Trace-element evidence for the origin of desert varnish by direct aqueous atmospheric deposition

Smooth rock surfaces in arid environments are often covered with a thin coating of Fe–Mn oxyhydroxides known as desert varnish. It is debated whether such varnish is formed (a) by slow diagenesis of dust particles deposited on rock surfaces, (b) by leaching from the underlying rock substrate, or (c) by direct deposition of dissolved constituents in the atmosphere. Varnishes collected from smooth rock surfaces in the Mojave Desert and Death Valley, California are shown here to have highly enriched and fractionated trace-element abundances relative to upper continental crust (UCC). They are highly enriched in Co, Ni, Pb and the rare-earth elements (REEs). In particular, they have anomalously high Ce/La and low Y/Ho ratios. These features can only be explained by preferential scavenging of Co, Ni, Pb and the REEs by Fe–Mn oxyhydroxides in an aqueous environment. High field strength elements (HFSEs: Zr, Hf, Ta, Nb, Th), however, show only small enrichments despite the fact that these elements should also be strongly scavenged by Fe–Mn oxyhydroxides. This suggests that their lack of enrichment is a feature inherited from a solution initially poor in HFSEs. The first two scenarios for varnish formation can be ruled out as follows. The high enrichment factors of Fe, Mn and many trace elements cannot be generated by mass loss associated with post-depositional diagenesis of dust particles because such a process predicts only a small increase in concentration. In addition, the highly fractionated abundance patterns of particle reactive element pairs (e.g., Ce/La and Y/Ho) rules out leaching of the rock substrate. This is because if leaching were to occur, varnishes would grow from the inside to the outside, and thus any particle-reactive trace element leached from the substrate would be quantitatively sequestered in the Fe–Mn oxyhydroxide layers, prohibiting any significant elemental fractionations. One remaining possibility is that the Fe, Mn and trace metals in varnish are derived from leaching of dust particles entrained in rain or fog droplets either in the atmosphere or during wet atmospheric deposition. The high trace metal enrichment factors require that most of the dust was physically removed before or during varnish formation. The remaining aqueous counterpart would be depleted in HFSEs and Th relative to the REEs, Co, Ni and Pb because the former are more insoluble and hence largely retained in the removed dust fraction. The high Ce/La ratios suggest that precipitation of trace metals may have been governed by equilibrium partitioning in an excess of wet atmospheric deposition. If varnishes are indeed derived from wet atmospheric deposition, they may provide a record of the aqueous component of atmospheric dust inputs to various environments. D 2004 Elsevier B.V. All rights reserved.