Measurement of Dissolved Organic Carbon in Seawater: A Comparison between the High-Temperature Catalytic Oxidation Method and the Persulfate Oxidation Method

A high-temperature catalytic oxidation method (HTCO) has been developed to measure dissolved organic carbon (DOC) in natural seawater. The method employs Pt! A/203 as a catalyst to oxidize DOC at 680°C into C02 which is detected by a non­ dispersive IR detector. The detection limit of the method was estimated to be 29. 5 µMC on the basis of a sample volume of 66 µ/. The precision was generally better than 7% with respect to selected organics added to natural seawater. The detection limit and precision are expected to be lower and greater, respectively, if the in j ection volume is increased. The HTCO method is faster, more reliable, but higher in analytical cost than the persulfate oxidation method for the analysis of DOC in seawater samples. To obtain reliable values of DOC in a iand laboratory, the seawater sample should be filtered, acidified and stored at 4°C immediately after sampling. The HTCO method measured the higher values of DOC than did the persulfate method in shelf waters off northeast­ ern and southwestern Taiwan, indicating the incomplete oxidation of the persulfate method. The persulfate method, however, is still recommended for use along with the HTCO method because each method for anaiyzing DOC provides different geochemical information. The DOC value increases as salinity decreases in surface water off north­ eastern Taiwan reflecting the influence of terrestrial sources on the distribution of DOC. A vertical decrease of DOC with depth was also found at a station off southwestern Taiwan, indicating that there may be a relationship between DOC distribution biological turn over of organic carbon. The priorities extended from this study will focus on the spatial and temporal distributions of DOC and their geochemical significance reflected from the use of both methods for marginal seawater under the influence of the Kuroshio.


INTRODUCTION 165
Measurement of dissolved organic carbon in seawater has been made for over a half centul)', but no reliable method with direct and/or shipboard performance has been fully recognized yet (Toggweiler, 1990;Sharp, 1991). Previous results obtained from different Institute of Marine Geology, National Sun Yat-Sen University, Kaohsiung, Taiwan, R.O.C. methods revealed that the high-temperature combustion method has higher oxidation effi ciency and gives higher DOC values than do the chemical (eg. persulfate) and UV oxida tion methods (Menzel and Vaccaro, 1964;Armstrong et al., 1966;Gordon and Sutcliffe, 1973;Sharp, 1973;Gershey et al., 1979;Kumar et al., 1990;Romankevich and Ljutsarev, 1990). However, the high-temperature combustion method has not been widely applied in seawater analysis largely due to problems with contamination and poor repro ducibility (MacKinnon, 1978;Sharp, 1991).
Recently, Sugimura and Suzuki (1988) used the high-temperature catalytic oxidation method (HTCO) to measure total concentration of DOC in seawater by means of a Sumigraph N-200 instrument. The method employs Pt!A/203 as catalyst to oxidize dis solved organic matter at 680°C into C02 which was subsequently detected by a non dispersive IR. They reported much higher values in both surface and deep water than those previously measured. The results also indicated that the DOC concentration de creases strongly with depth and correlates inversely with apparent oxygen utilization and nitrate concentrations. These striking features suggest that DOC is actively involved in the biogeochemical cycles of carbon (Togweiller, 1989;Bacastow and Maier-Reimer, 1991).
The method, however, has not been fully confirmed because no identical results have been reproduced (Toggweiler, 1989;Sharp, 1991). Nevertheless, the HTCO is still the most attractive and potential method to date in measuring DOC in seawater because of its high efficiency in extraction of DOC from seawater.
On the other hand, the persulfate oxidation, which was commonly used to measure DOC previously, has been reported to be far less efficient in recovering DOC from seawater (Sharp, 1973;Sugimura and Suzuki, 1988). Sugimura and Suzuki (1988) also ascribed the persulfate-resistant fraction of DOC to the newly produced, high molecular weight organic compounds in the upper layer of the ocean. The findings may suggest that the simultaneous measurements of DOC with HTCO and persulfate methods are required in order to distinguish the phases of DOC. Nevertheless, the boundary of new and old phases of DOC may not be clear-cut if optimal conditions of persulfate oxidation, which were rarely mentioned before, are not rigorously set for seawater. Therefore, the analytical precision of both methods should be evaluated prior to application on the study of DOC distribution in the oceans.
Although a new set of commercialized HTCO instruments (Sumigraph model TOC-90) has been available recently ·and some DOC results have been revised (Suzuki et al., 1991 ), we were not inclined to employ this instrument for DOC analysis because of its inacceptably high international price and unknown reliability. Thus we employed the Shimadzu TOC analyzer with a HTCO device for DOC measurements in seawater. The principles of this instrument are basically similar to those of the Sumigraph analyzer, but the price is reasonable enough for us to aff ord it. The methodology and analytical proce dures of the method are addressed as well as the results obtained from the HTCO and persulfate I 100 ° C methods are compared and discussed.

