WO2017005511A1

WO2017005511A1 - Multichannel spectrometer equipped with a photosensitive layer with a lateral composition gradient

Info

Publication number: WO2017005511A1
Application number: PCT/EP2016/064718
Authority: WO
Inventors: Jean-Yves Duboz
Original assignee: Centre National De La Recherche Scientifique Cnrs
Priority date: 2015-07-08
Filing date: 2016-06-24
Publication date: 2017-01-12
Also published as: FR3038714B1; FR3038714A1

Abstract

The present invention relates to a spectrometer (100) comprising: a photosensitive layer made of a semiconductor alloy, which layer is divided into a plurality of photodetectors (110), the photodetectors being organised into at least one column extending in a first direction (210); and a circuit for processing the signals delivered by the photodetectors (110) in response to incident electromagnetic radiation, and wherein the semiconductor alloy of the photosensitive layer has a composition gradient in the first direction (210) so that at least some of the photodetectors (110) of said column have spectral responses to the incident electromagnetic radiation that are shifted in wavelength.

Description

MULTI-CHANNEL SPECTROMETER WITH PHOTOSENSITIVE LAYER WITH COMPOSITE LATERAL GRADIENT

TECHNICAL AREA

The present invention relates to a multichannel spectrometer, that is to say a spectrometer for simultaneously detecting several spectral components of electromagnetic radiation. More particularly, the invention relates to a spectrometer comprising a plurality of photodetectors, in the form of a bar or matrix, and devoid of dispersive element.

STATE OF THE ART

A standard technique for analyzing the spectrum of electromagnetic radiation is to disperse (or "spread") the electromagnetic radiation with the aid of a dispersive element, in order to separate the different components of the spectrum and then to measure the intensity of the electromagnetic radiation. each component using an optical detector. For example, a prism or optical diffraction grating may be used to disperse the visible light.

Detection of the spectral components can be performed sequentially. A slot isolates a quasi-monochromatic component of the scattered spectrum and a single photodetector measures the intensity of this component. The rotation of the dispersive element allows the photodetector to successively collect all the components of the spectrum. The disadvantage of this technique is that it takes a long time to measure the entire spectrum of electromagnetic radiation. The photodetector arrays have drastically reduced the measurement time of an electromagnetic spectrum. The incident polychromatic radiation is always dispersed by a dispersive element, but the monochromatic beams from the dispersive element are all transmitted to the photodetector array. he no longer use the slot and the rotation mechanism of the dispersive element to select one by one the beams. The photodetectors of the matrix are positioned so that each photodetector captures a single monochromatic beam. Thus, each photodetector corresponds to an angle of incidence of the beam received by the photodetector, and therefore a wavelength (the incidence angle varies monotonically with the wavelength). We can then simultaneously detect all the wavelengths corresponding to the range of available angles. The larger the array of photodetectors, the wider the range of detected wavelengths.

Spectrometers devoid of dispersive element have been developed to reduce their size and weight. These spectrometers also comprise a matrix of photodetectors, above which is arranged a set of filters. Each photodetector is thus equipped with a filter, in order to capture only a portion of the spectrum. The filters have different cutoff frequencies, so that the photodetectors can detect different spectral bands.

Document US5166755 describes an example of a spectrometer with filters, in which the filters are of the band-pass, high-pass or low-pass type. A processing circuit associated with the photodetector matrix calculates the intensity of a spectral component by subtracting the signals from two adjacent photodetectors. The operation is then repeated for each pair of adjacent photodetectors to determine the entire spectrum. The filters are for example of the interference type. They consist of an alternation of dielectric and metal layers deposited on the matrix of photodetectors and whose thicknesses vary according to their position within the matrix. Their manufacture is complex because each filter has a particular geometry, and requires a large number of technological steps. Spectrometers equipped with filters therefore have a significant manufacturing cost. SUMMARY OF THE INVENTION

There is therefore a need to provide a compact and lightweight spectrometer at a lower cost. According to the invention, this need is satisfied by providing a spectrometer comprising:

a photosensitive layer made of a semiconductor alloy divided into a plurality of photodetectors, the photodetectors being organized in at least one column extending in a first direction; and

a circuit for processing signals delivered by the photodetectors in response to incident electromagnetic radiation;

characterized in that the semiconductor alloy of the photosensitive layer has a compositional gradient in the first direction such that at least a portion of the photodetectors of said column have spectral responses to incident electromagnetic radiation shifted in wavelength. .

