Article pubs.acs.org/est
Silver Dissolution and Release from Ceramic Water Filters Anjuliee M. Mittelman, Daniele S. Lantagne, Justine Rayner, and Kurt D. Pennell* Department of Civil and Environmental Engineering, Tufts University, Medford, Massachusetts 02155, United States S Supporting Information *
ABSTRACT: Application of silver nanoparticles (nAg) or silver nitrate (AgNO3) has been shown to improve the microbiological efficacy of ceramic water filters used for household water treatment. Silver release, however, can lead to undesirable health effects and reduced filter effectiveness over time. The objectives of this study were to evaluate the contribution of nanoparticle detachment, dissolution, and cation exchange to silver elution, and to estimate silver retention under different influent water chemistries. Dissolved silver (Ag+) and nAg release from filter disks painted with 0.03 mg/g casein-coated nAg or AgNO3 were measured as a function of pH (5−9), ionic strength (1−50 mM), and cation species (Na+, Ca2+, Mg2+). Silver elution was controlled by dissolution as Ag+ and subsequent cation exchange reactions regardless of the applied silver form. Effluent silver levels fell below the drinking water standard (0.1 mg/L) after flushing with 30−42 pore volumes of pH 7, 10 mM NaNO3 at pH 7. When the influent water was at pH 5, contained divalent cations or 50 mM NaNO3, silver concentrations were 5−10 times above the standard. Our findings support regular filter replacement and indicate that saline, hard, or acidic waters should be avoided to minimize effluent silver concentrations and preserve silver treatment integrity.
■
INTRODUCTION An estimated 768 million people do not have access to an improved drinking water source, and an additional 1.2 billion are estimated to drink microbiologically contaminated water from improved sources.1 Household water treatment and safe storage (HWTS) options, such as chlorine or filtration methods, can serve as an effective means for providing safe drinking water in areas where water sources are untreated, insufficiently treated, or become contaminated during distribution or storage.1 Ceramic “pot” water filters (CWFs) are one HWTS technology, and consist of an approximately 10 L capacity ceramic filter that rests on its rim in a lidded receptacle. The receptacle serves as a safe storage container and is fitted with a tap for dispensing filtered water.2 CWFs are locally produced in more than 50 countries by firing a mixture of clay, burn-out material, and water to produce a porous ceramic. Silver is added to the filters either after firing or directly into the clay/burn-out material mixture. Silver addition improves the microbiological efficacy of CWFs3−5 and inhibits biological growth within filters and plastic receptacles,3,6 potentially reducing filter clogging. Casein-coated silver nanoparticles (nAg) are the most commonly used silver formulation for CWFs, whereas some factories use silver nitrate (AgNO3) due to lower cost and/or local availability.7 Water treatment in CWFs is achieved through three mechanisms: physical straining of larger particles, such as protozoa, due to pore size exclusion; physical-chemical attachment of particles (some bacteria and viruses) to the ceramic surface; and the antimicrobial action of silver. While © 2015 American Chemical Society
silver released into the treated water may prevent recontamination during storage, silver release over time may also decrease the effective lifespan of the filter and result in increased silver exposure. The World Health Organization and the U.S. Environmental Protection Agency have established a secondary standard of 0.1 mg/L for silver in drinking water.8,9 Thus, there are competing requirements of applying sufficient silver to ensure disinfection during treatment and storage while not exceeding drinking water standards. Prior research has shown that after painting, dipping, or firing-in of nAg or AgNO3, silver concentrations in effluent water fall below 0.1 mg/L after 8−24 h of filter throughput.3,4,10−12 A limited number of studies have examined the impact of different water chemistries on silver fate in CWFs, and have shown that pH, ionic strength, dissolved organic matter, and chlorine content can influence silver elution.11−13 To date, no studies have examined the impact of transient water chemistry on silver retention, the mechanisms governing silver transport and release (e.g., nanoparticle dissolution or cation exchange) under different influent chemistries, or consistently distinguished between Ag+ and nAg in effluent samples. Based on observations from prior nAg transport studies conducted in water-saturated sands,14 we hypothesized that nAg would be strongly retained within the ceramic filter Received: Revised: Accepted: Published: 8515
