EP3000132A1

EP3000132A1 - Photodiode array having adjustable charge-absorption

Info

Publication number: EP3000132A1
Application number: EP14725184.7A
Authority: EP
Inventors: Yang Ni
Original assignee: New Imaging Technologies SAS
Current assignee: New Imaging Technologies SAS
Priority date: 2013-05-22
Filing date: 2014-05-21
Publication date: 2016-03-30
Also published as: FR3006105B1; FR3006105A1; WO2014187840A1

Abstract

The invention concerns a photodiode array comprising a first common electrode with a P-N junction, comprising at least one substrate layer (4) and one active layer (5), - a passivation layer (6), and at least two kinds of doped areas of the same type: - first doped areas (3) at least partially formed in the active layer (5), defining second electrodes to form, with the first common electrode, photodiodes connected to reading circuits to form images, - at least one second doped area (10) forming a third electrode capable of absorbing excess charge carriers in order to discharge same, polarisation means being capable of applying, to said second doped area, an electrical potential (Vring) by means of which the absorption of the charge carriers by said second doped area (10) can be adjusted, said at least one second doped area (10) being formed in the passivation layer (6) and being separated from the active layer (5) by a portion of said passivation layer (6).

Description

PHOTODIODE MATRIX WITH ADJUSTABLE LOAD ABSORPTION

FIELD OF THE INVENTION The invention relates to photodiode arrays, and more particularly photodiode arrays based on gallium indium arsenide (In GaAs) and indium phosphide (InP) layers, as well as to their photodiode process. manufacturing.

BACKGROUND OF THE INVENTION

One of the photodiode array manufacturing methods in semiconductor materials with a narrow band gap for infrared light detection is to insert the active low-band gap detection layer between two semiconductor materials with large band gap. The two large band gap semiconductor layers provide effective protection / passivation while remaining transparent to the wavelength of the radiation to be detected by the photodiodes.

In addition, with appropriate doping, the two heterojunctions between the active layer and the two protection / passivation layers confine the photoelectric charges in the active detection layer and improve the quantum yield of the photodiode thus constructed.

An InGaAs photodiode is a typical example of this physical structure. The active detection layer made of the InGaAs material can have an adjustable band gap depending on the indium and gallium composition in the InGaAs, ideal for operating in the SWIR (Short Wave Infrared) short-wave infrared band. wave), of the order of 1, 4 to 3 μιη. Indium phosphide and gallium-indium arsenide share the same face-centered cubic crystal structure. The most used composition is lno.53Gao.47As. The crystal mesh size is then comparable to that of the InP substrate, in particular the mesh parameters. This crystal compatibility allows the epitaxial growth of an InGaAs active layer of excellent quality on an InP substrate. The band gap of ln ₀ .53Ga ₀ .47As is about 0.73eV, capable of detecting up to a wavelength of 1.68 μιη in the SWIR band. It has a growing interest in the fields applications such as spectrometry, night vision, sorting used plastics, etc.

Both protection / passivation layers are usually made in InP. Especially the lnO.53GaO.47As composition, having the same crystal mesh size as InP, this allows a very low dark current from room temperature.

Figure 1 illustrates the physical structure of a matrix 1 of photodiodes. An active layer 5 composed of InGaAs is sandwiched between two layers of In P. The lower layer is in fact the substrate 4 on which the InGaAs layer is formed by vapor phase epitaxy in organometallic (or MO-CVD for metalorganic vapor phase epitaxy according to the Anglo-Saxon terminology). This InGaAs layer is then protected by a thin passivation layer 6 composed of InP, also deposited by epitaxy. The InP layers are generally N type, doped with silicon. The active layer 5 of InGaAs may be slightly N-doped or remain quasi-intrinsic. Thus, the two lower / upper InP layers and the InGaAs active layer 5 form the common cathode of the photodiodes in this matrix.

Individual anodes 3 are formed by local diffusion of zinc (Zn). The Zn dopant passes through the thin InP passivation layer 6 and enters the InGaAs active layer.

FIG. 2 illustrates an InGaAs image sensor consisting of a matrix 1 of InGaAs photodiodes connected in flip-chip mode with a reading circuit 2. In a matrix InGaAs sensor, the The photodiode array is connected to a reading circuit generally made of silicon in order to read the photoelectric signals generated by these InGaAs photodiodes. This interconnection is generally done by the flip-chip process via indium balls 8, as illustrated in FIG. 2. The radiation SWIR 9 arrives on the matrix of photodiodes through the substrate 4 of indium phosphide, transparent in this optical band.

