Free Electron Lasers 2002: Proceedings of the 24th International Free Electron Laser Conference and the 9th FEL Users Workshop, Argonne, Illinois, U.S.A., September 9-13, 2002

This book contains the Proceedings of the 24th International Free Electron Laser Conference and the 9th Free Electron Laser Users Workshop, which were held on September 9-13, 2002 at Argonne National Laboratory. Part I has been reprinted from Nucl. Instr. and Meth. A 507 (2003), Nos. 1-2.

• Language: English
• Publisher: Elsevier Science
• Release date: Dec 2, 2012
• ISBN: 9780080930428

Book preview

Part I

Section 1

FEL Prize and New Lasing

Chaos studies on the super-ACO free electron laser

M.E. Couprieab*,     aService de Photons, Atomes et Molécules, CEA/DSM/DRECAM, 91 191 Gif-sur-Yvette, France; bLaboratoire pour l’Utilisation du Rayonnement Électromagnétique, Université Paris-Sud, Bât. 209D, BP 34, 91898 Orsay Cedex, France. E-mail address: [e-mail]

*Corresponding author. Laboratoire pour l’Utilisation du Rayonnement Électromagnétique, Université Paris-Sud, Bât. 209D, BP 34, Orsay Cedex 91898, France. Tel.: + 33-1-64-46-80-44; fax: [phone]. E-mail address: (M.E. Couprie).

A laser with parameters modulated at a frequency f may respond not only at f and its harmonics, but also at is sub-harmonic frequencies f/n. The laser behaviour can become irregular though remaining deterministic. A gain modulation has been applied on the Super-ACO Free Electron Laser close to its natural frequency, via a change of the synchronization between the electron bunch and the optical pulse (detuning). Different macrotemporal structures can be observed such as 1T regime in which the laser is pulsed at the modulation period, 2T, chaos and so on. Such transitions can also be measured on the Free Electron Laser pulse temporal position. Bifurcation diagrams and attractors have also been recorded. Comparisons with simulations will be also given.© 2003 Elsevier Science B.V. All rights reserved.PACS: 41.60.Cr; 42.65. Sf

Keywords

FEL

Chaos

Storage ring

1 Introduction

The emergence of lasers, being temporally and spatially coherent light sources, led to a wide variety of interest in many domains. The stability of a laser source is then a crucial element for fundamental applications in science and use in technology. The non-linearity of a laser gain system can nevertheless induce complex regimes, including stable period and harmonic regimes, limit cycles, and even chaotic regimes. Different analysis have been carried out on conventional laser sources, such as CO2 laser [1–3], He-Ne laser [4], Yag laser [5, 6], laser diode [7]. The gain modulation can be introduced by different means: electro-optical modulator [1], tilt of the resonator mirror [4], saturable absorber pressure or laser frequency [2]. Chaos can also appear in presence of an optical feedback [7]. Apart from the spiking structure of the Self-Amplified Spontaneous Emission (SASE) where chaotic polarized radiation has been observed [8], chaos has been studied on different Free Electron Lasers (FELs) sources, such as Raman devices [9], and Compton FELs. On infra-red LINAC based FELs, Hopf bifurcations and transition to chaos were reported on ELSA [10], limit cycles and period doubling versus cavity detuning were discussed for short pulse FELs such as FELIX [11], chaotic regimes were reported on the JAERI FEL for a modification of the detuning [12]. Sidebands generation and transition to chaos were seen on ELSA [13], on the Stanford FEL [14] with various modelisations [15]. First studies on chaos on a storage ring FEL on ACO and Super-ACO showed that the FEL macrotemporal structure in the millisecond time scale follows the laws of deterministic chaos [16]. Besides, it has often been observed on storage ring FELs that the macrotemporal structure could present a pulsed regime with random intensities [17, 18]. Further theoretical investigation carried out with a simplified model allowed the Lyapunov characteristic exponents to be determined [19]. More recent numerical analysis based on a complete modelisation of the longitudinal detuning of the FEL was performed and compared to experimental results [20]. New results obtained on the Super-ACO Free Electron Laser are here presented.

2 The Longitudinal Dynamics of a Storage Ring FEL

In addition to the microtemporal structure reproducing the recurrence of the electron bunches in the ring at a high repetition rate (8 MHz for Super-ACO), the FEL exhibits a macrotemporal structure at the millisecond time scale [21, 22]. Depending on the gain and of the losses of the laser system, this regime appears systematically for given values of the synchronisation between the optical wave in the optical resonator and the electron bunches stored in the ring. The detuning Δ is defined as Δ = Tel - Tph where Tel is the time interval between two adjacent electron bunches and Tph is the round-trip time of the light wave in the optical cavity. CW regimes at the millisecond time scale are observed around perfect tuning and for large detunings, as shown in Fig. 1a. Pulsed regimes appear for intermediate detunings, as illustrated in Fig. 1b.

Fig. 1 (a) FEL Intensity, Measured with a Photomultiplier, Versus the Tuning Condition, Modified by a sweep of the RF frequency of Super-ACO. 1 Hz frequency change corresponds to 1.2 fs detuning change. (b) Intensity versus time for a pulsed FEL, exhibiting a natural frequency of 200 Hz. Case of Super-ACO operated at 800 MeV, with 38.2 mA, the FEL being at 350 nm, with a gain of 1.5% and cavity losses of 0.8%. (c) Simulations of the FEL evolution for different detunings. Parameters of the simulation: 800 MeV, 40 mA, gain of 2%, cavity losses of 0.5%.

