In this paper, we present an alternative approach to neuromorphic systems based on multi-level resistive memory (RRAM) synapses and deterministic learning rules. We demonstrate an original methodology to use conductive-bridge RAM (CBRAM) devices as, easy to program and low-power, binary synapses with stochastic learning rules. New circuit architecture, programming strategy and probabilistic STDP learning rule for two different CBRAM configurations 'with-selector (1T-1R)' and 'without-selector (1R)' are proposed. We show two methods (intrinsic and extrinsic) for implementing probabilistic STDP rules. Fully unsupervised learning with binary synapses is illustrated with the help of two example applications: (i) real-time auditory pattern extraction (inspired from a 64-channel silicon cochlea emulator) and (ii) visual pattern extraction (inspired from the processing inside visual cortex). High accuracy (audio pattern sensitivity>2, video detection rate>95%) and low synaptic-power dissipation (audio 0.55µW, video 74.2µW) are shown. The robustness and impact of synaptic parameter variability on system performance is also analyzed.

The missing link: Ferrocene and porphyrin monolayers are tethered on silicon surfaces with short (see picture, left) or long (right) linkers. Electron transfer to the silicon substrate is faster for monolayers with a short linker.Ferrocene and porphyrin derivatives are anchored on Si(100) surfaces through either a short two-carbon or a long 11-carbon linker. The two tether lengths are obtained by using two different grafting procedures: a single-step hydrosilylation is used for the short linker, whereas for the long linker a multistep process involving a 1,3-dipolar cycloaddition is conducted, which affords ferrocene-triazole-(CH(2))(11)-Si or Zn(porphyrin)-triazole-(CH(2))(11)-Si links to the surface. The modified surfaces are characterized by X-ray photoelectron spectroscopy and Fourier transform infrared spectroscopy. Cyclic voltammetry experiments show that the redox activity of the tethered ferrocene or porphyrin is maintained for both linker types. Microelectrode capacitor devices incorporating these modified Si(100) surfaces are designed, and their capacitance-voltage (C-V) and conductance-voltage (G-V) profiles are investigated. Capacitance and conductance peaks are observed, which indicates efficient charge transfer between the redox-active monolayers and the electrode surface. Slower electron transfer between the ferrocene or porphyrin monolayer and the electrode surface is observed for the longer linker, which suggests that by adjusting the linker length, the electrical properties of the device, such as charging and discharging kinetics and retention time, could be tuned.