1 Introduction

ELBE is an acronym for Electron Linear accelerator with high Brilliance and low Emittance and comprises a superconducting radiofrequency (RF) linear accelerator capable of operating in a continuous wave (CW) mode, and a suite of secondary radiation sources driven by the electron beam. Among these are the coherent infrared and THz photon sources FELBE and TELBE, which are the focus of this article. As a multipurpose machine, ELBE is also used to produce gamma photons (MeV bremsstrahlung), positrons, and neutrons for user experiments. However, those will not be discussed here. Figure 1 shows a schematic depiction of ELBE with all of the secondary sources and in Fig. 2 the entire ELBE facility is sketched in more detail. Note that there is one optical beamline (upper left) that brings the FEL radiation to the adjacent pulsed high-magnetic field laboratory (HLD) [1]. The far right part of the building contains petawatt laser systems used for electron and ion acceleration experiments [2]. These operate independently of the electron accelerator, but can be combined for Compton backscattering.

Fig.1
figure 1

Schematic layout of ELBE, including all types of secondary radiation

Full size image
Fig. 2
figure 2

Floor plan of the ELBE building indicating the laboratories and caves dedicated to different applications

Full size image

2 Present status

ELBE is based on a 40 MeV, 1 mA superconducting RF accelerator that can be operated in full CW mode. Two accelerator modules, each equipped with two TESLA RF cavities, are used. For radiation generation, two injectors with different parameters are available. A 250 keV thermionic DC gun followed by two-stage bunch compression is mainly used for FEL operation, positron and bremsstrahlung generation. The repetition rate is 13 MHz with a maximum bunch charge of 77 pC. An in-house developed superconducting RF photo injector is used for THz generation, but also for pulsed neutrons. The maximum bunch charge here is 250 pC with a maximum repetition rate of 500 kHz. The ELBE accelerator feeds several long-wavelength photon sources: two cavity-based free-electron laser (FEL) oscillators named FELBE and two single-pass, superradiant THz sources named TELBE.

FELBE is based on 11.53 m long optical cavities, corresponding to the pulse repetition rate of 13 MHz. Two undulators of 37 and 100 mm period (with acronym U37 and U100) serve as infrared/THz sources tunable from 5 to 40 µm and 18–250 µm wavelength, respectively, covering an overall frequency range of 1.2–60 THz or photon energies of 5–250 meV. First lasing was reported in 2004. Typical pulse energies reach up to 2 µJ, corresponding to an average power of 26 W. The number of optical cycles in a pulse is approximately equal to the number of undulator periods, thus the pulse length is proportional to the wavelength. At FELBE it varies from ~ 1 ps for the shortest to ~ 30 ps for longest wavelength, but can slightly be adjusted by cavity detuning.

As can be seen from the above numbers, a desired large pulse energy in the μJ range inevitably leads to a high average power beyond ten Watts, which is unacceptable for most solid-state spectroscopic experiments. One way around this is to chop the continuous pulse sequences into macropulses, similar to the operation mode of normal-conducting accelerators, yet this eliminates the advantages of CW operation. Another option is to select single FEL pulses at kHz repetition rates using an optical switch outside [3] or potentially inside [4] the cavity.

The above difficulty was one motivation to build the single-pass, superradiant THz source TELBE [5], operated since 2015. Here one takes advantage of the fact that wavelengths longer than the electron bunch length are emitted coherently with a power proportional to N2, the square of the number of electrons in a bunch. This can lead to gigantic enhancements by a factor of 108, up to a frequency limited by the achievable bunch length. TELBE consists of an 8-period electromagnetic undulator of 300 mm period, emitting multi-cycle, semi-narrowband (10–20% bandwidth) radiation up to 2.5 THz. In order to obtain high bunch charges (which quadratically increases the THz pulse energy), a superconducting RF photogun operated with either Mg or Cs2Te cathode [6] presently provides up to 250 pC bunches at a repetition rate up to 500 kHz, which generates THz pulses with energy up to 10 µJ. Present research on improved photocathode materials aim at bunch charges of 500 pC or more, which could lead to THz pulse energies in the 50 µJ range.

