Free Electron Lasers 2003: Proceedings of the 25th International Free Electron Laser Conference and the 10th FEL Users Workshop, Tsukuba, Ibaraki, Japan, 8-12 September 2003

Section 1

FEL Prize and New Lasing

Viability of infrared FEL facilities

H. Alan Schwettman*,     Department of Physics and W. W. Hansen Experimental Physics Laboratory, Stanford University, Stanford, CA 94305, USA. E-mail address: has@stanford.edu

*Tel: + 1-650-723-0305; fax: +1-650-725-8311.

Abstract

Infrared FELs have broken important ground in optical science in the past decade. The rapid development of optical parametric amplifiers and oscillators, and THz sources, however, has changed the competitive landscape and compelled FEL facilities to identify and exploit their unique advantages. The viability of infrared FEL facilities depends on targeting unique world-class science and providing adequate experimental beam time at competitive costs.© 2004 Elsevier B.V. All rights reserved.PACS: 41.60.Cr;07.57.Hm

Keywords

Superconducting linac

FEL facility

Optical science

1 Introduction

Infrared FELs have broken important ground in optical science in the past decade, opening new vistas in physics, chemistry, biology, and medicine [1]. There are now a number of FEL facilities throughout the world committed to infrared (IR) optical science; and the capabilities of these facilities are remarkably different, one from the other. To cite an example, linac-driven IR FELs have many attractive features, most notably wavelength tunability, control of spectral and temporal pulse width, and exceptional beam quality and stability. However, the normal conducting version is capable of delivering a high average power optical pulse train for a few microseconds, while the superconducting version is capable of delivering a modest average power optical pulse train that continues indefinitely. These differences, in fact, represent important capabilities that individual FEL facilities can exploit in scientific pursuits.

One decade ago the most important viability issue for IR FEL facilities was FEL reliability. As the facilities have matured, this issue has thankfully passed and in its stead are the central issues of providing sufficient experimental time at competitive costs and providing world-class scientific opportunities. These issues have everything to do with attracting outstanding scientific talent to the facility and with attracting adequate financial support that is stable over time. In this paper, I will use the superconducting linac driven FEL to illustrate viability issues and will derive most of my observations from our experience at Stanford.

2 Superconducting linacs for infrared FELs

The very first superconducting linac was built at Stanford in the early 1970s, now more than 30 years ago [2]. The landmark FEL oscillator experiment [3] depended critically on the unique capabilities of that linac, and that same linac has served well in support of our FEL Optical Science Center. Our linac, is located in a tunnel some 10 m below the main floor of the laboratory. In the tunnel there are two undulators, one to provide mid-IR radiation and the other to provide far-IR radiation. On the main floor of the laboratory there are 10 experimental rooms for optical science fed by the two undulators. In the early history of our Center, we were forcefully reminded by our colleagues that a laser is not an experiment! In fact, the 10 experimental rooms are filled with a vast array of optics, detectors, electronics, superconducting magnets, lasers, and other sophisticated optical instrumentation. The total investment in these probably exceeds that in the FEL itself.

In the past decade superconducting technology has progressed enormously and thus we are now replacing the historic superconducting structures with state-of-the-art structures, following the recent DESY design [4].¹ In the DESY design the helium vessel is an integral part of a 9-cell structure and the liquid helium capacity of this vessel is a mere 251. This is a significant departure from the original Stanford design and from the Jlab design. For continuous operation, as envisioned for the IR FEL, realistic energy gradients for the one meter long DESY structures are 10−15MeV/m and at 2K the refrigeration required is 10−20 W. The cryomodule requirements are dramatically different from those at DESY and thus a satisfactory design has been generated in a collaboration between Stanford and FZR Rossendorf. The cryomodule holds two structures and a cutaway view of one half of a cryomodule is shown in Fig. 1.

Fig. 1 Cutaway view of half a cryomodule indicating one of the 9-cell superconducting structures with integral helium vessel. Liquid helium is fed into the bottom of the vessel. Cold helium gas is collected in the tube located directly above the vessel and then transported to the refrigerator. RF power is fed to the structures from below at the center of the cryomodule.

The new superconducting structures will have a dramatic impact on FEL operations at Stanford. The peak and average FEL power will increase by a factor of 5, and we will once again be able to deliver a continuous optical pulse train. But perhaps most important, liquifying a mere 1501 will be sufficient to bring the FEL into operation, rather than the few thousand liters required now, thus providing great flexibility in scheduling experiments. One of the new cryomodules is already installed on the linac and all cryomodules will be replaced this year.

3 Viability issue: providing experimental time

Providing adequate experimental time at an IR FEL facility is not as simple as it sounds. Existing facilities typically provide 3000 h of experimental time per year and typically 25% of this time is allocated to a core experimental group. At Stanford in the mid-1990s a broad collaboration built around Professor Fayer’s group carried out the first comprehensive vibrational dynamics experiments on glass-forming liquids and proteins. For these pioneering experiments Fayer was awarded the Earl K. Plyler Prize for Molecular Spectroscopy in 2000. Unfortunately, the scientific interest in such experiments far exceeded the 750 h of time that could be allocated, and thus as optical parametric amplifiers and oscillators improved, Fayer sought and gained funding to build three mid-IR laser systems which now provide for his group approximately 7500 h of experimental time per year. It is clear that a scientifically successful FEL facility providing 10,000 h per year is still not providing enough experimental time.

