Ferroelectric tunnel junction (FTJ) has attracted considerable attention for its potential applications in nonvolatile memory and neuromorphic computing. However, the experimental exploration of FTJs with high ON/OFF ratios is a challenging task due to the vast search space comprising of ferroelectric and electrode materials, fabrication methods and conditions and so on. Here, machine learning (ML) is demonstrated to be an effective tool to guide the experimental search of FTJs with high ON/OFF ratios. A dataset consisting of 152 FTJ samples with nine features and one target attribute (i.e., ON/OFF ratio) is established for ML modeling. Among various ML models, the gradient boosting classification model achieves the highest prediction accuracy. Combining the feature importance analysis based on this model with the association rule mining, it is extracted that the utilizations of {graphene/graphite (Gra) (top), LaNiO₃ (LNO) (bottom)} and {Gra (top), Ca0.96Ce0.04MnO₃ (CCMO) (bottom)} electrode pairs are likely to result in high ON/OFF ratios in FTJs. Moreover, two previously unexplored FTJs: Gra/BaTiO₃ (BTO)/LNO and Gra/BTO/CCMO, are predicted to achieve ON/OFF ratios higher than 1000. Guided by the ML predictions, the Gra/BTO/LNO and Gra/BTO/CCMO FTJs are experimentally fabricated, which unsurprisingly exhibit ≥1000 ON/OFF ratios (∼8540 and ∼7890, respectively). This study demonstrates a new paradigm of developing high-performance FTJs by using ML.Ferroelectric tunnel junction (FTJ) has attracted considerable attention for its potential applications in nonvolatile memory and neuromorphic computing. However, the experimental exploration of FTJs with high ON/OFF ratios is a challenging task due to the vast search space comprising of ferroelectric and electrode materials, fabrication methods and conditions and so on. Here, machine learning (ML) is demonstrated to be an effective tool to guide the experimental search of FTJs with high ON/OFF ratios. A dataset consisting of 152 FTJ samples with nine features and one target attribute (i.e., ON/OFF ratio) is established for ML modeling. Among various ML models, the gradient boosting classification model achieves the highest prediction accuracy. Combining the feature importance analysis based on this model with the association rule mining, it is extracted that the utilizations of {graphene/graphite (Gra) (top), LaNiO₃ (LNO) (bottom)} and {Gra (top), Ca0.96Ce0.04MnO₃ (CCMO) (bottom)} electrode pairs are likely to result in high ON/OFF ratios in FTJs. Moreover, two previously unexplored FTJs: Gra/BaTiO₃ (BTO)/LNO and Gra/BTO/CCMO, are predicted to achieve ON/OFF ratios higher than 1000. Guided by the ML predictions, the Gra/BTO/LNO and Gra/BTO/CCMO FTJs are experimentally fabricated, which unsurprisingly exhibit ≥1000 ON/OFF ratios (∼8540 and ∼7890, respectively). This study demonstrates a new paradigm of developing high-performance FTJs by using ML.

为便于工业应用，设计了具有人机交互界面的步进电机控制器。该控制器具有独立的编程指令。用户可通过PC机运行的程序编译软件完成应用程序的编写、自定义图像的选取及寄存器参数的设置。程序编译软件将上述信息通过串行总线传送给步进电机控制器，并保存在非易失性存储器中。步进电机控制器读取、解译并执行应用程序。该系统中配置有液晶显示器，实时显示操作信息；同时配有键盘，便于用户在工业现场修改应用程序或寄存器参数。整个系统以片上外设丰富的Cortex-M3核ARM芯片为核心，对系统指令、人机交互键盘、非易失性存储及LCD显示模块的设计进行了详细分析，并给出了系统设计显示结果。结果显示：LCD显示信息与用户按键命令同步，非易失性模块数据存储可靠，系统运行结果良好。

为了提高晶体的非易失存储性能，采用双波长存储技术实验研究了LiNbO3∶In∶Fe∶Cu晶体中的非易失存储，折射率调制度为1.04×10－4，记录灵敏度达到0.965 cm/J.与传统的双色非易失记录相比，该方法大幅度地提高了光栅的强度和记录的灵敏度.利用Kukhtarev带输运模型对双波长非易失记录过程中光栅的动态演化过程进行了数值模拟，同时讨论了氧化还原程度对双波长非易失全息记录的影响.理论与实验符合的较好.

