Method and apparatus for detecting semiconductor wafer bow

Abstract

The application relates to the technical field of semiconductor detection, and provides a semiconductor wafer warping degree detection method and device.The method comprises the following steps: emitting a laser beam to an observation window of a semiconductor equipment reaction cavity; making part of the laser beam partially reflect through the observation window to form a reference reflection light beam, and making another part of the laser beam transmit to the surface of a wafer in the semiconductor equipment reaction cavity, and then reflecting the wafer surface to form a main reflection light beam; acquiring a plurality of light spots formed by the main reflection light beam and the reference reflection light beam; identifying a main light spot formed by the main reflection light beam and a reference light spot formed by the reference reflection light beam according to the motion characteristics of the plurality of light spots; compensating the displacement amount of the main light spot based on the displacement amount of the reference light spot relative to the initial reference position of the reference light spot; and calculating the wafer warping degree according to the displacement amount of the compensated main light spot.The application is used for non-contact, online and real-time warping degree measurement of the wafer in the reaction cavity in the semiconductor process.

Description

Technical Field

[0001] This invention relates to the field of semiconductor testing technology, and in particular to a method and apparatus for detecting the warpage of semiconductor wafers. Background Technology

[0002] As semiconductor manufacturing processes evolve towards smaller feature sizes, larger wafer diameters, and more complex three-dimensional structures, wafer flatness and stress control become crucial. In high-temperature processes (such as epitaxial growth, thermal oxidation, and annealing), significant thermal stress is generated within the wafer due to the difference in thermal expansion coefficients between the silicon wafer and the thin film material, as well as the non-uniformity of the process temperature. This leads to warp or bowing. Wafer warp not only reduces film thickness and doping uniformity but also severely affects the alignment accuracy of subsequent photolithography processes, and can even cause chip performance degradation, reduced reliability, and overall yield loss. Therefore, real-time, online monitoring of wafer warp during the process is a key step in achieving process control and yield improvement.

[0003] Existing technologies for measuring wafer warpage offer various solutions, primarily categorized into contact and non-contact methods. Contact methods typically employ capacitive sensors or mechanical probes to directly contact the wafer surface. However, within semiconductor process cavities (such as chemical vapor deposition, oxidation furnaces, or rapid thermal annealing equipment), high temperatures, corrosive gases, or plasma environments can severely damage contact sensors. Furthermore, their introduction can lead to particulate or metal contamination risks, affecting process purity and device reliability, thus limiting their application in practical high-temperature process in-situ monitoring. Non-contact optical methods, due to their lack of physical contact and rapid response, have become the mainstream approach. Common techniques include laser triangulation, laser interferometers, and deformation analysis based on moiré fringes or digital image correlation. These methods typically place the optical measurement system outside the process cavity, probing the wafer surface through a viewport. However, such solutions still face the following significant challenges in practical applications: First, common-mode noise caused by environmental disturbances. Semiconductor manufacturing equipment inevitably experiences mechanical vibrations (e.g., pump and valve operation), thermal drift (minor displacements of the observation window or optical components due to temperature changes), and micro-motions of optical components during operation. These environmental disturbances can cause overall offset or jitter in the measurement optical path, introducing common-mode noise into the probe signal. Since the wafer warpage signal itself is a minute change, it is easily submerged in this type of noise, leading to distorted measurement results and poor stability and repeatability. Secondly, there is interference from stray reflection signals. The measurement optical path typically passes through the observation window of the process cavity, which is composed of multiple layers of glass or quartz. Light undergoes multiple reflections at the two interfaces of the observation window and at any contamination layers that may be present, forming fixed stray reflection signals mixed with the primary reflection light from the wafer. If the signal processing algorithm fails to effectively distinguish between the primary reflection signal from the moving wafer surface and the reference reflection signal from the fixed interface, it may mistakenly treat the stray signal as a valid signal, thus significantly reducing the accuracy of the warpage measurement.

[0004] In summary, there is an urgent need to develop a new method and device for detecting the warpage of semiconductor wafers to overcome the aforementioned limitations and provide a reliable in-situ monitoring method for advanced semiconductor manufacturing. Summary of the Invention

[0005] This invention provides a method and apparatus for detecting the warpage of semiconductor wafers. This invention is used to perform non-contact, online, real-time warpage measurement of wafers in the reaction chamber during semiconductor processing.

[0006] According to a first aspect of the present invention, a method for detecting the warpage of a semiconductor wafer is provided, comprising: emitting a laser beam into an observation window of a reaction cavity of a semiconductor device; causing a portion of the laser beam to be partially reflected by the observation window to form a reference reflection beam, and another portion of the laser beam to be transmitted to the surface of a wafer in the reaction cavity of the semiconductor device, and then reflected by the surface of the wafer to form a main reflection beam; acquiring a plurality of light spots formed by the main reflection beam and the reference reflection beam; identifying, based on the motion characteristics of the plurality of light spots, a main light spot formed by the main reflection beam and a reference light spot formed by the reference reflection beam; compensating for the displacement of the main light spot based on the displacement of the reference light spot relative to its initial reference position; and calculating the wafer warpage based on the displacement of the compensated main light spot.

[0007] In one embodiment, identifying a main light spot formed by a main reflected beam and a reference light spot formed by a reference reflected beam based on the motion characteristics of multiple light spots includes: setting a displacement amplitude threshold; acquiring the displacement of multiple light spots during wafer rotation; identifying light spots with displacement values ​​higher than the displacement amplitude threshold as main light spots, and identifying light spots with displacement values ​​lower than the displacement amplitude threshold as reference light spots.

