New Method for Sampling Groundwater Colloids under Natural Gradient Flow Conditions NOAM WEISBROD, † DANIEL RONEN,* ,‡ AND RONIT NATIV † Seagram Center for Soil and Water Sciences, The Hebrew University of Jerusalem , P.O. Box 12, Rehovot 76100, Israel, and Department of Environmental Sciences and Energy Research, Weizm ann Institute of Science, Rehovot 76100, Israel This paper describes a new method for the passive sampling of groundwater colloids using a multilayer sampler (MLS). It is based on the use of dialysis cells with large pore size (10 μm) membranes that are in dynamic equilibrium with the mobile colloid and liquid phases in the aquifer. Under laboratory conditions, the dialysis cells reached equilibrium with a suspension of latex microspheres (5 mg/L) after 44- 100h and with a suspension of kaolinite (16-41 NTU; 20-50 mg/L) after 50-180 h. No fractionation was detected in the particle-size distribution between the kaolinite suspensions inside and outside the dialysis cells. Field profiles, obtained under natural gradient flow conditions in a sand and sandstone aquifer, showed large variability (up to 1 order of magnitude) in the colloid content within profiles (e.g., variation of 7 NTU (∼45 mg/L) between cells located at a vertical distance of 40 cm) and between them. The predominant colloidal particles found in the cells were aluminosilicates, CaCO 3 , silica, and organic matter. The membranes are suitable for sampling groundwater colloids over long periods of time, at least 36 days, in very turbid solutions (up to 50 NTU; ∼550 mg/L). Introduction Colloids are an important mobile phase in aquifers and are carriers of contaminants (1-7). However, field data on colloid concentration,size distribution,and their chemical and mineralogical composition are still scarce due to unreliable sampling protocols (8, 9). Pumping has been shown to produce colloids and associated contaminants that are not naturally present in groundwater (10). Even at the recommended very slow pumping rate of about 100 mL/min (8, 9, 11), the resulting shear rate is 1 order of magnitude higher than that found under natural gradient flowconditions(12). Therefore,colloids maybe artificially formed in the forced gradient fields caused by pumping, and thesamplesobtained maynotrepresentnaturalcolloids or their distribution in the aquifer. Microscale heterogeneity (scale order of centimeters) in the chemicalcomposition and flow field ofgroundwater isalso a well-known phenomenon (13, 14)that can influence colloid generation, stabilization, and transport. Sampling intervals achieved by pumping are generally >0.5 m, and a vertical profile of simultaneous samples is quite difficult to obtain. It is expected that a more adequate method for sampling colloids will enable (a) collection of colloids that are transported by the natural gradient flow field without the need for an external source of energy (pumping); (b) dynamic equilibrium with groundwater so that the con- centration, composition, and size distribution of colloids in the sample vary if they vary in groundwater; (c) simultaneous sampling, at small vertical intervals (centi- meters), of the microscale environments; and (d) sample integrity not to be biased either by the insertion of the sampling device or the retrieval of the sample. Ronen et al.(15)developed a passive method ofsampling groundwater that employs dialysis cells to obtain, simul- taneously,undisturbed discrete microenvironmentswithin the aquifer. The present paper reports on the further development of the multilayer sampler (MLS) technique forthe passive samplingofgroundwatercolloids. It isbased on the use of dialysis cells covered with large pore size membranesthat are in dynamicequilibrium with the mobile colloid and liquid phases in the aquifer. The methodology wastested for artificialand naturalcolloidsin the laboratory in batch and flow experiments. It was also used to obtain field profiles of colloids and groundwater chemical com- position in a sandy Coastal Plain aquifer. Methodology Sampling Principle. The MLS is composed of a sequence ofcriss-crossed dialysis cells separated byseals that fit flush to the well screen (Figure 1). When the MLS is lowered into the groundwater,colloids percolate through large pore size membranes that cover the dialysis cells at both sides. Transport across a dialysis cell membrane is driven mostly bydiffusion. However,for large pore size membranes,the influence of advective transport cannot be ruled out. Hereafter, the term “percolation” is used to denote such combined effects. In the field,the impact ofMLSinsertion and retrieval from a well on sample integrity is negligible. This is due to (a) the relatively long time the MLS is immersed in groundwater after deployment (weeks),which allows for re-equilibration with the natural conditions of the aquifer, and (b) the short time needed (minutes) for MLS retrieval as compared to the long equilibration time between the cell solution and groundwater (days; see LaboratoryEquilibration Experimentssection). Asthe MLS is retrieved from the well, the sample is preserved inside the cell by surface tension at the membrane -air interface. The MLS (15) enables acquisition of vertical profiles of the concentration and composition ofcolloidsand groundwater chemistry at variable vertical intervals (e.g., 3.5 cm Figure 1a and 12.5 cm Figure 1b). *Corresponding author e-mail address: cidaniel@weizmann. weizmann.ac.il; fax: 972-8-9344124. † The Hebrew University of Jerusalem. ‡ Weizmann Institute of Science. Environ. Sci. Technol. 1996, 30, 3094-3101 3094 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 10, 1996 S0013-936X(96)00197-6 CCC: $12.00  1996 American Chemical Society