System and method for semiconductor defect detection based on light-induced perturbation impedance imaging

Abstract

The application discloses a kind of semiconductor defect detection system and method based on light-induced disturbance impedance imaging, it is related to semiconductor material and device detection technical field, in system, sample bearing and environment control module is used to bear semiconductor sample to be detected, and make semiconductor sample to be detected in air environment, vacuum environment or liquid phase environment;Imaging probe light output module is used to output imaging probe light to semiconductor sample to be detected;Periodic disturbance light output module is used to output periodic modulated disturbance light to semiconductor sample to be detected;Microscopic imaging module is used to image probe light response to image acquisition module;Image acquisition module is used to collect time sequence optical image of probe light response;Defect determination module is used to determine the position and type of defect in semiconductor sample to be detected.This system can carry out defect detection in air, vacuum or liquid phase environment, and does not need fluorescent label or electrical contact.

Description

Technical Field

[0001] This invention relates to the field of semiconductor material and device testing technology, and in particular to a semiconductor defect detection system and method based on photoinduced perturbation impedance imaging. Background Technology

[0002] The performance, reliability, and yield of semiconductor devices are highly dependent on the defect states within and on the surface of the material. Defects such as point defects, vacancies, impurity aggregation, dislocations, grain boundaries, stacking faults, microcracks, passivation layer defects, surface contamination, and localized leakage channels can all lead to shortened carrier lifetime, enhanced local recombination, increased leakage current, decreased breakdown voltage, reduced luminous efficiency, or thermal runaway. With the development of third-generation semiconductors, two-dimensional semiconductors, wide-bandgap power devices, micro-LEDs, and advanced integrated circuits, defect detection needs to simultaneously meet requirements such as large-area coverage, high detection sensitivity, non-contact operation, and high spatial resolution.

[0003] Existing semiconductor defect detection technologies mainly include bright-field optical detection, dark-field optical detection, fluorescence or photoluminescence imaging, Raman imaging, infrared thermal imaging, electron microscopy, scanning probe microscopy, deep-level transient spectroscopy, and electrical testing. Bright-field and dark-field optical detection are fast, but they mainly rely on morphology or scattering differences. For electrically active defects, subsurface defects, or weakly absorbed defects that do not produce obvious morphological fluctuations, their detection sensitivity is insufficient. Photoluminescence imaging can reflect carrier recombination processes, but not all semiconductor materials or device structures have sufficiently strong emission signals, and the emission intensity is easily affected by surface condition, thickness, and optical interference effects. Electron microscopy and scanning probe microscopy have high spatial resolution, but the detection area is limited, the detection throughput is low, and some techniques require complex sample preparation or vacuum operation conditions.

[0004] In many semiconductor samples, defect regions do not necessarily exhibit obvious brightness differences in static optical images. However, when subjected to external light, electricity, heat, or field disturbances, parameters such as carrier trapping and release, nonradiative recombination, local absorption, local refractive index, dielectric constant, and temperature at the defect site will show dynamic responses different from those of defect-free regions. Current technologies lack a detection system capable of operating in air, vacuum, or liquid environments, and switching between low-magnification large-area rapid screening and high-magnification fine localization. Furthermore, there is a lack of a universal semiconductor defect microscopic imaging detection method capable of converting dynamic changes in absorption, reflection, dielectric constant, or temperature caused by defects into pixel-level amplitude and phase maps. Summary of the Invention

[0005] This invention aims to address the shortcomings of existing technologies, specifically providing a semiconductor defect detection system and method based on optically induced perturbation impedance imaging, as detailed below: 1) In a first aspect, the present invention provides a semiconductor defect detection system based on optically induced perturbation impedance imaging, the specific technical solution of which is as follows: The system includes: a sample support and environmental control module, a microscopic imaging module, an imaging probe light output module, a periodic perturbation light output module, an image acquisition module, a synchronization control module, and a defect determination module. The sample support and environmental control module supports the semiconductor sample to be tested and maintains it in an air, vacuum, or liquid environment. The imaging probe light output module outputs imaging probe light to the semiconductor sample, which generates a probe light response on the sample. The periodic perturbation light output module outputs periodically modulated perturbation light to the semiconductor sample, causing changes in the photoelectric properties of defect areas within the sample. The microscopic imaging module is used to image the probe light response to the image acquisition module; the image acquisition module is used to acquire time-series optical images of the probe light response during the period when the perturbation light applies a perturbation to the semiconductor sample to be tested; the synchronization control module is used to control the synchronization of the periodic perturbation light output module and the image acquisition module; the defect determination module is connected to the image acquisition module, used to receive the time-series optical images, and perform pixel-level demodulation on the time-series optical images, extracting the signal components of each pixel or each region of interest that are at the same frequency as the perturbation light or have a fixed phase relationship with the perturbation light, so as to obtain amplitude maps and phase maps, and determine the location and type of defects in the semiconductor sample to be tested based on the amplitude maps and phase maps.

[0006] The beneficial effects of the semiconductor defect detection system based on optically induced perturbation impedance imaging provided by this invention are as follows: The imaging probe light output module outputs an imaging probe light, while the periodic perturbation light output module outputs periodically modulated perturbation light, causing changes in the photoelectric properties of the defect region. During the perturbation, the image acquisition module acquires time-series optical images of the probe light response. The defect determination module performs pixel-level demodulation on the time-series optical images to extract the signal component with the same frequency as the perturbation light, obtaining amplitude and phase maps. This allows for the extraction of weak dynamic responses at defects from a strong static background, improving the detection sensitivity for electrically active defects, subsurface defects, and weak absorption defects. The sample carrying and environmental control module enables the semiconductor sample to be tested to be placed in an air environment, a vacuum environment, or a liquid phase environment. An air environment is suitable for rapid screening and routine online detection; a vacuum environment is suitable for reducing the impact of air convection, water vapor adsorption, oxygen adsorption, and environmental refractive index fluctuations on weak photothermal or electrical responses; and a liquid phase environment is suitable for detecting wet process residues, surface passivation defects, and ion adsorption-related defects, giving the system multiple environmental adaptability. The microscopic imaging module images the probe light response to the image acquisition module, and can switch between low-magnification and high-magnification imaging modes. The low-magnification imaging mode is used for large-area rapid scanning, while the high-magnification imaging mode is used for fine localization and high spatial resolution retesting of suspected defect areas, thus balancing detection throughput and spatial resolution. The amplitude and phase maps output by the defect determination module provide intensity and time delay information of the defect response, which is helpful in distinguishing different types of defects such as morphological defects, electrical defects, and thermal defects.

[0007] Furthermore, the sample support and environmental control module includes a sample support stage, a sample chamber, an electric displacement stage, and a temperature control unit. The sample support stage is located within the sample chamber and is used to support the semiconductor sample to be tested. The electric displacement stage is connected to the sample support stage and is used to drive the sample support stage to move horizontally and vertically. The temperature control unit is connected to the sample support stage and is used to regulate the temperature of the semiconductor sample to be tested. The sample chamber is a sealed cavity, and a vacuum pump is connected to the sample chamber to create a vacuum environment within the sample chamber. The sample chamber is also connected to a gas inlet to introduce gas into the sample chamber to create an air environment. It is also equipped with a liquid sample cell or flow cell, which is used to contain liquid medium so that the semiconductor sample to be tested is immersed in the liquid medium to form a liquid environment. The liquid sample cell or flow cell is equipped with a transparent window, a seal, an inlet and outlet and a flow rate control unit. The transparent window allows imaging detection light and disturbance light to pass through. The seal is set between the transparent window and the sample chamber to ensure the sealing of the sample chamber. The inlet and outlet are connected to the liquid sample cell or flow cell and are used to introduce or discharge the liquid medium. The flow rate control unit is connected to the inlet and outlet and is used to control the flow rate of the liquid medium in the liquid sample cell or flow cell.

[0008] The beneficial effects of adopting the above-mentioned further scheme are as follows: The sample chamber is connected to a vacuum pump to create a vacuum environment, connected to a gas inlet to create an air environment, and a liquid sample cell or flow cell is set up to create a liquid environment. This allows the sample carrier and environmental control module to switch between vacuum, air, and liquid environments. The vacuum environment reduces the interference of air convection, water vapor adsorption, oxygen adsorption, and environmental refractive index fluctuations on the measurement. The liquid environment allows the sample to be detected in a liquid medium, expanding the system's application range. An electric displacement stage drives the sample carrier stage to move horizontally and vertically. Horizontal movement enables scanning and detection of the sample surface, while vertical movement adjusts the distance between the sample and the objective lens. A temperature control unit regulates the temperature of the semiconductor sample under test, enabling defect detection to be performed under different temperature conditions. The transparent window of the liquid sample cell or flow cell allows imaging probe light and disturbance light to pass through, while a seal ensures the sample chamber's airtightness. The inlet / outlet and flow rate control units control the introduction, discharge, and flow rate of the liquid medium, making the detection process in the liquid environment stable and controllable.

[0009] Furthermore, the imaging probe light output module includes a probe light source and a beam combining and splitting optical path. The probe light source can be a continuous laser light source, a pulsed laser light source, a light-emitting diode light source, a supercontinuum light source, or a broadband light source. The beam combining and splitting optical path is set on the output optical path of the probe light source. The beam combining and splitting optical path guides the imaging probe light to the microscopic imaging module. The imaging probe light is irradiated onto the surface of the semiconductor sample to be tested by the microscopic imaging module. The imaging probe light undergoes reflection, transmission, scattering, interference, or polarization response on the surface of the semiconductor sample to be tested to form a probe light response. The beam combining and splitting optical path also guides the probe light response to the image acquisition module. The wavelength of the imaging probe light is different from the wavelength of the perturbation light.

[0010] The advantages of adopting the above-mentioned further scheme are as follows: The detection light source can be a continuous laser source, a pulsed laser source, a light-emitting diode source, a supercontinuum source, or a broadband source, providing a variety of light source options. This allows the system to select a suitable light source based on the different characteristics of the semiconductor sample material and the detection requirements. The beam combining and splitting optical path is set on the output optical path of the detection light source, guiding the imaging detection light to the microscopic imaging module. The imaging detection light is then irradiated onto the surface of the semiconductor sample under test by the microscopic imaging module, where reflection, transmission, scattering, interference, or polarization responses occur to form the detection light response. The beam combining and splitting optical path also guides the detection light response to the image acquisition module, realizing the optical path guidance of the imaging detection light and the signal transmission of the detection light response, ensuring the stability of the optical path and the imaging quality. The wavelength of the imaging detection light is different from the wavelength of the perturbation light, allowing the beam combining and splitting optical path to separate the imaging detection light and the detection light response from the perturbation light based on the wavelength difference, preventing the perturbation light from entering the image acquisition module and interfering with the acquisition of the detection light response.

[0011] Furthermore, the periodic perturbation light output module includes a periodic perturbation light source and a modulator. The output end of the periodic perturbation light source is connected to the input end of the modulator. The modulator receives the light beam output by the periodic perturbation light source and periodically modulates the light beam to output perturbation light. The perturbation light is irradiated onto the surface of the semiconductor sample to be tested by the microscopic imaging module. The modulator is an acousto-optic modulator, an electro-optic modulator, a mechanical chopper, or a function generator. The modulation waveform of the perturbation light is a sine wave, a square wave, a triangular wave, a pulse wave, or a pseudo-random sequence. The wavelength of the perturbation light is located in the ultraviolet band, the visible band, the near-infrared band, or the mid-infrared band.

[0012] The beneficial effects of adopting the above-mentioned further scheme are as follows: A periodically disturbed light source outputs a beam; the modulator receives the beam and periodically modulates it to output disturbed light; the disturbed light is then irradiated onto the surface of the semiconductor sample under test by a microscopic imaging module, causing changes in the photoelectric properties of the defect region. This provides a reference signal synchronized with the dynamic response of the defect for subsequent pixel-level demodulation. The modulator can be an acousto-optic modulator, electro-optic modulator, mechanical chopper, or function generator, providing a variety of modulation methods, allowing the system to flexibly select the modulation method according to different detection requirements. The modulation waveform of the disturbed light can be a sine wave, square wave, triangular wave, pulse wave, or pseudo-random sequence. The response signal under sine wave excitation is easy to extract amplitude and phase through orthogonal demodulation; the transient transition of the square wave can excite the transient response signal of the defect region; triangular wave excitation can be used to measure the linearity of the defect response; short pulse pulse excitation can be used to excite the transient carrier dynamics process in the defect region; and pseudo-random sequence excitation has strong anti-interference capabilities and can simultaneously excite responses of multiple frequency components. The wavelength of the perturbation light is located in the ultraviolet, visible, near-infrared, or mid-infrared bands. Perturbation light in the ultraviolet and visible bands is suitable for wide bandgap semiconductors to excite carrier generation, while perturbation light in the near-infrared and mid-infrared bands is suitable for narrow bandgap semiconductors to excite carrier transitions on defect state energy levels or generate local thermal effects, enabling the system to be adapted to semiconductor materials with different bandgap widths.

[0013] Furthermore, the microscopic imaging module includes a low-magnification objective lens, a high-magnification objective lens, an objective lens switching mechanism, and an autofocus module. The low-magnification and high-magnification objectives are fixedly mounted on the objective lens switching mechanism. The objective lens switching mechanism drives the low-magnification and high-magnification objectives to switch so that the imaging probe light and disturbance light are irradiated onto the surface of the semiconductor sample to be tested through the low-magnification or high-magnification objective lens. The probe light response is returned to the image acquisition module through the low-magnification or high-magnification objective lens. The autofocus module is connected to the objective lens switching mechanism and controls the objective lens switching mechanism or the sample support and environment control module according to the image clarity feedback of the probe light response to adjust the relative distance between the semiconductor sample to be tested and the low-magnification or high-magnification objective lens.

[0014] The beneficial effects of adopting the above-mentioned further solution are as follows: The low-magnification objective and the high-magnification objective are respectively fixedly mounted on the objective switching mechanism. The objective switching mechanism drives the low-magnification objective and the high-magnification objective to switch, so that the imaging probe light and the disturbance light are irradiated onto the surface of the semiconductor sample to be tested through the low-magnification objective or the high-magnification objective. The probe light response is returned to the image acquisition module through the low-magnification objective or the high-magnification objective. The low-magnification objective has a large field of view, which can quickly scan the sample over a large area. The high-magnification objective has high spatial resolution, which can perform fine imaging of local areas. The switching between the low-magnification objective and the high-magnification objective allows the system to meet the needs of both large-area rapid screening and high-resolution fine detection. The autofocus module is connected to the objective switching mechanism and controls the objective switching mechanism or the sample carrier and environment control module according to the image clarity feedback of the probe light response to adjust the relative distance between the semiconductor sample to be tested and the low-magnification objective or the high-magnification objective, ensuring that the probe light response image remains clear after the objective switching.

[0015] Furthermore, the image acquisition module is an area array camera, a line array camera, an avalanche photodiode array, a photomultiplier tube array, an infrared camera, or a high-speed camera. The image acquisition module is connected to the synchronization control module and receives the trigger signal output by the synchronization control module. The image acquisition module continuously acquires time-series optical images of the probe light response at a preset frame rate throughout the entire process of the perturbation light applying perturbation to the semiconductor sample to be detected, based on the trigger signal. The preset frame rate is at least twice the modulation frequency of the perturbation light.

[0016] The beneficial effects of adopting the above-mentioned further scheme are: it enables the system to select an appropriate image acquisition module according to different detection scenarios and requirements. The image acquisition module is connected to the synchronization control module and receives the trigger signal output by the synchronization control module, ensuring that the acquisition action of the image acquisition module is synchronized with the periodic perturbation light output module, thus guaranteeing the correspondence between each frame of the time-series optical image and the modulation phase of the perturbation light. Based on the trigger signal, the image acquisition module continuously acquires time-series optical images of the probe light response at a preset frame rate throughout the entire process of the perturbation light applying perturbation to the semiconductor sample under test, providing complete time-series optical image data for the defect determination module. The preset frame rate is at least twice the modulation frequency of the perturbation light. According to the Nyquist sampling theorem, when the preset frame rate is greater than or equal to twice the modulation frequency of the perturbation light, the acquired time-series optical images can completely retain the signal components with the same frequency as the perturbation light, enabling the defect determination module to accurately extract amplitude and phase information from the time-series optical images.

[0017] Furthermore, when the semiconductor sample to be tested is in an air environment, the defect determination module performs air background subtraction and flat field correction on the time-series optical image to obtain a first time-series optically corrected image; when the semiconductor sample to be tested is in a vacuum environment, the defect determination module performs vacuum background subtraction and flat field correction on the time-series optical image to obtain a second time-series optically corrected image; when the semiconductor sample to be tested is in a liquid phase environment, the defect determination module performs liquid phase background subtraction and flat field correction on the time-series optical image to obtain a third time-series optically corrected image; the defect determination module then performs pixel-level demodulation on the first time-series optically corrected image, the second time-series optically corrected image, or the third time-series optically corrected image.

[0018] The beneficial effects of adopting the above-mentioned further scheme are as follows: When the semiconductor sample to be tested is in an air environment, the defect determination module performs air background subtraction and flat field correction on the time-series optical image to obtain a first time-series optically corrected image; when the semiconductor sample to be tested is in a vacuum environment, the defect determination module performs vacuum background subtraction and flat field correction on the time-series optical image to obtain a second time-series optically corrected image; when the semiconductor sample to be tested is in a liquid phase environment, the defect determination module performs liquid phase background subtraction and flat field correction on the time-series optical image to obtain a third time-series optically corrected image. The defect determination module performs corresponding background subtraction and flat field correction on the time-series optical image according to the different environmental types of the sample. Air background subtraction eliminates the interference of static optical signals and dark current noise in the air environment; vacuum background subtraction eliminates the interference of static optical signals and dark current noise in the vacuum environment; liquid phase background subtraction eliminates the interference of static optical signals, liquid medium absorption, and scattering in the liquid phase environment; and flat field correction eliminates the effects of illumination non-uniformity and optical element transmittance non-uniformity. The defect determination module then performs pixel-level demodulation on the corrected time-series optically corrected image, so that the amplitude map and phase map obtained by demodulation can truly reflect the dynamic response signal of the defect region in the semiconductor sample to be tested.

[0019] Furthermore, the process by which the defect determination module obtains the first time-series optical correction image includes: the defect determination module acquiring multiple frames of air background images acquired by the image acquisition module when the periodic perturbation light output module is in the off state, and performing pixel-level averaging on the multiple frames of air background images to obtain an average air background image, and subtracting the pixel value of the corresponding position in the average air background image from each frame of the time-series optical image pixel by pixel; acquiring multiple frames of uniform illumination images acquired by the image acquisition module when the periodic perturbation light output module is in the on state and the semiconductor sample to be tested is not loaded, and performing pixel-level averaging on the multiple frames of uniform illumination images to obtain a flat field correction reference image, and dividing each frame of the image after air background subtraction pixel by the pixel value of the corresponding position in the flat field correction reference image and performing normalization processing to obtain the first time-series optical correction image; The process of the defect determination module obtaining the second time-series optical correction image includes: the defect determination module acquiring multiple frames of vacuum background images acquired by the image acquisition module when the sample cavity is in a vacuum state and the periodic perturbation light output module is in a closed state, and performing pixel-level averaging on the multiple frames of vacuum background images to obtain a vacuum background average image, and subtracting the pixel value of the corresponding position in the vacuum background average image from each frame of the time-series optical image pixel by pixel; acquiring multiple frames of uniform illumination images acquired by the image acquisition module when the sample cavity is in a vacuum state, the periodic perturbation light output module is in a closed state and the semiconductor sample to be tested is not loaded, and performing pixel-level averaging on the multiple frames of uniform illumination images to obtain a flat field correction reference image, and dividing each frame of the image after vacuum background subtraction pixel by the pixel value of the corresponding position in the flat field correction reference image and performing normalization processing to obtain the second time-series optical correction image; The process of the defect determination module obtaining the third time-series optically corrected image includes: the defect determination module acquiring multiple frames of liquid background images acquired by the image acquisition module when the liquid sample cell or flow cell is filled with liquid medium and the semiconductor sample to be tested is not immersed in the liquid medium and the periodic perturbation light output module is in the off state; and performing pixel-level averaging on the multiple frames of liquid background images to obtain a liquid background average image; and subtracting the pixel value of the corresponding position in the liquid background average image from each frame of the time-series optical image pixel by pixel; acquiring multiple frames of uniform illumination images acquired by the image acquisition module when the liquid sample cell or flow cell is filled with liquid medium and the semiconductor sample to be tested is not immersed in the liquid medium and the periodic perturbation light output module is in the on state; and performing pixel-level averaging on the multiple frames of uniform illumination images to obtain a flat field correction reference image; and dividing each frame of the image after liquid background subtraction pixel by the pixel value of the corresponding position in the flat field correction reference image and performing normalization processing to obtain the third time-series optically corrected image.

[0020] The beneficial effects of adopting the above-mentioned further solutions are as follows: The defect determination module acquires multiple frames of air background images in an air environment and performs pixel-level averaging to obtain an average air background image. It then subtracts the corresponding pixel value from the average air background image pixel by pixel from each frame of the time-series optical image, eliminating interference from static optical signals and dark current noise in the air environment. Similarly, in a vacuum environment, it acquires multiple frames of vacuum background images and performs pixel-level averaging to obtain an average vacuum background image. It then subtracts the corresponding pixel value from the average vacuum background image pixel by pixel from each frame of the time-series optical image, eliminating interference from static optical signals and dark current noise in the vacuum environment. Furthermore, in a liquid phase environment, it acquires multiple frames of liquid phase background images and performs pixel-level averaging to obtain an average liquid phase background image. It then subtracts the corresponding pixel value from the average liquid phase background image pixel by pixel from each frame of the time-series optical image, eliminating interference from static optical signals, liquid medium absorption, and scattering in the liquid phase environment. The defect determination module acquires multiple frames of uniformly illuminated images under three different environments and performs pixel-level averaging to obtain a flat-field correction reference image. Each frame, after background subtraction, is then divided pixel-by-pixel by the corresponding pixel value in the flat-field correction reference image and normalized to eliminate the effects of illumination non-uniformity and optical element transmittance non-uniformity. Pixel-level averaging, pixel-by-pixel subtraction, and pixel-by-pixel division are all performed at the single-pixel level, ensuring signal accuracy and spatial resolution at each pixel location in the corrected image.

[0021] Furthermore, when the defect determination module performs pixel-level demodulation on the first, second, or third time-series optically corrected images, it extracts the intensity sequence of each pixel as it changes over time in the first, second, or third time-series optically corrected images. Using the modulation signal of the perturbed light as a reference signal, it performs orthogonal demodulation on the intensity sequence to extract the cosine and sine components with the same frequency as the perturbed light. The cosine component is the integral average of the product of the intensity sequence and the cosine function of the reference signal over a complete modulation period, and the sine component is the product of the intensity sequence and the reference signal. The integral average of the product of the sine functions over the complete modulation period is used to determine the magnitude and phase of the pixel. The defect determination module calculates the magnitude and phase of the pixel based on the cosine and sine components. The magnitude is the square root of the sum of the squares of the cosine and sine components, and the phase is the inverse function of the tangent of the ratio of the sine to the cosine components. The defect determination module performs the above processing on all pixels to obtain magnitude and phase maps, and performs spatial filtering and threshold segmentation on the magnitude and phase maps to extract magnitude and phase abnormal regions. The spatial intersection of the magnitude and phase abnormal regions is marked as the location of the defect, and the type of defect is determined based on the magnitude and phase corresponding to the location of the defect.

[0022] The beneficial effects of the above-mentioned further scheme are as follows: The defect determination module extracts the intensity sequence of each pixel in the time-series optically corrected image, which varies over time. Using the modulation signal of the perturbed light as a reference signal, the intensity sequence is orthogonally demodulated to extract the cosine and sine components with the same frequency as the perturbed light. The cosine and sine components are the integral averages of the cosine and sine functions of the intensity sequence and the reference signal, respectively, over a complete modulation period. Integral averaging over a complete modulation period eliminates noise and interference components in the intensity sequence that have a different modulation frequency than the perturbed light, ensuring that the cosine and sine components contain only the response components with the same frequency as the perturbed light. The defect determination module calculates the amplitude and phase of each pixel based on the cosine and sine components. The amplitude is the square root of the sum of the squares of the cosine and sine components, representing the response intensity with the same frequency as the perturbed light in the probe light response signal. The phase is the inverse function of the tangent function of the ratio of the sine to the cosine components, representing the phase delay of the probe light response signal relative to the modulation signal of the perturbed light. The defect determination module processes all pixels separately to obtain amplitude and phase maps. Spatial filtering and thresholding are then applied to both maps to extract amplitude and phase anomaly regions. The spatial intersection of these regions is marked as the defect location. By utilizing information from both amplitude and phase anomalies, defect localization is reduced due to single-dimensional anomalies, improving accuracy. Finally, the module determines the defect type based on the amplitude and phase corresponding to the defect's location, enabling the system to distinguish between defects with different physical mechanisms.