At a synchrotron radiation facility tens of experimental stations are located around the perimeter of the synchrotron and experiments can be performed at each of these simultaneously. The radiation produced by a synchrotron covers a broad spectral range and each User can select an appropriate portion of the spectrum. The crux of the problem here is that the FEL by its nature provides a narrow spectral line, and each individual User will want to control the wavelength of that line. There are means, however, for an IR FEL to effectively serve multiple experiments, for instance by switching from one experiment to another on a macropulse basis. In an FEL driven by a superconducting linac, the linac parameters, and therefore the e-beam and optical beam parameters can be changed in a time the order of 10 ms. Thus it is possible to switch between two linac states, ultimately providing a 40 ms optical beam to each of two experiments at a 10 Hz rate, and each can have independent control of its own beam. At Stanford we implemented such macro-pulse switching between two undulators a few years ago. One might imagine that there is no need to switch linac parameters if there are two undulators, but that proved to be naïve. If the two experiments require widely different wavelengths or widely different pulse lengths, macro-pulse control of e-beam parameters is essential. The Stanford system [5] provides macropulse-to-macropulse control of the gun current, the phase of the sub-harmonic buncher, the amplitude and phase of the accelerating structures and the electron beam steering and focussing. The switching time in the Stanford system is approximately 40 ms, limited by switching of the quadrupole settings. Careful design of the quadrupoles would permit switching in the 10 ms time referred to above.

It is also possible to macropulse switch in a single undulator. The interleaved optical macro-pulses must then be separated and delivered to the proper experiment. At Stanford in the past month, we have demonstrated an optical switching system [6] that time-sorts the interleaved macropulses and transports them to separate experimental rooms. In principle there is no reason an FEL could not switch between four different states, providing an independently controllable 15 ms optical beam to each of four different experiments. To accomplish this, electron beam macropulses in two different states must be delivered to each of two different undulators. The nominal 3000 h of experimental time could then be expanded to 12,000 h. Increasing experimental time in this way would diminish the availability issue, and would significantly reduce the cost. At Stanford the operations cost for experimental time could decrease from $600 per hour to $150 per hour.

4 Viability issue: scientific opportunities

In the early 1990s FELs were virtually alone in providing high quality picosecond (ps) optical beams in the mid- and far-IR. At FEL facilities, experiments have been performed exploring vibrational dynamics, timescales and pathways of energy flow, IR near-field microscopy, THz spectroscopy of semiconducting nanostructures, and many other topics. The rapid development of optical parametric amplifiers and oscillators, and THz sources, however, has changed the landscape and has encouraged Users to identify and exploit the special advantages of the FEL.

4.1 Vibrational dynamics experiments

Superconducting linac-driven FELs, as an example, have a special advantage that can be exploited in vibrational dynamics experiments. They provide an optical pulse train at 10−20 MHz that continues indefinitely, and from this pulse train it is possible to select optical pulses at a repetition rate that matches the experiment at hand. To illustrate the experimental problem consider two extreme situations. The first is a pump–probe experiment performed at the JLab FEL to study vibrational relaxation of hydrogen defects in silicon [7], a topic of great technological importance. There are many absorption features in proton-implanted crystalline silicon, corresponding to specific hydrogen defects, as illustrated in Fig. 2. Characterization of such defects requires a detailed understanding of timescales and pathways of energy flow from the defects, to local modes, and then to the phonon bath of the host material. The lifetimes of the Si–H stretch modes, determined by the pump–probe experiments at JLab, represent a first step in this detailed understanding, and show that the lifetime is strongly dependent on local structure, ranging from ps for interstitial-like hydrogen to hundreds of ps for hydrogen-vacancy complexes.

Fig. 2 FTIR spectrum of protein-implanted crystalline silicon, illustrating absorption features corresponding to specific hydrogen defects.

Let us be certain we understand the nature of this vibrational relaxation measurement. In the mid-IR at room temperature the bound hydrogen impurities are in the vibrational ground state. The pump beam excites some of these to the first excited state and they subsequently relax back to the ground state by thermalization processes. That vibrational relaxation process can be monitored by the probe beam. The attenuation of the probe beam as it passes through the sample is proportional to the population difference between the ground state and the first excited state. Thus by measuring the probe beam attenuation as a function of the delay between pump and probe, one can determine the vibrational relaxation time constant, T1.

The problem in this pump–probe experiment is that the concentration of hydrogen impurities in the experiment is small and the cross-section of vibrational transitions is small as well. The saving grace is that in the wavelength region of interest, silicon is nearly transparent. Thus it is possible to utilize the full 20 MHz rep rate of the FEL, delivering an average power of 10W to the experiment. The high rep rate, high average power and exceptional beam stability of the FEL transformed an experiment that was marginal using an optical parametric amplifier at 1 KHz and 5mW average power to an eminently viable experiment.

At the other extreme, we have the situation encountered in the vibrational dynamics experiments on glass-forming liquids and proteins performed at Stanford. Here the sample concentrations are much higher, but the solvent is rather strongly absorbing. In vibrational dynamics one is interested in the pure dephasing time which is a measure of ps fluctuations of the local environment. This measurement is a bit more complicated than the simple relaxation measurement discussed above. To determine the pure dephasing time one must perform a photon echo experiment, the optical analogue of the spin echo, in addition to the vibrational relaxation experiment. At Stanford pure dephasing rates as a function of temperature have been measured for a number of systems. Results are shown in Fig. 3 for myoglobin, a small protein that has the biological function of reversible binding and transport of oxygen in muscle tissue [8]. The observed temperature dependence below 200 K is reminiscent of the two level system dynamics of conventional low-temperature glasses, and suggests that the observed behavior arises from tunneling dynamics of an ensemble of protein two-level systems. Above 200 K, however, the data indicate the emergence of an exponentially activated process.

