Anti-interference detection method of wafer temperature measurement system

Abstract

The invention discloses an anti-interference detection method of a wafer temperature measurement system, and belongs to the technical field of semiconductor rapid heat treatment process, the anti-interference detection method comprises the following steps: a temperature sensor detects the temperature of a wafer, an original sampling value detected by the temperature sensor is firstly subjected to noise interference filtering processing through a temperature detector, and then the original sampling value is subjected to noise interference filtering processing; sending the processed data as the detection temperature of the wafer to a temperature controller, comparing the detection temperature with the formula set temperature in the temperature controller, and then adjusting the output power to form closed-loop control; the noise interference filtering processing comprises the steps that an original sampling value obtained by the temperature sensor is subjected to data verification, the verified data is subjected to extreme value removal filtering, then real-time estimation is carried out by adopting a noise estimation algorithm, steady-state noise is identified, and the steady-state noise is removed to obtain the actual temperature of the wafer. According to the method, extreme value removal filtering is adopted in the aspect of random disturbance, a noise estimation mode is adopted in the aspect of stable sine disturbance, and the temperature measurement accuracy is remarkably improved.

Description

Technical Field

[0001] The invention belongs to the technical field of semiconductor rapid thermal processing technology, and in particular relates to an anti-interference detection method for a wafer temperature measurement system. Background Art

[0002] Rapid Thermal Processing (RTP) is an advanced monolithic heat treatment process and a key technology in semiconductor manufacturing. It can change the microstructure and performance of materials. During the rapid thermal processing, the temperature of the wafer is quickly raised to the temperature required for the process and can be quickly cooled. Accurate temperature control is extremely important for the rapid thermal processing process, and accurate measurement of wafer temperature has a decisive influence on the temperature control effect and process yield.

[0003] Rapid thermal processing usually uses non-contact temperature measurement technology. In view of the complex system structure of the RTP equipment, there will be electromagnetic interference such as static electricity and radio frequency in the surrounding environment. At the same time, there will be rotation of magnetic levitation and other mechanical disturbances in the system. These interferences will have a serious impact on the temperature measurement system, resulting in inaccurate temperature measurement, and then the actual temperature control will not meet expectations. Therefore, these interferences must be shielded and filtered so that the temperature measurement system can achieve strong anti-interference capabilities. At the same time, since the rapid thermal processing process requires the temperature to rise and fall rapidly, the real-time temperature anti-interference detection is particularly necessary. The existing solutions do not have anti-interference designs or the effects are not good. There is obvious noise interference in the temperature data, which is difficult to filter out and affects the control effect. Therefore, a solution that can effectively resist interference is needed. Summary of the invention

[0004] Based on the technical problems existing in the prior art, the present invention provides an anti-interference detection method for a wafer temperature measurement system to solve the problems that the temperature data measured in the existing semiconductor equipment has obvious noise and affects the control effect, and can effectively filter out interference and obtain accurate temperature data, thereby improving the accuracy of temperature control in the semiconductor process.

[0005] According to the technical solution of the present invention, the present invention provides an anti-interference detection method for a wafer temperature measurement system, wherein the wafer temperature measurement system adopted by the present invention comprises a temperature sensor for detecting the wafer temperature, the temperature sensor is sequentially connected with a temperature detector, a temperature controller and a power controller, the temperature controller is connected with a host computer, and the power controller is connected with a wafer heating device of a semiconductor device; The anti-interference detection method of the wafer temperature measurement system comprises the following steps: The temperature controller receives the recipe parameters set by the host computer and adjusts the output power of the power controller to control the heating power of the wafer heating device; the temperature sensor detects the temperature of the wafer, and the original sampling value obtained by the temperature sensor is first subjected to noise interference filtering through the temperature detector to obtain the wafer detection temperature, and then the wafer detection temperature is sent to the temperature controller, and the temperature controller compares the wafer detection temperature with the recipe setting temperature, and then adjusts the output power to form a closed-loop control; Among them, the noise interference filtering process includes that the original sampling value obtained by the temperature sensor is first subjected to data verification to prevent data errors, the verified data is filtered out by de-extreme value filtering to remove possible random interference, and then the noise estimation algorithm is used to perform real-time estimation and identify steady-state noise. After removing the steady-state noise through real-time filtering, the wafer detection temperature is obtained.