[0023] 2) In a second aspect, the present invention also provides a semiconductor defect detection method based on optically induced perturbation impedance imaging, the specific technical solution of which is as follows: The semiconductor defect detection system based on optically induced perturbation impedance imaging, as described above, comprises the following steps: placing the semiconductor sample to be tested in a sample support and environmental control module, and subjecting the semiconductor sample to an air environment, a vacuum environment, or a liquid environment; outputting an imaging probe light to the semiconductor sample to be tested by activating the imaging probe light output module, wherein the imaging probe light forms a probe light response on the semiconductor sample to be tested; outputting periodically modulated perturbation light to the semiconductor sample to be tested by activating the periodic perturbation light output module, wherein the perturbation light is used to induce photoelectric property changes in the defect region of the semiconductor sample to be tested; controlling the synchronization of the periodic perturbation light output module and the image acquisition module by a synchronization control module, wherein the image acquisition module acquires a time-series optical image of the probe light response during the period when the perturbation light applies perturbation to the semiconductor sample to be tested; receiving the time-series optical image by a defect determination module, performing pixel-level demodulation on the time-series optical image, extracting the signal component of each pixel or each region of interest that is at the same frequency as the perturbation light or has a fixed phase relationship with the perturbation light, thereby obtaining an amplitude map and a phase map, and determining the location and type of defect in the semiconductor sample to be tested based on the amplitude map and the phase map.

[0024] It should be noted that the beneficial effects of the technical solution of the second aspect of the present invention and the corresponding possible implementation can be found in the above description of the technical effects of the first aspect and its corresponding possible implementation, and will not be repeated here. Attached Figure Description

[0025] Figure 1 This is one of the structural schematic diagrams of a semiconductor defect detection system based on optically induced perturbation impedance imaging according to an embodiment of the present invention; Figure 2 This is a schematic flowchart of a semiconductor defect detection method based on optically induced perturbation impedance imaging according to an embodiment of the present invention. Figure 3 This is a second schematic diagram of a semiconductor defect detection system based on optically induced perturbation impedance imaging, according to an embodiment of the present invention. Figure 4 This is a schematic diagram illustrating the workflow for low-magnification rapid screening and high-magnification precise localization. Detailed Implementation

[0026] like Figure 1 As shown, a semiconductor defect detection system based on light-induced perturbation impedance imaging according to an embodiment of the present invention includes: a sample support and environment control module, a microscopic imaging module, an imaging detection light output module, a periodic perturbation light output module, an image acquisition module 14, a synchronization control module 1, and a defect determination module 15. The sample carrier and environment control module is used to carry the semiconductor sample 10 to be tested and to place the semiconductor sample 10 to be tested in an air environment, a vacuum environment, or a liquid phase environment, specifically: The sample support and environmental control module includes a sample support stage 9, a sample chamber 11, an electric displacement stage 12, and a temperature control unit. The sample support stage 9 is disposed within the sample chamber 11 and is used to support the semiconductor sample 10 to be tested. The electric displacement stage 12 is connected to the sample support stage 9 and is used to drive the sample support stage 9 to move in the horizontal and vertical directions. The temperature control unit is connected to the sample support stage 9 and is used to regulate the temperature of the semiconductor sample 10 to be tested. The sample chamber 11 is a sealed cavity. The sample chamber 11 is connected to a vacuum pump to create a vacuum environment within the sample chamber 11. The sample chamber 11 is also connected to a gas inlet to introduce gas into the sample chamber 11 to form air. The sample chamber 11 is also equipped with a liquid sample cell or a flow cell. The liquid sample cell or flow cell is used to contain a liquid medium so that the semiconductor sample 10 to be tested is immersed in the liquid medium to form a liquid environment. The liquid sample cell or flow cell is equipped with a transparent window, a seal, an inlet and outlet, and a flow rate control unit. The transparent window allows imaging detection light and disturbance light to pass through. The seal is set between the transparent window and the sample chamber 11 to ensure the sealing of the sample chamber 11. The inlet and outlet are connected to the liquid sample cell or flow cell and are used to introduce or discharge the liquid medium. The flow rate control unit is connected to the inlet and outlet and is used to control the flow rate of the liquid medium in the liquid sample cell or flow cell.

[0027] The sample stage 9 is disposed within the sample cavity 11 and is used to hold the semiconductor sample 10 to be tested. The semiconductor sample 10 to be tested includes silicon, germanium, silicon carbide, gallium nitride, gallium arsenide, indium phosphide, gallium oxide, diamond, perovskite semiconductors, oxide semiconductors, two-dimensional semiconductors, semiconductor epitaxial wafers, thin-film solar cell materials, light-emitting diode epitaxial wafers, power device wafers, integrated circuit wafers, or chips before and after packaging. In specific applications, For silicon carbide power device wafers, the perturbation light is selected at a wavelength higher than or close to the absorption edge of the silicon carbide material to induce carrier generation, and the imaging probe light is selected as visible light or near-infrared light that will not strongly excite the sample. Carrier trapping and non-radiative recombination enhancement at dislocations, stacking faults or local recombination centers in silicon carbide result in local reflectivity, refractive index or photothermal response that differs from the surrounding defect-free region. The amplitude map and phase map obtained by pixel-level demodulation are used to locate the electrical active defects in the silicon carbide power device wafer.

[0028] For gallium nitride epitaxial wafers or micro-LED epitaxial wafers, nonradiative recombination, localized stress, and thermal accumulation at defects cause coherent changes in reflected or interfering signals under periodic perturbations; amplitude and phase maps provide time delay information on defect response, which is used to distinguish between surface morphology defects and electrically active defects.

[0029] For silicon-based integrated circuit wafers or thin film structures, perturbation light induces local thermal expansion, changes in thermal reflectivity, or changes in dielectric function at defects in metal interconnects, passivation layers, or dielectric layers. Imaging probe light reads the local thermal expansion, changes in thermal reflectivity, or changes in dielectric function through reflection or interference. The thermal diffusion and local absorption in defective regions differ from those in normal regions, and their phase delay and frequency response curves are used to identify local hot spots, voids, cracks, or interlayer delamination.

[0030] For two-dimensional semiconductor or perovskite semiconductor thin films, grain boundaries, layer number variations, ion migration-related defects, localized trap states, or surface adsorption sites exhibit different absorption, reflection, or dielectric responses under periodic optical perturbations; low-magnification objective 6 is used for large-area thin film uniformity evaluation, while high-magnification objective 7 is used for defect localization in individual grains, grain boundaries, or micro / nano device channels.

[0031] The sample support stage 9 has a flat plate structure, with a smooth upper surface on which the semiconductor sample 10 to be tested is placed. The sample support stage 9 is made of a metal material with good thermal conductivity, such as aluminum alloy or oxygen-free copper, to ensure that the temperature control unit can transfer heat to the semiconductor sample 10 through the sample support stage 9. The upper surface of the sample support stage 9 is provided with vacuum adsorption holes or mechanical clamps. The vacuum adsorption holes are connected to an external vacuum pump via pipes, and the mechanical clamps are fixedly installed at the edge of the sample support stage 9. The vacuum adsorption holes and mechanical clamps are used to fix the semiconductor sample 10 to the bearing surface of the sample support stage 9, preventing displacement of the semiconductor sample 10 when the sample support stage 9 is moved by the electric displacement stage 12. The lower surface of the sample support stage 9 is fixedly connected to the upper end of the electric displacement stage 12.

[0032] The electric displacement stage 12 is connected to the sample support stage 9 and is used to drive the sample support stage 9 to move in the horizontal and vertical directions. The electric displacement stage 12 is a multi-axis precision displacement platform, which includes a horizontal motion mechanism and a vertical motion mechanism. The horizontal motion mechanism is located at the bottom of the sample cavity 11. The horizontal motion mechanism is a linear displacement platform orthogonally arranged along the X and Y axes. The X-axis linear displacement platform and the Y-axis linear displacement platform are respectively composed of a servo motor driven lead screw and nut pair. The servo motor is fixedly installed at the bottom of the sample cavity 11. The lead screw in the lead screw and nut pair extends horizontally, and the nut in the lead screw and nut pair is fixedly connected to the upper motion platform. When the servo motor drives the lead screw to rotate, the nut drives the motion platform to move along the extension direction of the lead screw. A vertical motion mechanism is positioned above the horizontal motion mechanism. This vertical motion mechanism is a Z-axis linear displacement platform, consisting of a servo motor-driven lead screw and nut assembly. The lead screw extends vertically, and its lower end is fixedly connected to the motion platform of the horizontal motion mechanism. The nut in the Z-axis linear displacement platform is fixedly connected to the lower surface of the sample support stage 9. When the servo motor drives the lead screw to rotate, the nut causes the sample support stage 9 to move vertically. The electric displacement stage 12, through the combined motion of the horizontal and vertical motion mechanisms, drives the sample support stage 9 to move horizontally and vertically, thereby achieving scanning movement of the semiconductor sample 10 under test in the horizontal direction and focusing movement in the vertical direction.

[0033] The temperature control unit is connected to the sample stage 9 and is used to regulate the temperature of the semiconductor sample 10 to be tested. The temperature control unit is located outside or inside the sample chamber 11. The temperature control unit includes a heating element, a cooling element, a temperature sensor, and a temperature controller. The heating element is a resistance heater or a thermoelectric heater, attached to the lower surface of the sample stage 9 or embedded inside the sample stage 9, and electrically connected to the temperature controller via wires. The cooling element is a thermoelectric cooler or a circulating liquid cooling pipe, attached to the lower surface of the sample stage 9, and the circulating liquid cooling pipe is embedded inside the sample stage 9, connected to an external refrigeration cycler via pipes. The temperature sensor is a thermistor or a thermocouple, fixedly installed on the upper surface of the sample stage 9 or embedded inside the sample stage 9, and electrically connected to the temperature controller via wires. The temperature controller is located outside the sample chamber 11 and is electrically connected to the heating element, the cooling element, and the temperature sensor via wires. The temperature controller receives the temperature signal of the sample stage 9 from the temperature sensor and controls the working state of the heating element or the cooling element according to the temperature signal to maintain the temperature of the sample stage 9 within the preset temperature range. The sample stage 9 transfers the temperature to the semiconductor sample 10 to be tested through heat conduction, thereby adjusting the temperature of the semiconductor sample 10 to be tested.

[0034] The sample chamber 11 is a sealed cavity, a box structure with an internal cavity. The side and bottom walls of the sample chamber 11 are made of metal, such as stainless steel or aluminum alloy. An openable sealing cover is provided on the top of the sample chamber 11, hinged to the side wall of the sample chamber 11. A sealing ring is provided between the sealing cover and the side wall of the sample chamber 11. When the sealing cover is closed, the sealing ring is pressed between the sealing cover and the side wall of the sample chamber 11 to ensure the airtightness of the sample chamber 11. A transparent window, made of optical glass or quartz glass, is also provided on the top or side wall of the sample chamber 11. The transparent window is fixedly installed at an opening on the top or side wall of the sample chamber 11. A seal is provided between the transparent window and the edge of the opening in the sample chamber 11. The transparent window allows imaging probe light and disturbance light to pass through and illuminate the surface of the semiconductor sample 10 to be tested inside the sample chamber 11. Simultaneously, it allows probe light to pass through and exit the sample chamber 11, entering the image acquisition module 14.

[0035] A vacuum pump is connected to the sample chamber 11 to create a vacuum environment within it. The vacuum pump is located outside the sample chamber 11, and its extraction port is connected to the interior of the sample chamber 11 via a vacuum pipe. A vacuum valve is installed on the vacuum pipe, and a sealing joint is provided at the connection between the vacuum pipe and the sample chamber 11 to ensure a tight seal. When the vacuum pump is started, it extracts gas from inside the sample chamber 11 through the vacuum pipe, reducing the pressure inside the sample chamber 11 to a preset vacuum level, thereby creating a vacuum environment within the sample chamber 11.

[0036] The sample chamber 11 is also connected to a gas inlet to introduce gas into the sample chamber 11 and create an air environment. The gas inlet is located on the side wall or top of the sample chamber 11 and is a through hole penetrating the side wall or top wall of the sample chamber 11. The outer end of the gas inlet is connected to an external gas source through a pipe, and the inner end of the gas inlet is connected to the interior of the sample chamber 11. The external gas source introduces dry air, nitrogen, or inert gas into the sample chamber 11 through the gas inlet to restore the gas pressure inside the sample chamber 11 to a normal atmospheric pressure, thereby creating an air environment inside the sample chamber 11. The air environment is suitable for rapid screening and routine online detection of the semiconductor sample 10 to be tested.

[0037] The sample chamber 11 also includes a liquid sample cell or a flow cell. The liquid sample cell or flow cell is used to contain a liquid medium so that the semiconductor sample 10 to be tested is immersed in the liquid medium to form a liquid environment. The liquid sample cell or flow cell is located inside the sample chamber 11 and above the sample support stage 9 or in a position that replaces the sample support stage 9. The liquid sample cell is a container structure with an open top, a flat bottom plate, and side walls perpendicular to the bottom plate. The bottom plate and side walls together form a cavity for containing the liquid medium. The bottom plate of the liquid sample cell is fixedly installed at the bottom of the sample chamber 11 or on the motion platform of the electric displacement stage 12. The flow cell is a closed or semi-closed flow channel structure with a liquid inlet and a liquid outlet. The interior of the flow cell has a flow channel for the liquid medium to flow through. The bottom of the flow cell is a flat bottom plate, which is fixedly installed at the bottom of the sample chamber 11 or on the motion platform of the electric displacement stage 12. The semiconductor sample 10 to be tested is placed in the detection area on the bottom plate of the flow cell.

[0038] The liquid sample cell or flow cell is equipped with a transparent window, a seal, inlet and outlet ports, and a flow rate control unit. The liquid sample cell or flow cell can also be further configured with electrodes or reference electrodes to apply an external electric field to the semiconductor sample or measure its electrochemical potential in a liquid environment, thereby expanding the detection capabilities. The transparent window allows imaging probe light and disturbance light to pass through. The transparent window is located at the top or bottom of the liquid sample cell or flow cell and is made of optical glass or quartz glass. The transparent window is fixedly installed at the opening of the liquid sample cell or flow cell. When the transparent window is located at the top of the liquid sample cell or flow cell, the imaging probe light and disturbance light pass downwards through the transparent window from above and then illuminate the surface of the semiconductor sample 10 immersed in the liquid medium. When the transparent window is located at the bottom of the liquid sample cell or flow cell, the imaging probe light and disturbance light pass upwards through the transparent window from below and then illuminate the surface of the semiconductor sample 10 immersed in the liquid medium.

[0039] A seal is positioned between the transparent window and the sample chamber 11 to ensure the airtightness of the sample chamber 11. The seal is an O-ring or a flat gasket, made of an elastic sealing material such as fluororubber or silicone rubber. The seal is positioned between the outer edge of the transparent window and the opening edge of the sample chamber 11. The seal is pressed tightly between the outer edge of the transparent window and the opening edge of the sample chamber 11. The elastic deformation of the seal fills the gap between the transparent window and the opening edge of the sample chamber 11, thereby preventing gas or liquid inside the sample chamber 11 from leaking to the outside of the sample chamber 11 through the gap between the transparent window and the sample chamber 11, while also preventing gas from the outside of the sample chamber 11 from entering the sample chamber 11 through this gap.

[0040] The inlet and outlet ports are connected to the liquid phase sample cell or flow cell and are used to introduce or discharge the liquid medium. The inlet and outlet ports are pipe fittings that penetrate the side wall or bottom plate of the liquid phase sample cell or flow cell. The inner end of the inlet and outlet port is connected to the receiving cavity of the liquid phase sample cell or the flow channel of the flow cell, and the outer end of the inlet and outlet port is connected to an external liquid source or waste liquid collection container via a pipe. The inlet and outlet ports include an inlet and an outlet. The inlet is used to introduce the liquid medium into the interior of the liquid phase sample cell or flow cell, and the outlet is used to discharge the liquid medium from the interior of the liquid phase sample cell or flow cell.

[0041] The flow rate control unit is connected to the inlet and outlet and is used to control the flow rate of the liquid medium in the liquid phase sample cell or flow cell. The flow rate control unit is located outside the sample chamber 11 and can be a peristaltic pump, syringe pump, or flow regulating valve. The inlet of the flow rate control unit is connected to an external liquid source via a pipe, and the outlet is connected to the inlet via a pipe. After the flow rate control unit is activated, it pumps the liquid medium from the external liquid source to the inlet at a preset flow rate. The liquid medium enters the interior of the liquid phase sample cell or flow cell through the inlet, flows through the containment cavity of the liquid phase sample cell or the flow channel of the flow cell, and is discharged from the outlet. The flow rate control unit controls the flow rate of the liquid medium by adjusting the pumping speed or valve opening, ensuring a stable flow state of the liquid medium in the liquid phase sample cell or flow cell. The detection principle of the semiconductor defect detection system using photoinduced perturbation impedance imaging of this invention does not depend on the specific composition of the liquid medium, nor on whether a chemical reaction occurs in the liquid phase environment, including electrochemical or photochemical reactions. The detection mode in the liquid phase environment is suitable for detecting defects in wet process residues, surface passivation defects, liquid wetting-induced trap states, ion adsorption or migration-related defects, and semiconductor materials in electrochemical or photoelectrochemical working environments.

[0042] The sample support and environmental control module also includes wafer or chip fixtures for fixing the semiconductor sample 10 to be tested, which is of a corresponding shape. The sample chamber 11 is also connected to an atmosphere control unit for adjusting the gas composition within the sample chamber 11. A vacuum environment is used to reduce the effects of air convection, water vapor adsorption, oxygen adsorption, and fluctuations in the ambient refractive index on weak photothermal or electrical responses. The vacuum environment can also be used in conjunction with a temperature control unit to adjust the temperature of the semiconductor sample 10 to be tested, thereby reducing environmental interference and enhancing the ability to distinguish deep-level defects, thermal conductivity anomalies, or subsurface defects.

[0043] The imaging probe light output module is used to output imaging probe light to the semiconductor sample 10 to be tested. The imaging probe light forms a probe light response on the semiconductor sample 10 to be tested, specifically: The imaging probe light output module includes a probe light source 4 and a beam combining / splitting optical path 5. The probe light source 4 can be a continuous laser source, a pulsed laser source, a light-emitting diode source, a supercontinuum source, or a broadband source. The beam combining / splitting optical path 5 is located on the output optical path of the probe light source 4. The beam combining / splitting optical path 5 guides the imaging probe light to the microscopic imaging module. The imaging probe light is then irradiated onto the surface of the semiconductor sample 10 to be tested by the microscopic imaging module. The imaging probe light undergoes reflection, transmission, scattering, interference, or polarization response on the surface of the semiconductor sample 10 to form a probe light response. The beam combining / splitting optical path 5 also guides the probe light response to the image acquisition module 14. The wavelength of the imaging probe light is different from the wavelength of the perturbation light. The imaging probe light output by the imaging probe light output module has different wavelengths, different incident angles, different polarization states, different spot sizes, or different optical path directions from the perturbation light. The imaging probe light does not need to strongly excite the semiconductor sample 10 to be tested. Its main function is to read the changes in the optical response of the sample under the action of the perturbation light. Therefore, the power density of the imaging probe light is lower than that of the perturbation light to avoid introducing additional photoinduced effects that interfere with defect determination.

[0044] The detection source 4 can be a continuous laser source, a pulsed laser source, a light-emitting diode source, a supercontinuum source, or a broadband source. The detection source 4 is fixedly mounted on the upper surface of the optical platform or frame, with its exit face facing the incident face of the beam combining / splitting optical path 5. The detection source 4 is connected to an external power supply via a power line, which provides operating current to drive the detection source 4 to emit imaging detection light. The continuous laser source is a laser that outputs a continuous and stable laser beam. The continuous laser source includes a laser resonant cavity, with a total reflection mirror and a partial reflection mirror at each end. The laser resonant cavity is filled with a laser gain medium. The pump source excites the laser gain medium to generate stimulated emission, which oscillates within the laser resonant cavity and then outputs a continuous laser beam from the partial reflection mirror. A pulsed laser source is a laser that outputs a pulsed laser beam. Internally, it includes a pulse generation circuit or a Q-switching element. The pulse generation circuit controls the pump source to operate in a pulsed manner. The Q-switching element is placed inside the laser resonant cavity and periodically changes the quality factor of the laser resonant cavity, causing it to release a high-energy pulsed laser beam in a short time. A light-emitting diode (LED) source is a light source that uses a semiconductor LED chip as the light emitter. The LED chip is fixedly mounted on a packaged substrate, which is connected to an external power supply via wires. When powered on, the LED chip emits imaging and detection light with a continuous or narrow-band spectrum. A supercontinuum source is a light source that outputs a supercontinuum broadband laser beam. It includes a mode-locked pulsed laser and a photonic crystal fiber. The mode-locked pulsed laser outputs an ultrashort pulse laser beam, which is coupled into the photonic crystal fiber. During propagation in the photonic crystal fiber, the ultrashort pulse laser beam undergoes nonlinear and dispersion effects, broadening its spectrum to a supercontinuum laser beam covering the visible to near-infrared band. A broadband light source is a light source that outputs a wide-band continuous spectrum. Broadband light sources can be halogen lamps, xenon lamps, or light-emitting diode arrays. When the filament of a halogen lamp is energized, it is heated and emits light, outputting a continuous spectrum covering the ultraviolet to infrared band. A xenon lamp is a high-pressure gas discharge light source. When the two electrodes of a xenon lamp are energized, xenon gas is excited to discharge and emit light, outputting a wide-band continuous spectrum. A light-emitting diode array is an array composed of multiple light-emitting diode chips with different emission wavelengths. When multiple light-emitting diode chips are energized simultaneously, they emit composite broadband light covering multiple bands.

[0045] The beam combining and splitting optical path 5 is disposed on the output optical path of the detection light source 4. The beam combining and splitting optical path 5 is an optical system composed of multiple optical elements arranged in a predetermined spatial position, and is fixedly mounted on the upper surface of the optical platform or frame. The beam combining and splitting optical path 5 includes a dichroic mirror, a reflector, a beam splitter, and a lens group.

[0046] A dichroic mirror is an optical lens with wavelength-selective transmission and reflection characteristics. The substrate of a dichroic mirror is an optical glass substrate, one surface of which is coated with a dielectric film. This dielectric film is composed of multiple layers of alternating high-refractive-index and low-refractive-index materials. The dielectric film reflects light beams within a specific wavelength range while transmitting light beams within another specific wavelength range. A reflector can be a plane mirror or a curved mirror. The substrate of the reflector is an optical glass substrate or a metal substrate. The reflective surface of the reflector is coated with a metallic reflective film or a dielectric reflective film. The metallic reflective film can be an aluminum film, a silver film, or a gold film. The dielectric reflective film is composed of multiple layers of alternating high-refractive-index and low-refractive-index materials. The reflective surface of the reflector reflects the incident light beam to change the direction of beam propagation. The beam splitter is either a semi-transparent, semi-reflective mirror or a polarizing beam splitter. The substrate of the semi-transparent, semi-reflective mirror is an optical glass substrate, with a semi-transparent, semi-reflective coating on one surface. This coating reflects a portion of the incident beam and transmits the other portion. The polarizing beam splitter is composed of two right-angle prisms cemented together, with a polarizing beam-splitting coating on the cemented surface. This coating transmits P-polarized light from the incident beam and reflects S-polarized light. The lens group consists of one or more optical lenses arranged in a predetermined spatial position. These lenses can be convex, concave, or achromatic lenses, and their substrate is also an optical glass substrate. Both surfaces of the optical lenses are spherical or aspherical. The lens group collimates, expands, or focuses the imaging probe light to control its beam diameter and divergence angle. Depending on the sample shape and detection requirements, the disturbance light can be coaxially incident on the surface of the semiconductor sample 10 to be detected from the objective lens of the microscopic imaging module, or obliquely incident on the sample surface from below, side or above the sample stage 9. The imaging probe light can adopt a reflection, transmission, dark field, interference or polarization imaging path.

