Low-noise semiconductor photoelectric detector

Abstract

The invention relates to a low-noise semiconductor photoelectric detector, and belongs to the technical field of photoelectric detection. The low-noise photoelectric detector is characterized by consisting of an avalanche photodiode (1) for normally receiving optical signals and a dark pixel avalanche photodiode (2). Through introduction of the dark pixel structure, background noise and interference are compensated under the condition that normal optical signal receiving is not affected, and the signal-to-noise ratio of the detector is improved. The method is suitable for measuring extremely weak optical signals, and can be widely applied to the fields of laser ranging, quantum communication, quantum key distribution, medical imaging and the like.

Description

(I) Technical field

[0001] The invention relates to a low-noise semiconductor detector, belonging to the technical field of photoelectric detection. (II) Background technology

[0002] Avalanche photodiode (APD) is a high-performance photodetector with higher gain than ordinary photodetectors, so it has higher sensitivity at low light intensity levels. It works by utilizing the avalanche effect, that is, under the action of an electric field, the electrons generated by a carrier collision interact with the lattice atoms to produce a large number of electron-hole pairs, causing the number of carriers to increase exponentially, and eventually generating an observable avalanche current. Single Photon Avalanche Diode (SPAD) is an avalanche photodiode operating in Geiger mode, which operates at a bias voltage greater than the breakdown voltage and is designed to detect single photon events. It is an extremely sensitive photodetector and is often used in applications that require high sensitivity and high resolution, such as quantum communication, quantum key distribution, fluorescence lifetime detection, single photon microscopy, and photon counting.

[0003] The dark count of a single-photon detector refers to the count of avalanche breakdown events caused by internal noise or other factors in the absence of external light sources. Dark counts have an important impact on the performance and application of single-photon avalanche photodiodes. First, dark counts will increase the background noise level of the system, resulting in a decrease in the signal-to-noise ratio, and the system's detection sensitivity to real photon signals will be reduced, making it difficult for the system to detect extremely weak light signals, thereby limiting its scope of application and performance. Secondly, dark counts will cause the system to generate false alarms, that is, the dark counts will be mistaken for real photon signals, reducing the reliability and accuracy of the system, especially in tasks that require high-precision detection. In addition, dark counts will also affect the imaging quality, resulting in an increase in image background, reducing the contrast and clarity of the image, thereby affecting the imaging quality. In order to compensate for the impact of dark counts, it may be necessary to increase the integration time to increase the ratio between the signal and the background, thereby extending the time of the imaging or detection process. Therefore, reducing dark counts is one of the keys to improving the performance and application reliability of single-photon avalanche photodiodes. The generation of dark counts mainly comes from several factors inside single-photon avalanche photodiodes: thermal excitation, lattice defects, and tunneling effects. At room temperature, thermal excitation may cause carriers to obtain enough energy to trigger an avalanche breakdown event, while lattice defects and impurities can cause local electric field inhomogeneities, which in turn promotes avalanche. In addition, the quantum mechanical tunneling effect can also lead to carrier avalanches under weak electric fields. Therefore, in order to reduce dark counts, it is necessary to consider these factors when preparing and designing SPADs, and take corresponding measures to improve the performance and reliability of the system.