There are several advantages of TELBE as compared to FELBE:

  • the single-pass design allows for a lower and flexible repetition rate, which can be adjusted for each particular experiment to avoid sample overheating or other thermal effects, and thus enables the use of the full THz pulse energy in experiments;

  • the carrier-envelope phase (CEP) stability, which provides sub-cycle temporal resolution; for example a 1 THz pulse can be sampled with 30 fs resolution, which corresponds to 33 measured data points per cycle.

In addition, a diffraction radiator is used to generate broad-band, single-cycle pulses at TELBE with high-frequency cut-off of about 1.5–2 THz. While its parameters are similar to table-top THz radiators based on the tilted-wavefront technique in LiNbO3 [7, 8], the single-cycle THz pulse is mainly used as a timing reference to enable the sub-cycle temporal resolution [9]. The spectral performance and typical electric-field traces of the TELBE sources are shown in Fig. 3. Table 1 gives an overview on the parameters of the three tunable infrared/THz sources at ELBE.

Fig. 3
figure 3

Typical TELBE undulator THz pulse energies as a function of THz frequency (left, 50 kHz repetition rate; the violet vs red points show the recently achieved improvement) and representative electric-field transients for the undulator source (top right, after 0.3 THz bandpass filter) and the coherent diffraction radiator (CDR) (lower right)

Full size image
Table 1 Parameters of the infrared/THz sources at ELBE
Full size table

3 Scientific highlights

The major fraction of experiments performed both at FELBE and TELBE are in solid state physics, and can be divided into three (albeit overlap**) types:

  • Pump-probe spectroscopy to explore the relaxation processes of highly excited nonequilibrium states, hereby exploiting the temporal resolution provided by the short pulses; here the probe radiation can originate from the same FEL beam (single-color pump-probe). However, often it is useful to probe with a different beam, stemming from one of the femtosecond table-top sources synchronized to the FEL. This can be visible or near-infrared (NIR) light, but also broad-band THz radiation. Using femtosecond laser systems for probing provides coherent spectroscopy with sub-THz-cycle temporal resolution, so the THz-induced coherent dynamics can be visualized in time domain.

  • Exploiting the high peak power to induce nonlinear, nonperturbative effects, such as generation of higher harmonics or even inducing new phases via the strong THz electric field.

  • Using the high photon flux to perform near-field nanospectroscopy using a scattering scanning near-field optical microscope (SNOM) [10]. Ideally such investigations can also be performed with sub-ps-temporal resolution utilizing techniques that are currently under development [9].

An important point of such strong infrared/THz sources is the ability to excite or pump low-energy excitations selectively and resonantly. Examples for such excitations are confined states in semiconductor nanostructures, collective quasi-particle excitations such as phonons, plasmons, magnons, excitons, or the interactions between them (e.g., polaritons [11]). Here we briefly present a few experimental highlights of the past decade for both FELBE and TELBE.

FELBE was used to resonantly pump the 1s–2p intra-excitonic transition in (In)GaAs quantum wells with few-meV photons, thus inducing dressed states evidenced by the Autler-Townes (or: Rabi) splitting of the involved levels [12]. The splitting of the 1s state was probed with a femtosecond broad-band NIR pulse. Figure 4 shows the observed line splitting under different electric-field values (left panel) and a color map as a function of time delay between NIR and FEL pulses [13].

Fig. 4
figure 4

Left: (NIR) absorption spectrum of the quantum well sample while the 1s-2p intra-exciton transition is pumped resonantly with the THz FEL at 1.6 THz (6.6 meV). The line splitting evolving for increasing THz electric fields can be clearly observed. Right: 2D color map of the NIR absorption as a function of photon energy (horizontal) and delay time between NIR and FEL pulses (vertical) for a THz photon energy of 8.2 meV (blue detuning) and a peak electric field of 12.5 kV cm−1. Reprinted from Ref. [13]

Full size image

The two-dimensional material graphene was predicted to exhibit very high THz nonlinearity due do its linear, Dirac-type band dispersion [14], but for a long time experiments did not show evidence for this. With TELBE we successfully could demonstrate harmonic generation up to the seventh order in doped single-layer graphene [15], as shown in Fig. 5. The THz fields were detected directly in the time domain using free-space electro-optic sampling.

