LONG-TERM CORROSION-INDUCED COPPER RUNOFF FROM NATURAL AND ARTIFICIAL PATINA AND ITS ENVIRONMENTAL IMPACT

Environmental Toxicology and Chemistry, Vol. 25, No. 3, pp. 891–898, 2006 q 2006 SETAC Printed in the USA 0730-7268/06 $12.00 1 .00

LONG-TERM CORROSION-INDUCED COPPER RUNOFF FROM NATURAL AND ARTIFICIAL PATINA AND ITS ENVIRONMENTAL IMPACT SOFIA BERTLING,*† INGER ODNEVALL WALLINDER,† DAN BERGGREN KLEJA,‡ and CHRISTOFER LEYGRAF† †Division of Corrosion Science, Royal Institute of Technology, Dr. Kristinas v. 51, SE-100 44 Stockholm, Sweden ‡Department of Soil Sciences, Swedish University of Agricultural Sciences, SLU (Sveriges Lantbruks Universitet), Ulvsva¨g 17, SE-750 07 Uppsala, Sweden ( Received 14 January 2005; Accepted 8 June 2005) Abstract—The overall objective of this paper is to present an extensive set of data for corrosion-induced copper dispersion and its environmental interaction with solid surfaces in the near vicinity of buildings. Copper dispersion is discussed in terms of total copper flows, copper speciation and bioavailability at the immediate release situation, and its changes during transport from source to recipient. Presented results are based on extensive field exposures (eight years) at an urban site, laboratory investigations of the runoff process, published field data, generated predictive site-specific runoff rate models, and reactivity investigations toward various natural and manmade surfaces, such as those in soil, limestone, and concrete. Emphasis is placed on the interaction of coppercontaining runoff water with different soil systems through long-term laboratory column investigations. The fate of copper is discussed in terms of copper retention, copper chemical speciation, breakthrough capacities, and future mobilization based on changes in copper concentrations in the percolate water, computer modeling using the Windermere Humic Aqueous Model, and sequential extractions. The results illustrate that, for scenarios where copper comes in extensive contact with solid surfaces, such as soil and limestone, a large fraction of released copper is retained already in the immediate vicinity of the building. In all, both the total copper concentration in runoff water and its bioavailable part undergo a significant and rapid reduction. Keywords—Runoff

Soil interaction

Copper

Retention

Risk assessment

data on runoff rates of a metal from a given structure, information on the chemical speciation of the released metal (which affects the bioavailability, a measure of the rate and extent of uptake by organisms), and information on changes during environmental entry. The potential environmental effect is also affected by dilution with storm water and rainwater from other solid surfaces and the possible retention on natural and manmade absorbing surfaces. Corrosion rates have erroneously been used to determine copper flows from external applications because of a lack of data on real runoff rates and the conservative assumption that the corrosion rate equals the runoff rate. Recent investigations have shown corrosion rates of copper to be significantly higher than corresponding runoff rates [4]. This is a result of different chemical and physical processes that govern each rate. The corrosion rate reflects the amount of bulk copper that is oxidized per time period and surface area, whereas the runoff rate reflects the amount of copper that can be dissolved and released from the patina as a result of precipitation. The former is a continuous process of varying rate, whereas the latter is determined by the frequency and duration of precipitation. The runoff rate is intimately related to prevailing environmental characteristics such as pollutant levels; dry and wet periods; precipitation characteristics such as rain volume, pH, and intensity; and material characteristics such as patina composition, thickness and morphology, surface inclination, and degree of sheltering (Fig. 1) [4,5]. On a short-term perspective, such as a rain episode, the runoff rate is time dependent, with initially high rates during the initial rain portion followed by lower and more constant rates with increasing rain volume and duration [6]. However, on a yearly perspective, the runoff rate can be considered to be fairly constant. Reported literature

INTRODUCTION

The environmental concern for diffuse emission sources of metals has gradually increased within the European community during the past decade. In the Brundtland report, it is stated that a sustainable society should ‘‘meet the needs of the present without compromising the ability of future generations to meet their own needs’’ [1]. Many activities have followed this statement within the framework of environmental control (legislative actions) based on the best knowledge available at any particular moment in time within the European community. A voluntary initiative was therefore implemented in 2001 by copper-producing organizations with the aim to cover gaps of knowledge and to generate relevant data for future environmental risk assessments of copper. Several studies exist to identify flows of copper from various sources in the society [2,3] with the aim to find sectors and applications that could be improved from a sustainable point of view. In 1996, the Swedish Environmental Protection Agency initiated a research program to identify diffuse sources of copper in the urban environment of Stockholm, Sweden [2]. From the study it was concluded that copper was dispersed predominantly from the traffic environment, from the tap-water system, and from urban buildings and constructions. As a result, a number of legislative actions toward the use of copper were implemented. However, information on total copper flows is not sufficient for environmental risk assessments, risk management decisions, or regulatory actions toward any metal. Such deliberations require site-specific scientifically sound * To whom correspondence may be addressed ([email protected]). Presented at the Symposium on Risk Assessment of Metals in Soils, 14th Annual Meeting, SETAC Europe Meeting, Prague, Czech Republic, April 18–22, 2004. 891