[0047] The beam combining and splitting optical path 5 guides the imaging probe light to the microscopic imaging module. The imaging probe light emitted from the probe light source 4 propagates along the outgoing optical path to the beam combining and splitting optical path 5. The mirrors and dichroic mirrors in the beam combining and splitting optical path 5 change the propagation direction of the imaging probe light, causing it to enter the incident optical path of the microscopic imaging module in a predetermined direction. The lens group in the beam combining and splitting optical path 5 collimates the imaging probe light, ensuring it enters the microscopic imaging module as a parallel beam. The beam splitter in the beam combining and splitting optical path 5 guides a portion of the imaging probe light beam to the microscopic imaging module, while the other portion is guided by the beam splitter to the power monitoring detector, which is used to monitor the output power of the imaging probe light in real time.

[0048] The imaging probe light is emitted from the microscopic imaging module and then onto the surface of the semiconductor sample 10 to be tested. The imaging probe light exits from the beam combining / splitting path 5 and enters the microscopic imaging module. The microscopic imaging module includes a low-magnification objective lens 6 and a high-magnification objective lens 7. The imaging probe light enters the low-magnification objective lens 6 or the high-magnification objective lens 7 via the optical path selected by the objective lens switching mechanism 8. The low-magnification objective lens 6 or the high-magnification objective lens 7 focuses the imaging probe light, and the focused imaging probe light exits from the exit facet of the low-magnification objective lens 6 or the high-magnification objective lens 7 and illuminates the surface of the semiconductor sample 10 to be tested. The imaging probe light forms a circular or elliptical illumination spot on the surface of the semiconductor sample 10 to be tested, and the illumination spot covers a local area of ​​the surface of the semiconductor sample 10 to be tested.

[0049] The imaging probe light undergoes reflection, transmission, scattering, interference, or polarization responses on the surface of the semiconductor sample 10 to form a probe light response. After the imaging probe light illuminates the surface of the semiconductor sample 10, a portion of the light undergoes specular reflection, propagating symmetrically with the incident light and carrying morphological and optical constant information about the surface. Another portion of the imaging probe light is transmitted into the interior of the semiconductor sample 10, undergoing transmission and refraction during its propagation within the sample. The transmitted light carries defect and structural information about the interior of the semiconductor sample 10. The imaging probe light is scattered at the microstructures on the surface and inside the semiconductor sample 10, propagating in various directions. The intensity and spatial distribution of the scattered light carry roughness information about the surface and particle information about the interior. The imaging probe light undergoes multiple reflections and transmissions between the multilayer thin film structures of the semiconductor sample 10, resulting in interference between the reflected and transmitted light. The interference light carries film thickness and interlayer interface information about the semiconductor sample 10. After passing through a polarization element, the imaging probe light becomes linearly polarized or circularly polarized. When this linearly or circularly polarized light illuminates the surface of the semiconductor sample 10 under test, its polarization state changes due to the optical anisotropy or birefringence of the sample. This polarization-state-changed imaging probe light forms a polarization response, which carries stress and crystal orientation information of the semiconductor sample 10. The reflected light, transmitted light, scattered light, interference light, and polarization response light together constitute the probe light response. This response exits from the surface and interior of the semiconductor sample 10 and propagates along a reverse path to the microscopic imaging module.

[0050] The beam combining and splitting optical path 5 also guides the probe light response to the image acquisition module 14. After exiting the surface of the semiconductor sample 10, the probe light response propagates along the reverse path of the incident light path into the microscopic imaging module. After being collected by the low-magnification objective lens 6 or the high-magnification objective lens 7, the probe light response propagates along the exit light path of the microscopic imaging module to the beam combining and splitting optical path 5. The dichroic mirror in the beam combining and splitting optical path 5 separates the probe light response from the incident imaging probe light according to the wavelength of the probe light response. The dichroic mirror causes the probe light response to reflect or transmit, thereby changing the propagation direction of the probe light response. The reflecting mirror in the beam combining and splitting optical path 5 further guides the probe light response to the incident end face of the image acquisition module 14. The lens group in the beam combining and splitting optical path 5 performs imaging processing on the probe light response, forming a clear microscopic image on the photosensitive surface of the image acquisition module 14. The beam splitter in the beam combining and splitting optical path 5 guides a portion of the probe light response beam to the image acquisition module 14, while the other portion of the probe light response beam is guided by the beam splitter to the eyepiece or observation camera for the operator to observe.

[0051] The wavelength of the imaging probe light differs from that of the perturbation light. The imaging probe light output from the probe light source 4 has a first wavelength band, while the perturbation light output from the periodic perturbation light output module has a second wavelength band; the first and second wavelength bands do not overlap. The imaging probe light is used to read the reflection or scattering image of the surface of the semiconductor sample 10 under test, while the perturbation light is used to induce changes in photoelectric properties inside or on the surface of the semiconductor sample 10 under test. The non-overlapping wavelengths of the imaging probe light and the perturbation light enable the dichroic mirror in the beam combining and splitting optical path 5 to separate the imaging probe light and probe light response from the perturbation light based on the wavelength difference, preventing the perturbation light from entering the image acquisition module 14 and simultaneously preventing the imaging probe light from interfering with the perturbation effect of the perturbation light on the sample.

[0052] The periodic perturbation light output module is used to output periodically modulated perturbation light to the semiconductor sample 10 under test. The perturbation light is used to induce photoelectric property changes in the defect region of the semiconductor sample 10 under test, specifically: The periodic perturbation light output module includes a periodic perturbation light source 2 and a modulator 3. The output end of the periodic perturbation light source 2 is connected to the input end of the modulator 3. The modulator 3 receives the light beam output by the periodic perturbation light source 2 and periodically modulates the light beam to output perturbation light. The perturbation light is irradiated onto the surface of the semiconductor sample 10 to be tested by the microscopic imaging module. The modulator 3 is an acousto-optic modulator, an electro-optic modulator, a mechanical chopper, or a function generator. The modulation waveform of the perturbation light is a sine wave, a square wave, a triangular wave, a pulse wave, or a pseudo-random sequence. The wavelength of the perturbation light is located in the ultraviolet band, visible band, near-infrared band, or mid-infrared band. The modulation frequency is generally from 0.001Hz to 150MHz. Other frequencies and custom waveforms are also applicable. The modulation frequency range is limited by the light source equipment and function generator used.

[0053] The modulation frequency of the perturbation light can be selected based on the dynamic timescale of the defects in the semiconductor sample 10 to be tested: a higher modulation frequency is selected for defects with rapid carrier capture and release; a lower modulation frequency is selected for thermal diffusion, deep level release, or surface adsorption-related responses. By scanning the sample multiple times at different modulation frequencies, the frequency response curve of each pixel position can be obtained, thereby distinguishing the defect types dominated by different physical mechanisms based on the characteristics of the frequency response.

[0054] The periodic perturbation source 2 is a light source that outputs a continuous or pulsed beam. It is fixedly mounted on the upper surface of an optical platform or rack. The exit face of the periodic perturbation source 2 faces the incident face of the modulator 3. Alternatively, since the modulator or function generator currently used can be multi-channel, the coaxial cable can be relatively long and does not need to be placed adjacent to the light source, as long as the perturbation light output by the modulator can be guided to the microscopic imaging module via the optical path and the reference signal can be transmitted to the synchronization control module. The periodic perturbation source 2 is connected to an external power supply via a power line. The external power supply provides operating current to the periodic perturbation source 2 to drive it to emit the beam. The periodic perturbation source 2 can be a continuous laser, a pulsed laser, a light-emitting diode, a supercontinuum source, or a broadband source. A continuous-wave laser includes a laser resonant cavity with a total reflection mirror and a partial reflection mirror at each end. The cavity is filled with a laser gain medium. A pump source excites the gain medium to produce stimulated emission, which oscillates within the cavity and outputs a continuous beam from the partial reflection mirror. A pulsed laser includes a pulse generation circuit or a Q-switching element. The pulse generation circuit controls the pump source to operate in a pulsed manner. The Q-switching element is located within the laser resonant cavity and periodically changes the cavity's quality factor, enabling the cavity to release a high-energy pulsed beam within a short time. A light-emitting diode (LED) is a light source that uses a semiconductor LED chip as the light emitter. The LED chip is fixedly mounted on a packaged substrate, which is connected to an external power supply via wires. When powered on, the LED chip emits a continuous-spectrum or narrow-band beam. The supercontinuum light source includes a mode-locked pulsed laser and a photonic crystal fiber. The mode-locked pulsed laser outputs an ultrashort pulse beam, which is coupled into the photonic crystal fiber. During propagation within the fiber, the ultrashort pulse beam undergoes nonlinear and dispersion effects, broadening its spectrum to a supercontinuum beam covering the ultraviolet to near-infrared bands. The broadband light source is a halogen tungsten lamp, a xenon lamp, or a light-emitting diode (LED) array. The filament of a halogen tungsten lamp emits light when energized, outputting a continuous spectrum covering the ultraviolet to infrared bands. A xenon lamp is a high-pressure gas discharge light source; energizing the two electrodes excites xenon gas discharge, emitting a wide-band continuous spectrum. An LED array is an array of multiple LED chips with different emission wavelengths; when multiple LED chips are energized simultaneously, they emit a composite broadband light covering multiple bands. The wavelength of the beam output by the periodically perturbed light source 2 is selected based on the band gap and defect energy levels of the semiconductor material being tested, inducing changes in photoelectric properties in the defect region.

[0055] Modulator 3 receives the beam output from the periodically disturbed light source 2 and periodically modulates the beam to output disturbed light. Modulator 3 is positioned in the output optical path of the periodically disturbed light source 2, with its incident end face facing the output end face of the periodically disturbed light source 2 and its output end face facing the incident optical path of the microscopic imaging module. Alternatively, since the modulators or function generators currently used can be multi-channel, the coaxial cable can be relatively long and does not need to be placed adjacent to the light source, as long as the disturbed light output by the modulator can be guided to the microscopic imaging module through the optical path and the reference signal can be transmitted to the synchronization control module. Modulator 3 is mounted on the upper surface of the optical platform or rack via a fixed bracket, which securely connects the housing of modulator 3 to the optical platform or rack. Modulator 3 is electrically connected to the synchronization control module 1 via a signal line. The synchronization control module 1 outputs a modulation drive signal to modulator 3, and modulator 3 periodically modulates the incident beam according to the modulation drive signal. The modulator 3 has a beam-passing region inside. The beam output from the periodically disturbed light source 2 enters the beam-passing region from the incident end face of the modulator 3. After being modulated in the beam-passing region, the beam is output from the exit end face of the modulator 3, and the output beam is the disturbed light. The modulation of the beam by the modulator 3 includes modulation of the beam intensity, modulation of the beam phase, or blocking and conducting modulation of the beam propagation direction.

[0056] The wavelength of the perturbation light is located in the ultraviolet, visible, near-infrared, or mid-infrared bands. The ultraviolet band wavelength ranges from 10 nm to 400 nm, the visible band from 400 nm to 780 nm, the near-infrared band from 780 nm to 2500 nm, and the mid-infrared band from 2500 nm to 25000 nm. The perturbation light is formed by modulating the beam output from the periodic perturbation source 2 after being modulated by the modulator 3, or by outputting periodic perturbation light after being modulated by the periodic perturbation source 2. Its wavelength falls within any of the four bands mentioned above, and can span multiple bands or combine multiple bands. The wavelength of the perturbation light is selected based on the bandgap width and defect energy level position of the semiconductor material to be tested. When the semiconductor material to be tested is a wide bandgap semiconductor, the wavelength of the perturbation light is selected in the ultraviolet band to provide photon energy higher than the material's bandgap, thereby exciting carrier generation in the wide bandgap semiconductor. When the semiconductor material to be detected is a narrow bandgap semiconductor, the wavelength of the perturbation light is selected in the near-infrared or mid-infrared band to provide photon energy below or close to the material's bandgap, causing the perturbation light to excite carrier transitions on defect state energy levels or generate localized thermal effects. The wavelength of the perturbation light is different from that of the imaging probe light so that the dichroic mirror in the beam combining and splitting optical path 5 can separate the perturbation light from the imaging probe light and the probe light response based on the wavelength difference.

[0057] Optically induced perturbation impedance imaging can be understood as follows: A periodic perturbation beam alters the local photoelectric properties of a sample, while another imaging probe beam reads the changes in optical responses related to reflection, transmission, scattering, interference, polarization, refractive index, dielectric constant, photothermal temperature, or carrier dynamics caused by the perturbation. Amplitude, phase, frequency response, and defect response maps are obtained through pixel-level synchronous demodulation. Specifically, the perturbation beam causes localized absorption, reflectivity, refractive index, dielectric constant, scattering intensity, photothermal temperature, carrier concentration, carrier mobility, carrier lifetime, local capacitance, or local impedance changes in the defect region of the semiconductor sample 10 under test. The wavelength, incident angle, polarization state, spot size, or optical path direction of the imaging probe beam differs from that of the perturbation beam to avoid crosstalk and improve the sensitivity of dynamic response reading. This term is not limited to electrochemical liquid interfaces or photocatalytic reaction processes in this invention. The core of this invention is not merely to acquire ordinary static images of the semiconductor sample 10 under illumination, nor does it rely on the signal generated by the excited light of the semiconductor sample 10. Instead, it applies periodic perturbation to the semiconductor sample 10 under periodic perturbation by periodic perturbation light, and reads the dynamic optical response generated by the semiconductor sample 10 under periodic perturbation by imaging probe light, and extracts the weak dynamic response signal with the same frequency as the perturbation light to identify defects.

[0058] The microscopic imaging module is used to image the probe light response to the image acquisition module 14. The imaging methods of the microscopic imaging module include at least one of bright-field reflection imaging, transmission imaging, dark-field imaging, interferometric reflection imaging, phase contrast imaging, polarization imaging, surface plasmon resonance imaging, total internal reflection imaging, or photothermal microscopy. Bright-field reflection imaging refers to an imaging method where the imaging probe light is reflected from the surface of the semiconductor sample 10 and enters the image acquisition module 14; the reflected light carries the morphological information and reflectivity distribution information of the sample surface. Transmission imaging refers to an imaging method where the imaging probe light passes through the semiconductor sample 10 and enters the image acquisition module 14; the transmitted light carries the absorption and structural information of the sample's interior. Dark-field imaging refers to an imaging method where the image acquisition module 14 only collects the light signal scattered from the semiconductor sample 10; the image acquisition module 14 does not collect specular reflection or transmitted light from the sample, and the scattered light carries the microstructural information of the sample. Interferometric reflection imaging refers to an imaging method where the imaging probe light interferes between the multilayer structures of the semiconductor sample 10 under test before entering the image acquisition module 14. The interfering light carries information about the film thickness and interlayer interfaces of the sample. Phase contrast imaging refers to an imaging method that converts the phase difference between the zero-order diffracted light and the higher-order diffracted light in the imaging probe light into a difference in light intensity. Phase contrast imaging is used to observe the phase information of transparent samples. Polarization imaging refers to an imaging method where the imaging probe light becomes linearly polarized or circularly polarized light after passing through a polarization element and illuminates the sample. The image acquisition module 14 then acquires the response of the probe light after the polarization state change. The polarization response carries information about the stress and crystal orientation of the sample. Surface plasmon resonance imaging refers to an imaging method where the imaging probe light excites surface plasmon resonance on the surface of a metal thin film, and the image acquisition module 14 acquires the change in the intensity of the reflected light under resonance conditions. Surface plasmon resonance imaging has a high sensitivity response to changes in the refractive index of the sample surface. Total internal reflection (TIV) imaging refers to an imaging method where the imaging probe light is incident on the sample surface at an angle greater than the critical angle, generating an evanescent wave on the sample surface. The image acquisition module 14 acquires the probe light response formed after the evanescent wave is scattered or absorbed by the sample. TIV imaging has a selective response to changes in the refractive index in the region near the sample surface. Photothermal microscopy refers to an imaging method that detects changes in optical properties caused by local temperature changes in the sample due to perturbation light. Photothermal microscopy has a high sensitivity response to weak absorption defects and thermal anomalies in the sample. Specifically: The microscopic imaging module includes a low-magnification objective lens 6, a high-magnification objective lens 7, an objective lens switching mechanism 8, and an autofocus module 13. The low-magnification objective lens 6 and the high-magnification objective lens 7 are fixedly mounted on the objective lens switching mechanism 8. The objective lens switching mechanism 8 drives the low-magnification objective lens 6 and the high-magnification objective lens 7 to switch so that the imaging probe light and disturbance light are irradiated onto the surface of the semiconductor sample 10 under test via the low-magnification objective lens 6 or the high-magnification objective lens 7. The probe light response is returned to the image acquisition module 14 via the low-magnification objective lens 6 or the high-magnification objective lens 7. The autofocus module 13 is connected to the objective lens switching mechanism 8 and controls the objective lens switching mechanism 8 or the sample support and environment control module based on the image sharpness feedback of the probe light response to adjust the relative distance between the semiconductor sample 10 under test and the low-magnification objective lens 6 or the high-magnification objective lens 7. In another possible implementation, the relative distance between the semiconductor sample 10 under test and the low-magnification objective lens 6 or the high-magnification objective lens 7 is determined by human eye or camera image recognition.

[0059] The low-magnification objective 6 is an optical lens group with low magnification and a large numerical aperture, used for large-area rapid scanning of wafers, chips, epitaxial wafers, or thin film samples. The low-magnification objective 6 has a cylindrical metal tube with external threads on its outer wall. Inside the tube, multiple optical lenses are arranged sequentially along the optical axis. These lenses are made of optical glass or optical crystal materials, and their surfaces are coated with anti-reflection coatings. The low-magnification objective 6 has a magnification of 1x to 20x, a numerical aperture of 0.02 to 0.50, and a working distance of 5mm to 50mm. The low-magnification objective 6 has a large object-side field of view, enabling it to cover a large area of ​​the surface of the semiconductor sample 10 under inspection in a single imaging operation. The low-magnification objective lens 6 has an entrance aperture and an exit aperture at the upper end of its tube. Both the entrance aperture and the exit aperture are circular through holes. The entrance aperture is located at the center of the upper end face of the tube, and the exit aperture is located at the center of the lower end face of the tube. Imaging probe light and disturbance light enter the tube of the low-magnification objective lens 6 through the entrance aperture. After being refracted and focused by multiple optical lenses, the imaging probe light and disturbance light exit through the exit aperture and illuminate the surface of the semiconductor sample 10 to be tested. The probe light response exits from the surface of the semiconductor sample 10 to be tested and enters the tube of the low-magnification objective lens 6 through the exit aperture. After being collected and imaged by multiple optical lenses, the probe light response exits through the entrance aperture and propagates to the image acquisition module 14.

[0060] The high-magnification objective 7 is an optical lens group with high magnification and a large numerical aperture, used for high spatial resolution re-examination of suspected defect areas obtained from low-magnification scanning. The high-magnification objective 7 has a cylindrical metal tube with external threads on its outer wall. Inside the tube, multiple optical lenses are arranged sequentially along the optical axis. These lenses are made of optical glass or optical crystal materials, and their surfaces are coated with anti-reflective coatings. The high-magnification objective 7 has a magnification of 20x to 150x, a numerical aperture of 0.40 to 1.49, and a working distance of 0.07 mm to 5 mm. The high-magnification objective 7 has a smaller object-side field of view diameter, enabling it to cover a small local area on the surface of the semiconductor sample 10 in a single image. The spatial resolution of the high-magnification objective 7 is higher than that of the low-magnification objective 6. The high-magnification objective lens 7 has an entrance aperture and an exit aperture at the upper end of its tube. Both the entrance aperture and the exit aperture are circular through holes. The entrance aperture is located at the center of the upper end face of the tube, and the exit aperture is located at the center of the lower end face of the tube. Imaging probe light and disturbance light enter the tube of the high-magnification objective lens 7 through the entrance aperture. After being refracted and focused by multiple optical lenses, the imaging probe light and disturbance light exit through the exit aperture and illuminate the surface of the semiconductor sample 10 to be tested. The probe light response exits from the surface of the semiconductor sample 10 to be tested and enters the tube of the high-magnification objective lens 7 through the exit aperture. After being collected and imaged by multiple optical lenses, the probe light response exits through the entrance aperture and propagates to the image acquisition module 14. After high-magnification re-measurement, the fine location, size, response phase, and frequency response characteristics of the defect are obtained.

[0061] The low-magnification objective lens 6 and the high-magnification objective lens 7 are fixedly mounted on the objective lens switching mechanism 8. The objective lens switching mechanism 8 is either a rotary table type or a sliding table type. The rotary table type switching mechanism includes a rotary table, a rotating shaft, and a drive motor. The rotary table has a circular disc structure made of metal, and its upper surface is flat. Multiple mounting holes are evenly distributed at equal angles along the circumference of the rotary table on its upper surface. The external thread of the low-magnification objective lens 6 is screwed into the internal thread of the first mounting hole, and the low-magnification objective lens 6 is fixedly connected to the rotary table via these threads. The external thread of the high-magnification objective lens 7 is screwed into the internal thread of the second mounting hole, and the high-magnification objective lens 7 is fixedly connected to the rotary table via these threads. The center of the rotary table is fixedly connected to the upper end of the rotating shaft, which extends vertically. The lower end of the rotating shaft is fixedly connected to the output shaft of the drive motor via a coupling. The housing of the drive motor is mounted on the upper surface of the optical platform or frame via a fixed bracket, and the drive motor is electrically connected to the synchronous control module 1 via wires. The sliding switching mechanism includes a sliding table, a slide rail, and a drive motor. The sliding table is a rectangular plate structure made of metal, with a flat upper surface and multiple mounting holes evenly spaced along its length. The external thread of the low-magnification objective lens 6 is screwed into the internal thread of the first mounting hole, securing it to the sliding table. The external thread of the high-magnification objective lens 7 is screwed into the internal thread of the second mounting hole, securing it to the sliding table. The lower surface of the sliding table slides against the upper surface of the slide rail, which is fixedly mounted on the upper surface of the optical platform or frame and extends horizontally. The drive motor housing is fixedly mounted to one end of the slide rail, and its output shaft is fixedly connected to one end of a lead screw. The lead screw extends along the slide rail and passes through a nut seat on the lower surface of the sliding table, threadedly engaging with it. The drive motor is electrically connected to the synchronization control module 1 via a wire, but the process can also be achieved manually.

[0062] The objective lens switching mechanism 8 drives the low-magnification objective lens 6 and the high-magnification objective lens 7 to switch so that the imaging probe light and the disturbance light are irradiated onto the surface of the semiconductor sample 10 under test through the low-magnification objective lens 6 or the high-magnification objective lens 7. The synchronous control module 1 outputs a switching control signal to the drive motor of the objective lens switching mechanism 8. The drive motor generates driving force according to the switching control signal. This process can also be achieved manually. When the objective lens switching mechanism 8 is a turntable switching mechanism, the drive motor drives the rotating shaft to rotate, and the rotating shaft drives the turntable to rotate around the vertical axis. The low-magnification objective lens 6 and the high-magnification objective lens 7, which are fixedly mounted on the turntable, rotate synchronously with the turntable. When the low-magnification objective lens 6 rotates to be directly below the common optical path of the imaging probe light and the disturbance light, the imaging probe light and the disturbance light propagate downward along the common optical path and enter the entrance aperture of the low-magnification objective lens 6. After being focused by the low-magnification objective lens 6, the imaging probe light and the disturbance light exit from the exit aperture of the low-magnification objective lens 6 and irradiate the surface of the semiconductor sample 10 under test. When the high-magnification objective lens 7 rotates to be directly below the common optical path of the imaging probe light and the disturbance light, the imaging probe light and the disturbance light propagate downward along the common optical path and enter the entrance aperture of the high-magnification objective lens 7. After being focused by the high-magnification objective lens 7, the imaging probe light and the disturbance light exit from the exit aperture of the high-magnification objective lens 7 and illuminate the surface of the semiconductor sample 10 to be tested. When the objective lens switching mechanism 8 is a sliding stage switching mechanism, the drive motor drives the lead screw to rotate, and the lead screw drives the slide to move horizontally along the length of the slide rail through the nut seat. The low-magnification objective lens 6 and the high-magnification objective lens 7, which are fixedly mounted on the slide, move horizontally synchronously with the slide. When the low-magnification objective lens 6 moves to be directly below the common optical path of the imaging probe light and the disturbance light, the imaging probe light and the disturbance light propagate downward along the common optical path and enter the entrance aperture of the low-magnification objective lens 6. After being focused by the low-magnification objective lens 6, the imaging probe light and the disturbance light exit from the exit aperture of the low-magnification objective lens 6 and illuminate the surface of the semiconductor sample 10 to be tested. When the high-power objective lens 7 moves directly below the common optical path of the imaging probe light and the disturbance light, the imaging probe light and the disturbance light propagate downward along the common optical path and enter the entrance aperture of the high-power objective lens 7. After being focused by the high-power objective lens 7, the imaging probe light and the disturbance light exit from the exit aperture of the high-power objective lens 7 and illuminate the surface of the semiconductor sample 10 to be tested.

