Environ. Sci. Technol. 2010, 44, 7808–7813
Evaluation of Nanocopper Removal and Toxicity in Municipal Wastewaters R A J A G O P A L A N G A N E S H , * ,† JOSH SMERALDI,‡ TURAJ HOSSEINI,‡ LEILA KHATIB,† BETTY H. OLSON,‡ AND DIEGO ROSSO‡ Kennedy/Jenks Consultants, 2355 Main Street, Suite 140, Irvine, California 92614, and Department of Civil and Environmental Engineering, University of California at Irvine, California 92697
Received April 30, 2010. Revised manuscript received August 27, 2010. Accepted September 2, 2010.
Bench scale studies were performed to evaluate removal and toxicity of copper nanoparticles (CuNPs) and copper ions in activated sludge biomass. The data indicated that, under the test conditions, copper nanoparticles were removed more effectively (∼95%) than copper ions (30-70%) from the wastewater. Mechanisms of CuNP removal were further investigated by equilibrating CuNP and copper ion in activated sludge filtrate (0.45 µm). The predominant mechanisms of copper removal appear to be aggregation and settling (CuNP) or precipitation (copper ion) rather than biosorption. Most probable number (MPN) test data indicated that addition of 10 mg/L of copper ion was toxic to both coliform and ammonia oxidizing bacteria in the wastewater while no inhibitory effects were observed with the addition of the same amount of copper nanoparticles. Respirometry data indicated a 55% decrease in respiration rate when 10 mg/L ionic copper was added. However, no significant decrease in respiration rate was observed in the presence of copper nanoparticles. The toxicity of copper to activated sludge microorganisms appears to be a function of the concentration and characteristics of copper remaining in solution/suspension.
Introduction Use of manufactured nanomaterials (1-100 nm in at least one dimension) in everyday products is increasing exponentially (1). While the source elements are often the same as the ions already used in commercial products, nanomaterials are highly reactive and often differ in many physical and chemical characteristics compared to their ionic counterparts (1, 2). Furthermore, their elevated area per unit volume amplifies their reactive properties (3, 4). These different characteristics make them suitable for improvement or replacement of commercial products and applications (5). With increasing use of nanomaterials, wastewater treatment plants are among the last barriers prior to their environmental release. The fate, transport, and toxicity of nanomaterials in wastewaters may be different than those * Corresponding author phone: (949) 261-1577; fax: (949) 2612134; e-mail:
[email protected]. † Kennedy/Jenks Consultants. ‡ University of California at Irvine. 7808
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 20, 2010
of their ionic counterparts. The presence of nanomaterials in biosolids and model wastewater effluents has already been reported (6, 7). Nanocopper (CuNP) is currently used in a wide range of applications including fungicides, cosmetics, printers, and electronics (5). In the past, several studies have investigated the removal of copper ions in wastewater. A predominant mechanism of copper ion removal is biosorption and sorption to materials such as zeolites or bentonite (8-14). Furthermore, depending on the wastewater characteristics, copper can also be removed by coagulation, chemical precipitation, or ion exchange in wastewaters (14-16). CuNP, being colloidal in nature, may or may not adsorb to biomass as effectively. Rather, nanotitanium dioxide and fullerenes are reported to be removed by aggregation/sedimentation in high ionic strength solutions (17, 18). Copper ion toxicity to wastewater and other microorganisms is well documented. Under the National Toxics Rule (NTR) criteria, the wastewater discharge limit for copper into inland waters is 8.9 µg/L (19). The toxicity effects of CuNP to wastewater microorganisms have not been studied extensively. However, toxicity studies using zebrafish and aquatic invertebrates indicate that CuNP are less toxic than ionic copper in many instances (20). Without comparison to ionic copper, toxicity of CuNP to bacterial enzymes increased with its concentration in suspension. However, CuNP toxicity to C. dubia did not follow any obvious trend (21). Such conflicting data on toxic effects have also been observed with silver and other nanoparticles. For example, in one study ionic silver appeared to inhibit the growth of P. fluorescence more than nanosilver (22). In a different study evaluating nanosilver toxicity, while ionic silver inhibited growth of ammonia oxidizing bacteria (100%) more than nanosilver (mean size ∼14 nm; 54% inhibition), nanosilver inhibited the respiration rate (86%) more than ionic silver (42%) (23). The differences observed in toxicity effects may be due to differences in the properties of nanomaterials used (size, charge density), chemical composition of media (pH, organics, ionic strength), test conditions, and organisms evaluated (22, 24, 25). From a wastewater treatment industry perspective, any of the above factors that favor the removal of nanomaterials by adsorption, precipitation, aggregation, or other means will likely lower the inhibitory effects during the wastewater treatment, but may increase the risk during biosolids management. Also, suspension and subsequent dissolution of nanomaterials may increase or decrease the inhibitory effects depending on the relative toxicity of the two forms. A detailed evaluation of the extent to which CuNPs were removed, characteristics of CuNPs in suspension, and toxicity to major wastewater microorganisms, and a comparison of the above with ionic salts, is currently not available. This study, through bench scale tests, was designed to compare inhibition to wastewater microorganisms (coliform bacteria, ammonia oxidizers) when similar concentrations of CuNPs or ionic copper were released to wastewater. The copper removed and those remaining in suspension were also evaluated to help explain the toxicity data obtained.
