Fig. 2

Power dependence of the photoreaction probed by transient absorption (TA) spectroscopy. a Ultraviolet/Visible TA spectra at 15 ps delay time in purple membranes. The 520–570 nm range (grey) is dominated by very strong pump light scatter. With increasing pump intensities, the photoproduct absorption band shifts to longer wavelengths (650–680 nm), and a long-lived weak absorption band appears peaking at ~460 nm. b Semi-log plot of kinetic traces at 463 nm, highlighting two decay components. For higher pump intensities, the dominant sub-picosecond excited-state absorption of the retinal protonated Schiff base increases slightly in duration, and a second smaller ~20 ps component appears (dashed line). c Isomerization dynamics tracked in the spectral range of the retinal C=C stretching vibration for different delay times. Excitation at 535 nm with 20 GW cm⁻² (or 0.5 µJ). The retinal bleaching signal at 1527 and the excited state around 1480 cm⁻¹ are formed instantaneously. The rise of the J and K product signal is reflected by the signal increase around 1505 and 1513 cm⁻¹, respectively. d Maximal signal of the isomerization product (red dots) tracked at 1513 cm⁻¹ as a function of excitation intensity (delay time 10–20 ps, see Supplementary Fig. 1) at 535 nm in the picosecond range; transients are shown in Supplementary Fig. 1a–e. At 180 GW cm⁻², other signals overlap with the product band at 1513 cm⁻¹, decreasing its intensity (blue dot), thus preventing its use to quantify photoproduct yield. The maximal negative signal of the picosecond tryptophan bleaching band (black squares) is also presented (delay time 2–5 ps, see Supplementary Fig. 1). All four measurements were performed under the same experimental conditions without changing the laser set-up, keeping the non-systematic errors small. The samples were taken from the same batch with nearly identical concentration (~1 optical density at 535 nm and 25 µm thickness). Deviation from linearity occurs at about 30 GW cm⁻². The errors of the absorption signals (y axis) were determined by the noise on the transients around the time point used to read out the signal strength. The intensity errors (x axis) are given by error propagation of the focal radius (strongest contribution), pulse length and pulse energy. Source data are provided for Fig. 2a–d as a Source Data file