[0063] The probe light response returns to the image acquisition module 14 via the low-magnification objective lens 6 or the high-magnification objective lens 7. After being emitted from the surface of the semiconductor sample 10, the probe light response propagates upward along the common optical path, entering the low-magnification objective lens 6 through its exit aperture or the high-magnification objective lens 7 through its exit aperture. After being collected and imaged by the imaging lens group inside the low-magnification objective lens 6 or the high-magnification objective lens 7, the probe light response exits through the entrance aperture of either lens. The emitted probe light response continues to propagate upward along the common optical path, and after being guided by the beam combining and splitting optical path 5, it enters the incident end face of the image acquisition module 14. The photosensitive surface of the image acquisition module 14 receives the probe light response and forms a microscopic image.

[0064] In practical testing scenarios, the imaging probe light and the perturbation light can also be incident in opposite directions. For example, the imaging probe light is incident perpendicularly from above the sample through the objective lens onto the surface of the semiconductor sample 10 to be tested, while the perturbation light is incident from below, to the side, or above the sample stage 9 at a certain angle onto the same detection area of ​​the sample. The two beams overlap on or near the sample surface, and the imaging probe light reads the optical response caused by the photoelectric property change induced by the perturbation light. In this case, there is no need to couple the perturbation light into the common optical path of the microscopic imaging module. The beam combining and splitting optical path 5 is only responsible for guiding the imaging probe light to the microscopic imaging module and guiding the probe light response to the image acquisition module 14. The perturbation light is directly irradiated onto the sample through an independent optical path system, and its spatial overlap with the imaging probe light is achieved by aligning the light spots on the sample surface. Accordingly, the installation position and emission direction of the periodic perturbation light output module are adjusted according to the incident angle of the perturbation light. It is no longer limited to the incident light path with the emission end face facing the microscopic imaging module. As long as the perturbation light can illuminate the detection area of ​​the sample through an independent light path and spatially overlap with the imaging detection light on the sample surface, it is acceptable.

[0065] It should also be noted that the arrangement of the objectives is not limited to the upright mode (where the objectives are above the sample). In the inverted objective mode, the microscopic imaging module is positioned below the sample stage 9. The low-magnification objective 6 and the high-magnification objective 7 are positioned below the semiconductor sample 10 to be tested and facing upwards toward the lower surface of the sample. The imaging probe light is emitted from below the sample through the objective and then onto the lower surface of the sample. The probe light response is also collected by the lower objective and returned to the image acquisition module 14. The perturbation light can be incident on the sample from above, below, or to the side, which can also realize defect detection by light-induced perturbation impedance imaging. The inverted objective mode is particularly suitable for detection scenarios in a liquid environment. In this case, the objective is immersed in the liquid medium or images are taken from below the sample through a transparent window, which avoids the risk of direct contact between the objective and the liquid medium. At the same time, it is convenient to arrange the perturbation light source and liquid environment control components above the sample. The upright mode and the inverted mode can be selected according to the shape of the semiconductor sample 10 to be tested, the detection environment, and the ease of operation. Both modes can realize the defect detection function of this invention.

[0066] The autofocus module 13 is connected to the objective lens switching mechanism 8 and controls the objective lens switching mechanism 8 or the sample support and environment control module based on the image sharpness feedback of the probe light response to adjust the relative distance between the semiconductor sample 10 to be tested and the low-magnification objective lens 6 or the high-magnification objective lens 7. The autofocus module 13 is an image processing-based autofocus unit, including an image analysis unit and a drive control unit. The image analysis unit is electrically connected to the image acquisition module 14 via a signal line. The image analysis unit receives the microscopic image signal output by the image acquisition module 14, which is digital image data. The image analysis unit performs sharpness evaluation calculations on the digital image data. The sharpness evaluation calculation method is an edge detection algorithm or a contrast detection algorithm. The edge detection algorithm calculates the gray-level gradient values ​​between adjacent pixels in the digital image. The sum of the squares or absolute values ​​of all pixel gray-level gradient values ​​is used as the sharpness evaluation value; a higher sharpness evaluation value indicates a sharper image. The contrast detection algorithm calculates the difference between the maximum and minimum values ​​of pixel gray-level values ​​or the standard deviation of pixel gray-level values ​​in the digital image as the sharpness evaluation value; a higher sharpness evaluation value indicates a sharper image. The drive control unit is electrically connected to the synchronization control module 1 via a signal line. The drive control unit is also electrically connected to the drive motor of the objective lens switching mechanism 8 via the synchronization control module 1, and to the motorized stage 12 of the sample support and environment control module via the synchronization control module 1. The image analysis unit compares the calculated sharpness evaluation value with a preset sharpness threshold. When the sharpness evaluation value is lower than the preset sharpness threshold, the image analysis unit outputs a defocus signal to the drive control unit. The drive control unit generates a focus control command based on the defocus signal. The focus control command is output to the drive motor of the objective lens switching mechanism 8 via the synchronization control module 1 or to the motorized stage 12 of the sample support and environment control module. When the focus control command is output to the drive motor of the objective lens switching mechanism 8, the drive motor drives the turntable or slide of the objective lens switching mechanism 8 to move vertically. The turntable or slide drives the low-magnification objective lens 6 or high-magnification objective lens 7, which is fixedly mounted thereon, to move vertically, thereby changing the relative distance between the exit aperture of the low-magnification objective lens 6 or high-magnification objective lens 7 and the surface of the semiconductor sample 10 to be tested. When the focusing control command is output to the electric displacement stage 12 of the sample carrier and environmental control module, the vertical motion mechanism of the electric displacement stage 12 drives the sample carrier stage 9 to move in the vertical direction. The sample carrier stage 9 drives the semiconductor sample 10 to be tested placed on it to move in the vertical direction, thereby changing the relative distance between the surface of the semiconductor sample 10 to be tested and the exit aperture of the low-magnification objective lens 6 or the high-magnification objective lens 7.After the objective lens switching mechanism 8 completes the switching between the low-power objective lens 6 and the high-power objective lens 7, the autofocus module 13 performs an autofocus operation. Through the above-mentioned feedback control process, the autofocus module 13 ensures that the surface of the semiconductor sample 10 to be tested is always located on the focal plane of the low-power objective lens 6 or the high-power objective lens 7, thereby ensuring that the probe light response image acquired by the image acquisition module 14 remains clear.

[0067] Besides the statement that "autofocus module 13 is an image processing-based autofocus unit, including an image analysis unit and a drive control unit," autofocus module 13 can also be implemented using an active focus locking method. For example, a near-infrared light source is provided, and the emitted near-infrared light is guided by an optical path and projected onto the surface of the semiconductor sample 10 to be tested or its reference interface, forming a reference light spot on the sample surface; simultaneously, a linear array CCD sensor or a position-sensitive detector is provided to continuously monitor the position or intensity changes of the near-infrared light signal reflected from the sample surface. When the sample surface experiences focal shift due to thermal drift, mechanical vibration, or the stepping motion of the electric displacement stage 12, the imaging position or spot shape of the reflected light signal on the linear CCD sensor changes accordingly. The autofocus module 13 calculates the defocus direction and amount in real time based on this change and outputs compensation control signals to the drive motor of the objective lens switching mechanism 8 or the sample support and environment control module with a millisecond-level response speed. This drives the low-magnification objective lens 6 or the high-magnification objective lens 7 to move vertically, or drives the sample support stage 9 to move vertically, dynamically compensating for the focal shift and ensuring that the focus remains locked at the target position throughout the long scanning and detection process. Compared to passive autofocus methods based on image sharpness evaluation, this active focus locking method does not rely on the sharpness calculation of the probe light response image. It can independently monitor and compensate for the focus during image acquisition, making it particularly suitable for scenarios where the sample surface height changes frequently or where there is vibration in the detection environment.

[0068] The autofocus module 13 can also adjust the relative distance between the semiconductor sample 10 to be tested and the low-magnification objective lens 6 or the high-magnification objective lens 7 based on feedback of reflection intensity, interference contrast or demodulation amplitude.

[0069] Image acquisition module 14 is used to acquire time-series optical images of the probe light response during the period when the perturbation light applies a perturbation to the semiconductor sample 10 to be tested, specifically: The image acquisition module 14 is an area array camera, a line array camera, an avalanche photodiode array, a photomultiplier tube array, an infrared camera, or a high-speed camera. The image acquisition module 14 is connected to the synchronization control module 1 and receives the trigger signal output by the synchronization control module 1. The image acquisition module 14 continuously acquires time-series optical images of the probe light response at a preset frame rate throughout the entire process of the perturbation light applying perturbation to the semiconductor sample 10 to be tested. The preset frame rate is at least twice the modulation frequency of the perturbation light.

[0070] The image acquisition module 14 is connected to the synchronization control module 1 and receives the trigger signal output by the synchronization control module 1. The trigger signal input interface of the image acquisition module 14 is electrically connected to the trigger signal output interface of the synchronization control module 1 via a signal line. The synchronization control module 1 internally generates a trigger signal, which is either a digital level pulse signal or a transistor-to-transistor logic level pulse signal. The high-level voltage of the digital level pulse signal is 3.3 volts or 5 volts, and the low-level voltage is 0 volts. The high-level voltage of the transistor-to-transistor logic level pulse signal is 5 volts, and the low-level voltage is 0 volts. The pulse width of the trigger signal is on the order of microseconds to milliseconds (less than 1 ms in high-speed synchronization applications, with a rise time less than 1 μs or even less than 1 ns), and the rise time of the trigger signal is less than one-tenth of the pulse width of the trigger signal. The synchronous control module 1 transmits the trigger signal to the trigger signal input interface of the image acquisition module 14 via the signal line. After the controller inside the image acquisition module 14 detects the rising edge or falling edge of the trigger signal, the controller starts the exposure operation of the image acquisition module 14 according to the detected edge signal.

[0071] The image acquisition module 14 continuously acquires time-series optical images of the probe light response at a preset frame rate throughout the entire process of applying perturbation to the semiconductor sample 10 under test by the perturbation light, based on the trigger signal. The synchronization control module 1 starts outputting a trigger signal sequence simultaneously with activating the periodic perturbation light output module. The trigger signal sequence is a pulse sequence composed of multiple trigger pulses arranged at fixed time intervals, with each trigger pulse corresponding to one image acquisition. The image acquisition module 14 begins exposure and acquisition of the first frame image upon receiving the first trigger pulse, and begins exposure and acquisition of the second frame image upon receiving the second trigger pulse. The image acquisition module 14 continuously acquires one frame image upon each trigger pulse in the above manner until the synchronization control module 1 stops outputting the trigger signal sequence. The time-series optical image is a data set of multiple frames arranged in chronological order of acquisition time. Each frame image corresponds to an acquisition moment, and all frames together constitute the time-series optical image. Each frame image in the time-series optical image is either a two-dimensional grayscale image or a two-dimensional color image. In the two-dimensional grayscale image, the value of each pixel represents the light intensity at that pixel location. In the two-dimensional color image, each pixel contains red channel values, green channel values, and blue channel values.

[0072] The preset frame rate is at least twice the modulation frequency of the perturbation light. The preset frame rate is the number of image frames acquired per second by the image acquisition module 14, measured in frames per second. The modulation frequency of the perturbation light is the number of complete cycles of the periodic change in the intensity of the perturbation light per second, measured in Hertz. The relationship between the preset frame rate and the modulation frequency of the perturbation light satisfies that the preset frame rate is greater than or equal to the modulation frequency of the perturbation light multiplied by 2. When the preset frame rate is equal to the modulation frequency of the perturbation light multiplied by 2, the image acquisition module 14 acquires two images within one perturbation light modulation cycle, with the time interval between the two images being half of the modulation cycle. When the preset frame rate is greater than the modulation frequency of the perturbation light multiplied by 2, the image acquisition module 14 acquires more than two images within one perturbation light modulation cycle, with the time interval between two adjacent images being less than half of the modulation cycle. Preferably, the preset frame rate can be set to 4 times, 8 times, or a higher multiple of the modulation frequency of the perturbation light to improve the accuracy of phase and amplitude extraction. For perturbation light with a low modulation frequency, the image acquisition module 14 can continuously acquire time-series optical images within multiple modulation cycles and improve the signal-to-noise ratio through multi-cycle averaging. For perturbation light with a high modulation frequency, the image acquisition module 14 can use external trigger synchronization, exposure gating, or hardware phase-locked loop to acquire images synchronously, ensuring that accurate time-series images can still be obtained under rapid modulation conditions.

[0073] The specific value of the preset frame rate is determined by the synchronization control module 1 based on the preset modulation frequency value. When generating the trigger signal sequence, the synchronization control module 1 sets the time interval of the trigger pulses to the reciprocal of the preset frame rate, so that the time interval between two adjacent trigger pulses is equal to the reciprocal of the preset frame rate. The image acquisition module 14 continuously acquires time-series optical images according to the above-mentioned time interval of the trigger pulses, so that the phase position of each frame of the acquired image is evenly distributed within the modulation period of the perturbation light, thereby ensuring that the defect judgment module 15 can accurately extract the signal component with the same frequency as the perturbation light from the time-series optical images.

[0074] In addition, a bandwidth-saving synchronous acquisition scheme can be adopted. In this scheme, the synchronization control module 1 receives multi-channel signals (including the camera's frame synchronization signal, the laser's pulse synchronization signal, the modulator's reference signal, etc.) from the image acquisition module 14, the periodic perturbation light source 2, and the modulator 3, and synchronously samples each channel signal at a sampling rate of twice or more than the highest frequency among the multi-channel signals (in actual operation, the default sampling rate is 100 times the highest frequency to ensure accurate reconstruction of phase information). Based on the phase relationship of each channel signal obtained from the sampling, the synchronization control module 1 dynamically calculates and adjusts the output time and phase offset of the trigger signal, so that the exposure start time of the image acquisition module 14 is precisely aligned with the specific phase (such as the peak phase or zero-crossing phase) of the perturbation light. This scheme does not require the image acquisition module 14 to continuously acquire all time-series images at a fixed high frame rate. Instead, it triggers the image acquisition module 14 to selectively expose within the key phase window based on the real-time phase information of each channel signal. This significantly reduces the amount of image data while ensuring phase alignment accuracy, reducing data transmission bandwidth and storage pressure, making it particularly suitable for long-term continuous monitoring or large-area scanning detection scenarios. In the specific implementation, the synchronization control module 1 performs digital phase-locked loop processing on the channel signals obtained by sampling, calculates the real-time phase difference between the disturbance light modulation signal and the 14-frame synchronization signal of the image acquisition module, and dynamically adjusts the delay time of the next frame trigger according to the phase difference, so that each subsequent frame of image acquisition maintains a fixed phase relationship with the preset phase of the disturbance light, ensuring that the defect judgment module 15 can still accurately extract the signal component with the same frequency as the disturbance light under a limited number of image frames.

[0075] Synchronization control module 1 is used to control the synchronization of the periodic perturbation light output module and the image acquisition module 14, specifically: The synchronization control module 1 is electrically connected to the periodic perturbation light output module, the image acquisition module 14, the motorized stage 12, the objective lens switching mechanism 8, and the autofocus module 13. The synchronization control module 1 receives a reference signal output from the modulator 3. The frequency and phase of the reference signal are consistent with the modulation frequency and phase of the perturbation light. The synchronization control module 1 generates a trigger signal based on the reference signal and outputs the trigger signal to the image acquisition module 14. Upon receiving the trigger signal, the image acquisition module 14 begins acquiring time-series optical images. The synchronization control module 1 also outputs a displacement control signal to the motorized stage 12 according to a preset scanning path. The moving stage 12 drives the sample carrier stage 9 to move stepwise in the horizontal direction according to the displacement control signal. After the moving stage 12 completes each step, the synchronous control module 1 triggers the image acquisition module 14 to perform an image acquisition. The synchronous control module 1 also outputs a switching control signal to the objective lens switching mechanism 8. The objective lens switching mechanism 8 drives the low-magnification objective lens 6 and the high-magnification objective lens 7 to switch according to the switching control signal. After the objective lens switching mechanism 8 completes the switching, the synchronous control module 1 triggers the autofocus module 13 to perform autofocus. After the autofocus module 13 completes focusing, the synchronous control module 1 triggers the image acquisition module 14 to start acquiring time-series optical images.

[0076] The synchronization control module 1 is electrically connected to the periodic perturbation light output module, image acquisition module 14, motorized stage 12, objective lens switching mechanism 8, and autofocus module 13. The synchronization control module 1 is an electronic controller unit, internally comprising a central processing unit (CPU), a clock generator, input / output interfaces, and memory. The CPU is a microcontroller chip or a digital signal processor chip, fixedly mounted on a circuit board inside the synchronization control module 1. The CPU is electrically connected to the clock generator, input / output interfaces, and memory via printed circuit traces on the circuit board. The clock generator is a crystal oscillator or a phase-locked loop clock synthesizer, generating a high-precision reference clock signal with a frequency stability on the order of one part per million. This reference clock signal is input to the clock input terminal of the CPU. The input / output interfaces include digital input interfaces and digital output interfaces. The digital input interfaces are general-purpose input / output interfaces or serial peripheral interfaces, and the digital output interfaces are general-purpose input / output interfaces or serial peripheral interfaces. The memory is a random access memory (RAM) and a non-volatile flash memory (NSF). The RAM is used to temporarily store running programs and data, while the NSF stores preset scan paths and configuration parameters. The housing of the synchronization control module 1 is a metal shielded box, which is fixedly installed inside the rack or control cabinet. The front panel of the metal shielded box is equipped with multiple signal interface sockets, which are electrically connected to the periodic disturbance light output module, the image acquisition module 14, the electric displacement stage 12, the objective lens switching mechanism 8, and the autofocus module 13 via signal lines.

[0077] Synchronization control module 1 receives a reference signal output from modulator 3. The frequency and phase of the reference signal are consistent with the modulation frequency and phase of the perturbed light. Modulator 3 can be an acousto-optic modulator, electro-optic modulator, mechanical chopper, or function generator. When modulator 3 is an acousto-optic modulator, the radio frequency (RF) driver generates an RF signal. The frequency and phase of the RF signal determine the frequency and phase of the ultrasonic field inside the acousto-optic crystal. The ultrasonic field causes the intensity of the diffracted light to change periodically according to the frequency and phase of the RF signal. While outputting the RF signal, the RF driver also outputs an electrical signal with the same frequency and phase as the RF signal as a reference signal. This reference signal is output to the digital input interface of synchronization control module 1 through a signal line. When modulator 3 is an electro-optic modulator, the electro-optic drive source generates a modulation voltage signal. The frequency and phase of the modulation voltage signal determine the frequency and phase of the electric field inside the electro-optic crystal. The electric field causes the intensity of the light beam passing through the electro-optic crystal to change periodically according to the frequency and phase of the modulation voltage signal. Simultaneously with outputting the modulation voltage signal, the electro-optic drive source outputs an electrical signal with the same frequency and phase as the modulation voltage signal as a reference signal. This reference signal is output to the digital input interface of the synchronization control module 1 via a signal line. When modulator 3 is a mechanical chopper, the motor driver controls a rotary motor to rotate at a predetermined speed. The rotary motor drives the chopper disk to rotate. The rotation of the chopper disk causes the light beam to be periodically blocked and opened. The frequency of the light beam intensity change is equal to the product of the chopper disk's rotational speed and the number of fan-shaped light-transmitting apertures. Simultaneously with outputting the drive signal, the motor driver outputs an electrical signal with a frequency equal to the product of the chopper disk's rotational frequency and the number of fan-shaped light-transmitting apertures as a reference signal. This reference signal is output to the digital input interface of the synchronization control module 1 via a signal line. When modulator 3 is a function generator, it generates a periodic electrical signal waveform. The frequency of this waveform is the modulation frequency of the perturbation light, and the phase is the modulation phase. Simultaneously, the function generator outputs the periodic electrical signal waveform to the driving circuit of the periodic perturbation light source 2, and also outputs an electrical signal with the same frequency and phase as the periodic electrical signal waveform as a reference signal. This reference signal is output to the digital input interface of the synchronization control module 1 via a signal line. The central processing unit (CPU) of the synchronization control module 1 receives the reference signal through the digital input interface. The CPU detects the rising and falling edges of the reference signal, calculates the frequency of the reference signal based on the time interval between adjacent effective edges (rising or falling edges), and determines the phase of the reference signal based on its instantaneous voltage value.

[0078] The synchronization control module 1 generates a trigger signal based on the reference signal and outputs the trigger signal to the image acquisition module 14. After receiving the reference signal, the central processing unit (CPU) of the synchronization control module 1 inputs the reference signal to the synchronization input terminal of the clock generator. The clock generator regenerates its internal reference clock based on the reference signal, locking the frequency and phase of the internal reference clock with the frequency and phase of the reference signal. The CPU generates the trigger signal based on the internal reference clock. The trigger signal is a digital level pulse sequence, with the rising edge of each pulse aligned with a predetermined edge of the internal reference clock, ensuring a constant phase difference between the trigger signal and the reference signal. The frequency of the trigger signal is set by the CPU according to a preset frame rate, which is at least twice the modulation frequency of the perturbation light. The CPU obtains the trigger signal frequency by dividing the frequency of the internal reference clock by a set division coefficient using a frequency divider circuit. The CPU outputs the generated trigger signal to the trigger signal input interface of the image acquisition module 14 through a digital output interface. The trigger signal input interface of the image acquisition module 14 is electrically connected to the digital output interface of the synchronization control module 1 via a signal line, and the trigger signal is transmitted to the controller inside the image acquisition module 14 via the signal line.

[0079] Upon receiving a trigger signal, the image acquisition module 14 begins acquiring time-series optical images. The controller inside the image acquisition module 14 continuously monitors the level of the trigger signal input interface. When the controller detects a valid edge (rising or falling edge) of the trigger signal, it initiates the exposure timing sequence of the image acquisition module 14. The exposure timing sequence includes three consecutive stages: shutter opening, charge accumulation, and charge readout. The shutter opening stage lasts for microseconds, during which the probe light response enters the photosensitive surface of the image sensor chip after the shutter opens. The charge accumulation stage lasts for a preset exposure time, which is set by the synchronization control module 1 through an independent exposure control signal. During the charge accumulation stage, each photosensitive unit of the image sensor chip converts the received light signal into accumulated charge. After the charge accumulation stage ends, the controller enters the charge readout stage, sequentially reading out the accumulated charge of each photosensitive unit and converting it into a digital image signal. After acquiring one frame of image, the controller waits for the next rising edge of the trigger signal. When the controller detects the next rising edge of the trigger signal, it repeats the above exposure timing sequence to acquire the next frame of image. The image acquisition module 14 continuously acquires time-series optical images in the manner described above. Each frame of the time-series optical image corresponds to a unique trigger pulse in the trigger signal sequence, and the acquisition start time of each frame is consistent with the rising edge time of the corresponding trigger pulse.

[0080] The synchronous control module 1 also outputs a displacement control signal to the electric displacement stage 12 according to a preset scanning path. The preset scanning path is a coordinate sequence stored in the non-volatile flash memory of the synchronous control module 1, where each coordinate value is the X-axis and Y-axis position coordinates in the horizontal direction. The central processing unit (CPU) reads the coordinate sequence of the preset scanning path from the non-volatile flash memory. The CPU compares the coordinate value corresponding to the current acquisition position with the next coordinate value in the coordinate sequence to calculate the position deviation in the X-axis and Y-axis directions. The CPU generates a displacement control signal based on the X-axis and Y-axis position deviations. The displacement control signal is either a pulse direction signal or a digital communication protocol data frame. When the displacement control signal is a pulse direction signal, it includes two signals: the first signal is a pulse signal, where each pulse corresponds to a fixed angle rotation of the servo motor in the horizontal motion mechanism of the electric displacement stage 12; the second signal is a direction signal, where a high level indicates forward movement and a low level indicates reverse movement. When the displacement control signal is a digital communication protocol data frame, the digital communication protocol data frame includes the target position coordinates and the motion velocity. The target position coordinates and the motion velocity are represented in digital code form. The central processing unit outputs the displacement control signal to the control signal input interface of the electric displacement stage 12 through the digital output interface.

