decreases so that the product QxF remains approximately constant, of the order of 10 ¹² Hz.

In addition, the multiplication of the frequency requires energy, which is not desirable for nomadic systems.

Finally, an alternative to quartz based resonators is highly desirable, especially for silicon embedded technology.

To avoid the multiplication of the frequency and to work in the latter technology, it is possible to use a resonator which is smaller than a quartz-based structure and whose resonant frequency is already of the order of 1 GHz, namely a FBAR type structure (for Bulk Acoustic Resonator Film), comprising a thin layer piezoelectric which is deposited on a substrate and often based on aluminum nitride.

The quality factor is then only of the order of 1000, as is for example described in the article by R. C. Ruby et al., Thin film bulk acoustic wave resonators for wireless applications,

IEEE Ultrasound. Symp. pp. 813-821, 2001). But the maximum QxF product is of the order of 10 ¹² Hz.

However, this solution requires the use of new materials in the integrated silicon technology, namely materials such as AIN, ZnO, Mo and Pt for example.

Another possibility is to use a resonator of HBAR (High-tone Bulk Acoustic Resonator) type, comprising an acoustic resonant cavity, in other words an acoustic quality substrate, on which an electrode, a piezoelectric thin film and a other electrode.

In such a resonator, the electrical excitation of the piezoelectric thin film by means of the electrodes makes it possible to generate a whole series of harmonics, due to the presence of the acoustic cavity.

With regard to HBAR resonators, reference can be made to the article by M. Pijolat et al.

Large Qxf product for HBAR using Smart Cut ™ transfer of LiNb0 ₃ thin layers onto LiNb0 ₃ substrate, 2008 IEEE International Ultrasonics Symposium, 2008: 201-4 IEEE, Piscataway, NJ, USA, Conference Paper.

The use of a HBAR resonator has the advantages of the two previous solutions (Quartz resonator and FBAR type resonator): the quality factor is very high and the resonance frequencies of the different harmonics persist up to frequencies of the order of several gigahertz.

In addition, the use of this type of resonator makes it possible to achieve QxF products of the order of 10 ¹⁴ Hz (especially in the case where the thin layer is in AlN and the sapphire substrate). However, the presence of very many harmonics, whose frequencies are very close to each other, often induces instability of the oscillator.

It should be noted that the quality factor of a HBAR type resonator results mainly from the presence of the substrate which constitutes the acoustic cavity and must be made of a monocrystalline material with very low acoustic losses, such as sapphire. The existence of resonances at frequencies of the order of 1 GHz is due, for its part, to the presence of the piezoelectric thin film.

Since the latter is deposited on the substrate, it does not exhibit an intrinsically very high quality factor.

But since the acoustic energy is mainly concentrated in the cavity, the low quality factor of the thin layer has little influence on the effective quality factor of the resonator.

STATEMENT OF THE INVENTION

The subject of the present invention is an electromechanical device, in particular a acoustic resonator, which overcomes the previous disadvantages.

The invention relates to an electromechanical device comprising:

a dielectric stack having two opposite faces and comprising at least one electret layer, and

two electrodes respectively carried by the two opposite faces of the stack,

characterized in that the electret is a mineral electret.

The present invention is therefore limited to inorganic electrets (which may be amorphous or crystalline): the polymeric electrets are excluded.

Until now, non-polymeric electrets have never been used to make resonators.

Mineral electrets, such as SiO ₂ , SiN, Al ₂ O ₃ and SrTiO _3, for example, have the advantage of being much more rigid than polymer electrets.

US 2007/063793 discloses acoustic resonators which comprise electrets made of polymeric materials. But such materials can not be suitable for the manufacture of resonators intended to operate at frequencies of the order of a few gigahertz: they are too viscous. On the contrary, electrets based on mineral materials are well suited to such frequencies.

It is specified that the dielectric stack forming part of the device, object of the invention, may comprise a single layer which is then the electret layer, or several dielectric layers, at least one of which is the electret layer.

Note that a dielectric material is an electrostrictive material; on the other hand, the dielectric materials are not necessarily piezoelectric.

