Environ. Sci. Technol. 2004, 38, 515-521
Emission, Fate and Effects of Soluble Silicates (Waterglass) in the Aquatic Environment H E N N O P . V A N D O K K U M , * ,§ JAN H. J. HULSKOTTE,# KEES J. M. KRAMER,‡ AND J O E¨ L W I L M O T † Department for Ecological Risk Studies, TNO Environment, Energy and Process Innovation, P.O. Box 57, 1780 AB, Den Helder, Netherlands, Department for Emissions Assessment, TNO Environment, Energy and Process Innovation, P.O. Box 342, 7300 AH, Apeldoorn, Netherlands, Mermayde, P.O. Box 109, 1860 AC, Bergen, Netherlands, and CEFIC-CEES, Avenue E. van Nieuwenhuyse 4, bte 1, B-1160 Bruxelles, Belgium
Soluble silicates, commercially known as waterglass, are among the largest volume synthetic chemicals in the world. Silicon from waterglass is rapidly transformed to the biologically active orthosilicic acid (referred to as dissolved silicate). This paper aims to assess the impact of waterglass on the aquatic environment in Western Europe. The emission to surface waters from the four most relevant application areas, household detergents, pulp and paper production, water and wastewater treatment, and soil stabilization, is estimated to be ca. 88-121 kton of SiO2 per year. This is a small fraction ( 10 m from stabilized soil. Fraction Removal in Municipal Sewagewater Treatment Plant (frs). To assess the fraction removal of dissolved silicate in a STP, samples were taken from the influent and effluent of 6 STPs in northwestern Netherlands in spring 2002. Samples (proportional sampling over 24 h) were taken in duplicate, three times with two-week intervals. The samples were left to settle for 24 h. The supernatant was treated with HF, and silicate monomer concentrations (C) were measured by AAS. The fraction removal was calculated as
frs ) (Cinfluent - Ceffluent)/(Ceffluent) Fraction Removal in Wastewater Treatment Plant of a Papermill (frw). The fraction removal of dissolved silicate in the WWTP of a paper mill was determined by measurements in the influent and effluent of three different German paper mills. Concentrations (C) were determined by AAS, after settling of the sample. The fraction removal was calculated as
frw ) (Cinfluent - Ceffluent)/(Ceffluent) Chemical Speciation Modeling. The chemical speciation of dissolved silicate in fresh surface water, seawater, and (municipal) wastewater was modeled with the program CHEAQS (12) to gain understanding of the biological activity of waterglass when it disassociates to dissolved silicate. Mass balance equations and stability constants K can be entered in the model, which will calculate the speciation at equilibrium (13). Next to Si and H, the cations Mg, Ca and Fe were included in the model and the anions Cl, CO3, HCO3, and SO4. Stability constants were based on refs 12, 14, and 15. The composition of representative wastewater was based on refs 16 and 17; representative freshwater composition is based on average Rhine and Meuse water composition (18, 19); and representative seawater composition is derived from ref 18. Scenarios with and without solid species were evaluated, at pH values of 7 to 9.
3. Results and Discussion Utilization of Waterglass in Water-Relevant Applications. The total waterglass production by CEES member companies in 2000 is ca. 700 kton of SiO2/year. Of the total amount of waterglass produced only 21% (145 kton of SiO2/year) is used in one of the four water-relevant application areas (see Table 1). The remainder of the waterglass produced is used as precursor chemical to produce silica, silica gel and silica sols. Detergents are the most important group in water related applications (largest volume: 78 kton of SiO2/year) followed by pulp and paper production (53 kton of SiO2/year). Together, these two groups make up 90% of the waterglass used in water-relevant application areas. The results from
TABLE 1. Consumption of Waterglass in Western Europe for Four Relevant Application Areas and Estimated Emission to the Surface Watera
a
application area
consumption based on CEES sales/ deliveries
consumption based on consumption data
emission to surface water
detergents pulp and paper production (waste-)water treatment soil stabilization total
78 52.5 2.8 11.7 145
91 107 unknown unknown 198+
65-76 21-43 2.3 negligible 88-121
All numbers are in kton of SiO2/year.
