Multi-gas recognition method based on carbon-based field effect transistor type gas sensor

Abstract

The application discloses a multi-gas identification method based on a carbon-based field effect transistor type gas sensor. First, a carbon-based field effect transistor type gas sensor is prepared, and then a transfer characteristic curve of the carbon-based field effect transistor type gas sensor in multiple gases is tested. Based on the transfer characteristic curve and a change mode, a characteristic variable representing the change mode of the transfer characteristic curve in each gas is screened. For different types of gases, the above gas sensor is used to detect and obtain the transfer characteristic curve, the corresponding characteristic variable is extracted, a second identification algorithm is used, and a gas detection model for detecting and identifying different gases is established. Finally, when in use, for a gas to be detected, the above carbon-based field effect transistor type gas sensor is used to detect and obtain the transfer characteristic curve, and the corresponding characteristic variable is extracted, and then the established gas detection model is used, so that the rapid detection and identification of multiple gases in a complex environment can be realized through the use of a single sensor.

Description

Technical Field

[0001] This invention relates to the field of gas sensor technology, and more specifically to a carbon-based field-effect transistor gas sensor and gas identification method for multi-gas identification. Background Technology

[0002] With the widespread application of gas sensors across various industries, the demand for sensors capable of identifying multiple gases is becoming increasingly urgent. This efficient detection method can significantly reduce the size and manufacturing cost of sensor devices. Currently, the main focus of research in the field of gas sensors is on improving sensitivity and extending the detection limit. However, it should be noted that there is still room for improvement in the selectivity of sensors, especially in the detection of multiple gases using only a single gas sensor.

[0003] For resistive gas sensors, which have received considerable attention and application in recent years, various methods, such as noble metal catalysis and surface modification, have been extensively explored to enhance their selectivity. However, these selectivity-enhancing methods are only beneficial for detecting a single target gas, and have limited effect on promoting the identification of multiple gases by a single gas sensor, and may even hinder the identification of multiple gases, greatly limiting the application of the sensor. The underlying reason for this bottleneck is that the electrical parameters currently used to evaluate gas-sensing performance are too simplistic. Typically, gas sensors rely on changes in a single parameter (the resistivity of the sensing material) to detect various information about the target gas. Besides changes in gas concentration, different gas types also cause changes in resistivity, making it difficult to establish a clear relationship between resistivity and a specific target gas.

[0004] Constructing an array of multiple sensors can be used to detect various gases, but it essentially involves mechanically stacking multiple individual gas sensors together. It still relies on individual sensors to specifically identify particular gases, achieving the identification and detection of different gases within the same sample (the gas to be detected). Furthermore, signals from each sensor in the array may interfere with each other, affecting the accuracy of gas identification.

[0005] Currently, achieving the detection of multiple gases using a single sensor remains a significant challenge in the field of gas sensor research. To achieve multi-target gas detection, multiple resistive gas sensors are typically used individually, or multiple sensors are configured into a sensor array. However, the mechanism by which resistive gas sensors detect gases relies on changes in resistivity to reflect information about the target gas. This results in complex design, large size, difficult fabrication processes, and increased costs for the gas sensor array. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-gas identification method based on a carbon-based field-effect transistor gas sensor. It aims to achieve the detection and identification of multiple gases in complex environments by using a single sensor, and solve the technical problems of large sensor requirements, high manufacturing difficulty, and large sensor component size caused by the reliance on sensor arrays for multi-gas identification in complex gas environments.

[0007] In some aspects of the present invention, a multi-gas identification method based on a carbon-based field-effect transistor gas sensor is provided. This method utilizes multiple electrical parameters of the carbon-based field-effect transistor for gas identification, effectively reducing the number of sensors required for identification and achieving the goal of classifying and identifying multiple target gases with a single sensor. This reduces the amount of sensitive materials modified by the sensor, lowers the manufacturing cost, and reduces the size of the equipment.

[0008] According to a first aspect of the present invention, a multi-gas identification method based on a carbon-based field-effect transistor gas sensor is proposed, comprising the following steps:

[0009] Step 1: Fabricate a carbon-based field-effect transistor gas sensor by using carbon nanotubes with a mesh-like thin film structure as the channel layer and depositing Pd nanoparticles of a certain thickness on the surface of the channel layer.

