2.2. Future information technologies
2.2.1. Magnetism dynamics
Controlling magnetic states of matter is crucial for understanding the underlying fundamental physical phenomena and for engineering the next-generation magnetic devices combining ultrafast data processing with ultrahigh-density data storage. The elementary building blocks governing magnetism are the magnetic moment, the spin-orbit coupling and the exchange interaction - the short-range force that couples the constituent spins. Thus, in order to generate, manipulate and eventually control such magnetic states of matter one needs to understand the dynamical behavior of these key parameters on their characteristic time and length-scales i.e. from femtoseconds to nanoseconds and from local spin moments to extended nanostructures on the nanometer length scale.
Being equally a science- and technologically-driven research area, the field of ultrafast magnetization dynamics has witnessed an intense activity, both experimentally and theoretically, leading to a rapid development over the past decade. Despite of this progress there are several fundamental, yet unanswered questions that form the ‘terra incognita’ of ultrafast magnetism, as for instance:
- What is the ultimate speed at which magnetic order can be manipulated and eventually controlled
- What are the microscopic processes responsible for magnetization dynamics on elementary length- and time-scales?
- What are the channels of ultrafast angular momentum transfer to and from the spin system?
Next to purely fundamental interest, the answers to these questions are essential for the development of future technologies for ultrafast and energy-saving recording and processing of magnetically stored information (Figure 6).
Figure 6. BESSY-VSR will provide novel and fundamental knowledge about the microscopic processes responsible for magnetization dynamics, which will allow for developing and designing novel materials and approaches that could boost the magnetic recording write/read rates of the magnetic bits to unprecedented speeds. (Courtesy I. Radu, HZB).
The outline below summarizes the scientific case for magnetism dynamics going from femtosecond phenomena at the atomic scale to nanosecond phenomena in extended structures at the nanometer length scale.
Ultrafast magnetism on elementary time and length scales
The use of femtosecond soft X-ray pulses with variable polarization, available at the femtosecond slicing facility FEMTOSPEX of BESSY II Berlin has provided unprecedented insights into the physics of ultrafast magnetism as demonstrated, for instance, by the latest investigations of various material systems [1-5]. Although groundbreaking, these measurements suffered due to the relatively low photon flux achieved in the slicing mode (106 photons/s) while momentum-resolved investigations were limited to specific samples showing high diffraction efficiency. Moreover, experiments imaging the nanoscale with fs time resolution, a crucial ingredient in understanding ultrafast magnetism, were impossible.
The envisioned capabilities of BESSY-VSR are indispensable in addressing all these fundamental questions, which will ultimately provide us with the required fundamental knowledge about the genuine magnetization and spin dynamics. In particular, using its high photon flux and short X-ray pulses we will obtain quantitative information on ultrafast dynamics of the spin and orbital magnetic moments as well as on the transient dynamics of the spin-orbit coupling and of the exchange interaction – the key ingredients governing the magnetic order. Moreover, BESSY-VSR will allow us to monitor the elementary length scales pertinent to magnetization dynamics in highly non-equilibrium states, i.e. comparable with the exchange length of 10 nm and below, using X-ray scattering techniques. In addition, imaging of the magnetic domains/macrospins dynamics at the nanoscale on picosecond to femtosecond timescales and with element specificity will become feasible with BESSY-VSR .
In addition to the fs-X-ray studies, the ultrashort electron bunches and their flexible time structure will provide us with a powerful spectroscopic tool covering the far-IR and THz spectral range. Such a tool will enable us to identify and disentangle the various quasiparticles (e.g. phonons, magnons) which are driving and/or involved the ultrafast magnetization dynamics on ultrashort timescales as described in the following.
Ultrafast magnetization dynamics upon resonant excitation of phonon and magnon modes
In a pilot experiment at the free-electron laser FLASH (Hamburg) a collaboration including I. Radu, M. Gensch, F. Radu, R. Abrudan, T. Kampfrath, and A. Kimel studied the ultrafast magnetization dynamics of magnetic garnets upon resonant excitation of the relevant phonon and magnon modes using intrinsically synchronized THz and X-ray pulses. the Terahertz radiation served as pump beam and the X-ray radiation was used to probe the system [6].
This is a new type of experiment with the aim to understand the role of spin and lattice excitations during the THz-induced magnetization dynamics; in particular during demagnetization and magnetization reversal processes. Similar experiments would be enabled BESSYVSR, as the intensity of THz radiation at BESSY-VSR will be very high and the THz and X-ray pulses will be intrinsically synchronized.
