Dual-gas response organic transistor sensor simulation design and gas concentration decoupling method thereof

Abstract

The application relates to a dual-gas response organic transistor sensor simulation design and a gas concentration decoupling method thereof, and belongs to the field of organic semiconductor device simulation and electronic technology. The method aims at the problem that existing gas monitoring technology cannot simultaneously consider low cost, room temperature operation and multi-component synchronous detection, and realizes the synchronous response and concentration accurate detection of a single organic transistor device to dual-component gas through the following steps: selection of gas-sensitive functional materials and their gas response key parameters, simulation design and electrical performance analysis of the dual-gas response organic transistor sensor, calculation of device electrical characteristic parameters and sensitivity, dual-gas signal data preprocessing and data set construction, dual-gas signal decoupling model construction and dual-gas concentration detection. The method improves the sensor integration and design efficiency, and realizes high-precision decoupling of mixed gas by using an algorithm, thereby providing technical support for the development of a sensing-storage-computing integrated intelligent gas sensing terminal related to gas monitoring of power system equipment.

Description

Technical Field

[0001] This invention belongs to the field of organic semiconductor device simulation and electronic technology, and relates to a simulation design of a dual-gas-response organic transistor sensor and its gas concentration decoupling method. Background Technology

[0002] With the rapid development and construction of smart grids, real-time monitoring and maintenance of the operating status of power system equipment is crucial. Under fault conditions, the insulating medium in power equipment decomposes to produce characteristic gases such as hydrogen and sulfur dioxide, the concentration of which is directly related to the type and severity of the fault. Therefore, real-time and accurate monitoring of these gas concentrations is of great significance for assessing the health status of equipment.

[0003] Existing gas monitoring technologies are mainly divided into two categories. One is spectroscopic detection technology, which offers high accuracy but requires expensive and bulky equipment. Furthermore, when detecting multi-component gas mixtures, it is prone to cross-interference due to overlapping spectral lines, making accurate gas identification difficult. The other category, detection technologies based on gas-sensitive functional materials, offers advantages such as low cost and small size. However, traditional metal-oxide-semiconductor (MOS) gas sensors typically operate in high-temperature environments, making them unsuitable for online room temperature monitoring requirements in power equipment.

[0004] Furthermore, traditional gas sensors are mostly designed for single gases. To simultaneously identify and distinguish multiple gases, complex sensor arrays are typically required. This approach not only significantly increases the hardware cost and integration complexity of the system, but its subsequent data processing also relies on a large amount of experimental data and tedious feature extraction engineering, resulting in high algorithm complexity and high computational consumption.

[0005] In summary, existing technologies, when applied to environments such as power systems, struggle to simultaneously achieve low cost, room temperature operation, high integration, and accurate resolution of multi-component gases. Currently, there is also a lack of research on using computer-aided design (TCAD) simulations to construct dual-gas response models for organic transistor sensors and integrating deep learning algorithms to decouple mixed signals. Therefore, providing a TCAD-based method for optimizing the design and signal decoupling of lightweight organic transistor sensors with dual-gas responses has become one of the urgent technical challenges to be addressed in this field. Summary of the Invention

[0006] In view of this, the purpose of this invention is to provide a simulation design for a dual-gas-response organic transistor sensor and a method for decoupling its gas concentration.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A simulation design of a dual-gas-responsive organic transistor sensor and its gas concentration decoupling method, comprising the following steps: S1. Selection of key parameters for gas-sensitive functional materials and their gas response; S2. Construct an organic field-effect transistor model in the computer-aided design technology TCAD software, import the key parameter values ​​selected in S1 as variables into the organic field-effect transistor model, and obtain the transfer characteristic curve of the device by applying the gate-source voltage and source-drain voltage, so as to realize the simulation design and electrical performance analysis of the dual-gas response organic transistor sensor. S3. Calculate the electrical characteristic parameters of the device based on the transfer characteristic curve obtained in S2, and calculate the sensitivity of the device to gas sensing based on the electrical characteristic parameters. S4. Perform data preprocessing on the transfer characteristic curves obtained in S2 and construct a dataset; S5. Construct a deep learning model and configure the input layer, intermediate layer and output layer of the deep learning model to match the dataset input constructed in S4 and the output of the two gas types. S6. Using the dataset constructed in S4, the deep learning model constructed in S5 is trained to establish a nonlinear mapping relationship between the response signal of the organic transistor sensor transfer characteristic curve and the concentration of each gas component, thereby achieving decoupled detection of the concentration of each component in the mixed gas.

[0008] Furthermore, S1 includes: screening gas-sensitive functional materials that are highly sensitive to the target gas, and determining key response parameters characterizing the gas-sensitive properties of the gas-sensitive functional materials; establishing a quantitative correspondence model between the target gas concentration and the key response parameters; and converting the quantitative correspondence model and material property parameters into an input parameter configuration file that can be recognized by TCAD simulation software.

