What Don Quixote thought were giants turned out to be windmills, and what researchers five years ago thought was giant piezoresistance appears, at least for now, to have been just as illusory. Unfortunately, that’s bad news for everyone who hoped to take advantage of this exciting property of miniscule silicon wires. But it is a very nice example of the scientific process at work.
What is piezoresistance anyway? And what good is it?
Normal-sized piezoresistance (or, more formally, “the piezoresistive effect”) is a change in the electrical resistivity of a material that results from mechanical stress — such as stretching or compression.
Piezoresistance was first discovered in the 1950s, and it occurs only in semiconductors. Unlike the considerably higher resistivity of insulators and considerably lower resistivity of conductors, the intermediate resistivity of semiconductors is sensitive to the tiny changes in the atomic structure of a material that occur when it is stretched or compressed.
There are lots of applications of the piezoresistive effect — again, this is normal-sized piezoresistance. Many commercial pressure sensors are based on the effect; they use the change in resistivity to measure the force on the sensor. At the forefront of research, the piezoresistive effect can provide an electronic measurement of nanoscale motion, such as that of a cantilever; this can be used both on man-made nanoscale systems and biological structures. It has also been found that mechanical stress can improve the performance of transistors via the same phenomenon.
The ephemeral promise of giant piezoresistance
Given all that piezoresistance can already do, there was significant excitement in 2006, when researchers at UC Berkeley’s Lawrence Berkeley National Laboratory published measurements of a piezoresistive effect nearly 40 times larger than it is in normal, bulk silicon. This “giant piezoresistance” was seen by stretching or compressing silicon wires only 50 to 350 nanometers wide and 2,000 nanometers long. (For reference, the average human hair is 80,000 nanometers wide.)
In addition to increasing the benefits that could be gained from the higher sensitivity to mechanical motion, the authors pointed out that this result “may have significant implications in nanowire-based flexible electronics.” Also, it seemed to be another fascinating example of the way material properties can be surprisingly different at small scales. For example, nanoscale silicon wires have already been found to exhibit an enhanced thermoelectric effect (in which a difference in temperature between the two ends produces a difference in voltage, and vice versa).
At least for now, though, giant piezoresistance has turned out to be quite normal-sized after all. In the November 2010, researchers in France (at Ecole Polytechnique and Institut d’Electronique, de Microelectronique et de Nanotechnologie) and Switzerland (at the University of Geneva) reported in Physical Review Letters that they reproduced the same changes in resistivity described by the LBNL group . . . but found them to be variations in time that are completely independent of the applied stress!
So what’s really going on?
Consider that the normal approach in these experiments goes something like this: Measure the conducting properties of the wire, increase the applied mechanical stress, measure again, increase again, and so on – gradually stepping up to higher stresses. The French and Swiss collaborators, however, used a “stress modulation technique” in which they measured the conducting properties of the wire for 8000 seconds straight, but flipped between high stress and zero stress conditions every 10 seconds. This allowed them to separate the piezoresistivity from other resistivity changes.
And what changes there were! The amount of electricity the wire conducted decreased 27 percent in the first 1000 seconds, then gradually ramped back up over the rest of the time. Turning the stress on and off, meanwhile, only changed the conduction in the wire by at most a few percent. In other words, the actual piezoresistivity is 10 to 100 times smaller than the time-varying resistivity changes.
What’s causing that time variation? The researchers show evidence that charge is building up on the surfaces of the wires, effectively changing the voltage being applied to them – and, as a result, the amount of electricity flowing through. The rest of the paper presents measurements in good agreement with a charge-buildup model.
Getting back to the issue of giant piezoresistance: What would happen if you didn’t know about that time variation, but rather were stepping up the applied mechanical stress in the usual fashion? Turns out, you’d get graphs just like the ones shown in the 2006 paper, demonstrating a falsely inflated piezoresistance measurement.
That doesn’t mean giant piezoresistance isn’t a possibility, as the authors of the PRL paper readily admit. However, they caution, “future claims must conclusively demonstrate that any measured resistance change be solely due to [the applied mechanical stress].”