Semiconductor device simulation model calibration method based on physical feature extraction and device thereof

Abstract

The application relates to the technical field of computer-aided design, and discloses a semiconductor device simulation model calibration method and device based on physical feature extraction, which are used to solve the technical problems that the prior art cannot perform data fitting based on the physical features of semiconductor devices, and it is difficult to truly reflect the physical characteristics of real semiconductor devices and the error is large. The method comprises the following steps: acquiring a measured data sequence and a simulation data sequence, and calculating the physical feature vectors at each data point; constructing a multi-dimensional physical feature distance operator and a nonlinear physical correction adjustment function, and dynamically weighting and compensating the multi-dimensional physical feature distance operator; calling an asynchronous cumulative distance algorithm, calculating a cumulative distance value based on the measured data sequence and the simulation data sequence, and judging whether the cumulative distance value is less than a cumulative distance threshold value; if not, finding a physical working area with the largest distance deviation based on the multi-dimensional physical feature distance operator, calculating a correction value of a semiconductor device simulation model parameter, and performing feedback adjustment for calibration.

Description

Technical Field

[0001] This invention relates to the field of computer-aided design technology, and in particular to a method, apparatus, electronic device, computer storage medium, and computer program product for calibrating semiconductor device simulation models based on physical feature extraction. Background Technology

[0002] In the research and development and manufacturing of semiconductor devices, technology computer-aided design (TCAD) physical simulation is a key technology for predicting device electrical performance and optimizing process conditions. To ensure that the simulation model can accurately reflect the electrical behavior of the device after actual manufacturing, it is necessary to adjust the physical parameters in the simulation (such as doping distribution, interface state density, carrier mobility model parameters, etc.) to match the simulated current-voltage (IV) characteristic curve with the measured curve. This process is called simulation model calibration, and its core lies in establishing a fitting index that can accurately evaluate the difference between the simulated curve and the measured curve, and thereby drive the automatic optimization of model parameters.

[0003] Currently, widely used fitting evaluation metrics in engineering practice mainly include point-by-point relative error, mean squared error (MSE), and relative mean squared error. However, the current-voltage curve (IV curve) of semiconductor devices has extremely strong nonlinear characteristics, covering multiple operating regions with distinct physical mechanisms, such as the subthreshold region, linear region, and saturation region. In the subthreshold region, the current changes exponentially with the voltage, directly determining the device's leakage power consumption and subthreshold swing (SS). The linear and saturation regions involve complex physical effects such as carrier mobility degradation, velocity saturation, and channel length modulation, which determine the device's driving capability. Existing evaluation metrics have the following technical shortcomings when processing such multi-order-of-magnitude, multi-physical-mechanism curve data: (1) Lack of physical feature capture: Existing indicators (such as mean square error) tend to compensate for the weight in the high current value region, resulting in extremely low fitting accuracy in the subthreshold region where the current order of magnitude is small, and thus failing to accurately extract the turn-on voltage. Key physical parameters such as ) and subthreshold swing (SS) affect the reliability of low-power device design; (2) Poor ability to describe the "inflection point" of the curve: The "knee region" of a semiconductor device transitioning from the linear region to the saturation region contains important physical information about the degradation of carrier mobility and the change in output resistance. Existing indicators lack sensitivity to the curvature of the curve (i.e., the second derivative), and therefore cannot effectively evaluate the accuracy of the simulation model at key physical inflection points, resulting in the failure of mobility model parameter calibration.

[0004] (3) Data processing introduces human distortion: In order to take into account different orders of magnitude of current, existing technologies often use logarithmic scaling or manual weighting based on experience. Logarithmic scaling will mask the subtle physical differences in the linear region, while manual weighting is highly dependent on the experience of engineers, resulting in low automation and poor versatility of simulation model calibration, which seriously restricts the iterative efficiency of semiconductor process and technology co-optimization (DTCO).

[0005] (4) Calculation bias caused by sampling point mismatch: Experimental data and simulation data often have different voltage sampling distributions. Traditional evaluation indicators rely on interpolation to approximate the data for alignment. In regions where physical characteristics change drastically (such as the inflection point of the subthreshold region and the transition section between the linear region and the saturation region), interpolation will introduce additional numerical errors, causing the extracted physical parameters to deviate from the true physical nature.

[0006] In summary, the existing technology lacks a fitting and evaluation mechanism that can deeply couple semiconductor physical characteristics and take into account both full-scale current changes and local morphological features. This results in low accuracy and poor automation of physical parameter extraction during TCAD simulation model calibration, making it difficult to meet the actual engineering needs of advanced semiconductor process development. Summary of the Invention

[0007] The main objective of this invention is to solve the technical problem that existing semiconductor simulation models cannot perform data fitting based on the physical characteristics of semiconductor devices, making it difficult to truly reflect the physical characteristics of real semiconductor devices, thus leading to large simulation errors.