[0006] Furthermore, the de-extreme value filtering is to use a cache array to simulate a queue to store the temperature measurement data obtained by the temperature sensor, the queue length is N, and the calculation starts only after the queue is filled. The latter data always replaces the earliest data in the queue to maintain the real-time nature of the data; the values ​​in the queue are arranged in order of size, and the middle value is taken as the benchmark T mid , according to the current formula heating rate or cooling rate to determine the theoretical maximum difference between two adjacent temperature values ​​err r , define T mid ±(a×err r ) is the de-extreme value judgment condition at the current moment. If the actual sampling value at the current moment exceeds the range of the de-extreme value judgment condition at the current moment, the temperature measurement data value at the current moment is refreshed according to the de-extreme value judgment condition at the current moment.

[0007] Furthermore, the noise estimation algorithm includes the following contents: Define the residual Loss: , Among them, T' i is the measured value after de-extreme filtering, TN i is the sinusoidal noise value, T i is the current prediction value of the system, M is the number of iterations; , in, τ is the time constant of the system, K is the transfer coefficient, T i-1 is the predicted value of the system at the previous sampling moment; TN i =A i sin(2πfT s +φ i ), Where f is the system noise frequency, which is determined by the magnetic suspension rotation frequency; T s is the sampling time; A i is the amplitude of the sinusoidal noise at the current sampling moment; φ i is the sinusoidal noise phase angle at the current sampling moment; Loss to A i and φ i Calculate the partial derivatives respectively and get the following iterative formula: A i = A i-1 – b×sin(2πfT s +φ i-1 ), φ i = φ i-1 – b×cos(2πfT s +φ i-1 ) ; Among them, A i is the amplitude of the sinusoidal noise at the current sampling moment; A i-1 is the amplitude of the sinusoidal noise at the previous sampling moment; φ i is the sinusoidal noise phase angle at the current sampling moment; φ i-1 is the sinusoidal noise phase angle at the previous sampling moment; b is the sinusoidal noise estimation iteration update rate; The optimization goal is to minimize Loss, so the sinusoidal noise value TN at the current sampling time is obtained i .

[0008] Furthermore, the noise estimation algorithm further includes: When secondary filtering is observed or the amplitude of the second harmonic is detected to be higher than the set threshold, the second harmonic filtering is turned on; the second harmonic filtering specifically includes the following contents: Define the residual Loss' of the second harmonic filtering process: , Among them, T'' i TNS is the measured value after removing the sinusoidal noise. i is the second harmonic noise value, T i is the current prediction value of the system, M is the number of iterations; TNS i =A' i sin(2πf'T s +φ' i ), Where, f' is the second harmonic noise frequency, f'=2f, f is the system noise frequency; T s is the sampling time; A' i is the second harmonic noise amplitude at the current sampling moment; φ' iis the second harmonic noise phase angle at the current sampling moment; Loss' to A' i and φ' i Calculate the partial derivatives respectively and get the following iterative formula: A' i = A' i-1 – b'×sin(2πf'T s +φ' i-1 ), φ' i = φ' i-1 – b'×cos(2πf'T s +φ' i-1 ) ; Among them, A' i-1 is the second harmonic noise amplitude at the previous sampling moment; φ' i-1 is the second harmonic noise phase angle at the previous sampling moment; b' is the second harmonic estimation iteration update rate; The optimization goal is to minimize Loss', so the second harmonic noise value TNS at the current sampling time is obtained i .

[0009] Furthermore, a sensor control box is connected between the temperature sensor and the temperature detector; the temperature sensor and the sensor control box are connected via an optical fiber cable, and the sensor control box and the temperature detector are connected via a shielded twisted pair cable, and the shielding layer is grounded.

[0010] Furthermore, an interference suppressor is provided at a communication interface on the temperature detector for connecting with the sensor control box.

[0011] Furthermore, the temperature detector has an isolated regulated power supply with a filter circuit inside.

[0012] Furthermore, the temperature detector is arranged in a sealed shielding shell, and the temperature detector is well grounded through the shielding shell.

[0013] Compared with the prior art, the beneficial technical effects of the present invention are as follows: 1. In the anti-interference detection method of the wafer temperature measurement system of the present invention, the main disturbances of the semiconductor equipment system on the temperature measurement results are divided into random disturbances and steady-state noise through analysis and research. De-extreme value filtering is used for random disturbances, and noise estimation is used for steady-state noise (stable sinusoidal disturbances). In this way, the original sampling values ​​of the temperature sensor are filtered out for noise interference, thereby improving the accuracy of wafer temperature measurement.