Fig. 3 Pure dephasing rate of native myoglobin (CO), indicating transition from region governed by the tunneling dynamics of protein two level systems to region governed by an exponentially activated process.

But what is an appropriate optical pulse rep rate and single pulse energy for such an experiment? Among other things that depends on the choice of solvent, but let us consider water, the solvent of preference for biological systems. If the sample is confined in a chamber, let us say with a CaF2 window at both the front and back surface, and if you assume the heat conduction path is through the CaF2 to a heat sink, one can calculate the temperature distribution in the water, given the pulse rep rate, the single pulse energy, and the thermal conductivity, heat capacity and absorption coefficient of water at the wavelength of interest. At 6.45 μm, the wavelength of the Amide II mode in proteins, the absorption coefficient of water is 825 cm−1 corresponding to a penetration depth of approximately 12 μm. For a water sample 20 μm thick and a train of 1 μJ pulses repeating at 10 kHz (10 mW average power), the steady state temperature at the water/CaF2 interface rises by 2.5 K and the maximum temperature in the sample rises by 5.2 K. The calculated thermal relaxation time of the system is approximately 200 μs and thus it is clear that the assumed repetition rate is not quite fast enough to smooth the sample temperature in time. In fact, a single 1 μJ optical pulse elevates the water temperature at the surface by 1.4 K. The incremental temperatures calculated above are acceptable for biological experiments, but increasing either the single pulse energy or the average power by an order of magnitude will lead to unacceptable conditions. The Stanford experiments on proteins, in fact, utilized a rep rate 2.5 times larger and a single pulse energy 2.5 times smaller than assumed in the calculation.

4.2 Gas phase experiments

At the time we wrote the proposal for our optical science center at Stanford gas phase experiments with our FEL seemed marginal at best. My interest in gas phase experiments was rekindled, however, in the early 1990s when I served as chair of a DOE committee to review a proposed 600 W IR FEL to be constructed at Berkeley as part of the Chemical Dynamics Research Laboratory. I suggested to Professor Yuan Lee, the principal scientist behind the proposed facility, that if the project were approved it would be interesting to consider doing these experiments either as FEL intracavity experiments or as synchronously pumped external cavity experiments. Either would greatly reduce the power required in the FEL. These suggested possibilities, illustrated in Fig. 4, were motivated in part by two projects underway at Stanford. At that time a graduate student, Ken Berryman, was designing our FIR FEL system [9] and we were already committed to building a three mirror FEL cavity that reached out of the shielded area into an experimental room. It did not seem out of the question to make a four mirror FEL cavity that could then accommodate a gas-phase experiment. Another graduate student, Paul Haar, was beginning experiments, to synchronously pump an external cavity [10], and he soon achieved pulse energies in the external cavity that were nearly 100 times the energy of the individual incident pulses, demonstrating remarkable coherence in the optical pulse train.

Fig. 4 Possible intracavity and external cavity configurations that could accommodate gas-phase optical experiments. Each of these would dramatically increase the optical fluence available for experiments.

Unfortunately, the FEL plans at Berkeley never materialized. At Stanford no intracavity experiment was attempted, however, the external cavity was reconfigured and used to demonstrate the feasibility of synchronously pumped cavity ring-down spectroscopy. Cavity ringdown has often been used to measure absorption in gaseous systems that are very dilute or very weakly absorbing. In this technique, light from a laser source is coupled into a high Q optical cavity that encloses the gas of interest. When the light source is interrupted, optical energy stored in the cavity decays exponentially due to cavity losses and absorption by the gaseous medium. The absorption spectrum of the gas is obtained by measuring the decay rate with medium and subtracting the decay rate of the empty cavity. High sensitivity is achieved since the method provides a very long effective pathlength for absorption by the gas and is insensitive to amplitude fluctuations of the laser source.

The high pulse repetition rate of the FEL pulse train has a dramatic impact on cavity ringdown spectroscopy (CRDS). To achieve high sensitivity, the mirror transmission must be small, let us assume 10−3. Thus a single pulse, coupled through the front mirror into the cavity will be smaller than the incident pulse by this factor. The exponential decay of this pulse will be monitored by the energy coupled out of the cavity through the back mirror and this is smaller than the incident pulse by the factor 10−6. Synchronous pumping addresses this problem. The pulse energy in the cavity builds to 10+3 times the incident pulse energy and the energy coupled out of the cavity through the back mirror initially is equal to the incident pulse energy and then decays exponentially in time. To demonstrate the feasibility of the technique we measured the (8,2,6)−(9,3,7) transition of H2O diluted in helium [11]. The measured transition spectrum is compared to HITRAN96 in Fig. 5.

Fig. 5 Measured spectrum of the (8,2,6)−(9,3,7) transition of H2O diluted in helium compared to HITRAN96.