[0008] The first aspect of this invention provides a method for calibrating a semiconductor device simulation model based on physical feature extraction, comprising: Obtain the measured data sequence and simulation data sequence of the current-voltage characteristic curve of semiconductor devices; Calculate multiple physical feature vectors at each data point for the measured data sequence and the simulated data sequence, respectively; wherein, the physical feature vectors include at least numerical feature vectors, first derivative feature vectors, and curvature feature vectors; A multidimensional physical feature distance operator is constructed by weighting and combining the physical feature vectors, and a nonlinear physical correction adjustment function is constructed to dynamically weight and compensate the multidimensional physical feature distance operator. The asynchronous cumulative distance algorithm is invoked to find the globally optimal physical matching path based on the measured data sequence and the simulated data sequence, and the cumulative distance value is calculated. Determine whether the cumulative distance value is less than a preset cumulative distance threshold; If not, the physical working region with the largest distance deviation is found based on the multi-dimensional physical feature distance operator, and the correction value of the semiconductor device simulation model parameters is calculated according to the regional characteristics corresponding to the physical working region. Based on the correction value, feedback adjustment is performed on the semiconductor device simulation model, and the semiconductor device simulation model is calibrated repeatedly. If so, the calibration is completed, and a semiconductor device simulation model is obtained.

[0009] Optionally, in a first implementation of the first aspect of the present invention, the calculation expression of the multidimensional physical feature distance operator is: ; in, Represents a multidimensional physical feature distance operator. The weights of the numerical feature vectors, The weights of the first-order derivative eigenvectors are... The weights of the curvature eigenvector are... The index of the data points in the measured data sequence. The index of the data points in the simulation data sequence.

[0010] Optionally, in a second implementation of the first aspect of the present invention, the numerical feature vector includes voltage and current values ​​at each data point; The first derivative eigenvector includes the first derivatives of the voltage and current values ​​at each data point along the curve direction. The curvature feature vector includes the curvature value at each data point, which is calculated by the first and second derivatives of the voltage and current values ​​at each data point along the curve direction.

[0011] Optionally, in a third implementation of the first aspect of the present invention, the expression for the nonlinear physical correction adjustment function is: ; in, Represents the weight operator, , and These represent the current or voltage values ​​of three adjacent data points, respectively. This represents the error function.

[0012] Optionally, in a fourth implementation of the first aspect of the present invention, the step of invoking the asynchronous cumulative distance algorithm to find the globally optimal physical matching path based on the measured data sequence and the simulated data sequence, and calculating the cumulative distance value, includes: The local distance between the starting data point pairs of the measured data sequence and the simulated data sequence is calculated based on the multidimensional physical feature distance operator to set the initial deviation of the cumulative path. Based on the initial deviation, dynamic weights are determined based on the index positions of each data point pair in their respective sequences. The weighted local distance value is obtained, and the weighted local distance value is added to the historical cumulative cost with the smallest cumulative deviation in the preceding path to obtain the cumulative cost at the current matching data point pair. Perform recursive calculations, and when the data endpoints of the measured data sequence and the simulated data sequence are reached, use the cumulative cost at the endpoint as the global cumulative distance value.

[0013] Optionally, in a fifth implementation of the first aspect of the present invention, the physical working area includes: a subthreshold region, a linear region, a saturation region, and a bend between the linear region and the saturation region; The process of finding the physical working region with the largest distance deviation based on a multi-dimensional physical feature distance operator, and calculating the correction value of the semiconductor device simulation model parameters based on the regional characteristics corresponding to the physical working region, includes: The corresponding physical model parameters are determined based on the physical working area with the largest distance deviation, and the correction values ​​of the corresponding physical model parameters are calculated based on the multi-dimensional physical feature distance operator at each data point within the physical working area. The step of performing feedback adjustment on the semiconductor device simulation model based on the correction value and repeating the simulation includes: The corrected value is written into the simulation software via API interface or script control instructions to trigger the next round of simulation iteration.

[0014] A second aspect of the present invention provides a semiconductor device simulation model calibration device based on physical feature extraction, comprising: The data acquisition module is used to acquire the measured data sequence and simulation data sequence of the current-voltage characteristic curve of semiconductor devices; The feature extraction module is used to calculate multiple physical feature vectors at each data point of the measured data sequence and the simulated data sequence, respectively; wherein, the physical feature vectors include at least numerical feature vectors, first derivative feature vectors, and curvature feature vectors; The first calculation module is used to construct a multidimensional physical feature distance operator by weighting and combining each of the physical feature vectors, and to construct a nonlinear physical correction adjustment function to dynamically weight and compensate the multidimensional physical feature distance operator. The second calculation module is used to call the asynchronous cumulative distance algorithm to find the globally optimal physical matching path based on the measured data sequence and the simulation data sequence, and to calculate the cumulative distance value. A calibration judgment module is used to determine whether the cumulative distance value is less than a preset cumulative distance threshold; The calibration execution module is used to find the physical working region with the largest distance deviation based on the multi-dimensional physical feature distance operator when the judgment result output by the calibration judgment module is negative, calculate the correction value of the semiconductor device simulation model parameters according to the regional characteristics corresponding to the physical working region, perform feedback adjustment on the semiconductor device simulation model based on the correction value, and repeatedly perform calibration on the semiconductor device simulation model; it is also used to complete the calibration when the judgment result output by the calibration judgment module is positive, and obtain the semiconductor device simulation model.

[0015] Optionally, in the first implementation of the second aspect of the present invention, the calculation expression of the multidimensional physical feature distance operator is: ; in, Represents a multidimensional physical feature distance operator. The weights of the numerical feature vectors, The weights of the first-order derivative eigenvectors are... The weights of the curvature eigenvector are... The index of the data points in the measured data sequence. The index of the data points in the simulation data sequence.

