EP4094069A1 - System for capturing point values for constituting an image with terahertz radiation - Google Patents

Abstract

The device (40) for capturing point values for constituting an image, comprises: - an incoherent source (100) of rays, the frequency of which is between 0.075 THz and 10 THz for illuminating an object, - a sensor (415) of radiation coming from the object, which comprises an area sensitive to the radiation coming from the source and which emits an electrical signal representative of the intensity of the rays coming from the source and reaching the sensitive area of the sensor, and - at least one optical focusing system (400, 410) with aperture number (F-Number) less than one, situated on the optical path of rays emitted by the source and propagating from the source to the sensor of rays, passing via the object. Preferably, the source (100) illuminates the object with a sufficiently broad emission spectrum to scan the standing wave in a period shorter than the acquisition time of the sensor. Preferably, the incoherent source has a bandwidth of several GHz, preferably at least equal to 12 GHz at -100 dB.

Description

 SYSTEM FOR CAPTURING PICTURE VALUES TO CONSTITUTE AN IMAGE WITH TERAHERTZ RADIATION

TECHNICAL AREA

 The present invention aims a system for capturing point values, to constitute an image with Terahertz radiation. It concerns, in particular, the field of imaging, for example for the control of qualities of manufactured parts.

STATE OF THE ART

 The Terahertz domain (THz) designates electromagnetic waves whose frequency is between 0.075 THz and 10 THz, in terms of wavelength from 4 mm to 30 μηι. These waves are said to be millimeters, they are located between far infrared (or FIR for Far InfraRed), and radar waves (microwave). The invention more particularly relates to the range 75-700 GHz. Indeed, in this frequency range, a large part of the non-conductive plastic and composite materials are transparent to the radiation. And the radiation sources are relatively accessible in terms of the ratio between the price of the source and the available optical power.

 All optical principles used in visible or infrared system design apply to THz radiation with the same proportion ratio of wavelength phenomena. For example, the diffraction limit follows the Airy law: Rdiff = 1, 22 λ ON, in which λ is the wavelength and ON is the numerical aperture. For a numerical aperture ON of 2, and a wavelength of 1 mm, a diffraction limit of 2.44 mm. This diffraction limit of the order of one millimeter corresponds to the average spatial resolution of a THz imaging system.

 However, the known systems do not reach this resolution, which would make it possible to reach the optimal performances.

 Existing THz systems use off-axis parabolic mirrors as a means of beam propagation. This choice is explained because, in many applications, the power level of the source is so low that the losses introduced by each lens are not acceptable (losses by absorption, and especially losses by reflections on the interfaces).

 However, the use of these mirrors poses many problems:

 - difficulty of alignment with many angle references,

 - very high price of optics,

 - restricted focal length range especially on large openings and

- obligation to work in infinite-focus conjugation and on the optical axis. Moreover, a THz beam is invisible to the eye, and, unlike the near infrared, there is no photosensitive map allowing the conversion of radiation to wavelengths visible to the eye. This aspect has an impact on the tuning and alignment of the systems.

 The types of sources available in the THz domain are described below.

Several THz radiation generation methods for continuous THz radiation exist and can be used in a device that is the subject of the invention, for example:

 Gunn diode associated with a frequency multiplier to go beyond 0.2 THz, for example 2 THz,

 - Impatt diode associated with a frequency multiplier to go beyond 0.2

THz

 Oscillator fixed or variable between 0 and 0.02 THz associated with a frequency multiplier,

 Frequency conversion by beat of two lasers in the near infrared, - HT tube called BWO (acronym for Backward-wave oscillator),

 Excitation of a gas cavity by a far infrared laser or

 Quantum Cascade Laser (QCL).

 Chernomyrdin Nikita et al: "Wide-aperture Aspherical Lens for High-resolution Terahertz Imaging" (Review of Scientific Instruments, AIP, Melville, NY, US, Vol 88, No. 1, January 12, 2017, XP012215296, is known from ISSN: 0034-6748, DOI: 10.1063 / 1 .4973764.

 In this document, the signal generated by the sources usually used in the THz domain has a very pure coherence, the width of the emission line is very fine. This coherent radiation is not conducive to image formation because interference fringes are generated with a very large amplitude on each system allowing it. Typically, these sources have an accuracy of 0.001 Hz with a bandwidth of 1 kHz to -118dBc / Hz.

 Using a coherent source in an imaging system while the scene to be observed is not perpendicular at any point in the field generates an amplitude modulation proportional to modulo λ / 4 altitudinal variations. Which is detrimental to the quality of the image formed.

SUMMARY OF THE INVENTION

 The present invention aims to remedy all or part of these disadvantages.

 For this purpose, according to a first aspect, the present invention aims a device for capturing point values for constituting an image, which comprises:

an incoherent source of rays whose frequency is between 0.075 THz and 10 THz to illuminate an object, a radiation sensor originating from the object, which comprises a zone sensitive to the radiation coming from the source and which emits an electrical signal representative of the intensity of the rays coming from the source reaching the sensitive zone of the sensor and

at least one less than one aperture focusing optical system (F-Number) located on the optical path of the rays emitted by the source from the source to the ray sensor passing through the object.

 Thanks to these provisions, the imaging device is not subject to the presence of standing waves due to the inconsistency of the source. The inventor has determined that this combination of optical and electronic means makes it possible to perform precise point measurements, the relative displacement of these means and the object making it possible to produce linear or matrix images.

 The source used has a bandwidth of several GHz, preferably:

- between 100 and 200 GHz the bandwidth is 12 GHz to -1 OOdB

 - between 220 and 320 GHz the bandwidth is 15 GHz to -1 OOdB

 - between 560 and 640 GHz the bandwidth is 15 GHz to -1 OOdB.

 In embodiments, the source illuminates the object with a broad enough emission spectrum to scan the standing wave in a shorter time than the acquisition time of the sensor.

 In embodiments, at least one optical system has an aspheric optical lens.

 In embodiments, the dispersion (in percentage) of the indices of the materials used on the frequency ranges of the source is less than 1%.

 For example, the dispersion in HDPE (high density polyethylene) is of the order of 0.5% over the frequency range and, more particularly:

 - 0.2%, 100 to 300 GHz and

 - 0.5% from 100 to 700 GHz

 The aspheric lens simplifies optical design, compared to the use of parabolic mirrors.

 In embodiments, at least one optical component of an optical system has an antireflection coating with cone or crater-like microstructures.

 The advantage of this method is that it makes it possible to perform anti-reflective treatments very broadband and insensitive to the orientation of the surface.

In embodiments, at least one optical system includes an optical lens and wherein the incoherent ray source is configured to illuminate the entirety of the optical lens closest to said source. In embodiments, the transmission frequency of the incoherent source is modulated.

 In embodiments, the incoherent source includes a noise source of thermal noise type in a resistor or impedance diode.

