Temperature compensation method, temperature compensation device and process chamber

Abstract

The invention provides a temperature compensation method applied to a process chamber and the process chamber, and the method comprises the steps: determining a first temperature compensation value corresponding to each thermodetector according to a film thickness value of a film layer at a corresponding position of the thermodetector and a temperature sensitive value of the process chamber; inputting a difference value between the film thickness value corresponding to each thermodetector and an average value of the plurality of film thickness values, a difference value between the uniform value of the film layer and a preset uniform value of the film layer, and a first temperature compensation value corresponding to each thermodetector into a pre-trained temperature compensation model to obtain a second temperature compensation value corresponding to each thermodetector; and obtaining a target temperature compensation value corresponding to the thermodetector according to the first temperature compensation value and the second temperature compensation value corresponding to the thermodetector. According to the temperature compensation method provided by the invention, the temperature compensation value meeting the requirement can be found only through one-step debugging, the consumption of wafers for testing is saved, the operation of a machine is simplified, and the temperature compensation value can be automatically determined by the machine.

Description

Technical Field

[0001] The present invention relates to the field of process chambers, and in particular to a temperature compensation method applied to a process chamber, a temperature compensation device and a process chamber. Background Art

[0002] In the process of IC manufacturing, according to special requirements, a thin oxide film needs to be formed between the silicon substrate and other deposited layers. For example, in the preparation of flash memory, an oxide-nitride-oxide (ONO) layer needs to be formed between the control gate and the word line for isolation to allow the normal flow of electrons and improve the stability of the device. The oxide layer in the ONO layer needs to be obtained by rapid thermal oxidation (RTO) process. By rapidly raising and lowering the temperature, a layer of silicon dioxide (SiO2) with controllable thickness and thickness uniformity is formed on the surface of silicon (Si). The rapid thermal oxidation process needs to be completed in a rapid thermal processing (RTP) machine.

[0003] The rapid thermal processing machine usually includes a chamber, multiple heating lamps arranged in the chamber, a carrier plate and multiple infrared thermometers. The heating lamps are used to heat the wafers carried on the carrier plate, and the infrared thermometers are used to monitor the temperature of various locations on the wafer in real time. In order to ensure that the thickness uniformity of the oxide film generated by the rapid thermal oxidation process is within the required range, after each chamber opening operation, it is necessary to perform a tempMatch operation on the 7 infrared thermometers in the chamber, that is, to perform single-point temperature calibration on the 7 infrared thermometers through a standard light source to achieve recalibration of the rapid thermal oxidation process. The specific meaning of recalibration is to adjust the thickness uniformity of the oxide film generated by the rapid thermal oxidation process to an appropriate value by adding appropriate compensation values ​​(offset values) to the 7 infrared thermometers.

[0004] In the related art, the process engineer usually manually adjusts the appropriate compensation value in the recalibration step, and needs to perform multiple rapid thermal oxidation processes to confirm whether the debugging direction is correct. Therefore, a certain number of wafers are required to find the appropriate compensation value combination, and the adjustment amount of each compensation value depends on the process engineer's own experience and feeling, which is not conducive to promotion on the client side.

[0005] Therefore, how to provide a temperature compensation method for a process chamber that can accurately and efficiently determine the compensation value of a thermometer has become a technical problem that needs to be solved urgently in the field. Summary of the invention

[0006] The present invention aims to provide a temperature compensation method applied to a process chamber, a temperature compensation device and a process chamber capable of implementing the method. The temperature compensation method can accurately and efficiently determine the compensation value of a temperature measuring instrument.

[0007] To achieve the above object, as one aspect of the present invention, a temperature compensation method for a process chamber is provided, wherein the process chamber comprises a support base for supporting a substrate, a heater, and a plurality of temperature measuring instruments sequentially arranged along the radial direction of the support base, wherein the temperature

[0008] Compensation methods include:

[0009] For each of the thermometers, a first temperature compensation value corresponding to the thermometer is determined according to a film thickness value of the film layer at a position corresponding to the thermometer and a temperature sensitivity value of the process chamber; wherein the film layer is a film layer formed on the substrate by controlling the heater by a plurality of the thermometers at an initial set temperature; and the temperature sensitivity value refers to a thickness change of the film layer under a unit temperature change;

[0010] Input the difference between the film thickness value corresponding to each of the thermometers and the average value of the plurality of film thickness values, the difference between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value corresponding to each of the thermometers into a pre-trained temperature compensation model to obtain a second temperature compensation value corresponding to each of the thermometers;

[0011] For each of the thermometers, a target temperature compensation value corresponding to the thermometer is obtained according to the first temperature compensation value and the second temperature compensation value corresponding to the thermometer.

