Extended Data Fig. 1: Schematic of the experimental setup.
From: Attosecond clocking of correlations between Bloch electrons
a, The 8.6-fs-NIR pulse train is split into two branches. The first one passes a quarter wave plate (λ/4) and is overlapped with the THz field with an indium tin oxide-coated window (ITO) before both pulses are focused onto a GaSe crystal for EOS or the WSe2 sample to generate HSB. The remaining excitation light and the emerging HSB radiation are collimated and separated from the THz radiation with a germanium wafer (Ge). With a dichroic mirror (DM), the HSB radiation is sent into a spectrometer (CCD), while the polarization state of the transmitted NIR pulse can be analysed by a quarter wave plate, a Wollaston prism (W), and a balanced pair of photodiodes (BPD). The remaining THz field is overlapped with the second NIR branch to constantly monitor the electro-optic signal (EOM). b, Measured SHG-FROG spectrogram and c, reconstructed temporal shape of the NIR light with a pulse duration of 8.6 fs. d, Comparison between excitation spectrum (orange) and absorbance (blue) of monolayer WSe2. The high-energy edge of the excitation spectrum is chosen to be resonant with the monolayer 1s transition. e, Waterfall plot of successive EOS measurements with the reference beam line. The electric field of the reference THz waveform corresponds to the colour scale. Fitting the temporal position of the recorded transients reveals a standard deviation of only σ = 84 as, attesting to the excellent stability of the setup during a measurement.