In the invention, the electret layer may comprise permanent electrical charges, forming an electromechanical coupling.

It is specified that when the dielectric stack comprises in addition to the electret layer one or more non-piezoelectric layers, the use of charges in the electret layer makes it possible to create an electromechanical coupling that can be used to form an acoustic resonator as well. an electromechanical actuator, or even an electromechanical sensor.

The dielectric stack outside the electret may or may not be piezoelectric. The electret can be, for its part, piezoelectric or non-piezoelectric.

The thickness of the electret layer can be chosen from a few nanometers to a few tens of micrometers; preferably, it is less than or equal to about 1 μm.

The electromechanical device, object of the invention may further comprise a substrate on one side of which are arranged the stack and the electrodes. It is specified that whatever the thickness of the electret layer, the electromechanical device may comprise a substrate.

According to a particular embodiment of the invention, the substrate is provided with a cavity, or hole, which opens at least on the face of the substrate on which one of the two electrodes rests, said electrode being at least partly above the cavity.

Of course, between the substrate and the electrode, there may be an intermediate layer, in particular to improve the adhesion or to isolate the electrode.

It is specified that the use of a substrate having a hole (blind or not) allows the stack, provided with the electrodes, to move perpendicular to the plane of the layers.

This embodiment is particularly interesting for the production of electromechanical actuator type structures or electromechanical sensor or acoustic resonator such as, for example, a FBAR (for Bulk Acoustic Resonator Film).

When the substrate of the device according to the invention does not have a hole, the corresponding device is particularly interesting for the production of acoustic resonators and in particular resonators HBAR type (for High overtone Bulk Acoustic Resonator).

According to another particular embodiment of the invention, the electromechanical device further comprises a Bragg grating acoustic having two opposite faces, one of which rests on one side of the substrate and the other carries one of the two electrodes.

It is specified that the use of a substrate associated with a Bragg grating under the stack and the electrodes makes it possible in particular to produce an acoustic resonator.

In the invention, the mineral electret layer may be crystalline or, on the contrary, amorphous.

The present invention also relates to a method of manufacturing an electromechanical device, comprising:

the formation of a dielectric stack having two opposite faces, said stack comprising at least one layer of dielectric material,

permanently electrically charging said layer of dielectric material to form an electret layer, and

the formation of first and second electrodes respectively on these two opposite faces, characterized in that the dielectric material is a mineral dielectric material.

In this method, the thickness of the electret layer may be chosen from a few nanometers to a few tens of micrometers; preferably, it is less than or equal to about 1 μm.

The dielectric stack and the electrodes can be made above a substrate, either directly or indirectly (for example above an intermediate layer that can serve as an etch stop layer).

According to a first particular embodiment of this method, from a substrate provided with a sacrificial layer and from the first electrode which rests above the sacrificial layer,

forming on the first electrode, the dielectric stack comprising at least the layer of dielectric material,

the permanent electrical charging of the layer of dielectric material is carried out,

the second electrode is formed on said stack, and

at least part of the sacrificial layer is removed to form a hole, or cavity, under the first electrode.

According to a second particular embodiment:

at a face of the substrate, an etching stop layer is formed, and then the first electrode which rests on this face of the substrate, above the stop layer when it exists,

the dielectric stack comprising at least the layer of dielectric material is formed on the first electrode,

the second electrode is formed on said stack, and the substrate is etched from the face opposite the first face to the stop layer when it exists, so as to make a hole, or cavity, under the first electrode.

According to a third particular embodiment:

an acoustic Bragg grating is formed on a substrate, at one face of this substrate,

the first electrode and the dielectric stack comprising at least the layer of dielectric material are then formed on this network, the permanent electrical charging of the layer of dielectric material is carried out, and

the second electrode is formed on said stack.

In the process, the permanent electrical charging can be carried out by a method selected from ion implantation and / or electronic implantation and / or corona discharge and / or the wet electrode method.

In this method of the wet electrode, the electric charging is done by contact with a liquid, before the formation of the second electrode.

We will return to this method at the end of the present description.

