EP4271973A1

EP4271973A1 - Method for manufacturing a detection device comprising an encapsulating structure comprising an opaque thin layer resting on a peripheral mineral wall

Info

Publication number: EP4271973A1
Application number: EP21844791.0A
Authority: EP
Inventors: Jean-Jacques Yon; Geoffroy Dumont; Thomas PERRILLAT-BOTTONET
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: 2021-01-04
Filing date: 2021-12-30
Publication date: 2023-11-08
Also published as: WO2022144427A1; IL303960A; FR3118663A1; CA3203028A1; KR20230128301A; CN116829914A

Abstract

The invention relates to a method for manufacturing a detection device (1), comprising the steps of: o producing thermal detectors distributed in a detection array (2p) and a compensation array (2s) by means of sacrificial mineral layers (41, 42); o producing an encapsulating structure (30s) comprising an opaque thin layer (33) extending above the compensation array (2s); o partially removing the sacrificial mineral layers (41, 42) by chemical etching, so as to release the detection array (2p) and the compensation array (2s), and to obtain a peripheral wall (31s) then formed of a non-etched portion of the sacrificial mineral layers (41, 42) and surrounding the compensation array (2s), the opaque thin layer (33) then being suspended above the compensation array (2s) and resting on the peripheral wall.

Description

METHOD FOR MANUFACTURING A DETECTION DEVICE COMPRISING AN ENCAPSULATION STRUCTURE COMPRISING A THIN OPAQUE LAYER RESTING ON A MINERAL PERIPHERAL WALL

TECHNICAL AREA

[001] The field of the invention is that of devices for detecting electromagnetic radiation, in particular infrared or terahertz, comprising an encapsulation structure in which is located a matrix of thermal compensation detectors, the encapsulation structure comprising a thin upper layer opaque to the radiation to be detected. The invention applies in particular to the field of infrared or terahertz imaging, thermography, or even gas detection.

PRIOR ART

[002] A device for detecting electromagnetic radiation may comprise a matrix of sensitive pixels each containing a thermal detector. Thermal detectors are made from a reading substrate containing a reading and control integrated circuit (ROIC). The thermal detectors can be of the type with an absorbent membrane thermally insulated from the reading substrate. The absorbent membrane comprises an absorber of the electromagnetic radiation to be detected associated with a thermometric transducer whose electrical property varies in intensity as a function of its heating.

[003] The temperature of the thermometric transducer being however greatly dependent on its environment, the absorbent membrane is usually thermally insulated from the substrate and from the reading circuit, which is placed in the substrate. Thus, the absorbent membrane is generally suspended above the substrate by anchoring pillars, and is thermally insulated therefrom by holding and thermal insulation arms. These anchoring pillars and insulation arms also have an electrical function by ensuring the electrical connection of the suspended membrane to the reading circuit arranged in the substrate.

[004] However, when reading an electrical signal from the thermal detector during the absorption of electromagnetic radiation, the useful part of the measured electrical signal, which is associated with heating of the thermometric transducer (induced by the absorption of to be detected), remains low compared to the intensity of the electrical signal measured. Also, the detection device usually comprises detectors so-called compensation thermal signals intended to measure the non-useful part of the electrical signal, also called common mode, associated with the environment of the thermal detector, which is then subtracted from the response signal to deduce the useful part therefrom.

[005] The detection device, in particular when it operates in 'rolling shutter' mode, can then comprise a matrix of sensitive pixels, and a matrix of compensation pixels usually smaller than the matrix of sensitive pixels, and an integrator CTIA type placed at the foot of each column of pixels. In operation, the matrix of sensitive pixels is read line by line. The integrator receives the response signal k from the thermal detector and subtracts from it the common mode electrical signal l _c measured by the corresponding compensation detector. Thus, the non-useful part contained in the response signal k is compensated by the common mode l _c . The useful part ld-lc associated with the absorption of the electromagnetic radiation to be detected is thus obtained, without it being necessary to specifically regulate the temperature of the substrate.

[006] The document WO2012/056124A1 describes an example of a detection device comprising a matrix of sensitive pixels and a matrix of compensation pixels. Each pixel comprises an absorbent membrane thermal detector suspended above the reading substrate. The thermal detectors are made on and through a first sacrificial layer of polyimide, and are covered by a second sacrificial layer. A compensation structure forms a cavity in which the array of compensation pixels is located. It comprises a thin opaque layer making it possible to screen the compensation pixels, that is to say not to transmit the electromagnetic radiation to be detected. For this, the thin opaque layer is produced by conformal deposition so as to continuously cover the upper face and the sides of the two sacrificial layers. Also, the thin opaque layer comprises an upper wall extending above the compensation pixels and a peripheral wall, which rests on the reading substrate, extending around the latter.

[007] However, there is a need to have a method for manufacturing such a detection device that makes it possible to improve the mechanical strength of the encapsulation structure of the compensation matrix, without degrading the quality of the screening. optics, and without complicating the production steps.

[008] Document US2009/0146059A1 is also known, which describes a detection device comprising thermal detectors and compensation elements. However, the latter are structurally different from the thermal detectors which ensure the detection of the electromagnetic radiation of interest, which can affect the quality of the compensation carried out. DISCLOSURE OF THE INVENTION

[009] The object of the invention is to remedy, at least in part, the drawbacks of the prior art. For this, the object of the invention is a method of manufacturing a device for detecting electromagnetic radiation, comprising the following steps: o production, on and through a first sacrificial layer resting on a substrate of reading, of a detection matrix formed of thermal detectors intended to detect the electromagnetic radiation, and of at least one so-called compensation matrix formed of thermal detectors intended not to detect the electromagnetic radiation, a second sacrificial layer covering the thermal detectors and the first sacrificial layer; the thermal detectors (20s) of the compensation matrix (2s) being adapted to detect electromagnetic radiation and structurally identical to the thermal detectors (20p) of the detection matrix (2p); o production of a so-called secondary encapsulation structure delimiting a secondary cavity in which the compensation matrix is located, and comprising a peripheral wall as well as an opaque upper wall resting on the peripheral wall and formed of at least one thin layer opaque.

According to the invention, the first and second sacrificial layers are made of an inorganic material. Furthermore, the step of producing the encapsulation structure comprises the following steps: producing the thin opaque layer so that it extends in a continuously planar manner only over an upper surface of the second mineral sacrificial layer; o production, in the thin opaque layer, of vents arranged facing the compensation matrix; o partial removal of the first and second mineral sacrificial layers through the vents, by chemical etching, so as to release the detection matrix and the compensation matrix, and to obtain the peripheral wall then formed of a non-etched portion of the sacrificial layers minerals and surrounding the compensation matrix, the thin opaque layer then being suspended above the compensation matrix and resting on the peripheral wall.

[0011] Certain preferred, but non-limiting aspects of this method are as follows.

[0012] The first and second sacrificial layers can be made of the same mineral material based on an oxide or a silicon nitride. [0013] The thermal detectors of the detection matrix, like the thermal detectors of the compensation matrix, can each comprise an absorbent membrane capable of absorbing the electromagnetic radiation to be detected and can comprise a thermometric transducer, suspended above the reading substrate by anchor pillars and support and thermal insulation arms.