With a detector operating in integration mode, an output signal is obtained that is proportional to the product of the flux and the duration of exposure. However, the output signal is limited by the maximum integration capability of the detector. For scenes with high contrast, it is often impossible to get a good rendering of dark areas and at the same time keep bright areas without saturation. This problem is all the more more serious for night vision to which an InGaAs photodiode array sensor is often intended.

Another way of reading the photodiode signals of the photodiodes in general is proposed by the document EP1354360 and illustrated in principle in FIG. 3 of the attached drawings. EP1354360 proposes a solar cell mode operation of a photodiode 51 in order to obtain a logarithmic response as a function of the intensity of the incident optical radiation 59. In this mode of operation, the photodiode 51 receives no external polarization and it is polarized in direct by the photoelectric charges generated in its junction. The forward bias voltage observed on the photodiode is proportional to the logarithm of the incident optical flux. This logarithmic response makes it possible to cover, without any electrical and optical adjustment, a dynamic operating range greater than 120dB, which is essential for the use of a SWIR InGaAs sensor in natural conditions outside. EP1354360 also proposes to associate a read circuit 55 with switching to the photodiode.

The principle of use of the image sensor array illustrated in FIG. 3 is as follows:

a) Activate the selection signal SEL in order to select the desired photodiode 51 by closing the switch 54. Once this photodiode has been selected, the first read signal RD1 is activated which will close the corresponding controlled switch in order to memorize the voltages of a first reading in the memory 56. This first reading records both the image and the fixed spatial noise.

b) Then activates the reset signal RSI, which signal will cause the closing of the switch 53. The photodiode 51 is thus short-circuited, thus simulating a reference image in absolute darkness.

c) Then activates the second read signal RD2 to thereby record in the memory element 57 the voltages of the second reading. The fixed spatial noise alone has thus been memorized.

d) The difference between the result of the two memorizations contained in the respective memory elements 56 and 57 is calculated by a differential amplifier 58. The output signal of this amplifier 58 then corresponds to an image free of fixed spatial noise.

With the second reading, a zero voltage corresponding to the dark condition is generated. This electronic darkness signal makes it possible to eliminate the offsets in the reading chain in a matrix detector.

The principle proposed by EP1354360 has been applied in an InGaAs sensor and works perfectly. But a phenomenon of dazzling ("blooming" in Anglo-Saxon terminology) is observed for diurnal scenes. This phenomenon can be simply described as a loss of spatial resolution in an image. The detector nevertheless continues to be sensitive to the variation of light according to the logarithmic law.

French patent application FR2977982 proposes electrical insulation by etching around each photodiode. This approach makes it possible to effectively suppress this blooming phenomenon, but at the cost of a very high current of darkness in the photodiodes because of defects created by this etching. Another problem of this approach lies in the fact that the etching and diffusion stages of the photodiode anodes constitute two distinct stages of the manufacturing process, requiring different masks. Mask alignment errors can create additional nonuniformities between photodiodes in a matrix.

In a structure of the state of the art as illustrated in FIG. 1, it can be seen that each photodiode contains several PN junctions, one of which is wanted and a certain number which are parasitic. These PN junctions are illustrated in FIG. 4. The PN junctions 31 between the anodes 3 and the active layer 5 are desired and constitute the diodes of the photodiode array.

The side parasitic PN junctions 32 between the anodes 3 and the passivation layer 6 constitute a possible electrical pathway between the neighboring photodiodes via the passivation layer.

A conventional read circuit integrates, in a capacitance, the reverse current into the photodiode by applying an inverse bias on the photodiode. In this configuration, the side parasitic junctions 32 in the photodiodes are reverse biased at the same time with the effect of adding an additional parasitic current in the integration capacity. This parasitic current degrades the image quality, but generates almost no crosstalk between neighboring photodiodes. These parasitic currents can be partially compensated by complex image processing on the raw image coming out of the read circuit.

When a photodiode is operating in solar cell mode, the junction is forward biased by the incident light. In this case, the side parasitic junctions 32 are also forward biased and they constitute a passage of electric current between neighboring photodiodes. This direct polarization becomes all the more important as the incident optical intensity increases, thus creating a blooming phenomenon which considerably degrades the spatial resolution of the sensor.