It is specified that the order of the steps of the method, object of the invention, can be modified. In particular, the charging of the electret can be performed after the formation of the second electrode.

BRIEF DESCRIPTION OF THE DRAWINGS

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

FIGS. 1 to 3 diagrammatically illustrate a first particular embodiment of the method, object of the invention,

FIGS. 4 to 7A schematically illustrate a second particular embodiment of the method, object of the invention,

FIG. 7B schematically illustrates an alternative embodiment of the method illustrated in FIG. 7A,

FIGS. 8 to 10 schematically illustrate a third particular embodiment of the method, object of the invention,

FIGS. 11 to 13 schematically illustrate a fourth particular embodiment of the method, object of the invention,

FIGS. 14 to 17 schematically illustrate a fifth particular embodiment of the method, object of the invention, and

Figures 18, 19 and 20 illustrate schematically the principle of the method of the wet electrode, which is usable in the invention.

DETAILED DESCRIPTION OF PARTICULAR EMBODIMENTS With reference to the "Smart Cut ™" technique, reference may be made, for example, to the article by M. Bruel, Application of hydrogen ions to Silicon on Insulator material technology, Nuclear Instruments and Methods in Physics Research, Section B Beam Interactions with Materials and Atoms, February 1996, B 108 (3): 313-19, or patent FR 2,681,472.

This method is also described for example in the article by B. Aspar et al., "Silicon Wafer Bonding Technology for VLSI and MEMS applications", edited by SS Iyer and AJ Auberton-Herve, 2002, INSPEC, London, Chapter 3, pages 35-52.

It is also mentioned in the following the implementation of a Corona discharge.

With regard to the Corona discharge, reference may be made for example to the article by J.A. Giacometti et al., Corona Charging of Polymers, IEEE Transactions on Electrical Insulation, vol. 27, No. 5, October 1992.

According to one aspect of the invention, a non-piezoelectric dielectric material with very low acoustic losses, for example sapphire or STO (SrTiOs), is used and a permanent electrical charge is made of this material in order to transform it into electret.

As we have seen, this loading can be done by ion implantation, by implantation electronic or by the technique known as Corona discharge.

When the permanently charged material is between two electrodes, an electro-acoustic resonance occurs when an electric field is imposed on the electrodes.

In other words, the material becomes piezoelectric induced.

Various examples of the method, object of the invention, are described below to obtain a resonator with very low acoustic losses.

The first example concerns the manufacture of a solid sapphire resonator.

In this example, a thick sapphire layer 2 having a thickness of at least 100 μm is used, for example 500 μm (FIG. 1).

The permanent electrical charge of the layer 2 is then carried out by ion implantation, by electronic implantation or by corona discharge, to transform it into a 2E electret layer.

Figure 2 illustrates this step by taking the example of the Corona discharge.

To implement this technique, the layer 2 rests on an electrode 4 which is grounded.

A tip-shaped electrode 6 is placed above the layer 2.

A gate 8 is disposed between the electrode 6 and the layer 2. The discharge is obtained by bringing the electrode 6 to a high positive potential, for example 10 kV, by means of a suitable voltage source 10.

As seen in FIG. 2, the current of the discharge is measured by an ammeter 12 and controlled by adjusting the potential of the gate 8 by means of a suitable voltage source 14.

An electrode 16 is then formed on one face 18 of the layer 2E and another electrode 20 on the opposite face 22 of the layer 2E. To form these electrodes, it is possible to deposit a metal, for example aluminum, on these faces 18 and 22 (FIG. 3).

In FIG. 3, the reference 23 designates the charged zone (whose thickness is non-zero).

The second example concerns the manufacture of a monocrystalline thin-film resonator using a dielectric, non-piezoelectric material.

This resonator does not require the application of a DC voltage between its electrodes, during its operation.

In this example, a monocrystalline sapphire thin film 24 is used (FIG. 4). The thickness of this layer is for example 1 μm.

Furthermore, a substrate 26 is formed, for example made of silicon, provided with an electrode 28 and a sacrificial layer 30 at a face 32 of the substrate.

