Tyrosine-mediated analog resistive switching for artificial neural networks

Min Kyu Song, Seok Daniel Namgung, Hojung Lee, Jeong Hyun Yoon, Young Woong Song, Kang Hee Cho, Yoon Sik Lee, Jong Seok Lee, Ki Tae Nam, Jang Yeon Kwon

Research output: Contribution to journalArticlepeer-review

7 Citations (Scopus)


The fourth industrial revolution indispensably brings explosive data processing and storage; thus, a new computing paradigm based on artificial intelligence-enabling device structure is urgently required. Memristors have received considerable attention in this regard because of their ability to process and store data at the same location. However, fundamental problems with abrupt switching characteristics limit their practical application. To address this problem, we utilized the concept of metaplasticity inspired by biosystems and observed gradual switching in the peptide-based memristor at high proton conductivity. An unexpectedly high slope value > 1.7 in the log/-V curve at low voltage (≤ 400 mV) was considered the main origin, and it might arise from the modulatory response of proton ions on the threshold of Ag ion migration in the peptide film. With the obtained gradual switching property at high proton conductivity, the device showed significantly increased accuracy of image recognition (∼ 82.5%). We believe that such a demonstration not only contributes to the practical application of neuromorphic devices but also expands the bioinspired functional synthetic platform. [Figure not available: see fulltext.].

Original languageEnglish
Pages (from-to)858-864
Number of pages7
JournalNano Research
Issue number1
Publication statusPublished - 2023 Jan

Bibliographical note

Publisher Copyright:
© 2022, Tsinghua University Press.

All Science Journal Classification (ASJC) codes

  • Atomic and Molecular Physics, and Optics
  • General Materials Science
  • Condensed Matter Physics
  • Electrical and Electronic Engineering


Dive into the research topics of 'Tyrosine-mediated analog resistive switching for artificial neural networks'. Together they form a unique fingerprint.

Cite this