[0009] Furthermore, the gas-sensitive functional material includes the gate material of an organic transistor or the channel layer material of an organic semiconductor; the key response parameters include the gate work function and the carrier mobility.

[0010] Furthermore, in S3, the electrical characteristic parameters of the device include threshold voltage, on-state current, off-state current, on-off current ratio, subthreshold slope, and mobility. The threshold voltage The extraction method is linear extrapolation; The on-state current is the drain current value corresponding to the preset maximum gate-source voltage; The off-state current is the drain current value corresponding to the preset minimum gate-source voltage; The switching current ratio is the ratio of the on-state current to the off-state current; The subthreshold slope is calculated as follows: within the subthreshold working interval of the transfer characteristic curve, according to the formula SS=d V GS / d(log10 (| I DS |))Calculation, where I DS Drain current, V GS Gate-source voltage; The mobility is calculated as follows: under saturated working region conditions, i.e., | V DS |greater than or equal to| V GS - V th |, according to the formula Calculation; where L The length of the channel. W The width of the channel. C i The capacitance of the gate dielectric layer per unit area.

[0011] Furthermore, in S3, the sensitivity S The calculation formula is: ,in, These are the measured values ​​of the variables after the target gas is adsorbed. The variable is a baseline value for a target gas environment, and the variable is any one of the device electrical characteristic parameters that can characterize changes in gas concentration.

[0012] Further, S4 includes: extracting the drain current value under a specific gate-source voltage from the transfer characteristic curve obtained in S2 to construct an input feature vector, and performing Min-Max normalization processing on the input feature vector and the output gas concentration variable; the formula for Min-Max normalization is: ,in, The value is the normalized value. x The original data, It is the minimum value in this feature dataset. This is the maximum value in the feature dataset.

[0013] Furthermore, in S5, the deep learning model is a residual multilayer perceptron model, which includes an input projection layer, multiple cascaded residual blocks, and an output regression layer. The input projection layer is used to receive the normalized drain current feature vector and map it to a high-dimensional latent space. Each residual block contains a fully connected layer, and there is a skip connection between the input and output of each residual block. The output regression layer is used to map the high-dimensional features output by the last residual block to a two-dimensional output space, and the two output nodes of the two-dimensional output space correspond to the first target gas concentration detection value and the second target gas concentration detection value, respectively.

[0014] Furthermore, after S3, there is also an optimization step for the non-gas-sensitive functional layer material: setting multiple candidate schemes for non-gas-sensitive functional layer materials, repeatedly executing S2 and S3 for each candidate scheme, calculating the sensing sensitivity of the device under each scheme, and selecting the non-gas-sensitive functional layer material corresponding to the highest sensing sensitivity as the optimal material.

[0015] Furthermore, in S6, the dataset constructed in S4 is used to train multiple different deep learning models constructed in S5, and the deep learning model with the best detection accuracy is selected as the final gas concentration decoupling model based on the mean absolute error and coefficient of determination on the validation set.

[0016] A dual-gas-responsive organic transistor sensor, constructed using the method described above, is applied to gas monitoring scenarios in artificial electronic nose design, sensing-storage-computing intelligent terminal development, and power system equipment maintenance.

[0017] The beneficial effects of this invention are as follows: (1) This invention constructs a device architecture with multiple gas-sensitive response mechanisms by introducing multiple gas-sensitive functional layers into the same organic transistor device. This design enables a single device to simultaneously capture and respond to bicomponent or multicomponent gases, thereby overcoming the limitations of traditional single-function devices. Compared with traditional solutions that rely on complex sensor arrays, this invention achieves effective detection of multicomponent gases while significantly reducing hardware integration complexity and manufacturing costs.

[0018] (2) This invention uses TCAD simulation technology to replace traditional trial-and-error experiments, and establishes a quantitative mapping model between the microstructure parameters of the device and the macroscopic electrical performance and sensing parameters. This method allows for rapid simulation and performance evaluation of gas sensors with different gas-sensitive materials and structures, providing a theoretical basis for the targeted design and optimization of the device, effectively avoiding the cumbersome process iteration, and significantly shortening the sensor development cycle.

[0019] (3) This invention utilizes a deep learning model to construct a highly robust nonlinear signal decoupling algorithm, which can effectively solve the coupling and cross-interference problems of dual-gas mixed signals. This model can automatically extract deep features directly from the transfer characteristic curve of the device without relying on complex and cumbersome manual feature engineering, thereby achieving accurate inversion and quantitative analysis of the concentration of each component in the mixed gas, thus improving the accuracy and intelligence level of gas detection in complex environments.

