Improving solid-phase extraction of dissolved organic matter (DOM)

Abstract

In current dissolved organic matter (DOM) research, solid-phase extraction (SPE) with Bond Elut PPL resins is the gold standard for desalination and concentration. This method routinely recovers 50-65% of marine and terrestrial DOM. However, in unusual samples, for example from hydrothermal vents, early-stage microalgal cultures, the sea-surface microlayer, or sediment leachates, DOM recovery drops to 20% or less, presumably because the prevalent organic compounds are outside the polarity or molecular size window of the PPL sorbent. Here, we adjusted individual steps of the standard protocol, such as acid type, acid strength, and solvent polarity to increase DOM yields in notoriously low-recovery samples. We compared acidification of a coastal marine sample with either HNO3 or HCl at pH 1 and pH 2. From samples of the sea-surface microlayer, the underlying seawater, and macroalgal and intertidal sediment leachates, we performed two subsequent extractions at environmental pH and at pH 2. Following DOM extraction, we compared several eluents, including methanol, acidic methanol, tetrahydrofuran, and dichloromethane. We evaluated the different approaches by determining dissolved organic carbon (DOC) and dissolved black carbon (DBC) yields as well as changes in molecular DOM composition using ultra-high resolution mass spectrometry.Lowering the pH from 2 to 1 increased the DBC yield but resulted in slightly lower DOC recoveries, indicating a shift in molecular composition of the retained compounds rather than expansion of the SPE window. HNO3 may furthermore cause nitration, nitrosation and oxidation of DOM. However, combining sequential extraction at neutral and acidic pH with a series of eluents increased DOC recoveries substantially in comparison to the standard protocol. The inter-sample differences in the molecular composition (as detected via ultrahigh-resolution mass spectrometry on a molecular formula level) of the extracted DOM exceeded those caused by variations in the protocol and followed similar trends compared to the standard procedure. The co-variance of molecular degradation indicators with PPL performance emerged as a general trend: Recoveries with the standard protocol increased with the DOM degradation state. We conclude that PPL is much more versatile than its regular use indicates and encourage exploring a larger bandwidth of experimental setups for a better representation of natural DOM from a wide variety of sources.