892

Environ. Toxicol. Chem. 25, 2006

S. Bertling et al.

Fig. 1. Key parameters and information required to obtain a comprehensive picture of the environmental interaction of corrosion-induced copper runoff from external applications in the society.

values of annual runoff rates of copper from patinated copper vary between 0.8 and 8.6 g/m2/year for urban and rural sites of varying pollutant levels and annual rainfall quantities [7]. Copper-containing runoff water from a given building is diluted with storm water from other solid surfaces and either collected and transported to a wastewater plant or introduced to solid surfaces in the near vicinity of buildings, often soil, concrete, or limestone [8]. Previous laboratory investigations have shown soil to exhibit high retention capacities of copper (.99%) at constant flow and supply of copper (5 mg/L) during three years (600 mm/year) of simulated exposure [8]. Copper is preferentially and relatively strongly retained by organic complexes. Also, concrete and limestone exhibit a high capacity to retain copper, forming naturally occurring minerals such as malachite [9,10]. A recent combined field and laboratory investigation with limestone and copper-containing runoff water showed that more than 98% copper can be retained by optimizing the specific surface area of limestone or increasing the contact time between copper-containing runoff water and limestone either by reducing the flow rate of runoff water or by increasing the amount of limestone [11,12]. A large research effort was initiated at the division of Corrosion Science at Royal Institute of Technology (KTH), Stockholm, Sweden, in response to several important circumstances: the increased awareness in society of the potential effect that release of copper from outdoor constructions may have on the environment, the lack of long-term data on copper runoff rates induced by atmospheric exposure, and the limited knowledge on the environmental interaction of released copper with soils. The unique approach has been to combine cross-disciplinary research skills to provide scientifically sound data for sustainable decisions and legislative actions. The work has been performed in close collaboration with ecotoxicologists (Laboratory of Environmental Toxicology and Aquatic Ecology, Gent University, Gent, Belgium) and soil scientists (Swedish University of Agricultural Sciences, Uppsala, Sweden), together with copper-producing organizations. The most important scientific outcome from this research investigation, to be used within the framework of environmental risk assessment and risk management, is compiled in Figure 1. The aim of this paper is to elucidate the environmental fate of corrosion-induced copper runoff from patinated copper sheet on external structures by providing quantitative data on long-term annual runoff rates of total copper, chemical spe-

ciation and bioavailability of copper at the immediate release situation by using computer modeling and tests with different bioassays such as bacteria and algae, and retention capacities of copper in soil and its potential for future mobilization. MATERIALS AND METHODS

Urban field exposure Naturally and artificially patinated copper sheet of varying patina thickness and composition was exposed unsheltered between April 25, 1996, and April 25, 2004, at an urban test site located within the campus area at the Royal Institute of Technology in Stockholm, Sweden. The test site is located on the roof of an eight-story building and is surrounded by built-up areas with relatively high traffic intensity. Mean annual environmental and meteorological data (temperature [T] and relative humidity [RH], SO2, NO2, O3, PM10 [aerosol particles ,10 mm]) were supplied by the Stockholm Environment and Health Protection Administration, Stockholm, Sweden. Total annual rain quantities were measured at KTH. Mean data are given for each year of exposure in Table 1. The following four panels were investigated and supplied by Outokumpu Copper, Va¨stera˚s, Sweden: naturally brownpatinated copper, that is, fresh copper sheet (99.98% Cu, 0.02% P) that forms a patina composed primarily of Cu2O within the eight-year time period; naturally green-patinated copper (i.e., copper sheet of varying age, 40 and 130 years, preexposed in the urban environment of Stockholm, Sweden, in which the patina is composed mainly of an inner layer of cuprite-Cu2O and an outer layer of posnjakite-Cu4SO4[OH]6·H2O and brochantite-Cu4SO4[OH]6, respectively); artificially brown-patinated copper sheet (Nordic Brown), which is chemically oxidized using chlorite at 708C during 2 to 24 h to obtain an adherent copper oxide (mainly Cu2O and CuO) with a thickness of typically 1 mm; and artificially green-patinated copper sheet that is chemically oxidized using chlorite at 708C during 2 to 24 h, followed by a subsequent application of a mixture of basic copper salts to accomplish the greenish layer (typically 18 mm) formed on top of the oxidized brown layer. Single panels of each material (300 cm2) were exposed single sided, mounted on Plexiglas fixtures at 458 from the horizontal facing south from which runoff water was collected, and transported to polyethylene vessels using silicon hoses. The sampling periods varied between 1 d and one month,