a. Instruments
Analysis of DOC in seawater was established on a commercially available instrument with HTCO oxidation system (Shimadzu TOC 5000), and the results were compared with those from the persulfate/l 00°C oxidation system (0. I. Corporation, model 700). The HTCO system primarily consists of a combustion unit, a gas purification system, a non dispersive IR (NDIR) detector and a recording system. The. water sample was injected into an oxidation column of quartz material containing either normal catalyst (Pt-alumina) or high sensitivity catalyst (Pt-quartz wool), depending on the type of measurement. The column is housed in an electric furnace which is continuously heated at 680°C under the atmosphere of purified air. The generated C02 was carried by purified air to the NDIR detector, and the signal was digitally converted into carbon concentration according to the calibration of DOC standards. In case of the model 700 0. I. analyzer, the instrument is widely known in the oceanographic community and detailed description is not necessary.
Usually the seawater is directly injected into the system, and dissolved C02 is expelled from the acidified seawater by purging with nitrogen. The remaining DOC is then oxi dized by persulfate at I 00°C and the released C02 is measured by a NDIR detector .. Standard calibration should be made and recorded in the analyzer prior to analysis of seawater samples.

b. Calibration
A standard stock solution (1000 ppm C) was prepared for both methods by dissolving analytical-grade potassium hydrogen phthalate (Merck) in 0.4% HCI solution (vHC/:vQ H20), stored at 4°C in the dark. Working standard solutions were prepared by dilution of stock solution in 0.4% HCI solution. Standard stock and working standard solutions should be prepared weekly and daily, respectively. The injection volume of each sample for the HTCO method can be selected (100 µ/ max. for using normal catalyst) and oper ated by an automatic injector, thus the standard solution and seawater were injected with 66 µl throughout this experiment. Peak areas (integrated counts) of standards were least square fitted and the concentration of seawater sample was calculated from this calibra tion.
The calibration of persulfate method was made froi;n a single standard. The working standard (10 ppm) was run repeatly by introducing 0. 5 ml into the model 700 analyzer to react with 3 ml of persulfate for 13 min. until ihe detector response was constant; the blank-corrected scaling factor (µgC!m V) was used to calculate the DOC concentration in seawater.

c. Verification
The recovery of organic compounds spiked in natural seawater was examined to realize the precision of the HTCO method. The same experiment was also conducted for the persulfate/100°C oxidation method.