[0014] 2. In the anti-interference detection method of the wafer temperature measurement system of the present invention, a closed-loop controlled wafer temperature measurement system is adopted, which can automatically analyze and adjust the heating power according to the situation in real time, and then adjust the wafer heating / cooling rate, so that the change of wafer temperature better meets expectations and finally more accurately reaches the recipe setting temperature set by the host computer.

[0015] 3. In the anti-interference detection method of the wafer temperature measurement system of the present invention, a combination of hardware anti-interference and software anti-interference is also adopted. The software anti-interference is to filter out noise interference from the original sampling value, and the hardware anti-interference is achieved by optimizing and improving the hardware of each device and the overall wafer temperature measurement system, thereby effectively suppressing the interference of electromagnetic fields, pulse noise signals and other aspects on the sampling value and data transmission process.

[0016] 4. The anti-interference detection method of the wafer temperature measurement system of the present invention significantly reduces the impact of disturbances on temperature data through multi-angle research and improvement, and has better real-time performance, thereby improving the temperature control accuracy and improving the semiconductor process yield and process quality with rapid thermal treatment process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a structural schematic diagram of the wafer temperature measurement system provided by the present invention.

[0018] Figure 2 It is a structural block diagram of a simplified wafer temperature measurement system provided by the present invention.

[0019] Figure 3 It is a flow chart of the method provided by the invention.

[0020] Description of reference numerals in the accompanying drawings: 1. Temperature sensor; 2. Sensor control box; 3. Interference suppressor; 4. Temperature detector; 5. Temperature controller; 6. Host computer; 7. Power controller; 8. Wafer heating device; 9. Wafer. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other.

[0023] It should be noted that the concepts such as "first" and "second" mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0024] It should be noted that the modifications of "one" and "plurality" mentioned in the present invention are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0025] During the rapid heat treatment process in semiconductor manufacturing, the precise measurement of wafer temperature directly affects the temperature control effect, which in turn has a decisive impact on the process yield. The existing solutions do not have anti-interference design or the effect is not good. There is obvious noise and interference in the temperature data, which is difficult to filter out, affecting the temperature measurement and control effect. The study found that due to the complexity of the process equipment, interference will be caused to the temperature measurement system, that is, the original sampling value = actual wafer temperature value + random interference value + noise value, where further, the noise value = sinusoidal noise value + high-order harmonic noise value; usually when the system is stable, there is no or negligible high-order harmonic noise value, and when it is unstable, the high-order harmonic noise value will affect the temperature detection. Among the high-order harmonic noise values, the most influential one is the second harmonic noise value.

[0026] See also Figures 1 to 3 In the anti-interference detection method of the wafer temperature measurement system provided by the invention, the wafer temperature measurement system used includes a temperature sensor 1 for detecting the temperature of a wafer 9 in a semiconductor device. The temperature sensor 1 is connected to a temperature detector 4, a temperature controller 5 and a power controller 7 in sequence. The temperature controller 5 is connected to a host computer 6, and the power controller 7 is connected to a wafer heating device 8 of the semiconductor device. There are generally multiple temperature sensors 1, which monitor the temperature of different positions of the wafer respectively, in order to ensure that the wafer is heated evenly. For example, there are L temperature sensors in total.

[0027] When working, the temperature controller receives the recipe parameters set by the host computer, and automatically adjusts the output power of the power controller through the control algorithm to control the heating power of the wafer heating device (such as the brightness of the lamp). The lamp irradiates the wafer to make the wafer heat up at the designed rate (or, lower the heating power / turn off the heating device to cool the wafer at the designed rate). The temperature sensor detects the temperature of the wafer. The original sampling value detected by the temperature sensor is first processed by the temperature detector for noise interference filtering to obtain the wafer detection temperature, and then the wafer detection temperature is sent to the temperature controller. The temperature controller compares the wafer detection temperature with the recipe setting temperature, and then adjusts the output power adaptively to form a closed-loop control, so that the wafer temperature changes in a set manner, so that the change of the wafer temperature better meets the expectation and finally reaches the recipe setting temperature set by the host computer (that is, the target temperature value set in the recipe). The anti-interference detection method in the wafer temperature measurement system of the present invention preferably adopts a combination of hardware anti-interference and software anti-interference; of course, for example, in other embodiments, only part of the structure / method can be used.

