Spin and orbital angular momenta are two intrinsic properties of an electron and are responsible for the physics of a solid. How the spin and orbital evolve with respect to each other on several hundred femtoseconds is largely unknown, but it is at t
he center of laser-induced ultrafast demagnetization. In this paper, we introduce a concept of the spin-orbital correlation diagram, where spin angular momentum is plotted against orbital angular momentum, much like the position-velocity phase diagram in classical mechanics. We use four sets of highly accurate time-resolved x-ray magnetic circular dichroism (TR-XMCD) data to construct four correlation diagrams for iron and cobalt. To our surprise, a pattern emerges. The trace on the correlation diagram for iron is an arc, and at the end of demagnetization, it has a pronounced cusp. The correlation diagram for cobalt is different and appears more linear, but with kinks. We carry out first-principles calculations with two different methods: time-dependent density functional theory (TDDFT) and time-dependent Liouville density functional theory (TDLDFT). These two methods agree that the experimental findings for both Fe and Co are not due t experimental errors. It is the spin-orbit coupling that correlates the spin dynamics to the orbital dynamics.Microscopically, Fe and Co have different orbital occupations, which leads to distinctive correlation diagrams. We believe that this correlation diagram presents a useful tool to better understand spin and orbital dynamics on an ultrafast time scale. A brief discussion on the magnetic anisotropy energy is also provided.
Laser-induced ultrafast demagnetization has puzzled researchers around the world for over two decades. Intrinsic complexity in electronic, magnetic, and phononic subsystems is difficult to understand microscopically. So far it is not possible to expl
ain demagnetization using a single mechanism, which suggests a crucial piece of information still missing. In this paper, we return to a fundamental aspect of physics: spin and its change within each band in the entire Brillouin zone. We employ fcc Ni as an example and use an extremely dense {bf k} mesh to map out spin changes for every band close to the Fermi level along all the high symmetry lines. To our surprise, spin angular momentum at some special {bf k} points abruptly changes from $pm hbar/2$ to $mp hbar/2$ simply by moving from one crystal momentum point to the next. This explains why intraband transitions, which the spin superdiffusion model is based upon, can induce a sharp spin moment reduction, and why electric current can change spin orientation in spintronics. These special {bf k} points, which are called spin Berry points, are not random and appear when several bands are close to each other, so the Berry potential of spin majority states is different from that of spin minority states. Although within a single band, spin Berry points jump, when we group several neighboring bands together, they form distinctive smooth spin Berry lines. It is the band structure that disrupts those lines. Spin Berry points are crucial to laser-induced ultrafast demagnetization and spintronics.
Harmonic generation in atoms and molecules has reshaped our understanding of ultrafast phenomena beyond the traditional nonlinear optics and has launched attosecond physics. Harmonics from solids represent a new frontier, where both majority and mino
rity spin channels contribute to harmonics.} This is true even in a ferromagnet whose electronic states are equally available to optical excitation. Here, we demonstrate that harmonics can be generated {mostly} from a single spin channel in half metallic chromium dioxide. {An energy gap in the minority channel greatly reduces the harmonic generation}, so harmonics predominantly emit from the majority channel, with a small contribution from the minority channel. However, this is only possible when the incident photon energy is well below the energy gap in the minority channel, so all the transitions in the minority channel are virtual. The onset of the photon energy is determined by the transition energy between the dipole-allowed transition between the O-$2p$ and Cr-$3d$ states. Harmonics {mainly} from a single spin channel can be detected, regardless of laser field strength, as far as the photon energy is below the minority band energy gap. This prediction should be tested experimentally.
High harmonic generation (HHG) has unleashed the power of strong laser physics in solids. Here we investigate HHG from a large system, solid C$_{60}$, with 240 valence electrons engaging harmonic generation at each crystal momentum, the first of this
kind. We employ the density functional theory and the time-dependent Liouville equation of the density matrix to compute HHG signals. We find that under a moderately strong laser pulse, HHG signals reach 15th order, consistent with the experimental results from C$_{60}$ plasma. The helicity dependence in solid C$_{60}$ is weak, due to the high symmetry. In contrast to the general belief, HHG is unsuitable for band structure mapping in C$_{60}$. However, we find a window of opportunity using a long wavelength, where harmonics are generated through multiple-photon excitation. In particular, the 5th order harmonic energies closely follow the transition energy dispersion between the valence and conduction bands. This finding is expected to motivate future experimental investigations.
