A cross-correlator-based timing tool for FemtoMAX

Abstract

We report on the commissioning of an ultrafast timing diagnostic for measuring a time-offset signal between two different synchronized ultrashort light pulses. The method is based on sum-frequency generation in a nonlinear crystal. The setup is similar to an auto/cross-correlator setup. In this case, one of the beams is a relatively weak diagnostic beam of visible light from a bending magnet at FemtoMAX (10⁶ photons pulse^-1 in a 0.2 mm high and 5 mm wide beam) while the other is a relatively intense laser beam (200 µJ pulse^-1) derived from the same laser that is used to pump the sample in pump/probe experiments. This enables online monitoring of the relative timing of a linear accelerator-based, short-pulse, hard X-ray source and a synchronized visible laser. We show that for a <50 fs full width at half-maximum (FWHM) light pulse from the accelerator and a 50 fs (FWHM) long laser pulse, we can determine the relative timing of the two pulses with an accuracy below 30 fs in a time interval of 4 ps. The advantages and limitations of the method are discussed.

Keywords: femtosecond laser; synchronization; time-resolved pump/probe experiments; timing jitter monitor; timing tool.

Figure 1

Overview of the timing diagnostics. The lower part with bold lines shows the cross-correlator, including a path to create laser-generated white light in order to set up the sum-frequency generation process. The interference filter is generally only used for setup. The FROG is for monitoring the temporal amplitude and phase of the laser pulse. The photodiodes (PD) are used for the RF-based timing monitor. A UV pulse is generated for an X-ray streak camera that is triggered by a photo-conductive switch. The abbreviations in the figure are as follows. RR: retroreflector; WLG: white-light generation; WL: white light; BBO: beta barium borate non-linear crystal; 2ω indicates frequency doubling and 3ω indicates mixing of the fundamental and the doubled radiation; DCX: double convex lens; CL: cylindrical lens; 2D: two-dimensional detector; BS: beam splitter; M: mirror; FROG: frequency-resolved optical gating; WP: wave-plate; GD: group delay compensation plate; PD: photodiode; PS: photo-conductive switch; IF355: 355 nm interference filter used to select the cross-correlator signal.

Figure 2

Schematic of the RF-cavity based timing tool at FemtoMAX which was used as a reference in this study. The electron bunch interacts with a button antenna, which also acts as a beam position monitor. The laser is detected by a fast photodiode (Fast PD). The two signals are combined and excite an RF cavity resonance. The two signals are amplified and measured on the same oscilloscope channel using a 80 Gs s⁻¹ oscilloscope.

Figure 3

Conceptual set-up for extraction of visible light from the dump magnet of the LINAC.

Figure 4

The SFG signal generated for two different time delays between the weak pulse from the bending magnet (orange) and the intense femtosecond light-pulse from the laser (red) is marked as a blue arrow. In the lower image the bending magnet pulse arrives sooner by a distance of cΔt resulting in a change of position of the generated signal (blue arrow). After calibration, the position of the SFG signal is a direct measure of the relative time between the laser pulse and the light pulse from the bending magnet.

Figure 5

(a) Raw data from the 2D CCD detector where the strong feature is the cross-correlator signal. The background has been subtracted. A height of about 10 pixels correspond to 0.2 mm and the width of the beam is a measure of the temporal convolution of the pulse-shapes of the laser and the bending magnet radiation. The horizontal position of the signal is a measure of the relative timing of the electron beam and the laser. The calibration factor is approximately 16 fs per pixel. (b) Horizontal lineout of the cross-correlator signal in (a) calibrated in femtoseconds.

Figure 6

Data comparing the timing information from the ping timing monitor and the cross-correlator. These data are used for the temporal calibration. The noise is due to the 200 fs uncertainty of the ping. Ideally the fit would be linear, but a small second-order term has been added in order to obtain a flat residual over the full 4 ps. The residual, which is defined as the fit subtracted from the data, is plotted in gray.

Figure 7

In (a) we see a rapid increase of the diffuse scattering from laser-excited Ge₂Sb₂Te₅. The data are sorted by an RF-based timing tool. In (b) the speed of the dynamics is revealed to be significantly faster when the data have been sorted by the ultrafast cross-correlator timing tool. In both cases an error function, shown in black, has been fitted to the rapid rise. Using the RF-based ping monitor the 10% to 90% rise time of the error function is 330 fs while the corresponding time for the cross-correlator data is 70 fs.