Research Aqueous Stability of Gadolinium in Surface Waters Receiving Sewage Treatment Plant Effluent, Boulder Creek, Colorado P H I L I P L . V E R P L A N C K , * ,† HOWARD E. TAYLOR,‡ D. KIRK NORDSTROM,‡ AND LARRY B. BARBER‡ U.S. Geological Survey, Box 25046, MS973, Denver, Colorado 80225, and U.S. Geological Survey, 3215 Marine Street, Boulder, Colorado 80303
In many surface waters, sewage treatment plant (STP) effluent is a substantial source of both regulated and unregulated contaminants, including a suite of complex organic compounds derived from household chemicals, pharmaceuticals, and industrial and medical byproducts. In addition, STP effluents in some urban areas have also been shown to have a positive gadolinium (Gd) anomaly in the rare earth element (REE) pattern, with the Gd derived from its use in medical facilities. REE concentrations are relatively easy to measure compared to many organic wastewater compounds and may provide a more widely utilized tracer of STP effluents. To evaluate whether sewage treatment plant-associated Gd is a useful tracer of treatment plant effluent, an investigation of the occurrence, fate, and transport of rare earth elements was undertaken. The rare earth element patterns of four of five STP effluents sampled display positive Gd anomalies. The one site that did not have a Gd anomaly serves a small community, population 1200, with no medical facilities. Biosolids from a large metropolitan STP are not enriched in Gd even though the effluent is, suggesting that a substantial fraction of Gd remains in the aqueous phase through routine treatment plant operation. To evaluate whether STP-derived Gd persists in the fluvial environment, a 14km study reach downstream of an STP was sampled. Gadolinium anomalies were present at all five downstream sites, but the magnitude of the anomaly decreased. Effluent from STPs is a complex mixture of organic and inorganic constituents, and to better understand the chemical interactions and their effect on REEs, the aqueous speciation was modeled using comprehensive chemical analyses of water samples collected downstream of STP input. These calculations suggest that the REEs will likely remain dissolved because phosphate and carbonate complexes dominate over free REE ions. This study supports the application of Gd anomalies as a useful tracer of urban wastewater.
* Corresponding author phone: (303) 236-1902; fax: (303) 2363200; e-mail:
[email protected]. † U.S. Geological Survey, Denver, CO. ‡ U.S. Geological Survey, Boulder, CO. 10.1021/es048456u Not subject to U.S. Copyright. Publ. 2005 Am. Chem. Soc. Published on Web 08/09/2005
Introduction Due to the Oddo-Harkins effect, Gd (atomic number 64) is naturally enriched compared to its nearest neighbors Eu (atomic number 63) and Tb (atomic number 65), because during solar system evolution even-atomic-number elements were more stable than neighboring odd-number elements. By normalizing the rare earth element (REE) concentrations to a reference material, this effect is removed. Urban rivers that receive treated effluent from a sewage treatment plant (STP) have been shown to have unique REE patterns. Bau and Dulski (1) first reported positive Gd anomalies in shalenormalized REE patterns in waters draining urban and industrial areas in Germany and Sweden. Because of its unique paramagnetic properties, Gd has a variety of industrial applications. Gadolinium is one of the most magnetic elements and has the highest neutron-absorbing ability (4.9 × 104 barns) of all the elements. Industrial uses include phosphors in television tubes, gadolinium-yttrium garnets in microwave applications, an alloy, a superconductor, and a contrasting agent in magnetic resonance imaging (MRI). Bau and Dulski (1) concluded that the Gd enrichment results from the use of gadopentetic acid [Gd(DTPA)2-, where DPTA ) diethylenetriaminepentaacetate] as a contrasting agent for MRIs. During the past 10 years, studies in Italy, Japan, France, and the Czech Republic have documented positive Gd anomalies in rivers that receive treated effluent (2-5), and this study is the first documentation in the United States (6). Since 1988, Gd organic compounds have been used in the United States as contrasting agents to enhance images in patients undergoing MRIs and are taken either orally or intravenously (7). Gadolinium is an ideal element to use as a contrasting agent because it has a large magnetic moment due to multiple unpaired electrons and, as an organic complex, is quite inert; it is excreted through the kidneys with a half-life of 2 h. Organic Gd compounds are extremely stable, and traditional STP practices may not break them down (4). Studies have shown that urban STP effluents and receiving waters contain a suite of complex organic compounds derived from household chemicals, pharmaceuticals, and industrial and commercial activities (8-12). These organic compounds can be difficult to measure because of their low concentrations and complex extraction procedures. In contrast, REE are relatively easy to measure using an inductively coupled plasma-mass spectrometer, and for many waters, including results presented here, no preconcentration or separation is necessary. In this study we examine the occurrence, fate, and transport of STP-derived REEs upon entering a river system; to use the positive Gd anomaly as a tracer of STP effluent, an understanding of the nature of REEs in fluvial systems is needed. Two conditions need to be met to use REEs as tracers in aquatic systems: (1) an understanding of the sources of REEs to rivers and (2) an understanding of the processes that control their fate and transport. Solution chemistry likely plays an important role in REE behavior, and since sewage treatment plant effluent is a complex mixture of organic and inorganic constituents, comprehensive water analyses were performed to describe the solution and to identify chemical interactions. Our primary study site, a 14-km reach of Boulder Creek, CO, is ideally suited for this investigation because of the following: (1) at low flow the STP effluent dominates the discharge, (2) upstream of the STP no other Gd-rich effluent VOL. 39, NO. 18, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6923
FIGURE 1. Map depicting sampling sites. enters, and (3) in the study reach only one small tributary enters. To further investigate the role of solution chemistry on the fate of REEs, aqueous speciation was examined using the analyzed concentrations of the field samples.