 In embodiments, the device comprises a proximity electronics for biasing a nano-transistor comprising the photosensitive zone, by a gate voltage close to its swing voltage where the conventional operation of the transistor is the most nonlinear.

 In embodiments, the rectified signal from the nano-transistor is amplified by forcing asymmetry of the charges in the nano-transistor channel by injecting a current in the transistor channel, between the drain and the source and / or using metallic patterns acting as antennas.

 In embodiments, the rectified signal is a continuous potential difference between the drain and the source of the nano-transistor measured in common or differential mode.

 In embodiments, the proximity electronics comprise a compensation circuit of the offset generated by the injection of the current between the drain and the source, for example by using a subtractor mounting.

 In embodiments, the photosensitive area is a nano-transistor, the signal generated by the radiation THz is a continuous potential difference between the drain and the source of the nano-transistor measured in common or differential mode.

 In embodiments, the device comprises a proximity electronics for biasing the nano-transistor by a gate voltage close to its swing voltage where the conventional operation of the transistor is the most nonlinear.

 In embodiments, the rectified signal from the nano-transistor is amplified by forcing asymmetry of the charges in the nano-transistor channel by injecting a current in the transistor channel, between the drain and the source and / or using metallic patterns acting as antennas.

 This increases the sensitivity of the device.

 In embodiments, the device comprises at least one low noise and low drift amplifier that amplifies the signal on the dynamics of an analog digital converter.

 In embodiments, the device comprises means for synchronizing the demodulation of the signal with the amplitude modulation signal of the source.

This provides a maximized signal-to-noise ratio by eliminating all additive noises at a frequency other than the modulation frequency of the source. In embodiments, the device comprises means for synchronizing the digital output of the device with the THz source.

 The sensor and the source are synchronized by a digital clock signal. The source is modulated in amplitude by this clock signal which makes it possible to measure and compensate for the signal residue (offset) present at the terminals of the sensor in the absence of THz. The compensation is both digital by high frequency subtraction of the signal measured with THz and without THz, but also by adjusting the voltage to subtract from the sensor signal to maintain the overall level within a certain voltage range.

 This architecture guarantees greater sensitivity, improved signal-to-noise ratio.

 In embodiments, the device comprises an encoder of the absolute positions of the object to be imaged, these measured position data being synchronized with the data from the sensor for use over the entire stroke of the displacement of the object. , including in the zones of accelerations, and an exploitation of the displacement back and forth of the object, by the setting of the even and odd lines avoiding a phenomenon of shearing.

 A averaging of the data calibrated on the absolute positions and not on the time allows an adaptation at any speed of displacement

 In embodiments, the device according to the invention comprises a splitter plate, the polarization incident on the splitter plate being the polarization in TM mode (acronym for transverse mode for transverse mode) in which the field is parallel to the plane of impact.

 The separating blade is, for example, a blade made of Si-HRFZ whose thickness is calculated to ensure optimal separation. Thanks to these provisions, it is possible to perform measurements in reflectometry. Polarization in TM mode provides better stability of the separation efficiency when the thickness of the separating blade varies in its manufacturing tolerance.

 In embodiments, the device includes means for measuring the intensity of radiation from the object, and / or polarization of the rays from the object, and / or the difference in ray steps from the object.

 The device object of the invention thus allows several types of measurements.

BRIEF DESCRIPTION OF THE FIGURES

Other advantages, aims and features of the present invention will emerge from the description which follows, for an explanatory and non-limiting purpose, with reference to the appended drawings, in which: FIG. 1 represents, schematically and in section, a first embodiment of an incurent terahertz source,

 FIG. 2 represents, schematically and in section, a second embodiment of an incurent terahertz source,

 FIG. 3 represents, schematically and in section, a first embodiment of an imaging objective implemented in devices that are the subject of the invention,

 FIG. 4 represents, schematically and in section, a second embodiment of an imaging objective implemented in devices that are the subject of the invention,

FIG. 5 represents, schematically and in section, a third embodiment of an imaging objective implemented in devices that are the subject of the invention,

FIG. 6 represents, schematically and in section, a fourth embodiment of an imaging objective implemented in devices that are the subject of the invention,

FIG. 7 represents, schematically and in perspective, a first means of linear illumination of an object, implemented in devices that are the subject of the invention; FIG. 8 represents a non-uniform illumination curve;

 FIG. 9 represents an emission curve of a light source to compensate for the non-uniform illumination shown in FIG. 8,

 FIG. 10 represents an optical system providing substantially the emission curve illustrated in FIG. 9,

 FIG. 11 represents a substantially uniform illumination curve provided by the optical system illustrated in FIG. 10, with the linear illumination means illustrated in FIG. 7,

 FIG. 12 represents, schematically and in perspective, a second linear illumination means having non-symmetrical surfaces,

 FIG. 13 represents, schematically and in section, a first particular embodiment of the device which is the subject of the invention,

 FIG. 14 represents, schematically and in section, a second particular embodiment of the device which is the subject of the invention,

 FIG. 15 represents, schematically and in section, a third particular embodiment of the device which is the subject of the invention,

 FIG. 16 represents, schematically, a point sensor,

 FIG. 17 represents, schematically, a multi-pixel sensor and

 FIG. 18 represents an electronic diagram for a THz microscope. DESCRIPTION OF EMBODIMENTS OF THE INVENTION

As of now, we note that the figures are not to scale. In the image capture systems that are the subject of the invention, refractive optics (lenses) are used, which have greater modularity than mirrors, and give access to the use of powerful optical design tools. computer assisted.

 In the frequency range 100-700 GHz, the preferred field of use of the present invention, the indices of the materials (high density polyethylene (HDPE), polymethylpentene (PMP), polycarbonate (PC), silicon HRFZ (Si)) are constant. within a few percent, or less as discussed above, for HDPE. The systems that are the subject of the invention therefore do not have a problem of chromaticity, that is to say that there is no change in the optical behavior of the diopters when the wavelength changes.

 Preferably, the imaging device that is the subject of the invention uses an incoherent terahertz source (THz) 100.

 An incoherent THz source 100 is, according to a first example, constituted, as illustrated in FIG.

 a broadband TeraHertz 105 (f <200 GHz) transmitter (0.1 GHz <Ai <40 GHz), a filter 1 10 for limiting the emission band,

 amplifiers or attenuators 1 15,

 multipliers 120 of frequency and

 of a transmitting antenna 125.

 An incoherent THz source 100 is, according to a second example, constituted, as illustrated in FIG. 2:

 a transmitter 130 in the low band, that is to say in the frequency range from 0.1 GHz to 40 GHz, emitting on a narrow band (f <1 GHz),

 a modulator 135 on a band 0.1 GHz <Af <40 GHz,

 amplifiers or attenuator 140,

 - frequency multipliers 145 and

 a transmitting antenna 150.