[0004] In order to reduce the noise of avalanche photodetectors, researchers have continuously optimized the device structure to reduce the dark count rate and improve the signal-to-noise ratio of the detector. For example, Veerappan et al. designed a CMOS single-photon avalanche diode with a vertical pin diode structure, with a dark count rate of 1.5 cps / μm2 under 11V over-bias (Veerappan, Chockalingam, and Edoardo Charbon. "A low dark count pin diode based SPAD in CMOS technology." IEEE transactions on electron devices 63.1 (2015): 65-71.); Abbas proposed the first 3D stacked back-illuminated SPAD image sensor, with a dark count rate of less than 200 cps under 1V over-bias at room temperature (Al Abbas, T., et al. "Backside illuminated SPAD image sensor with 7.83μm pitch in 3D-stackedCMOS technology." 2016IEEE International Electron Devices Meeting (IEDM). IEEE, 2016.). In addition, Jin Xiangliang et al. proposed a new structure of SPAD, which uses P+ / LNW (Light N-well doping) junction to detect photons, and suppresses edge breakdown through a low-concentration N-type diffused circular guard ring. When the overbias voltage is 2V, the dark count rate is 260Hz (Yang Jia, et al. "Design and analysis of a novel low dark count rate SPAD." Journal of Infrared and Millimeter Waves 35.4 (2016): 394-397.); Xu Yue used polysilicon field plates to reduce the dark count rate of SPAD devices. The experimental results show that the introduction of polysilicon field plates reduces the dark count rate of SPAD devices by an order of magnitude (Dong, Han, et al. "Reducing dark count of single-photonavalanche diode detector with polysilicon field plate." Acta Physica Sinica 69.14 (2020).). Although the noise of single-photon detectors can be reduced to a certain extent by optimizing the structure of existing models, the current dark count rate still reaches tens to hundreds of Hz, and it is difficult to reduce it further.The defects of these designs are: 1. Due to the thermal excitation effect of the semiconductor material itself, dark carriers cannot be completely eliminated, and due to the limitations of the device manufacturing process, the dark current cannot be further reduced, which makes it impossible to further improve the device performance; 2. As the bias voltage increases, the field strength inside the device increases accordingly. At this time, the photon detection efficiency will increase due to the increased probability of avalanche breakdown, but the probability of dark carriers triggering avalanches will also increase. Therefore, the dark count rate and photon detection efficiency need to be compromised, which greatly affects the detection performance of the detector; 3. The number of carriers inside the device is affected by the ambient temperature. For example, as the temperature increases, the breakdown voltage increases, which makes the device stability worse.

[0005] In order to solve the above problems, the present invention discloses a silicon-based avalanche photodiode detector with integrated dark pixels, which can be used for the detection of extremely weak light such as laser ranging, fluorescence lifetime detection, and quantum communication. The structure of the present invention integrates two avalanche photodiodes with exactly the same semiconductor structure parameters by vertical stacking, and the bottom is used as a dark pixel, so that it can only operate under lightless conditions to provide dark noise. The avalanche diode operating in a dark environment has the same structure, size and process manufacturing parameters as the avalanche diode operating normally in the integrated detection light signal mode, and will have the same dark count rate. The dark counts of the two can compensate and offset to reduce the noise of the entire system. In theory, the dark count rate can be reduced to close to 0, which can improve the sensitivity and accuracy of the detection system. The avalanche diode used for light signal detection in the semiconductor photodetector is connected to the dark pixel operating under lightless conditions through the same electrode and works under the same bias voltage, so it has the same (or similar) operating temperature and performance, which makes its noise reduction effect not affected by the change of the temperature of the avalanche photodetector chip. The present invention can be applied to avalanche photodiodes in single-photon detection modes, including gated mode and free-running mode, and can also be applied to avalanche diodes operating below a breakdown voltage, greatly improving the scope of application. (III) Summary of the invention

[0006] The object of the present invention is to provide a low-noise semiconductor photodetector, which is composed of an avalanche photodiode (1) for receiving light signals normally and a dark pixel avalanche photodiode (2), wherein the avalanche photodiode (1) for receiving light signals normally is on the top and the dark pixel avalanche photodiode (2) is placed at the bottom, and the two are stacked and integrated into one. The avalanche photodiode (1) for receiving light signals normally comprises, from top to bottom, a top electrode (10), a top SiO2 layer (11), a top epitaxial layer (12), a first substrate (13) and a common electrode (3); the dark pixel avalanche photodiode (2) comprises, from top to bottom, a common electrode (3), a second substrate (23), a bottom epitaxial layer (22), a bottom SiO2 layer (21), and a bottom electrode (20); the common electrode (3) belongs to the common area of ​​the avalanche photodiode (1) and the avalanche photodiode (2).