[0014] The thermal detectors of the detection matrix and/or the thermal detectors of the compensation matrix can each comprise a reflective layer, which rests on the reading substrate, below each absorbent membrane.

[0015] The opaque upper wall may comprise an interference stack that absorbs the electromagnetic radiation to be detected

[0016] The thin opaque layer can be a layer that reflects or absorbs the electromagnetic radiation to be detected.

[0017] The thin opaque layer may have a uniform thickness.

[0018] The upper opaque wall may further comprise at least one thin reinforcing layer covering the thin opaque layer, and may have a border projecting vis-à-vis the peripheral wall in a plane parallel to the reading substrate, the projecting border comprising the thin opaque layer and/or the thin reinforcing layer.

The secondary cavity may have a length and a width in a plane parallel to the reading substrate, the width being less than or equal to 200 μm. The width is less than the length.

[0020] The opaque top wall may not include reinforcing pillars, made in one piece and of the same material with a thin layer of the opaque top wall, located in the secondary cavity and coming to rest on the reading substrate .

[0021] The mineral sacrificial layers can be made of a material that absorbs the electromagnetic radiation to be detected.

[0022] The thin opaque layer can be made of a material with a getter effect.

[0023] The manufacturing process may comprise the following steps: o before the partial elimination step, production of the opaque upper wall formed of a stack comprising a thin protective layer made of amorphous carbon inert to an etchant used during the partial removal step and located in contact with the second mineral sacrificial layer, the thin opaque layer extending only over and in contact with the thin protective layer, so that, during the partial removal step, the thin opaque layer is protected by the thin protective layer; o after the partial removal step, removal of at least part of the thin protective layer by chemical etching, so as to free a lower face of the opaque thin layer.

[0024] The manufacturing method may comprise a step of producing a main encapsulation structure delimiting a main cavity in which the detection matrix is located, and comprising a main upper wall comprising a thin encapsulation layer resting on a main peripheral wall, by the following steps: o deposition of the thin encapsulation layer on the second mineral sacrificial layer, extending above the detection matrix and the compensation matrix; o production, in the thin encapsulation layer, of main vents arranged facing the detection matrix; o the partial removal of the first and second mineral sacrificial layers being carried out so as to form the main peripheral wall then formed of an unetched portion of the mineral sacrificial layers and surrounding the detection matrix, the thin encapsulation layer then being suspended from the above the detection matrix and resting on the main peripheral wall.

[0025] The manufacturing method may comprise a step of producing a communication chamber connecting the secondary cavity and the main cavity, the communication chamber being delimited laterally by a non-etched portion of the first and second mineral sacrificial layers.

[0026] The manufacturing method may include a step of producing reinforcing pillars of the thin encapsulation layer, resting on the reading substrate, preferably by means of anchoring pillars of the thermal detectors of the matrix of detection.

The chemical etching in an acid medium can be carried out with hydrofluoric acid in the vapor phase, and the first and second mineral sacrificial layers can be made of a silicon-based mineral material, and preferably of a silicon oxide. It can be performed using a fluorocarbon etchant in the gas phase, in particular when the mineral sacrificial layers are made based on a silicon nitride.

The invention also relates to a device for detecting electromagnetic radiation, comprising: o a reading substrate; o a detection matrix formed of thermal detectors intended to detect the electromagnetic radiation; o at least one so-called compensation matrix formed of thermal detectors intended not to detect electromagnetic radiation, adapted to detect electromagnetic radiation, and structurally identical to the thermal detectors (20p) of the detection matrix (2p); o a so-called secondary encapsulation structure delimiting a secondary cavity in which the compensation matrix is located, and comprising a peripheral wall as well as an opaque upper wall resting on the peripheral wall and formed of at least one opaque thin layer; o the opaque thin layer extending continuously flat, and the peripheral wall being made of a mineral material.

BRIEF DESCRIPTION OF DRAWINGS

Other aspects, objects, advantages and characteristics of the invention will appear better on reading the following detailed description of preferred embodiments thereof, given by way of non-limiting example, and made with reference in the appended drawings in which: FIGS. 1A to 1F are schematic and partial views illustrating various stages of a method of manufacturing a detection device according to one embodiment; Figure 2 is a top view, schematic and partial, of a detection device according to a variant of the embodiment illustrated in fig.lF, in which it comprises several secondary cavities; FIGS. 3A to 3F are schematic and partial views illustrating various steps of a method of manufacturing a detection device according to an embodiment variant, in which the thin opaque layer is made of a material with a getter effect; FIGS. 4A to 4D are schematic and partial views illustrating various steps of a method of manufacturing a detection device according to another embodiment, in which the encapsulation structure of the detection matrix comprises an attached cover and assembled to the reading substrate. DETAILED DISCUSSION OF PARTICULAR EMBODIMENTS

In the figures and in the following description, the same references represent identical or similar elements. In addition, the various elements are not shown to scale so as to favor the clarity of the figures. Furthermore, the different embodiments and variants are not mutually exclusive and can be combined with each other. Unless otherwise indicated, the terms “substantially”, “approximately”, “of the order of” mean to within 10%, and preferably within 5%. Furthermore, the terms "between ... and ..." and equivalents mean that the terminals are included, unless otherwise stated.

The invention generally relates to a method of manufacturing a device for detecting infrared or terahertz electromagnetic radiation.

This detection device comprises a plurality of thermal detectors, which are distributed so as to form at least one so-called sensitive matrix, or detection matrix, of thermal detectors intended to detect electromagnetic radiation, and at least one so-called compensation of thermal detectors intended not to detect electromagnetic radiation.

[0033] The manufacturing method comprises a step of producing the matrix of thermal detectors by means of so-called mineral sacrificial layers, made of a mineral or inorganic material, these sacrificial layers being intended to form a peripheral wall of a structure of encapsulation. This is a dielectric material based on silicon also allowing the production of an inter-metal dielectric layer of the read circuit, that is to say an electrically insulating material, with for example a dielectric constant, or relative permittivity, less than or equal to 3.9, thus limiting parasitic capacitance between the interconnects. This mineral material does not comprise carbon chains, and it can be based on a silicon oxide, for example be a silicon oxide SiO _x , possibly organosilicon such as SiOC, SiOCH, or a material of the fluoride glass type. such as SiOF. It can also be based on a silicon nitride, for example be a silicon nitride Si _x N _y . It is preferably a silicon oxide SiO _x .

[0034] The manufacturing process also includes a step of partially removing the mineral sacrificial layers by chemical etching, optionally chemical etching in an acid medium, for example with hydrofluoric acid in the vapor phase (HF vapour), in particular when the material mineral is based on a silicon oxide. In the case of a mineral material based on a silicon nitride, the partial etching can be carried out using a chemical gas-phase fluorocarbon. Be that as it may, other etchants can be used depending on the nature of the mineral material used.

