Environ. Sci. Technol. 2009, 43, 8604–8609
Measuring the Photochemical Production of Carbon Dioxide from Marine Dissolved Organic Matter by Pool Isotope Exchange W E I W A N G , * ,† C A R L G . J O H N S O N , † KAZUHIKO TAKEDA,‡ AND O L I V E R C . Z A F I R I O U * ,† Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543, and Hiroshima University, 1-7-1 Kagamiyama, Higashi-Hiroshima, 739-8521, Japan
Received May 27, 2009. Revised manuscript received September 25, 2009. Accepted October 2, 2009.
CO2 is the major known product of solar photolysis of marine dissolved organic matter (DOM). Measuring the rate of this globally significant process is hindered by low rates per unit volume, high background CO2 in seawater, and ubiquitous contamination. Current methods utilize CO2-free seawater matrices, possibly introducing artifacts. Alternatively, pool isotope exchange (PIE) replaces most of the sample’s DI12C with DI13C at natural pH and temperature, so that 12CO2 from DOM photooxidation elevates 12 CO2/13CO2 ratios in irradiated samples compared to dark controls. 12CO2/13CO2 ratios are then measured using a modified GC-IRMS. The minimum detectable concentration change (three standard deviations) is 300 nmol DI12C/kg. Methods for minimizing contamination while exchanging, transferring, sealing, and irradiating samples, and for recovering and purifying CO2 are presented. Results from PIE agree within uncertainties with those from CO2-free coastal seawater, suggesting that both methods apply to river-dominated coastal waters. However, photooxidation in the open ocean, which likely dominates the global flux despite lower rates per unit volume, involves DOM that differs from coastal DOM, so that coastal agreement cannot validate open-ocean studies. Major advantages of PIE are use of nearly unperturbed seawater matrices, potential to incubate samples in situ to obtain depth-integrated rates directly, and potential to use larger samples to measure openocean waters.
Introduction Are there significant effects of photochemical production of CO2 from dissolved organic matter (DOM) on the oceanic CO2 cycle or on the fate of marine DOM (1)? In seawater, analytical difficulties have thus far prevented direct measurements of the rates in unmodified seawater. Comparisons of current estimates of CO2 photoproduction with other major terms in the oceanic carbon cycle (2, 3) are shown in Figure 1. These CO2 flux estimates are based on various combinations of data, none direct, as summarized by White and coauthors (4). Shown in red are three estimates utilizing * Address correspondence to either author. Phone: 508-289-2342 (O.C.Z.). Fax: 508-457-2164 (W.W.; O.C.Z.). E-mail: wwang@ whoi.edu (W.W.);
[email protected] (O.C.Z.). † Woods Hole Oceanographic Institution. ‡ Hiroshima University. 8604
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 22, 2009
FIGURE 1. Estimations of carbon fluxes. Annual oceanic sources and sinks (2, 3) are shown in black bars. Estimates of photoproduction of DIC from terrestrial or estuarine organic matter, scaled globally using various global marine CO flux estimates, are shown in reds. Shown in blue colors are DIC photoproduction from marine organic matter in a modified seawater matrix (“pool depletion”, see text), scaled globally by optical modeling (cyan) (5-7) or by recent, low marine CO photoproduction estimates (7, 8). measurements of CO2 photoproduction in terrestrial DOM samples, extrapolated to the ocean assuming a CO2:CO production ratio of 20:1. Their large discrepancy is due to widely varying estimates of the global marine CO flux (4). The estimates shown in blue utilize measurements of CO2 photoproduction in marine DOM samples in altered seawater matrices (5, 6). The larger value extrapolates the data globally using an optical model that involves integrating photoproduction over solar wavelengths (see calculation method in ref 7), an approach requiring assumptions about DOM photodynamics and underwater light fields. The smaller estimate is based on the same CO2 photoproduction data, scaled to recent, relatively low marine CO flux estimates (7, 8). Though quite variable, these estimates all suggest that CO2 photoproduction rates are comparable to other CO2 cycle terms (Figure 1) and therefore significant. However, “terrestrial DOM” estimates implicitly assume that oceanic and terrestrial DOM react identically. Paradoxically, these estimates yield CO2 photoproduction rates exceeding the rate of input of riverine DOM to the sea, 0.25 Pg C y-1 (3). They are thus either overestimates, or else marine-derived or atmospherically derived DOM are also major CO2 source(s). Difficulties in measuring these rates directly arise from low expected rates, 98.8% of the DI12C in seawater for DI13C at near-ambient pH and ΣCO2. Samples are incubated in light or dark and then their ΣCO2 is sparged out, dried, and sealed. The CO2 isotopic ratio is then analyzed by gas chromatography-isotope ratio mass spectrometry (GC-IRMS). Since photochemically formed CO2 derives from isotopically natural, nonexchangeable DOM, CO2 photoproduction rates can be calculated from the 12C/13C ratios of irradiated (light) and dark-control (dark) samples. We also compare PIE results with those from PD.
