Thermal Memories

Thermal Memories
Story by Ruth Beran

Many materials have memory. If a rubber band is stretched, for example, and then let go, it will be slightly larger than before it was stretched. This is called elastic hysteresis. Similarly, materials like iron can be magnetised, and will remain magnetised indefinitely, until demagnetised by heat or a magnetic field in the opposite direction. This type of memory is used in everyday items like credit cards and hard drives, and is called magnetic hysteresis.





Figure 1. Figure 1: The structure of the iron (II) compound at 90K



Computers have shrunk dramatically in size over the last 70 years, and advances from valves to transistors to chips have also driven massive growth in computer power. The race is on though, to make computers even smaller, more powerful and more energy efficient, and scientists are looking at molecules for the answers.

Memory is one of the most important components of any computer. At the moment, computer hard drive disks use particles of iron to store memory, but these particles are relatively large. One way to make hard drives smaller and more powerful is to make them molecule based, and synthetic chemist Sally Brooker and her team have created an iron compound which has thermal memory, in the same way that those particles of iron in today’s hard drives have magnetic memory.


Most iron compounds are either high spin (four unpaired electrons) or low spin (no unpaired electrons), no matter what temperature they are.

“It’s only a special group of compounds that can do this switching in the first place: when you’ve got the Goldilocks size just right – not too big and not too small,” says Brooker. “Within that subclass is an even smaller subclass that can do this hysteresis, where the switching isn’t quite the same in the two modes of heating versus cooling.” The molecule that Brooker and her team have created is an iron (II) compound which not only switches between the two different states – high spin or low spin – but switches at different temperatures, depending on whether the compound is being heated or cooled.








Figure 2. The hysteresis loop at different scan rates

Hysteresis is when a system depends not only on its current environment but also on its previous history. So for materials that exhibit thermal hysteresis, the heating and cooling modes track different paths, called the hysteresis loop. Within the temperature range of this hysteresis loop, a compound can be in either one of two states, depending on whether it is being heated or cooled. In this case, the iron compound can either be high spin (if it’s hot and being cooled down), or low spin (if it’s cool and being heated up). As Brooker says, “it knows where it comes from”.

In the low spin state, this compound is diamagnetic and a dramatically different colour, purple, to the high spin state, which is paramagnetic and almost colourless. As such, it could be used in the future in displays. The spin states of the iron centres are also very sensitive to small changes in the environment, so they could be used in sensors.

Another application is in computing. “If the operating temperature of your device is in the middle of the hysteresis loop, and you can be either high spin or low spin depending on whether it is cooling down or warming up to get there, then you have a binary code. An on/off,” says Brooker.

For applications like long term storage of data, the wider the temperature range of the hysteresis loop the better, as it’s more difficult for the material to accidentally change “states” and lose its memory. The compound that Brooker and her team have created has a thermal hysteresis loop width of 22°C.

“This is modest, but it’s equal to the best reported to date for a dimetallic compound. A lot of hysteresis loops are only a couple of degrees wide, which is practically nothing really,” says Brooker.




Figure 3. A plot of the scan rate dependence of the spin crossover temperatures (in heating and cooling modes) and hysteresis loop width.

If compounds like this one are to be used in memory components, then Brooker is adamant that studies on thermal hysteresis loops need to include scan rate data as well, in order to develop understanding of the potential life time of the phenomenon.

Scan rate refers to the rate of change in temperature over time. Magnetometers, like the SQUID machine used in this study to measure the unpaired electrons as temperature changes over time, can either be used to scan in either sweep or settle mode. “In settle mode, it’s taking small temperature steps and waiting at each new temperature until it stabilises, before it takes a reading. So it’s very slow,” says Brooker. “In sweep mode, you set a scan speed and it just keeps changing the temperature as a function of time at that rate, taking measurements as it goes.”

With high demands on machine time, scientists may be tempted to scan materials faster by using the sweep mode. However, because thermal hysteresis is a kinetic effect, usually the faster you scan the wider the hysteresis loop. Conversely, if you scan through the temperature range slowly enough, the hysteresis loop may disappear. “If you’re going to make a practical device, that’s a fairly serious thing to happen!” says Brooker. “A memory component needs to last in the order of ten years, so scan rate studies and relaxation studies in the loop region are critical to probe this.”

For this reason, Brooker makes it clear that researchers should be reporting their method of data collection, such as the mode of scanning used, and how fast the scan rate was.



Figure 4. Generalised view of a possible explanation for the unusual kinetic data, with ligands in gray, iron ions in blue (low spin) and red (high spin; circle vs small ellipse vs large ellipse show increasing distortion).

Brooker says that the scan rate studies for this particular iron compound turned out to be the most interesting part of the research. In this case, the hysteresis loop was scan rate dependent, but only when the compound was cooled, not when it was heated. So the hysteresis loop closed from one side only.

“But we have now seen other examples in which it closes from both sides, which is more what you’d expect,” says Brooker –  this is why it’s important to report scan rate studies.

Another unexpected result for this particular compound was the hysteresis loop didn’t close completely, remaining 22°C wide even at the slowest scan. Indeed the lifetime was found to be greater than 24 hours, which is a small step towards the aim of designing compounds that can retain memory for ten years or more.

However, there are also a number of practical issues to overcome before compounds like this one can be used in computer memory devices. “If you’re going to use temperature to do this switching you’d need an array that would allow point heating,” says Brooker. “The compound would also have to be on some sort of surface or support.” Hence her team is collaborating with surface scientists as they look to immobilise their molecular switches on solid supports.


Figure 5. Scan rate DSC study showing the exo and endothermic transitions seen on cooling and heating.

A number of the instruments used in this study were funded by the MacDiarmid Institute, including the Quantum Design MPMS SQUID and PPMS magnetometers based at the Robinson Research Institute (Jeff Tallon), the Mössbauer spectrometer at Otago University (Guy Jameson), and the low temperature attachment for the Differential Scanning Calorimeter at Otago University.

“A lot of capability has been added in recent years, so we are now able to do these studies within New Zealand, and this is primarily thanks to the MacDiarmid Institute,” says Brooker.

For more detailed findings go to: Kulmaczewski, R.; Olguín, J., Kitchen, J.A.; Feltham, H. L. C.; Jameson, G. N. L.; Tallon, J. L. and Brooker, S. Remarkable Scan Rate Dependence for a Highly Constrained Dinuclear Iron (II) Spin Crossover Complex With a Wide Thermal Hysteresis Loop, J. Am. Chem. Soc. 2014 136, 878−881. Images published with permission from The Journal of the American Chemical Society. Copyright 2014 American Chemical Society.