[0010] Step 2: Test the transfer characteristic curves of the carbon-based field-effect transistor gas sensor in various gases;

[0011] Step 3: Based on the transfer characteristic curve and the change pattern of the transfer characteristic curve, use the first identification algorithm to screen feature variables used to characterize the change pattern of the transfer characteristic curve in each gas;

[0012] Step 4: For different types of gases, the above-mentioned carbon-based field-effect transistor gas sensor is used to detect and obtain the transfer characteristic curve, extract the corresponding feature variables, and use the second recognition algorithm to establish a gas detection model for detecting and recognizing different gases.

[0013] Step 5: For the gas to be tested, the above-mentioned carbon-based field-effect transistor gas sensor is used to detect and obtain the transfer characteristic curve, and the corresponding feature variables are extracted. Then, the established gas detection model is used to identify different gases.

[0014] As an optional embodiment, the variation mode of the transfer characteristic curve includes at least one of translation, tilting, and rotation.

[0015] As an optional embodiment, the first identification algorithm includes at least one of independent component analysis, ANN, clustering algorithm, particle swarm classification, and SVM.

[0016] As an optional embodiment, the clustering algorithm includes one of k-means clustering and hierarchical clustering.

[0017] As an optional embodiment, the feature variables selected using the first recognition algorithm are:

[0018] Δg m ΔV th ΔI ds(-60V) ΔI ds(-40V) ΔI ds(-20V) ΔI ds(0V) ΔI ds(20V) and ΔI ds(40V)

[0019] Feature variable Δg m The transconductance change in the curve is used to characterize the change in carrier mobility of CNTs, and the calculation formula is Δg. m =g m(gas) -g m(air) , where g m(gas) Indicates g in the target gas m g m(air) G represents the amount of air. m ;

[0020] Feature variable ΔV th The threshold voltage change used to characterize the curve reflects the change in the number of charge carriers in CNTs, and is calculated using the formula ΔV. th =V th(gas) -V th(air) V th(gas) V in the target gas th V th(air) V represents the amount of air th ;

[0021] Feature variable ΔI ds The formula used to characterize the translation, tilt, and rotation patterns of a curve is ΔI. ds =I ds(gas) -I ds(air) , where I ds(gas) Indicates I in the target gas ds I ds(air) Indicates I in the air ds ,in:

[0022] ΔI ds(-60V) Gate-source voltage V gs I calculated at -60V ds Change;

[0023] ΔI ds(-40V) Gate-source voltage V gs I calculated at -40V ds Change;

[0024] ΔI ds(-20V) Gate-source voltage V gs I calculated at -20V ds Change;

[0025] ΔI ds(0V) Gate-source voltage V gs I calculated when = 0V ds Change;

[0026] ΔI ds(20V) Gate-source voltage V gs I calculated at 20V ds Change;

[0027] ΔI ds(40V) Gate-source voltage V gs I calculated at 40V ds Change.

[0028] As an optional embodiment, the second recognition algorithm is the PCA algorithm.

[0029] As an optional embodiment, the various gases are NO2, SO2, CO, H2S, H2 and NH3.

[0030] As an optional embodiment, the fabrication of the carbon-based field-effect transistor gas sensor includes:

[0031] Step 1-1: Deposit a network of carbon nanotube films on a Si-SiO2 substrate to obtain a CNT substrate;

[0032] Steps 1-2: Use laser direct writing to define the deposition areas of source and drain electrodes on the CNTs substrate, and then use electron beam deposition to deposit Ti / Pd / Au metal layers in the designated areas to form source and drain electrodes;

[0033] Steps 1-3: Define the channel region using laser direct writing, and then use reactive ion etching process to etch away the CNT thin film outside the channel region to obtain a carbon-based FET with a back gate structure.

[0034] Steps 1-4: Pd nanoparticles are deposited in the channel region by electron beam deposition to form a carbon-based FET gas sensor (Pd / CNTs-GS);

[0035] Steps 1-5: The prepared device is encapsulated onto a printed circuit board by wire bonding and polydimethylsiloxane curing.

[0036] As an optional embodiment, in steps 1-4, the deposition thickness is set to 1 nm.

[0037] As an optional embodiment, in the design of the carbon-based field-effect transistor gas sensor, the silicon substrate is a P-type silicon substrate with a resistivity of 0.001 to 0.005 Ω·cm and a thickness of 500 μm;

[0038] A dielectric layer of SiO2 with a thickness of 500 nm was prepared on a silicon-based substrate to form a Si-SiO2 substrate;

[0039] The thickness of the mesh carbon nanotube film, used as a channel material, is 1–2 nm.