This behaviour can be described by a phenomenological model, following the evolution of the FEL intensity profile yn(t) at each pass n [23, 24]. The spontaneous emission is is represented as a monochromatic wave and the optical wave is assumed to be sharply centered on the resonant wavelength. At each pass n, the light wave is amplified according to the gain term g(t), reflected on the cavity mirror of reflectivity R, and the spontaneous emission from the undulator adds up. The slippage being extremely small, it is here neglected. It comes [21]:

(1)

(2)

where In is the total laser intensity and Ieq the FEL equilibrium intensity and τ the longitudinal coordinate. Because of the FEL heating, the normalised electron bunch energy spread at pass n evolves as

(3)

where τs is the synchrotron damping time, ∑n ), σγn being the energy spread at pass n, σγoff laser off, and σγeq at equilibrium (∑ = I = 1). The FEL saturation leads to a reduction of the gain at pass n gn,0 for the centre of the bunch distribution, according to [20]:

(4)

where goff the is the gain at the laser start-up, P are the cavity losses. The gain dependence versus τ can be written as g(τ) = g0e−τ²/2σ²ln, assuming a Gaussian distribution of RMS dimension σln, go being the maximum gain for τ = 0. For Δ = 0 and considering a small perturbation to the equilibrium state, the energy spread evolution reduces to a second-order differential equation [25], allowing to understand the pulsed regimes of the FEL. The insertion of the detuning [23–26] in the model showed CW regimes at the millisecond time scale around perfect tuning whereas pulsed regimes are obtained for intermediate detunings. For larger desynchronisation between the electron bunches circulating in the ring and the optical pulses in the resonator, the FEL is again CW, but with larger temporal and spectral distributions. The model can properly reproduce the experimental behaviour of the FEL intensity versus detuning, as shown in Fig. 1c.

An external modulation of the tuning condition can be added to the model, such as

(5)

where f and a are the frequency and the amplitude of the modulation, and b is the detuning around which the modulation is performed. f and a are the control parameters of the system in our experiment.

3 The Response of the Super-ACO FEL to a Detuning Modulation

The Super-ACO FEL presents a stable CW regime around perfect synchronism. Various studies on the different sources of perburbations have been extensively carried out in order to provide a source as stable as possible for the users [27]. A longitudinal feedback system has been developed in order to compensate drifts in synchronisation [28]. An external modulation can be experimentally applied to the RF frequency pilot generator, allowing a controlled change of the FEL detuning.

The FEL intensity, measured with a photomultiplier, is recorded for different values of the amplitude and the frequency of the modulation. When the amplitude of the applied modulation is increased, in the case of a modulation at 660 Hz presented in Fig. 2a, the FEL still adopts a pseudo CW regime for very low amplitudes of modulations. Then, for a = 7.3 Hz, a 1T regime appears, for which the FEL is pulsed at the same period as the one of the applied modulation. Then, for larger amplitudes (a = 12 Hz), the FEL adopts a 2T regime for which the period of the FEL is twice that of the modulation. Then (a = 20 Hz), the FEL bifurcates into a chaotic regime, where neither the period nor the intensity of the FEL seems to be periodic. A signature of the deterministic chaos can be derived from the analysis of trajectories having very similar behaviours. Sequences in which the trajectories are identical for several periods of modulation can then diverge, while the difference is growing exponentially. For even larger amplitudes of the modulation (a = 46 Hz), a 3T regime is adopted. The frequency analysis of the FEL intensity allows a simple determination of the regime, which is adopted by the FEL. As shown in Fig. 3, the 2T regime corresponds to a clear sharp peak at f/2, and less intense peaks at f/4 and f/8 besides the peaks at f and its harmonics. In the chaotic regime, a broad background suddenly appears, even if some peaks of sub-harmonic frequencies temporally remain above the continuous spectrum. The FEL response can also depend on f. When the frequency of the modulation is more than twice the natural frequency of the FEL, T/2 regimes are also seen. A scan of the amplitude of the modulation can be applied to the RF pilot, allowing the different regimes to be followed. In addition, the system is not completely reversible, and the FEL response is not symmetrical neither for an increase or a decrease of the amplitude of modulation nor for positive and negative detuning. Such a scenario of a period-doubling cascade leading to chaos and periodic regimes was also reported in CO2 lasers with loss modulation obtained by an intra-cavity electro-optical modulator [1]. The period doubling cascade, and inverse cascade, which is also observed, are a signature that the irregular part of the evolution is due to deterministic chaos [29]. Besides, the FEL response can also be studied for a given amplitude while the frequency of modulation is varied.

Fig. 2 FEL Intensity for Different Amplitudes. Case of Super-ACO Operated at 800 MeV, the FEL being at 350 nm, with a Gain of 2% and Cavity Losses of 0.84%. (a) Experimental Results for an External Modulation at 660 Hz. (b) Numerical Simulations Corresponding to the Model presented in Eq. (1)–(5). b = 2 Hz, f = 320 Hz.

Fig. 3 Frequency Domains for the Evolution of the FEL Corresponding to Fig. 2: (a) 2T Regime; (b) 3T Regime; and (c) Chaotic Regime. The Intensity Fourier Transform (IFT) of the Laser Intensity Versus Time has been Performed.

First simulations on the theoretical behaviour in response to a gain modulation performed in Ref. [16] allowed to reproduce qualitatively the experimental behaviour, but without a proper description of the detuning. Further analysis based on this simplified model were carried out analytically in Ref. [17]. Nevertheless, the importance played by the detuning on the FEL dynamics [21] led to more recent analysis with a more complete description, including the detuning [20], and allowed to reproduce qualitatively the experimental results. For b = 0 and for small values of the amplitude a (corresponding to the central CW zone), the FEL adopts a stable 1T regime. For larger values of a, the FEL usually presents a chaotic regime. For small values of b (but not zero) corresponding to a slight desynchronisation between the electron bunches and the optical pulses bouncing in the optical resonator, the FEL behaviour is rather different. In general, when a is not larger than half the width of the detuning curve, the FEL exhibits a 1T regime (see Fig. 2b). The cascades seem to appear for larger values of a. The behaviour versus the frequency of the modulation shows also some cascades.

The dynamics of the Super-ACO FEL has been actively studied by the mean of a double sweep streak camera [30] allowing the determination of the FEL pulse position with respect to a reference one, and the FEL pulse width, besides the FEL intensity. Fig. 4 illustrates the response of the FEL in different cases for a modulation at 320 Hz with the following sequence: 1T, 3T, chaos, 3T, 1T, 2T, 4T, 3T, 5T, 3T, 2T, 3T chaos 1T, T/2. In Fig. 4a, a clear 1T regime is seen, with a periodically stable intensity, position and width of the FEL. Fig. 4b shows sudden jumps of the FEL longitudinal position. In the middle appears a pattern in which the intensity seems to be T/2, but in fact there are two distinct positions of the FEL, corresponding to a 1T regime. In Fig. 4c, the 3T regime is presented, with two different positions and drift direction of the FEL pulse inside the 3T pattern. In Fig. 4d, the image shows irregular intensity, position and FEL width, and corresponds to a chaotic behaviour.

