- Issue 2 |
- January 2010
Profile: Peter Maksymovych
“Can you get close?” “Sure.” Over the next hour or so, we talk about the research he’s been doing, his background and how it led him to ORNL, and what he enjoys outside of work – though he’s quick to add “I don’t want to [give you] the impression that [work] is torture, this is pleasure… There’s a lot of torture in it involved, but isn’t this true for everything, this is life, right?”
As Maksymovych explains, silicon and other semiconductors surround us. They’re in “that recorder you have, the computer I turned on” to show me graphics; in fact with the exception of magnetic hard drives, semiconductors are used for almost everything that contains electronic information. This is possible because of a decades long effort to learn not only how to make these materials, but how to use doping to control their properties. For example, conductivity is tuned by ionizing or deionizing the doping.
This has worked well so far, but there are serious challenges in moving forward. The semiconductor industry is working of scaling electronics, attempting to shrink them to the nano scale to both increase their speed and reduce the amount of power wasted through dissipation. Doping, however, leads to a minimum size for the electronics. Eventually, there will only be a single atom of dopant – or no atom. Other things also limit the size, like the complexity of the devices. After all, “if you need to put three electrons in a point of space, that point of space has to have finite dimensions.”
Ferroelectric oxides have also been used. They are wide bandgap semiconductors, almost insulators. Devices like Play Stations currently use them through FeRAM (Ferro-electric Random Access Memory.) FeRAM is useful – the memory is in the polarization of the material, so it can be removed from the machine without losing memory, and when it’s no longer needed, it can be written over. Unfortunately, this method is physically large, and can’t be scaled. It was suggested years ago that conductance should be measured instead, since it can be detected at smaller scales and will hopefully be modulated by the polarization, giving two states (up and down), enough for use in memory. The problem is that these materials are practically insulators – you can’t run current across them. This can be fixed by shrinking it down – but then you lose the polarization property and with it memory functionality.
Maksymovych and his colleagues have found a way around that: “You don’t actually need to shrink it down. All you have to do is [use] a somewhat leakier medium.” These ‘leaky’ thin films are usually avoided due to their interference with other measurements, like polarization. To avoid this problem, the researchers designed a new machine. “Those who do characterize leakiness have little means to differentiate intrinsic and extrinsic behaviors. Our machine, by virtue of its low-dimensional design, lets us identify the intrinsic and controllable properties. If we so desire, we can subsequently go and find defects in the material to see what influence they have on the intrinsic properties, or whether they dominate them.”
They used this machine to study an everyday material – “ an off the shelf, PZT (lead, zircon and titonate) material which is sort of the prototypical ferroelectric oxide.” Ferroelectric oxides are unique in that they like to polarize themselves. This causes charges to form on opposite sides of the oxide, and these charges can be switched by applying an electric field.
As described in the Science paper, Maksymovych and his colleagues showed that the local conductance can be controlled by switching this ferroelectric polarization. The polarization provides two memory states, and the conductance provides the way to read this memory. As Maksymovych puts it, “It’s a nondestructive readout, it exists in conventional ferroelectric films, and even more importantly, it’s tunable. …It’s not just that you can shrink the size of the material to nanometer scales – which would already be difficult for any doped material – but it’s also that you can control when and how much it switches.” This is possible because we already know the thermodynamic equations that govern the behavior of its phase transitions. These laws allow the tunability of ferroelectrics. Using these, both the current and the voltage requirements can be changed to adapt to different environments and uses.
This new form of memory – unlike the current flash memory – will be highly efficient since it’s driven by electric fields, not by current. It is also has shown that it isn’t necessary to go to the extreme sizes of 1-2 nanometers, which are “difficult to achieve in [an] academic setting” and “almost impossible to achieve in any practical application.” Though this new technology is anticipated to affect much more than electric applications alone, Maksymovych maintains the research (both already done and further work) “would pay off if we were to offer something real small, fast, efficient, tunable;” which he believes this work will lead to.
But Maksymovych says that he’s driven to study these materials not mainly for the benefits they might give but to discover the fundamental physics behind them, and he predicts his colleagues would say the same. “Deep inside we are explorers, right? And we’re so excited when we see something new….” While scientists always try to anticipate the results of any experiment, “if you uncover exactly what you intended that means either the problem wasn’t very interesting – it was predictable – or the material didn’t surprise you with anything. It’s always exciting to find something we didn’t really think about.”
Peter earned a B.S. in chemistry from the Kiev Tara Shevchenko University in Ukraine and graduated from the University of Pittsburg with a doctorate in physical chemistry. He arrived at ORNL in 2007 as a Eugene P. Wigner Fellow. When not working, Peter enjoys spending time with his wife, Liliya, and his hobbies of reading, unconventional live music, hiking, and diving.