Method for manufacturing back-illuminated CMOS optical sensor
By forming an isolation structure and a low resistivity layer in the back-illuminated optical sensor, the problem that the sensing element cannot share the bias voltage is solved, and higher photon detection efficiency and lower crosstalk are achieved, which improves the performance of the sensor.
Patent Information
- Application Number
- CN202080065330.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2019-09-18
- Filing Date
- 2020-09-14
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2040-09-14
AI Technical Summary
Existing back-illuminated (BSI) optical sensors cannot connect all sensing elements to a common bias when trying to reduce crosstalk through optical isolation structures, resulting in reduced fill factor and poor photon detection efficiency (PDE).
By forming an isolation structure, such as a trench grid structure or an interleaved trench row structure, the deep trench isolation structure is filled with insulating and opaque materials and on the backside surface while keeping all sensing elements connected to the common bias voltage, the deep trench isolation structure is formed, such as a trench grid structure or an interleaved trench row structure, using insulating materials and opaque materials, and on the backside surface. Deposit a low resistivity layer to ensure isolation and electrical continuity between sensing elements.
It is achieved without affecting the fill factor, reducing crosstalk and improving photon detection efficiency (PDE), while allowing all sensing elements to share voltage, improving the performance of the optical sensor.
Smart Images

Figure CN114402436B_ABST
Abstract
Description
Technical Field
[0001] The present invention generally relates to methods for manufacturing back-side illuminated (BSI) optical sensors, and more specifically to methods for manufacturing BSI optical sensors with improved detection parameters, such as reduced crosstalk and enhanced photon detection efficiency (PDE). In particular, the present invention provides methods for isolating the sensing elements of a BSI optical sensor and applying a common bias voltage to the sensing elements. Background Art
[0002] Optical sensors have achieved a level of miniaturization and detection accuracy that has enabled them to be used in many fields.
[0003] For example, optical sensors are currently used in autonomous driving systems, such as ADAS (Advanced Driver Assistance Systems) to provide distance measurements. Among this type of optical sensor, there is a so-called time-of-flight (TOF) device, which is designed to detect the distance of a target object by measuring the time required for an electromagnetic wave (e.g., a laser pulse) to return from the target object. The TOF device uses a very sensitive sensing element because the sensing element is a single photon avalanche diode (SPAD). The main components of the SPAD are: an avalanche photodiode operating above the breakdown voltage (Geiger mode); a quenching resistor; a first electrode (cathode) and a second electrode (anode). Because of the multiplication process caused by the impact ionization of the space charge region of the photodiode, the electron-hole pairs generated by the incident photons in the photodiode region (e.g., a reverse biased pn junction) with a bias higher than the breakdown voltage (multiplication region) can trigger an avalanche. In this way, a macroscopic current is generated in response to the incident photon stream.
[0004] The use of SPADs is not limited to the automotive market, but is widely used in other markets, such as the medical market. SPAD arrays (often referred to as silicon photomultipliers or SiPMs) have been adopted to replace photomultiplier tube (PMT) technology in photon counting applications, such as positron emission tomography (PET) technology, to observe metabolic processes in the body. The implementation of SPAD technology in CMOS technology further enhances the advantages of SiPM technology, thereby providing high dynamic range photon counting in picosecond time.
[0005] Referring again to the automotive market, LIDAR (Light Detection and Ranging) technology, for example, employs near infrared (NIR) light detection to measure the time of flight of an emitted laser pulse. In order to be sensitive to the NIR wavelength range, a SPAD array optical sensor must have a photo detection efficiency (PDE) greater than 10% in the electromagnetic spectrum in the range of 850nm to 910nm. This is difficult to achieve using conventional front-illuminated (FSI) CMOS optical sensors, due to the deep penetration distance of NIR photons, which are absorbed deeply in semiconductor substrates (e.g., silicon substrates), with the result that the generated electron / hole pairs have little chance of being collected in the avalanche photodiode region. In contrast, the above requirement can be achieved by a back-illuminated (BSI) CMOS optical sensor configured so that all sensing elements are arranged close to the backside surface, from which the incident photons strike, and for this reason, most of the generated electron / hole pairs have a greater chance of being collected in the avalanche photodiode region.