[0040] The channel is designed to be 310μm long and 600μm wide.

[0041] Combining the multi-gas identification method based on carbon-based field-effect transistor gas sensors in the above embodiments, this method utilizes the strategy of gas identification based on multiple electrical parameters of field-effect transistors. By extracting eight feature variables from the sensor transfer characteristic curve and combining them with principal component analysis pattern recognition algorithms, a single sensor can classify and identify multiple target gases. This effectively reduces the number of sensors required for identification, the amount of sensitive materials modified by the sensor, and lowers the manufacturing cost and equipment size. It solves the technical problem of requiring a large number and size of sensors for multi-gas identification in complex gas environments.

[0042] It should be understood that all combinations of the foregoing concepts and the additional concepts described in more detail below may be considered part of the inventive subject matter of this disclosure, provided that such concepts do not contradict each other. Furthermore, all combinations of the claimed subject matter are considered part of the inventive subject matter of this disclosure.

[0043] The foregoing and other aspects, embodiments, and features of the teachings of the present invention will be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the invention, such as features and / or beneficial effects of exemplary embodiments, will become apparent from the following description or may be learned through practice of specific embodiments according to the teachings of the present invention. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating the fabrication process of a carbon-based field-effect transistor gas sensor (Pd / CNTs-GS) according to an embodiment of the present invention.

[0045] Figure 2 This is the structure of a carbon-based field-effect transistor gas sensor according to an embodiment of the present invention.

[0046] Figure 3This is a schematic diagram showing the variation of the transfer characteristic curves of a carbon-based field-effect transistor gas sensor (Pd / CNTs-GS) according to an embodiment of the present invention in NO2(a), NH3(b), H2(c), H2S(d), CO(e) and SO2(f).

[0047] Figure 4 This is a heat map plotted by extracting eight characteristic variables from multiple sets of experimental data using a carbon-based field-effect transistor gas sensor (Pd / CNTs-GS) according to an embodiment of the present invention.

[0048] Figure 5 This is a schematic diagram illustrating the three-dimensional identification results of six gases obtained by using eight feature variables as inputs for principal component analysis according to an embodiment of the present invention. Detailed Implementation

[0049] To better understand the technical content of the present invention, specific embodiments are described below in conjunction with the accompanying drawings.

[0050] Various aspects of the invention are described in this disclosure with reference to the accompanying drawings, in which numerous illustrative embodiments are shown. The embodiments of this disclosure are not necessarily intended to encompass all aspects of the invention. It should be understood that the various concepts and embodiments described above, as well as those described in more detail below, can be implemented in any of many ways, because the concepts and embodiments disclosed herein are not limited to any particular implementation. Furthermore, some aspects of the invention disclosed may be used alone or in any suitable combination with other aspects of the invention disclosed.

[0051] Multi-gas identification method based on carbon-based field-effect transistor gas sensor

[0052] The multi-gas identification method based on a carbon-based field-effect transistor gas sensor according to embodiments of the present invention aims to use multiple electrical parameters of a carbon-based field-effect transistor for gas identification, effectively reducing the number of sensors required for identification, achieving the purpose of classifying and identifying multiple target gases with a single sensor, reducing the amount of sensitive materials modified by the sensor, and lowering the manufacturing cost and equipment size.

[0053] As an optional embodiment, the multi-gas identification method based on a carbon-based field-effect transistor gas sensor includes the following steps:

[0054] Step 1: Fabricate a carbon-based field-effect transistor gas sensor by using carbon nanotubes with a mesh-like thin film structure as the channel layer and depositing Pd nanoparticles of a certain thickness on the surface of the channel layer.

[0055] Step 2: Test the transfer characteristic curves of the carbon-based field-effect transistor gas sensor in various gases;

[0056] Step 3: Based on the transfer characteristic curve and the change pattern of the transfer characteristic curve, use the first identification algorithm to screen feature variables used to characterize the change pattern of the transfer characteristic curve in each gas;

[0057] Step 4: For different types of gases, the above-mentioned carbon-based field-effect transistor gas sensor is used to detect and obtain the transfer characteristic curve, extract the corresponding feature variables, and use the second recognition algorithm to establish a gas detection model for detecting and recognizing different gases.

[0058] Step 5: For the gas to be tested, the above-mentioned carbon-based field-effect transistor gas sensor is used to detect and obtain the transfer characteristic curve, and the corresponding feature variables are extracted. Then, the established gas detection model is used to identify different gases.