The combination of resonant THz excitation and resonant x-ray magnetic scattering at BESSY-VSR can be expected to open up a new field exploring phase transitions on ultrashort time scales in transient and metastable states. These non-equilibrium states may unleash new and exciting properties, as has recently been demonstrated in the case of ferrimagnetic GeFe alloys, which show a transient ferromagnetic state in an otherwise antiferromagnetically coupled system [4].
BESSYVSR will thus provide unique tools, both in the frequency and time domain, which will allow observing, understanding and ultimately controlling the fundamental interactions driving magnetic order at the nanoscale with femtosecond time resolution. These could be uniquely complemented with BESSY-VSR by investigations focusing on the ps time scale.
New frontiers in spin-dependent band mapping – Magnetization dynamics in lanthanides and their compounds
The magnetization dynamics of Gadolinium and Terbium have recently been analyzed by the group of M. Weinelt (FU Berlin) by time- and angle-resolved photoemission with VUV radiation from a HHG-source [7] (Figure 7) as well as using X-ray magnetic circular dichroism (XMCD) at the BESSY femtoslicing facility FEMTOSPEX [8]. The results indicate that the magnetic anisotropy plays an important role in ultrafast demagnetization.
Figure 7. Time- and angle-resolved photoemission result with VUV radiation from a HHG-source s on the magnetization dynamics of Gadolinium and Terbium [7]. An asymmetric collapse of the exchange splitting of the valence bands is observed; the upshift of the majority band (blue) is delayed by one picosecond (ps) with respect to the instantaneous response of the minority spin band (red). This indicates that Gd valence and 4f spin-systems are out of equilibrium within the first few picoseconds after laser excitation. (Courtesy M. Weinelt, FU Berlin).
Using the spin-filter momentum microscope that is currently being built for installation at the Russian-German beamline at BESSY II will offer the opportunity to simultaneously analyze the transient, spin-polarized band structure and Fermi surface in 4 dimensions, i.e. for parallel momenta kx and ky, at a given binding energy and delay time. Thereby, it will be possible to discriminate between spin transport and spin-flip phenomena and unravel the controversial details of laser-driven magnetization dynamics. At BESSY-VSR , the repetition rate of short 1.5 ps pulses in the low MHz regime fits perfectly well to the acceptance of the delay-line detector used in the time-of-flight spectrometer. Together with the highly efficient spin detector, the described experiment will make accessible new frontiers in spin-dependent band mapping. The reduced pulse length down to 1.5 ps will allow for time-resolved measurements with dramatically increased flux and availability compared to the current situation at BESSY II.
The analysis of the valence band-structure of ferromagnets will be fostered with time-resolved measurements of linear and circular dichroism in 4f (resonant) photoemission [9] and X-ray scattering at the lanthanide M-edges. Both techniques are element specific and thus generally applicable to compounds and multilayer systems, i.e., material systems of technological relevance. All of the above experiments will greatly profit from the increased flux and ps time-resolution of the proposed BESSY-VSR mode.
Precessional Magnetization Dynamics in Complex Magnetic Layer Structures and Spintronic Materials
Complex magnetic layer stacks are the building blocks of spintronics devices, such as magnetic tunneling junctions, spin transistors, and they play an important role in the magnetic data storage or processing such as in complex circuits and logical gates. The dynamic response of these layer stacks to external stimuli, such as magnetic field pulses, spin polarized currents or light pulses differs considerably from the single layer response. It is determined by a complicated interplay of interactions and magnetic coupling mechanisms in the stack. Such structures in fact take advantage of the combination of the different magnetic properties of the different coupled layers either by direct or indirect exchange coupling, or of magnetoresistive effects. The influence of a certain coupling mechanism such as the interlayer exchange coupling, e.g., may depend strongly on the time scale of the magnetic excitation. The transient change in magnetic properties following the optical excitation of the electronic system can lead to a change in the magnetization state, either by magnetization precession or by all-optical magnetization reversal. An understanding of the dynamic response of the stack therefore requires an understanding of the behavior of the individual magnetic and nonmagnetic entities in the layered structure.
Figure 8. Time-resolved photoemission microscopy (TR-PEEM) used to disentangle the individual magnetic responses of the different layers, NiFe and CoFe, in a stack. (Courtesy C. M. Schneider, Forschungszentrum Jülich).
