It’s one thing to design a 3D microchip on paper, but it’s quite another to manufacture such a device in practice. We designed the device itself quite quickly; within a couple of weeks, in fact. However, actually building the chip – getting all of the atoms into the desired positions and performing how we wanted them to – took the best part of two years to achieve.
Professor Russell Cowburn FRS
A team of researchers from the University of Cambridge
has succeeded in creating a microchip capable of transferring information across three dimensions. The scientists, whose work has been detailed in the journal Nature
, believe that in addition to being cost effective, their 3D spintronic chip will open up new possibilities in terms of storage capacity.
Conventional microchips are only able to move information across a single plane; left to right and front to back. However, by exploiting the properties of component materials such as cobalt, platinum and ruthenium, the researchers have been able to spread data amongst multiple layers.
To learn more about this shapely new microchip, I spoke to lead researcher Professor Russell Cowburn FRS, Director of Research at the University of Cambridge’s Cavendish Laboratory
. I began by asking whether his team had encountered any significant challenges during the course of the chip’s development.
"It’s one thing to design a 3D microchip on paper, but it’s quite another to manufacture such a device in practice," Professor Cowburn replied. "We designed the device itself quite quickly; within a couple of weeks, in fact. However, actually building the chip – getting all of the atoms into position and making them perform how we wanted them to – took the best part of two years to achieve. Material-fabrication processes can be very difficult to get right, but once you do
get them right, they tend to keep on working. Even so, there was an awful lot of trial and error involved in getting these processes to work in the first place."
Indeed, the path that led to the successful creation of this 3D microchip was littered with unforeseeable challenges. However, Professor Cowburn believes that the time and effort invested by his team has ultimately proven worthwhile. The researchers’ hard work should help things to run smoothly in the future.
"Sometimes, things just stop working for no obvious reason," he explained. "At times, we found it difficult to get the atoms to do what they were supposed to do. However, by and large, we now know the best tricks to use. We have learned which things work and which things don't. For instance, the length of time for which your vacuum chamber is pumped can impact whether or not your ruthenium does its job properly. If the chamber has only been pumped for a couple of days, this element will not perform effectively. However, if you pump the chamber for a week, the ruthenium will work. I should mention that there was nothing out of the ordinary about the difficulties that we encountered. These are the obstacles that one typically has to overcome when developing a novel microchip. I am confident that the process can be industrialised."
Professor Cowburn has said that this project represents a great example of advanced materials science. By exploiting the naturally occurring properties of the microchip’s component materials, his team was able to avoid certain difficulties at the outset.
"To construct our microchip, we used three basic elements: cobalt, platinum and ruthenium," he explained. "As the cobalt is magnetic, it can be used to represent the data. The platinum, which sets the direction of north and south, allows us to transfer information in the third dimension. Finally, ruthenium enables the different magnets of the microchip to communicate with one another. We used this element to pass data from one layer to another.
"These are some of our favourite materials from the field of spintronics," Professor Cowburn continued. "Spintronic chips offer a greater potential to create functionality from the basic atomic properties. Conventional silicon electronic chips will probably always need to have transistors integrated within them. Spintronic chips, on the other hand, do not require transistors at all. The general feeling is that during the coming years, most of the memory within computers will be provided by these chips. This is why we opted to use spintronics. It is important for us to align our research with technological trends."
Whilst the accomplishments of the Cambridge-based team are certainly impressive, such projects can only be judged as successful if they prove to be useful. In light of this, I asked Professor Cowburn what advantages are offered by his 3D spintronic chip. For example, why not simply use a series of single-layer microchips?
"That's a really important issue," he replied. "I can give two answers to your question: one economic and one technical. Economically speaking, the cost of producing a microchip is proportional to its surface area. For example, you could produce a device that incorporates 10 individual layers – the kind that are included within conventional microchips. However, the cost of doing so would simply be 10 times that of producing a single microchip; it wouldn’t be worthwhile. The main driver behind the shrinking of microchips is not
to keep our mobile phones small; it’s to make them cheaper
. 3D microchips would be of no interest if cost rose proportionally with layers. We have created a microchip that, from an economic perspective, looks like we are only making one layer, whereas in actuality, we are producing several.
"As I said, 3D spintronic chips could also offer some technical advantages. Take, for example, heat dissipation. Most silicon microchips are at their limit in terms of how quickly heat can be removed from them. This is why the clock rate has gone down slightly over the last few years; the chips would melt if they got any faster. If you were to make a 3D chip in the conventional manner, it would be even more difficult to remove heat from its centre. Spintronic chips, however, are extremely energy efficient. There is no problem associated with cooling them down because they hardly generate any heat in the first place."
Finally, I asked Professor Cowburn about the next steps for his research. As he explained, he and his colleagues now need to build upon the design principles that they have established.
"The electronic functionality that we have developed, in the form of our little shift register, is very, very basic," Professor Cowburn concluded. "We have only really gone so far as to prove that this technology is functional. We now need to increase the complexity of our device in order to achieve levels of performance that will be required for a genuine industrial application."