March 20, 2015 June 8, 2015 June 11, 2015 June 11, 2015 DOI: 10.1021/acs.est.5b01428 Environ. Sci. Technol. 2015, 49, 8515−8522
Article
Environmental Science & Technology
UK),24 and sealed with silicone and neoprene O-rings (McMaster-Carr, Elmhurst, IL) at disk top and bottom to ensure that applied solutions did not preferentially flow around the disks. Nonreactive Tracer Tests. Nonreactive tracer tests were performed in water-saturated disks to confirm the total pore volume (PV), characterize water flow and hydrodynamic dispersion in each disk, and evaluate variation in disks from the same batch (see Supporting Information). Effluent bromide concentration data were simulated using a dimensionless form of the advective-dispersive transport equation, which can be written as
matrix, and would slowly dissolve, releasing Ag+ into effluent water over time. The form of silver released from filters could have implications related to disinfection performance, as the biological activity of nAg results primarily from release of silver ions, with the degree of nAg dissolution correlated with antimicrobial activity.15−17 In the silver oxidation equation, protons and dissolved oxygen are required for dissolution to occur. Ag(s) +
1 1 + + O2 + H(aq) ↔ Ag(aq) + H 2O 4 2
(1)
Batch reactor studies have also demonstrated that nAg dissolution rates are dependent upon solution temperature, ionic strength, electrolyte species, and dissolved organic matter content, as well as the concentration, particle size and surface coating of nAg.18−21 For example, acidic conditions are known to promote nAg dissolution,21 whereas other water properties can inhibit dissolution reactions and reduce the antibacterial effectiveness of nAg. For example, dissolved organic matter (e.g., humic acid) has been shown to coat particle surfaces, thereby inhibiting dissolution reactions and preventing physical interaction between nAg and microorganisms.21−23 The objective of this study was to evaluate the fate and transport of nAg and Ag+ in ceramic filter disks painted with 0.03 mg/g of nAg or AgNO3, and exposed to influent waters of varying pH (5, 7, 9), ionic strength (1, 10, 50 mM), and cation species (Na+, Ca2+, Mg2+). Physical and chemical properties of the ceramic filters were characterized using mercury porosimetry, nonreactive tracer tests, cation exchange capacity (CEC) measurements, and scanning electron microscopy (SEM). Derjaguin, Landau, Verwey, Overbeek (DLVO) theory was used to qualitatively evaluate the effects of changes in solution chemistry on nAg attachment and detachment behavior. In addition, a set of experiments utilized filters painted on only the top surface, rather than both the top and bottom, to determine the relative amounts of silver released from each surface. The experimental results informed recommendations for CWF use and longevity as a function of source water chemistry.
RF
∂C* 1 ∂ 2C* ∂C* = − 2 ∂PV Pe ∂X ∂X
RF = 1 +
(2)
vpt vpL KDρb x C ; C* = ; PV = ; Pe = ; X* = θw C0 L DH L (3)
where RF is the solute retardation factor, ρb is the bulk density of the matrix, C is the aqueous concentration, C0 is the influent or applied aqueous concentration, PV is the dimensionless pore volume, vp is the pore-water velocity, t is time, L is the length of the domain in the direction of flow (filter disk thickness), Pe is the Peclet number, DH is the hydrodynamic dispersion coefficient, X* is dimensionless distance, and x is distance in the direction of flow. The CXTFIT program (ver. 2.0),25 which employs a leastsquares fitting routine, was used to fit effluent breakthrough curve (BTC) data, expressed as relative concentration (C/C0) versus the number of PV applied. To account for nonequilibrium mass transfer processes, the solute transport equations (eq 1) can be reformulated to include a “two-region” model that accounts for physical mass transfer limitations.26,27 In the case of physical nonequilibrium, due to diffusional mass transfer of a nonreactive solute between regions of mobile and immobile water, the following dimensionless terms β and ω, are incorporated into the analytical solution for eq 1:
■
β=
