Etching endpoint determination method, device, detection apparatus, and medium

Abstract

The application provides an etching endpoint judgment method and device, a detection equipment and a medium, and relates to the technical field of semiconductor detection. The light intensity information of the characteristic spectrum line corresponding to the target product is continuously obtained while the etching process is carried out, and the light intensity information is used to generate an excitation signal. Then, the excitation signal is used for the operation of the excitation of the vibrator, so that the change of the characteristic spectrum line, which is originally weak and easily submerged by noise in the etching process, is no longer judged by linear indicators such as the absolute value of the light intensity, the modulation amplitude or the frequency domain energy, but is converted into the evolution characteristics of the vibrator state. Since the vibrator state can jump from the periodic state to the chaotic state when the excitation signal indicates that the target product has disappeared, the mutation of the vibrator state can be used to represent the process from the existence to the disappearance of the target product, so as to judge whether the etching endpoint has been reached.

Description

Technical Field

[0001] This invention relates to the field of semiconductor testing technology, and more specifically to a method, apparatus, testing equipment, and medium for determining the etching endpoint. Background Technology

[0002] Through Silicon Via (TSV) deep silicon etching is a key process in 3D integrated circuit packaging. It typically employs the Bosch periodic etching method, which alternately introduces SF6 etching gas and C4F8 passivation gas to form high aspect ratio vias in the silicon wafer. However, the aperture ratio of the TSV process is usually very low, even as low as 0.1% to 0.5%, resulting in extremely weak emission spectral signals of characteristic products such as SiF in the plasma generated by the etching gas (used for etching). The signal-to-noise ratio is often close to the background noise, making it difficult to extract the endpoint signal stably. Existing intensity thresholding, modulation depth, spectral analysis, and time-domain synchronous averaging methods, although all attempt to improve the signal-to-noise ratio through filtering, averaging, and frequency domain transformation, are essentially linear signal processing methods that cannot overcome the noise floor and often suffer from high misjudgment rates. Summary of the Invention

[0003] The purpose of this invention is to provide a method, apparatus, detection equipment, and medium for determining the etching endpoint, aiming to break through the existing linear signal processing methods for determining the etching endpoint and improve the accuracy of etching endpoint determination.

[0004] In a first aspect, the present invention provides a method for determining the etching endpoint, the method comprising: While the etching process is in progress, the light intensity information of the characteristic spectral lines corresponding to the target product at time t is continuously acquired. The target product is the product generated by etching in the plasma used for etching. The excitation signal at time t is generated using the light intensity information at time t; Determine the fluctuation of the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, and the noise intensity in the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, wherein the end time of the first time period is time t, and the length of the first time period is the first preset duration. Based on the fluctuation and noise intensity, the critical coefficients corresponding to time t are determined, where the critical coefficients are the adjustable coefficients in the calculation formula on which the external excitation operation of the oscillator is based. Based on the excitation signal at time t and the key coefficients corresponding to time t, an external excitation operation is performed on the oscillator to obtain the oscillator output at time t. The oscillator state at time t is then updated based on the oscillator output at time t. The oscillator state at time t is used to characterize the state type of the oscillator output obtained up to time t. The state type is either a periodic state or a chaotic state. When the excitation signal at time t indicates that the target generated object has disappeared, the oscillator state at time t transitions from a periodic state to a chaotic state. Based on the state of the oscillator at time t, determine whether the etching endpoint has been reached.

[0005] Thus, while the etching process is underway, the light intensity information of the characteristic spectral lines corresponding to the target product is continuously acquired. This light intensity information is then used to generate an excitation signal, and an external excitation operation is performed on the oscillator based on this excitation signal. This transforms the changes in the characteristic spectral lines, which were originally weak and easily submerged by noise during the etching process, from being judged solely by linear indicators such as absolute light intensity, modulation amplitude, or frequency domain energy, into the evolution characteristics of the oscillator state. Since the oscillator state can transition from a periodic state to a chaotic state when the excitation signal indicates that the target product has disappeared, the abrupt change in the oscillator state can be used to characterize the process of the target product from existence to disappearance, thereby determining whether the etching has reached its endpoint. By reducing the dependence on the amplitude and signal-to-noise ratio of the characteristic spectral intensity signal, the endpoint criterion can still be extracted relatively stably even when the emission spectral signal of characteristic products such as SiF is extremely weak due to the low aperture ratio of the TSV. This reduces the misjudgment caused by the intensity threshold method, modulation depth method, spectrum analysis method, and time-domain synchronous averaging method due to the noise background being close to the effective signal, thus improving the reliability and stability of etching endpoint judgment. Moreover, the adjustable coefficient in the calculation formula on which the external excitation calculation of the oscillator is based is not fixed, but dynamically determined by the detection device according to the fluctuation of the intensity information of the characteristic spectral line corresponding to the target product and the noise intensity within the first time period, making the oscillator excitation calculation more compatible with the current signal quality.

[0006] In some implementations, the calculation formula on which the external excitation calculation of the oscillator is based is:

[0007] in, For time markers, for The oscillator output at any given moment, for right The second derivative, for right The first derivative, for cubed, for The fifth power, for Momentary stimulus signal Driven by cycles, The frequency of the cycle drive is the angular frequency corresponding to the gas switching frequency during the etching process, and the cycle of the cycle drive is... The gas is the gas used for etching. The amplitude is driven by the period. The damping coefficient is adjustable. The periodic state is the state in which the oscillator output changes periodically with the periodic drive. In the periodic state, the oscillator output changes repeatedly within multiple consecutive periodic drives. The chaotic state is the state in which the oscillator output does not change periodically with the periodic drive. In the chaotic state, the oscillator output changes non-periodically within multiple consecutive periodic drives.

[0008] Thus, the excitation signal is not used as a threshold judgment object alone, but together with the periodic drive in the calculation formula (a Duffing equation) it constitutes the excitation sum, and affects the oscillator output through the excitation sum. If the target product still exists, the excitation signal generated by the characteristic spectral line corresponding to the target product still has a certain effective component, and the excitation sum can keep the oscillator in a dynamic response corresponding to the state in which the target product exists. If the target product has disappeared, the effective component in the excitation signal changes, the excitation sum changes accordingly, and thus the oscillator output in the calculation formula changes accordingly, and may further cause the oscillator state to transition from a periodic state to a chaotic state.

[0009] In some implementations, the amplitude of the periodic drive is the product of the critical amplitude and a first preset multiple, where the first preset multiple is less than 1 and greater than 0, and the absolute value of the difference between the critical amplitude and the amplitude of the periodic drive is less than a preset difference threshold. The critical amplitude is determined through the following steps: Set the excitation signal to 0; Multiple preset amplitudes are determined, wherein the numerical difference between any two adjacent preset amplitudes is a preset step size; Perform a calibration pre-step, which includes determining the oscillator output at multiple times corresponding to each preset amplitude in the calculation formula in ascending order of numerical values, and using the oscillator output at multiple times corresponding to the Poincaré cross-section discrete points, and determining the Poincaré cross-section variance corresponding to the preset amplitude based on the Poincaré cross-section discrete points. Based on the Poincaré cross-sectional variances corresponding to multiple preset amplitudes, a critical amplitude is determined among the multiple preset amplitudes. The critical amplitude is the first preset amplitude that makes the corresponding Poincaré cross-sectional variance less than a first preset variance threshold during the execution of the calibration pre-step.

[0010] Thus, the amplitude of the periodic drive is set near the critical amplitude, making the calculation formula more sensitive to the weak same-frequency periodic components in the characteristic spectrum of the target product; at the same time, since the critical amplitude is obtained by Poincaré cross-sectional variance calibration under no external excitation conditions, it can improve the reliability of subsequent etching endpoint determination and reduce the possibility of misjudgment caused by noise fluctuations or background drift.

[0011] In some implementations, the critical coefficient is the damping coefficient in the calculation formula. The critical coefficient corresponding to time t is determined based on the fluctuation conditions and noise intensity, including: The fluctuation characterization value is determined based on the fluctuation situation. The fluctuation characterization value is used to characterize the degree of fluctuation of light intensity information within the first time period. The noise characterization value is determined based on the noise intensity, whereby the noise characterization value is used to characterize the degree of noise interference in the light intensity information within the first time period. The fluctuation characterization value and the noise characterization value are normalized to obtain the normalized fluctuation characterization value and the normalized noise characterization value. Based on the preset benchmark damping coefficient, normalized fluctuation characterization value, and normalized noise characterization value, the damping coefficient corresponding to time t is determined. The damping coefficient corresponding to time t is positively correlated with the normalized noise characterization value and negatively correlated with the normalized fluctuation characterization value.

[0012] In this way, the detection equipment can adaptively determine the damping coefficient in the calculation formula based on the fluctuation of light intensity information and the degree of noise interference within the first time period. When the effective fluctuation is strong, it can improve the response capability of the external excitation calculation of the oscillator and enhance the suppression of noise when the noise intensity is high, thereby improving the reliability of the oscillator output at time t and the oscillator state update at time t.

[0013] In some implementations, determining whether the etching endpoint has been reached based on the oscillator state at time t includes: The oscillator outputs at multiple periodic times in the oscillator state at time t are extracted as discrete points of the Poincaré section, where the periodic times are positive integer multiples of the gas switching period during the etching process, and the gas is the gas used for etching. Determine the length of the sliding window, and group the discrete points of the Poincaré section according to the length of the sliding window to obtain multiple groups of sliding windows; For each set of sliding windows, the variance of the discrete points of the Poincaré cross section within the sliding window is determined as the variance of the Poincaré cross section of the sliding window. When the variance of the Poincaré cross section is less than the second preset variance threshold, it indicates that the oscillator state of the corresponding period of the sliding window is in a periodic state. When the variance of the Poincaré cross section is greater than or equal to the second preset variance threshold, it indicates that the oscillator state of the corresponding period of the sliding window is in a chaotic state. Determine whether the etching endpoint has been reached based on the variance of multiple Poincaré sections.

[0014] In this way, the detection equipment can transform the "whether the trajectory output by the oscillator is stable and convergent" into a numerical judgment of the variance of the Poincaré section, thereby identifying periodic and chaotic states with a small amount of computation. The detection equipment does not rely on the light intensity threshold at a single moment, but determines the endpoint based on the topological changes of the oscillator state, which is beneficial to improving the reliability of etching endpoint determination under low signal-to-noise ratio conditions.

[0015] In some implementations, determining whether the etching endpoint has been reached is based on the variance of multiple Poincaré sections, including: If the duration of the oscillator state in a chaotic state at time t of multiple Poincaré cross-section variance indicators has exceeded a preset duration threshold, it is determined that the etching endpoint has been reached. The first cycle time that satisfies the condition that the duration of the oscillator state in a chaotic state has exceeded the preset duration threshold is determined as the etching endpoint arrival time. The preset duration threshold is N gas switching cycles, where N is a positive integer greater than 1.

