WorksPhysics and Astronomy
State-resolved ultrafast charge and spin dynamics in [Co/Pd] multilayers
Abstract
We use transient absorption spectroscopy with circularly polarized x-rays to detect laser-excited hole states below the Fermi level and compare their dynamics with that of unoccupied states above the Fermi level in ferromagnetic [Co/Pd] multilayers. While below the Fermi level an instantaneous and significantly stronger demagnetization is observed, above the Fermi level the demagnetization is delayed by 35+/-10 fs. This provides a direct visualization of how ultrafast demagnetization proceeds via initial spin-flip scattering of laser-excited holes to the subsequent formation of spin waves.
Fig. 2 shows transient XAS and XMCD of laser-induced holes below and above the Fermi level. Figure 2(a) XAS and Figure 2(b) XMCD spectra were measured at a fixed time delay of 0.4 ps at the Co L edge. The pump induced changes are shown as green symbols and shading. While the change in XMCD appears to be mostly an homogeneous reduction at all photon energies, the change in XAS clearly displays a derivative-like shape with a zero crossing at an x-ray energy of 777.2 eV (see top axis of Figure 2) as indicated by the dashed vertical line. At lower x-ray energy the XAS signal is increased as expected for fs laser-induced hole states. At higher energy XAS transitions into previously unoccupied states are blocked by laser-excited electrons leading to the observed intensity reduction. It is, therefore, possible to identify 777.2 eV as the position of the Fermi level (see bottom axis of Figure 2).41

Figure 2. (Color online) Pumped, unpumped and their differences in (a) XAS and (b) XMCD at the Co L edge at a delay of 0.4 ps. The vertical dashed line indicates the position of the observed zero crossing at the Fermi level . The photon energy is shown on the top axis, while the energy with respect to the Fermi level is shown on the bottom. The differences (pumped-unpumped) are shown on a separate vertical axis on the right. For the XMCD difference the sign was reversed to ease visual comparison with the unpumped XMCD profile.
Figure 3 displays time-delay traces obtained for various state energies relative to the Fermi level, . Below the curves display initial increases in the XAS intensity followed by subsequent decays on timescales longer than several 100 fs. Above the transient XAS changes are negative while directly at a more complex behavior emerges. In the following we describe these observations in terms of changes in the hole population, , at state of energy . The small contribution due to spin-orbit coupling of the state-resolved XAS intensity42,43 will be neglected here. It is important to emphasise that can also include time dependent changes in the electronic structure.44 It is apparent in Figure 2(a) that such electronic structure changes indeed occur. For instance at near 4 eV, i.e. much higher than the pump photon energy, the observed variations of are unlikely to be caused by the population dynamics of electrons in these states.

Figure 3. (Color online) Time-resolved change in state resolved (a) charge and (b) relative polarization change around the at the edge. In (b), the data are shifted vertically for clarity and the gray dashed curves are the fit as explained in the text at eV.
The curves for in Figure 3(a) were fitted with a double exponential to describe an excitation and a relaxation process. The fit parameters are summarized in Table II in the supplementary information. Far above and below the Fermi level, the initial rise times of are essentially determined by the length of the pump pulses. The subsequent decay time scales are shorter further away from the Fermi level as one would expect from a Fermi liquid behavior of the electronic system.
The XMCD spectra can be described as being proportional to the product of state-dependent population, and a polarization term, . The latter contains both spin and orbital polarization with the orbital contribution being significantly smaller than the spin polarization as shown in the sum rule analysis detailed in the supplementary information.42,43 Similar to conventional sum-rule-analysis of time-resolved XMCD spectra, the magnetic dipole term can be neglected for our poly-crystalline samples45,46 and was found to be negligible in similar samples.47 Using the results for from Figure 3(a) we can separate the state polarization from state-dependent charge dynamics. The time-resolved polarization dynamics, normalized to the ground-state polarization of the respective states, are shown in Figure 3(b). The individual experimental results (symbols) are shifted vertically for clarity and are compared to exponential polarization decays that include the magnitude of the decay, , the decay time constant, , and a delayed demagnetization onset, , as fit parameters. All parameters are summarized in Table II in the supplementary information. To highlight the difference between the curves below and above the Fermi level, the same relative polarization dynamics for = 0.88 eV is shown as a dashed grey curve together with each trace. This allows two visual observations. Firstly, the amount of demagnetization is not the same
at each value of . Clearly, the demagnetisation is significantly stronger below than above. Secondly, there is a time delay, , apparent in the response, with faster dynamics for states below . The complete fitting model and analysis of the uncertainties on the fitted values is presented in Table II in the supplementary information. In Figure 4 the relative change in polarization and the time lag are shown as function of . We stress here that this time lag is determined by the delayed apparent response of the relative change in polarization with respect to the charge dynamics , for the same given photon energy. It can also be visualized by comparing the relative change in polarization at different photon energy but this is not how we extracted it.