Rather than using filters that "select" narrow ranges of wavelengths, and thus expose the photodetectors underlying only a portion of the incident electromagnetic radiation, the invention proposes to act directly on the sensitivity photodetectors so that their detection is limited to these same ranges of wavelengths. In other words, the photodetectors of the spectrometer according to the invention are simultaneously exposed to the entire electromagnetic radiation and themselves fulfill this selection function.

Photodetectors belonging to the same column have wavelength-shifted spectral responses due to the fact that the composition of the semiconductor alloy forming the photodetectors varies from one pixel to the other of the column. Indeed, by varying the composition of the semiconductor alloy, that is to say by modifying the distribution between at least two of the constituents of the alloy, the width of the forbidden band of the semi-material is modified. driver. The photodetectors of the column therefore do not all have the same detection threshold, ie the threshold (in energy or in wavelength) from which the photons of the electromagnetic radiation are absorbed by the material.

Thanks to this column of non-uniform photodetectors, it is possible to measure the different spectral components of the incident electromagnetic radiation by dispensing with the filters. The manufacture of the spectrometer according to the invention is simplified, and therefore less expensive.

Advantageously, the semiconductor alloy is a direct bandgap semiconductor material, preferably a ternary alloy III-V or II-VI. A III-V material (respectively 11-VI) is an alloy of one or more elements of column III (respectively II) of the Mendeleev table with one or more elements of column V (respectively VI) of the Mendeleev table. This alloy is said to be ternary when it comprises two elements in competition belonging to the same column (for example indium (In) and gallium (Ga) - column III) and a third element of the other column (for example nitrogen (N) - column V).

Preferably, the composition of the semiconductor alloy varies monotonically from one end to the other of the photodetector column. In one embodiment of the spectrometer according to the invention, the photodetectors are organized in a matrix comprising several columns extending in the first direction and several rows extending in a second direction, the semiconductor alloy of the photosensitive layer. having a compositional gradient in the first direction for each column of photodetectors and a constant composition in the second direction for each row of photodetectors.

In an alternative embodiment, the photodetectors are organized into a matrix comprising a plurality of columns extending in the first direction and several rows extending in a second direction, the semiconductor alloy of the photosensitive layer having a composition gradient in the first direction for each column of photodetectors and a composition gradient in the second direction for each row of photodetectors. The processing circuit is advantageously configured to calculate a signal difference between two consecutive photodetectors of each column, which makes it possible to determine the intensity of each component of the electromagnetic spectrum.

Preferably, the processing circuit is configured to measure the intensity of several spectral components of the incident electromagnetic radiation.

BRIEF DESCRIPTION OF THE FIGURES

Other characteristics and advantages of the invention will emerge clearly from the description which is given below, by way of indication and in no way limiting, with reference to the appended figures, among which: FIG. 1 represents a first embodiment of FIG. a multichannel spectrometer according to the invention;

FIG. 2 shows the spectral responses of the photodetectors belonging to the multichannel spectrometer of FIG. 1;

FIG. 3 represents a second embodiment of multichannel spectrometer according to the invention;

FIG. 4 schematically represents a first method for obtaining a photosensitive layer with a lateral gradient of composition; and

- Figure 5 schematically shows a second method for obtaining a photosensitive layer with a lateral gradient of composition.

For the sake of clarity, identical or similar elements are identified by identical reference signs throughout the figures.

DETAILED DESCRIPTION OF AT LEAST ONE EMBODIMENT

In the following description, the term "bar" is used to refer to a column or row of pixels, and "matrix" to an arrangement defined by several columns and several rows of pixels. Each pixel constitutes a unitary photodetector, or photosensitive element, capable of measuring a more or less extended band of the spectrum of incident electromagnetic radiation. We can therefore speak indifferently of bar / matrix of pixels or barrette / matrix of photodetectors.

FIG. 1 represents a first embodiment of multichannel spectrometer 00, in which the different photodetectors 1 10 are organized in the form of a strip. The strip 100 of photodetectors 1 10 here extends in a vertical direction 210, that is to say in the direction of the columns.

The strip 100 comprises a number M of photodetectors 1 10 (in the example shown, M = 6). The photodetector of rank n is denoted "1 10n", n being a positive integer ranging from 1 to M. The photodetector of rank n + 1, that is to say the photodetector which follows the photodetector of rank n in the direction 210, bears the reference "1 10n + i".

The photodetectors 1 10 of the strip 100 are formed from a substrate on which is deposited a photosensitive layer of a semiconductor alloy. The photosensitive layer is the active layer of photodetectors 1 10, where the absorption of photons occurs.