[0008] In one embodiment, identifying a main spot formed by a main reflected beam and a reference spot formed by a reference reflected beam based on the motion characteristics of multiple light spots includes: acquiring displacement signals of multiple light spots and performing spectral analysis to obtain a displacement spectrum; identifying light spots whose displacement spectrum contains frequency components consistent with the rotation frequency of the wafer as main light spots; and identifying light spots whose displacement spectrum does not contain frequency components as reference light spots.

[0009] In one embodiment, identifying a main spot formed by a primary reflected beam and a reference spot formed by a reference reflected beam based on the motion characteristics of multiple light spots includes: setting a displacement amplitude threshold; acquiring the displacement of multiple light spots during wafer rotation; acquiring the displacement signals of multiple light spots and performing spectral analysis to obtain a displacement spectrum; identifying light spots with displacement values ​​higher than the displacement amplitude threshold and whose displacement spectrum contains frequency components consistent with the wafer rotation frequency as main light spots; and identifying light spots with displacement values ​​lower than the displacement amplitude threshold and whose displacement spectrum does not contain frequency components as reference light spots.

[0010] In one implementation, identifying the main light spot and the reference light spot based on the motion characteristics of multiple light spots includes: calling a machine learning model or classification algorithm to divide the coordinates of the multiple light spots into a first coordinate set and a second coordinate set; placing the coordinates of the light spots that satisfy periodic motion into the first coordinate set as the historical coordinate set of the main light spot; and placing the coordinates of the light spots that are stationary or do not satisfy periodic motion into the second coordinate set as the historical coordinate set of the reference light spot.

[0011] In one implementation, the displacement of the compensated main spot (ΔX) main_corrected ΔY main_corrected )satisfy:

[0012]

[0013]

[0014] Among them, (X) main Y main ) is the current coordinate of the main light spot; (X) main0 Y main0 ) is the coordinate of the initial reference position of the main light spot; (ΔX) ref ΔY ref Let be the reference drift amount of the reference spot, satisfying:

[0015]

[0016]

[0017] Among them, (X) ref Y ref (X) represents the current coordinates of the reference spot; ref0 Y ref0 ) represents the coordinates of the initial reference position of the reference spot.

[0018] In one embodiment, when another portion of the laser is transmitted to the surface of the wafer, a current measurement point is formed on the surface of the wafer; the wafer warpage is calculated based on the displacement of the compensated main spot, including: substituting the displacement of the compensated main spot into the warpage calculation model to obtain the wafer warpage K, satisfying:

[0019]

[0020] Where C is the calibration coefficient, Φ is the azimuth angle of the current measurement point relative to the center point of the wafer, Φ0 is the initial value of the azimuth angle, and the azimuth angle Φ of the current measurement point satisfies δ is the preset minimum angle threshold; L0 is the distance from the current measurement point to the axis of rotation.

[0021] In one embodiment, the method further includes: real-time monitoring of the signal quality of the main spot; when the signal quality of the main spot is lower than a preset reliability threshold, generating a replacement displacement based on the historical displacement data of the reference spot and a preset process model, and calculating the wafer warpage based on the replacement displacement.

[0022] According to a second aspect of the present invention, an apparatus for detecting the warpage of a semiconductor wafer is provided, comprising the method of any one of the first aspects. The apparatus includes: a laser emitting unit for emitting a laser beam toward an observation window of a reaction cavity of a semiconductor device; causing a portion of the laser beam to be partially reflected by the observation window to form a reference reflection beam, and another portion of the laser beam to be transmitted to the surface of a wafer in the reaction cavity of the semiconductor device, and then reflected by the surface of the wafer to form a main reflection beam; a position sensing unit for acquiring a plurality of light spots formed by the main reflection beam and the reference reflection beam; a light spot identification unit for identifying, based on the motion characteristics of the plurality of light spots, a main light spot formed by the main reflection beam and a reference light spot formed by the reference reflection beam; a displacement compensation unit for compensating the displacement of the main light spot based on the displacement of the reference light spot relative to its initial reference position; and a calculation unit for calculating the wafer warpage based on the compensated displacement of the main light spot.

[0023] In one embodiment, the laser emitting unit and the position sensing unit are coaxially arranged so that the main reflected beam and the reference reflected beam return along the incident optical path and are received by the position sensing unit.

[0024] In one embodiment, the device further includes a beam splitting unit disposed between the laser emitting unit and the observation window; the laser emitting unit is used to emit laser light along a first direction; the laser light emitted along the first direction is transmitted through the beam splitting unit to form a main transmitted beam; the main transmitted beam is transmitted through the observation window along the first direction to the wafer surface; the main reflected beam formed by reflection from the wafer surface is transmitted in the opposite direction of the first direction to the position sensing unit to form a main spot; the laser light emitted along the first direction is reflected by the observation window to form a reference reflected beam; the reference reflected beam is transmitted in the opposite direction of the first direction to the position sensing unit to form a reference spot; the main spot and the reference spot do not overlap.

[0025] In one embodiment, the device further includes a focusing lens disposed between the beam splitting unit and the position sensing unit, for focusing the returned laser beam onto the photosensitive surface of the position sensing unit to form a main spot and a reference spot.

[0026] In one embodiment, the position sensing unit includes at least one of a position-sensitive detector, an area array charge-coupled device (CCD) image sensor camera, and an area array complementary metal-oxide-semiconductor (CMOS) image sensor camera.

