R. Siepmann, F. von der Kammer and U. Förstner Technical University Hamburg Harburg, Dept. of Environmental Science and Technology, Eißendorfer Str. 40, D-21073 Hamburg (E-mail:
[email protected];
[email protected];
[email protected]) Abstract The efficiency of road run-off filtration facilities based on ion-exchange materials is reduced by pollutants which are transported bound to particles. To quantify the factors governing particle transport phenomena, a simplified model consisting of quartz sand-filled columns representing the filter/soil was set up. Suspensions of artificial clays, cold water-extracted natural clays, and real run-off were used as model effluents. Five experiments were performed: breakthrough of a natural soil suspension, remobilization of a natural soil suspension after ionic strength-drop, the same two experiments with a suspension of the artificial clay mineral Laponite, and the remobilization of run-off accumulated on a column at high ionic strength with an ionic strength down-gradient. Short-interval effluent fractions were analysed by flow-field-flowfractionation (F4) to obtain the size distributions of the colloids present. The size distributions of subsequent fractions were then plotted in a staggered arrangement to give three-dimensional graphs that are time- and particle size-resolved. With this method the subsequent release of different agglomerate sizes formed on the column could be shown for the artificial clay mineral, questioning its use as a model colloid. The combined particle size- and time-resolved plots proved to be a powerful tool for monitoring colloidal solids in column effluents. Keywords Colloids; field-flow fractionation; monitoring; run-off; stormwater
Water Science and Technology Vol 50 No 12 pp 95–102 © IWA Publishing 2004
Colloidal transport and agglomeration in column studies for advanced run-off filtration facilities – particle size and time resolved monitoring of effluents with flow-field-flowfractionation
Introduction
Pollutants in the subsurface environment do not only originate from point-source plumes on heavily contaminated sites. The diffuse and large area immissions from power plants, heatings and traffic exhaust plays an important role in the general degradation of soil and groundwater quality especially in urban areas (Fuchs et al., 1997). Non-point source pollutants may be different heavy metals like Zn, Cu, Pb, Cd which originate from e.g. industrial, household and engine exhausts, metal roofs (Förster, 1999) and car body corrosion, and different organic compounds like PAHs from incomplete combustion. It is well known that the subsurface transport of these pollutants does not only occur in solution, but also to a certain extent bound to mobile colloidal particles, depending on the pollutant, immission, soil and groundwater characteristics. Extensive reviews on this subject were published by McCarthy and Zachara (1989), Ryan and Elimelech (1996) and Kretzschmar et al. (1999). One major source of pollution in densely populated areas is the uncontrolled discharge of road runoff into surface waters, retention ponds or infiltration ditches. State-of-the-art filtration consists of flat bed sand filters with or without plantation, vertical sand filters or different systems of direct infiltration into the vadoze zone. Colloid bound transport of pollutants or the genesis of particular pollutants may occur first on the road by the mixing of dust from dry and wet deposition with tyre and brake-pads wear, soot particles from car exhaust, or metal oxides from car body corrosion (Grout et al.,
95
R. Siepmann et al. 96