[0006] The main disadvantage generated in the SPAD array is the crosstalk between adjacent sensing elements. Crosstalk may be caused by a parasitic avalanche in one sensing element, which is caused by an avalanche in another sensing element, thereby generating a parasitic output current pulse (dark current) in the absence of light. In order to limit the presence of photons propagating from one SPAD to an adjacent SPAD, thereby limiting the possibility of triggering a second avalanche, (crosstalk) and thus limiting the dark count rate (DCR), the individual sensing elements must be isolated from each other. Different isolation structures can be used. Generally, the isolation structure can be formed in the form of a trench etched from the front to the substrate adjacent to the sensing element area. When the trench is formed, the trench is usually filled with a dielectric material. For example, according to U.S. Patent 9,741,759, the trench isolation structure has the form of a deep trench isolation (DTI) structure formed from the back side of the semiconductor substrate toward the front side. According to the same U.S. patent, the trench isolation structure is filled with a first isolation material conformally formed on the inner wall of the trench and a second conductive material conformally formed to fill the trench isolation structure. In addition, according to the same US patent, the trench isolation structure can apply a voltage by connecting a voltage applying device to the trench isolation structure. Generally, problems arise with this isolation method because the trench isolation structure isolates the sensing elements from each other, but makes it impossible to connect all sensing elements to a common bias. For example, in the case of SiPM, the anode electrode of each sensing element (SPAD) is connected to a certain potential (e.g., ground potential) so as to be reverse biased at the junction. When the SPADs are isolated from each other, it is impossible to connect the anode electrode of the SPAD to a common potential. The only way to overcome this problem is to bias each SPAD individually, such as by adding a single contact to connect each SPAD to the same potential to bias each SPAD, but this will result in the use of additional space for the array layout, thereby reducing the fill factor, which is defined as the ratio of the sensing area of the sensing element to the total area, and thus reducing the photon detection efficiency (PDE).
[0007] Therefore, there is a need to provide a BSI optical sensor with improved characteristic parameters, such as reduced crosstalk and improved photon detection efficiency (PDE), without affecting the fill factor. Summary of the invention
[0008] Therefore, a general object of the present invention is to overcome the above-mentioned technical problems that arise when trying to reduce crosstalk by implementing an optical isolation structure that isolates the sensing elements of an optical sensor from each other, thereby prohibiting each sensitive element from being connected to a common bias voltage.
[0009] Furthermore, a specific object of the present invention is to provide a method for manufacturing a backside illuminated (BSI) CMOS optical sensor with improved detection parameters, the method comprising the step of forming an isolation structure between adjacent sensing elements while keeping all sensing elements connected to a common bias voltage.
[0010] These and other objects are achieved by the method of manufacturing a back-illuminated CMOS optical sensor of the present invention.
[0011] According to a first embodiment of the present invention, a method for manufacturing a back-illuminated (BSI) CMOS optical sensor comprises the following steps: forming an isolation structure in the form of a trench grid structure around each sensing element of a pixel array, the isolation structure comprising a first isolation structure (e.g., a shallow trench isolation (STI) structure or a LOCOS isolation structure or an implanted region) formed from the front side surface of the optical sensor and a deep trench isolation (DTI) structure, the deep trench isolation (DTI) structure being etched from the back side surface of the optical sensor after the silicon substrate has been thinned from the back side and reaching the corresponding bottom surface of the front side first isolation structure; filling the back side DTI structure with a first insulating material and a second opaque material (e.g., metal) suitable for reducing crosstalk; planarizing the back side surface of the optical sensor from the back side, thereby exposing the silicon surface corresponding to each sensing element; depositing an additional low resistivity layer on the back side surface, thereby connecting the exposed silicon surface and providing a common voltage application structure applied to all sensing elements of the pixel array. Preferably, the additional low resistivity layer comprises an amorphous silicon material, which has the advantage of filtering UV and visible light, thereby allowing only NIR light to be detected. According to a second embodiment of the present invention, a method for manufacturing a back-illuminated (BSI) CMOS optical sensor includes the following steps: forming an isolation structure in the form of staggered trench rows around each sensing