Four years after scientists in the US reported seeing "giant piezoresistance" in silicon nanowires, a team of researchers in France and Switzerland claims that this phenomenon may not exist after all.

Giant piezoresistance is a large change in electrical resistance that occurs when a material is stretched. After it was first reported in tiny silicon wires, claims were made that it could significantly improve nanoelectronic devices, such as nanoscale transistors, and help make ultrasensitive nanosensors.

Now, new work by Jason Milne and Alistair Rowe at the Ecole Polytechnique, Steve Arscott of IEMN-CNRS and Christoph Renner at the University of Geneva calls such applications into question.

Physicists have known about piezoresistance (PZR) – whereby the electrical resistance of a semiconductor changes when a small mechanical stress is applied to it – for many years. Giant piezoresistance occurs when there is a much larger change in resistance for the same applied strain. For example, the change in resistance per unit of strain (the "gauge factor") typically ranges up to 100 in bulk silicon but in giant PZR, this value can reach several thousand.

Practical applications

Giant PZR would find many practical applications. For example, it might be used to detect motion in nanomechanical systems (NEMS) because traditional detectors lose their sensitivity at these length scales. Furthermore, because mechanical stress is currently employed to enhance the performance of electronic devices (in so-called "strain engineering"), it might also help enhance nanoscale transistors too.

The very act of measuring the resistance changes its value Alistair Rowe, Ecole Polytechnique

Four years ago Peidong Yang's team at the University of California at Berkeley first observed giant PZR in silicon nanowires and the discovery created a flurry of interest in labs worldwide. Indeed, the researchers measured gauge factors up to almost 6000. The effect was thought to be a new phenomenon occurring in an otherwise well-characterized material resulting from the sample's reduced size and characteristic surface states.

In a paper just published in Physical Review Letters, the France-Switzerland team claims that these observations were probably artefacts in no way related to the mechanical stress applied to the silicon nanowires. They were, instead, caused by surface trapping of charges induced by the voltage applied to measure the resistance. "In other words, the very act of measuring the resistance changes its value," explained Rowe.

Non-stress-related drift

PZR is usually measured by performing a standard resistance measurement on a sample while gradually changing the applied mechanical stress on it. The trouble is that any non-stress-related drift in the value of the resistance cannot be separated from that caused by the applied stress.

The France-Switzerland team says it overcame this problem by applying an oscillating stress to its samples. In this way, stress repeatedly increases and then decreases as function of time. "This is a fairly standard technique (called heterodyne detection) in physics and engineering and is used to separate two or more signals and give artefact-free measurements," said Rowe.

According to Rowe, scientists had never applied heterodyne techniques to PZR measurements before, so previous measurements revealed large (but not stress-related) resistance changes in the silicon nanowires. "This meant that the resistance drift due to charge trapping (also known as dielectric relaxation) was assumed to be the result of the applied stress", he added. "This now appears to have been an incorrect assumption."

Top-down or bottom-up?

Yang himself disagrees: "They are reporting PZR measurements on a collection of top-down micro- and nano-wires while our measurements were on bottom-up grown nanowires. Their results might not actually be that surprising as we now all know that bottom-up synthetic bridging nanowires have quite different strain levels, surface states and dopant profiles from those of top-down fabricated ones. In fact, the lack of giant PZR effect in such nanowires was already reported back in 2003. However, the lack of giant PZR effect in these new fabricated samples should not automatically imply the same in our synthetic bridging nanowires.

The observed PZR effect in our nanowires, whether it is intrinsic or from the surface states effect, has already proven to be useful Peidong Yang University of California at Berkeley

"After all, the observed PZR effect in our nanowires, whether it is intrinsic or from the surface states effect, has already proven to be useful," he added. "For example, we recently demonstrated the first piezoresistively transduced very high frequency silicon nanowire resonators with on-chip electronic actuation at room temperature. We clearly showed that, for very thin silicon nanowires, their time-varying strain can be exploited for self-transducing the devices' resonant motions at frequencies as high as 100 MHz. This simply would not be possible without the enhanced PZR effect."

The debate looks set to continue.