The compensation matrix is located in a cavity, preferably hermetic, formed by an encapsulation structure which extends above and around the compensation thermal detectors. The encapsulation structure comprises at least: a mineral peripheral wall which extends around the compensation matrix and laterally delimits the cavity. As explained below, the mineral peripheral wall is formed of a non-etched portion of the mineral sacrificial layers; an opaque upper wall, which extends above the compensation matrix and vertically delimits the cavity. This opaque upper wall comprises at least one thin opaque layer of a material opaque to the electromagnetic radiation to be detected, that is to say whose transmission is less than or equal to 5%, or even less than or equal to 1%.

By thin layer is meant a layer formed by microelectronic material deposition techniques, the thickness of which is preferably less than or equal to 10 μm. Furthermore, a thin layer is said to be transparent when it has a transmission rate greater than or equal to 50%, preferably 75%, or even 90% for a central wavelength of the spectral range of the electromagnetic radiation to be detected. . The absorption rate of the thin layer is preferably less than or equal to 50%, preferably 25%, and more preferably 10%.

Different embodiments are illustrated below, and differ essentially with regard to the encapsulation structure defining the main cavity in which the detection matrix is located. It may thus be an encapsulation structure entirely made by depositing thin transparent layers on and through mineral sacrificial layers; or an encapsulation structure of which at least a part is transferred and assembled to the reading substrate.

[0038] Figures IA to 1F illustrate, schematically and partially, different steps of a method of manufacturing a detection device 1 according to one embodiment, in which the encapsulation structures 30s, 30p of the matrix compensation 2s and the detection matrix 2p are produced by depositing thin layers on and through the mineral sacrificial layers 41, 42. For the sake of clarity, only part of the detection matrix 2p and of the encapsulation structure 30p corresponding are shown in the figures. In this example, the detection device 1 comprises a compensation matrix 2s located in a cavity secondary 3s, but it can alternatively comprise several compensation matrices each located in a dedicated secondary 3s cavity (see fig.2).

We define here and for the rest of the description a direct three-dimensional reference XYZ, where the XY plane is substantially parallel to the plane of the reading substrate 10, the Z axis being oriented in a direction substantially orthogonal to the plane of the substrate of reading 10 towards thermal detectors 20p, 20s. The terms “vertical” and “vertically” are understood as being relative to an orientation substantially parallel to the Z axis, and the terms “horizontal” and “horizontally” as being relative to an orientation substantially parallel to the XY plane. Furthermore, the terms “lower” and “upper” are understood as being relative to an increasing positioning as one moves away from the reading substrate 10 along the direction +Z.

The detection device 1 comprises: a so-called sensitive matrix 2p of thermal detectors 20p intended to receive and detect the electromagnetic radiation of interest, the detection matrix 2p preferably being located in a main cavity 3p defined by a main encapsulation structure 30p; at least one so-called compensation matrix 2s of thermal detectors 20s intended not to receive the electromagnetic radiation of interest, the compensation matrix 2s being located in a secondary cavity 3s defined by a secondary encapsulation structure 30s. This secondary encapsulation structure 30s comprises an opaque upper wall 32s resting on a mineral peripheral wall 31s and comprising at least one thin opaque layer 33.

By way of example, the thermal detectors 20p are here suitable for detecting infrared radiation in the LWIR (Long Wavelength Infrared) range, the wavelength of which is between approximately 8 μm and 14 μm. Thermal detectors 20p and 20s are connected to a read circuit 14 located in substrate 10 (then called read substrate). The sensitive thermal detectors 20p thus form sensitive pixels preferably arranged periodically, and may have a lateral dimension in the plane of the reading substrate 10, of the order of a few tens of microns, for example equal to approximately 10 μm or even less.

The thermal compensation detectors 20s are structurally similar or identical to the sensitive thermal detectors 20p in the sense that they comprise a suspended membrane 22 by holding arms (not shown) and anchoring pillars 21. The suspended membrane 22 may also include a thermometric transducer. They can then provide the reading circuit 14 with an electrical signal representative of heating by the Joule effect During the lecture. It should also be noted that certain compensation thermal detectors can also supply read circuit 14 with an electrical signal that is also representative of the temperature of read substrate 10 (common mode). For this, these thermal detectors are thermalized to the read substrate 10 insofar as the holding arms do not ensure the thermal insulation of the absorbent membrane 22 vis-à-vis the read substrate 10.

With reference to fig.lA, the detection matrix 2p and the compensation matrix 2s are produced from the reading substrate 10, on and through a first mineral sacrificial layer 41. The reading substrate 10 is made from silicon, and is formed of a support substrate 11 containing the read circuit 14 suitable for controlling and reading the thermal detectors 20p, 20s. The read circuit 14 is presented here in the form of a CMOS integrated circuit. It comprises, among other things, portions of conductive lines separated from each other by inter-metal insulating layers made of a dielectric material, for example a silicon-based mineral material such as a silicon oxide SiO _x , a silicon nitride SiN _x , among others. Conductive portions 12 are flush with the surface of the support substrate 11, and ensure the electrical connection of the anchoring pillars 21 of the thermal detectors 20p, 20s to the reading circuit 14. In addition, one or more connection portions or pads 7 (cf. fig.lF) are flush with the surface of the support substrate 11, and make it possible to connect the reading circuit 14 to an external electronic device (not shown). In this example, the reading circuit 14 is adapted to read an electrical signal emitted by the thermal compensation detectors 20s, which is representative of heating by Joule effect during reading (and possibly representative of the temperature of the reading substrate 10) . Thus, by making a differential reading of the sensitive thermal detector 20p and of the compensation thermal detector 20s, it is possible to subtract from the 'raw' electrical signal the parasitic component linked to heating by Joule effect (and possibly the component linked to the temperature of the substrate) to keep only the useful part linked to the detection of the electromagnetic radiation of interest.

Each sensitive thermal detector 20s, and preferably each compensation thermal detector 20p, comprises a reflective layer 23 (reflector), which rests on the reading substrate 10 and is located opposite (and therefore below) each absorbent membrane 22. The reflector 23 may be formed by a portion of a conductive line of the last interconnection level, the latter being made of a material suitable for reflecting the electromagnetic radiation to be detected, or be a layer deposited on the layer protection 13 shown below. It extends opposite the absorbent membrane 22 of the sensitive thermal detector 20p, and is intended to form therewith a quarter-wave interference cavity vis-à-vis electromagnetic radiation to be detected. It preferably also extends opposite the absorbent membrane 22 of the thermal compensation detector 20s.

Finally, the reading substrate 10 here comprises a protective layer 13 so as to cover in particular the upper inter-metal insulating layer. This protective layer 13 here corresponds to an etching stop layer made of a material substantially inert to the chemical etching agent subsequently used to remove the various mineral sacrificial layers, for example in the HF medium in the vapor phase. This protective layer 13 thus forms a hermetic and chemically inert layer, and electrically insulating to avoid any short-circuit between the anchoring pillars 21. It thus makes it possible to prevent the underlying inter-metal insulating layers from being etched during this step of removing the mineral sacrificial layers. It can be formed from an aluminum oxide or nitride, or even from aluminum trifluoride, or else from unintentionally doped amorphous silicon.