[0012] Optionally, a fully connected neural network algorithm is used to construct the temperature compensation model, wherein the temperature compensation model includes an input layer, an output layer and at least one hidden layer;

[0013] The step of inputting the difference between the film thickness value corresponding to each of the thermometers and the average value of the plurality of film thickness values, the difference between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value corresponding to each of the thermometers into a pre-trained temperature compensation model to obtain the second temperature compensation value corresponding to each of the thermometers comprises:

[0014] Inputting the difference between the film thickness value corresponding to each of the thermometers and the average value of the plurality of film thickness values, the difference between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value corresponding to each of the thermometers into the input layer;

[0015] A second temperature compensation value corresponding to each of the temperature measuring instruments is obtained from the output layer.

[0016] Optionally, the expression of the input layer is: a1=(x-offset1)×gain1+x min , where x is the difference between the film thickness value and the average value, the difference between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value, offset1, gain1 and x min are all training parameters, and a1 is the output result of the input layer.

[0017] Optionally, the expression of the hidden layer is: n=b1+LW1×a1, Among them, LW1 is the weight parameter of the first layer, b1 is the training parameter, and a2 is the output result of the input layer.

[0018] Optionally, the expression of the output layer is: a3=b2+LW2×a2, y=(a3-y min )÷gain2+offset2, where LW2 is the second layer weight parameter, b2, gain2, offset2 and y min are all training parameters, and y is the second temperature compensation value.

[0019] Optionally, determining a first temperature compensation value corresponding to the thermometer according to a film thickness value of a film layer at a position corresponding to the thermometer and a temperature sensitivity value of the process chamber includes:

[0020] For each of the thermometers, the difference between the film thickness value of the film layer at the position corresponding to the thermometer and the average value of the plurality of film thickness values ​​is divided by the temperature sensitivity value to obtain the first temperature compensation value corresponding to the thermometer.

[0021] Optionally, obtaining a target temperature compensation value corresponding to the thermometer according to the first temperature compensation value and the second temperature compensation value corresponding to the thermometer includes:

[0022] The first temperature compensation value and the second temperature compensation value are weightedly summed to obtain the corresponding target temperature compensation value.

[0023] Optionally, the weight factor of the first temperature compensation value is greater than the weight factor of the second temperature compensation value, and the weight factor of the first temperature compensation value and the weight factor of the second temperature compensation value are both greater than 0 and less than or equal to 1.

[0024] As a second aspect of the present invention, a temperature compensation device is provided, which includes a processor and a memory, wherein a computer program is stored in the memory, and when the processor executes the computer program, the temperature compensation method described above is executed.

[0025] As a third aspect of the present invention, a process chamber is provided, comprising a cavity, a controller, a carrier plate, a plurality of heating lamps and a plurality of thermometers, wherein the carrier plate is arranged in the cavity, the plurality of heating lamps are used to heat the wafers carried on the carrier plate, and the plurality of thermometers are used to detect the temperature of various locations on the wafers; the controller comprises a processor and a memory, a computer program is stored in the memory, and when the processor executes the computer program, the temperature compensation method described above is executed.

[0026] In the temperature compensation method, temperature compensation device and process chamber provided by the present invention and applied to the process chamber, the controller first preliminarily determines the first temperature compensation value corresponding to each thermometer according to the difference corresponding to each thermometer, and then determines the second temperature compensation value corresponding to each thermometer based on the neural network algorithm according to the previously calculated data, and finally determines the target temperature compensation value for compensating the thermometer according to the first temperature compensation value and the second temperature compensation value.

[0027] The temperature compensation method provided by the present invention only requires one film deposition process and debugging to find a temperature compensation value that meets the requirements, which saves the consumption of test wafers and simplifies the machine operation required after each tempMatch operation. In addition, the controller uses a neural network algorithm to replace the process engineer's operation of selectively enlarging and reducing the temperature compensation value according to his own understanding, which can realize the automatic determination of the temperature compensation value by the machine, thereby improving the machine debugging efficiency, and enhancing the machine production capacity and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present invention but do not constitute a limitation of the present invention. In the accompanying drawings:

[0029] Figure 1 is a schematic structural diagram of a process chamber provided by an embodiment of the present invention;

[0030] Figure 2 is a schematic flow chart of a temperature compensation method applied to a process chamber provided by an embodiment of the present invention;

[0031] Figure 3 It is a schematic diagram of the structure of a neural network in a temperature compensation method applied to a process chamber provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the present invention, and is not used to limit the present invention.