[0028] In terms of hardware anti-interference, the wafer temperature measurement system of the present invention includes several temperature sensors (probes), a sensor control box, a transmission cable, and a temperature detector. Compared with a simpler solution, a sensor control box 2 is also connected between the temperature sensor 1 and the temperature detector 4 in the present invention. The temperature sensor 1 (probe) and the sensor control box 2 are connected by an optical fiber cable, which has the ability to resist electromagnetic interference. The sensor control box 2 outputs temperature measurement data, and the sensor control box 2 and the temperature detector 4 are connected by a shielded twisted pair to transmit data. The shielding layer is grounded and has good anti-interference ability; the shielded twisted pair is used as a transmission cable, more specifically, a shielded differential twisted pair. The communication interface on the temperature detector 4 used to connect to the sensor control box 2 is provided with an interference suppressor 3 to suppress pulse noise signals and electrostatic interference that may be coupled to the interface. The temperature detector 4 has an isolated regulated power supply with a filter circuit inside, which effectively filters out the noise of the switching power supply input signal and shields the line conduction interference. The temperature detector 4 is placed in a sealed shielding shell (of course, the shielding shell can allow the corresponding cables to pass through the temperature detector interface), and the temperature detector 4 is well grounded through the shielding shell. For example, the signal ground of the temperature detector 4 is connected to the shielding shell (that is, the ground wire of the temperature detector is connected to the shielding shell). The shielding shell is well grounded, which can effectively suppress the interference of space radiation electromagnetic field and improve the anti-interference ability.

[0029] Digital filters are used for software anti-interference. There are random interference and steady-state noise in the temperature measurement data. The design of digital filters needs to filter out these two parts at the same time. Since the temperature sensors are installed in different positions and are subject to different degrees of external interference, each temperature sensor requires an independent digital filter. The data of the temperature sensor is first verified (for example, CRC verification, accumulation and verification, etc. are all possible, not limited to a certain type) to prevent data errors caused by interference. The verified data is first filtered to remove possible random interference. The steady-state noise mainly comes from the periodic steady-state noise caused by the rotation of the magnetic levitation. After analysis, the steady-state noise is a sinusoidal noise with the same frequency as the rotation of the magnetic levitation and its higher harmonics. The noise estimation algorithm based on the gradient descent method is used to estimate and identify the steady-state noise in real time, and then the real-time filtering is performed, that is, the identified steady-state noise is further removed from the current data value, and finally the wafer detection temperature is obtained.

[0030] More specifically, the de-extreme value filtering is to use a cache array to simulate a queue to store the temperature measurement data obtained by the temperature sensor. The queue length is N, where N is a positive integer and is selected according to the actual situation. The operation starts only after the queue is filled. The later data always replaces the earliest data in the queue to maintain the real-time nature of the data. The values ​​in the queue (array) are arranged in order of size (for example, from small to large), and the middle value is taken as the benchmark T mid The theoretical maximum difference between two adjacent temperature values ​​err is determined based on the heating rate or cooling rate of the current formula. r , take a times of the theoretical maximum difference as the amplitude limit (theoretically a=(N+1)÷2, in actual use, a certain amplification margin is required to prevent false triggering, that is, in actual use, a=(N+1)÷2×c can be taken, c is an empirical parameter determined according to actual tests, c≥1), that is, define T mid ±(a×err r ) is the de-extreme value judgment condition at the current moment. If the actual sampling value at the current moment exceeds this range, the temperature measurement data value at the current moment is refreshed according to the de-extreme value judgment condition at the current moment (the actual sampling value at the current moment is replaced by the theoretical extreme value at the current moment).

[0031] For example, in the heating process, if a newly acquired actual sampling value is greater than the T of the previous N data mid +(a×err r ), then the newly acquired actual sampling value is considered to be wrong and replaced with the T of the previous N data. mid +(a×err r); if the new actual sampling value is less than or equal to the de-extreme value judgment condition value of the previous N data, it remains unchanged; before or after a new actual sampling value enters the queue (in short, before the following de-extreme value judgment condition is calculated), delete the earliest acquired data in the queue (or replace the earliest acquired data with the new data), so that when the de-extreme value judgment condition is calculated, the cache array simulation queue always has a total of N data of the previous N sampling moments; then calculate the T of the current N data mid +(a×err r ), as the next actual sampling value de-extreme value judgment condition, and so on. The cooling process is similar, and the de-extreme value judgment condition is T mid -(a×err r ). In this way, the random interference value T is removed by de-extreme filtering. random As a supplementary explanation, the N temperature measurement data at the initial stage of heating / cooling are the original sampling values, and then the filtering operation process actually begins. The subsequent temperature measurement data are the sampling values ​​that are judged to meet the conditions or refreshed and replaced in the above manner; the "size" in "arranging the values ​​in the queue in order of size" refers to the corresponding temperature values ​​in the temperature measurement data; the temperature measurement data also corresponds to the sampling time or sequence data of the data, and the "adjacent" in "determining the theoretical maximum difference between two adjacent temperature values" refers to the sampling times being adjacent in order.