To proceed quickly with a demonstration of this CRDS technique, a number of compromises were made. A single detector was used at the output of the spectrometer instead of a commercially available 30 element array. Furthermore, our 30 year old FEL delivered optical pulses at 10 Hz, and without cavity length stabilization, synchronized pumping actually occurred at less than 1 Hz. Despite these compromises, the sensitivity demonstrated was comparable to the best achieved in the literature. Now, however, with the new linac structures we expect a pumping rate of 100 Hz. This development and the array detector will dramatically improve the data collection rate and the sensitivity and thus open new opportunities in gas-phase spectroscopy. But is it realistic to synchronously pump an external cavity at 100 Hz? With the new linac we will once again be able to operate the FEL continuously. If we modulate the cavity length at 50 Hz, we will pass through the synchronous pumping resonance condition at a 100 Hz rate. And each time we pass through resonance, we will have acquired both a measured decay and a measurement of drift in the cavity length which can be used to stabilize operation. The remaining question is whether passage through resonance is adiabatic. For reasonable cavity parameters, the resonance width (change in cavity length) is 4nm and the cavity decay time is 2 μS. With a modulation amplitude of 300 nm and a frequency of 50 Hz the velocity is 10−4 m/s which satisfies the adiabatic condition by an order of magnitude.

The scheme for modulating the length of an external cavity to achieve periodic synchronous pumping might also be useful in other gas phase spectroscopic techniques such as FT Ion Cyclotron Resonance Mass Spectroscopy. With the new linac installed the single optical pulse energy will increase from 1 to 5 μJ and the pulse circulating in the cavity will be 500 μJ. For the modulation conditions described above the FWHM time on resonance is approximately 40 μs and the total energy passing through the interaction region is more than 1 J in that time.

5 Conclusions

With the emergence of optical parametric amplifiers and oscillators, and THz sources, the domain of infrared FELs is less that it had been one decade ago. None-the-less, FELs have unique advantages that can be exploited in important classes of optical experiments. The task at hand for IR FEL facilities is to identify those opportunities and provide adequate experimental time at competitive costs.

Acknowledgements

The author would like to acknowledge the remarkable contributions of the many individuals that have been part of the FEL program at Stanford over the past 30 years, but most particularly, his long term colleague Todd Smith. This work was supported in part by the Air Force Office of Scientific Research, grant number F49620-00-1-0349.

References

1. W.B. Colson, E.D. Johnson, M.L. Kelley, H.A. Schwettman, Phys. Today (2003) 35.

2. McAshan, M.S., Mittag, K., Schwettman, H.A., Suelzle, L.R., Turneaure, J.P. Appl. Phys. Lett. 1973; 22(11):605.

3. Deacon, D.A.G., Elias, L.R., Madey, J.M.J., Ramian, G.J., Schwettman, H.A., Smith, T.I. Phys. Rev. Lett. 1977; 38(16):892.

4. TESLA reports describing the DESY structure are available at http://tesla.desy.de.

5. Crosson, E.R., James, G.E., Schwettman, H.A., Smith, T.I., Swent, R.L. Multi-user operation at an FEL facility. Nucl. Instr. and Meth. B. 1998; 144:25.

6. Private communications from Doug King, George Marcus, Richard Swent.

7. G. Luepke, N.H. Tolk, L.C. Feldman, J. Appl. Phys. 93 (5) (2003) 2317.

8. C.W. Rella, A. Kwok, K. Rector, J.R. Hill, H.A. Schwettman, D.D. Dlott, M.D. Fayer, Phys. Rev. Lett. 77 (8) (1996) 1648.

9. K.W. Berryman, Design operation, and applications of a far-infrared free electron laser, Ph.D. Dissertation, Stanford University, CA, USA, 1995.

10. P. Haar, Pulse stacking in the Stanford external cavity and photo-induced reflectivity in the infrared, Ph.D. Dissertation, Stanford University, CA, USA, 1997.

11. Crosson, E.R., Haar, P., Marcus, G.A., Schwettman, H.A., Paldus, B.A., Spence, T.G., Zare, R.N. Rev. Sci. Instrum. 1999; 70(1):4.

¹The author would particularly like to thank Dercy Pooch and his group at DESY for advise and assistance in producing the Stanford structures.

FELs, nice toys or efficient tools?

A.F.G. van der Meer,     FOM Institute for Plasma Physics ‘Rijnhuizen’, Edisonbaan 14, 3439 MN Nieuwegein, The Netherlands. E-mail address: a.f.g.vandermeer@rijnh.nl

Abstract

An FEL is an intrinsically interesting device and pushing its performance presents a natural challenge to a physicist. Nonetheless, the main justification for doing FEL research is of course its potential as a unique, versatile source of radiation to be employed for something useful. After 25 years of FEL research, one may wonder how efficient these tools have become. In this paper, I will reflect on this issue from the perspective of 10 years of operation of FELIX as a user facility.© 2004 Elsevier B.V. All rights reserved.PACS: 41.60Cr; 42.62Fi; 82.80Gk; 82.50Bc

Keywords

Free-electron lasers

Infrared

Spectroscopy

1 Introduction

Whereas the (technological) challenges of a free electron laser probably present the main personal motivation for the people active in the field, its potential as a unique and versatile source of radiation that could be used for a variety of purposes, has always been the main justification for the research on FELs. More than 25 years after the ‘invention’ and first demonstration of operation of an FEL [1], it seems appropriate to ask ourselves how far we have come on the way to realizing this potential. In this paper, I will address this question with the experience gained in 10 years of operation of the IR User Facility FELIX. I will therefore start with an evaluation of the efficiency of FELIX as a tool for scientific research, before attempting to generalize to other (types of) FELs and applications.

