Sunday, May 08, 2016

Interacting Quantum Systems Driven Out of Equilibrium - day 1 (updated - complete)

Our workshop was fun and interesting.   There are multiple ways to drive physical systems out of equilibrium - you can take some system and push on it with some force, for example.  In the case of a condensed matter system (whether solid state or trapped cold atoms), you can apply a bias - some difference in population (or chemical potential or pressure) that drives the system, either by adding kinetic energy to it or encouraging the flow of matter and/or charge.  You can apply a temperature difference across the system, driving some average flow of energy through the system's degrees of freedom.  You can shine light on the system, adding energy and momentum either at a steady rate or in a sudden pulse.  One favorite piece of vocabulary these days is a quench - suddenly (compared with relaxation rates of the system) changing some condition like the potential energy of the particles, and then watching the response of the system's degrees of freedom.  Does the system "thermalize"?  That is, do the microscopic pieces of the system interact with each other and redistribute energy so that there seems to be some effective temperature?  Or does the system fail to thermalize, and instead slosh around in some non-thermal configuration for a long time?  There are many open issues.

We had 13 talks on the first day, and I don't want to write exhaustive summaries of all of them.  We will eventually be posting pdf files of the relevant slides.  That being said, I will give a super-brief description of each, and link to a relevant paper or two so that you can see what was discussed.  Here are the 13 talks we had on the first day.