Methods Sampling. This study was part of a larger investigation of the Boulder Creek watershed (13), and detailed collection and analytical procedures are presented in Verplanck et al. (6) and Barber et al. (14). To evaluate the fate of STP-derived Gd in a river system, a 14-km reach of Boulder Creek was sampled (Figure 1). In the study reach, the creek is low-gradient, flows through grass lands, and is partially shaded with willows and cottonwoods. During late summer, much of the upstream water is diverted for irrigation purposes such that the STP effluent dominates the flow and chemistry. This reach was chosen for study because there are no upstream inputs of Gd and within the reach only one small tributary enters Boulder Creek. Most of the land adjacent to the study reach is either privately owned conservation land or part of Boulder County open space parkland so that non STP-derived Gd sources are unlikely. Within the study reach some of the water is removed for agricultural use and groundwater enters the creek (13). Seven sites sampled include a site 60 m upstream of the outflow from the Boulder STP (BC-BKGD), the STP effluent (BLD-EFF), and sites 0.5-, 4.2-, 5.6-, 8.1-, and 13.4-km downstream of the STP effluent outflow (BC-0.5, BC-4.2, BC5.6, BC-8.1, and BC-13.4). The site at 0.5 km was chosen to allow for mixing of the effluent with Boulder Creek water. Sampling occurred in October 2000 during low-flow conditions. Samples were collected from upstream sites to downstream sites to try to sample the same parcel of water. The one exception was that the most downstream site was sampled at the end of the day prior to sampling the other sites. Discharge from the STP varies throughout the day because of community water usage, so the relative proportion of stream water and effluent can vary. Prior to sampling, stream-flow velocity was not known precisely; thus, the same parcel of water may not have been sampled at each site. Field measurements at the sampling sites included water temperature, pH, specific conductance, and stream velocity. Water samples were collected using a depth-integrated sampler following the equal-width-increments method (15). Water samples for major, minor, and trace element determinations were filtered with a 142-mm diameter, 0.1-µm pore-size tortuous path, filter membrane. Anion samples were filtered and not acidified, cation samples were filtered and acidified with concentrated HNO3 to a pH of < 2, and samples for iron speciation were filtered and acidified with 6 M HCl to a pH of < 2. In this paper, analyses of filtrate from the 0.1-µm pore-size filter membrane will be termed “dissolved” concentrations. Total recoverable samples were unfiltered aliquots from the same sample-collection bottle as the filtered samples, which were acidified with concentrated nitric acid. Samples for ethylenediaminetetraacetic acid (EDTA) were 6924
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 18, 2005
filtered on site through 0.7-µm glass fiber filters, collected in precleaned amber glass bottles, and preserved with 2% by volume formalin (14). Analytical Method. Concentrations of minor and trace elements, including REEs, were determined by inductively coupled plasma-mass spectrometry using a Perkin-Elmer Elan 6000 instrument with a “cone-spray” pneumatic nebulizer for sample introduction (16). Samples were determined in triplicate using a direct calibration curve quantification technique with 103Rh, 115In, and 193Ir used as internal standards. Calibration standards were gravimetrically prepared from high-purity salts, dissolved in HNO3-acidified deionized water. Corrections for isobaric interferences were made when necessary (17). All REEs with masses greater than Sm were routinely corrected for isobaric interferences because barium and the light REEs (LREEs) are prone to form isobaric oxides during analysis. Two new REE standard reference water samples (PPREE1 and SCREE1) were used to check for accuracy (18). Major cations and silica were determined using a Leeman Labs Direct Reading Echelle inductively coupled plasmaoptical emission spectrometer. Iron redox species were determined using a modification of the FerroZine colorimetric method (19, 20) with a Hewlett-Packard 8453 diode array UV/vis spectrophotometer. Concentrations of major anions were determined by ion chromatography (21) using a Dionex 2010i ion chromatograph. Alkalinity (as HCO3-) was determined using an Orion 960 autotitrator and standardized H2SO4 (22). Data for all samples with complete analyses were checked for charge imbalance using the computer program WATEQ4F (23). A complete tabulation of the analytical results can be found in Verplanck et al. (6). EDTA analytical procedures are presented in Barber et al. (14, 24). In summary, samples were oven-evaporated to dryness, derivatized with acetyl chloride/propanol, and analyzed by gas chromatography/mass spectrometry. Because of limited time, fewer sites were sampled for EDTA and included BC-BKGD, BLD-EFF, BC-0.5, and BC-13.4. Modeling. Rare earth element speciation was calculated with the PHREEQE code using the MINTEQ thermodynamic database and the ion association model (25). The MINTEQ thermodynamic database was updated to include the REEs using data from Millero (26) for Cl- and SO42- complexes, Klungness and Byrne (27) for OH- complexes, Schijf and Byrne (28) for F- species, Lee and Byrne (29, 30) for CO32and PO43- complexes, and Smith and Martell (31) for EDTA. All calculations were performed at 25 °C. Following the approach outlined in Gimeno Serrano et al. (32), activity coefficients were calculated using the Davies (33) approximation.