[0016] Optionally, in a second implementation of the second aspect of the present invention, the numerical feature vector includes voltage and current values ​​at each data point; The first derivative eigenvector includes the first derivatives of the voltage and current values ​​at each data point along the curve direction. The curvature feature vector includes the curvature value at each data point, which is calculated by the first and second derivatives of the voltage and current values ​​at each data point along the curve direction.

[0017] Optionally, in a third implementation of the second aspect of the present invention, the expression for the nonlinear physical correction adjustment function is: ; in, Represents the weight operator, , and These represent the current or voltage values ​​of three adjacent data points, respectively. This represents the error function.

[0018] Optionally, in a fourth implementation of the second aspect of the present invention, the second calculation module is specifically used for: The local distance between the starting data point pairs of the measured data sequence and the simulated data sequence is calculated based on the multidimensional physical feature distance operator to set the initial deviation of the cumulative path. Based on the initial deviation, dynamic weights are determined based on the index positions of each data point pair in their respective sequences. The weighted local distance value is obtained, and the weighted local distance value is added to the historical cumulative cost with the smallest cumulative deviation in the preceding path to obtain the cumulative cost at the current matching data point pair. Perform recursive calculations, and when the data endpoints of the measured data sequence and the simulated data sequence are reached, use the cumulative cost at the endpoint as the global cumulative distance value.

[0019] Optionally, in a fifth implementation of the second aspect of the present invention, the physical working area includes: a subthreshold region, a linear region, a saturation region, and a bend between the linear region and the saturation region; The calibration execution module is specifically used to determine the corresponding physical model parameters based on the physical working area with the largest distance deviation, and to calculate the correction value of the corresponding physical model parameters based on the multi-dimensional physical feature distance operator at each data point in the physical working area; And to write the correction value into the simulation software via API interface or script control instructions to trigger the next round of simulation iteration.

[0020] A third aspect of the present invention provides a semiconductor device simulation model calibration device based on physical feature extraction, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the semiconductor device simulation model calibration device based on physical feature extraction to perform the steps of the above-described semiconductor device simulation model calibration method based on physical feature extraction.

[0021] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the steps of the above-described semiconductor device simulation model calibration method based on physical feature extraction.

[0022] A fifth aspect of the present invention provides a computer program product comprising a computer program / instruction that, when executed by a processor, implements the steps of the semiconductor device simulation model calibration method based on physical feature extraction as described above.

[0023] The technical solution provided by this invention involves acquiring measured data sequences and simulated data sequences of the current-voltage characteristic curves of semiconductor devices; calculating multiple physical feature vectors at each data point in the measured and simulated data sequences; wherein the physical feature vectors include at least numerical feature vectors, first-order derivative feature vectors, and curvature feature vectors; constructing a multi-dimensional physical feature distance operator by weighted combination of each physical feature vector, and constructing a nonlinear physical correction adjustment function to dynamically compensate the multi-dimensional physical feature distance operator; calling an asynchronous cumulative distance algorithm to find the globally optimal physical matching path based on the measured and simulated data sequences, and calculating the cumulative distance value; determining whether the cumulative distance value is less than a preset cumulative distance threshold; if not, finding the physical working region with the largest distance deviation based on the multi-dimensional physical feature distance operator, and calculating the correction value of the semiconductor device simulation model parameters according to the regional characteristics corresponding to the physical working region, performing feedback adjustment on the semiconductor device simulation model based on the correction value, and repeatedly performing calibration on the semiconductor device simulation model; if yes, completing the calibration and obtaining the semiconductor device simulation model.

[0024] This method can perform data fitting based on the physical characteristics of semiconductor devices, achieve high-precision calibration of simulation models, and enable the calibrated models to truly reflect the physical characteristics of real semiconductor devices, greatly reducing simulation errors.

[0025] The device, electronic device, computer-readable storage medium, and computer program product provided by this invention also solve the corresponding technical problems. Attached Figure Description

[0026] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart illustrating the first embodiment of the semiconductor device simulation model calibration method based on physical feature extraction in this invention. Figure 2 This is a schematic diagram of an embodiment of the semiconductor device simulation model calibration device based on physical feature extraction in this invention. Figure 3 This is a schematic diagram of an embodiment of a semiconductor device simulation model calibration device based on physical feature extraction according to the present invention; Figure 4 This is a schematic diagram illustrating the principle of a computer-readable medium according to an embodiment of the present invention. Detailed Implementation

[0027] Exemplary embodiments of the invention will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limiting the invention to the embodiments set forth herein. Rather, these exemplary embodiments are provided to make the invention more comprehensive and complete, and to facilitate a full communication of the inventive concept to those skilled in the art. The same reference numerals in the drawings denote the same or similar elements, components, or parts, and therefore repeated descriptions of them will be omitted.

[0028] Subject to the technical concept of this invention, the features, structures, characteristics or other details described in a particular embodiment may be combined in one or more other embodiments in a suitable manner.

[0029] In the description of specific embodiments, the features, structures, characteristics, or other details described in this invention are intended to enable those skilled in the art to fully understand the embodiments. However, it is not excluded that those skilled in the art can practice the technical solutions of this invention without one or more of the specific features, structures, characteristics, or other details.

[0030] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0031] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0032] The terms “and / or” or “and / or” include all combinations of any one or more of the listed items.

