Current-assisted photonic demodulator comprising vertically arranged modulating and collecting doped regions located in a compression zone
The vertical arrangement of doped regions and mechanical stresses in the photonic demodulator enhance near-infrared detection by reducing dark current and improving AC demodulation contrast and bandwidth.
Patent Information
- Application Number
- FR2022013208
- Authority / Receiving Office
- FR · FR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-12-12
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2042-12-12
AI Technical Summary
Current-assisted photonic demodulators struggle to detect light radiation in the near infrared effectively while minimizing dark current and maintaining good performance in terms of AC demodulation contrast and bandwidth.
A current-assisted photonic demodulator design featuring a detection portion made of germanium with doped regions arranged vertically and a peripheral lateral portion made of a material with a lower thermal expansion coefficient, inducing mechanical stresses in tension and compression, reducing dark current and enhancing AC demodulation contrast and bandwidth.
The vertical arrangement of doped regions and mechanical stresses improve the demodulator's performance by reducing dark current and increasing AC demodulation contrast and bandwidth, enabling detection beyond 1.55 μm cut-off wavelength.
Smart Images

Figure 00000023_0000 
Figure 00000023_0001 
Figure 00000024_0000
Abstract
Description
Title of the invention: Current-assisted photonic demodulator comprising doped modulation and collection regions arranged vertically and located in a compressed zone Technical field
[0001] The field of the invention is that of current-assisted photonic demodulators (CAPD) suitable for detecting light radiation in the near infrared. The invention finds application in particular in telemetry, biological analysis, and industrial inspection (contactless detection of surface defects). STATE OF THE PRIOR ART
[0002] Current-assisted photonic demodulators are photodetectors in which the distribution of a drift electric field is modulated. They were initially described in particular in the scientific article by Van Nieuwenhove et al. entitled Novel Standard CMOS Detector using Majority Carrent for giding Photo-Generated Electrons towards Detecting Junctions, Proc. Symp. IEEE / LEOS Benelux Chapter, pp. 229-232, 2005. This type of optoelectronic device is particularly used in time-of-flight (TOF) telemetry.
[0003] Such a demodulator usually comprises a detection portion made from a lightly doped p-type crystalline semiconductor material, which has, at one of its faces, two p+ doped regions for generating and modulating a drift current, as well as two n+ doped regions located near the p+ doped regions for collecting the photocurrent. An electric potential difference is applied between the p+ doped regions, which generates a drift electric field in the detection portion. Also, when the light radiation is absorbed in the detection portion, an electron-hole pair is generated, then the photogenerated hole propagates under the effect of the drift field towards the p+ doped region having the lowest electric potential, while the photogenerated electron is directed towards the opposite p+ doped region, then is collected by the adjacent n+ doped region.Thus, the photocurrent (minority electrons) can be efficiently measured by the demodulator.
[0004] Due to the separation between the majority hole current and the photocurrent (minority electrons), the contribution of the majority hole current to the Schottky noise (shot noise in English) as well as to the thermal noise (thermal noise in English) is thus limited. However, there is a need for such a current-assisted photonic demodulator, suitable for detecting in the near infrared, while showing good performance. Statement of the invention
[0005] The invention aims to propose a current-assisted photonic demodulator, suitable for detecting in the near infrared with a cut-off wavelength greater than 1.55 pm, and having a reduced dark current. Such a demodulator can then have good performance, particularly in terms of AC demodulation contrast and bandwidth.
[0006] For this, the object of the invention is a current-assisted photonic demodulator, suitable for detecting light radiation of interest, comprising: - a detection portion, having a first face and a second face opposite each other and parallel to a main plane, and comprising a central zone for detecting the incident light radiation. The detection portion is made of a first crystalline semiconductor material based on germanium, and comprises doped regions, including: at least two doped modulation regions, p-type doped, and intended to generate and modulate a drift current in the detection portion; and at least one doped collection region, n-type doped, and intended to collect the minority charge carriers photogenerated during the absorption of the light radiation of interest; - a dielectric passivation layer, made from an electrically insulating material, and covering the first face; and electrodes, located at the first face and passing through the dielectric passivation layer, including modulation electrodes in electrical contact with the doped modulation regions; and at least one collection electrode in electrical contact with at least the doped collection region.
[0007] According to the invention, the photonic demodulator comprises a peripheral lateral portion, surrounding the detection portion in the main plane, made from a second material having a thermal expansion coefficient lower than that of the first material, the detection portion then having mechanical stresses in tension in the main plane and mechanical stresses in compression along a vertical axis orthogonal to the main plane.
[0008] In addition, said doped regions are arranged in a vertical arrangement such that: at least one region doped according to a first type of conductivity, called the upper doped region, is flush with the first face and is located close to the central zone; and at least two regions doped according to a second type of conductivity opposite to the first type, called the lower doped regions, are located at a distance along the vertical axis from the upper doped region.
[0009] Some preferred but non-limiting aspects of this photonic demodulator are as follows.
[0010] Each lower doped region may extend parallel to the main plane towards the central zone and the upper doped region, and has a distal end distant from the central zone and a proximal end oriented towards the central zone.
[0011] The demodulator may comprise conductive vias extending along the vertical axis of the first face and in electrical contact with the lower doped regions.
[0012] The conductive vias may be distinct from the peripheral lateral portion.
[0013] Alternatively, the conductive vias may be parts, distinct from each other, of the peripheral lateral portion, the latter then being made of an electrically conductive material.
[0014] The electrode(s) in contact with the upper doped region(s) may be closer to the central area, in the main plane, than the electrodes in contact with the conductive vias.
