Realizing Performance Balance by Band Offset and Defect Concentration Engineering in HfO<i> _x </i>/ZrO<i> _y </i> Self-Rectifying Memristor

High-performance self-rectifying memristor (SRM) is an ideal candidate for in-memory computing (IMC). On-off ratio, nonlinearity (NL), and on-current of SRM are the key metrics for IMC hardware implementation. Here, we propose an SRM optimization method with a well-defined physical mechanism. As a demonstration, we fabricate a Ta/HfO<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>x</i></sub>/ZrO<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>y</i></sub>/Pt with well-balanced performance including a large on/off ratio (>10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sup>), NL (>7×10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup>), rectification ratio (>8×10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sup>), and high on-current (100 nA) by band offset and defect concentration engineering. The switching energies of the Set and Reset processes are calculated as 7.02 fJ and 1.24 aJ. High NL and on-current are achieved by utilizing electron conduction in the defective HfO<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>x</i></sub> resistive switching layer, which is difficult to satisfy Poole-Frenkel (P-F) emission at small voltage bias and enables the P-F emission at large voltage bias. A large on/off ratio is attributed to the Pt/ZrO<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>y</i></sub> interfacial barrier and numerous defects in HfO<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>x</i></sub> layer. Our work provides a promising device prototype for constructing IMC hardware platform.