French Patent Application No. 1350830 proposes a matrix of photodiodes, a sectional view of which is illustrated in FIG. 5, comprising:

a cathode comprising at least one substrate layer 4 made of a material of the indium phosphide family and an active layer 5 made of a material of the gallium-indium arsenide family, and

at least two kinds of doped zones of the same type formed at least in part in the active layer 5:

first doped zones 3 forming with the cathode photodiodes for the formation of images,

at least one second doped zone 20 absorbing excess charge carriers to evacuate them. FIG. 6 schematically illustrates the structure of energy bands according to section AA 'of the photodiode array of FIG. 5, that is to say in a section crossing the substrate 4, the active layer 5 and the layer of Passivation 6. The different levels of energy are represented as a function of depth according to arbitrary scales of depth and energy, for purely illustrative purposes: the energy of the valence band E _v , the energy of the band of conduction E _c , and the level of Fermi E _F.

There is thus an area 15 corresponding to the active layer 5 having a higher valence energy E _v than the two zones surrounding it, ie an area 14 corresponding to the substrate 4 and an area 16 corresponding to the layer of passivation 6. The holes 9 constituting here the charge carriers are thus confined in the active layer 5. In a similar manner to FIG. 6, FIG. 7 schematically illustrates the structure of energy bands according to the BB 'and CC sections of the photodiode array of FIG. 5, that is to say in a cross-section through the substrate. 4, the active layer 5 and a first doped zone 3 or a second doped zone 20, these two doped zones 3, 20 having the same level of doping at the same depth.

It can thus be seen that a zone 13 corresponding to a doped zone 3, has a valence energy E _v greater than the zones 15 and 14, respectively corresponding as above to the active layer 5 and to the substrate 4. The holes 9 constituting here the charge carriers are not confined in the active layer 5, and their passage in the doped zones 3, 20 is possible.

This approach effectively suppresses this blooming phenomenon, by decreasing the lateral conductivity in the photodiode matrix, but at the cost of a loss of collection efficiency (quantum efficiency). Indeed, a portion of the photoelectric charges are absorbed by the absorption zones constituted by the second doped zones 20, and, in the case of a low level of light, this loss becomes unacceptable. Thus, these approaches, despite their effectiveness, deteriorate the performance at low light level.

The patent application US 2010/0258894 A1 discloses a matrix of photodiodes in which a P-doped zone of an anti-crosstalk part advances more deeply towards the substrate than the P doped zone of the photodiodes, in order to form between the photodiodes of the photodiodes. depletion zones acting as anti-crosstalk walls. This approach also deteriorates the performance at low light level, since too much charge is absorbed by the anti-crosstalk part.

PRESENTATION OF THE INVENTION

The present invention provides a matrix of photodiodes comprising

a first common electrode of a PN junction, comprising at least one substrate layer made of a material of the indium phosphide family and an active layer made of a material of the gallium-indium arsenide family,

a passivation layer made of a material of the indium phosphide family, the active layer being situated between the substrate layer and the passivation layer, and

at least two kinds of doped zones of the same type: first doped regions formed at least partly in the active layer, defining second electrodes for forming, with the first common electrode, photodiodes connected to reading circuits and adapted for image formation,

at least one second doped zone forming a third electrode adapted to absorb excess charge carriers to evacuate them, biasing means being adapted to apply to said second doped zone an electric potential (Vring) by which the absorption of charge carriers by said second doped zone,

said at least one second doped region being formed in the passivation layer and being separated from the active layer by a portion of said passivation layer.

The photodiode matrix according to the invention is advantageously completed by the following characteristics, taken alone or in any of their technically possible combinations:

the matrix is configured so that the electric potential applied to the second doped zone is modulated according to the level of illumination on the photodiode array;

the second doped zone is located between at least some of the first doped zones;

the second doped zone individually surrounds the first doped zones;

a plurality of second doped zones are distributed parallel to each other and interspersed with first doped zones;

the matrix comprises a plurality of second doped zones distributed between the first doped zones along the diagonals of the photodiode array;

the second doped zone is separated from the first doped zones by a sufficient distance so that the space charge zones respectively associated with the second doped zone and the first doped zones are separated;

a metal grid on the surface of said matrix connects different points of the second doped zone (s) in order to homogenize the electric potential of the second doped zone (s) (s); ).

The invention also relates to an image sensor incorporating a matrix of photodiodes according to the invention.

The present invention provides an improvement by electrically controlling the degree of absorption of the photoelectric charges. In case of low level of light, the degree of absorption by the second doped zones will be minimized in order to favor the collection charge by the photodiodes. But in case of high level of illumination, the degree of absorption by the second doped zones will be reinforced in order to minimize the crosstalk between the photodiodes. The invention also relates to a method for manufacturing a matrix of photodiodes according to the invention, said method comprising the steps according to which, from a first common electrode comprising at least one substrate layer made of a material of the family of the indium phosphide and an active layer made of a material of the gallium-indium arsenide family, and a passivation layer made of a material of the indium phosphide family, the active layer being located between the layer of substrate and the passivation layer:

selective etching of the passivation layer is carried out,

the first doped zones and said at least one second doped zone are formed, said first doped zones being formed at the zones of the selective etching of the passivation layer previously produced.