1999). The colloid-bound or colloidal pollutants are washed into and through filters or discharge facilities. Afterwards in-situ mobilized soil particles take up contaminants and accelerate them on their way towards and inside groundwater (Set and Karathanasis, 1997). High sodium chloride contents of the run-off due to the use of de-icing salts in winter and a rapid lowering of the ionic strength in the following thawing period were discovered to be main factor for particle remobilization along roadsides. (Amrhein et al., 1993; Norrström and Bergstedt, 2001). Recent approaches for road run-off treatment are based on reactive low-cost filter materials such as natural zeolite, which remove dissolved heavy metal species by ion exchange. The retention efficiency of such systems is lowered by pollutants that are strongly sorbed on particles that are mobile in the filter. They simply pass the filter without being exchanged since cation exchange mechanisms are only effective for the truly dissolved or weakly bound fraction of the cations (Förstner et al., 2001). Main factors facilitating colloidal enhanced transport mechanisms in surface run-off infiltration systems are: • elevated concentrations of particulate matter and colloidal particles in run-off • a large fraction of e.g. heavy metals is bound to particles and colloids • transport distances to the groundwater surface are short (typical < 1 m to some metres) • variability of hydraulic loads and resulting flow rates • extreme variability of ionic strength at roads in winter (e.g. 2–16,000 mg L–1 Na+). In order to quantify the main factors that govern particle transport in these advanced filtration facilities, a simplified laboratory-scale column model was set up, which consists of glass columns filled with quartz sand of different grain sizes. Breakthrough, retardation and remobilization experiments are performed with artificial, well defined model colloid suspensions like Laponite, a small sized clay used to form gels in the cosmetic industry, as well as with stable suspensions of natural soil colloids (stabilized cold water soil extracts), and collected road run-off. The experimental conditions can be varied and adjusted separately with regard to the colloidal particles present in the tested effluents. Chemical factors include ionic strength, solution composition, and pH, while physical factors include flow speed and water saturation. The effluents from the column experiments are further analyzed by symmetrical flowfield-flow-fractionation (F4), a hydrochromatographic method that separates different particle sizes by applying a field (here: cross-flow) vertical to the flow direction in the separation channel (Beckett and Hart, 1993). The size separation is followed by multidetector analysis (MDA) which applies UV-VIS-DAD, 3D-fluorescence and combined online static (SLS) and dynamic (DLS) light scattering detectors (v.d. Kammer and Förstner, 1998). The UV absorption/turbidity data, fluorescence and 90° angle scattering intensities give a suitable approximation of particle mass concentration over the FFF elution profile which can be calibrated against particle size standards delivering particle size distributions. Natural organic compounds as humic acids can be distinguished qualitatively by comparing fluorescence and turbidity/ scattering signals. Possible interferences of the ideal FFF particle size fractionation by particle shape or particle-membrane interactions can be evaluated with SLS and the derivation of particles radius of gyration independently from SLS data (Wyatt, 1997) The method on which we focus in the presented work combines short-interval effluent sampling from the test columns and applying the F4-analysis to obtain the size distribution of each eluting sub-sample. These are combined in a staggered arrangement to give threedimensional plots that are time- and particle size-resolved. In contrast to column filtration studies applying only UV-VIS turbidity or DLS detection/sizing of particles at the column outlet, the presented method obtains high-resolution effluent volume to size distribution
Materials and methods Laboratory column facility
R. Siepmann et al.
profiles that give a dynamic picture of the changes in particle size and composition during different column experiments performed. The final goal of these experiments is to find low-cost, passive techniques to alter the physico-chemical conditions in a run-off treatment plant by applying suitable reactive materials. These materials should be capable of maintaining hydrochemical conditions (e.g. by release of Ca2+) for retaining run-off particles by destabilization and surface deposition of particles/ colloids on coarse filter grains rather than physical filtration with subsequent blocking of filter pores.