d. Analysis of DOC in seawater
Seawater samples were collected on board the R/V Ocean Researcher 1 with Niskin bottles (2.5 1) mounted on a Rosette sampler from a station (st. A) off southwestern Taiwan and stations (sts. 1-11) of a transect off northeastern Taiwan (Figure 1) during the periods of 20-26 August and 4-7 October, 1991, respectively. One liter of each sample was filtered through the pre-combusted (450°C, 4 h) glass-fiber filter (Whatman GF/F) which was mounted on a Swin-Lock filtration holder (Nucleopore). After discarding the first 800 ml of filtrate, a subsequent 100 ml of filtrate was transferred to the seawater pre- HCl and then stored at 4°C until analyses at a land-based laboratory. The determination of DOC in seawater by the HTCO method was carr ied out by purging 20 ml of filtered and acidified seawater in a Pyrex test tube with purified air for 5 min. to remove dissolved C02 '. The 66 µI of decarbonated seawater was injected into the oxidation column by means of an automatic syringe. Each sample was run for at least 5 repeated measure ments, and abnormal signals were deleted to ensure that the standard deviation of repeats was less than 2%. The averaged running time for a sample was estimated to be 12 min.
If DOC was determined by the persulfate method, a volume of 0.5 ml acidified and purged seawater was allowed to react with 3 ml of 10% (w/w) potassium persulfate for 13 min. At least three repeats, which usually last longer than 45 min., were performed for a sample ensuring that the standard deviation of repeats was less than 5 %.

a. Calibration
Calibration of the HTCO method was based on four-point (2, 4, 6 and 8 ppm C) standards. Output signals (integrated counts) of the four standards were perfactly linear with respect to carbon concentrations as shown in Figure 2. The intercept of regression line on the y-axis indicates the blank signal derived from the instrument (system) itself and the matrix of standard solution (acidified Q-H20). The matrix blank was not evaluated 39059 STANDARD I ppm directly because organic-carbon free Q-H20 was not available. The effect of different matrices on the DOC measurement was not thought significant, because a comparable method has been proved to be effective in the oxidation of DOC in natural seawater (Sugimura and Suzuki, 1988). The non-target gases such as C/2 and S02 produced from the combustion of seawater, which may affect the DOC detection, have been removed by a halogen scrubber before C02 entered the NDIR detector. The system blank, which may come from the carrier gas, catalysts (Pt-alumina) and carry-over (memory.) residue during consecutive injections, however, must be known because it is included in the measuring signals of the following natural seawater samples. In practical analysis, the signal of a seawater sample (including system blank) was converted to carbon concentration accord ing to the calibration of net response by shifting the y-intercepted calibration line to. pass the origin. The concentration of DOC in seawater sample, therefore, contains a part of carbon derived from the blank Furthermore, the memoiy blank, which may be a part of system blank, can stem from the incomplete combustion of organic carbon in the oxidation column and/or adsorption of carbon on the surface of catalyst. This cariy-over blank may not be consistently same throughout the measurements and this may result in the variabil ity of the system blank during the experiment. To evaluate this memoiy blank of HTCO analyzer, samples of seawater and Q-H20 were alternatively measured with five injections for each determination. From Table 1, we found that the signal of the first injection is usually but not always higher than those of the other injections for the Q-H20 sample. It should be mentioned that standard deviations of repeated measurements are generally greater than 2% because the data were presented as original counts, and no abnormal count has been deleted. The results indicate that the memoiy blank may not be neglected when the measurement is switched from the sample of relatively high carbon to the sample of relatively low carbon. In order to avoid the daily variability of system blank, we thus did not consider the value of first injection between samples to eliminate the possibil- inj.: injection number ity of memory effect for samples of natural seawater. Under these conditions, the system blank is estimated from the y-intercepts of zero volume extrapolated from regression lines between integrated counts and different volumes of injections for various solutions ( Figure  3). The greater sign als of solution C (2 ppmC in 3% NaCl) over solution B (2 ppmC in Q H20) are ascribed to be largely derived from the organic carbon contamination of NaCl solution rather than from the matrix effect, because the solution of KHP(2 ppm) prepared from pre-combusted NaCl (450°C, 6 hr) gave the less carbon signal than did the same concentration ·of KHP solution prepared from non-combusted NaCl. The averaged inte grated counts of the system blank were estimated at 762, which is equivalent to 11.69 ngC. This value is equal to be 14.75 µMC in the seawater volume of 66 µ!,which was considered as the minimum running system blank. It was always subtracted from DOC concentration of seawater samples converted from the origin-passing calibration. The blank of the persulfate method, which comes mainly from reagents, gas, tubing and digestion chamber, is dependent upon the introduced reagent volume. It was generally found to be around 10 m Von the basis of 3 ml of oxidant, which should be subtracted from the signals of standards and samples. The blank was roughly equivalent to 0.45 µgC.
This blank is subject to variability and should be checked prior to running standards and samples. The instrument uses µgC!mV as scaling factor to convert the measured voltages from samples to DOC concentrations. The linear range of the NDIR detector was reported up to 50 µgC (Operating Manual of 0. I. Instrument), which far exceeds the 5 µgC in the standard solution.

