Potential pitfalls when using popular chemical extractions to characterize Al‐ and Fe‐containing soil constituents

Wet‐chemical extraction of soil to quantify pedogenic species or to remove specific compounds prior to other analyses is an established approach in analytical soil mineralogy and soil chemistry. Interpretation and informational value of data derived from long‐established and frequently used extractions, for instance involving dithionite, oxalate/oxalic acid in the dark (AOD), and pyrophosphate (PYR), suffers from nonuniform practical regulation and missing knowledge about potential methodical limitations. In this review, we analyzed potential pitfalls of these frequently used extractions, with the focus on selectivity and completeness of the methods as derived from effects of time dependency and of phase separation. Major problems we identified comprised that time‐dependency of extraction differed between analytical targets, that a multitude of species is attacked, reducing the selectivity for the original analytical target, and that studies on extraction from model compounds, including analytical targets and nontargets, are not universally present. The latter aspect is crucial for the completeness of AOD and PYR extraction that has not been proven for all potential analytical targets of the methods yet. We practically tested citrate (CIT) extraction of aluminum (Al) and iron (Fe) in organic association, using selected models of soil constituents. Apart from a synthesized poorly ordered Si‐rich short‐range ordered aluminosilicate, CIT did not extract Al from nontarget phases, confirming previous studies, but did extract Al and Fe completely from organic associations. In addition to recommendations on the practical use of dithionite‐based, AOD, citrate‐ascorbate (CA), and CIT extraction, we suggest replacing highly problematic PYR extraction by CIT extraction for metals in organic association in soil and using AOD extraction in combination with CA and CIT extraction to avoid potential misinterpretation of ambiguous data.