The Study Of Ferroelectric Phase Transitions And Electrocaloric Effect Near The Morphotropic Phase Boundary

K_1-xNa_xNbO₃(KNN)-based lead-free piezoelectric ceramics have attracted widespread scientific interest due to comparable properties to that of PZT and restrictive use of lead and lead containing materials.However,there are few related researches on KNN ferroelectric thin films,so this paper fits the thermodynamic coefficients of KNN thin films and carries out related research work.In addition,(1-x)PMN-x PT materials have excellent physical properties,but there are few reports on different oriented(1-x)PMN-x PT films.Thus,this paper takes KNN films and(1-x)PMN-x PT films as examples,conducts detailed theoretical research on different oriented single-domain ferroelectric films based on LGD thermodynamic theory.The main contents and conclusions of this paper are as follows:The thermodynamic theoretical functions of(001),(110)and(111)oriented single-domain thin film are established by taking the coordinate transformation and taking into account the misfit strain and boundary conditions.The temperature-misfit strain phase diagrams and physical properties of different oriented ferroelectric films are calculated based on the established thermodynamic functions.The results show that the phase structures and physical properties vary greatly depending on the different strain symmetry.The(001)oriented thin films have strong in-plane and out-of-plane dielectric responses at the tetragonal-rhombohedral(TR)and orthorhombic-rhombohedral(OR)phase boundaries,respectively.Additionally,the(001)and(110)oriented films have better in-plane and out-of-plane dielectric and piezoelectric properties in the compressive strain and tensile strain ranges,respectively.The polarization components are selected as features,and three machine learning algorithms are used to predict and classify the phase diagrams of the KNN films.The comprehensive performance of DNNs is the best among the three algorithms,because DNNs is highly adaptive and adjustable,and the prediction accuracy is higher than that of k-NN and SVM.DNNs can play a greater advantage in the machine learning process of medium and large amounts of data.The k-NN is a simple,effective and easy to implement algorithm.However,the disadvantage of k-NN is that it needs to calculate the distance between the predicted samples and all the training samples,and this process is time-consuming.The SVM algorithm has the advantages of short running time and high efficiency,so it is more suitable for the machine learning process with small amount of data.These three classification algorithms have very good prediction accuracy rates after training,which reflects the feasibility and credibility of machine learning in the prediction and classification of ferroelectric thin film phase diagrams.The different oriented ferroelectric films are applied with[001]direction electric field,and the results show that the[001]direction electric field makes the phases whose polarization component P₃ is equal to 0 no longer stable in the original phase diagram.The electric field reduces the out-of-plane polarization,dielectric and piezoelectric properties,but does not have much effect on the in-plane related properties.In addition,the misfit strain u_m can significantly shift the peak values of electrocaloric effect.The(001)oriented KNbO₃ thin film has a temperature change of about 3 K near room temperature under electric field change?E=20 MV/m and tensile strain u_m=0.5%.The(111)oriented 0.3PMN-0.7PT has a temperature change of about 3.3 K near room temperature under electric field change?E=20 MV/m and tensile strain u_m=0%.In conclusion,it is possible to use electrocaloric effect to achieve room temperature cooling under reasonable electric fields and misfit strain.