The specific reactive surface area of granular zero-valent iron in metal contaminant removal: Column experiments and modelling
Graphical abstract
Introduction
Zero-valent iron (ZVI) is a widely used reactive material for contaminant removal in remediation and water treatment technologies (Gillham, 2010). Despite its successful commercial application, the relative importance of different removal mechanisms is known to vary with type and concentration of ZVI, contaminants, and other medium components (Alowitz and Scherer, 2002, Wilkin and McNiel, 2003, Rangsivek and Jekel, 2005, Noubactep, 2008). A detailed understanding of removal mechanisms is required in order to optimise the design of in situ permeable reactive barriers (PRBs) and other low-cost water treatment systems (Gillham, 2010, Rangsivek and Jekel, 2011, Hussam and Munir, 2007). Whether contaminant removal is (i) dominated by reactions at localised defects in a developing iron oxyhydroxide layer or (ii) distributed across the entire iron oxyhydroxide film has broad implications for technology development (Gaspar et al., 2002).
To date, experimental investigation of contaminant removal mechanisms has focused on qualitative understanding of the reaction site and quantitative characterisation of kinetic rate constants. Surface analysis techniques, such as auger electron microscopy, have identified localised regions of reactive sites and removal products on the iron oxyhydroxide surface (Gaspar et al., 2002, Rangsivek and Jekel, 2005). Results suggest that contaminant removal may occur at defects in an iron oxyhydroxide layer surrounding an elemental iron core (Gaspar et al., 2002). Multi-site models have been proposed to conceptualise the iron surface area as a collection of different reactive sites (Bandstra and Tratnyek, 2004, Burris et al., 1998, Bi et al., 2010). Bi et al. (2010) investigated reactive and non-reactive sorption sites for the treatment of trichloroethene by ZVI using the transient and steady-state sections of column breakthrough curves. The modelling results indicated that only 2% of sorption sites controlled reactivity (Bi et al., 2010); which could further emphasise the importance of variations in the iron oxyhydroxide layer. Contaminant treatment kinetic rate constants have also been determined using rotating-disc cementation (Lee et al., 1978, Ku and Chen, 1992) and column experiments with varying solution velocities and/or multiple sampling points (Komnitsas et al., 2007, Wüst et al., 1999). These observations have aided in the qualitative understanding of reactive surfaces, however quantification of ZVI reactive proportion of the total surface area from experimental data has not yet been undertaken.
Transport models have also been applied to predict contaminant removal by ZVI (Mayer et al., 2002, Yabusaki et al., 2001, Li et al., 2006, Jeen et al., 2007, O et al., 2009, Carniato et al., 2012). These models use laboratory-defined rate constants scaled to reactive media surface areas (Mayer et al., 2002). Currently, due to varying solution parameters and ZVI source, calibration of some rate constants (based on gas generation and other reaction parameters) is required to model different systems (Jeen et al., 2007, Carniato et al., 2012). Developing new methods of determining reactive surface areas may help in the further understanding of contaminant removal and development of these predictive models.
This study forms part of a broader research program investigating the application of ZVI-based PRBs at contaminated coastal Antarctic sites (Statham et al., 2015a, Statham et al., 2015b), where design of remediation technologies is limited by environmental and logistical considerations. Antarctic PRB systems have been implemented previously (Mumford et al., 2013), using media other than ZVI, in the context of hydrocarbon remediation. These previous studies have elucidated unique design challenges, for example diurnal and seasonal variation in water flow rates and associated variations in the degree of contaminant loading (Mumford et al., 2014). Hence, this study emulates site-specific parameters, including operating temperature, soil salinity, and dissolved oxygen concentration, and quantifies their influence on contaminant removal performance. This process will aid placement, media selection and sizing of future ZVI remediation technologies. While the current research focuses on melt water management, general insights into ZVI behaviour can potentially aid in the development of nano-ZVI based remediation techniques (Zhang, 2003, Tratnyek and Johnson, 2006). The model contaminants Cu2+ and Zn2+ have been selected as these represent potential metals of concern at coastal Antarctic sites (Snape et al., 2001, Northcott et al., 2003); the selection also enables a comparison of contaminants that are, respectively, reducible and non-reducible by metallic iron. A combination of experimental and numerical modelling techniques is used to evaluate the contributions of species diffusion and reaction mechanisms to net contaminant removal rates. Comparison of model-determined reactive surface areas with the total column ZVI surface area can provide further insight into contaminant interaction with iron oxyhydroxide surfaces. While the current findings are readily applicable to cold-climate PRB design, the fundamental results also outline the quantification of ZVI reactive surface areas, which will aid in the general development of ZVI-based water treatment systems.