[0007] The purpose of the present invention is achieved in this way: the detector as a whole can be regarded as being composed of two avalanche photodiodes stacked up, the avalanche photodiode (1) at the top that normally receives light signals is responsible for receiving light to generate a response current, the dark pixel avalanche photodiode (2) at the bottom only generates dark noise, and the common electrode (3) is the common anode of the two avalanche photodiodes. Under the action of a reverse bias voltage, the avalanche photodiode (1) at the top that normally receives light signals receives light signals to generate photogenerated carriers. Due to the blocking of the middle metal common electrode (3), photons cannot be absorbed by the dark pixel avalanche photodiode (2) at the bottom. The top electrode (10) outputs a total current containing dark current and photogenerated current, and the bottom electrode (20) outputs a dark current generated by dark noise excitation. The difference between the output currents of the two is the light signal current excited by the incident photons.

[0008] The avalanche photodiode (1) in the detector that normally receives light signals comprises, from top to bottom, a top electrode (10), a top SiO2 layer (11), a top epitaxial layer (12), a first substrate (13) and a common electrode (3), wherein the top electrode is a metal cathode. Preferably, the SiO2 layer (11) has a thickness of 100-300 nm, the top epitaxial layer (12) has a thickness of 1500-3000 nm, and the common electrode (3) is greater than 2000 nm. The top epitaxial layer is heavily doped to form N + The contact layer (121) is preferably N + The doping concentration of the contact layer (121) is 10 17 -10 19 cm -3 . N + The outer side of the contact layer (121) is lightly doped to form an N-Well region (122), and the height of the N-Well region (122) is greater than that of the N + Thickness of the contact layer (121). N + A Center-Nwell (123) is lightly doped below the contact layer (121) to form a PN junction with the top epitaxial layer (12). Below the top epitaxial layer (12) is a first substrate (13) in contact with the common electrode (3). Preferably, the doping concentration of the first substrate (13) is 10 18 -10 20 cm -3 .

[0009] The dark pixel avalanche photodiode (2) in the detector is used as a dark pixel, and from top to bottom are a common electrode (3), a second substrate (23), a bottom epitaxial layer (22), a bottom electrode (20), and a bottom SiO2 layer (21). The bottom epitaxial layer (22) includes a dark pixel N +The contact layer (221), the dark pixel N-Well region (222), the dark pixel Center-Nwell (223) and the dark pixel electrode (20). The shape, size and doping concentration of each region in the dark pixel are consistent with those of each region of the corresponding avalanche photodiode (1) that normally receives light signals.

[0010] When the detector is working, under the action of reverse bias voltage, photons are absorbed by the avalanche photodiode (1) that normally receives light signals, and the excited photogenerated carriers enter the PN junction depletion region, and the photogenerated electron-hole pairs are separated and accelerated. When they have sufficient energy, they collide with lattice atoms to ionize and generate new electron-hole pairs. The new electron-hole pairs will continue to collide and ionize to generate more electron-hole pairs, and the carriers will increase exponentially. The multiplied carriers are finally collected by the top electrode (10) to form a multiplied current. The Center-Nwell (223) of the dark pixel and the bottom epitaxial layer (22) form a dark pixel PN junction, and the dark carriers undergo a multiplication effect in this region, and finally form a dark noise current that is output by the bottom electrode (20).

[0011] Beneficial effects of the present invention: The semiconductor photodetector is composed of an avalanche photodiode (1) that normally receives light signals and a dark pixel avalanche photodiode (2). Since the dark pixel avalanche photodiode (2) and the avalanche photodiode (1) that normally receives light signals in the light signal detection mode have the same structure, size and manufacturing parameters, the dark count rates of the two are the same, and their dark counts can offset each other, significantly reducing system noise. In theory, the dark count rate can be reduced to close to zero, thereby improving the sensitivity and accuracy of the detection system. The avalanche photodiode (1) that normally receives light signals in the semiconductor photodetector and the dark pixel avalanche photodiode (2) that operates under lightless conditions are connected through the same electrode and operate under the same bias voltage, so they have the same internal electric field and the same avalanche triggering probability. Because the dark counts can be offset and eliminated, the problem of increasing the dark count rate while increasing the bias voltage to improve the detection efficiency does not need to be considered. The avalanche photodiode (1) that normally receives light signals in the semiconductor photodetector and the dark pixel avalanche photodiode (2) that operates under lightless conditions are connected together through a common electrode (3), so they have the same operating temperature, and the amount of change in internal noise affected by temperature is also the same, so that the noise reduction effect will not be affected by changes in the operating environment temperature of the avalanche photodetector. (IV) Description of the drawings