Environ. Toxicol. Chem. 25, 2006

Corrosion-induced copper runoff

893

Table 1. Annual mean environmental parameters during the field exposure in Stockholm, Swedena Exposure year Time period 1 2 3 4 5 6 7 8 a

9604–9704 9704–9804 9804–9904 9904–0004 0004–0104 0104–0204 0204–0304 0304–0404

T (8C) mean 6.8 7.6 6.7 8.3 7.9 8.2 7.3 7.8

SO2 RH (%) (mg/m3) 69 72 78 75 80 80 78 82

3.5 3.1 3.2 2.9 2.6 2.4 2.8 2

NO2 (mg/m3)

O3 (mg/m3)

PM10 (mg/m3)

Rain mm (mm/year)

42 42 44 46 42 43 45 43

49 50 47 53 46 51 56 54

18 18 16 17 18 17 19 17

405 553 563 428 582 517 340 432

Rain mm from collected rain volume impinging an inert sample holder inclined 458 from the horizontal. RH 5 relative humidity; PM10 5 concentration of particles smaller than 10 mm measured at the roof level. Rain pH varied from 4.18 to 6.38, with a medium of 5.01.

depending on variations in occurrence and quantity of rain events. The total rainwater volume impinging on the samples was collected and measured in addition to pH and total copper concentration. Copper concentrations were measured by means of atomic absorption spectroscopy (Instrumental Laboratory model IL 551, Instrumentation Lab, Wilmington, DE, USA). Prior to analysis, all collected runoff water samples were acidified with HNO3 to reach a pH of approximately 2 in order to avoid any formation of copper complexes on the walls of the collecting vessels. The copper concentration was determined as the mean value of 15 individual measurements on each sample, with residual standard deviation value of less than 0.01. The detection limit of the current setup is 0.01 mg/L.

Soil column investigation The long-term interaction of corrosion-induced copper runoff water with soil systems was investigated in laboratory column studies using three different topsoils representative of European conditions (Table 2). The aim was to determine the breakthrough capacity of each soil. Artificial runoff water, containing 4.8 mg Cu/L, was continuously introduced to each soil core during a simulated time period of approximately 30 years (assuming an infiltration rate of 250 mm/year). Three centimeters (27 g) of each soil were exposed to a constant runoff water intensity of 2 mm/h (equivalent to a supply of 1.5 ml/ h for this column size), with a total volume of approximately 6,500 ml runoff water percolating each system. A detailed description of the soil column investigation and setup is given by Bertling et al. [13]. Percolate water samples were collected

every fifth day for pH and total copper concentration measurements. Total copper was determined by means of inductively coupled plasma–atomic emission spectroscopy. Measurements of the chemical composition of some percolate water samples, including measurements of anions, cations, and dissolved organic carbon (DOC), were performed for chemical speciation modeling. The chemical speciation of copper in percolate waters was predicted using version 5 of the Windermere Humic Aqueous Model (WHAM [V]) [14]. This model allows calculations of the equilibrium speciation of copper under various environmental conditions characterized by, for example, pH, hardness, alkalinity, presence of anions, and competing cations together with organic ligands. Exposed soil columns of the Ho¨gbytorp and the Kalmthout topsoils were cut into three vertical segments. Each segment was extracted in sequence starting with extraction in artificial rainwater (1 g soil/16 ml rainwater) followed by extraction in ethylenediaminetetraacetic acid (EDTA) (0.02 M EDTA 1 0.5 M NH4Ac). RESULTS AND DISCUSSION

The environmental interaction of corrosion-induced copper runoff from naturally and artificially patinated planar surfaces of copper used for roofing applications is presented in the following sections. This top-down/bottom-up approach focuses on aspects related to long-term annual release rates of copper (eight years), copper chemical speciation and ecotoxicity at the immediate release situation, retention capacities, and potential groundwater leaching during environmental interaction of copper-containing runoff water with urban soils.

Table 2. Characteristics of soils used for percolation studies

Annual runoff rates of Cu at the immediate runoff situation Ho¨gbytorp Kalmthout Ko¨vlinge topsoil topsoil topsoil pH (H2O) OC (%)a Clay (%) CECeff (cmolc/kg)b Cuaq-regia (mg/kg)c CuEDTA (mg/kg)

