Fig. 1. Experimental setup.

a Haemoglobin microcrystals were injected into the XFEL beam using a gas dynamic virtual nozzle injector¹¹. The first ~30 fs X-ray pulse (photon energy above the iron K-edge (7.112 keV) and a pulse energy of up to ~0.1 mJ, average value 0.03 mJ) was used as a pump, isochorically heating the jet and launching a shockwave (depicted in grey) that propagates upstream and downstream of the interaction region. The shock wave propagates at supersonic speeds and outruns the explosive gap formed in the jet. The scattered X-rays were absorbed by a thin iron filter and did not reach the detector. After 122.5 ns a sample segment upstream of the pump pulse was hit by a second ~30 fs X-ray pulse (photon energy just below the iron K-edge and a pulse energy of ~0.9 mJ), which was displaced by ~5 µm towards the nozzle. In this case, the scattered X-rays passed through the iron filter, reaching the detector. The setup differs from previous two-colour X-ray pump/X-ray probe experiments^33,34 due to the displacement of the pump pulse. b Femtosecond snapshot image of the jet a few nanoseconds after the probe pulse had interacted with the jet. Explosions induced by the pump (orange arrow) and probe (blue arrow) pulse are clearly visible. Although present in the jet, the shockwaves are not visible in the camera images. In jets this small, the optical path difference induced by shock compression does not suffice to observe the shock. A movie showing consecutive pump–probe pairs hitting the jet is available as Supplementary Movie 1. The jet instabilities (wiggling) are ascribed to the presence of crystals in the jet. These instabilities are largely absent in homogeneous (crystal-free) jets and become more pronounced with increasing crystal concentration.