Fig. 4 Double Sweep Streak Camera Image of the Super-ACO FEL in Response to a Gain Modulation Applied on the Detuning Condition, f = 320 Hz, b ∼ 0: (a) 2a = 28.7 Hz, 1T regime; (b) 2a = 52.3 Hz, 3T Regime, Δpos = 48 ps; (c) 2a = 98.7 Hz Chaos, Δpos = 208 ps; (d) 2a = 98.7 Hz, chaos, Δpos = 212 ps. Δpos Corresponds to the Different Vertical Positions in the Streak Camera Image.

Simulations performed with the above model also clearly show that the bifurcation happens at the same time for the FEL intensity, the FEL position and width, as illustrated in Fig. 5.

Fig. 5 Simulations of the Evolution of the Laser Intensity, the Laser Position and Pulse Width.

4 Reconstruction of the Attractors and Bifurcation Diagrams

The measurements of all the variables of the phase space in studies on a laser being often impossible, trajectories, which are topologically identical to those in the original phase space are usually reconstructed from time series of a single measured quantity, X(t), being currently the laser intensity I(t) [1]. For instance, the representation of the attractor can be drawn from the representation of I(t), I(t + t), I(t + 2t) or I(t) and its successive derivatives, with t being an arbitrary time. The latter method requiring electronic processing of the signal, it is easier to reconstruct the attractors via successive values of I(t) with a simple processing of the data. Typically, t which is clearly a multiple of the sampling time Δ t, is chosen to be 12Δ t, or one-fifth of the modulation period. The attractors I(t)cos(2πft), I(t)sin(2πft), I(t + t) in cylindrical coordinates for the intensity records of Fig. 2a are shown in Fig. 6a. The stratification seen in the chaotic regime is a property of strange attractors. Corresponding PoincaréeA sections are illustrated in Fig. 6b, for which the number of intercepts n of the attractor with the plane of the PoincaréeA section is a signature of the "nT" regime, and where a folded shape which does not fill the whole space corresponds to the chaos. From the observation of the attractors, the origin of the irregular patterns of the FEL can be attributed to the deterministic chaos rather than to noise.

Fig. 6 (a) I(t)cos(2πft), I(t)sin(2πft), I(t + τ) Attractors Reconstructed from Intensity Evolution of Fig. 3. 2T Regime for a = 12 Hz, Chaos for a = 20 Hz, 3T Regime for a = 46 Hz. (b) Corresponding Poincaré Sections.

The evolution of the system when a given control parameter is varied can be followed by the observation of the bifurcation diagram, in which the intensity sampled at a definite time of the modulation period is plotted versus the control parameter. As the stroboscopy is synchronized with the modulation, a nT regime delivers a n-valued output and random answers reflect the statistics of the chaos. Fig. 7 illustrates a 1T–2T bifurcation for a change of the frequency of the modulation in (a), and a 1T, 2T, chaos, 2T, 3T sequence for a modification of the amplitude of modulation. Bifurcation diagrams are powerful tools for the observation of the transitions between the different regimes.

Fig. 7 Bifurcation Diagrams (a) for a Change of f between 200 and 400 Hz, a = 14.3 Hz and (b) for a Change of a between 0 and 60 Hz, f = 600 Hz.

5 Conclusion

Clear bifurcation and chaos sequences have been observed in the response of the FEL to a detuning modulation. In a further work, the sequences (period doubling, inverse cascade, logistic map, etc.) corresponding to various parameters set-ups should be studied, for the determination of the basins of attraction. Clear analysis of the data will allow the determination of the Lyapunov coefficients, and the comparison with the values from a simplified model in Ref. [17]. Besides, the slight asymmetry in the electron bunch distribution due to the microwave instability [31] should be considered in the model for the simulations. It would also be interesting to perform a gain modulation without a change of the detuning condition, and to compare the observed sequences. A further step will consist on elaborating a stabilization of unstable steady states of the system [32].

Acknowledgements

This work is a prolongation of an activity established with my former collaborator M. Bill-ardon, with whom I started in the FEL field and carried out with great pleasure joint activity. This work has been stimulated by active discussions with my colleagues from Lille University, P. Glorieux, S. Randoux and D. Hennequin. The results achievements have also benefit from the work of my team collaborators, D. Garzella, T. Hara, L. Nahon, R. Bakker, D. Nutarelli, R. Roux, B. Visentin, E. Renault, G. De Ninno and his friend D. Fanelli, C. Thomas, C. Bruni, G. L. Orlandi and from my collaborator from ENEA, R. Bartolini, thanks to the TMR European Collaboration Towards a storage ring Free Electron Laser at 200 nm (No. ERB 4061 PL 97-0102).

References

1. Dangoisse, D., et al. Phys. Rev. A. 1987;36(10):4775. Arecchi, T., et al. Phys. Rev. Lett. 1982; 49:1217.

2. Dangoisse, D., et al. Europhys. Lett. 1998; 6:335.

3. Meucci, R., et al. Phys. Rev. E. 1994; 49(4):R2528.

4. Weissand, C.O., King, H. Opt. Commun. 1982; 44(1):59.

5. Bauer, T.J. Opt. Soc. Am. B. 1990; 3:1175.

6. James, G., Harell, E., Roy, R. Phys. Rev. A. 1990; 41:2778.

7. Otsuka, K., Chern, J.Y., Lih, J.S. Opt. Lett. 1997; 22:292.

8. Saldin, E.L., et al. Opt. Commun. 1998; 148(4–6):383.

9. Park, E.H., et al. Nucl. Instr. and Meth. A. 1995;358(1–3):448. Kawasaki, S., et al. Nucl. Instr. and Meth. A. 1995; 358(1–3):114.

10. Chaix, P., et al. Phys. Rev. E. 1999; 59(1):1136.

11. Jaroszynski, D.A., et al. Nucl. Instr. and Meth. A. 1998; 407(1–3):407.

12. Hajima, R., et al. Nucl. Instr. and Meth. A. 2001; 0.475(1–3):270.

13. Chaix, P., Guimbal, P. Nucl. Instr. and Meth. A. 1995;358(1–3):96. Chaix, P., Iracane, D., Benoist, C. Phys. Rev. E. 1993; 48(5):3259.