[0033] See Figure 1 The first embodiment of the semiconductor device simulation model calibration method based on physical feature extraction in this invention includes: It is understood that the executing entity of this invention can be a semiconductor device simulation model calibration device based on physical feature extraction, or it can be a terminal or a server; the specific implementation is not limited here. This embodiment of the invention will be described using a server as an example.

[0034] First, construct a TCAD (Technology Computer Aided Design) simulation model of the semiconductor device to be simulated, as well as the actual semiconductor device.

[0035] S101. Obtain the measured data sequence and simulation data sequence of the current-voltage characteristic curve of the semiconductor device; Obtain the measured data sequence of current-voltage characteristic curves (IV characteristic curves) of semiconductor devices. and simulation data sequence .in, and The measured data points and the simulated data points are respectively, and they may not be equal; that is, the number of sampling points and the voltage sampling distribution of the measured data sequence and the simulated data sequence may not be the same. The measured data sequence is obtained by scanning the current-voltage characteristics of the actual device using semiconductor testing equipment, while the simulated data sequence is obtained by performing simulation calculations using TCAD simulation software that calls the parameters of the currently constructed semiconductor device simulation model. The IV characteristic curve covers multiple physical operating regions of the semiconductor device, including at least the subthreshold region, the linear region, and the saturation region; in a preferred embodiment, the physical operating region may further include the bend at the boundary between the linear region and the saturation region.

[0036] S102. Calculate the multiple physical feature vectors of the measured data sequence and the simulated data sequence at each data point, respectively. In this embodiment, the physical feature vector includes at least a numerical feature vector, a first-order derivative feature vector, and a curvature feature vector.

[0037] For the measured data sequence of the first The data point and the simulation data sequence in the nth data point For each of the data points, calculate the physical feature vector as follows: (1) Numerical eigenvectors: Directly obtain the voltage and current values ​​at each data point, i.e. and The numerical feature vector reflects the absolute numerical state of the device at that bias point.

[0038] (2) First derivative eigenvectors: The transconductance of the device is calculated using a combination of center-difference and one-sided-difference methods. ) and output conductance ( The specific calculation formula is as follows: (This refers to the local variation characteristics, such as...) ; ; ; ; (3) Curvature eigenvector: The curvature of the measured curve and the simulated curve at each data point was calculated using the discrete curvature formula. and The morphological features of the "knee region" used to identify the transition of a device from the linear region to the saturation region are specifically calculated using the following formula: ; ; The second derivative term is calculated using the second-order central difference formula: ; ; ; ; For the boundary points of the data sequence (i.e. , ; , The derivatives of each order are calculated using the one-sided difference formula to ensure the effectiveness of the calculation at the boundary.

[0039] S103. A multidimensional physical feature distance operator is constructed by weighting and combining each physical feature vector, and a nonlinear physical correction adjustment function is constructed to dynamically compensate the multidimensional physical feature distance operator. In this step, a weighted linear combination of the above physical feature vectors is performed to construct a multidimensional physical feature distance operator that describes the closeness between measured data points and simulated data points in the physical state space. .

[0040] ; in, The weights of the numerical feature vectors, The weights of the first-order derivative eigenvectors are... The weights of the curvature eigenvector are... The index of the data points in the measured data sequence. The index of the data points in the simulation data sequence.

[0041] In a preferred embodiment, , , These weights can be adaptively adjusted according to different device types or calibration focuses.

[0042] In this embodiment, to address the issue of current spanning multiple orders of magnitude in the semiconductor IV curve (from... arrive To address the problem of severely unbalanced fitting weights when the order of magnitude is ampere (A), a nonlinear physical correction adjustment function was constructed to dynamically weight and compensate for the distance operator of multidimensional physical features.

[0043] In one specific implementation, when using a nonlinear physical correction adjustment function for dynamic weighted compensation, the nonlinear physical correction adjustment function is introduced when extracting the first-order derivative eigenvector in S102. right Dynamic weighted compensation is performed. After introducing a nonlinear physical correction adjustment function, the expression for the first-order derivative eigenvector is: ; ; ; ; in, This is a nonlinear physical correction and adjustment function, where x, y, and z represent the current or voltage values ​​of three adjacent data points, and x, y, and z should be the same type of physical quantity in the same call. As shown in the expression above... And others wait.

[0044] The specific expression for the nonlinear physical correction adjustment function is as follows: ; in, The Gaussian error function; To perform logarithmic operations, logarithmic domain compression of the current is used to enhance the operator's ability to capture minute current changes in the subthreshold region (low current region); the combination of the error function term and the curvature term enables automatic identification of regions where the curve undergoes abrupt physical changes. When the curve enters a region of drastic change in physical characteristics (such as the instant a device is turned on), Automatically increase weights to force the simulation model to prioritize the fitting accuracy of key physical nodes; in the calculation of the first derivative, , , Take the corresponding physical quantity values ​​(including voltage or current values) of three adjacent data points respectively.

