Membrane Resistance

Two types of polymeric membranes have been used for passive samplers. Non-porous membranes include LDPE [3,5,6,31,32], polypropylene and polyvinylchloride [3,33], PDMS [3,33-35], polyimide [36], polyacrylate (PA) [37,38] and other non-polar polymers [38]. Micropo-rous membranes include regenerated cellulose [4,39,40], polyethersulf-one (PES) [41], polysulfone (PS) [32] and polyacrylamide hydrogel [42]. Some other membranes used are discussed by Stuer-Lauridsen [43] in an extensive review of passive sampling techniques. In some applications, the membrane is also the primary accumulation site of the anal-ytes (TwisterTM bars, LDPE strip samplers, SPME, silicone strip samplers). In other applications, the membrane is meant to separate a sorption phase from the water (diffusive gradients in thin films (DGT), Chemcatcher, MESCO, SPMD) and to reduce the flux to the sorption phase.

The conductivity to mass transport through the membrane is given by kmKmw ^ DmKmW (7.31)

dm where dm is the thickness of the membrane (Eq. (7.5)). Both Dm and Kmw are compound-dependent. The role of Kmw in Eq. (7.31) may be appreciated by considering that compounds with high membrane-water partition coefficients will have similarly high concentrations at the membrane side of the membrane-water interface. As a result, the concentration gradient over the membrane is elevated compared with that found for compounds with low Kmw values, and the steeper concentration gradient results in a larger flux through the membrane. Conversely, the selection of a membrane for which the target analytes have a low affinity (e.g. hydrophilic membranes for sampling hydrophobic compounds) results in an enhanced transport resistance posed by the membrane and to reduced sampling rates. Several examples of this effect have been reported. A comparison between solvent-filled cellulose and polyethylene membranes showed that the uptake rates of organo-chlorine pesticides by the samplers with cellulose membranes were lower by two orders of magnitude [40]. Similarly, the uptake kinetics of hydrophobic contaminants by the MESCO and Chemcatcher were greatly enhanced by replacing the hydrophilic membrane by polyethylene [5,6], and the uptake rates of the polar compounds diazinon, ethynylestradiol and atrazine by the polar organic chemical integrative sampler (POCIS) were much larger with PES membranes than with polyethylene or Nylon-66 membranes. The choice of membrane material has an effect not only on the sampling rates, but also on the flow sensitivity of the sampler. When the membrane resistance becomes smaller, rate control switches more to side of the WBL, which is by nature dependent on the hydrodynamic conditions at the membrane-water interface. Therefore, attempts to reduce the flow sensitivity of passive samplers by installing membranes that have lower partition coefficients for the analytes, automatically reduce the sampling rates. Conversely, membranes with high Kmw values enhance sampling rates but also increase the sensitivity of these samplers to changing flow conditions [5]. Whether or not reduced sampling rates are problematic, depends of course on the aqueous concentration levels, the exposure time and the sensitivity of the analytical equipment. No general rule can be given, but in the light of the above, it seems unlikely that a flow-insensitive passive sampler can be developed that has sufficiently high sampling rates in all environments.

Estimating sampling rates of compounds for membrane-controlled uptake is hindered by the scarcity of data on diffusion coefficients, particularly for compounds of environmental interest. Diffusion coefficients (- m) in LDPE have been collected from the engineering literature by Hofmans [8]. She proposed to model —m as a function of molecular weight (M) according to log — m - -7.47 - 2.33 log M

where —m is in units of m2 s-1. Diffusion coefficients of PAHs in PDMS appear to be higher than in LDPE by about two to three orders of magnitude and —m values of PAHs in polyoxymethylene are about one order of magnitude lower than in LDPE (Tatsiana Rusina, Research Center for Environmental Chemistry and Ecotoxicology, Masaryk University, Czech Republic, personal communication). These observations are consistent with the theory that diffusion coefficients increase with increasing segmental mobility and free volume fraction of the polymer [14,44,45], and decrease with increasing glass-transition temperature of the polymer [14].

A large volume of data on PDMS-water and PA-water partition coefficients of organic contaminants can be found in the SPME literature [12,38,46-49,50]. A smaller data set is available for the case of LDPE [30,45,51]. Available log Kmw values are shown in Fig. 7.3 as a function of log Kow. Although the scatter is rather high, some general trends can be identified. Log Kmw values for LDPE are higher than for PDMS by 0.7 log units, on average. In the range 1<log Kow<4.5, the log Kmw values for PA are 0.3 log units higher than those for LDPE, but this trend does not seem to persist in the higher log Kow range. The log Kmw data could be modelled by

Ldpe Membrane
Fig. 7.3. Membrane-water partition coefficients and regression models for LDPE (filled circles, drawn line — linear fit), PDMS (open circles, dashed line — linear fit) and PA (asterisks, dotted line — quadratic fit).

PA : log Kmw - -0.0629 log KL + 1.341 log K, (R2 - 0.88, s - 0.50, n - 74)

0 0

Post a comment