Fig. 5
figure 5

Amplitude spectrum transmitted through graphene clearly showing 3rd, 5th and 7th harmonics (blue). The incident THz spectrum with a center frequency of 0.3 THz and 85 kV/cm peak field is shown in red

Full size image

Apart from energy and intensity, also the polarization of the THz radiation can be purposefully employed. For example, Landau levels (the electronic energy levels in a magnetic field) in graphene are non-equidistant, and considering the dipole selection rules, even the spectrally degenerate transitions − 1 → 0 and 0 → 1 can be distinguished by their opposite circular polarizations. Pump-probe experiments in all four combinations of pum** and probing with left or right circularly polarized radiation, respectively, reveal a surprising effect: in slightly n-doped material, pum** the − 1 → 0 transition results in a depopulation of the zeroth Landau level even though this level is optically pumped at the same time. This depopulation is directly evidenced by a sign change of the pump-probe signal when probing with opposite helicity and can be explained by a very efficient Auger scattering between the lowest Landau levels [16].

4 Technological development

Besides the development of the ELBE-based THz sources by itself, the timing stability between accelerator and external laser systems is a very important parameter, as it determines the temporal resolution of the studied dynamical processes and the sensitivity of coherent detection schemes. At ELBE we have reached an effective temporal resolution of a few femtoseconds, which has been enabled by a combination of a stable master-clock system with thermal drift-compensated fiber-link signal distribution and pulse-resolved signal detection schemes which enable post mortem data sorting [17]. Recently, we have demonstrated that such a pulse-resolved detection technique can be also applied to the CEP-unstable FELBE source, which enables CEP phase corrections up to the full bandwidth of the FEL (13 MHz). At the same time, for experimental techniques requiring data averaging or demodulation, the pulse-resolved techniques are not feasible. For this, we are develo** passive all-optical jitter-free synchronization schemes based on THz slicing [9].

Recently the capabilities for near-field nanoscopy at FELBE have been upgraded by implementing a commercial (Neaspec) instrument with a mid-infrared difference-frequency generation source (5–15 µm) and a nano-Fourier-transform spectroscopy (nano-FTIR) detection option. The synchronization of the internal source to FELBE enables pump-probe experiments on the nanoscale, where FELBE allows tunable nonlinear narrowband excitation of the samples and the response can be probed broadband utilizing the mid-infrared source.

Combining the THz slicing synchronization scheme with a near-field nanoscopy end station at TELBE we have recently demonstrated that high-field THz coherent dynamics in time-domain can be combined with nm-size spatial resolution. Thereby, THz scattering-type near-field optical microscopy on THz-driven nonlinear processes in nanostructures is making a decisive step closer to realization.

THz-induced changes to a material’s electronic structure are routinely measured using optical techniques, such as transient absorption, which give only indirect access to the actual electronic populations and their absolute energy levels. Combining the high-duty-cycle TELBE source with a time- and angle-resolved photoelectron spectroscopy (trARPES) setup is therefore highly desirable, as it enables the measurement of ultrafast movies of THz-induced carrier dynamics and their absolute energy and dispersion. At the TELBE laboratory, a femtosecond XUV source with a photon energy of 21 eV and a repetition-rate of 100 kHz has recently been set up. This source will act as probe source for the THz pump—trARPES probe experiment that is currently under construction.

Since the radiated THz power is proportional to the square of the bunch charge, it is of key importance to generate as many electrons as possible by the injector. There are two technical challenges to realize this: the first is to enhance the extracting electric field on the cathode surface, which allows one to compensate for higher space-charge forces and thus to generate more electrons per bunch; and the second is to employ highly efficient and robust photocathodes in combination with a perfectly synchronized drive laser system. To address the first point, HZDR is currently assembling its third SRF gun, housing a refurbished cavity (in cooperation with DESY and HZB) that is expected to achieve an electric field 50% higher (30 MV/m on axis) than the presently running SRF gun. In addition, we are continuously improving the preparation process of our current Cs2Te photocathodes, but also advancing the investigation of GaN as a promising new cathode material [18].

