Control of the amplification process in baseline XFEL undulator with mechanical SASE switchers

The recent achievement of LCLS LCLS1 , LCLS2 relies on a high-performance beam formation system, which works as in the ideal operation scenario described in the conceptual design report LCLS1 . In particular, the small electron-beam emittance achieved allows saturation within $20$ undulator modules, out of the $33$ available. A similar scenario is also foreseen for the European XFEL. One has, then, the possibility of taking advantage of a long, unused part of the SASE undulators to upgrade the facility.

In this paper we describe a method for providing simultaneous multi-color capability at three different wavelengths. For the sake of exemplification we will consider radiation around $0.2$ nm, $0.15$ nm and $0.1$ nm, produced by an electron beam with $0.4\mu$m normalized emittance and $0.25$ nC charge. After integration with a photon-beam distribution system, this method can lead to an effcient multi-user facility.

In its essence, the method is based on controlling the amplification process in the baseline XFEL undulator with the help of SASE switchers constituted by mechanical shutters to be installed at the position of a weak magnetic chicanes at specific locations down the SASE undulator. A sketch of the setup is presented in Fig. 1. Three different modes of operations are foreseen, based on a long ($60$ fs) pulse, depending on what shutters are off and on (see Fig. 2). In the first part of the undulator the beam undergoes the SASE process in the linear regime. After that, it passes through the first switcher, which consists of two devices. First, a weak magnetic chicane creating a transverse offset for the electrons and washing out the electron beam modulation. Second, a mechanical shutter, which has two positions: ”on” when absorbing SASE radiation and ”off”, when SASE radiation from the first undulator passes unperturbed to the second undulator. The electron bunch passes through the magnetic chicane, and the beam modulation is washed out. Note that in the chicane, the electron bunch is also delayed with respect to the radiation but the extra-path length can be chosen small enough to provide a shift of $6$ fs, much shorter than the long electron bunch ($30$ fs rms). If the shutter is off (see Fig. 2 top and Fig. 3), at the position of the second undulator almost all the ”fresh” (washed out) bunch is seeded by the radiation coming from the first part, and the seeded main part of electron bunch reaches saturation with ten GW power level at $0.2$ nm wavelength. In this case, energy losses and energy spread within the electron bunch are important, and the bunch cannot further be used for generation of high intensity SASE radiation at wavelength comparable to $0.1$ nm. If the shutter is on (see Fig. 2 middle and bottom, and Fig. 4) instead, a fresh electron beam will enter the second part of the undulator, and will start radiate according to the usual SASE process. The second part of the undulator is short enough so that the SASE amplification process ends, once more, in the linear regime. In this case energy losses and energy spread within the electron bunch are negligible and the electron bunch is still a good ”active medium” for the next color pulse generation in the following undulator parts. These parts are tuned to a different wavelength, namely $0.15$ nm and $0.1$ nm. When the second shutter is off (see Fig. 2 middle), the seeded SASE process reaches saturation at $0.15$ nm. When the second shutter is on, instead, one has saturation in the final part of the undulator only at $0.1$ nm (see Fig. 2 bottom). The distribution of photons can be achieved on the basis of pulse trains, thereby allowing many users working in parallel at different wavelengths. For the temporal structure of the radiation produced at the European XFEL tdr-2006 , this means that we have no restrictive requirement on the switching time of the mechanical shutter, which should be simply shorter than $100$ msec, a very suitable condition for mechanical system.

In the next section we present a feasibility study of the method, and we make exemplifications with the parameters of the SASE2 line of the European XFEL (see Table 1). With the help of this scheme it will be possible to provide in parallel X-rays at three different wavelengths around 0.2 nm, 0.15 nm and 0.1 nm. In the following section we will see how, combining this method with a photon-beam distribution system, one can provide an efficient way to generate a multi-user facility.

In the following we describe the outcomes of computer simulations using the code Genesis $1.3$ GENE .