Preferably, the first doped zones and said at least one second doped zone are formed during the same selective doping step. Preferably, the selective etching of the passivation layer removes a thickness of the passivation layer greater than the thickness of the portion of said passivation layer ultimately separating the second doped regions of the active layer.

The invention also relates to a method for manufacturing a matrix of photodiodes according to the invention, said method comprising the steps according to which, from a first common electrode comprising at least one substrate layer made of a material of the family of the indium phosphide and an active layer made of a material of the gallium-indium arsenide family, and a passivation layer made of a material of the indium phosphide family, the active layer being located between the layer of substrate and the passivation layer:

a first selective doping is carried out in order to start forming the first doped zones,

a second selective doping is then performed to finish forming the first doped zones and to form said at least one second doped zone. BRIEF DESCRIPTION OF THE FIGURES

Other aspects, objects and advantages of the present invention will become more apparent upon reading the following detailed description. The invention will also be better understood with reference to this description considered in conjunction with the accompanying drawings, given as non-limiting examples and in which:

FIG. 1, already commented on, is a diagram illustrating the structure of a matrix of InGaAs photodiodes of the state of the art;

FIG. 2, already commented on, illustrates an InGaAs image sensor consisting of a matrix of InGaAs photodiodes connected in flip-chip with a silicon substrate reading circuit;

- Figure 3, already commented, is a schematic diagram of realization of a logarithmic sensor with photodiodes solar cell mode;

FIG. 4 illustrates the different junctions in a matrix of photodiodes of the state of the art;

- Figure 5 illustrates a sectional view of a photodiode array according to French Patent Application No. 1350830, comprising absorption zones;

FIG. 6 schematically illustrates the structure of energy bands according to section AA 'of the photodiode array of FIG. 5;

FIG. 7 schematically illustrates the structure of energy bands according to sections BB 'and CC of the photodiode array of FIG. 5;

FIG. 8 illustrates a sectional view of a matrix of photodiodes according to the invention;

FIG. 9 schematically illustrates the structure of energy bands according to section AA 'of the photodiode array of FIG. 8;

FIG. 10 schematically illustrates the energy band structure according to section BB 'of the photodiode array of FIG. 8;

FIGS. 11, 12 and 13 schematically illustrate the structure of energy bands according to section CC of the photodiode array of FIG. 8 under the influence of three different polarizations;

FIGS. 14, 15, 16 and 17 are top views of various possible embodiments of the photodiode array according to the invention;

FIGS. 18, 19 and 20 schematically illustrate successive steps of a method of possible fabrication of the photodiode matrix according to the invention;

- Figures 21, 22 and 23 schematically illustrate successive steps of a possible method of manufacturing the photodiode matrix according to the invention. DETAILED DESCRIPTION

The present invention provides a structure for varying the absorption of charge carriers by an absorption zone. A matrix of photodiodes manufactured according to the present invention can be exploited in solar cell mode as described in EP1354360, without loss of spatial resolution, even in the presence of very high optical intensities. Such a matrix also provides an improvement in image quality with a conventional reading circuit in integration mode, such as, for example, the different ISC9705 and ISC9809 CMOS reading circuits marketed by Indigo / FLIR in the USA. The ISC9705 circuit integrates the photoelectric current of a photodiode directly onto a capacitor (direct injection mode) and the ISC9809 circuit integrates the photoelectric current through an operational amplifier (CTIA mode). The CTIA mode allows a higher charge-to-voltage conversion gain that promotes detection sensitivity.

With reference to FIG. 8, a photodiode array comprises a first common electrode comprising at least one substrate layer 4 made of a material of the indium phosphide family and an active layer 5 made of a material of the arsenide family. of gallium-indium.

A passivation layer 6, for example made of indium phosphide family material, is provided above the active layer of gallium indium arsenide. The active layer 5 is thus located between the substrate layer 4 and the passivation layer 6.

A material of the family of indium phosphide means a semiconductor material composed mainly, or almost exclusively, of indium phosphide, and possibly other components in a much smaller quantity, for example dopants. This material will therefore be designated by its main component, that is to say indium phosphide, or InP.

Similarly, the material of the family of gallium-indium arsenide is a semiconductor material composed mainly or exclusively of gallium-indium arsenide, and possibly other components in a much smaller amount, by examples of dopants. This material will therefore be designated by its main component, ie gallium-indium arsenide, or InGaAs.