[0020] (4) The dual-gas-response organic transistor sensor proposed in this invention features simple structure, excellent performance, and low power consumption. This type of lightweight gas sensor is easy to manufacture, has low-cost manufacturing advantages, and is easy to integrate into arrays, making it suitable for advanced designs such as biomimetic olfactory chips. This not only significantly improves the processing speed of gas sensing data but also lays the foundation for the development of integrated sensing, storage, and computing intelligent terminals in the future, making it particularly suitable for gas monitoring scenarios such as power system equipment maintenance.

[0021] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0022] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating the simulation design of the dual-gas-response organic transistor sensor and its gas concentration decoupling method according to the present invention. Figure 2 This is a simulation diagram of the dual-gas-response organic transistor sensor based on a Pt gate / CuPc semiconductor channel layer and a PMMA dielectric layer in Embodiment 1 of the present invention. Figure 3 The transfer characteristic curves of H2 and SO2 in the simulation model of the dual-gas response organic transistor sensor based on PMMA dielectric layer in Embodiment 1 of the present invention are shown. Figure 4 The simulation model of the dual-gas response organic transistor sensor based on the PMMA dielectric layer in Embodiment 1 of the present invention shows the sensing sensitivity curves for H2 and SO2. Figure 5 This is a simulation diagram of the dual-gas response organic transistor sensor based on a Pt gate / CuPc semiconductor channel layer and a Si3N4 dielectric layer in Embodiment 2 of the present invention. Figure 6 The transfer characteristic curves of H2 and SO2 in the simulation model of the dual-gas response organic transistor sensor based on the Si3N4 dielectric layer in Embodiment 2 of the present invention are shown. Figure 7 The above are the sensing sensitivity curves of H2 and SO2 for the simulation model of the dual-gas response organic transistor sensor based on the Si3N4 dielectric layer in Embodiment 2 of the present invention. Figure 8 This is a structural diagram of the signal decoupling model based on residual multilayer perceptron in Embodiment 3 of the present invention.

[0023] Figure 9 This is a histogram comparing the evaluation metrics of different algorithm models in Embodiment 3 of the present invention. Detailed Implementation

[0024] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0025] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual pictures, and should not be construed as limiting the invention. To better illustrate the embodiments of the invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0026] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "front," and "rear" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0027] Example 1 This embodiment provides a simulation design for a dual-gas-response organic transistor sensor and its gas concentration decoupling method. The overall process is as follows: Figure 1 As shown, the method covers six core steps from material selection to gas decoupling detection. Specifically, taking SO2 and H2 sensing as examples, the method includes the following steps: Step 1: Selection of gas-sensitive functional materials and their key gas response parameters: To achieve dual-gas detection and facilitate decoupling, both the gate and the organic semiconductor channel layer were chosen as gas-sensitive sensing layers. Platinum exhibits catalytic adsorption of hydrogen, which leads to a decrease in the work function of platinum; therefore, platinum was chosen as the gate material to achieve hydrogen sensing. To directly expose the gate to the target gas environment, a top-gate-bottom-contact structure was adopted. However, the adsorption of SO2 by the platinum gate induces an increase in the work function. Based on the functional relationship between the target gas concentration and the work function, and considering the concentration levels of the two target gases in the actual application scenario, the combined effect of these two target gases on the platinum gate work function was approximated as a linear superposition model. By calculating the effects of each target gas on the platinum work function separately and then superimposing the two effects, the total change in the platinum gate work function in the mixed gas was finally obtained. The effects of different concentrations of the two target gases on the platinum work function are shown in Tables 1 and 2.

[0028] Table 1. Changes in platinum grid work function at different H2 concentrations

[0029] Table 2. Changes in platinum gate work function for different SO2 concentrations

[0030] On the other hand, H2 has no modulation effect on the organic semiconductor channel layer copper phthalocyanine (CuPc), so CuPc is only used for sensing SO2 in this sensor. SO2, as a polar molecule, has positively charged sulfur atoms that directly adsorb onto the carrier transport core region of the CuPc semiconductor / dielectric interface, compensating for shallow trap states at the interface. Therefore, the mobility of the organic semiconductor channel layer can be modified to accommodate different concentrations of SO2 gas in the environment: when the SO2 concentration increases from 0 ppm to 30 ppm, the hole mobility of the organic semiconductor channel layer increases from 0.0275 cm⁻¹. 2 / (V·s) increases linearly to 0.0386 cm 2 / (V·s).