2 The IR user facility ‘FELIX’

FELIX consists of a normal conducting, 12−45 MeV linear accelerator that alternatively drives a far-IR FEL with partial waveguide that covers the spectral range from 25 to 250 μm, or a mid-IR FEL with a spectral range from 5 μm (3 μm on 3rd harmonic) to 40 μm. The general layout of FELIX is shown in Fig. 1. As usual for this type of linear accelerator, the output consists of bursts (macropulses) of micropulses. The spacing between the micropulses is either 1 ns (1 GHz-mode) or 40 ns (25 MHz-mode), the latter corresponding to the roundtrip time of the 6 m cavity. Using a transient optical switch, it is possible to slice a single micropulse out of the pulse train with an efficiency of more than 50%. The main characteristics of the output are listed in Table 1. The facility is operated in two-shift mode, 5 days a week, providing more than 3000 h of beam time for user experiments.

Table 1

Characteristic parameters of FELIX

Fig. 1 General layout of FELIX showing the two beamlines for far- and mid-infrared generation.

3 User experiments at FELIX

Presently, these experiments fall predominantly in one of two classes: relaxation phenomena in condensed matter or spectroscopy of gas phase species, (bio)molecules and clusters, either neutral or ionized. Experiments in the first class will use the 25 MHz or single-micropulse mode in view of the relaxation times involved, not only of the primary process but also of the temperature transient due to the energy absorbed. Especially in the case of biological samples the latter is usually more limiting. The second class is characterized by (very) low absorption cross-sections and, because detection is based on dissociation or ionization of the species, typically a strongly nonlinear dependence of the signal on the laser fluence, implying the use of the 1 GHz-mode.

For illustration, two examples of either class will be discussed briefly. The first example of the first class concerns the relaxation of the stretch vibration of hydrogen and deuterium in amorphous silicon. Hydrogen is often used to passivate the dangling bonds, thereby enhancing the characteristics of this technologically important material, but the beneficial effect strongly reduces with time. Recently, it was found that this aging effect is much smaller when deuterium is used instead. The experimental result for hydrogen is given in Fig. 2 [2] and for deuterium in Fig. 3 [3]. Whereas the decay for deuterium is clearly single-exponential, it is not for hydrogen. This observation can be related to the striking difference in energy decay channel for both cases: whereas the relaxation for H occurs mainly via an almost resonant energy transfer to the bending mode at the same site and is therefore a local event, the relaxation of D primarily involves phonons, so modes of the bulk. This non-localized energy release is now believed to be the main reason for the strongly reduced ‘aging’ effect.

Fig. 2 Relaxation of the Si–H stretch vibration in amorphous Si: time dependence (upper panel) and temperature-dependent rate (lower panel) [2].

Fig. 3 Relaxation of the Si–D stretch vibration in amorphous Si: time dependence (upper panel) and temperature-dependent rate (lower panel) [3].

In the second example, far-IR radiation of FELIX was used to probe the time-dependent exciton density in a GaAs quantum well after excitation across the bandgap with a synchronized pump laser. The result was compared to a conventional measurement, i.e. by monitoring the luminescence of the sample in the spectral interval associated with the presence of excitons (Fig. 4) [4].

Fig. 4 The transient transmission change of the FIR probe pulse is shown in the upper panel for three cases: exciting at resonance (middle), 36 meV below resonance (bottom) and 62 meV above resonance (top). In the lower panel the transient photoluminescence at the ‘exciton peak’ is plotted while exciting 62 meV above resonance [4].

Whereas there is a prompt absorption signal, resulting from 1s−2p transitions of the bound electron–hole pairs, when the sample is excited very close to the bandgap, there is no signal if the excitation is well above the bandgap. In the latter case, the conventional method does show a rapid signal though. This result seems to provide experimental evidence for the possibility of exciton-like emission without excitons being present. Recently, this behaviour was predicted theoretically and is attributed to a correlation of the unbound electron and hole population. The rise time of the photo-luminescence reflects the time for the electrons and holes to relax into the low-lying k-states at the bottom of the band.

As a first example of an experiment of the second class, a measurement of the IR spectrum of titanium-carbide clusters is shown in Fig. 5. A YAG-laser is used to ablate material from a titanium rod and subsequently a gas-puff, in this case methane, is applied, resulting in a molecular beam containing titanium-carbide clusters.

Fig. 5 The infrared spectra of Ti8C12, Ti8C11 and Ti13C14 clusters measured using IR-REMPI as a function of the FELIX wavelength (three upper traces) [5]. The insets show the previously reported structures. As a comparison the lowest trace shows the EELS spectrum of bulk TiC(100).

Previous work, using UV-lasers to ionize these clusters, showed that clusters consisting of 8 titanium and 12 carbon atoms were particularly stable, as well as clusters with, respectively, 14 and 13 atoms. Due its high fluence, it proved to be possible to ionize the clusters with the IR beam of FELIX and record an infrared spectrum [5]. A very strong signal around 8 μm, characteristic for a C–C stretch vibration, is present for the Ti8C12 cluster and absent for the Ti14C13 cluster, in accordance with the ordering of the C-atoms in the structures that had been proposed (see insets in Fig. 5). The signal around 20 μm for the Ti14C13 cluster shows a great similarity with a spectrum recorded with EELS on bulk TiC, suggesting that it is indeed a nano-crystal. Moreover, this spectral feature is now believed to be the source of a hitherto unexplained strong emission around post asymptotic giant-branch stars [6]. In a similar manner, by looking for dissociation rather than ionization, the IR spectrum of ions can be recorded. Usually the ions are confined in a trap in order to increase the density. Recently, the IR spectrum of two water molecules bridged by a proton was measured using a tandem ion trap separated by a mass selector (Fig. 6) [7]. Using an FTICR high-resolution mass spectrometer, proton bridging of larger molecules was investigated [8]. The similarity of the main features of the spectra shows that these are really characteristic for proton bridging.