Typically the fastest magnetization reversal is obtained by precessional motion in the 100 ps regime, but even faster modes may become within reach with ultrafast laser pulse excitation. The transient response of the system can be studied by pump-probe experiments on an element-resolved basis by employing the XMCD effect. Such experiments can either laterally integrate over a wide area of the sample, or microscopically image the sample as in PEEM. The latter has the advantage that also local effects, caused by variations in the local effective field by, for example, domain walls, may be studied.
Time-resolved photoemission microscopy (TR-PEEM) has proven to be a very powerful technique to disentangle the individual magnetic responses of the different layers in a stack. The use of aberration-corrected electron optics enhances the transmission and increases the sensitivity of this approach. The group of C. M. Schneider (Forschungszentrum Jülich) recently implemented a temporal gating and a laser-based excitation in their aberration-corrected PEEM to enable time-resolved studies of magnetization dynamics (Figure 8). In earlier studies they already demonstrated a best time-resolution of 10 ps in the low-alpha mode, which allowed them to directly map precessional modes in simple Permalloy elements.
BESSY-VSR will offer much better conditions for time-resolved magnetodynamic imaging down to the picosecond limit and may even provide a bridge to the sub-ps regime to access the time span shortly after the excitation with TR-PEEM. This would allow identifying the mechanisms starting the reversal mechanisms on an element-resolved, and thus layer-resolved basis. Compared to the femtosecond slicing facility FEMTOSPEX at BESSY II, BESSY-VSR would allow for higher flux at short time scales, thus enabling imaging experiments with, e.g., TR-PEEM.
Another example of precessional dynamics by the group of H. Zabel (Universität Bochum) addresses the free precessional dynamics and damping of spins in spintronic materials (Figure 9).
Figure 9. Time resolved precessional dynamics of Fe in a Py/Cu/Fe spin valve determined by pump-probe methods at the Fe L3 - edge. Taken from ref. 14, R. Salikhov et al., Phys. Rev. B 86, 144422 (2012).
From these experiments, which were performed at BESSY II, the g-factor and the Gilbert damping was determined in an element-selective way and as a function of external magnetic field, opening angle, temperature, doping concentration, etc. [10-14]. The bunch length at BESSY II determined the time resolution. With present pulse length of about 50 ps in single bunch mode at BESSY II it was possible to resolve the precessional frequency but the investigations are limited to magnetic materials with low exchange coupling and therefore low precessional frequencies of below 1-2 GHz. Because of this, Fe and Ni spins in Py can very well be probed, but not, for instance, Fe spins in Fe, which has much higher exchange energy. A pulse length of 15 ps or even down to 300 fs as provided by BESSY-VSR would greatly enhance the frequency range that we can probe. This in turn, would allow investigating the free precession and the damping in a much wider range of materials, including those with high relevance for magnetic storage devices, such as FePtx and CoPtx alloys and the spin precession of magnetic quantum dots. The higher repetition rate and higher current would, in addition, greatly expedite the described experiments.
Time-Resolved Investigations of Resistive Switching Dynamics
Nonvolatile memory concepts involving resistive switching phenomena promise the ultimate scaling behavior. The microscopic mechanisms behind resistive switching and particularly the dynamics are far from being fully understood. Depending on the material system, the resistive switching involves either crystalline phase changes, the formation of metallic filaments, or local valence changes (Figure 10). Disentangling the physical processes, which govern the voltage time-dilemma and determining the limits of the switching speed are pivotal for a future use of RRAM’s in microelectronics. For this purpose, time-resolved studies in the nano- and picosecond regime are needed.
Figure 10. Soft x-ray PEEM investigations of chemical changes in Fe:SrTiO3 – a prototypical resistive switching material – after removal of the top electrode used for electroforming. In the switched region (“forming crater”) the photoemission spectra of the Sr 3d states exhibit a significant change, as compared to the surrounding area. This spectral signature can be used for time-resolved experiments. (Courtesy C. M. Schneider, Forschungszentrum Jülich).
The experimental challenge in time-resolved studies of the resistive switching dynamics is to obtain precise chemical information from a buried layer underneath a metallic electrode with high lateral resolution. This requires a combination of hard x-ray photoemission with photoemission microscopy, i.e. hard x-ray photoemission (HAXPEEM). Since preliminary investigations suggest the dynamics to take place on the subnanosecond time scale, a time-resolution in the low picosecond regime will be needed to address the time-dependent changes in the electronic/chemical structure. Based on state-of-the-art time-resolved PEEM and HAXPEEM results at BESSY II a dedicated pump-probe experiment at BESSY-VSR would allow for mapping the chemical changes underneath the top electrode taking place during the application of short current/voltage pulses. The changes will be determined by means of hard x-ray photoemission from characteristic core levels in the resistive switching material.
