LCLS CDR 2/16/16 9.2.3.3 FEL commissioning and physics The FEL experimental enclosure, and the diagnostic enclosure, will be where optics and diagnostics component R&D is performed prior to application to specific individual experiments. Below are listed some of the techniques that will be tested. 9.2.3.3.1 Pulse slice, compression and transform-limiting systems 9.2.3.3.1.1 Pulse slice Shorter than 200 fs pulses are of interest, e.g. to the biological experiments. A chirped electron pulse creates a chirped photon pulse, so that slicing the pulse in energy also slices the pulse in time. Three techniques have been proposed. First is to use an xray monochrometer with a bandpass << the width of the chirp (2%); the optimal choice is for a bandpass that just corresponds to the zero bandwidth pulse duration, 6 fs. A possible scheme uses a crystal (Si) monochrometer where the crystals are cut asymmetrically to increase the energy acceptance. By adjusting the asymmetry, the Si (111) reflection can be tuned to match that needed to slice the chirped x-ray pulse. Second, reflection off multi-layers will allow a slice out of a chirped pulse [1]. The efficiency, bandwidth and band shape can be controlled by the multilayer choice, and should allow < 10 fs pulses with > 50% efficiency. Third is an off – axis zone plate in combination with a slit [2]. This last is proposed for the first experiments on LCLS, because of the similarity with of the required lens with what has already been achieved. The experimental program outlined for the LCLS points toward scientific opportunities with photon pulses that are shorter than the baseline design. The most straightforward way to approach this problem is to introduce an photon energy-time correlation (chirp) to the x-ray pulse and then select a small part of the energy width yielding a small time width. Full 6D Simulations of chirped electron beams provide photon beams with typically 1-2% energy bandwidths for the 230 fs baseline pulse length. Thus an x-ray optic that can select a fraction of that bandwidth, say 0.1-0.2 % would provide a pulse that is commensurately shorter, perhaps approaching the 10 fs range. The needed bandwidths, of say 0.1% are perhaps 10 times wider than typical bandwidths for conventional perfect crystal optics. However the need for broad bandwidth crystal optics has been foreseen already for conventional synchrotron radiation applications [ref Matsushita]. This approach of a pair of asymmetric reflections in the traditional ‘channel cut’ arrangement using a Si monolith can provide the needed bandwidth enhancement. It also provides a simple and robust optic that will preserve the beam quality. -1- LCLS CDR 2/16/16 l > l l 0 = l l 0 < l 0 Z one plate axis FEL Axis Figure 1 - Image formation by zone plate struck off axis at different wavelengths. During a wavelength chirpped FEL pulse the focused spot will move along the zone-plate axis causing the beam to sweep across the plane of the central focus. The concept, shown schematically in figure 1, is that a circular zone plate focuses light to a point on its optical axis whose distance from the zone plate is a function of wavelength. If the illumination is restricted to a small area near the outer edge of the zone plate, it will still converge to the focal point on the zone plate axis, giving the appearance that the small beam has been deflected sideways as if by a grating (but unlike a grating, the light is also focused). In the case of a chirped FEL pulse, where the wavelength decreases monotonically over the duration of the pulse, the focal spot will move outward along the zone plate axis. An observer looking at a screen placed perpendicular to the zone plate axis at the distance of the mid-pulse focus will see a small spot of light moving across the screen during the pulse. Placing a small pinhole through the screen in the path of the spot produces a brief pulse of light downstream whose duration is roughly equal to the original pulse duration times the ratio of the width of the pinhole to the total path length of the moving spot. Note that the entire zone plate does not have to be manufactured, only the portion intercepted by the FEL beam. For large offsets, the required off-axis zone plate portion closely resembles a grating having zones of nearly equal thickness across its aperture. We propose fabrication of a system for 8.271 keV, in the FEL physics hall, using alternating layers of Be and B4C. At 8.271 keV, the optimal slice thickness for the Be and B4C system is 33 m which gives a 1 phase shift between x rays emerging from the Be and B4C layers. The attenuation through the 33 m thick optic is negligible, the