The thermal detectors 20p, 20s are then produced on the reading substrate 10. These production steps are identical or similar to those described in particular in the document EP3239670A1. The sensitive thermal detectors 20p and the compensation thermal detectors 20s here advantageously have a similar structure. They are here microbolometers each comprising an absorbent membrane 22, i.e. capable of absorbing the electromagnetic radiation to be detected, suspended above the reading substrate 10 by anchoring pillars 21 and holding arms (not shown). The holding arms also ensure the thermal insulation of the absorbent membranes vis-à-vis the reading substrate 10. This is of course the case of the sensitive thermal detectors 20p, but also of the thermal compensation detectors 20s which thus supply a signal electricity representative of heating by the Joule effect during reading.

The production of absorbent membranes 22 is conventionally carried out by surface micro-machining techniques consisting in producing the anchoring pillars 21 through a first mineral sacrificial layer 41, and the holding arms as well as the membranes absorbents 22 on the upper face of the mineral sacrificial layer 41. Each absorbent membrane 22 further comprises a thermometric transducer, for example a thermistor material, connected to the reading circuit 14 by electrical connections provided in the thermal insulation arms and in anchor pillars 21.

The sensitive thermal detectors 20p are located in a main zone of the surface of the reading substrate 10 intended to correspond to the main cavity 3p (detection cavity), and the compensation thermal detectors 20s are located in a secondary zone of this surface intended to correspond to the secondary cavity 3s (compensation cavity). Note that the detection matrix 2p can contain a large number of thermal detectors 20p, for example 640×480. The 2s compensation matrix can, for example, contain 4×480 20s thermal detectors. The main zone therefore has a larger surface than the secondary zone.

A second mineral sacrificial layer 42 is then deposited, preferably of the same nature as the mineral sacrificial layer 41. The mineral sacrificial layer 42 thus covers the mineral sacrificial layer 41 as well as the sensitive thermal detectors 20p and the compensating thermal detectors 20s. It has a substantially planar free upper face. In general, the various mineral sacrificial layers 41, 42 can be a silicon oxide obtained from a TEOS compound (tetraethyl orthosilicate) deposited by PECVD. Mineral sacrificial layers 41, 42 can be made of the same mineral material.

With reference to fig.lB, the thin opaque layer 33 is produced intended to screen the compensation matrix 2s, that is to say to avoid the transmission of the electromagnetic radiation to be detected in the direction of the thermal compensation detectors 20s. Thin opaque layer 33 can be a layer that reflects or absorbs the electromagnetic radiation of interest. The thin opaque layer 33 is made so that it extends in a continuously flat manner only on an upper surface (on part of the upper face) of the mineral sacrificial layer 42.

In this example, a thin opaque layer 33 is deposited on and in contact with the second sacrificial mineral layer 42. Alternatively, one or more thin layers may have been previously deposited on the second sacrificial mineral layer 42. The layer thin opaque 33 is deposited so as to extend in a continuously planar manner above the compensation matrix 2s. By continuously planar, it is meant that the thin opaque layer 33 extends planarly in the XY plane over its entire surface area. It is deposited so that it has a substantially constant thickness.

In the case of a reflective material, it may be aluminum, gold, tungsten, copper or titanium, with a constant thickness for example between 1000 nm and a few hundred nanometers. , for example equal to approximately 300 nm. Preferably, the thickness of the thin opaque layer 33 is less than or equal to lpm so as not to complicate the manufacturing process. These materials of the opaque thin layer 33 are advantageously substantially inert (or weakly reactive) to the chemical etching implemented to partially remove the mineral sacrificial layers 41, 42. In the case where the material is weakly reactive to the etching agent used, the thickness of the material deposited will be slightly greater than the desired final thickness, to take into account a slight partial etching (thinning) during the chemical etching step.

The thin opaque layer 33 can be deposited by thin layer deposition techniques guaranteeing uniformity of its thickness, for example by physical vapor deposition (PVD, for Physical Vapor Deposition, in English), of the sputtering type. cathode of a metal target or by vacuum evaporation of a metal heated in a crucible.

The thin opaque layer 33 is then structured by lithography and localized etching, so that it does not extend above the detection matrix 2p. It can thus extend everywhere on the second mineral sacrificial layer 42 (except above the detection matrix 2p - as illustrated in FIG. 4A), or can only extend above the matrix of 2s compensation and secondary zone (as shown in fig.lF). Be that as it may, the thin opaque layer 33 extends at least partly above the area where the mineral peripheral wall 31s of the secondary encapsulation structure 30s will be located.

Preferably, indentations 43 are made here and preferably insulating portions 44 for the production of reinforcing pillars 35 of thin encapsulation layer 34 of the main encapsulation structure 30p. Initially, several indentations 43 are made which extend from the upper face of the second mineral sacrificial layer 42 along the Z axis to lead to the anchoring pillars 21 of the sensitive thermal detectors 20p. Then, a plurality of insulating portions 44 are advantageously made in the notches 43. These insulating portions 44 are portions of a thin layer made of an electrically insulating material. They make it possible to avoid electrical contact between the sensitive thermal detectors 20p and the thin encapsulation layer 34 via its reinforcing pillars 35. To do this, a thin insulating layer is deposited on the free surface of the anchoring pillars 21 to inside the notches 43. The thin insulating layer is here advantageously etched locally above the sensitive thermal detectors 20p, so as not to disturb or reduce the transmission of the electromagnetic radiation to be detected, but it could not be etched. It may have a thickness of between 10 nm and approximately 200 nm. It is made of a material inert to the chemical etching implemented during the removal of the mineral sacrificial layers, which can be chosen from AIN, ALOs, HfO ₂ .

With reference to fig.lC, the thin encapsulation layer 34 of the main encapsulation structure 30p is produced, this thin encapsulation layer 34 being formed of a portion upper extending above the detection matrix 2p, and comprising reinforcing pillars 35 located in the main zone, distinct from each other, and resting on the reading substrate 10 via the anchoring pillars 21 of the detectors sensitive thermals 20p. For this, the conformal deposition of the thin encapsulation layer 34 is carried out, made of a material transparent to the electromagnetic radiation of interest and inert to the chemical etching implemented subsequently, with a thickness comprised for example between 200 nm and 2 pm , for example equal to approximately 800 nm or even less, for example amorphous silicon, amorphous germanium, an amorphous silicon-germanium alloy, among others. The thin encapsulation layer 34 is deposited on the mineral sacrificial layer 42 as well as in the notches 43, for example by a technique of chemical vapor deposition (CVD for Chemical Vapor Deposition, in English). Preferably, it also covers the thin opaque layer 33, thus ensuring a reinforcement of the mechanical strength of what will be the upper opaque wall 32s of the secondary encapsulation structure 30s. The thin encapsulation layer 34 thus comprises, made in one piece and from the same material or materials: an upper portion, substantially planar in the XY plane, which extends above along the Z axis of the matrix of detection 2p, and reinforcement pillars 35 which rest on the reading substrate 10, here indirectly via the anchoring pillars 21.