[0016] In this way, the detection equipment bases its judgment of "whether the etching endpoint has been reached" on the time continuity condition that the oscillator is in a continuous chaotic state, rather than on a single variance value or a single sampling point. The detection equipment can reduce the risk of misjudgment caused by instantaneous disturbances and can output the endpoint arrival signal to the main controller of the etching equipment in a timely manner through the digital I / O interface after confirming the endpoint.

[0017] In some implementations, the second preset variance threshold is determined by the following steps: The variance of the Poincaré section of the sliding window in which at least one oscillator is in a periodic state is determined as at least one calibration variance; Determine the mean of at least one calibration variance; Multiply the mean by a second preset factor to obtain a second preset variance threshold, wherein the second preset factor is greater than 1.

[0018] In this way, the detection equipment can adaptively determine the second preset variance threshold based on the Poincaré cross-sectional variance when the oscillator is in a periodic state during the etching stabilization period. This makes the distinction between periodic and chaotic states more consistent with the actual process signal characteristics. This scheme can reduce the risk of misjudgment caused by setting the threshold too low, and also avoid the endpoint response delay caused by setting the threshold too high, which is beneficial to improving the reliability of etching endpoint determination under low signal-to-noise ratio conditions.

[0019] In some implementations, the excitation signal at time t is generated using the light intensity information at time t, including: The light intensity information at time t is processed to remove DC, thus obtaining the excitation signal at time t.

[0020] In this way, the detection equipment can convert the light intensity information containing the background intensity component into an excitation signal after DC removal, so that the signal input to the oscillator detection module can more concentratedly reflect the periodic changes of the characteristic spectral lines of the target product, and can also reduce the influence of the absolute value drift of light intensity on the oscillator state identification, thereby improving the stability and reliability of the determination of the low aperture ratio TSV etching endpoint.

[0021] Secondly, the present invention provides an etching endpoint determination device, the device comprising: The acquisition module is used to continuously acquire the light intensity information of the characteristic spectral lines corresponding to the target product at time t while the etching process is in progress. The target product is the product generated by etching in the plasma used for etching. The generation module is used to generate the excitation signal at time t using the light intensity information at time t; The first determining module is used to determine the fluctuation of the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, and the noise intensity in the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, wherein the end time point of the first time period is time t, and the length of the first time period is the first preset duration. The second determining module is used to determine the critical coefficients corresponding to time t based on the fluctuation and noise intensity. The critical coefficients are the adjustable coefficients in the calculation formula on which the external excitation calculation of the oscillator is based. The calculation module is used to perform external excitation calculation on the oscillator based on the excitation signal and key coefficients at time t, to obtain the oscillator output at time t, and to update the oscillator state at time t based on the oscillator output at time t. The oscillator state at time t is used to characterize the state type of the oscillator output obtained up to time t. The state type is either periodic or chaotic. When the excitation signal at time t indicates that the target generated object has disappeared, the oscillator state at time t transitions from a periodic state to a chaotic state. The determination module is used to determine whether the etching endpoint has been reached based on the oscillator state at time t.

[0022] Thirdly, the present invention provides a detection device, including a spectral signal acquisition unit and a calculation unit. The spectral signal acquisition unit is used to acquire the characteristic spectral lines corresponding to the target product, and the target product is a product generated by etching in a plasma used for etching. The computing unit includes a processor and a memory. The memory stores programs or instructions that can run on the processor. When the program or instructions are executed by the processor, they implement the etching endpoint determination method provided in the first aspect of the present invention.

[0023] Fourthly, the present invention provides a machine-readable storage medium storing instructions that, when executed by a processor, cause the processor to implement the etching endpoint determination method provided in the first aspect of the present invention.

[0024] The etching endpoint determination method, apparatus, detection equipment, and medium provided by this invention convert the light intensity information of the characteristic spectral lines corresponding to the target product into an excitation signal, and perform external excitation calculations on the oscillator based on this excitation signal. This transforms the weak and noise-saturated characteristic spectral line changes into oscillator state evolution characteristics, thereby enabling the characterization of the target product's existence and disappearance through the transition of the oscillator state from a periodic state to a chaotic state, reducing dependence on the amplitude and signal-to-noise ratio of the characteristic spectral line light intensity signal. By setting the amplitude of the periodic drive near the critical amplitude, adaptively determining the damping coefficient based on the fluctuation of light intensity information and the degree of noise interference, identifying the periodic and chaotic states based on the Poincaré cross-sectional variance, and determining the etching endpoint by combining the duration of the chaotic state, the sensitivity to weak same-frequency periodic components can be improved, the suppression of noise and background drift can be enhanced, and the risk of misjudgment caused by instantaneous disturbances can be reduced. Therefore, even when the emission spectral signal of the target product is extremely weak due to the low aperture ratio of the TSV, the endpoint criterion can still be extracted relatively stably, improving the reliability, stability, and timeliness of etching endpoint determination. Attached Figure Description

[0025] Figure 1 This is a flowchart illustrating the etching endpoint determination method provided in an embodiment of the present invention; Figure 2 This is another flowchart illustrating the etching endpoint determination method provided in this embodiment of the invention; Figure 3 This is a schematic diagram of the etching endpoint determination device provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the detection device provided in an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the scope of protection of the present invention.

[0027] The terms "first," "second," etc., used in the specification and claims of this invention are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0028] This invention provides a method, apparatus, detection equipment, and medium for determining the etching endpoint, aiming to break through the existing linear signal processing approach to determine the etching endpoint and improve the accuracy of etching endpoint determination.

[0029] Figure 1 This is a flowchart illustrating the etching endpoint determination method provided in an embodiment of the present invention, as shown below. Figure 1 As shown, the first aspect of the present invention provides a method for determining the etching endpoint. The method for determining the etching endpoint provided by the present invention can be executed by a detection device for the etching endpoint, and includes the following steps S100-S600: Step S100: While the etching process is in progress, continuously acquire the light intensity information of the characteristic spectral lines corresponding to the target product at time t, wherein the target product is the product generated by etching in the plasma used for etching.

[0030] In this step, specifically, during the etching process performed by the etching equipment, the detection equipment can continuously collect the light signal emitted by the plasma through the spectral signal acquisition unit; the light signal contains the characteristic spectral lines corresponding to the target product. The target product is a product generated by the reaction between the etched material and the etching gas in the plasma environment during the etching process. Therefore, the presence, weakening or disappearance of the target product can reflect whether the etching process is close to or has reached the etching endpoint. The spectral signal acquisition unit can transmit the acquired plasma emission signal to the spectral detection unit, which selects the characteristic spectral line at the wavelength corresponding to the target product. Subsequently, the detection device can convert the optical signal of the characteristic spectral line into an electrical signal through the photoelectric conversion unit, and then convert the electrical signal into a digital signal through the analog-to-digital conversion unit. At time t, the detection device acquires the light intensity sample value corresponding to the target product in the digital signal, which is used as the light intensity information of the characteristic spectral line corresponding to the target product at time t. As the etching process continues, the detection equipment can continuously acquire light intensity information at different times according to the preset sampling period, thereby forming a time-updated light intensity information sequence inside the detection equipment. Thus, without interfering with the etching process, the current light intensity change of the characteristic spectral lines corresponding to the target product can be obtained in real time, enabling subsequent steps to determine whether the target product is still being generated or exists in the plasma based on the light intensity information at time t. Since the target product usually continues to be generated before the etching reaches its endpoint, its characteristic spectral lines still have a certain light intensity; while the target product will significantly weaken or disappear when the etching reaches its endpoint, continuously acquiring the light intensity information at time t can provide a real-time signal basis for determining the etching endpoint. For example, in the TSV deep silicon etching process, the target product can be SiF generated by the reaction of silicon with fluorine-containing etching gas; after the etching equipment forms plasma in the reaction chamber, the silicon material is etched and SiF is generated. SiF has corresponding characteristic spectral lines in the plasma emission spectrum; the detection equipment can collect the plasma emission signal in the reaction chamber through an optical probe, lock the characteristic spectral line corresponding to SiF through a spectrometer, and then convert the light intensity of the characteristic spectral line into an electrical signal and a digital signal; the detection equipment reads the light intensity sampling value of the digital signal at time t, for example, I(t), and uses I(t) as the light intensity information of the characteristic spectral line corresponding to SiF at time t.

[0031] Step S200: Generate the excitation signal at time t using the light intensity information at time t.

[0032] In this step, specifically, after acquiring the light intensity information at time t, the detection device can perform signal processing on the light intensity information to adapt to the external excitation calculation of the oscillator, so as to generate the excitation signal corresponding to time t. The signal processing can include one or more of the following: background subtraction, amplitude normalization, noise suppression, outlier removal, phase correction synchronized with the etching cycle, and sliding window processing based on historical light intensity information. Since the single light intensity information at time t may be affected by factors such as random noise, background light fluctuation, optical window contamination, and plasma power fluctuation, the detection device can also combine multiple light intensity information before time t to dynamically correct the current light intensity information, thereby generating an excitation signal that is more suitable as an external excitation input. From the perspective of the signal transformation process, the characteristic spectral lines corresponding to the target product first manifest as optical signals in the plasma; these optical signals are converted into electrical signals after photoelectric conversion; these electrical signals are converted into digital light intensity information that can be processed by the detection equipment after analog-to-digital conversion; the detection equipment then performs preprocessing, scale transformation, or feature extraction based on this digital light intensity information to generate the excitation signal required for the external excitation operation of the oscillator; thus, the excitation signal may not be the original light intensity information itself, but rather a digital signal converted from the original light intensity information that can characterize whether the target product still exists or has disappeared at time t. In this way, the weak light intensity information of the characteristic spectral lines corresponding to the target generated object is converted into an excitation signal suitable for external excitation calculation of the input oscillator. Especially in the low aperture ratio etching scenario, the characteristic spectral line signal of the target generated object may be very weak, and directly judging the endpoint based on the original light intensity information is easily affected by noise. By using the light intensity information to generate an excitation signal, the influence of irrelevant noise and background drift can be weakened, so that the subsequent external excitation calculation of the oscillator can respond more sensitively to the presence or disappearance of the characteristic signal of the target generated object. For example, at a certain time t, the detection device acquires the light intensity information I(t) of the SiF characteristic spectral lines. Since the background light of the reaction cavity and the contamination of the detection window may cause I(t) to contain a slowly changing background, the detection device can calculate the background reference based on the light intensity information over a period of time before time t, and subtract the background reference from I(t). Then, the signal after subtracting the background is normalized to obtain an excitation signal s(t) with a numerical range suitable for the external excitation operation of the oscillator. This excitation signal s(t) can be used as the external excitation input at time t in the subsequent step S300.

[0033] Step S300: Determine the fluctuation of the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, and the noise intensity in the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, wherein the end time of the first time period is time t, and the length of the first time period is the first preset duration.