Materials and Methods Activated Sludge Samples. Activated sludge for this test was collected from Orange County Sanitation District’s Plant 1 (OCSD, Fountain Valley, CA). At the time of sampling, the activated sludge process was operating in ammonia bypass mode (i.e., carbon oxidation only) with approximately 1.2 day sludge retention time with average mixed liquor sus10.1021/es101355k
2010 American Chemical Society
Published on Web 09/20/2010
pended solids (MLSS) of 650 mg/L. The copper removal studies, described below, were designed to generally reflect the OCSD’s operating conditions. Please see Supporting Information (SI) for available activated sludge water quality data. Copper Nanoparticles. CuNPs were obtained from QSI Company, Santa Ana, CA. The CuNP obtained from the vendor is a mixture of copper and copper oxide particles with a copper content of approximately 51%. CuNPs were further characterized at the Nanotech Analytics Lab, NEI Corporation, Somerset, NJ. Primary particle size analyses were performed by scanning electron microscopy. Specific surface area was measured with the Brunauer, Emmett, and Teller method. Particle density was measured using a Quantachrome Autotap (Syossett, NY). Stock CuNP suspensions (200 mg/L of Cu) were prepared by adding the appropriate amount of copper nanoparticles to DI water (preadjusted to pH 8). The suspension was then well mixed and sonicated (VWR 75T Aquasonic sonicator) for 1 h prior to copper removal studies. Particle size (hydrodynamic diameter) distribution in the suspensions was measured using Zeta Sizer Nano (Malvern Instruments). Chemical compositions of the powder as well as the stock suspension were determined using X-ray diffraction analyses. Details of the analyses are provided in SI. Copper Removal Studies. Two sets of experiments were performed to evaluate copper nanoparticles and copper ion removal. First, studies were performed in the presence of activated sludge microorganisms. Flasks (250 mL) containing 100 mL of activated sludge (MLSS ∼ 650 mg/L, as in OCSD treatment plant) were spiked with 2-10 mg/L of copper, added as CuNPs or copper ions (using CuCl2 salt). The flasks were equilibrated at 25 °C, at100 rpm using a Gyratory Water Bath shaker. An equilibration time of 20 h (rather than typical 3-4 h) was chosen to maintain a time frame comparable to the sludge retention time of approximately 28 h at OCSD. The samples were then filtered using 0.45 µm filters, and the filtrates were analyzed for copper concentrations. The pH of the activated sludge was approximately 7.6. Addition of copper nanoparticles or copper ions altered the pH within 0.2 units. No further pH adjustments were made. Currently no data are available on current or projected concentrations of CuNPs in actual wastewaters. A maximum concentration of 10 mg/L was selected for this study since it is unlikely that concentrations higher than this will enter wastewater treatment plants. In the second set of tests, activated sludge filtrates (0.45 µm), rather than activated sludge containing biomass, were spiked with copper nanoparticles or copper ions and equilibrated for 20 h. The objective of this set of experiments was to identify the fraction of copper removed without biosorption. These mechanisms may include chemical precipitation or physical aggregation and settling. After equilibration, the supernatant samples were again filtered (0.45 µm), prior to analyses. Most Probable Number (MPN) Tests. MPN tests were performed to evaluate whether the presence of nanomaterials affected the growth of coliform and ammonia oxidizing populations in OCSD wastewater. Series of dilution in the respective growth media were made using activated sludge exposed to 10 mg/L CuNP or copper ion, and triplicate samples were incubated at 25 °C for MPN estimates (26, 27). Additional details are provided in SI. Evaluation of Respiration Rates. Specific oxygen uptake rate was measured using a PF-8000 aerobic/anaerobic respirometer from Respirometer Systems and Applications, LLC (Springdale, AR). Activated sludge exposed to 10 mg/L of Cu from CuNPs or ionic copper were used. Briefly, 500 mL of samples spiked with 10 mg/L nano or ionic copper were placed in serum bottles with an integrated carbon dioxide
TABLE 1. Reported Properties of Copper Nanoparticles Used in This Study property a
primary size (nm) surface areab (m2/g) tap densityd (g/cm3) particle size distribution in suspensione (nm) average particle diameter in suspensione (nm)
value 50-100 14.27 ( 0.43c 1.65 100-200 125
a Measured by SEM. b BET method. c Mean ( range. Measured using Quantachrome Autotap. e Zeta Sizer Nano (Malvern Instruments).
