Linac Coherent Light Source: The first five years

A new scientific frontier opened in 2009 with the start of operations of the world’s first x-ray free-electron laser (FEL), the Linac Coherent Light Source (LCLS), at SLAC National Accelerator Laboratory. LCLS provides femtosecond pulses of x rays (270 eV to 11.2 keV) with very high peak brightness to access new domains of ultrafast x-ray science. This article presents the fundamental FEL physics and outlines the LCLS source characteristics along with the experimental challenges, strategies, and instrumentation that accompany this novel type of x-ray source. The main part of the article reviews the scientific achievements since the inception of LCLS in the five primary areas it serves: atomic, molecular, and optical physics; condensed matter physics; matter in extreme conditions; chemistry and soft matter, and biology.

Figure 1

Schematic of a high-gain self-amplified spontaneous emission (SASE) free-electron laser, in which both the radiation power and the electron beam microbunching grow as a function of the undulator distance until saturation.

Figure 2

Linac Coherent Light Source accelerator layout from the electron gun to the beam dump, including a laser heater, two bunch compressors, three linac sections, and a 132-m-long undulator. Also included are the slotted foil in the middle of bunch compressor 2 for x-ray pulse length control and two transverse deflecting cavities (TCAV3 and XTCAV) for pulse length measurements.

Figure 3

Linac Coherent Light Source free-electron laser power (red points) at 1.5 Å vs active undulator length. The measured gain length is 3.3 m and a genesis simulation is overlaid in blue with electron beam parameters measured. The x-ray image on the yttrium aluminum garnet screen is shown in the inset. Adapted from [120].

Figure 4

Overview of hard x-ray self-seeding operation at the Linac Coherent Light Source (LCLS). (a) The LCLS hard x-ray self-seeding setup at 1.5 Å (seeding diamond is not drawn to scale). (b) The self-amplified spontaneous emission (SASE) and (c) seeded spectra recorded on single shots. The seeded bandwidth (0.5 eV FWHM) is about a factor of 40 narrower than the SASE bandwidth (20 eV FWHM).

Figure 5

$X$-band deflecting cavity (XTCAV) measurements of sub-5 fs, 1 keV x-ray pulse. The electron bunch charge is 20 pC with electron beam energy of 4.7 GeV. (a) One single-shot lasing-off image, (b) one single-shot lasing-on image, and (c) reconstructed x-ray profile with a calculated 2.6 fs FWHM duration. Adapted from [30].

Figure 6

The diffraction-before-destruction concept showing how nuclear motions caused by an x-ray FEL pulse can occur primarily after a suitably short pulse has passed through the sample. From [312].

Figure 7

Schematics of typical time scales and length scales of physical, chemical, and biological processes. Shaded regions indicate the time scales and length scales accessed by different light sources.

Figure 8

The three instruments in the Near Experimental Hall (NEH) were commissioned first in a phased approach starting in October of 2009 with user science experiments on the AMO instrument, followed by SXR in July 2010 and XPP in October 2010. In the Far Experimental Hall (FEH) experiments began on CXI in February 2011 followed by XCS in November 2011 and MEC in April 2012 to form the complete facility. The x-ray energy range in keV is specified for each instrument in brackets. Third harmonic radiation is used for photon energies greater than 12.7 keV.

Figure 9

Diagram of sequential x-ray multiphoton ionization in neon. The highest possible charge states are the ones energetically accessible by one-photon ionization. Below the Ne $1s$ threshold only valence electrons are ionized (V process); above the threshold, photoionization (P) and Auger (A) decay processes alternate. For increasing charge states the core-hole lifetimes increase, leading to subsequent photoionization (P–P) processes and x-ray induced x-ray transparency. From [453].

Figure 10

Sankey diagram illustrating the transition probabilities of argon atoms illuminated with intense x-ray pulses at 480 eV. Vertical bars present possible electronic configurations, and the width of each branch indicates the transition probability. The most prominent ionization pathway to reach the highest observed charge state ${\mathrm{Ar}}^{11+}$ from the Ar ground state is highlighted in colors other than green. The inset shows the electronic configurations involved in the exemplary pathway. From [201].

Figure 11

Measured Auger spectra and their comparison to calculated double core-hole Auger spectra. Inset: The data are taken with different intensities and for reference at a synchrotron radiation source. Main panel: The difference of the two FEL x-ray intensities (black line) shows spectral intensity similar to the calculated spectra (yellow). Adapted from [33].