[0045] S104. Call the asynchronous cumulative distance algorithm to find the globally optimal physical matching path based on the measured data sequence and the simulation data sequence, and calculate the cumulative distance value. Considering that measured and simulated sampling points often cannot completely overlap, this scheme adopts an asynchronous physical data alignment algorithm, which uses a cumulative distance recursive formula. Without interpolation approximation (to avoid introducing additional numerical errors), the globally optimal physical matching path is sought. Based on the multi-dimensional physical feature distance operator, the local distance values ​​of the starting data point pairs of the measured data sequence and the simulated data sequence are calculated to set the initial deviation of the cumulative path. Based on the initial deviation, dynamic weights are determined based on the index positions of each data point pair in its respective sequence, and weighted local distance values ​​are obtained. These weighted local distance values ​​are added to the historical cumulative cost with the smallest cumulative deviation in the preceding path to obtain the cumulative cost at the current matching data point pair. Recursive calculation is performed, and when the data endpoints of the measured data sequence and the simulated data sequence are reached, the cumulative cost at the endpoint is used as the global cumulative distance value.

[0046] In one specific implementation, the physical deviation of the starting point is first set to initialize the boundary. The specific expression is: ; ; ; When performing recursive calculations, the core expression for the recursion is as follows: when hour, ; Among them, weighting factors This is used to balance the path cost under different sampling densities and ensure the continuity of the global physical topography.

[0047] The above recursive process follows arrive The dynamic programming path expansion is used to solve for the final cumulative distance (Dynamic IV Distance, DIVD) of the corresponding globally optimal physical matching path: ; in, This represents the cumulative distance value.

[0048] S105. Determine whether the cumulative distance value is less than the preset cumulative distance threshold; A cumulative distance threshold is preset, and it is determined whether the cumulative distance value is less than the preset cumulative distance threshold. In one specific embodiment, the cumulative distance threshold can be... .

[0049] S106. If not, then find the physical working region with the largest distance deviation based on the multi-dimensional physical feature distance operator, and calculate the correction value of the semiconductor device simulation model parameters according to the regional characteristics corresponding to the physical working region. Based on the global cumulative distance matrix, the optimal matching path is backtracked, and the data point region with the largest point-to-point physical feature distance value corresponding to each matching point pair on the optimal path is extracted. This region is the physical working region with the largest deviation in the current iteration. Subsequently, the regional characteristics corresponding to this physical working region are analyzed: If the deviation is concentrated in the subthreshold region, then the simulation model is determined to be in the subthreshold swing (SS) or turn-on voltage ( There is a deviation in the model parameters, and the corresponding model parameter correction values ​​are for the interface state density. Adjust parameters such as threshold voltage; If the deviation is concentrated in the transition "bend" of the linear or saturated region, it is determined that the simulation model has deviations in effects such as carrier velocity saturation and mobility degradation. The corresponding model parameter correction value is for the mobility degradation coefficient. ), saturation velocity ( ) or channel length modulation coefficient ( Adjustments will be made accordingly. If the deviation is concentrated in the linear region (dominated by the deviation of the first derivative characteristic term), then the simulation model is determined to have a deviation in the linear response of the transconductance or output conductance. The corresponding model parameter correction values ​​are applied to the carrier mobility model parameters (such as low-field mobility). Adjustments will be made accordingly. The model parameter correction values ​​for each region are calculated proportionally based on the magnitude of the physical feature distance deviation, and are converted into parameter adjustment amounts recognized by the TCAD simulation software using a linear or nonlinear mapping method, and then used as correction values. Proceed to the next step with the output.

[0050] S107. Perform feedback adjustment on the semiconductor device simulation model based on the correction value, and repeatedly perform calibration on the semiconductor device simulation model; Correction values ​​of simulation model parameters calculated by S106 By writing control commands into the TCAD simulation software through the API interface or automated script, the corresponding physical model parameters in the current semiconductor device simulation model are updated, completing a parameter feedback adjustment.

[0051] After the parameters are updated, the TCAD simulation software is called again to generate a new simulation data sequence based on the updated simulation model parameters. And return to S101 with a new simulation data sequence. Replace the original simulation data sequence Repeat the calibration process from S101 to S105; thus forming a closed-loop automatic iterative calibration mechanism until the cumulative distance value meets the convergence condition.

[0052] Based on this feedback, the corresponding physical model parameters in the software are adjusted to enter the next iteration, thus achieving closed-loop automatic calibration.

[0053] S108. If so, the calibration is completed, and a semiconductor device simulation model is obtained.

[0054] When the cumulative distance value When the system determines that all physical parameters of the current semiconductor device simulation model accurately reflect the current-voltage characteristics of the actual device within the full bias range of the subthreshold region, linear region, and saturation region, the calibration process terminates. The final values ​​of the physical parameters of each simulation model are output, resulting in a semiconductor device simulation model that has been physically verified. This model can be directly used for subsequent device performance prediction, process parameter optimization, and Design Technology Co-optimization (DTCO) processes.

[0055] The device provided in this embodiment of the invention can perform data fitting based on the physical characteristics of semiconductor devices, realize high-precision automated calibration of simulation models, and enable the calibrated model to truly reflect the physical characteristics of real semiconductor devices. It realizes a fitting evaluation mechanism that can deeply couple semiconductor physical characteristics and take into account the full range of current changes and local morphological features, which greatly reduces simulation errors. In turn, it can meet the actual engineering needs of simulation models in semiconductor process development.

[0056] Specifically, compared to traditional fitting evaluation metrics, it has significant technical advantages in the fields of semiconductor device R&D and TCAD simulation: (1) High-fidelity alignment of physical characteristics across the entire operating range is achieved: This scheme effectively solves the fitting distortion problem caused by the semiconductor IV curve spanning multiple orders of magnitude by introducing a nonlinear physical correction adjustment function. Especially in the subthreshold region, this scheme can accurately capture minute current fluctuations, ensuring the turn-on voltage ( The high accuracy of extracting key physical parameters such as ) and subthreshold swing (SS) provides a reliable simulation basis for the design of low-power devices.