d
trap. Control tests were also performed where no copper was spiked to the activated sludge samples. A magnetic stir bar was placed in each bottle which was then sealed with a cap and integrated septum and placed in a water bath at 25 °C. The serum bottle was connected to a pressure sensing system via a needle through the septum and tubing. The pressure sensor continually monitored pressure changes and controlled the release of necessary oxygen to the bottle as needed while also recording the oxygen uptake in the bottles. Copper Analyses. Copper ions as well as CuNPs in the filtrates were analyzed by ICP (EPA Method 200.7), by an external, state certified laboratory. The samples were digested using EPA Method 200.2 prior to ICP analyses. Scanning Electron Microscopy of CuNPs in Wastewater. SEM analyses were performed on CuNPs and ionic copper (10 mg/L as Cu) suspended in wastewater filtrate using a Ziess EVO scanning electron microscope at 10 kV voltage. Samples were collected immediately after addition and after 4 h of incubation, filtered using 0.2 µm nuclepore filters for the analyses. Additional details are in SI. Analyses of Submicrometer Particles. A Zatasizer Nano (Malvern Instruments, Westborough, MA) was used to analyze submicrometer/nanoscale particle size distribution and particle count in the filtrate (0.45 µm) samples. The instrument uses a dynamic light scattering (DLS) technology to measure particle distribution in the range 0.6 nm to 6 µm. Statistical analyses performed using the photon count rate data indicated that this data can be a useful technique to qualitatively measure relative nanoparticle removal in wastewaters (28). Hence, the photon count rate measured during particle analyses was used to determine relative particle count in various samples. The submicrometer particles measured during the filtrate analyses include naturally occurring submicrometer particles in the wastewater, CuNPs, colloidal fraction of copper ions.
Results CuNP Particle and Stock Suspension Characterization. Table 1 summarizes data from CuNP characterization. X-ray diffraction analyses indicated the presence of copper and copper oxide in the CuNP particles received from the vendor. The predominant copper oxide phases are CuO and Cu2+1O. During preparation of stock suspensions the dark brown elemental CuNPs oxidized to copper oxides within 30 min of sonication, as indicated by the green color and X-ray diffraction analyses (SI). Analyses of these suspensions by Zeta Sizer indicated that the particles aggregated to a size (measured as hydrodynamic diameter) distribution of 100-200 nm with an average particle size of approximately 125 nm. Removal of Copper Ions and CuNPs in Activated Sludge. Figure 1 shows the percent removal of copper ion and CuNPs in activated sludge samples. Approximately 35-70% of copper ions were removed in the presence of VOL. 44, NO. 20, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7809
FIGURE 1. Percent removal of copper nanoparticles and copper ions in OCSD activated sludge samples. Activated sludge samples from OCSD wastewater treatment plant were spiked with 2-10 mg/L copper using copper ions (copper chloride) or CuNPs. After 20 h of incubation, filtrates (0.45 µm) were analyzed to estimate percent copper removed. biomass. The percent removal increased with the initial copper concentration. CuNPs were removed more effectively than copper ions in this test. More than 90% of the CuNPs were removed in all of the flasks. Residual copper concentrations in the ionic copper spiked samples varied from approximately 1.2 to 3.5 mg/L, and those in the CuNPs spiked samples were less than 0.5 mg/L. Copper Removal in Activated Sludge Filtrate. Table 2 shows the copper levels in the filtered supernatant after equilibration for 20 h with activated sludge filtrate. Relevant data from unfiltered activated sludge studies are also shown for comparison. A significant amount of copper (ion or nano) was removed in the samples containing no biomass. Nearly 25-55% of copper ions and 75-80% of CuNPs were removed in the biomass-free solution. The copper removed in the biomass-free filtrate was nearly 65-85% of copper removed in the presence of activated sludge biomass. This indicated that only about 15-35% of copper removed in the presence of biomass was due to adsorption of copper to activated sludge biomass. Remaining copper was probably removed due to other mechanisms such as precipitation or aggregation and settling. Even in the absence of biomass, the CuNPs were removed more effectively than ionic copper. Nearly 50% more CuNPs than copper ions were removed in the activated sludge filtrate. SEM