[0059] As an optional implementation, the variation mode of the transfer characteristic curve includes at least one of translation, tilting, and rotation.

[0060] As an optional implementation, the first identification algorithm includes at least one of independent component analysis, ANN, clustering algorithm, particle swarm optimization, and SVM. The clustering algorithm includes one of k-means clustering and hierarchical clustering.

[0061] As an optional implementation, the feature variables selected using the first recognition algorithm are:

[0062] Δg m ΔV th ΔI ds(-60V) ΔI ds(-40V) ΔI ds(-20V) ΔI ds(0V) ΔI ds(20V) and ΔI ds(40V)

[0063] Feature variable Δg m The transconductance change in the curve is used to characterize the change in carrier mobility of CNTs, and the calculation formula is Δg. m =g m(gas) -g m(air) , where g m(gas) Indicates g in the target gas m g m(air) G represents the amount of air. m ;

[0064] Feature variable ΔV th The threshold voltage change used to characterize the curve reflects the change in the number of charge carriers in CNTs, and is calculated using the formula ΔV. th =V th(gas) -V th(air) V th(gas)V in the target gas th V th(air) V represents the amount of air th ;

[0065] Feature variable ΔI ds The formula used to characterize the translation, tilt, and rotation patterns of a curve is ΔI. ds =I ds(gas) -I ds(air) , where I ds(gas) Indicates I in the target gas ds I ds(air) Indicates I in the air ds ,in:

[0066] ΔI ds(-60V) Gate-source voltage V gs I calculated at -60V ds Change;

[0067] ΔI ds(-40V) Gate-source voltage V gs I calculated at -40V ds Change;

[0068] ΔI ds(-20V) Gate-source voltage V gs I calculated at -20V ds Change;

[0069] ΔI ds(0V) Gate-source voltage V gs I calculated when = 0V ds Change;

[0070] ΔI ds(20V) Gate-source voltage V gs I calculated at 20V ds Change;

[0071] ΔI ds(40V) Gate-source voltage V gs I calculated at 40V ds Change.

[0072] As an optional implementation, the second recognition algorithm is the PCA algorithm.

[0073] As an optional implementation, the various gases are NO2, SO2, CO, H2S, H2 and NH3.

[0074] As an optional implementation, the fabrication of the carbon-based field-effect transistor gas sensor includes:

[0075] Step 1-1: Deposit a network of carbon nanotube films on a Si-SiO2 substrate to obtain a CNT substrate;

[0076] Steps 1-2: Use laser direct writing to define the deposition areas of source and drain electrodes on the CNTs substrate, and then use electron beam deposition to deposit Ti / Pd / Au metal layers in the designated areas to form source and drain electrodes;

[0077] Steps 1-3: Define the channel region using laser direct writing, and then use reactive ion etching process to etch away the CNT thin film outside the channel region to obtain a carbon-based FET with a back gate structure.

[0078] Steps 1-4: Pd nanoparticles are deposited in the channel region by electron beam deposition to form a carbon-based FET gas sensor (Pd / CNTs-GS);

[0079] Steps 1-5: The prepared device is encapsulated onto a printed circuit board by wire bonding and polydimethylsiloxane curing.

[0080] Combining the methods of the above embodiments, the multi-gas identification method based on carbon-based field-effect transistor gas sensors includes the preparation of carbon-based field-effect transistor gas sensors Pd / CNTs-GS, the acquisition of transfer characteristic curves of Pd / CNTs-GS in different gases, the screening of characteristic variables, the establishment of a gas detection model based on the characteristic variables, and finally, for complex gases to be detected, the transfer characteristic curve is output by Pd / CNTs-GS, and after extracting the characteristic variables, the gas is identified using the gas detection model.

[0081] Example 1

[0082] Pd / CNTs-GS preparation

[0083] Combination Figure 1 The flowchart shown is a fabrication and packaging process for a carbon-based FET gas sensor (Pd / CNTs-GS).

[0084] The carbon-based FET gas sensor (Pd / CNTs-GS) is based on a carbon nanotube (CNTs) substrate on which a semiconducting CNTs mesh film is uniformly deposited, which facilitates the mass production of the device without the need for precise positioning of the photolithography area.