  • Nadya Mason from UIUC spoke about her group's work on engineered superconducting/normal metal structures in magnetic fields.  These devices allow studies of current-driven motion of trapped magnetic flux.  In some sense this is an old, established problem, but traditional models actually do a poor job of reproducing the experimental data.  The experiments are here, and it looks like it's important to include some "delayed friction" to understand vortex motion.
  • Jonathan Bird from Buffalo spoke about his group's studies of quantum point contacts in semiconductors, where it's long been known how to measure electronic conduction down to the limit of discrete quantum channels, where the devices act like waveguides for the electrons.   His group has developed some high speed techniques for making sub-ns electronic measurements, and what really gets interesting is when systems are driven hard, so that the electronic bias is the largest energy scale in the problem - you have to worry quite a bit about exciting phonons and what they do.  A key result is the apparent formation of a specific, somewhat heating-immune transport mode when such a point contact is driven really hard.
  • David Goldhaber-Gordon from Stanford spoke about his group's recent experiments looking at quantum dots, some building on work looking at the so-called two-channel Kondo effect.  An unpaired electron is placed in the position of trying to couple to two (carefully tuned to be) independent baths of electrons.  Some of the not-yet-published results look at interesting scaling as one tunes through the accessible regimes, and involved some stunningly pretty device fabrication done at the Weizmann Institute.  Other experiments looked at the apparent emergence of symmetry in systems comprising two quantum dots.
  • Tilman Esslinger of ETH presented his group's great work on using cold atoms to look at systems rather analogous to the ones Prof. Bird had mentioned.  They can create blobs of fermionic cold atom fluids of unequal populations, and link them by a carefully controlled constriction, and then they can image transport.  If they squeeze the contact to be effectively one dimensional, they can see quantized conductance of atoms (just as solid state folks can do with charge in a quantum point contact).  They can use atomic physics methods to dial around the interactions between the particles, and can then look at how this affects dissipation in the out of equilibrium situation.  Gorgeous stuff.
  • Takashi Oka of the Max Planck Institutes in Dresden talked about Floquet theory and using lasers to control the topology of the band structure of materials.  There was a lot to this talk, and it's not easy to summarize.  In Floquet theory, you apply a periodic driving potential to a quantum system.  Just like a spatially periodic potential energy picks out certain spatial periodicities and gives you a compact way of looking at band structure, temporal periodicity creates what you could call replicas of the band structure but shifted in energy by multiples of \( \hbar \omega\), where \(\omega\) is the driving frequency.  If you do this right, the driven system can have topological edge states.  You can also use periodic driving to reorient the magnetization of materials as if you had a whopping huge effective magnetic field.
  • Andrew Millis of Columbia University has worked on many relevant topics, and in this case chose to speak about theory he and collaborators have done regarding a recent experiment looking at vanadium dioxide.  That material has a structural phase transition at 65 C that separates a low temperature, monoclinic, insulating state from a high temperature, tetragonal, metallic state.  In the experiment, optical excitation puts the material into a metallic state without actually leaving the monoclinic crystal structure.  The theory suggests that this is a correlation effect - scoop electrons out of the lower Hubbard band and drop them into the upper band, and interorbital interaction effects can stabilize a new, metastable electronic structure that's a metal.
  • Alessandra Lanzara of Berkeley gave a really nice talk about her group's work on time-resolved angle-resolved photoemission.  You hit a material of interest with an ultrafast, time-resolved pump pulse of near-infrared light (1.5 eV photons), and then at some known delay you smack the system with a 6 eV probe pulse at a particular polarization and orientation, and measure the energy and momentum distribution of the electrons that get kicked out.  This lets you measure the transient electronic structure.  They've been able to use this approach to study the dynamics of quasiparticles in cuprate superconductors, how Cooper pairs respond to such pumping, etc.
  • N. Peter Armitage at Johns Hopkins articulated nicely three reasons to "go nonequilibrium":  to learn about elementary excitations of an equilibrium phase; to access "phases" not possible in equilibrium material configurations; and to look for new "phases" that have no equilibrium analog.  He then gave a fun talk about using optical spectroscopy techniques to look at many-body relaxations (older paper here) in the Coulomb glass phase of lightly doped semiconductors - when there are strongly interacting, localized electrons in a disordered configuration so that screening is poor.  Interestingly, these systems relax more slowly when the carrier densities get higher, in physics related to the orthogonality catastrophe
  • My faculty colleague Jun Kono from Rice spoke about so-called Dicke phenomena (such as superradiance, superfluorescence) in semiconductors.  These effects are great examples of nonequilibrium physics, when a driven system (say a semiconductor in a magnetic field illuminated by THz radiation that spans the energy scale of the cyclotron resonance, \(\omega_{\mathrm{c}} = e B/m^{*}\)) spontaneously develops coherence among the many electron-hole excitations in the system.  You can put such a system in a clever kind of 1d optical cavity, and approach the "strong coupling" regime so that the energetic coupling between the charge carriers and the photons in the cavity is comparable to the cyclotron energy.
  • Christof Weitenberg from Hamburg then spoke about exciting results in simulating condensed matter systems using cold atoms in optical lattices.  One piece of physics that's very in vogue right now because of the rise of topology and various 2d materials is Berry curvature.  It's hard to explain this in brief - if you look at how the energy bands of a material as a function of crystal momentum \(E(\mathbf{k})\) are curved, the wavefunction of a particle traversing some closed trajectory in \(\mathbf{k}\)-space can pick up a phase factor related to that curvature.  In Weitenberg's experiments, cleverly arranged laser beams can create designer lattices.  Shaking the lasers periodically as a function of time can lead to the same Floquet physics discussed above, changing the effective band structure for atoms confined in those lattices, and through cool imaging techniques the experimentalists can reconstruct the Berry curvature that they have designed into that effective band structure.
  • Another colleague Kaden Hazzard from Rice gave a nice theoretical talk about different nonequilibrium collective phenomena in ultracold atomic matter.  One aspect involved dilute molecules with electric dipoles (KBr) trapped in an optical lattice.  Because of their dipole moments, the molecules interact with each other over long ranges (dipole-dipole interactions scale like \(1/r^{3}\)), and their relaxation after getting dinged is governed by many-body interaction effects.  Another system is trapped Rydberg atoms, where dipolar interactions scale like the principal quantum number to the eleventh power (!).  
  • Andrea Cavalleri from the Max Planck in Hamburg (and also spending time at Oxford) spoke about his group's very high profile work that I've already described here.  The central question here is really can driving a quantum material stabilize collective states like superconductivity that have coherence, correlations, and remarkable physical properties that would be absent without the drive.  Both Cavalleri and Oka made reference to this video, which shows how driving a classical pendulum can render the inverted position of the pendulum stable.  The experiments themselves are truly remarkable.
  • In the last talk of Day 1, Sarang Gopalakrishnan of Cal Tech gave a theory talk again examining the response of driven many-body quantum systems, focusing particularly on the issue of many-body localization.  That is, when do the quantum dynamics of a many-body system lead to a real breakdown of quantum ergodicity, so that the degrees of freedom get "stuck", having large variability of local observables (instead of things being smoothed out and looking thermally smeared) and comparatively weak entanglement (which grows more slowly with system size than in the effectively thermal case).  He pointed out experimental challenges, that experiments probe dynamics rather than quantum eigenstates and that everything really is coupled (however weakly) to some thermal "bath", but argued that these issues aren't fatal to the interesting physics.

3 comments:

Anonymous said...

A naive question that constantly bugs me when I read about pump-probe experiments is that how do we know we are not just heating up the system. How do we keep sample cool when we dump a lot of energy in a small spot?

Douglas Natelson said...

Anon., that's not naive at all - it's a subtle point. Some ultrafast pump/probe experiments simply examine the system only on timescales that are so short that the energy injected by the "pump" pulse has only had time to get to the degrees of freedom seen by the "probe" pulse. On longer timescales (say tens to hundreds of nanoseconds) enough time has passed that the injected energy has cascaded down, with excited charge carriers scattering off each other and off of optical and acoustic phonons, and those phonons scattering off each other - the really macroscopically irreversible part of the heating process. In many pump/probe experiments, the time between repetitions of the pump/probe pulse sequence is long enough that this heat is conducted away, so that each pump pulse finds a fresh, cool system as its initial state. In the Cavalleri work, for example, they smack the system with near-IR light, which excites (via Raman transitions) a particular optical phonon of interest, and then they look at the transient electronic response via the optical reflectivity. If they don't put enough time between the NIR pulses, they do just heat the whole thing, and their optical probes only look at quite short timescales.

Anonymous said...

Thanks for the explanation and for taking time to write up the summary. Looking forward to the summary of Day 2.