[0034] In this step, specifically, the detection device continuously acquires the light intensity information of the characteristic spectral lines corresponding to the target product while the etching process is in progress. Therefore, at any time t, the detection device can acquire multiple light intensity information over a period of time before time t. The detection device can determine the time period with time t as the end point and a length of the first preset duration as the first time period, and extract multiple light intensity information continuously collected within the first time period. Since the end time of the first time period is time t, the light intensity information within the first time period can reflect the recent changes in the characteristic spectral lines of the target product up to time t. Since the length of the first time period is the first preset duration, the detection device can evaluate the change characteristics of light intensity information on a fixed time scale, avoiding instability in judgment caused by determining subsequent calculation parameters based solely on the light intensity information at a single moment. The detection equipment can determine the fluctuation of light intensity information based on the degree of change among multiple light intensity information within a first time period. This fluctuation can characterize whether the light intensity information of the characteristic spectral lines corresponding to the target product exhibits significant fluctuations, the magnitude of these fluctuations, the stability of the changes, and the clarity of the trend. For example, if multiple light intensity information fluctuates around a certain level with small amplitude within the first time period, the detection equipment can determine that the fluctuation is weak. Conversely, if multiple light intensity information exhibits significant periodic fluctuations or continuous changes within the first time period, the detection equipment can determine that the fluctuation is strong. By determining the fluctuation, the detection equipment can ascertain whether the light intensity information near time t contains significant effective change components, thus providing a basis for subsequently determining the key coefficients corresponding to time t. The detection equipment also determines the noise intensity in the light intensity information of the characteristic spectral lines corresponding to the target generated within the first time period; the noise intensity can be used to characterize the strength of random fluctuations, spike interference, acquisition jitter, or non-target change components in the light intensity information within the first time period; the detection equipment can determine the noise intensity based on the degree of dispersion, short-term abrupt change, deviation from the smoothed light intensity information within the first time period, or other parameters that can reflect the degree of random interference. If the light intensity information exhibits numerous irregular fluctuations, short-term spikes, or high-frequency random disturbances within the first time period, the detection equipment can determine that the noise intensity is high; if the light intensity information changes smoothly and exhibits fewer random disturbances within the first time period, the detection equipment can determine that the noise intensity is low. Thus, before performing external excitation calculations on the oscillator, the detection equipment first evaluates the quality of the light intensity information near time t. Since the light intensity information during the etching process may simultaneously contain effective change components related to the target product and ineffective interference components caused by the acquisition environment, plasma disturbances, or equipment noise, if the detection equipment does not distinguish between the two and directly performs external excitation calculations on the oscillator, it may cause the oscillator output to be overly sensitive to noise, thereby affecting the accuracy of the oscillator state update. Therefore, by determining the fluctuation situation and noise intensity, the detection equipment can provide a data basis for the subsequent dynamic determination of key coefficients, enabling the subsequent external excitation calculations on the oscillator to adapt to the light intensity information characteristics near the current time t.

[0035] Step S400: Determine the critical coefficients corresponding to time t based on the fluctuation and noise intensity, where the critical coefficients are the adjustable coefficients in the calculation formula on which the external excitation operation of the oscillator is based.

[0036] In this step, specifically, after determining the fluctuation of light intensity information and noise intensity within the first time period, the detection device can combine the two to determine the key coefficient corresponding to time t. The key coefficient is an adjustable coefficient in the calculation formula on which the external excitation operation of the oscillator is based. That is to say, the key coefficient is not a fixed constant, but an operation parameter that the detection device can adjust according to the light intensity information characteristics near time t. By adjusting the key coefficient, the detection device can change the response mode of the external excitation operation of the oscillator to the excitation signal at time t, so that the external excitation operation of the oscillator has different response characteristics under different signal conditions. In one implementation, the detection device can determine the key coefficient corresponding to time t based on the effective change represented by the fluctuation and the interference represented by the noise intensity. When the fluctuation indicates a significant effective change in light intensity information within the first time period and the noise intensity is low, the detection device can determine the key coefficient to a value that is beneficial to improving the sensitivity of the oscillator's external excitation operation response, so that the excitation signal at time t can more fully influence the oscillator output at time t. When the noise intensity is high, the detection device can determine the key coefficient to a value that is beneficial to suppressing the influence of noise, so as to reduce the impact of random interference on the oscillator output and oscillator state update at time t. Thus, the detection device can achieve a dynamic balance between "enhancing the response to effective changes" and "suppressing noise interference". In this way, the detection equipment can adaptively determine the key coefficients used in the external excitation calculation of the oscillator based on the fluctuations in the current light intensity information and the noise intensity. Through this method, when the effective changes in light intensity information are significant, the detection equipment can maintain a high response capability in the external excitation calculation of the oscillator; when the noise intensity in the light intensity information is high, the detection equipment can make the external excitation calculation of the oscillator more effective at suppressing noise. Therefore, the detection equipment can reduce the problem of insufficient adaptability of fixed key coefficients under different etching stages and signal quality conditions, and improve the reliability of the oscillator output at time t and the oscillator state update at time t.

[0037] Step S500: Perform external excitation calculation on the oscillator based on the excitation signal and key coefficients at time t to obtain the oscillator output at time t, and update the oscillator state at time t according to the oscillator output at time t. The oscillator state at time t is used to characterize the state type of the oscillator output obtained up to time t. The state type is either periodic or chaotic. When the excitation signal at time t indicates that the target generated object has disappeared, the oscillator state at time t transitions from a periodic state to a chaotic state.

[0038] Specifically, in this step, the detection device can use the excitation signal at time t as an external excitation term and input it into the preset oscillator operation model. Based on the key coefficients determined in step S400, the external excitation operation of the oscillator is performed to obtain the oscillator output at time t. The external excitation operation of the oscillator can be understood as a nonlinear dynamic response operation, that is, simulating the dynamic response of the oscillator under external excitation through digital operation. The oscillator output can include the oscillator's displacement output, velocity output, phase trajectory point, combination of state variables, or other output quantities that can reflect the oscillator's motion characteristics; the detection device can update the oscillator state at time t based on the oscillator output at time t and the historical oscillator outputs before time t (the oscillator state reflects the oscillator output in a continuous state); the oscillator state can be a periodic state or a chaotic state (characterizing the state type presented by the oscillator output obtained up to time t). Among them, the periodic state can represent that the oscillator output has a relatively stable, repeatable or highly regular dynamic response within a number of adjacent periods; the chaotic state can represent that the oscillator output exhibits a discrete, non-periodic, irregular or highly sensitive dynamic response to input changes. In this embodiment of the invention, when the target generated object still exists, the excitation signal still contains effective signal components related to the characteristic spectral lines corresponding to the target generated object, and the external excitation operation of the oscillator can keep the oscillator in a periodic state; when the target generated object has disappeared, the effective signal components corresponding to the target generated object in the excitation signal are significantly weakened or disappeared, the external excitation conditions change, and the oscillator state transitions from a periodic state to a chaotic state; therefore, the transition of the oscillator state can be used to characterize whether the target generated object has disappeared.

[0039] In this way, the light intensity changes that are not easy to be directly identified from the noisy background are transformed into oscillator state changes that are easier to judge. Compared with directly judging whether the light intensity is lower than the threshold, the external excitation operation of the oscillator uses the nonlinear response of the oscillator to the excitation signal change to make the weak signal change caused by the disappearance of the target product manifest as a transition from a periodic state to a chaotic state, thereby improving the stability of etching endpoint identification.

[0040] For example, when the TSV deep silicon etching has not yet reached its end, SiF continues to be generated in the plasma. The excitation signal s(t) generated by the detection device based on the intensity information of the SiF characteristic spectral lines still has an effective component that can maintain the periodic response of the oscillator. At this time, the detection device performs external excitation calculation on the oscillator based on s(t). The oscillator output at time t and the oscillator output at previous times together exhibit a relatively stable repeating trajectory. Therefore, the oscillator state at time t is updated to a periodic state. As the etching continues, when the generation of SiF decreases significantly or even disappears, the excitation signal s(t) can no longer maintain the original periodic response. The oscillator output calculated by the detection device gradually exhibits irregular and discrete dynamic characteristics. At this time, the oscillator state at time t is updated to a chaotic state.

[0041] Step S600: Determine whether the etching endpoint has been reached based on the oscillator state at time t.

[0042] In this step, specifically, after updating the oscillator state at time t, the detection device can determine whether the oscillator state indicates that the target product has disappeared; if the oscillator state is still in a periodic state, it can be determined that the effective excitation signal corresponding to the target product still exists, and thus it can be determined that the etching has not yet reached the etching endpoint; if the oscillator state transitions from a periodic state to a chaotic state, it can be determined that the excitation signal indicates that the target product has disappeared or significantly weakened, and thus it can be determined that the etching has reached the etching endpoint. To avoid misjudgment caused by transient noise, the detection device can further determine whether the chaotic state continues to meet preset conditions after detecting that the oscillator state has transitioned from a periodic state to a chaotic state. For example, it can determine whether the chaotic state is continuously maintained for a preset number of sampling times, whether it lasts for a preset time length, or whether it remains chaotic within a preset number of etching cycles. When the preset conditions are met, the detection device can output an etching endpoint arrival signal. When the preset conditions are not met, the detection device can continue to execute steps S100 to S500 to continue monitoring the etching process.

[0043] The etching endpoint arrival signal can be used to alert operators or sent to the etching equipment controller to stop the current etching step, shut off or switch the etching gas, adjust the RF power, proceed to the next process step, or perform other endpoint control operations. In this way, the transition result of the oscillator state is converted into the endpoint judgment result in the etching process, realizing the real-time determination of the etching endpoint. Since the judgment is based on whether the oscillator state transitions from a periodic state to a chaotic state, rather than simply based on the absolute light intensity of the characteristic spectral lines of the target product, it can reduce the influence of weak light intensity, low signal-to-noise ratio and background noise on the endpoint judgment. For example, during a TSV deep silicon etching process, the detection device continuously acquires the light intensity information of SiF characteristic spectral lines at each sampling time and generates corresponding excitation signals sequentially. Before t=30, the detection device determines that the oscillator state is continuously in a periodic state based on the oscillator output, and therefore judges that the etching has not yet reached the etching endpoint. After t=30, the detection device determines that the oscillator state transitions from a periodic state to a chaotic state based on the oscillator output, and this chaotic state is maintained for multiple consecutive sampling times. At this time, the detection device can determine that SiF has disappeared or significantly weakened, and thus determine that the etching has reached the etching endpoint, and outputs an etching endpoint arrival signal.

