Influence of carrier solution ionic strength and injected sample load on retention and recovery of natural nanoparticles using Flow Field-Flow Fractionation

Natural nanoparticles, including both natural organic matter (NOM) and inorganic mineral-like phases, have been broadly characterized using Flow Field-Flow Fractionation (FlowFFF). Calibration with polystyrene sulfonate (PSS) standards was generally carried out in order to determine the molecular weight distribution of the NOM, however if the analyzed sample has a different charge density compared to the PSS standards, the resulting molecular weight distribution may become meaningless. The presented study therefore investigates and compares the influences of ionic strength and sample load on the retention time and recovery of both PSS standards and natural nanoparticles from a variety of sources. The minimum ionic strength in the carrier solution and the maximum injected sample load required for satisfactory separation depend on the molecular weight of the PSS standards and on the nature of the NOM. The degree to which results depend on conditions and parameters within the FlowFFF varies significantly between the different natural nanoparticle samples. We found that it may be necessary to calibrate the channel under different conditions from the actual sample runs. Under well controlled and documented conditions this could represent an important move away from the paradigm of “same conditions for standards and sample”. From all conditions tested, the most reliable molecular weight calibrations were obtained at elevated ionic strengths in the carrier solution (>0.04 M) and low injected mass of PSS. However, even under these optimized conditions variations of up to 20% occur in the calculated molecular weights, and the recovery of NOM falls by up to 50% at high ionic strengths. Many applications aim for both correct molecular weight distribution and the measurement of low concentrations of elements bound to natural nanoparticles. We conclude, however, that finding conditions that are equally optimal for both of these analytical tasks is not always feasible.