The electrode 28 rests on this face, above the sacrificial layer 30. The electrode 28 is for example obtained by evaporation deposition of a metal, for example aluminum, above the layer 30.

The thin layer 24 is then transferred to the electrode 28. To do this, one can use the "Smart Cut" (TM) technique which will be discussed at the end of the present description.

Alternatively, a thick monocrystalline sapphire layer is used; it is bonded to the electrode 28, for example by molecular bonding; then the thick layer is thinned, for example by grinding, until the desired thickness for the thin layer is obtained.

Then, the permanent electric charge of the layer 24 is carried out by ionic or electrical implantation or by corona discharge (FIG. 5).

The electrical charges sent into the layer 24 bear the reference 34 in FIG.

Then another electrode 36 is formed on the electret layer 24E (FIG. 6) which results from the charging of the layer 24.

To do this, for example, a layer of aluminum is deposited on this layer 24E; then we proceed to the structuring of the aluminum layer, by photolithography and etching.

Then, the sacrificial layer 30 is removed, for example by chemical etching (FIG. 7A), hence the formation of a cavity 37 under the resonator.

Thus, one "releases" the formed resonator: it is then surrounded by air. The release of the membrane constituted by the resonator can also be obtained by an etching of the silicon substrate 26, by the rear face of the latter. It is then a deep etching in this substrate.

An intermediate layer may be added between the lower electrode 28 and the substrate 26. This layer may be used as an etch stop layer.

This is schematically illustrated by FIG. 7B in which the etch stop layer 37a, formed between the substrate 26 and the electrode 28, and a through hole 37b, formed through the substrate, under the coating layer are seen. etching stop 37a.

It is specified that the etching stop layer is not essential; its use depends on the etching process and the materials used.

In a third example, a resonator having a HBAR type structure is manufactured. As before, a thin sapphire layer can be used. And if the substrate itself is sapphire, we obtain a structure "all sapphire".

The substrate 38 is provided with an electrode 40, for example made of aluminum, which rests on a face 42 of the substrate (FIG. 8).

As previously, the sapphire thin film 44 is formed on the electrode 40; then (FIG. 9) the permanent electrical charging of the layer 44 (the arrows 46 symbolizing this charging) is carried out, to transform the layer 44 into an electret layer 44E; and forming another electrode 48 on the layer 44E (FIG. 10).

For example, if the layer 44 has a thickness of 1 μm and the electrodes 40 and 48 have a thickness of 100 nm respectively, a substrate 38 with a thickness of 50 μm can be used.

In a fourth example, a resonator is made with a thin layer of STO (SrTiOs) on an acoustic Bragg grating.

In this case, this network is used to confine the acoustic energy of the resonator which consists of the SrTiO3 layer and two electrodes.

With regard to such a network, reference is made, for example, to the article by K. M. Lakin et al.

Development of miniature filters for wireless applications, IEEE Trans. Microwave Theory Tech., Vol. 43, No. 12, pp. 2933-2939, 1995.

The acoustic Bragg grating 50 (FIG. 11), for example in W / SiO 2 / W / SiO 2, is formed on a substrate 52, for example made of silicon. Then an electrode 53, for example platinum, is formed on the acoustic Bragg grating 50.

We will return to such a network at the end of the present description.

Then, the thin film 54 of STO is formed on the electrode 53. As before, the layer 54 can be postponed or refined.

Then the permanent electrical charging of the layer 54 is carried out to transform it into a layer electret, as shown in Figure 12 where the arrows 56 symbolize this loading.

Then an electrode 58, for example platinum, is formed on the electret layer 54E by deposition and then structuring (FIG. 13).

The method, which is the subject of the invention, can also be implemented with dielectric, non-piezoelectric materials of lower acoustic quality than sapphire or STO, for example materials deposited by any possible technique (in particular sputtering, evaporation, CVD, MOCVD, ALD, MBE), in particular the following materials: S1O _2, Si _x N _y, Al2O3, Hf0 _2, Y2O3, Zr0 _2, Ti0 _2, SrTi0 ₃ deposited (Ba, Sr) Ti0 _3.