[0012] Figure 1 It is a schematic diagram of the structure of a low-noise semiconductor photodetector of the present invention.

[0013] Figure 2 It is a spectrum response curve diagram of a low-noise semiconductor photodetector of the present invention.

[0014] Figure 3 : is a current characteristic curve of a low-noise semiconductor photodetector of the present invention. ph is the photocurrent curve, I dark is the dark current curve. (V) Specific implementation methods

[0015] The present invention will be further described below in conjunction with specific embodiments.

[0016] Embodiment 1: Figure 1 As shown, a low-noise semiconductor photodetector includes an avalanche photodiode (1) for receiving light signals normally and an avalanche photodiode (2) operating in a dark environment. The avalanche photodiode (1) for receiving light signals normally in the detector includes a top electrode (10), a top SiO2 layer (11), a top epitaxial layer (12), a first substrate (13) and a common electrode (3). The top electrode is a metal cathode, and the top epitaxial layer is heavily doped with N + The contact layer (121) is connected to N + The contact layer (121) is a cathode contact layer, and the doping concentration is 10 17 -10 19 cm -3 . N + Both sides of the contact layer (121) are lightly doped to form an N-Well region (122), which serves as a guard ring to prevent premature edge breakdown. + The contact layer (121) is lightly doped to form a Center-Nwell (123) which is used to form a PN junction with the top epitaxial layer (12). The top epitaxial layer (12) has a thickness of 1500-3000nm and a P-type doping concentration of 10 15 -10 17 cm -3 The electric field gathers in the PN junction depletion region, where photogenerated carriers and dark carriers are multiplied. Below the top epitaxial layer (12) is the first substrate (13), which has a doping concentration of 10 18 -10 20 cm -3 , as an anode contact layer connected to the common electrode (3). The common electrode (3) is a metal anode with a thickness greater than 2000nm.

[0017] The dark pixel avalanche photodiode (2) serves as a dark pixel, and includes a common electrode (3) serving as an anode of the dark pixel. A second substrate (23) below the common electrode (3) in contact with the common electrode (3) is a P-type heavily doped substrate with a concentration of 10 18 -10 20 cm -3The bottom epitaxial layer (22) is below the second substrate (23) and has a doping concentration of 10 15 -10 17 cm -3 , a P-type epitaxial layer with a thickness of 1500-3000nm; wherein the bottom epitaxial layer (22) comprises dark pixels N + The bottom electrode (20) is connected to the contact layer (221) for contacting the dark pixel cathode and the contact layer (221). At the same time, the bottom electrode (20) also isolates the bottom photons so that the dark pixel operates in a lightless environment. The dark pixel N-Well region (222) prevents the dark pixel edge from breaking down prematurely. The dark pixel Center-Nwell (223) and the bottom epitaxial layer (22) form a dark pixel PN junction, where dark carriers multiply. The bottom SiO2 layer (21) has a thickness of 100-300nm and is a protective layer. The shapes, sizes, and doping concentrations of the various regions in the above-mentioned dark pixel are consistent with those of the corresponding avalanche photodiode (1) that normally receives light signals.