The dynamics of information storage – Going to extended nanostructures
One example for the future use of variable-duration pulses from BESSY-VSR for studying the dynamics of magnetic data storage and manipulation is the investigation of magnetic bubble states. Such a circular magnetic domain in perpendicular anisotropy material is shown in Figure 11.
Figure 11. (a) Two-bubble state in a circular thin film medium of 550 nm diameter with perpendicular magnetic anisotropy, imaged via Fourier Transform holography with XMCD contrast at BESSY II in single bunch mode. The trajectory of the bubble indicated by an arrow is traced. (b) Trajectory of the bubble displacement (data points) after excitation via the magnetic field of a microcoil together with a theoretical model (line). The entire trajectory has a duration of 15 ns (color coded purple to yellow). (Courtesy S. Eisebitt, Technische Universität Berlin).
The image was obtained via x-ray holography, using the high brightness of BESSY II to record a hologram of the sample with magnetic contrast. Such magnetic bubbles have topologic properties that make them of interest for both basic science research and future information storage units. To study the dynamic properties of this system, it was crucial to initially bring the system into a well-defined two bubble state, close to nucleation of an additional domain. In this study, this was achieved via the (iterative) adjustment of the external magnetic field and analysis of the resulting state via holography. As such adjustments in a (in general multi-dimensional) parameter space are time consuming, the ability to quickly characterize the resultant state is crucial. Here, multibunch acceptance of all pulses of BESSY-VSR would enable this steering of the system to the desired initial state. Once prepared, BESSY-VSR will allow switching to the specific temporal resolution required to study the dynamic phenomenon in question. In the example here, the magnetic bubble was excited via an external field pulse administered via a microcoil, and the resulting dynamics of the bubble was traced holographically. The results shown here were achieved very tediously in single bunch mode and allow e.g. the determination of the topological mass of the soliton-like bubble [AG Eisebitt et al, in preparation]. BESSY-VSR will enable much more efficient preparation of the desired exited state of a system to be studied, even in more complex phase diagrams (B,T, sample composition), and thus enable new studies of intrinsic an controlled dynamic behavior, e.g. in nanomagnetic systems for information storage and processing.
References for section 2.2.1.
[1] C. Stamm et. al., Nature Materials 6, 740 (2007).
[2] K. Holldack et. al., Appl. Phys. Lett. 97, 62502 (2010).
[3] C. Boeglin et. al., Nature 465, 458 (2010).
[4] I. Radu et. al., Nature 472, 205 (2011).
[5] A. Eschenlohr et. al., Nature Materials 12, 332 (2013).
[6] I. Radu, M. Gensch, F. Radu, R. Abrudan, T. Kampfrath, A. Kimel, FLASH Proposal 2012.
[7] R. Carley, K. Döbrich, B. Frietsch, C. Gahl, M. Teichmann, O. Schwarzkopf, P. Wernet, M. Weinelt, Phys. Rev. Lett.
109, 057401(2012).
[8] M. Wietstruk, A. Melnikov, C. Stamm, T. Kachel, N. Pontius, M. Sultan, C. Gahl, M. Weinelt, H. A. Dürr, U.
Bovensiepen, Phys. Rev. Lett. 106, 127401 (2011).
[9] A. Melnikov, H. Prima-Garcia, M. Lisowski, T. Gießel, R. Weber, R. Schmidt, C. Gahl, U. Bovensiepen, M. Weinelt, Phys.
Rev. Lett. 100, 107202 (2008).
[10] S. Buschhorn, F. A. Brüssing, R. Abrudan, H. Zabel, J. Synchrotron Rad. 18, 212 (2011).
[11] S. Buschhorn, F. Brüssing, R. Abrudan, H. Zabel, J. of Physics D: Applied Physics 44, 165001 (2011).
[12] R. Salikhov, R. Abrudan, F. Brüssing, St. Buschhorn, M. Ewerlin, D. Mishra, F. Radu, I. A. Garifullin, H. Zabel, Appl. Phys.
Lett. 99, 092509 (2011).
[13] S. Buschhorn, F. Brüssing, M. Ewerlin, R. Salikhov, R. Abrudan, H. Zabel, J. Appl. Phys. 112, 103914 (2012).
[14] R. Salikhov, R. Abrudan, F. Brüssing, K. Gross, C. Luo, K. Westerholt, and H. Zabel, F. Radu, I. A. Garifullin, Phys. Rev. B
86, 144422 (2012).