Environ. Sci. Technol. 2004, 38, 4496-4502
An FTIR-DRIFT Study on River Sediment Particle Structure: Implications for Biofilm Dynamics and Pollutant Binding T O M G A L L EÄ , * , † B A R E N D V A N L A G E N , ‡ ANDREAS KURTENBACH,§ AND REINHARD BIERL§ Resource Centre for Environmental Technologies (CRTE), CRP-Henri Tudor, 66 rue de Luxembourg, BP 144 L-4002 Esch-sur-Alzette, Grand Duchy of Luxembourg, Laboratory of Soil Science and Geology, Wageningen Agricultural University, Wageningen, The Netherlands, and Department of Hydrology, Faculty of Geosciences, University of Trier, D-54286 Trier, Germany
Diffuse reflectance infrared Fourier transform (DRIFT) spectrometry was applied to a set of sediment samples collected by traps over one and a half years in a midmountainous river. Dynamic changes in hydrological and lifecycle conditions generated sediment particles of different Corg content and organic composition. Periods in the midst of or shortly after flood events left particles poor in Corg content with spectral features that were enriched in carboxylic and aromatic signals. These are characteristic of terrestrial oxidized vascular plant debris. Low-flow conditions saw the consequent build-up of amide, aliphatic, and polysaccharide moieties as expected for autochthonous biofilm derived material. A peak ratio of two bands representing the alternation of these two types of organic matter showed that flood particle Corg had a higher affinity for metals than the high Corg of mature biofilms, probably owing to higher COO- contents in the first. The relative dietary bioavailability of the metals from sediment Corg, which is related to the nutritional value of the substrate, is therefore probably lower in the aftermath of a flood than in prolonged low-flow situations. This needs to be accounted for in future metal speciation and bioavailability modeling approaches.
Introduction Cohesive sediment, consisting of particulate matter smaller than 63 µm, provides an important fraction of the transported sediment in most rivers (1). Trace metals of environmental relevance are adsorbed in different proportions to particulate matter in rivers, underlining the importance of particulate transport for metals such as Pb, Zn, or Cu (2, 3). Sequential extraction techniques as well as metal distribution modeling to different inorganic and organic phases demonstrated the impact of structural particulate matter composition on metal binding (4-6). Investigations on cohesive suspended sediments with microscopic techniques such as transmission electron mi* Corresponding author phone: +49 651 201 3074; fax: +49 651 201 3080; e-mail:
[email protected]. † Resource Centre for Environmental Technologies. ‡ Wageningen Agricultural University. § University of Trier. 4496
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 17, 2004
croscopy (TEM) and scanning confocal laser microscopy (SCLM) showed that this fraction is commonly transported in flocculated form (7). These observations led to a conceptual model of flocs as inverted biofilms consisting of living (bacteria, algae) and nonliving (detritus) biological material as well as inorganic constituents (clays, iron/manganese oxides), held together by extracellular polymeric substances (EPS)(8). Suspended and benthic organic particles constitute an important food source for aquatic animals, especially for filter- or deposit-feeding invertebrates in rivers (9). Dietary uptake has been shown to be a major bioaccumulation pathway for certain metals, and the assimilation efficiency from different floc constituents proved variable (10, 11). A myriad of methods to characterize the chemical nature of particulate matter in natural waters has been developed in the fields of sediment transport research and investigations on the carbon cycle. Biogeochemical characterizations often rely on very specific measurements such as chlorophyll a, δ13C, lignin-phenols, or fatty acids to cite but a few (12-14). Owing to their specificity, these methods are useful as molecular markers but fail to provide a quantitative carbon distribution within the particulate matter (15, 16). A more general organic characterization is possible with pyrolysisgas chromatography-mass spectrometry (Py-GC-MS), which has been successfully applied by several researchers to natural water particles (17-19). Nuclear magnetic resonance spectroscopy (NMR) can provide quantitative carbon distributions under certain conditions but its insensitivity makes measurements at low carbon contents time-consuming and less reliable (19, 20). Fourier transform infrared spectroscopy (FTIR) has rarely been used in water particle investigation, except as a complementary technique with little endeavor to interpret the spectra intensively (18, 20, 21). Nevertheless FTIR, especially in diffuse reflectance mode (DRIFT), offers some important advantages, such as the ability to assess both mineral and organic structures in particles, good sensitivity, and high throughput. The multiple layers of information featured in an infrared spectrum are a major challenge for the interpretation and quantification of the data. This applies particularly to the 1000-1600 cm-1 region, where many minor mineral peaks are hidden under a multitude of organic bands from the so-called fingerprint region (22). Furthermore, performing reproducible quantitative DRIFT measurements requires strict attention to experimental detail, especially to particle size distribution and packing density of the sample (23, 24). On the other hand, quality FTIR-spectra allow the extensive statistical treatment of the datasets both among each other and the inclusion of external data from other measurement techniques. Partial least squares (PLS) analysis has been applied to DRIFT spectra from a set of soil samples. They exhibited good correlation with mineralogical and elemental measurements (25). Here we present an investigative DRIFT study carried out on a set of 57 sediment samples collected by sediment traps on a weekly basis as part of a river sediment contamination monitoring campaign lasting one and a half years. The main focus rested on the ability of DRIFT to detect differences in sediment composition reflecting frequent dynamic changes in hydrological or life-cycle situations characteristic of river systems. The impact of these changes on metal binding by sediment particles was further investigated. Furthermore, the consequences for metal bioavailability and speciation modeling were discussed. 10.1021/es040005m CCC: $27.50
2004 American Chemical Society Published on Web 07/28/2004
Experimental Section Study Site. The Ruwer River drains a basin of 238 km2 situated in the northern Hunsru ¨ ck Mountains near the city of Trier, Germany. The bedrock is dominated by quartzite and Devonian schist. Consequently, the water chemistry is of low specific conductance (mean of 184 µS/cm) and water hardness (mean Ca concentration 12 mg/L; mean Mg concentration 5 mg/L). Land use in the catchment area is dominated by forest (57%) mainly in the southern part, while the sampling site is surrounded by steep vineyards, pastures, and urban areas. Sampling. Sediment traps were made of concrete slabs (30 × 30 × 5 cm) cast using egg cardboard moulds. These were designed to present a rough surface comparable to a consolidated natural river-bed. A set of traps was exposed over a period of one week in a shaded low-current part of the river. The sampling procedure consisted of washing the loose material off the trap with river water and collecting it in a glass bottle with a large funnel. Parallel samplings of natural sediments with a sediment grab in the vicinity of the trap showed no significant difference in Corg, N, and metal contents as well as DRIFT spectra (n ) 5). In the laboratory, sediments were wet-sieved to the fraction of