[0033] In the related art, the process engineer usually calculates the difference Δd between the film thickness value at the corresponding position of the temperature measuring instrument and the average value of multiple film thickness values. i Calculate the required temperature compensation value, and then selectively enlarge or reduce the calculated temperature compensation value based on your understanding of the machine, that is, fine-tune the calculation result based on "feeling" to achieve the effect of leveling the film thickness. This process requires process engineers to perform multiple rapid thermal oxidation processes to confirm whether the direction of their debugging based on feeling is correct, and the adjustment amount is heavily dependent on the process engineer's own experience and feeling, which is not conducive to promotion on the client side.

[0034] To solve the above technical problems, as one aspect of the present invention, a temperature compensation method for a process chamber is provided. The process chamber comprises a support base for supporting a substrate, a heater, and a plurality of temperature measuring instruments sequentially arranged along the radial direction of the support base, such as Figure 2 As shown, the temperature compensation method includes:

[0035] Step S1: for each thermometer, determine the first temperature compensation value offset corresponding to the thermometer according to the film thickness value of the film layer at the corresponding position of the thermometer and the temperature sensitivity value δ of the process chamber. i (i=1,2,…n, where n is the number of thermometers); wherein the film layer is formed on the substrate by controlling the heater by multiple thermometers at the initial set temperature; the temperature sensitivity value δ refers to the thickness change of the film layer under unit temperature change;

[0036] Step S2: calculate the difference Δd between the film thickness value corresponding to each temperature measuring instrument and the average value of multiple film thickness values. i , the difference Δr between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value offset corresponding to each temperature measuring instrument i Input the pre-trained temperature compensation model to obtain the second temperature compensation value Δt corresponding to each thermometer i ;

[0037] Step S3: for each thermometer, obtain a target temperature compensation value ΔT corresponding to the thermometer according to the first temperature compensation value and the second temperature compensation value corresponding to the thermometer. i .

[0038] In the temperature compensation method for a process chamber provided by the present invention, firstly, according to the difference Δd corresponding to each temperature measuring instrument, i Preliminarily determine the first temperature compensation value offset corresponding to each thermometer i (corresponding to the step of calculating the temperature compensation value in the related art), and then determining the second temperature compensation value Δt corresponding to each thermometer based on the neural network algorithm according to the previously calculated data i(corresponding to the steps in the related art where process engineers enlarge and reduce the temperature compensation value according to their own understanding), and finally according to the first temperature compensation value offset i and the second temperature compensation value Δt i Determine the target temperature compensation value ΔT used to compensate the thermometer i .

[0039] The temperature compensation method for a process chamber provided by the present invention can find a target temperature compensation value ΔT that meets the requirements by only one film deposition process and debugging. i , saving the consumption of test wafers and simplifying the machine operation required after each tempMatch operation. In addition, the temperature compensation method provided by the present invention uses a neural network algorithm to replace the process engineer's operation of selectively enlarging and reducing the temperature compensation value according to his own understanding, which can realize the machine's automatic determination of the temperature compensation value, thereby improving the machine debugging efficiency, and enhancing the machine's production capacity and user experience.

[0040] As an optional implementation of the present invention, the preset film layer uniformity value (ie, the target value of film thickness uniformity) may be set to 0.3%.

[0041] As an optional implementation of the present invention, a temperature compensation model is constructed using a fully connected neural network algorithm, and the temperature compensation model includes an input layer, an output layer, and at least one hidden layer.

[0042] The difference Δd between the film thickness value corresponding to each temperature measuring instrument and the average value of multiple film thickness values i , the difference Δr between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value offset corresponding to each temperature measuring instrument i Input the pre-trained temperature compensation model to obtain the second temperature compensation value Δt corresponding to each thermometer i ,include:

[0043] The difference Δd between the film thickness value corresponding to each temperature measuring instrument and the average value of multiple film thickness values i , the difference Δr between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value offset corresponding to each temperature measuring instrument i Enter the input layer;

[0044] The second temperature compensation value Δt corresponding to each thermometer is obtained from the output layer i .

