# Century-old law on electric noise overturned

Electric noise can be useful for scientists but inconvenient for chip manufacturers. They do share a wish to predict the amount of noise. PhD student Sumit Tewari overturns a century-old law relating noise to current. He defends his thesis on March 27th.

We usually see random fluctuations in an electric current as an unwanted part of a signal. However, this noise can provide useful information, for example about the transmission of electrons across a tiny conductor. In that case physicists speak of shot noise. Scientists gather information about electrons while they produce shot noise much like you can get information about hail stones while they produce acoustic noise on a metal roof. For this, researchers do need to know the correct relationship between shot noise and current. Since the early 1900s, they have assumed that shot noise always increases linearly with current. Sumit Tewari now shows in his PhD thesis that this is false.

## Surprise

In the lab of his professor Jan van Ruitenbeek, Tewari stretched a gold wire up to the point that only a single atom is left in the middle, and applied a voltage. To his surprise, he saw that the shot noise did not keep on increasing linearly as he increased the current. In fact, at some point the noise even dropped. ‘This came as a total surprise,’ says Tewari. ‘The data came in and we thought: "Did we measure something wrong?" But we were able to verify the data. Then the only thing left to do was to think of anything that can explain this nonlinearity.’

## Imperfections

Tewari worked out a model in which local imperfections within the wire cause electrons to interfere with each other, thereby messing up the neatly linear relationship between noise and current. This is possible because electrons are not just particles; they are waves at the same time. The model shows that as long as there are imperfections, you will never have a linear relationship.

## Minimize noise

Contrary to physicists, manufacturers see noise purely as inconvenient. In the future they might be able to play around with imperfections within their devices using predictions from Tewari’s model. That way they could minimize the noise precisely at a device’s particular voltage.