the “top-down” estimate (based on consumption figures) are somewhat higher (Table 1). The amount of waterglass for household detergents is 91 kton of SiO2/year (1.2 times volume based on production data); the amount for pulp and paper production is 107 kton of SiO2/year (2 times the production volume). There were no consumption data available to estimate the volume of waterglass used for (waste)water treatment and soil stabilization. There are several possible reasons for the observed differences: not all European waterglass production is for the European market, and European countries may import waterglass from the U.S. or Japan. Also, there may be some errors in attributing production data to end-uses, as waterglass producers do not prepare the final formulations that are sold on the market. Taking this into account, the fit between bottom-up and top-down assessment is reasonably good. Emission Factors and Emissions. Waterglass in household detergents will enter the sewer system after use in washing machines, dishwashers or hand wash. The fraction of households that are connected to a STP varies from 27% in Belgium to 97% in The Netherlands (5, 10). The removal of waterglass in STPs was measured at six plants in The Netherlands. The average influent concentration of all 36 samples was 7.3 mg/L of SiO2, and the average effluent concentration was 6.7 mg/L of SiO2 (92% of influent concentration). There was no significant difference between three STPs that were equipped with a chemical phosphorus removal step, and three STPs that were not. Three STPs, all equipped with an aeration basin, were considered as most representative. The average removal (( SD) was 11.8% ( 0.5%, 8.6% ( 1.4% and 9.5% ( 4.1%, respectively. The other three STPs showed a strong variation in fraction removal and occasionally showed an increase of dissolved silicate concentration during passage through the STP. Two of the STPs are plug-flow systems, the third an aeration basin type which is overloaded (Water Board Uitwaterende Sluizen, personal communication). A fraction removal of 10% was used as a representative value for the emission assessment. The fraction that is lost in the sewer system before the wastewater reaches the STP is not known. Sedimentation of dissolved silicate adsorped to suspended solids, or sedimentation of amorphous silica or silicate polymers, may lead to losses. In the emission assessment, a fraction removal of 10%, equal to the removal in the STP, is assumed as a best guess. From these data, an overall emission factor of 84% for household detergents was calculated. An amount of 65 (bottom-up approach) to 76 (top-down approach) kton of SiO2/year is emitted to surface water systems. The majority will be emitted to freshwaters and a small fraction directly into coastal waters. A fraction of the waterglass used in pulp and paper production is retained in the process (incorporated in produced paper or pulp). The remainder is discharged to a WWTP, where another fraction is removed and the rest is discharged to the surface water. To estimate the fraction of dissolved silicate that is retained in the process, mass balances for three paper mills in Germany were made (Dr. Rieber,
Woellner Silikat, personal communication). From data on the paper production, raw water intake, raw water dissolved silicate concentration, water volume discharged to WWTP, influent concentration in WWTP and the input per tonne pulp produced, the fraction of dissolved silicate input that is retained in the process was calculated. For three paper mills, this fraction is 61%, 41% and 69% of the dissolved silicate input, respectively. The fraction removal in the WWTP can be calculated from measurements in the influent and effluent. Average influent concentrations ranged from 32 to 91 mg/L of SiO2, and average effluent concentrations ranged from 28 to 70 mg/L of SiO2. The average fraction removal at the three plants was 2, 13 and 30%. For the emission assessment, a typical emission factor of 40% was calculated which accounts for both retention in the production process and in the WWTP. From these numbers, an emission to the surface water of 21 (bottom up approach)- 43 kton (topdown approach) of SiO2/year can be estimated. For dissolved silicate used in (waste-)water treatment, an emission factor of 84% is assumed (comparable to household detergents). This results in an annual emission of 2.3 kton of SiO2/year (bottom-up approach). The emission from the fourth application area, soil stabilization, is expected to be negligible in comparison to the first two application areas. No data are available, but from the emission route (weathering of the matrix f leaching to groundwater f diffusive or advective transport with groundwater flow to surface water) it is assumed that only a negligible amount (arbitrarily set at 0%) of the 11.7 kton of SiO2/year (bottom-up approach) will enter the surface water. In conclusion, the volume of silicates from waterglass that is used in the four water-relevant application areas in Western Europe is estimated at ca. 88-121 kton of SiO2/ year. Natural Fluxes and Anthropogenic Emissions. An estimate of the natural dissolved silicate fluxes may help to gain an understanding of the relative magnitude of the Western European anthropogenic emission flux (88-121 kton of SiO2/ year). Natural sources of dissolved silicate arise from the chemical weathering of sedimentary and crystalline rock in watersheds, aolian erosion of land surface followed by longdistance atmospheric transport, and weathering of submarine basalt (20). The most recent estimate of the input of dissolved silicate to the world oceans was made by Tre´guer et al. (20) with a total riverine input of 301 Mtonnes of SiO2/year, 20% of which originates from temperate regions. Assuming that the temperate regions contain Europe, the former Soviet Union, China, the United States and Canada, the surface area of Western Europe is ca. 6.5% of the temperate regions, resulting in an estimated flux of dissolved silicate to the Atlantic Ocean of 3.9 Mtonnes/year. A second estimate can be obtained from multiplying average river discharge and average dissolved silicate concentrations. Discharge data for major European rivers are summarized by Kempe et al. (21). A total flux of dissolved silicate of 2.0 to 8.8 Mtonnes of SiO2/ year transported by rivers to the sea was calculated assuming a range in the average dissolved silicate concentrations of 3 VOL. 38, NO. 2, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