5 Upgrade plans: DALI—Dresden advanced light infrastructure

While FELBE and TELBE have fulfilled many requirements of researchers, and many exciting scientific results have been and are being obtained, it would be desirable to combine the advantages of FELBE and TELBE while avoiding their drawbacks. Specifically, this points to an infrared/THz source

  • with flexible repetition rates like TELBE (not fixed at 13 MHz like for FELBE),

  • but not limited to frequencies lower than 2–3 THz (like TELBE), but reaching up to 20–30 THz into the mid-infrared, like FELBE.

It is important to be able to cover especially the core range between 3 and 15 THz, for which no convenient and strong table-top sources exist.

At the same time significantly increased pulse energies of 100 µJ to 1 mJ should be achieved.

Presently we are working on a conceptual design report for such a new facility called DALI: Dresden Advanced Light Infrastructure. It should replace ELBE which will approach an operation time of 20 years soon. That facility should even include a VUV FEL to enable combined experiments with high scientific gain, as described below.

DALI is envisioned to be an integrated, accelerator-based THz and VUV source that covers the frequency range from 0.1 to 30 THz with coherent, semi-narrowband (≈ 10% spectral bandwidth) few-cycle radiation pulses and the VUV spectral range from 50 to 250 nm with intense, quasi-transform-limited sub-picosecond pulses. Driven by a superconducting 50 MeV electron linear accelerator (LINAC), the THz source will provide radiation with high pulse energy (100 µJ–1 mJ) and flexible repetition rate (10 kHz to 1 MHz), an unprecedented combination of both parameters being a factor of at least 100 larger compared to what is available to date. The VUV FEL source will be based on a 300 MeV superconducting accelerator, implemented as a 150 MeV recirculation LINAC, and provide up to 30 µJ pulse energy at 0.1–5 MHz repetition rate. With a sub-100 fs synchronization, this combined, worldwide unique THz/VUV facility will open a wealth of new avenues for investigations of nonlinear and high-field-driven phenomena.

Calculations show that it is not possible to properly cover the whole spectral range from 0.1 to 30 THz using one and the same accelerator-based mechanism. Due to constraints related to electron bunch charge and bunch length compression, the customary methods for producing high-power coherent undulator radiation from 50 MeV electron bunches will be limited to frequencies lower than ~ 2 THz. To achieve higher frequencies up to 30 THz, the electron beam has to be longitudinally density-modulated by an external laser (optical klystron). To this end, the most appropriate external laser will effectively be a free-electron laser (FEL) oscillator, as it provides continuous tunability and high peak fields at high repetition rate.

At the new DALI facility, THz radiation will therefore be generated by:

  • A superradiant undulator (0.1–2 THz, 10–20% bandwidth, repetition rate 10 kHz–1 MHz, carrier-envelope phase stable)

  • A superradiant undulator driven by a modulated electron beam (2–30 THz, few-percent bandwidth, repetition rate 10 kHz–1 MHz)

  • An FEL oscillator, which can act either as a modulator or as a narrowband THz source (with 10 MHz repetition rate, 1–2% bandwidth, and 1–10 µJ pulse energy). An additional cavity dumper should enable an increase in pulse energy to 10 – 100 µJ accompanied by a reduction of the repetition rate down to 100 kHz.

  • In addition to the multicycle narrowband sources, a single-cycle broadband source based on coherent diffraction radiation (or a similar mechanism) will be built. Instead of selectively exciting specific modes, it will exploit the achievable ultrahigh electric field of several MV/cm, which is comparable to the intrinsic electric fields in atoms.

Two separate electron injectors will be needed, one accelerating 1 nC bunches at a repetition rate of 10 kHz–1 MHz for the superradiant sources and one accelerating 100 pC bunches at 10 MHz for the FEL oscillator that will provide the electron bunch modulation or cavity-dumped FEL pulses.