[0015] The demodulator may comprise two lower doped modulation regions, and may comprise a single upper doped collection region, located in the central zone and at a distance, in the main plane, from the lower doped modulation regions.
[0016] The demodulator may comprise two upper doped regions, each located perpendicular to one of the lower doped regions.
[0017] For an upper doped region and a vertically adjacent lower doped region, a vertical distance, along the vertical axis, between said doped regions may be less than a horizontal distance, in the main plane, between the electrode connected to the upper doped region and the electrode connected to the conductive via of the adjacent lower doped region.
[0018] The upper doped region(s) may be collection regions, and the lower doped regions may then be modulation regions.
[0019] The peripheral lateral portion can be made from silicon. The detection portion can then comprise a lateral zone made from SiGe, located at the interface with the peripheral lateral portion.
[0020] The invention also relates to a method of manufacturing a photonic demodulator according to any one of the preceding characteristics, comprising the following steps: - production, by epitaxy from a support layer, of a first layer based on doped germanium; - production of the lower doped regions by localized etching of the first layer based on doped germanium; - production by epitaxy of a main layer based on unintentionally doped germanium, on and in contact with the lower doped regions; - creation of a trench through the main layer surrounding the detection portion; - production of the peripheral lateral portion in the trench, at a temperature above ambient temperature, so that, after returning to ambient temperature, the detection portion is in tension in the main plane and in compression along the vertical axis; - production of one or more upper doped regions in the detection portion; - deposition of a dielectric passivation layer on the detection portion; - production of modulation electrodes and collection electrodes.
[0021] The method may comprise, before the creation of the trench, a step of creating two first notches opening onto the lower doped regions, followed by a step of creating conductive vias in the first notches.
[0022] The method may be such that, when the trench is made, it opens onto the lower doped regions, so that the peripheral lateral portion is then in contact with them.
[0023] The upper doped region(s) may be produced by localized ion implantation. Brief description of the drawings
[0024] Other aspects, aims, advantages and characteristics of the invention will appear better on reading the following detailed description of preferred embodiments thereof, given by way of non-limiting example, and made with reference to the appended drawings in which:
[0025] [Fig.lA] and [Fig.lB] are schematic and partial views, in cross-section ([Fig.lA]) and in top view ([Fig.lB]), of a current-assisted photonic demodulator according to one embodiment;
[0026] [Fig.1C] illustrates an evolution of the absorption, transmission and reflection rates of the detection portion, as a function of its thickness;
[0027] [Fig.2A] is a schematic and partial view, in cross-section, of a current-assisted photonic demodulator according to an alternative embodiment;
[0028] [Fig.2B] and [Fig.2C] are schematic and partial views, in cross-section ([Fig.2B]) and in top view ([Fig.2C]), of a current-assisted photonic demodulator according to another embodiment variant;
[0029] Figures 3A to 3L illustrate different steps of a method of manufacturing a demodulator similar to that of [Fig.lA].
[0030] DETAILED DESCRIPTION OF PARTICULAR EMBODIMENTS
[0031] In the figures and in the remainder of the description, the same references represent identical or similar elements. Furthermore, the different elements are not shown to scale so as to enhance the clarity of the figures. Furthermore, the different embodiments and variants are not mutually exclusive and may be combined with each other. Unless otherwise indicated, the terms "substantially", "approximately", "of the order of" mean to within 10%, and preferably to within 5%. Furthermore, the terms "between ... and ..." and equivalents mean that the limits are included, unless otherwise indicated.
[0032] The invention relates to a current-assisted photonic demodulator, suitable for detecting light radiation in the near infrared spectral band (SWIR) corresponding to the spectral range from approximately 0.8 pm to 1.7 pm, or even approximately 2.5 pm. Furthermore, it preferably belongs to a matrix photodetector comprising a matrix of detection pixels identical to each other, where each detection pixel comprises a planar demodulator.
[0033] Generally speaking, the demodulator comprises: • a detection portion, made from germanium, of which a so-called central zone is the main absorption site of the light radiation to be detected. It extends vertically between an upper face F1 and a lower face F2, and horizontally in the main plane XY. It comprises: at least two modulation regions, p-type doped, and adapted to generate and modulate the drift current; and at least one collection region, n-type doped, and adapted to collect the photogenerated minority carriers; • a dielectric passivation layer, made of at least one electrically insulating material, which covers the upper face Fl; • electrodes located at the level of the upper face Fl and passing through the dielectric passivation layer, including modulation electrodes Ml, M2 in electrical contact with the doped modulation regions; and at least one collection electrode Cl, C2 in electrical contact with the doped collection region(s).
[0034] According to the invention, as detailed below, the germanium-based detection portion has tensile mechanical stresses in the main XY plane, and compressive mechanical stresses oriented along the vertical Z axis. These horizontal tensile stresses of the germanium of the detection portion allow the demodulator to detect in the near infrared spectral band up to a cut-off wavelength greater than 1.55 pm (value of the cut-off wavelength of relaxed germanium). In addition, by Poisson effect, the detection portion undergoes vertical compression.
[0035] To generate these mechanical stresses, the demodulator comprises a peripheral lateral portion, made from a material whose coefficient of thermal expansion is lower than that of the material of the detection portion, and which surrounds the detection portion in the XY plane. Thus, following the production of the peripheral lateral portion, the mechanical stresses are generated in the detection portion.