[0027] In one embodiment, the laser emitting unit and the position sensing unit are fixed to the outside of the semiconductor device's reaction cavity or encapsulated in the same housing, so that the optical axis of the laser forms a preset angle with the observation window.

[0028] In one embodiment, the device further includes an evaluation module for real-time monitoring of the signal quality of the main spot; when the signal quality of the main spot is lower than a preset reliability threshold, the calculation unit generates an alternative displacement based on the historical displacement data of the reference spot and a preset process model, and calculates the wafer warpage based on the alternative displacement.

[0029] According to a third aspect of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory is used to store a computer program executable by the processor; and the processor is used to execute the computer program in the memory to implement the method described above.

[0030] According to a fourth aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, enables the implementation of the above-described method.

[0031] According to a fifth aspect of the present invention, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method described above.

[0032] Compared with existing technologies, the advantages of this invention are as follows: This invention emits a laser beam into the observation window of the semiconductor device's reaction cavity, utilizing partial reflection from the front surface of the observation window to form a reference reflected beam, while simultaneously allowing the transmitted portion to irradiate the wafer surface and form the main reflected beam. This allows for the simultaneous acquisition of both the main beam and the reference beam on the position sensing unit. Since the observation window is fixed to the reaction cavity, the reference beam generated by its reflection primarily reflects external environmental disturbances, such as common-mode displacement caused by mechanical vibration, thermal drift, or micro-motions in the optical system. The displacement of the main beam, however, is superimposed with the wafer warpage deformation signal and the aforementioned common-mode disturbances. By analyzing the motion characteristics of multiple beams during wafer rotation, the main beam and the reference beam can be accurately distinguished, and the original displacement of the main beam can be compensated using the displacement of the reference beam relative to the initial reference position, effectively eliminating the influence of common-mode noise. Based on this, the wafer warpage is calculated according to the compensated displacement of the main beam, improving the accuracy and stability of the measurement results. This invention eliminates the need to install sensors inside the reaction chamber, avoiding interference with the process environment. It also enables online detection of wafer warpage under harsh conditions such as high temperature, vacuum, or plasma, making it suitable for real-time monitoring needs in semiconductor manufacturing processes. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating a method for detecting the warpage of a semiconductor wafer according to an exemplary embodiment.

[0034] Figure 2 This is a schematic diagram of an apparatus for detecting the warpage of a semiconductor wafer, according to an exemplary embodiment.

[0035] Figure 3 This is a block diagram illustrating an electronic device according to an exemplary embodiment.

[0036] Explanation of the reference numerals in the figure:

[0037] 1. Laser emitting unit; 2. Position sensing unit; 3. Spot recognition unit; 4. Displacement compensation unit; 5. Calculation unit; 6. Device for detecting semiconductor wafer warpage;

[0038] 900. Electronic device; 922. Processing component; 926. Power supply component; 932. Memory; 950. Network interface; 958. Input / output interface. Detailed Implementation

[0039] Unless otherwise defined, the technical or scientific terms used in this specification should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. Specific embodiments of the invention will be described below with reference to the accompanying drawings. It should be noted that, in order to provide a concise description, this specification cannot provide a detailed description of all features of the actual embodiments. Without departing from the spirit and scope of the invention, those skilled in the art can make modifications and substitutions to the embodiments of the invention, and the resulting embodiments are also within the protection scope of the invention.

[0040] like Figure 1 As shown, the first embodiment of the present invention provides a method for detecting the warpage of a semiconductor wafer, comprising the following steps S1-S5:

[0041] Step S1: A laser beam is emitted into the observation window of the semiconductor device reaction cavity; part of the laser beam is partially reflected by the observation window to form a reference reflected beam, and the other part of the laser beam is transmitted to the surface of the wafer in the semiconductor device reaction cavity, and then reflected by the surface of the wafer to form the main reflected beam.

[0042] Step S2: Obtain multiple light spots formed by the main reflected beam and the reference reflected beam;

[0043] Step S3: Based on the motion characteristics of multiple light spots, identify the main light spot formed by the main reflected beam and the reference light spot formed by the reference reflected beam.

[0044] Step S4: Based on the displacement of the reference spot relative to its initial reference position, compensate for the displacement of the main spot.

[0045] Step S5: Calculate the wafer warpage based on the displacement of the compensated main spot.

[0046] In one embodiment, identifying a main light spot formed by a main reflected beam and a reference light spot formed by a reference reflected beam based on the motion characteristics of multiple light spots includes: setting a displacement amplitude threshold; acquiring the displacement of multiple light spots during wafer rotation; identifying light spots with displacement values ​​higher than the displacement amplitude threshold as main light spots, and identifying light spots with displacement values ​​lower than the displacement amplitude threshold as reference light spots.

[0047] In some specific embodiments, the process of identifying the main beam and reference beam based on the motion characteristics of multiple beams includes: pre-setting a displacement amplitude threshold; acquiring the displacement of each beam relative to its initial position during wafer rotation; identifying beams with displacements higher than the threshold as main beams, as their motion is mainly modulated by wafer surface deformation and exhibits significant dynamic changes; and identifying beams with displacements lower than the threshold as reference beams, as their position remains relatively stable due to reflections from the surface in front of the observation window. Through this displacement amplitude-based discrimination mechanism, the system can effectively distinguish the beams corresponding to the main reflected beam and the reference reflected beam. This implementation does not rely on spectral analysis or complex classification algorithms; it only requires a simple comparison of displacement amplitudes to identify the beams, reducing computational complexity and improving recognition speed and the system's applicability in industrial settings.

