Dataset: Multi Angle Laser Light Scattering

Methodology:
Flow field-flow fractionation (FlFFF) is a chromatography-like elution technique based on hydrodynamic principles, where colloid particles are separated due to their interaction with a cross-flow carrier liquid, applied over the cross section of a ribbon-like channel that is thin (~250 μm) and flat (~centimeter length). Here, the carrier solution was artificial seawater (ASW) that mimics the matrix of the sample solutions, such that the sample remains unaltered by the FlFFF processing. The ASW was first run through a column of Chelex resin (to remove metals) and then ultrafiltered (at 5 kDa nominal pore size, to remove any resin released during cleaning). This study is the first to use ASW as the carrier solution while coupling FlFFF to ICPMS, since previous studies worked only with lower salinity samples where 15 mmol L−1 NH4NO3 buffer (pH 7) was used as the carrier for freshwater and 55 mmol L-1 NH4Cl(aq) (pH 8) was used in saltier (estuarine) water to resemble ~10% of sweater concentration to not clog the ICPMS cones.

The bottom wall of the FlFFF channel has a permeable (10 kDa) ultrafiltration membrane where the cross flow and soluble-sized material pass through to waste so that only the colloidal sized particles are retained within the channel. Two flows govern the FlFFF size exclusion chromatography elution: the (parabolic) sample flow and the cross flow. The sample in the ASW carrier solution flows tangentially to the ultrafiltration membrane, and a parabolic flow profile is created by a function of the thin, ribbon-like channel, as the walls drag the flow on the top and the bottom of the channel, producing a faster laminar flow in the center of the channel. The cross flow through the membrane creates an opposing diffusion force, where smaller compounds with higher diffusivities can diffuse against the cross-flow better and thus remain in the center of the chamber, while larger colloids remain lower towards the wall of the parabolic flow profile. The unequal velocities in laminar flow created by the parabolic flow profile down-channel, with constant Brownian motion based on diffusion, effectively separates the colloids in a 20 mL sample by producing a gradient of different size fractions (~eighteen 1.5 mL aliquots) that elute through the channel, with the smaller colloids eluting earlier than the larger colloids. Based on the diffusion coefficients of the separately eluted colloid size fractions, the hydrodynamic diameter, e.g. the diameter of a compact sphere, can be calculated using Stokes Law.

Sampling and analytical procedures:

The procedure for FlFFF includes three steps: sample loading and preconcentration, relaxation or focusing, and elution or colloid separation (Lyvén et al., 1997; Baalousha et al., 2011). The AF-2000 FlFFF instrument that was employed in this project has been modified to allow large volumes (in our case 20 mL) of ultrafiltered seawater to be back-loaded into the channel, where the accumulating colloids are focused at the head of the channel (into ~70-140 μL) during the first 50 minutes of run time. This small volume allows most colloids to start down the channel at the same time, which minimizes the issue of “blurring” of the size partitioning. We had some concern that the 286x pre-concentration at the channel head would allow a change in colloidal size and shape distribution during pre-concentration, though others have found that focusing even larger volumes of up to 500 ml does not appreciably alter the size distribution of organic colloids (Floge and Wells, 2007).

Next during the relaxation stage, the sample no longer experiences a large carrier focusing force pushing it up to the head of the channel. Then, the colloids balance the forces of Brownian diffusion and crossflow velocity, resulting in the separation of colloids along the membrane. This process of separation also occurs during preconcentration but is allowed to continue after for a short period after sample focusing but before the channel flow starts. The third phase, elution, begins when the channel flow pumps the carrier solution from the head of the channel, pushing the separated colloids through so that they elute out of the end of the channel. This channel outflow then passes through both the UVvis absorbance detector and the MALLS detectors online before being fraction collected into ~eighteen 1.5 mL aliquots. Each 1.5 mL fraction was acidified to pH 2 using Optima-grade hydrochloric acid (to 0.012 M HCl) in order to mitigate metal adsorption onto vial walls.

Prior to running each sample through the FlFFF, a daily FlFFF procedural blank was run through the system using the chelexed and ultrafiltered artificial seawater as the carrier and ultrapure water (18.2 MΩ-cm Milli-Q, MQ, Millipore) as the sample. The entire FlFFF run takes ~100 minutes with the channel outflow rate totaling 0.5 mL/min.

In addition, time was minimized between sample collection and FlFFF processing (<3 hours) to limit the potential bottle effects of storage, such as colloid sorption to bottle walls and colloidal instability (Fitzsimmons and Boyle, 2012). However, it has also been found that bottle storage for as long as 24 hours has minimal effects on colloid size distribution and that freezing does not affect the integrity of the organic colloidal size spectrum (Floge and Wells, 2007; Wells and Goldberg, 1994). Therefore, we minimized storage time and directly evaluated the effect of freezing (-20 ̊C) on the colloidal size distribution in fresh-frozen comparison tests on the same sample (see below).

Before each new sample run, the entire FlFFF-UVvis-MALLs system experienced a cleaning procedure of ~20 minutes each of flushing with 5% methanol (to wash out the trapped organics) and then MQ, and then dilute acid (trace metal grade 0.1M HCl) was loaded into the system and left overnight before final flushing with MQ until the pH was >6 and the MALLS detector angles read <1.5 for the scattering intensity values. Finally, the chelexed and ultrafiltered ASW was flushed through the system to prime it for the sample.

Matlab codes were written to 1) import the raw PostNova UV and MALLS data, and 2) process these data to generate colloidal size outputs. The importation code is run first and then the processing code. See Supplemental Documents for example m-files.