 The emitter 110 or 130 generates an incoherent signal such that, in a time shorter than the acquisition time of the sensor, the emitter emits a sufficiently broad emission spectrum to scan the standing wave.

 The source used has a bandwidth of several GHz, preferably:

- between 100 and 200 GHz the bandwidth is 12 GHz to -1 OOdB

 - between 220 and 320 GHz the bandwidth is 15 GHz to -1 OOdB

 - between 560 and 640 GHz the bandwidth is 15 GHz to -1 OOdB.

In embodiments, the transmitter 110 or 130 is a noise source of thermal noise type in a resistance or impedance diode. In embodiments, the transmitter 110 or 130 is a tunable source such as a variable frequency oscillator or JIG.

 The modulator 135 is tuned so that, once multiplied by each frequency multiplier 145, the generated signal has a transmission bandwidth preferably in the 0.1 GHz <Af <10 GHz range. Note that the modulator 135 is flexible over a very narrow band (for example 200 MHz).

 The amplifiers 1 15, 140 or attenuators are used to adjust the input power of the signal before each frequency multiplier 120 or 145. The multipliers, amplifiers and attenuators can use structures on PCB (acronym of printed circuit board circuit board) or well guided (at the frequencies considered, we can manipulate these electromagnetic waves confined in metal waveguides).

 The choice of the antenna 125 or 150 makes it possible to parameterize the optical properties of the beam coming out of the last multiplier 120 or 145: polarization, TEM mode (transverse electromagnetic mode for transverse electromagnetic mode), divergence and size of the emission point.

 The source presented in FIGS. 1 and 2 uses an emission spectrum which ensures a scan at least equal to λ / 4, which makes it possible to reduce the sensitivity to standing waves in the image while allowing the choice of the working frequency and the polarization.

 For imaging optics the spatial resolution is related to the size of the smallest spot that can be focused through the system. The radius of the spot is given by the law of Airy Rdiff = 1, 22 λ ON, in which λ is the wavelength and ON the numerical aperture.

 The spatial resolution is therefore proportional to the wavelength λ, and to the numerical aperture ON used in the focusing system. This ON opening is calculated by means of the diameter D of the focusing optics and the focal length f of this optic ON = f / D.

 For a fixed wavelength λ, it is therefore possible to optimize the resolution of the system by choosing a larger optical diameter D, or a shorter optical distance f.

 The numerical aperture ON is physically limited to a value of 0.5. But in the air with realistic optical indices (of a value of 1, 5 for example), it is very difficult to reach openings smaller than 0.7.

 To maximize the diameter D of the beam, the divergence of the source is taken into account in order to choose the focal length f of the first lens so that the beam completely fills this first lens. Thus the diameter D of the beam is equal to the diameter of the first lens, and the numerical aperture parameter ON is optimized.

 Preferably, the imaging device that is the subject of the invention uses aspherical lenses (FIGS. 3 to 6 and 13 to 15).

We recall here that a spherical lens is a lens whose shape of each surface follows a sphere. When using spherical lenses over large diameters, it appears an aberration related to the curvature of the lens, called "spherical aberration". Marginal rays (passing near the edge of the lens) do not focus in the same place as the paraxial rays (passing in the center of the lens). This effect counterbalances the gain in resolution with the increase of the diameter of the beam, because the focused beam with spherical aberration no longer follows the Airy law mentioned above. In addition, the introduction of this spherical aberration disrupts the entire propagation of the THz beam by creating rings of light instead of a point of focus ("spot") well circular: there is a loss of energy in addition to 'a loss of information.

 There are several solutions to solve this phenomenon, the simplest is to use so-called aspherical lenses. For an aspheric lens the shape of the curvature follows a sphere at the center, but out of the center, the curvature is corrected to cancel the spherical aberration. In visible and infrared optics, these lenses are very expensive to manufacture because of their complex shape.

 In THz, given the manufacturing methods used, it is as simple to manufacture a spherical lens as an aspherical lens.

 3D printing and / or micromachining with a residual roughness between 0.02 and 0.1 mm is carried out. The measurement of the source propagation parameters makes it possible to define the illumination at the input of the system: divergence, distribution of the energy

 In embodiments, the dispersion (in percentage) of the indices of the materials used for the lens, on the frequency ranges of the source, is less than 1%.

 For example, the dispersion in HDPE is of the order of 0.5% over the frequency range and, more particularly:

 - 0.2%, 100 to 300 GHz and

 - 0.5% from 100 to 700 GHz

 The system is optimized by varying the distances between the optics, the radius of curvature of the surfaces as well as the conicity coefficients (hence the asphericity)

 The system optimization criterion can be defined for example on the minimization of the spot size at several points of the field, or the minimization of the wavefront difference at several points of the field.

 The wavelength of the beam being millimetric, a roughness of 1/100 is sufficient to ensure good transmission efficiency. These roughnesses are compatible with microfraises used on modern lathes and milling machines and 3D wire deposition printing.

In the design of the imager all lenses are therefore aspherical designed to achieve the maximum optical performance in terms of spherical aberration and limit the number of lenses used. In the case of point imaging devices, the focus lenses are designed to work in infinite focus mode, ie the incident beam must be collimated (parallel rays) and its diameter must be the same. than the active diameter of the lens. At the focus lens output, the focal point where the beam is the smallest is defined and then the beam diverges again is captured by another lens.

 Thanks to the use of high numerical aperture focusing lenses, images are made with the highest spatial resolution allowed by physical laws, with a view of details of dimensions smaller than the wavelength used.

 Concerning the antireflection treatments of the lenses and, where appropriate, of the separating plate, their structure consists of a surface structure on the refractive optics which has the effect of reducing the reflection of each diopter by acting on the optical index at the surface of the lens.

 A first method consists in depositing one or more layers of dielectric material whose optical index is lower than the optical index of the treated lens, for example the optical index of the treatment material is close to the square root of the optical index of the lens. The thickness of the layer to be deposited is of the order of a quarter of the average wavelength used. This thickness of the treatment material is relatively large in THz, compared to the infra-red, which poses problems on the deposition technique.

 A second method, preferably used in the imaging device of the invention is to form cone-shaped or crater-shaped microstructures using a micro-machining method, for example "laser drilling" ( laser milling) using a second femto laser. These microstructures of dimensions smaller than the wavelength, create an index gradient apparent on the surface of the diopters which cancels the reflection on the face because there is no longer the index discontinuity that causes the reflection. The advantage of this method is that it makes it possible to perform anti-reflective treatments very wide band and less sensitive to the orientation of the surface, unlike the case of layer deposition.

 The use of this type of treatment increases the energy efficiency of multi-lens optical systems by increasing the transmission of each optical diopter. Thanks to antireflection treatments, the energy projected on the sensor is increased by several tens of percent. Antireflection treatments also help eliminate unwanted images and standing waves.

