Washington, DC - Imitation may be a form of flattery, but sometimes exact duplication is just common sense. That was the case recently when Intel Corporation asked NIST researchers to help the company precisely replicate a measurement system invented and developed at NIST. This month Intel scientists published their first results from use of the NIST-model system in the Journal of Applied Physics.
“It’s sort of the next stage in tech transfer,” says Tom Silva of NIST’s Physical Measurement Laboratory. “Not only are they using it, but they’re publishing papers on measurements taken with it.”
The device in question allows scientists to study the dynamic properties of materials for eventual use in spintronic random-access memory. This promising technology is a matter of urgent interest worldwide because of its inherent speed and minimal energy demand. Conventional computer memory (e.g., DRAM) works by storing bits as charge states in tiny electronic structures. That requires a significant amount of current – producing a substantial amount of heat in chips with billions of individual components – to perform memory operations, and then more current to continuously refresh the system so that the data will not be lost as the structures discharge, a condition called volatility.
Spintronics, by contrast, stores information by manipulating the spin states* of electrons, procedures that can be accomplished with vanishingly small amounts of power in a very short time. Moreover, once written, spintronic memory in stored in the nanoscale magnetic orientation of ferromagnetic materials, so it is non-volatile.
But the performance of spintronic memory depends critically on the degree to which the magnetic nanostructures resist changing spin orientation. That property is called damping, and it has been nearly impossible to study for the kinds of extremely thin magnetic films that are required for memory applications.
So starting in 2009, Silva and colleague Hans Nembach designed, built, and iteratively improved a system that could measure damping and reveal how it varies with material composition, mode of excitation, and layer thicknesses.
A sample is placed in a magnetic field and exposed to microwaves that travel down a waveguide. When the microwaves interact with the material, the electron spins tend to resonate with the microwave frequency and the applied field, and changes in that resonance are monitored by a sensor. Adjusting the parameters of the system, called a ferromagnetic resonance (FMR) spectrometer, identifies key factors in the material’s damping properties.
With collaborators, the NIST scientists published the initial results from the system in 2011. “I and Hans Nembach basically beat on the instrument and figured out all the ways we could get rid of the noise sources for measuring very challenging, single-layer samples," Silva says. "We modified it to eliminate most of the noise, and pushed the sensitivity up to where we can measure films in the range of 0.5 nanometers or so in thickness.”
In 2012, a representative of Intel came to the Silva/Nembach lab and asked to see the device. “They didn’t know if anybody had the sensitivity to be able to look at these very thin films, intended for incorporation in nonvolatile memory,” Silva says. “They left some samples with us for characterization. We got the data, analyzed it, and sent back results. They in turn sent back more samples.”
“It very soon became clear that Intel needed its own dedicated system, and they wanted an exact copy of ours. We helped them with parts lists and the custom components that we ended up buying. They even used our software.
“So they have an exact duplicate. We’re still collaborating with them in doing certain kinds of measurements. But their FMR capabilities have advanced to the point where they’re doing cutting-edge research on their own.”
Meanwhile, the number of similar FMR systems at NIST has now grown to half a dozen, with more on the way to satisfy the growing interest in spintronic applications.
According to the Intel-authored publication, "Spin-transfer torque random access memory (STT-RAM) is one of the most promising future memories due to its non-volatility, low power consumption, and fast switching. STT-RAM with perpendicular magnetic tunnel junction has great potential in that the switching current is lower than that of in-plane molecular tunnel junction, while the thermal stability can be kept the same."
"The damping parameter is effectively the only parameter that can be adjusted to reduce the threshold current. Therefore, measuring the damping parameter accurately for various thin film structures, using the NIST-originated FMR system, is critical in optimizing spintronic random-access memory performance."