517
mg/L of SiO2 (average of the Rhine in Holland) (18) to 13 mg/L of SiO2 (average concentration world rivers) (22). As these estimates are based on actual data, they include both natural and anthropogenic dissolved silicate loads. The natural flux to the aquatic environment is higher than these estimates, because net removal of dissolved silicate occurs in river systems (see later). On the scale of western Europe, the anthropogenic emissions (ca. 100 ktonnes of SiO2/year) are therefore 9). At pH ) 10, the dominance of orthosilicic acid is taken over by H(H2SiO4). Si from waterglass that has entered the water system is not only affected by chemical processes but also by biological processes. Diatoms play a key role, because they utilize dissolved silicate to build their frustule (silicious cell wall). The role of diatoms in the biogeochemical cycle of silicon in oceans has been reasonably well studied. Once dissolved
TABLE 3. Expected Eco(toxico)logical Effects of an Increase of the Concentration Dissolved Silicate with a Magnitude of 0.1-10 mg/L of SiO2 in the Aquatic Environment in Europe no. i ii iii iv v vi vii
effect
remarks
Extension of spring and fall diatom blooms
Likely; esp. in temperate shallow lakes. Function of Si:P molar ratio Decrease algal biomass (green algae, bluegreen algae, Possible; increased spring diatom bloom will leave less P flagellates) in summer available for summer/fall blooms. Some cyanobacteria or flagellates are considered as noxious or harmful. Shift of primary production from summer to spring and Possible, due to effects (i) and (ii). Depends on mineralization fall rate of phosphorus Overall increase of phytoplankton biomass over the year Possible, when increase in spring and fall bloom exceeds decrease of summer nondiatom bloom Increased zooplankton, benthos, fish production and/or Possible; due to changes in phytoplankton production (detritus). temporal changes in production pattern Depends on P and N in systems. Shift in oxygen consumption from water column to Possible, opposite recorded for decreased Si:N ratios sediments Ecotoxicity Unlikely; toxic thresholds higher than expected environmental concentrations; pH effects not expected in natural systems
silicate is incorporated into a diatom frustule as an amorphous solid known as biogenic silica, upon death of the diatom it will take some time (on average: 50 days for 50%) for the biogenic silica to dissolve (4). In the oceans, a large fraction (60% on average) of biogenic silica production is recycled via dissolution in the upper 100 m of the water column (28, 29). Because of this removal process, surface water in oceans is undersaturated with respect to Si (27). Dissolution continues during sedimentation, and the specific surface area and solubility of biogenic silica decrease with depth (28). Once on the sea floor, the frustules will dissolve slowly, and the biogenic silica deposits (referred to as opal, or SiO2‚nH2O) may persist from a few months to a few hundred years at the sediment-water interface (20). Processes on and in the sediment are complex and involve interaction with Al(III) from dissolved detrital minerals, reducing the solubility and dissolution kinetics of biogenic silica and inducing precipitation of aluminosilicates (28, 30). The result of the various processes is an asymptotic decreasing concentration profile in marine sediments. The net removal of dissolved silicate from the water column also occurs in freshwaters, especially in lakes, but also behind impoundments or dams which has been termed the “artificial lake effect” (31). For example, 50% of inputs of dissolved silicate were removed or retained in three French reservoirs (32). The retention represented only a small fraction of the internal fluxes of phytoplankton uptake and recycling at the sediment-water interface. The dissolved silicate removal by the “Iron Gate” dam in the Danube River was described by Humborg et al. (33). Admiraal et al. (23) describe removal of dissolved silicate in man-made sedimentation areas in the Rhine Delta. Dissolved silicate concentration in rivers decreases strongly during bloom periods, up to a level where a limitation of diatoms by dissolved silicate may occur (34). Meybeck (35) reports a marked decrease in dissolved silicate concentrations in the river Seine basin during the period 1848-1990, coinciding with a marked increase of phosphorus and nitrogen levels and the creation of reservoirs in the 1960s. As the dissolved silicate flux to the river probably has not changed much in this period (see ref 36), the decrease is likely due to removal processes in the river basin. The importance for this study is that part of the dissolved silicate from waterglass emission is likely to be removed from the aquatic environment in sinks such as lakes, reservoirs and dammed river sections. Ecological Effects. Silicon is a naturally occurring substance, which is present in all environmental compartments and all biota (3) (see Table 2). It is an essential element for diatoms (algae) but not for other algae (such as green or bluegreen algae). The concentrations of dissolved silicate
