Electrochemical random-access memory: recent advances in materials, devices, and systems towards neuromorphic computing

Nano Convergence

Table 1 Comparison of Li ion-based ECRAM for in-memory computing

Mobile ions	Channel	Electrolyte	Reservoir	Effective size	Min. Write time (operation V)	Retention @RT	Max. G Range	On/off ratio	Linearity	Multilevel	Endur	Refs
Li	Li_1-xCoO₂	LiPON	Si	L: 2 μm	2 s (± 75 mV)	–	4.5–270 μS	60	–	200	10⁵	[34]
Li	α-MoO₃	LiClO₄/PEO	–	L: 10 μm	10 ms (± 2.5 V)	τ1: 110 ms τ2: 2624 ms	42–75 nS	1.8	–	50	4 × 10³	[46]
Li	WO₃	LiPON	–	L: 300 nm W: 300 nm	5 ns (± 1 mA)	–	0–24 nS	10³	0.347/−0.268	1000	10⁵	[54]
Li	Li_xTiO₂	LiClO₄/PEO	Li_xTiO₂	L: 10 μm W: 8 μm	10 ms (± 300 mV)	7 h	45–75 μS	1.7	–	250	> 10⁶	[53]
Li	WO₃	LiClO₄-PVA with graphene	–	L: 100 μm W: 10 μm	5 ms (± 3 V)	–	0.8–22 μS	27.5	0.96/−0.11	50	500	[37]
Li	LiCoO₂	Li₃PO_xSe_x	Si	L: 50 μm W: 20 μm	1 s (± 1.5 V)		2.1–40.6 nS	19.3	1.33/−0.34	90	720	[50]
Li	WO₃	Li₃PO₄	Si	L: 5 μm W: 5 μm	1 s (+ 3 V/−2.5 V)	–	0.5–3.5 μS	6.4	0.60/−0.58	30	420	[38]