Figure 4. (Color online) Fitted relative change in polarization and time lag in polarization change response as function of at edge.
The data we presented in this letter conclusively demonstrate a vastly different magnetization dynamics above and below the Fermi level for Co 3d levels in [Co/Pd] multilayers. Below , the ultrafast drop in magnetic polarization is up to 32% larger than above (see Figure 4). This is clearly outside any experimental uncertainty as demonstrated in Figure 3(b). Moreover, the onset of the polarization dynamics occurs simultaneously to the charge dynamics, i.e. fs. This is the behavior expected for individual electrons/holes being scattered between different electronic states as depicted in Figure 1(a). This also leads to Stoner excitations where electrons/holes are scattered between the spin up and down states.28 In strong ferromagnets such as [Co/Pd] multilayers, spin-flip scattering can only occur for states below where spin up and down states are hybridized via spin-orbit coupling.31 The same Elliot-Yaffet-type spin-flip scattering processes are thought to also transfer spin angular momentum to the lattice.29–31
Figure 1. (Color online) (a) Schematic of the experiment where the unoccupied spin-resolved density of states (DOS) are probed by core-level absorption spectroscopy. Upon excitation by a femtosecond laser pulse, electrons are promoted from below to above the Fermi level, , in a spin-conserving process (purple arrow). In a strong ferromagnet such as [Co/Pd], spin relaxation can only occur below by a hole spin-flip (green arrow). (b) After the localized hole spin-flip excitation, spin-waves are generated and correspondingly the spin-resolved DOS are partially mirrored.
However, for ultrafast demagnetization to occur, the flipped spins of individual electrons/holes need to be transferred to the whole electronic system. This usually takes place via the formation of collective spin excitations, i.e. spin waves. In the Heisenberg model, spin waves lead to slight changes of the atomic spin quantization axis and result in a mixing of spin up and down states as observed in photoemission spectroscopy.48 This situation is depicted in Figure 1(b). Since the formation of spin waves takes time34 we expect a characteristic time delay relative to the instantaneous demagnetization of individual electrons/holes. We assign the observed delayed onset of demagnetization above of fs (see Figure 4) to this effect. This is observable above the Fermi level since there the majority of unoccupied states reflects the atomic magnetic moments.49,50
In summary, taking advantage of the high FEL brightness and improved normalization scheme, we were able to show that time-resolved XAS and XMCD spectroscopy can provide detailed information in the microscopic mechanism at play during ultrafast laser excitation. In particular we report on different dynamics of the spin system below and above the Fermi level. This is manifested by both a 32% larger change in spin dynamics below the Fermi level and a 35 ± 10 fs delayed response above it. Both of these effects suggest a scenario for a strong ferromagnet where spin-flips occur preferentially below the Fermi level where spin up and down states are hybridized. Moreover, we also report on initial evidence that indicates effects beyond a simple electronic redistribution and demagnetization with changes in XAS observed 4 eV above the Fermi level, suggestive of band structure dynamics. With ever improved normalization schemes and higher repetition rate FELs, transient near-edge soft X-ray spectroscopy promises to be a valuable tool in understanding out-of-equilibrium phenomena.
Supplementary Material
See supplementary material for the sample preparation, sum-rules analysis, timing drift correction, pump and probe absorption profiles in the sample and finally the complete fitting model with uncertainty estimation.
