CN-121655700-B - Method and device for detecting semiconductor temperature field

CN121655700BCN 121655700 BCN121655700 BCN 121655700BCN-121655700-B

Abstract

The invention relates to the technical field of semiconductor detection, and provides a method and a device for detecting a semiconductor temperature field. The method comprises the steps of carrying out point-by-point scanning on the surface of a wafer to obtain radiation information of all spatial position points on the surface of the wafer, sequentially separating radiation energy with different wavelengths according to the radiation information of each spatial position point, generating radiation data of each spatial position point under a plurality of wavelengths based on the radiation energy with each wavelength, coordinating the time sequence of the point-by-point scanning and the radiation energy with each wavelength separation to enable each spatial position point to be matched with corresponding multi-wavelength radiation data, obtaining multi-wavelength radiation data, constructing a radiation equation set based on the multi-wavelength radiation data of the same spatial position point, and synchronously inverting and calculating temperature values of each point on the surface of the wafer according to the radiation equation set. The invention is used for carrying out self-adaption and high-reliability non-contact detection on the surface temperature field without depending on preset parameters of emissivity in the process of manufacturing semiconductors.

Inventors

LIU XINYANG
XING CHENCHEN
ZHANG LIMING

Assignees

上海车仪田科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260206

Claims (9)

1. A method for detecting a semiconductor temperature field, which is applied to a device for detecting the semiconductor temperature field, and is characterized in that the method comprises the following steps: carrying out point-by-point scanning on the surface of the wafer to obtain radiation information of each spatial position point on the surface of the wafer; Sequentially separating out radiation energy with different wavelengths according to the radiation information of each spatial position point; Generating radiation data for each spatial location point at a plurality of wavelengths based on the radiation energy at each wavelength; Coordinating the time sequence of the point-by-point scanning and the radiation energy of each wavelength separation, so that each spatial position point is matched with the corresponding multi-wavelength radiation data; The method comprises the steps of obtaining multi-wavelength radiation data, constructing a radiation equation set based on the multi-wavelength radiation data of the same spatial position point, synchronously inverting and calculating temperature values of all points on the surface of a wafer according to the radiation equation set, establishing an emissivity wavelength model for each spatial position point, wherein the emissivity wavelength model is expressed by adopting a primary function of wavelength, constructing a theoretical radiation intensity model based on Planckian radiation law and combining the emissivity wavelength model, defining an optimization criterion, solving a temperature parameter and an emissivity model coefficient by taking the least square sum of deviation of the theoretical radiation intensity and the actually measured radiation intensity as a solution target, and stopping iteration when the iteration parameter variation or residual error is smaller than a preset threshold value, and calculating the emissivity value of the spatial position point under all wavelengths based on the emissivity model parameter.
2. The method of claim 1, wherein coordinating the point-wise scanning with the timing of separating the radiant energy of each wavelength comprises: The method comprises the steps of carrying out space scanning according to a preset scanning path, obtaining an actual feedback value of a scanning position in real time, converting the actual feedback value into a space coordinate through a geometric calibration relation, continuously detecting an angle of a filter wheel, converting the angle of the filter wheel into a corresponding wavelength parameter through a preset mapping relation, generating a sampling trigger signal when the scanning position is judged to reach a stable state at a target point and the center of the filter exactly passes through an optical axis, collecting a radiation signal based on the sampling trigger signal, and associating the corresponding space coordinate and the wavelength parameter with the radiation signal to establish a corresponding relation between the space position and multi-wavelength radiation data.
3. The method of claim 1, further comprising the step of preprocessing the data prior to constructing the set of radiation equations: collecting original data comprising space coordinates, wavelength identifiers and original radiation signals; subtracting dark noise at a corresponding wavelength from the original radiation signal to obtain a net radiation signal; correcting the net radiation signal based on a pre-calibrated system spectral response function to obtain accurate radiation intensity values under each wavelength; A corrected multi-wavelength radiation intensity dataset is generated for each spatial location point as input data for constructing a system of radiation equations.