To exploit the full research potential of an intense THz source, various probing techniques are of key importance. Different probing techniques, all synchronized to the THz source with minimum timing jitter, deliver targeted information on the reaction of materials upon excitation with intense THz radiation. For example, probing with femtosecond white-light pulses allows monitoring bandgaps of semiconductors and dielectrics. From THz time-domain spectroscopy (THz-TDS), utilizing (weak) broadband pulses, the complex dynamic conductivity of materials can be obtained, which will for example indicate superconducting gaps. Time- and angle-resolved photoemission spectroscopy (trARPES) directly reveals the energy dispersion of occupied states and can thus indicate how the electronic band structure and occupation are altered by intense THz fields. Ultrafast electron diffraction (UED), on the other hand, provides insights into THz-driven structural changes. Such an UED probe, based on the successful superconducting radiofrequency (SRF) electron gun design currently operated at ELBE, will also be implemented at DALI. Thus, the selective addressing of low-energy excitations by intense THz pulses, combined with specific tools to monitor the changes induced in the electronic or structural properties using trARPES and UED, respectively, will enable a new quality of solid-state research.

To advance the understanding of the dynamic processes in molecular systems and chemical reactions, we furthermore propose to build a VUV FEL source, which would operate in the wavelength range of 250–50 nm (5–25 eV) and provide capabilities (e.g., spectral brightness, transform-limited pulses, high pulse energy, and high repetition rate) orders of magnitude beyond what is possible with table-top VUV sources.

The proposed DALI VUV FEL would operate as a seeded high-gain harmonic generation (HGHG) FEL, with pulse energies of tens of µJ at MHz repetition rate. Such a project generates tremendous interest in the physical chemistry, atomic physics, and molecular physics communities, as it would bring transformative changes to these research fields and their use of synchrotron radiation sources. For instance, it would lead to a better microscopic understanding of chemical reaction dynamics, as relevant e.g., in catalysis, combustion, astrophysics, and astrochemistry, as well as in the science of planetary atmospheres and of Earth’s environment. Of particular interest is the ability to selectively control chemical reactions with the tunable high-field THz radiation from DALI and observe the dynamics of these reactions with ultrashort VUV FEL pulses.

Additionally, the third harmonic of the FEL would deliver VUV radiation down to 17 nm wavelength (75 eV photon energy), albeit with 100–1000 times less pulse energy (tens of nJ). This would be an ideal source to perform trARPES on condensed-matter samples with an unprecedented signal-to-noise ratio and spectral resolution, thus hel** to solve long-standing puzzles in solid-state physics, such as high-temperature superconductivity and topological phase transitions.

The VUV FEL source would be synchronized with the THz and UED sources, so that combined pump–probe experiments would be possible. Such a combination of high-pulse-energy, high-repetition-rate THz and VUV sources would be unique worldwide. Fig. 6 shows an envisioned layout of the DALI facility.

Fig. 6
figure 6

Envisioned layout of DALI. The labels for the user labs are only examples

Full size image

The new DALI facility will thus encompass light sources with unmatched characteristics that are optimally suited for research related to questions of high societal impact. The combination of intense and versatile THz beams, which can be exactly tuned to the processes of interest, with VUV beams and ultrafast electron beams will allow unprecedented scientific insights. Realizing this combination of sources essentially requires a dedicated large-scale facility, making DALI unique for the foreseeable future.

Compact light sources based on advanced laser plasma accelerators may complement the high-field sources of the DALI facility for dedicated low-repetition-rate applications. Ultrafast high-power laser-driven sources with variable photon energy enable unprecedented probing capabilities as well as highest transient fields and serve as a technology development platform for future applications.