b. Verification of DOC measurements in seawater
The reliability of DOC measurements in seawater by means ofHTCO and persulfate/ I 00°C methods was examined from the recovery of various amounts of organic com pounds (KHP, EDT A, oxalic acid and caffeine) spiked in the offshore seawater of south western Taiwan. Recovery of added compounds by the HTCO method was found to be nearly 100% for most measurements, with the exception of EDTA where the uncertainty was up to 7% for a added concentration. The precision is, therefore, believed to be better than 7% with respect to added compounds in natural seawater ( Table 3).
As far as the oxidization efficiency of the persulfate/100°C method is concerned, it is dependent on the chloride content of sample, oxidant amount and reaction time. The analytical conditions, which can completely oxidize KHI' in Q-H20, did not oxidize KHP efficiently in NaCl solutions of various chlorinity (Figure 4). Experimental results also showed that increases of oxidant amount and reaction time can greatly enhance the recov ery of KHP in NaCl (3.5%) solutions, thus an operational condition (seawater: 0.5 ml; persulfate (10% ): 3 ml; reaction time: 13 min.) was set for measuring DOC in natural seawater. However, the optimal condition found from synthetic solutions may not be also optimal for DOC analyses in natural seawater. The data in Table 3 show that the added organic compounds in natural seawater were not totally recovered, indicating the incom plete oxidation of DOC in natural seawater by the persulfate oxidation.

c. Detection limits of methods
The system blank of the HTCO method was estimated to be 11.69 ngC. The detection limit of the instrument (35.07 ngC) is defi ned as a three-fold intensity of system blank This signal is equal to that produced from a seawater sample containing 23.38 ngC because the system blank (11.69 ngC) is always present and added to any determination. The concentration of detection limit in seawater is therefore equivalent to 29.5 µMC on the basis of an injection volume of 66 µ/ (Table 3). This value is relatively smaller than the lowest concentration of DOC found from the deep open ocean (Sugimura and Suzuki, 1988). Obviously, the method is suitable for the measurement of DOC in the ocean.
Because the signal of system blank was assumed to be independent of injection volume, * Detection limit of the instrument is defined as three-fold intensity of the system noises.
# Detection limit in seawater is about one half less than that of the instrument (see explanation in text) and the concentration is calculated on the basis of injection volume, 66 µl and 0.5 ml, respectively, for the HTCO and persulfate methods.  the increase in injection volume will decrease the concentration of blank in solution, which · will result in a decrease of the detection limit. However, this increase of injection volume will shorten the life of catalysts which should be replaced more frequently.
On the other hand, we did not triple the total blank of the persulfate method as the detection limit of the method because the blank was quite high (0.45 µgC), the major fraction of which was derived from reagents rather than from system noises. Meanwhile, the total blank was initially recorded in the instrument and signals of standards and samples were then automatically offset by this value before they were used for calibration or concentration derivation. Therefore, we still use the triple intensity of system noises as the detection limit of the instrument, although the signal has been blanketed by the reagent blank (personal communication with Dr. S. R. Huang, professor of analytical chemistty, National Sun Yat-sen Univ.). It should be kept in mind, however, the concentration of DOC in seawater, which can be reliably measured, may be greater than the detection limit because the effect of a relatively high reagent blank.
d. Analysis of DOC in seawater Sugimura and Suzuki (1988) Table 4