Figure 12

Calculated fluorine $1s$ photoelectron angular distributions with 45 eV kinetic energy for a pFAB molecule in (a) the equilibrium geometry, (b) with doubled F-C bond length, and in (c) and (d) two proposed transition states. The molecules are one dimensionally aligned and the x-ray polarization is along the arrow. The distributions show a profound sensitivity to the local chemical environment. Adapted from [43].

Figure 13

Time-resolved fragmentation of ${\mathrm{CH}}_3\mathrm{I}$. (a) Map of the delay-dependent kinetic-energy distribution of ${\mathrm{I}}^{6+}$ ions following near-infrared dissociation and x-ray ionization of iodomethane. Channel 1 stems from x-ray induced dissociation of the bound molecules, channel 2 from Coulomb explosion of charged fragments, and channel 3 from x-ray ionization of the ${\mathrm{CH}}_3{\mathrm{I}}^{+}$ dissociative channel. (b) Delay-dependent yields of channels 1 and 3. Channel 3 appears only when charge-transfer processes between the iodine and methyl fragments are effectively suppressed in the dissociating molecule. Adapted from [122].

Figure 14

Time-coincidence map between x-ray pulses and detected fluorescence from Fe XVII in an electron beam ion trap. The x-ray induced fluorescence is measured in a small time window determined only by the time resolution of the Ge detector. Gating on the x-ray time window increases the ratio of signal-to-electron-beam induced background by a factor of 14 000  allowing one to obtain high-quality spectra from ultradilute systems. From [32].

Figure 15

Comparison of (a) sideband generation and (b) streaking in mixed optical and x-ray laser fields. Sidebands are symmetrically distributed around the main line and they reflect the exchange of photons from both fields during more than one optical period. In streaking experiments (b) the photoelectrons experience a redistribution of kinetic energy that depends on the magnitude of the electric field at the instance of their generation during a subcycle of the optical-laser field. Adapted from [110], and [195].

Figure 16

X-ray and optical wave mixing signal. The additional peak at 2.878° on the sloping background of the main peak stems from sum frequency generation of the x ray and optical pulse. The inset shows the spectral analysis of this peak, confirming that the energy of these photons equaled the sum of the incoming x-ray and optical photon frequencies. Adapted from [169].

Figure 17

X-ray pumped x-ray lasing. Selected single-shot spectra of the neon x-ray lasing line at 850 eV and simultaneously transmitted FEL pulse around 960 eV. The x-ray lasing line width is much narrower and jitter free. Note that the FEL pulse is attenuated by the neon gas by about 3 orders of magnitude compared to the incoming pulse energy. The inset displays the exponential gain with increasing pump power. Adapted from [344].

Figure 18

Measured single-shot emission spectra as a function of the (a) incoming photon energy and three distinct line-outs with (b) as-measured and (c) normalized intensities. The x-ray FEL pulses appear as a diagonal line in (a) which is heavily attenuated for photon energies above the Ne $K$ edge at 870 eV. The dominant emission line appears at 850 eV. Below 867.5 eV excitation energy the emission is absent and above 870 eV the line is stable. For the intermediate emission energy regime, the line profile and peak position vary strongly, owing to stimulated electronic x-ray Raman processes. Adapted from [438].

Figure 19

Single-shot ion time-of-flight spectra of clusters irradiated with intense x-ray pulses, taken in coincidence with their diffraction image. In the three representative shots the clusters were located in the (a) center or (c) outer wings of the focus or (b) in between. The ionization dynamics depend strongly on the exposure intensity. The focal volume averaged data (bottom) are dominated by the low-intensity signal in the outer wings of the focus. Adapted from [172].

Figure 20

Single-shot diffraction images (in logarithmic scale) from superfluid helium droplets doped with xenon atoms. The xenon atoms gather at the quantum vortices in the superfluid droplets, giving rise to Bragg peaks. The data suggest that the quantum vortices organize in a regular lattice with high densities. From [171].

Figure 21

Nonlinear lattice dynamics. Time-dependent Bragg reflections for four peaks in YBCO. The displacive distortion of the lattice shows a decrease in two peaks while the intensity increases in the other two. The data are for the x-ray diffraction pump-probe experiment, while the fitted curves are for the DFT with local-density approximation (LDA) theory. The green shows calculations of all 11 $A_g$ phonon modes, while the red shows the most important four: Those that couple most strongly to the optically driven, $B_{1u}$ mode. From [277].