[0045] Considering that the use scenario of the neural network algorithm in the present invention is not complicated (the so-called complexity refers to the factors in the scenario that affect the same input but the system will give different outputs. For example, in autonomous driving, the same destination is input, but there are often some unexpected situations on the road, such as pedestrians suddenly rushing out, and the car next to you suddenly not obeying the traffic rules, resulting in a different path to the destination each time. For example, in the image recognition task, to recognize the same face, the light will cause different light and dark areas, and different accessories on the face (such as wearing different glasses, changing hairstyles, etc.) will cause different recognition results. In the current scenario, the same offset value is input, and the film thickness value obtained after the chamber runs the RTO process must be the same, or fluctuates within a very small range. The conditions for each RTO process operation are also the same, so it can be considered that this use scenario is not complicated), so the fully connected neural network algorithm with a relatively simple structure is selected, and the deep convolutional neural network with a large number of layers, ResNet and other neural network algorithms with more complex structures are not considered.

[0046] As an optional embodiment of the present invention, Figure 3 As shown, the neural network algorithm includes an input layer, a hidden layer and an output layer.

[0047] As an optional implementation of the present invention, the expression of the input layer is: a1 = (x-offset1) × gain1 + x min , where x is the difference Δd between the film thickness value and the average value i , the difference Δr between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value offset i , offset1, gain1, and x min are all training parameters (the values ​​of the training parameters are automatically updated by the algorithm during the model training process, and the training parameters are fixed to constants after the training is completed), and a1 is the output result of the input layer.

[0048] The input value x of the input layer is:

[0049] [Δd1,Δd2,…Δd n ,Δr,offset1,offset2,…offset n That is, the input value of the input layer includes a plurality of first temperature compensation values ​​Δd1, Δd2, ... Δd corresponding to a plurality of temperature measuring instruments. n , a difference Δr between the average value and the target film thickness average value, and a plurality of first temperature compensation values ​​offset1, offset2, ... offset n .

[0050] As an optional embodiment of the present invention, Figure 1As shown, the process chamber includes 7 temperature measuring instruments 300. Then the input value x of the input layer is:

[0051] [Δd1,Δd2,Δd3,Δd4,Δd5,Δd6,Δd7,Δr,offset1,offset2,offset3,offset4,offset5,offset6,offset7].

[0052] As an optional implementation of the present invention, the hidden layer (Hidden Layer) adopts the Sigmoid activation function, and the expression of the hidden layer is: n = b1 + LW1 × a1, Among them, LW1 is the weight parameter of the first layer (that is, the weight parameter of this layer, which is also automatically updated by the algorithm during the training process and fixed as a constant after the training is completed), b1 is the training parameter, and a2 is the output result of the input layer.

[0053] As an optional implementation of the present invention, the output layer (Output Layer) adopts a linear regression output function, and the expression of the output layer is: a3=b2+LW2×a2, y=(a3-y min )÷gain2+offset2, where LW2 is the second layer weight parameter, b2, gain2, offset2 and y min are all training parameters, y is the second temperature compensation value Δt i .

[0054] As an optional embodiment of the present invention, Figure 1 As shown, the process chamber includes 7 temperature measuring instruments 300. Then the output value y of the output layer is:

[0055] [Δt1,Δt2,Δt3,Δt4,Δt5,Δt6,Δt7].

[0056] As an optional embodiment of the present invention, the temperature compensation method applied to the process chamber also includes:

[0057] The parameters of the neural network algorithm are initialized by the Xavier initialization algorithm, and the neural network algorithm is trained by the Levenberg-Marquardt algorithm.

[0058] As an optional embodiment of the present invention, the data set used in the training process of the neural network algorithm is collected from actual machine data. Specifically, the data set used for training can be all the film thickness data and temperature compensation value change data of the machine that have been debugged from zero compensation (0 offset) in the past.

[0059] As an optional embodiment of the present invention, in step S1, the first temperature compensation value offset corresponding to the thermometer is determined according to the film thickness value of the film layer at the corresponding position of the thermometer and the temperature sensitivity value δ of the process chamber. i Specifically include:

[0060] For each of the temperature measuring instruments, the difference Δd between the film thickness value of the film layer at the corresponding position of the temperature measuring instrument and the average value of the plurality of film thickness values ​​is calculated. i Divide by the temperature sensitivity value δ to get the first temperature compensation value offset corresponding to the thermometer i .

[0061] That is, the first temperature compensation value offset i =Δd i ÷δ, where the temperature sensitivity value δ is a constant that can be determined in advance.

[0062] As an optional implementation of the present invention, in step S3, the target temperature compensation value ΔT corresponding to the thermometer is obtained according to the first temperature compensation value and the second temperature compensation value corresponding to the thermometer. i Specifically include:

[0063] Each first temperature compensation value offset i The corresponding second temperature compensation value Δt i Weighted summation to obtain the corresponding target temperature compensation value ΔT i .