[0085] As an example, the fabrication of a carbon-based field-effect transistor (FET) type gas sensor includes:

[0086] In step I, laser direct writing is first used to define the deposition areas of the source and drain electrodes on the CNTs substrate (CNTs deposited on a Si-SiO2 substrate); then, electron beam deposition is used to deposit Ti / Pd / Au metal layers (0.6nm Ti, 20nm Pd, and 40nm Au) in the designated areas to form the source and drain electrodes, with deposition rates of 0.1, 1, and 1 for Ti, Pd, and Au, respectively. Ti has good wettability with CNTs and can be used as an adhesion layer between the source / drain electrodes and CNTs; Pd is a high work function metal, and the Schottky barrier formed when it contacts CNTs is beneficial for hole transport; Au can increase the thickness of the source / drain electrodes, which is beneficial for device packaging.

[0087] In step II, a channel region with a length of 310 μm and a width of 600 μm was first defined using laser direct writing. To prevent mutual interference between devices and affect their electrical performance, reactive ion etching (reactive gases O2 and CF4) was then used to etch away the CNTs thin film outside the channel region to obtain a carbon-based FET (CNTs-FET) with a back gate structure.

[0088] In step III, Pd nanoparticles are deposited in the channel region by electron beam deposition to form a carbon-based FET gas sensor (Pd / CNTs-GS), with the deposition thickness parameter set to 1 nm.

[0089] In step IV, to ensure the stability and reliability of the device, the prepared device is encapsulated onto a printed circuit board by wire bonding (WB) and polydimethylsiloxane (PDMS); a dust shield is placed on top of the device to prevent external dust from affecting the device.

[0090] like Figure 2 The diagram shows the fabricated carbon-based FET-type gas sensor (Pd / CNTs-GS). Using the methods described above, the resulting carbon-based field-effect transistor-type gas sensor is used for the classification and identification of multiple target gases. Figure 1 As shown, label 1 represents a P-type silicon substrate with a resistivity of 0.001–0.005 Ω·cm and a thickness of 500 μm; label 2 represents a SiO2 dielectric layer with a thickness of 500 nm; label 3 represents the source stage; label 4 represents the drain stage; label 5 represents the channel region, which uses a semiconductor-type carbon nanotube network film with a thickness of 1–2 nm, a channel length of 310 μm, and a width of 600 μm; label 6 represents Pd nanoparticles with a thickness of 1 nm deposited on the channel surface using electron beam evaporation deposition.

[0091] In embodiments of the present invention, in order to achieve the detection of multiple target gases, the prepared Pd / CNTs-GS needs to generate an effective response to various gases present in the environment. Therefore, the following technical solutions are used in the device design: CNTs are used as channel materials to achieve the detection of such gases through electronic interaction with oxidizing and reducing gases (e.g., NO2 and NH3); 1nm Pd nanoparticles are used to modify the channel, and Pd metal is used to form the source and drain electrodes to achieve the detection of such gases through catalytic reaction with H-containing gases (e.g., H2 and H2S); the work function of Pd nanoparticles and Pd metal in the source and drain electrodes is easily affected by gas molecules, thereby changing the height and width of the Schottky barrier formed with CNTs, thus achieving the detection of such gases.

[0092] To overcome the problems of large quantities and bulky sensitive materials required when using resistive gas sensors to construct sensor arrays for gas identification, the following technical solution is adopted for gas identification: First, the transfer characteristic curves of Pd / CNTs-GS in various gases are tested. Then, eight representative characteristic variables are extracted from them. Finally, principal component analysis (PCA) is combined to identify the gas.

[0093] Gas testing

[0094] The prepared Pd / CNTs-GS sensor was tested using NO2, SO2, CO, H2S, H2, and NH3 as examples.

[0095] Combination Figure 3 As shown, the transfer characteristic curves of the Pd / CNTs-GS sensor in six target gases (NO2, SO2, CO, H2S, H2 and NH3) were tested.

[0096] Figure (a) shows the transfer characteristic curve of Pd / CNTs-GS in NO2. The curve exhibits a rightward shift and tilt with increasing concentration. This rightward shift is attributed to the following reasons: when NO2 diffuses to the surface of CNTs in the Pd / CNTs-GS channel, electrons are transferred from CNTs to NO2, leading to an increase in the hole concentration in the CNTs. Furthermore, the change in the Schottky barrier between Pd nanoparticles and CNTs also affects the hole concentration in the CNTs. This is because when the diameter of Pd particles is at the nanometer scale, the work function of Pd nanoparticles is easily affected by gas molecules. The increase / decrease in the hole concentration in CNTs will cause I... dsThe changes in these characteristics are manifested as a rightward / leftward shift in the transfer characteristic curve. The tilt of the transfer characteristic curve can be explained by the fact that the space charge region formed between Pd nanoparticles and CNTs can act as scattering sites, affecting carrier mobility. Compared to the reducing gas NH3, the transfer characteristic curve of Pd / CNTs-GS changes in the opposite way to that of NO2. Specifically, the transfer characteristic curve in Figure (b) exhibits a combination of leftward shift and tilt due to the decrease in hole concentration and carrier mobility. Figures (c)-(d) depict the changes in the transfer characteristic curves of Pd / CNTs-GS in H2 and H2S, respectively, showing an overall rotational pattern. This rotation can be seen as a superposition of leftward shifts and rightward tilts caused by changes in carrier concentration and mobility.

