Vanadium dioxide and ionic liquids

Before posting about the APS topics I'd mentioned, I wanted to write a bit about a recent paper that has received quite a bit of attention.  Stuart Parkin's group at IBM Research in Almaden reported in Science last week that they had used an ionic liquid to alter the properties of the strongly correlated oxide VO₂.  As I'd written previously, VO₂ has a metal-insulator transition at 67 C in the bulk, and there has been much interest in using some kind of electrical means to trigger the transition from the insulating to the metallic state in such correlated oxides.  If the insulating state results from Mott physics (i.e., the on-site electron-electron repulsion in the half-filled d band is so strong that the electrons cannot doubly occupy any of the transition metal sites, and the system acts like an insulator even though it would be a metal in the absence of the Coulomb repulsion), then electronically tuning the system away from half-filling (e.g., with a gate using the field effect) could switch the system into the metallic state.  Such a MottFET would be a switch operating on principles very different from those used in conventional Si transistors, and could in principle have superior operating characteristics (for the experts, things like subthreshold swing).   In the Parkin group's work, however, they report pretty compelling evidence that the switching in their system is driven by electrochemistry rather than simple electrostatic doping.  They argue that the metallic state that they produce results from chemical reactions between the VO₂ and the gate-biased ionic liquid that pull oxygen atoms out of the VO2 film.  The evidence for this is that a film treated this way recovers the insulating state when properly exposed to oxygen.  Moreover, they can expose the device to a specific isotope, ¹⁸O.  They can then sputter material off the film and use mass spectrometry to "weigh" the fragments (secondary ion mass spectrometry), and they find evidence that ¹⁸O makes it into the film to a depth of tens of nanometers (many unit cells). 

I find three things interesting about this paper.  First, the actual science is very nice, and I like the isotope tagging/SIMS quite a bit.  Second, I found the perspective put on this by IBM (and the resulting media coverage) a bit surprising - that using liquids and ion motion was a major advance because it would allow chips that operate more like the brain (history dependence = learning, + nonvolatile state retention).   I think that's a surprising spin to put on this.  That brings me to my third point, the true significance of the paper in (part of) the CM community:  This shows that you have to be very very careful when playing with ionic liquids to avoid electrochemistry!  There are previous papers out there that show very modest response of VO₂ in some forms to ionic liquid gating (here and here, for example), and a high profile Nature paper from last summer that reports a huge response.  The present work places these prior publications in an important context, calling into question the relative importance of electrostatics vs. electrochemistry.