Glass columns of 25 mm inner diameter were wet-packed with pure (99%) quartz sand with a mean grain size of 0.25 mm to a height of 30 cm, resulting in a porosity of 0.38 and a pore volume of 57 ml. Saturated columns were operated with bottom to top flow to avoid air bubbles being trapped. The effluent of each column was monitored by an online analytic chain consisting of a three-angle laser-light-scattering photometer (Wyatt Mini-Dawn), UV/VIS, conductivity and pH. Start of sampling of effluent fractions from multiple columns was triggered by detecting predefined threshold levels in the analytic chain. The effluent mixtures were provided by peristaltic pumps, programmed hydrochemical effluent gradients were provided by connected flow paths of two computer controlled peristaltic pumps. Model colloids and natural colloids
1. The artificial, industrially produced clay mineral Laponite, a magnesium lithium silicate with the formula Na+0.7 [(Si8 Mg5.5 Li0.3) O20 (OH)4]0.7–, which consists of uniform, small, equal-sized platelets of approximately 25 nm diameter, was used as a simplified model colloid for breakthrough and remobilization experiments. Laponite was chosen because it is artificially produced and therefore free from contaminants, it has a definite chemical composition, and a small but uniform particle size. The disadvantage of Laponite was, however, that due to the small particle size a fairly large concentration had to be used in order to match the sensitivity of the analytic instruments. The stable Laponite suspension was at pH 9.5. 2. The water-extracted soil suspension named “reference 3” was produced from a 1:10 w/w mixture of soil from an aquifer matrix and a till by multi-step 1:10 cold water extraction including a washing step with 0.1 mol L–1 NaNO3, shaking/ sonication with ultra pure water and sufficient sedimentation duration (12 months) until a stable colloidal suspension in the supernatant was achieved. The resulting suspension (“reference 3”) contained 76 mg L–1 colloids below 500 nm (solid mass retained on Whatman Anopore 20 nm membranes). 3. Road run-off taken at a run-off treatment pilot plant and testing site, located at a highly frequented federal road (approx. 20000 vehicles/day), located near the highway A7 in Hamburg Harburg. The sample used was collected from the plants sand trap after a 14day period with temperatures constantly below 0°C, so the complete inventory of the sand trap was generated by thawing due to de-icing salts. It contained about 61 mg L–1 suspended solids at a sodium content of 3,38 g L–1, equivalent to a 147 mmol L–1 sodium chloride solution. Apparent pH was 6.5. F4-analysis
Symmetrical F4 (F1000 Flow-FFF channel from FFFractionation, Utah) followed by MDA (v.d. Kammer and Förstner, 1998) was used to determine the size distribution of colloids in
97
R. Siepmann et al.
the sampled fractions of the column effluents. The size distributions depicted refer to the distribution of relative intensities at 90° scattering angle over the particle size calibrated FFF elution volume. Due to the dependency of scattering intensity from particle size the given distributions are intensity weighted approximations of the actual size distribution. The single Laponite particles are true Rayleigh scatterers which prevents independent particle size analysis of the FFF effluent by the used SLS equipment, while agglomerates of Laponite extend into Debye scattering and can be evaluated for particle size applying Rayleigh-Gans-Debye approximation in the SLS software (Wyatt, 1997). In FFF size distributions concentrations of single Laponite particles may be underestimated. The channel was calibrated using five different sizes of monodisperse latex beads (Duke Scientific) of defined sizes between 50 and 350 nm. Size distributions of each collected fraction were arranged to give a three-dimensional time resolved plot of size distributions. Dynamic changes in the size distribution of the particles eluted could be monitored (see Figure 1). The following experiments give an example of the use of time- and particle-sizeresolved monitoring. Experiment 1: Breakthrough of a natural colloid extract at low ionic strength: Columns were fed with ultrapure water at a velocity 1 m d–1. 3 ml undiluted natural colloid extract “reference 3” was injected. Experiment 2: Remobilization of a natural colloid extract deposited on the column by stepping down the ionic strength: The column was preconditioned with 20 mmol L–1 sodium chloride solution, 3 ml “reference 3” were injected at 1 m d–1 flow speed, no breakthrough occurred. Feed solution was changed to ultrapure water while maintaining the flow rate and the subsequent release of the colloids was monitored. Experiment 3: Breakthrough of a Laponite suspension: columns were fed with ultrapure water at a velocity 1 m d–1. 400 µl of Laponite suspension (10 g L–1) was injected. Experiment 4: Remobilization of Laponite deposited on the column by stepping down the ionic strength: The column was preconditioned with 20 mmol L–1 sodium chloride solution, 3 ml Laponite suspension was injected at 1 m d–1 flow speed, no breakthrough occurred. Feed solution was changed to ultra pure water while maintaining the flow rate and the subsequent release of the Laponite was monitored. Experiment 5: Thawing simulation with a winter run-off sample. A column preconditioned with 147 mmol L–1 sodium chloride solution was fed with 10 pore volumes of freshly collected road run-off with a similar sodium content. The feed solution was changed to online breakthrough monitoring
event-contolled sampling
int.