[0048] In another embodiment, based on the motion characteristics of multiple light spots, identifying the main light spot formed by the main reflected beam and the reference light spot formed by the reference reflected beam includes: acquiring displacement signals of multiple light spots and performing spectral analysis to obtain a displacement spectrum; identifying light spots whose displacement spectrum contains frequency components consistent with the rotation frequency of the wafer as main light spots; and identifying light spots whose displacement spectrum does not contain frequency components as reference light spots.

[0049] In some specific embodiments, the process of identifying the main spot and reference spot based on the motion characteristics of multiple light spots includes: during wafer rotation, synchronously acquiring the coordinate changes of each light spot on the position sensing unit to form its own displacement signal; performing spectral analysis on the displacement signal of each light spot to obtain the corresponding displacement spectrum; subsequently, determining whether each light spot's displacement spectrum contains a frequency component consistent with the wafer rotation frequency—if the displacement spectrum of a certain light spot contains this frequency component, it indicates that its motion is modulated by the wafer rotation and originates from reflection from the wafer surface, thus being identified as the main light spot; if the displacement spectrum of a certain light spot does not contain this frequency component, it indicates that its position is basically unaffected by the wafer rotation and originates from reflection from the surface in front of the observation window, thus being identified as the reference light spot. This implementation uses rotation frequency as a feature identifier, which can effectively distinguish between dynamic main spots and static reference spots. Even when multiple light spots coexist or there are slight environmental disturbances, it can still achieve highly reliable light spot classification, thereby providing accurate input signals for subsequent displacement compensation and warpage calculation.

[0050] In another embodiment, based on the motion characteristics of multiple light spots, identifying the main light spot formed by the main reflected beam and the reference light spot formed by the reference reflected beam includes: setting a displacement amplitude threshold; acquiring the displacement of multiple light spots during wafer rotation; acquiring the displacement signals of multiple light spots and performing spectral analysis to obtain a displacement spectrum; identifying the light spot whose displacement is higher than the displacement amplitude threshold and whose displacement spectrum contains a frequency component consistent with the wafer rotation frequency as the main light spot; and identifying the light spot whose displacement is lower than the displacement amplitude threshold and whose displacement spectrum does not contain a frequency component as the reference light spot.

[0051] In one implementation, identifying the main light spot and the reference light spot based on the motion characteristics of multiple light spots includes: calling a machine learning model or classification algorithm to divide the coordinates of the multiple light spots into a first coordinate set and a second coordinate set; placing the coordinates of the light spots that satisfy periodic motion into the first coordinate set as the historical coordinate set of the main light spot; and placing the coordinates of the light spots that are stationary or do not satisfy periodic motion into the second coordinate set as the historical coordinate set of the reference light spot.

[0052] In some specific embodiments, the process of identifying the main spot and reference spot based on the motion characteristics of multiple spots includes: acquiring the coordinate sequence of each spot during wafer rotation, and extracting motion features based on the sequence, such as displacement amplitude, time-domain variation patterns, or periodic indicators; calling a pre-trained machine learning model or rule-based classification algorithm to automatically classify all spots according to the motion features, dividing the coordinates of the spots into a first coordinate set and a second coordinate set; wherein, the coordinates of spots exhibiting periodic motion characteristics are assigned to the first coordinate set as the historical coordinate set of the main spot; while the coordinates of spots exhibiting static or non-periodic motion characteristics are assigned to the second coordinate set as the historical coordinate set of the reference spot. In this way, the system can dynamically and adaptively complete spot identification without relying on fixed thresholds or manually set parameters. This implementation utilizes a data-driven classification method, which can effectively cope with changes in spot behavior under complex working conditions, improve the generalization ability and stability of identification, and provide structured spot trajectory data for subsequent real-time compensation and warpage calculation.

[0053] In one implementation, the displacement of the compensated main spot (ΔX) main_corrected ΔY main_corrected )satisfy:

[0054]

[0055]

[0056] Among them, (X) main Y main ) is the current coordinate of the main light spot; (X) main0Y main0 ) is the coordinate of the initial reference position of the main light spot; (ΔX) ref ΔY ref Let be the reference drift amount of the reference spot, satisfying:

[0057]

[0058]

[0059] Among them, (X) ref Y ref (X) represents the current coordinates of the reference spot; ref0 Y ref0 ) represents the coordinates of the initial reference position of the reference spot.

[0060] It is worth noting that this compensation operation essentially transforms the displacement of the main light spot from the global coordinate system of the laboratory to a relative coordinate system with the observation window as a local reference, thereby eliminating spurious displacements caused by common-mode disturbances such as platform vibration, thermal expansion, or optical system drift. Through this compensation mechanism, the true wafer deformation signal can be effectively separated from environmental interference, significantly improving the accuracy and repeatability of warpage calculations, making it particularly suitable for long-term, high-stability semiconductor online monitoring scenarios.

[0061] In one embodiment, when another portion of the laser is transmitted to the surface of the wafer, a current measurement point is formed on the surface of the wafer; the wafer warpage is calculated based on the displacement of the compensated main spot, including: substituting the displacement of the compensated main spot into the warpage calculation model to obtain the wafer warpage K, satisfying:

[0062]

[0063] Where C is the calibration coefficient, Φ is the azimuth angle of the current measurement point relative to the center point of the wafer, Φ0 is the initial value of the azimuth angle, and the azimuth angle Φ of the current measurement point satisfies δ is the preset minimum angle threshold; L0 is the distance from the current measurement point to the axis of rotation.