A high-intensity positron source based on bremsstrahlung production will complement the DALI light sources, thus opening up unique experimental conditions for materials science, chemistry, solid-state physics, and semiconductor research. Positron annihilation in matter serves as an excellent tool for depth-resolved studies of open-volume defects and nanoscale porosity in bulk samples and thin films. The experience gained at the existing ELBE positron source [19] in defect studies and porosimetry in materials will be extended by new types of in-situ defect analysis and characterization, focusing on dynamical effects of defect migration and development by means of laser-induced charge carrier manipulation, high- and low-temperature annealing, and ion irradiation [20, 21]. High electron bunch charges and pulse repetition rates of 1–5 MHz constitute ideal conditions for positron annihilation lifetime measurements of depth-resolved open-volume defects, such as point defects, precipitations, and grain boundaries in metals, alloys, ceramics, polymers, and (ionic) liquids, and of material porosities with sizes ranging from 0.2 to 35 nm.

6 Next scientific challenges

With a THz photon source such as DALI one could apply electric fields resembling or even exceeding the internal fields in atoms, Volts per nanometer, or tens of MV/cm, with sub-cycle time resolution and repetition rates adapted to the respective experiment. This way researchers will be able to study and actively manipulate the interplay between spin, orbital and lattice degrees of freedom and the ultrafast dynamics of phase transitions [22, 23]. By selective excitation of low-energy modes one can massively influence relevant macroscopic properties. An emerging field is nonlinear phononics [24, 25] where strong pum** of specific phonon modes induces radical changes of electronic properties including phase transitions, e.g., through changing a bond angle in the crystal lattice. Electron diffraction could provide insights into such THz-driven structural changes. Recently it was shown that transient superconductivity can be induced in cuprates and in doped fullerides in such a way [26]. This wide and fast evolving field can be described as the non-equilibrium coherent control of emergent phenomena in quantum materials [27].

Furthermore one could access new, non-equilibrium and “hidden” exotic phases. In particular so-called Floquet engineering has attracted a lot of interest recently [28]. Here a non-resonant time-periodic driving field modifies the band structure, giving rise to quasi-energy bands. Typically such modified band structures (and populations) can be investigated via time-resolved ARPES, requiring VUV pulses which will be available from the 3rd harmonic of the VUV part of DALI. A beautiful experiment of this kind (albeit with table-top sources) has recently been performed, revealing the THz-induced Dirac current in a topological insulator [29].

Additionally, THz spintronics is a currently very active research area. The DALI sources will provide unique instrumentation for studying coherent spin (transport) dynamics at ultrahigh frequencies and in the nonlinear regime, which are among the fundamental effects governing the performance of upcoming spin-based technology [30].

Apart from the above mentioned perspectives in the solid state physics of quantum materials, there is a wide range of challenges and opportunities for discovery in surface chemistry, liquid-state and gas-phase spectroscopy. THz fields could accelerate or modify chemical reactions while staying in the electronic ground state (“ground state chemistry”). High THz fields can induce ionic motions in crystals, liquids, and even single molecules. On surfaces, these motions have a strong impact on the chemical bonds, both intramolecular and between molecules and surface. Consequently, this makes THz excitation a highly promising tool for investigating catalytic processes. The complex and highly dynamic network of intermolecular hydrogen bonds in liquids, important for many biological processes, could be unveiled by THz nonlinear spectroscopy, for example in liquid jets.

An especially promising field, employing both infrared and VUV sources, is the threshold IR-VUV double-resonance photoionization of molecules and radicals. Here the VUV FEL is tuned just below the threshold for the vibrational ground state (or a populated vibrationally excited state) of the species of interest. When an IR laser is on-resonance with a transition to a higher excited state, the combined IR + VUV energy is sufficient to overcome the ionization potential and yields a photo-ion that can be sensitively and mass-selectively detected. This will enable access to important, but usually hidden reactive intermediate species and vibrational state populations (Barratt Park and Alec Wodtke, private communication). The VUV FEL with its unprecedented repetition rate (and thus photon flux) will open up entirely new possibilities for research in astrochemistry, environmental chemistry, combustion science and similar fields highly relevant for understanding and controlling the climate change.

7 Note added in proof

After finishing this article it has been decided to not pursue the project of the VUV FEL further at the present time. The scientific potential presented here however remains valid.