The thin encapsulation layer 34 here forms a quarter-wave plate with respect to the electromagnetic radiation of interest. Thus, in the case of amorphous silicon and for a spectral detection band ranging from 8 μm to 14 μm, it advantageously has a thickness of approximately 800 nm. Thus, the opaque upper wall 32s comprising the thin opaque layer 33 and the thin encapsulation layer 34 (quarter-wave plate) forms an interference stack which, while remaining opaque to the electromagnetic radiation of interest, makes it possible to reduce the reflection of the latter by absorption in the quarter-wave plate liable to form parasitic images by the detection device 1. It should be noted here that the opaque upper wall 32s can of course comprise additional thin layers, thus improving the interference properties of this stack .

More broadly, the opaque thin layer 33 can be an absorbent multilayer such as a stack (multilayer) formed by alternating elementary metallic and dielectric thin layers, thus reducing parasitic reflections. The thin encapsulation layer 34 can also be a stack formed of alternating metallic and dielectric layers, which however remains transparent to the electromagnetic radiation of interest when it extends above the detection matrix 2p (cf. fig.1E, 3E) or which can be opaque (absorbent multilayer) when it does not extend above the detection matrix 2p (cf. fig.4C). Be that as it may, one and/or the other of the multilayers of the opaque upper wall 32s can thus form an absorbing interference stack reducing parasitic reflections and thus improving the performance of the detection device 1. Furthermore, the thin opaque layer 33 and/or the part of the thin encapsulation layer 34 located above the , and more broadly the opaque upper wall 32s, may also have lateral structuring, in the XY plane, improving the opacity properties, in particular by absorption of the electromagnetic radiation of interest.

The reinforcing pillars 35 have dimensions in the XY plane of the order of those of the anchor pillars 21. Thus, the anchor pillars 21 can each comprise a vertical portion of dimensions in the XY plane of the order of 0.5pm to lpm surmounted by an upper portion projecting laterally of the order of 0.2pm to 0.5pm vis-à-vis the vertical portion. The reinforcing pillars 35 may here have dimensions in the XY plane of the order of 0.5 μm to 2 μm approximately.

[0060] Vents 36p, 36s are then made, making it possible to produce the main 3p and secondary 3s cavities. These vents 36p, 36s lead to the sacrificial mineral layer 42 and are intended to allow the evacuation of the various sacrificial mineral layers 41, 42 out of the main cavity 3p and the secondary cavity 3s. First vents 36p are made through the thin encapsulation layer 34 and are intended for the formation of the main cavity 3p. Second vents 36s are made through the thin encapsulation layer 34 and the thin opaque layer 33, and are intended for the formation of the secondary cavity 3s. The vents 36p, 36s are arranged only opposite the main zone and the secondary zone, for example at the rate of one vent per thermal detector. Thus, they will make it possible to completely free the surface of the reading substrate 10 in the main and secondary zones, and to form the mineral peripheral wall 31s. In this example, the vents 36p, 36s are located perpendicular to the suspended membranes of the sensitive thermal detectors 20p and the compensation thermal detectors 20s, but they can be arranged differently, in particular perpendicular to their anchoring pillars 21. vents 36p, 36s can have different shapes in the XY plane, for example a circular shape with a diameter of 0.4 μm or even less. Thus, the first vents 36p do not or only slightly disturb the transmission of the electromagnetic radiation of interest, and the second vents 36s do not or only slightly disturb the screening vis-à-vis this electromagnetic radiation of interest.

[0061] With reference to fig.ID, a chemical etching is carried out adapted to partially remove the mineral sacrificial layers 41, 42. The chemical etching is an etching, for example with hydrofluoric acid in the vapor phase, in particular when the layers sacrificial minerals 41, 42 are made based on a silicon oxide. The products of the chemical reaction are evacuated through vents 36p, 36s. Due to the arrangement of the vents 36p, 36s located only opposite the 2p detection and 2s compensation matrices, the etching agent completely removes the mineral sacrificial layers 41, 42 located in these areas, but the chemical etching is performed so that the etching agent does not etch a peripheral portion of the mineral sacrificial layers 41, 42 which extends around the compensation matrix 2s, and here also around the detection matrix 2p. Thus, the mineral peripheral wall 31s surrounds the compensation matrix 2s and laterally delimits the secondary cavity 3s. A mineral peripheral wall also surrounds the detection matrix 2p and laterally delimits the main cavity 3p (cf. fig.LF). The two mineral peripheral walls 31s, 31p coincide between the main cavity 3p and the secondary cavity 3s.

Thus, the thin opaque layer 33 and the thin encapsulation layer 34 together form an upper opaque wall 32s, suspended above the compensation matrix 2s, which rests on the mineral peripheral wall 31s. It participates in delimiting, with the latter, the secondary cavity 3s. And the thin encapsulation layer 34 is suspended above the detection matrix 2p, and participates in delimiting the main cavity 3p. It rests on the mineral peripheral wall 31p.

Unlike document WO2012/056124A1, the encapsulation structure 30s of the secondary cavity 3s does not include a peripheral wall formed of a thin layer which would extend above and around the compensation matrix 2s , and would come to the reading substrate 10. In the context of the invention, the encapsulation structure 30s of the secondary cavity 3s comprises the mineral peripheral wall 31s and an opaque upper wall 32s which rests on the latter, and s 'extends continuously planar above the compensation matrix 2s.

[0065] With reference to fig.lE, a sealing layer 37 is deposited on the thin encapsulation layer 34 with a sufficient thickness to ensure the sealing, ie the plugging, of the vents 36s, 36p. It extends at least opposite the main cavity 3p and the secondary cavity 3s. The sealing layer 37 is transparent to the electromagnetic radiation to be detected, and can be made of germanium with a thickness of approximately 1.7 μm by vacuum deposition for placing the thermal detectors under vacuum. It is also possible to deposit an antireflection layer (not represented) making it possible to optimize the transmission of electromagnetic radiation through the main encapsulation structure 30p. This anti-reflective layer can be made of zinc sulphide with a thickness of approximately 1.2 μm. [0066] FIG. IF illustrates in top view, schematically and partially, the detection device 1. The compensation matrix 2s is located in the secondary cavity 3s, which is delimited laterally by the mineral peripheral wall 31s (whose inner border is represented by a dotted line) and vertically by the thin opaque layer 33 (solid line). The latter extends in a continuously planar manner above the compensation matrix 2s, with a constant thickness, and rests on the mineral peripheral wall 31s. The detection matrix 2p is located in the main cavity 3p, which is delimited laterally by the mineral peripheral wall 31p (dotted line) and vertically by the thin encapsulation layer 34 (not shown). Connection pads 7 are here located at the edge of the matrices of thermal detectors 20p, 20s, and make it possible to connect the read circuit 14 to an external electrical circuit (not shown). They are accessible from the outside through openings made in non-etched portions of the mineral sacrificial layers (through layers 37, 34, 42 then 41). It is noted that the secondary cavity 3s has a lateral dimension smaller than the dimensions of the main cavity 3p. It can thus be less than or equal to 200pm. This width is defined so that the secondary encapsulation structure 30s does not require reinforcing pillars 35, unlike here the main encapsulation structure 30p, which would be made in one piece and from the same material with a thin layer (here the thin layer 34) of the opaque upper wall 32s.