F4 + multi detection
size and time resolved breakthrough plot int.
time
pore volumes (time axis)
start level
start time
98
: hydrodynamic diameter (size axis)
Figure 1 Principle of time- and particle-size resolved monitoring of column effluents
particle-free sodium chloride solution and the ionic strength was gradually lowered to zero within 32 pore volumes. Partial remobilization of the colloids was monitored shortly before the input ionic strength reached zero. Results and discussion
Experiment 1: The time- and particle size-resolved plot of a “reference 3” break-through at zero ionic strength shows a just slightly retarded elution profile (see Figure 3). The size distribution of the eluted particles does not change during the breakthrough, and no size exclusion effects, which would mean a higher ratio of larger particles coming out at the beginning of the breakthrough, could be observed. The size distribution observed is similar to that in the original suspension.
R. Siepmann et al.
Figure 2 shows the size distribution of the natural colloid extract used. It contains particles in a range from 20 to 500 nm with a maximum around 150 to 200 nm. These data from the original material are compared to the distribution plots depicted below.
Experiment 2: The plot in Figure 4 depicts the release of natural clay colloids from the column at stepping down the ionic strength to de-ionized water. The breakthrough occurs slightly delayed after the pore volume that followed the step-down. The breakthrough is very fast, complete, and no preferred particles sizes are eluted first. The size distribution is in analogy to the fed-in original suspension. Figure 5 depicts the size distribution of a stable suspension containing 10 g L–1 of the artificial clay mineral Laponite. The sizes measured correspond exactly to the specifications given by the manufacturer. The mean size of the platelets being 25 nm.
Figure 2 “Reference 3” size distribution measured with F4
Figure 3 Breakthrough of “reference 3”
99
R. Siepmann et al.
Experiment 3: The breakthrough after injection of a small amount of Laponite suspension is shown in Figure 6. The breakthrough is more retarded and less compact than that of the natural colloid extract. While a large amount of colloids is released as single particles, a second peak shows up with a maximum at 200 to 300 nm, which can only be explained by the forming of agglomerates on the column. The retarded breakthrough and the simultaneous elution of single particles and agglomerates suggests a dynamic sorption-desorption process of the mobile Laponite with the stationery quartz grains. This is possible because of the lower pH of 7 on the column compared to the pH of 9.5 of the bulk suspension. This lowers the negative surface charge of the Laponite and facilitates weak sorption of the positively charged crystal edges on the negatively charged quartz surfaces as well as the forming of agglomerates. Experiment 4: If a Laponite suspension is fed into a column at a sodium content of 20 mmol L–1, which is then switched to de-ionized water, only agglomerates are released very
Figure 4 Remobilization of “reference 3”
Figure 5 Laponite, size distribution measured with F4
100
Figure 6 Breakthrough of Laponite
early beginning at 0.84 pore volumes from the step down of the ionic strength (see Figure 7). The subsequent lowering of agglomerate sizes during the elution can be clearly seen. This indicates a size exclusion effect or the early mobilization of bigger agglomerates at the beginning of the ionic strength drop zone that moves through the column, and the subsequent break-up of larger agglomerates at lower ionic strengths. R. Siepmann et al.
Experiment 5: Simulation of thawing conditions with a winter run-off sample deposited on the column at high ionic strength. A linear gradient for lowering the ionic strength was applied in the experiment (see Figure 8). Partial remobilization of mainly small particles begins at the comparably low Na+ input concentration of 15 mmol L–1 (see Figure 9). The
Figure 7 Remobilization of Laponite
Figure 8 Down-gradient of ionic strength in a column preloaded with run-off, remobilization
Figure 9 Thawing simulation with real run-off
101
release is very sharp and due to the large fractions collected, no change in the size distribution can be monitored at the peak. Conclusions