[0064] It is worth noting that by introducing azimuth angle and geometric relationship to normalize the compensated spot displacement, the two-dimensional displacement information is mapped into a warp parameter related to the local curvature of the wafer. This not only improves the physical consistency of the calculation results, but also effectively suppresses the systematic deviation caused by different measurement point positions, thereby achieving high-precision and quantifiable characterization of the overall warp morphology of the wafer.

[0065] In one embodiment, the method further includes: real-time monitoring of the signal quality of the main spot; when the signal quality of the main spot is lower than a preset reliability threshold, generating a replacement displacement based on the historical displacement data of the reference spot and a preset process model, and calculating the wafer warpage based on the replacement displacement.

[0066] In some specific embodiments, the signal-to-noise ratio, sharpness, positional stability, and energy intensity of the main spot on the position sensing unit are continuously evaluated as quantitative indicators of signal quality. When the signal quality of the main spot falls below a preset reliability threshold, such as when the spot becomes blurred or even lost due to wafer surface contamination, plasma interference in the reaction cavity, or laser power fluctuations, the system automatically activates a fault-tolerance mechanism. This mechanism is based on recorded historical displacement data of the reference spot and a preset process model to calculate and generate a replacement displacement of the main spot at the current moment. The preset process model reflects the pattern of wafer warpage over time or temperature under typical process conditions. The replacement displacement can be obtained by extrapolating historical warpage trends or constructed based on the statistical correlation between reference spot drift and the normal response of the main spot. Subsequently, the system uses this replacement displacement for subsequent warpage calculations to maintain the continuity of the measurement output. This implementation effectively enhances the robustness of the measurement system under abnormal signal conditions, ensuring that effective warpage monitoring results can still be provided even when the main spot is temporarily unreliable, thereby supporting stable process control and improving the practicality and reliability of the online inspection system in complex semiconductor manufacturing environments.

[0067] like Figure 2 As shown, according to a second embodiment of the present invention, a device 6 for detecting the warpage of a semiconductor wafer is provided, which is used in any of the methods described in the above embodiments. The device includes: a laser emitting unit 1, for emitting a laser beam into an observation window of a semiconductor device reaction cavity; causing a portion of the laser beam to be partially reflected by the observation window to form a reference reflection beam, and another portion of the laser beam to be transmitted to the surface of a wafer in the semiconductor device reaction cavity, and then reflected by the surface of the wafer to form a main reflection beam; a position sensing unit 2, for acquiring multiple light spots formed by the main reflection beam and the reference reflection beam; a light spot identification unit 3, for identifying the main light spot formed by the main reflection beam and the reference light spot formed by the reference reflection beam according to the motion characteristics of the multiple light spots; a displacement compensation unit 4, for compensating the displacement of the main light spot based on the displacement of the reference light spot relative to its initial reference position; and a calculation unit 5, for calculating the wafer warpage based on the displacement of the compensated main light spot.

[0068] In some specific embodiments, the laser emitting unit 1 can be configured as a single laser emitter, such as a semiconductor laser diode or a solid-state laser, to output a continuous laser beam with stable wavelength and good beam quality; the position sensing unit 2 can be configured as a position-sensitive detector (PSD) or an area array image sensor (such as CMOS or CCD), to receive the main reflected beam and the reference reflected beam and form multiple light spots on the photosensitive surface, while outputting the real-time two-dimensional coordinates of each light spot; the light spot recognition unit 3 can be configured as a recognition logic module running on a processor, which, based on the displacement trajectory of the light spot during wafer rotation, uses threshold comparison, spectrum analysis, or... The classification algorithm determines the motion characteristics, identifying spots exhibiting periodic motion as the main spot and spots that are essentially stationary as the reference spot. Displacement compensation unit 4 can be configured as a processing module to perform displacement correction calculations. Based on the drift of the reference spot relative to the initial reference position, it performs differential calculations on the original displacement of the main spot to eliminate spurious displacements caused by common-mode disturbances such as vibration and thermal drift. Calculation unit 5 can be configured as a calculation module to perform warpage calculations. It combines the compensated main spot displacement with the azimuth and radial distance of the current measurement point, inputs it into a pre-calibrated geometric optics model, and finally outputs the quantified result of the wafer warpage. The functions of the above units can be implemented using a general-purpose microprocessor, a digital signal processor, a field-programmable gate array, or a combination thereof.

[0069] In one embodiment, the laser emitting unit 1 and the position sensing unit 2 are coaxially arranged so that the main reflected beam and the reference reflected beam return along the incident optical path and are received by the position sensing unit 2.

[0070] In some specific embodiments, the laser emitting unit 1 and the position sensing unit 2 adopt a coaxial optical path layout, that is, the laser emitting optical axis and the receiving optical axis of the position sensing unit 2 coincide or are substantially coincident. Specifically, after the laser is emitted from the laser emitting unit 1, it passes sequentially along the main optical axis through the beam splitting unit and the observation window of the reaction cavity, illuminating the wafer surface. The main reflected beam formed by the reflection from the wafer surface and the reference reflected beam formed by the reflection from the front surface of the observation window both return in the opposite direction of the original incident optical path, pass through the observation window and the beam splitting unit again, and are finally received by the position sensing unit 2 located on the same optical axis. This coaxial arrangement ensures that the main reflected beam and the reference reflected beam can return to the detection area efficiently and stably, while simplifying the optical structure, avoiding signal loss or spot distortion caused by optical path offset, and improving the repeatability and spatial resolution of the measurement. In addition, combined with the path separation effect of the beam splitting unit on the returning beam, the main spot and the reference spot can form a spatially separated and stable image on the position sensing unit 2, providing a reliable basis for subsequent identification and compensation.