[0067] A secondary hermetic cavity 3s is thus obtained, preferably under vacuum or at reduced pressure, in which the compensation thermal detectors 20s are housed. The secondary encapsulation structure 30s therefore comprises an opaque upper wall 32s formed here of the thin opaque layer 33, of the thin encapsulation layer 34, and of the thin sealing layer 37, this opaque upper wall 32s resting on the wall mineral device 31s.

[0068]Thus, the secondary encapsulation structure 30s does not comprise any support structure for the opaque upper wall 32s on the reading substrate 10 other than the mineral peripheral wall 31s, which comes from the mineral sacrificial layers 41, 42 necessary to the realization of thermal detectors 20p, 20s. It therefore does not include a peripheral wall in a thin layer, produced through the mineral sacrificial layers 41, 42, which would come to rest directly on the reading substrate 10, as in document WO2012/056124A1. Moreover, the mineral peripheral wall 31s is not reflective, which makes it possible to prevent stray light from being reflected towards the sensitive thermal detectors 20p.

Besides the fact that this makes it possible to reduce the complexity of the manufacturing process (in number of production steps, in particular), it also avoids carrying out localized etching of the mineral sacrificial layers 41, 42 which would lead to the protective layer 13 of the read substrate 10. This avoids any risk of degradation of this protective layer 13, in particular in terms of sealing, which eliminates the risks of degradation of the read substrate 10 during chemical etching with HF vapor. In addition, the mechanical strength of the encapsulation structure 30s is improved insofar as the opaque upper wall 32s is assembled to the reading substrate 10 by a mineral peripheral wall 31s which has an interface with the reading substrate 10 of greater surface than in the case of a peripheral wall in a thin layer.

Furthermore, the absence of reinforcing pillars 35 in the secondary cavity 3s makes it possible to avoid a variation in the topology of the thin opaque layer 33 in the XY plane, or even a variation in thickness. Such variations can result in a degradation of the optical property of opacity of the thin opaque layer 33. In addition, such reinforcing pillars would extend through openings made in the thin opaque layer 33; these openings would degrade the screening of the 2s compensation matrix. Also, by the absence of the reinforcing pillars 35 in the secondary cavity 3s, the thin opaque layer 33 can remain continuously flat and of constant thickness, thus preserving the good uniformity of its optical property of opacity.

Furthermore, the fact of producing the thin opaque layer 33 by PVD deposition opens up a greater choice of possible materials, in particular metallic, than in the case where the thin opaque layer 33 forms a peripheral wall in a thin layer, as in WO2012/056124A1. Indeed, in this case, it would be necessary to use specific deposition techniques, such as for example chemical vapor deposition (CVD, for Chemical Vapor Deposition, in English), which limits the choice of possible materials. Furthermore, a greater choice of possible materials makes it possible to choose an opaque material having an additional function, such as a getter function, as described later with reference to FIGS. 3A to 3F.

Furthermore, the thin opaque layer 33, resting on the mineral peripheral wall 31s, can overflow laterally vis-à-vis the compensation matrix 2s, which makes it possible to obtain good screening efficiency. The screening efficiency is also all the greater since the material of the mineral peripheral wall 31s can participate in laterally screening the electromagnetic radiation of interest. Indeed, by way of example, a silicon oxide has a high absorption in the spectral band between 8 and 14 μm.

Finally, the opaque top wall 32s is described in this example for illustrative purposes. Other configurations are of course possible. Thus, the upper opaque wall 32s can comprise other thin layers, located under or on the thin opaque layer 33. Furthermore, the arrangement thin layers in the opaque upper wall 32s can be chosen so as to take account of the differences in mechanical stresses in each of the thin layers.

[0074] Figure 2 is a top view, schematic and partial, of a detection device 1 according to a variant of that illustrated in fig.LF. In this example, the detection device 1 differs from that described in FIG. 1F essentially in that it comprises several secondary cavities, here two, which each house a compensation matrix 2s. The two secondary cavities are adjacent and are separated by the same mineral peripheral wall 31s. The thin opaque layer 33 here extends continuously above the two compensation matrices 2s. It therefore rests on the mineral peripheral wall 31s located between the two secondary cavities. As a variant, the secondary encapsulation structures 30 s can each comprise a dedicated thin opaque layer 33 . Be that as it may, it is advantageous to provide several secondary cavities when the required number of thermal compensation detectors 20s does not allow them all to be housed in the same secondary cavity without having to make reinforcing pillars 35 similar to those of the main cavity. In other words, it is advantageous to house the thermal compensation detectors 20s in several secondary cavities of a sufficiently small lateral dimension to avoid having to make reinforcement pillars 35, for example less than or equal to approximately 200 μm.

[0075] Figures 3A to 3F illustrate, schematically and partially, different steps of a manufacturing process according to a variant of the embodiment illustrated in fig.lA to 1F. In this example, thin opaque layer 33 is made of a material having a getter function. In general, a material with a getter effect is a material intended to be exposed to the atmosphere of the hermetic cavity and capable of performing gas pumping by absorption and/or adsorption. It may be the metallic material reflecting the electromagnetic radiation of interest, for example titanium.

In this example, the metallic material is sensitive to the etchant used during the chemical etching used for the partial removal of the mineral sacrificial layers 41, 42. Also, it is protected from this etchant by a layer sacrificial protection 38 made of amorphous carbon.

The amorphous carbon may optionally be of the DLC (Diamond Like Carbon) type, that is to say it has a high sp ³ carbon hybridization rate. It is substantially inert with respect to the chemical etching carried out to partially remove the layers sacrificial minerals 41, 42, that is to say that it reacts little or not with the chemical etching agent. Also, after this partial removal step, it still protects the getter material. The sacrificial protective layer 38 is adapted to be removed by a second chemical etching such as dry chemical etching, an etching agent of which is for example oxygen contained in a plasma.

With reference to FIG. 3A, the detection matrix 2p and the compensation matrix 2s are produced on and through the first mineral sacrificial layer 41. The second mineral sacrificial layer 42 covers the two matrices of thermal detectors 20p, 20s as well as the first mineral sacrificial layer 41. It has a flat upper face. This step is identical to that described previously.

With reference to FIG. 3B, an opaque stack is produced formed of a thin protective layer 38 and of the thin opaque layer 33. This stack extends in a planar and continuous manner above the matrix of 2s compensation, and does not extend above the 2p detection matrix. It is intended to rest on the mineral peripheral wall 31s.