element of a pixel array, each trench row including trench portions separated by gaps filled with a silicon substrate, the trench portions being staggered relative to trench portions belonging to parallel trench rows; the isolation structure including a first isolation structure (e.g., a shallow trench isolation (STI) structure or a LOCOS isolation structure or an implanted region) formed from a front side surface of the optical sensor and a second deep trench isolation (DTI) structure, the deep trench isolation (DTI) structure being etched from a thinned back side surface of the optical sensor and reaching a corresponding bottom surface of the front side first isolation structure; filling the back side DTI structure with an opaque material suitable for reducing crosstalk; planarizing the back side surface of the optical sensor to expose the silicon back side surface corresponding to each sensing element and corresponding to the silicon gap, the exposed silicon back side surface providing a common voltage application structure applied to all sensing elements of the pixel array. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] For a better understanding of the present invention, preferred embodiments will now be described with reference to the accompanying drawings (not to scale), which are shown by way of example only and are not to be construed as limiting, in which:
[0013] · Figure 1 A schematic diagram of a pixel array of a known semiconductor optical sensor is shown, wherein sensing elements are connected in parallel to a common bias voltage;
[0014] · Figure 2 A schematic diagram of a pixel array of a known semiconductor optical sensor is shown, wherein sensing elements are separated by isolation structures (solid bars);
[0015] · Figure 3 shows a schematic diagram of a pixel array of a known semiconductor optical sensor, wherein the sensing elements are separated by isolation structures (solid bars) and connected to a common bias voltage via a single contact included in each sensing element;
[0016] · Figure 4 shows a cross-sectional view of a semiconductor BSI optical sensor obtained using the claimed method according to a first embodiment;
[0017] · Figure 5 A flow chart illustrating a method of manufacturing a semiconductor BSI optical sensor according to a first embodiment is shown;
[0018] · Figure 6 shows a top view and a cross-sectional view of a semiconductor BSI optical sensor obtained using the claimed method according to a second embodiment;
[0019] · Figure 7 A flow chart illustrating a method of manufacturing a semiconductor BSI optical sensor according to a second embodiment is shown. DETAILED DESCRIPTION
[0020] The following discussion is presented to enable those skilled in the art to make and use the invention. Various modifications to the various embodiments will be apparent to those skilled in the art without departing from the scope of the invention as claimed. Therefore, the present invention is not intended to be limited to the embodiments shown and described, but is to be accorded the widest scope consistent with the principles and features disclosed herein and defined in the appended claims.
[0021] As will be discussed in detail below, aspects of the present invention provide a method of manufacturing a BSI optical sensor that is configured such that all sensing elements (e.g., SPADs) of a pixel array (e.g., SiPM) are separated by isolation structures to reduce crosstalk, with the individual elements connected to a common voltage application structure provided from the back side of the optical sensor in a manner that does not affect the fill factor.
[0022] Figures 1 to 3 Three different pixel arrays of an optical sensor are schematically shown, comprising a plurality of sensing elements designed according to known layouts.
[0023] For example, Figure 1 The schematic diagram depicted in shows a pixel array 100 of an optical sensor implemented in CMOS technology, the pixel array 100 comprising a plurality of sensing elements 110 (e.g., SPADs) arranged in a two-dimensional matrix and having a first electrode (110a) connected to a potential V1 and a second electrode (110b) connected to a common potential node V2 (115). In particular, each sensing element is configured to include a photodiode active region (110C) which, in its simplest form, is implemented in a pn (or np) reverse biased junction such that the n (or p) region consumes charge carriers (e.g., electron / hole pairs) and, therefore, incident photons generate electron / hole pairs that are collected by the consumption region of the photodiode. The pn junction is reverse biased above the breakdown voltage so that the electrons (and holes) gain sufficient energy to form further pairs and initiate an avalanche process before being collected, thereby amplifying the signal generated by the incident single photon. A quenching resistor 120 is used to stop the avalanche process. In this way, each incident photon can be detected, resulting in a strong current pulse with a duration of a few nanoseconds. In the standard method, the potential V1 is applied continuously to each first electrode, while the potential V2 (e.g., ground potential) is applied to all second electrodes simultaneously. Therefore, all sensing elements of the pixel array are connected in parallel to a common bias voltage V1-V2.