 In the case of the use of a point sensor 600 in the imaging device of the invention, its structure is constituted, as represented in FIG.

a THz detector 605. The THz detector is an RF nano-transistor in GAAS, GAN or INP in a technology less than 0.25 μηι which by plasma effect in its channel straightens the THz waves in the 0.1 THz bands. at 2.5 THz; - A proximity electronics 610 for biasing the detector. The nano-transistor to be an efficient rectifier must be biased by a gate voltage close to its swing voltage where the conventional operation of the transistor is the most nonlinear. The rectified signal is amplified by forcing asymmetry of the charges in the channel by injecting a current in the transistor channel between the drain and the source, which has the effect of increasing the sensitivity of the transistor;

 a proximity electronics 615 for shaping the signal THz. The rectified signal is a continuous potential difference between the drain and the source of the nano-transistor measured in common or differential mode. The proximity electronics 615 has a high impedance so as not to attenuate the amplitude of the signal and to force the injection of the bias current into the transistor and not into the amplification circuit. A subtractor assembly compensates for the offset generated by the current injection between the drain and the source. One or more low noise and low drift amplifiers amplifies the signal on the dynamics of the Digital Analog Converter ("CAN") 620;

 an electronic device 620 for digitizing the signal. The analog-to-digital converter has a large dynamic amplitude (> 12 bits) and is fast (> 100 thousand samples per second);

 a central processing unit 625 implementing a signal demodulation algorithm. In order to limit the noise BF and HF ("offset") the useful signal is placed at a frequency greater than 100 Hz by modulating the THz source. The useful signal at the output of the sensor is derived from the demodulation of the signal received by the electronics.

 The function of the point sensor 600 and its proximity electronics is to straighten the THz wave and to provide a suitable impedance signal for the following electronic stages. In the electronics 615, the subtracter followed by an amplifier, or the differential amplifier, makes it possible to amplify the useful signal differentially with respect to the offset signal generated by the injection of the current into the transistor.

 The signal is digitized at a high rate (acquisition frequency> 1 kHz) as close as possible to the analog output. Signal demodulation is performed synchronously with the source modulation signal, which provides a maximized signal-to-noise ratio by eliminating all additive noises at a frequency other than the source modulation frequency. The digital output of the sensor is synchronized with the absolute positions of the object to be imaged.

With regard to the electronics of the single or multi-pixel sensor, for the analog circuit, an injection of a current into the channel of the transistor is carried out. To increase the sensitivity of the THz sensor based on HEMT, a current of the order of a few tens of μΑ is injected into the transistor channel. This has the effect of increasing by several orders of magnitude the electrical response in the presence of THz radiation. The injection is done by a current generating component directly connected to the transistor by its drain, the component may be for example a LM334 associated with a resistor, the choice of the value of the resistor allows to set the amount of current injected.

 The useful signal at the terminals of the transistor in the presence of radiation THz being the voltage between its drain and its source, it is necessary to isolate the transistor and the component injecting the current from the rest of the amplification chain by a follower mounting having a very strong impedance. Thus the injected current can propagate only in the transistor and not in the amplification chain.

 Analog correction of the offset generated by the current

 Since the resistance of the transistor channel is of the order of kOhm, the injection of the current into the channel generates a continuous offset across it. This offset is present whether there is THz or not. This offset can be of the order of several hundred mV which is problematic for the amplification chain. Indeed the amplifiers amplify as much the useful signal generated by the THz as the continuous offset: beyond a certain level the offset causes a saturation of the inputs of the different amplifiers. It is therefore necessary to subtract a DC voltage of the same order of magnitude as the offset voltage to the signal generated by the transistor to allow the amplification chain to remain in its input operating range.

 Amplification

 The amplification is chosen such that the noise generated by it is low, it has a drift as low as possible, and a sufficient bandwidth for the working frequencies. For example, an OPA735 amplifier from Texas Instruments has the required performance. Two amplification stages are conventionally chosen: the first of fixed gain of 10 or 100 near the sensitive surface, which ensures the transport of a signal several orders of magnitude higher than the level of noise induced by the circuit. Then the second variable gain amplifier (for example between 1 and 10) near the analog-digital converter (ADC). This allows the total to reach a maximum amplification factor of 1000 while being able to adapt the amplification to fill the conversion range of the ADC.

FIG. 18 shows a transistor 815 receiving a gate voltage 805 and a current injection 810 on its drain, the voltage on the drain being supplied to a follower 820. At the output of the follower 820, a subtractor 830 subtracts a voltage 825, then two amplifiers 835 and 840 amplify the signal. Signal modulation / demodulation

 The noise generated by the transistor increases greatly when current is injected. It is no longer interesting to operate in continuous mode (DC) because the noise level is too important, especially the low frequency noise related to the offset voltage. However, by placing the useful signal at a frequency of the order of kHz, a signal-to-noise ratio greater than 60 dB is obtained, compared to approximately 45 dB in continuous mode without injected current. The advantage of current injection is in this signal to noise ratio which is better with the current for a working frequency between 100Hz and 100kHz.

 There are several methods for modulating and demodulating a signal, particularly that of synchronous detection which is widely used in THz systems. This method has a disadvantage: it requires long integration times which greatly reduce the acquisition speed.

 In the present case, the procedure is as follows: the source is modulated externally by a programmable electronic component (for example an FPGA) which controls a radiofrequency amplitude modulator. The programmable component controls the high and low states of the modulation: on one, the THz is emitted by the source; on the other, no signal is emitted. Alternatively, the signal received by the sensor is either the useful signal related to the presence of THz radiation, or the background noise in the absence of THz. The programmable component samples the signal from the sensor during the high and low states, it produces two averages: one corresponding to the average signal level on the high state, the other at the average signal level on the state low. By subtracting these two averages, we obtain the signal corresponding to the filtered THz alone of its low frequency components thanks to the difference between high and low, and filtered from its high frequency components thanks to the average over each high and low state. It is this averaged differential signal that is produced at the output of the sensor.

 The advantage of this method is that high acquisition rates can be maintained provided fast ADCs (200,000 samples per second, for example) have adequate amplification bandwidths. Moreover this method is inexpensive in computations, which makes it possible to implement it for a whole matrix of detector, the computation being made independently on each sensor.

 This architecture ensures greater sensitivity, reduced signal noise and signal synchronization with the optimal object position.