Section snippets
Experimental method
Experiments were conducted using a vertical glass column of length L = 12.8 cm and diameter dc = 2.8 cm. Columns were dry-filled with a homogeneous mixture of 0–8 wt% Peerless iron (Peerless Metal Powders & Abrasive, cast iron aggregate −8/50 mesh, average diameter 0.50 mm; for other iron properties refer to Statham et al. (2015b)) and an inert packing of ballotini glass spheres (sieve-determined average diameter of 0.58 mm). Dry media was homogenised by combining enough ZVI and ballotini for
Longitudinal dispersion coefficient
Consistent with literature results (Freeze and Cherry, 1979, Woinarski et al., 2006, Steefel, 2008), the longitudinal dispersion coefficient, Dl, was confirmed to depend linearly on the particle Reynolds number, Rep. The resultant correlation is presented in Table 1 (complete data are shown in Fig. S4). As the volume of ZVI used was approximately 0–2vol%, any error associated with determining the dispersion coefficient in a ballotini-only column has been neglected.
Time dependence
Representative changes in
Conclusion
To continue the development of remediation systems for contaminated sites in Antarctica a series of column experiments were conducted under varying experimental conditions. A mass transfer model, using a differential transport equation, was developed to assess kinetic parameters and determine reactive surface areas.
Results from this study suggest that, in de-aerated solutions, film diffusion controls Cu2+ removal but a first-order surface reaction provides a more accurate model for Zn2+ removal
Acknowledgements
Financial support from Australian Antarctic Science Grant 4029 is gratefully acknowledged. The Particulate Fluids Processing Centre, a Special Research Centre of the Australian Research Council, is thanked for provision of experimental equipment. T. S. and L. M. acknowledge support from the Australian Postgraduate Award.
References (66)
- et al.
Lead removal from wastewater by cementation utilising a fixed bed of iron spheres
Environ. Pollut.
(1988) - et al.
Predicting longevity of iron permeable reactive barriers using multiple iron deactivation models
J. Contam. Hydrol.
(2012) - et al.
Kinetics of reductive denitrification by nanoscale zero-valent iron
Chemosphere
(2000) - et al.
Cementation treatment of copper in wastewater: mass transfer in a fixed bed of iron spheres
Chem. Eng. Process. Process Intensif.
(2004) - et al.
Copper cementation on zinc and iron mixtures: Part 1: results on rotating disc electrode
Hydrometallurgy
(2011) - et al.
Effects of dissolved oxygen on formation of corrosion products and concomitant oxygen and nitrate reduction in zero-valent iron systems with or without aqueous Fe2+
Water Res.
(2005) - et al.
Liquid holdup and flow dispersion in zeolite packed beds
Chem. Eng. Sci.
(2001) - et al.
Long-term efficiency and kinetic evaluation of ZVI barriers during clean-up of copper containing solutions
Min. Eng.
(2007) - et al.
Effect of precipitant surface roughness on cementation kinetics
Hydrometallurgy
(1978) - et al.
Modeling porosity reductions caused by mineral fouling in continuous-wall permeable reactive barriers
J. Contam. Hydrol.
(2006)