5.95 5.1 20–40 13.4 13 3

4.93 4.1 1 2.6 7 1.8

5.65 5.6 7 6 6.6 1.7

Ko¨vlinge subsoil 5.3 2 4 1.6 1.8 0.1

OC 5 organic carbon. Cation exchange capacity (CEC) equals extractable (0.1 M BaCl2) amounts of Ca, Mg, Na, and K in the soil. c The initial Cu concentration in the soils was determined as the exchangeable concentration of Cu (extraction with 0.5M NH4Ac 1 0.02 M ethylenediaminetetraacetic acid [EDTA], and the total concentration of Cu, obtained through extraction with aqua regia according to International Organization for Standardization standard 11 466:1995. All extractions were performed for 24 h. a

b

Average annual runoff rates of Cu are compiled in Figure 2a for naturally and artificially patinated copper during eight years of urban exposure. The variations between different years of exposure are a result primarily of differences in annual rainfall quantities (see Table 1). The rates are calculated on the basis of the sum of release rates for individual sampling periods during each year of exposure and normalized in relation to the exposed surface area. Naturally and artificially brown-patinated copper (natural: 1.22 6 0.29 g/m2/year; artificial: 1.22 6 0.30 g/m2/year) show similar runoff rates as a result of similarities in patina composition (mainly Cu2O), thickness, and morphology, whereas green-patinated copper surfaces (inner layer: mainly Cu2O; outer green layer: mainly Cu4SO4[OH]6) show slightly higher runoff rates (natural: 1.47 6 0.30 g/m2/year; artificial: 1.68 6 0.44 g/m2/year) compared

894

Environ. Toxicol. Chem. 25, 2006

Fig. 2. Mean annual release rates of Cu from naturally and artificially patinated copper sheet exposed unsheltered during eight years (artificially brown copper sheet: six years) in the urban environment of Stockholm, Sweden (a) and correlation between the annual rainfall quantity plotted against the annual runoff rate for naturally and artificially brown- and green-patinated copper sheet (b).

to brown-patinated copper. However, within the framework of risk assessment and risk management, these small differences are negligible when estimating flows of copper from external constructions. The natural background deposition of copper was measured to approximately 0.02 g/m2/year. The reason for the slight discrepancy in runoff rates between brown- and green-patinated copper has been thoroughly described elsewhere [15]. The discrepancy is related primarily to differences in runoff rates during the initial rainfall portion (first flush) as a result of differences in patina thickness and porosity of the green patina layer. This layer has a high ability to retain corrosive species and humidity, hence increasing the possibility for copper to be dissolved from the patina. However, no significant difference in runoff rate is observed between brown and green patina during prolonged rainfall quantities (steady state) [6]. The first flush effect is more pronounced for greenpatinated copper and is intimately related to differences in prevailing environmental conditions. This is illustrated in Figure 2b, showing a good correlation between the annual runoff rate and the rainfall quantity for naturally brown patina, whereas the correlation for artificially brown and green patinas is not as evident. The reason for worse correlation for artificial compared to natural patinas is probably that the artificial patina has not yet reached its equilibrium composition and thickness in the environmental conditions where it is exposed. Models have recently been developed with the aim to predict runoff rates of copper from surfaces of various inclinations at any urban site of varying rain pH or SO2 concentration and annual rain quantity [7]. Presented runoff rates of copper and literature data (0.8–8.6 g/m2/year) [7] can be used to calculate copper flows from any building in urban environments. However, these calculations must consider the fact that real roofs and buildings show large variations in, for example, orientation, inclination, deposition of corrosive species, and degree of rain sheltering, which result in runoff rates at least 30% lower compared to the standardized 458 orientation facing south in the Northern Hemisphere [16].

Copper chemical speciation, bioavailability, and ecotoxicity at the immediate runoff situation Total copper runoff rates are not sufficient for environmental risk assessment and management since they do not consider the copper chemical speciation, which determines the bioavailability (a measure of the rate and extent of absorption into organisms) and its changes during environmental entry. As previously discussed and thoroughly described by He et al. [6], the released copper concentration (and runoff rate) decreases with time and rain quantity during any rain event.

S. Bertling et al.

Fig. 3. Copper concentration in collected runoff water samples at the immediate runoff situation from naturally and artificially patinated copper (a) and corresponding collected rainfall quantities impinging the same surfaces (b).