MATERIALS AND METHODS Disk Fabrication and Pore Size Distribution. Ceramic disks (∼10 cm diameter, 1.5 cm thick) were manufactured by Advanced Ceramics Manufacturing (ACM, Tucson, AZ) following established protocols. Clay was imported from a Tanzanian factory and pine sawdust was purchased from Pallet Recyclers (Tucson, AZ). Sawdust was sieved between US no. 30 and no. 60 mesh (0.595 and 0.251 mm openings, respectively) sieves before mixing with clay. The mass ratio of sawdust to clay was ∼4%, which provided an initial flow rate equivalent to 2−3 L/h in a full-sized filter. Disks were formed using a hydraulic press and fired to a temperature of 900 °C. Porosity and pore size distribution of fragments from two intact disks were measured in duplicate using mercury intrusion porosimetry (Micromeritics Instrument Corp., Norcross, GA). X-ray diffraction measurements (H&M Analytical, Allentown, NJ) were used to determine the primary mineral phases present in the ceramic material. Upon receipt from ACM, disks were reduced to 5.1 cm in diameter to fit in existing holders using a hole saw blade.3 Before use in experiments, disks were sterilized by heating at 550 °C for 30 min. After cooling, disks were placed in modified polycarbonate membrane holders (model 420400, Whatman,
ω=
θw,m + fρb KD θw + ρb KD
α1L θw,mvp,m
(4)
(5)
In eq 4, θw,m and θw represent the mobile and total volumetric water content, respectively, f refers to the fraction of solid phase in contact with mobile water, and KD is the distribution coefficient. For a nonreactive tracer, the value of KD = 0, and β reduces to θw,m/θw, or simply the fraction of mobile water. Thus, when the value of β approaches 1 all of the pore water is considered to be “mobile”, whereas, when the value of β approaches 0 all of the pore water is considered to be “immobile.” In eq 5, α1 represents the first-order mass transfer rate between mobile and immobile water, and vp,m is the velocity of the mobile pore water. Cation Exchange Capacity. The CEC of the disk material was determined using both crushed and intact disks. CEC of crushed samples was determined by a Ca/Mg batch exchange method using the methods of Pennell et al.28 CEC of the intact disk was measured by first flushing the filter disk with 25 PV of 1 M Ca(NO3)2 to saturate exchange sites with Ca2+. The disk was rinsed with 10 PV of ethanol to remove excess salt, and 8516
DOI: 10.1021/acs.est.5b01428 Environ. Sci. Technol. 2015, 49, 8515−8522
Article
Environmental Science & Technology then flushed with 60 PV of 0.5 M Mg(NO3)2 to exchange and recover Ca2+. In addition, the capacity of an intact filter disk to exchange Ag+ was determined by introducing 25 PV of 1 M AgNO3, followed by a DI water rinse (AgNO3 is reduced by ethanol to produce precipitates) and 60 PV of 0.5 M Mg(NO3)2. Concentrations of Mg2+ and Ca2+ in aqueous samples obtained from the CEC experiments were analyzed by inductively coupled plasma-optical emission spectroscopy (ICP-OES) (7300 DV, PerkinElmer, Waltham, MA) in axial view mode. Samples were introduced without further modification through a cross-flow nebulizer at 0.5 mL/min, and Mg and Ca were detected at wavelengths of 279.07 and 317.93 nm, respectively. Silver concentrations were determined by ICP-OES at 328.068 nm. Silver Application and Release from Filter Disks. A 200 mg/L (total Ag) silver solution or suspension was prepared as follows: (a) AgNO3 (AlfaAesar, Ward Hill, MA) or caseincoated nAg (Argenol Laboratories, Spain) was diluted in DI water; (b) adjusted to pH 7.00 ± 0.05 using dilute HNO3 or NaOH; and (c) 3 mL was brushed on each of the top and bottom disk surfaces using a paintbrush with Teflon-coated bristles (Shur-Line, Mooresville, NC). This application method results in the currently recommended silver application of 0.03 mg/g of the total ceramic pot weight.2 Disks were dried at room temperature for 12 h and sealed into holders as described previously. The mean diameter of nAg in the painting solution as determined by dynamic light scattering (see Supporting Information) was ca. 50 nm, in agreement with prior transmission electron microscopy (TEM) images of caseincoated silver nanoparticles.29 Scanning electron microscopy (SEM) imaging was performed to visualize nAg deposition in the ceramic filters following painting. A filter disk was painted with nAg suspension (0.03 mg/g), allowed to dry for 24 h, and then broken into ca. 1 cm2 fragments for SEM analysis. SEM images were collected using a Supra 55VP (Zeiss, Germany) with an SE2 detector at an accelerating voltage of 3 kV. Nine separate disk experiments (Table 1) were