Demagnetization Dynamics on a Femtosecond Time Scale
March 20, 2014
But progress in this area has been impeded by lack of a clear understanding – and precision measurements – of the fundamental physics involved when materials become demagnetized on femtosecond time scales and nanometer spatial scales.
A productive collaboration between scientists at PML and at JILA,* however, is revealing the fine-grained dynamics of the process. The experimental results to date, obtained from a novel ultra-fast probe technique, have challenged some long-held assumptions about magnetic behavior in alloys; and anticipated improvements in short-wavelength probe output will leave the team poised to provide unprecedented insights into condensed-matter metrology at ever smaller dimensions.
“Our goal is to characterize how materials change their magnetization through the interactions of spins, electrons, photons, and phonons on the smallest scales,” says Tom Silva of the Electromagnetics Division’s Magnetics Group. “Right now, nobody knows exactly what actually takes place in these materials after an intense optical excitation, or what the important length scales are. It’s really wide open.”
One reason for that situation is that scientists have only recently begun to interrogate their samples in small enough dimensions of space and time. Microscopy at smaller scales requires shorter wavelengths, and resolution on the nanometer scale typically requires either electrons or X-rays. A decade ago, the customary sources for probe X-rays were synchrotrons, with inherently limited access time for each research group. Using them was often time-consuming (taking data one wavelength at a time), the signal-to-noise ratio was too low for many applications, and the probe pulse duration was orders of magnitude slower than desired, obliging the researchers to further subdivide the X-ray pulses to femtosecond time segments in a process known as “slicing.”
“After the monochromator, you get a few photons per slice,” says PML collaborator Justin Shaw. “The repetition rate is about 6000 photons per second. Needless to say, that’s not much to work with.”
The PML-JILA collaboration, by contrast, employs a tabletop technique – developed and improved by JILA Fellows Margaret Murnane and Henry Kapteyn over the course of a decade – called high-harmonic-generation (HHG) to probe materials with very short (< 10 fs) pulses of extreme ultraviolet/X-ray radiation containing multiple wavelengths.
The sample is fabricated in the form of a grating, and repeatedly demagnetized by a 780 nm laser pulse that heats the material up to and beyond its Curie temperature (the maximum temperature at which ferromagnetic order is maintained), prompting a cascade of microscopic interactions. While that is happening, the broadband HHG probe beam is directed onto, and reflected from, a spot on the grating. The intensity of the reflected beam’s components, spectrally spread by the diffraction grating, changes depending on both the amplitude and direction of the magnetic ordering of the electron spins in the sample at any given time over the few hundred femtoseconds during which demagnetization takes place. In addition, by measuring the change of reflectivity at different energies, the magnetization dynamics of the different constituent atomic species in a magnetic compound can be independently queried.
So by examining the difference in the reflected beams at regular intervals, the scientists have a sensitive measure of the dynamics of demagnetization – in effect, a microscope for tracking the changes – with the potential for spatial resolution on the scale of 30 nm or 40 nm. They hope to halve that number in coming years.
“Around 2010, when the HHG technique was improved sufficiently,” Silva says, “we immediately started seeing things that nobody had ever seen before.”
One of them was the surprising difference in response of two different atomic species – nickel and iron – that are contained in the metallic alloy Permalloy, which Silva calls “the Drosophila of magnetism.” It was generally thought that the dynamic magnetic response of each atomic species in a ferromagnetic alloy was essentially indistinguishable, and that their average collective responses to demagnetization were adequately described by mean field theory.
But when the team began irradiating the samples with short, high energy (tens of electronvolts) pulses, “we learned that it’s not actually true if you’re talking about very fast processes,” Silva says. “It turns out that nickel and iron behave differently. Iron reacts to the light more rapidly than nickel. In particular, the Ni response was delayed relative to the Fe response. It appears that optically-pumped angular momentum is driven out of Fe and into Ni during the demagnetization process. That was a complete surprise.”
Many of the bulk magnetic properties of the alloy are determined by the interatomic exchange-coupling between the Fe and Ni electrons, and that coupling has a characteristic time-scale associated with it. Measuring the alloy’s response on shorter timescales revealed the difference in response between the two elements, which the experiments indicated was an intrinsic property of the material. The team tested their results by adding copper to the alloy – thus reducing the exchange-coupling between the Fe and Ni electrons – which further delayed the Ni demagnetization in proportion to the percentage of copper in the mix.
The next step is to examine the demagnetization process on finer scales. The researchers anticipate that the Kapteyn-Murnane labs will eventually develop a laser that can produce photons with kiloelectronvolts of energy, and wavelengths on the order of 1 nm.
“Ultimately, what we want to do is provide a description of these phenomena that spans the range from quantum-mechanical to macroscopic,” Silva says. “Can we understand something about the actual fundamental quantum-mechanical structure in the magnetic material during this demagnetizing process? Can we actually provide a macroscopic picture?
“In a way, it’s like analyzing an election. When people vote, it’s either one party or another that wins. It’s a binary outcome – in our case, a material that is either magnetized or not.
“But before the voting, there’s actually all this polling that looks at the fine structure of political behavior, which sub-populations are inclined to go one way or the other, and where are they located, city by city, block by block. We want to get that kind of picture, and examine the details of the electron population during the entire process.”