Since the concentration also is intimately related to prevailing environmental conditions in terms of, for example, rain pH, pollutant levels, and dry and wet periods, the concentration varies significantly between different sampling periods. Copper concentrations, measured in collected runoff water samples at the immediate runoff situation, are compiled in Figure 3a for both naturally and artificially patinated copper from all samplings periods (546 periods) during the exposure period. The median copper concentration for naturally and artificially patinated copper is 3.2 mg Cu/L (between 0.85 and 18 mg Cu/L). Corresponding rainfall quantities impinging the same surfaces show a median value of 27 mm (between 1.7 and 51 mm) (Fig. 3b). The concentration of released copper in the runoff water is, hence, related to a combination of rain characteristics and duration, prevailing environmental conditions, and patina characteristics. Dilution with storm water from other solid surfaces and retention on natural and manmade absorbing surfaces should also be considered when considering the potential environmental interaction. The bioavailability of any substance is different for different organisms and is related to the chemical speciation, which in turn depends on a large number of physical and chemical parameters such as pH, temperature, presence of organic and inorganic ligands, and so on [17] as well as different uptake routes. The free hydrated ion Cu(H2O)261 is considered to be the most bioavailable form of copper in aqueous media [18]. Its concentration in natural surface waters is low as a result of its strong affinity to organic complexants [19,20]. Measurements of the fraction of the free hydrated copper ion (Cu[H2O]261) at the immediate release situation is presented in Figure 4 for naturally and artificially patinated copper. The median fraction varies between 64 and 73% for all patinas, with 50% of all data between approximately 60 and 80%. Periods of low fractions are most probably a result of higher organic content of the runoff water, such as pollen (no measurements of the organic content of the runoff water were performed). Measurements of the pH of the runoff water from both naturally and artificially patinated copper showed median values of 6.2 as a result of the buffering capacity of released species from the patina. Computer modeling using MinteqA2 (http://www.lwr.kth.se/English/OurSoftware/vminteq/index. htm) of the chemical speciation of copper in runoff water from different sampling periods during the first three years of exposure, measurements of the bioavailability toward bacteria, and acute ecotoxicity testing on the green algae Raphidocelis subcapitata at the immediate release situation have previously been published with consistent observations [18]. The results show copper to be predominantly present as the free hydrated copper ion with minor constituents of Cu(OH)1, Cu(OH2) 221, CuCO3(aq), and CuSO4(aq) and hence in a chemical form that

Environ. Toxicol. Chem. 25, 2006

Corrosion-induced copper runoff

Fig. 4. Fraction (%) of the free cupric ion concentration, as measured with an ion selective electrode, compared to the total copper concentration released from naturally and artificially (Art.) brown- and greenpatinated copper. Each box comprises the inner 50% of the data closest to the median (shown as a line). The lines extending from the top and the bottom of each box mark the minimum and maximum values that fall within the median value 61.5 times the concentration span of the box. Outliers are displayed as individual points.

is bioavailable toward bacteria and ecotoxic toward the algae R. subcapitata (72-h EbC50 acute effect concentrations from 6 to 24 mg Cu/L). Recent studies in surface waters show an optimum copper concentration for R. subcapitata between 1 and 35 mg Cu/L [21]. The observed growth inhibition in runoff water and not in the surface water is probably a result of lower complexation toward dissolved organic carbon in the runoff water. Results on chemical speciation and bioavailability at the immediate release situation should not be simply extrapolated to real environmental effects since both total concentrations and the chemical speciation of released copper in the runoff water will undergo major changes during environmental entry. Dilution with surface water or storm water from other solid surfaces decreases the concentration with orders of magnitude, and the interaction with, for example, organic and inorganic species and surfaces will decrease the fraction of the free cupric ion of runoff water and the copper bioavailability through, for example, complexation. Similar effects will occur during interaction with soil. These processes are discussed in the next section.

Environmental interaction of copper-containing runoff water with urban soils The interaction of copper-containing runoff water with three different topsoil systems of varying characteristics (Table 2) was evaluated in long-term laboratory soil column investigations with respect to retention capacities, changes in total copper concentration, chemical speciation, and potential groundwater leaching. The column investigation simulates approximately 30 years (assuming an annual infiltration rate of 250 mm/year) of continuous outdoor exposure to copper-containing runoff water by using artificial runoff water with a pH of 6.2 and a copper concentration of 4.8 mg/ L (Table 3). The

Fig. 5. Copper retention capacities of the Ho¨gbytorp (m), Ko¨vlinge (X), and Kalmthout (v) topsoils (each 27 g) (a). Concentration of total copper in percolate water from the Ko¨vlinge topsoil after interaction with rainwater (0 mg Cu/L) and runoff water (4.8 mg Cu/L) (b).

copper concentration in artificial runoff water was slightly higher than the observed/natural median copper concentration in runoff water (3 mg/L; see the previous discussion). Soil column investigations were performed in parallel with artificial rainwater (no copper addition, pH 6.2) in order to distinguish between natural percolation of copper from the soil itself and percolation as a result of the interaction of copper-containing runoff water. A high retention capacity for copper (.99%) was observed for all topsoils investigated during approximately 25 years of simulated exposure to copper-containing runoff water (Fig. 5a). The retention capacity is defined as the fraction of introduced copper (in runoff water) that has reacted with the soil. The retention capacity is decreasing after approximately 26 years of runoff water interaction for the Kalmthout soil (clay content, 1%; pH, 4.93; organic material, 4.1%; cation exchange capacity, 2.6 cmol/kg). The Ko¨vlinge topsoil shows the same behavior after approximately the same time period (clay content, 7%; pH, 5.65; organic material, 5.6%; cation exchange capacity, 6 cmol/kg), whereas the Ho¨gbytorp topsoil after 30 years of simulated exposure still shows a high retention capacity (clay content, 20–40%; pH, 5.95; organic material, 5.1%; cation exchange capacity, 13.4 cmol/kg). The main reason for the lower retention capacity of the Kalmthout and Ko¨vlinge topsoils compared to the Ho¨gbytorp topsoil is probably a lower clay content and pH, resulting in lower cation exchange capacity. The data are applicable in cases where the downward transport of runoff water predominates compared to surface runoff (because of high flow rates from downspouts, some runoff water is transported on the soil and vegetation surface instead of penetrating downward) before a recipient is reached. Laboratory-obtained data were implemented into the sorption and transport model software Hydrus-1Dt (http:// www.ussl.ars.usda.gov/models/hydr1d1.htm) in order to predict different realistic outdoor scenarios. This approach is presented in its full context in Bertling et al. [13]. The time period for the Ho¨gbytorp topsoil to show a decreasing retention capacity, not seen during the laboratory exposure, was predicted to approximately 70 years (after percolation of ;17,500 mm