[0036] In addition, the vertical compression of the detection portion is used to improve the performance of the demodulator. For this, instead of arranging the doped collection region(s) in a coplanar manner with the doped modulation regions, as in the prior art, these are arranged in a vertical configuration, while maintaining an electrical polarization from the same upper face F1. More precisely, the doped regions are arranged in the following manner: • at least one doped region is flush with the upper face Fl and is located close to the central zone. It is then called the 'upper doped region'. The upper doped region(s) are doped according to a first type of conductivity; • at least two doped regions are located at a distance along the vertical Z axis from the upper doped region(s). They are then called 'lower doped regions'. Thus, the upper doped region(s) are located above the lower doped regions along the vertical Z axis. They are doped according to a second type of conductivity opposite to the first type. The lower doped regions are in electrical contact with the corresponding electrodes by means of conductive vias, which may be either separate from the peripheral lateral portion or parts of the peripheral lateral portion.
[0037] Thus, this vertical arrangement of the doped collection and modulation regions, within a detection portion in vertical compression, makes it possible in particular to reduce the dark current between the doped collection region(s) and the doped modulation regions, due to an increase in the indirect bandgap energy of the detection portion, thus improving the performance of the demodulator. In addition, it is possible to reduce the (vertical) distance between the doped collection region(s) and the doped modulation regions, which contributes to improving the AC demodulation contrast as well as the bandwidth. In the case of a matrix photodetector, it is then possible to reduce the pixel pitch, thus improving the resolution of the matrix photodetector.
[0038] Generally speaking, the collection doped region(s) may be the upper doped regions, and the modulation doped regions are then the lower doped regions. The reversed configuration is possible, where the doped regions of collection are the lower doped regions, and the modulation doped regions are the upper doped regions.
[0039] Figures 1A and 1B are schematic and partial views, in cross-section ([Fig.1A]) and in top view ([Fig.1B]), of a current-assisted photonic demodulator 1 according to one embodiment, here belonging to a matrix of identical planar demodulators 1.
[0040] In this example, the detection portion 10 comprises two upper doped regions 11 which are here n+ doped collection regions, and two lower doped regions 12 which are then p+ doped modulation regions. Furthermore, the doped modulation regions are connected to the modulation electrodes M1 and M2 by dedicated conductive vias 3 (therefore distinct from the peripheral lateral portion 2) and made from germanium. Furthermore, the lower face F2 is the one which here receives the light radiation to be detected.
[0041] Here and for the remainder of the description, a three-dimensional direct reference frame XYZ is defined, where the X and Y axes are parallel to the main plane of the demodulators 1, and where the vertical direction +Z is oriented along the thickness of the detection portion 10 of the demodulator 1, from the lower face F2 towards the upper face F1. The terms “lower” and “upper” refer to an increasing positioning along the +Z direction defined in [Fig. 1A]. Furthermore, the term “horizontal” refers to an orientation parallel to the XY plane and the term “vertical” refers to an orientation parallel to the Z axis.
[0042] In this example, the demodulators 1 have a planar configuration insofar as they are made here from the same main semiconductor layer 23 (see [Fig. 3C]). The latter extends in a main plane XY between the upper faces F1 and lower faces F2 opposite each other and parallel to the main plane. The two faces F1, F2 therefore extend along identical planes for each of the demodulators 1, and delimit vertically (along the thickness axis Z) the detection portions 10 of the demodulators 1. Also, the demodulators 1 do not have a mesa structure insofar as they are made from the same main semiconductor layer 23.
[0043] Each demodulator 1 comprises a detection portion 10 made from germanium and therefore adapted to detect light radiation in the near infrared (SWIR). Insofar as the detection portion 10 is made from germanium and undergoes mechanical stresses in tension in the XY plane, the demodulator 1 is adapted to detect light radiation at a cut-off wavelength greater than 1.55 pm.
[0044] The thickness of the detection portion 10, defined along the vertical axis Z between the upper face F1 and lower face F2, is here substantially constant for a demodulator 1 to the other, for example is between a few hundred nanometers and a few microns, for example between approximately 1 pm and 5 pm. The thickness is chosen so as to obtain good absorption in the wavelength range of the light radiation to be detected. The detection portion 10 has a transverse dimension in the XY plane which can be between a few hundred nanometers and a few tens of microns, for example between approximately 1 pm and 20 pm, for example equal to 1 pm.
[0045] The detection portion 10 is made from germanium, that is to say that the semiconductor material(s) are germanium or a compound (binary or ternary, etc.) formed from at least germanium. Thus, the detection portion 10 can be made, for example, from germanium Ge, silicon germanium SiGe, germanium tin GeSn, or even silicon germanium tin SiGeSn. It can thus be made from the same semiconductor material and have regions of different types of conductivity (homojunction) so as to form a pn or pin junction. It can alternatively be a stack of sub-layers of different semiconductor materials (heterojunction), which are then formed from germanium.
[0046] The detection portion 10 comprises an intermediate region 13, not intentionally doped (with possible residual p-type doping) or lightly p-doped. The intermediate region 13 extends between the faces F1 and F2, as well as in the XY plane, and forms the main absorption region of the light radiation to be detected.
[0047] A central zone Zc of the detection region 10 is defined as being the main place of absorption of the light radiation to be detected. This central zone Zc is advantageously delimited in the XY plane by an optical mask located on the side of the face F1 or F2 which receives the light radiation to be detected. Preferably, the light radiation to be detected is incident on the side of the lower face F2 (called the front face), the lower doped regions 12 can then form this optical mask and thus participate in defining the optical window of the demodulator 1, in particular when they are doped with p-type boron (modulation), as explained below.
[0048] The upper doped regions 11 are therefore here n-type doped regions, preferably n+ doped. They are adapted to collect the photogenerated minority carriers (photocurrent) resulting from the absorption of the light radiation to be detected in the central zone Zc of the detection portion 10 (in the intermediate region 13), via the electrical potential imposed on them by the collection electrodes C1 and C2. They are here n+ doped, and may have a doping which may be between 5x1018 and 2x1020 at / cm3 approximately.