Such a method is particularly interesting for the integrated silicon technology.

Indeed, it is a method that can be fully compatible with CMOS structures and that uses only materials well known in the field of CMOS.

This is illustrated by way of example, with reference to Figs. 14-17.

This example uses amorphous S1O ₂ , which is deposited by CVD.

The method implemented in this example is particularly advantageous in the case of a microelectronic integration.

Indeed, as we will see, it requires only materials that are fully compatible with CMOS technology: Si, S1O ₂ and Al or Cu. The steps of this example correspond to the steps of the second example which has been given above with reference to FIGS. 4 to 7 (the latter respectively corresponding to FIGS. 14 to 17), except that the thin sapphire layer is replaced by a thin layer of Si0 ₂ .

This layer is loaded by ionic or electronic implantation or Corona discharge.

Then, the upper electrode is deposited and structured.

More specifically, a structure is formed comprising a substrate 60, for example made of silicon, a sacrificial layer 62, an electrode 64, for example made of aluminum or copper, on the substrate 60, above the sacrificial layer; and forming the thin layer 66 in S1O ₂ on the electrode 64 (Figure 14).

Then the layer 66 is electrically charged permanently to form the electret layer (FIG. 15).

The loading is symbolized by the arrows

68.

Next, an electrode 70, for example aluminum or copper, is formed on the electret layer 66E (FIG. 16); then removing the sacrificial layer 62 (hence the formation of a cavity 72) to release the resonator which is thus surrounded by air (Figure 17).

In the foregoing, therefore, a variant of the second example has been described. But one can also implement comparable variants for the third and fourth examples. And, in the first example, one could also replace the thick layer of sapphire with a thick layer of a dielectric material, non-piezoelectric, amorphous, for example a thick layer of amorphous S102.

We will come back to the Corona discharge. The Corona discharge is generally used in photocopiers, for the production of ozone or to improve the wettability of certain materials.

In the case of the present invention, its purpose is to inject charges in a dielectric material which is capable of keeping them for a long time (typically several years): it is an electret. This results in the appearance of a surface potential and the creation of an electric field within the material. It is an electric dipole, just as the permanent magnet is a magnetic dipole.

In order to control the value of the permanent electric field, it is generally easier to control the surface potential of the sample. Indeed, thanks to the triode Corona system (tip 6 / gate 8 / electrode 4 - see FIG. 2), the surface potential of the sample 2 (V _s ) takes the value of the potential imposed on the gate (V _g ) and so :

V V

E = - ^ = ^

dd where d is the thickness of the sample and E is the electric field.

The quantity of charges that are present in the sample material depends on the thickness of this material (d), its dielectric constant (ε) and the capacity of the material to preferentially keep the charges on its surface, at depth. , or, for multilayer systems, at the interfaces.

It is thus estimated that if a material retains its charges (Q) at surface (S), then the charge density (σ) is: σ = - Q = - ^εε- o - (ε ₀ : permittivity of vacuum) .

In the case of deep storage, it is more difficult to determine the charge density (p) in the material.

A typical order of magnitude is a load of 2 mC / m ² which corresponds to a surface potential of 200 V over 500 nm of Si0 ₂ .

The charging can be carried out under standard conditions of temperature and pressure (20 ° C to 10 ⁵ Pa); nevertheless, it is not excluded to achieve it under other conditions and in particular at higher temperatures or lower pressures and vice versa. Nor is it possible to heat the sample while it is being loaded, which generally has the effect of increasing the penetration depth of the charges in the material as well as the stability. Charging is usually done in ambient air (0 ₂ : 20%, 2: 80%). Nevertheless, it is not excluded to change these ratios, or even to change gas.

The voltage of the tip (V _p ) is of the order of magnitude of a few kilovolts. The voltage of the gate (V _g ) can vary between 0 V and 500 V. The two values can be positive (Corona +) or negative (Corona-). These voltages can, for example, be obtained using DC / HT converters.

Generally, the space between the tip and the grid, as well as the space between the grid and the sample, is of the order of 1 cm. The holes in the grid have a millimeter size.