[0097] Pd, as a catalytic metal, is distributed in the channel and source / drain electrodes of Pd / CNTs-GS. Its work function is easily affected by gases such as H2 and H2S. Changes in the work function of Pd nanoparticles in the channel affect the number and mobility of charge carriers, while changes in the work function of Pd in ​​the source / drain electrodes mainly cause changes in the Schottky barrier between the electrodes and CNTs, thereby altering the mobility of charge carriers. Figure 3 Figures (e)-(f) show the changes in the transfer characteristics of Pd / CNTs-GS in CO and SO2. The slope of the curves is closely related to the changes in carrier mobility. Although Pd / CNTs-GS responds similarly to the two gases, the degree of slope of the curves differs. Figure 3 The results in (a)-(f) show that the transfer characteristic curves of Pd / CNTs-GS in different gases exhibit specific variation patterns, which preliminarily demonstrates the feasibility of using a single Pd / CNTs-GS for multi-gas identification. This will provide a good foundation for subsequent strategies using multiple electrical parameters of Pd / CNTs-GS for gas identification.

[0098] Extraction of characteristic variables from transfer characteristic curves

[0099] In an embodiment of the present invention, eight feature variables (Δg) are extracted. m ΔV th ΔI ds(-60V) ΔI ds(-40V) ΔI ds(-20V) ΔI ds(0V) ΔI ds(20V) and ΔI ds(40V) This is used to characterize the variation pattern of the transfer characteristic curves in each gas.

[0100] It should be understood that, according to the gas reaction mechanism, the effect of gas on the transfer characteristic curve is expressed as g m and V thWhen determining characteristic variables, these two variables are chosen as the basis for analysis. However, when using only these two variables for multi-parameter identification, quantitative and non-overlapping gas differentiation cannot be achieved.

[0101] Feature variable Δg m The curve is used to characterize the transconductance change and to reflect the change in carrier mobility of CNTs.

[0102] The calculation formula is Δg m =g m(gas) -g m(air) , where g m(gas) Indicates g in the target gas m g m(air) G represents the amount of air. m .

[0103] Feature variable ΔV th The threshold voltage change used to characterize the curve reflects the change in the number of charge carriers in CNTs.

[0104] The calculation formula is ΔV th =V th(gas) -V th(air) V th(gas) V in the target gas th V th(air) V represents the amount of air th .

[0105] Feature variable ΔI ds Used to characterize the translation, tilting, and rotation patterns of a curve.

[0106] The calculation formula is ΔI ds =I ds(gas) -I ds(air) , where I ds(gas) Indicates I in the target gas ds I ds(air) Indicates I in the air ds .

[0107] ΔI ds(-60V) Gate-source voltage V gs I calculated at -60V ds Change;

[0108] ΔI ds(-40V) Gate-source voltage V gs I calculated at -40V ds Change;

[0109] ΔI ds(-20V) Gate-source voltage V gs I calculated at -20V ds Change;

[0110] ΔI ds(0V) Gate-source voltage V gs I calculated when = 0V ds Change;

[0111] ΔI ds(20V) Gate-source voltage V gs I calculated at 20V ds Change;

[0112] ΔI ds(40V) Gate-source voltage V gs I calculated at 40V ds Change.

[0113] In the embodiments of the present invention, heatmaps and hierarchical clustering analysis algorithms are used to verify and analyze eight feature variables, thereby achieving preliminary identification (qualitative identification and classification) of multiple gases and further verifying the feasibility of multi-gas identification based on these eight feature variables.

[0114] By compiling the calculation methods of the above feature variables into a batch file, it is beneficial to extract feature vectors in the subsequent process.

[0115] Will Figure 3 The transition characteristic curves shown are input into the Origin software, and feature variables are extracted in batches using a pre-written batch processing file.