[0049] The upper doped regions 11 are flush with the upper face F1 and are located close to the central zone Zc. By flush is meant "reaching the level of", or "extending from". They extend along the vertical axis Z over a depth predefined. Also, each upper doped region 11 is here surrounded in the XY plane and along the -Z direction by the intermediate region 13. It opens directly onto the upper face Fl, without being spaced from it by a zone of the intermediate region 13.
[0050] The two upper doped regions 11 are arranged in the XY plane on either side of the central zone Zc, and close to it. Thus, the collection electrodes C1 and C2 are located close to the central zone Zc, while the modulation electrodes M1 and M2 are distant from it. More generally, the electrodes of the upper doped regions 11 are located close to the central zone Zc, while the electrodes of the lower doped regions 12 are distant from the central zone Zc.
[0051] Note that the upper doped regions 11 are here produced by localized ion implantation in the detection portion 10 from the upper face F1. Alternatively, they can be produced by growth doping during epitaxy resumption in notches formed from the upper face F1 of the detection portion 10.
[0052] The lower doped regions 12 are here p-type doped regions, preferably p+ doped. They are adapted to generate and modulate the drift current via the electrical potential imposed on them by the modulation electrodes M1 and M2. They are here p+ doped, and have for example a doping of between 1018 and 1020 at / cm3 approximately, preferably 1019 at / cm3.
[0053] The lower doped regions 12 are located at a distance along the vertical axis Z from the upper doped regions 11. In other words, the upper doped regions 11 are located above the lower doped regions 12 along the vertical axis Z (with or without overlapping thereof). The lower doped regions 12 are therefore spaced from the upper doped regions 11 by a zone of the intermediate region 13 in vertical compression. Furthermore, the lower doped regions 12 are here spaced from the lower face F2 by a non-zero distance, but alternatively, they could be flush with the lower face F2 (see variants of [Fig.2A] and 2B).
[0054] The lower doped regions 12 are located in the XY plane on either side of the central zone Zc of the detection portion 10. Also, the electrodes (here M1 and M2) of the lower doped regions 12 are further away from the central zone Zc than those of the upper doped regions 11.
[0055] The lower doped regions 12 extend parallel to the XY plane towards the central zone Zc and the upper doped regions 11, between a distal end 12d (i.e. far from the central zone) and a proximal end 12p (i.e. close to the central zone). The terms 'distal' and 'proximal' are defined with respect to the central zone Zc. The distal end 12d is the area of the electrical connection to a modulation electrode, here by means of a dedicated conductive via 3, while the proximal end 12p is the main drift current generation area.
[0056] In this example, the lower doped regions 12 extend in the XY plane in the direction of the central zone Zc until they become perpendicular to the upper doped regions 11. In other words, in projection in the XY plane, there is overlap of at least a part of the lower doped region 12 by the adjacent upper doped region 11. Alternatively, as illustrated in [Fig.2B], the lower doped regions 12 may not be perpendicular to the upper doped regions 11.
[0057] Note that the lower doped regions 12 are produced by growth doping during an epitaxy resumption taking place during the production of the detection portion 10. Each lower doped region 12 has the same predefined thickness along the vertical axis Z, and substantially constant in the XY plane, for example equal to a few tens or hundreds of nanometers. The thickness is here equal to approximately 200nm, but it can be between 100 and 500nm.
[0058] The lower doped regions 12 (here modulation) are electrically connected to the electrodes (here M1, M2) by means of conductive vias 3 which extend along the vertical axis Z between the electrodes concerned and the distal end 12d of the lower doped regions 12. In this example, the conductive vias 3 are distinct from the peripheral lateral portion 2. They are made of at least one electrically conductive material, here of a material based on p-type doped germanium, deposited in notches 24 made in the main semiconductor layer 23 (see [Fig.3F]) from the upper face F1 and opening onto the lower doped regions 12.
[0059] The detection portion 10 is delimited laterally, in the XY plane, by a peripheral lateral portion 2, made from a material which has a coefficient of thermal expansion lower than that of the germanium-based material of the detection portion 10. The material is preferably silicon-based, and is preferably polysilicon. It may also be monocrystalline silicon, amorphous silicon, or even silicon germanium. The material may optionally be doped. The peripheral lateral portion 2 surrounds the detection portion 10 in the XY plane, either continuously (see [Fig. 1B]), or discontinuously (see [Fig. 2C]).
[0060] Due to the thermal expansion differential, following the production of the peripheral lateral portion 2 by deposition at a temperature higher than ambient temperature, for example by epitaxy at a temperature between 500°C and 700°C, the detection portion 10 has mechanical stresses in horizontal tension, and by Poisson effect, mechanical stresses in vertical compression.
[0061] The peripheral lateral portion 2 further ensures lateral optical isolation of the demodulators 1 in the XY plane. Here, it preferably extends over the entire thickness from the detection portion 10 to open onto the support layer 5.
[0062] When the material of the peripheral lateral portion 2 is made from silicon, a lateral zone 14 made from silicon germanium may be present between and in contact with the detection portion 10 and the peripheral lateral portion 2. The lateral zone 14 has a bandgap energy greater than that of the detection portion 10 made from germanium. This lateral “gap opening” makes it possible to reduce the sensitivity of the demodulator 1 to defects present near the peripheral lateral trench. This also improves the performance of the demodulator 1.
[0063] Furthermore, the detection portion 10 rests on a support layer 5. This can be made of a semiconductor crystalline material suitable for the epitaxy of the germanium of the detection portion 10. This material can be doped or not. This support layer 5 can be an upper layer of a SiGeOi substrate or a GeOi substrate. Here it is made of silicon.