[0044] Through steps S100-S600, the intensity sequence of the characteristic spectral lines corresponding to the target product is continuously acquired during the etching process. This intensity sequence is used to generate an excitation signal, and then an external excitation operation is performed on the oscillator based on this excitation signal. This transforms the changes in the characteristic spectral lines, which were originally weak and easily submerged by noise during the etching process, from being judged solely by linear indicators such as absolute intensity, modulation amplitude, or frequency domain energy, into the evolution characteristics of the oscillator state. Since the oscillator state can transition from a periodic state to a chaotic state when the excitation signal indicates that the target product has disappeared, the abrupt change in the oscillator state can be used to characterize the process of the target product from existence to disappearance, thereby determining whether the etching has reached the etching endpoint. This reduces the dependence on the amplitude and signal-to-noise ratio of the characteristic spectral line intensity signal. Even when the emission spectrum signal of characteristic products such as SiF is extremely weak due to the low aperture ratio of the TSV, the endpoint criterion can still be extracted relatively stably. This reduces the misjudgment caused by the intensity threshold method, modulation depth method, spectrum analysis method, and time-domain synchronous averaging method due to the noise background being close to the effective signal, thus improving the reliability and stability of the etching endpoint determination. Moreover, the adjustable coefficients in the calculation formula on which the external excitation calculation of the oscillator is based are not fixed, but dynamically determined by the detection device based on the fluctuation of light intensity information and noise intensity of the characteristic spectral lines corresponding to the target generated within the first time period, so that the oscillator excitation calculation can be better adapted to the current signal quality.

[0045] Figure 2 This is another flowchart illustrating the etching endpoint determination method provided in this embodiment of the invention; please refer to it as well. Figures 1-2 .

[0046] In some implementations, the calculation formula on which the external excitation calculation of the oscillator is based is:

[0047] in, For time markers, for The oscillator output at any given moment, for right the second derivative of, is the first derivative of , is to the power of three, is to the power of five, is the excitation signal at time , is periodic driving, is the frequency of periodic driving, wherein the frequency of periodic driving is the angular frequency corresponding to the gas switching frequency in the etching process, and the period of periodic driving is , and the gas is a gas used for etching, is the amplitude of periodic driving, is an adjustable damping coefficient; A periodic state is a state in which the oscillator output changes periodically with the periodic driving. In the periodic state, the oscillator output changes repeatedly over a plurality of consecutive periods of the periodic driving. A chaotic state is a state in which the oscillator output does not change periodically with the periodic driving. In the chaotic state, the oscillator output changes aperiodically over a plurality of consecutive periods of the periodic driving.

[0048] In this embodiment, when performing oscillator external excitation calculation based on an excitation signal to obtain an oscillator output at time t, the excitation signal at time t may be added to the periodic driving in a calculation formula (which is a Duffing equation) based on which the oscillator external excitation calculation is performed ( Figure 2 expressed as "inputting into the Duffing oscillator" herein), to obtain a sum of excitation , and determine the oscillator output at time t in the calculation formula according to the sum of excitation; Specifically, the Duffing equation is a type of equation used to describe the dynamic response of a nonlinear oscillator; compared with an ordinary linear oscillator, the Duffing equation contains a nonlinear term, so its oscillator state is relatively sensitive to changes in an external excitation signal, and can exhibit a periodic state, a chaotic state, and a transition between the two states under specific parameter conditions; based on this characteristic, the excitation signal generated from the light intensity information of the characteristic spectral line corresponding to the target product can be used as an external excitation and superimposed on the periodic driving of a preset Duffing equation, so that the excitation signal participates in the calculation of the dynamic response of the oscillator; in this way, the change of the excitation signal caused by whether the target product exists or disappears can be converted into the change of the oscillator output and the oscillator state; In the above formula (1), The periodic drive set for the Duffing equation is used to ensure the oscillator has a predetermined periodic response under preset operating conditions. Since etching processes such as Bosch etching typically involve periodic switching between etching and passivation gases, the frequency of the periodic drive can be set to the angular frequency corresponding to the gas switching frequency during etching (correspondingly, the period of the periodic drive can be...). Divide by that frequency to express, i.e. The reason for this configuration is that in Bosch etching, the alternating introduction of SF6 etching gas and C4F8 passivation gas causes the intensity of the characteristic spectral lines of the target product to exhibit periodic fluctuations related to the gas switching cycle. Aligning the periodic drive frequency of the Duffing equation with this process cycle allows the oscillator to maintain a stable periodic response (output) under "normal etching cycle fluctuations". In this way, when the target product disappears and the excitation signal no longer conforms to this periodic change, the oscillator is more likely to transition from the periodic state to the chaotic state, thereby highlighting the abnormal change corresponding to the etching endpoint. Therefore, those skilled in the art will understand that the periodic state represents a stable periodic correspondence between the oscillator output and the periodic drive: Because the periodic drive is When the detection device performs external excitation calculations on the oscillator based on the periodic drive and the excitation signal, if the oscillator output can change periodically with the periodic drive and repeat within multiple consecutive periodic drive cycles, it indicates that the oscillator output can form a stable and repetitive response under the action of the periodic drive. At this time, the detection device can determine that the oscillator is in a periodic state. That is, the periodic state does not only indicate that there is a numerical change in the oscillator output, but also that the change pattern of the oscillator output can correspond to the change rhythm of the periodic drive and repeat within multiple consecutive cycles. Correspondingly, the chaotic state is used to characterize the unstable periodic correspondence between the oscillator output and the periodic drive: If the oscillator output does not change periodically with the periodic drive and exhibits non-repetitive changes within multiple consecutive periodic drives, it indicates that the oscillator output has failed to form a stable repetitive response under the action of the periodic drive. In this case, the detection device can determine that the oscillator is in a chaotic state. In a chaotic state, the oscillator output exhibits irregular changes relative to the periodic drive, and the oscillator output within multiple consecutive periods is unlikely to show the same or similar change pattern. Therefore, the detection device can distinguish whether the oscillator is in a periodic state or a chaotic state by judging whether the oscillator output changes periodically with the periodic drive and whether the oscillator output exhibits repetitive changes within multiple consecutive periodic drives. Specifically, the detection equipment can receive an external trigger signal (gas switching synchronization signal) from the etching equipment and accurately extract the angular frequency corresponding to the gas switching frequency. ( (for the gas switching cycle), and apply it to the above Duffing equation as the frequency driven by the cycle; This represents the excitation signal generated from the intensity information of the characteristic spectral lines corresponding to the target product, which is used to incorporate the spectral changes of the target product into the external excitation calculation of the oscillator; and The result after addition , that is, the excitation and the combined effect on the oscillator; When the detection equipment performs calculations, it can acquire the excitation signal at time t. And calculate the periodic drive Then add the two together to get the excitation sum at time t; Subsequently, the detection equipment substitutes the excitation into the aforementioned Duffing equation, and combines it with the oscillator output from the previous moment, the oscillator's first derivative from the previous moment, and preset coefficients. Amplitude driven by period Frequency driven by period The oscillator output at time t is determined by numerical iteration using parameters such as [parameter t]. The numerical iteration method can be a discretization operation method suitable for solving differential equations, such as a recursive solution method based on the state variables of the current time and the previous time. This embodiment of the invention does not specifically limit this method. Thus, the excitation signal is not used as a threshold judgment object alone, but together with the periodic drive in the Duffing equation, it constitutes the excitation sum, and affects the oscillator output through the excitation sum. If the target product still exists, the excitation signal generated by the characteristic spectral line corresponding to the target product still has a certain effective component, and the excitation sum can keep the oscillator in a dynamic response corresponding to the state in which the target product exists. If the target product has disappeared, the effective component in the excitation signal changes, the excitation sum changes accordingly, and thus the oscillator output in the calculation formula changes accordingly, and may further cause the oscillator state to transition from a periodic state to a chaotic state.

[0049] In some implementations, the amplitude of the periodic drive is the product of the critical amplitude and a first preset multiple, where the first preset multiple is less than 1 and greater than 0, and the absolute value of the difference between the critical amplitude and the amplitude of the periodic drive is less than a preset difference threshold. The critical amplitude is determined through the following steps: Set the excitation signal to 0; Multiple preset amplitudes are determined, wherein the numerical difference between any two adjacent preset amplitudes is a preset step size; Perform a calibration pre-step, which includes determining the oscillator output at multiple times corresponding to each preset amplitude in the calculation formula in ascending order of numerical values, and using the oscillator output at multiple times corresponding to the Poincaré cross-section discrete points, and determining the Poincaré cross-section variance corresponding to the preset amplitude based on the Poincaré cross-section discrete points. Based on the Poincaré cross-sectional variances corresponding to multiple preset amplitudes, a critical amplitude is determined among the multiple preset amplitudes. The critical amplitude is the first preset amplitude that makes the corresponding Poincaré cross-sectional variance less than a first preset variance threshold during the execution of the calibration pre-step.

[0050] In this embodiment, in order to enable the calculation formula to maintain high sensitivity to extremely weak periodic feature changes during the etching process, the amplitude of the periodic drive can be calibrated first, and the amplitude of the periodic drive can be located near the critical amplitude. Specifically, the amplitude of the periodic drive can be set as the product of the critical amplitude and a first preset multiple, where the first preset multiple is less than 1 and greater than 0. For example, the first preset multiple can be 0.98. In this way, the amplitude of the periodic drive is not directly equal to the critical amplitude, but slightly lower than the critical amplitude. At the same time, the absolute value of the difference between the critical amplitude and the amplitude of the periodic drive is less than a preset difference threshold, so that the amplitude of the periodic drive is sufficiently close to the critical amplitude (the first preset multiple and the preset difference threshold can be determined by technicians through repeated adjustments). Therefore, the Duffing equation expressed by the calculation formula (1) can work at a position close to the state transition boundary. That is, it is not easy to enter the periodic state stably when there is no effective same-frequency excitation input, but it can be induced into the periodic state by the weak same-frequency periodic component when there is a weak same-frequency periodic component. The critical amplitude can be determined through pre-calibration. During calibration, the frequency of the periodic drive can be set to the angular frequency corresponding to the gas switching frequency during the etching process. For example, in the Bosch etching process, the etching gas and passivation gas alternate periodically. If the gas switching period is T, the angular frequency corresponding to the periodic drive can be set to... This allows the cycle drive to match the gas switching cycle during the etching process; simultaneously, the damping coefficient... The value is set at 0.5; In addition, when calibrating the critical amplitude, the excitation signal can be set to 0. That is, the external excitation obtained from the characteristic spectral intensity information of the target product is not introduced for the time being, and only the influence of the periodic drive itself on the state of the oscillator detection module is examined. In determining the critical amplitude, the detection device can first determine multiple preset amplitudes. The preset amplitude with the smallest value among the multiple preset amplitudes can be 0, and the numerical difference between any two adjacent preset amplitudes is the preset step size; for example, the preset step size can be 0.001, then the multiple preset amplitudes can be 0, 0.001, 0.002, 0.003, etc. in sequence. Subsequently, the detection equipment can input each preset amplitude as a candidate amplitude for periodic driving into the above Duffing equation in ascending order of numerical values, and calculate the corresponding oscillator output by numerical simulation. For example, the fourth-order Runge-Kutta method can be used to solve the above Duffing equation to obtain the oscillator output of the above Duffing equation at multiple times. The detection equipment uses the oscillator output obtained at these moments as discrete points of the Poincaré cross section, and calculates the variance of the Poincaré cross section corresponding to the preset amplitude based on these discrete points. If the discrete points of the Poincaré cross section are relatively scattered, it indicates that the phase trajectory of the oscillator output is still in a chaotic state, and the corresponding variance of the Poincaré cross section is large. If the discrete points of the Poincaré cross section gradually converge or concentrate, it indicates that the phase trajectory of the oscillator output has entered a large-scale periodic state, and the corresponding variance of the Poincaré cross section is small. As the detection equipment gradually increases the preset amplitude according to the preset step size and inputs it into the above Duffing equation, it can compare the relationship between the Poincaré cross-sectional variance corresponding to each preset amplitude and the first preset variance threshold. When a certain preset amplitude first makes the corresponding Poincaré cross-sectional variance less than the first preset variance threshold, it indicates that the above Duffing equation has for the first time transitioned from a chaotic state to a periodic state under the action of the preset amplitude. Therefore, the preset amplitude can be determined as the critical amplitude. The first preset variance threshold can be preset through experiments or simulations, for example, it can be set to 0.02; the critical amplitude determined therefrom can characterize the critical periodic driving amplitude required for the oscillator detection module to transition from a chaotic state to a periodic state under no external excitation conditions. Thus, the amplitude of the periodic drive is set near the critical amplitude, making the Duffing equation more sensitive to the weak same-frequency periodic components in the characteristic spectrum of the target product. At the same time, since the critical amplitude is obtained by Poincaré cross-sectional variance calibration under no external excitation conditions, it can improve the reliability of subsequent etching endpoint determination and reduce the possibility of misjudgment caused by noise fluctuations or background drift.