Analyses of Copper Removed. Figure 2 shows the picture of CuNPs and ionic copper immediately after suspension and after 4 h of incubation in the filtrate. The CuNP and ionic copper particles showed distinctly different morphologies in the two samples. The CuNPs appeared to have transformed to larger size aggregates (>1 µm) during their removal. Ionic copper precipitated from the wastewater as finer solids, completely covering the filter surface. These observations support the hypothesis that the mechanisms governing the removal of CuNP and ionic copper are different, and hence, the extent of copper removed may also be different. DLS Analyses. The DLS analyses were performed to evaluate possible differences in the size distribution and count rate of the copper remaining in the filtrate samples. The data from the analyses showed different trends for CuNP and ionic copper spiked samples. The control samples (no copper spike) had the lowest residual copper concentration, average particle size, and particle count (Table 3). For the same amount of initial copper added, the copper ion spiked samples had higher residual copper than those in CuNP spiked samples. The average nanoscale particle size as well 7810
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 20, 2010
as nanoscale particle count increased with increasing residual copper concentration in the copper ion spiked samples (Table 3, Figure 3). The average particle size (hydrodynamic diameter) increased from about 23 nm at a residual copper concentration of 1.5 mg/L to ∼100 nm at a residual concentration of 4 mg/L. In the CuNP suspended samples, the average nanoscale particle size did not vary with the residual copper level. It was approximately 100 nm at residual copper levels of 1 or 2 mg/L. The particle count rate, however, increased with residual copper level. The nanoparticle size as well as count in samples containing lower residual CuNPs (∼1.1 mg/L) were higher than samples containing larger (∼2 mg/L) ionic copper levels. This suggested the possibility that the residual CuNPs in suspension did not completely dissolve. Most Probable Number (MPN) Tests for Coliform and Ammonia Oxidizing Bacteria. Coliform bacteria, which are more prevalent in wastewater and other aquatic environments, and ammonia oxidizing bacteria, that are often more sensitive to upsets during wastewater treatment, were chosen for toxicity analyses (29). Table 4 shows the MPN values obtained using nano and ionic copper. Significant growth of coliform bacteria was observed within 24 h in control and copper nanoparticle spiked samples. Although MPNs in CuNP samples were higher than those for control samples, these were not statistically different at 95% confidence interval (27). These variations are not uncommon for MPN tests due to various intrinsic and extrinsic variables such as clumping of cells and cell damage (30, 31). However, the MPNs for ionic copper suspended samples were significantly lower than those of the control and CuNP suspended samples (at 95% confidence interval), indicating toxicity of ionic copper to coliform bacteria under the test conditions. Similar trends were found with MPN data for ammonia oxidizing bacteria. Since the initial ammonia oxidizing bacterial population was low in the OCSD wastewater, nearly 11 days of incubation were required to grow ammonia oxidizing bacteria. As observed with the coliform bacteria, compared to control samples, the bacterial growth in the samples spiked with CuNP did not vary significantly. However, the growth was significantly lower in the copper ion spiked samples. Respirometer Tests. The respiration (oxygen uptake) rate measurements can provide an estimate of short-term toxicity of nanomaterials to microorganisms (23). The respiration rate trends using CuNPs and ionic copper, in general, were consistent with the MPN toxicity test data. The respiration rates for activated sludge microorganisms in the CuNP spiked samples were similar to that of the control sample (Figure 4). This indicated that addition of CuNPs did not induce significant short-term toxicity to microorganisms. However, the respiration rate for copper ion suspended samples was always lower than that of the control and CuNP suspended samples (reduced slope for copper ions in Figure 4). After 10 h of incubation the respiration rate for the copper ion spiked samples was approximately 55% of the control and CuNP spiked samples. These data, again, suggested that, under the test conditions, ionic copper remaining solution was more responsible for toxicity to activated sludge microorganisms than the CuNP (which were mostly removed from suspension).