14. Leemans, W.P., et al. Nucl. Instr. and Meth. A. 1995; 358(1–3):208.

15. Hahn, S.J., et al. Nucl. Instr. and Meth. A. 1994;341(1–3):200. Sang-June-Hahn, Jae-Koo-Lee. Phys. Rev. E. Sept. 1993;48(3):2162–2171. Elgin, J.N., Sharma, A. Opt. Commun. 1993; 103(1–2):100.

16. Billardon, M. Phys. Rev. Lett. 1990; 65(6):713.

17. Yamada, K., et al. Nucl. Instr. and Meth. A. 2002; 483:162.

18. Walker, R.P., et al. Nucl. Instr. and Meth. A. 2001; 475(1–3):20.

19. Wang, W.J., Wang, G.R., Chen, S.G. Phys. Rev. E. 1995; 51(1):653.

20. De Ninno, G., Fanelli, D., Bruni, C., Couprie, M.E., Euro, J. Phys. D. 2003; 22(2):269–277.

21. Couprie, M.E., et al. Nucl. Instr. and Meth. A. 1993; 331:37.

22. Hama, H., et al. Nucl. Instr. and Meth. A. 1996; 375:39.

23. Billardon, M., Garzella, D., Couprie, M.E. Phys. Rev. Lett. 1992; 69(16):2368.

24. Litvinenko, V.N., et al. Nucl. Instr. and Meth. A. 1995; 358:334.

25. Elleaume, P. J. Phys. 1984; 45:997.

26. Hara, T., Couprie, M.E., Billardon, M. Nucl. Instr. and Meth. A. 1996; 375:39.

27. Couprie, M.E., et al. IEEE J. Quant. Electron. 1994; 30(3):781.

28. Couprie, M.E., et al. Nucl. Instr. and Meth. A. 1995; 358:374.

29. Eckman, J.P., Ruelle, D. Rev. Mod. Phys. 1985; 57:617.

30. Roux, R., et al. Nucl. Instr. and Meth. A. 1997; 393:33.

31. Dattoli, G., et al. Nucl. Instr. and Meth. A. 2001; 471:403.

32. Bielawski, S., et al. Phys. Rev. A. 1993; 47(4):3276.

Two-color experiments with infrared lasers

J.M. Ortega,     Université de Paris Sud, LURE, 209d, BP 34, 91898 Orsay cedex, France. E-mail address: [e-mail]

Various pump–probe application experiments that can be done with infrared FELs and with table-top lasers are described. The spectral ranges that can be reached with these lasers and their spectral width and pulse length are briefly discussed. Many interesting experiments can be performed by using different wavelengths either with a two-color FEL or with various lasers synchronized with an FEL. A few examples are given.© 2003 Elsevier Science B.V. All rights reserved.PACS: 42.62; 41.40.C; 82.80.G

Keywords

FEL

Free-electron lasers

Infrared

Spectroscopy

1 Introduction

Molecular species generally exhibit a typical infrared absorption that constitutes their chemical signature. They lie in the domain spanning typically 2–50 μm (5000–200 cm−1). This is also the case of various other systems such as pseudo-atoms made by artificial nanostructures (5–100 μm) or elementary excitations in solids like magnons and Cooper pairs in superconductors. The absorption bands of these may extend as well in the millimeter wave region. Spectroscopy from near- to far-infrared is, therefore, an essential tool in many fields of science.

Linear spectroscopy can be performed using thermal sources or, in some cases synchrotron radiation. But the study of the dynamics and the non-linear spectroscopy of these systems can be performed only using tunable high-power short-pulse infrared lasers. These lasers can be either tunable classical lasers or infrared free-electron lasers or, preferably, a combination of two, or more, of these. Here, we are interested in mode-locked lasers which have the largest field of applications. The main types of tunable infrared short-pulse lasers are the following:

• Optical parametric oscillators (OPOs) pumped by mode-locked solid-state lasers, like Nd:YAG lasers. They cover usually from UV to about 10 μm. The pulse lengths are typically 10 ps, peak power 1 MW and average power a few milliWatts. The bandwidth can be made reasonably small (about 1 cm−1). A comparison between a typical FEL [1] and a typical OPO [2] is given in Fig. 1.

Fig. 1 Energy and Average Power of the CLIO FEL and OPO.

• Optical parametric amplifiers (OPAs) pumped by Ti:S lasers [3]. They cover upto approximately 18 μm with low average power (∼1 mW), high peak power but large bandwidth (∼100 cm−1). This is due to the fact that the pump laser pulse is necessarily very short (<1 ps) in order to get enough gain in the amplifier.

• RF linac-based infrared FELs, that cover the whole infrared up to the millimeter waves. They have a rather high average power, of the order of 1W, or more in case of recirculating accelerators, and very high peak power, of the order of 100 MW, or more. Their relative bandwidth ranges between 0.1% and 1%. For example, at CLIO Δλ/λ = 0.2% at 5 μm and 0.8% at 75 μm, corresponding to pulse lengths of 5 and 70 ps. It is also possible to obtain larger bandwidths, and thus a better temporal resolution (down to 300 fs) by tuning the optical cavity length [5]. This unique flexibility makes the FEL an excellent tool for time-resolved spectroscopy as it offers an interesting range of operating parameters in terms of spectral width and pulse length. It also provides an average power, typically, 10² times higher than other lasers in the range 5–10 μm and more than 10³ times higher at longer wavelengths.

The OPO/OPA types of lasers are commercially available, manufactured by many laser companies. Their performances may vary from the average values indicated above. They can be synchronized with an FEL with an accuracy better than 0.5 ps [3]. In addition, there exist Q-switched (dyes or OPOs) nanosecond lasers that can reach the mid-infrared and various low-power cw lasers with some limited tunability (diode and cascade lasers). Also, electrostatic accelerator-based FELs [4] work in this intermediate regime of pulse duration.

Dynamic studies are made by pump–probe and photon echo techniques which require absorption of two or more photons by the same system. In pump–probe experiments, it is often more interesting to be able to tune independently the frequency of the two beams which can be done with a two-color FEL or a combination of different lasers. For each system under study, one has to look for the best combination of pulse length and linewidth.

Non-linear spectroscopy is particularly suited to study the properties of matter. Also, it allows to perform linear spectroscopy of very dilute systems; in this cases the non-linearity is used to drastically enhance the sensitivity of the detection. For example, multiphoton ionization of molecules allows to detect the products with the extreme sensitivity of the Fourier transform-ion cyclotron resonant-mass spectrometer (FT-ICR-MS) and sum frequency generation (SFG) at surfaces allows to detect less than a monolayer of molecules. The SFG is more powerful when doubly resonant, with both laser beams tunable, and kinetics can be performed, with three colors.

