Study On Graphene-based Thin Film Gas Sensor And Its Array

With the development of economy and technology,gas sensors are widely used to monitor toxic and harmful gases.However,a single gas sensor is often affected by cross sensitivity,so a sensor array consisting of multiple sensing units,and pattern recognition algorithms need to be developed for accurate qualitative identification or quantitative pre-diction of single gas or complex gas mixture.The traditional sensing materials and devices usually operate at high temperature,and cannot meet the requirement for room tempera-ture detection of toxic and harmful gases.A novel gas-sensitive material,graphene,has a large specific surface area and a high electron mobility at room temperature,which is favorable for the adsorption of gas molecules and the gas detection at room temperature.In this paper,various graphene nanocomposites were prepared,and sprayed on interdigi-tated electrodes（IDEs）to fabricate the corresponding gas sensors.The sensing properties of the single sensor and the as-constructed sensing array based on them were investigated.The main research contents are divided into three parts:1.The dynamic response model of graphene gas sensor was established,and a new method for calibrating gas concentration was proposed.The sensors were prepared with pure reduced graphene oxide（RGO）by air-brushing process.The effects of the film thick-ness and reduction degree of RGO were studied.It was found that increasing the film thickness and reduction degree result in the exponential decrease in the resistance of the sensor.This is because of the large specific surface area and abundant surface functional groups of RGO,therefore,the gas molecules are easy to be adsorbed and difficult to des-orb on the as-prepared RGO film,bring out an obvious baseline drift during the test.The experimental results indicated that the response speed and acceleration of the sensor in-creased with the increase of gas concentration.By further analysis of the first and second derivative curves of the response,it was found that the extreme values of the first and second derivatives were significantly correlated with the gas concentration.Based on the characteristics of first and second derivative,the dynamic response model of gas adsorp-tion process was established.According to the model,the first derivative extrema（FDE）and the square root of absolute value of the second derivative extrema are proportional to the gas concentration,which can be used to calibrate the gas concentration.This method is not affected by the baseline drift,and the gas concentration can be identified quickly due to the earlier occurrence of the derivative extrema.2.A room temperature chemoresistive NO₂sensor based on an RGO/SnO₂nanocom-posite film was fabricated via an air-brush spray deposition process.Therein,the aqueous suspensions of RGO/SnO₂composites acted as the spraying raw materials,which were prepared by mixing the diluted RGO dispersion and the SnO₂nanopowder,and then air-brushed on interdigitated electrodes to obtain the corresponding gas sensors.The char-acterizations of SEM,TEM,Raman,XRD and FTIR were used to characterize the mi-crostructures,morphologies and surface chemical compositions of the nanocomposites,indicating that the two materials coexist in the composite films and the concentration of surface defects is affected by the amount of SnO₂nanoparticles.It is also found that the room temperature sensing performance of RGO to NO₂can be improved by introducing appropriate amount of SnO₂nanoparticles.The enhanced NO₂sensing properties are at-tributed to the rough surface structure and increased surface area and surface defects of the nanocomposite films.Since further reduction of RGO,heat treating the sensing films resulted in a decrease in the response and recovery times of the sensors.Furthermore,the sensor annealed at 200^?C exhibited a small baseline drift,wide detection range,good linearity,high stability and better selectivity.Finally,we demonstrate the superiority of the First Derivative Extrema（FDE）and the square Root of Maxima of Second Derivative（RMSD）of the response curves in detecting NO₂.3.The gas sensor array based on RGO nanocomposite films was prepared for mixed gas detection,and the parameters of GA-BP algorithm were optimized by statistical method,which improved the accuracy and stability of pattern recognition.Four gas sensors,based on RGO,SnO₂and CuCl,were prepared to construct an array for detecting mixed gas（NO₂,NH₃）.The characteristic indexes were extracted according to the response of the array,and it was found that there was a non-linear relationship between the response and the gas concentration through principal component analysis（PCA）.Therefore,BP net-work was selected to identify the mixed gas,and the key parameters of BP network were optimized by statistical method.According to the box-plots of fitting error and prediction error made after a series of simulations,it was found that if the expected error of BP net-work was set as 10^-4and the initial weight and threshold of BP network were generated by uniformly distributing U（-3,3）,the mean square error of prediction can be reduced greatly.To avoid BP network falling into local optimum,genetic algorithm（GA）was employed to select the initial weights and thresholds globally,and then BP network was used to analyze the gas mixture.The results of 300 simulations showed that the accuracy of the optimized GA-BP algorithm for qualitative identification of mixed gases was 100%,and the relative error of quantitative prediction in the worst case was less than 30%.The algorithm has good stability.