[0057] (2) Enhanced ability to capture inflection points of complex physical effects: By integrating second derivative (curvature) features into the operator, this scheme can keenly identify the physical morphology features of the device when it transitions from the linear region to the saturation region. This enables the simulation model to more accurately calibrate deep submicron physical effects such as carrier mobility degradation and velocity saturation, and solves the technical problem of fitting traditional indicators at the "bends" of curves.

[0058] (3) Elimination of physical information loss caused by data preprocessing: This scheme adopts an asynchronous data alignment algorithm, which directly adapts to the situation where the sampling distribution of measured data and simulation data is inconsistent, without the need for traditional interpolation approximation or manual experience scaling. This not only avoids numerical noise introduced by interpolation, but also preserves the original physical details of the IV curve, ensuring the objectivity of the calibration process and the authenticity of the data.

[0059] (4) Significantly improves the automation and efficiency of TCAD simulation calibration: Due to its strong physical sensitivity and versatility, this index can provide a clear and monotonic convergence guide for the automatic optimization algorithm. Experiments have shown that using this scheme as a feedback index can significantly improve the iterative calibration speed of TCAD physical parameters compared to the traditional manual weighted fitting method, and improve the accuracy of key physical parameters (such as turn-on voltage). The extraction error of (etc.) is stable within 1%, which greatly shortens the cycle of semiconductor process development (DTCO) and reduces R&D costs.

[0060] (5) Balancing global trends with local details: The DIVD index defined in this scheme not only focuses on the numerical closeness of each data point, but also focuses on the overall evolution trend of the curve through recursive cumulative logic. This characteristic of "balancing the whole and the part" enables the calibrated simulation model to maintain a high degree of physical consistency across the entire bias range, enhancing the model's predictive ability under extreme process corners.

[0061] The above describes the semiconductor device simulation model calibration method based on physical feature extraction in the embodiments of the present invention. The following describes the semiconductor device simulation model calibration device based on physical feature extraction in the embodiments of the present invention. (See reference...) Figure 2 One embodiment of the semiconductor device simulation model calibration device based on physical feature extraction in this invention includes: The data acquisition module 201 is used to acquire the measured data sequence and the simulation data sequence of the current-voltage characteristic curve of the semiconductor device; The feature extraction module 202 is used to calculate multiple physical feature vectors at each data point of the measured data sequence and the simulated data sequence, respectively; wherein, the physical feature vectors include at least numerical feature vectors, first derivative feature vectors, and curvature feature vectors; The first calculation module 203 is used to construct a multidimensional physical feature distance operator by weighting and combining each of the physical feature vectors, and to construct a nonlinear physical correction adjustment function to dynamically weight and compensate the multidimensional physical feature distance operator. The second calculation module 204 is used to call the asynchronous cumulative distance algorithm to find the globally optimal physical matching path based on the measured data sequence and the simulation data sequence, and to calculate the cumulative distance value. The calibration judgment module 205 is used to determine whether the cumulative distance value is less than a preset cumulative distance threshold. The calibration execution module 206 is used to find the physical working area with the largest distance deviation based on the multi-dimensional physical feature distance operator when the judgment result output by the calibration judgment module is negative, calculate the correction value of the semiconductor device simulation model parameters according to the regional characteristics corresponding to the physical working area, perform feedback adjustment on the semiconductor device simulation model based on the correction value, and repeatedly perform calibration on the semiconductor device simulation model; and to complete the calibration when the judgment result output by the calibration judgment module is positive, thereby obtaining the semiconductor device simulation model.

[0062] The device provided in this embodiment of the invention can perform data fitting based on the physical characteristics of semiconductor devices, realize high-precision calibration of simulation models, and enable the calibrated models to truly reflect the physical characteristics of real semiconductor devices, greatly reducing simulation errors.

[0063] In another embodiment of this application, the calculation expression of the multidimensional physical feature distance operator is: ; in, Represents a multidimensional physical feature distance operator. The weights of the numerical feature vectors, The weights of the first-order derivative eigenvectors are... The weights of the curvature eigenvector are... The index of the data points in the measured data sequence. The index of the data points in the simulation data sequence.

[0064] In another embodiment of this application, the numerical feature vector includes voltage and current values ​​at each data point; The first derivative eigenvector includes the first derivatives of the voltage and current values ​​at each data point along the curve direction. The curvature feature vector includes the curvature value at each data point, which is calculated by the first and second derivatives of the voltage and current values ​​at each data point along the curve direction.

[0065] In another embodiment of this application, the expression for the nonlinear physical correction adjustment function is: ; in, Represents the weight operator, , and These represent the current or voltage values ​​of three adjacent data points, respectively. This represents the error function.

[0066] In another embodiment of this application, the second calculation module 204 is specifically used for: The local distance between the starting data point pairs of the measured data sequence and the simulated data sequence is calculated based on the multidimensional physical feature distance operator to set the initial deviation of the cumulative path. Based on the initial deviation, dynamic weights are determined based on the index positions of each data point pair in their respective sequences. The weighted local distance value is obtained, and the weighted local distance value is added to the historical cumulative cost with the smallest cumulative deviation in the preceding path to obtain the cumulative cost at the current matching data point pair. Perform recursive calculations, and when the data endpoints of the measured data sequence and the simulated data sequence are reached, use the cumulative cost at the endpoint as the global cumulative distance value.