relative to nitrogen and phosphorus, more specifically the ratios of Si:P and Si:N ratio, play a key role in the succession of phytoplankton species through the year. Diatoms dominate at high Si:P ratios and green algae (high N:P ratio) or bluegreen algae (low N:P ratio, high temperature) at low Si:P ratios (4). Schleyer and Blumberg (22) report a concentration of 0.6 mg/L of SiO2 as limiting for diatom growth, but the limiting concentration varies considerably between diatom species: from 0.17 to 4.3 mg/L of SiO2 (4). In temperate lakes, increasing temperature and light intensity in the spring will induce a diatom bloom. When the pool of dissolved silicate is depleted, other algae will dominate the phytoplankton until, in fall when mineralization processes have partly replenished the dissolved silicate a second diatom bloom may occur. Therefore, the effects of increased dissolved silicate concentration due to the emission of soluble silicates are expected to be (i) an extension of the spring (and fall) blooms of diatoms (which often ends when the dissolved silicate pool is depleted) and (ii) a possible reduction in summer green or bluegreen algae blooms (because a larger amount of phosphorus is used up in the spring bloom) (Table 3). This could lead to (iii) a shift in biomass production from summer to spring and fall, and, possibly, (iv) an overall increase of phytoplankton biomass over the year (when the increase in summer and fall bloom is larger than the decrease in summer density). Changes in phytoplankton species composition and changes in phytoplankton production (increase in total production and/or shifts in time) will have their effect on the entire aquatic foodweb, as primary production is the starting point for many foodchains (v). These effects, predicted from theoretical considerations, are difficult to verify with reported experimental and/or field studies, because enrichment of ecosystems with dissolved silicate (without other nutrients) is poorly studied. Several studies have addressed the eutrophication of coastal waters with N and P (i.e., decreasing instead of increasing Si:P and Si:N ratios). Billen and Garnier (36) modeled the impacts of human activities in a watershed causing changes in the delivery of nutrients in a hypothetical river system and their effect on coastal phytoplankton communities. In the pristine situation, diatoms are the dominant algae throughout the year. However, in scenarios with increased supplies of P and N, diatom dominance in early spring in spring is quickly followed by a dominance of nondiatoms during summer periods (36). Turner et al. (37) studied the effects of decreasing Si:N ratios on the Mississippi river continental shelf. They found a decreased fraction of copepods in the mesozooplankton at lower Si:N ratios and a lower transfer of primary production to zooplankton faecal pellets. Because diatomrich faecal pellets sink to the bottom faster, oxygen conVOL. 38, NO. 2, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
519
proposed as a measure to reduce the negative effects of sewage effluent discharge and reduced Si:P and Si:N ratios in marine ecosystems (39).
TABLE 4. Acute Toxicity of Waterglassa species
threshold value (EC50) (mg/L of SiO2)
references
algae crustaceans fish microorganisms
213 160-895 215-2320 10 - >1000
(41) (41, 42) (22, 41, 42) (41)
a Effects may well be due to increased pH of the test medium, see text.
sumption in the water column is lower, but oxygen consumption near the bottom is higher and more uniformly distributed at high Si:N ratios than at lower Si:N ratios (vi). The authors suggest that a decrease of Si:N ratios leads to a smaller risk of bottom water hypoxia but to a larger risk of flaggelated algal blooms, including harmful algae blooms. The potential effects of the relative decrease of dissolved silicate in coastal waters on the occurrence of harmful algal blooms and changes in zooplankton communities and the entire coastal ecosystem are also discussed in ref 33. The potential effects are large, but much more research is necessary to obtain a better understanding of the consequences of reductions in dissolved silicate. The ecological effects in rivers are less likely to be as important, because primary production is often light-limited due to the high turbidity (33). Toxicity from dissolved silicate from waterglass (vii) is not likely to occur in natural water systems. According to ecotoxicity data (Table 4), waterglass has been classified as “slightly toxic” to “very slightly toxic” (5). The toxicity varies with molar ratio and pH (1, 21). However, toxicity is not realistic for natural water systems, because, as discussed earlier, precipitation of amorphous silica will occur at a dissolved silicate concentration of ca. 95 mg/L of SiO2 (pH < 9; T ) 25 °C), and natural waters are greatly undersaturated with respect to amorphous silica. The observed effects in older toxicity tests (without a pH-buffered medium) may well be caused by the high pH of the test medium and not necessarily due to high waterglass concentrations. For example, the upper pH at which waterfleas (crustaceans) Chydorus sphaericus, Eurycercus lamellatus, Ceriodaphnia reticulata, Daphnia magna and D. pulex can survive is 10.511 with decreasing densities of daphnid species at pH > 9 observed in enclosure studies (38). The overall, long-term ecological effects of waterglass emissions are not known but are expected to be small. The total emission from Europe is