[0024] To isolate SPADs from each other in order to reduce crosstalk, the standard approach is to use trench isolation structures ( Figure 2 ). Figure 2 As shown in , the formation of the trench isolation structure provides an optical isolation structure, but at the same time, the trench isolation structure also electrically isolates the sensing elements from each other, thereby making it impossible to connect all the sensing elements to a common potential V2.
[0025] Figure 3A standard approach for overcoming this problem is shown in . With this approach, an additional contact 310 is added to each sensing element so that a potential V2 (e.g., ground potential) is continuously applied to each second electrode of each sensing element. Adding a contact 310 to each sensing element requires more silicon space to be used for the pixel array layout, thereby reducing the fill factor and therefore the PDE.
[0026] To overcome the above problems, the proposed solution provides a method of manufacturing a BSI optical sensor according to the first and second embodiments.
[0027] To better understand the present invention, Figure 4 An example of an optical sensor obtained according to a first preferred embodiment of the proposed method is shown. In particular, Figure 4 is a cross-sectional view of a BSI optical sensor 400 at a particular step in the manufacturing process, the BSI optical sensor 400 comprising:
[0028] · a front side FS surface 410 and a back side BS surface 411;
[0029] A semiconductor substrate (eg, a silicon substrate) 401 having a first surface 401 a and a second surface 401 b facing each other;
[0030] At least a first sensing element 402a and a second sensing element 402b, wherein the at least first sensing element 402a and the second sensing element 402b are formed in the semiconductor substrate 401 and separated by a front side isolation structure (e.g., a shallow trench isolation (STI) structure) 403;
[0031] A front-side voltage applying device 404 electrically connected from the FS surface 401a of the semiconductor substrate to the at least first sensing element or the second sensing element;
[0032] An insulating layer 406 formed as a multilayer, the insulating layer 406 comprising at least a first interconnect layer 407 and a metal contact 408;
[0033] a carrier semiconductor substrate 409 attached to a front side surface 410 of the optical sensor;
[0034] A backside deep trench isolation (DTI) structure 405, which surrounds at least the first sensing element and the second sensing element, is etched from the BS surface 401b and reaches the bottom side of the FS isolation structure 403, and the BSDTI structure is filled with a first insulating material 405a (e.g., silicon oxide) covering the sidewall and bottom of the trench and a second conductive material 405b (e.g., tungsten) filling the inside of the BSDTI structure;
[0035] • An additional low resistivity layer 412 (eg an amorphous silicon layer) deposited on the BS surface 401b and connecting at least the first sensing element 402a and the second sensing element 402b from the semiconductor second surface.
[0036] According to a first preferred embodiment of the present invention, the BSI optical sensor 400 is based on Figure 5 The semiconductor substrate is manufactured by the manufacturing process steps shown in the flowchart A of . According to the preferred embodiment, the manufacturing process includes the following manufacturing steps: 501) performing a standard CMOS front-side manufacturing process, which includes the steps of forming a FS isolation structure (such as a shallow trench isolation structure (STI)), multiple interconnection layers and multiple contact / via structures arranged in the form of a multi-layer isolation structure; 502) attaching a carrier wafer to the front-side surface of the multi-layer isolation structure; 503) partially thinning the semiconductor substrate from the back-side surface; 504) forming a BSI deep trench isolation (DTI) structure from the back side through a photolithography / etching process step, the BSI deep trench isolation (DTI) structure reaching the bottom side of the FS isolation structure and having a mesh 505) depositing a first isolation layer to cover the sidewalls and bottom wall of the DTI structure, and depositing a second conductive material such as tungsten, tungsten titanium (Ti / W) or aluminum to fill the interior of the BSIDTI structure; 506) planarizing the back surface (e.g., by chemical mechanical polishing, CMP) to remove all excess conductive materials on the back surface, and forming a low resistivity layer by depositing an amorphous silicon layer, doping the amorphous silicon layer (e.g., by implantation), and annealing the amorphous silicon layer (e.g., by thermal laser annealing, LTA or microwave annealing or other methods that do not affect the underlying layer) to convert the amorphous silicon layer into a low resistivity layer. Finally, an appropriate ARC layer can be applied on top of the low resistivity layer to improve the transmission of near infrared (NIR) radiation.