Discussion Removal of CuNP in Wastewater. More copper was removed in CuNP added samples than in ionic copper added samples. Less than 20% of the copper was removed through adsorption or other biomass mediated processes in either samples. Removal of nanomaterials in aqueous solutions is dictated by chemical composition of the water and physical/chemical properties of the nanomaterials (17, 18, 24). In an aqueous environment, ionic strength appears to facilitate aggregation
TABLE 2. Estimate of Copper Removed by Adsorption and Other Meansa residual copper (mg/L) copper
copper added
when equilibrated with biomass
when equilibrated in the filtrate
copper removed by biomass (mg/L)b,c
copper removed by other means (mg/L)d
copper ion
2 10 3.52 8.8
1.22 3.5 0.21 0.4
1.5 4.5 1 2
0.28 (14%) 1 (10%) 0.79 (19.8%) 1.6 (16%)
0.5 (25%) 5.5 (55%) 2.52 (72%) 6.8 (77%)
CuNP
a 25% of samples were randomly selected for replicate analyses. The replicate data varied from 0 to 6.5%. paranthesis indicate % of copper removed. c Column (4) - column (3). d Column (2) - column (4).
b
Values in
FIGURE 2. SEM of (a) CuNPs and (b) ionic copper removed by filtration (a1, b1) immediately after addition to and (a2, b2) after 4 h of incubation. Activated sludge filtrate (0.45 µm) was spiked with 10 mg/L of CuNP or ionic copper (copper chloride) and refiltered for this analyses.
TABLE 3. Residual Copper Levels and Submicrometer Particle Characteristics after Equilibration in Biomass-Free Filtrates
sample control copper ion CuNP
initial conc (mg/L)
final conc (mg/L)
average sizeb (nm)
particle countc (kCPS)
a 2 10 4 10
0.18 1.5 4.1 1.1 2
28 23 105.7 98 105.7
158.5 ((0.3) 182 ((3.8) 283 ((3.25) 207 ((0.4) 341 ((0.9)
a Typical residual copper concentration in OCSD filtrate is less than 0.2 mg/L. b Size measured as hydrodynamic diameter. c Values in parentheses show range from duplicate samples.
(through surface charge screening and reducing energy barrier) and settling of metal oxide nanomaterials, whereas the organic content appears to stabilize the nanomaterials in suspension (21, 24). Furthermore, the extent of aggregation and dispersion varies with isoelectric point and other physical/chemical characteristics of the nanomaterials. For example, Gao et al. (21) evaluated dispersion and toxicity of CuNP in three different Suwannee River water samples (headwaters, river midsection, and off-delta). Their study showed that aggregation and settling of CuNPs were lower in headwaters which had higher organic content and low ionic strength (DOC 45.7 mg C/L; I 0.94 mM and residual
FIGURE 3. Residual copper concentrations and particle count rates in wastewater filtrates after spiking with copper ion or CuNPs. Samples were incubated for 20 h and refiltered for analyses. Residual copper concentrations in the filtrate were measured by the ICP method. Particle count rates were measured using the DLS Method (Zeta Sizer Nano, Malvern Instruments). Count rates in control samples are shown for comparison. CuNP 12.58 mg/L). Removal of CuNP increased with an increase in ionic strength and decrease in organic content (DOC 10.18 mg C/L, I 3.34 mM, residual CuNP 1.45 mg/L for VOL. 44, NO. 20, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7811
TABLE 4. MPN Test Data for OCSD Activated Sludge Microorganisms in CuNP and Ionic Copper Suspended Samplesa
sample control copper ion CuNP
coliform bacteria MPNb (av/mL) 930 000 (230 000-3 800 000)