Fig. 6 IR-spectra of protonated water dimers [7] and p-methyl ether dimer, p-ethyl ether dimer and p-diglyme, respectively [8].

4 Performance assessment

From the above, I hope it is obvious that FELIX can be used for high-quality research, but what about productivity? A typical 3rd generation synchrotron has some 40 beam lines and an annual output of some 500 papers, whereas the number of user papers for the FELIX facility is only 20−25 per year. Given the difference in investment and running costs, typically a factor of 15, the relative output of FELIX however comes close to that of a synchrotron. So it seems justified to conclude that an FEL can be an efficient tool for scientific research. On the other hand, the total number of user papers that have been produced at FEL facilities is substantially less than twice the annual production of a single synchrotron! So, on the whole, the impact of FELs on scientific research has until now been almost negligible and does not yet justify the efforts put into their development.

5 Outlook

May we expect this balance to improve in the future, either by more and high-impact scientific research applications or applications in different areas, for example industrial processing? Following the successful demonstration of a 1 kW average power IR FEL, based on superconducting accelerator technology combined with energy recovery at Jefferson labs, very high average power (> 100 kW) units are now under study. At these power levels, military applications once again appear at the horizon, but history-based skepticism still seems justified. High average power would in principle also greatly increase the number of applications in industrial processing, and a successful application in a billion dollar market could already affect the balance. An industrial application that is clearly economically sound still needs to be demonstrated though. Returning to the use of FELs for scientific research, it should be noted that to my opinion the future of IR FELs is not very bright. Primarily because of the rapid development of alternative sources over the past 10 years, especially those based on today’s work horse, the Ti:Sapphire laser. Table-top sources based on parametric generation are very competitive in the mid-IR, especially in the field of relaxation studies for which the pulse structure is often better suited for the time scales involved. In the far-IR, the micropulse energies available from FELs are still unchallenged and applications requiring high peak powers to pump the system far from equilibrium will also in the foreseeable future require the availability of an FEL. If the far-IR radiation is just used for probing, it is however not power that counts but sensitivity. Broad-band, ‘single-cycle’ THz pulses can be generated by focussing a short-pulse laser, again usually a Ti:Sapphire laser, onto a semiconductor such as GaAs. Typically these pulses have rather low energy, at the pJ level, but by using detection schemes that are based on (e.g. electro-optic) sampling with part of the pulse used for the generation, a very high sensitivity has been obtained: about 1 part in 10⁸ for a 1 s integration time [9]. By varying the delay between the THz pulse and the sampling beam, very high time resolution can also be obtained. As the detection is often sensitive to the electric field of the THz pulse rather than its intensity, it has the added advantage that also phase information of the different frequency components within the bandwidth is obtained. As a matter of fact, the use of an FEL for the experiments given above as examples of the first class can no longer be justified. FELs will most likely continue to have an advantage for experiments of the second class. This field of research is rapidly expanding at our facility and recently funding was obtained for the construction of a third beam line [10] that will allow this class of experiments to benefit from the much higher powers present within the FEL cavity. A schematic of FELICE, the Free Electron Laser for Intra-Cavity Experiments, that should cover the wavelength range from 3 to 100 μm, is shown in Fig. 7. Nonetheless, the niche for IR FELs has decreased quite significantly compared to when FELIX started operation and even though there will certainly be a need for a number of FEL-based IR facilities in the future, it is not realistic to expect an increase of the impact of FELs in the IR. Also in this respect the recent progress made towards very short-wavelength operation of FELs is of course very encouraging for the FEL community. But based on the experience in the infrared it should be realized that the success of X-FELs is not necessarily guaranteed merely by their realization.

Fig. 7 General layout of the FELICE cavity including two intra-cavity setups: for IR-REMPI and IR-MPD experiments a high-resolution FTICR mass spectrometer and a molecular beam machine.

References

1. Elias, L.R., et al. Phys. Rev. Lett. 1976; 36:717.

2. van der Voort, M., et al. Phys. Rev. Lett. 2000; 84:1236.

3. Wells, J-P.R., et al. Phys. Rev. Lett. 2002; 89:125504.

4. R. Chari, et al., Phys. Rev. Lett., submitted for publication.

5. van Heijnsbergen, D., et al. Phys. Rev. Lett. 1999; 83:4983.

6. von Helden, G., et al. Science. 2000; 288:313–316.