4. The method of claim 1, further comprising, after calculating temperature values for each point on the wafer surface from the simultaneous inversion of the set of radiation equations: Integrating the temperature values of all the spatial position points on the surface of the wafer based on the multi-wavelength radiation data of the same spatial position point and the emissivity values of all the points on the surface of the wafer calculated by inversion to generate full-field temperature distribution data; Performing spatial interpolation processing on the full-field temperature distribution data to form continuous temperature field distribution; Converting the continuous temperature field distribution into a visual image for output; And generating an emissivity distribution map of the wafer surface based on the calculated emissivity values of each point of the wafer surface by inversion, wherein the emissivity distribution map is used for representing the state characteristics of the wafer surface.
5. A method according to claim 3, wherein the subtraction of dark noise at the corresponding wavelength from the original radiation signal yields a net radiation signal The following formula is adopted for calculation: ; Wherein, the Is the background voltage of the corresponding wavelength channel measured by shielding the light path, Is the original radiation signal value; the net radiation signal is corrected based on a pre-calibrated system spectral response function to obtain radiation intensity values under each wavelength The following formula is adopted for calculation: ; Wherein, the The system is obtained by calibrating a blackbody furnace at the wavelength The spectral response function below.
6. The method of claim 1, wherein the theoretical radiation intensity The model construction of (2) adopts the following expression: ; Wherein, the Indicating that the ideal blackbody is at wavelength And the intensity of the spectral radiation at a temperature T, And And the emissivity model coefficients to be inverted.
7. The device is characterized by comprising a space scanning mechanism, a wavelength time sequence separating mechanism, a single-point radiation detecting unit, a synchronous control mechanism and a data processing mechanism; The space scanning mechanism is used for scanning the surface of the wafer point by point to acquire radiation information of each space position point on the surface of the wafer; The wavelength time sequence separating mechanism is arranged in the light path of the space scanning mechanism and is used for sequentially separating out radiation energy with different wavelengths according to the radiation information of each space position point; the single-point radiation detection unit is used for receiving the radiation energy of each wavelength separated by the wavelength time sequence separation mechanism and generating radiation data of each spatial position point under a plurality of wavelengths; The synchronous control mechanism is used for coordinating the working time sequences of the space scanning mechanism and the wavelength time sequence separating mechanism, so that each space position point is accurately matched with the corresponding multi-wavelength radiation data; The data processing mechanism is used for receiving multi-wavelength radiation data generated by the single-point radiation detection unit, constructing a radiation equation set based on the multi-wavelength radiation data of the same space position point, synchronously inverting and calculating temperature values of each point on the surface of a wafer according to the radiation equation set, and comprises the steps of establishing an emissivity wavelength model for each space position point, wherein the emissivity wavelength model is expressed by adopting a primary function of wavelength, constructing a theoretical radiation intensity model based on Planckian radiation law and combining with the emissivity wavelength model, defining an optimization criterion, solving a temperature parameter and an emissivity model coefficient by using the minimum sum of squares of deviation of the theoretical radiation intensity and the actually measured radiation intensity as a solution target, and calculating the temperature parameter and the emissivity model coefficient by an iterative optimization method, wherein the iterative optimization method comprises the steps of parameter initialization, calculating an update quantity, parameter updating and convergence judging, ending iteration when the iteration parameter change quantity or residual error is smaller than a preset threshold value, outputting the temperature value and the emissivity model parameter of the space position point, and calculating the emissivity value of the space position point under each wavelength based on the emissivity model parameter.
8. The device according to claim 7, wherein the wavelength timing separation mechanism comprises a filter wheel, a rotation driving unit and an angle detection unit, wherein a plurality of narrow-band filters with different center wavelengths are installed on the filter wheel, the narrow-band filters with different wavelengths sequentially enter an optical path when the filter wheel rotates to perform wavelength selective separation on radiation information of the same spatial position point, the rotation driving unit is used for driving the filter wheel to rotate at a constant angular velocity, and the angle detection unit is installed at the coaxial position of the filter wheel to continuously detect the angle of the filter wheel.
9. The apparatus of claim 7, wherein the single point radiation detection unit has a pre-amplification circuit with an adjustable gain, the gain of the pre-amplification circuit being dynamically adjusted according to the radiation intensity to adapt the intensity of the radiant energy of different wavelengths.

