Forget Flash, time for a phase-change!

Flash memory is everywhere – in your phone, tablet, netbook and camera. But this type of memory has some drawbacks, so semiconductor manufacturing companies are looking around for a successor. Phase-change memory may be in with a chance. PhD student Jasper Oosthoek put this new type of memory under the microscope. Literally.

The pictures he shows do not seem spectacular. Grey goo with specks, that’s what it looks like. But they are two different forms of a phase-change material that may one day allow you to store information faster.

‘This material can be crystalline or amorphous’, Oosthoek explains. In the crystalline form, the atoms are aligned, like neatly stacked marbles. But after heating and a quick cooling-down, the atoms are no longer ordered. ‘And this affects the conductivity of the material. The crystalline phase is a much better conductor.’

Materials that can be switched from conductor to insulator are perfectly suited to building digital memories. Indeed, they are already being used. ‘Rewritable CDs and DVDs work with phase-change materials that are switched by laser pulses’, says Oosthoek. But the sort of memory he has been working on is switched using an electric pulse.

The Dutch company NXP Semiconductors had been working on phase-change memory cells and needed more information on the behaviour of the material during the phase-changes. They contacted The Zernike Institute for Advanced Materials of the University of Groningen to visualize this process. This formed the basis of Oosthoek’s PhD project, which is also financed by the Materials innovation institute (M2i). Oosthoek defends his PhD thesis on 11 April.

‘NXP provided us with a series of memory cells that could be switched between the crystalline and amorphous phases.’ Oosthoek had to prepare them for imaging using an electron microscope. This meant the cells couldn’t be more than 200 nanometers thick. That’s much thinner than your average computer chip. ‘A lot of the supporting material had to be etched off to make the material translucent. This made the memory cells very delicate.’

It also made it virtually impossible to get a working cell in an electron microscope. But Oosthoek did manage to image preset cells. Such memory cells resemble the letter H, with two thick bars linked by a thin line. The phase-change is induced in this line, which determines the electrical resistance of the cell.

One thing Oosthoek noticed was that the amorphous phase wasn’t uniform across the line in the memory cell. ‘The explanation is that the electrons used in the pulse to induce the phase-change carry heat with them. This means that the heating isn’t the same across the line.’

Another discovery he made with his electron microscope is that repeated switching can induce material to leak from the thin line. ‘These findings suggest that the design of these memory cells can be improved’, says Oosthoek.

An important advantage of phase-change memory over flash memory is that it is much more resistant to frequent use, as Oosthoek confirmed in his studies. ‘A flash memory cell can be switched around 10,000 times. But the phase-change cells worked ten to a hundred times longer.’ At the moment, flash memory banks or hard drives need elaborate algorithms to make sure that all cells are used with the same frequency. ‘Otherwise, files that are never changed would increase the pressure on the remaining memory space, and that would reduce the lifespan of the memory device.’

Will phase-change replace flash? ‘At the moment, the investment in flash research is many times higher than for phase-change. Seven years ago, companies feared that flash memory couldn’t be miniaturized any further and would run into trouble within a couple of years. That didn’t happen.’ On the other hand, Samsung has produced a prototype solid state drive using phase-change memory. ‘And at a conference, I saw a demonstrator video showing that phase-change memory is faster than flash.’ Also, Nokia has produced a low-end telephone with phase-change memory.

However, work needs to be done to optimize phase-change memory. ‘The biggest drawback at the moment is that it isn't heat resistant enough.’ Although it is stable for one or two weeks at 100 degrees Celsius, the norm is 10 years at 80 degrees. ‘But there is still plenty room for improvement’, says Oosthoek. ‘And phase-change memory would be ideal for space missions, as it is extremely resistant to radiation, while flash memory is very vulnerable.’

Oosthoek is himself no longer working on phase-change memory. ‘I’ve joined a company called Sentron in the town of Roden. We are developing sensors for medical applications. That’s also an exciting field!‘

Jasper Oosthoek will be awarded a PhD for a thesis entitled ‘Structural and electrical characterization of phase-change memory line cells’ on 11 April. The research was carried out in the Nano-structured Materials and Interfaces Department of the Zernike Institute for Advanced Materials. Prof. Bart Kooi headed the research, which was funded by the Materials Innovation Institute (M2i) and NXP.