7. Asmis, K.R., et al. Science. 2003; 299:1375.

8. D.T. Moore, et al., Chem. Phys. Chem., (2004) in press.

9. Leitenstorfer, A., et al. Physica. 2002; B:248. [314].

10. Militsyn, B.L., et al. Nucl. Instr. and Meth. A. 2003; 507:494.

First lasing at the high-power free electron laser at Siberian center for photochemistry research

E.A. Antokhin, R.R. Akberdin, V.S. Arbuzov, M.A. Bokov, V.P. Bolotin, D.B. Burenkov, A.A. Bushuev, V.F. Veremeenko, N.A. Vinokurov*, P.D. Vobly, N.G. Gavrilov, E.I. Gorniker, K.M. Gorchakov, V.N. Grigoryev, B.A. Gudkov, A.V. Davydov, O.I. Deichuli, E.N. Dementyev, B.A. Dovzhenko, A.N. Dubrovin, Yu.A. Evtushenko, E.I. Zagorodnikov, N.S. Zaigraeva, E.M. Zakutov, A.I. Erokhin, D.A. Kayran, O.B. Kiselev, B.A. Knyazev, V.R. Kozak, V.V. Kolmogorov, E.I. Kolobanov, A.A. Kondakov, N.L. Kondakova, S.A. Krutikhin, A.M. Kryuchkov, V.V. Kubarev, G.N. Kulipanov, E.A. Kuper, I.V. Kuptsov, G.Ya. Kurkin, E.A. Labutskaya, L.G. Leontyevskaya, V.Yu. Loskutov, A.N. Matveenko, L.E. Medvedev, A.S. Medvedko, S.V. Miginsky, L.A. Mironenko, S.V. Motygin, A.D. Oreshkov, V.K. Ovchar, V.N. Osipov, B.Z. Persov, S.P. Petrov, V.M. Petrov, A.M. Pilan, I.V. Poletaev, A.V. Polyanskiy, V.M. Popik, A.M. Popov, E.A. Rotov, T.V. Salikova, I.K. Sedliarov, P.A. Selivanov, S.S. Serednyakov, A.N. Skrinsky, S.V. Tararyshkin, L.A. Timoshina, A.G. Tribendis, M.A. Kholopov, V.P. Cherepanov, O.A. Shevchenko, A.R. Shteinke, E.I. Shubin and M.A. Scheglov,     Budker Institute of Nuclear Physics, Acad. Lavrentyev Prospect 11, 630090 Novosibirsk, Russia. E-mail address: vinokurov@inp.nsk.su

*

Corresponding author. Tel.: +7-3832-394003; fax: +7-3832-342163.

Abstract

The first lasing near wavelength 140 μm was achieved in April 2003 on a high-power free electron laser (FEL) constructed at the Siberian Center for Photochemical Research. In this paper, we briefly describe the design of FEL driven by an accelerator-recuperator. Characteristics of the electron beam and terahertz laser radiation, obtained at the first experiments, are also presented in the paper.© 2004 Elsevier B.V. All rights reserved.PACS: 41.60.Cr

Keywords

Free electron laser

Energy recovery linac

1 Introduction

A new source of terahertz radiation was commissioned recently in Novosibirsk. It is CW FEL based on an accelerator–recuperator, or an energy recovery linac (ERL). It differs from the earlier ERL-based FELs [1,2] in the low-frequency non-superconducting RF cavities and longer wavelength operation range. The terahertz FEL is the first stage of a bigger installation, which will be built in 3 years and will provide shorter wavelengths and higher power. The facility will be available for users in 2004. The first radiation study results are discussed in this paper.

2 Accelerator–recuperator

Full-scale Novosibirsk free electron laser is to be based on multi-orbit 50 MeV electron accelerator–recuperator. It is to generate radiation in the range 3 μm–0.3mm [3,4]. The first stage of the machine contains a full-scale RF system, but has only one orbit. Layout of the accelerator-recuperator is shown in Fig. 1. The 2 MeV electron beam from an injector passes through the accelerating structure, acquiring 12 MeV energy, and comes to the FEL, installed in the straight section. After interaction with radiation in the FEL the beam passes once more through the accelerating structure, returning the power, and comes to the beam dump at the injection energy. Main parameters of the accelerator are listed in Table 1.

Table 1

Accelerator parameters (the first stage)

Fig. 1 Scheme of the first stage of Novosibirsk high-power free electron laser.

The FEL is installed in a long straight section of a single-orbit accelerator–recuperator. It consists of two undulators, a magnetic buncher, two mirrors of the optical resonator, and an out-coupling system. Both electromagnetic planar undulators are identical. The length of an undulator is 4 m, period is 120 mm, the gap is 80 mm, and deflection parameter K is up to 1.2. One can use one or both undulators with or without a magnetic buncher. The buncher is simply a three-pole electromagnetic wiggler. It is necessary to optimize the relative phasing of undulators and is used now at low longitudinal dispersion Nd<1. Both laser resonator mirrors are identical, spherical, 15 m curvature radius, made of gold plated copper, and water cooled. In the center of each mirror there is a 3.5 mm diameter hole. It serves for mirror alignment (using He–Ne laser beam) and output of small amount of radiation. The distance between mirrors is 26.6 m. The outcoupling system contains four adjustable planar 45 copper mirrors (scrapers). These mirrors cut the tails of Gaussian eigenmode of the optical resonator and redirect radiation to calorimeters. This scheme preserves the main mode of optical resonator well and reduces amplification of higher modes effectively.

3 FEL commissioning

For FEL commissioning, we used both undulators. Beam average current was typically 5 mA at repetition rate 5.6 MHz, which is the round-trip frequency of the optical resonator and 32th subharmonics of the RF frequency f≈180 MHz. Most of measurements were performed without scrapers recording radiation flux from one of the mirror apertures. Instead of fine tuning of the optical resonator length we tuned the RF frequency. The tuning curve is shown in Fig. 2.