[0081] It should be noted that the motorized stage 12 also retains a manual operation mode in actual operation to meet the needs of flexible positioning and manual re-inspection. In this mode, the operator can directly control the movement of the motorized stage 12 in the horizontal and vertical directions through the manual control interface or physical joystick, without relying on the preset scanning path output by the synchronous control module 1. Specifically, the operator can manually input the approximate XY coordinates of the known contamination or suspected defects on the wafer based on existing wafer inspection records, and the motorized stage 12 will drive the sample carrier stage 9 to move to the vicinity of the target area according to the coordinate values; alternatively, the operator can also manually adjust the vertical position of the motorized stage 12 to complete the focus by visually observing the real-time imaging screen under the low-magnification objective lens 6, and identify the area of ​​interest by eye under the low-magnification field of view, and then manually move the operating lever of the motorized stage 12 to move the sample carrier stage 9 to that area, and then switch to the high-magnification objective lens 7 through the objective lens switching mechanism 8 for fine observation and re-inspection. The manual operation mode and the automatic scanning mode controlled by the synchronous control module 1 can be switched and complement each other: the automatic scanning mode is suitable for large-area, standardized batch inspection processes, while the manual operation mode is suitable for rapid positioning of specific coordinate points, manual verification of suspected defects, and flexible inspection of non-standardized samples. In manual operation mode, the servo motor driver of the electric displacement stage 12 remains connected to the synchronous control module 1 via a signal line. The synchronous control module 1 reads the current position coordinates of the electric displacement stage 12 in real time and feeds them back to the manual control interface to assist the operator in positioning. When the operator switches back to automatic scanning mode, the synchronous control module 1 can continue to execute the remaining part of the preset scanning path from the currently manually positioned position without having to restart the scan.

[0082] The electric displacement stage 12 drives the sample support stage 9 to move stepwise in the horizontal direction according to the displacement control signal. The control signal input interface of the electric displacement stage 12 receives the displacement control signal output by the synchronous control module 1. When the displacement control signal is a pulse direction signal, the servo motor driver of the horizontal motion mechanism of the electric displacement stage 12 receives the pulse signal and the direction signal. The servo motor driver determines the rotation direction of the servo motor according to the direction signal, and determines the rotation angle of the servo motor according to the number of pulses of the pulse signal. The servo motor drives the lead screw to rotate according to the determined rotation direction and rotation angle. The lead screw drives the nut to move in the horizontal direction. The nut drives the motion platform of the horizontal motion mechanism to move in the horizontal direction. The motion platform drives the sample support stage 9 to move in the horizontal direction through the vertical motion mechanism. When the displacement control signal is a digital communication protocol data frame, the servo motor driver of the horizontal motion mechanism of the electric displacement stage 12 receives the digital communication protocol data frame. The servo motor driver parses the target position coordinates and motion speed values ​​from the digital communication protocol data frame. The servo motor driver calculates the position deviation based on the target position coordinates and the current actual position coordinates. The servo motor driver generates rotation direction and rotation angle control signals for the servo motor based on the position deviation. The servo motor drives the lead screw to rotate according to the control signals. The lead screw drives the nut to move horizontally. The nut drives the motion platform of the horizontal motion mechanism to move horizontally to the position corresponding to the target position coordinates. The motion platform drives the sample carrier 9 to move to the position corresponding to the target position coordinates through the vertical motion mechanism. After the electric displacement stage 12 completes each step of movement, the motion platform of the horizontal motion mechanism reaches a coordinate position in the preset scanning path, and the sample carrier 9 follows the motion platform to the corresponding coordinate position.

[0083] After the electric displacement stage 12 completes each step of movement, the synchronization control module 1 triggers the image acquisition module 14 to perform an image acquisition. After completing the displacement action corresponding to the displacement control signal, the servo motor driver of the horizontal motion mechanism of the electric displacement stage 12 outputs a position signal to the digital input interface of the synchronization control module 1 via a signal line. The position signal is a digital level pulse signal, and the effective edge (rising or falling edge) of the position signal indicates that the electric displacement stage 12 has completed the current step movement. After the central processing unit of the synchronization control module 1 detects the effective edge (rising or falling edge) of the position signal, it generates a trigger signal based on a reference signal and outputs it to the image acquisition module 14 via a digital output interface. Upon receiving the trigger signal, the image acquisition module 14 performs an image acquisition, acquiring a time-series optical image at the current coordinate position. After completing the time-series optical image acquisition at the current coordinate position, the synchronization control module 1 reads the next coordinate value from the preset scanning path and generates the next displacement control signal. The electric displacement stage 12 drives the sample support stage 9 to move to the next coordinate position according to the next displacement control signal. After the electric displacement stage 12 completes the next displacement action, the synchronization control module 1 triggers the image acquisition module 14 to perform another image acquisition. The synchronization control module 1 and the electric displacement stage 12 work together in this manner until all coordinate positions in the preset scanning path have been scanned.

[0084] The synchronization control module 1 also outputs a switching control signal to the objective lens switching mechanism 8. The drive motor of the objective lens switching mechanism 8 is electrically connected to the digital output interface of the synchronization control module 1 via a signal line. The central processing unit of the synchronization control module 1 generates the switching control signal according to a preset scanning procedure. The switching control signal is either a digital level signal or a digital communication protocol data frame. When the switching control signal is a digital level signal, a high level indicates switching to the low-magnification objective lens 6, and a low level indicates switching to the high-magnification objective lens 7. When the switching control signal is a digital communication protocol data frame, the digital communication protocol data frame includes the target objective lens number, which is represented in digital encoding form. The central processing unit outputs the switching control signal to the control signal input interface of the drive motor of the objective lens switching mechanism 8 via the digital output interface.

[0085] The objective lens switching mechanism 8 drives the low-magnification objective lens 6 and the high-magnification objective lens 7 to switch according to the switching control signal. When the objective lens switching mechanism 8 is a rotary switching mechanism, the control signal input interface of the drive motor receives the switching control signal. The drive motor determines the rotation direction and rotation angle according to the switching control signal, and drives the rotating shaft to rotate. The rotating shaft drives the turntable to rotate around the vertical axis, and the low-magnification objective lens 6 and the high-magnification objective lens 7, which are fixedly mounted on the turntable, rotate synchronously with the turntable. When the switching control signal indicates switching to the low-magnification objective lens 6, the drive motor drives the turntable to rotate until the low-magnification objective lens 6 is directly below the common optical path of the imaging probe light and the disturbance light. When the switching control signal indicates switching to the high-magnification objective lens 7, the drive motor drives the turntable to rotate until the high-magnification objective lens 7 is in the common optical path of the imaging probe light and the disturbance light. When the objective lens switching mechanism 8 is a sliding table type switching mechanism, the control signal input interface of the drive motor receives the switching control signal. The drive motor determines the rotation direction and rotation angle according to the switching control signal, and drives the lead screw to rotate. The lead screw drives the slide table to move horizontally along the length of the slide rail through the nut seat. When the switching control signal indicates switching to the low-magnification objective lens 6, the drive motor drives the slide table to move horizontally until the low-magnification objective lens 6 is in the common optical path of the imaging probe light and the disturbance light. When the switching control signal indicates switching to the high-magnification objective lens 7, the drive motor drives the slide table to move horizontally until the high-magnification objective lens 7 is in the common optical path of the imaging probe light and the disturbance light. After the objective lens switching mechanism 8 completes the switching between the low-magnification objective lens 6 and the high-magnification objective lens 7, the drive motor of the objective lens switching mechanism 8 outputs a switching completion signal to the digital input interface of the synchronization control module 1 through the signal line. The switching completion signal is a digital level pulse signal, and the effective edge (rising edge or falling edge) of the switching completion signal indicates that the objective lens switching mechanism 8 has completed the objective lens switching.

[0086] After the objective lens switching mechanism 8 completes the switching, the synchronization control module 1 triggers the autofocus module 13 to perform autofocus. Upon detecting a valid edge (rising or falling edge) of the switching completion signal, the central processing unit (CPU) of the synchronization control module 1 generates an autofocus start signal, which is a digital level pulse signal. The CPU outputs the autofocus start signal to the control signal input interface of the autofocus module 13 via a digital output interface. The control signal input interface of the autofocus module 13 is electrically connected to the digital output interface of the synchronization control module 1 via a signal line. Upon receiving the autofocus start signal, the autofocus module 13 begins autofocusing. The image analysis unit of the autofocus module 13 receives the probe light response image currently output by the image acquisition module 14. The image analysis unit performs a sharpness evaluation calculation on the probe light response image and obtains a sharpness evaluation value. The drive control unit of the autofocus module 13 compares the sharpness evaluation value with a preset sharpness threshold. When the sharpness evaluation value is lower than the preset sharpness threshold, the drive control unit generates a focus control command and outputs it through the synchronization control module 1 to the drive motor of the objective lens switching mechanism 8 or to the electric displacement stage 12 of the sample carrier and environment control module to adjust the relative distance between the semiconductor sample 10 to be tested and the low-magnification objective lens 6 or the high-magnification objective lens 7. The autofocus module 13 executes this feedback control process cyclically until the sharpness evaluation value calculated by the image analysis unit reaches or exceeds the preset sharpness threshold.

[0087] After the autofocus module 13 completes focusing, the synchronization control module 1 triggers the image acquisition module 14 to begin acquiring time-series optical images. Upon completion of focusing, the autofocus module 13 outputs a focus completion signal to the digital input interface of the synchronization control module 1 via a signal line. This focus completion signal is a digital level pulse signal, and its valid edge (rising or falling edge) indicates that the autofocus module 13 has completed the focusing operation. After detecting the valid edge (rising or falling edge) of the focus completion signal, the central processing unit (CPU) of the synchronization control module 1 generates a trigger signal based on a reference signal and outputs it to the image acquisition module 14 via a digital output interface. Upon receiving the trigger signal, the image acquisition module 14 begins acquiring time-series optical images. The synchronization control module 1 triggers the image acquisition module 14 to begin acquisition in the manner described above after objective lens switching and autofocus are completed, ensuring that the image acquisition module 14 acquires images under the correct focusing state after objective lens switching.

[0088] The defect determination module 15 is connected to the image acquisition module 14 and is used to receive time-series optical images, perform pixel-level demodulation on the time-series optical images, extract the signal components of each pixel or each region of interest that are at the same frequency as the perturbed light or have a fixed phase relationship with the perturbed light, so as to obtain amplitude maps and phase maps, and determine the location and type of defects in the semiconductor sample 10 to be inspected based on the amplitude maps and phase maps. Specifically: 1) Acquire a first time-series optically corrected image, a second time-series optically corrected image, or a third time-series optically corrected image, specifically: ① When the semiconductor sample 10 to be tested is in an air environment, the defect determination module 15 performs air background subtraction and flat field correction on the time-series optical image to obtain a first time-series optically corrected image, specifically: The defect determination module 15 acquires multiple frames of air background images acquired by the image acquisition module 14 when the periodic disturbance light output module is in the off state, and performs pixel-level averaging on the multiple frames of air background images to obtain an average air background image. The pixel value of the corresponding position in the average air background image is subtracted from each frame of the time series optical image. The module 15 also acquires multiple frames of uniform illumination images acquired by the image acquisition module 14 when the periodic disturbance light output module is in the on state and the semiconductor sample 10 to be tested is not loaded, and performs pixel-level averaging on the multiple frames of uniform illumination images to obtain a flat field correction reference image. The pixel value of each frame of the image after air background subtraction is divided by the pixel value of the corresponding position in the flat field correction reference image and normalized to obtain a first time series optical correction image. The defect determination module 15 acquires multiple frames of air background images captured by the image acquisition module 14 when the periodic disturbance light output module is in the off state. The periodic disturbance light output module being in the off state means that the power supply to the periodic disturbance light source 2 is cut off or the modulator 3 stops working, and the periodic disturbance light output module does not output any disturbance light to the semiconductor sample 10 under test. When the periodic disturbance light output module is in the off state, the image acquisition module 14 continuously acquires multiple frames of optical images of the probe light response at a preset frame rate. At this time, the sample cavity 11 is an air environment, meaning the sample cavity 11 is filled with air and the air pressure is equal to the external atmospheric pressure. The multiple frames of air background images acquired by the image acquisition module 14 record the images formed when, under conditions of no disturbance light illumination, the imaging probe light is irradiated onto the surface of the sample stage 9 via the beam combining / splitting optical path 5 and the microscopic imaging module, and the probe light response returns to the image acquisition module 14 via the microscopic imaging module and the beam combining / splitting optical path 5. Since there is no irradiation, the air background image contains static optical signals generated by the reflection, transmission, or scattering of the imaging probe light in the air environment and on the surface of the sample stage 9, as well as the dark current noise and readout noise of the image acquisition module 14 itself. These signals are unrelated to the defects of the semiconductor sample 10 under test and are interference signals for subsequent defect determination. The defect determination module 15 receives multiple frames of air background images from the image acquisition module 14 via a data cable. Each frame of the multiple air background images is a two-dimensional digital image. The two-dimensional digital image contains multiple pixels, each pixel has a corresponding pixel position and pixel value. The pixel position is determined by the horizontal and vertical coordinates, and the pixel value is the light intensity value at that pixel position.

[0089] The defect determination module 15 performs pixel-level averaging on multiple frames of atmospheric background images to obtain an average atmospheric background image. Pixel-level averaging refers to summing all pixel values ​​at the same pixel location in multiple frames of atmospheric background images and dividing by the number of frames to obtain the average pixel value at that location. The central processing unit (CPU) of the defect determination module 15 reads digital image data of multiple frames of atmospheric background images from memory. The CPU processes each frame of the image one by one according to pixel location. For a pixel location with horizontal coordinate i and vertical coordinate j, the CPU reads the pixel value at that location from the first frame of the atmospheric background image, the pixel value at that location from the second frame of the atmospheric background image, and so on, up to the Nth frame of the atmospheric background image, where N is the total number of frames of the acquired atmospheric background images. The CPU adds the N pixel values ​​read to obtain a total pixel value and divides the total pixel value by N to obtain the average pixel value at that location. The CPU repeats the above calculation process for all pixel locations in the image to obtain the average pixel value corresponding to each pixel location. All pixel locations and their corresponding average pixel values ​​together constitute the average atmospheric background image. The size of the average air background image is exactly the same as that of a single-frame air background image. The pixel value at horizontal coordinate i and vertical coordinate j in the average air background image is equal to the average pixel value at the same pixel position in multiple frames of air background images. The average air background image represents the average static optical signal distribution formed by the imaging probe light in the air environment and on the surface of the sample stage 9 under undisturbed light illumination, as well as the average dark current noise and readout noise distribution of the image acquisition module 14.

[0090] The defect determination module 15 subtracts the pixel value of the corresponding position in the average air background image from each frame of the time-series optical image. The time-series optical image is a sequence of probe light response images continuously acquired by the image acquisition module 14 at a preset frame rate throughout the entire process of the perturbation light applying perturbation to the semiconductor sample 10 under test, when the periodic perturbation light output module is in the on state and the semiconductor sample 10 under test is loaded on the sample support stage 9. Each frame of the time-series optical image is a two-dimensional digital image, and each pixel in each frame has a pixel position determined by horizontal and vertical coordinates, as well as a pixel value at that pixel position. The central processing unit of the defect determination module 15 reads the first frame of the time-series optical image and the average air background image from the memory, and processes each pixel position in the first frame image sequentially. For a pixel position with horizontal coordinate i and vertical coordinate j, the CPU reads the pixel value of that pixel position in the first frame image and simultaneously reads the pixel value of the same pixel position in the average atmospheric background image. The CPU subtracts the pixel value in the average atmospheric background image from the pixel value in the first frame image to obtain the difference pixel value for that pixel position. After performing the above subtraction operation on all pixel positions in the first frame image, the CPU obtains the background-subtracted image of the first frame image. The CPU repeats the above pixel-by-pixel subtraction operation on the second, third, and finally last frames of the time-series optical images to obtain the background-subtracted image for each frame. The pixel-by-pixel subtraction of the average atmospheric background image eliminates static optical signals unrelated to defects in the semiconductor sample 10 under test, surface reflection signals from the sample stage 9, and dark current noise and readout noise from the image acquisition module 14 in the time-series optical images, so that the pixel values ​​in the background-subtracted image mainly reflect the dynamic optical response of the semiconductor sample 10 under test under the action of perturbation light.

[0091] The defect determination module 15 acquires multiple frames of uniformly illuminated images captured by the image acquisition module 14 when the periodic perturbation light output module is in the on state and the semiconductor sample 10 to be tested is not loaded. The periodic perturbation light output module being in the on state means that the power supply to the periodic perturbation light source 2 is turned on and the modulator 3 is working normally, with the periodic perturbation light output module outputting periodically modulated perturbation light into the sample cavity 11. The semiconductor sample 10 to be tested being unloaded means that no semiconductor sample 10 to be tested is placed on the bearing surface of the sample stage 9, and the upper surface of the sample stage 9 is directly exposed to the optical paths of the imaging probe light and the perturbation light. When the periodic perturbation light output module is in the on state and the semiconductor sample 10 to be tested is unloaded, the image acquisition module 14 continuously acquires multiple frames of optical images of the probe light response at a preset frame rate. At this time, the sample cavity 11 is an air environment, meaning that the sample cavity 11 is filled with air and the air pressure is equal to the external atmospheric pressure. The image acquisition module 14 acquires multiple frames of uniform illumination images, recording the images formed when, under conditions of perturbation light on and no sample, the imaging probe light is irradiated onto the upper surface of the sample stage 9 via the beam combiner / splitter path 5 and the microscopic imaging module, and the probe light response is returned to the image acquisition module 14 via the microscopic imaging module and the beam combiner / splitter path 5. Since no sample is present, the pixel value distribution in the uniform illumination image reflects the non-uniformity of the imaging probe light in the optical path, the non-uniformity of the intensity distribution of the illumination spot, the non-uniformity of the transmittance of each optical element in the beam combiner / splitter path 5, the non-uniformity of the transmittance of each objective lens in the microscopic imaging module, and the non-uniformity of the response sensitivity of each pixel in the image acquisition module 14. These non-uniformities are superimposed on the actual optical response signal of the semiconductor sample 10 under test in the time-series optical images and need to be corrected before defect determination.

[0092] The defect determination module 15 performs pixel-level averaging on multiple frames of uniform illumination images to obtain a flat-field correction reference image. Pixel-level averaging refers to summing all pixel values ​​at the same pixel location in multiple frames of uniform illumination images and dividing by the number of frames to obtain the average pixel value at that location. The central processing unit (CPU) of the defect determination module 15 reads the digital image data of the multiple frames of uniform illumination images from memory. The CPU processes each frame of the image one by one according to the pixel location. For a pixel location with a horizontal coordinate of i and a vertical coordinate of j, the CPU reads the pixel value at that location from the first frame of the uniform illumination image, then from the second frame, and so on, up to the Mth frame, where M is the total number of frames of the acquired uniform illumination images. The CPU adds the M pixel values ​​to obtain the total pixel value and divides the total pixel value by M to obtain the average pixel value at that location. The CPU repeats the above calculation process for all pixel locations in the image to obtain the average pixel value corresponding to each pixel location. All pixel locations and their corresponding average pixel values ​​together constitute the flat-field correction reference image. The size of the flat-field correction reference image is exactly the same as that of a single-frame uniform illumination image. The pixel value at horizontal coordinate i and vertical coordinate j in the flat-field correction reference image is equal to the average pixel value at the same pixel location in multiple frames of uniform illumination images. The flat-field correction reference image represents the combined spatial distribution of the intensity distribution non-uniformity of the imaging probe light throughout the optical system, the transmittance non-uniformity of the beam combining and splitting paths 5, the transmittance non-uniformity of the objective lens, and the pixel response sensitivity non-uniformity of the image acquisition module 14 under the condition that the perturbation light is on and there is no sample.

[0093] The defect determination module 15 divides each frame of the image after background subtraction by the pixel value of the corresponding position in the flat-field correction reference image and performs normalization processing to obtain the first time-series optical correction image. Each frame of the image after background subtraction is the background-subtracted image obtained by subtracting the average image of the background from the corresponding frame image in the time-series optical image. Each pixel in the background-subtracted image has a pixel position determined by horizontal and vertical coordinates, as well as a difference pixel value at that pixel position. The central processing unit of the defect determination module 15 reads the first frame of background-subtracted image and the flat-field correction reference image from the memory. The central processing unit processes each pixel position in the first frame of background-subtracted image sequentially. For a pixel position with horizontal coordinate i and vertical coordinate j, the central processing unit reads the difference pixel value of that pixel position in the first frame of background-subtracted image and simultaneously reads the average pixel value of the same pixel position in the flat-field correction reference image. The central processing unit divides the difference pixel value in the first frame of background subtraction image by the average pixel value in the flat-field correction reference image to obtain the preliminary correction pixel value of that pixel position. After performing the aforementioned division operation on all pixel positions in the first frame of the background-subtracted image, the CPU obtains the preliminary corrected image of the first frame. The CPU repeats the pixel-by-pixel division operation on the second, third, and final frames of the background-subtracted image to obtain the preliminary corrected image for each frame. After completing the pixel-by-pixel division, the CPU performs normalization processing on the preliminary corrected image. Normalization processing involves transforming the preliminary corrected pixel value of each pixel in the preliminary corrected image to a predetermined numerical range according to a linear mapping relationship; the predetermined numerical range is 0 to 1 or 0 to 255. The CPU calculates the maximum and minimum pixel values ​​of all pixels in the preliminary corrected image. For each pixel position in the preliminary corrected image, the CPU subtracts the minimum pixel value from the preliminary corrected pixel value of that pixel position and divides it by the difference between the maximum and minimum pixel values ​​to obtain the normalized pixel value. After performing the above normalization calculation on all pixel positions in the preliminary corrected image, the CPU obtains the normalized image. The central processing unit repeats the above normalization process for each frame of the preliminary corrected image to obtain the normalized image corresponding to each frame. All normalized images are arranged in chronological order of the original frames in the time-series optical image, together forming the first time-series optical corrected image. Each frame in the first time-series optical corrected image eliminates the influence of static signals from the air background, illumination inhomogeneity, transmittance inhomogeneity of optical elements, and pixel response inhomogeneity of the image acquisition module 14. The pixel values ​​in the first time-series optical corrected image can truly reflect the spatial distribution and temporal variation of the local optical response signal generated by the semiconductor sample 10 under periodic perturbation light.

[0094] ② When the semiconductor sample 10 to be tested is in a vacuum environment, the defect determination module 15 performs vacuum background subtraction and flat field correction on the time-series optical image to obtain a second time-series optically corrected image, specifically: The defect determination module 15 acquires multiple frames of vacuum background images acquired by the image acquisition module 14 when the sample cavity 11 is in a vacuum state and the periodic perturbation light output module is in a closed state, and performs pixel-level averaging on the multiple frames of vacuum background images to obtain a vacuum background average image. The pixel value of the corresponding position in the vacuum background average image is subtracted from each frame of the time series optical image. The module 15 acquires multiple frames of uniform illumination images acquired by the image acquisition module 14 when the sample cavity 11 is in a vacuum state, the periodic perturbation light output module is in a closed state and the semiconductor sample 10 to be tested is not loaded, and performs pixel-level averaging on the multiple frames of uniform illumination images to obtain a flat field correction reference image. The pixel value of each frame of the image after vacuum background subtraction is divided by the pixel value of the corresponding position in the flat field correction reference image and normalized to obtain a second time series optical correction image. The defect determination module 15 acquires multiple frames of vacuum background images captured by the image acquisition module 14 when the sample cavity 11 is in a vacuum state and the periodic perturbation light output module is in the off state. A vacuum state in the sample cavity 11 means that the sample cavity 11 is continuously evacuated by a vacuum pump, reducing the internal air pressure to a predetermined vacuum level below atmospheric pressure, where the predetermined vacuum level is less than 1 Pascal. The periodic perturbation light output module being in the off state means that the power supply to the periodic perturbation light source 2 is cut off or the modulator 3 stops working, and the periodic perturbation light output module does not output any perturbation light to the semiconductor sample 10 under test. When the sample cavity 11 is in a vacuum state and the periodic perturbation light output module is in the off state, the image acquisition module 14 continuously acquires multiple frames of optical images of the probe light response at a preset frame rate. The image acquisition module 14 acquires multiple frames of vacuum background images, recording the images formed when the imaging probe light, after passing through the beam combiner / splitter path 5 and the microscopic imaging module to illuminate the surface of the sample stage 9 in a vacuum environment without irradiation, returns to the image acquisition module 14 via the microscopic imaging module and the beam combiner / splitter path 5. Since the sample cavity 11 is in a vacuum state, there is no scattering or absorption effect of air molecules on the imaging probe light in the vacuum background image. The vacuum background image contains static optical signals generated by the reflection and transmission of the imaging probe light in the vacuum environment and on the surface of the sample stage 9, as well as the dark current noise and readout noise of the image acquisition module 14 itself. The difference between the vacuum background image and the air background image is that the vacuum background image does not contain scattering signals from air molecules, and the signal-to-noise ratio of the vacuum background image is higher than that of the air background image. The defect judgment module 15 receives multiple frames of vacuum background images from the image acquisition module 14 via a data cable. Each frame of the multiple vacuum background images is a two-dimensional digital image. The two-dimensional digital image contains multiple pixels. Each pixel has a corresponding pixel position and pixel value. The pixel position is determined by the horizontal and vertical coordinates, and the pixel value is the light intensity value at that pixel position.