[0037] The advantages of the present invention are clear from the above. In particular, it is emphasized that, by the above method, all of the at least first sensing elements and second sensing elements of the optical sensor pixel array (e.g., SPAD array) are connected by the low resistivity layer to the front side ( Figure 4The back side potential (e.g. ground potential) is connected to the same back side potential (e.g. ground potential) by contacting one or very few contacts (not shown). It is also advantageous to use an amorphous silicon layer as an additional low resistivity material because it helps to filter UV / visible radiation in the case of near infrared (NIR) detection.
[0038] According to a second embodiment of the present invention, a BSI deep trench isolation (DTI) structure is fabricated in the form of staggered trench rows, thereby ensuring optical isolation between adjacent sensing elements of a pixel array while maintaining electrical continuity between the sensing elements. For better understanding, Figure 6 An example of a BSI optical sensor obtained according to the second embodiment of the proposed method is shown. In particular, Figure 6 The top portion of the diagram represents a top view of the BSI optical sensor 600, and the bottom portion represents a view along the BSI optical sensor 600. Figure 6 The optical sensor obtained according to the second embodiment comprises:
[0039] A front FS surface 610 and a back BS surface 611;
[0040] A semiconductor substrate 601 having a first surface 601 a and a second surface 601 b facing each other;
[0041] At least a first sensing element 602a and a second sensing element 602b formed in the semiconductor substrate and separated by a FS isolation structure (e.g., a shallow trench isolation (STI) structure) 603;
[0042] A front-side voltage applying device 604 electrically connected from the FS surface 601a of the semiconductor substrate to the at least first sensing element or the second sensing element;
[0043] An insulating layer 606 formed as a multilayer, the insulating layer 606 comprising at least a first interconnect layer 607 and a metal contact 608;
[0044] a carrier semiconductor substrate 609 attached to a front side surface 610 of the optical sensor;
[0045] A backside BS deep trench isolation (DTI) structure, which surrounds the at least first sensing element and the second sensing element, is formed in the form of staggered trench rows 605, is etched from the backside surface 601b and reaches the bottom side of the FS isolation structure 603, and the BSDTI structure is filled with a first insulating material 605a (e.g., silicon oxide) and a second conductive material 605b (e.g., tungsten, tungsten titanium or aluminum). The first insulating material 605a covers the sidewalls and the bottom of the trench, and the second conductive material 605b fills the interior of the BSDTI structure.
[0046] According to the second embodiment of the present invention, the BSI optical sensor 600 is based on Figure 7 The manufacturing process steps shown in the flowchart B are used for manufacturing. According to the method, the manufacturing process includes the following manufacturing steps: 701) performing a standard front-side manufacturing process, which includes the steps of forming a FS isolation structure (e.g., a shallow trench isolation structure (STI)), multiple interconnect layers, and multiple contact / via structures arranged in the form of a multi-layer isolation structure; 702) attaching a carrier wafer to the front-side surface of the multi-layer isolation structure; 703) partially removing the silicon substrate from the back-side silicon surface; 704) forming a BSI deep trench isolation (DTI) structure from the back side through a photolithography / etching process step, which BSI deep trench isolation (DTI) structure reaches the bottom side of the FS isolation structure, has the form of staggered trench rows separated by gaps, and surrounds each sensing element area; 705) depositing a first isolation layer to cover the trench sidewalls and bottom walls of the DTI structure and depositing a second conductive material such as tungsten, tungsten titanium or aluminum to fill the interior of the BSIDTI structure; 706) planarizing the back-side surface (e.g., by chemical mechanical polishing, CMP) to remove all excess conductive layers on the back-side surface.
[0047] The advantages of the present invention are clear from the above. In particular, it is emphasized that the proposed solution consists in constructing a backside deep trench isolation structure in the form of staggered trench rows extending parallel to each other and surrounding adjacent sensing elements, the staggered trench rows comprising trench portions separated by gaps filled with silicon substrate, thereby ensuring electrical continuity between sensing elements of a pixel array (e.g. SPAD) through these silicon gaps and simultaneously forming an optical barrier to crosstalk. According to the above method, the exposed silicon substrate in the photodiode region acts as a second electrode of the sensing element, and each second electrode ( Figure 7 ) applies a common potential.