Acknowledgments
L.L.G. acknowledges the Volkswagen-Stiftung for the financial support through the Peter-Paul-Ewald Fellowship. Work at SLAC and the operation of LCLS are supported by the U.S. Department of Energy, Office of Science.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
References
- [1]¹A. Kirilyuk, A. V. Kimel, and T. Rasing, “Ultrafast optical manipulation of magnetic order,” Reviews of Modern Physics 82, 2731–2784 (2010).DOI
- [2]2F. Hellman, A. Hoffmann, Y. Tserkovnyak, G. S. Beach, E. E. Fullerton, C. Leighton, A. H. MacDonald, D. C. Ralph, D. A. Arena, H. A. Dürr, P. Fischer, J. Grollier, J. P. Heremans, T. Jungwirth, A. V. Kimel, B. Koopmans, I. N. Krivorotov, S. J. May, A. K. Petford-Long, J. M. Rondinelli, N. Samarth, I. K. Schuller, A. N. Slavin, M. D. Stiles, O. Tchernyshyov, A. Thiaville, and B. L. Zink, “Interface-induced phenomena in magnetism,” Reviews of Modern Physics 89 (2017), 10.1103/revmodphys.89.025006.
- [3]C. D. Stanciu, F. Hansteen, A. V. Kimel, A. Kirilyuk, A. Tsukamoto, A. Itoh, and T. Rasing, “All-optical magnetic recording with circularly polarized light,” Physical Review Letters 99 (2007), 10.1103/physrevlett.99.047601.pubmed.ncbi.nlm.nih.gov/17678404
- [4]T. A. Ostler, J. Barker, R. F. L. Evans, R. W. Chantrell, U. Atxitia, O. Chubykalo-Fesenko, S. E. Moussaoui, L. L. Guyader, E. Mengotti, L. J. Heyderman, F. Nolting, A. Tsukamoto, A. Itoh, D. Afanasiev, B. A. Ivanov, A. M. Kalashnikova, K. Vahaplar, J. Mentink, A. Kirilyuk, T. Rasing, and A. V. Kimel, “Ultrafast heating as a sufficient stimulus for magnetization reversal in a ferrimagnet,” Nature Communications 3, 666 (2012).DOI
- [5]S. Mangin, M. Gottwald, C.-H. Lambert, D. Steil, V. Uhlíř, L. Pang, M. Hehn, S. Alebrand, M. Cinchetti, G. Malinowski, Y. Fainman, M. Aeschlimann, and E. E. Fullerton, “Engineered materials for all-optical helicity-dependent magnetic switching,” Nature Materials 13, 286–292 (2014).DOI
- [6]C.-H. Lambert, S. Mangin, B. S. D. C. S. Varaprasad, Y. K. Takahashi, M. Hehn, M. Cinchetti, G. Malinowski, K. Hono, Y. Fainman, M. Aeschlimann, and E. E. Fullerton, “All-optical control of ferromagnetic thin films and nanostructures,” Science 345, 1337–1340 (2014).
- [7]M. Battiato, K. Carva, and P. M. Oppeneer, “Superdiffusive spin transport as a mechanism of ultrafast demagnetization,” Phys. Rev. Lett. 105, 027203 (2010).pubmed.ncbi.nlm.nih.gov/20867735
- [8]M. Battiato, K. Carva, and P. M. Oppeneer, “Theory of laser-induced ultrafast superdiffusive spin transport in layered heterostructures,” Phys. Rev. B 86, 024404 (2012).DOI
- [9]B. Pfau, S. Schaffert, L. Müller, C. Gutt, A. Al-Shemmary, F. Büttner, R. Delaunay, S. Düsterer, S. Flewett, R. Frömter, J. Geilhufe, E. Guehrs, C. Günther, R. Hawaldar, M. Hille, N. Jaouen, A. Kobs, K. Li, J. Mohanty, H. Redlin, W. Schlotter, D. Stickler, R. Treusch, B. Vodungbo, M. Kläui, H. Oepen, J. Lüning, G. Grübel, and S. Eisebitt, “Ultrafast optical demagnetization manipulates nanoscale spin structure in domain walls,” Nature Communications 3 (2012), 10.1038/ncomms2108.DOI
- [10]D. Rudolf, C. La-O-Vorakiat, M. Battiato, R. Adam, J. M. Shaw, E. Turgut, P. Maldonado, S. Mathias, P. Grychtol, H. T. Nembach, T. J. Silva, M. Aeschlimann, H. C. Kapteyn, M. M. Murnane, C. M. Schneider, and P. M. Oppeneer, “Ultrafast magnetization enhancement in metallic multilayers driven by superdiffusive spin current,” Nature Communications 3 (2012), 10.1038/ncomms2029.