[0031] Step 2: Simulation Design and Electrical Performance Analysis of Dual-Gas Response Organic Transistor Sensor In TCAD software, appropriate bandgap, electron affinity, relative permittivity, trap density, and electron and hole cross-sections are set according to the organic semiconductor channel layer used in the device. An organic field-effect transistor (OFET) is constructed, comprising a gate (G), dielectric layer, organic semiconductor channel layer, source (S), drain (D), and substrate. The dielectric layer of the device is made of polymethyl methacrylate (PMMA), and the specific simulation parameters are shown in Table 3. The device adopts a top-gate bottom-contact structure, where the platinum gate is placed on the top layer of the device, undertaking the H2 and SO2 sensing functions; the organic semiconductor channel layer is exposed to the detection environment to achieve a response to changes in SO2 concentration. The simulated device structure is shown below. Figure 2 As shown, the response parameter values ​​are imported as variables into the constructed organic field-effect transistor model to simulate the dual-gas sensing process. By applying appropriate gate-source voltage and drain current in the simulation, the transfer characteristic curves of the device are obtained. I DS - V GS .

[0032] Table 3 Simulation parameters of OFET based on PMMA dielectric layer

[0033] After completing the simulation model construction, the transfer characteristic curve of the device is obtained by applying appropriate gate-source voltage and source-drain voltage in the simulation. Specifically, the simulation parameters are configured as follows: gate-source voltage... V GS The scan range is set to 0 V to -10 V, and the source-drain voltage is... V DS The voltage was fixed at -15 V. Regarding the gate and organic semiconductor channel layer parameter settings, the hydrogen concentration gradient was set to 0 ppm, 26 ppm, 52 ppm, 111 ppm, and 257 ppm, with these concentration changes corresponding to a gradual decrease in the platinum gate work function. The sulfur dioxide concentration gradient was set to 0 ppm, 2 ppm, 5 ppm, 10 ppm, 20 ppm, and 30 ppm, and simulation was performed by simultaneously increasing the CuPc mobility and gate work function. After completing the above data settings, the controlled variable method was used to simulate different concentrations of mixed gases one by one, and the transfer characteristic curves under different mixed gas environments were collected. The transfer characteristic curves of the device under different gas concentrations are shown below. Figure 3 As shown. Figure 3 The figures show the transfer characteristic curves of H2 and SO2 for the simulation model of the dual-gas response organic transistor sensor based on the PMMA dielectric layer in Embodiment 1 of the present invention; where a is the transfer characteristic curve of the sensor under different SO2 concentrations, and b is the transfer characteristic curve of the sensor under different H2 concentrations.

[0034] Step 3: Calculation of device electrical characteristic parameters and sensitivity: Based on the transfer characteristic curves obtained from simulations under different mixed gas concentrations, the threshold voltage, on-state current, off-state current, on / off current ratio, subthreshold slope, and mobility of the organic transistor are extracted. The threshold voltage is extracted using a linear extrapolation method: within the linear operating region or saturation operating region of the transfer characteristic curve, the data points of the linear segment where the drain current changes with the gate-source voltage at the largest rate are selected for linear fitting to obtain a fitted straight line equation; this fitted straight line equation is extrapolated to the position where the drain current is zero, and the voltage value at the intersection of the fitted straight line and the gate-source voltage axis is the threshold voltage of the organic field-effect transistor. The method for obtaining the on-state current, off-state current, and on / off current ratio includes: in the transfer characteristic curve, the drain current value corresponding to a preset maximum gate-source voltage is determined as the on-state current, and the drain current value corresponding to a preset minimum gate-source voltage is determined as the off-state current; subsequently, the ratio of the on-state current to the off-state current is calculated, and this ratio is used as the on / off current ratio characterizing the electrical switching performance of the device. Subthreshold slope (SS) and mobility are also included. μ The calculation formula is as follows.

[0035]

[0036]

[0037] In the formula, I DS Drain current, V GS Gate-source voltage, L The length of the channel. W The width of the channel. C i It is the dielectric capacitance per unit area.

[0038] Sensitivity is a core indicator for evaluating the gas-sensitive performance of a sensor. In this embodiment, sensitivity is defined as shown in the following formula.

[0039]

[0040] In the formula, S The variable to be measured X Sensitivity, X gas These are the measured values ​​of the variables after the target gas is adsorbed. X ref These are the baseline values ​​for variables in a target gas environment.

[0041] This embodiment calculates the sensitivity indices of various electrical parameters of the device under different target gas concentrations, and the specific sensitivity curves are shown below. Figure 4 As shown. Figure 4 The above are the sensing sensitivity curves of H2 and SO2 for the simulation model of the dual-gas-response organic transistor sensor based on the PMMA dielectric layer in Embodiment 1 of the present invention; where a is the sensing sensitivity curve of the sensor for H2 and b is the sensing sensitivity curve of the sensor for SO2.