 In the case of the use of a linear sensor having one or more lines of photosites (or pixels), this multipixel sensor 700 is constituted, as illustrated in FIG. 17:

one or more rows 705 of THz detectors 710. Each detector THz is a nano-RF transistor in GAAS, GAN, or INP in a technology less than 0.25 μηι which, by plasma effect, in its channel straightens the waves THz in the 0.1 THz bands at 2.5 THz. Each nano-transistor is packaged in a micro surface packaging <0.5 mm ² ;

 - A proximity electronics 715 for polarizing each detector. A nano-transistor for being an efficient rectifier is biased by a gate voltage close to its swing voltage where the conventional operation of the transistor is the most nonlinear. The rectified signal is amplified by forcing asymmetry of the charges in the channel by injecting a current between the drain and the source; a proximity electronics 720 for formatting the THz signal. The rectified signal is a continuous potential difference between the drain and the source of the nano-transistor measured in common or differential mode. The proximity electronics 720 has a high impedance so as not to attenuate the amplitude of the signal and to force the injection of the bias current into the transistor and not into the amplification circuit. A subtractor assembly compensates for the offset generated by the current injection between the drain and the source. One or more low noise and low drift amplifiers amplifies the signal on the dynamics of the Digital Analog Converter (ADC) 725;

 an electronic 725 to digitally parallel the signal. This digital-to-digital converter 725 has a large dynamic amplitude (> 12 bits) and is fast (> 100 kSamples / s or thousands of samples per second);

a system 730 for acquiring and processing the signal for noise reduction.

 In the same way as in the case of the point sensor, the THz source is modulated in amplitude which makes it possible to perform an averaged differential measurement.

 a high-speed communication system 735 via an Ethernet, USB or cameralink communication protocol (registered trademarks).

 The sensor and the source are synchronized by a digital clock signal. The source is modulated in amplitude by this clock signal which makes it possible to measure and compensate for the signal residue (offset) present at the terminals of the sensor in the absence of THz. The compensation is both digital by high frequency subtraction of the signal measured with THz and without THz, but also by adjusting the voltage to subtract from the sensor signal to maintain the overall level within a certain voltage range.

In the case of the monopixel sensor, the absolute positions of the object to be imaged are measured by an encoder. These position data are then synchronized with the data from the THz sensor for use over the entire travel path including in the acceleration zones, but also the exploitation of the outward and return displacement by the calibration of the lines. even and odd, avoiding a shear phenomenon. And lastly, a averaging of the data calibrated on the absolute positions and not on the time allows an adaptation at any speed of displacement This architecture guarantees the highest sensitivity, reduced signal noise and synchronization of the signal with the optimal object position.

 The multipixel 700 sensor makes it possible to enlarge the field of view spatially and thus to produce high resolution images with a high refresh rate (line frequency> 1 kHz). Each sensor 700, including its proximity electronics is in charge of straightening the THz wave it receives locally and provide a suitable impedance signal for the following electronic stages. In electronics 720, each subtracter circuit followed by an amplifier, or differential amplifier, amplifies the useful signal.

 The signal is digitized at a high rate (acquisition frequency> 1 kHz) as close as possible to the analog output.

 The signal demodulation is performed synchronously with the source modulation signal on the signals from each pixel, providing a maximized signal-to-noise ratio by eliminating all additive noises at a frequency other than the modulation frequency of the signal. source.

 The multipixel sensor 700 is located in the focal plane of the imaging optical system. Each point sensor is in charge of straightening the THz wave, amplifying the useful signal. The signal is digitized at a high rate (acquisition frequency> 1 kHz) as close as possible to the analog output

 The small surface of the sensors allows a great compactness and thus it results a high spatial resolution. The photosensitive components are arranged in line with, for example, a space between two components of the order of 300 μηι and an interpixel of 800 μηι, each component having a dimension of 500 μηι × 1 mm. The individual polarization of each sensor makes it possible to guarantee uniform sensor-to-sensor sensitivity while maintaining high sensitivity. Parallel management of all signals ensures a very high acquisition speed. Embedded signal processing provides real-time filtering for the best signal-to-noise ratio. Communication via an Ethernet, USB or cameralink communication protocol allows packet management to send high-speed lines to link several muiti pixels sensors without loss of data.

 The starting point for the design of linear sensor imaging systems is the spatial resolution imposed by the sensor. The imaging device object of the invention has, in its multipixel embodiments, an inter-pixel of 800 μηι, and according to the model, a line of 64, 1 28 or 192 pixels. By choosing the field of view on the side of the object to be imaged, the optimal spatial resolution of the sensor and lens assembly is imposed.

For example, in the case of a 30mm field of view on the object side, it is necessary to ensure a correct sampling of the scene according to Shannon criteria (four pixels for a pair of lines at the maximum resolution). Thus the optimal spatial resolution of the camera (sensor and lens) is 0.53 pl / mm (pairs of lines per millimeter) in frequency. A higher optical resolution of the system could not be sampled correctly by the camera, a lower optical resolution would lead to having an excessive pixel / mm number so not to benefit from all the available resource. This approach also directly indicates the magnification required for the optical system, here 1, 71 x.

 Following this prerequisite, one is able to give a performance target to the optical system. Knowing the working frequency, for example 255 GHz, it is calculated by the law of Airy that the objective to be designed must have a numerical aperture object of 1, 31, a numerical aperture image side of 2.23. We can also estimate the dimension of the diameter of the objective necessary to reach this numerical aperture: here 75 mm.

 Given the optical parameters determined above, we see that the objective to be designed to answer the problem is a microscope-type objective. In fact, contrary to the objective of photography or to a telescope (distant telescope), the microscope objective has the following overall characteristics: the scene to be imaged is smaller than the size of the sensor, the pixels of the sensor are of the same order of magnitude as the average details of the scene. The reproduction ratio is much higher than one (a scene of 100 μηι x 100 μηι will be reproduced on a 10 mm x 10mm sensor: magnification of 100). The objective works at a short distance from the object, and its object aperture can reach values such as 0.7.

 As regards the use of a microscope objective, it is recalled that microscopic optics makes it possible both to grow the object to be observed, and to separate the details of this object so that it is observable. It is the combination of magnification and resolving power that characterizes such an optical system.

 In the context of microscope objective imaging systems in THz imaging, on the one hand the scene is generally smaller than the size of the detector array, and the detector is on the order of magnitude of the scene details it is therefore necessary to have a magnification of the scene and its details to image it. On the other hand, the image of the scene must be formed by ensuring sufficient resolution of the object to be able to observe its details, which generally implies working with large openings, so close to the object and with large diameter optics compared to the stage.

 To answer the problem of THz field imaging, the microscope objective is the best candidate:

 the pixels of the sensor are large compared to the details of the image to be formed,

 the sensor is large:> 150 mm for 192 pixels and

the resolution to be achieved requires very large openings. The consequence of this choice is that the lens is large, of a similar size, or larger than the objects to be imaged.

 The results in terms of the number of lenses of the optics, their shapes, and their dimensions greatly depends on the desired magnification, and the expected resolution, as well as the draw of the objective (distance between object and first lens)

 The lenses have at least two lens groups:

 1st group: near the object, allows the collection of the light coming from the object with the targeted opening and

 - 2nd group: close to the image, allows to form the image of the object with the target image opening.

 For weakly constrained objectives (eg two magnification, and low wavelength), it is possible to design a lens on this principle with only one lens per group, so only two lenses in all. This is made possible in particular by the ability to manufacture aspherical lenses.