The semiconductor alloy of the photosensitive layer comprises at least two semiconductor materials whose distribution is likely to vary. The photosensitive layer is divided into several portions, preferably by etching trenches 130. Each portion thus obtained constitutes the photon absorption zone of a photodetector 1 10. The trenches 130 are filled with dielectric material, in order to isolate electrically the photodetectors 1 10 from each other.

The photodetectors 1 10 are preferably made of PN diodes, Schottky diodes (Metal / Semiconductor) or MSM diodes (Metal / Semiconductor / Metal).

The semiconductor alloy of the photosensitive layer is preferably a direct bandgap material, like most semiconductor materials III-V or II-VI. In this type of semiconductor material, a photon is absorbed as soon as its energy is greater than or equal to the width of the forbidden band of the material. Consequently, a photodetector whose active layer consists of such a material can only capture photons of energy greater than the forbidden bandwidth, ie radiation having a wavelength less than the wavelength, so-called of cutoff, corresponding to the forbidden band of the semiconductor material (at each energy E corresponds a wavelength λ, according to the relation

The composition of the semiconductor alloy within the array 00 varies so that at least a portion of the photodetectors 110 have different cutoff wavelengths. When chosen from the family of materials III-V and II-VI, the semiconductor alloy is advantageously a ternary alloy comprising two elements of the same column of the periodic table. These two elements, in competition with each other, make it possible to vary the composition of the semiconductor alloy by varying their distribution. According to the wavelength range that one wishes to detect, that is to say according to the nature of the incident electromagnetic radiation (for example ultraviolet, visible, near infrared, far infrared ...), the ternary alloy can be chosen from AlxGa i xN, ln _x Gai-xN, A Gai xAs, In _x Gai-xAs, ln _x Asi xSb, HgxCdi xTe (some of the alloys cited have indirect forbidden bands for certain compositions) ...

FIG. 2 represents the spectral response R of the photodetectors 1 10 of the strip 100 as a function of the wavelength λ of the incident radiation. The spectral response of a photodetector is defined as the ratio of the electric current at the output of the photodetector (in A.crrr ² ) to the power of the electromagnetic radiation at the input (in W.crrr ² ) at a given wavelength λ . It is therefore usually expressed in ampere by AW ^{~ 1} . In FIG. 2, the spectral responses of photodetectors 1 10 have been normalized and are therefore expressed in relative value.

As explained above, the photodetectors 1 10 all have a wavelength low-pass behavior: the spectral response R (A) of each photodetector 1 10 is maximum (R = 1) for the wavelengths λ less than the cut-off wavelength AG. For wavelengths longer than the cut-off wavelength AG, the spectral response R (A) of the photodetectors 1 1 0 decreases sharply. For example, for a photosensitive layer of ALGai-xN, the intensity of the electric current at the output of the photodetectors 110 is divided by every 8 nm wavelength after the cut-off wavelength AG.

As illustrated in FIG. 2, the photodetectors 110 of the array 100 preferably have different cut-off wavelengths. The cut-off wavelength of the photodetectors also varies monotonically from one end to the other of the strip 100. For example, the photodetector 1 1 On (of rank n) has a cut-off wavelength Acn less than the cut-off wavelength Acn + i of the photodetector 1 10n + i (of rank n + 1), whatever the value of n. Outside the cut-off wavelength AG, the spectral responses R (A) are preferably substantially identical from one photodetector to the other. They are therefore simply offset from each other in abscissa, along the axis of wavelengths A (Figure 2). The extraction of the different spectral bands of the incident electromagnetic radiation is effected by means of a processing circuit, which subtracts the signals from the photodetectors of the column, in groups of two consecutive pixels having a composition gradient. The electrical current of the photodetector of rank n for a flux radiation φ as a function of wavelength λ is given by the following relation:

where R _n (λ) is the spectral response of the photodetector of rank n. In the same way, the electrical current of the photodetector of rank n + 1 is written

where R _{n + 1} (λ) is the spectral response of the photodetector of rank n + 1. The difference between the currents of two consecutive pixels is therefore:

'ri + 1 n n + 1) -? _Tl (A)) 0 (A) dl

Knowing, as described above in connection with FIG. 2, that the spectral responses of two consecutive pixels are just wavelength shifted by a difference λλ, such that R _{n + λ} (A) = R _n (λ + Δλ), the previous equation becomes:

f ^∞ f ^∞ dR _n (X)

l _n = JA "+ X) - R _n W) φ (Χ) άλ = J - -jj-. Δλ. φ (Χ) άλ

As a first approximation, the spectral response R _n (λ) of the photodetector 1 1 can be assimilated to a Heaviside function H (λ), that is to say a function such that H = 1 for A ≤ _Gn and // = 0 for λ> At _Gn , where _Gn is the cut-off wavelength of the photodetector of rank n. The derivative of the Heaviside H (A) function is a Dirac pulse δ (λ) centered on the cut-off wavelength _λn . We obtain then: l _n = Αλ. φ (Χ) άλ = ΔΑ. φ (λ _0η )

where φ (λ _6η ) is the flux received by the pixel n at the cut-off wavelength _λn .

With this approximation, the processing circuit therefore calculates an average value of the flux received by the pixel n in a wavelength range between X _Gn and À _{Gn + 1} , that is to say over a range of width AÀ. . The gap AÀ also defines the spectral resolution of the spectrometer. It depends on the compositional gradient of the semiconductor alloy in the vertical direction 210 and the difference between two consecutive pixels of the strip 100. Thus, for the same composition gradient, the greater the number of pixels in the bar 100 is large, the lower the resolution parameter AÀ (and therefore the greater the separation power of the spectrometer is strong).

Rather than a Heaviside function, we can consider a spectral response R _n (A) having a less abrupt cut, for example a function characterized by an exponential decay for wavelengths greater than the cut-off wavelength than :

R _n (X) = 1 for A <X _Gn ;

R _n {X) = exp (- ^ γή for A _Gn <λ X X _Gn + 2β, and

R _n (X) ~ 0 for A> At _Gn + 2β;

where 2β denotes the cutoff width ( _η (2β) = 1 / e ² ) and is typically a few nanometers (for AIGaN, 2β = 8 nm).

In this case, the difference of the signals between two consecutive pixels is:

One then sees that the processing circuit integrates the light flux over a spectral band ranging from TO to _Gn to _Gn + 2β. Therefore, it is useless to choose a resolution parameter AÀ less than 2β, because the difference of the signals integrates the flux over a width equal to 2β, even if AÀ is smaller. In this case, it is not A that limits the resolution but 2β. At best, the spectral resolution parameter Δλ will be chosen equal to 2β. This illustrates the value of having materials with abrupt cuts, hence the preferred use of direct bandgap semiconductors.

By replacing the flux φ (Χ) by its mean value <φ (λ)> over a width AÀ between _Gn and _Gn + 2β, we obtain: άΙ _η = άλ Αλ. <φ (λ)>

When the flux is constant, we thus obtain the same value as with the first approximation, with the restriction that the spectral resolution parameter Αλ is limited to 2β (ie 8 nm in the case of AIGaN).

By repeating this calculation for each pair of consecutive pixels of the column, the processing circuit is able to determine the intensity of each spectral band (of width Αλ) between the two extreme cut-off wavelengths of the column, A _{G 1} and _GM .

For example, with the ln _x Gai _× N system, it is easy to obtain cutoff frequencies ranging from 400 nm to 500 nm, by varying the indium (In) concentration x from about 7% to about 20%. With the A Gai xN system, cut-off wavelengths that can be reached without encountering manufacturing difficulties are between 250 nm and 350 nm, a concentration x of aluminum (Al) varying from 70% to approximately 0%. Since the spectral resolution parameter Δλ is limited to 8 nm, about one or two tens of pixels are sufficient to scan the above wavelength ranges.

If it is not possible to reduce the spectral resolution parameter Αλ, the array 00 may contain a larger number of pixels in order to increase the quality of the spectrometer signal. These pixels are advantageously distributed in a group of consecutive pixels having substantially the same cut-off wavelength. The difference of the electric currents, for the extraction of the different spectral bands, is then no longer between two consecutive pixels of the column, but between two groups of adjacent pixels. It deals with the sums of electric currents of the pixels belonging to the same group.

By way of example, the strip 100 may comprise 128 pixels having an area of 50x50 μηι ² . It then measures 6.4 mm in length, without taking into account the spacing between the pixels (trenches 130). As its thickness is of the same order of magnitude as that of the substrate on which it is formed, of a few hundred micrometers, the bar 100 occupies a very small volume.