Fig. 2 Laser intensity vs RF frequency detuning f-180400 kHz (diamonds at repetition rate 5.6 MHz, triangles at repetition rate 2.8 MHz).

Typical results of spectrum measurement with rotating Fabri–Perot interferometer are shown in Fig. 3. They were used to find both wavelength and line width of radiation. Radiation wavelengths were in the range 120−180 μm depending on the undulator field amplitude. The shortest wavelength is limited by the gain decrease at a low undulator field, and the longest one—by the optical resonator diffraction loss increase. Relative line width (FWHM) was near 3 · 10−3. The corresponding coherence length λ²/2Δλ = 2 cm is close to the electron bunch length, therefore we, probably, achieved the Fourier-transform limit.

Fig. 3 Results of the Fabri–Perot interferometer rotation angle scanning (laser wavelength λ = 136 μm).

The loss of the optical resonator was measured with a fast Schottky diode detector [5]. Its typical output is the pulse sequence with 5.6 MHz repetition rate. Switching off the electron beam, we measured the decay time (see Fig. 4). The typical round-trip loss values were from 5% to 8%.

Fig. 4 Time dependence of the output radiation power after switching the electron beam off.

The FEL oscillation was obtained not only at f0 = 5.6 MHz bunch repetition rate, but at f0/2, f0/3, f0/4 and 2f0/3. The time dependence of intensity at bunch repetition rate f0/4 is shown in Fig. 5. Radiation decay time (and therefore resonator loss) can also be measured from this dependence. The dependence of power on loss is shown in Fig. 6. For example, operation at bunch repetition rate f0/4 corresponds to four time more loss per one amplification. It indicates that our maximum gain is about 30%.

Fig. 5 The output radiation time dependence. Electron bunch repetition rate 1.4 MHz is four time less then the optical resonator round-trip frequency 5.6 MHz.

Fig. 6 Average intra-cavity power vs loss per one amplification.

The absolute power measurements were performed in two ways. First we measured the power coming through the hole in the mirror without scrapers. Output coupling is very weak in this case, so the power was about 10 W. It corresponds to intra-cavity average power near 2 kW. Other measurements were performed with two (right and left) scrapers inserted. The insertion depth was chosen to decrease intra-cavity power twice. The measured power in each calorimeter was 20 W. Taking into account other resonator loss one can estimate the total power loss as 100W. The electron beam power was 50kW. Therefore, an electron efficiency is about 0.2%. The possible explanation of such a low value is too long undulator and high electron energy spread. Attempts to get oscillation with one undulator switched off are in progress. Possible way for decreasing of the energy spread—the installation of a 3rd harmonic (540 MHz) cavity—is under examination.

4 Further development

A beamline for transport radiation out of the accelerator hall to the user station rooms is under construction. The first experimental station is designed. The facility is to start operation for users in 2004. Expected radiation parameters for users are shown in Table 2.

Table 2

Expected radiation parameters for users

References

1. Neil, G.R., et al. Phys. Rev. Lett. 2000; 84:662.

2. Minehara, E.J. Nucl. Instr. and Meth. A. 2002; 483:8.

3. Gavrilov, N.G., et al. IEEE J. Quantum Electron. 1991; QE-27:2626.

4. V.P. Bolotin, et al., Proceedings of FEL-2000, Durham, USA, 2000, p. 11-37.

5. Kubarev, V.V., Kazakevich, G.M., Jeong, Y.V., Lee, B.C. Nucl. Instr. and Meth. A. 2003; 507:523.

First lasing of the IR upgrade FEL at Jefferson lab

C. Behrea, S. Bensona*, G. Biallasa, J. Boycea, C. Curtisa, D. Douglasa, H.F. Dyllaa, L. Dillon-Townesa, R. Evansa, A. Grippoa, J. Gubelia, D. Hardya, J. Heckmana, C. Hernandez-Garciaa, T. Hiatta, K. Jordana, L. Mermingaa, G. Neila, J. Preblea, H. Ruttb, M. Shinna, T. Sigginsa, H. Toyokawac, D.W. Waldmana, R. Walkera, N. Wilsona, B. Yunna and S. Zhanga,     aJefferson Lab, MS 6A, TJNAF, Newport News, 12000 Jefferson Avenue, VA 23606, USA; bSouthampton University, Southampton, UK; cKEK, Tsukuba, Japan. E-mail address: felman@jlab.org

*

Corresponding author. Tel.: 1-757-269-5026; fax: 1-757-269-5519.

Abstract

We report initial lasing results from the IR Upgrade FEL at Jefferson Lab (Proceedings: 2001 Particle Accelerator Conference, IEEE, Piscataway, NJ, 2001). The electron accelerator was operated with low average current beam at 80 MeV. The time structure of the beam was 120 pC bunches at 4.678 MHz with up to 750 μ8 pulses at 2 Hz. Lasing was established over the entire wavelength range of the mirrors (5.5-6.6 μpm). The detuning curve length, turn-on time, and power were in agreement with modeling results assuming a 1 ps FWHM micropulse. The same model predicts over 10 kW of power output with 10 mA of beam and 10% output coupling, which is the ultimate design goal of the IR Upgrade FEL. The behavior of the laser while the dispersion section strength was varied was found to qualitatively match predictions. Initial CW lasing results also will be presented.© 2004 Elsevier B.V. All rights reserved.PACS: 29.20.c; 29.27.–a; 41.60.Cr;