Major- and trace element analysis

Major- and trace elements were determined using X-ray fluorescence analysis (XRF) and inductively coupled plasma mass spectrometry (ICP-MS). XRF analyses were carried out on powdered and fused samples using a Philips Panalytical PW2400 at the Institute of Applied Geosciences at the Technical University Berlin. The accuracy of XRF analyses was better than 5% RSD for most elements estimated using reference materials JR-1 and JR-2 (rhyolite). The uncertainty on trace element concentrations determined by ICP-MS (Actlabs, Canada) was estimated be < 10% based on analyses of reference materials DNC-1 (dolerite), W-2a (diabase), and BIR-1a (basalt). The uncertainty of REE + Y analyses is typically < 10% RSD, estimated from analyses of reference materials DNC-1, W-2a, BIR-1a and NCS DC70014 8 (ore).

We calculate Ce* for Ce-anomalies (Ce_N/Ce*_N) according to Lawrence et al.⁷⁰ (Equation ²). Using Nd and Pr to calculate Ce* precludes artificial anomalies that may arise from elevated La concentrations that result from the higher stability of La relative to other REE in solution. Ce-anomalies result from Ce(III) to Ce(IV) oxidation and enhanced scavenging onto reactive surfaces such as Fe-Mn-oxyhydroxides. As a result, Ce is becoming depleted in residual seawater and authigenic mineral phases precipitated thereof.

The subscript N denotes element concentrations normalized to post PAAS⁵³.

We calculate the Ge/Si ratio of illite based on the illite stoichiometry, measured bulk rock Al₂O₃ concentrations and based on constant Ge/Si ratios of the pure silica end-member of 0.5 (Supplementary Fig. ²).

We report normalized element concentrations as element (El) enrichment factors (EF), defined as:

To preclude artefacts from normalization, EFs were calculated only for samples with  > 3% detritus⁴³. Moreover, we note that coefficients of variation are typically smaller for Al than for the redox-sensitive elements reported (Cu, Ni, U, V, Cr, P and Ba) except Mo.