[0064] In the case here where the lower doped regions 12 are not in contact with the support layer 5 (see [Fig. 1A]), the material of the support layer 5 can be doped and brought to an electrical potential, for example via the peripheral lateral portion 2. In the opposite case where the lower doped regions 12 are in contact with the support layer 5 ([Fig. 2A]), the material of the latter is not intentionally doped. As a variant, the detection portion 10 can rest on a support layer 5 made of an insulating material.
[0065] The demodulator 1 comprises a dielectric passivation layer 4, made of at least one electrically non-conductive material, such as an insulator or an intrinsic semiconductor. It covers the upper face F1 and makes it possible to passivate the detection portion 10 and to electrically isolate the electrodes M1, M2, C1, C2. It is thus in contact with the intermediate region 13, the upper doped regions 11, the conductive vias 3, and the peripheral lateral portion 2. It is preferably made of an oxide, such as an oxide of silicon, aluminum, germanium, hafnium, etc. or even intrinsic silicon. It has a thickness, for example, between 2 nm and 500 nm.
[0066] The demodulator 1 comprises electrodes M1, M2, Cl, C2 located on the side of the upper face Fl, which pass through the dielectric passivation layer 4 and are electrically connected to the upper 11 and lower 12 doped regions. The modulation electrodes M1, M2 make it possible to generate and modulate the drift current. Here, they are in contact with the conductive vias 3, and thus make it possible to apply a positive or zero electrical potential to the lower doped regions 12 (here modulation). They are connected to an electrical control circuit. The collection electrodes Cl, C2 make it possible to collect the photogenerated electrons (photocurrent). Here, they are at contact of the upper doped regions 11 (here collection) to apply a positive electrical potential to them. They are connected to an electrical reading circuit. As indicated above, the electrodes of the lower doped regions 12 are further away from the central zone Zc than the electrodes of the upper doped regions 11.
[0067] The demodulator 1 is thus adapted to detect in the near infrared, with a cut-off wavelength greater than 1.55 pm. This is due to the tensioning in the XY plane of the germanium-based detection portion 10, which leads to a reduction in the optical gap of the material of the detection portion 10. In addition, this horizontal tension of the detection portion 10 makes it possible to increase the mobility of the charge carriers, and therefore improves the bandwidth of the demodulator 1.
[0068] In addition, the vertical compression of the detection portion 10 leads to an increase in the indirect gap of the germanium-based material of the detection portion 10, and therefore to a reduction in the dark current between each doped modulation region and the adjacent doped collection region, which are arranged vertically.
[0069] It is then possible to reduce the vertical distance between each doped modulation region and the adjacent doped collection region. This improves the AC demodulation contrast and the bandwidth of the demodulator 1, while eliminating the risk of an increase in the leakage current. Furthermore, it is possible to reduce the pixel pitch, in the case of a matrix photodetector.
[0070] In the case of rear face lighting (lower face F2), and when the lower doped regions 12 are the p+ doped regions (germanium doped for example with boron with a concentration of the order of 1019 cm3), these same lower doped regions 12 form a reflector and therefore an optical mask defining in the XY plane the optical window and therefore the central zone Zc. It is then not necessary to produce a dedicated optical mask.
[0071] Furthermore, it is advantageous to optimize the thickness of the detection portion 10, in order to effectively absorb the infrared light radiation at the desired wavelength, while generating sufficient mechanical stresses in tension and compression.
[0072] In this respect, [Fig. 1C] illustrates an example of the evolution of the absorption rates A, transmission T and reflection R of the detection portion 10 with respect to light radiation of wavelength 1.55 pm.
[0073] In this example, the passivation dielectric layer 4 is made of SiO2 with a thickness of 280 nm, thus providing good antireflection. The detection portion 10 is made of germanium, and the intermediate region 13 has p-type doping of the order of approximately 1015 cm3. The lower doped portions 12 (for modulation) are in boron-doped germanium with a concentration of approximately 1019 cm3 and have a thickness of 200 nm. The upper doped portions 11 (collection) are in phosphorus-doped germanium (or with arsenic, or antimony) with a concentration of approximately 1019 cm3. The detection portion 10 has a thickness (between the two faces Fl and F2) which is varied. It is surrounded by a peripheral lateral portion 2 in polycrystalline silicon.
[0074] The absorption rates A, transmission T and reflection R illustrated in [Fig. 1C] are obtained by numerical simulation of the electromagnetic equations using the COMSOL Multiphysics software. It appears that the absorption peaks are located for thicknesses of the detection portion 10 of approximately 300nm, 480nm, 670nm, and 860nm. Thus, for a thickness of 295nm, the absorption rate A is equal to approximately 85%. It rises to approximately 88% for the thickness of 480nm, and to approximately 90% for the thickness of 670nm.
[0075] In addition, the mechanical stresses are determined using the same software, for different thickness values of the detection portion 10. Thus, for a thickness of 295nm, a horizontal tensile strain of +0.36% is obtained as well as a vertical compressive strain of -0.14%. For a thickness of 480nm, +0.30% horizontal tension and -0.12% vertical compression are obtained. And for a thickness of 670nm, +0.25% horizontal tension and -0.11% vertical compression are obtained. Thus, the mechanical stresses decrease in intensity as the thickness of the detection portion 10 increases.
[0076] Also, in this example, a thickness of 480nm appears to be a good compromise between optical absorption and mechanical deformations. It thus makes it possible to obtain a high absorption rate (here 88%) as well as sufficient mechanical constraints (horizontal tension of +0.30%, and vertical compression of -0.12%).