conducted to assess the impact of influent water chemistry on silver dissolution and transport in filter disks with AgNO3 or casein-nAg applied to the top and bottom surfaces. Each silver release experiment began with 12−15 h (36−45 PV) of flushing with a pH 7 10 mM NaNO 3 background solution, representative of the ionic strength and pH of a typical groundwater,30 followed by the introduction of aqueous solutions with different pH (5, 7, 9), IS (1, 10, 50 mM), and cation species (Na+, Ca2+, Mg2+). These conditions were selected to investigate the effects of specific changes in solution chemistry on silver release from the ceramic filters for a range of source waters that may be used with filters. Filter effluent samples were collected continuously for 30−60 min (17.4 mL34.8 mL or 1.5−3 PV per sample) and analyzed for total Ag (nAg + Ag+) and dissolved Ag (Ag+). Ag+ was separated from nAg by centrifugation at 2500g in ultrafiltration units (Amicon 3k, Millipore, Billerica, MA). Silver concentrations were analyzed by graphite furnace-atomic absorption spectroscopy (GF-AAS) (ThermoScientific), with an Ag+ detection limit of 0.8 μg/L.31 Longevity Analysis. Time to silver depletion was estimated based on the silver mass remaining in the disk (mass applied less cumulative effluent mass) after the initial flushing period (∼30 PV) assuming a constant flow rate of 0.6
Table 1. Experimental Parameters for Silver Release Experiments and Estimated Duration of Silver Retention application AgNO3both sides nAg - both sides AgNO3both sides nAg - both sides nAg - both sides AgNO3both sides nAg - both sides nAg - both sides AgNO3top surface nAg - top surface
variable of interest
sequence
silver retention (years)
baseline
10 mM NaNO3 (pH 7)
0.58 ± 0.05
baseline
10 mM NaNO3 (pH 7)
1.1 ± 0.2
ionic strength ionic strength ionic strength pH
3.4 (1 mM)
cation species baseline
10 mM→ 1 mM → 10 mM NaNO3 (pH 7) 10 mM → 1 mM → 10 mM NaNO3 (pH 7) 10 mM→ 1 mM → 10 mM→ 50 mM → 10 mM NaNO3 pH 7 → pH 5 → pH 7 → pH 9 → pH 7 pH 7 → pH 5 → pH 7 → pH 9 → pH 7 Na+ → Mg2+ → Na+ → Ca2+ → Na+ 10 mM NaNO3
baseline
10 mM NaNO3
10.4
pH
3.5 (1 mM) 0.04 (50 mM) 0.07 (pH 5)/ 5.6 (pH 9) 0.12 (pH 5)/ 7.5 (pH 9) 0.22 (Ca)/ 0.16 (Mg) 1.3
mL/min under two conditions; (a) baseline case of pH 7 10 mM NaNO3 and (b) worst or best case concentration for each chemical condition (e.g., peak concentration at pH 7 50 mM NaNO3 or minimum concentration at pH 9 10 mM NaNO3). The volume of water required to reach silver depletion in filter disks was converted to full-sized filter volume (10 L capacity) by scaling with the cross-sectional area exposed to flow and equating the PV (∼12 mL in disk and ∼1.2 L in pot). The total volume of water was converted to time (years) assuming a constant flow rate of 3.6 L/h and 1 full filter/day. Details of longevity estimates are provided in Supporting Information.
■
RESULTS AND DISCUSSION Ceramic Filter Characterization. The average porosity of the intact filter disk material measured by mercury porosimetry was 41.9 ± 0.8%. The ceramic material contained a wide range of pore sizes, with approximately 35% of pore diameters 30 μm. To evaluate the potential effects of this pore size distribution on water flow through the disks, nonreactive tracer tests were performed on each disk prior to application of nAg or AgNO3. Effluent BTCs obtained from 0.1 PV and 3.7 PV pulse injections of tracer are shown in Figure 1. For the smaller pulse, the maximum relative concentrations (C/C0) approached 0.11 and the BTC was asymmetrical (Figure 1A), consistent with nonreactive tracer data reported by Oyanedel-Craver and Smith3 and Kallman et al.,29 who investigated E. coli transport in ceramic disks. Nonreactive tracer BTCs obtained for a 3.7 PV pulse injection for three separate disks (10, 12, and 13) (Figure 1B) all exhibited early breakthrough of bromide, indicating that the solute did not fully interrogate the pore volume. In addition, all of the BTCs exhibited tailing, which was particularly pronounced for disks 10 and 12. Mass recovery ranged from 98 to 100% in these experiments, indicating that exchange of nonreactive solute between regions of mobile− immobile water was reversible, even in disks that exhibited prolonged tailing. 8517
DOI: 10.1021/acs.est.5b01428 Environ. Sci. Technol. 2015, 49, 8515−8522
Article
Environmental Science & Technology
Figure 1. Representative tracer testing breakthrough curves with equilibrium and nonequilibrium solute transport model fits at (a) 0.09 PV and (b) 3.7 PV.