transmission through the Be is 99.5% while through the B4C the transmission is 98.2%. The surface dose to the Be is 0.002 eV/atom and the surface dose to the B4C is 0.007 eV/atom at a position 15 meters downstream of the undulator exit. The design parameters are shown in Figure 2, based on a 2% energy chirp and a 50 fs slice. An -2- LCLS CDR 2/16/16 additional requirement is that the focused spot should be suitable for experiments, that require 200 photons /A2. Parameter Value Units Object Distance Image Distance Focal length Offset Core radius Inner zone width 53.99 2.67 2.587 277.5 177.5 1.08 meters meters meters microns microns microns Outer zone radius Outer zone width Number of fabricated zones 377.3 0.51 291 microns microns Figure 2: design parameters for the time-slice lens The zone plate pattern is fabricated by sputtering alternating layers of Be and B4C onto a wire core having a 177.5 m radius. The deposit is 200 m thick extending to an outer radius of 377 m and contains 291 active zones. This zone plate is about 2 x larger in diameter and active radius than previously fabricated sputtered-sliced circular zone plates but it also has fewer and larger zones so its fabrication is probably within the current state-of-the-art. A numerical simulation of the system is based on the Gaussian properties of the FEL beam and Kirchhoff’s diffraction theory. Figure 3 shows an overexposed gray level intensity distribution of this modified electric field overlaid with a drawing outlining the active area of the lens. The image was overexposed by altering the intensity to gray level map to show the active boundaries of the lens. -3- LCLS CDR 2/16/16 Outli ne o f cyli ndrical zone plate X, m Figure 3 An illustration of the xone plate and LCLS beam Figure 4 shows the focused spots in the same area around the sample position for 3 of the 52 calculated diffraction patterns. The top image is of the t = 0 fs step, the middle is of the central t=125 fs step, and the bottom image is of the t = 255 fs step. The 3 images clearly show the movements of the spot due to the wavelength change over the chirp and the change in spot size as the z position of the focus moves from downstream to upstream of the sample plane. The line shows the position of the samples fired from the injector Figure 4 Focused spot at sample plane at beginning, middle, and end of chirp. -4- LCLS CDR 2/16/16 9.2.3.3.1.2 Pulse compression Pulse compression can be achieved by dispersing successive intervals of a chirped photon pulse along trajectories of different lengths, and then recombining them within a common time-space volume. A proposal that will be researched is that of a strained crystal. The chirped beam diffracts from a crystal in which the strain varies with depth. Low-Z crystals are favored to reduce absorption over the required path lengths and small Bragg angles are favored to increase the diffraction efficiency of the strained crystal. The best compression ratio found for a 230 fs, 05% chirped beam at a wavelength of 1.54 A is achieved with a graphite crystal. References [1] R. Tatchyn, in "Workshop on Methods and Instrumentation for X-FEL," DESY, Hamburg, June 26-27, 2000 [2] R. Bionta, “A transmissive optics approach to time-slicing the LCLS x-ray pules”, LCLS-TN-00-7, and UCRL-ID-139011 (2000) 9.2.3.3.2 Wavelength selection 9.2.3.3.2.1 Monochrometers Crystals and crystal-based instrumentation will undoubtedly prove to be critical to the ultimate productivity of the LCLS [1,2]. Depending on the unit cell structure of the material and its lattice spacing, a number of candidates exist that can be used to efficiently diffract LCLS radiation in the <3-5 Å wavelength range, and for which a broad depth of experience has been acquired at storage ring SR and other facilities over the last twenty years, and for which photon absorption and consequent damage estimates suggest they can survive at the shorter wavelengths (< 3 Å). The most common materials used as single crystal monochromators for synchrotron radiation are silicon, germanium, diamond and beryllium (in order of decreasing perfection). Single crystals of silicon and germanium exist that have lattice planes straight to better than 10-9 rad with their lattice plane distances defined to the same accuracy: ∆d/d < 10-9. Also diamond single crystals of excellent perfection can be manufactured. Thus these crystals are better with respect to "slope errors" than the best man-made mirrors and the only problem is that they must be mounted with great care; the same, of course, is true for mirrors. In contrast, beryllium is a mosaic crystal with a reflection width of ~ 200 µrad and is chosen only for cases