[0077] The performance of such a demodulator 1 is particularly high, whether in terms of time impulse response and bandwidth, insofar as a maximum value of 95% is obtained for the contrast in Cac demodulation in AC, and a bandwidth of the order of 400 to 500 MHz. Here, the doped collection regions (upper regions 11) are brought to an electrical bias of +1V, and the doped modulation regions (lower regions 12) are brought one to 0V and the other to +0.7V. In operation, these two modulators are in phase opposition and frequency modulated.
[0078] Recall that the time impulse response corresponds to the demodulation contrast Cac in AC. The contrast CAc is defined by the following relation (simplified here to a two-dimensional configuration): Cac(f) = ( Qci(t)-Qc2(t) ) / Qtot, where QCi and QC2 are the electrical charges at the electrodes Cl and C2, and Qtot is the total charge in the demodulator 1. The demodulation contrast Cac in AC is a contrast which is extracted from the temporal variation of the currents collected by the n+ doped collection regions. The integration of these two currents gives the quantities of charges in the demodulator 1. On the other hand, the demodulation contrast Cdc in DC does not involve the frequency (contrast in static).
[0079] As described in particular in the scientific article by Dalla Betta et al. entitled Design and Characterization of Current-Assisted Photonic Demodulators in 0.18-pm CMOS Technology, IEEE Trans. Electron Devices, vol. 58, no. 6, pp. 1702-1707, June 2011, the demodulation contrast Cd is defined as: C ( = X 1^ / 1^' ll™ is the total photocurrent collected at the electrodes Cl and C2: Itot = ICi + IC2; and fci is the fraction of the photocurrent Itot collected by the electrode Cl. In the case where the incident light radiation is modulated sinusoidally in power and the electrical voltage applied between the modulation electrodes Ml and M2 is also sinusoidal (the two modulators are always in phase opposition to each other), of the same frequency as the light radiation, the collected photocurrents can be written as follows:
[0080] iCJ - [ i / 2 +f^sin(27Tft) ] x [ljt+&n(2îrft-q>) ] Here = [V2 -fcisin (27ift)] x [ljt + CtSin (2jrft - (p)]
[0081] where and are the DC and AC components of the measured photocurrent Itot, f^j is the amplitude of the AC component of the fraction fCi, f is the modulation frequency, and q> is the phase shift between the emitted light and the received light. From these two equations, we can access the average of the currents ICi and IC2 which directly contain the phase q> from which we can thus determine a distance of an object located in the illuminated scene. These electric currents make it possible to determine the demodulation contrast Cac in AC.
[0082] [Fig.2A] is a schematic and partial view, in cross-section, of a current-assisted photonic demodulator 1 according to an alternative embodiment.
[0083] In this example, the demodulator 1 differs from that of [Fig.lA] in that the detection portion 10 was produced by epitaxy from an SOI substrate. The detection portion 10 is then in contact with a support layer 5 made of silicon, preferably monocrystalline. This support layer 5 is a nucleation layer for the germanium of the detection portion. It rests on an insulating layer 20 made of a silicon oxide (BOX). A thick layer of silicon (not shown) may be present in contact with the oxide layer 20. Here, the lower doped regions 12 are in contact with the support layer 5. Also, the latter is made of silicon that is not intentionally doped, so as to avoid any short circuit between the lower doped regions.
[0084] Figures 2B and 2C are schematic and partial views, in cross-section ([Fig.2B]) and in top view ([Fig.2C]), of a current-assisted photonic demodulator 1 according to an alternative embodiment.
[0085] In this example, the demodulator 1 differs from those of [Fig. 1A] and 2A in particular in that it only comprises a single doped collection region, which is then an upper doped region 11 located in the center of the central zone Zc. The doped modulation regions are then lower doped regions 12.
[0086] Furthermore, the conductive vias 3 which ensure the electrical polarization of the lower doped regions 12 are not vias distinct from the peripheral lateral portion 2, but are parts (distinct from one another) of the peripheral lateral portion 2. This is here formed from a plurality of distinct parts in the XY plane (without continuity of material between them). The peripheral lateral portion 2 is made of a doped conductive material of the same conductivity type as the lower doped regions 12, here of the p type. In this example, it is made of p+ doped polycrystalline silicon with a boron concentration of approximately 1020 cm3. The electrodes are then in contact with the parts of the peripheral lateral portion 2.
[0087] A p-type doped lateral region 15 may be present in the detection portion 10, along the peripheral lateral portion 2, when the latter is p-type doped as here. This p+ doped lateral region 15 makes it possible to prevent the space charge region of the demodulator 1 from extending to the lateral edge of the detection portion 10. Thus, the contribution of this region (potentially not free from defects linked to the production of the trenches) to the dark current is limited.
[0088] Figures 3A to 3L illustrate different steps of a method for manufacturing a matrix of current-assisted photonic demodulators 1 similar to that illustrated in [Fig.lA]. The demodulators 1 here have a planar configuration. They are made from germanium, and are therefore suitable for detecting infrared radiation in the near infrared. The detection portions 10 are made from germanium. In this example, they are made from a growth SOI substrate.
[0089] With reference to [Fig.3A], a growth SOI substrate is provided, the upper layer 5 of which is a layer of monocrystalline silicon 10 to 100nm thick. A layer 21 of unintentionally doped germanium (nest) of a few tens of nanometers, for example 50nm, is then produced by epitaxy. The growth temperature can be 400°C. Then, a layer 22 of p-type doped germanium with a concentration of approximately 1019 cm3, with a thickness of approximately 200nm, is produced by epitaxy at a temperature of 400°C.
[0090] With reference to [Fig.3B], the lower doped regions 12 are then produced by lithography and localized etching of the p++ doped layer 22, with etching stopped on the lower layer 21 in nested germanium. Each lower doped region 12 extends between a proximal end 12p and a distant end 12e in the XY plane. The proximal ends 12p delimit along the X axis the optical window. The dimension along the X axis of the optical window depends on the pixel pitch, and can thus range from 1.5pm to several tens of microns.