[0042] Step 4: Dual-gas signal data preprocessing and dataset construction: First, the gate-source voltage is extracted from the transfer characteristic curve obtained from the TCAD simulation. V GS = Drain current at 11 equally spaced sampling points within the range of 0 to -10 V I DS As the input feature vector, this feature vector fully encodes the electrical response information of the device under different gate voltage drives. Subsequently, the input features are normalized using the Min-Max method, linearly mapping each feature value to the [0, 1] interval to eliminate the adverse effects of differences in physical dimensions and numerical ranges on model training. Simultaneously, the output variables, namely H2 concentration and SO2 concentration, are also normalized. The specific normalization formulas are as follows.

[0043]

[0044] In the formula, x norm The value is the normalized value. x It is a value in the original data. x min It is the minimum value in this feature dataset. x max This is the maximum value in the feature dataset.

[0045] For data partitioning, considering the limited sample size of the simulation data, a stratified K-Fold Cross Validation strategy was adopted to construct the dataset. Specifically, K was set to 4, dividing all samples into four mutually exclusive subsets. While maintaining the consistency of H2 concentration distribution in each subset, three subsets were selected alternately as the training set for iterative optimization of model parameters, and the remaining subset was used as an independent validation set to evaluate the model's generalization performance. This process was repeated four times to ensure that each sample was used for both training and validation. Finally, the statistical average of the four validation metrics was used to comprehensively evaluate the model's generalization ability and robustness.

[0046] Step 5: Construction of the dual-gas signal decoupling model: A standard multilayer perceptron model was constructed to decouple the gases. This multilayer perceptron consists of an input layer, two hidden layers, and an output layer. The hidden layers have 128 and 64 neurons respectively, using the ReLU activation function, while the output layer uses a linear activation function. The model takes the raw signals collected by the gas sensor as input, calculates the decoupled concentrations of the two gases through forward propagation, and minimizes the mean square error between the detected values ​​and the true values ​​using a backpropagation algorithm.

[0047] Step 6: Dual gas concentration detection: Using the dual-gas coupled response simulation dataset constructed in step 4, the standard multilayer perceptron model constructed in step 5 was iteratively trained. The model parameters were updated using the Adam optimizer, with a learning rate of 0.001, a batch size of 36, and 3000 training epochs.

[0048] To comprehensively evaluate the effectiveness of the proposed standard multilayer perceptron model, this embodiment selects multiple evaluation metrics to assess the model, including the coefficient of determination. R 2 Mean Absolute Error (MAE). R 2 It is used to measure how well the detected value explains the variance of the true data; MAE intuitively represents the mean absolute deviation between the detected concentration and the true concentration.

[0049] After training, the standard multilayer perceptron model exhibited good decoupling performance for dual-gas signals on the test set. Experimental results show that for the SO2 concentration detection task, its coefficient of determination is [missing information]. R 2 The coefficient of determination reached 0.92, and the MAE was 1.99 ppm. For H2 concentration detection, the coefficient of determination... R 2 The value was 0.67, and the MAE value was 32.04 ppm.

[0050] Example 2 The difference between this embodiment and Embodiment 1 is that silicon nitride (Si3N4) is used as the dielectric layer of the dual-gas-response organic transistor sensor in step 2. Replacing the dielectric layer material will cause corresponding differences in the device's transfer characteristic curve, thereby altering its gas sensing sensitivity. Simultaneously, the change in transfer characteristics will directly lead to a change in the feature distribution of the constructed dataset, ultimately affecting the accuracy of gas concentration detection.

[0051] Step 1 in this embodiment is the same as in embodiment 1; Step 2: Simulation Design and Electrical Performance Analysis of Dual-Gas Response Organic Transistor Sensor In TCAD software, appropriate characteristic parameters such as bandgap, electron affinity, relative permittivity, trap density, and electron and hole cross-sections are set according to the organic semiconductor channel layer used in the device to construct an organic field-effect transistor including a gate, dielectric layer, organic semiconductor channel layer, source and drain, and substrate. The dielectric layer of the device is Si3N4, so its relative permittivity is set to 7.5. The remaining parameter settings are consistent with step 2 of Example 1, and the device simulation structure is as follows. Figure 5 As shown, the response parameters of the gas-sensitive component are imported as variables into the constructed organic field-effect transistor model to simulate the dual-gas sensing process. By applying appropriate gate-source voltage and drain current in the simulation, the transfer characteristic curves of the device are obtained. I DS - V GS ,like Figure 6 As shown. Figure 6 The figures show the transfer characteristic curves of H2 and SO2 for the simulation model of the dual-gas response organic transistor sensor based on the Si3N4 dielectric layer in Embodiment 2 of the present invention; where a is the transfer characteristic curve of the sensor under different SO2 concentrations, and b is the transfer characteristic curve of the sensor under different H2 concentrations.