For charging, another technique can be used, namely the wet electrode method (also called the liquid electrode method), instead of the corona discharge.

With respect to this method, reference may be made to the article by K. Ikezaki et al., Thermally Stimulated Currents from Ion-Injected Teflon-FEP Film Electrets, Jpn. J. Appl. Phys. 20 (1981) pp. 1741-1747.

The principle of this method is schematically illustrated by Figures 18, 19 and 20.

In these figures, the reference 72 designates a platinum upper electrode, the reference 74 a cotton pad, the reference 76 an aqueous solution of electrolytes, the reference 78 a layer constituting a sample, and the reference 80 a lower electrode which carries the sample and which is grounded. The upper electrode is negatively charged (by appropriate means, not shown) and surrounded by the buffer 74. And the upper electrode is on the sample.

In a first step (FIG. 18), the upper electrode of the solution is approached. The latter and the sample are then loaded positively by influence.

In a second step (FIG. 19), the solution is absorbed by the buffer and the positive charges remain on the sample. The upper electrode provided with the impregnated buffer is then removed from the solution.

Once the upper electrode is retracted, there is therefore a positively charged sample on the electrode 80 (Figure 20).

We come back now to the Smart process

Cut ™.

In order to produce a thin layer of monocrystalline material, it is advantageous to use two types of existing thin-film transfer techniques, making it possible to preserve the monocrystalline character: the Smart Cut ™ technology, based on the implantation of gaseous ions ( typically hydrogen ions), and the technique of gluing / thinning.

These techniques are unique techniques that allow a monocrystalline layer to be postponed onto a host substrate. These techniques are perfectly controlled on silicon and allow, among other things, the industrial manufacture of SOI wafers (for Silicon On Insulator). These two techniques are differentiated by the range of material thicknesses that we seek to report, the Smart Cut ™ process to achieve very small thicknesses that can typically be less than about 0.5 ym.

The Smart Cut ™ process (see the article by M. Bruel, Silicon on insulator material technology, Electronic Letters, 31 (14), p.1201-1202, 1995), allows the production of SOI substrates, comprising silicon on a insulating.

Smart Cut ™ technology can be schematically summarized by the following four essential steps:

Step 1: Hydrogen implantation is performed on an oxidized Si substrate A. The oxide layer is then the future buried insulation film of the SOI structure. This implantation step generates the formation of a zone weakened in depth, which consists of microcavities whose growth is at the base of the phenomenon of detachment.

Step 2: The molecular bonding bonding makes it possible to secure the implanted plate A on the support plate (against plate or base) B which is not necessarily oxidized. A surface preparation is necessary to obtain a good quality bonding.

Step 3: The fracture step is performed at the weakened zone by means of a heat treatment in the range of 400 ° C-600 ° C. On the one hand, the SOI structure is obtained, and on the other hand the initially implanted substrate A, peeled from the layer transferred. The latter can then be recycled for the realization of another transfer.

Step 4: Final treatments consist, on the one hand, of a high temperature annealing to consolidate the bonding interface between the transferred thin film and the support substrate and, on the other hand, of a polishing which allows to obtain the final thickness of silicon surface film desired, as well as a good surface finish.

The thickness of the transferred layer is directly related to the ion beam implantation energy, and thus makes it possible to obtain good flexibility in the thickness combinations (thin film and buried oxide). For example, the thickness of transferred silicon may range from a few tens of nanometers to about 2 μm using a conventional implanter (energy less than 210 keV).

The transferred layers are uniform and homogeneous in thickness because they are defined by a depth of implantation and not by a mechanical thinning.

Manufacturing costs are reduced, on the one hand, by the recycling of the substrates (the initially implanted plates can be reused after transfer of the thin film), and on the other hand by the use of standard equipment in microelectronics.