[0116] exist Figure 4 In the figure, the heat map illustrates the selected characteristic variables (Δg) of Pd / CNTs-GS for various gases (1ppm NO2, 1ppm SO2, 1ppm CO, 1ppm H2S, 10ppm H2 and 200ppm NH3). m ΔV th ΔI ds(-60V) ΔI ds(-40V) ΔI ds(-20V) ΔI ds(0V) ΔI ds(20V) and ΔI ds(40V) The difference.

[0117] Different shades of color represent specific numerical values, making the differences between gases more intuitive. For each gas, five replicate experiments were performed using the same Pd / CNTs-GS, generating five sets of data (set-1, set-2, set-3, set-4, and set-5). To facilitate comparisons between gases in the heatmap, z-score normalization was applied to the data. Since different characteristic variables have different orders of magnitude, the z-score used can transform multiple sets of data into unitless data for comparison; its calculation formula is as follows:

[0118] Where X represents the original data, Z represents the z-score, μ represents the mean of the total sample space, and σ represents the standard deviation of the total sample space. As can be seen from the heatmap, the extracted feature variables can effectively represent… Figure 3 The variation patterns of transfer characteristic curves in various gases.

[0119] Multi-gas recognition

[0120] Combination Figure 5 We used eight characteristic variables combined with principal component analysis to classify and identify six gases. To visually observe the effectiveness of using eight characteristic variables for gas identification, principal component analysis was performed on the characteristic variable data. For example... Figure 5 The results show that, in a three-dimensional plane, the six gases are visually separated and do not overlap with each other. The cumulative variance contribution of PC1 and PC2 is as high as 97.9%, explaining the variation of each gas.

[0121] Detailed information on each eigenvalue, percentage of variance, and cumulative value in PCA is provided. To verify the recognition performance of Pd / CNTs-GS in unknown gases, we prepared 18 sets of experimental data as a test set (6 gases, 3 sets of data for each gas) for gas prediction, achieving a recognition accuracy of 94.4%. These results collectively demonstrate the effective differentiation of 6 gases in three-dimensional space using 8 feature variables extracted from the transfer characteristic curves.

[0122] Table 1. Detailed information on each PCA parameter.

[0123]

[0124] Complex Gas Detection

[0125] The above implementation process verifies the feasibility and accuracy of the proposed method for detecting and identifying multiple gases based on the above eight characteristic variables.

[0126] Based on this, we established a gas detection model by testing multiple gas samples. In subsequent actual environmental gas detection, we can directly use this model to output the detection results, thereby achieving rapid and accurate detection and identification of multiple gases.

[0127] The significant advantages of the device, in combination with the embodiments of the various aspects of the present invention described above, are as follows:

[0128] (1) The method of the present invention uses only one sensor to classify and identify multiple target gases, which overcomes the problems of multiple target gas detection requiring a large number of sensors, a large number of sensitive materials and a large volume. Moreover, the sensor fabrication process is compatible with CMOS processing technology, which is beneficial to the integration of the sensor with the back-end circuit.

[0129] (2) The method of the present invention breaks through the previous test mode of sensors. It reflects the detection information of the target gas by testing the transfer characteristic curve of FET type sensor in different gases. Multiple feature variables are extracted from the transfer characteristic curve through batch processing script. The feature variables are analyzed and screened by heat map and cluster analysis. The screened feature variables are combined with the pattern recognition algorithm PCA to perform the final fast classification and identification of multiple gases. This overcomes the shortcomings of the previous sensor data processing methods that are too simple and fail to extract effective information from the data.