The photodiode matrix further comprises at least two kinds of doped zones of the same type:

first doped regions 3 formed at least partly in the active layer 5, defining second electrodes for forming, with the first common electrode, photodiodes for forming images,

at least one second doped zone forming a third electrode absorbing excess charge carriers to evacuate them.

Preferably, the first doped zones 3 and the second doped zone 10 have the closest possible doping characteristics, and are preferably formed by the same dopants.

A plurality of second doped regions 10 may be provided for absorbing excess charge carriers and discharging them from the photodiode array.

The second doped zone 10 is formed in the passivation layer 6 and is separated from the active layer 5 by a portion of said passivation layer 6. The second doped zone 10 is therefore not in contact with the active layer 5, while that the first doped zones 3 extend from the passivation layer 6 into the active layer 5. Preferably, the thickness of the portion of the passivation layer 6 separating the second doped zone 10 from the active layer 5 is less than at 0.5 μιη, and is preferably between 0.1 μιη and 0.5 μιη.

The two kinds of doped zones are of the same type, that is to say N or P. For reasons of simplicity, we will present here the case where the two kinds of doped zones are of the P type.

The InP layers are then of the N type, for example doped with silicon. The active layer 5 of InGaAs may be slightly N-doped or remain quasi-intrinsic. Thus the two lower / upper InP layers, that is to say the substrate 4 and the passivation layer 6, and the active layer 5 of InGaAs form a common cathode of the photodiodes in this matrix, said common cathode therefore being the first common electrode already mentioned. The first doped zones 3 then constitute a plurality of anodes formed at least in part in the active layer 5, the cooperation between an anode and the cathode forming a photodiode.

Each of the first doped zones 3 is connected to a read circuit which makes it possible to read the photoelectric signals generated by the photodiodes constituted by said first doped zones 3 and the first common electrode. In particular, the photodiodes are connected to circuits of readings similar to that illustrated in FIG. 3, and the electric potentials Vpd1, Vpd2 that they exhibit, as a function, in particular, of the exposure to which they are subjected and of their polarization before the exposure, are read by these read circuits to determine an image.

The second doped zone 10 is connected by an electrical connection to polarization means configured to apply an adjustable electric potential to said second doped zone 10. Polarization means therefore apply to said second doped zone 10 an electric potential Vring by which is adjusted the absorption of the charge carriers by said second doped zone 10. The Vring electrical potential of the second doped zone 10 is chosen to be lower than the lowest potential among the potentials Vpd1, Vpd2 of the first doped zones 3 so that Vring <min ( Vpd1, Vpd2).

Typically, it is an electrical connection connecting the second doped zone 10 to a power supply by which the Vring electrical potential is imposed and through which the excess charges absorbed by the second doped zone 10 are discharged.

The Vring electric potential applied by said biasing means to said second doped zone 10 may vary within a value range comprising at least:

a first polarization value at which the charge carriers are confined in the active layer due to an energy barrier corresponding to the portion of the passivation layer separating said second doped zone from said active layer; 5; and

a second polarization value at which the portion of the passivation layer 6 separating said second doped zone 10 from said active layer 5 does not cause an energy barrier for the charge carriers of the active layer 5. FIG. 9 schematically illustrates the structure of energy bands according to section AA 'of the photodiode array of FIG. 8, that is to say in a section crossing the substrate 4, the active layer 5 and the layer of Passivation 6. The different levels of energy are represented as a function of depth according to arbitrary scales of depth and energy, for purely illustrative purposes: the energy of the valence band E _v , the energy of the band of conduction E _c , and the level of Fermi E _F.

A zone 15 corresponding to the active layer 5 having an energy of valence E _v greater than that of the two zones bordering it, ie an area 14 corresponding to the substrate 4 and an area 16 corresponding to the passivation layer 6. The holes 9 constituting here the charge carriers are thus confined in the active layer 5.

In a similar manner to FIG. 9, FIG. 10 schematically illustrates the structure of energy bands according to the section BB 'of the photodiode array of FIG. 8, that is to say in a section crossing the substrate 4. the active layer 5 and a first doped zone 3.

It is then clear that a zone 13 corresponding to the first doped zone 3 has a higher valence energy E _v than the zones 15 and 14, respectively corresponding as above to the active layer 5 and to the substrate 4. The holes 9 constituting here the charge carriers are not confined in the active layer 5, and their passage through the first doped zone 3 is possible. FIGS. 11, 12 and 13 schematically illustrate the structure of energy bands according to the section CC of the photodiode array of FIG. 8, that is to say in a section crossing the substrate 4, the active layer 5 , the passivation layer 6 and a second doped zone 10, in a manner similar to that of FIGS. 9 and 10 for their respective cuts.