Figure 22

Universal curve showing the speed of the metal-to-insulator transition in magnetite. X-ray data (blue and green symbols and brown dashed line) taken at the ($0,0,1/2$) reflection (Fe $L_3$ edge) shown together with the optical reflectivity (red line, representing an exponential fit to the data). The reduction in the electronic ordering coherence length (${\lambda }_{\mathrm{coh}}$, green squares) and the change in diffraction angle ($\mathrm{\Delta }\beta$, blue diamonds) occur concomitantly with the increase of the optical reflectivity ($\mathrm{\Delta }R/R$, red curve) with a time constant of $1.5\pm 0.2\mathrm{ps}$. This decay occurs with a delayed onset as clearly indicated by the unchanged ${\lambda }_{\mathrm{coh}}$ and $\mathrm{\Delta }\beta$ values at 200 fs time delay and reflects the structural modifications toward the cubic lattice where metallic charge fluctuations reduce the coherence of the remnant patches of electronic order. For comparison, the trimeron lattice diffraction intensity vs time (brown dashed line) shows a fast quench, reflecting the time resolution of the experiment. From [97].

Figure 23

The x-ray diffraction peak shown as an intensity as a function of time, describing the dynamics of the atomic structure of a ${\mathrm{Pr}}_{0.5}{\mathrm{Ca}}_{0.5}{\mathrm{MnO}}_3$ manganite single crystal. The bottom panel shows a simulation using a simple model based on Landau theory for a second-order phase transition. These are both shown as a function of fluence, the analog of the order parameter, used for the theoretical model. From [28].

Figure 24

Time-dependent behavior of an electromagnon. The magnetic diffraction of the ($0k0$) peak of ${\mathrm{TbMnO}}_3$ (blue symbols) is compared with the electric field of the THz pump (red solid lines) as a function of the time delay. (a) The response of the sample in the multiferroic phase at $T=13\mathrm{K}$ and for (b) an inverted electric-field driving pulse. (c) The response for the system in the spin-density wave phase for $T=30\mathrm{K}$, where the effect vanishes. From [239].

Figure 25

Reversal of Gd spins within nanoscale Gd-rich regions as measured by resonant x-ray diffraction. (a) Temporal evolution of the magnetic structure factor $S_q$ for Gd $4f$ spins (red) and Fe $3d$ spins (blue) measuring the local deviations from the average magnetization. (b) The same $S_q$ data but on a longer time scale. The $S_q$ signals were integrated over the low-$q$ region of the magnetic diffraction pattern ($q<0.2{\mathrm{nm}}^{-1}$) and high-$q$ ($q>0.2{\mathrm{nm}}^{-1}$) regions and are normalized to the equilibrium values. Therefore, $|S_q(t<0)|=1$ for both $q$ ranges and sublattices, although high-$q$ $S_q$ for Gd is $\sim 4$ times greater than for Fe. Dashed and solid lines represent fits of exponential decay and rise functions to the data. The observed delayed onset at $\mathrm{\Delta }t=360\pm 50\mathrm{fs}$ indicates that angular momentum transfer effectively occurs only after this time delay. Beyond 1 ps, the Gd (Fe) data are well described by an exponential decay (rise) with a time constant of $6\pm 2\mathrm{ps}$. The Fe data are fit with an additional decay with time constant 8 ps. From [175].

Figure 26

Time-resolved diffuse scattering measurements of Ge. (a)–(c) Typical frames depicting the oscillatory component of the diffuse scattering collected from the (022) Brillouin zone in Ge, after background correction, at 0.4, 1.2 and 2.8 ps. (d) A demonstration of the variation in the normalized diffuse scattering intensity as a function of time delay at select wave vectors from $u=(-0.1,0.00,-0.08)$ to $v=(-0.33,0.15,-0.27)$ in reciprocal lattice units. From [415].

Figure 27

Map of the high-energy density (HED) regime above $P=1\mathrm{Mbar}$. HED science includes study of warm dense matter (WDM), hot dense matter (HDM), interior of planets, astrophysics, relativistic laser plasmas, and inertial confinement fusion (ICF) research.

