Atomically Precise Ag<sub>11</sub> and Ag<sub>12</sub> Nanocluster-Assembled 2D Materials for Memristive and Neuromorphic Functionality.

Alam, Noohul; Rahaman, Towhidur; Besnard, Céline; Rosspeintner, Arnulf; Daku, Latevi Max Lawson; Khan, Saood; Ray, Soumya Jyoti; Bürgi, Thomas · Adv Mater · 2026

basic_science · Level V

Where this comes from

Abstract

Designing two-dimensional (2D) materials from atomically precise nanocluster building blocks is a challenging task, yet crucial for next-generation electronic devices. By tuning cluster nuclearity and ligand environment, well-defined assemblies can be designed, enabling control over optoelectronic properties at the atomic level. Herein, we report the bottom-up synthesis of two structurally distinct 2D nanocluster-assembled materials constructed from Ag<sub>12</sub> (2D Ag<sub>12</sub>-CAM) and Ag<sub>11</sub> (2D Ag<sub>11</sub>-CAM) nanocluster building blocks. The change in cluster nuclearity arises solely from ligand tuning; benzenethiol yields the Ag<sub>12</sub>-based framework, whereas 3,5-bis(trifluoromethyl)benzenethiol (BTFMBT) directs formation of the Ag<sub>11</sub> analogue. Lateral memristive devices fabricated from these materials (D-1: 2D Ag<sub>12</sub>-CAM; D-2: 2D Ag<sub>11</sub>-CAM) exhibit coexistence of negative differential resistance (NDR) and robust resistive switching with endurance stable over 4000 switching cycles. Device D-2, based on 2D Ag<sub>11</sub>-CAM, shows a markedly enhanced on-off (I<sub>on</sub>/I<sub>off</sub>) ratio (∼ 642) compared to device D-1 (∼ 200), and more pronounced NDR. Furthermore, the 2D Ag<sub>11</sub>-CAM (D-2)-based device was evaluated for neuromorphic functionality, mimicking the adaptive behavior of biological synapses. It exhibits analogue potentiation and depression under pulse stimulation, while experimentally extracted synaptic weight updates enable an artificial neural network to achieve a test accuracy of 84.2%, approaching that of a machine learning (ML)-trained network (93.2%).