Table 3. Chemical composition (mg/L) of artificial runoff water used in the percolation study

Artificial rain Artificial runoff water

895

pH

Cu

SO4

Cl

NO3

NH4

Na

K

Mg

Ca

6.2 6.2

0 4.8

3.6 3.6

0.36 0.36

2.48 2.48

0.72 0.72

0.23 0.23

0.12 0.12

0.12 0.12

0.2 0.2

896

Environ. Toxicol. Chem. 25, 2006

S. Bertling et al.

Table 4. Chemical input data for modeling speciation of Cu in percolate waters. Data are given as the span representative of the simulated exposure period Ko¨vlinge top pH TOCa (mg/L) Fe (mg/L) Al (mg/L) Cu (mg/L) Zn (mg/L) Cd (mg/L) Ca (mg/L) K (mg/L) Mg (mg/L) Na (mg/L) Cl (mg/L) NO3 (mg/L) SO4 (mg/L) F (mg/L) a

5.01–5.57 1–13 6–107 100–726 0.5–388 16–102 0–0.2 3.8–10 1–3.5 0.2–1.5 1.5–8.8 0.2 11–13 0.2 0.3–0.4

TOC 5 total dissolved organic carbon in percolate water.

runoff water). Modeling of various realistic scenarios of the interaction of copper-containing runoff water (0.01–10 mg Cu/ L, 1–50 mg DOC/L) with a 10-cm-thick topsoil (2–10% organic carbon, pH 4–6) and 40-cm-thick subsoil (2% organic carbon, pH 5) showed the time for breakthrough for this kind of soil system to vary between 170 and more than 8,000 years. The results are supported by field analysis of soil systems at sites of known copper introduction for as long as 25 years, showing only the top 40 cm of the soil system to have elevated concentrations of copper [22]. The copper runoff water interaction with soil results in an immediate decrease in copper concentration from 4,800 mg Cu/L (introduced) to approximately 10 mg Cu/L. With time, the concentration decreases even more and reaches a level between 1 and 2 mg Cu/L after 1,000 mm of runoff water percolation (about four years of simulated exposure). This concentration level is maintained until the retention capacity decreases and breakthrough is initiated for the Kalmthout and the Ko¨vlinge topsoils. An initially high concentration of copper in combination with a high concentration of DOC is also seen in percolate waters when exposed to rainwater only (artificial rain with no addition of copper), exemplified for the Ko¨vlinge topsoil in Figure 5b. This initial decrease in copper and DOC concentration is probably a result of drying of the soils before exposure, and the majority of copper detected in the percolate water during the first 1,000 mm of simulated exposure to copper-containing runoff water is hence due to natural leaching

Fig. 6. Chemical speciation calculations, using version 5 of the Windermere Humic Aqueous Model, of copper as free cupric ions (Cu21) and organically complexed copper in percolate water from the Ko¨vlinge topsoil. The results are presented as the concentration of each species (a) and corresponding fraction in the percolate water (b).

Fig. 7. Vertical profile of retained copper for the Ho¨gbytorp and the Kalmthout topsoils by extracting segments of the soil core in sequence starting with artificial rainwater (a measure of easily mobilized copper) followed by ethylenediaminetetraacetic acid (EDTA) (a measure of more strongly retained copper).