[0095] The defect determination module 15 performs pixel-level averaging on multiple frames of vacuum background images to obtain an average vacuum background image. Pixel-level averaging refers to summing all pixel values ​​at the same pixel location in multiple frames of vacuum background images and dividing by the number of frames to obtain the average pixel value at that location. The central processing unit (CPU) of the defect determination module 15 reads the digital image data of multiple frames of vacuum background images from memory. The CPU processes each frame of the image one by one according to pixel location. For a pixel location with horizontal coordinate i and vertical coordinate j, the CPU reads the pixel value at that location from the first frame of the vacuum background image, the pixel value at that location from the second frame of the vacuum background image, and so on, up to the Nth frame of the vacuum background image, where N is the total number of frames of the acquired vacuum background images. The CPU adds the N pixel values ​​read to obtain the total pixel value and divides the total pixel value by N to obtain the average pixel value at that location. The CPU repeats the above calculation process for all pixel locations in the image to obtain the average pixel value corresponding to each pixel location. All pixel locations and their corresponding average pixel values ​​together constitute the average vacuum background image. The size of the vacuum background average image is exactly the same as that of a single-frame vacuum background image. The pixel value at horizontal coordinate i and vertical coordinate j in the vacuum background average image is equal to the average pixel value at the same pixel position in multiple frames of vacuum background images. The vacuum background average image represents the average static optical signal distribution formed by the imaging probe light on the vacuum environment and the surface of the sample stage 9 under vacuum conditions and undisturbed light illumination, as well as the average dark current noise and readout noise distribution of the image acquisition module 14 under vacuum conditions.

[0096] The defect determination module 15 subtracts the pixel value of the corresponding position in the vacuum background average image from each frame of the time-series optical image. The time-series optical image is a sequence of probe light response images continuously acquired at a preset frame rate by the image acquisition module 14 when the sample cavity 11 is in a vacuum state, the periodic perturbation light output module is in the on state, and the semiconductor sample 10 to be tested is loaded on the sample support stage 9, during the entire process of the perturbation light applying perturbation to the semiconductor sample 10 to be tested. Each frame of the time-series optical image is a two-dimensional digital image, and each pixel in each frame has a pixel position determined by horizontal and vertical coordinates, as well as a pixel value at that pixel position. The central processing unit of the defect determination module 15 reads the first frame of the time-series optical image and the vacuum background average image from the memory, and processes each pixel position in the first frame image sequentially. For a pixel position with horizontal coordinate i and vertical coordinate j, the CPU reads the pixel value of that pixel position in the first frame image and simultaneously reads the pixel value of the same pixel position in the vacuum background average image. The CPU subtracts the pixel value in the vacuum background average image from the pixel value in the first frame image to obtain the difference pixel value for that pixel position. After performing the above subtraction operation on all pixel positions in the first frame image, the CPU obtains the vacuum background subtracted image of the first frame image. The CPU repeats the above pixel-by-pixel subtraction operation on the second, third, and last frames of the time-series optical images to obtain the vacuum background subtracted image for each frame image. The pixel-by-pixel subtraction of the vacuum background average image eliminates static optical signals of the vacuum environment, surface reflection signals of the sample stage 9, and dark current noise and readout noise of the image acquisition module 14 that are unrelated to defects in the semiconductor sample 10 under test in the time-series optical images. This ensures that the pixel values ​​in the vacuum background subtracted image mainly reflect the dynamic optical response of the semiconductor sample 10 under test in a vacuum environment under the action of perturbed light. The signal-to-noise ratio of the image after vacuum background subtraction is higher than that of the image after air background subtraction because the effects of air molecule scattering and absorption are eliminated.

[0097] The defect determination module 15 acquires multiple frames of uniformly illuminated images captured by the image acquisition module 14 when the sample cavity 11 is in a vacuum state, the periodic perturbation light output module is turned on, and the semiconductor sample 10 to be tested is not loaded. A vacuum state in the sample cavity 11 means that the sample cavity 11 is continuously evacuated by a vacuum pump to maintain the internal air pressure at a predetermined vacuum level. The periodic perturbation light output module being turned on means that the power supply to the periodic perturbation light source 2 is connected, and the modulator 3 is working normally, with the periodic perturbation light output module outputting periodically modulated perturbation light into the sample cavity 11. The semiconductor sample 10 to be tested being unloaded means that no semiconductor sample 10 to be tested is placed on the bearing surface of the sample support stage 9, and the upper surface of the sample support stage 9 is directly exposed to the optical paths of the imaging probe light and the perturbation light. When the sample cavity 11 is in a vacuum state, the periodic perturbation light output module is turned on, and the semiconductor sample 10 to be tested is unloaded, the image acquisition module 14 continuously acquires multiple frames of optical images of the probe light response at a preset frame rate. The image acquisition module 14 acquires multiple frames of uniformly illuminated images, recording the images formed under vacuum conditions with the perturbation light on and no sample present. The imaging probe light is irradiated onto the upper surface of the sample stage 9 via the beam combiner / splitter path 5 and the microscopic imaging module. The probe light response then returns to the image acquisition module 14 via the microscopic imaging module and the beam combiner / splitter path 5. Since the sample cavity 11 is in a vacuum state and no sample is present, the pixel value distribution in the uniformly illuminated images reflects the non-uniformity of the imaging probe light in the optical path, the non-uniformity of the illumination spot intensity distribution, the non-uniformity of the transmittance of each optical element in the beam combiner / splitter path 5, the non-uniformity of the transmittance of each objective lens in the microscopic imaging module, and the non-uniformity of the response sensitivity of each pixel in the image acquisition module 14 under vacuum conditions. The uniformly illuminated images acquired under vacuum conditions eliminate the additional non-uniformity caused by air molecule scattering and absorption, making the flat-field correction reference image more accurately reflect the non-uniformity of the optical system itself.

[0098] The defect determination module 15 performs pixel-level averaging on multiple frames of uniform illumination images to obtain a flat-field correction reference image. Pixel-level averaging refers to summing all pixel values ​​at the same pixel location in multiple frames of uniform illumination images and dividing by the number of frames to obtain the average pixel value at that location. The central processing unit (CPU) of the defect determination module 15 reads the digital image data of the multiple frames of uniform illumination images from memory. The CPU processes each frame of the image one by one according to the pixel location. For a pixel location with a horizontal coordinate of i and a vertical coordinate of j, the CPU reads the pixel value at that location from the first frame of the uniform illumination image, then from the second frame, and so on, up to the Mth frame, where M is the total number of frames of the acquired uniform illumination images. The CPU adds the M pixel values ​​to obtain the total pixel value and divides the total pixel value by M to obtain the average pixel value at that location. The CPU repeats the above calculation process for all pixel locations in the image to obtain the average pixel value corresponding to each pixel location. All pixel locations and their corresponding average pixel values ​​together constitute the flat-field correction reference image. The size of the flat-field correction reference image is exactly the same as that of a single-frame uniform illumination image. The pixel value at horizontal coordinate i and vertical coordinate j in the flat-field correction reference image is equal to the average pixel value at the same pixel position in multiple frames of uniform illumination images. The flat-field correction reference image obtained in a vacuum environment represents the combined spatial distribution of the intensity non-uniformity of the imaging probe light in the entire optical system, the transmittance non-uniformity of the beam combining and splitting paths 5, the transmittance non-uniformity of the objective lens, and the pixel response sensitivity non-uniformity of the image acquisition module 14 under vacuum, perturbation light on, and sample-free conditions. This flat-field correction reference image does not contain the effects of scattering and absorption by air molecules.

[0099] The defect determination module 15 divides each frame of the image after vacuum background subtraction pixel by the corresponding pixel value in the flat-field correction reference image and performs normalization processing to obtain the second time-series optical correction image. Each frame of the image after vacuum background subtraction is the vacuum background subtracted from the corresponding frame image in the time-series optical image. Each pixel in the vacuum background subtracted image has a pixel position determined by horizontal and vertical coordinates, as well as a difference pixel value at that pixel position. The central processing unit of the defect determination module 15 reads the first frame of vacuum background subtracted image and the flat-field correction reference image from the memory. The central processing unit processes each pixel position in the first frame of vacuum background subtracted image sequentially. For a pixel position with horizontal coordinate i and vertical coordinate j, the central processing unit reads the difference pixel value of that pixel position in the first frame of vacuum background subtracted image and simultaneously reads the average pixel value of the same pixel position in the flat-field correction reference image. The central processing unit divides the difference pixel value in the first frame of vacuum background subtracted image by the average pixel value in the flat-field correction reference image to obtain the preliminary correction pixel value of that pixel position. After performing the aforementioned division operation on all pixel positions in the first frame image after vacuum background subtraction, the CPU obtains the preliminary corrected image of the first frame. The CPU repeats the pixel-by-pixel division operation on the second, third, and final frames after vacuum background subtraction to obtain the preliminary corrected image for each frame. After completing the pixel-by-pixel division, the CPU performs normalization processing on the preliminary corrected image. Normalization processing involves transforming the preliminary corrected pixel value of each pixel in the preliminary corrected image to a predetermined numerical range according to a linear mapping relationship. The predetermined numerical range is 0 to 1 or 0 to 255. The CPU calculates the maximum and minimum pixel values ​​of all pixels in the preliminary corrected image. For each pixel position in the preliminary corrected image, the CPU subtracts the minimum pixel value from the preliminary corrected pixel value of that pixel position and divides it by the difference between the maximum and minimum pixel values ​​to obtain the normalized pixel value. After performing the above normalization calculation on all pixel positions in the preliminary corrected image, the CPU obtains the normalized image. The central processing unit repeats the above normalization process for each frame of the preliminary corrected image to obtain the normalized image corresponding to each frame. All normalized images are arranged in chronological order of the original frames in the time-series optical image to form the second time-series optical corrected image. Each frame in the second time-series optical corrected image eliminates the effects of static background signals in a vacuum environment, illumination non-uniformity and transmittance non-uniformity of the optical system in a vacuum environment, and 14-pixel response non-uniformity of the image acquisition module.Compared with the first time-series optically corrected image, the second time-series optically corrected image eliminates noise introduced by air molecule scattering and absorption. The second time-series optically corrected image has a higher signal-to-noise ratio, and the pixel values ​​in the second time-series optically corrected image can more accurately reflect the spatial distribution and temporal variation of the local optical response signal generated by the semiconductor sample 10 under vacuum environment and periodic perturbation light.

[0100] ③ When the semiconductor sample 10 to be tested is in a liquid phase environment, the defect determination module 15 performs liquid phase background subtraction and flat-field correction on the time-series optical image to obtain a third time-series optically corrected image, specifically: The defect determination module 15 acquires multiple frames of liquid background images acquired by the image acquisition module 14 when the liquid sample cell or flow cell is filled with liquid medium, the semiconductor sample 10 to be tested is not immersed in the liquid medium, and the periodic perturbation light output module is in the off state. The module then performs pixel-level averaging on the multiple frames of liquid background images to obtain an average liquid background image. Each frame of the time-series optical image is subtracted pixel by pixel from the corresponding position of the average liquid background image. The module also acquires multiple frames of uniform illumination images acquired by the image acquisition module 14 when the liquid sample cell or flow cell is filled with liquid medium, the semiconductor sample 10 to be tested is not immersed in the liquid medium, and the periodic perturbation light output module is in the on state. The module then performs pixel-level averaging on the multiple frames of uniform illumination images to obtain a flat field correction reference image. Each frame of the image after liquid background subtraction is divided pixel by pixel by the corresponding position of the flat field correction reference image and normalized to obtain a third time-series optical correction image.

[0101] The defect determination module 15 acquires multiple frames of liquid background images captured by the image acquisition module 14 when the liquid sample cell or flow cell is filled with liquid medium, the semiconductor sample 10 to be tested is not immersed in the liquid medium, and the periodic perturbation light output module is in the off state. The liquid sample cell is a container structure with an open top, a flat bottom plate, and side walls perpendicular to the bottom plate. The bottom plate and side walls together form a cavity for containing the liquid medium. The bottom plate of the liquid sample cell is fixedly installed at the bottom of the sample cavity 11 or on the motion platform of the electric displacement stage 12. The flow cell is a closed or semi-closed flow channel structure with a liquid inlet and a liquid outlet. The interior of the flow cell has a flow channel for the liquid medium to flow through. The bottom of the flow cell is a flat bottom plate, which is fixedly installed at the bottom of the sample cavity 11 or on the motion platform of the electric displacement stage 12. The semiconductor sample 10 to be tested is placed in the detection area on the bottom plate of the flow cell. The liquid medium is a liquid substance filling the containment cavity of the liquid phase sample cell or the flow channel of the flow cell. The liquid medium can be deionized water, acidic solution, alkaline solution, buffer solution, salt solution, electrolyte solution, alcohol solvent, nitrile solvent, carbonate solvent, ether solvent, amide solvent, ionic liquid, or a mixture of the above liquid media. "The liquid phase sample cell or flow cell is filled with liquid medium" means that the liquid medium fills from the bottom of the liquid phase sample cell upwards to a predetermined liquid level, or that the liquid medium continuously flows through the flow channel of the flow cell and fills the entire flow channel space. "The semiconductor sample 10 to be tested is not immersed in the liquid medium" means that the semiconductor sample 10 to be tested is not placed on the sample support stage 9, or that although the semiconductor sample 10 to be tested is placed on the sample support stage 9, it is not in contact with the liquid medium in the liquid phase sample cell or flow cell; the semiconductor sample 10 to be tested is located above the liquid surface of the liquid medium in the liquid phase sample cell or flow cell or outside the flow channel of the flow cell. The periodic perturbation light output module being in the off state means that the power supply to the periodic perturbation light source 2 is cut off or the modulator 3 stops working, and the periodic perturbation light output module does not output any perturbation light. When the liquid sample cell or flow cell is filled with liquid medium and the semiconductor sample 10 to be tested is not immersed in the liquid medium, and the periodic perturbation light output module is in the off state, the image acquisition module 14 continuously acquires multiple frames of optical images of the probe light response at a preset frame rate. The multiple frames of liquid background images acquired by the image acquisition module 14 record the images formed when, under conditions of a liquid environment and no perturbation light irradiation, the imaging probe light is irradiated to the bottom of the liquid sample cell or flow cell via the beam combining / splitting optical path 5 and the microscopic imaging module, and the probe light response returns to the image acquisition module 14 via the microscopic imaging module and the beam combining / splitting optical path 5. The liquid background images contain static optical signals generated by reflection, transmission, and scattering of the imaging probe light as it passes through the transparent window, liquid medium, and the bottom of the liquid sample cell or flow cell in the liquid environment, as well as the dark current noise and readout noise of the image acquisition module 14 itself.The liquid background image also includes the signal attenuation distribution caused by the absorption and scattering of the imaging probe light by the liquid medium, as well as the local scattering signal caused by suspended particles or bubbles in the liquid medium. The defect judgment module 15 receives multiple frames of liquid background images from the image acquisition module 14 via a data cable. Each frame of the multiple liquid background images is a two-dimensional digital image. The two-dimensional digital image contains multiple pixels, each pixel has a corresponding pixel position and pixel value. The pixel position is determined by the horizontal and vertical coordinates, and the pixel value is the light intensity value at that pixel position.

[0102] The defect determination module 15 performs pixel-level averaging on multiple frames of liquid background images to obtain an average liquid background image. Pixel-level averaging refers to summing all pixel values ​​at the same pixel location in multiple frames of liquid background images and dividing by the number of frames to obtain the average pixel value at that location. The central processing unit (CPU) of the defect determination module 15 reads digital image data of multiple frames of liquid background images from memory, and processes each frame of image one by one according to pixel location. For a pixel location with horizontal coordinate i and vertical coordinate j, the CPU reads the pixel value at that location from the first frame of the liquid background image, the pixel value at that location from the second frame of the liquid background image, and so on, up to the Nth frame of the liquid background image, where N is the total number of frames of the acquired liquid background images. The CPU adds the N pixel values ​​read to obtain the total pixel value, and divides the total pixel value by N to obtain the average pixel value at that location. The central processing unit repeats the above calculation process for all pixel positions in the image to obtain the average pixel value for each pixel position. All pixel positions and their corresponding average pixel values ​​together constitute the average liquid phase background image. The size of the average liquid phase background image is exactly the same as the size of a single-frame liquid phase background image. The pixel value located at the horizontal coordinate i and vertical coordinate j in the average liquid phase background image is equal to the average pixel value at the same pixel position in multiple frames of liquid phase background images. The average liquid phase background image represents the average static optical signal distribution formed after the imaging probe light passes through the transparent window, the liquid medium, and the bottom of the liquid phase sample cell or the bottom of the flow cell under undisturbed light conditions in a liquid phase environment, as well as the average distribution of absorption and scattering by the liquid medium, and the average dark current noise and readout noise distribution of the image acquisition module 14.

[0103] The defect determination module 15 subtracts the pixel value of the corresponding position in the average liquid phase background image from each frame of the time-series optical image. The time-series optical image is a sequence of probe light response images continuously acquired at a preset frame rate by the image acquisition module 14 when the liquid phase sample cell or flow cell is filled with liquid medium, the semiconductor sample 10 to be tested is immersed in the liquid medium, and the periodic perturbation light output module is in the on state, during the entire process of the perturbation light applying perturbation to the semiconductor sample 10 to be tested. Immersion of the semiconductor sample 10 to be tested in liquid medium means that the semiconductor sample 10 to be tested is completely immersed in the liquid medium within the containment cavity of the liquid phase sample cell or completely immersed in the liquid medium within the flow channel of the flow cell, with both the upper and lower surfaces of the semiconductor sample 10 in contact with the liquid medium. Each frame of the time-series optical image is a two-dimensional digital image, and each pixel in each frame has a pixel position determined by horizontal and vertical coordinates, as well as a pixel value at that position. The CPU of the defect determination module 15 reads the first frame image and the average liquid background image of the time-series optical image from the memory. The CPU processes each pixel position in the first frame image sequentially. For a pixel position with a horizontal coordinate of i and a vertical coordinate of j, the CPU reads the pixel value of that pixel position in the first frame image and simultaneously reads the pixel value of the same pixel position in the average liquid background image. The CPU subtracts the pixel value in the average liquid background image from the pixel value in the first frame image to obtain the difference pixel value for that pixel position. After performing the above subtraction operation on all pixel positions in the first frame image, the CPU obtains the liquid background subtracted image of the first frame image. The CPU repeats the above pixel-by-pixel subtraction operation on the second frame image, the third frame image, and so on up to the last frame image of the time-series optical image to obtain the liquid background subtracted image corresponding to each frame image. The operation of subtracting the liquid background average image pixel by pixel eliminates static optical signals of the liquid environment, transparent window reflection signals, liquid medium absorption and scattering signals, reflection signals from the bottom of the liquid sample cell or the bottom of the flow cell, as well as dark current noise and readout noise of the image acquisition module 14 that are unrelated to defects in the semiconductor sample 10 under test in the time-series optical image. This makes the pixel values ​​in the image after liquid background subtraction mainly reflect the dynamic optical response of the semiconductor sample 10 under test in the liquid environment and under the action of perturbed light.

[0104] The defect determination module 15 acquires multiple frames of uniform illumination images captured by the image acquisition module 14 when the liquid sample cell or flow cell is filled with liquid medium, the semiconductor sample 10 to be tested is not immersed in the liquid medium, and the periodic perturbation light output module is in the on state. The state where the liquid sample cell or flow cell is filled with liquid medium and the semiconductor sample 10 to be tested is not immersed in the liquid medium is the same as the acquisition conditions for the aforementioned liquid background image. The periodic perturbation light output module being in the on state means that the power supply of the periodic perturbation light source 2 is turned on and the modulator 3 is working normally, and the periodic perturbation light output module outputs periodically modulated perturbation light to the liquid sample cell or flow cell. When the liquid sample cell or flow cell is filled with liquid medium, the semiconductor sample 10 to be tested is not immersed in the liquid medium, and the periodic perturbation light output module is in the on state, the image acquisition module 14 continuously acquires multiple frames of optical images of the probe light response at a preset frame rate. The image acquisition module 14 acquires multiple frames of uniformly illuminated images, recording the images formed under conditions of a liquid environment, with the perturbation light on and no sample present. These images are formed when the imaging probe light, after passing through the beam combiner / splitter path 5 and the microscopic imaging module, illuminates the bottom of the liquid sample cell or the bottom of the flow cell. The probe light response then returns to the image acquisition module 14 via the microscopic imaging module and the beam combiner / splitter path 5. Since no sample is present, the pixel value distribution in the uniformly illuminated images reflects the non-uniformity of the imaging probe light in the optical path, the non-uniformity of the illumination spot intensity distribution, the non-uniformity of the transmittance of the transparent window, the non-uniformity of absorption and scattering in the liquid medium, the non-uniformity of reflectance at the bottom of the liquid sample cell or the bottom of the flow cell, the non-uniformity of transmittance of each optical element in the beam combiner / splitter path 5, the non-uniformity of transmittance of each objective lens in the microscopic imaging module, and the non-uniformity of the response sensitivity of each pixel in the image acquisition module 14. The uniformly illuminated images acquired in the liquid environment exclude the optical response of the semiconductor sample 10 itself, but fully preserve the spatial non-uniformity caused by the absorption and scattering of the imaging probe light by the liquid medium.

[0105] The defect determination module 15 performs pixel-level averaging on multiple frames of uniform illumination images to obtain a flat-field correction reference image. Pixel-level averaging refers to summing all pixel values ​​at the same pixel location in multiple frames of uniform illumination images and dividing by the number of frames to obtain the average pixel value at that location. The central processing unit (CPU) of the defect determination module 15 reads the digital image data of the multiple frames of uniform illumination images from memory. The CPU processes each frame of the image one by one according to pixel location. For a pixel location with horizontal coordinate i and vertical coordinate j, the CPU reads the pixel value at that location from the first frame of uniform illumination images, the pixel value at that location from the second frame of uniform illumination images, and so on, up to the Mth frame of uniform illumination images, where M is the total number of frames of uniform illumination images acquired. The CPU adds the M pixel values ​​read to obtain the total pixel value and divides the total pixel value by M to obtain the average pixel value at that location. The CPU repeats the above calculation process for all pixel locations in the image to obtain the average pixel value corresponding to each pixel location. All pixel locations and their corresponding average pixel values ​​together constitute the flat-field correction reference image. The size of the flat-field correction reference image is exactly the same as that of a single-frame uniform illumination image. The pixel value at horizontal coordinate i and vertical coordinate j in the flat-field correction reference image is equal to the average pixel value at the same pixel position in multiple frames of uniform illumination images. The flat-field correction reference image obtained in a liquid phase environment represents the combined spatial distribution of the imaging probe light intensity non-uniformity, transparent window transmittance non-uniformity, liquid medium absorption and scattering non-uniformity, liquid sample cell bottom or flow cell bottom reflectance non-uniformity, beam combining and splitting light path transmittance non-uniformity, objective lens transmittance non-uniformity, and image acquisition module pixel response sensitivity non-uniformity in a liquid phase environment with perturbation light on and no sample.