The thin protective layer 38 rests on and in contact with the second sacrificial mineral layer 42. It is intended to protect the thin opaque layer 33 during the chemical etching implemented during the partial removal of the sacrificial mineral layers 41 , 42. It is intended to be removed during a second chemical etching, in which the thin opaque layer 33 is substantially inert, for example by dry chemical etching. It is made of amorphous carbon and has a thickness for example of between 50 nm and 500 nm.

The thin opaque layer 33 rests on and in contact with the thin protective layer 38, and is therefore not in contact with the second mineral sacrificial layer 42. It is made of a metallic material that reflects the electromagnetic radiation to be detected. and has a getter effect, for example in titanium.

The indentations 43 and the insulating portions 44 are also produced, intended for the production of the reinforcing pillars 35 of the thin encapsulation layer 34 of the main encapsulation structure 30p, in the same manner as described above.

[0083] With reference to FIG. 3C, the thin encapsulation layer 34 is then deposited, so as to cover the opaque stack here and to extend above the detection matrix 2p. It fills the notches 43 and forms the reinforcing pillars 35. The first and second vents 36p, 36s are also produced.

With reference to fig.3D, the chemical etching is carried out so as to partially remove the mineral sacrificial layers 41, 42, and thus form the main cavities 3p and secondary 3s delimited by the mineral peripheral walls 31s, 31p (cf. fig.3F). The upper opaque wall 32s is then suspended above the compensation matrix 2s and rests on the mineral peripheral wall 31s. The thin protective layer 38 has a part of its underside which has been made free. However, it protected the thin opaque layer 33 against the etchant used. The structural integrity of the thin opaque layer 33 has therefore been preserved, and therefore also its optical properties and its getter effect.

With reference to fig.3E, a second chemical etching is carried out, to which the thin protective layer 38 is sensitive, for example a dry chemical etching, to remove the part having its free lower surface. Lateral over-etching may also occur. A part of the lower face of the thin opaque layer 33 is thus made free. The sealing layer is then deposited to close the vents 36p, 36s. The chemisorption of the getter-effect material of the thin opaque layer 33 is then activated by subjecting the detection device 1 to an appropriate heat treatment, for example in an oven or an oven.

The fig.3F is a top view, schematic and partial, of the detection device 1 thus obtained. So that the getter material of the thin opaque layer 33 can ensure the gaseous pumping of the secondary and main cavities, a communication chamber 6 is made, which ensures the gaseous communication between the two cavities. It is delimited laterally by a non-etched portion of the mineral sacrificial layers and is delimited here vertically by the opaque upper wall 32s. To obtain this communication chamber 6 during the step of partial removal of the mineral sacrificial layers, vents 36s, 36p were previously made through the opaque upper wall 32s, and placed above the area intended to form the communication chamber 6. Vents 36s here pass through layers 34, 33 and 38, while vents 36p only pass through layer 34.

Figures 4A to 4D illustrate, schematically and partially, different steps of a manufacturing method according to another embodiment. It differs from those described previously essentially in that the main encapsulation structure 30p does not comprise a thin encapsulation layer 34, but an attached rigid cover 9, that is to say a cover made beforehand then added and assembled to the reading substrate 10 so as to encapsulate the detection matrix 2p (the cover 9 here also encapsulates the compensation matrix 2s). The main encapsulation structure 30p is here similar or identical to that described in the document EP3239670A1. The cover 9 can be made from a silicon substrate, and structured so as to include a peripheral wall intended to be assembled to the reading substrate 10. The peripheral wall is fixed to the reading substrate 10 by the intermediary of a hermetic seal 8, the latter preferably being in contact with an attachment portion of a metallic layer. The hermetic seal 8 can be obtained by the recasting of a fusible metal or by the formation of an intermetallic alloy.

The method then includes a step of producing (FIG. 4A) the detection matrix 2p and the compensation matrix on the read substrate 10, as described above. The upper opaque wall 32s is here formed of a stack comprising the thin opaque layer 33 and a thin reinforcing layer 39. The thin reinforcing layer 39 can form a quarter-wave plate, as described above. It participates here in reinforcing the mechanical strength of the opaque upper wall 32s. The opaque upper wall 32s extends in the secondary zone, and possibly around the main zone, but does not extend above the detection matrix 2p. In this example, it is intended to extend beyond the mineral peripheral wall 31s, so as to form a cantilevered portion (a part which projects laterally beyond the mineral peripheral wall 31s along a opposite direction to the secondary cavity 3s).

The partial removal of the mineral sacrificial layers 41, 42 is then carried out (fig.4B) by chemical etching. The detection matrix 2p is thus released, as well as the compensation matrix 2s, which is surrounded by the mineral peripheral wall 31s. The opaque upper wall 32s then presents the cantilevered portion located between the compensation matrix 2s and the detection matrix 2p. In this example, this cantilevered portion is formed of two thin layers 33, 39, but as a variant, it may be formed only of the thin reinforcing layer 39 (the opaque thin layer 33 stopping at the above the mineral peripheral wall 31s). The choice between these two configurations may depend on the difference in mechanical stresses between these thin layers 33, 39. Thus, a configuration where the cantilevered portion would be formed only of the thin reinforcing layer 39 is advantageous to avoid an imbalance of the mechanical stresses between the two layers 33, 39 and to correct any deflection of the cantilevered portion. Note that it appears that a chemical attack in an acid medium of the mineral sacrificial layers 41, 42 in a confined medium (ie under the opaque upper wall 32s) has a lateral etching speed (in the XY plane) greater than the etching speed vertical (along the Z axis). Also, the release of the detection matrix 2p and the formation of the secondary cavity 3s are obtained at the same time (etching of the layers 41 and 42 and evacuation through the vents 36s). A line of sealing material intended to form the hermetic seal 8 is then deposited (fig.4C), which rests on the reading substrate 10 and surrounds the detection matrix 2p. It also surrounds the 2s compensation matrix here. This line of sealing material was deposited before the partial removal of the mineral sacrificial layers 41, 42, for example in a peripheral trench crossing the mineral sacrificial layers and surrounding the detection matrix 2p. The cover 9 is then transferred to the hermetic seal 8 and it is assembled to the reading substrate 10. It is noted that the cover 9 ensures the sealing of the vents 36s. Thus, the cavity 3s is contained in the cavity 3p.

[0092] The fig.4D is a top view, schematic and partial, of the detection device 1 obtained after the step of producing the hermetic seal 8 and before the postponement of the cover 9. The mineral peripheral wall 31s of the structure of the secondary encapsulation 30s here has a width less than its length, and extends longitudinally around the compensation matrix 2s. The thin opaque layer 33 extends above the compensation matrix 2s, rests on the mineral peripheral wall 31s, and here has a cantilevered portion. The rest of the surface of reading substrate 10 is thus not covered by a non-etched portion of mineral sacrificial layers.

[0093] Particular embodiments have just been described. Different variants and modifications are possible while remaining within the scope of the invention.

Thus, the main encapsulation structure 30p can, as a variant, be similar or identical to that described in the document EP3399290A1. Such an encapsulation structure comprises a peripheral wall which surrounds the detection matrix 2p, and which is produced by a thin layer deposition technique. A top wall can be transferred and assembled on the peripheral wall by means of a temporary handle.