[0032] In addition, the temperature control in rapid thermal processing includes the heating process, the cooling process, and the stable or soak stage after the temperature reaches the set value. The ideal stable stage is that the temperature remains unchanged without heating or cooling. The de-extreme value judgment condition of this stable stage can adopt a fixed value obtained based on experience, or still adopt the above calculation method, in which the theoretical maximum difference err r That is, corresponding to the temperature fluctuation, the value can be determined according to experience or process requirements, and it is necessary to ensure that the de-extreme value judgment condition is within the normal range determined according to experience.

[0033] The noise estimation algorithm includes the following contents.

[0034] The present invention adopts the gradient descent method to perform noise estimation, defines the residual Loss (residual in mathematical statistics refers to the difference between the actual observed value and the estimated value (fitting value)), and uses the residual to evaluate the prediction model. The smaller the total residual, the more accurate the prediction.

[0035] , The optimization goal is to minimize the error between the measured value after removing the sinusoidal noise and the system predicted value. Among them, T' i is the measured value after de-extreme filtering, TNi is the sinusoidal noise value (the estimated value of the sinusoidal noise at the current sampling moment, or the estimated value of the steady-state noise at the current sampling moment), T i is the current predicted value of the system (the wafer temperature value at the current sampling moment estimated and determined by the equipment system, in other words, the predicted value of the wafer temperature), and M is the number of iterations; , in, τ is the time constant of the system, which is a fixed constant and is related to the control cycle and the delay inertia of the entire system; K is the transfer coefficient, which is a constant and is determined by the system characteristics; T i-1 is the predicted value of the system at the previous sampling moment; TN i =A i sin(2πfT s +φ i ), Where f is the system noise frequency, which is determined by the magnetic suspension rotation frequency; T s is the sampling time; A i is the amplitude of the sinusoidal noise at the current sampling moment, φ i is the sinusoidal noise phase angle at the current sampling moment.

[0036] Loss to A i and φ i Calculate the partial derivatives respectively and get the following iterative formula: A i = A i-1 – b×sin(2πfT s +φ i-1 ), φ i = φ i-1 – b×cos(2πfT s +φ i-1 ) ; Among them, A i is the amplitude of the sinusoidal noise at the current sampling moment; A i-1 is the amplitude of the sinusoidal noise at the previous sampling moment; φ i is the sinusoidal noise phase angle at the current sampling moment; φ i-1 is the sinusoidal noise phase angle at the previous sampling moment; b is the sinusoidal noise estimation iteration update rate, which is a constant.

[0037] The optimization goal is to minimize the Loss, that is, the obtained A i and φ i Substitute the above formula to get the expression of Loss, and the optimization goal is to minimize it. In this way, the sinusoidal noise value TN at the current sampling time can be obtained i, which can be removed by real-time filtering. i It changes continuously in real time. By iterating the above formula, it will gradually approach the real TN. i For each point, the current TN can be estimated based on the value at the previous moment. i , each calculation is a noise estimation.

[0038] When the system is stable, there is usually only a stable sinusoidal noise with the same speed as the magnetic suspension, that is, the aforementioned sinusoidal noise value TN i , and its frequency is f. However, when the interference intensity is large and the system is unstable, the amplitude of this noise will be large, resulting in the generation of higher-order harmonics. Usually, the amplitude (intensity) of the second harmonic will be relatively high, and the other higher-order harmonics will have little impact. Therefore, the second harmonic can be filtered out, and the frequency of the second harmonic is 2f. Further using the same form as above (the difference is that f in the formula is replaced by f', f'=2f) for noise estimation, the second harmonic noise value TNS can be obtained. i .

