University of Minnesota
School of Physics & Astronomy

Spotlight

When noise is a good thing

Dan Dahlberg
Dan Dahlberg
Richard Anderson
                                                       

Physicists at the University of Minnesota are using very small, magnetic particles to answer questions in fundamental physics that have broad reaching implications in understanding some difficult real-world problems.

In the 1990s, Professor Dan Dahlberg was attempting to do spectroscopy measurements on magnetic tunnel junctions but was discouraged by other researchers saying there would be too much noise. While pursuing the spectroscopy, he had another graduate student look into the noise problem. Based on the graduate student’s research, Dahlberg’s group decided that many of the measurements previously published on magnetic noise were likely incorrect. A rather subtle overlooked effect was what others had been measuring and calling noise. “Being a physicist and believing in the laws of thermodynamics and statistical mechanics, I knew there had to be magnetic fluctuations” which would produce measurable noise and that other scientists had not seen it.

Given this belief in the physics, Dahlberg and his group set out to develop a model system to observe and investigate magnetic noise. The model system they developed were small magnetic particles on a length scale that have different physics than that in both the macroscopic and the atomic regimes, a regime called mesoscopic. The smallest of the mesoscopic magnets they made are about 200 atoms on a side and 25 atoms thick. The different physics clearly showed magnetic noise that could be modeled accurately. “What’s important about the work, is that historically at this length scale, people had measured collections of particles and relied on averages. We developed a system in which we can throw away the averages and measure one particle at a time.” The devices which are built at the Nanocenter at the University of Minnesota are designed so that they can measure one particle at a time, by attaching four independent leads to each one. With four wires attached to a single particle, two wires can be used for passing current through the mesoscale magnet and the other two for measuring the voltage generated across it.

The magnetic particles exhibited random telegraph noise where the magnetization in a particle jumps back and forth between two directions; Dahlberg’s devices are the best model system of random telegraph noise ever investigated.

“Anytime you develop a new technique or model system, it puts you into a regime where everything opens up,” Dahlberg says. “I’d never studied noise in my life. Since then it’s all I’ve been doing.” For example, the mesoscale magnets are ideal to study stochastic resonance noise.

In certain systems, like his magnets, noise is generated by bouncing back and forth between two different states randomly. It turns out in this case the random noise can be used as an amplifier. “That’s where the resonance come in, you use the thermal noise, the random bouncing back and forth, to amplify a small signal.” This phenomenon hadn’t been thoroughly explored because no one had developed a good model system. Dahlberg’s magnetic particles ended up being ideal for this and his group has done the best studies to date on stochastic resonance. The testing of the stochastic resonance phenomena is important as it has been used as a fundamental model for brain function, signal analysis and in medicine.

Dahlberg’s group also studies 1/f noise, sometimes called “pink noise,” because of the appearance of visible light with this power spectrum. This noise is everywhere in nature from the fluctuations of the flood levels of the Nile to the traffic on the freeway. “We don’t have a general framework, no formula such as f=ma, for 1/f noise.” One theory is that it the noise is caused by a collection of random telegraph noise oscillators that act independently. In other words, even though the random telegraph noise of a single particle is not 1/f but when many are measured it is predicted to exhibit a 1/f type of noise. Dahlberg’s group is able to take their small magnetic particles and chain them together to measure the noise in each individual particle and the noise of all the particles in the chain at once. The group has found that 1/f noise does indeed arise from a chain of random telegraph noise oscillators. This is the first confirmation of that theory. “Now could you identify a random telegraph oscillator in the flood levels of the Nile? I’m not sure, but in semiconductors and in electronic transport you have a chance to identify the random telegraph noise oscillator and this would tell you how 1/f noise arises.”

Dahlberg’s group is also applying their noise studies to a long term problem called spin glasses in which a disordered magnetic system has a phase transition that is complex and not well understood. Dahlberg’s group are making measurements with their small systems where they can see behaviors in spin glasses that people have only been able to guess at because they’ve had to measure collections of particles.

Dahlberg’s future research continues in the area of stochastic resonance, 1/f noise, and spin glass noise.

More information at http://www.physics.umn.edu/people/dand.html