[0106] The defect determination module 15 divides each frame of the image after liquid background subtraction pixel by the corresponding pixel value in the flat-field correction reference image and performs normalization processing to obtain the third time-series optical correction image. Each frame of the image after liquid background subtraction is the liquid background subtracted from the corresponding frame image in the time-series optical image. Each pixel in the liquid background subtracted image has a pixel position determined by horizontal and vertical coordinates, as well as the difference pixel value at that pixel position. The central processing unit of the defect determination module 15 reads the first frame of liquid background subtracted image and the flat-field correction reference image from the memory, and processes each pixel position in the first frame of liquid background subtracted image sequentially. For a pixel position with horizontal coordinate *i* and vertical coordinate *j*, the CPU reads the difference pixel value of that pixel position in the first frame of the liquid background subtracted image, and simultaneously reads the average pixel value of the same pixel position in the flat-field correction reference image. The CPU divides the difference pixel value in the first frame of the liquid background subtracted image by the average pixel value in the flat-field correction reference image to obtain the preliminary corrected pixel value for that pixel position. After performing the above division operation on all pixel positions in the first frame of the liquid background subtracted image, the CPU obtains the preliminary corrected image of the first frame. The CPU repeats the above pixel-by-pixel division operation on the second, third, and last frames of the liquid background subtracted image to obtain the preliminary corrected image for each frame. After completing the pixel-by-pixel division, the CPU performs normalization processing on the preliminary corrected image. Normalization processing refers to converting the preliminary corrected pixel value of each pixel in the preliminary corrected image to a predetermined numerical range according to a linear mapping relationship. The predetermined numerical range is 0 to 1 or 0 to 255. The central processing unit (CPU) calculates the maximum and minimum pixel values ​​for all pixels in the preliminary corrected image. For each pixel location in the preliminary corrected image, the CPU subtracts the minimum pixel value from the preliminary corrected pixel value and divides the result by the difference between the maximum and minimum pixel values ​​to obtain the normalized pixel value. After performing the above normalization calculation on all pixel locations in the preliminary corrected image, the CPU obtains the normalized image. The CPU repeats the above normalization process for each frame of the preliminary corrected image to obtain the normalized image corresponding to each frame. All normalized images are arranged in chronological order of the original frames in the time-series optical image, together forming the third time-series optical corrected image. Each frame in the third time-series optical corrected image eliminates the effects of static background signals in the liquid environment, illumination inhomogeneity and transmittance inhomogeneity of the optical system in the liquid environment, absorption and scattering inhomogeneity of the liquid medium, and pixel response inhomogeneity of the image acquisition module.The pixel values ​​in the third time-series optically corrected image can truly reflect the spatial distribution and temporal variation of the local optical response signal generated by the semiconductor sample 10 under the action of liquid environment and periodic perturbation light. The response signal includes carrier capture and release, interface charge accumulation, local absorption change, reflectivity change, dielectric constant change, refractive index change and probe light response change caused by photothermal temperature change on and inside the surface of the semiconductor sample 10.

[0107] 2) When the defect determination module 15 performs pixel-level demodulation on the corrected time-series optical image, for each pixel, it extracts the intensity sequence of that pixel in the first, second, or third time-series optically corrected image. Using the modulation signal of the perturbed light as a reference signal, it performs orthogonal demodulation on the intensity sequence to extract the cosine and sine components with the same frequency as the perturbed light. The cosine component is the integral average of the product of the intensity sequence and the cosine function of the reference signal over a complete modulation period, and the sine component is the integral average of the product of the intensity sequence and the sine function of the reference signal over a complete modulation period. As an alternative to orthogonal demodulation, the defect determination module 15 can also use lock-in amplification demodulation, fast Fourier transform demodulation, or short-time Fourier transform demodulation. The system employs any one of the following methods: wavelet transform demodulation, cross-correlation analysis demodulation, orthogonal fitting demodulation, or multi-frequency fitting demodulation. It synchronously demodulates the intensity sequence of each pixel to extract signal components with the same frequency as the disturbing light. The defect determination module 15 calculates the amplitude and phase of the pixel based on the cosine and sine components. The amplitude is the square root of the sum of the squares of the cosine and sine components, and the phase is the inverse function of the tangent of the ratio of the sine to the cosine components. The defect determination module 15 performs the above processing on all pixels to obtain amplitude and phase maps. It then performs spatial filtering and threshold segmentation on the amplitude and phase maps to extract amplitude and phase anomaly regions. The spatial intersection of the amplitude and phase anomaly regions is marked as the location of the defect. The type of defect is determined based on the amplitude and phase corresponding to the location of the defect.

[0108] The first time-series optically corrected image is obtained by the defect determination module 15 after performing air background subtraction and flat-field correction on the time-series optical image when the semiconductor sample 10 to be tested is in an air environment. The second time-series optically corrected image is obtained by the defect determination module 15 after performing vacuum background subtraction and flat-field correction on the time-series optical image when the semiconductor sample 10 to be tested is in a vacuum environment. The third time-series optically corrected image is obtained by the defect determination module 15 after performing liquid phase background subtraction and flat-field correction on the time-series optical image when the semiconductor sample 10 to be tested is in a liquid phase environment. All three time-series optically corrected images are data sets of multiple frames of two-dimensional digital images arranged in chronological order of acquisition time. Each frame of the two-dimensional digital image contains multiple pixels, and each pixel has a pixel position determined by horizontal and vertical coordinates. After the central processing unit of the defect determination module 15 reads the first, second, or third time-series optically corrected image from the memory, the central processing unit processes each pixel position one by one. For the horizontal coordinate... And the vertical coordinate is The central processing unit (CPU) reads the pixel value at the specified pixel location from the first frame of the time-series optically corrected image, then from the second frame, then from the third frame, and so on, starting from the fourth frame. The pixel value at that pixel location is read from the frame, where The total number of frames for the time-series optically corrected images. The central processing unit will read the... The pixel values ​​are arranged in ascending order of frame number to form a sequence containing... A one-dimensional sequence of values ​​represents the intensity sequence of a pixel location over time. Each value in the intensity sequence corresponds to the light intensity at the corresponding acquisition time for that pixel location. The time interval between two adjacent values ​​is equal to the time interval between two adjacent frames acquired by the image acquisition module 14, which is the reciprocal of the preset frame rate. The central processing unit repeats the above reading and sorting operations for all pixel locations, ensuring that each pixel location obtains a corresponding intensity sequence.

[0109] The defect determination module 15 uses the modulation signal of the perturbation light as the reference signal. The modulation signal of the perturbation light is an electrical signal output by modulator 3 in the periodic perturbation light output module. The waveform of this electrical signal is consistent with the intensity change waveform of the perturbation light, the frequency of this electrical signal is equal to the modulation frequency of the perturbation light, and the phase of this electrical signal is equal to the modulation phase of the perturbation light. The reference signal is a sinusoidal signal with the same frequency and phase as the modulation signal of the perturbation light. The mathematical expression of the reference signal is: ,in The modulation frequency of the perturbation light. For time, The modulation phase of the perturbation light is determined. After receiving the reference signal output from the modulator 3, the synchronization control module 1 transmits the reference signal to the reference signal input interface of the defect determination module 15 via the signal line. The central processing unit (CPU) of the defect determination module 15 receives the reference signal through the reference signal input interface and stores it in its memory as a reference signal for subsequent quadrature demodulation. The CPU also performs phase delay processing on the reference signal to generate a quadrature reference signal with a phase difference of 90 degrees from the reference signal. The mathematical expression for the quadrature reference signal is: The reference signal and the quadrature reference signal together constitute the two reference components required for quadrature demodulation.

[0110] The defect determination module 15 performs orthogonal demodulation on the intensity sequence to extract the cosine and sine components with the same frequency as the disturbing light. Orthogonal demodulation refers to the process of multiplying and averaging the intensity sequence to be demodulated with a reference signal and an orthogonal reference signal, respectively, to extract the signal components in the intensity sequence with the same frequency as the reference signal. The central processing unit of the defect determination module 15 reads the intensity sequence at a specific pixel location from the memory. The intensity sequence contains... The intensity sequence contains discrete pixel values, each with a known acquisition time. The central processing unit (CPU) simultaneously reads waveform data of a reference signal and a quadrature reference signal from memory. These reference and quadrature reference signals are continuous-time functions. The CPU discretizes these continuous-time functions according to the same acquisition times as the intensity sequence, obtaining discrete sequences of the reference and quadrature reference signals that correspond one-to-one with the time points of the intensity sequence. The CPU multiplies each pixel value in the intensity sequence with the corresponding value in the discrete sequence of the reference signal at the same time, obtaining a product sequence. The CPU sums all the product values ​​in the product sequence and divides the sum by the length of the product sequence to obtain the cosine component. The mathematical expression for the cosine component is: ,in Time in the intensity sequence pixel values, For the reference signal at time The instantaneous value, This represents the total number of frames in the intensity sequence. The central processing unit (CPU) multiplies each pixel value in the intensity sequence with the corresponding value in the discrete sequence of the orthogonal reference signal at the same time, obtaining a product sequence. The CPU then sums all the product values ​​in the product sequence and divides the sum by the length of the product sequence to obtain the sine component. The mathematical expression for the sine component is: ,in, Time in the intensity sequence pixel values, For the orthogonal reference signal at time The instantaneous value, This represents the total number of frames in the intensity sequence. The cosine component represents the response component in the intensity sequence that is in phase with the modulation signal of the perturbed light, while the sine component represents the response component in the intensity sequence that is 90 degrees out of phase with the modulation signal of the perturbed light. The central processing unit repeats the above quadrature demodulation operation on the intensity sequence at each pixel location, so that each pixel location obtains the corresponding cosine component value and sine component value.

[0111] The cosine component is the integral average of the product of the intensity sequence and the cosine function of the reference signal over a complete modulation period. This definition of the cosine component is from the perspective of continuous-time signals. Mathematically, it means that within a time interval equal to a complete modulation period, the cosine function of the intensity signal and the reference signal is multiplied moment by moment, integrated over time, and the result is divided by the length of the modulation period to obtain the cosine component. Integrating and averaging over a complete modulation period can eliminate noise and interference components in the intensity sequence that have a different modulation frequency than the perturbed light, ensuring that the cosine component contains only the response component with the same frequency as the perturbed light. In practical digital signal processing, the integral average is calculated by summing discretely and then dividing by the number of discrete points, which is the number of image frames acquired by the image acquisition module 14 within a complete modulation period.

[0112] The sinusoidal component is the integral average of the product of the intensity sequence and the sine function of the reference signal over a complete modulation period. This definition of the sinusoidal component is from the perspective of a continuous-time signal. Mathematically, it means that within a time interval equal to a complete modulation period, the sinusoidal function of the intensity signal and the reference signal are multiplied at each time step, integrated over time, and the result is divided by the length of the modulation period to obtain the sinusoidal component. Integrating and averaging over a complete modulation period can eliminate noise and interference components in the intensity sequence that have a different modulation frequency than the perturbed light, ensuring that the sinusoidal component contains only the response component with the same frequency as the perturbed light. In practical digital signal processing, the integral average is calculated by summing discretely and then dividing by the number of discrete points, which is the number of image frames acquired by the image acquisition module 14 within a complete modulation period.

[0113] The defect determination module 15 calculates the amplitude and phase of a pixel based on the cosine and sine components. The amplitude is the square root of the sum of the squares of the cosine and sine components. The central processing unit reads the cosine component value of a specific pixel location from memory. and sine component values Central processing unit (CPU) computing Calculate the square value of . The square value of , The square value and The squares of the values ​​are added together to obtain a sum of squares. The central processing unit (CPU) calculates the square root of this sum to obtain the amplitude at that pixel location. The mathematical expression for the amplitude is: The amplitude represents the intensity of the response signal with the same frequency as the perturbation light in the probe light response signal at that pixel location. A larger amplitude indicates a stronger change in the photoelectric properties at that pixel location. The phase is the inverse function of the tangent of the ratio of the sine component to the cosine component. The central processing unit reads the cosine component value of a specific pixel location from memory. and sine component values Central processing unit (CPU) computing Divide by The central processing unit (CPU) calculates the arctangent function value of the quotient to obtain the phase at that pixel location. The mathematical expression for the phase is: Phase represents the phase delay of the probe light response signal at that pixel location relative to the modulation signal of the perturbation light. The magnitude of the phase reflects the degree of time lag between the change in photoelectric properties at that pixel location and the excitation by the perturbation light. The central processing unit repeats the above amplitude and phase calculation operations for the cosine and sine component values ​​of each pixel location, so that each pixel location obtains the corresponding amplitude and phase values.

[0114] The defect determination module 15 performs the above processing on all pixels to obtain amplitude and phase maps. After the central processing unit (CPU) performs intensity sequence extraction, orthogonal demodulation, cosine and sine component calculation, amplitude and phase calculation for each pixel location, each pixel location has an amplitude value and a phase value. The CPU arranges the amplitude values ​​of all pixel locations into a two-dimensional matrix according to the horizontal and vertical coordinates of the pixel location; this two-dimensional matrix is ​​the amplitude map. The amplitude map is located at the horizontal coordinate... and vertical coordinates The value at a given point equals the amplitude at the corresponding pixel location. The central processing unit arranges the phase values ​​of all pixel locations into a two-dimensional matrix according to the horizontal and vertical coordinates of the pixel location; this two-dimensional matrix is ​​the phase map. The phase map is located at the horizontal coordinates... and vertical coordinates The value at a given point is equal to the phase at the corresponding pixel location. The size of the amplitude map and phase map is the same as the size of a single frame in the time-series optically corrected image. The amplitude map reflects the intensity distribution of the photoelectric response generated at various locations on the surface of the semiconductor sample 10 under periodic perturbation light, and the phase map reflects the distribution of the response time delay at various locations on the surface of the semiconductor sample 10.

[0115] The defect determination module 15 performs spatial filtering, noise suppression, and threshold segmentation on the amplitude map and phase map respectively to extract amplitude anomaly regions and phase anomaly regions. Spatial filtering involves low-pass filtering or median filtering to remove high-frequency noise and isolated noise points from the image. Low-pass filtering smooths the image by weighting the value of each pixel with the values ​​of its neighboring pixels. The weighting coefficients of the weighted average are distributed according to a Gaussian function, and the half-width and height of the Gaussian function are preset values. Median filtering removes isolated noise points by replacing the value of each pixel with the median value of all pixels in its neighborhood window. The size of the neighborhood window is a preset value; the neighborhood window is a square window with a side length of 3 pixels, 5 pixels, or 7 pixels. The central processing unit of the defect determination module 15 inputs the amplitude map into the spatial filtering unit, which performs spatial filtering on the amplitude map and outputs a filtered amplitude map. The central processing unit inputs the phase map into the spatial filtering unit, which performs spatial filtering on the phase map and outputs a filtered phase map. Threshold segmentation compares the value of each pixel in the image with a preset threshold, classifying pixels into normal and abnormal pixels based on the comparison result. The preset amplitude threshold is determined based on the statistical distribution of amplitude in the defect-free region; it equals the average amplitude of the defect-free region plus three standard deviations. The preset phase threshold is determined based on the statistical distribution of phase in the defect-free region; it equals the average phase of the defect-free region plus three standard deviations. The central processing unit (CPU) of the defect determination module 15 judges the position of each pixel in the filtered amplitude image. When the amplitude value of a pixel is greater than the preset amplitude threshold, that pixel is marked as an amplitude abnormal pixel, and the continuous region formed by all amplitude abnormal pixels constitutes the amplitude abnormal region. The CPU judges the position of each pixel in the filtered phase image. When the phase value of a pixel is greater than the preset phase threshold, that pixel is marked as a phase abnormal pixel, and the continuous region formed by all phase abnormal pixels constitutes the phase abnormal region. The CPU takes the absolute value of pixels with negative phase values ​​and compares it with the preset phase threshold, because the absolute value of the phase offset reflects the degree of response delay.

[0116] The defect determination module 15 marks the spatial intersection of the amplitude anomaly region and the phase anomaly region as the location of the defect. The spatial intersection region refers to the set of pixel positions that simultaneously belong to both the amplitude anomaly region and the phase anomaly region. The central processing unit (CPU) of the defect determination module 15 creates a blank binary image with the same size as the amplitude map, and the initial value of each pixel position in the blank binary image is 0. The CPU traverses all pixel positions. For each pixel position, the CPU determines whether the pixel position is simultaneously marked as an amplitude anomaly pixel and a phase anomaly pixel. When the pixel position satisfies both the amplitude anomaly condition and the phase anomaly condition, the CPU sets the value of the pixel position in the blank binary image to 1. The continuous region formed by all pixel positions with a value of 1 is the spatial intersection region of the amplitude anomaly region and the phase anomaly region. This spatial intersection region is marked by the defect determination module 15 as the location of the defect in the semiconductor sample 10 to be inspected. The CPU outputs a list of defect location coordinates, which contains the horizontal and vertical coordinates of each pixel position in the spatial intersection region. By using the spatial intersection of amplitude anomaly regions and phase anomaly regions to mark the defect location, information from both amplitude and phase anomalies can be utilized simultaneously, reducing misjudgments caused by single-dimensional anomalies and improving the accuracy of defect location.

[0117] The defect determination module 15 determines the defect type based on the amplitude and phase corresponding to the defect's location. The central processing unit (CPU) of the defect determination module 15 reads the horizontal and vertical coordinates of each defect pixel from the defect location coordinate list. Based on this pixel location, the CPU reads the corresponding amplitude value from the amplitude map and the corresponding phase value from the phase map. The CPU compares the read amplitude and phase values ​​with standard parameters in the defect type determination rule base. The defect type determination rule base is stored in the non-volatile flash memory of the defect determination module 15 and contains various correspondences between defect types and amplitude and phase value ranges. Point defects correspond to a high amplitude range and a small phase delay range. Vacancy defects correspond to a medium amplitude range and a medium phase delay range. Impurity aggregation defects correspond to a high amplitude range and a medium phase delay range. Dislocation defects correspond to a medium amplitude range and a small phase delay range. The amplitude range corresponding to grain boundary defects is low, and the phase range corresponding to grain boundary defects is large phase delay. The amplitude range corresponding to crack defects is low, and the phase range corresponding to crack defects is large phase delay. The amplitude range corresponding to local leakage channels is high, and the phase range corresponding to local leakage channels is large phase delay. The amplitude range corresponding to hotspot defects is medium, and the phase range corresponding to hotspot defects is large phase delay. The CPU compares the amplitude value of the defect pixel location with the amplitude value range corresponding to each defect type in the defect type determination rule library, and compares the phase value of the defect pixel location with the phase value range corresponding to each defect type in the defect type determination rule library. When both the amplitude value and phase value fall within the amplitude value range and phase value range corresponding to a certain defect type, the CPU determines that the defect type is the defect type of that defect pixel location. The CPU repeats the above comparison and determination operation for all defect pixel locations, so that each defect pixel location obtains a corresponding defect type. The CPU outputs the defect determination result. The resulting defect response map can reveal defects in the semiconductor sample 10 that are not obvious in ordinary static optical images.

[0118] The defect determination module 15 repeatedly acquires time-series optical images under different perturbation frequencies, perturbation powers, perturbation wavelengths, probe light wavelengths, polarization states, atmospheric pressures, liquid media compositions, solvent polarities, ionic strengths, sample temperatures, or applied electric fields. For each condition, it performs pixel-level demodulation to obtain amplitude and phase maps. The defect determination module 15 extracts the amplitude of each defect pixel location from the amplitude map under each condition and the phase of each defect pixel location from the phase map under each condition, forming a first response sequence of amplitude variation with the aforementioned conditions and a second response sequence of phase variation with the aforementioned conditions for each defect pixel location. For each pixel location, the defect determination module 15 can also calculate a normalized response, which is the ratio of amplitude to the static intensity of the imaging probe light, where the static intensity is the intensity value of the probe light response acquired by the image acquisition module 14 when not disturbed by the perturbation light. The normalized response is used to eliminate the influence of imaging probe light illumination inhomogeneity and sample surface reflectivity differences on the amplitude.

[0119] The defect determination module 15 distinguishes defect types based on the changing trends of the first response sequence and the second response sequence: when the amplitude shows a peak as the perturbation frequency changes and the phase changes with the perturbation frequency and the sign reverses, the defect type is determined to be a carrier capture and recombination dominated defect; when the amplitude increases linearly with the perturbation power and the phase does not change with the perturbation power, the defect type is determined to be a light absorption dominated defect; when the amplitude decreases monotonically as the perturbation frequency increases and the phase increases linearly as the perturbation frequency increases, the defect type is determined to be a thermal diffusion dominated defect; when the amplitude changes slowly with the perturbation frequency and the phase changes abruptly at a specific frequency, the defect type is determined to be a dielectric response dominated defect; when the amplitude changes significantly with the liquid medium composition, solvent polarity, or ionic strength, the defect type is determined to be an ion adsorption or migration dominated defect. The defect determination module 15 compares the defect type determined based on the multi-condition response sequence with the defect type determined based on the amplitude value range and phase value range. When the two determination results are consistent, the defect type is output. When the two determination results are inconsistent, the determination result with lower confidence is output and marked as a defect to be confirmed.

[0120] The defect determination module 15 is further configured to classify defects based on amplitude threshold, phase threshold, frequency response curve, spatial continuity, defect morphology, machine learning classification model, or a combination thereof. Spatial continuity refers to the degree to which defect pixels are continuously distributed in space, and is quantified by calculating the average distance between adjacent defect pixels in the defect region. Defect morphology refers to the geometric shape features of the defect region, including its area, perimeter, roundness, aspect ratio, and orientation angle. The machine learning classification model is a pre-trained classifier. Its input feature vector includes the amplitude value, phase value, frequency response curve feature parameters, spatial continuity index, and defect morphology parameters for each defect pixel location. The output of the machine learning classification model is the defect type label and corresponding confidence score for each defect pixel location. The machine learning classification model is trained using labeled data of known defect samples, which includes the defect pixel locations and their corresponding defect type labels. The defect types include at least one of the following: point defects, vacancies, impurity aggregates, dislocations, stacking faults, grain boundaries, cracks, localized oxidation defects, surface contamination, passivation layer defects, metal interconnect defects, localized leakage channels, hot spots, or subsurface defects.

[0121] The defect determination module 15 can also extract the second harmonic signal or multi-frequency response signal that is in the same frequency as the disturbing light, and obtain the real part, imaginary part, or time constant. The defect determination module 15 compares the above signal components with the background response of the defect-free area to output a defect response map.

[0122] The system also includes a test report output module 16, which is connected to the defect determination module 15. It receives the defect location and defect type output by the defect determination module 15 and generates a defect test report based on the defect location and defect type. The output format includes at least one of electronic document, image annotation and database record. At the same time, the test report output module 16 also generates a defect response map based on the defect determination result. The defect response map includes amplitude map, phase map, frequency response map and defect confidence map, and outputs alarm signal or classification result according to defect type.

[0123] Example 1, Defect Detection of Silicon Carbide Power Device Wafers: The sample to be tested is a silicon carbide power device wafer. The sample is fixed on the sample support stage 9, and air mode or vacuum mode is selected. The perturbation light uses a light source higher than or close to the absorption edge of the material to induce carrier generation, and the imaging probe light uses visible or near-infrared light that will not strongly excite the sample. The synchronization control module 1 modulates the perturbation light according to a preset frequency and triggers the image acquisition module 14 to acquire a time-series optical image of the probe light response. The defect determination module 15 performs synchronous demodulation on each pixel to obtain an amplitude map and a phase map. Due to carrier trapping and non-radiative recombination enhancement at dislocations, stacking faults, or local recombination centers, the local reflectivity, refractive index, or photothermal response differs from the surrounding area. Therefore, electrically active defects can be located through the demodulation map.

[0124] Example 2, Gallium Nitride Epitaxial Wafer or Micro-LED Epitaxial Wafer Detection: The sample to be tested is a gallium nitride epitaxial wafer or a micro-LED epitaxial wafer. A low-magnification objective lens 6 first performs a large-area scan of the sample to obtain the distribution of candidate defects; then, the objective lens switching mechanism 8 switches to a high-magnification objective lens 7 to re-test the candidate region. Non-radiative recombination, localized stress, and thermal accumulation at the defect site can cause coherent changes in the reflection or interference signals under periodic perturbations. Compared with traditional photoluminescence intensity maps, the amplitude and phase maps of this invention can further provide time delay information on the defect response, helping to distinguish between surface morphology defects and electrically active defects.

[0125] Example 3, Inspection of Silicon-based Integrated Circuit Wafers or Thin Film Structures: The sample to be inspected is a silicon-based integrated circuit wafer or thin film structure. For defects in metal interconnects, passivation layers, or dielectric layers, perturbation light can induce localized thermal expansion, changes in thermal reflectivity, or changes in dielectric function; imaging probe light reads these changes through reflection or interference. Since the thermal diffusion and localized absorption in defective regions differ from those in normal regions, their phase delay and frequency response curves can be used to identify localized hot spots, voids, cracks, or interlayer delamination.