It is a flexible process that allows for example the realization of heterostructures. For example, Smart Cut ™ technology allows you to couple the advantages of a solid Si support substrate (in particular cost, weight and mechanical characteristics) and of an active thin layer. It is thus possible to transfer layers of different materials such as:

- Sic - see L. DiCioccio et al.,

"Silicon carbide on insulator training by Smart Cut ™ process", Master, Sci. Eng. flight. B46, pp. 349-356 (1997);

GaAs - see E. Jalaguier et al., "Transfer on thin GaAs film on silicon substrate by proton implantation process", Electronic Letters, vol. 34, No. 4, pp. 408-409 (1998);

InP - see E. Jalaguier et al., "IEEE Proc.llth International Conference on Indium Phosphide and Related Materials, Davos, Switzerland (1999);

GaN - see A. Tauzin et al., "Transfers of 2-inch GaN films onto substrates using Smart CutTM technology," Electronics Letters 2005, vol. 41, No. 11;

- or the Ga - see C. Deguet et al. - "200 mm Germanium-On-Insulator (GeOI) structures realized from epitaxial Germanium wafers by the Smart Cut ™ technology", Electro Chemical Society 2005.

These transfers can be carried out on different substrates, in particular quartz, Si, Ge, GaAs and sapphire.

We are now back to Bragg's mirror. A solution for isolating the bulk acoustic resonator BAW (Bulk Acoustic Wave) of the substrate is based on a principle widely used in optics: the Bragg mirror.

Its acoustic transposition consists of producing a stack under the resonator in which quarter-wave layers of low acoustic impedance materials alternate with quarter-wave layers of high acoustic impedance materials. In this configuration, the resonators are also called SMR (for Solidly Mounted Resonator).

This idea, proposed as early as 1965 - see the article by WE Newell which, like the other documents cited later, is mentioned at the end of the present description - for quartz resonators, was taken up in the BAW SMR resonators made by KM Lakin et al. (1995).

In the case of the Bragg mirror, the reflection coefficient depends on the materials and the number of layers used and is not constant over the entire frequency band. We will therefore specify the key parameters and the characteristics of the response of a Bragg mirror.

It is possible to calculate the reflection coefficient that a Bragg mirror exhibits for a longitudinal wave using a transmission line model - see the article by KM Lakin (1991). This model makes it possible to represent the acoustic impedance Z _n presented by a layer as a function of the acoustic impedance of the lower layer Z _n _i by the expression: Ζ _η _ _γ . cos (9 _mat ) + iZ _mat . sin (9 _matte )

zma, · cos (0 _mat ) + ί.Ζ _η _ _γ . sin (0 _mat )

where ^u mat = = ^ma ' ^{is the} pulsation and Z _mat , e _mat and V _mat are respectively the acoustic impedance, the thickness and the speed of the longitudinal wave of the layer.

From this expression, it is possible to determine the impedance Z _Br agg presented by the Bragg mirror at the interface between the lower electrode and the Bragg mirror. The reflection coefficient R for the longitudinal wave is written:

elec Bragg where Z _e iec represents the acoustic impedance of the lower electrode.

The reflection coefficient presented by the Bragg mirror is a function of the number of layers. The pair of S1O ₂ / W materials is commonly used because it allows, from four layers, to fulfill the sound insulation function.

The number of layers required increases when materials with a lower acoustic impedance ratio are used. Thus, in the case of the couple SÎ0 ₂ / A1, which was one of the first to be exploited, two layers are needed to to reach a sufficient reflection - see the MA Dubois article.

The ratio of acoustic impedances also defines the width of the reflection bandwidth of the Bragg mirror. The higher the ratio of acoustic impedances, the greater the range of frequencies over which the Bragg mirror has a good reflection. Thus, this reflection band for a six-layer mirror reaches 1.5 GHz for the torque S1O ₂ / AIN and 2.8 GHz for the torque

The S1O ₂ / W couple has the advantage of using few layers and presenting a very wide reflection range. On the other hand, its integration into BAW filters requires burning the tungsten outside the active zones in order to avoid parasitic capacitive couplings.