[0091] With reference to [Fig.3C], the main semiconductor layer 23 is produced by epitaxy from the layer 21 of nested germanium and the lower doped regions 12. The main semiconductor layer 23 is made of unintentionally doped germanium and has a thickness of between approximately 700 and 800 nm. It can be produced as described in particular in the publication by Hartmann & Aubin entitled Assessment of the growth / etch back technique for the production of Ge strain-relaxed buffers on Si, Journal of Crystal Growth, 488 (2018), 43. The main semiconductor layer 23 then has a very low density of emerging dislocations (for example of the order of 107 dislocations / cm2), which contributes to reducing the dark current in the detection portion 10 of the demodulator 1.
[0092] Then, a chemical-mechanical planarization (CMP) step is carried out, to planarize the upper face F1, and to reduce the thickness of the main semiconductor layer 23 to an optimal value having been determined (for example equal to approximately 480 nm). The thickness of what will be the detection portion is defined between the lower face F2 (interface between the layer 21 of nested germanium and the nucleation support layer 5) and the upper face FL
[0093] With reference to [Fig.3D], a passivation dielectric layer 4 is deposited on the upper face F1 of the main semiconductor layer 23 (layer 21 is integrated into layer 23). This passivation dielectric layer 4 may be formed of a first passivation sub-layer made for example of an aluminum oxide deposited by atomic layer deposition (ALD) with a thickness of the order of 10 to 50nm, for example 10nm, or even by a thin layer of unintentionally doped silicon with a thickness ranging from 1 to 4nm, deposited by epitaxy on the germanium of the detection portion 10. Then, a second sub-layer made for example of a silicon oxide such as SiO2 TEOS (tetraethyl orthosilicate) with a thickness of the order of 100 to 500nm, is deposited on the first underlying sub-layer.
[0094] With reference to [Fig.3E], notches 24 are then made in the main layer 23, extending along the Z axis from the upper face F1 until they open onto the lower doped regions 12, at the distal ends 12d. These notches 24 are made by lithography and localized etching, with etching stopped on the p++ doped germanium of the lower doped regions 12. The transverse dimensions of these notches 24 can be of the order of approximately 0.5 to 1 pm.
[0095] With reference to [Fig.3F], the conductive vias 3 are then produced by depositing an electrically conductive material in the notches 24. Here, the notches 24 are filled by epitaxy at a temperature of approximately 600°C with boron-doped p-type germanium with a concentration of approximately 1019 cm3. A CMP planarization step is then carried out to eliminate the p+ doped germanium resting on the passivation dielectric layer 4.
[0096] With reference to [Fig.3G], a peripheral trench 25 is then made in the main layer 23, making it possible to pixelate the demodulators 1 by the peripheral lateral portions 2. A localized etching of the main layer 23 is thus carried out to open here onto the BOX oxide layer 20. Each trench 25 here extends continuously in the XY plane around a detection portion 10. A plurality of detection portions 10 are thus obtained, separated from each other by a continuous trench 25. They are preferably obtained by an anisotropic etching technique, so as to obtain a substantially vertical lateral border along the Z axis. The trenches 25 have a transverse dimension (width) in the XY plane which can be between 0.5 pm and 2 pm, for example equal to 1 pm. The detection portions 10 can thus have a shape in the XY plane, for example circular, oval, polygonal, for example square, or any other shape..
[0097] With reference to [Fig.3H], the peripheral lateral portion 2 is produced by epitaxy in the trenches 25 of a crystalline semiconductor material, here based on silicon, at a temperature between 500°C and 700°C. This is polycrystalline silicon. It can be p-type doped, for example with boron with a doping level of the order of 4x1019 cm3. This material has a thermal expansion coefficient lower than that of the germanium of the detection portion 10, so that upon returning to ambient temperature (after epitaxy of the polysilicon in the peripheral trench 25), the detection portion 10 has mechanical stresses in horizontal tension and vertical compression.
[0098] With reference to [Fig.31], an interdiffusion anneal is carried out to form the SiGe-based lateral zone 14. Finally, a chemical mechanical polishing (CMP) step is then carried out, with a stop on the upper face of the passivation dielectric layer 4, to remove the excess silicon-based material and planarize the upper face of the stack.
[0099] With reference to [Fig.3J], non-through notches 26 are made in the dielectric passivation layer 4, leaving an unetched part of, for example, 20nm. A resin 27 (implantation mask) is then deposited, leaving the implantation zones defined by the notches 26 free. Then, the upper doped regions 11, here n+ doped, are made by ion implantation of phosphorus (or even with arsenic or antimony) in the detection portion 10. The doping level can be of the order of 5xl018 to 1020 cm 3. The upper doped regions 11 are therefore flush with the upper face Fl and are located above the lower doped regions 12. Here, they are located perpendicular to the proximal ends 12p of the lower doped regions 12. The resin 27 is then removed, and an activation annealing of the phosphorus is carried out.
[0100] Note that the upper doped regions 11 can also be produced by resumption of germanium epitaxy in notches made in the detection portion 10 from the face F1, with n-type doping during growth. However, ion implantation is preferred so as to preserve the mechanical constraints present in the detection portion 10.
[0101] With reference to [Fig.3K], a new dielectric layer is deposited on the underlying dielectric layer 4, so as to cover the detection portion 10, the conductive vias 3 and the peripheral lateral portion 2.