 For more constrained objectives (for example, magnification of 1 and high wavelength) it is necessary to use more than one lens per group:

 - field compensator in the first group to limit the field curvature and

 - two or three lenses in the second group to limit the thicknesses and the aberrations

 In some embodiments, the lens actually has three lens groups, the second group being divided into two separate subgroups of a large air space, which is particularly the case when the lens is telecentric. side image.

 Examples of realization

 Example 1, Figure 3: 40 mm object field, magnification 1, 3x, 235 GHz

 The objective 200 has two groups 205 and 210 of a lens each, it is not telecentric side image. This 200 lens has been designed with the constraint of an object print larger than 50 mm. The MTF (acronym for "modulation transfer function" for modulation transfer function) shows a resolution of 0.4 pl / mm, ie 1.92 mm of resolution relative to the object. The spot diagram shows that the system is limited by diffraction on the whole field: the image is not disturbed by the aberrations.

 Example 2, Figure 4: object field 30 mm, magnification 1, 7x, 235 GHz

The objective 215 has three lenses 220, 225 and 230, the second group being divided into two subgroups of a lens. Telecentric lens on the image side. The MTF image shows a resolution of 0.45 pl / mm, ie 1.3 mm of resolution reported to object. The spot diagram shows that the system is limited by diffraction on the whole field: the image is not disturbed by the aberrations. Example 3, Figure 5: object field 52 mm, magnification 1 x, 140 GHz

Goal 235 has a second group divided into two subgroups. Four lenses 240, 245, 250 and 255 are necessary to obtain the magnification of 1 and the telecentricity on the image side. The MTF image shows a resolution of 0.40 pl / mm, or 2.5 mm resolution reported object. The spot diagram shows that the system is limited by diffraction except at the edge of the field (slight defect).

 Example 4, Figure 6: 50 mm object field, magnification 3.1, 235 GHz

 The objective 260, of great increase in two groups, with a compensator of field in the first group. This objective, with three lenses 265, 270 and 275 makes it possible to reach resolutions below the wavelength. The image-wise MTF shows a resolution of 0.41 pl / mm, ie 780 μηι of resolution referred to object. The spot diagram shows that the system is limited by diffraction except at the edge of the field (slight defect).

 The THz imaging lens is used to image an object in the field to an image plane located opposite the lens. The condition for operating in the specifications is to have a THz signal input sufficient for the detection sensitivity in the image plane. In a conventional embodiment, this objective is integrated in a string including a source, a projection system of the illumination of the source, the objective itself, a multi-pixel sensor, and a computerized analysis system (no represent).

 The objective is therefore an element is key in the chain described above because it is he who ensures the optical spatial resolution on the object to be imaged. The dimensions of the lens are intrinsically related to its performance: the higher the desired resolution, the larger the lenses, and the length of the lens is important.

 The use of a line of pixels as a sensor requires the use of lighting on a line of the scene in order to focus the energy of the source on the part of the object observed by the linear sensor. The source emits radiation on an emission cone determined by its horn. The distribution of energy in the emission cone presents a non-uniform and circular energy distribution because it is generally of Gaussian distribution and symmetry of revolution. The linearization system of the source has two objectives:

 - to make the illumination of the linear source and

 make the illumination of the source uniform on the stage.

In embodiments, the illumination of the scene is designed parallel in one direction, and focused in the perpendicular direction. The use of a cylindrical lens 290 is a solution to achieve this goal. The beam 285 from the source is enlarged to the same diameter as the scene to be illuminated, then with a lens whose curvature is aspherical with a cylindrical geometry (and not with a geometry of revolution) the radiation is focused in a single direction of the space, as shown in Figure 7. The linearization 280 by the preceding method has a non-uniform illumination because the distribution of the illumination of the source is not. The focusing of the energy distribution of the source in a single direction of space keeps the same inhomogeneous distribution carried along the line of illumination.

 FIG. 8 represents a power distribution 300 at source output 100.

 With optical design software, it is possible to calculate the optimal revolution geometry distribution to ensure uniform illumination once transformed by the cylindrical lens. Since this distribution is known, the software uses it as a target for optimizing the beam shaping lenses at the output of the source 100. For example, an illumination having a distribution 305, illustrated in FIG. 9, forms a uniform distribution after a cylindrical lens.

 To achieve a uniform illumination on the line observed by the linear image sensor, in two design steps, a three-lens system 310, 315, 320, 325 is formed:

 during a first step, the distribution 305 necessary for the entry of the cylindrical lens 325 is calculated,

 during a second step, optimization is made of a system with two aspherical lenses 315, 320 to form this distribution 305.

 FIG. 11 illustrates the distribution 330 resulting from the implementation of the optical system 310.

 In another linearization method, the two beamforming lenses 315 and 320 are replaced by a single lens, having at least one non-symmetrical surface of revolution (sphere, asphere), but representing splines or polynomials. In this case, a set of two lenses 335 and 340 makes the illumination linear and uniform, as illustrated in FIG. 12:

 the lens 335 is used for shaping the beam to prepare the energy distribution (XY polynomial surface lens, for example)

 the aspherical cylindrical lens 340 achieves the focusing of the beam in only one direction.

The advantage of this method is that it is simpler to implement in a computation software: a macro of passage of the Gaussian circular illumination of the source to the uniform linear illumination is directly implemented at the level of the optimization without going through an intermediate step of calculating the illumination necessary for the entry of the cylindrical lens 340. The computation of the macro, complex for a system with symmetry of revolution, breaks this symmetry from the first lens. This brings the advantage of reducing the number of lenses needed. The design of the illumination takes into account the source used (wavelength, divergence). The illumination system projects enough THz energy onto a plane of space and then uses a collection lens and a sensor to form an image of an object placed in that plane of space.

 The linearization system allows you to choose how to distribute the energy from the source to illuminate the scene most effectively. A system benefiting from this type of illumination has the greatest energy efficiency, which makes it possible to use less sensitive sensors, or to make images of very opaque or low-reflective objects.

 In another embodiment, the illumination may be a rectangle whose widths and heights may be chosen by the designer. In this case the illumination is collimated on the object to be studied which allows on the one hand to have a greater depth in which the illumination is optimal and on the other hand to have the exit pupil of the system. illumination to infinity to avoid having a part of the image of the source that is superimposed with the image of the object.

 To achieve a rectangular illumination is carried out according to the second method described above: a photometric calculation to determine a macro passage between a Gaussian illumination of revolution and uniform rectangular illumination (or rectangular with a particular intensity profile. , we impose the angle of incidence of zero point rays on the object to make the illumination collimated). This macro is implemented in the design software to optimize two non-symmetrically revolving surface lenses that generate rectangular illumination at the object as defined by the designer.