[0039] More specifically, the processing method may be to set a high-order harmonic filter switch, and when a second harmonic is observed in the temperature detection data, the second harmonic filter may be turned on; or, when the amplitude of the second harmonic is detected to be higher than a certain set threshold, the second harmonic filter may be turned on. The second harmonic filter specifically includes the following contents.

[0040] The optimization goal is to minimize the error between the measured value after further removing the second harmonic noise and the system predicted value.

[0041] , Among them, Loss' is the residual of the second harmonic filtering process, in other words, it is the error accumulation sum of the measured value after further removing the second harmonic noise and the system prediction value, T'' is the measured value after removing the sinusoidal noise, TNS i is the second harmonic noise value, T i is the current predicted value of the system; M is the number of iterations; TNS i =A' i sin(2πf'T s +φ' i ), Where, f' is the second harmonic noise frequency, f'=2f, f is the system noise frequency; T s is the sampling time; Loss' to A' i and φ' i Calculate the partial derivatives respectively and get the following iterative formula: A' i = A' i-1– b'×sin(2πf'T s +φ' i-1 ), φ' i = φ' i-1 – b'×cos(2πf'T s +φ' i-1 ) ; Among them, A' i is the second harmonic noise amplitude at the current sampling moment; A' i-1 is the second harmonic noise amplitude at the previous sampling moment; φ' i is the second harmonic noise phase angle at the current sampling moment; φ' i-1 is the second harmonic noise phase angle at the previous sampling moment; b' is the second harmonic estimation iteration update rate, which is a constant.

[0042] The optimization goal is to minimize Loss', so the second harmonic noise value TNS at the current sampling time is obtained i Then, the second harmonic noise value TNS can be reduced by real-time filtering. i Remove.

[0043] As a supplementary explanation for easier understanding, the temperature detection in the present invention undergoes the following changes: the temperature sensor measures the original acquisition value T measure , the original acquisition value T measure After the extreme value filtering, the random interference value T is removed. random , get the measured value T' after de-extreme filtering, and then remove the noise value (remove the sinusoidal noise value TN i , get T'', and according to the situation, further remove the second harmonic noise value TNS i ), after such processing, the wafer detection temperature is obtained. The wafer detection temperature is the wafer temperature measured by the anti-disturbance detection method of the present invention, which is closer to the actual temperature of the wafer than the prior art.

[0044] It should be noted that in a semiconductor device, multiple temperature sensors are usually installed to measure the temperatures of multiple different positions of the wafer respectively. The specific arrangement and comprehensive analysis of the data from multiple temperature sensors can follow the existing technical solutions and are not the focus of improvement of the present invention, so they will not be elaborated here.

[0045] To sum up, the present invention adopts de-extreme value filtering for random disturbances and adopts noise estimation method for stable sinusoidal disturbances. After the original sampling values ​​of the temperature sensor are subjected to noise interference filtering processing, the influence of the disturbance on the temperature data is significantly reduced, the accuracy of temperature measurement is improved, and then the temperature control accuracy is improved, thereby improving the quality of semiconductor process.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An anti-interference detection method for a wafer temperature measurement system, characterized in that: The wafer temperature measurement system used therein includes a temperature sensor for detecting the wafer temperature, the temperature sensor is sequentially connected to a temperature detector, a temperature controller and a power controller, the temperature controller is connected to a host computer, and the power controller is connected to a wafer heating device of a semiconductor device; The anti-interference detection method of the wafer temperature measurement system comprises the following steps: The temperature controller receives the recipe parameters set by the host computer and adjusts the output power of the power controller to control the heating power of the wafer heating device; the temperature sensor detects the temperature of the wafer, and the original sampling value obtained by the temperature sensor is first subjected to noise interference filtering through the temperature detector to obtain the wafer detection temperature, and then the wafer detection temperature is sent to the temperature controller, and the temperature controller compares the wafer detection temperature with the recipe setting temperature, and then adjusts the output power to form a closed-loop control; Among them, the noise interference filtering process includes: the original sampling value obtained by the temperature sensor is first subjected to data verification to prevent data errors, the verified data is filtered out by de-extreme value filtering to remove possible random interference, and then the noise estimation algorithm is used to perform real-time estimation and identify steady-state noise. After removing the steady-state noise through real-time filtering, the wafer detection temperature is obtained.