[0051] For example, in a certain TSV deep silicon etching process using Bosch periodic etching, with a gas switching period T of 120ms, the angular frequency corresponding to the periodic drive can be set to... When calibrating the critical amplitude, the excitation signal is set to 0 and the damping coefficient is set to 0.5; then, starting from the preset amplitude of 0, the preset amplitude is gradually increased in a preset step size of 0.001. For each preset amplitude, the Duffing equation is solved using the fourth-order Runge-Kutta method to obtain the oscillator output at multiple time points, forming discrete points of the Poincaré cross section. The variance of the Poincaré cross section corresponding to the preset amplitude is then calculated. For example, the detection equipment can calculate the corresponding variance of the Poincaré cross section using the following formula:

[0052] in, Preset amplitude The corresponding Poincaré section variance, The length of the sliding window. For the first in the sliding window Discrete points of a Poincaré section This is the mean of multiple discrete points on the Poincaré section within the sliding window; When the preset amplitude is 0.973, the discrete points of the Poincaré section obtained by the detection device at 5 time intervals are, for example, 1.050, 0.550, 0.922, 0.678, and 0.800; the detection device first determines the mean value of this set of discrete points of the Poincaré section:

[0053] Subsequently, the detection equipment calculates the squared deviation of each discrete point on the Poincaré section from the mean, and obtains:

[0054] Therefore, the variance of the Poincaré section corresponding to the preset amplitude of 0.973 is approximately 0.031; When the preset amplitude is 0.974, the discrete points of the Poincaré section obtained by the detection device at 5 time periods are, for example, 1.020, 0.580, 0.929, 0.671, and 0.800; the mean of this set of discrete points of the Poincaré section is 0.800, and the detection device calculates:

[0055] Therefore, the variance of the Poincaré section corresponding to the preset amplitude of 0.974 is approximately 0.026; When the preset amplitude is 0.975, the discrete points of the Poincaré section sampled by the detection device at five time periods are, for example, 1.000, 0.600, 0.912, 0.688, and 0.800. The mean of this set of discrete points is 0.800. The detection device calculates:

[0056] Therefore, the variance of the Poincaré section corresponding to the preset amplitude of 0.975 is approximately 0.021; When the preset amplitude is 0.976, the discrete points of the Poincaré section sampled by the detection device at 5 time periods are, for example, 0.990, 0.610, 0.894, 0.706, and 0.800; the mean of this set of discrete points of the Poincaré section is 0.800, and the detection device calculates:

[0057] Therefore, the variance of the Poincaré section corresponding to the preset amplitude of 0.976 is approximately 0.018; If the first preset variance threshold is 0.02, then the preset amplitude of 0.976 is the first preset amplitude that makes the variance of the Poincaré section less than the first preset variance threshold. Therefore, 0.976 can be determined as the critical amplitude. Further, if the first preset multiple is 0.98, then the amplitude of the periodic drive can be determined as 0.98 × 0.976 = 0.95648. If the preset difference threshold is 0.03, then the absolute value of the difference between the critical amplitude of 0.976 and the amplitude of the periodic drive of 0.95648 is 0.01952, which is less than the preset difference threshold, indicating that the amplitude of the periodic drive is near the critical amplitude. In the actual etching endpoint determination process, when the amplitude of the aforementioned periodic drive is used as the operating point, the Duffing equation is in a sensitive state close to the transition boundary between the chaotic state and the periodic state. When the etching has not yet reached the endpoint, the characteristic spectral lines corresponding to the target product still contain weak periodic components related to the gas switching cycle. These weak periodic components can work together with the periodic drive to maintain or transition the oscillator detection module to the periodic state. When the etching reaches the endpoint, the target product decreases or disappears, and the same-frequency periodic component in the excitation signal weakens or disappears. The oscillator detection module then transitions from the periodic state to the chaotic state, thereby determining the etching endpoint.

[0058] In some implementations, the critical coefficient is the damping coefficient in the calculation formula. The critical coefficient corresponding to time t is determined based on the fluctuation conditions and noise intensity, including: The fluctuation characterization value is determined based on the fluctuation situation. The fluctuation characterization value is used to characterize the degree of fluctuation of light intensity information within the first time period. The noise characterization value is determined based on the noise intensity, whereby the noise characterization value is used to characterize the degree of noise interference in the light intensity information within the first time period. The fluctuation characterization value and the noise characterization value are normalized to obtain the normalized fluctuation characterization value and the normalized noise characterization value. Based on the preset benchmark damping coefficient, normalized fluctuation characterization value, and normalized noise characterization value, the damping coefficient corresponding to time t is determined. The damping coefficient corresponding to time t is positively correlated with the normalized noise characterization value and negatively correlated with the normalized fluctuation characterization value.

[0059] In this embodiment, the key coefficient is the damping coefficient in the calculation formula (Duffing equation). The detection device adaptively determines the damping coefficient corresponding to time t based on the fluctuation of light intensity information and noise intensity within the first time period. Specifically, the detection equipment can first determine the fluctuation characterization value based on the fluctuation situation. The fluctuation characterization value is used to characterize the degree of fluctuation of light intensity information within the first time period. This degree of fluctuation can be understood as the degree of fluctuation in light intensity information within the first time period that can reflect the changes in the target generated object. If there are obvious, continuous, or regular changes in light intensity information within the first time period, the detection device can determine a large fluctuation value; if the changes in light intensity information are weak or remain basically stable within the first time period, the detection device can determine a small fluctuation value. The detection equipment can also determine the noise characterization value based on the noise intensity. The noise characterization value is used to characterize the degree of noise interference in the light intensity information within the first time period. If there are many irregular jitters, spike interferences or high-frequency random fluctuations in the light intensity information within the first time period, the detection equipment can determine a larger noise characterization value. If the light intensity information is relatively smooth and has fewer random disturbances within the first time period, the detection equipment can determine a smaller noise characterization value. By determining the fluctuation characterization value and the noise characterization value respectively, the detection equipment can distinguish the effective fluctuation degree related to the change of the target product in the light intensity information and the noise interference degree related to random disturbance, thereby avoiding misjudging the noise interference as the effective change corresponding to the target product. After obtaining the fluctuation and noise characterization values, the detection equipment can normalize these values ​​to obtain normalized fluctuation and noise characterization values. The purpose of normalization is to convert fluctuation and noise characterization values ​​with different dimensions and numerical ranges to a unified numerical range, such as between 0 and 1, so that they can be compared and combined in the same calculation formula. A larger normalized fluctuation characterization value indicates a stronger fluctuation in light intensity information within the first time period; a larger normalized noise characterization value indicates a stronger degree of noise interference in the light intensity information within the first time period. Subsequently, the testing equipment can determine the damping coefficient corresponding to time t based on the preset benchmark damping coefficient, normalized fluctuation characterization value, and normalized noise characterization value (the damping coefficient corresponding to time t is positively correlated with the normalized noise characterization value and negatively correlated with the normalized fluctuation characterization value). That is, when the normalized noise characterization value increases, the detection device can increase the damping coefficient at time t to enhance the ability of the oscillator corresponding to the Duffing equation to suppress noise interference; when the normalized fluctuation characterization value increases, the detection device can decrease the damping coefficient at time t to improve the response sensitivity of the oscillator corresponding to the Duffing equation to changes in effective light intensity; thus, the detection device can adaptively adjust the damping term of the Duffing equation according to the quality of the light intensity information in the current first time period, making the external excitation operation of the oscillator more stable when the noise is strong and more sensitive when the effective fluctuation is obvious. Specifically, the testing equipment can determine the candidate damping coefficient according to the following formula:

[0060] in, Candidate damping coefficients, The reference damping coefficient, This is the normalized noise characterization value. Here, α represents the normalized fluctuation characterization value, β represents the noise adjustment weight, and β represents the fluctuation adjustment weight. because The coefficient before the normalized noise characterization value is positive, therefore the larger the candidate damping coefficient is; because The coefficient before is negative, so the larger the normalized fluctuation characterization value, the smaller the candidate damping coefficient; the detection equipment can directly determine the candidate damping coefficient as the damping coefficient corresponding to time t, or after obtaining the candidate damping coefficient, it can perform amplitude limiting processing on the candidate damping coefficient according to the preset lower limit and the preset upper limit of the damping coefficient to obtain the damping coefficient corresponding to time t. Limiting can prevent the oscillator output from being overly sensitive to noise due to an excessively small damping coefficient, and it can also prevent the oscillator output from responding slowly to the excitation signal due to an excessively large damping coefficient. Once the damping coefficient at time t is determined, the detection device can use this damping coefficient in the Duffing equation of formula (1); In this way, the detection equipment can adaptively determine the damping coefficient in the Duffing equation based on the fluctuation of light intensity information and the degree of noise interference within the first time period. When the effective fluctuation is strong, it can improve the response capability of the external excitation operation of the oscillator and enhance the suppression of noise when the noise intensity is high, thereby improving the reliability of the oscillator output at time t and the oscillator state update at time t.

[0061] For example, the detection device determines the first time period as [t-50ms, t] at time t, and acquires multiple light intensity information of the characteristic spectral lines corresponding to the target product within the first time period; the detection device determines the fluctuation characterization value based on the fluctuation degree of the light intensity information within the first time period, and determines the noise characterization value based on the random jitter degree in the light intensity information within the first time period. Assuming the detection equipment normalizes the fluctuation and noise characterization values, it obtains the normalized fluctuation characterization value. Normalized noise characterization value Meanwhile, the testing equipment has a preset reference damping coefficient. Noise adjustment weight α = 0.20, fluctuation adjustment weight β = 0.10; The testing equipment calculates the candidate damping coefficient according to formula (3): ; If the preset lower limit of the damping coefficient is 0.40 and the preset upper limit of the damping coefficient is 0.70, then 0.47 is within this range. The detection device can determine 0.47 as the damping coefficient corresponding to time t. In this case, since the normalized fluctuation characterization value is large and the normalized noise characterization value is small, the detection device appropriately reduces the damping coefficient so that the external excitation calculation of the oscillator has a high response sensitivity to the excitation signal at time t.