Figure 28

The logarithm of the calculated emission spectrum from the Al plasma (blue dashed curve) and calculated emission spectrum at 100 fs after the simulated x-ray FEL pumps the He-like $n=1$ to $n=3$ transition (black solid curve). The major predicted effect of the intense x-ray pump is to cause photoionization. Adapted from [81].

Figure 29

Spectrally resolved Al $\mathrm{K}\alpha$ emission as a function of the x-ray FEL excitation photon energy from 1460 to 1680 eV. Emission lines from Al charge states from $4^{+}$ to ${10}^{+}$ are marked in Roman numerals: red for states with a single $K$-shell hole and blue for states with a double $K$-shell hole. Adapted from [427].

Figure 30

(a) A section of a Debye-Scherrer ring showing evolution of lattice dynamics with time delay. (b) A diffraction profile illustrating the characteristic lattice response. Adapted from [292].

Figure 31

A schematic of a dynamic structural change upon chemical transition. A model biatomic molecule is optically excited from the ground state to an electronic state with a different energy minimum from the ground state with respect to a structural degree of freedom.

Figure 32

(a) Representation of x-ray spectroscopy methods. The absorption cross section near an absorption edge of an inner-shell electron is sensitive to unoccupied electronic states just below the ionization energy $E_0$ and to the local structure around the excited atom. The generated hole can be occupied by a valence electron (e.g., from a highest occupied molecular orbital state, HOMO), thereby emitting x rays that represent electronic configurations. (b) Combination of experimental ultrafast x-ray spectroscopy setups. The absorption of the monochromatic incoming x-ray pulses is measured through the total fluorescence collected by an x-ray diode. Emission spectra are measured using a dispersive spectrometer. The kinetic energy of photoelectrons can be analyzed with a photoelectron spectrometer.

Figure 33

Spectroscopic pump-probe results from the photoexcited spin transition of $[{\text{Fe}}^{\mathrm{II}}{(\mathrm{bpy})}_3{]}^{2+}$ in solution. The increased iron-to-ligand distance is tracked by the XANES trace at 7125 eV [(a), black symbols, vertically offset by 0.5]. The structural HS rise time has been fitted by an exponential rise of $162\pm 6\mathrm{fs}$ (red) ([252]). (b) Time-dependent $K\beta$ emission difference spectra. The transient traces at 7061 and 7054 eV have been extracted [(a) green diamond symbols and (c) black symbols, respectively]. The data have been overlaid with results from global fits, including (blue lines) and excluding (green dashed lines) population of an intermediate ${}^{3}T$ triplet ligand field state. Adapted from [460].

Figure 34

Time-resolved studies of the photolysis of $\mathrm{Fe}{(\mathrm{CO})}_5$. (a) Chemical structure of $\mathrm{Fe}{(\mathrm{CO})}_5$; (b) ground-state and difference RIXS map between the integrated measurements after 266 nm excitation of $\mathrm{Fe}{(\mathrm{CO})}_5$ solution ethanol and the ground-state map. The transient intensities in the regions labeled in 1–4 [black dot symbols in (c)] have been modeled using a kinetic model using different potential electronic transition states [solid blue lines in (c)]. The result of the global fit is shown in the bottom panel of (c). The compatible model includes an excited ($E$), triplet ($T$), and ethanol ligated ($L$) $\mathrm{Fe}{(\mathrm{CO})}_4$ fragment population. Simulated signals excluding a triplet and a ligated population are indicated by dashed lines in panels 2 and 3. Adapted from [439].

Figure 35

Oxygen $K$-edge x-ray emission spectroscopy (left) and x-ray absorption spectroscopy (right) data of CO/Ru(0001) with corresponding fits (solid lines) for both before the excitation and for 12 ps afterward. At the bottom, the peak deconvolution resulting from the fit of the spectra is shown, with the elastic peak being indicated in light blue around 534 eV. (Middle) A schematic illustration of the excitation process from the $\mathrm{O}1s$ level to the unoccupied $2{\tilde{\pi }}^{*}$ resonance in XAS and the core-hole decay process from occupied molecular orbitals back to the $\mathrm{O}1s$ in XES. From [98].

Figure 36

(a) Diffuse scattering profiles of metastable water, measured at different temperatures. The nine lowest temperature curves were measured using ultrafast pulses from LCLS before nucleation of ice. (b) The increasing separation of the two peaks labeled $S_1$ and $S_2$, which approaches order peaks of low-density amorphous ice (LDA, dash-dotted blue lines) as well as hexagonal ice (ice ${\mathrm{I}}_h$, red dashed lines). From [372].

