An associated article published in Nature Nanotechnology can be found here: https://www.nature.com/article...
We are currently at an interesting juncture where the size of the digital universe is growing rapidly, far beyond the reliable handling ability of our current computing technologies. This necessitates invention of new, disruptive technologies through a culmination of fundamental science and application-oriented research. Over the last 50 years, starting with the advent of Si-electronics, there has been a single mantra to make a better computer: device miniaturization following Moore’s law. Today, Moore’s Law is either dead or dying, depending on who you ask. The effective limit of physical miniaturization of devices has already been reached and there is no longer ‘plenty of room at the bottom’. As hype about the imminent arrival of the fourth industrial revolution reaches fever pitch, the often-overlooked question is, do we really have the technology to implement at-scale AI?
On one hand this can be seen as an existential threat to conventional computing with far-reaching implications throughout all fields that computation touches today – from research and technology all the way to finance. On the other hand, though, this end of one era represents a window of radical opportunity. One that ushers in a new era of creativity where a confluence of new physics, chemistry, material science, predictive modelling and electrical engineering is fuelling the rational design and development of a new genre of devices that address and resolve fundamental scientific bottlenecks. Our research, reported in Nature Nanotechnology, is an effort in that direction .
The hero of our story is a molecular complex [RuIIL2] (PF6)2 where L=2,6-bis(phenylazo)pyridine. The ligand system, L, used here is a pincer ligand that Goswami’s group at Indian Association for the Cultivation of Science (IACS, Kolkata) discovered in 2014 . The design was inspired by their previously developed 2(phenylazo)pyridine (L0) ligand systems that turned out to be quite useful both as an electro-protic storage and a memory device. Our original idea was, in fact, very different. In a previous publication, we fabricated a binary memristor with a [RuII(L0)3] (PF6)2 complex comprising of three L0 ligand each containing one azo moiety . Based on in-situ spectroscopic measurements we concluded that the redox combination of the three ligands is the key to our observed switching. That is what motivated us to opt for the new molecular system [RuIIL2] (PF6)2 to explore the effect of an increased number of azo groups. However, during our attempt to understand the molecular states via in-situ spectroscopies, the off-state pushed us to our wit’s end. For months we had no clue about what we were looking at. We had nearly started to believe that we would never be able to explain our spectral measurements and that’s when a ‘eureka moment’ occurred. It was a casual evening discussion over tea where we suddenly realized that we had been making an assumption all along without questioning it. Instead of a pure redox state driving the off-state characteristics, what if it was in fact a mixed redox state? We jumped at the possibility and pulled out all stops to explore this path. All it took was a couple of sleepless nights and we arrived at the conclusion that we are now able to break the electronic symmetry in our film at room temperature just by applying voltage. A battery of control experiments, simulations and mathematical modelling all confirmed and strengthened this conclusion.
But what’s so great about symmetry breaking, anyway? From elementary physics we understand that every material is made of identical building block units known as molecules. This implies that if you locate one molecule in a material, the other molecules surrounding the first one would be identical – both structurally and electronically. This usual state is said to a have valence symmetry. Now if you break this symmetry, the electronic structure of a molecule will no longer be the same as its neighbour. That is a state with a broken valence symmetry which usually occurs via an internal electron transfer and this mechanism is called charge disproportionation (CD). Electronic symmetry breaking via CD has fascinated condensed matter physicists for decades as it results in multi-faceted changes in electronic, magnetic and optical properties of a material, triggering ferroelectricity, metal-insulator transition and colossal magnetoresistance. While fascinating to study and play with, symmetry breaking as a phenomenon has never been of any practical use because the observation of CD necessitates several material/physical conditions to co-exist, viz. (i) strongly correlated electronic materials, (ii) crystallinity, and (iii) specific physical parameters (e.g., a given value of temperature, pressure, strain, etc.). These conditions are impractical for any real device and hence, are roadblocks for technological deployment of CD. Our results establish a new field driven ion displacement CD mechanism requiring none of these specific conditions that can be realized in a spin-coated molecular film over a wide temperature range (240-350) K exploiting CD for the first time in a practical device application. The devices thus realized are robust (endurance of >1010 cycles), reproducible (tested over 295 samples), fast (<30 ns) and scalable (till ~60nm2) showing ultra-low energy switching – 4 orders better than the state-of-the-art memristors.
Now that we could achieve CD in a nano-device, we were also able to change the film resistance as well as capacitance at the same time resulting in a concomitant realization of a ternary memristor and binary memcapacitor . This demonstration opens the door to a sea of possibilities for physicists and circuit designers. Besides changing film resistance and capacitance, a CD mechanism could also switch magnetic and optical properties of a film, potentially enabling a new generation of multifunctional nano-devices in which all these properties can be simultaneously regulated at ambient conditions just by changing the applied voltage. In terms of applications, this could potentially enable ternary logic which, according to some mathematicians, is expected to offer the efficient most solution for digital computing. Additionally, our recent measurements suggest that, we can drive our devices to self-oscillate or even to a purely chaotic regime or interchangeably between them, closely replicating how our human brain functions. Computing scientists now recognize that our brain is the most energy efficient, intelligent and fault tolerant computing system ever made. Being able to emulate the brain’s best properties while running millions of times faster would change the face of computing as we know it. Based on our low switching energy, and dynamic tunability we have started believing that our molecular systems might eventually be able to deservedly compete with the state-of-the-art brain-inspired, brain-mimetic oxide and ‘ovonic’ materials demonstrated to date.
 Goswami, S. et al. Charge disproportionate molecular redox for discrete memristive and memcapacitive switching. Nat. Nanotechnol. (2020). https://doi.org/10.1038/s41565-020-0653-1
 Ghosh, P. et al. Introducing a New Azoaromatic Pincer Ligand. Isolation and characterization of redox events in its ferrous complexes. Inorganic chemistry. 53, 4678-4686 (2014).
 Goswami, S. et al. Robust resistive memory devices using solution-processable metal-coordinated azo aromatics. Nature materials. 16, 1216 (2017).