[0067] In another embodiment of this application, the physical working area includes: a subthreshold region, a linear region, a saturation region, and a bend between the linear region and the saturation region; The calibration execution module 206 is specifically used to determine the corresponding physical model parameters based on the physical working area with the largest distance deviation, and to calculate the correction value of the corresponding physical model parameters based on the multi-dimensional physical feature distance operator at each data point in the physical working area; And to write the correction value into the simulation software via API interface or script control instructions to trigger the next round of simulation iteration.

[0068] The device provided in this embodiment of the invention can perform data fitting based on the physical characteristics of semiconductor devices, realize high-precision automated calibration of simulation models, and enable the calibrated model to truly reflect the physical characteristics of real semiconductor devices. It realizes a fitting evaluation mechanism that can deeply couple semiconductor physical characteristics and take into account the full range of current changes and local morphological features, which greatly reduces simulation errors. In turn, it can meet the actual engineering needs of simulation models in semiconductor process development.

[0069] Based on the same inventive concept, this specification also provides an electronic device for calibrating a semiconductor device simulation model based on physical feature extraction. The electronic device for calibrating a semiconductor device simulation model based on physical feature extraction in this embodiment of the invention will be described in detail below from the perspective of hardware processing.

[0070] Figure 3 This is a schematic diagram of an electronic device provided as an embodiment of this specification. Refer to the following... Figure 3 The electronic device 300 according to this embodiment of the present invention will be described. Figure 3 The electronic device 300 shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0071] like Figure 3 As shown, the electronic device 300 is presented in the form of a general-purpose computing device. The components of the electronic device 300 may include, but are not limited to: at least one processing unit 310, at least one storage unit 320, a bus 330 connecting different system components (including storage unit 320 and processing unit 310), a display unit 340, etc.

[0072] The storage unit stores program code that can be executed by the processing unit 310, causing the processing unit 310 to perform the steps described in the processing method section of this specification according to various exemplary embodiments of the present invention. For example, the processing unit 310 can perform, for example... Figure 1 The steps are shown.

[0073] The storage unit 320 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 3201 and / or a cache storage unit 3202, and may further include a read-only memory unit (ROM) 3203.

[0074] The storage unit 320 may also include a program / utility 3204 having a set (at least one) program module 3205, such program module 3205 including but not limited to: an operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0075] Bus 330 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0076] Electronic device 300 can also communicate with one or more external devices 100 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 300, and / or with any device that enables electronic device 300 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 350. Furthermore, electronic device 300 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 360. Network adapter 360 can communicate with other modules of electronic device 300 via bus 330. It should be understood that, although... Figure 3As not shown, other hardware and / or software modules may be used in conjunction with electronic device 300, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0077] Through the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described in this invention can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this invention can be embodied in the form of a software product, which can be stored in a computer-readable storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, or network device, etc.) to execute the method described above according to this invention. When the computer program is executed by a data processing device, it enables the computer-readable medium to implement the method described above, i.e.: as... Figure 1 The method shown.

[0078] Figure 4 This is a schematic diagram of a computer-readable medium provided for embodiments of this specification.

[0079] accomplish Figure 1 The computer program of the method shown can be stored on one or more computer-readable media. A computer-readable medium can be a readable signal medium or a readable storage medium. A readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0080] The computer-readable storage medium may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0081] In addition, the present invention also provides a computer program product, including a computer program / instruction that, when executed by a processor, implements the semiconductor device simulation model calibration method based on physical feature extraction as described in any of the above embodiments.

[0082] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0083] In summary, the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that in practice, general-purpose data processing devices such as microprocessors or digital signal processors (DSPs) can be used to implement some or all of the functions of some or all of the components according to the embodiments of the present invention. The present invention can also be implemented as a device or apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such programs implementing the present invention can be stored on a computer-readable medium or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0084] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the present invention is not inherently related to any specific computer, virtual device, or electronic device, and various general-purpose devices can also implement the present invention. The above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0085] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0086] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for calibrating a semiconductor device simulation model based on physical feature extraction, characterized in that, include: Obtain the measured data sequence and simulation data sequence of the current-voltage characteristic curve of semiconductor devices; Calculate multiple physical feature vectors at each data point for the measured data sequence and the simulated data sequence, respectively; wherein, the physical feature vectors include at least numerical feature vectors, first derivative feature vectors, and curvature feature vectors; A multidimensional physical feature distance operator is constructed by weighting and combining the physical feature vectors, and a nonlinear physical correction adjustment function is constructed to dynamically weight and compensate the multidimensional physical feature distance operator. The asynchronous cumulative distance algorithm is invoked to find the globally optimal physical matching path based on the measured data sequence and the simulated data sequence, and the cumulative distance value is calculated. Determine whether the cumulative distance value is less than a preset cumulative distance threshold; If not, the physical working region with the largest distance deviation is found based on the multi-dimensional physical feature distance operator, and the correction value of the semiconductor device simulation model parameters is calculated according to the regional characteristics corresponding to the physical working region. Based on the correction value, feedback adjustment is performed on the semiconductor device simulation model, and the semiconductor device simulation model is calibrated repeatedly. If so, the calibration is completed, and a semiconductor device simulation model is obtained.