where more flux is needed at the expense of resolution. In Table 9.6.9.2.1 below, selected structural and thermal parameters, along with the absorption at 8 keV, are presented for all four materials. The quantity k/ma can be considered as a figure of merit. It is seen that diamond has by far the best performance both at room and cryogenic temperatures. -5- LCLS CDR 2/16/16 Table 9.6.9.2.1 Comparison of monochromator materials42: beryllium, diamond (C*), silicon and germanium Material Be C* Si Ge Atomic Number (Z) 4 6 14 32 Atomic Weight (A) 9 12 28 73 hcp diamond diamond diamond Lattice Constant a [Å] 2.286 3.567 5.431 5.658 Lattice Constant c [Å] 3.583 - - - 1188 1860 543 290 b)Absorption Coefficient m [cm-1] 1.8 7.5 1.41 402 a)Conductivity k [Wcm-1K-1] 1.93 23 1.5 0.64 a)Expansion Coefficient a [10-6K-1] 7.7 1.1 2.4 5.6 a)k/ma [MW] 0.14 2.78 4.4x10-3 2.8x10-4 c)k/ma [MW] 11 120 0.20 6.6x103 Crystal Structure a)Debye Temperature T D [K] a)297 K; b)at 8 keV; c)77 K A potential drawback of crystals compared to mirrors with respect to power loading is the much bigger reflection angle. However, this problem can be mitigated by asymmetrically cut crystals, where the lattice planes are not parallel to the crystal surface. In this way the beam footprint can be increased in the vertical or in the horizontal direction or both (the “rotated-inclined geometry”). The theoretical limit of asymmetry is determined by total reflection, so that in principle the glancing angle of mirrors can be approached. In that case a crystal diffracting hard X-rays is even able to specularly reflect away the softer part of the spectrum and thus to protect itself from the power of the soft radiation, which is much more heavily absorbed. Whereas meridional asymmetry (in the diffraction plane) affects angular and spectral beam parameters, sagittal asymmetry (inclined geometry) does not. For incident beam divergences smaller than the Darwin width, increase of the beam divergence after reflection must be taken into account. This is given by the sum of the Darwin widths at the entrance and at the exit of the crystal. This means that in addition to the increase of the positional beam width the beam will become progressively wider. At the same time a chirp of the beam occurs, i.e. the isophases and the isochrones of the reflected beam are no longer parallel and the isochrone is no longer normal to the beam propagation direction. The beam widening and the chirp could be used for time structure -6- LCLS CDR 2/16/16 experiments. On the other hand, if a second, antisymmetric crystal is inserted in the reflected beam, the increase of the beam divergence, the angle-energy dispersion and the chirp can be almost completely recovered. The extreme meridional asymmetry outlined above is able to transmit almost fully the coherent LCLS beam and will eliminate most of the unwanted photons and thus clean the beam, provided that the absorbed part of the incident radiation does not destroy the material. Detuning of the two crystals should be very efficient for eliminating harmonics. It is, however, desirable to verify the predictions based on simple geometrical diagrams in more detail with dynamical theory and to compare them experimentally, in addition to what has already been done. If we want to narrow the bandwidth delivered by the LCLS, we can no longer use meridional asymmetry to decrease the power density. In that case the sagittal asymmetry can provide a considerable spread of the beam footprint, by about the same amount as meridional asymmetry. Another means of decreasing the absorbed power would be to make the crystals very thin, approximately one extinction depth thick. Together with meriodanal asymmetry, the softer radiation would be totally reflected and the harder radiation transmitted. It has, however, turned out that it is very difficult to mount thin crystals without bending them. In the double reflection scheme outlined above the two crystals must be aligned to much better than the Darwin width at the exit of the first and the entrance of the second crystal (i.e., to within 2.2 µrad), so that even a tiny mounting strain will severely hamper the transmission. Thus there are reasons for withholding an overly pessimistic judgment. First, we have established that the best candidate materials for crystal optics are those of lowest Z, most probably Be and diamond, of which the latter has a widespread infrastructure that could support R&D critical to the LCLS. Second, as noted in the foregoing discussion, it is possible to utilize asymmetrically cut crystal geometries that could further decrease the absorbed power density, albeit at a probable cost in