Calculated performance of the monochromator beamline. (a) Beamline optical transmission (efficiency), which is dominated by the grating reflectivity. The lower scale gives the harmonic order in terms of the fundamental IR beam (800 nm), the corresponding photon energy is indicated on the upper scale. (b) Spectral width as a function of harmonic order for a range of slit widths and the appropriate grating. The plotted values include both the monochromator bandpass and the inherent width of the respective harmonic. (c) Temporal broadening due to the grating shown for the same parameters as (b).
Layout of the IR-pump – XUV-probe tr-ARPES experiment. Approximately 90% of the Ti:S laser output is used for HHG, while the remaining 10% is used for pumping the sample. Both beam paths have their own computer-controlled compressors for pulse duration control and a λ/2 −plate + polarizer combination for power control. HHG occurs in the gas cell, typically in Ar. XUV sensitive photodiodes are located just after the Al filter and after the second toroidal mirror. The IR pump beam is introduced into the beamline after the second toroidal mirror. A BBO crystal behind the analyzer chamber is used to find the temporal overlap of the pump pulse and an IR pulse allowed to pass through the beamline.
A high-order harmonic spectrum generated in argon at 110 mbar by a 1.5 mJ, 40 fs pulse with a center wavelength of 785 nm. The spectrum was measured by recording the current from an XUV sensitive photodiode just before the sample position while scanning the monochromator grating. The slit size of the entrance and exit slits was adjusted for a monochromator resolution of 210 meV at harmonic 23. The FWHM bandwidth of harmonic 23 is 500 meV.
Photoelectron spectra of the Cu(111) surface state at 100 K measured with the He Iα line at 21.2 eV photon energy (top) and with high-order harmonic radiation at 35.6 eV (bottom) with identical analyzer settings. The broad peak at ≈ 1.0 eV binding energy stems from the Cu d-band excited with the He-I satellite. The He Iα data are limited by the analyzer resolution. The solid red line is a convolution of these data with a Gaussian (with the satellite removed), which gives an energy bandwidth of 90 meV (FWHM) for the high-order harmonic radiation.
(a) IR pump – XUV probe arrangement for tr-ARPES. The IR and XUV beams impinge on the sample with a small angle between them in the horizontal plane, which forms an angle of 60° to the surface normal. Photoelectrons are detected along the surface normal. (b) Cross-correlation measurement between 50 fs IR and XUV at 32.6 eV. The monochromator was set to 150 meV photon-energy resolution. The vertical axis shows the population of electrons lifted transiently above the Fermi level of the W(110) crystal by side-band formation. It gives a cross-correlation of the IR and XUV pulses, which is fitted to a Gaussian of FWHM 125 fs (solid red line). Deconvolution from the IR pulse yields an XUV pulse duration of ≈115 fs FWHM.
Photoemission spectrum from a 10 nm thick Gd(0001) film on a W(110) substrate. The inset shows an ARPES image recorded for 5 min at one pump-probe delay using the 23rd harmonic at 35.6 eV. A photoelectron spectrum (blue points) is extracted by integrating over the highlighted stripe around k ∥ = 0. This is then repeated for all pump-probe delays. The energy bandwidth of the XUV was 150 meV. The electron analyzer, set to wide angle mode (±13° acceptance angle) at a pass energy of 60 eV and an entrance-slit width of 0.5 mm, has a calculated energy resolution of 150 meV FWHM. The upper limit of the total energy resolution is given by the 260 meV FWHM of the Gd surface state. The individual components used to fit the data (solid lines) are explained in the text.
Temporal evolution of the valence band minority and majority component in Gadolinium upon excitation with an s-polarized IR pulse stretched to 300 fs duration, delivering an absorbed IR fluence of 1.2 mJ cm−2. The binding energies were extracted from the tr-ARPES data and fitted as shown in Fig. 6 . The solid curves are fits to sigmoid functions serving to highlight the different dynamics of the two spin components, namely, a significant delay between the onset of the majority response compared to that of the minority. The recording time of the complete data set was 5 h.
Temporal evolution of the spin minority component of the valence band of Gd(0001) upon excitation with an s-polarized 50 fs IR pulse. The band position was fitted as shown in Fig. 6 . Deconvolution of the decay time of this curve from the IR pulse duration yields an XUV pulse duration of ⩽100 fs, confirming the sideband measurement.
Space-charge shift (circles) and broadening (triangles) of the spectra as measured from changes to the Gd(0001) surface state and the Gd 4f core level, (a) shows the probe pulse space charge of the harmonic radiation, (b) the pump-probe delay dependent contribution of the IR pump pulse, here at 4 ps pump-probe delay where it reaches a maximum.
Optical parameters of the monochromating XUV beamline. All optical elements are toroidal and gold coated to a depth of 40 nm for optimal reflectivity.
Article metrics loading...
Full text loading...