Experimental Section Extensive schematic diagrams, procedures, photographs of specialized apparatus, and procedural explanations are in the Supporting Information (SI). Seawater. Samples were collected in 2003 near Block Island on the U.S. east coast. Related data (4) are location, 41.04°N, 71.08°W; depth, 2 m; salinity, 32 psu. Water was
immediately 0.2 µm in-line filtered to greatly minimize bacteria, giving filtrate with pH, 8.1; aCDOM(330 nm), 0.506 m-1; spectral slope coefficient, 0.0175 nm-1; ΣCO2, 2.13 mmol/ kg (4). This water was then stored in borosilicate glass or fluorinated polyethylene jugs in darkness at ∼4 °C for ∼3 years, and refiltered (0.2 µm Nucleopore polycarbonate) before use. Exchange. In this step as much DI12C as feasible is replaced with DI13C at nearly constant pH and ΣCO2 in a continuously purged airtight system that has been cleaned, assembled, and leak-tested (see SI). Mass flow controllers (Hastings) provide an artificial air mixture containing ∼400 ppm 13CO2 that enters a sample exchange flask (∼500 cm3) at ∼140 cm3 STP min-1, then passes through a 16-port Peek manifold, through 16 quartz tubes (9 mm o.d.) with constrictions (∼1.5 mm i.d.) about 3 cm from the top, finally venting into the room through water bubbler seals functioning also as gas flow indicators. The gas mixture is prepared by mixing helium (UHP 5.0, Praxair), a preformed mixture of 1% CO2 (99.98% 13 C, Cambridge Isotope Laboratories) in UHP He, and 23% O2 (UHP 4.3, Praxair). Seawater (∼65 cm3) is transferred by a sterile syringe into the flask. Gently shaking the flask (∼180 rpm, Rock’n-Roller, Lab Mart) facilitates gas exchange without creating bubbles or foam. Surrounding the flask with a tap water cooling coil counteracts shaker-induced warming to prevent condensate forming and blocking the capillaries. The ΣCO2 exchange half-life is ∼150 min. We exchange for 36-48 h, approaching equilibrium closely (∼98.9% DI13C). During exchange, sample tubes have been extensively flushed by effluent gas close to equilibrium with samples. Note that any trace CO2 contamination before sample transfer is unimportant. After lowering a dip-tube that serves alternately as gas-out or sample-out path, exchanged seawater is distributed in parallel through silica capillaries into tubes, entering below their sealing constrictions that remain dry (SI). Gas flow is re-established after sample transfer, and the sample tubes are sealed with a H2/O2 torch using transfer/ sealing protocols (SI) to (1) clear water from the capillaries after the sample transfer, (2) maintain flushing gas flows over unsealed samples, (3) minimize contamination by CO2 outgassing from heated silica, and (4) delay any shattering of capillaries due to crystobalite formation. The resulting pressure inside the sealed tubes is ∼0.93 atm, with final pO2 ∼0.21 atm (SI). Although direct tests of the transfer/sealing blank and precision are lacking, there is no systematic trend in sample isotope ratio as a function of the order of sealing or of postincubation CO2 recovery, implying that the errors from these steps are random rather than cumulative. Sealed-Tube Integrity. Five sealed sample tubes were immersed to 200 m in the ocean on a R/V Cape Hatteras cruise. All survived intact. Since depth-integrated photoproduction measurements require 0.1-0.2 >0.03,