As said before, the distribution of photons is done on the basis of pulse-trains. The two mechanical shutters need to operate at a frequency of $1$ Hz for a single on-off-on cycle with switching time of less than $100$ ms (the temporal delay between two consecutive trains). Consider a temporal interval of $1$ second, i.e. $10$ trains of electron bunches. During the first $400$ ms, the first three trains of electron bunches are let through with both shutters off (see Fig. 2 upper part). Therefore, three trains of radiation at $0.2$ nm are produced. Then, the first shutter is switched on in less than $100$ ms, i.e. in the interval between the next two trains. During the next $300$ ms, three trains of electron bunches produce radiation at $0.15$ nm. Finally, the second shutter is switched on during the interval between the two following trains, and other four trains of radiation are produces during the final $300$ ms, this time at $0.1$ nm. Once separate color pulses are obtained in this way, they can be distributed to different users. Combining this method with a photon-beam distribution system based on movable multilayer X-ray mirrors, as discussed before in OUR1 , can provide an efficient way to generate a multi-user facility. This option, exemplified in Fig. 14, is not specific for the European XFEL and may be applied for LCLS and other XFELs.

An option for the distribution of photons specific for the European XFEL may also be considered. The original layout of the European XFEL includes two long undulators for spontaneous emission behind SASE2. These two undulators use the spent electron beam of SASE2. We speculate on the possibility of distributing the SASE2 undulator modules in three parts (14 cells, 12 cells and 16 cells), tuned at three different wavelengths as shown before, and of installing the second and the third part inside the U1 and U2 tunnels instead of the spontaneous emission undulators. The idea is sketched in Fig. 15. Combining this re-installation with mechanical SASE switchers for control of the FEL amplification process can provide an efficient way to generate a multi-user facility. The two mechanical shutters and the magnetic chicanes would be installed as shown in Fig. 15. Different colors may then be transported to different experimental halls. In principle then, at each experimental hall one may still take advantage of the multi-user scheme in Fig. 14.

Finally, similarly as in OUR0 we remark that the main difficulty concerning the distribution of photons consists in the separation of the three colors. Once this task is performed, mirrors in the photon beam transport system can be used to distribute the three-color photon beam among three independent user beam lines. As in OUR0 we propose to separate the three colors already in the undulator with the help of x-ray mirrors. The idea is sketched in Fig. 16, Fig. 17 and Fig. 18. The three colors can be separated horizontally in two stages by installing the two-mirror setup sketched in Fig. 17 after the undulator parts producing a given color, as specified in Fig. 16. Note that the installation of these setups also requires the presence of a weak magnetic chicane in order to create an offset in the electron trajectory to accomodate the mirrors. The horizontal offset should be chosen small enough to account for the presence of the spontaneous radiation absorbers in the vacuum chamber, which limits the effective aperture to a circle of $9$ mm diameter. The horizontal offset may be therefore chosen to be around $3$ mm, which is enough for separating the two-color pulses: in fact, at the position of the optical station the FWHM beam size is less than a millimeter. Additionally, mirrors can also be used to generate a few $\mu$rad deflection-angle, which is not important within the undulator but will create further a small extra-separation of a few millimeters at the position of the experimental station, as shown in Fig. 18.

We presented a novel method to control the SASE amplification process in the baseline XFEL undulator with the help of mechanical SASE switchers. After the lasing of LCLS LCLS2 , a new scenario where the beam formation system works as in the ideal case has become reality. This allows for a reduction of the gain length in the SASE process and for exploitation of the extra-available undulator modules. In particular, we show how it is possible to accommodate three FELs, lasing at three different wavelengths within the foreseen undulator length for the SASE2 beam line at the European XFEL. The scheme makes use of mechanical switchers capable of switching on and off the SASE process at a given particular wavelength. Three possible configurations of two switchers allows for separate production of each of the three wavelengths. The switchers should work on the basis of a train of pulses, with a frequency of $1$ Hz for an on-off-on cycle, and with a switching time of less than $100$ ms. In this way, simultaneous operation at three different wavelength is possible. Distribution of the photons to different stations is discussed.

We are grateful to Massimo Altarelli, Reinhard Brinkmann, Serguei Molodtsov and Edgar Weckert for their support and their interest during the compilation of this work.