2 The Two-Color FEL

The two-color operation of an FEL based on a two-section undulator has been demonstrated a few years ago with CLIO [6]. This is obtained simply by having two undulators in the optical cavity and tuning them to two different wavelengths. A high initial optical gain is required, since each undulator is lasing separately and the second undulator receives a beam that has experienced FEL interaction. Therefore, the beam has to be somewhat defocused in the first undulator, so that it may still produce gain in the second. This defocusing is also used to adjust the relative intensity of the two colors to the need of the user. Two-color operation is achieved from 3 to 18 μm, and with a two-color wavelength separation up to λ1 – λ2 = 5 μm. At longer wavelengths diffraction losses, reducing the net gain on each color, have prevented so far two-color lasing. This will be solved, to some extent, by using larger vacuum chambers in the optical cavity. From Fig. 2, it is clear that both colors can coexist at macropulse scale. Careful cross-correlation experiments have shown that the two colors can also overlap at least partly at the micropulse scale. As detailed, thereafter a pump–probe application experiment with an approximately 1 ps resolution has been performed with the two colors, and this is the best demonstration of the stability and reliability of this scheme.

Fig. 2 Time Evolution of the Wavelength Spectrum, during the Electron Beam Macropulse Duration of 10 μs, in the Two-Color Setup. The Macropulse Length is only about 6 μs, due to the Lower Gain in this Configuration. The Relative Intensities of the Two Colors can be Adjusted by Changing the Focusing into the Two Undulators.

At IFEL [7], a slightly different scheme is used: the two undulators are located in two different optical cavities where the same electron beam is sent successively. As a result, the second undulator is tuned to a longer wavelength where the gain will be less sensitive to the energy spread induced in the first one. Contrary to the CLIO scheme, such a scheme allows to lase only at two wavelengths very different from one another (lasing simultaneously at 5 and 18 μm has been demonstrated).

3 One-Color Experiments

Following the pioneering work at Stanford [8], it appeared to be interesting to study, both by pump–probe and by photon echoes [9] the vibrational relaxation of molecules in various environments (solid, liquid, sol–gel) or of nanostructures such as quantum dots [20]. In order to investigate molecular interactions in condensed phase, one commonly resorts to the analysis of vibrational transitions in the electronic ground state. Information is specifically derived from the transition linewidth. In addition to the contribution from the excited-state lifetime, it includes an homogeneous component which arises from fast fluctuations in the oscillator-environment coupling, and an inhomogeneous contribution that comes from site disorder or slow fluctuations in the coupling between the oscillator and its surroundings. The homogeneous width is expressed by

where T2 is the dephasing time, T1 the excited-state lifetime and T*2 the pure dephasing time due to environment effects.

With an adapted experimental setup, one can measure both T1 times—in one-color pump–probe experiments—and T2 times—in photon echo experiments, so that one obtains a view of the whole vibrational dynamics, giving information on the environment-induced perturbations. These techniques can as well be used to study elementary excitations in solids, such as Cooper pairs in superconductors, or artificial molecules, such as quantum dots [10]. Fig. 3 shows an example of such simultaneous measurements. It shows that T1 and T2 can be very different from each other and that the FEL allows to measure decay times as short as a few picoseconds. In that case, the absorption band spectral width was 14 cm−1 corresponding to a Δλ/λ of 0.7%, quite fitted with the FEL bandwidth.

Fig. 3 Measurement of T1 and T2 in W(CO)6 in Liquid n-Hexane at 300 K.

Another example of one-color experiments is the multiphoton dissociation: In this case, a non-linear phenomenon is used to perform linear spectroscopy with a great sensitivity. The IR spectrum of gaseous ions cannot be measured by classical means since the low ion density in the gas cells corresponds to an extremely small absorption. IR spectra are rather derived by monitoring the photo-fragments of the ion obtained through infrared multiphoton dissociation (IR-MPD). Dissociation occurs through a sequence of photon absorptions into a vibrational mode followed by redistribution of this IR energy into the vibrational bath. This particular spectroscopy requires the high flux, relatively small bandwidth and easy tunability of the FEL. The photo-fragments are analyzed through mass spectrometry. With this respect, the FT-ICR-MS approach has been demonstrated recently [11] to be particularly well suited. It provides an excellent mass resolution, together with the high selection capabilities over a large mass range and non-destructive ion detection, allowing for MSn type experiments. That is, one can prepare a given ion through a sequence combining successive ion-molecule reactions and mass selection of a reactive intermediate.

This is applied to gas phase studies of molecular ions and clusters, providing a comprehensive understanding of physical and chemical properties of great interest in chemistry, astrophysics and biochemistry. Future applications will also include studies of astrophysical and biochemical interest.

4 Two-Color Experiments

he MPI technique mentioned above is not a purely one-color technique since it is believed that the moderate bandwidth of the FEL helps the process of energy accumulation by a given molecule: the anharmonicity of successive absorptions results in small spectral shifts that remain inside the FEL spectral width. In some cases, it is interesting to use the FEL short pulses to produce wavelength chirping [12]. Furthermore, optical access to the reaction cell also provides the possibility in the future of combining UV–visible photo-dissociation steps to prepare ions of interest. Also, an interesting variant of this technique has been made at FELIX [13], where ionization is obtained after single photon absorption by irradiating with a UV laser, bringing the excited state above ionization limit. However, in these cases the lifetime of the excited molecular levels are rather long, of the order of a few microseconds or more, and we will now discuss two-color experiments where the time structure of the FEL is essential.