Figure 2. Exchange of (a) Mg for Ca in a Ca-saturated disk and (b) Ag for Ca in an Ag-saturated disk.
important role in Ag+ transport and release. Based on a Ca2+/ Mg2+ batch exchange method, the CEC of the crushed ceramic filter material was determined to be 5.2 ± 0.4 mequiv/100 g, consistent with CEC values reported for representative kaolinite clay minerals (2.4−4.2 mequiv/100g)28 and natural soils (2−50 mequiv/100 g).32 The CEC of intact ceramic disks, determined by saturation and extraction of exchanged Ca2+, was 4.2 mequiv/100g (Figure 2A). In these experiments, the introduction of 0.5 M Mg(NO3)2, readily displaced Ca2+ from exchange sites, and effluent Mg2+ approached the applied concentration (0.5 M) within 4 PV. The larger CEC value determined in batch exchange relative to intact filter experiments was anticipated due to the greater available surface area of the crushed material and greater mixing in batch reactors compared to the intact disks. To directly assess the capacity of the filter disks to exchange Ag+, an additional CEC experiment was performed by saturating an intact disk with 1 M Ag(NO)3. Following introduction of 0.5 M Mg(NO3)2, Ag+ concentrations reached 45 g/L in the first PV and then rapidly declined to nondetectable levels after 4 PV (Figure 2B). The resulting CEC of the filter disk was 11 mequiv/100g, which corresponds to a maximum Ag+ adsorption capacity of approximately 1.20 mg/g. Thus, CEC values obtained by multiple methods yielded CEC values ranging from 4 to 11 mequiv/100g, and indicate
Fitting the transport model (eq 2), assuming local equilibrium conditions and no solute sorption (RF = 1.0), to the 0.1 PV tracer BTC (Figure 1B, eq fit) yielded a Peclet number of 15.5, which corresponds to a hydrodynamic dispersion coefficient (DH) of 0.42 cm2/h. Fitting the “tworegion” physical nonequilibrium model (Figure 1B, Neq fit) resulted in a slightly improved fit with a β value of 0.22, indicating that 22% of the pore water was “mobile”. The fitted value of ω was 29.7, which corresponds to a mass transfer rate (α1) of 36.0 L/h, while the fitted Peclet number was 21.9 (DH = 1.32 cm2/h). The two-region model was able to capture the effluent BTC for disk 13 (Figure 2B), which yielded a β value of 0.89 (i.e., 11% of the pore water consisted of immobile water), an ω value of 0.015 (α1 = 0.015 1/h), and a Peclet number of 18.7 (DH = 0.36 cm2/h). However, tracer BTCs for disks 10 and 12 exhibited prolonged tailing, which could not be captured by the two-region model. These findings indicate that the ceramic disks contain variable pore structures with regions of immobile water. The observed differences among tracer BTCs obtained for filter disks produced from the same batch suggest a relatively high degree of variability in the pore structure of the manufactured filter material. Cation Exchange Capacity. Due to the high clay content of the filter material, cation exchange reactions could play an 8518
DOI: 10.1021/acs.est.5b01428 Environ. Sci. Technol. 2015, 49, 8515−8522
Article
Environmental Science & Technology that the ceramic filter has a substantial capacity to exchange cations. Silver Release from Ceramic Filters. Following the introduction of a 10 mM NaNO3 aqueous solution at pH 7, effluent silver concentrations from disks painted on both the top and bottom surfaces with AgNO3 or nAg reached maximum values of ∼1 mg/L (Figure 3A) and 0.4 mg/L
Figure 3. Silver release from duplicate disks (a) painted with AgNO3 and (b) painted with nAg exposed to pH 7 10 mM NaNO3. Horizontal dashed line indicates international drinking water standard. Total (nAg + Ag+) and dissolved (Ag+) silver release data are shown for nAg-painted disks.