R. Siepmann et al.
For the simulation of transport processes in filter facilities with simplified model colloids, Laponite is no suitable candidate. The agglomeration tendencies of Laponite differ very much from those of the natural colloid extract tested. Possible reasons for this are the almost mono-disperse and small size of the Laponite crystals, pH-shifts on the columns and the high input concentration that had to be used to ensure detection (SLS on-line at column and UV/VIS-turbidity), which is also far beyond possible conditions in the field. On the other side, these findings indicate that if favourable physical and chemical factors are present in a filter facility, agglomeration phenomena may occur among the present particles, accompany and affect the transport of particles along the flow path. This may result in reduced desorption rates of pollutants from the inside of agglomerates, retardation and filtration with subsequent filter blocking of big agglomerates, fast movement of small agglomerates due to size exclusion, and release of contaminated colloids after the break-up of agglomerates. The combination of small period sampling followed by F4 size separation and detection of colloids presents a powerful tool not only to monitor size exclusion effects, but also to visualize agglomeration and other phenomena that may alter the transport of colloidal particles. If F4 size separation is also followed by collecting fractions and these can be subjected to further elemental analysis, even breakthrough and release plots of the pollutants like heavy metals are possible. References
102
Amrhein, C., Mosher, P.A. and Strong, J.E. (1993). Colloid-Assisted Transport of Trace Metals in Roadside Soils Receiving Deicing Salts. Soil Sci. Am. J., 57, 1212–1217. Beckett, R. and Hart, B. (1993). Use of Field-Flow Fractionation Techniques to Characterize Aquatic Particles, Colloids, and Macromolecules. In: Buffle, J., van Leeuwen, H.P. (eds). Environmental Particles, 2, IUPAC Series on Environmental Analytical and Physical Chemistry, Lewis Publ., Chelsea, USA. Förster, J. (1999). Variability of Roof Runoff Quality. Wat. Sci. Tech., 39(5), 137–144. Förstner, U., Jacobs, P. and Kammer, F. v. d. (2001). Impact of Natural Nanophases on Heavy-Metal Retention in Zeolite-Supported Reactive Filtration Facilities for Urban Run-Off Treatment. Fresenius J. Anal. Chem., 371, 652–659. Fuchs, S., Haritopoulou, T., Schäfer, M. and Wilhelmi, M. (1997). Heavy Metals in Freshwater Ecosystems Introduced by Urban Rainwater Runoff – Monitoring of Suspended Solids, River Sediments and Biofilms. Wat. Sci. Tech., 36(8–9), 277–282. Grout, H., Wiesner, M.R. and Bottero, J.-Y. (1999). Analysis of Colloidal Phases in Urban Stormwater Runoff. Environ. Sci. Technol., 33, 831–839. v.d. Kammer, F. and Förstner, U. (1998). Colloidal Transport: Redox and pH Dependant Alteration of Colloid Characteristics Examined by Flow-Field-Flow-Fractionation-Multi-Detector Analysis. Contaminated Soil ’98, 1, Thomas Telford Publ., London. Kretzschmar, R., Borkovec, M., Grolimund, D. and Elimelech, M. (1999). Mobile Subsurface Colloids and Their Role in Contaminant Transport (Review). Adv. In Agronomy, 66, 121–193. McCarthy, J.F. and Zachara, J.M. (1989). Subsurface Transport of Contaminants. Environ. Sci. Technol., 23, 496–502. Norrström, A.-C. and Bergstedt, E. (2000). The Impact of Road De-Icing Salts (NaCl) on Colloid Dispersion and Base Cation Pools in Roadside Soils. Water, Air, and Soil Poll., 127, 281–299. Ryan, J.N. and Elimelech, M. (1996). Colloid Mobilization and Transport in Groundwater. Colloids Surfaces A,, 107, 1–56. Set, A.K. and Karathanasis, A.D. (1997). Stability and Transportability of Water-Dispersable Soil Colloids. Soil Sci. Am. J., 61, 604–611. Wyatt, P.J. (1998). Submicrometer Particle Sizing by Multiangle Light Scattering following Fractionation. Journ. of Colloid Interf. Sci., 197, 9–20.