2. The anti-interference detection method of the wafer temperature measurement system according to claim 1, characterized in that: The de-extreme value filtering is to use a cache array to simulate a queue to store the temperature measurement data obtained by the temperature sensor. The queue length is N. The operation starts only after the queue is filled. The latter data always replaces the earliest data in the queue to maintain the real-time nature of the data. The values ​​in the queue are arranged in order of size, and the middle value is taken as the benchmark T mid , according to the current formula heating rate or cooling rate to determine the theoretical maximum difference between two adjacent temperature values ​​err r , define T mid ±(a×err r ) is the de-extreme value judgment condition at the current moment. If the actual sampling value at the current moment exceeds the range of the de-extreme value judgment condition at the current moment, the temperature measurement data value at the current moment is refreshed according to the de-extreme value judgment condition at the current moment.

3. The anti-interference detection method of the wafer temperature measurement system according to claim 1, characterized in that: The noise estimation algorithm includes the following: Define the residual Loss: ， Among them, T' i is the measured value after de-extreme filtering, TN i is the sinusoidal noise value, T i is the current prediction value of the system, M is the number of iterations; ， in, τ is the time constant of the system, K is the transfer coefficient, T i-1 is the predicted value of the system at the previous sampling moment; TN i =A i sin(2πfT s +φ i ), Where f is the system noise frequency, which is determined by the magnetic suspension rotation frequency; T s is the sampling time; A i is the amplitude of the sinusoidal noise at the current sampling moment; φ i is the sinusoidal noise phase angle at the current sampling moment; Loss to A i and φ i Calculate the partial derivatives respectively and get the following iterative formula: A i = A i-1 – b×sin(2πfT s +φ i-1 ), f i = φ i-1 – b×cos(2πfT s +φ i-1 ); Among them, A i is the amplitude of the sinusoidal noise at the current sampling moment; A i-1 is the amplitude of the sinusoidal noise at the previous sampling moment; φ i is the sinusoidal noise phase angle at the current sampling moment; φ i-1 is the sinusoidal noise phase angle at the previous sampling moment; b is the sinusoidal noise estimation iteration update rate; The optimization goal is to minimize Loss, so the sinusoidal noise value TN at the current sampling time is obtained i .

4. The anti-interference detection method of the wafer temperature measurement system according to claim 3, characterized in that: The noise estimation algorithm also includes: When secondary filtering is observed or the amplitude of the second harmonic is detected to be higher than the set threshold, the second harmonic filtering is turned on; the second harmonic filtering specifically includes the following contents: Define the residual Loss' of the second harmonic filtering process: ， Among them, T'' i TNS is the measured value after removing the sinusoidal noise. i is the second harmonic noise value, T i is the current prediction value of the system, M is the number of iterations; TNS i =A' i sin(2πf'T s +φ' i ), Where, f' is the second harmonic noise frequency, f'=2f, f is the system noise frequency; T s is the sampling time; A' i is the second harmonic noise amplitude at the current sampling moment; φ' i is the second harmonic noise phase angle at the current sampling moment; Loss' to A' i and φ' i Calculate the partial derivatives respectively and get the following iterative formula: A' i = A' i-1 – b'×sin(2πf'T s +φ' i-1 ), f' i = φ' i-1 – b'×cos(2πf'T s +φ' i-1 ); Among them, A' i-1 is the second harmonic noise amplitude at the previous sampling moment; φ' i-1 is the second harmonic noise phase angle at the previous sampling moment; b' is the second harmonic estimation iteration update rate; The optimization goal is to minimize Loss', so the second harmonic noise value TNS at the current sampling time is obtained i .

5. The anti-interference detection method for a wafer temperature measurement system according to any one of claims 1 to 4, characterized in that: A sensor control box is also connected between the temperature sensor and the temperature detector; the temperature sensor and the sensor control box are connected via an optical fiber cable, and the sensor control box and the temperature detector are connected via a shielded twisted pair cable, and the shielding layer is grounded.

6. The anti-interference detection method of the wafer temperature measurement system according to claim 5, characterized in that: The temperature detector has an interference suppressor at the communication interface for connecting with the sensor control box.

7. The anti-interference detection method of the wafer temperature measurement system according to claim 5, characterized in that: The temperature detector has an isolated regulated power supply with a filtering circuit inside.

8. The anti-interference detection method of the wafer temperature measurement system according to claim 5, characterized in that: The temperature detector is arranged in a closed shielding shell, and the temperature detector is well grounded through the shielding shell.

Images