 In the case of the use of a point sensor (with a single photosite or pixel), the principle of the point imager rests on two groups of lenses:

 a first group realizes the shaping of the beam to reach the desired resolution and

 a second group carries out the collection of the stream transmitted through the object to be imaged. In this imaging mode, the spatial resolution is only related to the first optical group located before the object. The stream collection part has no impact on the spatial resolution of the formed image. The optimization of the second group is based on the principle of energy flow collection: the goal is to concentrate the energy on the sensor as much as possible in order to obtain the largest signal-to-noise ratio.

 Two imaging modes are used depending on the types of objects to be imaged.

 The transmission mode, through the object: transmitted measurement of the intensity, and / or polarization, or difference in operation by the sample.

The reflection mode on the object: reflected measurement of the intensity, and / or polarization, or difference in the surface or depth of the sample. The proposed punctual imager makes it possible to make images of the object in reflection and in transmission simultaneously. For this purpose, it uses two flow collection groups located upstream and downstream of the object. It can also control the amplitude of the source for calibration to perform absolute measurements.

 In addition to its optical components, the point imaging system can integrate the following elements:

 A source THz CW or modulated

 One or more point or multi-pixel sensors (linear)

 A Cartesian table to move the sample with at least one encoder on the fast axis.

 A computer system that creates the image by synchronizing with the absolute position of the Cartesian table the signal synchronized with the modulation of the source. The maximum resolution is obtained for a focusing lens with a very short focal length, of the order of 10 or 20 mm. As a result, the space available for inserting the sample may be very limited for large samples. The proposed imager presents a solution to this problem thanks to a modular system allowing to change resolution according to the dimension of the object to be imaged.

 The beam from the source is first collimated by a first lens. The collimated beam propagates with parallel rays as a result of this lens, which makes it possible to place the lens that focuses the radiation at any distance from the collimating lens.

 The plane of focus on the object is chosen such as a short focal lens allowing a high resolution, and thus placed close to the object, or a lens of long focal length allowing a lower resolution but making it possible to measure a larger object. voluminous, the lens being placed close to the source, focus the beam at the same place on the optical axis.

 Thanks to this system mechanical integration is facilitated and the ergonomics of the system are improved because the resolution of the system can be easily changed by changing the focusing lens but keeping the same distance between the source and the object.

 In transmission mode, the collection of flows is done by a group of two lenses placed after the object. The collector used has a numerical aperture at least equal to that of the first group, to ensure maximum efficiency.

The first lens may be identical to that of focusing. It collects the flux coming from the object, and makes the beam with parallel rays (collimated). The modularity of resolution described above is preserved thanks to this system by allowing to dispose the second lens at any distance from the first. The second lens in front of the detector focuses the THz beam on the photosensitive surface by concentrating the energy to maximize the signal-to-noise ratio. The aperture used for this lens is for example 1, 25. This value can be chosen to make a compromise between energy concentration (lower aperture value), and focusing stability: a larger aperture makes it possible to make the focusing on the sensor less dependent on any beam distortions associated with the images. mechanical dilatations or the simple disturbance of the beam by the object.

 In reflection mode, the collection of the flow is done by adding a splitter blade and a lens. The assembly is placed before the focusing lens on the sample. The separating blade makes it possible to let the incident THz beam pass between the source and the object. Once reflected on the object, the THz signal is collected by the focusing lens, then reflected on the splitter plate to be sent at 90 ° to the lens in front of the sensor. The lens in front of the sensor is the same as the lens used in transmission mode.

 The splitter blade introduces a loss of signal at each passage by a reflection of a portion of the energy of the source to the go, and by a transmission of a portion of the reflected energy of the object back.

 Regarding the polarization mode, the THz wave is an electromagnetic wave that can be polarized specifically to study a specific response of materials, such as birefringence.

 To do this, the incident wave must be circularly polarized. A delay component can be used to transform a linear polarization into a circular one, for example:

 a ¼ wave blade,

 - a horn or

 - 120 ° prism mounting.

 The analyzer analyzes the ellipticity, the phase shift of the ellipse and the direction of rotation of the ellipse. The techniques used are:

 - Rotation of a ½ wave + Polarizer retarder and acquisition of the energy at least 180 ° to reconstitute the parameters of the ellipse and

 - Decomposing the transmitted wave on at least three polarizer-equipped sensors that analyze at least three preferred directions of polarization. We perform the analysis on a single sensor that integrates:

 the measurement of amplitude on two perpendicular polarization directions using two plasmonic nano-transistors provided with their antenna and

the phase shift measurement of the ellipse and its direction of rotation using a nano-transistor provided with crossed antennas. Modularity offers a very flexible platform for measuring the thickness of samples (objects to be imaged) through print management, the need for adapted spatial resolution, transmission mode imaging for completely transparent materials, reflection for objects with one material opaque, polarization for materials that have a preferential axis of propagation (we speak of birefringence).

 The combined use of the THz source (modulation and broadband), the sensors (very sensitive, fast), the encoders of the Cartesian table (absolute position) and the acquisition system (synchronization, filtering) makes it possible to form an image with a very high sensitivity, no motion artifact, no standing wave and very low BF and HF noise.

 With regard to the splitter blade, for example the blades 435 and 455 illustrated in FIGS. 14 and 15, the high-resistivity silicon is very commonly used in the THz domain for the manufacture of optical windows, beam splitter or lens hemispherical collection. Preferably, the image capture system object of the invention uses this material for its semi-reflective blades. Indeed, the very high index of 3.41 makes it possible to obtain on a single material face a separation of the beam in two components reflected and transmitted by 45% and 55% respectively.

 To ensure the blade-separator function (ideally 50% transmission and 50% reflection) the blade must be carefully designed because the passage through both sides of the blade causes interference of the beam with itself which makes vary significantly the performance of the component.

 Preferably, the separating blade is the object of antireflection treatment on one side of its material. Thus there is no reflection on this face, and interference does not occur. For the treatment, a material with an index close to 1.8 is preferably chosen, and its thickness is adapted to a submultiple of the wavelength.

 In embodiments, a separation efficiency of 50% is obtained by choosing the thickness of the blade: by calculating the components of the incoming wave interference it is possible to determine which thickness leads to an efficiency of 50%.

 With regard to the incident polarization on the separating plate, preferentially the polarization is used in TM mode (acronym of transverse mode for transverse mode) in which the field is parallel to the plane of incidence. The inventor has in fact determined that the polarization in TM mode ensures a better stability of the separation efficiency when the thickness of the separating blade varies in its manufacturing tolerance.

The splitter blade is used on a weakly diverging beam to be most effective. Preferably, it is inserted into a collimated beam. The separating blade is crossed twice by the optical beam: once when it comes from the source, a second time when it returns from the object to the detector.

 The splitter blade is used to design an optical system in reflection, where the illumination of the object and the recovery of the flux from the object are on the same side of the object.

