Thermal expansion of the Martian mineral starkeyite (MgSO4·4H2O) and the effect of synchrotron X-ray radiation damage on its lattice parameters

Starkeyite, MgSO4·4H2O, is one of the few hydrated sulfate minerals identified by X-ray diffraction on the martian surface, yet its structure and lattice parameters are known only at room temperature, whereas the material is exposed to much colder conditions on the surface of Mars. We report the crystal structure of perdeuterated starkeyite at 10 K and 300 K from high-resolution neutron powder diffraction, including freely refined deuterium positions, and its lattice parameters over 10–370 K, providing the first reference data for this mineral across the full Martian temperature range. Thermal expansion is strongly anisotropic and near-uniaxial, with α1 : α2 : α3 = −0.12(4) : 0.89(5) : 1.00 at 305 K and the negative direction lying in the ac plane; the volume expansion is identical within uncertainty to that of the polymorph cranswickite despite their different topologies. At ~245 K, a temperature crossed daily at the surface of Gale crater on Mars a shoulder emerges on the 212 reflection that cannot be accounted for by the P21/n model, possibly indicating a hitherto unrecognised low-temperature modification. A synchrotron X-ray experiment on protiated starkeyite yields lattice parameters that diverge progressively from the neutron data, show cooling–heating hysteresis and negative expansion of the a-axis; snapshot patterns on irradiated and fresh sections of the capillary demonstrate that this is due to radiation damage. We describe the radiation-induced strain by a second-rank X-ray expansion tensor rij which in starkeyite is strongly anisotropic, reaches a volumetric value of 11.8 TGy-1 at an absorbed dose of 0.39 GGy, and largely anneals out on reheating to room temperature. Such doses cannot be reached on Mars, but they are routinely exceeded in synchrotron studies of minerals, where radiationinduced strain is indistinguishable from the thermal or pressure response within a single dataset and can bias lattice parameters, V–T relations and equation-of-state fits, particularly for hydrates and other beam-sensitive phases and in diamond-anvil-cell experiments where the sample cannot be moved to a fresh spot. We recommend that dose be recorded as a routine experimental variable and that thermal expansion be cross-checked against neutron or low-flux reference data.