[0071] In one embodiment, the device further includes a beam splitting unit disposed between the laser emitting unit 1 and the observation window; the laser emitting unit 1 is used to emit laser light along a first direction; the laser light emitted along the first direction is transmitted through the beam splitting unit to form a main transmitted beam; the main transmitted beam is transmitted through the observation window along the first direction to the wafer surface; the main reflected beam formed by reflection from the wafer surface is transmitted in the opposite direction of the first direction to the position sensing unit 2 to form a main spot; the laser light emitted along the first direction is reflected through the observation window to form a reference reflected beam; the reference reflected beam is transmitted in the opposite direction of the first direction to the position sensing unit 2 to form a reference spot; the main spot and the reference spot do not overlap.

[0072] In some specific embodiments, the beam splitting unit is disposed in the optical path between the laser emitting unit 1 and the observation window of the semiconductor device's reaction cavity. The laser emitting unit 1 emits a laser beam along a first direction. This laser beam first enters the beam splitting unit and passes through the beam splitting unit in a transmission manner, forming a main transmitted beam. The main transmitted beam continues to propagate along the first direction, passes through the observation window, and enters the reaction cavity, eventually illuminating the wafer surface. When the main transmitted beam reaches the wafer surface, it is reflected to form a main reflected beam. This main reflected beam returns in the opposite direction of the first direction, passes through the observation window and the beam splitting unit in sequence, and finally reaches the position sensing unit 2, forming a main light spot on its photosensitive surface.

[0073] Meanwhile, when the laser propagating along the first direction passes through the observation window, its front surface undergoes partial reflection due to the abrupt change in refractive index, thus forming a reference reflected beam. This reference reflected beam also returns in the opposite direction to the first direction, first returning to the beam splitting unit. At the beam splitting unit, the reference reflected beam is no longer transmitted but is reflected to a second direction different from the first direction (e.g., a vertical or inclined direction), and then propagates along this second direction to the position sensing unit 2, forming a reference spot on its photosensitive surface.

[0074] Since the main reflected beam and the reference reflected beam are guided by the beam splitter in the return path via transmission and reflection respectively, their imaging positions on the position sensing unit 2 are separated, thus ensuring that the formed main spot and the reference spot do not overlap in space. This common-path-splitter detection design based on the beam splitter not only ensures that the two beams experience highly consistent external optical paths, improving anti-interference capabilities, but also achieves natural differentiation of the spots through physical path separation, laying an optical foundation for subsequent motion-based identification and high-precision displacement compensation.

[0075] In one embodiment, the device further includes a focusing lens disposed between the beam splitting unit and the position sensing unit 2, for focusing the returned laser beam onto the photosensitive surface of the position sensing unit 2 to form a main spot and a reference spot.

[0076] In some specific embodiments, the primary reflected beam is transmitted through the beam-splitting unit in the opposite direction to the first direction, enters the focusing lens, and is focused onto the first region of the photosensitive surface; the reference reflected beam is reflected in the second direction at the beam-splitting unit, and then also passes through the same focusing lens and is focused onto the second region of the photosensitive surface. Because the two beams have different directions of spatial propagation, after being imaged by the focusing lens, their light spots naturally separate on the position sensing unit 2 and do not overlap.

[0077] By introducing a focusing lens, not only is the positioning accuracy and signal-to-noise ratio of the light spot improved, but the system's sensitivity to minute displacements is also enhanced, which is beneficial for subsequent high-resolution extraction of light spot motion information. Furthermore, using a single focusing lens to share the imaging of two returning beams simplifies the optical structure, reduces the number of components, and ensures that the main light spot and the reference light spot are detected under the same imaging conditions, thus improving measurement consistency and reliability.

[0078] In one embodiment, the position sensing unit 2 includes at least one of a position-sensitive detector, an area array charge-coupled device (CCD) image sensor camera, and an area array complementary metal-oxide-semiconductor (CMOS) image sensor camera.

[0079] In some specific embodiments, when a position-sensitive detector is used, it is based on the photocurrent distribution principle and can output the continuous coordinates of the light spot on the photosensitive surface in real time with high response speed and submicron resolution. It is suitable for scenarios where dynamic displacement is sampled at high frequency, and is especially suitable for real-time warp monitoring during the high-speed rotation of wafers.

[0080] In other specific embodiments, when using an area array CCD or CMOS image sensor, it captures the entire spot image through a pixel array. This not only allows for the simultaneous acquisition of the position information of the main spot and the reference spot, but also provides auxiliary features such as spot shape and intensity distribution, facilitating spot quality assessment, multi-spot identification, or anomaly detection. CCD sensors typically offer higher signal-to-noise ratios and imaging uniformity, making them suitable for high-precision measurements; while CMOS sensors offer advantages in integration, power consumption, and readout speed, making them more suitable for embedded or industrial online systems.

[0081] In one embodiment, the laser emitting unit 1 and the position sensing unit 2 are fixed to the outside of the semiconductor device's reaction cavity or encapsulated in the same housing, so that the optical axis of the laser forms a preset angle with the observation window.

