Arsenic in hydrothermal apatite : 2 oxidation state , mechanism of uptake , and comparison 3 between experiments and nature 4 5 6

23 Element substitution that occurs during fluid-rock interaction permits assessment of fluid 24 composition and interaction conditions in ancient geological systems, and provides a way to fix 25 contaminants from aqueous solutions. We conducted a series of hydrothermal mineral replacement 26 experiments to determine whether a relationship can be established between arsenic (As) 27 distribution in apatite and fluid chemistry. Calcite crystals were reacted with phosphate solutions 28 spiked with As(V), As(III), and mixed As(III)/As(V) species at 250 ̊C and water-saturated 29 pressure. Arsenic-bearing apatite rims formed in several hours, and within 48 hours the calcite 30 grains were fully replaced. X-ray Absorption Near-edge Spectroscopy (XANES) data show that 31 As retained the trivalent oxidation state in the fully-reacted apatite grown from solutions 32 containing only As(III). Extended X-ray Fine Spectroscopy (EXAFS) data reveal that these As(III) 33 ions are surrounded by about three oxygen atoms at an As-O bond length close to that of an 34 arsenate group (AsO4 ), indicating that they occupy tetrahedral phosphate sites. The three35 coordinated As(III)-O3 structure, with three oxygen atoms and one lone electron pair around 36 As(III), was confirmed by geometry optimization using ab initio molecular simulations. 37 The micro-XANES imaging data show that apatite formed from solutions spiked with 38 mixed As(III) and As(V) retained only As(V) after completion of the replacement reaction; in 39 contrast, partially reacted samples revealed a complex distribution of As(V)/As(III) ratios, with 40 As(V) concentrated in the center of the grain and As(III) towards the rim. Most natural apatites 41 from the Ernest Henry Iron Oxide Copper Gold deposit, Australia, show predominantly As(V), 42 but two grains retained some As(III) in their core. The As-anomalous amphibolite-facies gneiss 43 from Binntal, Switzerland, only revealed As(V), despite the fact that these apatites in both cases 44 formed under conditions where As(III) is expected to be the dominant As form in hydrothermal 45 fluids. 46 These results show that incorporation of As in apatite is a complicated process, and 47 sensitive to the local fluid composition during crystallization, and that some of the complexity in 48 As zoning in partially reacted apatite may be due to local fluctuations of As(V)/As(III) ratios in 49 the fluid and to kinetic effects during the mineral replacement reaction. Our study shows for the 50 first time that As(III) can be incorporated into the apatite structure, although not as efficiently as 51 As(V). Uptake of As(III) is probably highly dependent on the reaction mechanism. As(III)O3 352 moieties replace phosphate groups, but cause a high strain on the lattice; as a result, As(III) is 53 easily exchanged (or oxidized) for As(V) during hydrothermal recrystallization, and the fully 54 Arsenic in apatite Page 3 reacted grains only record the preferred oxidation state (i.e., As(V)) from mixed-oxidation state 55 solutions. Overall this study shows that the observed oxidation state of As in apatite may not reflect 56 the original As(III)/As(V) ratio of the parent fluid, due to the complex nature of As(III) uptake and 57 possible in-situ oxidation during recrystallization. 58 59