[0052] Step 3: Calculation of device electrical characteristic parameters and sensitivity: Based on the transfer characteristic curves obtained from simulations under different mixed gas concentrations, the threshold voltage, on-state current, off-state current, on / off current ratio, subthreshold slope, and mobility of the organic transistor were extracted. The specific calculation method is consistent with step 3 in Example 1. This example calculated the sensitivity indices of various electrical parameters of the device under different target gas concentrations, and the specific sensitivity curves are shown below. Figure 7 As shown. Figure 7 The above are the sensing sensitivity curves of H2 and SO2 for the simulation model of the dual-gas response organic transistor sensor based on the Si3N4 dielectric layer in Embodiment 2 of the present invention; where a is the sensing sensitivity curve of the sensor for H2 and b is the sensing sensitivity curve of the sensor for SO2.

[0053] Steps 4-5 in this embodiment are the same as in embodiment 1; Step 6: Dual gas concentration detection: Using the dual-gas coupled response simulation dataset constructed in step 4, the standard multilayer perceptron model constructed in step 5 is iteratively trained. The model training parameter settings and evaluation criteria in this embodiment are consistent with those in embodiment 1.

[0054] After training, the standard multilayer perceptron model exhibited good decoupling performance for dual-gas signals on the test set. Experimental results show that for the SO2 concentration detection task, its coefficient of determination is [missing information]. R 2The coefficient of determination reached 0.83, and the MAE was 2.54 ppm. For H2 concentration detection, the coefficient of determination... R 2 The value was 0.68, and the MAE value was 32.13 ppm.

[0055] To verify the superiority of Si3N4 determined in this embodiment compared to other dielectric layer materials, the inventors compared it with the PMMA scheme in Example 1. While maintaining consistent device structural parameters and testing environment, the calculations in steps S2 and S3 revealed that the device in Example 1 (PMMA) exhibited the highest H2 sensing sensitivity of 281.98% and the highest SO2 sensing sensitivity of 442.65%, while the device in this embodiment (Si3N4) achieved an H2 sensing sensitivity of 1157.46% and an SO2 sensing sensitivity of 1056.81%. These comparative results demonstrate that the non-gas-sensitive layer material selected through the optimization method of this invention can significantly improve the device performance of dual-gas-response organic transistor sensors, thus confirming the effectiveness and practicality of this method in material selection.

[0056] Example 3 The difference between this embodiment and Embodiment 2 lies in the use of a different gas decoupling model in step 6. Specifically, this embodiment uses a Residual Multilayer Perceptron (ResMLP) to construct the detection model, and uses this model for training and decoupling detection of gas concentration.

[0057] Steps 1-4 in this embodiment are the same as in embodiment 2; Step 5: Construction of the dual-gas signal decoupling model: A ResMLP model was constructed to decouple the gas. The model consists of three main parts: an input projection layer, multiple cascaded residual blocks, and an output regression layer. The overall structure of the ResMLP model is as follows: Figure 8 As shown, the model mainly consists of the following three layers: Input projection layer: Receives the preprocessed 11-dimensional drain current feature vector as input, and maps the sparse low-dimensional current features to a 128-dimensional high-dimensional latent space through a fully connected projection layer containing batch normalization and ReLU activation functions, thereby enhancing the model's ability to express complex nonlinear patterns.

[0058] Residual Blocks: The main body of the model consists of three cascaded residual blocks. Each residual block employs two fully connected layers, with batch normalization and ReLU activation functions embedded between them. Furthermore, skip connections are introduced between the input and output of each block, directly superimposing the inputs of each block onto its output. This connection mechanism effectively alleviates the vanishing gradient problem in deep network training, prompting each layer of the network to focus on learning the higher-order nonlinear residual features in the sensor response, thereby significantly improving the model's decoupling accuracy for coupled signals.

[0059] Output regression layer: The high-dimensional features output by the last residual block are processed by a dimension-reduced fully connected hidden layer and mapped to a two-dimensional output space. The two output nodes correspond to the SO2 concentration detection value and the H2 concentration detection value, respectively, thereby realizing the synchronous quantitative inversion of the concentrations of the two gas components.

[0060] Step 6: Dual gas concentration detection: Using the dual-gas coupled response simulation dataset constructed in step 4, the ResMLP model constructed in step 5 was iteratively trained. The training process employed the Adam optimization algorithm combined with the mean squared error (MSE) loss function, with a batch size of 9, a maximum training epoch count of 3000, and an initial learning rate of 1×10⁻⁶. -3 And configure the weight decay coefficient to be 1×10. -4 To mitigate the risk of overfitting, a cosine annealing warm restart strategy is introduced to dynamically adjust the learning rate, where the initial period... T Set 0 to 50, period multiplication factor T mult Setting it to 2 helps the model escape local optima. Gradient clipping was also implemented during training, limiting the upper bound of the gradient norm to 1.0, effectively preventing gradient explosion and ensuring convergence stability.