[0064] That is, the target temperature compensation value ΔT i =offset i ×a+Δt i ×b, where a and b are weight factors, which can be predetermined by experiments. The value range of the weight factor is 0 to 1. According to the experimental results, in general, the value of the weight factor a is greater than the value of the weight factor b.

[0065] As an optional implementation manner of the present invention, the weight factor of the first temperature compensation value may be set to 0.9, and the weight factor of the second temperature compensation value may be set to 0.2.

[0066] As an optional embodiment of the present invention, the temperature compensation method applied to the process chamber also includes:

[0067] Set the target temperature compensation value ΔT i Fill in the host computer software and perform the film deposition process again to verify the compensation effect.

[0068] As a second aspect of the present invention, a temperature compensation device is provided. The temperature compensation device includes a processor and a memory. The memory stores a computer program. When the processor executes the computer program, it executes the temperature compensation method provided in an embodiment of the present invention.

[0069] In the temperature compensation device provided by the present invention, the processor first calculates the temperature compensation value Δd corresponding to each temperature measuring instrument. i Preliminarily determine the first temperature compensation value offset corresponding to each thermometer i (corresponding to the step of calculating the temperature compensation value in the related art), and then determining the second temperature compensation value Δt corresponding to each thermometer based on the neural network algorithm according to the previously calculated data i (corresponding to the steps in the related art where process engineers enlarge and reduce the temperature compensation value according to their own understanding), and finally according to the first temperature compensation value offset i and the second temperature compensation value Δt i Determine the target temperature compensation value ΔT used to compensate the thermometer i .

[0070] The temperature compensation device provided by the present invention only needs one step of film deposition process and debugging to find the target temperature compensation value ΔT that meets the requirements. i , saving the consumption of test wafers and simplifying the machine operation required after each tempMatch operation. In addition, the controller uses a neural network algorithm to replace the process engineer's operation of selectively scaling up and down the temperature compensation value based on his own understanding, which can realize the machine's automatic determination of the temperature compensation value, thereby improving the user experience.

[0071] As a third aspect of the present invention, a process chamber is provided, such as Figure 1 As shown, the process chamber includes a cavity, a controller, a carrier plate 200, multiple heating lamps 100 and multiple thermometers 300. The carrier plate 200 is arranged in the cavity. The multiple heating lamps are used to heat the wafer 10 (wafer) carried on the carrier plate 200, and the multiple thermometers 300 are used to detect the temperature of various locations on the wafer 10. The controller includes a processor and a memory. The memory stores a computer program. When the processor executes the computer program, it executes the temperature compensation method provided in the embodiment of the present invention.

[0072] In the process chamber of the temperature measuring instrument provided by the present invention, the controller firstly determines the difference Δd corresponding to each temperature measuring instrument. iThe first temperature compensation value offset corresponding to each thermometer is preliminarily determined (corresponding to the step of calculating the temperature compensation value in the related art), and then the second temperature compensation value Δt corresponding to each thermometer is determined based on the neural network algorithm according to the previously calculated data (corresponding to the step of the process engineer in the related art enlarging and reducing the temperature compensation value according to his own understanding), and finally i and the second temperature compensation value Δt i Determine the target temperature compensation value ΔT used to compensate the thermometer i .

[0073] The process chamber provided by the present invention only needs one step of film deposition process and debugging to find the target temperature compensation value ΔT that meets the requirements. i , saving the consumption of test wafers and simplifying the machine operation required after each tempMatch operation. In addition, the controller uses a neural network algorithm to replace the process engineer's operation of selectively scaling up and down the temperature compensation value based on his own understanding, which can realize the machine's automatic determination of the temperature compensation value, thereby improving the user experience.

[0074] As an optional implementation of the present invention, the process chamber is a rapid thermal processing (RTP) device.

[0075] As an optional implementation of the present invention, the thermometer 300 is an infrared thermometer.

[0076] As an optional embodiment of the present invention, Figure 1 As shown, the process chamber includes seven temperature measuring instruments 300 .

[0077] In other embodiments of the present invention, the process chamber may also include other numbers of thermometers 300. For example, the process chamber may include three thermometers 300, and the controller needs to calculate three target temperature compensation values ​​(offset values) accordingly; or, the process chamber may include nine thermometers 300, and the controller needs to calculate nine target temperature compensation values ​​accordingly.

