An intercomparison of cross-flow filtration techniques used for sampling marine colloids: Overview and organic carbon results

An intercomparison was conducted between 14 different cross-flow filtration (CFF) systems. Each CFF membrane had a 1000 nominal molecular weight cut-off, and five different manufacturers' membranes were tested. The goal of this exercise was to examine whether the different CFF systems were behaving in a well-defined and operationally reproducible manner in marine applications. Surface seawater from Woods Hole and mid-depth waters from Hawaii were prefiltered (< 0.2 mu m) and subsamples were processed according to individual protocols. The main criterion for this intercomparison was the agreement of bulk organic carbon (OC) content of the permeate and colloidal fractions. OC blanks ranged from < 10 to > 100 mu M. The variation in OC blanks between CFF systems, even of a single type, suggested that cleaning and handling protocols were critical. One of the primary features of this intercomparison was the large range in apparent retention characteristics of the different CFF systems when used on natural seawater samples. In both settings, the quantity of colloidal material retained by the CFF systems followed the order: Amicon > Filtron greater than or equal to Osmonics > Membrex. For example, in the Hawaii sample, the Amicon CFF system retained on average 43% colloidal OC, while the Filtron, Osmonics and Membrex retained < 4%. Variations within a factor of 2-5 were found within a single membrane type, Other results from this intercomparison suggested that these same relative retention characteristics hold for Fe and Al, organic nitrogen and organic phosphorus, optical properties, and colloid standards. Time-series samples of permeate showed a generally increasing OC concentration with time or concentration factor which must be taken into account when interpreting CFF data. We recommend that considerable carl be taken in quantifying CFF blanks and in assessing the CFF cut-off for each system prior to use in marine applications. Time-series sampling and the use of standard molecules in controlled experiments are encouraged in order to further our understanding of the behavior of natural compound assemblages during CFF processing.