[0061] To comprehensively evaluate the effectiveness of the proposed ResMLP algorithm, this embodiment selects multiple evaluation metrics to assess the model, including the coefficient of determination. R 2 Mean Absolute Error (MAE). R 2 It is used to measure how well the detected value explains the variance of the true data; MAE intuitively represents the mean absolute deviation between the detected concentration and the true concentration.

[0062] After training, the ResMLP model exhibited excellent dual-gas signal decoupling performance on the test set. Experimental results show that for the SO2 concentration detection task, the ResMLP's determination coefficient is [missing information]. R 2The concentration reached 0.99, and the MAE was as low as 0.86 ppm. For H2 concentration detection tasks, ResMLP's... R 2 The concentration reached 0.91, and the MAE was 15.56 ppm.

[0063] To verify the superiority of the preferred residual multilayer perceptron of this invention compared to other candidate models, the inventors conducted a comparative analysis of the residual multilayer perceptron, the standard multilayer perceptron used in Example 2, and other classic machine learning models, such as... Figure 9 As shown. While maintaining consistency in the training dataset, validation set, and training parameters (such as learning rate and number of iterations), the residual multilayer perceptron selected in Example 3 exhibits a lower mean absolute error and the highest coefficient of determination on the validation set. This confirms the effectiveness of the evaluation metrics proposed in this invention ( R The effectiveness of model selection using (² and MAE) can accurately identify the deep learning model with the best detection accuracy, thereby ensuring the accuracy of data processing for the dual-gas response sensor. It also demonstrates that optimizing the deep learning model can improve detection accuracy. As can be seen from the above comparison, although the method in Example 2 is feasible, the model architecture described in this example is a more preferred implementation method, capable of achieving the best technical effect expected in this application.

[0064] Example 4 This embodiment provides a simulation design of a dual-gas-response organic transistor sensor and its gas concentration decoupling method, which can be applied to gas monitoring scenarios in artificial electronic nose design, sensing-storage-computing intelligent terminal development, and power system equipment maintenance.

[0065] The described artificial electronic nose design is based on a sensor array composed of multiple dual-gas-response organic transistor sensors as core components. By collecting the response of each device in the array to the target gas, qualitative identification and concentration inversion of mixed gases can be achieved. This design effectively simulates the response mechanism of the biological olfactory system, and can construct an artificial electronic nose that is small in size, lightweight, low in power consumption, and capable of distinguishing complex gases.

[0066] The aforementioned sensing-storage-computing intelligent terminal integrates the artificial electronic nose as a sensing unit with a semiconductor-based storage and computing unit. This allows for the simultaneous sensing, storage, and weighted calculation of gas signals within the same system. The terminal outputs gas category and concentration results without analog-to-digital conversion or data transfer, significantly reducing system latency and energy consumption. It is suitable for developing low-power intelligent gas identification nodes at the edge.

[0067] The gas monitoring scenario for power system equipment maintenance involves deploying the sensing-storage-computing intelligent terminal inside key power equipment such as transformers and gas-insulated switchgear to monitor characteristic gases generated by faults such as overheating or discharge in real time. The sensing-storage-computing intelligent terminal directly outputs early warning signals indicating the fault type and severity, eliminating the need for external host computer processing and enabling on-site analysis and decision-making. Through a distributed network of multiple sensing-storage-computing intelligent terminals, an autonomous gas monitoring system for power equipment can be constructed, providing a reliable technical means for condition-based maintenance and early fault warning.

[0068] In summary, the present invention discloses a simulation design of a dual-gas-response organic transistor sensor and its gas concentration decoupling method, which can accurately realize the rapid design of dual-gas-response sensors and the algorithmic decoupling of complex gas components. This has positive significance for improving sensor integration and device design efficiency, and provides strong theoretical basis and technical support for power system equipment maintenance, the design of multi-gas-response sensing-storage-computing intelligent terminals in complex environments, and gas decoupling.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A simulation design for a dual-gas-response organic transistor sensor and its gas concentration decoupling method, characterized in that: Includes the following steps: S1. Selection of key parameters for gas-sensitive functional materials and their gas response; S2. Construct an organic field-effect transistor model in the computer-aided design technology TCAD software, import the key parameter values ​​selected in S1 as variables into the organic field-effect transistor model, and obtain the transfer characteristic curve of the device by applying the gate-source voltage and source-drain voltage, so as to realize the simulation design and electrical performance analysis of the dual-gas response organic transistor sensor. S3. Calculate the electrical characteristic parameters of the device based on the transfer characteristic curve obtained in S2, and calculate the sensitivity of the device to gas sensing based on the electrical characteristic parameters. S4. Perform data preprocessing on the transfer characteristic curves obtained in S2 and construct a dataset; S5. Construct a deep learning model and configure the input layer, intermediate layer and output layer of the deep learning model to match the dataset input constructed in S4 and the output of the two gas types. S6. Using the dataset constructed in S4, the deep learning model constructed in S5 is trained to establish a nonlinear mapping relationship between the response signal of the organic transistor sensor transfer characteristic curve and the concentration of each gas component, thereby achieving decoupled detection of the concentration of each component in the mixed gas.