In the multichannel spectrometer of FIG. 1, the radiation illuminates all the pixels of the column. A pixel thus receives only a fraction of the total luminous flux φ _{(0 (} (λ), this fraction being equal to -. Φ _{(0 (} (λ) (M is the number of

M

photodetectors in the bar). The current produced by this pixel is then proportional to:

Conversely, in a conventional matrix spectrometer equipped with a dispersive element, the entire luminous flux at a length λ arrives on a single photodetector after passing through the dispersive element. The electric current produced by this photodetector is proportional to:

TF. <p _tot (X). AA

where TF is the transmittance of the dispersive element.

Thus, to determine which of the two spectrometers has the best signal-to-noise ratio, the ratio 1 / M should be compared to the transmission factor TF. The transmission factor of a dispersive system can be large (a fraction of a unit) and therefore greater than 1 / M as soon as M is greater than about ten. The signal / noise ratio will therefore be less good in the multichannel spectrometer of the invention, with comparable spectral resolution. However, dispersion matrix spectrometers are generally designed for high resolutions and therefore small pixels, and are therefore more expensive to produce than the spectrometer according to the invention.

Comparing the spectrometer according to the invention with a single-channel spectrometer using dispersion and slot, the transmission factor TF corresponds to the dispersive system and to the slot, and can always be larger than M. The spectrometer of the invention is in this case to simultaneously measure the entire spectrum while the single-channel dispersive system measures it sequentially and must limit its acquisition time at each wavelength. The lower signal amount at the output of the spectrometer of the invention can be easily compensated for by a longer acquisition time, thus arriving at equivalent signal-to-noise ratios.

FIG. 3 represents a second embodiment of the multichannel spectrometer making it possible to increase the signal / noise ratio of the spectrometer (if it is desired, for example, to maintain a high speed of measurement). In this second embodiment, the photodetectors 1 10 are arranged in matrix 300. The columns i of the matrix 300, here 6 in number, extend in the vertical direction 210 whereas the rows j of the matrix, also at the number of 6, extend in the horizontal direction 220.

As before, the photodetectors 1 10 of the matrix 300 are all formed by cutting a photosensitive layer having a lateral gradient of composition, here in the direction 210 of the columns. Several photodetectors belonging to each column have spectral responses shifted in wavelengths, as has been described previously, in particular in relation with FIG. 2. On the other hand, in the direction 220 of the rows, the composition of the semiconductor alloy forming the photodetectors 1 10 is advantageously constant. Thus, the columns i of the matrix 300 are identical to each other.

Such a configuration makes it possible to increase the quality of the signal representing each spectral band, because the difference between two consecutive pixels (having different cut-off wavelengths) can be averaged over several columns. This improves the signal / noise ratio of the spectrometer. The matrix-shaped spectrometer 300 also has the advantage of being able to be used simultaneously for a spectral analysis in the column direction 210 and for a geometrical analysis in the row direction 220, that is to say as a (visible, infrared ..., depending on the nature of the semiconductor alloy). In a third embodiment not shown in the figures, the photodetector array 1 10 has not only a composition gradient in the direction 210, but also a compositional gradient in the direction 220.

The photosensitive layer made of a semiconductor alloy can be obtained by known epitaxy techniques, in particular molecular beam epitaxy (or MBE), and vapor phase epitaxy. Depending on the nature of the epitaxial material, the substrate on which the photosensitive layer is deposited may be InP, GaAs, silicon or sapphire. In general, when growing a layer of a ternary alloy by epitaxy, during the manufacture of an infrared detector, for example, it is sought to avoid composition gradients. In contrast, to manufacture the spectrometer according to the invention, the epitaxial reactor is configured to promote the production of a composition gradient.

Figures 4 and 5 show schematically two epitaxial reactor configurations 400 for obtaining a lateral composition gradient in the photosensitive layer. The ternary alloy formed is, in this example, ln _x Gai xN. In FIG. 4, the reactor 400 comprises three sources 401, 402 and 403 respectively containing the elements In, Ga and N. So that the elements In and Ga have concentrations (respectively x and 1-x) that vary laterally in directions. opposed on the substrate 410, the sources 401 and 402 are inclined relative to the normal direction 420 to the substrate 410, respectively left and right. Due to the inclination of the source 401, the indium flux is greater on the left side of the substrate 410 than on its right side. Conversely, the source 402 being inclined in the other direction, the gallium flow arriving on the substrate 410 is greater on the right than on the left of the substrate 410. As a result, the photosensitive layer has an indium concentration x higher on the left than on the right and vice versa for the concentration of gallium 1 -x. The nitrogen source 403 is advantageously directed in the direction 420, ie perpendicular to the substrate 410.