[0078] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present invention, but the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A temperature compensation method for a process chamber, characterized in that: The process chamber comprises a supporting base for supporting a substrate, a heater, and a plurality of temperature measuring instruments sequentially arranged along a radial direction of the supporting base, and the temperature compensation method comprises: For each of the thermometers, a first temperature compensation value corresponding to the thermometer is determined according to a film thickness value of the film layer at a position corresponding to the thermometer and a temperature sensitivity value of the process chamber; wherein the film layer is a film layer formed on the substrate by controlling the heater by a plurality of the thermometers at an initial set temperature; and the temperature sensitivity value refers to a thickness change of the film layer under a unit temperature change; Input the difference between the film thickness value corresponding to each of the thermometers and the average value of the plurality of film thickness values, the difference between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value corresponding to each of the thermometers into a pre-trained temperature compensation model to obtain a second temperature compensation value corresponding to each of the thermometers; For each of the thermometers, a target temperature compensation value corresponding to the thermometer is obtained according to the first temperature compensation value and the second temperature compensation value corresponding to the thermometer.

2. The temperature compensation method for a process chamber according to claim 1, characterized in that: The temperature compensation model is constructed by using a fully connected neural network algorithm, wherein the temperature compensation model includes an input layer, an output layer and at least one hidden layer; The step of inputting the difference between the film thickness value corresponding to each of the thermometers and the average value of the plurality of film thickness values, the difference between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value corresponding to each of the thermometers into a pre-trained temperature compensation model to obtain the second temperature compensation value corresponding to each of the thermometers comprises: Inputting the difference between the film thickness value corresponding to each of the thermometers and the average value of the plurality of film thickness values, the difference between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value corresponding to each of the thermometers into the input layer; A second temperature compensation value corresponding to each of the temperature measuring instruments is obtained from the output layer.

3. The temperature compensation method for a process chamber according to claim 2, characterized in that: The expression of the input layer is: a1 = (x-offset1) × gain1 + x min , where x is the difference between the film thickness value and the average value, the difference between the uniform value of the film layer and the preset uniform value of the film layer, and the first temperature compensation value, offset1, gain1 and x min are all training parameters, and a1 is the output result of the input layer.

4. The temperature compensation method for a process chamber according to claim 2, characterized in that: The expression of the hidden layer is: n = b1 + LW1 × a1, Among them, LW1 is the weight parameter of the first layer, b1 is the training parameter, and a2 is the output result of the input layer.

5. The temperature compensation method for a process chamber according to claim 2, characterized in that: The expression of the output layer is: a3=b2+LW2×a2, y=(a3-y min )÷gain2+offset2, where LW2 is the second layer weight parameter, b2, gain2, offset2 and y min are all training parameters, and y is the second temperature compensation value.

6. The temperature compensation method for a process chamber according to any one of claims 1 to 5, characterized in that: The determining of a first temperature compensation value corresponding to the thermometer according to a film thickness value of a film layer at a position corresponding to the thermometer and a temperature sensitivity value of the process chamber comprises: For each of the thermometers, the difference between the film thickness value of the film layer at the position corresponding to the thermometer and the average value of the plurality of film thickness values ​​is divided by the temperature sensitivity value to obtain the first temperature compensation value corresponding to the thermometer.

7. The temperature compensation method for a process chamber according to any one of claims 1 to 5, characterized in that: The step of obtaining a target temperature compensation value corresponding to the thermometer according to the first temperature compensation value and the second temperature compensation value corresponding to the thermometer includes: The first temperature compensation value and the second temperature compensation value are weightedly summed to obtain the corresponding target temperature compensation value.

8. The temperature compensation method for a process chamber according to claim 7, characterized in that: The weight factor of the first temperature compensation value is greater than the weight factor of the second temperature compensation value, and the weight factor of the first temperature compensation value and the weight factor of the second temperature compensation value are both greater than 0 and less than or equal to 1.

9. A temperature compensation device, characterized in that: The temperature compensation device includes a processor and a memory, wherein a computer program is stored in the memory, and when the processor executes the computer program, the temperature compensation method according to any one of claims 1 to 8 is executed.

10. A process chamber, characterized in that: It includes a cavity, a controller, a carrying plate, multiple heating lamps and multiple thermometers, the carrying plate is arranged in the cavity, the multiple heating lamps are used to heat the wafers carried on the carrying plate, and the multiple thermometers are used to detect the temperature of various parts of the wafers; the controller includes a processor and a memory, the memory stores a computer program, and when the processor executes the computer program, it executes the temperature compensation method described in any one of claims 1-8.

Images