2. The simulation design of the dual-gas-response organic transistor sensor and its gas concentration decoupling method according to claim 1, characterized in that: S1 includes: Screening for gas-sensitive functional materials that are highly sensitive to the target gas, and determining the key response parameters characterizing the gas-sensitive properties of the gas-sensitive functional materials; Establish a quantitative correspondence model between the target gas concentration and the key response parameters; convert the quantitative correspondence model and material property parameters into an input parameter configuration file that can be recognized by TCAD simulation software.

3. The simulation design of the dual-gas-response organic transistor sensor and its gas concentration decoupling method according to claim 2, characterized in that: The gas-sensitive functional material includes the gate material of an organic transistor or the channel layer material of an organic semiconductor; the key response parameters include the gate work function and the carrier mobility.

4. The simulation design of the dual-gas-response organic transistor sensor and its gas concentration decoupling method according to claim 1, characterized in that: In S3, the electrical characteristic parameters of the device include threshold voltage, on-state current, off-state current, on-off current ratio, subthreshold slope, and mobility. The threshold voltage The extraction method is linear extrapolation; The on-state current is the drain current value corresponding to the preset maximum gate-source voltage; The off-state current is the drain current value corresponding to the preset minimum gate-source voltage; The switching current ratio is the ratio of the on-state current to the off-state current; The subthreshold slope is calculated as follows: within the subthreshold working interval of the transfer characteristic curve, according to the formula SS=d V GS / d(log 10 (| I DS |))Calculation, where I DS Drain current, V GS Gate-source voltage; The mobility is calculated as follows: under saturated working region conditions, i.e., | V DS |greater than or equal to| V GS - V th |, according to the formula Calculation; where L The length of the channel. W The width of the channel. C i The capacitance of the gate dielectric layer per unit area.

5. The simulation design of the dual-gas-response organic transistor sensor and its gas concentration decoupling method according to claim 4, characterized in that: In S3, the sensitivity S The calculation formula is: ,in, These are the measured values ​​of the variables after the target gas is adsorbed. The variable is a baseline value for a target gas environment, and the variable is any one of the device electrical characteristic parameters that can characterize changes in gas concentration.

6. The simulation design of the dual-gas-response organic transistor sensor and its gas concentration decoupling method according to claim 1, characterized in that: S4 includes: extracting the drain current value under a specific gate-source voltage from the transfer characteristic curve obtained in S2 to construct an input feature vector, and performing Min-Max normalization on the input feature vector and the output gas concentration variable; the formula for Min-Max normalization is: ,in, The value is the normalized value. x The original data, It is the minimum value in this feature dataset. This is the maximum value in the feature dataset.

7. The simulation design of the dual-gas-response organic transistor sensor and its gas concentration decoupling method according to claim 1, characterized in that: In S5, the deep learning model is a residual multilayer perceptron model, which includes an input projection layer, multiple cascaded residual blocks, and an output regression layer. The input projection layer receives the normalized drain current feature vector and maps it to a high-dimensional latent space. Each residual block contains a fully connected layer, and there is a skip connection between the input and output of each residual block. The output regression layer maps the high-dimensional features output by the last residual block to a two-dimensional output space, and the two output nodes of the two-dimensional output space correspond to the first target gas concentration detection value and the second target gas concentration detection value, respectively.

8. The simulation design of the dual-gas-response organic transistor sensor and its gas concentration decoupling method according to claim 1, characterized in that: Following S3, an optimization step for the non-gas-sensitive functional layer material is also included: setting multiple candidate schemes for non-gas-sensitive functional layer materials, repeatedly executing S2 and S3 for each candidate scheme, calculating the sensing sensitivity of the device under each scheme, and selecting the non-gas-sensitive functional layer material corresponding to the highest sensing sensitivity as the optimal material.

9. The simulation design of the dual-gas-response organic transistor sensor and its gas concentration decoupling method according to claim 1, characterized in that: In step S6, the dataset constructed in step S4 is used to train multiple different deep learning models constructed in step S5, and the deep learning model with the best detection accuracy is selected as the final gas concentration decoupling model based on the mean absolute error and coefficient of determination on the validation set.

10. A dual-gas responsive organic transistor sensor, characterized in that: The sensor is constructed using the method described in any one of claims 1 to 9 and is applied to gas monitoring scenarios in the design of artificial electronic noses, the development of sensing-storage-computing intelligent terminals, and the maintenance of power system equipment.

Images