Electronics built from ferroelectric materials have low power consumption and are well-suited for information storage. Their competitiveness depends on developing novel architectures on the nanoscale. A research team from Lund University and ETH Zurich in Switzerland has used the NanoMAX beamline at MAX IV to image through metal contacts on the ferroelectric material bismuth ferrite to see how they affect the material beneath them.
There are materials that spontaneously separate positive and negative charges, and thus create an electric field, within themselves, without an external bias being applied. They are known as ferroelectric materials, and are the electric version of the more well-known ferromagnetic materials that fridge magnets are made of. Ferroelectric materials are currently attracting a lot of attention for computing and information storage applications. Their use is not entirely new, however – ferroelectric RAM memories have been around commercially since the nineties and are used, for example, to store your video game result in the console – but they need development to become truly competitive.
Switching the polarisation of a ferroelectric memory bit and encoding a “1” or “0” is power-efficient, and the encoded bit stays in its state also when power is switched off, so-called non-volatile memory. The roadmap for ferroelectric RAM points towards more bits and storage per chip, meaning a smaller piece of the ferroelectric material is used for encoding the bit. Imperfections in manufacturing will thus have a larger impact, and there will be higher demands on the stability of the polarisation in ultrathin material films. This development requires different ferroelectric materials than the ones mainly used today.
“Scalability is quite important for ferroelectric applications. Most ferroelectrics suffer from a notorious depolarising field in the ultrathin regime; it is the internal field induced by bound charges at surfaces and interfaces. When the film thickness shrinks, the effect of such field becomes more noticeable and prevents it from having a net polarisation”, says Bixin Yan, one of the researchers.
Bismuth ferrite has been identified as a promising choice, and it is also something called a multiferroic, with both ferroelectric and magnetic ordering, which makes it even more powerful.
“One special benefit of bismuth ferrite is the coexistence of ferroelectric polarisation and net magnetisation, which are correlated through so-called magnetoelectric coupling. This enables the combination of the “easy-write” of ferroelectric and the “easy-readout” of magnetisation, which will further reduce the energy consumption of memories,” says Bixin Yan, one of the researchers in the team.” I’m also quite hyped about how it could be used for neuromorphic and in-memory computing applications.”
With X-ray-based methods, researchers can non-destructively image the ferroelectric material with high resolution. The resolution is important for investigating the structural and polarised domains that make up the smallest units of the material properties. Domains can switch polarity, and each bit is made up of several domains. As bits get smaller, fewer and fewer domains will make up each one, and the interface with a metal contact or electrode can affect their properties.
“We demonstrate in this study that nanometre-resolved X-ray diffraction allows high spatial resolution mapping of the domains beneath electrodes. In this way, we can investigate what effect the top electrodes have on the domains and their switching. This information is critical for understanding ferroelectric behaviour in real devices,” says Megan Landberg, beamline scientist at MAX IV and researcher in the team.
The technique used in the study of bismuth ferrite could also be used to study other materials or even working devices in so-called operando studies.
“It can be applied across many ferroelectric materials – particularly if they have domains on the tens of nanometers to a few microns’ length scale. We hope it can be adapted to be used across more materials systems. For us, the next step is obvious – operando studies! The higher energy X-rays allow us to image real devices, so we want to investigate what happens to the domains during the ferroelectric switching process. To do this, we must be able to switch the devices in-situ, which we are already pursuing,” concludes Landberg.
