Ferroelectric compounds reduce their energy by polarizing spontaneously, that is, by displacing their negative ions with respect to the positive ones. However, such ionic displacements ultimately result in an accumulation of electric charge at their interfaces with other compounds. For good or bad, when ferroelectrics are surrounded by media that does not enjoy such charge build-ups and the attendant electric fields (e.g., stiff dielectrics), the development of a polarization is energetically penalized. Nevertheless, ferroelectrics usually make do by breaking into regions (domains) of opposing polarization, so that the interfacial charges associated to different domains compensate each other and the energy penalty is minimized. The ferroelectric is frustrated in this state, though, as the boundaries (walls) between domains are energetically costly themselves. In fact, if we now apply an external electric bias that favors a particular polarization orientation, the ferroelectric readily forms bigger domains and minimizes the walls, thus approaching its preferred configuration and reducing its energy. This peculiar behavior (all but impossible among non-ferroelectrics) is characterized by a "negative capacitance". The energy released by the ferroelectric becomes available to do work in the surrounding materials, which effectively experience a bias that is larger than the one actually applied. This magic sounding "voltage amplification" has created a lot of technological excitement, as it may be a route towards the development of low-voltage (low-power) electronic devices. While waiting for those to come (which may not happen soon, because of many scientific and technical difficulties involved), we moved to the next obvious question: How large can this voltage amplification be? What can we do to maximize it? With Mónica Graf and others, we used second-principles simulation to predict viable strategies, based on sound physical mechanisms, that yield very large amplifications above 10-fold! Our results explain trends and point at directions for further improvement, suggesting that the intrinsic amplification (in ideal but not outlandish conditions) can indeed be very large. Admittedly, our job as theorists is probably quite easy compared to the device implementation of these concepts. But the important message is this: there is hope, and probably a big prize awaiting! To know more about our voltage-amplification treasure map, please read the freely available Graf et al., Nature Materials 21, 1252 (2022).

Antiferroelectric materials feature a peculiar response to applied electric fields, characterized by a structural transition between non-polar and polar states. Controllable structural changes are always interesting and potentially useful; in the case of antiferroelectrics, their field-driven transition allows to store high energy densities in them. Hugo Aramberri and I have spent several years thinking how to increase this stored energy density, either by optimizing known antiferroelectrics or by trying to discover new ones. While interesting, our results were hardly the expected ones: we ended up doubting the strict antiferroelectric nature of the model compound in the field (see Aramberri et al., npj Computational Materials 7, 196 (2021)) and did not manage to predict any viable alternative. As a last resort, we decided to consider artificial antiferroelectrics made of well-known ferroelectric compounds subject to suitable electrostatic constraints. We knew about the multidomain polar structures that ferroelectric layers usually display when sandwiched between dielectrics, where up- and down-polarization regions spontaneously appear to minimize spurious electric fields and the attendant energy cost of polarizing the dielectrics. It occurred to us that such multidomain structures look like an antiferroelectric (antipolar) phase, and that they can surely be transformed into a monodomain (polar) state under a field. Thus, we tackled a high-throughput computational investigation of hundreds of ferroelectric-dielectric superlattices, optimizing various design variables (layer thickness, dielectric response, ...) to get the best energy-storage behavior. The results exceeded our expectations, as we found that these artificial antiferroelecric-like materials compete with the best known compounds and, more importantly, can be tailored to display optimal properties under specific operation conditions. A happy ending to our struggles! You can read more about it in Aramberri et al., Science Advances 8, eabn4880 (2022).