2. The semiconductor device simulation model calibration method based on physical feature extraction according to claim 1, characterized in that, The calculation expression for the multidimensional physical feature distance operator is as follows: ； in, Represents a multidimensional physical feature distance operator. Indicates voltage. Represents current. This represents the measured data. Represents simulation data, Indicates curvature. The weights of the numerical feature vectors, The weights of the first-order derivative eigenvectors are... The weights of the curvature eigenvector are... The index of the data points in the measured data sequence. The index of the data points in the simulation data sequence.

3. The semiconductor device simulation model calibration method based on physical feature extraction according to claim 2, characterized in that: The numerical feature vector includes the voltage and current values ​​at each data point; The first derivative eigenvector includes the first derivatives of the voltage and current values ​​at each data point along the curve direction. The curvature feature vector includes the curvature value at each data point, which is calculated by the first and second derivatives of the voltage and current values ​​at each data point along the curve direction.

4. The semiconductor device simulation model calibration method based on physical feature extraction according to claim 1, characterized in that, The expression for the nonlinear physical correction adjustment function is: ； in, Represents the weight operator, , and These represent the current or voltage values ​​of three adjacent data points, respectively. Represents the error function; The nonlinear physical correction adjustment function is used to dynamically weight and compensate the first-order derivative eigenvector contained in the multidimensional physical feature distance operator.

5. The semiconductor device simulation model calibration method based on physical feature extraction according to claim 1, characterized in that, The asynchronous cumulative distance algorithm is invoked to find the globally optimal physical matching path based on the measured data sequence and the simulated data sequence, and the cumulative distance value is calculated as follows: The local distance between the starting data point pairs of the measured data sequence and the simulated data sequence is calculated based on the multidimensional physical feature distance operator to set the initial deviation of the cumulative path. Based on the initial deviation, dynamic weights are determined based on the index positions of each data point pair in their respective sequences. The weighted local distance value is obtained, and the weighted local distance value is added to the historical cumulative cost with the smallest cumulative deviation in the preceding path to obtain the cumulative cost at the current matching data point pair. Perform recursive calculations, and when the data endpoints of the measured data sequence and the simulated data sequence are reached, use the cumulative cost at the endpoint as the global cumulative distance value.

6. The semiconductor device simulation model calibration method based on physical feature extraction according to claim 1, characterized in that, The physical working area includes: a subthreshold region, a linear region, a saturation region, and a bend between the linear region and the saturation region; The process of finding the physical working region with the largest distance deviation based on a multi-dimensional physical feature distance operator, and calculating the correction value of the semiconductor device simulation model parameters based on the regional characteristics corresponding to the physical working region, includes: The corresponding physical model parameters are determined based on the physical working area with the largest distance deviation, and the correction values ​​of the corresponding physical model parameters are calculated based on the multi-dimensional physical feature distance operator at each data point within the physical working area. The step of performing feedback adjustment on the semiconductor device simulation model based on the correction value and repeating the simulation includes: The corrected value is written into the simulation software via API interface or script control instructions to trigger the next round of simulation iteration.

7. A semiconductor device simulation model calibration device based on physical feature extraction, characterized in that, The semiconductor device simulation model calibration device based on physical feature extraction includes: The data acquisition module is used to acquire the measured data sequence and simulation data sequence of the current-voltage characteristic curve of semiconductor devices; The feature extraction module is used to calculate multiple physical feature vectors at each data point of the measured data sequence and the simulated data sequence, respectively; wherein, the physical feature vectors include at least numerical feature vectors, first derivative feature vectors, and curvature feature vectors; The first calculation module is used to construct a multidimensional physical feature distance operator by weighting and combining each of the physical feature vectors, and to construct a nonlinear physical correction adjustment function to dynamically weight and compensate the multidimensional physical feature distance operator. The second calculation module is used to call the asynchronous cumulative distance algorithm to find the globally optimal physical matching path based on the measured data sequence and the simulation data sequence, and to calculate the cumulative distance value. A calibration judgment module is used to determine whether the cumulative distance value is less than a preset cumulative distance threshold; The calibration execution module is used to find the physical working region with the largest distance deviation based on the multi-dimensional physical feature distance operator when the judgment result output by the calibration judgment module is negative, calculate the correction value of the semiconductor device simulation model parameters according to the regional characteristics corresponding to the physical working region, perform feedback adjustment on the semiconductor device simulation model based on the correction value, and repeatedly perform calibration on the semiconductor device simulation model; it is also used to complete the calibration when the judgment result output by the calibration judgment module is positive, and obtain the semiconductor device simulation model.

8. A semiconductor device simulation model calibration device based on physical feature extraction, characterized in that, The semiconductor device simulation model calibration device based on physical feature extraction includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the physical feature extraction-based semiconductor device simulation model calibration device to perform the steps of the physical feature extraction-based semiconductor device simulation model calibration method as described in any one of claims 1-6.

9. A computer-readable storage medium storing a computer program / instructions thereon, characterized in that, When the program / instruction is executed by the processor, it implements the steps of the semiconductor device simulation model calibration method based on physical feature extraction as described in any one of claims 1-6.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, the steps of the semiconductor device simulation model calibration method based on physical feature extraction as described in any one of claims 1-6 are implemented.

Images