[0130] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A multi-gas identification method based on a carbon-based field-effect transistor gas sensor, characterized in that, Includes the following steps: Step 1: Fabricate a carbon-based field-effect transistor gas sensor by using carbon nanotubes with a mesh-like thin film structure as the channel layer and depositing Pd nanoparticles of a certain thickness on the surface of the channel layer. Step 2: Test the transfer characteristic curves of the carbon-based field-effect transistor gas sensor in various gases; Step 3: Based on the transfer characteristic curve and the change pattern of the transfer characteristic curve, use the first identification algorithm to screen feature variables used to characterize the change pattern of the transfer characteristic curve in each gas. Step 4: For different types of gases, the above-mentioned carbon-based field-effect transistor gas sensor is used to detect and obtain the transfer characteristic curve, extract the corresponding feature variables, and use the second recognition algorithm to establish a gas detection model for detecting and recognizing different gases. Step 5: For the gas to be tested, the above-mentioned carbon-based field-effect transistor gas sensor is used to detect and obtain the transfer characteristic curve, and the corresponding feature variables are extracted. Then, the established gas detection model is used to identify different gases. Among them, the variation mode of the transfer characteristic curve includes at least one of translation, tilting, and rotation; The feature variables selected using the first recognition algorithm are: Δg m , ΔV th , ΔI ds(-60 V) , ΔI ds(-40 V) , ΔI ds(-20 V) , ΔI ds(0 V) , ΔI ds(20 V) and ΔI ds(40 V) Feature variable Δg m The transconductance change in the curve is used to characterize the change in carrier mobility of CNTs, and the calculation formula is Δg. m =g m (gas) -g m (air) , where g m (gas) Indicates g in the target gas m g m (air) G represents the amount of air. m ; Feature variable ΔV th The threshold voltage change used to characterize the curve reflects the change in the number of charge carriers in CNTs, and is calculated using the formula ΔV. th =V th (gas) - V th (air) V th (gas) V in the target gas th V th (air) V in the air th ; Feature variable ΔI ds The formula used to characterize the translation, tilt, and rotation patterns of a curve is ΔI. ds =I ds (gas) - I ds (air) , where I ds (gas) Indicates I in the target gas ds I ds (air) Indicates I in the air ds ,in: ΔI ds(-60 V) Gate-source voltage V gs I calculated at -60 V ds Change; ΔI ds(-40 V) Gate-source voltage V gs I calculated at -40 V ds Change; ΔI ds(-20 V) Gate-source voltage V gs I calculated at -20 V ds Change; ΔI ds(0 V) Gate-source voltage V gs I calculated when = 0 V ds Change; ΔI ds(20 V) Gate-source voltage V gs I calculated at 20 V ds Change; ΔI ds(40 V) Gate-source voltage V gs I calculated at 40 V ds Change.

2. The multi-gas identification method based on a carbon-based field-effect transistor gas sensor according to claim 1, characterized in that, The first identification algorithm includes at least one of independent component analysis, ANN, clustering algorithm, particle swarm classification, and SVM.

3. The multi-gas identification method based on a carbon-based field-effect transistor gas sensor according to claim 2, characterized in that, The clustering algorithm includes one of k-means clustering and hierarchical clustering.

4. The multi-gas identification method based on a carbon-based field-effect transistor gas sensor according to claim 1, characterized in that, The second recognition algorithm is the PCA algorithm.

5. The multi-gas identification method based on a carbon-based field-effect transistor gas sensor according to any one of claims 1-4, characterized in that, The various gases are NO2, SO2, CO, H2S, H2, and NH3.

6. The multi-gas identification method based on a carbon-based field-effect transistor gas sensor according to claim 1, characterized in that, The fabrication of the carbon-based field-effect transistor gas sensor includes: Step 1-1: Deposit a network of carbon nanotube films on a Si-SiO2 substrate to obtain a CNT substrate; Steps 1-2: Use laser direct writing to define the deposition areas of source and drain electrodes on the CNTs substrate, and then use electron beam deposition to deposit Ti / Pd / Au metal layers in the designated areas to form source and drain electrodes; Steps 1-3: Define the channel region using laser direct writing, and then use reactive ion etching process to etch away the CNT thin film outside the channel region to obtain a carbon-based FET with a back gate structure. Steps 1-4: Pd nanoparticles are deposited in the channel region by electron beam deposition to form a carbon-based FET gas sensor (Pd / CNTs-GS). Steps 1-5: The prepared device is encapsulated onto a printed circuit board by wire bonding and polydimethylsiloxane curing.

7. The multi-gas identification method based on a carbon-based field-effect transistor gas sensor according to claim 6, characterized in that, In steps 1-4, the deposition thickness is set to 1 nm.

8. The multi-gas identification method based on a carbon-based field-effect transistor gas sensor according to claim 1, characterized in that, In the design of the carbon-based field-effect transistor gas sensor, the silicon substrate is a P-type silicon substrate with a resistivity of 0.001 to 0.005 Ω·cm and a thickness of 500 μm; A SiO2 dielectric layer with a thickness of 500 nm was prepared on a silicon-based substrate to form a Si-SiO2 substrate; The thickness of the mesh carbon nanotube film, used as a channel material, is 1–2 nm. The channel is designed to be 310 μm long and 600 μm wide.

Images