of copper from the soil itself. Similar copper concentrations are detected in the percolate water when exposed to both artificial rainwater (no copper addition) and artificial runoff water (copper addition). As a result, it can be concluded that percolate water with a copper concentration between 1 and 2 mg Cu/L can be transported through a soil system toward any given recipient. These copper concentrations are similar to what is found in pore waters of natural soils without any anthropogenic influence. Dilution effects of copper-containing runoff water with storm water and/or rainwater from other solid surfaces were modeled with the HYDRUS-1D sorption and transport model [23]. A dilution factor of 1,000 (a reduction of the copper concentration from 10 to 0.01 mg/L) resulted in a decreased time period to reach breakthrough from 170 years to more than 8,000 years for a 50-cm-thick soil profile. The potential risk for groundwater leaching was calculated to vary from 200 to about 500 years, depending on soil organic content and pH for a copper concentration of 5 mg/L. More details are given in Bertling et al. [13]. Computer modeling using WHAM(V) was performed to predict the chemical speciation of copper in percolate water. Input data for the modeling calculations were based on chemical analysis of percolate water (Table 4). The results are exemplified for the Ko¨vlinge topsoil in Figure 6, showing copper to be predominantly complexed to organic carbon and the rest to be present as free hydrated cupric ions in the percolate water. This fraction increases with decreasing concentration of dissolved organic carbon (.10 years of simulated exposure). The potential for future leaching of retained copper (.99%) was simulated for the Ho¨gbytorp and the Kalmthout topsoils by extracting vertical segments of the soil cores in sequence starting with artificial rainwater followed by EDTA. The vertical profile of retained copper is presented as mean values for both soils investigated in Figure 7. Minimum and maximum values are illustrated with the arrow for each segment. The results show that copper is strongly retained by soil, as only low concentrations of copper are found in equilibrium with the solution when extracting exposed soils with artificial rain (0.5–0.05%), which indicates a low transport of copper through

Corrosion-induced copper runoff

the soil. The major fraction of introduced copper needs a strong complexing agent such as EDTA for exchange from the solid to the solution phase. All retained copper was not exchangeable toward EDTA but seems to be irreversibly retained. More copper can be extracted from the upper soil segments, which suggests a relatively fast retention of copper within the soil core. Limitation of copper additions to soil and its effects on plants and organisms are more thoroughly discussed by Landner and Reuther [24]. The results from the soil column investigation illustrate that copper in runoff water is retained to a large extent already in the immediate vicinity of a roof, exemplified by soil. As a result, the total copper concentration undergoes a significant and rapid reduction before reaching a recipient such as a lake or the groundwater table. The concentration of copper after interaction with soil is at background levels, and the transport of copper through soil is slow. CONCLUSION

A cross-disciplinary research project, combining corrosion science, ecotoxicology, and soil science, has been implemented in an eight-year urban field investigation and in parallel laboratory investigations simulating approximately 30 years of exposure. The aim has been to provide quantitative data on copper release rates from naturally and artificially patinated copper sheets, commonly used for roofing applications, and to address the issues of chemical speciation, bioavailability, and potential ecotoxicity of released copper at the immediate release situation. Parallel laboratory soil-column investigations have aimed at illustrating the fate of copper-containing runoff water in contact with the environment. Annual release rates from naturally and artificially brown-patinated copper sheet are 1.22 6 0.29 and 1.22 6 0.30 g/m2/year, respectively. For naturally patinated green patina, the release rate is 1.47 6 0.30 g/m2/year and somewhat higher for artificially patinated green patina, 1.68 6 0.44 g/m2/year. No change with time is seen during long-term exposure; that is, model predictions can be made on long-term basis. At the immediate release situation, the total copper concentration varies between 0.85 and 18 (median value 3.2 mg/L). The dilution effect from rain impinging on other sources than copper sheets results in 10 to 1,000 times lower concentrations before environmental entry. Experimental observation showed 60 to 80% of the released copper to be present as hydrated copper ions. The 72-h EbC50 vales for R. subcapitata in runoff water varied between 6 and 24 mg/L. When copper in runoff water comes in contact with soil, .98% of the introduced copper is retained within the soil core. Capacities of more than 30 years for soil cores of (3 cm) have been found. The total copper concentrations after interaction with soil are in the range of background values for copper (1–10 mg Cu/L), and most of this copper is complexed to dissolved organic matter, leaving a very low fraction of the introduced copper available for organisms. Modeling field scenarios of a soil profile with a 10-cm-thick top layer and a 40cm sublayer resulted in breakthrough times ranging from 170 to more than 8,000 years, depending on the topsoil pH and organic carbon content and copper and DOC concentration of runoff water (i.e., very little risk for groundwater copper contamination). Acknowledgement—The financial support from the European Copper Institute, Belgium, is gratefully acknowledged. Jon Petter Gustavsson, Royal Institute of Technology, is acknowledged for his helpful dis-

