Total carbon analysis may overestimate organic carbon content of fresh waters in the presence of high dissolved inorganic carbon

Automated carbon analyzers often are configured to provide estimates of both total organic carbon (TOC) and nonpurgeable organic carbon (NPOC). We show there can be an overestimation of total carbon in the presence of moderate to large quantities of dissolved inorganic carbon. This leads to overestimates of TOC, which is measured as the difference between total carbon and inorganic carbon. Water samples were analyzed as both TOC and NPOC on a Shimadzu TC 5050 Carbon Analyzer. The difference between TOC and NPOC increased as a function of concentrations of dissolved inorganic carbon (DIC). Water samples spiked with DIC ranging from 0 to 100 mg DIC/L also reported increased TOC as large as 8 mg C/L. Our data suggest that the Shimadzu 5050 analyzer (and by analogy other instruments that estimate TOC by difference between TC and IC) overestimates total carbon (TC) when calibrated with an organic standard as recommended by the manufacturer. The magnitude of the overestimation varies both with the amount of DIC present in the sample and the extent to which measurement efficiency of the analyzer is less than 100%. The consequences will be most severe in analysis of samples from systems spanning a large range in DIC. Time series from individual systems are less likely to be affected because the necessary large change in DIC would be detected as changes in pH or other attributes well before any change in DOC. Systems with high DIC will, however, be susceptible to even small variations in measurement efficiency. *Corresponding author: E-mail: [email protected] Acknowledgments This research was supported by several grants from the Hudson River Foundation and the National Science Foundation (DEB 0414262, DEB 0423476, DEB 0454001, DBI-0521091, and DBI 0640300) and the US EPA (Connecticut River Airshed Watershed Consortium). DOI 10:4319/lom.2010.8.196 Limnol. Oceanogr.: Methods 8, 2010, 196–201 © 2010, by the American Society of Limnology and Oceanography, Inc. LIMNOLOGY and OCEANOGRAPHY: METHODS Findlay et al. Overestimation of total carbon in freshwater 197 Over the past several decades, methods for analyses of dissolved carbon pools in surface waters have shifted from benchtop wet chemistry to a variety of automated instruments. The instruments use some combination of acidification/sparging to remove inorganic carbon and catalyzed oxidation to convert organic carbon to CO2 before detection with infrared absorbance or gas chromatography. Many automated carbon analyzers determine organic carbon in a filtered water sample by two approaches. For instance, the Shimadzu® series instruments acidify and sparge a sample before analysis and any remaining carbon is then oxidized, detected, and is defined to be nonpurgeable organic carbon (NPOC). This approach is analogous to DOC measured by wet oxidation methods (e.g., McDowell et al. 1987). The alternate approach with the Shimadzu instruments is analysis for total carbon (TC) where an aliquot of the sample is oxidized, acidified, and sparged. The CO2 measured represents the total produced from both dissolved inorganic carbon (DIC) and oxidized organic carbon (OC) (Fig. 1). A separate aliquot of the sample is then acidified and sparged without oxidation to measure the CO2 derived solely from DIC, typically termed IC by instrument manufacturers. The difference between TC and IC is then taken as a measure of the CO2 produced from total DOC, including both the nonvolatile and volatile components (Fig. 1), and this quantity is referred to as total organic carbon (TOC). The difference between TOC and NPOC is, in theory, VOC lost to the atmosphere during the open, acidified sparging step of the NPOC analysis. (In this paper, we refer to the difference in carbon between TOC and NPOC as “apparent” VOC [abbreviated aVOC]). Compounds lost to the atmosphere during this sparging would not necessarily be volatile under environmental conditions, since the acidification step lowers the pH to around 2, much lower than the vast majority of natural waters. Some instruments allow for a separate analysis and capture of CO2 derived from both IC and DOC, which is useful in cases where both products are intended for further analyses such as determination of stable isotopes (see St-Jean 2003). Over the course of many measurements of dissolved carbon pools from several lakes, streams, and one larger river, we have noted surprisingly large differences between TOC and NPOC. As we show below, this difference is largely (if not completely), a consequence of an overestimation of the TC term; we suggest that this “estimate” of aVOC should not be construed as a real pool of carbon. In this article, we quantify the potential effect of DIC concentration on estimation of the total organic carbon pool during automated analyses and describe how measurement efficiency affects the magnitude of this DIC effect. We show the magnitude of aVOC for freshwater systems ranging from oligotrophic lakes to urban streams. Materials and procedures Sites—Data on TOC and NPOC were derived from six groups of sites that spanned urban to rural land covers, and Michigan to Baltimore, MD, USA. The groups included: (1) multiple locations in the tidal freshwater Hudson River, NY, USA (abbreviated HR); (2) streams draining suburban and urban watersheds associated with the Baltimore Ecosystem Study Long Term Ecological Research (BES LTER) site, MD, USA; (3) a set of 16 small (second to fourth order) streams in Dutchess County, NY, USA (DUT); (4) multiple locations in the Walkill River drainage, NY, USA (WAL); (5) a set of 15 lakes located at the University of Notre Dame Environmental Research Center in northern Michigan, USA (CAS), and (6) multiple tributaries that flow into Lake Sunapee in NH, USA (SUN). Timing of collection ranged from multiple years and seasons for the Hudson River to a single summer sampling of the Walkill River. Analysis—Samples were analyzed on a Shimadzu 5050 TC Analyzer at the Cary Institute of Ecosystem Studies, Millbrook, NY, USA. All water samples were filtered (either Gelman 934AH or Whatman GF/F) and stored cold before carbon analysis. With each analytical run of the Shimadzu, there were three interspersed sets of organic carbon standards (potassium phthalate, 2.5, 5, 10, and 20 mg C L–1) and blanks of Nanopure water. Samples were first run in Total Organic Carbon (TCIC) mode and then reanalyzed with the instrument in NPOC mode. For measurement of both TC and NPOC, the phthalate standard curve was used. For measurement of IC, a mixed carbonate/bicarbonate inorganic carbon standard was used, as recommended by the manufacturer. Several checks on Carbon Analyzer performance were suggested by large differences between TOC and NPOC. We analyzed standards containing only inorganic carbon as well as additions of inorganic carbon to natural water samples or phthalate standards. Fig. 1. Flow diagram of analytical scheme for Shimadzu 5050 Carbon Analyzer. In Total Carbon mode, the instrument estimates total organic carbon (TOC) as the difference between Total Carbon and Inorganic Carbon. In NPOC mode, IC is sparged from an acidified sample before the sample enters the instrument. The key difference is that the acidified sparging occurs in the sample train in the Total Carbon mode versus open to the atmosphere in the NPOC mode. Findlay et al. Overestimation of total carbon in freshwater 198 For most data presentations, we show only the mean for each freshwater system rather than all the data because the number of observations varies greatly among systems (see legend for Fig. 2). Model 2 regressions were calculated using the Geometric Mean approach described by Prepas in Downing and Rigler (1984).