In the configuration of Figure 5, the three sources are grouped around the normal direction 420, so that the fluxes are substantially perpendicular to the substrate 410. In this case, the composition gradient does not come from differences in flux, but because the substrate is subjected to a temperature gradient, for example by means of a heating substrate holder. Indeed, some elements do not incorporate with the same efficiency depending on the temperature. For example, indium (In) is incorporated less quickly when the temperature is high.

Thus, if the substrate 410 has a lateral temperature increase 430, from left to right in the example of FIG. 5, the right part of the substrate 410 is less rich in indium than its left part (and, by way of consequently, the right part of the substrate is richer in gallium than its left part).

In one or the other of these configurations, the substrate 410 can be immobile or rotated during growth, depending on whether a composition gradient is desired in a single direction (in the case of the strip 100 or the matrix 300 for example) or in several directions (the directions 210 and 220 for example). When the substrate 410 turns on itself at 360 ° at a constant speed, it is possible to obtain a composition gradient having a symmetry of revolution. The multichannel spectrometer described above can respond, with a spectral resolution generally between 5 nm and 8 nm, to many needs. In the field of the infrared for example, it can be used to measure the temperature of an object or a flame. The classical measurement of temperature consists in measuring the optical emission of the object at two different wavelengths, and then in deducing its temperature using the emission law of the black body. But the classical measurement is often distorted, because the emissivity of the body can vary between the two wavelengths. The complete measurement of the spectrum in the infrared, thanks to the multichannel spectrometer according to the invention, makes this measurement of temperature reliable. Another example of application is to analyze the spectrum of the radiation produced by a flame, in order to deduce the products of combustion, or the visible spectrum of radiation transmitted by glass objects, for the purpose of sorting them by color. The multichannel spectrometer also makes it possible to locate a laser line and to determine the wavelength of this line (with the resolution parameter AÀ near). Since in practice the spectral response of the photodetectors decreases exponentially after the cutoff frequency, several pixels will emit a signal in response to the laser line (their number depends on the spectral resolution). A calculation similar to that described previously will make it possible to obtain a characteristic signature of the line.

Many variations and modifications of the multichannel spectrometer will be apparent to those skilled in the art. The semiconductor alloy of the photodetectors to be a material different from the materials III-V and II-VI, and in particular an indirect gap semiconductor material, such as the Si _x Gei- _x alloy. Direct gap materials are preferred, however, because they have better sensitivity than indirect gap materials. In addition, signal processing is simplified in the case of direct gap materials because their absorption mechanism does not involve phonons and is steeper, leading to better spectral resolution.

Claims

1. A spectrometer (100, 300) comprising:

a photosensitive layer made of a semiconductor alloy divided into a plurality of photodetectors (1 10), the photodetectors being organized into at least one column extending in a first direction (210); and

a signal processing circuit delivered by the photodetectors (1 10) in response to incident electromagnetic radiation;

characterized in that the semiconductor alloy of the photosensitive layer has a composition gradient in the first direction (210) so that at least a portion of the photodetectors (1 10) of said column have spectral responses (R ( A)) to incident electromagnetic radiation shifted in wavelength.

The spectrometer of claim 1, wherein the semiconductor alloy is a direct bandgap semiconductor material.

3. The spectrometer according to claim 2, wherein the semiconductor alloy is a ternary alloy III-V or II-VI.

The spectrometer according to any one of claims 1 to 3, wherein the composition of the semiconductor alloy varies monotonically from one end to the other of said column of photodetectors (1 10).

The spectrometer according to any one of claims 1 to 4, wherein the photodetectors (100) are organized into a matrix (300) comprising a plurality of columns (i) extending in the first direction (210) and several rows (j). ) extending in a second direction (220), the semiconductor alloy of the photosensitive layer having a composition gradient in the first direction (210) for each column (i) of photodetectors and a constant composition in the second direction (220) for each row (j) of photodetectors.

The spectrometer according to any one of claims 1 to 4, wherein the photodetectors are organized into a matrix comprising a plurality of columns (i) extending in the first direction (210) and a plurality of rows (j) extending in a second direction (220), the semiconductor alloy of the photosensitive layer having a composition gradient in the first direction (210) for each column (i) of photodetectors and a compositional gradient in the second direction (220) for each row (j) of photodetectors.

The spectrometer according to any one of claims 1 to 6, wherein the processing circuit is configured to calculate a signal difference between two consecutive photodetectors (1 10) of each column (i).

8. Spectrometer according to any one of claims 1 to 7, wherein the processing circuit is configured to measure the intensity of several spectral components of the incident electromagnetic radiation.

5