Claims

24 CLAIMS

1. Method for manufacturing a device (1) for detecting electromagnetic radiation, comprising the following steps: o production of a detection matrix (2p) formed of thermal detectors (20p) intended to detect electromagnetic radiation, and at least one so-called compensation matrix (2s) formed of thermal detectors (20s) intended not to detect electromagnetic radiation, on and through a first sacrificial layer (41) resting on a reading substrate (10 ),

• the thermal detectors (20s) of the compensation matrix (2s) being suitable for detecting electromagnetic radiation and structurally identical to the thermal detectors (20p) of the detection matrix (2p); o production of a second sacrificial layer (42) covering the thermal detectors (20s, 20p) and the first sacrificial layer (41); o production of a so-called secondary encapsulation structure (30s) delimiting a secondary cavity (3s) in which the compensation matrix (2s) is located, and comprising a peripheral wall (31s) as well as an opaque upper wall (32s ) resting on the peripheral wall (31s) and formed of at least one thin opaque layer (33); o characterized in that: o the first and second sacrificial layers (41, 42) are made of a mineral material; o the step of producing the secondary encapsulation structure (30s) comprises the following steps:

• production of the opaque thin layer (33) so that it extends continuously flat only on an upper surface of the second mineral sacrificial layer (42);

• production, in the thin opaque layer (33), of vents (36s) arranged opposite the compensation matrix (2s);

• partial removal of the first and second mineral sacrificial layers (41, 42) through the vents (36s), by chemical etching, so as to release the detection matrix (2p) and the compensation matrix (2s), and to obtain the peripheral wall (31s) then formed of a non-etched portion of the mineral sacrificial layers (41, 42) and surrounding the compensation matrix (2s), the thin opaque layer (33) then being suspended above the compensation (2s) and resting on the peripheral wall (31s).

2. Manufacturing process according to claim 1, in which the first and second sacrificial layers (41, 42) are made of the same mineral material based on an oxide or a silicon nitride.

3. Manufacturing process according to claim 1 or 2, in which the thermal detectors (20p) of the detection matrix (2p) like the thermal detectors (20s) of the compensation matrix (2s) each comprise an absorbent membrane (22 ) capable of absorbing the electromagnetic radiation to be detected and comprising a thermometric transducer, suspended above the reading substrate (10) by anchoring pillars (21) and holding arms and thermal insulation.

4. Manufacturing process according to claim 3, in which the thermal detectors (20p) of the detection matrix (2p) and/or the thermal detectors (20s) of the compensation matrix (2s) each comprise a reflective layer (23 ), which rests on the reading substrate (10), below each absorbent membrane (22).

5. Manufacturing process according to any one of claims 1 to 4, in which the opaque upper wall (32s) comprises an interference stack absorbing the electromagnetic radiation to be detected.

6. Manufacturing process according to any one of claims 1 to 5, wherein the opaque upper wall (32s) further comprises at least one thin reinforcing layer (39) covering the thin opaque layer (33), and has a border protruding vis-à-vis the peripheral wall (31s) in a plane parallel to the reading substrate (10), the protruding border comprising the thin opaque layer (33) and/or the thin reinforcing layer (39) .

7. Manufacturing process according to any one of claims 1 to 6, in which the secondary cavity (3s) has a length and a width in a plane parallel to the reading substrate (10), the width being less than or equal to 200 μm , the opaque upper wall (32s) not comprising reinforcement pillars, made in one piece and of the same material with a thin layer of the opaque upper wall (32s), located in the secondary cavity (3s) and coming rest on the reading substrate (10).

8. Manufacturing process according to any one of claims 1 to 7, in which the mineral sacrificial layers (41, 42) are made of a material that absorbs the electromagnetic radiation to be detected.

9. Manufacturing process according to any one of claims 1 to 8, in which the thin opaque layer (33) is made of a material with a getter effect.

10. Manufacturing process according to claim 9, comprising the following steps: o before the step of partial elimination, production of the opaque upper wall (32s) formed of a stack comprising a thin protective layer (38) made of carbon inert to an etchant used during the partial removal step and located in contact with the second mineral sacrificial layer (42), the thin opaque layer (33) extending only over and in contact with the thin layer of protection (38);

• so that, during the partial removal step, the thin opaque layer (33) is protected by the thin protective layer (38), o after the partial removal step, removal of at least part of the thin protective layer (38) by chemical etching, so as to free an underside of the thin opaque layer (33).

11. Manufacturing method according to any one of claims 1 to 10, comprising a step of producing a main encapsulation structure (30p) delimiting a main cavity (3p) in which the detection matrix (2p) is located. , and comprising a main upper wall (32p) comprising a thin encapsulation layer (34) resting on a main peripheral wall (31p), by the following steps: o depositing the thin encapsulation layer (34) on the second mineral sacrificial layer (42), extending above the detection matrix (2p) and the compensation matrix (2s); o production, in the thin encapsulation layer (34), of main vents (36p) arranged facing the detection matrix (2p); o the partial removal of the first and second mineral sacrificial layers (41, 42) being carried out so as to form the main peripheral wall (31p) then formed of a non-etched portion of the mineral sacrificial layers (41, 42) and surrounding the matrix (2p), the thin encapsulation layer (34) then being suspended above the detection matrix (2p) and resting on the main peripheral wall (31p).

12. Manufacturing method according to claim 9 or 10, and according to claim 11, comprising a step of producing a communication chamber (6) connecting the secondary cavity (2s) and the main cavity (2p), the chamber communication (6) being delimited laterally by a non-etched portion of the first and second mineral sacrificial layers (41, 42). 27

13. Manufacturing process according to claim 11 or 12, comprising a step of producing reinforcement pillars (35) of the thin encapsulation layer (34), resting on the reading substrate (10), preferably by intermediary of anchoring pillars (21) of the thermal detectors (20p) of the detection matrix (2p).

14. Manufacturing process according to any one of claims 1 to 13, in which the chemical etching is carried out with hydrofluoric acid in the vapor phase, and the first and second mineral sacrificial layers (41, 42) are made of a material mineral based on a silicon oxide.

15. Device for detecting electromagnetic radiation, comprising: o a reading substrate (10); o a detection matrix (2p) formed of thermal detectors (20p) intended to detect the electromagnetic radiation; o at least one so-called compensation matrix (2s) formed of thermal detectors (20s) intended not to detect electromagnetic radiation, adapted to detect electromagnetic radiation, and structurally identical to the thermal detectors (20p) of the detection matrix (2p ); o a so-called secondary encapsulation structure (30s) delimiting a secondary cavity (3s) in which the compensation matrix (2s) is located, and comprising a peripheral wall (31s) as well as an opaque upper wall (32s) resting on the peripheral wall (31s) and formed of at least one thin opaque layer (33); o characterized in that:

• the thin opaque layer (33) extends continuously flat;

• the peripheral wall (31s) is made of a mineral material.