[0048] Finally, it is evident that the invention is susceptible of numerous modifications and variations, all of which are within the scope of the invention as defined in the appended claims.
Claims
1. A method for manufacturing a back-illuminated CMOS optical sensor having reduced crosstalk and increased photon detection efficiency, the method comprising the following operations: Providing a semiconductor wafer having a semiconductor substrate (401), a first surface (401a) and a second surface (401b) facing each other; performing a standard CMOS front-side manufacturing process, the standard CMOS front-side manufacturing process comprising the steps of forming at least a first sensing element (402a) and a second sensing element (402b) in the semiconductor substrate between the first surface and the second surface, forming at least one insulating layer (406) on the first surface of the semiconductor substrate, forming at least a first interconnect layer (407) and at least one metal contact portion (408), wherein the at least first interconnect layer and the at least one metal contact portion are embedded in the at least one insulating layer; providing a carrier semiconductor substrate (409) and attaching the carrier semiconductor substrate to the at least one insulating layer; thinning the semiconductor wafer by removing material from the second surface (401b) of the semiconductor substrate; forming a backside deep trench isolation structure (405) in the semiconductor substrate around at least the first sensing element and the second sensing element, the deep trench isolation structure extending from the thinned second surface; Filling the deep trench isolation structure with a first insulating material (405a) and a second conductive material (405b), wherein the first insulating material (405a) covers the sidewall and the bottom of the deep trench isolation structure, and the second conductive material (405b) fills the interior of the deep trench isolation structure; planarizing the semiconductor wafer from the thinned second surface to expose the semiconductor substrate (401) in the region of at least the first sensing element and the second sensing element, at least the first sensing element and the second sensing element being surrounded by the deep trench isolation structure; A common voltage applying structure in contact with the deep trench isolation structure and the semiconductor substrate in at least the first sensing element and the second sensing element is formed from the thinned second surface.
2. The method according to claim 1, wherein: The deep trench isolation structure (405) has a form of a trench grid structure surrounding at least the first sensing element and the second sensing element.
3. The method according to claim 2, wherein: The step of forming a common voltage applying structure in contact with all of the first sensing elements and the second sensing elements from the thinned second surface includes the following steps: forming a low resistivity layer (412) on the thinned second surface connected to the groove grid structure and the semiconductor substrate in at least the first sensing element and the second sensing element.
4. The method according to claim 3, wherein: The step of forming the low resistivity layer (412) on the thinned second surface comprises the following processing steps: depositing an amorphous silicon layer; doping the amorphous silicon layer by implantation; And annealing the amorphous silicon layer by thermal laser annealing or microwave annealing without affecting the underlying layer.
5. The method according to claim 1, wherein: The deep trench isolation structure (405) has the form of staggered trench rows, each trench row comprising trench portions separated by gaps filled with a semiconductor substrate, the trench portions belonging to one trench row being staggered relative to the trench portions belonging to parallel trench rows.
6. The method according to claim 5, wherein: The gaps between the groove portions are configured to provide a common voltage applying structure applied from the thinned second surface to at least the first sensing element and the second sensing element.
7. The method according to claim 6, wherein: The first insulating material (405a) is silicon oxide.
8. The method according to any one of claims 1 to 7, wherein: The second conductive material (405b) is an opaque material.
9. The method according to claim 8, wherein: The opaque material is made of metal.
10. The method according to claim 9, wherein: The metal is selected from titanium, tungsten and aluminum.
11. The method according to claim 1, wherein: The standard CMOS front-side manufacturing process further includes the step of forming a first isolation structure in the semiconductor substrate around at least the first sensing element and the second sensing element, the first isolation structure extending from the first surface.
12. The method according to claim 11, wherein: The first isolation structure formed in the semiconductor substrate around at least the first sensing element and the second sensing element is a shallow trench isolation structure.
13. The method according to claim 11, wherein: The first isolation structure formed in the semiconductor substrate around at least the first sensing element and the second sensing element is a LOCOS structure.
14. The method according to claim 11, wherein: The first isolation structure formed in the semiconductor substrate around at least the first sensing element and the second sensing element is an implanted region.