The articles mentioned above are:

W. E. Newell, Face-mounted piezoelectric resonators, Proc. of IEEE, pp. 575-581, 1965,

K. M. Lakin et al., Development of miniature filters for wireless applications, IEEE Trans. Microwave Theory. Tech., Vol. 43, No. 12, pp. 2933-2939, 1995

K.M. Lakin, Fundamental Properties of Thin Film Resonators, IEEE Freq. Against. Symp., Pp.201-206, 1991

MA Dubois, aluminum nitride and zirconate-titanate lead thin films for ultrasonics applications: integration, properties and devices, Thesis of EPFL, 1999.

Claims

An electromechanical device comprising:

a dielectric stack (2E, 24E, 44E, 54E, 66E) having two opposite faces and comprising at least one electret layer, and

two electrodes (16, 20, 28, 36, 40, 48, 53, 58, 64, 70) respectively carried by the two opposite faces of the stack,

characterized in that the electret is a mineral electret.

2. Electromechanical device according to claim 1, wherein the electret layer comprises permanent electrical charges, forming an electromechanical coupling.

3. Electromechanical device according to any one of claims 1 and 2, wherein the dielectric stack outside the electret is non-piezoelectric.

4. Electromechanical device according to any one of claims 1 to 3, wherein the thickness of the electret layer (2E) is selected from a few nanometers to a few tens of micrometers, and it is preferably less than or equal to about 1 ym.

5. Electromechanical device according to any one of claims 1 to 4, comprising in in addition to a substrate on one side of which are arranged the stack and the electrodes.

6. Electromechanical device according to claim 5, wherein the substrate (26, 60) is provided with a cavity (37, 72), or hole, which opens at least on the face of the substrate on which one (28) rests. 64) of the two electrodes, said electrode being at least partly above the cavity.

7. Electromechanical device according to claim 5, further comprising an acoustic Bragg grating (50) having two opposite faces, one of which rests on one side of the substrate and the other carries one (53) of the two electrodes.

An electromechanical device according to any one of claims 1 to 7, wherein the electret layer (2E, 24E, 44E, 54E) is crystalline.

An electromechanical device according to any one of claims 1 to 7, wherein the electret layer (66E) is amorphous.

A method of manufacturing an electromechanical device, comprising:

the formation of a dielectric stack (2, 24, 44, 54, 66), said stack comprising at least one layer of dielectric material, permanently electrically charging said layer of dielectric material to form an electret layer (2E, 24E, 44E, 54E, 66E), and

forming first and second electrodes (16,20; 28,36; 40,48; 53,58; 64,70) respectively on these two opposite faces;

characterized in that the dielectric material is a mineral dielectric material.

11. The method of claim 10, wherein the thickness of the electret layer (2E) is selected from a few nanometers to a few tens of micrometers and is preferably less than or equal to about 1 ym.

The method of any of claims 10 and 11, wherein the dielectric stack and the electrodes are formed over a substrate.

The method of claim 12, wherein, from a substrate (26, 60) having a sacrificial layer (30, 62) and the first electrode which rests above the sacrificial layer,

forming on the first electrode, the dielectric stack comprising at least the layer (24, 66) of dielectric material,

the permanent electrical charging of the layer (24, 66) of dielectric material is carried out, the second electrode (36, 70) is formed on said stack, and

at least part of the sacrificial layer (30, 62) is removed to form a hole, or cavity, under the first electrode.

The method of claim 12, wherein

forming, at a face of the substrate, optionally an etch stop layer and then the first electrode which rests on this face of the substrate, above the stop layer when it exists,

the permanent electrical charging of the layer (24, 66) of dielectric material is carried out,

the second electrode (36, 70) is formed on said stack, and

the substrate is etched from the face opposite the first face to the stop layer when it exists, so as to make a hole, or cavity, under the first electrode.

The method of claim 12, wherein

an acoustic Bragg grating (50) is formed on a substrate (52) at one side of this substrate, the first electrode and the dielectric stack comprising at least the layer of dielectric material are then formed on this network, the permanent electrical charging of the dielectric material layer (24, 66) is carried out, and

the second electrode (36, 70) is formed on said stack.

The method according to any one of claims 10 to 15, wherein the permanent electrical charging is carried out by a method chosen from ion implantation and / or electronic implantation and / or Corona discharge and / or the method of the wet electrode.