Volt-per-Ångstrom terahertz fields from X-ray free-electron lasers.

A rather modest upgrade of the GeV-level linear accelerator of modern X-ray free-electron lasers with purpose-built THz undulators allows generating intense, tunable and narrow-bandwidth THz pulses alongside the X-ray pulses. The enabling technological breakthroughs in accelerator physics are discussed and the arising new research possibilities are presented.

The electron linear accelerators driving modern X-ray free-electron lasers can emit intense, tunable, quasi-monochromatic terahertz (THz) transients with peak electric fields of V Å À1 and peak magnetic fields in excess of 10 T when a purpose-built, compact, superconducting THz undulator is implemented. New research avenues such as X-ray movies of THz-driven mode-selective chemistry come into reach by making dual use of the ultra-short GeV electron bunches, possible by a rather minor extension of the infrastructure.
X-ray free-electron lasers [XFELS; see, for example, Pellegrini (2016) for a recent review] are currently the brightest, tunable sources of short X-ray pulses available for basic scientific research. Various techniques based on X-ray scattering, X-ray diffraction, and X-ray spectroscopies enable element-specific probing of dynamical processes in materials on the timescales from femtoseconds to nanoseconds and/or with a spatial resolution down to the Å ngstrom level (Dunne, 2018;Marx, 2017;Fromme, 2015). As outlined in this Short communication, the kilometre-long linear accelerators (linacs) driving these facilities are also capable of producing narrowband, frequency-tunable terahertz (THz) transients with V Å À1 -level THz fields.
The THz generation is based on superradiant emission from GeV-scale electron bunches in a specifically designed THz undulator positioned between the X-ray undulators and the electron beam dump (see Fig. 1). A first prototype of such a scheme is operational in the electron linac of the free-electron laser FLASH (Gensch et al., 2008) and enables one to probe THz-driven processes in the XUV spectral range. However, implementation into the kilometre-long, GeV-level linacs of hard X-ray FELs was until recently considered unfeasible because of two major technological obstacles. Firstly, THz undulators based on conventional technology and sufficiently large gap would require an unreasonably long period length of 10 m or more. Secondly, the required beam transport over a few hundreds of meters would lead to unrealistic scenarios ISSN 1600-5775 such as complex all-optical X-ray delay lines. As we show here, several technological breakthroughs of recent years allow meanwhile to overcome these technological hurdles.
Taking the European XFEL with its particularly high beam energy as an example, we showed the feasibility of a compact eight-period undulator with a period length of 1 m, an on-axis peak field of 7.3 T and a magnetic gap of 50 mm, using NbTibased superconducting undulator technology (Casalbuoni et al., 2018;Tanikawa et al., 2019). The device has an overall magnetic length of 8 m and is hence by an order of magnitude shorter than normal conducting undulators based on copper coils or permanent magnets to produce the same frequency tuning range between 3 and 100 THz for the first harmonic. 1 The second obstacle, the significant delay of few tens of nanoseconds in the time of arrival of X-ray and THz pulses from the same electron bunch, can be overcome by recently successfully demonstrated double-bunch schemes (Zapolnova et al., 2018). In combination with the recent advances in pulseresolved detection and arrival-time measurements (see, for example, Kovalev et al., 2017;Bionta et al., 2014), THz and X-ray pulses can now be provided with a few femtoseconds timing precision despite independent transport beamlines of several 100 m (see Tanikawa et al., 2019, and references therein). Utilizing electron bunch forms calculated for the European XFEL allows one to predict the achievable pulse energies and electric and magnetic peak fields (Tanikawa et al., 2019).
As shown in Fig. 2, values of > 0.2 V Å À1 will be reached in a wide range of the emitted THz frequencies between 8 and 100 THz. A field strength of 0.5 V Å À1 can be achieved in a large part of the so-called molecular fingerprint region between 15 and 100 THz, which contains the vibrational and vibrational-rotational modes of highly relevant chemical bonds (e.g. C-H x , C-O, C O, NO 2 , NH 2 ), thereby opening up the opportunity to observe mode-selective chemistry in molecules and clusters (Zewail, 1980;Lee et al., 2012) in an X-ray molecular movie for the first time. Of fundamental relevance is the opportunity to investigate the molecular dynamics of water in the THz regime, where V Å À1 level THz pulses can induce structural non-equilibrium states, also extending across the recently discovered liquid-liquid phase transition (Kim et al., 2017). In solids, the provided THz frequency range covers the stretching and bending phonon modes of many metal-ion-oxygen bonds (e.g. Mn-O, Cu-O, Fe-O), which typically lie between 8 and 20 THz. When driven by sufficiently strong THz fields, these modes have been recently shown to play an important role in the control of transient electronic and magnetic phases in correlated quantum systems (Buzzi et al., 2018). Further, the achievable peak THz magnetic fields are in the few-Tesla range over the whole frequency spectrum from 3 to 100 THz. These field magnitudes are sufficient for selective excitation of magnetic resonances far into the nonlinear regime (Kampfrath et al., 2013), where the fastest predicted magnetization reversal mechanism (Tudosa et al., 2004), i.e. coherent magnetization reversal on a $ 1 ps timescale, is expected to occur. Combined with powerful THz streaking techniques, the V Å À1 THz electric fields allow to unravel short-wavelength FEL-induced correlated electron and nuclear dynamics in atoms, molecules, and nanoparticles on extreme time scales (Oelze et al., 2019).  Principal scheme (top) and expected pulse energies from the X-ray to THz regime (bottom) for the example of the European XFEL. Relativistic GeV ultra-short and highly charged electron bunches first generate X-ray pulses by self-amplified spontaneous emission (SASE) in soft and hard X-ray undulators in the few mJ regime (Pellegrini, 2016). After passing the X-ray undulator section the electron bunches enter an additional few-period THz undulator and generate tunable, narrowband THz pulses with up to few 100 mJ pulse energy by superradiant emission (Tanikawa et al., 2019). X-ray pulse energies are calculated at saturation for electron beam energies of 8.5, 10.5, 14 and 17.5 GeV and a bunch charge of 500 pC (Schneidmiller & Yurkov, 2017). (Note that when a certain peak current is fixed, the X-ray pulse duration, and therefore its energy, will scale roughly linearly with the charge. Deviations mainly occur due to the dependence of the electron beam parameters on the bunch charge. Currently, in operations electron energies of 11.5, 14 and 16.5 GeV with a charge of 250 pC are routinely used.) techniques such as hard X-ray angle-resolved photoemission spectroscopy. In summary, a modest extension by additional superconducting undulators and THz beamlines enables additional use of the GeV-scale electron beams in modern XFELs for the generation of V Å À1 -level THz fields. Thereby X-ray movies on femtosecond time and Å ngstrom length scales of long-debated phenomena become feasible, opening up completely new research avenues.

Funding information
VA acknowledges support from the Alexander von Humboldt Foundation, Germany. DT acknowledges the project 'Nonequilibrium dynamics in solids probed by terahertz fields' funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) -project number 278162697 -SFB 1242. TK acknowledges funding through the collaborative research center TRR 227 'Ultrafast spin dynamics' (project A05). SW was supported by European Union's Horizon 2020 research and innovation programme (Grant Agreement No. 758461), Spanish MINECO (Severo Ochoa grant SEV-2015-0522, SEV2015-0496) as well as Fundació Privada Cellex, and CERCA Programme/Generalitat de Catalunya.