[0062] In some implementations, determining whether the etching endpoint has been reached based on the oscillator state at time t includes: The oscillator outputs at multiple periodic times in the oscillator state at time t are extracted as discrete points of the Poincaré section, where the periodic times are positive integer multiples of the gas switching period during the etching process, and the gas is the gas used for etching. Determine the length of the sliding window, and group the discrete points of the Poincaré section according to the length of the sliding window to obtain multiple groups of sliding windows; For each set of sliding windows, the variance of the discrete points of the Poincaré cross section within the sliding window is determined as the variance of the Poincaré cross section of the sliding window. When the variance of the Poincaré cross section is less than the second preset variance threshold, it indicates that the oscillator state of the corresponding period of the sliding window is in a periodic state. When the variance of the Poincaré cross section is greater than or equal to the second preset variance threshold, it indicates that the oscillator state of the corresponding period of the sliding window is in a chaotic state. Determine whether the etching endpoint has been reached based on the variance of multiple Poincaré sections.

[0063] In this embodiment, when the detection device determines whether the etching has reached the etching endpoint based on the oscillator state at time t, it does not directly rely on the single-point value of the oscillator output at a certain time, but rather on the oscillator state based on the distribution characteristics of the oscillator output at multiple cycle times, thereby improving the stability of the endpoint determination. Specifically, the detection equipment can continuously obtain the oscillator output during the etching process, and extract the oscillator output corresponding to multiple cycle times from the oscillator state at time t, using the gas switching cycle (the gas used for etching) during the etching process as the time reference. Among them, the cycle times can be positive integer multiples of the gas switching cycle, such as the end time of the first gas switching cycle, the end time of the second gas switching cycle, the end time of the third gas switching cycle, etc. In actual implementation, the cycle times can also correspond to the rising or falling edge of the gas switching trigger signal. Therefore, by sampling the oscillator output at a time corresponding to the etching gas switching cycle, the detection equipment can match the sampling time with the periodic process changes during the etching process. The detection equipment uses the oscillator output sampled at multiple time points as discrete points of the Poincaré section. Figure 2 The Poincaré point is referred to as the "Poincaré point" in Chinese. That is, the detection device does not analyze the entire trajectory of the oscillator output in continuous time, but samples the oscillator output once at a fixed phase position in each gas switching cycle, which is equivalent to periodically intercepting the phase trajectory of the oscillator, thereby obtaining the discrete point sequence of the Poincaré section. If the oscillator is in a periodic state, the multiple Poincaré cross-section discrete points will usually converge within a small range, and the fluctuations between the discrete points will be small. If the oscillator is in a chaotic state, the multiple Poincaré cross-section discrete points will usually be more dispersed, and the fluctuations between the discrete points will be larger. Therefore, the detection device can use the degree of dispersion of the Poincaré cross-section discrete points (Poincaré cross-section variance) to characterize the oscillator state. The detection equipment can also determine the length of the sliding window and group the discrete points of the Poincaré section according to the length of the sliding window to obtain multiple groups of sliding windows. The length of the sliding window can be set according to the detection response speed and the stability of state recognition. For example, it can be set to 3 cycles, 4 cycles or 5 cycles. The shorter the sliding window length, the easier it is for the detection equipment to respond quickly to changes in the state of the oscillator. The longer the sliding window length, the easier it is for the detection equipment to suppress single-point fluctuations caused by accidental noise. Specifically, if the length of the sliding window is W, the detection device can select the (n-W+1)th to the nth Poincaré cross-section discrete points to form a set of sliding windows for the discrete points of the Poincaré cross-section corresponding to the nth period time. For each set of sliding windows, the detection device can determine the variance of each discrete point of the Poincaré section within the sliding window, and use this variance as the variance of the Poincaré section of the sliding window; for example, the detection device can determine the variance of the Poincaré section according to the following formula (4):

[0064] in, Indicates the length of the sliding window. Indicates the first The oscillator output obtained by sampling at each cycle time point For the variance of the Poincaré section, for (Variance) is an abbreviation for calculating the variance of discrete points on the Poincaré section within a sliding window. If the variance of the Poincaré section corresponding to a certain sliding window is less than the second preset variance threshold ( Figure 2 If the variance of the Poincaré cross section within a sliding window is greater than or equal to the second preset variance threshold, it indicates that the discrete points of the Poincaré cross section within the sliding window are relatively concentrated, and the detection device can determine that the oscillator state during the corresponding period of the sliding window is in a periodic state. If the variance of the Poincaré cross section corresponding to a certain sliding window is greater than or equal to the second preset variance threshold, it indicates that the discrete points of the Poincaré cross section within the sliding window are relatively dispersed, and the detection device can determine that the oscillator state during the corresponding period of the sliding window is in a chaotic state. The second preset variance threshold can be obtained through offline calibration so that it can distinguish between the low variance distribution in the periodic state and the high variance distribution in the chaotic state. Finally, the detection equipment can determine whether the etching has reached the etching endpoint based on the variance of the Poincaré cross section corresponding to the multiple sliding windows. For example, when the etching has not reached the etching endpoint, there are still periodic components related to the gas switching cycle in the characteristic spectral lines corresponding to the target product. These periodic components can keep the oscillator detection module in a periodic state. At this time, the variance of the multiple Poincaré cross sections obtained by the detection equipment is usually less than the second preset variance threshold. As etching reaches its endpoint, the target product decreases or its corresponding periodic spectral features weaken or even disappear. The oscillator detection module transitions from a periodic state to a chaotic state. At this time, the variance of the Poincaré section obtained by the detection device will increase, which is greater than or equal to the second preset variance threshold in one or more sliding windows. In this way, the detection equipment can transform the "whether the trajectory output by the oscillator is stable and convergent" into a numerical judgment of the variance of the Poincaré section, thereby identifying periodic and chaotic states with a small amount of computation. The detection equipment does not rely on the light intensity threshold at a single moment, but determines the endpoint based on the topological changes of the oscillator state, which is beneficial to improving the reliability of etching endpoint determination under low signal-to-noise ratio conditions.

[0065] For example, the gas switching cycle in a certain TSV deep silicon etching process is 120ms. The detection device uses this gas switching cycle as a reference and samples the oscillator output of the oscillator detection module at cycles of 120ms, 240ms, 360ms, 480ms, 600ms, 720ms, 840ms, and 960ms, and uses the sampled oscillator output as the discrete points of the Poincaré section. Assume the oscillator outputs sampled by the detection device in the first 5 cycles are as follows: , , , , Furthermore, if the detection device sets the sliding window length to 3 cycles, then the detection device can obtain the first set of sliding windows. The second set of sliding windows The third set of sliding windows ; The detection equipment calculates the variance of the discrete points of the Poincaré cross section within the aforementioned sliding windows. The variance of the Poincaré cross section in the first group of sliding windows is approximately 0.000067, the variance in the second group of sliding windows is approximately 0.000156, and the variance in the third group of sliding windows is approximately 0.000156. If the second preset variance threshold is 0.02, then the variances of the aforementioned Poincaré cross sections are all less than the second preset variance threshold. The detection equipment can determine that the oscillator state at the corresponding time of these sliding windows is in a periodic state and that the etching has not yet reached the etching endpoint. In another scenario, if the etching process is nearing or has reached its endpoint, the oscillator output sampled by the detection device at subsequent cycle times becomes... , , , , Then the detection equipment can form a sliding window. , , ; The detection equipment calculated that the variances of the Poincaré sections of the aforementioned sliding windows were approximately 0.0682, 0.1016, and 0.1061, respectively, all of which were greater than the second preset variance threshold of 0.02. Therefore, the detection equipment could determine that the oscillator state at the corresponding time of these sliding windows was in a chaotic state. If the detection equipment further determined that the chaotic state had been maintained for multiple gas switching cycles, the detection equipment could determine that the oscillator state had transitioned from a periodic state to a chaotic state, and thus determine that the etching had reached the etching endpoint.

[0066] In some implementations, determining whether the etching endpoint has been reached is based on the variance of multiple Poincaré sections, including: If the duration of the oscillator state in a chaotic state at time t of multiple Poincaré cross-section variance indicators has exceeded a preset duration threshold, it is determined that the etching endpoint has been reached. The first cycle time that satisfies the condition that the duration of the oscillator state in a chaotic state has exceeded the preset duration threshold is determined as the etching endpoint arrival time. The preset duration threshold is N gas switching cycles, where N is a positive integer greater than 1.

[0067] In this embodiment, after obtaining the Poincaré cross-sectional variances corresponding to multiple sliding windows, the detection device can compare each Poincaré cross-sectional variance with a second preset variance threshold, and determine the oscillator state of the corresponding sliding window period based on the comparison result. When the Poincaré cross-sectional variance corresponding to a certain sliding window is less than the second preset variance threshold, the detection device can determine that the oscillator state of the period corresponding to that sliding window is in a periodic state. When the Poincaré cross-sectional variance corresponding to a certain sliding window is greater than or equal to the second preset variance threshold, the detection device can determine that the oscillator state of the period corresponding to that sliding window is in a chaotic state. Since the variance of a single Poincaré section exceeding the second preset variance threshold may be caused by transient noise, sampling jitter, or short-term disturbances within the etching cavity, the detection device does not directly determine whether the etching has reached its endpoint based on the state result of a single sliding window. Instead, it further calculates the duration for which the variances of multiple Poincaré sections continuously indicate the duration of the oscillator's state in a chaotic state. The detection device can set a preset duration threshold to N gas switching cycles, where N is a positive integer greater than 1 (e.g., N=3). During continuous monitoring of the oscillator state, if the device detects a transition from a periodic state to a chaotic state, it can accumulate the duration of the chaotic state starting from that transition moment. If, in subsequent gas switching cycles, multiple Poincaré cross-sectional variances are greater than or equal to a second preset variance threshold, causing these Poincaré cross-sectional variances to continuously indicate that the oscillator state is in a chaotic state, and the duration of this chaotic state at time t is... If the duration exceeds the preset time threshold (N gas switching cycles), the detection device can determine that the etching has reached the etching endpoint and determine the first cycle time when the duration of the oscillator in a chaotic state exceeds the preset time threshold as the etching endpoint arrival time; that is, the detection device does not determine the entire duration of the chaotic state as the etching endpoint, but rather takes the corresponding cycle time (i.e., the first cycle time when the duration of the chaotic state exceeds the preset time threshold) as the etching endpoint arrival time when the duration of the chaotic state first exceeds the preset time threshold. Correspondingly, if the detection device does not detect the transition of the oscillator state from a periodic state to a chaotic state at time t, or if the detection device detects that part of the Poincaré cross-sectional variance indicates that the oscillator state at time t is in a chaotic state, but the chaotic state does not last longer than the preset duration threshold, then the detection device can determine that it is not yet sufficient to confirm that the etching has reached the etching endpoint, and continue to monitor the oscillator output at subsequent periodic times and the Poincaré cross-sectional variance corresponding to the subsequent sliding window; After the detection equipment determines that the etching has reached its endpoint, it can also output an endpoint arrival signal to the main controller of the etching equipment. For example, the detection equipment can send the endpoint arrival signal to the main controller of the etching equipment through a digital I / O (Input / Output) interface. This endpoint arrival signal can be represented by a TTL (Transistor-Transistor Logic) level transition, such as a TTL low-level transition. After receiving the endpoint arrival signal, the main controller of the etching equipment can stop the etching process or execute subsequent control actions corresponding to the etching endpoint. Thus, the detection equipment can not only determine the etching endpoint based on the duration of the oscillator's state in a chaotic state, but also convert the determination result into a control signal that the etching equipment can recognize, so as to realize the linkage between the endpoint determination result and the etching equipment control process. In this way, the detection equipment bases its judgment of "whether the etching endpoint has been reached" on the time continuity condition that the oscillator is in a continuous chaotic state, rather than on a single variance value or a single sampling point. The detection equipment can reduce the risk of misjudgment caused by instantaneous disturbances and can output the endpoint arrival signal to the main controller of the etching equipment in a timely manner through the digital I / O interface after confirming the endpoint.