Description

Method and device for detecting semiconductor temperature field Technical Field The present invention relates to the field of semiconductor detection technology, and in particular, to a method and apparatus for detecting a semiconductor temperature field. Background In the related art of semiconductor manufacturing processes, accurate detection of a temperature field is of decisive importance for process quality control. The current application of the non-contact infrared temperature measurement technology faces the core challenge that the emissivity of the surface of the semiconductor material is influenced by coupling of multiple factors such as the intrinsic characteristics of a wafer, the surface oxidation state, the microcosmic appearance, the process dynamic process and the like, and the surface of the semiconductor material presents obvious unknowns, spatial non-uniformity and time-varying characteristics. The traditional single-wavelength temperature measurement method is required to rely on preset emissivity parameters for temperature calculation, and when the actual emissivity deviates from a preset value, systematic temperature measurement errors are directly introduced, so that process monitoring misalignment is caused. To reduce emissivity dependence, dual wavelength thermometry techniques eliminate emissivity parameters by the ratio of the radiation intensities of specific wavelength combinations. However, this approach is based on the theoretical assumption that "emissivity satisfies a simplified functional relationship" at selected wavelengths, whereas semiconductor surfaces generally exhibit complex spectral selectivity characteristics in the infrared band, making this assumption difficult to hold in most practical process scenarios, and the inversion results are still affected by model mismatch. The multispectral temperature measurement technology has the potential of synchronously solving the temperature and the emissivity in principle, but the existing implementation path has structural defects that the adoption of a planar array multispectral imaging scheme is high in cost, sub-pixel level space registration is difficult to realize among images of different wavebands, so that space dislocation exists in multi-wavelength radiation data of the same physical position, and the scheme based on mechanical scanning and time sequence filtering is lack of an accurate time sequence cooperative mechanism of a scanning position and a wavelength switching process, so that time sequence offset exists in radiation acquisition moments of the same spatial point under different wavelengths, and when a temperature gradient exists on a measured surface or the temperature gradient is in a dynamic change process, the offset causes data mismatch and motion blurring, breaks a space-time consistency basis necessary for constructing a radiation equation set, and loses physical reliability of temperature inversion. Therefore, a method and apparatus for detecting a semiconductor temperature field are needed to improve the above-mentioned problems. Disclosure of Invention The invention provides a method and a device for detecting a semiconductor temperature field, which are used for carrying out self-adaptive and high-reliability non-contact detection on the surface temperature field without depending on preset parameters of emissivity in the process of semiconductor manufacturing. According to a first aspect of the embodiment of the invention, a detection method of a semiconductor temperature field is provided and applied to a detection device of the semiconductor temperature field, and the method comprises the steps of carrying out point-by-point scanning on the surface of a wafer to obtain radiation information of all spatial position points on the surface of the wafer, sequentially separating radiation energy with different wavelengths according to the radiation information of each spatial position point, generating radiation data of each spatial position point under a plurality of wavelengths based on the radiation energy with each wavelength, coordinating the point-by-point scanning with the time sequence of the radiation energy with each wavelength, enabling each spatial position point to be matched with corresponding multi-wavelength radiation data, obtaining multi-wavelength radiation data, constructing a radiation equation set based on the multi-wavelength radiation data of the same spatial position point, and synchronously inverting and calculating the temperature value of each point on the surface of the wafer according to the radiation equation set. In one embodiment, the coordination of point-by-point scanning and the time sequence of separating the radiation energy of each wavelength comprises the steps of carrying out space scanning according to a preset scanning path, acquiring an actual feedback value of a scanning position in real time, converting the actual feedback value into space coordinates