[0126] Example 4, Detection of Two-Dimensional Semiconductor or Perovskite Semiconductor Thin Films: The sample to be tested is a two-dimensional semiconductor or perovskite semiconductor thin film. Grain boundaries, layer number variations, ion migration-related defects, localized trap states, or surface adsorption sites will exhibit different absorption, reflection, or dielectric responses under periodic optical perturbation. Low-magnification mode can be used for large-area thin film uniformity assessment, while high-magnification mode can be used for defect localization in individual grains, grain boundaries, or micro / nano device channels.

[0127] Example 5, Semiconductor Surface Defect Detection in a Liquid Phase Environment: The sample to be tested is placed in a liquid sample cell or flow cell. The liquid medium can be deionized water, salt solution, buffer solution, electrolyte solution, or alcohols, nitriles, carbonates, ethers, amides, ionic liquids, or mixed solvents. Periodic perturbation light can induce photogenerated carrier capture and release, interfacial charge accumulation, local absorption, reflectivity, dielectric constant, refractive index, or photothermal temperature changes at the defect site. Imaging probe light synchronously reads these changes and generates a defect response map through pixel-level demodulation. This example is applicable to the detection of semiconductor materials under electrochemical or photoelectrochemical working environments, including wet process residues, surface passivation defects, liquid wetting-induced trap states, ion adsorption or migration-related defects, and defects caused by wet process residues, surface passivation defects, liquid wetting-induced trap states, ion adsorption or migration-related defects, and defects caused by wet process residues.

[0128] The system of this invention includes a sample support and environmental control module, a low-to-high magnification switchable microscopic imaging module, an imaging probe light output module, a periodic perturbation light output module, an image acquisition module 14, a synchronization control module 1, and a defect determination module 15. The imaging probe light is used to acquire wide-field or local optical images of the semiconductor sample surface. The periodic perturbation light is used to apply optical perturbations of preset frequency, wavelength, and intensity to the semiconductor sample, causing optical responses in defect areas due to carrier capture and recombination, local absorption, reflectivity, dielectric constant, or temperature changes. The image acquisition module 14 acquires time-series images corresponding to the probe light. The defect determination module 15 performs phase-locked, Fourier, or correlation demodulation on the time series of each pixel to obtain amplitude maps, phase maps, frequency response maps, and defect confidence maps. This method can be performed in air, vacuum, or liquid environments, and can achieve rapid screening of large areas using a low-magnification objective lens 6, followed by fine imaging and localization of suspected defects using a high-magnification objective lens 7. This invention eliminates the need for fluorescent markers or electrical contacts, enabling highly sensitive, non-contact, spatiotemporally resolved detection of defects, impurities, dislocations, grain boundaries, surface states, and localized photothermal anomalies in wafers, epitaxial wafers, semiconductor thin films, two-dimensional semiconductors, and micro / nano devices. This facilitates failure analysis and process quality control, and can be integrated into semiconductor process inspection equipment for wafer incoming inspection, post-epitaxy growth inspection, post-ion implantation inspection, post-annealing inspection, post-etching inspection, post-thin film deposition inspection, passivation layer inspection, device failure localization, or process window screening. In essence, this invention does not merely acquire ordinary images of samples under illumination, nor does it rely on sample emission. Instead, it images the sample using an imaging probe light, and a perturbation light periodically disturbs the sample, extracting the weak dynamic response synchronized with the perturbation light. Even if defects are invisible in ordinary bright-field images, they can still be identified by this invention as long as their dynamic response differs from the normal area.

[0129] like Figure 2As shown, an embodiment of the present invention provides a semiconductor defect detection method based on optically induced perturbation impedance imaging, employing any of the aforementioned semiconductor defect detection systems based on optically induced perturbation impedance imaging. The method includes: S1. Place the semiconductor sample 10 to be tested in the sample carrier and environmental control module, and place the semiconductor sample 10 to be tested in an air environment, a vacuum environment or a liquid phase environment. S2. By activating the imaging probe light output module, an imaging probe light is output to the semiconductor sample 10 to be tested, and the imaging probe light forms a probe light response on the semiconductor sample 10 to be tested. S3. Activate the periodic perturbation light output module to output periodically modulated perturbation light to the semiconductor sample 10 to be tested. The perturbation light is used to cause changes in the photoelectric properties of the defect region in the semiconductor sample 10 to be tested. S4. The synchronization of the periodic perturbation light output module and the image acquisition module 14 is controlled by the synchronization control module 1, so that the image acquisition module 14 acquires a time-series optical image of the probe light response during the period when the perturbation light applies a perturbation to the semiconductor sample 10 to be tested. S5. Receive time-series optical images through the defect determination module 15, perform pixel-level demodulation on the time-series optical images, extract the signal components of each pixel or each region of interest that are at the same frequency as the perturbed light or have a fixed phase relationship with the perturbed light, so as to obtain amplitude maps and phase maps, and determine the location and type of defects in the semiconductor sample 10 to be detected based on the amplitude maps and phase maps.

[0130] like Figure 3As shown, the semiconductor defect detection method of this embodiment of the invention first involves an operator fixing the semiconductor sample 10 to be tested onto the sample support stage 9, and selecting an air environment, a vacuum environment, or a liquid environment according to the detection requirements. The sample support stage 9 is disposed within the sample chamber 11, which is a sealed cavity. It can be connected to a vacuum pump to form a vacuum environment, or connected to a gas inlet to introduce gas to form an air environment. A liquid sample cell or a flow cell is also provided within the sample chamber 11 to form a liquid environment. After sample loading and environment selection are completed, the detection light source 4 in the imaging detection light output module outputs imaging detection light. The imaging detection light is guided through the beam combining and splitting optical path 5 to the low-magnification objective lens 6 or the high-magnification objective lens 7 in the microscopic imaging module, and after being focused by the objective lens, it illuminates the surface of the semiconductor sample 10 to be tested. The imaging probe light undergoes reflection, transmission, scattering, interference, or polarization responses on the surface of the semiconductor sample 10 to form a probe light response. The beam combining / splitting optical path 5 guides the probe light response to the image acquisition module 14. When the periodic perturbation light output module is off, multiple frames of optical images of the probe light response are acquired as the initial optical image. Subsequently, the periodic perturbation light source 2 in the periodic perturbation light output module outputs a beam, which is periodically modulated by the modulator 3 to form perturbation light. The perturbation light is irradiated onto the surface of the semiconductor sample 10 via the beam combining / splitting optical path 5 and the microscopic imaging module, causing changes in the photoelectric properties of the defect region. The synchronization control module 1 receives a reference signal output by the modulator 3, the frequency and phase of which are consistent with the modulation frequency and phase of the perturbation light. The synchronization control module 1 generates a trigger signal based on the reference signal and outputs the trigger signal to the image acquisition module 14. Upon receiving the trigger signal, the image acquisition module 14 continuously acquires time-series optical images of the probe light response at a preset frame rate throughout the entire process of the perturbation light applying perturbation to the semiconductor sample 10. The image acquisition module 14 transmits the acquired time-series optical images to the defect determination module 15. The defect determination module 15 (also known as the synchronous trigger-phase-locked FFT demodulation module) performs pixel-level demodulation on the time-series optical images. The defect determination module 15 first acquires the corresponding corrected time-series optical images based on the environmental type of the semiconductor sample 10 under test. When the semiconductor sample 10 is in an air environment, the defect determination module 15 performs air background subtraction and flat-field correction on the time-series optical images to obtain a first time-series optically corrected image. When the semiconductor sample 10 is in a vacuum environment, the defect determination module 15 performs vacuum background subtraction and flat-field correction on the time-series optical images to obtain a second time-series optically corrected image. When the semiconductor sample 10 is in a liquid phase environment, the defect determination module 15 performs liquid phase background subtraction and flat-field correction on the time-series optical images to obtain a third time-series optically corrected image.When the defect determination module 15 performs pixel-level demodulation on the corrected time-series optical image, for each pixel, it extracts the intensity sequence of that pixel in the time-series optical image as it changes over time. Using the modulation signal of the perturbed light as a reference signal, it performs orthogonal demodulation on the intensity sequence to extract the cosine and sine components with the same frequency as the perturbed light. The defect determination module 15 calculates the amplitude and phase of the pixel based on the cosine and sine components. It performs the above processing on all pixels to obtain amplitude and phase maps. The defect determination module 15 performs spatial filtering and threshold segmentation on the amplitude and phase maps respectively to extract amplitude and phase anomaly regions. The spatial intersection of the amplitude and phase anomaly regions is marked as the location of the defect. The defect determination module 15 determines the type of defect based on the amplitude and phase corresponding to the defect location. Finally, the defect determination module 15 outputs the defect determination result, which includes the defect location coordinates, the amplitude value of the defect pixel location, the phase value of the defect pixel location, and the defect type of the defect pixel location. The inspection report output module 16 generates a defect response diagram based on the judgment result of the defect judgment module 15. The defect response diagram includes an amplitude diagram, a phase diagram, a frequency response diagram, and a defect confidence diagram. It also outputs an alarm signal or classification result according to the defect type.

[0131] like Figure 4As shown, the process begins with low-magnification imaging scanning. The low-magnification objective lens 6 in the microscopic imaging module performs a large-area scan of the semiconductor sample 10 under test. The motorized stage 12 drives the sample stage 9 to move horizontally along a preset scanning path. The image acquisition module 14 acquires time-series optical images of the probe light response at each scanning position. After completing the large-area scan, the defect determination module 15 performs background subtraction and flat-field correction on the acquired time-series optical images. Depending on whether the semiconductor sample 10 is in an air, vacuum, or liquid environment, air background subtraction, vacuum background subtraction, or liquid background subtraction is performed respectively to eliminate interference from static background signals and dark current noise. Simultaneously, flat-field correction eliminates the effects of illumination non-uniformity and optical element transmittance non-uniformity. After background subtraction and flat-field correction, the defect determination module 15 performs pixel-level demodulation on the corrected time-series optical images. For each pixel, the intensity sequence of that pixel over time is extracted, and orthogonal demodulation is performed using the modulation signal of the perturbed light as a reference signal to obtain the amplitude and phase of each pixel, thereby acquiring the amplitude map and phase map. The defect determination module 15 performs threshold segmentation on the amplitude map and phase map respectively, marking pixels with amplitude values ​​greater than a preset amplitude threshold as amplitude abnormal pixels and pixels with phase values ​​greater than a preset phase threshold as phase abnormal pixels. A continuous region composed of all amplitude abnormal pixels is called an amplitude abnormal region, and a continuous region composed of all phase abnormal pixels is called a phase abnormal region. The defect determination module 15 marks the spatial intersection of the amplitude abnormal region and the phase abnormal region as a candidate defect and outputs the coordinates of the candidate defect. The synchronous control module 1 controls the electric displacement stage 12 to move the semiconductor sample 10 to be tested on the sample support stage 9 to the candidate defect coordinate position, and simultaneously controls the objective lens switching mechanism 8 to switch to the high-magnification objective lens 7. After the objective lens switching is completed, the autofocus module 13 controls the objective lens switching mechanism 8 or the sample support and environment control module according to the image clarity feedback of the probe light response to adjust the relative distance between the semiconductor sample 10 to be tested and the high-magnification objective lens 7 to ensure image clarity. After autofocusing is completed, the high-magnification objective lens 7 is used to re-measure the coordinates of the candidate defect. Time-series optical images of the probe light response are acquired again, and the background subtraction, flat-field correction, pixel-level demodulation, and threshold segmentation processes are repeated to obtain amplitude and phase maps under high-magnification imaging. The defect determination module 15 confirms the defect based on the amplitude and phase maps obtained from the high-magnification re-measurement, determining the precise location and size of the defect, and identifying the defect type based on the amplitude and phase corresponding to the defect location. Finally, the inspection report output module 16 generates an inspection report, which includes the defect's location coordinates, amplitude, phase, and defect type information, and outputs an alarm signal or classification result based on the defect type.

[0132] It should be noted that the beneficial effects of the semiconductor defect detection method based on optically induced perturbation impedance imaging provided in the above embodiments are the same as the beneficial effects of the semiconductor defect detection system based on optically induced perturbation impedance imaging provided in the above embodiments. In addition, the method and system embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0133] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0134] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0135] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A semiconductor defect detection system based on optically induced perturbation impedance imaging, characterized in that, include: The system includes a sample support and environmental control module, a microscopic imaging module, an imaging detection light output module, a periodic perturbation light output module, an image acquisition module, a synchronization control module, and a defect determination module. The sample carrying and environmental control module is used to carry the semiconductor sample to be tested and to place the semiconductor sample to be tested in an air environment, a vacuum environment, or a liquid phase environment. The imaging probe light output module is used to output imaging probe light to the semiconductor sample to be tested, and the imaging probe light forms a probe light response on the semiconductor sample to be tested. The periodic perturbation light output module is used to output periodically modulated perturbation light to the semiconductor sample to be tested, and the perturbation light is used to cause changes in the photoelectric properties of the defect region in the semiconductor sample to be tested. The microscopic imaging module is used to image the probe light response onto the image acquisition module; The image acquisition module is used to acquire a time-series optical image of the probe light response during the period when the perturbation light applies a perturbation to the semiconductor sample to be detected; The synchronization control module is used to control the synchronization of the periodic perturbation light output module and the image acquisition module; The defect determination module is connected to the image acquisition module and is used to receive the time-series optical image, perform pixel-level demodulation on the time-series optical image, extract the signal components of each pixel or each region of interest that are in the same frequency as the perturbation light or have a fixed phase relationship with the perturbation light, so as to obtain an amplitude map and a phase map, and determine the location and type of defects in the semiconductor sample to be detected based on the amplitude map and the phase map.

2. The semiconductor defect detection system based on optically induced perturbation impedance imaging according to claim 1, characterized in that, The sample carrying and environmental control module includes a sample carrying stage, a sample chamber, an electric displacement stage, and a temperature control unit. The sample carrying stage is disposed within the sample chamber and is used to carry the semiconductor sample to be tested. The electric displacement stage is connected to the sample carrying stage and is used to drive the sample carrying stage to move horizontally and vertically. The temperature control unit is connected to the sample carrying stage and is used to adjust the temperature of the semiconductor sample to be tested. The sample chamber is a sealed cavity, and a vacuum pump is connected to the sample chamber to create a vacuum environment within the sample chamber. The sample chamber is also connected to a gas inlet to introduce gas into the sample chamber to create an air environment. A liquid sample is also disposed within the sample chamber. The liquid sample cell or flow cell is used to contain a liquid medium so that the semiconductor sample to be tested is immersed in the liquid medium to form a liquid environment. The liquid sample cell or flow cell is equipped with a transparent window, a seal, an inlet and outlet, and a flow rate control unit. The transparent window allows the imaging detection light and the disturbance light to pass through. The seal is disposed between the transparent window and the sample cavity to ensure the sealing of the sample cavity. The inlet and outlet are connected to the liquid sample cell or flow cell and are used to introduce or discharge the liquid medium. The flow rate control unit is connected to the inlet and outlet and is used to control the flow rate of the liquid medium in the liquid sample cell or flow cell.

3. The semiconductor defect detection system based on optically induced perturbation impedance imaging according to claim 1, characterized in that, The imaging probe light output module includes a probe light source and a beam combining and splitting optical path. The probe light source is a continuous laser light source, a pulsed laser light source, a light-emitting diode light source, a supercontinuum light source, or a broadband light source. The beam combining and splitting optical path is disposed on the output optical path of the probe light source. The beam combining and splitting optical path guides the imaging probe light to the microscopic imaging module. The imaging probe light is irradiated onto the surface of the semiconductor sample to be tested by the microscopic imaging module. The imaging probe light undergoes reflection, transmission, scattering, interference, or polarization response on the surface of the semiconductor sample to be tested to form the probe light response. The beam combining and splitting optical path also guides the probe light response to the image acquisition module. The wavelength of the imaging probe light is different from the wavelength of the perturbation light.

4. The semiconductor defect detection system based on optically induced perturbation impedance imaging according to claim 1, characterized in that, The periodic perturbation light output module includes a periodic perturbation light source and a modulator. The output end of the periodic perturbation light source is connected to the input end of the modulator. The modulator receives the light beam output by the periodic perturbation light source and periodically modulates the light beam to output the perturbation light. The perturbation light is irradiated onto the surface of the semiconductor sample to be tested by the microscopic imaging module. The modulator is an acousto-optic modulator, an electro-optic modulator, a mechanical chopper, or a function generator. The modulation waveform of the perturbation light is a sine wave, a square wave, a triangular wave, a pulse wave, or a pseudo-random sequence. The wavelength of the perturbation light is located in the ultraviolet band, the visible band, the near-infrared band, or the mid-infrared band.

5. A semiconductor defect detection system based on optically induced perturbation impedance imaging according to claim 1, characterized in that, The microscopic imaging module includes a low-magnification objective lens, a high-magnification objective lens, an objective lens switching mechanism, and an autofocus module. The low-magnification objective lens and the high-magnification objective lens are respectively fixedly mounted on the objective lens switching mechanism. The objective lens switching mechanism drives the low-magnification objective lens and the high-magnification objective lens to switch so that the imaging probe light and the disturbance light are irradiated onto the surface of the semiconductor sample to be tested through the low-magnification objective lens or the high-magnification objective lens. The probe light response is returned to the image acquisition module through the low-magnification objective lens or the high-magnification objective lens. The autofocus module is connected to the objective lens switching mechanism and controls the objective lens switching mechanism or the sample support and environment control module according to the image sharpness feedback of the probe light response to adjust the relative distance between the semiconductor sample to be tested and the low-magnification objective lens or the high-magnification objective lens.

6. A semiconductor defect detection system based on optically induced perturbation impedance imaging according to claim 1, characterized in that, The image acquisition module is an area array camera, a line array camera, an avalanche photodiode array, a photomultiplier tube array, an infrared camera, or a high-speed camera. The image acquisition module is connected to the synchronization control module and receives the trigger signal output by the synchronization control module. The image acquisition module continuously acquires time-series optical images of the probe light response at a preset frame rate throughout the entire process of the perturbation light applying a perturbation to the semiconductor sample to be detected, according to the trigger signal. The preset frame rate is at least twice the modulation frequency of the perturbation light.

7. A semiconductor defect detection system based on optically induced perturbation impedance imaging according to claim 2, characterized in that, When the semiconductor sample to be tested is in the air environment, the defect determination module performs air background subtraction and flat field correction on the time-series optical image to obtain a first time-series optically corrected image; when the semiconductor sample to be tested is in the vacuum environment, the defect determination module performs vacuum background subtraction and flat field correction on the time-series optical image to obtain a second time-series optically corrected image; when the semiconductor sample to be tested is in the liquid phase environment, the defect determination module performs liquid phase background subtraction and flat field correction on the time-series optical image to obtain a third time-series optically corrected image; the defect determination module then performs pixel-level demodulation on the first time-series optically corrected image, the second time-series optically corrected image, or the third time-series optically corrected image.

8. A semiconductor defect detection system based on optically induced perturbation impedance imaging according to claim 7, characterized in that, The process by which the defect determination module obtains the first time-series optically corrected image includes: The defect determination module acquires multiple frames of air background images acquired by the image acquisition module when the periodic perturbation light output module is in the off state, and performs pixel-level averaging on the multiple frames of air background images to obtain an average air background image. Then, it subtracts the pixel value of the corresponding position in the average air background image from each frame of the time-series optical image pixel by pixel. The module also acquires multiple frames of uniform illumination images acquired by the image acquisition module when the periodic perturbation light output module is in the on state and the semiconductor sample to be tested is not loaded, and performs pixel-level averaging on the multiple frames of uniform illumination images to obtain a flat-field correction reference image. Finally, it divides each frame of the image after air background subtraction pixel by the pixel value of the corresponding position in the flat-field correction reference image and performs normalization processing to obtain the first time-series optical correction image. The process by which the defect determination module obtains the second time-series optically corrected image includes: The defect determination module acquires multiple frames of vacuum background images acquired by the image acquisition module when the sample cavity is in a vacuum state and the periodic perturbation light output module is in a closed state, and performs pixel-level averaging on the multiple frames of vacuum background images to obtain a vacuum background average image. Then, it subtracts the pixel value of the corresponding position in the vacuum background average image from each frame of the time-series optical image pixel by pixel. The module also acquires multiple frames of uniform illumination images acquired by the image acquisition module when the sample cavity is in a vacuum state, the periodic perturbation light output module is in a closed state, and the semiconductor sample to be tested is not loaded, and performs pixel-level averaging on the multiple frames of uniform illumination images to obtain a flat-field correction reference image. Finally, it divides each frame of the image after vacuum background subtraction pixel by the pixel value of the corresponding position in the flat-field correction reference image and performs normalization processing to obtain the second time-series optical correction image. The process by which the defect determination module obtains the third time-series optically corrected image includes: The defect determination module acquires multiple frames of liquid background images acquired by the image acquisition module when the liquid sample cell or the flow cell is filled with the liquid medium, the semiconductor sample to be tested is not immersed in the liquid medium, and the periodic perturbation light output module is in the off state. The module performs pixel-level averaging on these multiple frames of liquid background images to obtain an average liquid background image. Each frame of the time-series optical image is subtracted pixel-by-pixel from the pixel value of the corresponding position in the average liquid background image. The module also acquires multiple frames of uniform illumination images acquired by the image acquisition module when the liquid sample cell or the flow cell is filled with the liquid medium, the semiconductor sample to be tested is not immersed in the liquid medium, and the periodic perturbation light output module is in the on state. The module performs pixel-level averaging on these multiple frames of uniform illumination images to obtain a flat-field correction reference image. Each frame of the image after liquid background subtraction is divided pixel-by-pixel by the pixel value of the corresponding position in the flat-field correction reference image and normalized to obtain the third time-series optical correction image.

9. A semiconductor defect detection system based on optically induced perturbation impedance imaging according to claim 8, characterized in that, When the defect determination module performs pixel-level demodulation on the first time-series optically corrected image, the second time-series optically corrected image, or the third time-series optically corrected image, for each pixel, it extracts the intensity sequence of that pixel changing over time in the first time-series optically corrected image, the second time-series optically corrected image, or the third time-series optically corrected image. Using the modulation signal of the perturbation light as a reference signal, it performs orthogonal demodulation on the intensity sequence to extract the cosine component and sine component with the same frequency as the perturbation light. The cosine component is the integral average value of the product of the intensity sequence and the cosine function of the reference signal over a complete modulation period. The sine component is the integral average value of the product of the intensity sequence and the sine function of the reference signal over the complete modulation period. The defect determination module calculates the amplitude and phase of the pixel based on the cosine component and the sine component. The amplitude is the square root of the sum of the squares of the cosine component and the squares of the sine component. The phase is the inverse function of the tangent function of the ratio of the sine component to the cosine component. The defect determination module performs the above processing on all pixels to obtain the amplitude map and the phase map, and performs spatial filtering and threshold segmentation on the amplitude map and the phase map to extract amplitude abnormal regions and phase abnormal regions, marks the spatial intersection region of the amplitude abnormal region and the phase abnormal region as the location of the defect, and determines the type of the defect based on the amplitude and phase corresponding to the location of the defect.

10. A semiconductor defect detection method based on optically induced perturbation impedance imaging, characterized in that, The semiconductor defect detection system based on optically induced perturbation impedance imaging according to any one of claims 1 to 9 includes the following method: The semiconductor sample to be tested is placed in the sample carrier and environmental control module, and the semiconductor sample to be tested is placed in an air environment, a vacuum environment, or a liquid phase environment. By activating the imaging probe light output module, an imaging probe light is output to the semiconductor sample to be tested, and the imaging probe light forms a probe light response on the semiconductor sample to be tested. The periodic perturbation light output module is activated to output periodically modulated perturbation light to the semiconductor sample under test. The perturbation light is used to cause changes in the photoelectric properties of defect regions in the semiconductor sample under test. The synchronization control module controls the synchronization of the periodic perturbation light output module and the image acquisition module, so that the image acquisition module acquires a time-series optical image of the probe light response during the period when the perturbation light applies a perturbation to the semiconductor sample to be detected; The defect determination module receives the time-series optical image and performs pixel-level demodulation on the time-series optical image. It extracts the signal components of each pixel or each region of interest that are in the same frequency as the perturbation light or have a fixed phase relationship with the perturbation light to obtain an amplitude map and a phase map. Based on the amplitude map and the phase map, it determines the location and type of defects in the semiconductor sample to be detected.

Images