FIG. 11 illustrates a case in which the Vring electrical potential applied to the second doped zone 10 corresponds to a first polarization value at which the charge carriers are confined in the active layer 5 because of a barrier of energy corresponding to the portion of the passivation layer 6 separating said second doped zone 10 from said active layer 5. For example, it may be a weak polarization, applied in the case of low light to limit or prevent the absorption of charge carriers by the second doped zone 10.

FIG. 11 shows a zone 16 corresponding to the portion of the passivation layer 6 separating said second doped zone 10 from said active layer 5, said zone 16 having a lower energy E _v than the two zones 1 framing, that is to say the zone 17 corresponding to the second doped zone 10 and the zone 15 corresponding to the active layer 5. This zone 16 thus makes it possible to confine the holes 9 in the active layer 5 by defining a barrier of potential preventing them from joining the second doped zone 10.

FIG. 12 has the same configuration as FIG. 11, but in the case of a Vring electric potential applied to the second doped zone 10 by the biasing means whose value is situated between the first value and the second potential value. mentioned above, for example a more negative voltage than that applied in the case of Figure 1 1.

There is a decrease in the height of the potential barrier of the zone 16 corresponding to the portion of the passivation layer 6 separating said second doped zone 10 from said active layer 5, while the valence energies E _v and conduction E _c of the zone 17 corresponding to the second doped zone 10 rise. Some of the holes 9 are still confined in the active layer 5 because of the presence of the potential barrier of the zone 16, while some holes 9 cross this barrier to join the second doped zone 10.

FIG. 13 has the same configuration as FIGS. 11 and 12, but in the case of a Vring electric potential applied to the second doped zone 10 by the polarization means corresponding to the second polarization value at which the portion of the passivation layer 6 separating said second doped zone 10 from said active layer 5 does not cause an energy barrier for the charge carriers of the active layer 5. For example, it is a more negative voltage than those of Figures 1 1 and 12.

Thus, the disappearance of the potential barrier of the zone 16 corresponding to the portion of the passivation layer 6 separating said second doped zone 10 from said active layer 5. The holes 9 are no longer confined in the active layer 5 because of the disappearance of this barrier and the high energy level of the valence band at the zone 17 corresponding to the second doped zone 10, and can therefore join said second doped zone 10. It is clear that the applied potential Vring allows adjust the passage of the charge carriers from the active layer 5 to the second doped zone 10, and thus modulate the absorption of the charges by said second doped zone 10.

Preferably, the potential of the second doped zone 10 is modulated according to the level of illumination on the photodiode array. For this purpose, an illumination measurement can be provided on the photodiode array, in particular by means of the readout circuit as illustrated in FIG. 3. This illumination measurement makes it possible to determine which potential must be applied to the second doped zone 10. It is also possible to reduce the resistivity of the second doped zone by seconding it by a metal grid covering said second doped zone 10 so that the application of the potential, as well as the drainage of the charges, is uniform. This metal grid can also be used to connect together several second doped zones 10, thus fulfilling the role of connection and polarization means for applying the Vring potential.

The second doped zone 10 is located between at least some of the first doped zones 3 in order to separate them. Thus, in FIG. 8, the sectional view shows an alternation between the first doped zones 3 and one or more second doped zones 10. Thus, in the direction of the section, the second doped zone or zones 10 separate the first doped zones. 3 constituting the anodes of the photodiodes in order to absorb the excess charges likely to pass via the active layer 5 from a first doped zone 3 to the other. FIG. 14 shows a view from above of an embodiment in which first doped zones 3 are each at least partially surrounded by a doped zone 10 of the same type, here of N type, as said first doped zones 3, and formed at least partly in the active layer 5, to separate each of the anodes formed by said first doped areas 3 of the other anodes of said matrix. FIG. 15 shows a view from above of an embodiment in which the second doped zone 10 forms a grid between first doped zones 3 in order to individually surround first doped zones 3. Preferably, in order to reduce the complexity of manufacturing as well as interconnections, a single doped zone 10 is distributed on the surface of the matrix of photodiodes. However, it is possible to choose to have a plurality of second doped zones 10, as for example in the embodiment illustrated in FIG. 16, in which a plurality of second doped zones 10 are distributed parallel to each other and interspersed with first doped zones. FIG. 17 shows another example, in which the matrix comprises a plurality of second doped zones 10 distributed between the first doped zones 3 along the diagonals of the photodiode array, so that the majority of said second doped zones 10 are each adjacent to four first doped zones 3.