The first set of experiments deals with time-resolved pump–probe experiment with the two-color FEL. For example, stimulated emission under optical pumping has been observed in a three-level quantum well sample of GaAs/Al-GaAs, at 77K temperature at CLIO [14]. Fig. 4 displays the layout of the experiment. With a proper separation and filtering, the initial two-color beam gives one monochromatic beam at λ1 = 10 μm acting as the pump and another at λ2 acting as the probe. The wavelength λ2 of the probe scans the range 12–15 μm to estimate the gain in the sample. As shown in Fig. 4, stimulated emission occurs with an optical gain between 3 and 4. During a second set of experiments, with a high-quality sample, an optical gain of 50–500 has been measured with λ2 = 14 μm. The measurement of the lifetime for level three has also been carried out. This experiment demonstrates the stability of both colors during several hours, the flexibility of wavelength scanning and also the good geometric alignment of both colors. In the future, experiments are planned in HTc superconductors: It is interesting to use light in the mid-infrared (λ<20 μm) to dissociate Cooper pairs and look at the photo-induced reflectivity changes in far-infrared. This will give information on whether these pairs exist in the normal state (T > Tc), which is one of the challenges in understanding these materials. It is also possible to perform pump–probe experiments with an external laser synchronized with the FEL, which is possible with an accuracy better than 0.5 ps, although few have been made until now [3].

Fig. 4 Optical Pumping of Quantum Wells with a Two-Color FEL.

SFG on surfaces is a technique used to perform the spectroscopy of adsorbed species on surfaces. It also uses a non-linear process to perform linear spectroscopy. A frequency-doubled YAG laser is mixed with a tunable high-power infrared laser: if the substrate is centrosymetric only the surface contributes to the process, which becomes resonant if the infrared wavelength corresponds to an absorption from the surface. Therefore the non-linearity of the process, requiring high peak power, is efficient in discriminating the surface from the bulk. The FEL is particularly suited for this application since it is widely tunable and of high power. Moreover, YAG pumped OPOs between 1 and 8 μm, are also proposed in the FEL center CLIO, since their power is sufficient in this range, to run the SFG setup independently.

The SFG allows one to characterize the molecular structure and the coupling of the molecule to the substrate, surface or interface. In practice, this technique is the more demanding user experiment in terms of FEL intensity and wavelength stability, in order to obtain good signal-to-noise ratio spectra. It is therefore particularly important to monitor in real time the beam intensity, wavelength and linewidth. Many results have been obtained since 1992 with the SFG setup, in the fields of surface physics or electrochemistry [15–17]. It is also possible to use the difference frequency signal arising from the non-linear mixing at the surface. In some cases, difference frequency generation (DFG) can be more efficient than SFG [16]. The SFG/DFG signal depends on the orientation of the molecule and theoretical calculations allow to assign precisely a molecular orientation to each spectrum. Therefore, the orientation of the molecule during the various steps arising during the electro-chemistry can be completely characterized [17].

Kinetics have also been performed on the vibrational dynamics of CO at the (100) platinum electrochemical interface [18], with a resolution of approximately 0.5 ps. In that case, the pump and probe were at the same wavelength, the CLIO beam being split into a pump and a delayed probe beam.

In the future, it is planned to study biological molecules by SFG [19]. By using Langmuir–Blodgett film, one should be able to study the chemical bond between particular molecules and structures such as membranes or virus envelopes.

In the past, the CLIO center was using a pulsed YAG laser. In this case, the laser possesses the same time structure as the FEL, except that the micropulses were about 100 ps long, not matched to the FEL. Presently, an YLF laser oscillator cw mode-locked provides pulses as short as 8 ps. This will improve the signal over noise ratio in the SFG experiments by two orders of magnitude, since the SFG signal is proportional to the product of the visible by the infrared intensity and the detection is made in the visible region. Also, visible OPOs synchronized with the FEL will allow to perform doubly resonant SFG, adding the electronic to the vibrational spectroscopy at surfaces and interfaces. Indeed, the SFG intensity reads

ISFG = [χ(2)]²IvisIIR

where χ² is the second-order surface susceptibility:

(vIR, vvis)

(vIR, vvis).

resonant in vis non-resonnant and contributes only weakly to the signal. Then, only the IR frequency is swept, in order to characterize the adsorbate. However, if the visible frquency, vvis, is also adjustable, one can as well study the electronic states characterizing the surface of the substrate and of the adsorbate. It will also be possible to use either OPO light synchronized with the FEL, or the two-colors FEL, simultaneously with the visible light, providing three-color experiments. This will allow measurements of the dynamics of these systems, the study of the coupling of the vibrational modes of the adsorbates and the interactions between different adsorbed molecules. The use of several colors will also be of great utility to study the reactivity of molecules by selecting various excited states either in matrices (pump–probe) or in gas phase (detection by IR-MPD).

5 Conclusion

IR FELS are an invaluable tool for linear/nonlinear spectroscopy and kinetics. This is due to the combination of the FEL’s unique high average and peak power, broad tunability and spectral properties. Multicolor experiments add many possibilities to the use of FELs. Therefore, IR FEL facilities tend to become multilaser facilities. In addition, they are offering not only various laser beams but also sophisticated experimental setups to external users, serviced either by local researchers or in collaboration with various teams. This is analogous to the beam lines in synchrotron radiation facilities, the various setups becoming often too cumbersome and too sophisticated to be brought by external teams.

References

1. Prazeres, R., et al. Nucl. Instr. and Meth A. 2000; 445:204.

2. Peremans, A., et al. Ann. Phys. 1995; 20:527.

3. Knippels, G., et al. Opt. Lett. 1998; 23(22):1754.

4. Ramian, G. Nucl. Instr. and Meth A. 1992; 318:225.

5. Colson, W.B., Renieri, A. J. Phys. 1983; C44:C1–11. Hooper, B., et al. Nucl. Instr. and Meth . 1988; 272:96. Glotin, F., et al. Phys. Rev. Lett. 1993; 71:2587. Knippels, G., et al. Phys. Rev. Lett. 1995; 75:1755. Piovella, N., et al. Phys. Rev. E. 1995; 52:5470.