(Figure 3B), respectively, within 3 h (5 PV). Silver levels fell below the drinking water standard (0.1 mg/L) after 10−15 h (30−45 PV) and 3−4 h (18−24 PV) in AgNO3- and nAgpainted disks, respectively. Silver release from disks painted with AgNO3 was higher than nAg, in general agreement with the findings of Rayner11 who reported 20% greater release of silver from AgNO3-painted disks during the first 24 h (72 PV) of throughput with a phosphate buffer solution (pH 7.4, 16.3 mM). In nAg-painted disks (Figure 3B), differences between total (nAg + Ag+) and dissolved (Ag+) silver release were minimal (90% of silver eluted from the disks in dissolved form rather than the nanoparticle (nAg) form. The greatest amount of nAg release occurred within the first 5 h (15 PV) of flushing, and represented less than 1% of total silver mass released. Silver elution profiles obtained from duplicate nAg- and AgNO3-painted disks were similar (Figure 3), which indicates that variability observed in the physical pore structure of the disks (e.g., tracer tests, SEM images), did not strongly influence silver transport and release from the filter disks. Effects of Solution Chemistry on Silver Release. To evaluate the effects of source water chemistry on silver elution from disks painted with either AgNO3 or nAg, the IS of influent solution was maintained at 1, 10, or 50 mM (as NaNO3) at pH 7. When the IS of the influent solution was reduced from 10 mM to 1 mM NaNO3 for 4.5 h (14 PV), a nearly 10-fold decrease in effluent silver concentrations was observed (Figure 4A). When IS was subsequently increased to 10 mM NaNO3,
Figure 4. Silver release from disks (a) painted with nAg or AgNO3 exposed to ionic strength reduction (10 to 1 to 10 mMNaNO3, pH 7); (b) painted with nAg exposed to sequential changes in influent water ionic strength (10 to 1 to 10 to 50 to 10 mMNaNO3, pH 7); (c) painted with nAg exposed to varying divalent and monovalent cation species (Na+ to Ca2+ to Na+ to Mg2+ to Na+) at pH 7 10 mM; and (d) painted with AgNO3 and nAg exposed to variable influent water pH (7 to 9 to 7 to 4 to 7) at 10 mMNaNO3. Horizontal dashed line indicates international drinking water standard. Total (nAg + Ag+) and dissolved (Ag+) silver release data are shown for nAg-painted disks.
effluent silver concentrations rapidly rebounded to previous levels. Similar trends in silver elution were observed for an nAgpainted disk exposed to sequential changes in IS (10 mM to 1 mM to 10 mM to 50 mM to 10 mM NaNO3) (Figure 4B). When the IS was increased from 10 to 50 mM NaNO3 for 4 h (12 PV), total silver concentrations in the effluent reached levels as high as 0.8 mg/L. Small amounts (0.002−0.005 mg/L) of nAg detachment from nAg-painted disks were observed when IS was decreased to 1 mM. This response to changes in IS is consistent with DLVO interaction energy profiles calculated using the approach of Guzman et al.,30 which show an increasing primary energy barrier and loss of the secondary minimum as the IS was reduced from 50 to 1 nm (Supporting Information, Figure S1). The observed elution of nAg is consistent with prior column transport studies,33−35 where negatively charged colloids and nanoparticles detached from quartz sands following a reduction 8519
DOI: 10.1021/acs.est.5b01428 Environ. Sci. Technol. 2015, 49, 8515−8522
Article
Environmental Science & Technology
Decreasing the pH from 7 to 5, increasing IS from 10 to 50 mM NaNO3, or changing from a monovalent to divalent cation all resulted in rapid Ag+ exchange, with release governed by competitive exchange of Ag+ for Na+, Ca2+, and Mg2+, and H+. Similarly, when fewer positively charged ions, or ions of lower valence, were present, Ag+ was more strongly retained in the ceramic, resulting in decreased silver elution. Data from all silver release experiments demonstrated that the filter disks responded rapidly (within ∼2 PV) to changes in influent solution chemistry, consistent with results obtained from cation exchange experiments. When influent water was returned to background conditions (pH 7, 10 mMNaNO3), silver effluent levels returned to baseline concentrations, indicating that silver exchange was a reversible process. Effects of One-Sided Application. The standard manufacturing protocol for CWFs involves painting both the internal (upper) and external (lower) surfaces with a silver solution; however, the relative contribution of each painted surface to silver elution is unknown. Thus, a set of experiments was conducted in which only the upper (inlet) surface of the disk was painted with either AgNO3 or casein-nAg. After flushing with 60 pore volumes (20 h) of background solution (10 mM NaNO3 at pH 7) no silver was detected in effluent samples of the disk painted with AgNO3 (Figure 5a). When the IS was increased to 50 mM (NaNO3) to promote Ag+ release, effluent silver concentrations approached a maximum value of 0.03 mg/L.
in IS. Such behavior has been attributed to reduced screening of the electrical double layer of nAg at lower IS, resulting in weaker particle-surface interactions.36 Although moderate nAg release (