Figure 37

Structure of rhodopsin in complex with arrestin. TM1TM7 indicates rhodopsin transmembrane helices 17; H8 is intracellular helix 8. Adapted from [218].

Figure 38

Interface between synaptotagmin and the neuronal SNARE complex. (a) The primary interface along with interacting residues. (b) Electrostatic potential map of the primary interface shows how two polar regions I and II are connected by a hydrophobic patch (SNAP-25 I44, L47, and V48 and Syt1 V292, L294, and A402). (c), (d) Close-up views of regions I and II. Labels indicate interacting residues. Dashed lines indicate hydrogen bonds or salt bridges. $2{\mathrm{mF}}_o$ ${\mathrm{DF}}_c$ electron density maps of the interacting residues are superimposed (gray mesh; $\text{contour level}=1.5\sigma$). Adapted from [462].

Figure 39

Difference electron density maps [F(light)–F(dark)] in pump-probe time-resolved serial femtosecond crystallography on carbonmonoxy-myoglobin (MbCO). (a) Results at time delay of 10 ps showing the bound ($3\sigma$ contour level, red) and photodissociated ($+3\sigma$, contour level green) CO, the doming of the heme, the out-of-plane movement of the iron, and the associated movement of Histidine 93 (His93) away from the heme. Also shown is the rotation of Histidine 64 (His64) and the movement of Phenylalanine 43 (Phe43). (b) Same as (a) rotated by 90°. (c) Same as (b) but for a time delay of 0.5 ps. Adapted from [23].

Figure 40

$K\beta$ x-ray emission in photosystem II (PSII). (a) X-ray emission spectra collected on PSII solution for the doubly illuminated state ($2F$ in black), triply illuminated state after $250\mu \mathrm{s}$ delay ($3F^{\prime }$ in red) and triply illuminated state after 0.5 s delay ($3F$ in blue). (b) Difference between the LCLS-collected $3F$ and $2F$ spectra. (c) Difference between the $3F$ (LCLS) and $2F$ spectra collected at a synchrotron source at 15 K. The similarities between the LCLS and synchrotron results indicate that a similar electronic state of the manganese cluster is being probed in both cases. Adapted from [227].

Figure 41

Time-resolved structural changes in CO:myoglobin. (a) Damped oscillatory behavior of the radius of gyration ($\mathrm{\Delta }{\mathrm{R}}_{\mathrm{G}}$) showing a 1 Å increase in 1 ps measured in carbonmonoxy-myoglobin (MbCO) after laser excitation. (b) Time evolution of the volume ($\mathrm{\Delta }{\mathrm{V}}_{\mathrm{p}}$) of the molecule after laser excitation showing a delayed response of the volume compared to the radius of gyration. This delayed response indicates an ultrafast redistribution of the mass toward the outside solvent from the active site, verifying the existence of the protein quake. Adapted from [255].

Figure 42

Bragg termination under x-ray FEL illumination. (a) Integrated powder patterns showing the falloff of the measured Bragg intensity as a function of pulse duration for photosystem I crystals at a photon energy of 2 keV. The integrated intensity is shown relative to the shortest pulse duration (solid lines), along the simulated ratios from damage modeling (dashed lines). (b) Comparison of the simulated damage-driven disorder (solid lines) with the best fit Debye-Waller factor. Reasonable agreement between the two indicates that damage can be corrected to some extent through the use of a standard resolution-dependent correction factor to the intensities. Adapted from [26].

Figure 43

Difference electron density between a synchrotron structure and an LCLS high-dose structure in ferredoxin. Clear differences in the electron density are visible from the green positive density and the red negative density contoured at the $3\sigma$ level. (a) Cluster 1 and (b) cluster 2 at 7.36 keV. $S1$, $S2$, $S3$, and $S4$ represent the sulfur atoms labeled 1 to 4 ($S3$ not shown). The cysteine (CYS) amino acids around the clusters are labeled with the residue number in the protein sequence. Adapted from [308].

Figure 44

Assembled three-dimensional mimivirus diffraction pattern from 198 2D diffraction patterns. (a) Subset of ten diffraction patterns shown in their best orientation. (b) All 198 diffraction patterns shown with a cutout showing the origin in reciprocal space. Adapted from [116].