In the examples illustrated in FIGS. 1 and 15, all the anodes 3 are surrounded by one or more second doped zones 10. However, it is not strictly necessary, although preferable and coherent, for all the photodiodes to be surrounded. Nevertheless, in order to obtain a significant reduction in the crosstalk between photodiodes, preferably the majority of the photodiodes are surrounded by at least a second doped zone 10.

Similarly, in the examples illustrated in FIGS. 1 and 15, the first zones 3 are completely surrounded by doped second zones 10. However, a doped zone 10 around a first doped zone 3 may have openings, and thus only partially surround a first doped zone 3.

The fact of not completely surrounding first doped areas 3 with at least one second doped zone 10 may be dictated by manufacturing considerations but also to optimize the operation of the photodiode array. Indeed, the second doped zones compete with the photodiodes at the level of the charge carriers. In order to limit this competition, it can be expected that the second or second doped zones 10 do not completely surround the anodes, but nevertheless sufficiently to significantly reduce the crosstalk between photodiodes. The second doped zone 10 is separated from the first doped zones 3 by a sufficient distance so that the space charge areas associated respectively with the second doped zone 10 and the first doped zones 3 are separated. Thus, preferably, the second doped zone 10 is distant from the anode and surrounds it with at least 0.5 μιη.

Preferably, a second doped zone 10 has a width (top view) of at least 0.5 μιη in order to sufficiently insulate the photodiodes from each other. The width, (top view) of a doped zone 10 can thus extend to for example 2 μιη, or even reach 5 μιη.

A matrix of photodiodes according to the invention can naturally be manufactured by means of two selective doping steps:

a step of deep doping by means of a mask exposing the parts of the substrate intended to receive the first doped zones, and

- Another step of less deep doping by means of another mask exposing the parts of the substrate for receiving the second doped zones, the order of these two steps being unimportant. If such a manufacturing method has the advantage of simplicity, it is however not optimal, insofar as the operations of mask change and / or successive doping are long, risky and therefore expensive. It is therefore presented below two manufacturing processes with substantial advantages in terms of speed and reliability.

Thus, according to a second aspect, the invention also relates to a method for manufacturing a photodiode matrix according to the first aspect. Referring to Figures 18, 19 and 20, from a first electrode comprising at least one substrate layer 4 of a material of the family of indium phosphide and an active layer 5 of a material of the family of the gallium-indium arsenide, and a passivation layer 6 made of a material of the indium phosphide family, the active layer 5 being situated between the substrate layer 4 and the passivation layer 6, said process comprising the steps whereby :

a selective etching of the passivation layer 6 is carried out (FIG. 19), the first doped zones 3 and the said at least one second doped zone 10 are formed during the same selective doping step, the said first doped zones 3 being formed at the zones 11 of the selective etching of the passivation layer 6 previously made (FIG. 20).

Selective etching of the passivation layer allows removal of material at the zones 11 intended to form the first doped zones 3. During the subsequent step of selective doping, the dopants at the level of the etched zones 11, thus forming the first doped zones 3, thus penetrate further into the stack constituted by the passivation layer 6 and the active layer 5, until it reaches the latter. On the other hand, the dopants outside these etched zones 11, thus forming the second doped zones 6, do not reach the active layer 5 because of the additional thickness of the passivation layer 6 in the non-etched zones. It is therefore necessary that this additional thickness of the passivation layer 6 is sufficient so that, during the same doping step, the first doped zones 3 reach the active layer 5 while the second doped zones 10 do not reach this layer active 5.

As a result, the selective etching of the passivation layer 6 must remove a thickness of the passivation layer 6 greater than the thickness of the portion of said passivation layer 6 finally separating the second doped zones 10 from the active layer 5 .

Once the etching has been performed, the doping step can then be performed at the same time for the formation of the first doped zones 3 and the second doped zones 10, for example by means of a mask 12 with recessed zones corresponding to the first zones doped 3 and the second doped zones 10.

With reference to FIGS. 21, 22 and 23, from a first common electrode comprising at least one substrate layer 4 made of a material of the indium phosphide family and an active layer 5 made of a material of the family of gallium-indium arsenide, and a passivation layer 6 of a material of the indium phosphide family, the active layer 5 being located between the substrate layer 4 and the passivation layer 6, another method comprises the steps according to which:

a first selective doping is performed to begin forming the first doped zones 3 (FIG. 22);

a second selective doping is then performed to finish forming the first doped zones 3 and to form the second doped zone 10 (FIG. 23). To this end, it is possible to use a so-called "hardmask" film 12 deposited on the surface of the passivation layer 6 and made of a polymer that can be etched, for example by plasma, in order to create hollow zones through which doping zones are doped. -jacentes. Thus, for the first selective doping, the film 12 has recessed areas corresponding to the location of the first doped areas 3. Then, in the film 12 are made other recessed areas corresponding to the location of the second doped zones 10 in order to that during the second selective doping, the film 12 called "hardmask" has recessed areas corresponding to the location of the first doped areas 3 and the location of the second doped zones 10.