为了研究近化学计量比Ce:Fe:LiNbO3晶体的非挥发全息存储固定,测试了晶体的光谱特性,发现近化学计量比Ce:Fe:LiNbO3晶体较同成分Ce:Fe:LiNbO3晶体的红外透射谱变窄,峰值位于3466cm-1处;而紫外光谱发生紫移。并采用单、双光子存储方法进行了理论分析和实验验证。结果表明,单光子照射实验中,用汞灯预照晶体比不用汞灯预照的衍射效率要高;而双光子存储的性能参量明显优于单光子存储的性能参量。

为了对在LiNbO3:Ce:Cu晶体中绿光作为记录光的非挥发全息记录进行优化, 联立求解了双中心物质方程和双光束耦合波方程, 数值分析了平均空间电荷场(SCF)和衍射效率随晶体的氧化还原态、记录光与敏化光的光强比以及深浅中心的掺杂浓度的变化。结果表明, 采用绿光作为记录光在LiNbO3:Ce:Cu晶体中进行非挥发全息记录, 可以记录得到强光折变光栅, 其空间电荷场高达107 V/m;获得高达80%以上的固定衍射效率, 各相关参量都有较大的优化空间。

根据双中心带输运模型,对(Ce,Cu)∶LiNbO3晶体双中心非挥发全息记录进行了理论研究与优化。推导了(Ce,Cu)∶LiNbO3晶体的微观参量,采用数值方法通过严格求解模拟双中心带输运方程来模拟全息记录过程。分析了记录过程中,记录与敏化光强、Ce和Cu掺杂浓度以及晶体微观参量对(Ce,Cu)∶LiNbO3晶体双中心全息记录的影响。发现(Ce,Cu)∶LiNbO3晶体非挥发全息记录中实现高衍射效率与固定效率的主导因素是深中心Cu,在记录过程中,深中心Cu建立起了很强的空间电荷场。数值模拟的结果经过实验验证,最高饱和与固定衍射效率别为60.5%和53.8%。

采用双中心记录方案在双掺杂LiNbO3∶Fe∶Rh晶体中实现了近红外非挥发全息记录,研究了LiNbO3∶Fe∶Rh晶体在633 nm,752 nm,799 nm波长下的全息记录性能。结果表明,在使用近红外记录光时,其记录灵敏度随敏化光强的变化趋势与双中心短波长记录时的不同。通过和LiNbO3∶Fe∶Mn等传统双掺杂铌酸锂晶体的近红外波段记录效果对比,发现同时掺杂Fe和Rh可增强晶体对近红外光的吸收,获得更高的浅中心Fe光生伏特系数,从而能够在LiNbO3∶Fe∶Rh晶体中实现近红外波段的光折变全息记录。

研究了LiNbO₃∶Cr∶Cu晶体的吸收特性，发现LiNbO₃∶Cr∶Cu(含0．14wt．%Cr₂O₃和0．011wt．%CuO)晶体存在两个明显的吸收峰，中心波长分别位于480nm和660nm；随着Cr的含量逐渐减小，Cu的含量逐渐增大，短波段不存在明显吸收峰，掺Cr的含量越大，中心波长在660nm处的吸收越大；633nm红光虽然位于中心波长为660nm的吸收峰内，但它无助于光折变过程．分别采用390nm紫外光和488nm蓝光作为敏化光，514nm绿光作为记录光的记录方案，实现了非挥发全息记录，掺入适量的Cr(比如N_Cr=2．795×10²⁵m^－3，N_Cr／N_Cu=1)有助于全息记录性能的提高．

分别采用514 nm绿光、488 nm蓝光和390 nm紫外光作为敏化光,633 nm红光作为记录光,详细研究了敏化光波长对氧化(Fe,Ni):LiNbO3晶体全息记录性能的影响。结果表明:随着敏化光波长的逐渐减小,氧化(Fe,Ni):LiNbO3晶体的非挥发全息记录性能逐渐优化,390 nm紫外光是这三种敏化光中最优的敏化光。考虑敏化光的吸收,为了在双中心全息记录中获得最优的性能,应当选择合适波长的敏化光:一方面短波长敏化光能有效地敏化深中心;另一方面短波长敏化光的吸收太强(如对光折变效应无用的基质吸收),不能沿厚度方向有效地敏化晶体,所以实际上需折衷考虑,并从理论上给予了解释。