Two new types of transistor may lead to simpler, more efficient computers

Replica-of-first-transistor — Image by Revolweb via Flickr

WHEN baking a cake it helps to have all the ingredients within reach, rather than wasting time and energy making frequent trips to the pantry.

Something similar is true of the logic circuits in computers’ microprocessors. These could be made faster, and would consume less energy, if they were able to store information themselves instead of fetching it from separate memory chips or hard drives.

The problem is that the transistors used to make logic circuits hold their electronic state, and therefore any data they contain, only when powered up. The choice engineers face is thus between supplying continuous power to a transistor, so that it can retain its memory (which costs energy), and ferrying data that would otherwise be lost to and from so-called non-volatile memory devices that do not require continuous power (which costs time). Cracking this problem—so that transistors can act as their own non-volatile memory—would make all computers faster. It would be particularly valuable, though, for mobile devices. These could be made smaller and lighter, since they would require fewer components. And they could go for longer between charges.

To this end, Hiroshi Mizuta of the University of Southampton, in England, and Tsuyoshi Hasegawa of the National Institute for Material Science in Tsukuba, Japan, are proposing a marriage between two novel types of transistor that could hardly be more different. One, the atomic transistor, draws on the latest advances in nanoscience. The other, the mechanical transistor, sounds as if it has been lifted from the annals of the industrial revolution.

The atomic transistor works, as its name suggests, by shuffling individual atoms around within the device. The atoms in question are copper, and the result of the shuffling is to create or destroy a conductive pathway between two crucial bits of the transistor, the source and the drain, thus switching the device on or off. That is possible because this part of the transistor is made of tantalum pentoxide, a material whose atoms are arranged in a lattice which contains holes large enough for copper atoms to squeeze through.

The mechanical transistor, more properly called a nano-electromechanical systems (NEMS) transistor, creates and destroys the connection between source and drain mechanically. When a voltage is applied across two beamlike electrodes made of aluminium, which are separated by a gap of around 50 nanometres, charge builds up on each, creating an attractive force between them, until a critical point is reached. At that moment one of the electrodes flips towards the other, causing the two to make contact. This closes the circuit and turns the transistor on. Apply the opposite charge and the electrode flips back, breaking the circuit.

What makes these two very different types of transistor attractive is that both the copper and the aluminium stay put when the power is turned off. They can thus act as memories as well as processors. The “on” state represents one type of binary digit (a one, say) and the “off” state represents the other (zero). Dr Mizuta and Dr Hasegawa are therefore using the novel transistors to try to make the world’s first non-volatile processor chip.