Results and Discussion Wastewater-Derived Gadolinium. As part of a reconnaissance study, REE concentrations in STP effluent serving areas of varying population sizes were analyzed to determine how common positive Gd anomalies are and in what settings they occur. Rare earth element compositions of effluent from STPs that serve a small community (1200 people) with no medical facilities (NED-EFF); two medium-size cities (110 000 and 170 000 people), one with one MRI facility (BLD-EFF) and one with multiple MRI facilities (TAL-EFF); and a large metropolitan area (1.5 million people) with multiple MRI facilities (DEN-EFF) were determined (Table 1). The North American Shale Composite (NASC; 34, 35)-normalized REE patterns of the three effluents from communities with MRI facilities all display positive Gd anomalies, and the REE pattern of the effluent from the community without any medical facility does not (Figure 2). All four of the communities receive their drinking water from pristine mountain lakes or deep groundwater sources, such that the Gd does not originate from input water but from community usage.
TABLE 1. Site Information and Rare Earth Element and Selected Chemical Concentrations of Effluent and Boulder Creek Water Smaples
VOL. 39, NO. 18, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
sample
TAL-EFF
DEN-EFF
NED-EFF
BLD-EFF
BC-BKGD
BC-0.5
BC-4.2
BC-5.6
BC-8.1
BC-13.4
location population served MRI facilities date sampled pH
Tallahassee, FL 170,000
Denver, CO 1,500,000
Nederland, CO 1,200
Boulder, CO 110,000
Boulder Creek -
Boulder Creek -
Boulder Creek -
Boulder Creek -
Boulder Creek -
Boulder Creek -
4 10/29/2003 7.01
23 6/11/2004 7.09
0 10/9/2000 7.24
1 10/11/2000 7.28
10/11/2000 7.97
10/11/2000 7.28
10/11/2000 7.60
10/11/2000 8.02
10/11/2000 8.34
10/10/2000 9.17
filtration (µm) alkalinity (as mg/L HCO3) mg/L SO4 Cl F P Fe µg/L La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu EDTAa
total -
0.45 -
0.45 154
0.45 119
0.1 89.4
total -
0.1 116
total -
0.1 108
total -
0.1 105
total -
0.1 126
total -
0.1 158
total -
-
-
73 0.1 26 -
41 1.0 70 0.092
14 0.2 24 0 0.041
0.35
35 0.9 62 1.7 0.044
0.22
31 0.7 51 1.4 0.039
0.35
29 0.7 59 1.3 0.048
0.27
28 0.8 64 1.0 0.043
0.19
32 0.9 69 0.8 0.011
0.14
0.011 0.017 0.0024 0.011 0.0029 0.0003 0.12 0.0004 0.0028 0.0008 0.0033 0.0005 0.0044 0.0009 -
0.017 0.035 0.0050 0.024 0.0063 0.0005 0.14 0.0013 0.0092 0.0025 0.012 0.0021 0.021 0.0045 -
0.0095 0.019 0.0028 0.012 0.0029 0.0005 0.0023 0.0002 0.0012 0.0005 0.0014 0.0001 0.0017 0.0003 4.4
0.0057 0.010 0.0014 0.0058 0.0010 0.0004 0.068 0.0002 0.0028 0.0008 0.0038 0.0005 0.0040 0.0009 240
0.029 0.047 0.0075 0.032 0.0063 0.0012 0.0059 0.0007 0.0040 0.0008 0.0029 0.0006 .0044 0.0010