[0082] In some examples, the preset angle is 90°, and the optical axis of the laser is perpendicular to the observation window.

[0083] In one embodiment, the device further includes an evaluation module for real-time monitoring of the signal quality of the main spot; when the signal quality of the main spot is lower than a preset reliability threshold, the calculation unit 5 generates a replacement displacement based on the historical displacement data of the reference spot and a preset process model, and calculates the wafer warpage based on the replacement displacement.

[0084] In some specific embodiments, the evaluation module quantifies the reliability of the main spot by analyzing multiple characteristic parameters on the position sensing unit 2. These parameters include, but are not limited to, the intensity of the spot, signal-to-noise ratio, outline sharpness, position stability, and the amplitude of jumps between consecutive frames. When any or a combination of the above indicators is lower than a preset reliability threshold, such as when the main spot is blurred, split, or temporarily lost due to wafer surface contamination, intracavitary plasma emission interference, laser power fluctuations, or viewing window fogging, the evaluation module determines that the current main spot signal is unreliable.

[0085] In this situation, calculation unit 5 no longer uses the original displacement data of the current main spot, but instead activates a fault-tolerant compensation mechanism: on the one hand, it calls the stored historical displacement sequence of the reference spot to extract the recent drift trend; on the other hand, it combines a preset process model (which is based on historical process data and describes the typical evolution of wafer warpage under specific temperature, pressure, or time conditions) to calculate the reasonable displacement value that the main spot should have at the current moment, as a substitute displacement. This substitute displacement can be obtained by linearly or nonlinearly extrapolating the historical warpage rate, or it can be reconstructed using the statistical correlation between the reference spot drift and the main spot response under normal operating conditions.

[0086] Subsequently, calculation unit 5 substitutes this alternative displacement into the standard warpage calculation process and continues to output the wafer warpage result. This mechanism effectively avoids measurement interruptions or abnormal outputs caused by instantaneous signal degradation, ensuring the continuity and robustness of the online monitoring system, and is especially suitable for long-term operation or high-interference semiconductor manufacturing environments.

[0087] A third embodiment of the present invention provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program executable by the processor; and the processor is used to execute the computer program in the memory to implement the method of any of the above embodiments.

[0088] Figure 3 This is a block diagram illustrating an electronic device according to an exemplary embodiment. For example, electronic device 900 may be provided as a server. (Refer to...) Figure 3The electronic device 900 includes a processing component 922, which further includes one or more processors, and memory resources represented by memory 932 for storing instructions, such as application programs, that can be executed by the processing component 922. The application programs stored in memory 932 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 922 is configured to execute instructions to perform the methods described above.

[0089] Electronic device 900 may also include a power supply component 926 configured to perform power management of electronic device 900, a wired or wireless network interface 950 configured to connect electronic device 900 to a network, and an input / output (I / O) interface 958. Electronic device 900 may operate on an operating system stored in memory 932, such as Windows Server™, MacOS X™, Unix™, Linux™, FreeBSD™, or similar.

[0090] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 932 including instructions, which can be executed by a processing component 922 of an electronic device 900 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device.

[0091] A fourth embodiment of the present invention provides a readable storage medium storing a program, which, when executed, implements the method of any of the above embodiments.

[0092] The fifth embodiment of the present invention provides a computer program product, including a computer program, which, when executed, implements the method of any of the above embodiments.

[0093] In this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance. The term "multiple" refers to two or more unless otherwise expressly defined.

[0094] The above description of the embodiments is intended to enable those skilled in the art to understand and apply the present invention. It will be apparent to those skilled in the art that various modifications can be made to these embodiments, and the general principles described herein can be applied to other embodiments without creative effort. Therefore, the present invention is not limited to the embodiments described herein, and any improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope and spirit of the invention are within the scope of the present invention.

Claims

1. A method for detecting the warpage of a semiconductor wafer, characterized in that, include: A laser beam is emitted into the observation window of the reaction chamber of the semiconductor device; A portion of the laser beam is partially reflected through the observation window to form a reference reflected beam, while another portion of the laser beam is transmitted to the surface of a wafer in the reaction chamber of a semiconductor device, and then reflected by the surface of the wafer to form a main reflected beam. Multiple light spots formed by the main reflected beam and the reference reflected beam are obtained; Based on the motion characteristics of the plurality of light spots, identify the main light spot formed by the main reflected beam and the reference light spot formed by the reference reflected beam; The displacement of the main spot is compensated based on the displacement of the reference spot relative to its initial reference position. When another portion of the laser is transmitted to the surface of the wafer, a current measurement point is formed on the surface of the wafer; The wafer warpage is calculated based on the compensated displacement of the main light spot, including: The displacement of the compensated main spot (ΔX) main_corrected ΔY main_corrected Substituting into the warpage calculation model, we obtain the wafer warpage K, which satisfies: Where C is the calibration coefficient, Φ is the azimuth angle of the current measurement point relative to the center point of the wafer, Φ0 is the initial value of the azimuth angle, and the azimuth angle Φ of the current measurement point satisfies δ is the preset minimum angle threshold; L0 is the distance from the current measurement point to the axis of rotation.

2. The method according to claim 1, characterized in that, Based on the motion characteristics of the plurality of light spots, the main light spot formed by the main reflected beam and the reference light spot formed by the reference reflected beam are identified, including: Set the displacement amplitude threshold; The displacement of the multiple light spots during wafer rotation is obtained; Spots with displacement greater than the displacement amplitude threshold are identified as main spots, and spots with displacement less than the displacement amplitude threshold are identified as reference spots.