 A multipixel imaging device according to the invention integrates the following elements:

a source THz,

 an optical system consisting of lenses for performing linear illumination,

an optical system consisting of lenses for collecting the illumination after transmission through the object or reflection on the object,

 a separating blade in the case of reflection,

 - a multipixel sensor and

 a signal processing unit leaving the sensor.

 In general, it is complicated to use the transmission mode either because of the dimensions of the object or because of industrial environment constraints. The reflection mode is therefore commonly used. There are several ways to arrange the lighting and the lens to set up this mode.

 In transmission mode, the complete optical system has two separate parts 400 and 410, one for linear lighting, the other for the imaging lens. For example, in the optical system of the device 40 illustrated in FIG. 13, the three left lenses, part 400, for illumination, the two on the right, part 410, for the objective. Between the two, the free space makes it possible to insert the object 405 whose images must be captured by the sensor 415.

 The opening of the lighting is adapted to the opening of the lens to not lose energy outside the lens.

 In one embodiment of the device 42 illustrated in FIG. 14, with reflection through the entire objective, the three right-hand lenses 420 form the illumination which is projected at the level of the image plane of the objective 430. then the objective 430 to be projected onto the object 425. Finally the objective 430 forms the image of the object 425 which is projected at 90 ° on the sensor 415 through a semi-reflective plate 435.

 The lighting is then designed to uniformly illuminate the object 425 through the lens

430.

In one embodiment of the device 44 illustrated in FIG. 15, with reflection inside the lens, the illumination 440 takes into account an additional lens 460 after the cylindrical lens 465: the last lens of the lens 450. Thus, the illumination is inserted directly at the exit of the objective 450. The objective 450 operates thanks to a semi-reflective plate 455 which reflects the light reflected at 90 ° to the other two lenses and the linear sensor 470. The advantage of this design is to be more compact, and to introduce fewer losses with fewer lenses crossed by the THz beam between the source and the sensor.

 The illumination system provides a uniform illumination of the scene, the collection system a very high resolution exploited by the extreme compactness of the pixels of the multi-pixel sensor.

Claims

1. Device (40, 42, 44) for capturing point values to constitute an image, characterized in that it comprises:

 an incoherent source (100) of rays whose frequency is between 0.075 THz and 10 THz to illuminate an object,

a sensor (415, 470) of radiation coming from the object, which comprises a zone sensitive to the radiation coming from the source and which emits an electrical signal representative of the intensity of the rays coming from the source reaching the sensitive zone of the sensor and

 at least one optical (F-Number) optics (400, 410, 420, 430, 440, 450) of less than one, located on the optical path of the rays emitted by the source from the source to the ray sensor through the object.

2. Device (40, 42, 44) according to claim 1, wherein the source (100) illuminates the object with a sufficiently large emission spectrum to scan the standing wave in a shorter time than the time of acquisition of the sensor.

3. Device (40, 42, 44) according to one of claims 1 or 2, wherein the incoherent source has a bandwidth of several GHz.

4. Device (40, 42, 44) according to one of claims 1 to 3, wherein the incoherent source has a bandwidth at least equal to 12 GHz at -100dB.

5. Device (40, 42, 44) according to one of claims 1 to 4, wherein at least one optical system comprises an aspheric optical lens.

The device (40, 42, 44) according to claim 5, wherein the percentage dispersion of the indices of the materials used for the optical lens over the frequency ranges of the source is less than 1%.

7. Device (40, 42, 44) according to one of claims 5 or 6, wherein the dispersion, in percentage, of the indices of the materials used for the optical lens on the frequency ranges of the source is 0.2 %, from 100 to 300 GHz.

8. Device (40, 42, 44) according to one of claims 5 to 7, wherein the dispersion, in percentage, of the indices of the materials used for the optical lens over the frequency ranges of the source is 0.5 % from 100 to 700 GHz

9. Device (40, 42, 44) according to one of claims 1 to 8, wherein at least one optical component of an optical system has an antireflection treatment comprising cone-shaped microstructures or craters.

The device (40, 42, 44) according to one of claims 1 to 9, wherein at least one optical system comprises an optical lens and wherein the incoherent ray source (100) is configured to illuminate the entire the optical lens closest to said source.

1 1. Device (40, 42, 44) according to one of claims 1 to 10, wherein the emission frequency of the source (100) incoherent radiation is modulated.

12. Device (40, 42, 44) according to one of claims 1 to 1 1, wherein the source (100) incoherent radius comprises a noise source of thermal noise type in a resistance or impatted diode.

13. Device (40, 42, 44) according to one of claims 1 to 12, which comprises a proximity electronics for biasing a nano-transistor having the photosensitive area, by a gate voltage close to its swing voltage where the Transistor's classic operation is the most nonlinear.

14. Device (40, 42, 44) according to claim 13, wherein the rectified signal from the nano-transistor is amplified by forcing a charge asymmetry in the nano-transistor channel by injecting a current into the nano-channel. transistor, between the drain and the source and / or using metallized patterns acting as antennas.

15. Device (40, 42, 44) according to one of claims 13 or 14, wherein the rectified signal is a continuous potential difference between the drain and the source of the nano-transistor measured in common or differential mode.

16. Device (40, 42, 44) according to one of claims 13 to 15, wherein the proximity electronics comprises a compensation circuit of the offset generated by the injection of the current between the drain and the source of the nano-transistor for example using a subtractor assembly.

17. Device (40, 42, 44) according to one of claims 1 to 16, wherein the photosensitive area is a nano-transistor, the signal generated by the THz radiation is a difference continuous potential between the drain and the source of the nano-transistor measured in common or differential mode.

18. Device (40, 42, 44) according to claim 17, which comprises a proximity electronics for biasing the nano-transistor by a gate voltage close to its swing voltage where the conventional operation of the transistor is the most nonlinear.

19. Device (40, 42, 44) according to one of claims 17 or 18, wherein the rectified signal from the nano-transistor is amplified by forcing an asymmetry of the charges in the nano-transistor channel by injecting a current in the transistor channel, between the drain and the source and / or using metallized patterns acting as antennas.

20. Device (40, 42, 44) according to one of claims 1 to 19, which comprises at least one low noise amplifier and low drift that amplifies the signal on the dynamics of an analog digital converter.

21. Device (40, 42, 44) according to one of claims 1 to 20, which comprises means for synchronizing the demodulation of the signal with the amplitude modulation signal of the source (100) incoherent.

22. Device (40, 42, 44) according to one of claims 1 to 21, which comprises means for synchronizing the digital output of the device with the source (100) incoherent.

23. Device (40, 42, 44) according to one of claims 1 to 22, which comprises a separating plate, the polarization incident on the separating plate being the polarization in TM mode.

(acronym for transverse mode for transversal mode) in which the field is parallel to the plane of incidence.

24. Device (40, 42, 44) according to one of claims 1 to 23, which comprises means for measuring the intensity of radiation from the object, and / or polarization of the rays from the object , and / or the difference in steps of the rays coming from the object.