CMOL and cousins: Hybrid CMOS/nano circuit FAQs

Q: What is CMOL? A: The basic idea of hybrid CMOS/nanoelectronic circuits is to complement the CMOS stack with a few-layer nanoelectronic addon (Fig. 1a) in the form of a nanowire crossbar (Fig. 1b). This idea may be traced back at least to the pioneering paper by J. Heath et al. [1]; however, the authors of that work and several following works in which this concept has been developed (see, e.g., reviews [2–4]) have assumed the use of relatively complex, three-terminal nanoelectronic devices whose integration is still well beyond reach. The current stage of the hybrid circuit idea development (started in 2003 [5, 6], but having evolved substantially until the late 2005 [3, 7, 8]) is focused on hybrid circuits which do not use any active nanoelectronic components beyond similar, simple (two-terminal), bistable devices (Fig. 1c) formed at each crosspoint simultaneously with the crossbar patterning. Q: What are the main options for crosspoint device implementation? Does the acronym “CMOL” imply using molecular devices? A: The answer to the latter question is NO. This (admittedly, misleading) term was coined in 2003, when molecular electronics seemed the only option for the implementation of crosspoint devices. By now, two-terminal crosspoint devices with the necessary “latching switch” functionality (Fig. 1c) have been demonstrated using a broad variety of materials and fabrication techniques see, e.g. Refs. [9, 10] for recent reviews. For most of them, the deviceto-device reproducibility (which is, of course, necessary for integration) has not yet been documented; however, there are notable exceptions. For example, I. G. Baek et al. [11] have demonstrated a few-percent reproducibility of the effective ON resistance of metaloxide-based devices, while A. Chen et al. [12] have reported a (still acceptable) ∼ 30% r.m.s. spread of ON currents in copper-oxidebased junctions. Even more promising, J. Billen et al. [13] have achieved ∼7% and ∼20% r.m.s. scattering of the, respectively, OFF→ON and ON→OFF switching thresholds in (relatively thick) Cu-TCNQ layers, whereas S. Jo and W. Lu [14] have reported a ∼10% spread of the OFF→ON switching voltage in amorphousSi-based devices. The apparent bistability mechanism in all these devices is reversible field-induced drift of cations in amorphous oxide matrix, leading to conducting filament formation and dissolution. Prelim-