For example, it is possible to obtain the first common electrode for the implementation of the various processes by the following steps:

epitaxial growth of an active layer 5 in a material of the family of gallium indium arsenide InGaAs on a substrate 4 made of a material of the indium phosphorus family, and

epitaxial growth of a passivation layer 6 in a material of the indium phosphorus family InP on the active layer 5. The first doped zones 3 and said at least one second doped zone 10 can in turn be formed by a selective diffusion of zinc as a P-type dopant in the passivation layer 6 and, for the first doped zones 3, in the active layer 5, when said layers are N-type. The doping is preferentially by diffusion.

Claims

claims

1. Photodiode array comprising

a first common electrode of a PN junction, comprising at least one substrate layer (4) made of a material of the indium phosphide family and an active layer (5) made of a material of the family of the arsenide of gallium-indium, - a passivation layer (6) made of a material of the indium phosphide family, the active layer (5) being located between the substrate layer (4) and the passivation layer (6), and

at least two kinds of doped zones of the same type:

first doped zones formed at least partially in the active layer, defining second electrodes to form, with the first common electrode, photodiodes connected to reading circuits and adapted for image formation; ,

at least one second doped zone (10) forming a third electrode adapted to absorb excess charge carriers to evacuate them, polarization means being adapted to apply to said second doped zone an electric potential (Vring) by which is adjustable absorption of charge carriers by said second doped zone (10),

characterized in that said at least one second doped region (10) is formed in the passivation layer (6) and is separated from the active layer (5) by a portion of said passivation layer (6).

2. Matrix according to the preceding claim, wherein the electric potential (Vring) applied to the second doped zone (10) is modulated according to the level of illumination on the photodiode array.

3. Matrix according to one of the preceding claims, wherein the second doped zone (10) is located between at least some of the first doped zones (3).

4. Matrix according to one of the preceding claims, wherein the second zone (10) doped individually surrounds first doped areas (3).

5. Matrix according to one of claims 1 to 4, wherein a plurality of second doped zones (10) are distributed parallel to each other and interspersed with first doped zones (3).

6. Matrix according to one of claims 1 to 4, comprising a plurality of second doped zones (10) distributed between the first doped zones (3) along the diagonals of the photodiode array.

7. Matrix according to one of the preceding claims, wherein the second doped zone (10) is separated from the first doped zones (3) by a sufficient distance so that the space charge areas associated respectively with the second doped zone (10) and the first doped areas (3) are separated.

8. Matrix according to one of the preceding claims, wherein a metal gate on the surface of said matrix connects different points of the second or second zone (s) doped (s) (10) to homogenize the electrical potential the second doped zone (s) (10).

An image sensor incorporating a photodiode array according to any one of the preceding claims.

10. A method of manufacturing a photodiode array according to one of claims 1 to 8, said method comprising the steps according to which, from a first common electrode comprising at least one substrate layer (4) of a material of the family of indium phosphide and an active layer (5) of a material of the gallium-indium arsenide family, and a passivation layer (6) of a material of the phosphide-indium family indium, the active layer (5) being located between the substrate layer (4) and the passivation layer (6):

selective etching of the passivation layer (6) is carried out

the first doped zones (3) and said at least one second doped zone (10) are formed, said first doped zones (3) being formed at the zones of the selective etching of the passivation layer (6) previously produced.

1 1. Method according to the preceding claim, wherein the first doped zones (3) and said at least one second doped zone (10) are formed during the same selective doping step.

12. Method according to one of claims 10 to 1 1, wherein the selective etching of the passivation layer removes a thickness of the passivation layer (6) greater than the thickness of the portion of said passivation layer (6) finally separating the second doped zones (10) from the active layer (5).

13. A method of manufacturing a photodiode array according to one of claims 1 to 8, said method comprising the steps according to which, from a first common electrode comprising at least one substrate layer (4) of a material of the family of indium phosphide and an active layer (5) of a material of the gallium-indium arsenide family, and a passivation layer (6) of a material of the phosphide-indium family indium, the active layer (5) being located between the substrate layer (4) and the passivation layer (6):

a first selective doping is performed to start forming the first doped zones (3),

a second selective doping is then performed to finish forming the first doped zones (3) and to form said at least one second doped zone (10).