3. The method according to claim 1, characterized in that, Based on the motion characteristics of the plurality of light spots, the main light spot formed by the main reflected beam and the reference light spot formed by the reference reflected beam are identified, including: The displacement signals of the multiple light spots are acquired, and spectral analysis is performed to obtain the displacement spectrum; The light spot containing a frequency component in the displacement spectrum that matches the rotation frequency of the wafer is identified as the main light spot; The light spot that does not contain the frequency component in the displacement spectrum is identified as the reference light spot.

4. The method according to claim 1, characterized in that, Based on the motion characteristics of the plurality of light spots, the main light spot formed by the main reflected beam and the reference light spot formed by the reference reflected beam are identified, including: Set a displacement amplitude threshold; obtain the displacement of the multiple light spots during wafer rotation; The displacement signals of the multiple light spots are acquired, and spectral analysis is performed to obtain the displacement spectrum; The light spot whose displacement is higher than the displacement amplitude threshold and whose displacement spectrum contains a frequency component consistent with the rotation frequency of the wafer is identified as the main light spot; A light spot whose displacement is lower than the displacement amplitude threshold and whose displacement spectrum does not contain the frequency component is identified as a reference light spot.

5. The method according to claim 1, characterized in that, Identifying the main light spot and the reference light spot based on the motion characteristics of the multiple light spots includes: By calling a machine learning model or classification algorithm, the coordinates of the multiple light spots are divided into a first coordinate set and a second coordinate set; The coordinates of the light spot that satisfy the periodic motion are placed into the first coordinate set, which serves as the historical coordinate set of the main light spot; The coordinates of stationary or non-periodic light spots are placed into the second coordinate set as the historical coordinate set of the reference light spot.

6. The method according to claim 1, characterized in that, The displacement of the main light spot after compensation (ΔX) main_corrected ΔY main_corrected )satisfy: Among them, (X) main Y main (X) represents the current coordinates of the main light spot; main0 Y main0 (ΔX) represents the coordinates of the initial reference position of the main light spot; ref ΔY ref Let be the reference drift amount of the reference spot, satisfying: Among them, (X) ref Y ref (X) represents the current coordinates of the reference spot; ref0 Y ref0 ) represents the coordinates of the initial reference position of the reference spot.

7. The method according to claim 1, characterized in that, The method further includes: Real-time monitoring of the signal quality of the main light spot; When the signal quality of the main spot is lower than a preset reliability threshold, an alternative displacement is generated based on the historical displacement data of the reference spot and a preset process model, and the wafer warpage is calculated based on the alternative displacement.

8. A semiconductor wafer warpage detection device, used in the method according to any one of claims 1 to 7, characterized in that, The device includes: A laser emitting unit is used to emit a laser beam into the observation window of the reaction cavity of a semiconductor device; a portion of the laser beam is partially reflected by the observation window to form a reference reflected beam, and another portion of the laser beam is transmitted to the surface of a wafer in the reaction cavity of the semiconductor device, and then reflected by the surface of the wafer to form a main reflected beam; A position sensing unit is used to acquire multiple light spots formed by the main reflected beam and the reference reflected beam; A spot recognition unit is used to identify, based on the motion characteristics of the plurality of spots, a main spot formed by a main reflected beam and a reference spot formed by a reference reflected beam; The displacement compensation unit is used to compensate for the displacement of the main light spot based on the displacement of the reference light spot relative to its initial reference position. The calculation unit is used to calculate the wafer warpage based on the displacement of the compensated main spot.

9. The apparatus according to claim 8, characterized in that, The laser emitting unit is coaxially arranged with the position sensing unit so that the main reflected beam and the reference reflected beam return along the incident optical path and are received by the position sensing unit.

10. The apparatus according to claim 8, characterized in that, The device also includes a beam splitting unit disposed between the laser emitting unit and the observation window; The laser emitting unit is used to emit laser light along a first direction; the laser light emitted along the first direction is transmitted through the beam splitting unit to form a main transmitted beam; the main transmitted beam is transmitted through the observation window along the first direction to the wafer surface; The main reflected beam formed by the reflection from the wafer surface is transmitted in the opposite direction to the position sensing unit to form the main light spot; The laser emitted along the first direction is reflected by the observation window to form a reference reflected beam; The reference reflected beam is transmitted in the opposite direction to the first direction to the position sensing unit to form a reference spot; the main spot and the reference spot do not overlap. The device further includes a focusing lens disposed between the beam splitting unit and the position sensing unit, for converging the returned laser beam onto the photosensitive surface of the position sensing unit to form the main light spot and the reference light spot.

11. The apparatus according to claim 8, characterized in that, The position sensing unit includes at least one of a position-sensitive detector, a surface array charge-coupled device image sensor camera, and a surface array complementary metal-oxide-semiconductor image sensor camera; The laser emitting unit and the position sensing unit are fixed to the outside of the semiconductor device's reaction cavity or encapsulated in the same housing, so that the optical axis of the laser forms a preset angle with the observation window; The device also includes an evaluation module for real-time monitoring of the signal quality of the main light spot; When the signal quality of the main spot is lower than a preset reliability threshold, the calculation unit generates an alternative displacement based on the historical displacement data of the reference spot and a preset process model, and calculates the wafer warpage based on the alternative displacement.

Images