Environ. Toxicol. Chem. 25, 2006

897

cussions and review of the experimental work and evaluation of data. Many thanks to Gunilla Lundberg, Lise Gustafsson, and Kenth Andersson for performing all soil batch experiments and analysis. REFERENCES 1. Brundtland H. 1987. Our Common Future. Oxford University Press, Oxford, UK. 2. Bergba¨ck B, Johansson K, Molander U. 2001. Urban metal flows— A case study of Stockholm. Water Air Soil Pollut 1:3–24. 3. Landner L, Lindestro¨m L. 1999. Copper in Society and in the Environment. Swedish Environmental Research Group, Stockholm. 4. He W, Odnevall Wallinder I, Leygraf C. 2001. A comparison between corrosion rates and runoff rates from new and aged copper and zinc as roofing materials. Water Air Soil Pollut 1:67–82. 5. Odnevall Wallinder I, Verbiest P, He W, Leygraf C. 2000. Effects of exposure direction and inlclination on the runoff rates of zinc and copper roofs. Corrosion Science 42:1471–1487. 6. He W, Odnevall Wallinder I, Leygraf C. 2000. A laboratory study of copper and zinc runoff during first flush and steady-state conditions. Corrosion Science 43:127–146. 7. Odnevall Wallinder I, Bertling S, Xueyuan Z, Leygraf C. 2004. Predictive models of copper runoff from external structures. Journal of Environmental Monitoring 6:704–712. 8. Boulanger B, Nikolaidis NP. 2003. Mobility and aquatic toxicity of copper in an urban watershed. Journal of the American Water Resources Association 39:325–336. 9. Bertling S, Odnevall Wallinder I, Leygraf C, Berggren D. 2002. Immobilisation of copper in runoff water from roofing materials by limestone, soil and concrete. Proceedings, 15th International Corrosion Congress, September 22–27, Granada, Spain. Paper 44. 10. Sundberg R. 1999. The fate of copper released from the Vasa ship museum. Metall: Internationale Fachzeitschrift fu¨r Metallurgie 52:230–231. 11. Aziz HA, Othman N, Yusuff MS, Basri DRH, Ashaari FAH, Adlan MN, Othman F, Johari M, Perwira M. 2001. Removal of copper from water using limestone filtration technique: Determination of mechanism of removal. Environ Int 26:395–399. 12. Bertling S, Odnevall Wallinder I, Leygraf C. 2002. The capacity of limestone to immobilize copper in runoff water—A laboratory investigation. Proceedings, 15th International Corrosion Congress, September 22–27, Granada, Spain. Paper 42. 13. Bertling S, Degryse F, Odnevall Wallinder I, Smolders E, Leygraf C. 2006. Model studies of corrosion-induced copper runoff fate in soil. Environ Toxicol Chem 25:683–691. 14. Tipping E. 1998. Humic ion-binding model VI: An improved description of the interactions of protons and metal ions with humic substances. Aquatic Geochemistry 4:3–48. 15. Zhang X, He W, Odnevall Wallinder I, Pan J, Leygraf C. 2002. Determination of instantaneous corrosion rates of copper from naturally patinated copper during continuous rain events. Corrosion Science 44:2131–2151. 16. Odnevall Wallinder I, Bertling S, Leygraf C. Environmental interaction of copper and zinc released from building materials as a result of atmospheric corrosion. Metall: Internationale Fachzeitschrift fu¨r Metallurgie 58:557–560. 17. Di Toro DM, Allen HE, Bergman HL, Meyer JS, Paquin PR, Santore RC. 2001. Biotic ligand model of the acute toxicity of metals. 1. Technical basis. Environ Toxicol Chem 20:2838–2396. 18. Karle´n C, Odnevall Wallinder I, Heijerick D, Leygraf C. 2002. Runoff rates, chemical speciation and bioavailability of copper released from naturally patinated copper. Environ Pollut 120: 691–700. 19. De Schamphelaere K, Vasconcelos F, Tack F, Allen HA, Janssen CR. 2004. The effect of dissolved organic matter source on acute copper toxicity of Daphnia magna. Environ Toxicol Chem 23: 1248–1255. 20. De Schamphelaere K, Heijerick D, Jansen CR. 2002. Refinement and field validation of a biotic ligand model predicting copper toxicity to Daphnia magna. Comp Biochem Physiology C Pharmacol Toxicol Endocrinol 133:243–258. 21. Bossuyt B, Janssen CR. 2004. Long-term acclimation to copper of Pseudokirchneriella subcapitata (Korshikov) Hindak: Changes in tolerance and physiology. Aquat Toxicol 68:61–74. 22. Jensen DL, Holm PE, Christensen TH. 2000. Soil and groundwater contamination with heavy metal at two scrap iron and metal recycling facilities. Waste Manag Res 18:52–63.

898

Environ. Toxicol. Chem. 25, 2006

23. Sˇimu˚nek J, Huang K, van Genuchten M.Th. 1998. The HYDRUS code for simulating the one-dimensional movement of water, heat and multiple solutes in variably saturated media. Research Report 144. U.S. Salinity Laboratory, U.S. Department of Agriculture, Riverside, CA.

S. Bertling et al. 24. Landner L, Ruther R. 2004. Metals in Society and in the Environment: A Critical Review of Current Knowledge in Fluxes, Speciation, Bioavailability and Risk for Adverse Effects of Copper, Chromium, Nickel and Zinc. Kluwer, Dordrecht, The Netherlands.