[0018] When the detector is working, under the action of reverse bias voltage, photons are absorbed by the avalanche photodiode (1) that normally receives light signals, and the excited photogenerated carriers enter the PN junction depletion region formed by the Center-Nwell (123) and the P-type epitaxial layer (12), and the photogenerated electron-hole pairs are separated and accelerated. When they have sufficient energy, they collide with lattice atoms to ionize and generate new electron-hole pairs. The new electron-hole pairs will continue to collide and ionize to generate more electron-hole pairs, and the multiplied carriers are finally collected by the top electrode (10) to form a multiplied current. Due to the isolation of the middle metal common electrode (3) and the bottom electrode (20), there are no photons inside the dark pixel avalanche photodiode (2), and there is a large electric field in the dark pixel PN junction depletion region formed by the dark pixel Center-Nwell (223) and the bottom epitaxial layer (22), where dark carriers undergo a multiplication effect, and the dark noise current is output by the bottom electrode (20). The avalanche diode (2) operating in a dark environment has the same structure, size and process manufacturing parameters as the avalanche diode (1) operating normally in the detection light signal mode integrated together, and will have the same dark count rate. The dark counts of the two can compensate and offset to reduce the noise of the entire system. In theory, the dark count rate can be reduced to close to 0, which can improve the sensitivity and accuracy of the detection system.

[0019] The embodiment of the present invention provides a low-noise semiconductor photodetector that can achieve high-precision measurement of light signals in weak light application scenarios, while effectively reducing noise in the detection process. Compared with the prior art, the present invention makes full use of the design of the integrated dark pixel structure, so that the detector has the characteristics of high signal-to-noise ratio, high dynamic range, etc., meeting the needs of low-noise photodetection.

[0020] The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited thereto. Any modifications and changes made by those skilled in the art within the spirit and scope of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A low-noise semiconductor photodetector. Its characteristics are: The invention is composed of an avalanche photodiode (1) for receiving light signals normally and a dark pixel avalanche photodiode (2); the avalanche photodiode (1) for receiving light signals normally in the detector comprises a top SiO2 layer (11), a top epitaxial layer (12), a first substrate (13), and a common electrode (3) which are stacked in sequence; the top epitaxial layer (12) is provided with an N + The contact layer (121) is led out through the top electrode (10), N + An N-Well region (122) is provided outside the contact layer (121), and a Center-Nwell region (123) is provided in the center region below. The dark pixel avalanche photodiode (2) comprises a common electrode (3), a second substrate (23), a bottom epitaxial layer (22), and a bottom SiO2 layer (21); the bottom epitaxial layer (22) is provided with a dark pixel N + The contact layer (221) is led out through the bottom electrode (20), and the dark pixel N + A dark pixel N-Well region (222) is provided outside the contact layer (221), and a dark pixel Center-Nwell region (222) is provided in the upper central region.

2. The low-noise semiconductor photodetector according to claim 1, characterized in that: The avalanche photodiode (1) for normally receiving light signals and the dark pixel avalanche photodiode (2) are integrated into one body on the upper and lower sides of a common electrode (3) in a vertically stacked manner.

3. The low-noise semiconductor photodetector according to claim 1, characterized in that: The thickness of the SiO2 layer (11) is 100-300 nm, the thickness of the top epitaxial layer (12) is 1500-3000 nm, the thickness of the common electrode (3) is greater than 2000 nm, and the thickness of the N + The doping concentration of the contact layer (121) is 10 17 -10 19 cm -3 , the doping concentration of the first substrate (13) is 10 18 -10 20 cm -3 , the doping concentration of the top epitaxial layer (12) is 10 15 -10 17 cm -3 .

4. The low-noise semiconductor photodetector according to claim 1, characterized in that: The SiO2 layer (11), the top epitaxial layer (12), the first substrate (13), the N + The contact layer (121), the N-Well region (122), the Center-Nwell region (123) and the bottom SiO2 layer (21), the bottom epitaxial layer (22), the second substrate (23), the dark pixel N + The contact layer (221), the dark pixel N-Well region (222), and the dark pixel Center-Nwell region (222) have the same material and doping concentration, and are mirror-symmetrical in shape with respect to the common electrode (3) and have the same size.

Images