optical performance. Third, it is in principle possible (although expensive) to adapt to overly rapid damage accrual by employing dynamic optics; viz., by displacing the crystal to successively new exposure points following a definite series of pulses. Ultimately, the actual longevity and performance of candidate crystals will need to be experimentally determined. References [1] A. Freund, "Crystal Optics for the LCLS," presented at the SLAC/DESY International Workshop on the Interactions of Intense Sub-Picosecond X-Ray Pulses with Matter, SLAC, Stanford, CA, Jan. 23-24, 1997. [2] R. Tatchyn, "LCLS Optics: Technological Issues and Scientific Opportunities," in Proceedings of the Workshop on Scientific Applications of Short Wavelength Coherent Light Sources, SLAC Report 414; SLAC-PUB 6064, March 1993. -7- LCLS CDR 2/16/16 9.2.3.3.2.2 Grazing incidence mirrors Grazing incidence mirrors can be used for wavelength filtering, and beam steering. An initial configuration developed for the mirrors and mirror chamber (tank) of the LCLS is discussed under Optical Enclosures. Multiple-mirror reflectors are stacked in a vertical bank to allow the appropriate reflector to intercept the beam with a vertical motion. For illustrative purposes, a set of nominal mirror dimensions and mirror bank parameters, is listed below in Table 9.6.9.1.1. Evidently, facet length will be a sensitive parameter for higher-Z reflecting materials, and the initial design parameters of the mirror enclosure may need to be revised, or additional enclosures added, to accommodate multiple reflectors. Table 9.6.9.1.1 LCLS mirror bank parameters. Dw ~100 m. Energy Range m L [m] i[rad] T[rad] Beryllium 3-30 keV 1 0.25 0.0004 0.0004 A (x10-4) [eV/atom] 6-0.12 Gold < 3 keV 2 1 0.0001 0.0004 200 Mirror Material In view of the ubiquitous importance of specular reflection, studies, coupled with selected experimental investigations, are needed to shed light on a number of physical effects with the potential of inducing damage. These include radiation field / surface interactions vs. surface roughness and surface contaminants; radiation/surface interactions as a function of polarization; photoemissive surface stresses induced by radiation pulses; and bulk and surface ionization-related damage effects. 9.2.3.3.3 Beam splitting Beam splitters (see figure 9.2.3.3.3.1) can be made from thin crystals (Bragg or Laue ) or multilayers, a technique that has been developed in recent years at LLNL [1,2]. Blurring of the temporal structure due to a splitter's dispersive effects [3] could to a certain extent be mitigated by pinhole aperturing of the incoming light. Splitting methods based on homogeneous or perforated (6) foils operating in transmission/reflection will also be investigated, particularly if the practical performance of the grating splitter proves to be overly dispersion-limited. In this context, the development of broad-band, highquality multilayers as alternatives to the specular reflectors assumed here could be pursued as a means of scaling down the length of the instrument, particularly for FTS or source-analysis applications for which the bandwidth reduction would be acceptable. Incident beam Output:delayed beams -8- change delay LCLS CDR 2/16/16 Figure 9.2.3.3.3.1: An illustration of a beam splitter and delay line, based on Bragg diffraction. References [1] Bionta, R. M., Jankowski, A. F., Makowiecki, D. M., "Fabrication and Evaluation of Transmissive Multilayer Optics for 8 keV X Rays," Mat. Res. Soc. Symp. Proc. Vol. 103, 1993, pp. 257-263. [2] Bionta, R. M., "Transmission gratings that diffract 8 keV x rays," Appl. Phys. Lett. 51(10), 725-727 (1987). [3] Tatchyn, R., Materlik, G., Freund, A., Arthur, J., eds., Proceedings of the SLAC/DESY International Workshop on the Interactions of Intense Sub-Picosecond X-Ray Pulses with Matter, SLAC, Stanford, CA, Jan. 23-24, 1997, SLAC WP-12. 9.2.3.3.4 Pulse delay Single-crystal diffraction can be used to construct pulse delay lines [Joksch et al., Rev Sci Instrum 63 1114 (1993)], as needed for example in the nano-scale dynamics campaigns (see Figure 9.2.3.3.3.1). Part of the full beam is transmitted through a beam splitter (e.g. a thin diamond crystal), then doubly reflected by a pair of crystals at 900, and recombined with the direct beam using a second splitter. Adjustment to the time delay is made by moving the crystal pair in a direction perpendicular to the incident beam. Suitable reflections exist around 1.5 A. 9.2.3.3.5 Polarization control To be provided by John Arthur 9.2.3.3.6 Layout and other equipment The layout and specific hardware to be provided is found in the requirements document, hardware list -9-