6. Jaroszynski, D., Prazeres, R., Glotin, F., Ortega, J.M. Phys. Rev. Lett. 1994; 72:2387.

7. Zako, A., et al. Nucl. Instr. and Meth. A. 1999; 429:136.

8. Tokmakoff, A., Fayer, M.D. J. Chem. Phys. 1995; 103:2810.

9. Crépin, C., et al. Laser Chem. 1999; 19:65.

10. Sauvage, S., et al. Phys. Rev. B. 2001; 63:113312.

11. Lemaire, J., et al. Phys. Rev. Lett. 2002; 89:273002.

12. Maas, D.J., et al. Chem. Phys. Lett. 1998; 290:75.

13. Putter, M., et al. Chem. Phys. Lett. 1996; 258:118.

14. Gauthier-Lafaye, O., et al. J. Appl. Phys. 1998; 83:2920.

15. Tadjeddine, A., et al. J. Electroanal. Chem. 1999; 473:25.

16. Le Rille, A., et al. Chem. Phys. Lett. 1997; 271:95.

17. Pluchery, O., Tadjeddine, A. J. Electroanal. Chem. 2001; 500:379.

18. Tadjeddine, A., Peremans, A., Guyot-Sionnest, P. Chem. Phys. Lett. 1995; 247(3):243.

19. B. Busson, private communication.

20. Sauvage, S., et al. Phys. Rev. B 2002; 66:153312. Sauvage, S., et al. Phys. Rev. Lett. 2002; 88:177401–177402.

First SASE and seeded FEL lasing of the NSLS DUV FEL at 266 and 400 nm

L. DiMauro, A. Doyuran, W. Graves, R. Heese, E.D. Johnson, S. Krinsky, H. Loos, J.B. Murphy, G. Rakowsky, J. Rose, T. Shaftan*, B. Sheehy, J. Skaritka, X.J. Wang and L.H. Yu,     BNL-NSLS, Upton, NY 11973, USA. E-mail address: [e-mail]

*Corresponding author. (T. Shaftan)

The Deep Ultra-Violet Free Electron Laser (DUVFEL) at the National Synchrotron Light Source consists of a 5 MeV photoinjector, a 200 MeV S-band linear accelerator, a four-magnet chicane compressor and a 10 m wiggler with a 3.9 cm period. The commissioning of the SDL accelerator was completed recently and it is routinely producing a high-quality electron beam with a peak current of ∼400 A and a normalised emittance of 3–4 mm mrad. The first SASE lasing of the DUV FEL has been demonstrated at both 266 and 400 nm. The gain length of the SASE experiments was measured to be 66 cm in both cases and up to 20 nJ per pulse was obtained. A laser seeded FEL at 266 nm is in the very early stages of commissioning and amplification of the laser seed has been observed. The goal of the seeded laser FEL is to saturate the FEL and thereby generate sufficient third harmonic at 89 nm for experimental applications. Here we report the observation and measurements of SASE at 400 and 266 nm and the first results of seeded FEL.© 2003 Elsevier Science B.V. All rights reserved.PACS:41.60.Cr

Keywords

Free Electron Laser

UV radiation

SASE

1 Introduction

The High Gain Harmonic Generation (HGHG) technique of high gain Free Electron Laser (FEL) has been intensively studied in recent experiments at the Accelerator Test Facility at BNL [1]. For the ATF experiments, the 10.6-μm seed wavelength was generated by a CO2 laser and measurements of the FEL radiation at the radiator fundamental wavelength (5.3 μm), as well as at the second harmonic (2.65 μm) and the third harmonic (1.77 μm) were made. The radiation was fully characterised, and the pulse length, coherence, harmonic content, etc. were found to be consistent with theoretical predictions [2]. As a result, HGHG was proved to be effective in generating high peak power as well as fully longitudinally coherent and stable radiation with a FEL.

The next milestone in the development of short wavelength FEL is the Deep Ultra-Violet FEL (DUVFEL) project at NSLS (BNL). The scientific program of DUVFEL includes several major steps. The first step is to obtain and optimise SASE at 400 nm. This is important for the initial commissioning of the FEL, allowing us to optimise accelerator performance with respect to the output intensity. The saturation of SASE at this stage is not anticipated with the current electron beam parameters. The second step is a direct optical seeding of the electron beam in the radiator, using a 266 nm beam derived from the same laser that drives the photoinjector. At this stage the FEL output is expected to be saturated and coherent. This will make the first user experiments possible at the third harmonic (89 nm) of the resonant wavelength [3]. The third step requires installation of the modulator and dispersive section in the beamline. Using HGHG we will be able to obtain radiation at 200 nm with harmonics. In order to get to shorter wavelengths (100 nm) the energy of the SDL accelerator will need to be upgraded.

2 The deep ultra-violet FEL

The layout of the facility is shown in Fig. 1. The accelerator begins with a BNL/SLAC/UCLA 1.6 cell photocathode RF gun, illuminated by a frequency-tripled Ti: Sa laser at 266 nm. Currently, the RF gun is able to produce a 300 pC, 4.5 MeV, 1.7 ps (RMS) electron beam with emittances of 3–4 mm mrad. Following are two SLAC-type 2.856 GHz linac sections, which accelerate the electron beam up to 77 MeV. The second linac tank provides the energy chirp for the bunch compression by running the electron bunch off-crest of the RF wave. A four-magnet chicane with variable field strength converts this energy chirp into spatial bunching, increasing the peak current up to 300 A, which is appropriate for FEL operation. The third linac tank, installed after the chicane, performs the residual energy chirp cancellation, and also provides additional acceleration. The last tank is used for acceleration to the nominal energy and, in combination with the spectrometer dipole, for bunch length measurement, using the zero-phasing method [4]. Focusing triplet and a profile monitor are used for transverse emittance measurements via the quadrupole scan technique. The commissioning of the accelerator [5] was performed during the last year and current parameters are appropriate for successful FEL operation.

Fig. 1 The DUVFEL layout. 1—gun and seed laser system, 2—RF Gun, 3—linac tanks, 4—focusing triplets, 5—magnetic chicane, 6—spectrometers dipoles, 7—seed laser mirror, 8—mini-undulator, 9—dispersive section, 10—NISUS wiggler, 11—beam dumps, 12—FEL radiation measurements area.

The main component of the FEL magnetic system is a 10-m long undulator (NISUS) with 3.89 cm period and 0.31 T peak field. The undulator consists of 16 linked Sections (32 poles in each). Six poles in the centre of each section are alternately canted, providing the redistribution of the natural vertical focusing into the horizontal plane. Every section is also equipped with separate horizontal and vertical dipole and quadrupole correctors, using a 4-wire system. A transverse profile monitor which images the fluorescence induced by the e-beam as it traverses a Cerium-doped YAG-crystal into a CCD camera is located at the end of each section. Currently 13 monitors are available along the undulator for trajectory correction and beam size matching. Commissioning of the NISUS undulator required measuring and optimising the dipole and quadrupole components of the magnetic field. This was done using the measurements of the electron beam trajectory and envelope inside the undulator.