[0068] For example, in a certain TSV deep silicon etching process, the gas switching cycle is 120ms. The detection device sets the preset duration threshold to 3 gas switching cycles, i.e., 360ms, and sets the second preset variance threshold to 0.02. During the etching process, the detection device continuously extracts the oscillator output to form discrete points of the Poincaré cross section, and calculates multiple Poincaré cross section variances based on a sliding window. Assuming that when the etching is proceeding normally, the Poincaré cross section variances obtained by the detection device are 0.00008, 0.00012, 0.00010 and 0.00015, respectively, and these Poincaré cross section variances are all less than the second preset variance threshold of 0.02, the detection device determines that the oscillator state at the corresponding time is in a periodic state and determines that the etching has not yet reached the etching endpoint. Subsequently, as the etching neared its end, the variance of the Poincaré section obtained by the detection device successively changed to 0.034, 0.057, 0.081, 0.076, and 0.069. These variances were all greater than the second preset variance threshold of 0.02. Based on this, the detection device determined that the oscillator state had transitioned from a periodic state to a chaotic state and began accumulating the duration of the chaotic state. If the detection device detected the oscillator state transitioning from a periodic state to a chaotic state at the m-th cycle time, and at the corresponding m+1, m+2, m+3, and m+4-th cycle times... If the variance of the Poincaré cross section is greater than or equal to the second preset variance threshold, the detection device can determine that the duration of the chaotic state up to the (m+4)th cycle is 480ms. Since 480ms is greater than 360ms, the etching has reached its endpoint. Furthermore, even if the detection device completes the judgment at time t later than the (m+4)th cycle, the detection device will still determine the (m+4)th cycle as the etching endpoint arrival time. The detection device then sends an endpoint arrival signal to the etching device's main controller through the digital I / O interface, for example, by outputting a TTL low-level transition signal. For example, if the detection device obtains a Poincaré cross-sectional variance of 0.035 at a certain moment, briefly indicating that the oscillator is in a chaotic state, but the Poincaré cross-sectional variance decreases to 0.006 and 0.004 in the subsequent two gas switching cycles, the detection device can determine that the chaotic state at time t has not lasted for more than the preset duration threshold. Therefore, it does not determine this short-term change as the etching endpoint, but continues to monitor the subsequent oscillator state.

[0069] In some implementations, the second preset variance threshold is determined by the following steps: The variance of the Poincaré section of the sliding window in which at least one oscillator is in a periodic state is determined as at least one calibration variance; Determine the mean of at least one calibration variance; Multiply the mean by a second preset factor to obtain a second preset variance threshold, wherein the second preset factor is greater than 1.

[0070] In this embodiment, the detection device can calibrate a second preset variance threshold (which can be the same as the first preset variance threshold) used to distinguish between periodic and chaotic states before determining the etching endpoint. Since the second preset variance threshold is used to determine the oscillator state during the corresponding period of the sliding window, the second preset variance threshold can be determined based on the Poincaré cross-sectional variance when the oscillator state is in a periodic state, so that it can reflect the natural fluctuation range of the oscillator output during the normal etching stable period. Specifically, the detection equipment can collect optical signals of the characteristic spectral lines corresponding to the target product during the etching stabilization period, such as collecting SiF signals. The etching stabilization period can be understood as the period during which the etching reaction proceeds normally and the target product continues to be generated. During this period, there are still periodic modulation components related to the gas switching cycle in the characteristic spectral lines corresponding to the target product. Therefore, the oscillator state obtained by the detection equipment based on the signal during this period is in a periodic state. The detection device can generate an excitation signal based on the light intensity information of the characteristic spectral lines of SiF collected during the etching stabilization period, and input the excitation signal into the Duffing equation expressed by formula (1) to obtain the oscillator output of the Duffing equation at multiple periodic times; the detection device can use the oscillator output at multiple periodic times as the discrete points of the Poincaré section, and group the discrete points of the Poincaré section according to the preset sliding window length to obtain at least one sliding window in which the oscillator state is in a periodic state; Subsequently, the detection equipment can calculate the variance of discrete points of the Poincaré section within each sliding window and use these variances as calibration variances. Since these calibration variances originate from the etching stabilization period and the periodic modulation component exists during this period, these calibration variances can characterize the variance level of the oscillator detection module in the periodic state. After obtaining at least one calibration variance, the detection equipment can determine the mean of at least one calibration variance and multiply the mean by a preset multiple to obtain a second preset variance threshold, wherein the preset multiple is greater than 1, for example, the preset multiple can be 2; By amplifying the mean of the calibration variance under the periodic state by a certain factor as the second preset variance threshold, the detection device can leave a certain margin above the normal fluctuation range of the periodic state, thereby avoiding misjudging the oscillator state as chaotic due to slight noise jitter, sampling error or short-term disturbance of the cavity under the periodic state; in other words, the second preset variance threshold is not an arbitrarily set fixed empirical value, but is obtained by calibration of the actual signal during the etching stabilization period, and can be adapted to the current etching process, spectral acquisition conditions and output characteristics of the oscillator detection module; In this way, the detection equipment can adaptively determine the second preset variance threshold based on the Poincaré cross-sectional variance when the oscillator is in a periodic state during the etching stabilization period. This makes the distinction between periodic and chaotic states more consistent with the actual process signal characteristics. This scheme can reduce the risk of misjudgment caused by setting the threshold too low, and also avoid the endpoint response delay caused by setting the threshold too high, which is beneficial to improving the reliability of etching endpoint determination under low signal-to-noise ratio conditions.

[0071] For example, in a certain TSV deep silicon etching process, the gas switching cycle is 120ms. During the etching stabilization period, the detection device collects the light intensity information of the characteristic spectral lines of SiF and confirms that the light intensity information still contains periodic modulation components related to the gas switching cycle. Based on this light intensity information, the detection device drives a preset Duffing equation and samples the oscillator output at periods of 120ms, 240ms, 360ms, 480ms, 600ms, and 720ms to obtain multiple Poincaré cross-section discrete points. Assuming the detection device sets the sliding window length to 3 gas switching cycles and forms multiple points based on the Poincaré cross-section discrete points... The detection device calculates the variances of the Poincaré sections corresponding to these sliding windows as 0.00008, 0.00010, 0.00012, 0.00009, and 0.00011, respectively. The detection device can use these variances as calibration variances. The detection device further calculates the mean of these calibration variances, i.e., (0.00008+0.00010+0.00012+0.00009+0.00011) / 5=0.00010. If the preset multiple is 2, the detection device can determine 0.00010×2=0.00020 as the second preset variance threshold.

[0072] In some implementations, the detection device can also determine a second preset variance threshold through Monte Carlo simulation. Specifically, the detection device can introduce different noise levels, different SiF optical signal amplitudes, different background drift conditions, and different etching endpoint positions during the simulation process, calculate the false judgment rate under the candidate variance thresholds respectively, and select the variance threshold that minimizes the false judgment rate from multiple candidate variance thresholds as the second preset variance threshold. Thus, the detection device can obtain a second preset variance threshold that is more suitable for the current process conditions through simulation, based on actual signal calibration or in the absence of sufficient measured data during the stabilization period.

[0073] In some implementations, the excitation signal at time t is generated using the light intensity information at time t, including: The light intensity information at time t is processed to remove DC, thus obtaining the excitation signal at time t.

[0074] In this embodiment, when the detection device generates the excitation signal at time t using the light intensity information at time t, it can perform DC removal processing on the light intensity information at time t and use the light intensity information after DC removal processing as the excitation signal at time t. Specifically, taking SiF as the target product, the detection device can acquire the light intensity information of the SiF characteristic spectral line at time t and represent the light intensity information as I(t), which is the light intensity value of the SiF characteristic spectral line collected at time t. Due to factors such as window contamination, cavity temperature changes, gas flow fluctuations, RF power drift, or spectral acquisition link drift that may occur during the etching process, the light intensity information obtained by the detection device usually contains not only periodic change components related to the etching reaction, but also slowly changing background intensity components. These background intensity components can be understood as DC components or low-frequency background components superimposed on the effective light intensity changes. If the detection device directly inputs the unprocessed light intensity information as an excitation signal into the Duffing equation expressed by formula (1), the Duffing equation may be affected by the background intensity drift, causing deviations in the subsequent judgment of periodic and chaotic states. Therefore, the detection device can first perform DC removal processing on the light intensity information to weaken or remove the background intensity components in the light intensity information. Specifically, the detection device can determine the average value of light intensity information at multiple times, and subtract this average value from the light intensity information at time t to obtain the excitation signal at time t; this average value can be determined based on the light intensity information at multiple times within a preset time range, or it can be determined based on all the light intensity information obtained at time t. In this way, the detection equipment can convert the light intensity information containing the background intensity component into an excitation signal after DC removal, so that the signal input to the oscillator detection module can more concentratedly reflect the periodic changes of the characteristic spectral lines of the target product, and can also reduce the influence of the absolute value drift of light intensity on the oscillator state identification, thereby improving the stability and reliability of the determination of the low aperture ratio TSV etching endpoint.