[0102] With reference to [Fig.3L], the modulation electrodes M1 and M2 are produced, which extend through the dielectric passivation layer 4 and come into contact with the conductive vias 3, as well as the collection electrodes C1 and C2, which extend through the dielectric passivation layer 4 and come into contact with the upper doped regions 11.
[0103] We thus obtain a matrix of current-assisted photonic demodulators 1, here in planar configuration, which is suitable for detecting in the near infrared with a cut-off wavelength greater than 1.55 pm, and which has a dark current between the doped modulation region and the adjacent doped collection region, due to their vertical arrangement and the mechanical constraints in vertical compression.
[0104] Particular embodiments have just been described. Different variants and modifications will appear to those skilled in the art.
Claims
Claims
1. Current-assisted photonic demodulator (1), suitable for detecting light radiation of interest, comprising: a detection portion (10), having a first face (Fl) and a second face (F2) opposite each other and pa connected to a main plane, and comprising a central zone (Zc) for detecting incident light radiation, made of a first crystalline semiconductor material based on germanium, comprising doped regions, including: at least two doped modulation regions, p-type doped, and intended to generate and modulate a drift current in the detection portion (10); at least one doped collection region, n-type doped, and intended to collect the minority charge carriers photogenerated during the absorption of the light radiation of interest; a dielectric passivation layer (4), made from an electrically insulating material, and covering the first face (Fl); electrodes (Ml, M2, Cl, C2), located at the level of the first face (Fl) and passing through the dielectric passivation layer (4), including modulation electrodes (Ml, M2) in electrical contact with the doped modulation regions; and at least one collection electrode (Cl, C2) in electrical contact with at least the doped collection region; characterized in that it comprises a peripheral lateral portion (2), surrounding the detection portion (10) in the main plane, made from a second material having a coefficient of thermal expansion lower than that of the first material, the detection portion (10) then having mechanical stresses in tension in the main plane and mechanical stresses in compression along a vertical axis or- thogonal to the main plane; • and in that said doped regions are arranged in a vertical arrangement such that: • at least one region doped according to a first type of conductivity, called the upper doped region (11), is flush with the first face (Fl) and is located near the central zone (Zc); • at least two regions doped according to a second type of conductivity opposite to the first type, called the lower doped regions (12), are located at a distance along the vertical axis from the upper doped region (11).
2. Photonic demodulator (1) according to claim 1, wherein each lower doped region (12) extends parallel to the main plane towards the central zone (Zc) and the upper doped region (11), and has a distal end (12d) distant from the central zone (Zc) and a proximal end (12p) oriented towards the central zone (Zc).
3. Photonic demodulator (1) according to claim 1 or 2, comprising conductive vias (3) extending along the vertical axis of the first face (Fl) and in electrical contact with the lower doped regions (12).
4. Photonic demodulator (1) according to claim 3, in which the conductive vias (3) are distinct from the peripheral lateral portion (2).
5. Photonic demodulator (1) according to claim 3, in which the conductive vias (3) are parts, distinct from each other, of the peripheral lateral portion (2), the latter then being made of an electrically conductive material.
6. Photonic demodulator (1) according to any one of claims 3 to 5, in which the electrode(s) in contact with the upper doped region(s) (11) are closer to the central zone (Zc) than the electrodes in contact with the conductive vias (3).
7. Photonic demodulator (1) according to any one of claims 1 to 6, comprising two lower doped modulation regions (12), and comprising a single upper doped collection region (11), located in the central zone (Zc) and at a distance, in the main plane, from the lower doped modulation regions (12).
8. Photonic demodulator (1) according to any one of claims 1 to 6, comprising two upper doped regions (11), each being located perpendicular to one of the lower doped regions (12).
9. Photonic demodulator (1) according to any one of claims 1 to 8, wherein, for an upper doped region (11) and a lower doped region (12) vertically adjacent, a vertical distance, along the vertical axis, between said doped regions (11, 12) is less than a horizontal distance, in the main plane, between the electrode connected to the upper doped region (11) and the electrode connected to the conductive via (3) of the adjacent lower doped region (12).
10. A photonic demodulator (1) according to any one of claims 1 to 9, wherein the upper doped region(s) (11) are collection regions, and the lower doped regions (12) are modulation regions.
11. Photonic demodulator (1) according to any one of claims 1 to 10, in which the peripheral lateral portion (2) is made from silicon, and in which the detection portion (10) comprises a lateral zone (14) made from SiGe, located at the interface with the peripheral lateral portion (2).
12. A method of manufacturing a photonic demodulator (1) according to any one of the preceding claims, comprising the following steps: • production, by epitaxy from a support layer (5), of a first layer (22) based on doped germanium; • production of the lower doped regions (12) by localized etching of the first layer (22) based on doped germanium; • production by epitaxy of a main layer (23) based on unintentionally doped germanium, on and in contact with the lower doped regions (12); • production of a trench (25) through the main layer (23) surrounding the detection portion (10); • production of the peripheral lateral portion (2) in the trench (25), at a temperature above ambient temperature, so that, after returning to ambient temperature, the detection portion (10) is in tension in the main plane and in compression along the vertical axis; • production of one or more upper doped regions (11) in the detection portion (10); • deposition of a dielectric passivation layer (4) on the detection portion (10); • production of the modulation electrodes and the collection electrodes.
13. Manufacturing method according to claim 12, comprising, before producing the trench (25), a step of producing two first notches (24) opening onto the lower doped regions (12), followed by a step of producing conductive vias (3) in the first notches (24).
14. Manufacturing method according to claim 12, such that, when producing the trench (25), the latter opens onto the lower doped regions (12), so that the peripheral lateral portion (2) is then in contact with them.
15. A manufacturing method according to any one of claims 12 to 14, wherein the upper doped region(s) (11) are produced by localized ion implantation.