Here we carry out a first-principles time-dependent calculation to investigate how fast electrons actually move under laser excitation and how large the electron transport affects demagnetization on the shortest time scale. To take into account the t
ransport effect, we implement the intraband transition in our theory. In the bulk fcc Ni, we find the effect of the spin transport on the demagnetization is extremely small, no more than 1%. The collective electron velocity in Ni is 0.4 $rm AA/fs$, much smaller than the Fermi velocity, and the collective displacement is no more than 0.1 $rm AA$. But this does not mean that electrons do not travel fast; instead we find that electron velocities at two opposite crystal momenta cancel each other. We follow the $Gamma$-X line and find a huge dispersion in the velocities in the crystal momentum space. In the Fe/W(110) thin film, the overall demagnetization is larger than Ni, and the Fermi velocity is higher than Ni. However, the effect of the spin transport is still small in the Fe/W(110) thin film. Based on our numerical results and existing experimental findings, we propose a different mechanism that can explain two latest experimental results. Our finding sheds new light on the effect of ballistic transport on demagnetization.
All-optical spin reversal presents a new opportunity for spin manipulations, free of a magnetic field. Most of all-optical-spin-reversal ferromagnets are found to have a perpendicular magnetic anisotropy (PMA), but it has been unknown whether PMA is
necessary for the spin reversal. Here we theoretically investigate magnetic thin films with either PMA or in-plane magnetic anisotropy (IMA). Our results show that the spin reversal in IMA systems is possible, but only with a longer laser pulse and within a narrow laser parameter region. The spin reversal does not show a strong helicity dependence where the left- and right-circularly polarized light lead to the identical results. By contrast, the spin reversal in PMA systems is robust, provided both the spin angular momentum and laser field are strong enough while the magnetic anisotropy itself is not too strong. This explains why experimentally the majority of all-optical spin-reversal samples are found to have strong PMA and why spins in Fe nanoparticles only cant out of plane. It is the laser-induced spin-orbit torque that plays a key role in the spin reversal. Surprisingly, the same spin-orbit torque results in laser-induced spin rectification in spin-mixed configuration, a prediction that can be tested experimentally. Our results clearly point out that PMA is essential to the spin reversal, though there is an opportunity for in-plane spin reversal.
The exchange interaction among electrons is one of the most fundamental quantum mechanical interactions in nature and underlies any magnetic phenomena from ferromagnetic ordering to magnetic storage. The current technology is built upon a thermal or
magnetic field, but a frontier is emerging to directly control magnetism using ultrashort laser pulses. However, little is known about the fate of the exchange interaction. Here we report unambiguously that photoexcitation is capable of quenching the exchange interaction in all three $3d$ ferromagnetic metals. The entire process starts with a small number of photoexcited electrons which build up a new and self-destructive potential that collapses the system into a new state with a reduced exchange splitting. The spin moment reduction follows a Bloch-like law as $M_z(Delta E)=M_z(0)(1-{Delta E}/{Delta E_0})^{frac{1}{beta}}$, where $Delta E$ is the absorbed photon energy and $beta$ is a scaling exponent. A good agreement is found between the experimental and our theoretical results. Our findings may have a broader implication for dynamic electron correlation effects in laser-excited iron-based superconductors, iron borate, rare-earth orthoferrites, hematites and rare-earth transition metal alloys.
Superatomic molecular orbitals (SAMO) in C60 are ideal building blocks for functional nanostructures. However, imaging them spatially in the gas phase has been unsuccessful. It is found experimentally that if C60 is excited by an 800-nm laser, the ph
otoelectron casts an anisotropic velocity image, but the image becomes isotropic if excited at a 400-nm wavelength. This diffuse image difference has been attributed to electron thermal ionization, but more recent experiments (800 nm) reveal a clear non-diffuse image superimposed on the diffuse image, whose origin remains a mystery. Here we show that the non-diffuse anisotropic image is the precursor of the $f$ SAMO. We predict that four 800-nm photons can directly access the $1f$ SAMO, and with one more photon, can image the orbital, with the photoelectron angular distribution having two maxima at 0$^circ$ and 180$^circ$ and two humps separated by 56.5$^circ$. Since two 400-nm photons only resonantly excite the spherical $1s$ SAMO and four 800-nm photon excite the anisotropic $1f$ SAMO, our finding gives a natural explanation of the non-diffuse image difference, complementing the thermal scenario.