[0075] Figure 3 This is a schematic diagram of the etching endpoint determination device provided in an embodiment of the present invention, as shown below. Figure 3 As shown, a second aspect of the present invention provides an etching endpoint determination device 10, the device 10 comprising: The acquisition module 11 is used to continuously acquire the light intensity information of the characteristic spectral line corresponding to the target product at time t while the etching process is in progress, wherein the target product is the product generated by etching in the plasma used for etching. The generation module 12 is used to generate the excitation signal at time t using the light intensity information at time t; The first determining module 13 is used to determine the fluctuation of the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, and the noise intensity in the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, wherein the end time point of the first time period is time t, and the length of the first time period is the first preset duration. The second determining module 14 is used to determine the critical coefficient corresponding to time t based on the fluctuation and noise intensity, wherein the critical coefficient is the adjustable coefficient in the calculation formula on which the external excitation calculation of the oscillator is based; The calculation module 15 is used to perform external excitation calculation on the oscillator based on the excitation signal at time t and the key coefficients corresponding to time t, to obtain the oscillator output at time t, and to update the oscillator state at time t based on the oscillator output at time t. The oscillator state at time t is used to characterize the state type of the oscillator output obtained up to time t. The state type is either periodic or chaotic. When the excitation signal at time t indicates that the target generated object has disappeared, the oscillator state at time t transitions from a periodic state to a chaotic state. The third determining module 16 is used to determine whether the etching endpoint has been reached based on the oscillator state at time t.

[0076] The etching endpoint determination device 10 provided in the second aspect of the present invention can realize the various processes implemented in the above method embodiments and achieve the same beneficial effects. To avoid repetition, it will not be described again here.

[0077] Please see Figure 4 This is a schematic diagram of the structure of the detection device provided in the embodiment of the present invention. The third aspect of the present invention provides a detection device, including a spectral signal acquisition unit and a calculation unit. The spectral signal acquisition unit is used to acquire the characteristic spectral lines corresponding to the target product. The target product is a product generated by etching in a plasma used for etching.

[0078] The computing unit may include a processor (e.g., a central processing unit, a graphics processing unit, etc.) 401, which can perform various appropriate actions and processes according to a program stored in read-only memory 402 or a program loaded from memory 408 into random access memory 403. The read-only memory may be ROM, and the random access memory may be RAM. The random access memory 403 also stores various programs and data required for the operation of the detection device. The processor 401, read-only memory 402, and random access memory 403 are interconnected via bus 404. An input / output interface 405 is also connected to bus 404.

[0079] Typically, the following devices can be connected to the input / output interface 405: input devices 406 including, for example, a touchscreen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 407 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; memory devices 408 including, for example, magnetic tape, hard disk, etc.; and communication devices 409. Communication device 409 allows the detection device to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 4 The detection apparatus is shown with various devices, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0080] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 409, or installed from a memory 408 (the memory being a magnetic tape, hard disk, etc.), or installed from a read-only memory 402. When the computer program is executed by the processor 401, it performs the functions defined in the etching endpoint determination method of the embodiments of the present invention.

[0081] Figure 4 The detection device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0082] A fourth aspect of the present invention provides a machine-readable storage medium storing instructions that, when executed by a processor, cause the processor to implement the above-described etching endpoint determination method.

[0083] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0084] The above are merely embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

[0085] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the present invention, they should also be regarded as the content disclosed by the present invention.

Claims

1. A method for determining the etching endpoint, characterized in that, The method includes: While the etching process is underway, the light intensity information of the characteristic spectral lines corresponding to the target product at time t is continuously acquired, wherein the target product is the product generated by etching in the plasma used for etching; The excitation signal at time t is generated using the light intensity information at time t; Determine the fluctuation of the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, and the noise intensity in the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, wherein the end time point of the first time period is time t, and the length of the first time period is a first preset duration. Based on the fluctuation and the noise intensity, the key coefficient corresponding to time t is determined, wherein the key coefficient is an adjustable coefficient in the calculation formula on which the external excitation operation of the oscillator is based; The oscillator is externally excited based on the excitation signal at time t and the key coefficients corresponding to time t to obtain the oscillator output at time t. The oscillator state at time t is then updated based on the oscillator output at time t. The oscillator state at time t is used to characterize the state type of the oscillator output obtained up to time t. The state type is either a periodic state or a chaotic state. When the excitation signal at time t indicates that the target generator has disappeared, the oscillator state at time t transitions from the periodic state to the chaotic state. Based on the state of the oscillator at time t, determine whether the etching endpoint has been reached.

2. The method according to claim 1, characterized in that, The calculation formula on which the external excitation calculation of the oscillator is based is: in, For time markers, For the The oscillator output at any given moment, for right The second derivative, for right The first derivative, for cubed, for The fifth power, For the Momentary stimulus signal Driven by cycles, The frequency of the cycle drive is the angular frequency corresponding to the gas switching frequency during the etching process, and the period of the cycle drive is... The gas mentioned is a gas used for etching. The amplitude of the periodic drive, The damping coefficient is adjustable. The periodic state is the state in which the oscillator output changes periodically with the periodic drive. In the periodic state, the oscillator output changes repeatedly within multiple consecutive periodic drive cycles. The chaotic state is the state in which the oscillator output does not change periodically with the periodic drive. In the chaotic state, the oscillator output changes non-periodically within multiple consecutive periodic drive cycles.

3. The method according to claim 2, characterized in that, The amplitude of the periodic drive is the product of the critical amplitude and the first preset multiple, where the first preset multiple is less than 1 and greater than 0, and the absolute value of the difference between the critical amplitude and the amplitude of the periodic drive is less than a preset difference threshold. The critical amplitude is determined through the following steps: Set the excitation signal to 0; Multiple preset amplitude values ​​are determined, wherein the numerical difference between any two adjacent preset amplitude values ​​is a preset step size; Perform a calibration pre-step, wherein the calibration pre-step includes determining the oscillator output at multiple times corresponding to each preset amplitude in the calculation formula in ascending order of numerical values, and using the oscillator output at the multiple times corresponding to the Poincaré cross section discrete points, and determining the Poincaré cross section variance corresponding to the preset amplitude based on the Poincaré cross section discrete points; Based on the Poincaré cross-sectional variances corresponding to multiple preset amplitudes, a critical amplitude is determined among the multiple preset amplitudes, wherein the critical amplitude is the first preset amplitude that makes the corresponding Poincaré cross-sectional variance less than a first preset variance threshold during the execution of the calibration pre-step.

4. The method according to claim 2, wherein the critical coefficient is the damping coefficient in the calculation formula, and determining the critical coefficient corresponding to time t based on the fluctuation situation and the noise intensity includes: A fluctuation characterization value is determined based on the fluctuation situation, wherein the fluctuation characterization value is used to characterize the degree of fluctuation of the light intensity information within the first time period; A noise characterization value is determined based on the noise intensity, wherein the noise characterization value is used to characterize the degree of noise interference in the light intensity information within the first time period; The fluctuation characterization value and the noise characterization value are normalized to obtain normalized fluctuation characterization value and normalized noise characterization value. Based on the preset benchmark damping coefficient, the normalized fluctuation characterization value, and the normalized noise characterization value, the damping coefficient corresponding to time t is determined, wherein the damping coefficient corresponding to time t is positively correlated with the normalized noise characterization value and negatively correlated with the normalized fluctuation characterization value.

5. The method according to claim 1, characterized in that, The step of determining whether the etching endpoint has been reached based on the oscillator state at time t includes: Extract the oscillator outputs at multiple periodic times from the oscillator state at time t as discrete points of the Poincaré section, wherein the periodic times are positive integer multiples of the gas switching period during the etching process, and the gas is the gas used for etching. Determine the length of the sliding window, and group the discrete points of the Poincaré section according to the length of the sliding window to obtain multiple groups of sliding windows; For each set of sliding windows, the variance of the discrete points of the Poincaré cross section within the sliding window is determined as the variance of the Poincaré cross section of the sliding window. Wherein, if the variance of the Poincaré cross section is less than a second preset variance threshold, it indicates that the oscillator state of the corresponding period of the sliding window is in the periodic state. If the variance of the Poincaré cross section is greater than or equal to the second preset variance threshold, it indicates that the oscillator state of the corresponding period of the sliding window is in the chaotic state. The etching endpoint is determined based on the variance of the multiple Poincaré sections.

6. The method according to claim 5, characterized in that, The step of determining whether the etching endpoint has been reached based on the variances of multiple Poincaré sections includes: If the duration of the oscillator state in a chaotic state at time t, as indicated by the variance of multiple Poincaré sections, is greater than a preset duration threshold, it is determined that the etching endpoint has been reached. The first cycle time in which the duration of the oscillator state in a chaotic state is greater than the preset duration threshold is determined as the etching endpoint arrival time. The preset duration threshold is N gas switching cycles, where N is a positive integer greater than 1.

7. The method according to claim 5, characterized in that, The second preset variance threshold is determined through the following steps: The variance of the Poincaré section of at least one oscillator state in the periodic state is determined as at least one calibration variance; Determine the mean of the at least one calibration variance; The mean is multiplied by a second preset factor to obtain the second preset variance threshold, wherein the second preset factor is greater than 1.

8. The method according to claim 1, characterized in that, The step of generating the excitation signal at time t using the light intensity information at time t includes: The light intensity information at time t is subjected to DC removal processing to obtain the excitation signal at time t.

9. An etching endpoint determination device, characterized in that, The device includes: The acquisition module is used to continuously acquire the light intensity information of the characteristic spectral line corresponding to the target product at time t while the etching process is in progress, wherein the target product is the product generated by etching in the plasma used for etching; The generation module is used to generate the excitation signal at time t using the light intensity information at time t; The first determining module is used to determine the fluctuation of the light intensity information of the characteristic spectral line corresponding to the target generated object within a first time period, and the noise intensity in the light intensity information of the characteristic spectral line corresponding to the target generated object within the first time period, wherein the end time point of the first time period is time t, and the length of the first time period is a first preset duration. The second determining module is used to determine the key coefficient corresponding to time t based on the fluctuation situation and the noise intensity, wherein the key coefficient is an adjustable coefficient in the calculation formula on which the external excitation operation of the oscillator is based; The calculation module is used to perform external excitation calculation on the oscillator based on the excitation signal at time t and the key coefficients corresponding to time t, to obtain the oscillator output at time t, and to update the oscillator state at time t according to the oscillator output at time t. The oscillator state at time t is used to characterize the state type of the oscillator output obtained up to time t. The state type is either a periodic state or a chaotic state. When the excitation signal at time t indicates that the target generator has disappeared, the oscillator state at time t transitions from the periodic state to the chaotic state. The third determining module is used to determine whether the etching endpoint has been reached based on the oscillator state at time t.

10. A testing device, characterized in that, It includes a spectral signal acquisition unit and a computing unit. The spectral signal acquisition unit is used to acquire the characteristic spectral lines corresponding to the target product, which is a product generated by etching in a plasma used for etching. The computing unit includes a processor and a memory, the memory storing programs or instructions that can run on the processor, and the programs or instructions, when executed by the processor, implement the etching endpoint determination method as described in any one of claims 1-8.

11. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores instructions that, when executed by a processor, cause the processor to implement the etching endpoint determination method as described in any one of claims 1-8.

Images