Dissolved organic carbon measurement using a modified high-temperature combustion analyzer

Abstract

The performance of a dissolved organic carbon (DOC) analyzer operating on the principle of high-temperature combustion (HTC) is subject to numerous design and operation characteristics. Here we describe modifications and performance tests of a commercial HTC analyzer (MQ Scientific, model MQ-1001), many of which are applicable to other HTC instruments. Design improvements include a new combustion column, auto-sampler needle assembly, and a smaller sparger/water trap that automatically maintains constant water level and pH. Techniques for monitoring and compensating for carrier gas flow rate fluctuations, as well as electronic improvements to the auto-sampler advance control and the high-pressure injection gas pulse, are also described. A new model LICOR 7000, nondispersive IR (NDIR) detector is shown to provide a 50-fold increase in sensitivity over the previous LICOR 6252 model. The total blank for the modified instrument is initially ∼27 ng C and declines during use to ∼9 ng C as the combustion column ages. For an instrument with a well-conditioned combustion column, approximately half of this background is resolvable into a reagent component coming largely from the deionized, UV-irradiated (DUV) water used to rinse the sample onto the combustion column: the other half is intrinsic to the instrument and appears associated with the quartz bead packing. Injection of varying volumes of DUV water with and without an added constant C background, indicates that our DUV water contains between 2 and 9 μM DOC, depending on the variable performance of the water purifier. Similar experiments indicate that the intrinsic instrument blank is variable over time and depends complexly on both the wait time between individual water injections and the overall time that the combustion column has been conditioned. The modified instrument fitted with the new LICOR 7000 detector measures 46.1±1.3 μM DOC in Sargasso Sea water reference material (44–45 μM) against a total instrument blank equivalent to about a fourth of this value. Overall, the modified MQ-1001 analyzer is capable of dependable, automated analysis of relatively challenging deep seawater samples with an average accuracy of about ±3.8% of the consensus value.