For the trajectory optimisation two trims and two monitors are installed in front of NISUS. This system allows launching of the beam into the undulator with reproducible transverse coordinates and removing of residual betatron oscillations, if necessary. For the HGHG experiments, an in-vacuum seed laser mirror and associated optics are located before the spectrometer dipole. A combination of four trims performs a local bump of the trajectory to bend the beam around the seed mirror.

For the FEL output radiation measurements, 5 optical UV detectors are available. They are spaced at approximately equal distances along the undulator, so one can measure the gain length of SASE in the UV range. The end optical station contains two spectrometers and a streak camera with picosecond time resolution for the time synchronization of the electron and seed laser pulses. For this purpose we use the sub-picosecond SASE pulse, which is emitted while the electron beam is traversing the undulator, and the 5 ps (FWHM) seed laser pulse, so light beams are obtained at the same location in the undulator. The UV spectrometer serves not only for the measurements of the radiation spectra, but also as a tool for the electron beam energy tuning for proper overlap with the central wavelength of the seed laser radiation.

An electron beam spectrometer with a quadrupole doublet upstream, installed after NISUS, gives information about energy spectra of the electron beam exiting the undulator. The current set-up of the diagnostics allows us to obtain a nearly complete set of data about the FEL radiation pulse and electron beam before and after the undulator.

3 Experimental Results

One of the criteria for the trajectory improvement was the spectral width of the spontaneous radiation spectrum. Currently, the central cone spectral width is about 1.7 nm at 400 nm, which is close to the estimated value of 1.4 nm. For lasing at 400 nm the electron beam energy is 144 MeV. The obtained SASE signal was on the order of 5000 times brighter than the spontaneous emission. Measurements of the SASE pulse energy along the undulator show exponential growth of the intensity, with a gain length of 90 cm (Fig. 2), which is in a good agreement with analytical estimates using the measured electron beam parameters (150 A peak current, 4 mm mrad normalised emittance, 0.3% energy spread). The value of the peak current corresponds to the case of mild compression (bunch length was measured to be 0.7 ps [RMS] for the electron bunch with a relatively smooth longitudinal profile). Transverse electron beam parameters have been obtained measuring the beam sizes along the undulator. Transverse emittances and Twiss parameters were calculated from the analysis of the beam envelope in the undulator.

Fig. 2 Measured dependence of the SASE energy at 400 nm vs distance along the undulator.

After increasing the amount of compression, the measured SASE energy was significantly improved. The gain length decreased to 68 cm. This value is close to the one for saturation of SASE (50 cm). The explanation of this dramatic improvement is, probably, a very strong modulation of the longitudinal electron bunch density, caused during the compression process. The longitudinal profile of the compressed electron bunch shows a spiky structure with the peak current in the maxima of the modulation approaching 500 A (Fig. 3). The origin of this modulation is currently under study [6]. The corresponding SASE spectrum consists of several (2–3) spikes; the spike width gives an estimate of the bunch length of the radiation pulse (0.4 ps FWHM). This value, obtained from a single-shot spectrum, is close to the temporal width of one spiky part of the electron beam longitudinal profile, measured by the zero-phasing method.

Fig. 3 The modulation in longitudinal bunch profile.

After the installation of the UV detectors, the energy of the electron beam was increased up to 172 MeV and 266 nm SASE radiation has been obtained. The measured gain length is 66 cm, the SASE output pulse energy is 5 nJ (average over 200 shots) and the maximum value for the single shot can achieve 20 nJ.

The experiments with the seeded FEL at 266 nm are in progress. First, seeded FEL measurements have been performed and the output FEL signal is measured to be 10 μJ, which is about 1000 times larger than the SASE signal. The spectra of SASE, seed laser and seeded FEL at 266 nm are shown in Fig. 4. The shot-to-shot fluctuations of the spectral shape are greatly reduced for seeded FEL light relative to SASE alone.

Fig. 4 Spectra of SASE (a) and seed laser and seeded FEL (b) at 266 nm.

4 Conclusion

The results of the experiment show a good agreement with theoretical expectations. The detailed description of the measurements can be found in [7]. Our developed diagnostics setup allows us to obtain a complete set of data for the electron beam and the FEL radiation. The study of the observed longitudinal modulation is in progress for better understanding of the results.

Acknowledgements

Authors wish to thank Peter Paul at BNL, Chuck Roberson at ONR and Howard Schlossberg at AFOSR for their support of this research. Funding was provided by DOE Contract No. DE-AC02-98CH10886 and AFOSR/ONR MFEL Program #NMIPR01520375.

References

1. Yu, L.H.-, et al. Phys. Rev. A. 1991; 44:5178.

2. T. Shaftan, et al., Proceedings of PAC 2001, pp. 246–248.

3. Suits, A., et al. Science. 2001; 294:5551. [2527].

4. Ricci, K.N., et al. Nucl. Instr. and Meth. A. 2000; 445:333.

5. W.S. Graves, et al., Proceedings of PAC 2001, p. 2860.

6. H. Loos, et al., Proceedings of EPAC 2002, pp. 837–839.

7. A. Doyuran, et al., Nucl. Instr. and Meth. A (2003) these proceedings.

Section 2

FEL Theory

Some issues and subtleties in numerical simulation of X-ray FELs

W.M. Fawley,     Center for Beam Physics, Lawrence Berkeley National Laboratory, University of California, MS 47-112, 1 Cyclotron Road, MS 47-112, Berkeley, CA 94720-8211, USA. E-mail address: [e-mail]

Part of the overall design effort for X-ray FELs such as the LCLS and TESLA projects has involved extensive use of particle simulation codes to predict their output performance and underlying sensitivity to various input parameters (e.g. electron beam emittance). This paper discusses some of the numerical issues that must be addressed by simulation codes in this regime. We first give a brief overview of the standard approximations and simulation methods adopted by time-dependent (i.e. polychromatic) codes such as GINGER (LBNL Report No. LBNL-49625, 2002), GENESIS (Nucl. Instr. and Meth. A 429 (1999) 243), and FAST3D (Nucl. Instr. and Meth. A 429 (1999) 233), including the effects of temporal discretization and the resultant limited spectral bandpass, and then discuss the accuracies and inaccuracies of these codes in predicting incoherent spontaneous emission (i.e. the extremely low gain regime).© 2003 Elsevier Science B.V. All rights